qemu-e2k/block.c

8245 lines
242 KiB
C
Raw Normal View History

/*
* QEMU System Emulator block driver
*
* Copyright (c) 2003 Fabrice Bellard
* Copyright (c) 2020 Virtuozzo International GmbH.
*
* Permission is hereby granted, free of charge, to any person obtaining a copy
* of this software and associated documentation files (the "Software"), to deal
* in the Software without restriction, including without limitation the rights
* to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
* copies of the Software, and to permit persons to whom the Software is
* furnished to do so, subject to the following conditions:
*
* The above copyright notice and this permission notice shall be included in
* all copies or substantial portions of the Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
* THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
* OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
* THE SOFTWARE.
*/
#include "qemu/osdep.h"
#include "block/trace.h"
#include "block/block_int.h"
#include "block/blockjob.h"
#include "block/fuse.h"
#include "block/nbd.h"
#include "block/qdict.h"
#include "qemu/error-report.h"
#include "block/module_block.h"
#include "qemu/main-loop.h"
#include "qemu/module.h"
#include "qapi/error.h"
#include "qapi/qmp/qdict.h"
#include "qapi/qmp/qjson.h"
#include "qapi/qmp/qnull.h"
#include "qapi/qmp/qstring.h"
#include "qapi/qobject-output-visitor.h"
#include "qapi/qapi-visit-block-core.h"
#include "sysemu/block-backend.h"
#include "qemu/notify.h"
#include "qemu/option.h"
#include "qemu/coroutine.h"
#include "block/qapi.h"
#include "qemu/timer.h"
#include "qemu/cutils.h"
#include "qemu/id.h"
block: block-status cache for data regions As we have attempted before (https://lists.gnu.org/archive/html/qemu-devel/2019-01/msg06451.html, "file-posix: Cache lseek result for data regions"; https://lists.nongnu.org/archive/html/qemu-block/2021-02/msg00934.html, "file-posix: Cache next hole"), this patch seeks to reduce the number of SEEK_DATA/HOLE operations the file-posix driver has to perform. The main difference is that this time it is implemented as part of the general block layer code. The problem we face is that on some filesystems or in some circumstances, SEEK_DATA/HOLE is unreasonably slow. Given the implementation is outside of qemu, there is little we can do about its performance. We have already introduced the want_zero parameter to bdrv_co_block_status() to reduce the number of SEEK_DATA/HOLE calls unless we really want zero information; but sometimes we do want that information, because for files that consist largely of zero areas, special-casing those areas can give large performance boosts. So the real problem is with files that consist largely of data, so that inquiring the block status does not gain us much performance, but where such an inquiry itself takes a lot of time. To address this, we want to cache data regions. Most of the time, when bad performance is reported, it is in places where the image is iterated over from start to end (qemu-img convert or the mirror job), so a simple yet effective solution is to cache only the current data region. (Note that only caching data regions but not zero regions means that returning false information from the cache is not catastrophic: Treating zeroes as data is fine. While we try to invalidate the cache on zero writes and discards, such incongruences may still occur when there are other processes writing to the image.) We only use the cache for nodes without children (i.e. protocol nodes), because that is where the problem is: Drivers that rely on block-status implementations outside of qemu (e.g. SEEK_DATA/HOLE). Resolves: https://gitlab.com/qemu-project/qemu/-/issues/307 Signed-off-by: Hanna Reitz <hreitz@redhat.com> Message-Id: <20210812084148.14458-3-hreitz@redhat.com> Reviewed-by: Eric Blake <eblake@redhat.com> Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> [hreitz: Added `local_file == bs` assertion, as suggested by Vladimir] Signed-off-by: Hanna Reitz <hreitz@redhat.com>
2021-08-12 10:41:44 +02:00
#include "qemu/range.h"
#include "qemu/rcu.h"
#include "block/coroutines.h"
#ifdef CONFIG_BSD
#include <sys/ioctl.h>
#include <sys/queue.h>
#if defined(HAVE_SYS_DISK_H)
#include <sys/disk.h>
#endif
#endif
#ifdef _WIN32
#include <windows.h>
#endif
#define NOT_DONE 0x7fffffff /* used while emulated sync operation in progress */
/* Protected by BQL */
static QTAILQ_HEAD(, BlockDriverState) graph_bdrv_states =
QTAILQ_HEAD_INITIALIZER(graph_bdrv_states);
/* Protected by BQL */
static QTAILQ_HEAD(, BlockDriverState) all_bdrv_states =
QTAILQ_HEAD_INITIALIZER(all_bdrv_states);
/* Protected by BQL */
static QLIST_HEAD(, BlockDriver) bdrv_drivers =
QLIST_HEAD_INITIALIZER(bdrv_drivers);
static BlockDriverState *bdrv_open_inherit(const char *filename,
const char *reference,
QDict *options, int flags,
BlockDriverState *parent,
const BdrvChildClass *child_class,
BdrvChildRole child_role,
Error **errp);
static bool bdrv_recurse_has_child(BlockDriverState *bs,
BlockDriverState *child);
static void bdrv_child_free(BdrvChild *child);
static void bdrv_replace_child_noperm(BdrvChild **child,
BlockDriverState *new_bs,
bool free_empty_child);
static void bdrv_remove_file_or_backing_child(BlockDriverState *bs,
BdrvChild *child,
Transaction *tran);
static void bdrv_remove_filter_or_cow_child(BlockDriverState *bs,
Transaction *tran);
static int bdrv_reopen_prepare(BDRVReopenState *reopen_state,
BlockReopenQueue *queue,
Transaction *change_child_tran, Error **errp);
static void bdrv_reopen_commit(BDRVReopenState *reopen_state);
static void bdrv_reopen_abort(BDRVReopenState *reopen_state);
static bool bdrv_backing_overridden(BlockDriverState *bs);
/* If non-zero, use only whitelisted block drivers */
static int use_bdrv_whitelist;
#ifdef _WIN32
static int is_windows_drive_prefix(const char *filename)
{
return (((filename[0] >= 'a' && filename[0] <= 'z') ||
(filename[0] >= 'A' && filename[0] <= 'Z')) &&
filename[1] == ':');
}
int is_windows_drive(const char *filename)
{
if (is_windows_drive_prefix(filename) &&
filename[2] == '\0')
return 1;
if (strstart(filename, "\\\\.\\", NULL) ||
strstart(filename, "//./", NULL))
return 1;
return 0;
}
#endif
size_t bdrv_opt_mem_align(BlockDriverState *bs)
{
if (!bs || !bs->drv) {
block: align bounce buffers to page The following sequence int fd = open(argv[1], O_RDWR | O_CREAT | O_DIRECT, 0644); for (i = 0; i < 100000; i++) write(fd, buf, 4096); performs 5% better if buf is aligned to 4096 bytes. The difference is quite reliable. On the other hand we do not want at the moment to enforce bounce buffering if guest request is aligned to 512 bytes. The patch changes default bounce buffer optimal alignment to MAX(page size, 4k). 4k is chosen as maximal known sector size on real HDD. The justification of the performance improve is quite interesting. From the kernel point of view each request to the disk was split by two. This could be seen by blktrace like this: 9,0 11 1 0.000000000 11151 Q WS 312737792 + 1023 [qemu-img] 9,0 11 2 0.000007938 11151 Q WS 312738815 + 8 [qemu-img] 9,0 11 3 0.000030735 11151 Q WS 312738823 + 1016 [qemu-img] 9,0 11 4 0.000032482 11151 Q WS 312739839 + 8 [qemu-img] 9,0 11 5 0.000041379 11151 Q WS 312739847 + 1016 [qemu-img] 9,0 11 6 0.000042818 11151 Q WS 312740863 + 8 [qemu-img] 9,0 11 7 0.000051236 11151 Q WS 312740871 + 1017 [qemu-img] 9,0 5 1 0.169071519 11151 Q WS 312741888 + 1023 [qemu-img] After the patch the pattern becomes normal: 9,0 6 1 0.000000000 12422 Q WS 314834944 + 1024 [qemu-img] 9,0 6 2 0.000038527 12422 Q WS 314835968 + 1024 [qemu-img] 9,0 6 3 0.000072849 12422 Q WS 314836992 + 1024 [qemu-img] 9,0 6 4 0.000106276 12422 Q WS 314838016 + 1024 [qemu-img] and the amount of requests sent to disk (could be calculated counting number of lines in the output of blktrace) is reduced about 2 times. Both qemu-img and qemu-io are affected while qemu-kvm is not. The guest does his job well and real requests comes properly aligned (to page). Signed-off-by: Denis V. Lunev <den@openvz.org> Reviewed-by: Kevin Wolf <kwolf@redhat.com> Message-id: 1431441056-26198-3-git-send-email-den@openvz.org CC: Paolo Bonzini <pbonzini@redhat.com> CC: Kevin Wolf <kwolf@redhat.com> CC: Stefan Hajnoczi <stefanha@redhat.com> Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
2015-05-12 16:30:56 +02:00
/* page size or 4k (hdd sector size) should be on the safe side */
return MAX(4096, qemu_real_host_page_size());
}
IO_CODE();
return bs->bl.opt_mem_alignment;
}
size_t bdrv_min_mem_align(BlockDriverState *bs)
{
if (!bs || !bs->drv) {
block: align bounce buffers to page The following sequence int fd = open(argv[1], O_RDWR | O_CREAT | O_DIRECT, 0644); for (i = 0; i < 100000; i++) write(fd, buf, 4096); performs 5% better if buf is aligned to 4096 bytes. The difference is quite reliable. On the other hand we do not want at the moment to enforce bounce buffering if guest request is aligned to 512 bytes. The patch changes default bounce buffer optimal alignment to MAX(page size, 4k). 4k is chosen as maximal known sector size on real HDD. The justification of the performance improve is quite interesting. From the kernel point of view each request to the disk was split by two. This could be seen by blktrace like this: 9,0 11 1 0.000000000 11151 Q WS 312737792 + 1023 [qemu-img] 9,0 11 2 0.000007938 11151 Q WS 312738815 + 8 [qemu-img] 9,0 11 3 0.000030735 11151 Q WS 312738823 + 1016 [qemu-img] 9,0 11 4 0.000032482 11151 Q WS 312739839 + 8 [qemu-img] 9,0 11 5 0.000041379 11151 Q WS 312739847 + 1016 [qemu-img] 9,0 11 6 0.000042818 11151 Q WS 312740863 + 8 [qemu-img] 9,0 11 7 0.000051236 11151 Q WS 312740871 + 1017 [qemu-img] 9,0 5 1 0.169071519 11151 Q WS 312741888 + 1023 [qemu-img] After the patch the pattern becomes normal: 9,0 6 1 0.000000000 12422 Q WS 314834944 + 1024 [qemu-img] 9,0 6 2 0.000038527 12422 Q WS 314835968 + 1024 [qemu-img] 9,0 6 3 0.000072849 12422 Q WS 314836992 + 1024 [qemu-img] 9,0 6 4 0.000106276 12422 Q WS 314838016 + 1024 [qemu-img] and the amount of requests sent to disk (could be calculated counting number of lines in the output of blktrace) is reduced about 2 times. Both qemu-img and qemu-io are affected while qemu-kvm is not. The guest does his job well and real requests comes properly aligned (to page). Signed-off-by: Denis V. Lunev <den@openvz.org> Reviewed-by: Kevin Wolf <kwolf@redhat.com> Message-id: 1431441056-26198-3-git-send-email-den@openvz.org CC: Paolo Bonzini <pbonzini@redhat.com> CC: Kevin Wolf <kwolf@redhat.com> CC: Stefan Hajnoczi <stefanha@redhat.com> Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
2015-05-12 16:30:56 +02:00
/* page size or 4k (hdd sector size) should be on the safe side */
return MAX(4096, qemu_real_host_page_size());
}
IO_CODE();
return bs->bl.min_mem_alignment;
}
/* check if the path starts with "<protocol>:" */
int path_has_protocol(const char *path)
{
const char *p;
#ifdef _WIN32
if (is_windows_drive(path) ||
is_windows_drive_prefix(path)) {
return 0;
}
p = path + strcspn(path, ":/\\");
#else
p = path + strcspn(path, ":/");
#endif
return *p == ':';
}
int path_is_absolute(const char *path)
{
#ifdef _WIN32
/* specific case for names like: "\\.\d:" */
if (is_windows_drive(path) || is_windows_drive_prefix(path)) {
return 1;
}
return (*path == '/' || *path == '\\');
#else
return (*path == '/');
#endif
}
/* if filename is absolute, just return its duplicate. Otherwise, build a
path to it by considering it is relative to base_path. URL are
supported. */
char *path_combine(const char *base_path, const char *filename)
{
const char *protocol_stripped = NULL;
const char *p, *p1;
char *result;
int len;
if (path_is_absolute(filename)) {
return g_strdup(filename);
}
if (path_has_protocol(base_path)) {
protocol_stripped = strchr(base_path, ':');
if (protocol_stripped) {
protocol_stripped++;
}
}
p = protocol_stripped ?: base_path;
p1 = strrchr(base_path, '/');
#ifdef _WIN32
{
const char *p2;
p2 = strrchr(base_path, '\\');
if (!p1 || p2 > p1) {
p1 = p2;
}
}
#endif
if (p1) {
p1++;
} else {
p1 = base_path;
}
if (p1 > p) {
p = p1;
}
len = p - base_path;
result = g_malloc(len + strlen(filename) + 1);
memcpy(result, base_path, len);
strcpy(result + len, filename);
return result;
}
2017-05-22 21:52:16 +02:00
/*
* Helper function for bdrv_parse_filename() implementations to remove optional
* protocol prefixes (especially "file:") from a filename and for putting the
* stripped filename into the options QDict if there is such a prefix.
*/
void bdrv_parse_filename_strip_prefix(const char *filename, const char *prefix,
QDict *options)
{
if (strstart(filename, prefix, &filename)) {
/* Stripping the explicit protocol prefix may result in a protocol
* prefix being (wrongly) detected (if the filename contains a colon) */
if (path_has_protocol(filename)) {
GString *fat_filename;
2017-05-22 21:52:16 +02:00
/* This means there is some colon before the first slash; therefore,
* this cannot be an absolute path */
assert(!path_is_absolute(filename));
/* And we can thus fix the protocol detection issue by prefixing it
* by "./" */
fat_filename = g_string_new("./");
g_string_append(fat_filename, filename);
2017-05-22 21:52:16 +02:00
assert(!path_has_protocol(fat_filename->str));
2017-05-22 21:52:16 +02:00
qdict_put(options, "filename",
qstring_from_gstring(fat_filename));
2017-05-22 21:52:16 +02:00
} else {
/* If no protocol prefix was detected, we can use the shortened
* filename as-is */
qdict_put_str(options, "filename", filename);
}
}
}
/* Returns whether the image file is opened as read-only. Note that this can
* return false and writing to the image file is still not possible because the
* image is inactivated. */
bool bdrv_is_read_only(BlockDriverState *bs)
{
IO_CODE();
return !(bs->open_flags & BDRV_O_RDWR);
}
int bdrv_can_set_read_only(BlockDriverState *bs, bool read_only,
bool ignore_allow_rdw, Error **errp)
{
IO_CODE();
/* Do not set read_only if copy_on_read is enabled */
if (bs->copy_on_read && read_only) {
error_setg(errp, "Can't set node '%s' to r/o with copy-on-read enabled",
bdrv_get_device_or_node_name(bs));
return -EINVAL;
}
/* Do not clear read_only if it is prohibited */
if (!read_only && !(bs->open_flags & BDRV_O_ALLOW_RDWR) &&
!ignore_allow_rdw)
{
error_setg(errp, "Node '%s' is read only",
bdrv_get_device_or_node_name(bs));
return -EPERM;
}
return 0;
}
/*
* Called by a driver that can only provide a read-only image.
*
* Returns 0 if the node is already read-only or it could switch the node to
* read-only because BDRV_O_AUTO_RDONLY is set.
*
* Returns -EACCES if the node is read-write and BDRV_O_AUTO_RDONLY is not set
* or bdrv_can_set_read_only() forbids making the node read-only. If @errmsg
* is not NULL, it is used as the error message for the Error object.
*/
int bdrv_apply_auto_read_only(BlockDriverState *bs, const char *errmsg,
Error **errp)
{
int ret = 0;
IO_CODE();
if (!(bs->open_flags & BDRV_O_RDWR)) {
return 0;
}
if (!(bs->open_flags & BDRV_O_AUTO_RDONLY)) {
goto fail;
}
ret = bdrv_can_set_read_only(bs, true, false, NULL);
if (ret < 0) {
goto fail;
}
bs->open_flags &= ~BDRV_O_RDWR;
return 0;
fail:
error_setg(errp, "%s", errmsg ?: "Image is read-only");
return -EACCES;
}
/*
* If @backing is empty, this function returns NULL without setting
* @errp. In all other cases, NULL will only be returned with @errp
* set.
*
* Therefore, a return value of NULL without @errp set means that
* there is no backing file; if @errp is set, there is one but its
* absolute filename cannot be generated.
*/
char *bdrv_get_full_backing_filename_from_filename(const char *backed,
const char *backing,
Error **errp)
{
if (backing[0] == '\0') {
return NULL;
} else if (path_has_protocol(backing) || path_is_absolute(backing)) {
return g_strdup(backing);
} else if (backed[0] == '\0' || strstart(backed, "json:", NULL)) {
error_setg(errp, "Cannot use relative backing file names for '%s'",
backed);
return NULL;
} else {
return path_combine(backed, backing);
}
}
/*
* If @filename is empty or NULL, this function returns NULL without
* setting @errp. In all other cases, NULL will only be returned with
* @errp set.
*/
static char *bdrv_make_absolute_filename(BlockDriverState *relative_to,
const char *filename, Error **errp)
{
char *dir, *full_name;
if (!filename || filename[0] == '\0') {
return NULL;
} else if (path_has_protocol(filename) || path_is_absolute(filename)) {
return g_strdup(filename);
}
dir = bdrv_dirname(relative_to, errp);
if (!dir) {
return NULL;
}
full_name = g_strconcat(dir, filename, NULL);
g_free(dir);
return full_name;
}
char *bdrv_get_full_backing_filename(BlockDriverState *bs, Error **errp)
{
GLOBAL_STATE_CODE();
return bdrv_make_absolute_filename(bs, bs->backing_file, errp);
}
void bdrv_register(BlockDriver *bdrv)
{
assert(bdrv->format_name);
GLOBAL_STATE_CODE();
QLIST_INSERT_HEAD(&bdrv_drivers, bdrv, list);
}
BlockDriverState *bdrv_new(void)
{
BlockDriverState *bs;
int i;
GLOBAL_STATE_CODE();
bs = g_new0(BlockDriverState, 1);
QLIST_INIT(&bs->dirty_bitmaps);
for (i = 0; i < BLOCK_OP_TYPE_MAX; i++) {
QLIST_INIT(&bs->op_blockers[i]);
}
qemu_co_mutex_init(&bs->reqs_lock);
qemu_mutex_init(&bs->dirty_bitmap_mutex);
bs->refcnt = 1;
bs->aio_context = qemu_get_aio_context();
qemu_co_queue_init(&bs->flush_queue);
block: block-status cache for data regions As we have attempted before (https://lists.gnu.org/archive/html/qemu-devel/2019-01/msg06451.html, "file-posix: Cache lseek result for data regions"; https://lists.nongnu.org/archive/html/qemu-block/2021-02/msg00934.html, "file-posix: Cache next hole"), this patch seeks to reduce the number of SEEK_DATA/HOLE operations the file-posix driver has to perform. The main difference is that this time it is implemented as part of the general block layer code. The problem we face is that on some filesystems or in some circumstances, SEEK_DATA/HOLE is unreasonably slow. Given the implementation is outside of qemu, there is little we can do about its performance. We have already introduced the want_zero parameter to bdrv_co_block_status() to reduce the number of SEEK_DATA/HOLE calls unless we really want zero information; but sometimes we do want that information, because for files that consist largely of zero areas, special-casing those areas can give large performance boosts. So the real problem is with files that consist largely of data, so that inquiring the block status does not gain us much performance, but where such an inquiry itself takes a lot of time. To address this, we want to cache data regions. Most of the time, when bad performance is reported, it is in places where the image is iterated over from start to end (qemu-img convert or the mirror job), so a simple yet effective solution is to cache only the current data region. (Note that only caching data regions but not zero regions means that returning false information from the cache is not catastrophic: Treating zeroes as data is fine. While we try to invalidate the cache on zero writes and discards, such incongruences may still occur when there are other processes writing to the image.) We only use the cache for nodes without children (i.e. protocol nodes), because that is where the problem is: Drivers that rely on block-status implementations outside of qemu (e.g. SEEK_DATA/HOLE). Resolves: https://gitlab.com/qemu-project/qemu/-/issues/307 Signed-off-by: Hanna Reitz <hreitz@redhat.com> Message-Id: <20210812084148.14458-3-hreitz@redhat.com> Reviewed-by: Eric Blake <eblake@redhat.com> Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> [hreitz: Added `local_file == bs` assertion, as suggested by Vladimir] Signed-off-by: Hanna Reitz <hreitz@redhat.com>
2021-08-12 10:41:44 +02:00
qemu_co_mutex_init(&bs->bsc_modify_lock);
bs->block_status_cache = g_new0(BdrvBlockStatusCache, 1);
for (i = 0; i < bdrv_drain_all_count; i++) {
bdrv_drained_begin(bs);
}
QTAILQ_INSERT_TAIL(&all_bdrv_states, bs, bs_list);
return bs;
}
2016-08-12 15:27:03 +02:00
static BlockDriver *bdrv_do_find_format(const char *format_name)
{
BlockDriver *drv1;
GLOBAL_STATE_CODE();
2016-08-12 15:27:03 +02:00
QLIST_FOREACH(drv1, &bdrv_drivers, list) {
if (!strcmp(drv1->format_name, format_name)) {
return drv1;
}
}
2016-08-12 15:27:03 +02:00
return NULL;
}
2016-08-12 15:27:03 +02:00
BlockDriver *bdrv_find_format(const char *format_name)
{
BlockDriver *drv1;
int i;
GLOBAL_STATE_CODE();
2016-08-12 15:27:03 +02:00
drv1 = bdrv_do_find_format(format_name);
if (drv1) {
return drv1;
}
/* The driver isn't registered, maybe we need to load a module */
for (i = 0; i < (int)ARRAY_SIZE(block_driver_modules); ++i) {
if (!strcmp(block_driver_modules[i].format_name, format_name)) {
block_module_load_one(block_driver_modules[i].library_name);
break;
}
}
return bdrv_do_find_format(format_name);
}
static int bdrv_format_is_whitelisted(const char *format_name, bool read_only)
{
static const char *whitelist_rw[] = {
CONFIG_BDRV_RW_WHITELIST
NULL
};
static const char *whitelist_ro[] = {
CONFIG_BDRV_RO_WHITELIST
NULL
};
const char **p;
if (!whitelist_rw[0] && !whitelist_ro[0]) {
return 1; /* no whitelist, anything goes */
}
for (p = whitelist_rw; *p; p++) {
if (!strcmp(format_name, *p)) {
return 1;
}
}
if (read_only) {
for (p = whitelist_ro; *p; p++) {
if (!strcmp(format_name, *p)) {
return 1;
}
}
}
return 0;
}
int bdrv_is_whitelisted(BlockDriver *drv, bool read_only)
{
GLOBAL_STATE_CODE();
return bdrv_format_is_whitelisted(drv->format_name, read_only);
}
2016-03-21 15:11:48 +01:00
bool bdrv_uses_whitelist(void)
{
return use_bdrv_whitelist;
}
typedef struct CreateCo {
BlockDriver *drv;
char *filename;
QemuOpts *opts;
int ret;
Error *err;
} CreateCo;
static void coroutine_fn bdrv_create_co_entry(void *opaque)
{
Error *local_err = NULL;
int ret;
CreateCo *cco = opaque;
assert(cco->drv);
GLOBAL_STATE_CODE();
ret = cco->drv->bdrv_co_create_opts(cco->drv,
cco->filename, cco->opts, &local_err);
error_propagate(&cco->err, local_err);
cco->ret = ret;
}
int bdrv_create(BlockDriver *drv, const char* filename,
QemuOpts *opts, Error **errp)
{
int ret;
GLOBAL_STATE_CODE();
Coroutine *co;
CreateCo cco = {
.drv = drv,
.filename = g_strdup(filename),
.opts = opts,
.ret = NOT_DONE,
.err = NULL,
};
if (!drv->bdrv_co_create_opts) {
error_setg(errp, "Driver '%s' does not support image creation", drv->format_name);
ret = -ENOTSUP;
goto out;
}
if (qemu_in_coroutine()) {
/* Fast-path if already in coroutine context */
bdrv_create_co_entry(&cco);
} else {
co = qemu_coroutine_create(bdrv_create_co_entry, &cco);
qemu_coroutine_enter(co);
while (cco.ret == NOT_DONE) {
aio_poll(qemu_get_aio_context(), true);
}
}
ret = cco.ret;
if (ret < 0) {
if (cco.err) {
error_propagate(errp, cco.err);
} else {
error_setg_errno(errp, -ret, "Could not create image");
}
}
out:
g_free(cco.filename);
return ret;
}
/**
* Helper function for bdrv_create_file_fallback(): Resize @blk to at
* least the given @minimum_size.
*
* On success, return @blk's actual length.
* Otherwise, return -errno.
*/
static int64_t create_file_fallback_truncate(BlockBackend *blk,
int64_t minimum_size, Error **errp)
block: separate raw images from the file protocol We're running into various problems because the "raw" file access, which is used internally by the various image formats is entangled with the "raw" image format, which maps the VM view 1:1 to a file system. This patch renames the raw file backends to the file protocol which is treated like other protocols (e.g. nbd and http) and adds a new "raw" image format which is just a wrapper around calls to the underlying protocol. The patch is surprisingly simple, besides changing the probing logical in block.c to only look for image formats when using bdrv_open and renaming of the old raw protocols to file there's almost nothing in there. For creating images, a new bdrv_create_file is introduced which guesses the protocol to use. This allows using qemu-img create -f raw (or just using the default) for both files and host devices. Converting the other format drivers to use this function to create their images is left for later patches. The only issues still open are in the handling of the host devices. Firstly in current qemu we can specifiy the host* format names on various command line acceping images, but the new code can't do that without adding some translation. Second the layering breaks the no_zero_init flag in the BlockDriver used by qemu-img. I'm not happy how this is done per-driver instead of per-state so I'll prepare a separate patch to clean this up. There's some more cleanup opportunity after this patch, e.g. using separate lists and registration functions for image formats vs protocols and maybe even host drivers, but this can be done at a later stage. Also there's a check for protocol in bdrv_open for the BDRV_O_SNAPSHOT case that I don't quite understand, but which I fear won't work as expected - possibly even before this patch. Note that this patch requires various recent block patches from Kevin and me, which should all be in his block queue. Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2010-04-07 22:30:24 +02:00
{
Error *local_err = NULL;
int64_t size;
int ret;
block: separate raw images from the file protocol We're running into various problems because the "raw" file access, which is used internally by the various image formats is entangled with the "raw" image format, which maps the VM view 1:1 to a file system. This patch renames the raw file backends to the file protocol which is treated like other protocols (e.g. nbd and http) and adds a new "raw" image format which is just a wrapper around calls to the underlying protocol. The patch is surprisingly simple, besides changing the probing logical in block.c to only look for image formats when using bdrv_open and renaming of the old raw protocols to file there's almost nothing in there. For creating images, a new bdrv_create_file is introduced which guesses the protocol to use. This allows using qemu-img create -f raw (or just using the default) for both files and host devices. Converting the other format drivers to use this function to create their images is left for later patches. The only issues still open are in the handling of the host devices. Firstly in current qemu we can specifiy the host* format names on various command line acceping images, but the new code can't do that without adding some translation. Second the layering breaks the no_zero_init flag in the BlockDriver used by qemu-img. I'm not happy how this is done per-driver instead of per-state so I'll prepare a separate patch to clean this up. There's some more cleanup opportunity after this patch, e.g. using separate lists and registration functions for image formats vs protocols and maybe even host drivers, but this can be done at a later stage. Also there's a check for protocol in bdrv_open for the BDRV_O_SNAPSHOT case that I don't quite understand, but which I fear won't work as expected - possibly even before this patch. Note that this patch requires various recent block patches from Kevin and me, which should all be in his block queue. Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2010-04-07 22:30:24 +02:00
GLOBAL_STATE_CODE();
ret = blk_truncate(blk, minimum_size, false, PREALLOC_MODE_OFF, 0,
&local_err);
if (ret < 0 && ret != -ENOTSUP) {
error_propagate(errp, local_err);
return ret;
}
size = blk_getlength(blk);
if (size < 0) {
error_free(local_err);
error_setg_errno(errp, -size,
"Failed to inquire the new image file's length");
return size;
}
if (size < minimum_size) {
/* Need to grow the image, but we failed to do that */
error_propagate(errp, local_err);
return -ENOTSUP;
}
error_free(local_err);
local_err = NULL;
return size;
}
/**
* Helper function for bdrv_create_file_fallback(): Zero the first
* sector to remove any potentially pre-existing image header.
*/
static int create_file_fallback_zero_first_sector(BlockBackend *blk,
int64_t current_size,
Error **errp)
{
int64_t bytes_to_clear;
int ret;
GLOBAL_STATE_CODE();
bytes_to_clear = MIN(current_size, BDRV_SECTOR_SIZE);
if (bytes_to_clear) {
ret = blk_pwrite_zeroes(blk, 0, bytes_to_clear, BDRV_REQ_MAY_UNMAP);
if (ret < 0) {
error_setg_errno(errp, -ret,
"Failed to clear the new image's first sector");
return ret;
}
}
return 0;
}
/**
* Simple implementation of bdrv_co_create_opts for protocol drivers
* which only support creation via opening a file
* (usually existing raw storage device)
*/
int coroutine_fn bdrv_co_create_opts_simple(BlockDriver *drv,
const char *filename,
QemuOpts *opts,
Error **errp)
{
BlockBackend *blk;
QDict *options;
int64_t size = 0;
char *buf = NULL;
PreallocMode prealloc;
Error *local_err = NULL;
int ret;
GLOBAL_STATE_CODE();
size = qemu_opt_get_size_del(opts, BLOCK_OPT_SIZE, 0);
buf = qemu_opt_get_del(opts, BLOCK_OPT_PREALLOC);
prealloc = qapi_enum_parse(&PreallocMode_lookup, buf,
PREALLOC_MODE_OFF, &local_err);
g_free(buf);
if (local_err) {
error_propagate(errp, local_err);
return -EINVAL;
}
if (prealloc != PREALLOC_MODE_OFF) {
error_setg(errp, "Unsupported preallocation mode '%s'",
PreallocMode_str(prealloc));
return -ENOTSUP;
}
options = qdict_new();
qdict_put_str(options, "driver", drv->format_name);
blk = blk_new_open(filename, NULL, options,
BDRV_O_RDWR | BDRV_O_RESIZE, errp);
if (!blk) {
error_prepend(errp, "Protocol driver '%s' does not support image "
"creation, and opening the image failed: ",
drv->format_name);
return -EINVAL;
}
size = create_file_fallback_truncate(blk, size, errp);
if (size < 0) {
ret = size;
goto out;
}
ret = create_file_fallback_zero_first_sector(blk, size, errp);
if (ret < 0) {
goto out;
}
ret = 0;
out:
blk_unref(blk);
return ret;
}
int bdrv_create_file(const char *filename, QemuOpts *opts, Error **errp)
{
QemuOpts *protocol_opts;
BlockDriver *drv;
QDict *qdict;
int ret;
GLOBAL_STATE_CODE();
drv = bdrv_find_protocol(filename, true, errp);
block: separate raw images from the file protocol We're running into various problems because the "raw" file access, which is used internally by the various image formats is entangled with the "raw" image format, which maps the VM view 1:1 to a file system. This patch renames the raw file backends to the file protocol which is treated like other protocols (e.g. nbd and http) and adds a new "raw" image format which is just a wrapper around calls to the underlying protocol. The patch is surprisingly simple, besides changing the probing logical in block.c to only look for image formats when using bdrv_open and renaming of the old raw protocols to file there's almost nothing in there. For creating images, a new bdrv_create_file is introduced which guesses the protocol to use. This allows using qemu-img create -f raw (or just using the default) for both files and host devices. Converting the other format drivers to use this function to create their images is left for later patches. The only issues still open are in the handling of the host devices. Firstly in current qemu we can specifiy the host* format names on various command line acceping images, but the new code can't do that without adding some translation. Second the layering breaks the no_zero_init flag in the BlockDriver used by qemu-img. I'm not happy how this is done per-driver instead of per-state so I'll prepare a separate patch to clean this up. There's some more cleanup opportunity after this patch, e.g. using separate lists and registration functions for image formats vs protocols and maybe even host drivers, but this can be done at a later stage. Also there's a check for protocol in bdrv_open for the BDRV_O_SNAPSHOT case that I don't quite understand, but which I fear won't work as expected - possibly even before this patch. Note that this patch requires various recent block patches from Kevin and me, which should all be in his block queue. Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2010-04-07 22:30:24 +02:00
if (drv == NULL) {
return -ENOENT;
block: separate raw images from the file protocol We're running into various problems because the "raw" file access, which is used internally by the various image formats is entangled with the "raw" image format, which maps the VM view 1:1 to a file system. This patch renames the raw file backends to the file protocol which is treated like other protocols (e.g. nbd and http) and adds a new "raw" image format which is just a wrapper around calls to the underlying protocol. The patch is surprisingly simple, besides changing the probing logical in block.c to only look for image formats when using bdrv_open and renaming of the old raw protocols to file there's almost nothing in there. For creating images, a new bdrv_create_file is introduced which guesses the protocol to use. This allows using qemu-img create -f raw (or just using the default) for both files and host devices. Converting the other format drivers to use this function to create their images is left for later patches. The only issues still open are in the handling of the host devices. Firstly in current qemu we can specifiy the host* format names on various command line acceping images, but the new code can't do that without adding some translation. Second the layering breaks the no_zero_init flag in the BlockDriver used by qemu-img. I'm not happy how this is done per-driver instead of per-state so I'll prepare a separate patch to clean this up. There's some more cleanup opportunity after this patch, e.g. using separate lists and registration functions for image formats vs protocols and maybe even host drivers, but this can be done at a later stage. Also there's a check for protocol in bdrv_open for the BDRV_O_SNAPSHOT case that I don't quite understand, but which I fear won't work as expected - possibly even before this patch. Note that this patch requires various recent block patches from Kevin and me, which should all be in his block queue. Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2010-04-07 22:30:24 +02:00
}
if (!drv->create_opts) {
error_setg(errp, "Driver '%s' does not support image creation",
drv->format_name);
return -ENOTSUP;
}
/*
* 'opts' contains a QemuOptsList with a combination of format and protocol
* default values.
*
* The format properly removes its options, but the default values remain
* in 'opts->list'. So if the protocol has options with the same name
* (e.g. rbd has 'cluster_size' as qcow2), it will see the default values
* of the format, since for overlapping options, the format wins.
*
* To avoid this issue, lets convert QemuOpts to QDict, in this way we take
* only the set options, and then convert it back to QemuOpts, using the
* create_opts of the protocol. So the new QemuOpts, will contain only the
* protocol defaults.
*/
qdict = qemu_opts_to_qdict(opts, NULL);
protocol_opts = qemu_opts_from_qdict(drv->create_opts, qdict, errp);
if (protocol_opts == NULL) {
ret = -EINVAL;
goto out;
}
ret = bdrv_create(drv, filename, protocol_opts, errp);
out:
qemu_opts_del(protocol_opts);
qobject_unref(qdict);
return ret;
block: separate raw images from the file protocol We're running into various problems because the "raw" file access, which is used internally by the various image formats is entangled with the "raw" image format, which maps the VM view 1:1 to a file system. This patch renames the raw file backends to the file protocol which is treated like other protocols (e.g. nbd and http) and adds a new "raw" image format which is just a wrapper around calls to the underlying protocol. The patch is surprisingly simple, besides changing the probing logical in block.c to only look for image formats when using bdrv_open and renaming of the old raw protocols to file there's almost nothing in there. For creating images, a new bdrv_create_file is introduced which guesses the protocol to use. This allows using qemu-img create -f raw (or just using the default) for both files and host devices. Converting the other format drivers to use this function to create their images is left for later patches. The only issues still open are in the handling of the host devices. Firstly in current qemu we can specifiy the host* format names on various command line acceping images, but the new code can't do that without adding some translation. Second the layering breaks the no_zero_init flag in the BlockDriver used by qemu-img. I'm not happy how this is done per-driver instead of per-state so I'll prepare a separate patch to clean this up. There's some more cleanup opportunity after this patch, e.g. using separate lists and registration functions for image formats vs protocols and maybe even host drivers, but this can be done at a later stage. Also there's a check for protocol in bdrv_open for the BDRV_O_SNAPSHOT case that I don't quite understand, but which I fear won't work as expected - possibly even before this patch. Note that this patch requires various recent block patches from Kevin and me, which should all be in his block queue. Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2010-04-07 22:30:24 +02:00
}
int coroutine_fn bdrv_co_delete_file(BlockDriverState *bs, Error **errp)
{
Error *local_err = NULL;
int ret;
IO_CODE();
assert(bs != NULL);
if (!bs->drv) {
error_setg(errp, "Block node '%s' is not opened", bs->filename);
return -ENOMEDIUM;
}
if (!bs->drv->bdrv_co_delete_file) {
error_setg(errp, "Driver '%s' does not support image deletion",
bs->drv->format_name);
return -ENOTSUP;
}
ret = bs->drv->bdrv_co_delete_file(bs, &local_err);
if (ret < 0) {
error_propagate(errp, local_err);
}
return ret;
}
void coroutine_fn bdrv_co_delete_file_noerr(BlockDriverState *bs)
{
Error *local_err = NULL;
int ret;
IO_CODE();
if (!bs) {
return;
}
ret = bdrv_co_delete_file(bs, &local_err);
/*
* ENOTSUP will happen if the block driver doesn't support
* the 'bdrv_co_delete_file' interface. This is a predictable
* scenario and shouldn't be reported back to the user.
*/
if (ret == -ENOTSUP) {
error_free(local_err);
} else if (ret < 0) {
error_report_err(local_err);
}
}
/**
* Try to get @bs's logical and physical block size.
* On success, store them in @bsz struct and return 0.
* On failure return -errno.
* @bs must not be empty.
*/
int bdrv_probe_blocksizes(BlockDriverState *bs, BlockSizes *bsz)
{
BlockDriver *drv = bs->drv;
BlockDriverState *filtered = bdrv_filter_bs(bs);
GLOBAL_STATE_CODE();
if (drv && drv->bdrv_probe_blocksizes) {
return drv->bdrv_probe_blocksizes(bs, bsz);
} else if (filtered) {
return bdrv_probe_blocksizes(filtered, bsz);
}
return -ENOTSUP;
}
/**
* Try to get @bs's geometry (cyls, heads, sectors).
* On success, store them in @geo struct and return 0.
* On failure return -errno.
* @bs must not be empty.
*/
int bdrv_probe_geometry(BlockDriverState *bs, HDGeometry *geo)
{
BlockDriver *drv = bs->drv;
BlockDriverState *filtered = bdrv_filter_bs(bs);
GLOBAL_STATE_CODE();
if (drv && drv->bdrv_probe_geometry) {
return drv->bdrv_probe_geometry(bs, geo);
} else if (filtered) {
return bdrv_probe_geometry(filtered, geo);
}
return -ENOTSUP;
}
/*
* Create a uniquely-named empty temporary file.
* Return 0 upon success, otherwise a negative errno value.
*/
int get_tmp_filename(char *filename, int size)
{
#ifdef _WIN32
char temp_dir[MAX_PATH];
/* GetTempFileName requires that its output buffer (4th param)
have length MAX_PATH or greater. */
assert(size >= MAX_PATH);
return (GetTempPath(MAX_PATH, temp_dir)
&& GetTempFileName(temp_dir, "qem", 0, filename)
? 0 : -GetLastError());
#else
int fd;
const char *tmpdir;
tmpdir = getenv("TMPDIR");
if (!tmpdir) {
tmpdir = "/var/tmp";
}
if (snprintf(filename, size, "%s/vl.XXXXXX", tmpdir) >= size) {
return -EOVERFLOW;
}
fd = mkstemp(filename);
if (fd < 0) {
return -errno;
}
if (close(fd) != 0) {
unlink(filename);
return -errno;
}
return 0;
#endif
}
block: separate raw images from the file protocol We're running into various problems because the "raw" file access, which is used internally by the various image formats is entangled with the "raw" image format, which maps the VM view 1:1 to a file system. This patch renames the raw file backends to the file protocol which is treated like other protocols (e.g. nbd and http) and adds a new "raw" image format which is just a wrapper around calls to the underlying protocol. The patch is surprisingly simple, besides changing the probing logical in block.c to only look for image formats when using bdrv_open and renaming of the old raw protocols to file there's almost nothing in there. For creating images, a new bdrv_create_file is introduced which guesses the protocol to use. This allows using qemu-img create -f raw (or just using the default) for both files and host devices. Converting the other format drivers to use this function to create their images is left for later patches. The only issues still open are in the handling of the host devices. Firstly in current qemu we can specifiy the host* format names on various command line acceping images, but the new code can't do that without adding some translation. Second the layering breaks the no_zero_init flag in the BlockDriver used by qemu-img. I'm not happy how this is done per-driver instead of per-state so I'll prepare a separate patch to clean this up. There's some more cleanup opportunity after this patch, e.g. using separate lists and registration functions for image formats vs protocols and maybe even host drivers, but this can be done at a later stage. Also there's a check for protocol in bdrv_open for the BDRV_O_SNAPSHOT case that I don't quite understand, but which I fear won't work as expected - possibly even before this patch. Note that this patch requires various recent block patches from Kevin and me, which should all be in his block queue. Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2010-04-07 22:30:24 +02:00
/*
* Detect host devices. By convention, /dev/cdrom[N] is always
* recognized as a host CDROM.
*/
static BlockDriver *find_hdev_driver(const char *filename)
{
int score_max = 0, score;
BlockDriver *drv = NULL, *d;
GLOBAL_STATE_CODE();
block: separate raw images from the file protocol We're running into various problems because the "raw" file access, which is used internally by the various image formats is entangled with the "raw" image format, which maps the VM view 1:1 to a file system. This patch renames the raw file backends to the file protocol which is treated like other protocols (e.g. nbd and http) and adds a new "raw" image format which is just a wrapper around calls to the underlying protocol. The patch is surprisingly simple, besides changing the probing logical in block.c to only look for image formats when using bdrv_open and renaming of the old raw protocols to file there's almost nothing in there. For creating images, a new bdrv_create_file is introduced which guesses the protocol to use. This allows using qemu-img create -f raw (or just using the default) for both files and host devices. Converting the other format drivers to use this function to create their images is left for later patches. The only issues still open are in the handling of the host devices. Firstly in current qemu we can specifiy the host* format names on various command line acceping images, but the new code can't do that without adding some translation. Second the layering breaks the no_zero_init flag in the BlockDriver used by qemu-img. I'm not happy how this is done per-driver instead of per-state so I'll prepare a separate patch to clean this up. There's some more cleanup opportunity after this patch, e.g. using separate lists and registration functions for image formats vs protocols and maybe even host drivers, but this can be done at a later stage. Also there's a check for protocol in bdrv_open for the BDRV_O_SNAPSHOT case that I don't quite understand, but which I fear won't work as expected - possibly even before this patch. Note that this patch requires various recent block patches from Kevin and me, which should all be in his block queue. Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2010-04-07 22:30:24 +02:00
QLIST_FOREACH(d, &bdrv_drivers, list) {
if (d->bdrv_probe_device) {
score = d->bdrv_probe_device(filename);
if (score > score_max) {
score_max = score;
drv = d;
}
}
}
return drv;
}
2016-08-12 15:27:03 +02:00
static BlockDriver *bdrv_do_find_protocol(const char *protocol)
{
BlockDriver *drv1;
GLOBAL_STATE_CODE();
2016-08-12 15:27:03 +02:00
QLIST_FOREACH(drv1, &bdrv_drivers, list) {
if (drv1->protocol_name && !strcmp(drv1->protocol_name, protocol)) {
return drv1;
}
}
return NULL;
}
BlockDriver *bdrv_find_protocol(const char *filename,
bool allow_protocol_prefix,
Error **errp)
{
BlockDriver *drv1;
char protocol[128];
int len;
const char *p;
2016-08-12 15:27:03 +02:00
int i;
GLOBAL_STATE_CODE();
/* TODO Drivers without bdrv_file_open must be specified explicitly */
/*
* XXX(hch): we really should not let host device detection
* override an explicit protocol specification, but moving this
* later breaks access to device names with colons in them.
* Thanks to the brain-dead persistent naming schemes on udev-
* based Linux systems those actually are quite common.
*/
drv1 = find_hdev_driver(filename);
if (drv1) {
return drv1;
}
if (!path_has_protocol(filename) || !allow_protocol_prefix) {
return &bdrv_file;
block: separate raw images from the file protocol We're running into various problems because the "raw" file access, which is used internally by the various image formats is entangled with the "raw" image format, which maps the VM view 1:1 to a file system. This patch renames the raw file backends to the file protocol which is treated like other protocols (e.g. nbd and http) and adds a new "raw" image format which is just a wrapper around calls to the underlying protocol. The patch is surprisingly simple, besides changing the probing logical in block.c to only look for image formats when using bdrv_open and renaming of the old raw protocols to file there's almost nothing in there. For creating images, a new bdrv_create_file is introduced which guesses the protocol to use. This allows using qemu-img create -f raw (or just using the default) for both files and host devices. Converting the other format drivers to use this function to create their images is left for later patches. The only issues still open are in the handling of the host devices. Firstly in current qemu we can specifiy the host* format names on various command line acceping images, but the new code can't do that without adding some translation. Second the layering breaks the no_zero_init flag in the BlockDriver used by qemu-img. I'm not happy how this is done per-driver instead of per-state so I'll prepare a separate patch to clean this up. There's some more cleanup opportunity after this patch, e.g. using separate lists and registration functions for image formats vs protocols and maybe even host drivers, but this can be done at a later stage. Also there's a check for protocol in bdrv_open for the BDRV_O_SNAPSHOT case that I don't quite understand, but which I fear won't work as expected - possibly even before this patch. Note that this patch requires various recent block patches from Kevin and me, which should all be in his block queue. Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2010-04-07 22:30:24 +02:00
}
p = strchr(filename, ':');
assert(p != NULL);
len = p - filename;
if (len > sizeof(protocol) - 1)
len = sizeof(protocol) - 1;
memcpy(protocol, filename, len);
protocol[len] = '\0';
2016-08-12 15:27:03 +02:00
drv1 = bdrv_do_find_protocol(protocol);
if (drv1) {
return drv1;
}
for (i = 0; i < (int)ARRAY_SIZE(block_driver_modules); ++i) {
if (block_driver_modules[i].protocol_name &&
!strcmp(block_driver_modules[i].protocol_name, protocol)) {
block_module_load_one(block_driver_modules[i].library_name);
break;
}
}
2016-08-12 15:27:03 +02:00
drv1 = bdrv_do_find_protocol(protocol);
if (!drv1) {
error_setg(errp, "Unknown protocol '%s'", protocol);
}
return drv1;
}
/*
* Guess image format by probing its contents.
* This is not a good idea when your image is raw (CVE-2008-2004), but
* we do it anyway for backward compatibility.
*
* @buf contains the image's first @buf_size bytes.
* @buf_size is the buffer size in bytes (generally BLOCK_PROBE_BUF_SIZE,
* but can be smaller if the image file is smaller)
* @filename is its filename.
*
* For all block drivers, call the bdrv_probe() method to get its
* probing score.
* Return the first block driver with the highest probing score.
*/
raw: Prohibit dangerous writes for probed images If the user neglects to specify the image format, QEMU probes the image to guess it automatically, for convenience. Relying on format probing is insecure for raw images (CVE-2008-2004). If the guest writes a suitable header to the device, the next probe will recognize a format chosen by the guest. A malicious guest can abuse this to gain access to host files, e.g. by crafting a QCOW2 header with backing file /etc/shadow. Commit 1e72d3b (April 2008) provided -drive parameter format to let users disable probing. Commit f965509 (March 2009) extended QCOW2 to optionally store the backing file format, to let users disable backing file probing. QED has had a flag to suppress probing since the beginning (2010), set whenever a raw backing file is assigned. All of these additions that allow to avoid format probing have to be specified explicitly. The default still allows the attack. In order to fix this, commit 79368c8 (July 2010) put probed raw images in a restricted mode, in which they wouldn't be able to overwrite the first few bytes of the image so that they would identify as a different image. If a write to the first sector would write one of the signatures of another driver, qemu would instead zero out the first four bytes. This patch was later reverted in commit 8b33d9e (September 2010) because it didn't get the handling of unaligned qiov members right. Today's block layer that is based on coroutines and has qiov utility functions makes it much easier to get this functionality right, so this patch implements it. The other differences of this patch to the old one are that it doesn't silently write something different than the guest requested by zeroing out some bytes (it fails the request instead) and that it doesn't maintain a list of signatures in the raw driver (it calls the usual probe function instead). Note that this change doesn't introduce new breakage for false positive cases where the guest legitimately writes data into the first sector that matches the signatures of an image format (e.g. for nested virt): These cases were broken before, only the failure mode changes from corruption after the next restart (when the wrong format is probed) to failing the problematic write request. Also note that like in the original patch, the restrictions only apply if the image format has been guessed by probing. Explicitly specifying a format allows guests to write anything they like. Signed-off-by: Kevin Wolf <kwolf@redhat.com> Reviewed-by: Eric Blake <eblake@redhat.com> Reviewed-by: Max Reitz <mreitz@redhat.com> Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com> Message-id: 1416497234-29880-8-git-send-email-kwolf@redhat.com Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2014-11-20 16:27:12 +01:00
BlockDriver *bdrv_probe_all(const uint8_t *buf, int buf_size,
const char *filename)
{
int score_max = 0, score;
BlockDriver *drv = NULL, *d;
IO_CODE();
QLIST_FOREACH(d, &bdrv_drivers, list) {
if (d->bdrv_probe) {
score = d->bdrv_probe(buf, buf_size, filename);
if (score > score_max) {
score_max = score;
drv = d;
}
}
}
return drv;
}
static int find_image_format(BlockBackend *file, const char *filename,
BlockDriver **pdrv, Error **errp)
{
BlockDriver *drv;
uint8_t buf[BLOCK_PROBE_BUF_SIZE];
int ret = 0;
GLOBAL_STATE_CODE();
/* Return the raw BlockDriver * to scsi-generic devices or empty drives */
if (blk_is_sg(file) || !blk_is_inserted(file) || blk_getlength(file) == 0) {
*pdrv = &bdrv_raw;
return ret;
}
ret = blk_pread(file, 0, sizeof(buf), buf, 0);
if (ret < 0) {
error_setg_errno(errp, -ret, "Could not read image for determining its "
"format");
*pdrv = NULL;
return ret;
}
drv = bdrv_probe_all(buf, sizeof(buf), filename);
if (!drv) {
error_setg(errp, "Could not determine image format: No compatible "
"driver found");
*pdrv = NULL;
return -ENOENT;
}
*pdrv = drv;
return 0;
}
/**
* Set the current 'total_sectors' value
* Return 0 on success, -errno on error.
*/
int refresh_total_sectors(BlockDriverState *bs, int64_t hint)
{
BlockDriver *drv = bs->drv;
IO_CODE();
if (!drv) {
return -ENOMEDIUM;
}
/* Do not attempt drv->bdrv_getlength() on scsi-generic devices */
if (bdrv_is_sg(bs))
return 0;
/* query actual device if possible, otherwise just trust the hint */
if (drv->bdrv_getlength) {
int64_t length = drv->bdrv_getlength(bs);
if (length < 0) {
return length;
}
hint = DIV_ROUND_UP(length, BDRV_SECTOR_SIZE);
}
bs->total_sectors = hint;
block: introduce BDRV_MAX_LENGTH We are going to modify block layer to work with 64bit requests. And first step is moving to int64_t type for both offset and bytes arguments in all block request related functions. It's mostly safe (when widening signed or unsigned int to int64_t), but switching from uint64_t is questionable. So, let's first establish the set of requests we want to work with. First signed int64_t should be enough, as off_t is signed anyway. Then, obviously offset + bytes should not overflow. And most interesting: (offset + bytes) being aligned up should not overflow as well. Aligned to what alignment? First thing that comes in mind is bs->bl.request_alignment, as we align up request to this alignment. But there is another thing: look at bdrv_mark_request_serialising(). It aligns request up to some given alignment. And this parameter may be bdrv_get_cluster_size(), which is often a lot greater than bs->bl.request_alignment. Note also, that bdrv_mark_request_serialising() uses signed int64_t for calculations. So, actually, we already depend on some restrictions. Happily, bdrv_get_cluster_size() returns int and bs->bl.request_alignment has 32bit unsigned type, but defined to be a power of 2 less than INT_MAX. So, we may establish, that INT_MAX is absolute maximum for any kind of alignment that may occur with the request. Note, that bdrv_get_cluster_size() is not documented to return power of 2, still bdrv_mark_request_serialising() behaves like it is. Also, backup uses bdi.cluster_size and is not prepared to it not being power of 2. So, let's establish that Qemu supports only power-of-2 clusters and alignments. So, alignment can't be greater than 2^30. Finally to be safe with calculations, to not calculate different maximums for different nodes (depending on cluster size and request_alignment), let's simply set QEMU_ALIGN_DOWN(INT64_MAX, 2^30) as absolute maximum bytes length for Qemu. Actually, it's not much less than INT64_MAX. OK, then, let's apply it to block/io. Let's consider all block/io entry points of offset/bytes: 4 bytes/offset interface functions: bdrv_co_preadv_part(), bdrv_co_pwritev_part(), bdrv_co_copy_range_internal() and bdrv_co_pdiscard() and we check them all with bdrv_check_request(). We also have one entry point with only offset: bdrv_co_truncate(). Check the offset. And one public structure: BdrvTrackedRequest. Happily, it has only three external users: file-posix.c: adopted by this patch write-threshold.c: only read fields test-write-threshold.c: sets obviously small constant values Better is to make the structure private and add corresponding interfaces.. Still it's not obvious what kind of interface is needed for file-posix.c. Let's keep it public but add corresponding assertions. After this patch we'll convert functions in block/io.c to int64_t bytes and offset parameters. We can assume that offset/bytes pair always satisfy new restrictions, and make corresponding assertions where needed. If we reach some offset/bytes point in block/io.c missing bdrv_check_request() it is considered a bug. As well, if block/io.c modifies a offset/bytes request, expanding it more then aligning up to request_alignment, it's a bug too. For all io requests except for discard we keep for now old restriction of 32bit request length. iotest 206 output error message changed, as now test disk size is larger than new limit. Add one more test case with new maximum disk size to cover too-big-L1 case. Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Message-Id: <20201203222713.13507-5-vsementsov@virtuozzo.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2020-12-03 23:27:13 +01:00
if (bs->total_sectors * BDRV_SECTOR_SIZE > BDRV_MAX_LENGTH) {
return -EFBIG;
}
return 0;
}
/**
* Combines a QDict of new block driver @options with any missing options taken
* from @old_options, so that leaving out an option defaults to its old value.
*/
static void bdrv_join_options(BlockDriverState *bs, QDict *options,
QDict *old_options)
{
GLOBAL_STATE_CODE();
if (bs->drv && bs->drv->bdrv_join_options) {
bs->drv->bdrv_join_options(options, old_options);
} else {
qdict_join(options, old_options, false);
}
}
static BlockdevDetectZeroesOptions bdrv_parse_detect_zeroes(QemuOpts *opts,
int open_flags,
Error **errp)
{
Error *local_err = NULL;
char *value = qemu_opt_get_del(opts, "detect-zeroes");
BlockdevDetectZeroesOptions detect_zeroes =
qapi_enum_parse(&BlockdevDetectZeroesOptions_lookup, value,
BLOCKDEV_DETECT_ZEROES_OPTIONS_OFF, &local_err);
GLOBAL_STATE_CODE();
g_free(value);
if (local_err) {
error_propagate(errp, local_err);
return detect_zeroes;
}
if (detect_zeroes == BLOCKDEV_DETECT_ZEROES_OPTIONS_UNMAP &&
!(open_flags & BDRV_O_UNMAP))
{
error_setg(errp, "setting detect-zeroes to unmap is not allowed "
"without setting discard operation to unmap");
}
return detect_zeroes;
}
/**
* Set open flags for aio engine
*
* Return 0 on success, -1 if the engine specified is invalid
*/
int bdrv_parse_aio(const char *mode, int *flags)
{
if (!strcmp(mode, "threads")) {
/* do nothing, default */
} else if (!strcmp(mode, "native")) {
*flags |= BDRV_O_NATIVE_AIO;
#ifdef CONFIG_LINUX_IO_URING
} else if (!strcmp(mode, "io_uring")) {
*flags |= BDRV_O_IO_URING;
#endif
} else {
return -1;
}
return 0;
}
/**
* Set open flags for a given discard mode
*
* Return 0 on success, -1 if the discard mode was invalid.
*/
int bdrv_parse_discard_flags(const char *mode, int *flags)
{
*flags &= ~BDRV_O_UNMAP;
if (!strcmp(mode, "off") || !strcmp(mode, "ignore")) {
/* do nothing */
} else if (!strcmp(mode, "on") || !strcmp(mode, "unmap")) {
*flags |= BDRV_O_UNMAP;
} else {
return -1;
}
return 0;
}
/**
* Set open flags for a given cache mode
*
* Return 0 on success, -1 if the cache mode was invalid.
*/
int bdrv_parse_cache_mode(const char *mode, int *flags, bool *writethrough)
{
*flags &= ~BDRV_O_CACHE_MASK;
if (!strcmp(mode, "off") || !strcmp(mode, "none")) {
*writethrough = false;
*flags |= BDRV_O_NOCACHE;
} else if (!strcmp(mode, "directsync")) {
*writethrough = true;
*flags |= BDRV_O_NOCACHE;
} else if (!strcmp(mode, "writeback")) {
*writethrough = false;
} else if (!strcmp(mode, "unsafe")) {
*writethrough = false;
*flags |= BDRV_O_NO_FLUSH;
} else if (!strcmp(mode, "writethrough")) {
*writethrough = true;
} else {
return -1;
}
return 0;
}
static char *bdrv_child_get_parent_desc(BdrvChild *c)
{
BlockDriverState *parent = c->opaque;
return g_strdup_printf("node '%s'", bdrv_get_node_name(parent));
}
static void bdrv_child_cb_drained_begin(BdrvChild *child)
{
BlockDriverState *bs = child->opaque;
bdrv_do_drained_begin_quiesce(bs, NULL, false);
}
static bool bdrv_child_cb_drained_poll(BdrvChild *child)
{
BlockDriverState *bs = child->opaque;
return bdrv_drain_poll(bs, false, NULL, false);
}
block: Do not poll in bdrv_do_drained_end() We should never poll anywhere in bdrv_do_drained_end() (including its recursive callees like bdrv_drain_invoke()), because it does not cope well with graph changes. In fact, it has been written based on the postulation that no graph changes will happen in it. Instead, the callers that want to poll must poll, i.e. all currently globally available wrappers: bdrv_drained_end(), bdrv_subtree_drained_end(), bdrv_unapply_subtree_drain(), and bdrv_drain_all_end(). Graph changes there do not matter. They can poll simply by passing a pointer to a drained_end_counter and wait until it reaches 0. This patch also adds a non-polling global wrapper for bdrv_do_drained_end() that takes a drained_end_counter pointer. We need such a variant because now no function called anywhere from bdrv_do_drained_end() must poll. This includes BdrvChildRole.drained_end(), which already must not poll according to its interface documentation, but bdrv_child_cb_drained_end() just violates that by invoking bdrv_drained_end() (which does poll). Therefore, BdrvChildRole.drained_end() must take a *drained_end_counter parameter, which bdrv_child_cb_drained_end() can pass on to the new bdrv_drained_end_no_poll() function. Note that we now have a pattern of all drained_end-related functions either polling or receiving a *drained_end_counter to let the caller poll based on that. A problem with a single poll loop is that when the drained section in bdrv_set_aio_context_ignore() ends, some nodes in the subgraph may be in the old contexts, while others are in the new context already. To let the collective poll in bdrv_drained_end() work correctly, we must not hold a lock to the old context, so that the old context can make progress in case it is different from the current context. (In the process, remove the comment saying that the current context is always the old context, because it is wrong.) In all other places, all nodes in a subtree must be in the same context, so we can just poll that. The exception of course is bdrv_drain_all_end(), but that always runs in the main context, so we can just poll NULL (like bdrv_drain_all_begin() does). Signed-off-by: Max Reitz <mreitz@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2019-07-19 11:26:14 +02:00
static void bdrv_child_cb_drained_end(BdrvChild *child,
int *drained_end_counter)
{
BlockDriverState *bs = child->opaque;
block: Do not poll in bdrv_do_drained_end() We should never poll anywhere in bdrv_do_drained_end() (including its recursive callees like bdrv_drain_invoke()), because it does not cope well with graph changes. In fact, it has been written based on the postulation that no graph changes will happen in it. Instead, the callers that want to poll must poll, i.e. all currently globally available wrappers: bdrv_drained_end(), bdrv_subtree_drained_end(), bdrv_unapply_subtree_drain(), and bdrv_drain_all_end(). Graph changes there do not matter. They can poll simply by passing a pointer to a drained_end_counter and wait until it reaches 0. This patch also adds a non-polling global wrapper for bdrv_do_drained_end() that takes a drained_end_counter pointer. We need such a variant because now no function called anywhere from bdrv_do_drained_end() must poll. This includes BdrvChildRole.drained_end(), which already must not poll according to its interface documentation, but bdrv_child_cb_drained_end() just violates that by invoking bdrv_drained_end() (which does poll). Therefore, BdrvChildRole.drained_end() must take a *drained_end_counter parameter, which bdrv_child_cb_drained_end() can pass on to the new bdrv_drained_end_no_poll() function. Note that we now have a pattern of all drained_end-related functions either polling or receiving a *drained_end_counter to let the caller poll based on that. A problem with a single poll loop is that when the drained section in bdrv_set_aio_context_ignore() ends, some nodes in the subgraph may be in the old contexts, while others are in the new context already. To let the collective poll in bdrv_drained_end() work correctly, we must not hold a lock to the old context, so that the old context can make progress in case it is different from the current context. (In the process, remove the comment saying that the current context is always the old context, because it is wrong.) In all other places, all nodes in a subtree must be in the same context, so we can just poll that. The exception of course is bdrv_drain_all_end(), but that always runs in the main context, so we can just poll NULL (like bdrv_drain_all_begin() does). Signed-off-by: Max Reitz <mreitz@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2019-07-19 11:26:14 +02:00
bdrv_drained_end_no_poll(bs, drained_end_counter);
}
static int bdrv_child_cb_inactivate(BdrvChild *child)
{
BlockDriverState *bs = child->opaque;
GLOBAL_STATE_CODE();
assert(bs->open_flags & BDRV_O_INACTIVE);
return 0;
}
static bool bdrv_child_cb_can_set_aio_ctx(BdrvChild *child, AioContext *ctx,
GSList **ignore, Error **errp)
{
BlockDriverState *bs = child->opaque;
return bdrv_can_set_aio_context(bs, ctx, ignore, errp);
}
static void bdrv_child_cb_set_aio_ctx(BdrvChild *child, AioContext *ctx,
GSList **ignore)
{
BlockDriverState *bs = child->opaque;
return bdrv_set_aio_context_ignore(bs, ctx, ignore);
}
/*
* Returns the options and flags that a temporary snapshot should get, based on
* the originally requested flags (the originally requested image will have
* flags like a backing file)
*/
static void bdrv_temp_snapshot_options(int *child_flags, QDict *child_options,
int parent_flags, QDict *parent_options)
{
GLOBAL_STATE_CODE();
*child_flags = (parent_flags & ~BDRV_O_SNAPSHOT) | BDRV_O_TEMPORARY;
/* For temporary files, unconditional cache=unsafe is fine */
qdict_set_default_str(child_options, BDRV_OPT_CACHE_DIRECT, "off");
qdict_set_default_str(child_options, BDRV_OPT_CACHE_NO_FLUSH, "on");
/* Copy the read-only and discard options from the parent */
qdict_copy_default(child_options, parent_options, BDRV_OPT_READ_ONLY);
qdict_copy_default(child_options, parent_options, BDRV_OPT_DISCARD);
/* aio=native doesn't work for cache.direct=off, so disable it for the
* temporary snapshot */
*child_flags &= ~BDRV_O_NATIVE_AIO;
}
static void bdrv_backing_attach(BdrvChild *c)
{
BlockDriverState *parent = c->opaque;
BlockDriverState *backing_hd = c->bs;
GLOBAL_STATE_CODE();
assert(!parent->backing_blocker);
error_setg(&parent->backing_blocker,
"node is used as backing hd of '%s'",
bdrv_get_device_or_node_name(parent));
bdrv_refresh_filename(backing_hd);
parent->open_flags &= ~BDRV_O_NO_BACKING;
bdrv_op_block_all(backing_hd, parent->backing_blocker);
/* Otherwise we won't be able to commit or stream */
bdrv_op_unblock(backing_hd, BLOCK_OP_TYPE_COMMIT_TARGET,
parent->backing_blocker);
bdrv_op_unblock(backing_hd, BLOCK_OP_TYPE_STREAM,
parent->backing_blocker);
/*
* We do backup in 3 ways:
* 1. drive backup
* The target bs is new opened, and the source is top BDS
* 2. blockdev backup
* Both the source and the target are top BDSes.
* 3. internal backup(used for block replication)
* Both the source and the target are backing file
*
* In case 1 and 2, neither the source nor the target is the backing file.
* In case 3, we will block the top BDS, so there is only one block job
* for the top BDS and its backing chain.
*/
bdrv_op_unblock(backing_hd, BLOCK_OP_TYPE_BACKUP_SOURCE,
parent->backing_blocker);
bdrv_op_unblock(backing_hd, BLOCK_OP_TYPE_BACKUP_TARGET,
parent->backing_blocker);
}
static void bdrv_backing_detach(BdrvChild *c)
{
BlockDriverState *parent = c->opaque;
GLOBAL_STATE_CODE();
assert(parent->backing_blocker);
bdrv_op_unblock_all(c->bs, parent->backing_blocker);
error_free(parent->backing_blocker);
parent->backing_blocker = NULL;
}
static int bdrv_backing_update_filename(BdrvChild *c, BlockDriverState *base,
const char *filename, Error **errp)
{
BlockDriverState *parent = c->opaque;
bool read_only = bdrv_is_read_only(parent);
int ret;
GLOBAL_STATE_CODE();
if (read_only) {
ret = bdrv_reopen_set_read_only(parent, false, errp);
if (ret < 0) {
return ret;
}
}
ret = bdrv_change_backing_file(parent, filename,
base->drv ? base->drv->format_name : "",
false);
if (ret < 0) {
error_setg_errno(errp, -ret, "Could not update backing file link");
}
if (read_only) {
bdrv_reopen_set_read_only(parent, true, NULL);
}
return ret;
}
/*
* Returns the options and flags that a generic child of a BDS should
* get, based on the given options and flags for the parent BDS.
*/
static void bdrv_inherited_options(BdrvChildRole role, bool parent_is_format,
int *child_flags, QDict *child_options,
int parent_flags, QDict *parent_options)
{
int flags = parent_flags;
GLOBAL_STATE_CODE();
/*
* First, decide whether to set, clear, or leave BDRV_O_PROTOCOL.
* Generally, the question to answer is: Should this child be
* format-probed by default?
*/
/*
* Pure and non-filtered data children of non-format nodes should
* be probed by default (even when the node itself has BDRV_O_PROTOCOL
* set). This only affects a very limited set of drivers (namely
* quorum and blkverify when this comment was written).
* Force-clear BDRV_O_PROTOCOL then.
*/
if (!parent_is_format &&
(role & BDRV_CHILD_DATA) &&
!(role & (BDRV_CHILD_METADATA | BDRV_CHILD_FILTERED)))
{
flags &= ~BDRV_O_PROTOCOL;
}
/*
* All children of format nodes (except for COW children) and all
* metadata children in general should never be format-probed.
* Force-set BDRV_O_PROTOCOL then.
*/
if ((parent_is_format && !(role & BDRV_CHILD_COW)) ||
(role & BDRV_CHILD_METADATA))
{
flags |= BDRV_O_PROTOCOL;
}
/*
* If the cache mode isn't explicitly set, inherit direct and no-flush from
* the parent.
*/
qdict_copy_default(child_options, parent_options, BDRV_OPT_CACHE_DIRECT);
qdict_copy_default(child_options, parent_options, BDRV_OPT_CACHE_NO_FLUSH);
qdict_copy_default(child_options, parent_options, BDRV_OPT_FORCE_SHARE);
if (role & BDRV_CHILD_COW) {
/* backing files are opened read-only by default */
qdict_set_default_str(child_options, BDRV_OPT_READ_ONLY, "on");
qdict_set_default_str(child_options, BDRV_OPT_AUTO_READ_ONLY, "off");
} else {
/* Inherit the read-only option from the parent if it's not set */
qdict_copy_default(child_options, parent_options, BDRV_OPT_READ_ONLY);
qdict_copy_default(child_options, parent_options,
BDRV_OPT_AUTO_READ_ONLY);
}
/*
* bdrv_co_pdiscard() respects unmap policy for the parent, so we
* can default to enable it on lower layers regardless of the
* parent option.
*/
qdict_set_default_str(child_options, BDRV_OPT_DISCARD, "unmap");
/* Clear flags that only apply to the top layer */
flags &= ~(BDRV_O_SNAPSHOT | BDRV_O_NO_BACKING | BDRV_O_COPY_ON_READ);
if (role & BDRV_CHILD_METADATA) {
flags &= ~BDRV_O_NO_IO;
}
if (role & BDRV_CHILD_COW) {
flags &= ~BDRV_O_TEMPORARY;
}
*child_flags = flags;
}
static void bdrv_child_cb_attach(BdrvChild *child)
{
BlockDriverState *bs = child->opaque;
assert_bdrv_graph_writable(bs);
QLIST_INSERT_HEAD(&bs->children, child, next);
if (child->role & BDRV_CHILD_COW) {
bdrv_backing_attach(child);
}
bdrv_apply_subtree_drain(child, bs);
}
static void bdrv_child_cb_detach(BdrvChild *child)
{
BlockDriverState *bs = child->opaque;
if (child->role & BDRV_CHILD_COW) {
bdrv_backing_detach(child);
}
bdrv_unapply_subtree_drain(child, bs);
assert_bdrv_graph_writable(bs);
QLIST_REMOVE(child, next);
}
static int bdrv_child_cb_update_filename(BdrvChild *c, BlockDriverState *base,
const char *filename, Error **errp)
{
if (c->role & BDRV_CHILD_COW) {
return bdrv_backing_update_filename(c, base, filename, errp);
}
return 0;
}
block/vvfat: child_vvfat_qcow: add .get_parent_aio_context, fix crash Commit 3ca1f3225727419ba573673b744edac10904276f "block: BdrvChildClass: add .get_parent_aio_context handler" introduced new handler and commit 228ca37e12f97788e05bd0c92f89b3e5e4019607 "block: drop ctx argument from bdrv_root_attach_child" made a generic use of it. But 3ca1f3225727419ba573673b744edac10904276f didn't update child_vvfat_qcow. Fix that. Before that fix the command ./build/qemu-system-x86_64 -usb -device usb-storage,drive=fat16 \ -drive file=fat:rw:fat-type=16:"<path of a host folder>",id=fat16,format=raw,if=none crashes: 1 bdrv_child_get_parent_aio_context (c=0x559d62426d20) at ../block.c:1440 2 bdrv_attach_child_common (child_bs=0x559d62468190, child_name=0x559d606f9e3d "write-target", child_class=0x559d60c58d20 <child_vvfat_qcow>, child_role=3, perm=3, shared_perm=4, opaque=0x559d62445690, child=0x7ffc74c2acc8, tran=0x559d6246ddd0, errp=0x7ffc74c2ae60) at ../block.c:2795 3 bdrv_attach_child_noperm (parent_bs=0x559d62445690, child_bs=0x559d62468190, child_name=0x559d606f9e3d "write-target", child_class=0x559d60c58d20 <child_vvfat_qcow>, child_role=3, child=0x7ffc74c2acc8, tran=0x559d6246ddd0, errp=0x7ffc74c2ae60) at ../block.c:2855 4 bdrv_attach_child (parent_bs=0x559d62445690, child_bs=0x559d62468190, child_name=0x559d606f9e3d "write-target", child_class=0x559d60c58d20 <child_vvfat_qcow>, child_role=3, errp=0x7ffc74c2ae60) at ../block.c:2953 5 bdrv_open_child (filename=0x559d62464b80 "/var/tmp/vl.h3TIS4", options=0x559d6246ec20, bdref_key=0x559d606f9e3d "write-target", parent=0x559d62445690, child_class=0x559d60c58d20 <child_vvfat_qcow>, child_role=3, allow_none=false, errp=0x7ffc74c2ae60) at ../block.c:3351 6 enable_write_target (bs=0x559d62445690, errp=0x7ffc74c2ae60) at ../block/vvfat.c:3176 7 vvfat_open (bs=0x559d62445690, options=0x559d6244adb0, flags=155650, errp=0x7ffc74c2ae60) at ../block/vvfat.c:1236 8 bdrv_open_driver (bs=0x559d62445690, drv=0x559d60d4f7e0 <bdrv_vvfat>, node_name=0x0, options=0x559d6244adb0, open_flags=155650, errp=0x7ffc74c2af70) at ../block.c:1557 9 bdrv_open_common (bs=0x559d62445690, file=0x0, options=0x559d6244adb0, errp=0x7ffc74c2af70) at ... (gdb) fr 1 #1 0x0000559d603ea3bf in bdrv_child_get_parent_aio_context (c=0x559d62426d20) at ../block.c:1440 1440 return c->klass->get_parent_aio_context(c); (gdb) p c->klass $1 = (const BdrvChildClass *) 0x559d60c58d20 <child_vvfat_qcow> (gdb) p c->klass->get_parent_aio_context $2 = (AioContext *(*)(BdrvChild *)) 0x0 Fixes: 3ca1f3225727419ba573673b744edac10904276f Fixes: 228ca37e12f97788e05bd0c92f89b3e5e4019607 Reported-by: John Arbuckle <programmingkidx@gmail.com> Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Message-Id: <20210524101257.119377-2-vsementsov@virtuozzo.com> Tested-by: John Arbuckle <programmingkidx@gmail.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2021-05-24 12:12:56 +02:00
AioContext *child_of_bds_get_parent_aio_context(BdrvChild *c)
{
BlockDriverState *bs = c->opaque;
IO_CODE();
return bdrv_get_aio_context(bs);
}
const BdrvChildClass child_of_bds = {
.parent_is_bds = true,
.get_parent_desc = bdrv_child_get_parent_desc,
.inherit_options = bdrv_inherited_options,
.drained_begin = bdrv_child_cb_drained_begin,
.drained_poll = bdrv_child_cb_drained_poll,
.drained_end = bdrv_child_cb_drained_end,
.attach = bdrv_child_cb_attach,
.detach = bdrv_child_cb_detach,
.inactivate = bdrv_child_cb_inactivate,
.can_set_aio_ctx = bdrv_child_cb_can_set_aio_ctx,
.set_aio_ctx = bdrv_child_cb_set_aio_ctx,
.update_filename = bdrv_child_cb_update_filename,
block/vvfat: child_vvfat_qcow: add .get_parent_aio_context, fix crash Commit 3ca1f3225727419ba573673b744edac10904276f "block: BdrvChildClass: add .get_parent_aio_context handler" introduced new handler and commit 228ca37e12f97788e05bd0c92f89b3e5e4019607 "block: drop ctx argument from bdrv_root_attach_child" made a generic use of it. But 3ca1f3225727419ba573673b744edac10904276f didn't update child_vvfat_qcow. Fix that. Before that fix the command ./build/qemu-system-x86_64 -usb -device usb-storage,drive=fat16 \ -drive file=fat:rw:fat-type=16:"<path of a host folder>",id=fat16,format=raw,if=none crashes: 1 bdrv_child_get_parent_aio_context (c=0x559d62426d20) at ../block.c:1440 2 bdrv_attach_child_common (child_bs=0x559d62468190, child_name=0x559d606f9e3d "write-target", child_class=0x559d60c58d20 <child_vvfat_qcow>, child_role=3, perm=3, shared_perm=4, opaque=0x559d62445690, child=0x7ffc74c2acc8, tran=0x559d6246ddd0, errp=0x7ffc74c2ae60) at ../block.c:2795 3 bdrv_attach_child_noperm (parent_bs=0x559d62445690, child_bs=0x559d62468190, child_name=0x559d606f9e3d "write-target", child_class=0x559d60c58d20 <child_vvfat_qcow>, child_role=3, child=0x7ffc74c2acc8, tran=0x559d6246ddd0, errp=0x7ffc74c2ae60) at ../block.c:2855 4 bdrv_attach_child (parent_bs=0x559d62445690, child_bs=0x559d62468190, child_name=0x559d606f9e3d "write-target", child_class=0x559d60c58d20 <child_vvfat_qcow>, child_role=3, errp=0x7ffc74c2ae60) at ../block.c:2953 5 bdrv_open_child (filename=0x559d62464b80 "/var/tmp/vl.h3TIS4", options=0x559d6246ec20, bdref_key=0x559d606f9e3d "write-target", parent=0x559d62445690, child_class=0x559d60c58d20 <child_vvfat_qcow>, child_role=3, allow_none=false, errp=0x7ffc74c2ae60) at ../block.c:3351 6 enable_write_target (bs=0x559d62445690, errp=0x7ffc74c2ae60) at ../block/vvfat.c:3176 7 vvfat_open (bs=0x559d62445690, options=0x559d6244adb0, flags=155650, errp=0x7ffc74c2ae60) at ../block/vvfat.c:1236 8 bdrv_open_driver (bs=0x559d62445690, drv=0x559d60d4f7e0 <bdrv_vvfat>, node_name=0x0, options=0x559d6244adb0, open_flags=155650, errp=0x7ffc74c2af70) at ../block.c:1557 9 bdrv_open_common (bs=0x559d62445690, file=0x0, options=0x559d6244adb0, errp=0x7ffc74c2af70) at ... (gdb) fr 1 #1 0x0000559d603ea3bf in bdrv_child_get_parent_aio_context (c=0x559d62426d20) at ../block.c:1440 1440 return c->klass->get_parent_aio_context(c); (gdb) p c->klass $1 = (const BdrvChildClass *) 0x559d60c58d20 <child_vvfat_qcow> (gdb) p c->klass->get_parent_aio_context $2 = (AioContext *(*)(BdrvChild *)) 0x0 Fixes: 3ca1f3225727419ba573673b744edac10904276f Fixes: 228ca37e12f97788e05bd0c92f89b3e5e4019607 Reported-by: John Arbuckle <programmingkidx@gmail.com> Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Message-Id: <20210524101257.119377-2-vsementsov@virtuozzo.com> Tested-by: John Arbuckle <programmingkidx@gmail.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2021-05-24 12:12:56 +02:00
.get_parent_aio_context = child_of_bds_get_parent_aio_context,
};
AioContext *bdrv_child_get_parent_aio_context(BdrvChild *c)
{
GLOBAL_STATE_CODE();
return c->klass->get_parent_aio_context(c);
}
static int bdrv_open_flags(BlockDriverState *bs, int flags)
{
int open_flags = flags;
GLOBAL_STATE_CODE();
/*
* Clear flags that are internal to the block layer before opening the
* image.
*/
open_flags &= ~(BDRV_O_SNAPSHOT | BDRV_O_NO_BACKING | BDRV_O_PROTOCOL);
return open_flags;
}
static void update_flags_from_options(int *flags, QemuOpts *opts)
{
GLOBAL_STATE_CODE();
*flags &= ~(BDRV_O_CACHE_MASK | BDRV_O_RDWR | BDRV_O_AUTO_RDONLY);
block: Forbid trying to change unsupported options during reopen The bdrv_reopen_prepare() function checks all options passed to each BlockDriverState (in the reopen_state->options QDict) and makes all necessary preparations to apply the option changes requested by the user. Options are removed from the QDict as they are processed, so at the end of bdrv_reopen_prepare() only the options that can't be changed are left. Then a loop goes over all remaining options and verifies that the old and new values are identical, returning an error if they're not. The problem is that at the moment there are options that are removed from the QDict although they can't be changed. The consequence of this is any modification to any of those options is silently ignored: (qemu) qemu-io virtio0 "reopen -o discard=on" This happens when all options from bdrv_runtime_opts are removed from the QDict but then only a few of them are processed. Since it's especially important that "node-name" and "driver" are not changed, the code puts them back into the QDict so they are checked at the end of the function. Instead of putting only those two options back into the QDict, this patch puts all unprocessed options using qemu_opts_to_qdict(). update_flags_from_options() also needs to be modified to prevent BDRV_OPT_CACHE_NO_FLUSH, BDRV_OPT_CACHE_DIRECT and BDRV_OPT_READ_ONLY from going back to the QDict. Signed-off-by: Alberto Garcia <berto@igalia.com> Reviewed-by: Max Reitz <mreitz@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2018-09-06 11:37:06 +02:00
if (qemu_opt_get_bool_del(opts, BDRV_OPT_CACHE_NO_FLUSH, false)) {
*flags |= BDRV_O_NO_FLUSH;
}
block: Forbid trying to change unsupported options during reopen The bdrv_reopen_prepare() function checks all options passed to each BlockDriverState (in the reopen_state->options QDict) and makes all necessary preparations to apply the option changes requested by the user. Options are removed from the QDict as they are processed, so at the end of bdrv_reopen_prepare() only the options that can't be changed are left. Then a loop goes over all remaining options and verifies that the old and new values are identical, returning an error if they're not. The problem is that at the moment there are options that are removed from the QDict although they can't be changed. The consequence of this is any modification to any of those options is silently ignored: (qemu) qemu-io virtio0 "reopen -o discard=on" This happens when all options from bdrv_runtime_opts are removed from the QDict but then only a few of them are processed. Since it's especially important that "node-name" and "driver" are not changed, the code puts them back into the QDict so they are checked at the end of the function. Instead of putting only those two options back into the QDict, this patch puts all unprocessed options using qemu_opts_to_qdict(). update_flags_from_options() also needs to be modified to prevent BDRV_OPT_CACHE_NO_FLUSH, BDRV_OPT_CACHE_DIRECT and BDRV_OPT_READ_ONLY from going back to the QDict. Signed-off-by: Alberto Garcia <berto@igalia.com> Reviewed-by: Max Reitz <mreitz@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2018-09-06 11:37:06 +02:00
if (qemu_opt_get_bool_del(opts, BDRV_OPT_CACHE_DIRECT, false)) {
*flags |= BDRV_O_NOCACHE;
}
block: Forbid trying to change unsupported options during reopen The bdrv_reopen_prepare() function checks all options passed to each BlockDriverState (in the reopen_state->options QDict) and makes all necessary preparations to apply the option changes requested by the user. Options are removed from the QDict as they are processed, so at the end of bdrv_reopen_prepare() only the options that can't be changed are left. Then a loop goes over all remaining options and verifies that the old and new values are identical, returning an error if they're not. The problem is that at the moment there are options that are removed from the QDict although they can't be changed. The consequence of this is any modification to any of those options is silently ignored: (qemu) qemu-io virtio0 "reopen -o discard=on" This happens when all options from bdrv_runtime_opts are removed from the QDict but then only a few of them are processed. Since it's especially important that "node-name" and "driver" are not changed, the code puts them back into the QDict so they are checked at the end of the function. Instead of putting only those two options back into the QDict, this patch puts all unprocessed options using qemu_opts_to_qdict(). update_flags_from_options() also needs to be modified to prevent BDRV_OPT_CACHE_NO_FLUSH, BDRV_OPT_CACHE_DIRECT and BDRV_OPT_READ_ONLY from going back to the QDict. Signed-off-by: Alberto Garcia <berto@igalia.com> Reviewed-by: Max Reitz <mreitz@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2018-09-06 11:37:06 +02:00
if (!qemu_opt_get_bool_del(opts, BDRV_OPT_READ_ONLY, false)) {
*flags |= BDRV_O_RDWR;
}
block: Add auto-read-only option If a management application builds the block graph node by node, the protocol layer doesn't inherit its read-only option from the format layer any more, so it must be set explicitly. Backing files should work on read-only storage, but at the same time, a block job like commit should be able to reopen them read-write if they are on read-write storage. However, without option inheritance, reopen only changes the read-only option for the root node (typically the format layer), but not the protocol layer, so reopening fails (the format layer wants to get write permissions, but the protocol layer is still read-only). A simple workaround for the problem in the management tool would be to open the protocol layer always read-write and to make only the format layer read-only for backing files. However, sometimes the file is actually stored on read-only storage and we don't know whether the image can be opened read-write (for example, for NBD it depends on the server we're trying to connect to). This adds an option that makes QEMU try to open the image read-write, but allows it to degrade to a read-only mode without returning an error. The documentation for this option is consciously phrased in a way that allows QEMU to switch to a better model eventually: Instead of trying when the image is first opened, making the read-only flag dynamic and changing it automatically whenever the first BLK_PERM_WRITE user is attached or the last one is detached would be much more useful behaviour. Unfortunately, this more useful behaviour is also a lot harder to implement, and libvirt needs a solution now before it can switch to -blockdev, so let's start with this easier approach for now. Instead of adding a new auto-read-only option, turning the existing read-only into an enum (with a bool alternate for compatibility) was considered, but it complicated the implementation to the point that it didn't seem to be worth it. Signed-off-by: Kevin Wolf <kwolf@redhat.com> Reviewed-by: Eric Blake <eblake@redhat.com>
2018-10-05 18:57:40 +02:00
if (qemu_opt_get_bool_del(opts, BDRV_OPT_AUTO_READ_ONLY, false)) {
*flags |= BDRV_O_AUTO_RDONLY;
}
}
static void update_options_from_flags(QDict *options, int flags)
{
GLOBAL_STATE_CODE();
if (!qdict_haskey(options, BDRV_OPT_CACHE_DIRECT)) {
qdict_put_bool(options, BDRV_OPT_CACHE_DIRECT, flags & BDRV_O_NOCACHE);
}
if (!qdict_haskey(options, BDRV_OPT_CACHE_NO_FLUSH)) {
qdict_put_bool(options, BDRV_OPT_CACHE_NO_FLUSH,
flags & BDRV_O_NO_FLUSH);
}
if (!qdict_haskey(options, BDRV_OPT_READ_ONLY)) {
qdict_put_bool(options, BDRV_OPT_READ_ONLY, !(flags & BDRV_O_RDWR));
}
block: Add auto-read-only option If a management application builds the block graph node by node, the protocol layer doesn't inherit its read-only option from the format layer any more, so it must be set explicitly. Backing files should work on read-only storage, but at the same time, a block job like commit should be able to reopen them read-write if they are on read-write storage. However, without option inheritance, reopen only changes the read-only option for the root node (typically the format layer), but not the protocol layer, so reopening fails (the format layer wants to get write permissions, but the protocol layer is still read-only). A simple workaround for the problem in the management tool would be to open the protocol layer always read-write and to make only the format layer read-only for backing files. However, sometimes the file is actually stored on read-only storage and we don't know whether the image can be opened read-write (for example, for NBD it depends on the server we're trying to connect to). This adds an option that makes QEMU try to open the image read-write, but allows it to degrade to a read-only mode without returning an error. The documentation for this option is consciously phrased in a way that allows QEMU to switch to a better model eventually: Instead of trying when the image is first opened, making the read-only flag dynamic and changing it automatically whenever the first BLK_PERM_WRITE user is attached or the last one is detached would be much more useful behaviour. Unfortunately, this more useful behaviour is also a lot harder to implement, and libvirt needs a solution now before it can switch to -blockdev, so let's start with this easier approach for now. Instead of adding a new auto-read-only option, turning the existing read-only into an enum (with a bool alternate for compatibility) was considered, but it complicated the implementation to the point that it didn't seem to be worth it. Signed-off-by: Kevin Wolf <kwolf@redhat.com> Reviewed-by: Eric Blake <eblake@redhat.com>
2018-10-05 18:57:40 +02:00
if (!qdict_haskey(options, BDRV_OPT_AUTO_READ_ONLY)) {
qdict_put_bool(options, BDRV_OPT_AUTO_READ_ONLY,
flags & BDRV_O_AUTO_RDONLY);
}
}
static void bdrv_assign_node_name(BlockDriverState *bs,
const char *node_name,
Error **errp)
{
char *gen_node_name = NULL;
GLOBAL_STATE_CODE();
if (!node_name) {
node_name = gen_node_name = id_generate(ID_BLOCK);
} else if (!id_wellformed(node_name)) {
/*
* Check for empty string or invalid characters, but not if it is
* generated (generated names use characters not available to the user)
*/
error_setg(errp, "Invalid node-name: '%s'", node_name);
return;
}
/* takes care of avoiding namespaces collisions */
if (blk_by_name(node_name)) {
error_setg(errp, "node-name=%s is conflicting with a device id",
node_name);
goto out;
}
/* takes care of avoiding duplicates node names */
if (bdrv_find_node(node_name)) {
error_setg(errp, "Duplicate nodes with node-name='%s'", node_name);
goto out;
}
/* Make sure that the node name isn't truncated */
if (strlen(node_name) >= sizeof(bs->node_name)) {
error_setg(errp, "Node name too long");
goto out;
}
/* copy node name into the bs and insert it into the graph list */
pstrcpy(bs->node_name, sizeof(bs->node_name), node_name);
QTAILQ_INSERT_TAIL(&graph_bdrv_states, bs, node_list);
out:
g_free(gen_node_name);
}
static int bdrv_open_driver(BlockDriverState *bs, BlockDriver *drv,
const char *node_name, QDict *options,
int open_flags, Error **errp)
{
Error *local_err = NULL;
int i, ret;
GLOBAL_STATE_CODE();
bdrv_assign_node_name(bs, node_name, &local_err);
if (local_err) {
error_propagate(errp, local_err);
return -EINVAL;
}
bs->drv = drv;
bs->opaque = g_malloc0(drv->instance_size);
if (drv->bdrv_file_open) {
assert(!drv->bdrv_needs_filename || bs->filename[0]);
ret = drv->bdrv_file_open(bs, options, open_flags, &local_err);
} else if (drv->bdrv_open) {
ret = drv->bdrv_open(bs, options, open_flags, &local_err);
} else {
ret = 0;
}
if (ret < 0) {
if (local_err) {
error_propagate(errp, local_err);
} else if (bs->filename[0]) {
error_setg_errno(errp, -ret, "Could not open '%s'", bs->filename);
} else {
error_setg_errno(errp, -ret, "Could not open image");
}
goto open_failed;
}
ret = refresh_total_sectors(bs, bs->total_sectors);
if (ret < 0) {
error_setg_errno(errp, -ret, "Could not refresh total sector count");
return ret;
}
bdrv_refresh_limits(bs, NULL, &local_err);
if (local_err) {
error_propagate(errp, local_err);
return -EINVAL;
}
assert(bdrv_opt_mem_align(bs) != 0);
assert(bdrv_min_mem_align(bs) != 0);
assert(is_power_of_2(bs->bl.request_alignment));
for (i = 0; i < bs->quiesce_counter; i++) {
if (drv->bdrv_co_drain_begin) {
drv->bdrv_co_drain_begin(bs);
}
}
return 0;
open_failed:
bs->drv = NULL;
if (bs->file != NULL) {
bdrv_unref_child(bs, bs->file);
bs->file = NULL;
}
g_free(bs->opaque);
bs->opaque = NULL;
return ret;
}
/*
* Create and open a block node.
*
* @options is a QDict of options to pass to the block drivers, or NULL for an
* empty set of options. The reference to the QDict belongs to the block layer
* after the call (even on failure), so if the caller intends to reuse the
* dictionary, it needs to use qobject_ref() before calling bdrv_open.
*/
BlockDriverState *bdrv_new_open_driver_opts(BlockDriver *drv,
const char *node_name,
QDict *options, int flags,
Error **errp)
{
BlockDriverState *bs;
int ret;
GLOBAL_STATE_CODE();
bs = bdrv_new();
bs->open_flags = flags;
bs->options = options ?: qdict_new();
bs->explicit_options = qdict_clone_shallow(bs->options);
bs->opaque = NULL;
update_options_from_flags(bs->options, flags);
ret = bdrv_open_driver(bs, drv, node_name, bs->options, flags, errp);
if (ret < 0) {
qobject_unref(bs->explicit_options);
bs->explicit_options = NULL;
qobject_unref(bs->options);
bs->options = NULL;
bdrv_unref(bs);
return NULL;
}
return bs;
}
/* Create and open a block node. */
BlockDriverState *bdrv_new_open_driver(BlockDriver *drv, const char *node_name,
int flags, Error **errp)
{
GLOBAL_STATE_CODE();
return bdrv_new_open_driver_opts(drv, node_name, NULL, flags, errp);
}
QemuOptsList bdrv_runtime_opts = {
.name = "bdrv_common",
.head = QTAILQ_HEAD_INITIALIZER(bdrv_runtime_opts.head),
.desc = {
{
.name = "node-name",
.type = QEMU_OPT_STRING,
.help = "Node name of the block device node",
},
{
.name = "driver",
.type = QEMU_OPT_STRING,
.help = "Block driver to use for the node",
},
{
.name = BDRV_OPT_CACHE_DIRECT,
.type = QEMU_OPT_BOOL,
.help = "Bypass software writeback cache on the host",
},
{
.name = BDRV_OPT_CACHE_NO_FLUSH,
.type = QEMU_OPT_BOOL,
.help = "Ignore flush requests",
},
{
.name = BDRV_OPT_READ_ONLY,
.type = QEMU_OPT_BOOL,
.help = "Node is opened in read-only mode",
},
block: Add auto-read-only option If a management application builds the block graph node by node, the protocol layer doesn't inherit its read-only option from the format layer any more, so it must be set explicitly. Backing files should work on read-only storage, but at the same time, a block job like commit should be able to reopen them read-write if they are on read-write storage. However, without option inheritance, reopen only changes the read-only option for the root node (typically the format layer), but not the protocol layer, so reopening fails (the format layer wants to get write permissions, but the protocol layer is still read-only). A simple workaround for the problem in the management tool would be to open the protocol layer always read-write and to make only the format layer read-only for backing files. However, sometimes the file is actually stored on read-only storage and we don't know whether the image can be opened read-write (for example, for NBD it depends on the server we're trying to connect to). This adds an option that makes QEMU try to open the image read-write, but allows it to degrade to a read-only mode without returning an error. The documentation for this option is consciously phrased in a way that allows QEMU to switch to a better model eventually: Instead of trying when the image is first opened, making the read-only flag dynamic and changing it automatically whenever the first BLK_PERM_WRITE user is attached or the last one is detached would be much more useful behaviour. Unfortunately, this more useful behaviour is also a lot harder to implement, and libvirt needs a solution now before it can switch to -blockdev, so let's start with this easier approach for now. Instead of adding a new auto-read-only option, turning the existing read-only into an enum (with a bool alternate for compatibility) was considered, but it complicated the implementation to the point that it didn't seem to be worth it. Signed-off-by: Kevin Wolf <kwolf@redhat.com> Reviewed-by: Eric Blake <eblake@redhat.com>
2018-10-05 18:57:40 +02:00
{
.name = BDRV_OPT_AUTO_READ_ONLY,
.type = QEMU_OPT_BOOL,
.help = "Node can become read-only if opening read-write fails",
},
{
.name = "detect-zeroes",
.type = QEMU_OPT_STRING,
.help = "try to optimize zero writes (off, on, unmap)",
},
{
.name = BDRV_OPT_DISCARD,
.type = QEMU_OPT_STRING,
.help = "discard operation (ignore/off, unmap/on)",
},
{
.name = BDRV_OPT_FORCE_SHARE,
.type = QEMU_OPT_BOOL,
.help = "always accept other writers (default: off)",
},
{ /* end of list */ }
},
};
QemuOptsList bdrv_create_opts_simple = {
.name = "simple-create-opts",
.head = QTAILQ_HEAD_INITIALIZER(bdrv_create_opts_simple.head),
.desc = {
{
.name = BLOCK_OPT_SIZE,
.type = QEMU_OPT_SIZE,
.help = "Virtual disk size"
},
{
.name = BLOCK_OPT_PREALLOC,
.type = QEMU_OPT_STRING,
.help = "Preallocation mode (allowed values: off)"
},
{ /* end of list */ }
}
};
/*
* Common part for opening disk images and files
*
* Removes all processed options from *options.
*/
static int bdrv_open_common(BlockDriverState *bs, BlockBackend *file,
QDict *options, Error **errp)
{
int ret, open_flags;
const char *filename;
const char *driver_name = NULL;
const char *node_name = NULL;
const char *discard;
QemuOpts *opts;
BlockDriver *drv;
Error *local_err = NULL;
bool ro;
assert(bs->file == NULL);
assert(options != NULL && bs->options != options);
GLOBAL_STATE_CODE();
opts = qemu_opts_create(&bdrv_runtime_opts, NULL, 0, &error_abort);
if (!qemu_opts_absorb_qdict(opts, options, errp)) {
ret = -EINVAL;
goto fail_opts;
}
update_flags_from_options(&bs->open_flags, opts);
driver_name = qemu_opt_get(opts, "driver");
drv = bdrv_find_format(driver_name);
assert(drv != NULL);
bs->force_share = qemu_opt_get_bool(opts, BDRV_OPT_FORCE_SHARE, false);
if (bs->force_share && (bs->open_flags & BDRV_O_RDWR)) {
error_setg(errp,
BDRV_OPT_FORCE_SHARE
"=on can only be used with read-only images");
ret = -EINVAL;
goto fail_opts;
}
if (file != NULL) {
bdrv_refresh_filename(blk_bs(file));
filename = blk_bs(file)->filename;
} else {
block: Document -drive problematic code and bugs -blockdev and blockdev_add convert their arguments via QObject to BlockdevOptions for qmp_blockdev_add(), which converts them back to QObject, then to a flattened QDict. The QDict's members are typed according to the QAPI schema. -drive converts its argument via QemuOpts to a (flat) QDict. This QDict's members are all QString. Thus, the QType of a flat QDict member depends on whether it comes from -drive or -blockdev/blockdev_add, except when the QAPI type maps to QString, which is the case for 'str' and enumeration types. The block layer core extracts generic configuration from the flat QDict, and the block driver extracts driver-specific configuration. Both commonly do so by converting (parts of) the flat QDict to QemuOpts, which turns all values into strings. Not exactly elegant, but correct. However, A few places access the flat QDict directly: * Most of them access members that are always QString. Correct. * bdrv_open_inherit() accesses a boolean, carefully. Correct. * nfs_config() uses a QObject input visitor. Correct only because the visited type contains nothing but QStrings. * nbd_config() and ssh_config() use a QObject input visitor, and the visited types contain non-QStrings: InetSocketAddress members @numeric, @to, @ipv4, @ipv6. -drive works as long as you don't try to use them (they're all optional). @to is ignored anyway. Reproducer: -drive driver=ssh,server.host=h,server.port=22,server.ipv4,path=p -drive driver=nbd,server.type=inet,server.data.host=h,server.data.port=22,server.data.ipv4 both fail with "Invalid parameter type for 'data.ipv4', expected: boolean" Add suitable comments to all these places. Mark the buggy ones FIXME. "Fortunately", -drive's driver-specific options are entirely undocumented. Signed-off-by: Markus Armbruster <armbru@redhat.com> Message-id: 1490895797-29094-5-git-send-email-armbru@redhat.com [mreitz: Fixed two typos] Reviewed-by: Eric Blake <eblake@redhat.com> Signed-off-by: Max Reitz <mreitz@redhat.com>
2017-03-30 19:43:12 +02:00
/*
* Caution: while qdict_get_try_str() is fine, getting
* non-string types would require more care. When @options
* come from -blockdev or blockdev_add, its members are typed
* according to the QAPI schema, but when they come from
* -drive, they're all QString.
*/
filename = qdict_get_try_str(options, "filename");
}
if (drv->bdrv_needs_filename && (!filename || !filename[0])) {
error_setg(errp, "The '%s' block driver requires a file name",
drv->format_name);
ret = -EINVAL;
goto fail_opts;
}
trace_bdrv_open_common(bs, filename ?: "", bs->open_flags,
drv->format_name);
ro = bdrv_is_read_only(bs);
if (use_bdrv_whitelist && !bdrv_is_whitelisted(drv, ro)) {
if (!ro && bdrv_is_whitelisted(drv, true)) {
ret = bdrv_apply_auto_read_only(bs, NULL, NULL);
} else {
ret = -ENOTSUP;
}
if (ret < 0) {
error_setg(errp,
!ro && bdrv_is_whitelisted(drv, true)
? "Driver '%s' can only be used for read-only devices"
: "Driver '%s' is not whitelisted",
drv->format_name);
goto fail_opts;
}
}
/* bdrv_new() and bdrv_close() make it so */
assert(qatomic_read(&bs->copy_on_read) == 0);
if (bs->open_flags & BDRV_O_COPY_ON_READ) {
if (!ro) {
bdrv_enable_copy_on_read(bs);
} else {
error_setg(errp, "Can't use copy-on-read on read-only device");
ret = -EINVAL;
goto fail_opts;
}
}
discard = qemu_opt_get(opts, BDRV_OPT_DISCARD);
if (discard != NULL) {
if (bdrv_parse_discard_flags(discard, &bs->open_flags) != 0) {
error_setg(errp, "Invalid discard option");
ret = -EINVAL;
goto fail_opts;
}
}
bs->detect_zeroes =
bdrv_parse_detect_zeroes(opts, bs->open_flags, &local_err);
if (local_err) {
error_propagate(errp, local_err);
ret = -EINVAL;
goto fail_opts;
}
if (filename != NULL) {
pstrcpy(bs->filename, sizeof(bs->filename), filename);
} else {
bs->filename[0] = '\0';
}
pstrcpy(bs->exact_filename, sizeof(bs->exact_filename), bs->filename);
/* Open the image, either directly or using a protocol */
open_flags = bdrv_open_flags(bs, bs->open_flags);
node_name = qemu_opt_get(opts, "node-name");
assert(!drv->bdrv_file_open || file == NULL);
ret = bdrv_open_driver(bs, drv, node_name, options, open_flags, errp);
if (ret < 0) {
goto fail_opts;
}
qemu_opts_del(opts);
return 0;
fail_opts:
qemu_opts_del(opts);
return ret;
}
static QDict *parse_json_filename(const char *filename, Error **errp)
{
QObject *options_obj;
QDict *options;
int ret;
GLOBAL_STATE_CODE();
ret = strstart(filename, "json:", &filename);
assert(ret);
options_obj = qobject_from_json(filename, errp);
if (!options_obj) {
error_prepend(errp, "Could not parse the JSON options: ");
return NULL;
}
options = qobject_to(QDict, options_obj);
if (!options) {
qobject_unref(options_obj);
error_setg(errp, "Invalid JSON object given");
return NULL;
}
qdict_flatten(options);
return options;
}
static void parse_json_protocol(QDict *options, const char **pfilename,
Error **errp)
{
QDict *json_options;
Error *local_err = NULL;
GLOBAL_STATE_CODE();
/* Parse json: pseudo-protocol */
if (!*pfilename || !g_str_has_prefix(*pfilename, "json:")) {
return;
}
json_options = parse_json_filename(*pfilename, &local_err);
if (local_err) {
error_propagate(errp, local_err);
return;
}
/* Options given in the filename have lower priority than options
* specified directly */
qdict_join(options, json_options, false);
qobject_unref(json_options);
*pfilename = NULL;
}
/*
* Fills in default options for opening images and converts the legacy
* filename/flags pair to option QDict entries.
block: driver should override flags in bdrv_open() The BDRV_O_PROTOCOL flag should have an impact only if no driver is specified explicitly. Therefore, if bdrv_open() is called with an explicit block driver argument (either through the options QDict or through the drv parameter) and that block driver is a protocol block driver, BDRV_O_PROTOCOL should be set; if it is a format block driver, BDRV_O_PROTOCOL should be unset. While there was code to unset the flag in case a format block driver has been selected, it only followed the bdrv_fill_options() function call whereas the flag in fact needs to be adjusted before it is used there. With that change, BDRV_O_PROTOCOL will always be set if the BDS should be a protocol driver; if the driver has been specified explicitly, the new code will set it; and bdrv_fill_options() will only "probe" a protocol driver if BDRV_O_PROTOCOL is set. The probing after bdrv_fill_options() cannot select a protocol driver. Thus, bdrv_open_image() to open BDS.file is never called if a protocol BDS is about to be created. With that change in turn it is impossible to call bdrv_open_common() with a protocol drv and file != NULL, which allows us to remove the bdrv_swap() call. This change breaks a test case in qemu-iotest 051: "-drive file=t.qcow2,file.driver=qcow2" now works because the explicitly specified "qcow2" overrides the BDRV_O_PROTOCOL which is automatically set for the "file" BDS (and the filename is just passed down). Therefore, this patch removes that test case. Signed-off-by: Max Reitz <mreitz@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2015-03-19 19:53:16 +01:00
* The BDRV_O_PROTOCOL flag in *flags will be set or cleared accordingly if a
* block driver has been specified explicitly.
*/
static int bdrv_fill_options(QDict **options, const char *filename,
int *flags, Error **errp)
{
const char *drvname;
block: driver should override flags in bdrv_open() The BDRV_O_PROTOCOL flag should have an impact only if no driver is specified explicitly. Therefore, if bdrv_open() is called with an explicit block driver argument (either through the options QDict or through the drv parameter) and that block driver is a protocol block driver, BDRV_O_PROTOCOL should be set; if it is a format block driver, BDRV_O_PROTOCOL should be unset. While there was code to unset the flag in case a format block driver has been selected, it only followed the bdrv_fill_options() function call whereas the flag in fact needs to be adjusted before it is used there. With that change, BDRV_O_PROTOCOL will always be set if the BDS should be a protocol driver; if the driver has been specified explicitly, the new code will set it; and bdrv_fill_options() will only "probe" a protocol driver if BDRV_O_PROTOCOL is set. The probing after bdrv_fill_options() cannot select a protocol driver. Thus, bdrv_open_image() to open BDS.file is never called if a protocol BDS is about to be created. With that change in turn it is impossible to call bdrv_open_common() with a protocol drv and file != NULL, which allows us to remove the bdrv_swap() call. This change breaks a test case in qemu-iotest 051: "-drive file=t.qcow2,file.driver=qcow2" now works because the explicitly specified "qcow2" overrides the BDRV_O_PROTOCOL which is automatically set for the "file" BDS (and the filename is just passed down). Therefore, this patch removes that test case. Signed-off-by: Max Reitz <mreitz@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2015-03-19 19:53:16 +01:00
bool protocol = *flags & BDRV_O_PROTOCOL;
bool parse_filename = false;
BlockDriver *drv = NULL;
Error *local_err = NULL;
GLOBAL_STATE_CODE();
block: Document -drive problematic code and bugs -blockdev and blockdev_add convert their arguments via QObject to BlockdevOptions for qmp_blockdev_add(), which converts them back to QObject, then to a flattened QDict. The QDict's members are typed according to the QAPI schema. -drive converts its argument via QemuOpts to a (flat) QDict. This QDict's members are all QString. Thus, the QType of a flat QDict member depends on whether it comes from -drive or -blockdev/blockdev_add, except when the QAPI type maps to QString, which is the case for 'str' and enumeration types. The block layer core extracts generic configuration from the flat QDict, and the block driver extracts driver-specific configuration. Both commonly do so by converting (parts of) the flat QDict to QemuOpts, which turns all values into strings. Not exactly elegant, but correct. However, A few places access the flat QDict directly: * Most of them access members that are always QString. Correct. * bdrv_open_inherit() accesses a boolean, carefully. Correct. * nfs_config() uses a QObject input visitor. Correct only because the visited type contains nothing but QStrings. * nbd_config() and ssh_config() use a QObject input visitor, and the visited types contain non-QStrings: InetSocketAddress members @numeric, @to, @ipv4, @ipv6. -drive works as long as you don't try to use them (they're all optional). @to is ignored anyway. Reproducer: -drive driver=ssh,server.host=h,server.port=22,server.ipv4,path=p -drive driver=nbd,server.type=inet,server.data.host=h,server.data.port=22,server.data.ipv4 both fail with "Invalid parameter type for 'data.ipv4', expected: boolean" Add suitable comments to all these places. Mark the buggy ones FIXME. "Fortunately", -drive's driver-specific options are entirely undocumented. Signed-off-by: Markus Armbruster <armbru@redhat.com> Message-id: 1490895797-29094-5-git-send-email-armbru@redhat.com [mreitz: Fixed two typos] Reviewed-by: Eric Blake <eblake@redhat.com> Signed-off-by: Max Reitz <mreitz@redhat.com>
2017-03-30 19:43:12 +02:00
/*
* Caution: while qdict_get_try_str() is fine, getting non-string
* types would require more care. When @options come from
* -blockdev or blockdev_add, its members are typed according to
* the QAPI schema, but when they come from -drive, they're all
* QString.
*/
block: driver should override flags in bdrv_open() The BDRV_O_PROTOCOL flag should have an impact only if no driver is specified explicitly. Therefore, if bdrv_open() is called with an explicit block driver argument (either through the options QDict or through the drv parameter) and that block driver is a protocol block driver, BDRV_O_PROTOCOL should be set; if it is a format block driver, BDRV_O_PROTOCOL should be unset. While there was code to unset the flag in case a format block driver has been selected, it only followed the bdrv_fill_options() function call whereas the flag in fact needs to be adjusted before it is used there. With that change, BDRV_O_PROTOCOL will always be set if the BDS should be a protocol driver; if the driver has been specified explicitly, the new code will set it; and bdrv_fill_options() will only "probe" a protocol driver if BDRV_O_PROTOCOL is set. The probing after bdrv_fill_options() cannot select a protocol driver. Thus, bdrv_open_image() to open BDS.file is never called if a protocol BDS is about to be created. With that change in turn it is impossible to call bdrv_open_common() with a protocol drv and file != NULL, which allows us to remove the bdrv_swap() call. This change breaks a test case in qemu-iotest 051: "-drive file=t.qcow2,file.driver=qcow2" now works because the explicitly specified "qcow2" overrides the BDRV_O_PROTOCOL which is automatically set for the "file" BDS (and the filename is just passed down). Therefore, this patch removes that test case. Signed-off-by: Max Reitz <mreitz@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2015-03-19 19:53:16 +01:00
drvname = qdict_get_try_str(*options, "driver");
if (drvname) {
drv = bdrv_find_format(drvname);
if (!drv) {
error_setg(errp, "Unknown driver '%s'", drvname);
return -ENOENT;
}
/* If the user has explicitly specified the driver, this choice should
* override the BDRV_O_PROTOCOL flag */
protocol = drv->bdrv_file_open;
block: driver should override flags in bdrv_open() The BDRV_O_PROTOCOL flag should have an impact only if no driver is specified explicitly. Therefore, if bdrv_open() is called with an explicit block driver argument (either through the options QDict or through the drv parameter) and that block driver is a protocol block driver, BDRV_O_PROTOCOL should be set; if it is a format block driver, BDRV_O_PROTOCOL should be unset. While there was code to unset the flag in case a format block driver has been selected, it only followed the bdrv_fill_options() function call whereas the flag in fact needs to be adjusted before it is used there. With that change, BDRV_O_PROTOCOL will always be set if the BDS should be a protocol driver; if the driver has been specified explicitly, the new code will set it; and bdrv_fill_options() will only "probe" a protocol driver if BDRV_O_PROTOCOL is set. The probing after bdrv_fill_options() cannot select a protocol driver. Thus, bdrv_open_image() to open BDS.file is never called if a protocol BDS is about to be created. With that change in turn it is impossible to call bdrv_open_common() with a protocol drv and file != NULL, which allows us to remove the bdrv_swap() call. This change breaks a test case in qemu-iotest 051: "-drive file=t.qcow2,file.driver=qcow2" now works because the explicitly specified "qcow2" overrides the BDRV_O_PROTOCOL which is automatically set for the "file" BDS (and the filename is just passed down). Therefore, this patch removes that test case. Signed-off-by: Max Reitz <mreitz@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2015-03-19 19:53:16 +01:00
}
if (protocol) {
*flags |= BDRV_O_PROTOCOL;
} else {
*flags &= ~BDRV_O_PROTOCOL;
}
/* Translate cache options from flags into options */
update_options_from_flags(*options, *flags);
/* Fetch the file name from the options QDict if necessary */
if (protocol && filename) {
if (!qdict_haskey(*options, "filename")) {
qdict_put_str(*options, "filename", filename);
parse_filename = true;
} else {
error_setg(errp, "Can't specify 'file' and 'filename' options at "
"the same time");
return -EINVAL;
}
}
/* Find the right block driver */
block: Document -drive problematic code and bugs -blockdev and blockdev_add convert their arguments via QObject to BlockdevOptions for qmp_blockdev_add(), which converts them back to QObject, then to a flattened QDict. The QDict's members are typed according to the QAPI schema. -drive converts its argument via QemuOpts to a (flat) QDict. This QDict's members are all QString. Thus, the QType of a flat QDict member depends on whether it comes from -drive or -blockdev/blockdev_add, except when the QAPI type maps to QString, which is the case for 'str' and enumeration types. The block layer core extracts generic configuration from the flat QDict, and the block driver extracts driver-specific configuration. Both commonly do so by converting (parts of) the flat QDict to QemuOpts, which turns all values into strings. Not exactly elegant, but correct. However, A few places access the flat QDict directly: * Most of them access members that are always QString. Correct. * bdrv_open_inherit() accesses a boolean, carefully. Correct. * nfs_config() uses a QObject input visitor. Correct only because the visited type contains nothing but QStrings. * nbd_config() and ssh_config() use a QObject input visitor, and the visited types contain non-QStrings: InetSocketAddress members @numeric, @to, @ipv4, @ipv6. -drive works as long as you don't try to use them (they're all optional). @to is ignored anyway. Reproducer: -drive driver=ssh,server.host=h,server.port=22,server.ipv4,path=p -drive driver=nbd,server.type=inet,server.data.host=h,server.data.port=22,server.data.ipv4 both fail with "Invalid parameter type for 'data.ipv4', expected: boolean" Add suitable comments to all these places. Mark the buggy ones FIXME. "Fortunately", -drive's driver-specific options are entirely undocumented. Signed-off-by: Markus Armbruster <armbru@redhat.com> Message-id: 1490895797-29094-5-git-send-email-armbru@redhat.com [mreitz: Fixed two typos] Reviewed-by: Eric Blake <eblake@redhat.com> Signed-off-by: Max Reitz <mreitz@redhat.com>
2017-03-30 19:43:12 +02:00
/* See cautionary note on accessing @options above */
filename = qdict_get_try_str(*options, "filename");
if (!drvname && protocol) {
if (filename) {
drv = bdrv_find_protocol(filename, parse_filename, errp);
if (!drv) {
return -EINVAL;
}
drvname = drv->format_name;
qdict_put_str(*options, "driver", drvname);
} else {
error_setg(errp, "Must specify either driver or file");
return -EINVAL;
}
}
assert(drv || !protocol);
/* Driver-specific filename parsing */
if (drv && drv->bdrv_parse_filename && parse_filename) {
drv->bdrv_parse_filename(filename, *options, &local_err);
if (local_err) {
error_propagate(errp, local_err);
return -EINVAL;
}
if (!drv->bdrv_needs_filename) {
qdict_del(*options, "filename");
}
}
return 0;
}
typedef struct BlockReopenQueueEntry {
bool prepared;
bool perms_checked;
BDRVReopenState state;
QTAILQ_ENTRY(BlockReopenQueueEntry) entry;
} BlockReopenQueueEntry;
/*
* Return the flags that @bs will have after the reopens in @q have
* successfully completed. If @q is NULL (or @bs is not contained in @q),
* return the current flags.
*/
static int bdrv_reopen_get_flags(BlockReopenQueue *q, BlockDriverState *bs)
{
BlockReopenQueueEntry *entry;
if (q != NULL) {
QTAILQ_FOREACH(entry, q, entry) {
if (entry->state.bs == bs) {
return entry->state.flags;
}
}
}
return bs->open_flags;
}
/* Returns whether the image file can be written to after the reopen queue @q
* has been successfully applied, or right now if @q is NULL. */
static bool bdrv_is_writable_after_reopen(BlockDriverState *bs,
BlockReopenQueue *q)
{
int flags = bdrv_reopen_get_flags(q, bs);
return (flags & (BDRV_O_RDWR | BDRV_O_INACTIVE)) == BDRV_O_RDWR;
}
/*
* Return whether the BDS can be written to. This is not necessarily
* the same as !bdrv_is_read_only(bs), as inactivated images may not
* be written to but do not count as read-only images.
*/
bool bdrv_is_writable(BlockDriverState *bs)
{
IO_CODE();
return bdrv_is_writable_after_reopen(bs, NULL);
}
static char *bdrv_child_user_desc(BdrvChild *c)
{
GLOBAL_STATE_CODE();
return c->klass->get_parent_desc(c);
}
/*
* Check that @a allows everything that @b needs. @a and @b must reference same
* child node.
*/
static bool bdrv_a_allow_b(BdrvChild *a, BdrvChild *b, Error **errp)
{
const char *child_bs_name;
g_autofree char *a_user = NULL;
g_autofree char *b_user = NULL;
g_autofree char *perms = NULL;
assert(a->bs);
assert(a->bs == b->bs);
GLOBAL_STATE_CODE();
if ((b->perm & a->shared_perm) == b->perm) {
return true;
}
child_bs_name = bdrv_get_node_name(b->bs);
a_user = bdrv_child_user_desc(a);
b_user = bdrv_child_user_desc(b);
perms = bdrv_perm_names(b->perm & ~a->shared_perm);
error_setg(errp, "Permission conflict on node '%s': permissions '%s' are "
"both required by %s (uses node '%s' as '%s' child) and "
"unshared by %s (uses node '%s' as '%s' child).",
child_bs_name, perms,
b_user, child_bs_name, b->name,
a_user, child_bs_name, a->name);
return false;
}
static bool bdrv_parent_perms_conflict(BlockDriverState *bs, Error **errp)
{
BdrvChild *a, *b;
GLOBAL_STATE_CODE();
/*
* During the loop we'll look at each pair twice. That's correct because
* bdrv_a_allow_b() is asymmetric and we should check each pair in both
* directions.
*/
QLIST_FOREACH(a, &bs->parents, next_parent) {
QLIST_FOREACH(b, &bs->parents, next_parent) {
if (a == b) {
continue;
}
if (!bdrv_a_allow_b(a, b, errp)) {
return true;
}
}
}
return false;
}
static void bdrv_child_perm(BlockDriverState *bs, BlockDriverState *child_bs,
BdrvChild *c, BdrvChildRole role,
BlockReopenQueue *reopen_queue,
uint64_t parent_perm, uint64_t parent_shared,
uint64_t *nperm, uint64_t *nshared)
{
assert(bs->drv && bs->drv->bdrv_child_perm);
GLOBAL_STATE_CODE();
bs->drv->bdrv_child_perm(bs, c, role, reopen_queue,
parent_perm, parent_shared,
nperm, nshared);
/* TODO Take force_share from reopen_queue */
if (child_bs && child_bs->force_share) {
*nshared = BLK_PERM_ALL;
}
}
block: use topological sort for permission update Rewrite bdrv_check_perm(), bdrv_abort_perm_update() and bdrv_set_perm() to update nodes in topological sort order instead of simple DFS. With topologically sorted nodes, we update a node only when all its parents already updated. With DFS it's not so. Consider the following example: A -+ | | | v | B | | v | C<-+ A is parent for B and C, B is parent for C. Obviously, to update permissions, we should go in order A B C, so, when we update C, all parent permissions already updated. But with current approach (simple recursion) we can update in sequence A C B C (C is updated twice). On first update of C, we consider old B permissions, so doing wrong thing. If it succeed, all is OK, on second C update we will finish with correct graph. But if the wrong thing failed, we break the whole process for no reason (it's possible that updated B permission will be less strict, but we will never check it). Also new approach gives a way to simultaneously and correctly update several nodes, we just need to run bdrv_topological_dfs() several times to add all nodes and their subtrees into one topologically sorted list (next patch will update bdrv_replace_node() in this manner). Test test_parallel_perm_update() is now passing, so move it out of debugging "if". We also need to support ignore_children in bdrv_parent_perms_conflict() For test 283 order of conflicting parents check is changed. Note also that in bdrv_check_perm() we don't check for parents conflict at root bs, as we may be in the middle of permission update in bdrv_reopen_multiple(). bdrv_reopen_multiple() will be updated soon. Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Reviewed-by: Kevin Wolf <kwolf@redhat.com> Message-Id: <20210428151804.439460-14-vsementsov@virtuozzo.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2021-04-28 17:17:41 +02:00
/*
* Adds the whole subtree of @bs (including @bs itself) to the @list (except for
* nodes that are already in the @list, of course) so that final list is
* topologically sorted. Return the result (GSList @list object is updated, so
* don't use old reference after function call).
*
* On function start @list must be already topologically sorted and for any node
* in the @list the whole subtree of the node must be in the @list as well. The
* simplest way to satisfy this criteria: use only result of
* bdrv_topological_dfs() or NULL as @list parameter.
*/
static GSList *bdrv_topological_dfs(GSList *list, GHashTable *found,
BlockDriverState *bs)
{
BdrvChild *child;
g_autoptr(GHashTable) local_found = NULL;
GLOBAL_STATE_CODE();
block: use topological sort for permission update Rewrite bdrv_check_perm(), bdrv_abort_perm_update() and bdrv_set_perm() to update nodes in topological sort order instead of simple DFS. With topologically sorted nodes, we update a node only when all its parents already updated. With DFS it's not so. Consider the following example: A -+ | | | v | B | | v | C<-+ A is parent for B and C, B is parent for C. Obviously, to update permissions, we should go in order A B C, so, when we update C, all parent permissions already updated. But with current approach (simple recursion) we can update in sequence A C B C (C is updated twice). On first update of C, we consider old B permissions, so doing wrong thing. If it succeed, all is OK, on second C update we will finish with correct graph. But if the wrong thing failed, we break the whole process for no reason (it's possible that updated B permission will be less strict, but we will never check it). Also new approach gives a way to simultaneously and correctly update several nodes, we just need to run bdrv_topological_dfs() several times to add all nodes and their subtrees into one topologically sorted list (next patch will update bdrv_replace_node() in this manner). Test test_parallel_perm_update() is now passing, so move it out of debugging "if". We also need to support ignore_children in bdrv_parent_perms_conflict() For test 283 order of conflicting parents check is changed. Note also that in bdrv_check_perm() we don't check for parents conflict at root bs, as we may be in the middle of permission update in bdrv_reopen_multiple(). bdrv_reopen_multiple() will be updated soon. Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Reviewed-by: Kevin Wolf <kwolf@redhat.com> Message-Id: <20210428151804.439460-14-vsementsov@virtuozzo.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2021-04-28 17:17:41 +02:00
if (!found) {
assert(!list);
found = local_found = g_hash_table_new(NULL, NULL);
}
if (g_hash_table_contains(found, bs)) {
return list;
}
g_hash_table_add(found, bs);
QLIST_FOREACH(child, &bs->children, next) {
list = bdrv_topological_dfs(list, found, child->bs);
}
return g_slist_prepend(list, bs);
}
typedef struct BdrvChildSetPermState {
BdrvChild *child;
uint64_t old_perm;
uint64_t old_shared_perm;
} BdrvChildSetPermState;
static void bdrv_child_set_perm_abort(void *opaque)
{
BdrvChildSetPermState *s = opaque;
GLOBAL_STATE_CODE();
s->child->perm = s->old_perm;
s->child->shared_perm = s->old_shared_perm;
}
static TransactionActionDrv bdrv_child_set_pem_drv = {
.abort = bdrv_child_set_perm_abort,
.clean = g_free,
};
static void bdrv_child_set_perm(BdrvChild *c, uint64_t perm,
uint64_t shared, Transaction *tran)
{
BdrvChildSetPermState *s = g_new(BdrvChildSetPermState, 1);
GLOBAL_STATE_CODE();
*s = (BdrvChildSetPermState) {
.child = c,
.old_perm = c->perm,
.old_shared_perm = c->shared_perm,
};
c->perm = perm;
c->shared_perm = shared;
tran_add(tran, &bdrv_child_set_pem_drv, s);
}
static void bdrv_drv_set_perm_commit(void *opaque)
{
BlockDriverState *bs = opaque;
uint64_t cumulative_perms, cumulative_shared_perms;
GLOBAL_STATE_CODE();
if (bs->drv->bdrv_set_perm) {
bdrv_get_cumulative_perm(bs, &cumulative_perms,
&cumulative_shared_perms);
bs->drv->bdrv_set_perm(bs, cumulative_perms, cumulative_shared_perms);
}
}
static void bdrv_drv_set_perm_abort(void *opaque)
{
BlockDriverState *bs = opaque;
GLOBAL_STATE_CODE();
if (bs->drv->bdrv_abort_perm_update) {
bs->drv->bdrv_abort_perm_update(bs);
}
}
TransactionActionDrv bdrv_drv_set_perm_drv = {
.abort = bdrv_drv_set_perm_abort,
.commit = bdrv_drv_set_perm_commit,
};
static int bdrv_drv_set_perm(BlockDriverState *bs, uint64_t perm,
uint64_t shared_perm, Transaction *tran,
Error **errp)
{
GLOBAL_STATE_CODE();
if (!bs->drv) {
return 0;
}
if (bs->drv->bdrv_check_perm) {
int ret = bs->drv->bdrv_check_perm(bs, perm, shared_perm, errp);
if (ret < 0) {
return ret;
}
}
if (tran) {
tran_add(tran, &bdrv_drv_set_perm_drv, bs);
}
return 0;
}
typedef struct BdrvReplaceChildState {
BdrvChild *child;
block: Let replace_child_tran keep indirect pointer As of a future commit, bdrv_replace_child_noperm() will clear the indirect BdrvChild pointer passed to it if the new child BDS is NULL. bdrv_replace_child_tran() will want to let it do that, but revert this change in its abort handler. For that, we need to have it receive a BdrvChild ** pointer, too, and keep it stored in the BdrvReplaceChildState object that we attach to the transaction. Note that we do not need to store it in the BdrvReplaceChildState when new_bs is not NULL, because then there is nothing to revert. This is important so that bdrv_replace_node_noperm() can pass a pointer to a loop-local variable to bdrv_replace_child_tran() without worrying that this pointer will outlive one loop iteration. (Of course, for that to work, bdrv_replace_node_noperm() and in turn bdrv_replace_node() and its relatives may not be called with a NULL @to node. Luckily, they already are not, but now we should assert this.) bdrv_remove_file_or_backing_child() on the other hand needs to ensure that the indirect pointer it passes will stay valid for the duration of the transaction. Ensure this by keeping a strong reference to the BDS whose &bs->backing or &bs->file it passes to bdrv_replace_child_tran(), and giving up that reference only in the transaction .clean() handler. Signed-off-by: Hanna Reitz <hreitz@redhat.com> Message-Id: <20211111120829.81329-9-hreitz@redhat.com> Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> Message-Id: <20211115145409.176785-9-kwolf@redhat.com> Signed-off-by: Hanna Reitz <hreitz@redhat.com>
2021-11-15 15:54:04 +01:00
BdrvChild **childp;
BlockDriverState *old_bs;
bool free_empty_child;
} BdrvReplaceChildState;
static void bdrv_replace_child_commit(void *opaque)
{
BdrvReplaceChildState *s = opaque;
GLOBAL_STATE_CODE();
if (s->free_empty_child && !s->child->bs) {
bdrv_child_free(s->child);
}
bdrv_unref(s->old_bs);
}
static void bdrv_replace_child_abort(void *opaque)
{
BdrvReplaceChildState *s = opaque;
BlockDriverState *new_bs = s->child->bs;
GLOBAL_STATE_CODE();
block: Let replace_child_tran keep indirect pointer As of a future commit, bdrv_replace_child_noperm() will clear the indirect BdrvChild pointer passed to it if the new child BDS is NULL. bdrv_replace_child_tran() will want to let it do that, but revert this change in its abort handler. For that, we need to have it receive a BdrvChild ** pointer, too, and keep it stored in the BdrvReplaceChildState object that we attach to the transaction. Note that we do not need to store it in the BdrvReplaceChildState when new_bs is not NULL, because then there is nothing to revert. This is important so that bdrv_replace_node_noperm() can pass a pointer to a loop-local variable to bdrv_replace_child_tran() without worrying that this pointer will outlive one loop iteration. (Of course, for that to work, bdrv_replace_node_noperm() and in turn bdrv_replace_node() and its relatives may not be called with a NULL @to node. Luckily, they already are not, but now we should assert this.) bdrv_remove_file_or_backing_child() on the other hand needs to ensure that the indirect pointer it passes will stay valid for the duration of the transaction. Ensure this by keeping a strong reference to the BDS whose &bs->backing or &bs->file it passes to bdrv_replace_child_tran(), and giving up that reference only in the transaction .clean() handler. Signed-off-by: Hanna Reitz <hreitz@redhat.com> Message-Id: <20211111120829.81329-9-hreitz@redhat.com> Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> Message-Id: <20211115145409.176785-9-kwolf@redhat.com> Signed-off-by: Hanna Reitz <hreitz@redhat.com>
2021-11-15 15:54:04 +01:00
/*
* old_bs reference is transparently moved from @s to s->child.
*
* Pass &s->child here instead of s->childp, because:
* (1) s->old_bs must be non-NULL, so bdrv_replace_child_noperm() will not
* modify the BdrvChild * pointer we indirectly pass to it, i.e. it
* will not modify s->child. From that perspective, it does not matter
* whether we pass s->childp or &s->child.
* (2) If new_bs is not NULL, s->childp will be NULL. We then cannot use
* it here.
* (3) If new_bs is NULL, *s->childp will have been NULLed by
* bdrv_replace_child_tran()'s bdrv_replace_child_noperm() call, and we
* must not pass a NULL *s->childp here.
*
* So whether new_bs was NULL or not, we cannot pass s->childp here; and in
* any case, there is no reason to pass it anyway.
*/
bdrv_replace_child_noperm(&s->child, s->old_bs, true);
/*
* The child was pre-existing, so s->old_bs must be non-NULL, and
* s->child thus must not have been freed
*/
assert(s->child != NULL);
if (!new_bs) {
/* As described above, *s->childp was cleared, so restore it */
assert(s->childp != NULL);
*s->childp = s->child;
}
bdrv_unref(new_bs);
}
static TransactionActionDrv bdrv_replace_child_drv = {
.commit = bdrv_replace_child_commit,
.abort = bdrv_replace_child_abort,
.clean = g_free,
};
/*
* bdrv_replace_child_tran
*
* Note: real unref of old_bs is done only on commit.
*
* The function doesn't update permissions, caller is responsible for this.
block: Let replace_child_tran keep indirect pointer As of a future commit, bdrv_replace_child_noperm() will clear the indirect BdrvChild pointer passed to it if the new child BDS is NULL. bdrv_replace_child_tran() will want to let it do that, but revert this change in its abort handler. For that, we need to have it receive a BdrvChild ** pointer, too, and keep it stored in the BdrvReplaceChildState object that we attach to the transaction. Note that we do not need to store it in the BdrvReplaceChildState when new_bs is not NULL, because then there is nothing to revert. This is important so that bdrv_replace_node_noperm() can pass a pointer to a loop-local variable to bdrv_replace_child_tran() without worrying that this pointer will outlive one loop iteration. (Of course, for that to work, bdrv_replace_node_noperm() and in turn bdrv_replace_node() and its relatives may not be called with a NULL @to node. Luckily, they already are not, but now we should assert this.) bdrv_remove_file_or_backing_child() on the other hand needs to ensure that the indirect pointer it passes will stay valid for the duration of the transaction. Ensure this by keeping a strong reference to the BDS whose &bs->backing or &bs->file it passes to bdrv_replace_child_tran(), and giving up that reference only in the transaction .clean() handler. Signed-off-by: Hanna Reitz <hreitz@redhat.com> Message-Id: <20211111120829.81329-9-hreitz@redhat.com> Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> Message-Id: <20211115145409.176785-9-kwolf@redhat.com> Signed-off-by: Hanna Reitz <hreitz@redhat.com>
2021-11-15 15:54:04 +01:00
*
* (*childp)->bs must not be NULL.
*
block: Let replace_child_tran keep indirect pointer As of a future commit, bdrv_replace_child_noperm() will clear the indirect BdrvChild pointer passed to it if the new child BDS is NULL. bdrv_replace_child_tran() will want to let it do that, but revert this change in its abort handler. For that, we need to have it receive a BdrvChild ** pointer, too, and keep it stored in the BdrvReplaceChildState object that we attach to the transaction. Note that we do not need to store it in the BdrvReplaceChildState when new_bs is not NULL, because then there is nothing to revert. This is important so that bdrv_replace_node_noperm() can pass a pointer to a loop-local variable to bdrv_replace_child_tran() without worrying that this pointer will outlive one loop iteration. (Of course, for that to work, bdrv_replace_node_noperm() and in turn bdrv_replace_node() and its relatives may not be called with a NULL @to node. Luckily, they already are not, but now we should assert this.) bdrv_remove_file_or_backing_child() on the other hand needs to ensure that the indirect pointer it passes will stay valid for the duration of the transaction. Ensure this by keeping a strong reference to the BDS whose &bs->backing or &bs->file it passes to bdrv_replace_child_tran(), and giving up that reference only in the transaction .clean() handler. Signed-off-by: Hanna Reitz <hreitz@redhat.com> Message-Id: <20211111120829.81329-9-hreitz@redhat.com> Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> Message-Id: <20211115145409.176785-9-kwolf@redhat.com> Signed-off-by: Hanna Reitz <hreitz@redhat.com>
2021-11-15 15:54:04 +01:00
* Note that if new_bs == NULL, @childp is stored in a state object attached
* to @tran, so that the old child can be reinstated in the abort handler.
* Therefore, if @new_bs can be NULL, @childp must stay valid until the
* transaction is committed or aborted.
*
* If @free_empty_child is true and @new_bs is NULL, the BdrvChild is
* freed (on commit). @free_empty_child should only be false if the
* caller will free the BDrvChild themselves (which may be important
* if this is in turn called in another transactional context).
*/
block: Let replace_child_tran keep indirect pointer As of a future commit, bdrv_replace_child_noperm() will clear the indirect BdrvChild pointer passed to it if the new child BDS is NULL. bdrv_replace_child_tran() will want to let it do that, but revert this change in its abort handler. For that, we need to have it receive a BdrvChild ** pointer, too, and keep it stored in the BdrvReplaceChildState object that we attach to the transaction. Note that we do not need to store it in the BdrvReplaceChildState when new_bs is not NULL, because then there is nothing to revert. This is important so that bdrv_replace_node_noperm() can pass a pointer to a loop-local variable to bdrv_replace_child_tran() without worrying that this pointer will outlive one loop iteration. (Of course, for that to work, bdrv_replace_node_noperm() and in turn bdrv_replace_node() and its relatives may not be called with a NULL @to node. Luckily, they already are not, but now we should assert this.) bdrv_remove_file_or_backing_child() on the other hand needs to ensure that the indirect pointer it passes will stay valid for the duration of the transaction. Ensure this by keeping a strong reference to the BDS whose &bs->backing or &bs->file it passes to bdrv_replace_child_tran(), and giving up that reference only in the transaction .clean() handler. Signed-off-by: Hanna Reitz <hreitz@redhat.com> Message-Id: <20211111120829.81329-9-hreitz@redhat.com> Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> Message-Id: <20211115145409.176785-9-kwolf@redhat.com> Signed-off-by: Hanna Reitz <hreitz@redhat.com>
2021-11-15 15:54:04 +01:00
static void bdrv_replace_child_tran(BdrvChild **childp,
BlockDriverState *new_bs,
Transaction *tran,
bool free_empty_child)
{
BdrvReplaceChildState *s = g_new(BdrvReplaceChildState, 1);
*s = (BdrvReplaceChildState) {
block: Let replace_child_tran keep indirect pointer As of a future commit, bdrv_replace_child_noperm() will clear the indirect BdrvChild pointer passed to it if the new child BDS is NULL. bdrv_replace_child_tran() will want to let it do that, but revert this change in its abort handler. For that, we need to have it receive a BdrvChild ** pointer, too, and keep it stored in the BdrvReplaceChildState object that we attach to the transaction. Note that we do not need to store it in the BdrvReplaceChildState when new_bs is not NULL, because then there is nothing to revert. This is important so that bdrv_replace_node_noperm() can pass a pointer to a loop-local variable to bdrv_replace_child_tran() without worrying that this pointer will outlive one loop iteration. (Of course, for that to work, bdrv_replace_node_noperm() and in turn bdrv_replace_node() and its relatives may not be called with a NULL @to node. Luckily, they already are not, but now we should assert this.) bdrv_remove_file_or_backing_child() on the other hand needs to ensure that the indirect pointer it passes will stay valid for the duration of the transaction. Ensure this by keeping a strong reference to the BDS whose &bs->backing or &bs->file it passes to bdrv_replace_child_tran(), and giving up that reference only in the transaction .clean() handler. Signed-off-by: Hanna Reitz <hreitz@redhat.com> Message-Id: <20211111120829.81329-9-hreitz@redhat.com> Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> Message-Id: <20211115145409.176785-9-kwolf@redhat.com> Signed-off-by: Hanna Reitz <hreitz@redhat.com>
2021-11-15 15:54:04 +01:00
.child = *childp,
.childp = new_bs == NULL ? childp : NULL,
.old_bs = (*childp)->bs,
.free_empty_child = free_empty_child,
};
tran_add(tran, &bdrv_replace_child_drv, s);
/* The abort handler relies on this */
assert(s->old_bs != NULL);
if (new_bs) {
bdrv_ref(new_bs);
}
/*
* Pass free_empty_child=false, we will free the child (if
* necessary) in bdrv_replace_child_commit() (if our
* @free_empty_child parameter was true).
*/
bdrv_replace_child_noperm(childp, new_bs, false);
block: Let replace_child_tran keep indirect pointer As of a future commit, bdrv_replace_child_noperm() will clear the indirect BdrvChild pointer passed to it if the new child BDS is NULL. bdrv_replace_child_tran() will want to let it do that, but revert this change in its abort handler. For that, we need to have it receive a BdrvChild ** pointer, too, and keep it stored in the BdrvReplaceChildState object that we attach to the transaction. Note that we do not need to store it in the BdrvReplaceChildState when new_bs is not NULL, because then there is nothing to revert. This is important so that bdrv_replace_node_noperm() can pass a pointer to a loop-local variable to bdrv_replace_child_tran() without worrying that this pointer will outlive one loop iteration. (Of course, for that to work, bdrv_replace_node_noperm() and in turn bdrv_replace_node() and its relatives may not be called with a NULL @to node. Luckily, they already are not, but now we should assert this.) bdrv_remove_file_or_backing_child() on the other hand needs to ensure that the indirect pointer it passes will stay valid for the duration of the transaction. Ensure this by keeping a strong reference to the BDS whose &bs->backing or &bs->file it passes to bdrv_replace_child_tran(), and giving up that reference only in the transaction .clean() handler. Signed-off-by: Hanna Reitz <hreitz@redhat.com> Message-Id: <20211111120829.81329-9-hreitz@redhat.com> Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> Message-Id: <20211115145409.176785-9-kwolf@redhat.com> Signed-off-by: Hanna Reitz <hreitz@redhat.com>
2021-11-15 15:54:04 +01:00
/* old_bs reference is transparently moved from *childp to @s */
}
/*
* Refresh permissions in @bs subtree. The function is intended to be called
* after some graph modification that was done without permission update.
*/
static int bdrv_node_refresh_perm(BlockDriverState *bs, BlockReopenQueue *q,
Transaction *tran, Error **errp)
{
BlockDriver *drv = bs->drv;
BdrvChild *c;
int ret;
uint64_t cumulative_perms, cumulative_shared_perms;
GLOBAL_STATE_CODE();
bdrv_get_cumulative_perm(bs, &cumulative_perms, &cumulative_shared_perms);
/* Write permissions never work with read-only images */
if ((cumulative_perms & (BLK_PERM_WRITE | BLK_PERM_WRITE_UNCHANGED)) &&
!bdrv_is_writable_after_reopen(bs, q))
{
if (!bdrv_is_writable_after_reopen(bs, NULL)) {
error_setg(errp, "Block node is read-only");
} else {
error_setg(errp, "Read-only block node '%s' cannot support "
"read-write users", bdrv_get_node_name(bs));
}
return -EPERM;
}
/*
* Unaligned requests will automatically be aligned to bl.request_alignment
* and without RESIZE we can't extend requests to write to space beyond the
* end of the image, so it's required that the image size is aligned.
*/
if ((cumulative_perms & (BLK_PERM_WRITE | BLK_PERM_WRITE_UNCHANGED)) &&
!(cumulative_perms & BLK_PERM_RESIZE))
{
if ((bs->total_sectors * BDRV_SECTOR_SIZE) % bs->bl.request_alignment) {
error_setg(errp, "Cannot get 'write' permission without 'resize': "
"Image size is not a multiple of request "
"alignment");
return -EPERM;
}
}
/* Check this node */
if (!drv) {
return 0;
}
ret = bdrv_drv_set_perm(bs, cumulative_perms, cumulative_shared_perms, tran,
errp);
if (ret < 0) {
return ret;
}
/* Drivers that never have children can omit .bdrv_child_perm() */
if (!drv->bdrv_child_perm) {
assert(QLIST_EMPTY(&bs->children));
return 0;
}
/* Check all children */
QLIST_FOREACH(c, &bs->children, next) {
uint64_t cur_perm, cur_shared;
bdrv_child_perm(bs, c->bs, c, c->role, q,
cumulative_perms, cumulative_shared_perms,
&cur_perm, &cur_shared);
bdrv_child_set_perm(c, cur_perm, cur_shared, tran);
block: use topological sort for permission update Rewrite bdrv_check_perm(), bdrv_abort_perm_update() and bdrv_set_perm() to update nodes in topological sort order instead of simple DFS. With topologically sorted nodes, we update a node only when all its parents already updated. With DFS it's not so. Consider the following example: A -+ | | | v | B | | v | C<-+ A is parent for B and C, B is parent for C. Obviously, to update permissions, we should go in order A B C, so, when we update C, all parent permissions already updated. But with current approach (simple recursion) we can update in sequence A C B C (C is updated twice). On first update of C, we consider old B permissions, so doing wrong thing. If it succeed, all is OK, on second C update we will finish with correct graph. But if the wrong thing failed, we break the whole process for no reason (it's possible that updated B permission will be less strict, but we will never check it). Also new approach gives a way to simultaneously and correctly update several nodes, we just need to run bdrv_topological_dfs() several times to add all nodes and their subtrees into one topologically sorted list (next patch will update bdrv_replace_node() in this manner). Test test_parallel_perm_update() is now passing, so move it out of debugging "if". We also need to support ignore_children in bdrv_parent_perms_conflict() For test 283 order of conflicting parents check is changed. Note also that in bdrv_check_perm() we don't check for parents conflict at root bs, as we may be in the middle of permission update in bdrv_reopen_multiple(). bdrv_reopen_multiple() will be updated soon. Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Reviewed-by: Kevin Wolf <kwolf@redhat.com> Message-Id: <20210428151804.439460-14-vsementsov@virtuozzo.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2021-04-28 17:17:41 +02:00
}
return 0;
}
static int bdrv_list_refresh_perms(GSList *list, BlockReopenQueue *q,
Transaction *tran, Error **errp)
block: use topological sort for permission update Rewrite bdrv_check_perm(), bdrv_abort_perm_update() and bdrv_set_perm() to update nodes in topological sort order instead of simple DFS. With topologically sorted nodes, we update a node only when all its parents already updated. With DFS it's not so. Consider the following example: A -+ | | | v | B | | v | C<-+ A is parent for B and C, B is parent for C. Obviously, to update permissions, we should go in order A B C, so, when we update C, all parent permissions already updated. But with current approach (simple recursion) we can update in sequence A C B C (C is updated twice). On first update of C, we consider old B permissions, so doing wrong thing. If it succeed, all is OK, on second C update we will finish with correct graph. But if the wrong thing failed, we break the whole process for no reason (it's possible that updated B permission will be less strict, but we will never check it). Also new approach gives a way to simultaneously and correctly update several nodes, we just need to run bdrv_topological_dfs() several times to add all nodes and their subtrees into one topologically sorted list (next patch will update bdrv_replace_node() in this manner). Test test_parallel_perm_update() is now passing, so move it out of debugging "if". We also need to support ignore_children in bdrv_parent_perms_conflict() For test 283 order of conflicting parents check is changed. Note also that in bdrv_check_perm() we don't check for parents conflict at root bs, as we may be in the middle of permission update in bdrv_reopen_multiple(). bdrv_reopen_multiple() will be updated soon. Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Reviewed-by: Kevin Wolf <kwolf@redhat.com> Message-Id: <20210428151804.439460-14-vsementsov@virtuozzo.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2021-04-28 17:17:41 +02:00
{
int ret;
BlockDriverState *bs;
GLOBAL_STATE_CODE();
block: use topological sort for permission update Rewrite bdrv_check_perm(), bdrv_abort_perm_update() and bdrv_set_perm() to update nodes in topological sort order instead of simple DFS. With topologically sorted nodes, we update a node only when all its parents already updated. With DFS it's not so. Consider the following example: A -+ | | | v | B | | v | C<-+ A is parent for B and C, B is parent for C. Obviously, to update permissions, we should go in order A B C, so, when we update C, all parent permissions already updated. But with current approach (simple recursion) we can update in sequence A C B C (C is updated twice). On first update of C, we consider old B permissions, so doing wrong thing. If it succeed, all is OK, on second C update we will finish with correct graph. But if the wrong thing failed, we break the whole process for no reason (it's possible that updated B permission will be less strict, but we will never check it). Also new approach gives a way to simultaneously and correctly update several nodes, we just need to run bdrv_topological_dfs() several times to add all nodes and their subtrees into one topologically sorted list (next patch will update bdrv_replace_node() in this manner). Test test_parallel_perm_update() is now passing, so move it out of debugging "if". We also need to support ignore_children in bdrv_parent_perms_conflict() For test 283 order of conflicting parents check is changed. Note also that in bdrv_check_perm() we don't check for parents conflict at root bs, as we may be in the middle of permission update in bdrv_reopen_multiple(). bdrv_reopen_multiple() will be updated soon. Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Reviewed-by: Kevin Wolf <kwolf@redhat.com> Message-Id: <20210428151804.439460-14-vsementsov@virtuozzo.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2021-04-28 17:17:41 +02:00
for ( ; list; list = list->next) {
bs = list->data;
if (bdrv_parent_perms_conflict(bs, errp)) {
return -EINVAL;
block: use topological sort for permission update Rewrite bdrv_check_perm(), bdrv_abort_perm_update() and bdrv_set_perm() to update nodes in topological sort order instead of simple DFS. With topologically sorted nodes, we update a node only when all its parents already updated. With DFS it's not so. Consider the following example: A -+ | | | v | B | | v | C<-+ A is parent for B and C, B is parent for C. Obviously, to update permissions, we should go in order A B C, so, when we update C, all parent permissions already updated. But with current approach (simple recursion) we can update in sequence A C B C (C is updated twice). On first update of C, we consider old B permissions, so doing wrong thing. If it succeed, all is OK, on second C update we will finish with correct graph. But if the wrong thing failed, we break the whole process for no reason (it's possible that updated B permission will be less strict, but we will never check it). Also new approach gives a way to simultaneously and correctly update several nodes, we just need to run bdrv_topological_dfs() several times to add all nodes and their subtrees into one topologically sorted list (next patch will update bdrv_replace_node() in this manner). Test test_parallel_perm_update() is now passing, so move it out of debugging "if". We also need to support ignore_children in bdrv_parent_perms_conflict() For test 283 order of conflicting parents check is changed. Note also that in bdrv_check_perm() we don't check for parents conflict at root bs, as we may be in the middle of permission update in bdrv_reopen_multiple(). bdrv_reopen_multiple() will be updated soon. Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Reviewed-by: Kevin Wolf <kwolf@redhat.com> Message-Id: <20210428151804.439460-14-vsementsov@virtuozzo.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2021-04-28 17:17:41 +02:00
}
ret = bdrv_node_refresh_perm(bs, q, tran, errp);
if (ret < 0) {
return ret;
}
}
return 0;
}
void bdrv_get_cumulative_perm(BlockDriverState *bs, uint64_t *perm,
uint64_t *shared_perm)
{
BdrvChild *c;
uint64_t cumulative_perms = 0;
uint64_t cumulative_shared_perms = BLK_PERM_ALL;
GLOBAL_STATE_CODE();
QLIST_FOREACH(c, &bs->parents, next_parent) {
cumulative_perms |= c->perm;
cumulative_shared_perms &= c->shared_perm;
}
*perm = cumulative_perms;
*shared_perm = cumulative_shared_perms;
}
char *bdrv_perm_names(uint64_t perm)
{
struct perm_name {
uint64_t perm;
const char *name;
} permissions[] = {
{ BLK_PERM_CONSISTENT_READ, "consistent read" },
{ BLK_PERM_WRITE, "write" },
{ BLK_PERM_WRITE_UNCHANGED, "write unchanged" },
{ BLK_PERM_RESIZE, "resize" },
{ 0, NULL }
};
GString *result = g_string_sized_new(30);
struct perm_name *p;
for (p = permissions; p->name; p++) {
if (perm & p->perm) {
if (result->len > 0) {
g_string_append(result, ", ");
}
g_string_append(result, p->name);
}
}
return g_string_free(result, FALSE);
}
static int bdrv_refresh_perms(BlockDriverState *bs, Error **errp)
{
int ret;
Transaction *tran = tran_new();
g_autoptr(GSList) list = bdrv_topological_dfs(NULL, NULL, bs);
GLOBAL_STATE_CODE();
ret = bdrv_list_refresh_perms(list, NULL, tran, errp);
tran_finalize(tran, ret);
return ret;
}
int bdrv_child_try_set_perm(BdrvChild *c, uint64_t perm, uint64_t shared,
Error **errp)
{
Error *local_err = NULL;
Transaction *tran = tran_new();
int ret;
GLOBAL_STATE_CODE();
bdrv_child_set_perm(c, perm, shared, tran);
ret = bdrv_refresh_perms(c->bs, &local_err);
tran_finalize(tran, ret);
if (ret < 0) {
if ((perm & ~c->perm) || (c->shared_perm & ~shared)) {
/* tighten permissions */
error_propagate(errp, local_err);
} else {
/*
* Our caller may intend to only loosen restrictions and
* does not expect this function to fail. Errors are not
* fatal in such a case, so we can just hide them from our
* caller.
*/
error_free(local_err);
ret = 0;
}
}
return ret;
}
int bdrv_child_refresh_perms(BlockDriverState *bs, BdrvChild *c, Error **errp)
{
uint64_t parent_perms, parent_shared;
uint64_t perms, shared;
GLOBAL_STATE_CODE();
bdrv_get_cumulative_perm(bs, &parent_perms, &parent_shared);
bdrv_child_perm(bs, c->bs, c, c->role, NULL,
parent_perms, parent_shared, &perms, &shared);
return bdrv_child_try_set_perm(c, perms, shared, errp);
}
/*
* Default implementation for .bdrv_child_perm() for block filters:
* Forward CONSISTENT_READ, WRITE, WRITE_UNCHANGED, and RESIZE to the
* filtered child.
*/
static void bdrv_filter_default_perms(BlockDriverState *bs, BdrvChild *c,
BdrvChildRole role,
BlockReopenQueue *reopen_queue,
uint64_t perm, uint64_t shared,
uint64_t *nperm, uint64_t *nshared)
{
GLOBAL_STATE_CODE();
*nperm = perm & DEFAULT_PERM_PASSTHROUGH;
*nshared = (shared & DEFAULT_PERM_PASSTHROUGH) | DEFAULT_PERM_UNCHANGED;
}
static void bdrv_default_perms_for_cow(BlockDriverState *bs, BdrvChild *c,
BdrvChildRole role,
BlockReopenQueue *reopen_queue,
uint64_t perm, uint64_t shared,
uint64_t *nperm, uint64_t *nshared)
{
assert(role & BDRV_CHILD_COW);
GLOBAL_STATE_CODE();
/*
* We want consistent read from backing files if the parent needs it.
* No other operations are performed on backing files.
*/
perm &= BLK_PERM_CONSISTENT_READ;
/*
* If the parent can deal with changing data, we're okay with a
* writable and resizable backing file.
* TODO Require !(perm & BLK_PERM_CONSISTENT_READ), too?
*/
if (shared & BLK_PERM_WRITE) {
shared = BLK_PERM_WRITE | BLK_PERM_RESIZE;
} else {
shared = 0;
}
shared |= BLK_PERM_CONSISTENT_READ | BLK_PERM_WRITE_UNCHANGED;
if (bs->open_flags & BDRV_O_INACTIVE) {
shared |= BLK_PERM_WRITE | BLK_PERM_RESIZE;
}
*nperm = perm;
*nshared = shared;
}
static void bdrv_default_perms_for_storage(BlockDriverState *bs, BdrvChild *c,
BdrvChildRole role,
BlockReopenQueue *reopen_queue,
uint64_t perm, uint64_t shared,
uint64_t *nperm, uint64_t *nshared)
{
int flags;
GLOBAL_STATE_CODE();
assert(role & (BDRV_CHILD_METADATA | BDRV_CHILD_DATA));
flags = bdrv_reopen_get_flags(reopen_queue, bs);
/*
* Apart from the modifications below, the same permissions are
* forwarded and left alone as for filters
*/
bdrv_filter_default_perms(bs, c, role, reopen_queue,
perm, shared, &perm, &shared);
if (role & BDRV_CHILD_METADATA) {
/* Format drivers may touch metadata even if the guest doesn't write */
if (bdrv_is_writable_after_reopen(bs, reopen_queue)) {
perm |= BLK_PERM_WRITE | BLK_PERM_RESIZE;
}
/*
* bs->file always needs to be consistent because of the
* metadata. We can never allow other users to resize or write
* to it.
*/
if (!(flags & BDRV_O_NO_IO)) {
perm |= BLK_PERM_CONSISTENT_READ;
}
shared &= ~(BLK_PERM_WRITE | BLK_PERM_RESIZE);
}
if (role & BDRV_CHILD_DATA) {
/*
* Technically, everything in this block is a subset of the
* BDRV_CHILD_METADATA path taken above, and so this could
* be an "else if" branch. However, that is not obvious, and
* this function is not performance critical, therefore we let
* this be an independent "if".
*/
/*
* We cannot allow other users to resize the file because the
* format driver might have some assumptions about the size
* (e.g. because it is stored in metadata, or because the file
* is split into fixed-size data files).
*/
shared &= ~BLK_PERM_RESIZE;
/*
* WRITE_UNCHANGED often cannot be performed as such on the
* data file. For example, the qcow2 driver may still need to
* write copied clusters on copy-on-read.
*/
if (perm & BLK_PERM_WRITE_UNCHANGED) {
perm |= BLK_PERM_WRITE;
}
/*
* If the data file is written to, the format driver may
* expect to be able to resize it by writing beyond the EOF.
*/
if (perm & BLK_PERM_WRITE) {
perm |= BLK_PERM_RESIZE;
}
}
if (bs->open_flags & BDRV_O_INACTIVE) {
shared |= BLK_PERM_WRITE | BLK_PERM_RESIZE;
}
*nperm = perm;
*nshared = shared;
}
void bdrv_default_perms(BlockDriverState *bs, BdrvChild *c,
BdrvChildRole role, BlockReopenQueue *reopen_queue,
uint64_t perm, uint64_t shared,
uint64_t *nperm, uint64_t *nshared)
{
GLOBAL_STATE_CODE();
if (role & BDRV_CHILD_FILTERED) {
assert(!(role & (BDRV_CHILD_DATA | BDRV_CHILD_METADATA |
BDRV_CHILD_COW)));
bdrv_filter_default_perms(bs, c, role, reopen_queue,
perm, shared, nperm, nshared);
} else if (role & BDRV_CHILD_COW) {
assert(!(role & (BDRV_CHILD_DATA | BDRV_CHILD_METADATA)));
bdrv_default_perms_for_cow(bs, c, role, reopen_queue,
perm, shared, nperm, nshared);
} else if (role & (BDRV_CHILD_METADATA | BDRV_CHILD_DATA)) {
bdrv_default_perms_for_storage(bs, c, role, reopen_queue,
perm, shared, nperm, nshared);
} else {
g_assert_not_reached();
}
}
uint64_t bdrv_qapi_perm_to_blk_perm(BlockPermission qapi_perm)
{
static const uint64_t permissions[] = {
[BLOCK_PERMISSION_CONSISTENT_READ] = BLK_PERM_CONSISTENT_READ,
[BLOCK_PERMISSION_WRITE] = BLK_PERM_WRITE,
[BLOCK_PERMISSION_WRITE_UNCHANGED] = BLK_PERM_WRITE_UNCHANGED,
[BLOCK_PERMISSION_RESIZE] = BLK_PERM_RESIZE,
};
QEMU_BUILD_BUG_ON(ARRAY_SIZE(permissions) != BLOCK_PERMISSION__MAX);
QEMU_BUILD_BUG_ON(1UL << ARRAY_SIZE(permissions) != BLK_PERM_ALL + 1);
assert(qapi_perm < BLOCK_PERMISSION__MAX);
return permissions[qapi_perm];
}
/**
* Replace (*childp)->bs by @new_bs.
*
* If @new_bs is NULL, *childp will be set to NULL, too: BDS parents
* generally cannot handle a BdrvChild with .bs == NULL, so clearing
* BdrvChild.bs should generally immediately be followed by the
* BdrvChild pointer being cleared as well.
*
* If @free_empty_child is true and @new_bs is NULL, the BdrvChild is
* freed. @free_empty_child should only be false if the caller will
* free the BdrvChild themselves (this may be important in a
* transactional context, where it may only be freed on commit).
*/
static void bdrv_replace_child_noperm(BdrvChild **childp,
BlockDriverState *new_bs,
bool free_empty_child)
{
BdrvChild *child = *childp;
BlockDriverState *old_bs = child->bs;
int new_bs_quiesce_counter;
int drain_saldo;
assert(!child->frozen);
assert(old_bs != new_bs);
GLOBAL_STATE_CODE();
if (old_bs && new_bs) {
assert(bdrv_get_aio_context(old_bs) == bdrv_get_aio_context(new_bs));
}
new_bs_quiesce_counter = (new_bs ? new_bs->quiesce_counter : 0);
drain_saldo = new_bs_quiesce_counter - child->parent_quiesce_counter;
/*
* If the new child node is drained but the old one was not, flush
* all outstanding requests to the old child node.
*/
while (drain_saldo > 0 && child->klass->drained_begin) {
bdrv_parent_drained_begin_single(child, true);
drain_saldo--;
}
if (old_bs) {
/* Detach first so that the recursive drain sections coming from @child
* are already gone and we only end the drain sections that came from
* elsewhere. */
if (child->klass->detach) {
child->klass->detach(child);
}
assert_bdrv_graph_writable(old_bs);
QLIST_REMOVE(child, next_parent);
}
child->bs = new_bs;
if (!new_bs) {
*childp = NULL;
}
if (new_bs) {
assert_bdrv_graph_writable(new_bs);
QLIST_INSERT_HEAD(&new_bs->parents, child, next_parent);
/*
* Detaching the old node may have led to the new node's
* quiesce_counter having been decreased. Not a problem, we
* just need to recognize this here and then invoke
* drained_end appropriately more often.
*/
assert(new_bs->quiesce_counter <= new_bs_quiesce_counter);
drain_saldo += new_bs->quiesce_counter - new_bs_quiesce_counter;
/* Attach only after starting new drained sections, so that recursive
* drain sections coming from @child don't get an extra .drained_begin
* callback. */
if (child->klass->attach) {
child->klass->attach(child);
}
}
/*
* If the old child node was drained but the new one is not, allow
* requests to come in only after the new node has been attached.
*/
while (drain_saldo < 0 && child->klass->drained_end) {
bdrv_parent_drained_end_single(child);
drain_saldo++;
}
if (free_empty_child && !child->bs) {
bdrv_child_free(child);
}
}
/**
* Free the given @child.
*
* The child must be empty (i.e. `child->bs == NULL`) and it must be
* unused (i.e. not in a children list).
*/
static void bdrv_child_free(BdrvChild *child)
{
assert(!child->bs);
GLOBAL_STATE_CODE();
assert(!child->next.le_prev); /* not in children list */
g_free(child->name);
g_free(child);
}
typedef struct BdrvAttachChildCommonState {
BdrvChild **child;
AioContext *old_parent_ctx;
AioContext *old_child_ctx;
} BdrvAttachChildCommonState;
static void bdrv_attach_child_common_abort(void *opaque)
{
BdrvAttachChildCommonState *s = opaque;
BdrvChild *child = *s->child;
BlockDriverState *bs = child->bs;
GLOBAL_STATE_CODE();
/*
* Pass free_empty_child=false, because we still need the child
* for the AioContext operations on the parent below; those
* BdrvChildClass methods all work on a BdrvChild object, so we
* need to keep it as an empty shell (after this function, it will
* not be attached to any parent, and it will not have a .bs).
*/
bdrv_replace_child_noperm(s->child, NULL, false);
if (bdrv_get_aio_context(bs) != s->old_child_ctx) {
bdrv_try_set_aio_context(bs, s->old_child_ctx, &error_abort);
}
if (bdrv_child_get_parent_aio_context(child) != s->old_parent_ctx) {
GSList *ignore;
/* No need to ignore `child`, because it has been detached already */
ignore = NULL;
child->klass->can_set_aio_ctx(child, s->old_parent_ctx, &ignore,
&error_abort);
g_slist_free(ignore);
ignore = NULL;
child->klass->set_aio_ctx(child, s->old_parent_ctx, &ignore);
g_slist_free(ignore);
}
bdrv_unref(bs);
bdrv_child_free(child);
}
static TransactionActionDrv bdrv_attach_child_common_drv = {
.abort = bdrv_attach_child_common_abort,
.clean = g_free,
};
/*
* Common part of attaching bdrv child to bs or to blk or to job
*
* Resulting new child is returned through @child.
* At start *@child must be NULL.
* @child is saved to a new entry of @tran, so that *@child could be reverted to
* NULL on abort(). So referenced variable must live at least until transaction
* end.
*
* Function doesn't update permissions, caller is responsible for this.
*/
static int bdrv_attach_child_common(BlockDriverState *child_bs,
const char *child_name,
const BdrvChildClass *child_class,
BdrvChildRole child_role,
uint64_t perm, uint64_t shared_perm,
void *opaque, BdrvChild **child,
Transaction *tran, Error **errp)
{
BdrvChild *new_child;
AioContext *parent_ctx;
AioContext *child_ctx = bdrv_get_aio_context(child_bs);
assert(child);
assert(*child == NULL);
assert(child_class->get_parent_desc);
GLOBAL_STATE_CODE();
new_child = g_new(BdrvChild, 1);
*new_child = (BdrvChild) {
.bs = NULL,
.name = g_strdup(child_name),
.klass = child_class,
.role = child_role,
.perm = perm,
.shared_perm = shared_perm,
.opaque = opaque,
};
/*
* If the AioContexts don't match, first try to move the subtree of
* child_bs into the AioContext of the new parent. If this doesn't work,
* try moving the parent into the AioContext of child_bs instead.
*/
parent_ctx = bdrv_child_get_parent_aio_context(new_child);
if (child_ctx != parent_ctx) {
Error *local_err = NULL;
int ret = bdrv_try_set_aio_context(child_bs, parent_ctx, &local_err);
if (ret < 0 && child_class->can_set_aio_ctx) {
GSList *ignore = g_slist_prepend(NULL, new_child);
if (child_class->can_set_aio_ctx(new_child, child_ctx, &ignore,
NULL))
{
error_free(local_err);
ret = 0;
g_slist_free(ignore);
ignore = g_slist_prepend(NULL, new_child);
child_class->set_aio_ctx(new_child, child_ctx, &ignore);
}
g_slist_free(ignore);
}
if (ret < 0) {
error_propagate(errp, local_err);
bdrv_child_free(new_child);
return ret;
}
}
bdrv_ref(child_bs);
bdrv_replace_child_noperm(&new_child, child_bs, true);
/* child_bs was non-NULL, so new_child must not have been freed */
assert(new_child != NULL);
*child = new_child;
BdrvAttachChildCommonState *s = g_new(BdrvAttachChildCommonState, 1);
*s = (BdrvAttachChildCommonState) {
.child = child,
.old_parent_ctx = parent_ctx,
.old_child_ctx = child_ctx,
};
tran_add(tran, &bdrv_attach_child_common_drv, s);
return 0;
}
/*
* Variable referenced by @child must live at least until transaction end.
* (see bdrv_attach_child_common() doc for details)
*
* Function doesn't update permissions, caller is responsible for this.
*/
static int bdrv_attach_child_noperm(BlockDriverState *parent_bs,
BlockDriverState *child_bs,
const char *child_name,
const BdrvChildClass *child_class,
BdrvChildRole child_role,
BdrvChild **child,
Transaction *tran,
Error **errp)
{
int ret;
uint64_t perm, shared_perm;
assert(parent_bs->drv);
GLOBAL_STATE_CODE();
if (bdrv_recurse_has_child(child_bs, parent_bs)) {
error_setg(errp, "Making '%s' a %s child of '%s' would create a cycle",
child_bs->node_name, child_name, parent_bs->node_name);
return -EINVAL;
}
bdrv_get_cumulative_perm(parent_bs, &perm, &shared_perm);
bdrv_child_perm(parent_bs, child_bs, NULL, child_role, NULL,
perm, shared_perm, &perm, &shared_perm);
ret = bdrv_attach_child_common(child_bs, child_name, child_class,
child_role, perm, shared_perm, parent_bs,
child, tran, errp);
if (ret < 0) {
return ret;
}
return 0;
}
static void bdrv_detach_child(BdrvChild **childp)
{
BlockDriverState *old_bs = (*childp)->bs;
GLOBAL_STATE_CODE();
bdrv_replace_child_noperm(childp, NULL, true);
if (old_bs) {
/*
* Update permissions for old node. We're just taking a parent away, so
* we're loosening restrictions. Errors of permission update are not
* fatal in this case, ignore them.
*/
bdrv_refresh_perms(old_bs, NULL);
/*
* When the parent requiring a non-default AioContext is removed, the
* node moves back to the main AioContext
*/
bdrv_try_set_aio_context(old_bs, qemu_get_aio_context(), NULL);
}
}
/*
* This function steals the reference to child_bs from the caller.
* That reference is later dropped by bdrv_root_unref_child().
*
* On failure NULL is returned, errp is set and the reference to
* child_bs is also dropped.
*
* The caller must hold the AioContext lock @child_bs, but not that of @ctx
* (unless @child_bs is already in @ctx).
*/
BdrvChild *bdrv_root_attach_child(BlockDriverState *child_bs,
const char *child_name,
const BdrvChildClass *child_class,
BdrvChildRole child_role,
uint64_t perm, uint64_t shared_perm,
void *opaque, Error **errp)
{
int ret;
BdrvChild *child = NULL;
Transaction *tran = tran_new();
GLOBAL_STATE_CODE();
ret = bdrv_attach_child_common(child_bs, child_name, child_class,
child_role, perm, shared_perm, opaque,
&child, tran, errp);
if (ret < 0) {
goto out;
}
ret = bdrv_refresh_perms(child_bs, errp);
out:
tran_finalize(tran, ret);
/* child is unset on failure by bdrv_attach_child_common_abort() */
assert((ret < 0) == !child);
bdrv_unref(child_bs);
return child;
}
/*
* This function transfers the reference to child_bs from the caller
* to parent_bs. That reference is later dropped by parent_bs on
* bdrv_close() or if someone calls bdrv_unref_child().
*
* On failure NULL is returned, errp is set and the reference to
* child_bs is also dropped.
*
* If @parent_bs and @child_bs are in different AioContexts, the caller must
* hold the AioContext lock for @child_bs, but not for @parent_bs.
*/
BdrvChild *bdrv_attach_child(BlockDriverState *parent_bs,
BlockDriverState *child_bs,
const char *child_name,
const BdrvChildClass *child_class,
BdrvChildRole child_role,
Error **errp)
{
int ret;
BdrvChild *child = NULL;
Transaction *tran = tran_new();
GLOBAL_STATE_CODE();
ret = bdrv_attach_child_noperm(parent_bs, child_bs, child_name, child_class,
child_role, &child, tran, errp);
if (ret < 0) {
goto out;
}
ret = bdrv_refresh_perms(parent_bs, errp);
if (ret < 0) {
goto out;
}
out:
tran_finalize(tran, ret);
/* child is unset on failure by bdrv_attach_child_common_abort() */
assert((ret < 0) == !child);
bdrv_unref(child_bs);
return child;
}
/* Callers must ensure that child->frozen is false. */
void bdrv_root_unref_child(BdrvChild *child)
{
BlockDriverState *child_bs;
GLOBAL_STATE_CODE();
child_bs = child->bs;
bdrv_detach_child(&child);
bdrv_unref(child_bs);
}
typedef struct BdrvSetInheritsFrom {
BlockDriverState *bs;
BlockDriverState *old_inherits_from;
} BdrvSetInheritsFrom;
static void bdrv_set_inherits_from_abort(void *opaque)
{
BdrvSetInheritsFrom *s = opaque;
s->bs->inherits_from = s->old_inherits_from;
}
static TransactionActionDrv bdrv_set_inherits_from_drv = {
.abort = bdrv_set_inherits_from_abort,
.clean = g_free,
};
/* @tran is allowed to be NULL. In this case no rollback is possible */
static void bdrv_set_inherits_from(BlockDriverState *bs,
BlockDriverState *new_inherits_from,
Transaction *tran)
{
if (tran) {
BdrvSetInheritsFrom *s = g_new(BdrvSetInheritsFrom, 1);
*s = (BdrvSetInheritsFrom) {
.bs = bs,
.old_inherits_from = bs->inherits_from,
};
tran_add(tran, &bdrv_set_inherits_from_drv, s);
}
bs->inherits_from = new_inherits_from;
}
/**
* Clear all inherits_from pointers from children and grandchildren of
* @root that point to @root, where necessary.
* @tran is allowed to be NULL. In this case no rollback is possible
*/
static void bdrv_unset_inherits_from(BlockDriverState *root, BdrvChild *child,
Transaction *tran)
{
BdrvChild *c;
if (child->bs->inherits_from == root) {
/*
* Remove inherits_from only when the last reference between root and
* child->bs goes away.
*/
QLIST_FOREACH(c, &root->children, next) {
if (c != child && c->bs == child->bs) {
break;
}
}
if (c == NULL) {
bdrv_set_inherits_from(child->bs, NULL, tran);
}
}
QLIST_FOREACH(c, &child->bs->children, next) {
bdrv_unset_inherits_from(root, c, tran);
}
}
/* Callers must ensure that child->frozen is false. */
void bdrv_unref_child(BlockDriverState *parent, BdrvChild *child)
{
GLOBAL_STATE_CODE();
if (child == NULL) {
return;
}
bdrv_unset_inherits_from(parent, child, NULL);
bdrv_root_unref_child(child);
}
static void bdrv_parent_cb_change_media(BlockDriverState *bs, bool load)
{
BdrvChild *c;
GLOBAL_STATE_CODE();
QLIST_FOREACH(c, &bs->parents, next_parent) {
if (c->klass->change_media) {
c->klass->change_media(c, load);
}
}
}
block: Update BlockDriverState.inherits_from on bdrv_set_backing_hd() When a BlockDriverState's child is opened (be it a backing file, the protocol layer, or any other) inherits_from is set to point to the parent node. Children opened separately and then attached to a parent don't have this pointer set. bdrv_reopen_queue_child() uses this to determine whether a node's children must also be reopened inheriting the options from the parent or not. If inherits_from points to the parent then the child is reopened and its options can be changed, like in this example: $ qemu-img create -f qcow2 hd0.qcow2 1M $ qemu-img create -f qcow2 hd1.qcow2 1M $ $QEMU -drive if=none,node-name=hd0,file=hd0.qcow2,\ backing.driver=qcow2,backing.file.filename=hd1.qcow2 (qemu) qemu-io hd0 "reopen -o backing.l2-cache-size=2M" If the child does not inherit from the parent then it does not get reopened and its options cannot be changed: $ $QEMU -drive if=none,node-name=hd1,file=hd1.qcow2 -drive if=none,node-name=hd0,file=hd0.qcow2,backing=hd1 (qemu) qemu-io hd0 "reopen -o backing.l2-cache-size=2M" Cannot change the option 'backing.l2-cache-size' If a disk image has a chain of backing files then all of them are also connected through their inherits_from pointers (i.e. it's possible to walk the chain in reverse order from base to top). However this is broken if the intermediate nodes are removed using e.g. block-stream because the inherits_from pointer from the base node becomes NULL: $ qemu-img create -f qcow2 hd0.qcow2 1M $ qemu-img create -f qcow2 -b hd0.qcow2 hd1.qcow2 $ qemu-img create -f qcow2 -b hd1.qcow2 hd2.qcow2 $ $QEMU -drive if=none,file=hd2.qcow2 (qemu) qemu-io none0 "reopen -o backing.l2-cache-size=2M" (qemu) block_stream none0 0 hd0.qcow2 (qemu) qemu-io none0 "reopen -o backing.l2-cache-size=2M" Cannot change the option 'backing.l2-cache-size' This patch updates the inherits_from pointer if the intermediate nodes of a backing chain are removed using bdrv_set_backing_hd(), and adds a test case for this scenario. Signed-off-by: Alberto Garcia <berto@igalia.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2018-10-31 17:16:37 +01:00
/* Return true if you can reach parent going through child->inherits_from
* recursively. If parent or child are NULL, return false */
static bool bdrv_inherits_from_recursive(BlockDriverState *child,
BlockDriverState *parent)
{
while (child && child != parent) {
child = child->inherits_from;
}
return child != NULL;
}
/*
* Return the BdrvChildRole for @bs's backing child. bs->backing is
* mostly used for COW backing children (role = COW), but also for
* filtered children (role = FILTERED | PRIMARY).
*/
static BdrvChildRole bdrv_backing_role(BlockDriverState *bs)
{
if (bs->drv && bs->drv->is_filter) {
return BDRV_CHILD_FILTERED | BDRV_CHILD_PRIMARY;
} else {
return BDRV_CHILD_COW;
}
}
/*
* Sets the bs->backing or bs->file link of a BDS. A new reference is created;
* callers which don't need their own reference any more must call bdrv_unref().
*
* Function doesn't update permissions, caller is responsible for this.
*/
static int bdrv_set_file_or_backing_noperm(BlockDriverState *parent_bs,
BlockDriverState *child_bs,
bool is_backing,
Transaction *tran, Error **errp)
{
int ret = 0;
bool update_inherits_from =
bdrv_inherits_from_recursive(child_bs, parent_bs);
BdrvChild *child = is_backing ? parent_bs->backing : parent_bs->file;
BdrvChildRole role;
block: Update BlockDriverState.inherits_from on bdrv_set_backing_hd() When a BlockDriverState's child is opened (be it a backing file, the protocol layer, or any other) inherits_from is set to point to the parent node. Children opened separately and then attached to a parent don't have this pointer set. bdrv_reopen_queue_child() uses this to determine whether a node's children must also be reopened inheriting the options from the parent or not. If inherits_from points to the parent then the child is reopened and its options can be changed, like in this example: $ qemu-img create -f qcow2 hd0.qcow2 1M $ qemu-img create -f qcow2 hd1.qcow2 1M $ $QEMU -drive if=none,node-name=hd0,file=hd0.qcow2,\ backing.driver=qcow2,backing.file.filename=hd1.qcow2 (qemu) qemu-io hd0 "reopen -o backing.l2-cache-size=2M" If the child does not inherit from the parent then it does not get reopened and its options cannot be changed: $ $QEMU -drive if=none,node-name=hd1,file=hd1.qcow2 -drive if=none,node-name=hd0,file=hd0.qcow2,backing=hd1 (qemu) qemu-io hd0 "reopen -o backing.l2-cache-size=2M" Cannot change the option 'backing.l2-cache-size' If a disk image has a chain of backing files then all of them are also connected through their inherits_from pointers (i.e. it's possible to walk the chain in reverse order from base to top). However this is broken if the intermediate nodes are removed using e.g. block-stream because the inherits_from pointer from the base node becomes NULL: $ qemu-img create -f qcow2 hd0.qcow2 1M $ qemu-img create -f qcow2 -b hd0.qcow2 hd1.qcow2 $ qemu-img create -f qcow2 -b hd1.qcow2 hd2.qcow2 $ $QEMU -drive if=none,file=hd2.qcow2 (qemu) qemu-io none0 "reopen -o backing.l2-cache-size=2M" (qemu) block_stream none0 0 hd0.qcow2 (qemu) qemu-io none0 "reopen -o backing.l2-cache-size=2M" Cannot change the option 'backing.l2-cache-size' This patch updates the inherits_from pointer if the intermediate nodes of a backing chain are removed using bdrv_set_backing_hd(), and adds a test case for this scenario. Signed-off-by: Alberto Garcia <berto@igalia.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2018-10-31 17:16:37 +01:00
GLOBAL_STATE_CODE();
if (!parent_bs->drv) {
/*
* Node without drv is an object without a class :/. TODO: finally fix
* qcow2 driver to never clear bs->drv and implement format corruption
* handling in other way.
*/
error_setg(errp, "Node corrupted");
return -EINVAL;
}
if (child && child->frozen) {
error_setg(errp, "Cannot change frozen '%s' link from '%s' to '%s'",
child->name, parent_bs->node_name, child->bs->node_name);
return -EPERM;
}
if (is_backing && !parent_bs->drv->is_filter &&
!parent_bs->drv->supports_backing)
{
error_setg(errp, "Driver '%s' of node '%s' does not support backing "
"files", parent_bs->drv->format_name, parent_bs->node_name);
return -EINVAL;
}
if (parent_bs->drv->is_filter) {
role = BDRV_CHILD_FILTERED | BDRV_CHILD_PRIMARY;
} else if (is_backing) {
role = BDRV_CHILD_COW;
} else {
/*
* We only can use same role as it is in existing child. We don't have
* infrastructure to determine role of file child in generic way
*/
if (!child) {
error_setg(errp, "Cannot set file child to format node without "
"file child");
return -EINVAL;
}
role = child->role;
}
if (child) {
bdrv_unset_inherits_from(parent_bs, child, tran);
bdrv_remove_file_or_backing_child(parent_bs, child, tran);
}
if (!child_bs) {
goto out;
}
ret = bdrv_attach_child_noperm(parent_bs, child_bs,
is_backing ? "backing" : "file",
&child_of_bds, role,
is_backing ? &parent_bs->backing :
&parent_bs->file,
tran, errp);
if (ret < 0) {
return ret;
}
/*
* If inherits_from pointed recursively to bs then let's update it to
* point directly to bs (else it will become NULL).
*/
if (update_inherits_from) {
bdrv_set_inherits_from(child_bs, parent_bs, tran);
block: Update BlockDriverState.inherits_from on bdrv_set_backing_hd() When a BlockDriverState's child is opened (be it a backing file, the protocol layer, or any other) inherits_from is set to point to the parent node. Children opened separately and then attached to a parent don't have this pointer set. bdrv_reopen_queue_child() uses this to determine whether a node's children must also be reopened inheriting the options from the parent or not. If inherits_from points to the parent then the child is reopened and its options can be changed, like in this example: $ qemu-img create -f qcow2 hd0.qcow2 1M $ qemu-img create -f qcow2 hd1.qcow2 1M $ $QEMU -drive if=none,node-name=hd0,file=hd0.qcow2,\ backing.driver=qcow2,backing.file.filename=hd1.qcow2 (qemu) qemu-io hd0 "reopen -o backing.l2-cache-size=2M" If the child does not inherit from the parent then it does not get reopened and its options cannot be changed: $ $QEMU -drive if=none,node-name=hd1,file=hd1.qcow2 -drive if=none,node-name=hd0,file=hd0.qcow2,backing=hd1 (qemu) qemu-io hd0 "reopen -o backing.l2-cache-size=2M" Cannot change the option 'backing.l2-cache-size' If a disk image has a chain of backing files then all of them are also connected through their inherits_from pointers (i.e. it's possible to walk the chain in reverse order from base to top). However this is broken if the intermediate nodes are removed using e.g. block-stream because the inherits_from pointer from the base node becomes NULL: $ qemu-img create -f qcow2 hd0.qcow2 1M $ qemu-img create -f qcow2 -b hd0.qcow2 hd1.qcow2 $ qemu-img create -f qcow2 -b hd1.qcow2 hd2.qcow2 $ $QEMU -drive if=none,file=hd2.qcow2 (qemu) qemu-io none0 "reopen -o backing.l2-cache-size=2M" (qemu) block_stream none0 0 hd0.qcow2 (qemu) qemu-io none0 "reopen -o backing.l2-cache-size=2M" Cannot change the option 'backing.l2-cache-size' This patch updates the inherits_from pointer if the intermediate nodes of a backing chain are removed using bdrv_set_backing_hd(), and adds a test case for this scenario. Signed-off-by: Alberto Garcia <berto@igalia.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2018-10-31 17:16:37 +01:00
}
out:
bdrv_refresh_limits(parent_bs, tran, NULL);
return 0;
}
static int bdrv_set_backing_noperm(BlockDriverState *bs,
BlockDriverState *backing_hd,
Transaction *tran, Error **errp)
{
GLOBAL_STATE_CODE();
return bdrv_set_file_or_backing_noperm(bs, backing_hd, true, tran, errp);
}
int bdrv_set_backing_hd(BlockDriverState *bs, BlockDriverState *backing_hd,
Error **errp)
{
int ret;
Transaction *tran = tran_new();
GLOBAL_STATE_CODE();
bdrv_drained_begin(bs);
ret = bdrv_set_backing_noperm(bs, backing_hd, tran, errp);
if (ret < 0) {
goto out;
}
ret = bdrv_refresh_perms(bs, errp);
out:
tran_finalize(tran, ret);
bdrv_drained_end(bs);
return ret;
}
/*
* Opens the backing file for a BlockDriverState if not yet open
*
* bdref_key specifies the key for the image's BlockdevRef in the options QDict.
* That QDict has to be flattened; therefore, if the BlockdevRef is a QDict
* itself, all options starting with "${bdref_key}." are considered part of the
* BlockdevRef.
*
* TODO Can this be unified with bdrv_open_image()?
*/
int bdrv_open_backing_file(BlockDriverState *bs, QDict *parent_options,
const char *bdref_key, Error **errp)
{
char *backing_filename = NULL;
char *bdref_key_dot;
const char *reference = NULL;
int ret = 0;
block: Add BDS.auto_backing_file If the backing file is overridden, this most probably does change the guest-visible data of a BDS. Therefore, we will need to consider this in bdrv_refresh_filename(). To see whether it has been overridden, we might want to compare bs->backing_file and bs->backing->bs->filename. However, bs->backing_file is changed by bdrv_set_backing_hd() (which is just used to change the backing child at runtime, without modifying the image header), so bs->backing_file most of the time simply contains a copy of bs->backing->bs->filename anyway, so it is useless for such a comparison. This patch adds an auto_backing_file BDS field which contains the backing file path as indicated by the image header, which is not changed by bdrv_set_backing_hd(). Because of bdrv_refresh_filename() magic, however, a BDS's filename may differ from what has been specified during bdrv_open(). Then, the comparison between bs->auto_backing_file and bs->backing->bs->filename may fail even though bs->backing was opened from bs->auto_backing_file. To mitigate this, we can copy the real BDS's filename (after the whole bdrv_open() and bdrv_refresh_filename() process) into bs->auto_backing_file, if we know the former has been opened based on the latter. This is only possible if no options modifying the backing file's behavior have been specified, though. To simplify things, this patch only copies the filename from the backing file if no options have been specified for it at all. Furthermore, there are cases where an overlay is created by qemu which already contains a BDS's filename (e.g. in blockdev-snapshot-sync). We do not need to worry about updating the overlay's bs->auto_backing_file there, because we actually wrote a post-bdrv_refresh_filename() filename into the image header. So all in all, there will be false negatives where (as of a future patch) bdrv_refresh_filename() will assume that the backing file differs from what was specified in the image header, even though it really does not. However, these cases should be limited to where (1) the user actually did override something in the backing chain (e.g. by specifying options for the backing file), or (2) the user executed a QMP command to change some node's backing file (e.g. change-backing-file or block-commit with @backing-file given) where the given filename does not happen to coincide with qemu's idea of the backing BDS's filename. Then again, (1) really is limited to -drive. With -blockdev or blockdev-add, you have to adhere to the schema, so a user cannot give partial "unimportant" options (e.g. by just setting backing.node-name and leaving the rest to the image header). Therefore, trying to fix this would mean trying to fix something for -drive only. To improve on (2), we would need a full infrastructure to "canonicalize" an arbitrary filename (+ options), so it can be compared against another. That seems a bit over the top, considering that filenames nowadays are there mostly for the user's entertainment. Signed-off-by: Max Reitz <mreitz@redhat.com> Reviewed-by: Eric Blake <eblake@redhat.com> Reviewed-by: Alberto Garcia <berto@igalia.com> Message-id: 20190201192935.18394-5-mreitz@redhat.com Signed-off-by: Max Reitz <mreitz@redhat.com>
2019-02-01 20:29:08 +01:00
bool implicit_backing = false;
BlockDriverState *backing_hd;
QDict *options;
QDict *tmp_parent_options = NULL;
Error *local_err = NULL;
GLOBAL_STATE_CODE();
if (bs->backing != NULL) {
goto free_exit;
}
/* NULL means an empty set of options */
if (parent_options == NULL) {
tmp_parent_options = qdict_new();
parent_options = tmp_parent_options;
}
bs->open_flags &= ~BDRV_O_NO_BACKING;
bdref_key_dot = g_strdup_printf("%s.", bdref_key);
qdict_extract_subqdict(parent_options, &options, bdref_key_dot);
g_free(bdref_key_dot);
block: Document -drive problematic code and bugs -blockdev and blockdev_add convert their arguments via QObject to BlockdevOptions for qmp_blockdev_add(), which converts them back to QObject, then to a flattened QDict. The QDict's members are typed according to the QAPI schema. -drive converts its argument via QemuOpts to a (flat) QDict. This QDict's members are all QString. Thus, the QType of a flat QDict member depends on whether it comes from -drive or -blockdev/blockdev_add, except when the QAPI type maps to QString, which is the case for 'str' and enumeration types. The block layer core extracts generic configuration from the flat QDict, and the block driver extracts driver-specific configuration. Both commonly do so by converting (parts of) the flat QDict to QemuOpts, which turns all values into strings. Not exactly elegant, but correct. However, A few places access the flat QDict directly: * Most of them access members that are always QString. Correct. * bdrv_open_inherit() accesses a boolean, carefully. Correct. * nfs_config() uses a QObject input visitor. Correct only because the visited type contains nothing but QStrings. * nbd_config() and ssh_config() use a QObject input visitor, and the visited types contain non-QStrings: InetSocketAddress members @numeric, @to, @ipv4, @ipv6. -drive works as long as you don't try to use them (they're all optional). @to is ignored anyway. Reproducer: -drive driver=ssh,server.host=h,server.port=22,server.ipv4,path=p -drive driver=nbd,server.type=inet,server.data.host=h,server.data.port=22,server.data.ipv4 both fail with "Invalid parameter type for 'data.ipv4', expected: boolean" Add suitable comments to all these places. Mark the buggy ones FIXME. "Fortunately", -drive's driver-specific options are entirely undocumented. Signed-off-by: Markus Armbruster <armbru@redhat.com> Message-id: 1490895797-29094-5-git-send-email-armbru@redhat.com [mreitz: Fixed two typos] Reviewed-by: Eric Blake <eblake@redhat.com> Signed-off-by: Max Reitz <mreitz@redhat.com>
2017-03-30 19:43:12 +02:00
/*
* Caution: while qdict_get_try_str() is fine, getting non-string
* types would require more care. When @parent_options come from
* -blockdev or blockdev_add, its members are typed according to
* the QAPI schema, but when they come from -drive, they're all
* QString.
*/
reference = qdict_get_try_str(parent_options, bdref_key);
if (reference || qdict_haskey(options, "file.filename")) {
/* keep backing_filename NULL */
} else if (bs->backing_file[0] == '\0' && qdict_size(options) == 0) {
qobject_unref(options);
goto free_exit;
} else {
block: Add BDS.auto_backing_file If the backing file is overridden, this most probably does change the guest-visible data of a BDS. Therefore, we will need to consider this in bdrv_refresh_filename(). To see whether it has been overridden, we might want to compare bs->backing_file and bs->backing->bs->filename. However, bs->backing_file is changed by bdrv_set_backing_hd() (which is just used to change the backing child at runtime, without modifying the image header), so bs->backing_file most of the time simply contains a copy of bs->backing->bs->filename anyway, so it is useless for such a comparison. This patch adds an auto_backing_file BDS field which contains the backing file path as indicated by the image header, which is not changed by bdrv_set_backing_hd(). Because of bdrv_refresh_filename() magic, however, a BDS's filename may differ from what has been specified during bdrv_open(). Then, the comparison between bs->auto_backing_file and bs->backing->bs->filename may fail even though bs->backing was opened from bs->auto_backing_file. To mitigate this, we can copy the real BDS's filename (after the whole bdrv_open() and bdrv_refresh_filename() process) into bs->auto_backing_file, if we know the former has been opened based on the latter. This is only possible if no options modifying the backing file's behavior have been specified, though. To simplify things, this patch only copies the filename from the backing file if no options have been specified for it at all. Furthermore, there are cases where an overlay is created by qemu which already contains a BDS's filename (e.g. in blockdev-snapshot-sync). We do not need to worry about updating the overlay's bs->auto_backing_file there, because we actually wrote a post-bdrv_refresh_filename() filename into the image header. So all in all, there will be false negatives where (as of a future patch) bdrv_refresh_filename() will assume that the backing file differs from what was specified in the image header, even though it really does not. However, these cases should be limited to where (1) the user actually did override something in the backing chain (e.g. by specifying options for the backing file), or (2) the user executed a QMP command to change some node's backing file (e.g. change-backing-file or block-commit with @backing-file given) where the given filename does not happen to coincide with qemu's idea of the backing BDS's filename. Then again, (1) really is limited to -drive. With -blockdev or blockdev-add, you have to adhere to the schema, so a user cannot give partial "unimportant" options (e.g. by just setting backing.node-name and leaving the rest to the image header). Therefore, trying to fix this would mean trying to fix something for -drive only. To improve on (2), we would need a full infrastructure to "canonicalize" an arbitrary filename (+ options), so it can be compared against another. That seems a bit over the top, considering that filenames nowadays are there mostly for the user's entertainment. Signed-off-by: Max Reitz <mreitz@redhat.com> Reviewed-by: Eric Blake <eblake@redhat.com> Reviewed-by: Alberto Garcia <berto@igalia.com> Message-id: 20190201192935.18394-5-mreitz@redhat.com Signed-off-by: Max Reitz <mreitz@redhat.com>
2019-02-01 20:29:08 +01:00
if (qdict_size(options) == 0) {
/* If the user specifies options that do not modify the
* backing file's behavior, we might still consider it the
* implicit backing file. But it's easier this way, and
* just specifying some of the backing BDS's options is
* only possible with -drive anyway (otherwise the QAPI
* schema forces the user to specify everything). */
implicit_backing = !strcmp(bs->auto_backing_file, bs->backing_file);
}
backing_filename = bdrv_get_full_backing_filename(bs, &local_err);
if (local_err) {
ret = -EINVAL;
error_propagate(errp, local_err);
qobject_unref(options);
goto free_exit;
}
}
if (!bs->drv || !bs->drv->supports_backing) {
ret = -EINVAL;
error_setg(errp, "Driver doesn't support backing files");
qobject_unref(options);
goto free_exit;
}
if (!reference &&
bs->backing_format[0] != '\0' && !qdict_haskey(options, "driver")) {
qdict_put_str(options, "driver", bs->backing_format);
}
backing_hd = bdrv_open_inherit(backing_filename, reference, options, 0, bs,
&child_of_bds, bdrv_backing_role(bs), errp);
if (!backing_hd) {
bs->open_flags |= BDRV_O_NO_BACKING;
error_prepend(errp, "Could not open backing file: ");
ret = -EINVAL;
goto free_exit;
}
block: Add BDS.auto_backing_file If the backing file is overridden, this most probably does change the guest-visible data of a BDS. Therefore, we will need to consider this in bdrv_refresh_filename(). To see whether it has been overridden, we might want to compare bs->backing_file and bs->backing->bs->filename. However, bs->backing_file is changed by bdrv_set_backing_hd() (which is just used to change the backing child at runtime, without modifying the image header), so bs->backing_file most of the time simply contains a copy of bs->backing->bs->filename anyway, so it is useless for such a comparison. This patch adds an auto_backing_file BDS field which contains the backing file path as indicated by the image header, which is not changed by bdrv_set_backing_hd(). Because of bdrv_refresh_filename() magic, however, a BDS's filename may differ from what has been specified during bdrv_open(). Then, the comparison between bs->auto_backing_file and bs->backing->bs->filename may fail even though bs->backing was opened from bs->auto_backing_file. To mitigate this, we can copy the real BDS's filename (after the whole bdrv_open() and bdrv_refresh_filename() process) into bs->auto_backing_file, if we know the former has been opened based on the latter. This is only possible if no options modifying the backing file's behavior have been specified, though. To simplify things, this patch only copies the filename from the backing file if no options have been specified for it at all. Furthermore, there are cases where an overlay is created by qemu which already contains a BDS's filename (e.g. in blockdev-snapshot-sync). We do not need to worry about updating the overlay's bs->auto_backing_file there, because we actually wrote a post-bdrv_refresh_filename() filename into the image header. So all in all, there will be false negatives where (as of a future patch) bdrv_refresh_filename() will assume that the backing file differs from what was specified in the image header, even though it really does not. However, these cases should be limited to where (1) the user actually did override something in the backing chain (e.g. by specifying options for the backing file), or (2) the user executed a QMP command to change some node's backing file (e.g. change-backing-file or block-commit with @backing-file given) where the given filename does not happen to coincide with qemu's idea of the backing BDS's filename. Then again, (1) really is limited to -drive. With -blockdev or blockdev-add, you have to adhere to the schema, so a user cannot give partial "unimportant" options (e.g. by just setting backing.node-name and leaving the rest to the image header). Therefore, trying to fix this would mean trying to fix something for -drive only. To improve on (2), we would need a full infrastructure to "canonicalize" an arbitrary filename (+ options), so it can be compared against another. That seems a bit over the top, considering that filenames nowadays are there mostly for the user's entertainment. Signed-off-by: Max Reitz <mreitz@redhat.com> Reviewed-by: Eric Blake <eblake@redhat.com> Reviewed-by: Alberto Garcia <berto@igalia.com> Message-id: 20190201192935.18394-5-mreitz@redhat.com Signed-off-by: Max Reitz <mreitz@redhat.com>
2019-02-01 20:29:08 +01:00
if (implicit_backing) {
bdrv_refresh_filename(backing_hd);
pstrcpy(bs->auto_backing_file, sizeof(bs->auto_backing_file),
backing_hd->filename);
}
/* Hook up the backing file link; drop our reference, bs owns the
* backing_hd reference now */
ret = bdrv_set_backing_hd(bs, backing_hd, errp);
bdrv_unref(backing_hd);
if (ret < 0) {
goto free_exit;
}
qdict_del(parent_options, bdref_key);
free_exit:
g_free(backing_filename);
qobject_unref(tmp_parent_options);
return ret;
}
static BlockDriverState *
bdrv_open_child_bs(const char *filename, QDict *options, const char *bdref_key,
BlockDriverState *parent, const BdrvChildClass *child_class,
BdrvChildRole child_role, bool allow_none, Error **errp)
{
BlockDriverState *bs = NULL;
QDict *image_options;
char *bdref_key_dot;
const char *reference;
assert(child_class != NULL);
bdref_key_dot = g_strdup_printf("%s.", bdref_key);
qdict_extract_subqdict(options, &image_options, bdref_key_dot);
g_free(bdref_key_dot);
block: Document -drive problematic code and bugs -blockdev and blockdev_add convert their arguments via QObject to BlockdevOptions for qmp_blockdev_add(), which converts them back to QObject, then to a flattened QDict. The QDict's members are typed according to the QAPI schema. -drive converts its argument via QemuOpts to a (flat) QDict. This QDict's members are all QString. Thus, the QType of a flat QDict member depends on whether it comes from -drive or -blockdev/blockdev_add, except when the QAPI type maps to QString, which is the case for 'str' and enumeration types. The block layer core extracts generic configuration from the flat QDict, and the block driver extracts driver-specific configuration. Both commonly do so by converting (parts of) the flat QDict to QemuOpts, which turns all values into strings. Not exactly elegant, but correct. However, A few places access the flat QDict directly: * Most of them access members that are always QString. Correct. * bdrv_open_inherit() accesses a boolean, carefully. Correct. * nfs_config() uses a QObject input visitor. Correct only because the visited type contains nothing but QStrings. * nbd_config() and ssh_config() use a QObject input visitor, and the visited types contain non-QStrings: InetSocketAddress members @numeric, @to, @ipv4, @ipv6. -drive works as long as you don't try to use them (they're all optional). @to is ignored anyway. Reproducer: -drive driver=ssh,server.host=h,server.port=22,server.ipv4,path=p -drive driver=nbd,server.type=inet,server.data.host=h,server.data.port=22,server.data.ipv4 both fail with "Invalid parameter type for 'data.ipv4', expected: boolean" Add suitable comments to all these places. Mark the buggy ones FIXME. "Fortunately", -drive's driver-specific options are entirely undocumented. Signed-off-by: Markus Armbruster <armbru@redhat.com> Message-id: 1490895797-29094-5-git-send-email-armbru@redhat.com [mreitz: Fixed two typos] Reviewed-by: Eric Blake <eblake@redhat.com> Signed-off-by: Max Reitz <mreitz@redhat.com>
2017-03-30 19:43:12 +02:00
/*
* Caution: while qdict_get_try_str() is fine, getting non-string
* types would require more care. When @options come from
* -blockdev or blockdev_add, its members are typed according to
* the QAPI schema, but when they come from -drive, they're all
* QString.
*/
reference = qdict_get_try_str(options, bdref_key);
if (!filename && !reference && !qdict_size(image_options)) {
if (!allow_none) {
error_setg(errp, "A block device must be specified for \"%s\"",
bdref_key);
}
qobject_unref(image_options);
goto done;
}
bs = bdrv_open_inherit(filename, reference, image_options, 0,
parent, child_class, child_role, errp);
if (!bs) {
goto done;
}
done:
qdict_del(options, bdref_key);
return bs;
}
/*
* Opens a disk image whose options are given as BlockdevRef in another block
* device's options.
*
* If allow_none is true, no image will be opened if filename is false and no
* BlockdevRef is given. NULL will be returned, but errp remains unset.
*
* bdrev_key specifies the key for the image's BlockdevRef in the options QDict.
* That QDict has to be flattened; therefore, if the BlockdevRef is a QDict
* itself, all options starting with "${bdref_key}." are considered part of the
* BlockdevRef.
*
* The BlockdevRef will be removed from the options QDict.
*/
BdrvChild *bdrv_open_child(const char *filename,
QDict *options, const char *bdref_key,
BlockDriverState *parent,
const BdrvChildClass *child_class,
BdrvChildRole child_role,
bool allow_none, Error **errp)
{
BlockDriverState *bs;
GLOBAL_STATE_CODE();
bs = bdrv_open_child_bs(filename, options, bdref_key, parent, child_class,
child_role, allow_none, errp);
if (bs == NULL) {
return NULL;
}
return bdrv_attach_child(parent, bs, bdref_key, child_class, child_role,
errp);
}
/*
* TODO Future callers may need to specify parent/child_class in order for
* option inheritance to work. Existing callers use it for the root node.
*/
BlockDriverState *bdrv_open_blockdev_ref(BlockdevRef *ref, Error **errp)
{
BlockDriverState *bs = NULL;
QObject *obj = NULL;
QDict *qdict = NULL;
const char *reference = NULL;
Visitor *v = NULL;
GLOBAL_STATE_CODE();
if (ref->type == QTYPE_QSTRING) {
reference = ref->u.reference;
} else {
BlockdevOptions *options = &ref->u.definition;
assert(ref->type == QTYPE_QDICT);
v = qobject_output_visitor_new(&obj);
visit_type_BlockdevOptions(v, NULL, &options, &error_abort);
visit_complete(v, &obj);
qdict = qobject_to(QDict, obj);
qdict_flatten(qdict);
/* bdrv_open_inherit() defaults to the values in bdrv_flags (for
* compatibility with other callers) rather than what we want as the
* real defaults. Apply the defaults here instead. */
qdict_set_default_str(qdict, BDRV_OPT_CACHE_DIRECT, "off");
qdict_set_default_str(qdict, BDRV_OPT_CACHE_NO_FLUSH, "off");
qdict_set_default_str(qdict, BDRV_OPT_READ_ONLY, "off");
block: Add auto-read-only option If a management application builds the block graph node by node, the protocol layer doesn't inherit its read-only option from the format layer any more, so it must be set explicitly. Backing files should work on read-only storage, but at the same time, a block job like commit should be able to reopen them read-write if they are on read-write storage. However, without option inheritance, reopen only changes the read-only option for the root node (typically the format layer), but not the protocol layer, so reopening fails (the format layer wants to get write permissions, but the protocol layer is still read-only). A simple workaround for the problem in the management tool would be to open the protocol layer always read-write and to make only the format layer read-only for backing files. However, sometimes the file is actually stored on read-only storage and we don't know whether the image can be opened read-write (for example, for NBD it depends on the server we're trying to connect to). This adds an option that makes QEMU try to open the image read-write, but allows it to degrade to a read-only mode without returning an error. The documentation for this option is consciously phrased in a way that allows QEMU to switch to a better model eventually: Instead of trying when the image is first opened, making the read-only flag dynamic and changing it automatically whenever the first BLK_PERM_WRITE user is attached or the last one is detached would be much more useful behaviour. Unfortunately, this more useful behaviour is also a lot harder to implement, and libvirt needs a solution now before it can switch to -blockdev, so let's start with this easier approach for now. Instead of adding a new auto-read-only option, turning the existing read-only into an enum (with a bool alternate for compatibility) was considered, but it complicated the implementation to the point that it didn't seem to be worth it. Signed-off-by: Kevin Wolf <kwolf@redhat.com> Reviewed-by: Eric Blake <eblake@redhat.com>
2018-10-05 18:57:40 +02:00
qdict_set_default_str(qdict, BDRV_OPT_AUTO_READ_ONLY, "off");
}
bs = bdrv_open_inherit(NULL, reference, qdict, 0, NULL, NULL, 0, errp);
obj = NULL;
qobject_unref(obj);
visit_free(v);
return bs;
}
static BlockDriverState *bdrv_append_temp_snapshot(BlockDriverState *bs,
int flags,
QDict *snapshot_options,
Error **errp)
{
/* TODO: extra byte is a hack to ensure MAX_PATH space on Windows. */
char *tmp_filename = g_malloc0(PATH_MAX + 1);
int64_t total_size;
QemuOpts *opts = NULL;
BlockDriverState *bs_snapshot = NULL;
int ret;
GLOBAL_STATE_CODE();
/* if snapshot, we create a temporary backing file and open it
instead of opening 'filename' directly */
/* Get the required size from the image */
total_size = bdrv_getlength(bs);
if (total_size < 0) {
error_setg_errno(errp, -total_size, "Could not get image size");
goto out;
}
/* Create the temporary image */
ret = get_tmp_filename(tmp_filename, PATH_MAX + 1);
if (ret < 0) {
error_setg_errno(errp, -ret, "Could not get temporary filename");
goto out;
}
opts = qemu_opts_create(bdrv_qcow2.create_opts, NULL, 0,
&error_abort);
qemu_opt_set_number(opts, BLOCK_OPT_SIZE, total_size, &error_abort);
ret = bdrv_create(&bdrv_qcow2, tmp_filename, opts, errp);
qemu_opts_del(opts);
if (ret < 0) {
error_prepend(errp, "Could not create temporary overlay '%s': ",
tmp_filename);
goto out;
}
/* Prepare options QDict for the temporary file */
qdict_put_str(snapshot_options, "file.driver", "file");
qdict_put_str(snapshot_options, "file.filename", tmp_filename);
qdict_put_str(snapshot_options, "driver", "qcow2");
bs_snapshot = bdrv_open(NULL, NULL, snapshot_options, flags, errp);
snapshot_options = NULL;
if (!bs_snapshot) {
goto out;
}
ret = bdrv_append(bs_snapshot, bs, errp);
if (ret < 0) {
bs_snapshot = NULL;
goto out;
}
out:
qobject_unref(snapshot_options);
g_free(tmp_filename);
return bs_snapshot;
}
/*
* Opens a disk image (raw, qcow2, vmdk, ...)
*
* options is a QDict of options to pass to the block drivers, or NULL for an
* empty set of options. The reference to the QDict belongs to the block layer
* after the call (even on failure), so if the caller intends to reuse the
* dictionary, it needs to use qobject_ref() before calling bdrv_open.
*
* If *pbs is NULL, a new BDS will be created with a pointer to it stored there.
* If it is not NULL, the referenced BDS will be reused.
*
* The reference parameter may be used to specify an existing block device which
* should be opened. If specified, neither options nor a filename may be given,
* nor can an existing BDS be reused (that is, *pbs has to be NULL).
*/
static BlockDriverState *bdrv_open_inherit(const char *filename,
const char *reference,
QDict *options, int flags,
BlockDriverState *parent,
const BdrvChildClass *child_class,
BdrvChildRole child_role,
Error **errp)
{
int ret;
BlockBackend *file = NULL;
BlockDriverState *bs;
BlockDriver *drv = NULL;
BdrvChild *child;
const char *drvname;
const char *backing;
Error *local_err = NULL;
QDict *snapshot_options = NULL;
int snapshot_flags = 0;
assert(!child_class || !flags);
assert(!child_class == !parent);
GLOBAL_STATE_CODE();
if (reference) {
bool options_non_empty = options ? qdict_size(options) : false;
qobject_unref(options);
if (filename || options_non_empty) {
error_setg(errp, "Cannot reference an existing block device with "
"additional options or a new filename");
return NULL;
}
bs = bdrv_lookup_bs(reference, reference, errp);
if (!bs) {
return NULL;
}
bdrv_ref(bs);
return bs;
}
bs = bdrv_new();
/* NULL means an empty set of options */
if (options == NULL) {
options = qdict_new();
}
/* json: syntax counts as explicit options, as if in the QDict */
parse_json_protocol(options, &filename, &local_err);
if (local_err) {
goto fail;
}
bs->explicit_options = qdict_clone_shallow(options);
if (child_class) {
bool parent_is_format;
if (parent->drv) {
parent_is_format = parent->drv->is_format;
} else {
/*
* parent->drv is not set yet because this node is opened for
* (potential) format probing. That means that @parent is going
* to be a format node.
*/
parent_is_format = true;
}
bs->inherits_from = parent;
child_class->inherit_options(child_role, parent_is_format,
&flags, options,
parent->open_flags, parent->options);
}
ret = bdrv_fill_options(&options, filename, &flags, &local_err);
if (ret < 0) {
goto fail;
}
block: Document -drive problematic code and bugs -blockdev and blockdev_add convert their arguments via QObject to BlockdevOptions for qmp_blockdev_add(), which converts them back to QObject, then to a flattened QDict. The QDict's members are typed according to the QAPI schema. -drive converts its argument via QemuOpts to a (flat) QDict. This QDict's members are all QString. Thus, the QType of a flat QDict member depends on whether it comes from -drive or -blockdev/blockdev_add, except when the QAPI type maps to QString, which is the case for 'str' and enumeration types. The block layer core extracts generic configuration from the flat QDict, and the block driver extracts driver-specific configuration. Both commonly do so by converting (parts of) the flat QDict to QemuOpts, which turns all values into strings. Not exactly elegant, but correct. However, A few places access the flat QDict directly: * Most of them access members that are always QString. Correct. * bdrv_open_inherit() accesses a boolean, carefully. Correct. * nfs_config() uses a QObject input visitor. Correct only because the visited type contains nothing but QStrings. * nbd_config() and ssh_config() use a QObject input visitor, and the visited types contain non-QStrings: InetSocketAddress members @numeric, @to, @ipv4, @ipv6. -drive works as long as you don't try to use them (they're all optional). @to is ignored anyway. Reproducer: -drive driver=ssh,server.host=h,server.port=22,server.ipv4,path=p -drive driver=nbd,server.type=inet,server.data.host=h,server.data.port=22,server.data.ipv4 both fail with "Invalid parameter type for 'data.ipv4', expected: boolean" Add suitable comments to all these places. Mark the buggy ones FIXME. "Fortunately", -drive's driver-specific options are entirely undocumented. Signed-off-by: Markus Armbruster <armbru@redhat.com> Message-id: 1490895797-29094-5-git-send-email-armbru@redhat.com [mreitz: Fixed two typos] Reviewed-by: Eric Blake <eblake@redhat.com> Signed-off-by: Max Reitz <mreitz@redhat.com>
2017-03-30 19:43:12 +02:00
/*
* Set the BDRV_O_RDWR and BDRV_O_ALLOW_RDWR flags.
* Caution: getting a boolean member of @options requires care.
* When @options come from -blockdev or blockdev_add, members are
* typed according to the QAPI schema, but when they come from
* -drive, they're all QString.
*/
if (g_strcmp0(qdict_get_try_str(options, BDRV_OPT_READ_ONLY), "on") &&
!qdict_get_try_bool(options, BDRV_OPT_READ_ONLY, false)) {
flags |= (BDRV_O_RDWR | BDRV_O_ALLOW_RDWR);
} else {
flags &= ~BDRV_O_RDWR;
}
if (flags & BDRV_O_SNAPSHOT) {
snapshot_options = qdict_new();
bdrv_temp_snapshot_options(&snapshot_flags, snapshot_options,
flags, options);
/* Let bdrv_backing_options() override "read-only" */
qdict_del(options, BDRV_OPT_READ_ONLY);
bdrv_inherited_options(BDRV_CHILD_COW, true,
&flags, options, flags, options);
}
bs->open_flags = flags;
bs->options = options;
options = qdict_clone_shallow(options);
/* Find the right image format driver */
block: Document -drive problematic code and bugs -blockdev and blockdev_add convert their arguments via QObject to BlockdevOptions for qmp_blockdev_add(), which converts them back to QObject, then to a flattened QDict. The QDict's members are typed according to the QAPI schema. -drive converts its argument via QemuOpts to a (flat) QDict. This QDict's members are all QString. Thus, the QType of a flat QDict member depends on whether it comes from -drive or -blockdev/blockdev_add, except when the QAPI type maps to QString, which is the case for 'str' and enumeration types. The block layer core extracts generic configuration from the flat QDict, and the block driver extracts driver-specific configuration. Both commonly do so by converting (parts of) the flat QDict to QemuOpts, which turns all values into strings. Not exactly elegant, but correct. However, A few places access the flat QDict directly: * Most of them access members that are always QString. Correct. * bdrv_open_inherit() accesses a boolean, carefully. Correct. * nfs_config() uses a QObject input visitor. Correct only because the visited type contains nothing but QStrings. * nbd_config() and ssh_config() use a QObject input visitor, and the visited types contain non-QStrings: InetSocketAddress members @numeric, @to, @ipv4, @ipv6. -drive works as long as you don't try to use them (they're all optional). @to is ignored anyway. Reproducer: -drive driver=ssh,server.host=h,server.port=22,server.ipv4,path=p -drive driver=nbd,server.type=inet,server.data.host=h,server.data.port=22,server.data.ipv4 both fail with "Invalid parameter type for 'data.ipv4', expected: boolean" Add suitable comments to all these places. Mark the buggy ones FIXME. "Fortunately", -drive's driver-specific options are entirely undocumented. Signed-off-by: Markus Armbruster <armbru@redhat.com> Message-id: 1490895797-29094-5-git-send-email-armbru@redhat.com [mreitz: Fixed two typos] Reviewed-by: Eric Blake <eblake@redhat.com> Signed-off-by: Max Reitz <mreitz@redhat.com>
2017-03-30 19:43:12 +02:00
/* See cautionary note on accessing @options above */
drvname = qdict_get_try_str(options, "driver");
if (drvname) {
drv = bdrv_find_format(drvname);
if (!drv) {
error_setg(errp, "Unknown driver: '%s'", drvname);
goto fail;
}
}
assert(drvname || !(flags & BDRV_O_PROTOCOL));
block: Document -drive problematic code and bugs -blockdev and blockdev_add convert their arguments via QObject to BlockdevOptions for qmp_blockdev_add(), which converts them back to QObject, then to a flattened QDict. The QDict's members are typed according to the QAPI schema. -drive converts its argument via QemuOpts to a (flat) QDict. This QDict's members are all QString. Thus, the QType of a flat QDict member depends on whether it comes from -drive or -blockdev/blockdev_add, except when the QAPI type maps to QString, which is the case for 'str' and enumeration types. The block layer core extracts generic configuration from the flat QDict, and the block driver extracts driver-specific configuration. Both commonly do so by converting (parts of) the flat QDict to QemuOpts, which turns all values into strings. Not exactly elegant, but correct. However, A few places access the flat QDict directly: * Most of them access members that are always QString. Correct. * bdrv_open_inherit() accesses a boolean, carefully. Correct. * nfs_config() uses a QObject input visitor. Correct only because the visited type contains nothing but QStrings. * nbd_config() and ssh_config() use a QObject input visitor, and the visited types contain non-QStrings: InetSocketAddress members @numeric, @to, @ipv4, @ipv6. -drive works as long as you don't try to use them (they're all optional). @to is ignored anyway. Reproducer: -drive driver=ssh,server.host=h,server.port=22,server.ipv4,path=p -drive driver=nbd,server.type=inet,server.data.host=h,server.data.port=22,server.data.ipv4 both fail with "Invalid parameter type for 'data.ipv4', expected: boolean" Add suitable comments to all these places. Mark the buggy ones FIXME. "Fortunately", -drive's driver-specific options are entirely undocumented. Signed-off-by: Markus Armbruster <armbru@redhat.com> Message-id: 1490895797-29094-5-git-send-email-armbru@redhat.com [mreitz: Fixed two typos] Reviewed-by: Eric Blake <eblake@redhat.com> Signed-off-by: Max Reitz <mreitz@redhat.com>
2017-03-30 19:43:12 +02:00
/* See cautionary note on accessing @options above */
backing = qdict_get_try_str(options, "backing");
if (qobject_to(QNull, qdict_get(options, "backing")) != NULL ||
(backing && *backing == '\0'))
{
if (backing) {
warn_report("Use of \"backing\": \"\" is deprecated; "
"use \"backing\": null instead");
}
flags |= BDRV_O_NO_BACKING;
qdict_del(bs->explicit_options, "backing");
qdict_del(bs->options, "backing");
qdict_del(options, "backing");
}
/* Open image file without format layer. This BlockBackend is only used for
* probing, the block drivers will do their own bdrv_open_child() for the
* same BDS, which is why we put the node name back into options. */
if ((flags & BDRV_O_PROTOCOL) == 0) {
BlockDriverState *file_bs;
file_bs = bdrv_open_child_bs(filename, options, "file", bs,
&child_of_bds, BDRV_CHILD_IMAGE,
true, &local_err);
if (local_err) {
goto fail;
}
if (file_bs != NULL) {
/* Not requesting BLK_PERM_CONSISTENT_READ because we're only
* looking at the header to guess the image format. This works even
* in cases where a guest would not see a consistent state. */
file = blk_new(bdrv_get_aio_context(file_bs), 0, BLK_PERM_ALL);
blk_insert_bs(file, file_bs, &local_err);
bdrv_unref(file_bs);
if (local_err) {
goto fail;
}
qdict_put_str(options, "file", bdrv_get_node_name(file_bs));
}
}
/* Image format probing */
raw: Prohibit dangerous writes for probed images If the user neglects to specify the image format, QEMU probes the image to guess it automatically, for convenience. Relying on format probing is insecure for raw images (CVE-2008-2004). If the guest writes a suitable header to the device, the next probe will recognize a format chosen by the guest. A malicious guest can abuse this to gain access to host files, e.g. by crafting a QCOW2 header with backing file /etc/shadow. Commit 1e72d3b (April 2008) provided -drive parameter format to let users disable probing. Commit f965509 (March 2009) extended QCOW2 to optionally store the backing file format, to let users disable backing file probing. QED has had a flag to suppress probing since the beginning (2010), set whenever a raw backing file is assigned. All of these additions that allow to avoid format probing have to be specified explicitly. The default still allows the attack. In order to fix this, commit 79368c8 (July 2010) put probed raw images in a restricted mode, in which they wouldn't be able to overwrite the first few bytes of the image so that they would identify as a different image. If a write to the first sector would write one of the signatures of another driver, qemu would instead zero out the first four bytes. This patch was later reverted in commit 8b33d9e (September 2010) because it didn't get the handling of unaligned qiov members right. Today's block layer that is based on coroutines and has qiov utility functions makes it much easier to get this functionality right, so this patch implements it. The other differences of this patch to the old one are that it doesn't silently write something different than the guest requested by zeroing out some bytes (it fails the request instead) and that it doesn't maintain a list of signatures in the raw driver (it calls the usual probe function instead). Note that this change doesn't introduce new breakage for false positive cases where the guest legitimately writes data into the first sector that matches the signatures of an image format (e.g. for nested virt): These cases were broken before, only the failure mode changes from corruption after the next restart (when the wrong format is probed) to failing the problematic write request. Also note that like in the original patch, the restrictions only apply if the image format has been guessed by probing. Explicitly specifying a format allows guests to write anything they like. Signed-off-by: Kevin Wolf <kwolf@redhat.com> Reviewed-by: Eric Blake <eblake@redhat.com> Reviewed-by: Max Reitz <mreitz@redhat.com> Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com> Message-id: 1416497234-29880-8-git-send-email-kwolf@redhat.com Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2014-11-20 16:27:12 +01:00
bs->probed = !drv;
if (!drv && file) {
ret = find_image_format(file, filename, &drv, &local_err);
if (ret < 0) {
goto fail;
}
/*
* This option update would logically belong in bdrv_fill_options(),
* but we first need to open bs->file for the probing to work, while
* opening bs->file already requires the (mostly) final set of options
* so that cache mode etc. can be inherited.
*
* Adding the driver later is somewhat ugly, but it's not an option
* that would ever be inherited, so it's correct. We just need to make
* sure to update both bs->options (which has the full effective
* options for bs) and options (which has file.* already removed).
*/
qdict_put_str(bs->options, "driver", drv->format_name);
qdict_put_str(options, "driver", drv->format_name);
} else if (!drv) {
error_setg(errp, "Must specify either driver or file");
goto fail;
}
block: driver should override flags in bdrv_open() The BDRV_O_PROTOCOL flag should have an impact only if no driver is specified explicitly. Therefore, if bdrv_open() is called with an explicit block driver argument (either through the options QDict or through the drv parameter) and that block driver is a protocol block driver, BDRV_O_PROTOCOL should be set; if it is a format block driver, BDRV_O_PROTOCOL should be unset. While there was code to unset the flag in case a format block driver has been selected, it only followed the bdrv_fill_options() function call whereas the flag in fact needs to be adjusted before it is used there. With that change, BDRV_O_PROTOCOL will always be set if the BDS should be a protocol driver; if the driver has been specified explicitly, the new code will set it; and bdrv_fill_options() will only "probe" a protocol driver if BDRV_O_PROTOCOL is set. The probing after bdrv_fill_options() cannot select a protocol driver. Thus, bdrv_open_image() to open BDS.file is never called if a protocol BDS is about to be created. With that change in turn it is impossible to call bdrv_open_common() with a protocol drv and file != NULL, which allows us to remove the bdrv_swap() call. This change breaks a test case in qemu-iotest 051: "-drive file=t.qcow2,file.driver=qcow2" now works because the explicitly specified "qcow2" overrides the BDRV_O_PROTOCOL which is automatically set for the "file" BDS (and the filename is just passed down). Therefore, this patch removes that test case. Signed-off-by: Max Reitz <mreitz@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2015-03-19 19:53:16 +01:00
/* BDRV_O_PROTOCOL must be set iff a protocol BDS is about to be created */
assert(!!(flags & BDRV_O_PROTOCOL) == !!drv->bdrv_file_open);
/* file must be NULL if a protocol BDS is about to be created
* (the inverse results in an error message from bdrv_open_common()) */
assert(!(flags & BDRV_O_PROTOCOL) || !file);
/* Open the image */
ret = bdrv_open_common(bs, file, options, &local_err);
if (ret < 0) {
goto fail;
}
if (file) {
blk_unref(file);
file = NULL;
}
/* If there is a backing file, use it */
if ((flags & BDRV_O_NO_BACKING) == 0) {
ret = bdrv_open_backing_file(bs, options, "backing", &local_err);
if (ret < 0) {
goto close_and_fail;
}
}
/* Remove all children options and references
* from bs->options and bs->explicit_options */
QLIST_FOREACH(child, &bs->children, next) {
char *child_key_dot;
child_key_dot = g_strdup_printf("%s.", child->name);
qdict_extract_subqdict(bs->explicit_options, NULL, child_key_dot);
qdict_extract_subqdict(bs->options, NULL, child_key_dot);
qdict_del(bs->explicit_options, child->name);
qdict_del(bs->options, child->name);
g_free(child_key_dot);
}
/* Check if any unknown options were used */
if (qdict_size(options) != 0) {
const QDictEntry *entry = qdict_first(options);
if (flags & BDRV_O_PROTOCOL) {
error_setg(errp, "Block protocol '%s' doesn't support the option "
"'%s'", drv->format_name, entry->key);
} else {
error_setg(errp,
"Block format '%s' does not support the option '%s'",
drv->format_name, entry->key);
}
goto close_and_fail;
}
bdrv_parent_cb_change_media(bs, true);
qobject_unref(options);
options = NULL;
/* For snapshot=on, create a temporary qcow2 overlay. bs points to the
* temporary snapshot afterwards. */
if (snapshot_flags) {
BlockDriverState *snapshot_bs;
snapshot_bs = bdrv_append_temp_snapshot(bs, snapshot_flags,
snapshot_options, &local_err);
snapshot_options = NULL;
if (local_err) {
goto close_and_fail;
}
/* We are not going to return bs but the overlay on top of it
* (snapshot_bs); thus, we have to drop the strong reference to bs
* (which we obtained by calling bdrv_new()). bs will not be deleted,
* though, because the overlay still has a reference to it. */
bdrv_unref(bs);
bs = snapshot_bs;
}
return bs;
fail:
blk_unref(file);
qobject_unref(snapshot_options);
qobject_unref(bs->explicit_options);
qobject_unref(bs->options);
qobject_unref(options);
bs->options = NULL;
bs->explicit_options = NULL;
bdrv_unref(bs);
error_propagate(errp, local_err);
return NULL;
close_and_fail:
bdrv_unref(bs);
qobject_unref(snapshot_options);
qobject_unref(options);
error_propagate(errp, local_err);
return NULL;
}
BlockDriverState *bdrv_open(const char *filename, const char *reference,
QDict *options, int flags, Error **errp)
{
GLOBAL_STATE_CODE();
return bdrv_open_inherit(filename, reference, options, flags, NULL,
NULL, 0, errp);
}
/* Return true if the NULL-terminated @list contains @str */
static bool is_str_in_list(const char *str, const char *const *list)
{
if (str && list) {
int i;
for (i = 0; list[i] != NULL; i++) {
if (!strcmp(str, list[i])) {
return true;
}
}
}
return false;
}
/*
* Check that every option set in @bs->options is also set in
* @new_opts.
*
* Options listed in the common_options list and in
* @bs->drv->mutable_opts are skipped.
*
* Return 0 on success, otherwise return -EINVAL and set @errp.
*/
static int bdrv_reset_options_allowed(BlockDriverState *bs,
const QDict *new_opts, Error **errp)
{
const QDictEntry *e;
/* These options are common to all block drivers and are handled
* in bdrv_reopen_prepare() so they can be left out of @new_opts */
const char *const common_options[] = {
"node-name", "discard", "cache.direct", "cache.no-flush",
"read-only", "auto-read-only", "detect-zeroes", NULL
};
for (e = qdict_first(bs->options); e; e = qdict_next(bs->options, e)) {
if (!qdict_haskey(new_opts, e->key) &&
!is_str_in_list(e->key, common_options) &&
!is_str_in_list(e->key, bs->drv->mutable_opts)) {
error_setg(errp, "Option '%s' cannot be reset "
"to its default value", e->key);
return -EINVAL;
}
}
return 0;
}
/*
* Returns true if @child can be reached recursively from @bs
*/
static bool bdrv_recurse_has_child(BlockDriverState *bs,
BlockDriverState *child)
{
BdrvChild *c;
if (bs == child) {
return true;
}
QLIST_FOREACH(c, &bs->children, next) {
if (bdrv_recurse_has_child(c->bs, child)) {
return true;
}
}
return false;
}
/*
* Adds a BlockDriverState to a simple queue for an atomic, transactional
* reopen of multiple devices.
*
* bs_queue can either be an existing BlockReopenQueue that has had QTAILQ_INIT
* already performed, or alternatively may be NULL a new BlockReopenQueue will
* be created and initialized. This newly created BlockReopenQueue should be
* passed back in for subsequent calls that are intended to be of the same
* atomic 'set'.
*
* bs is the BlockDriverState to add to the reopen queue.
*
* options contains the changed options for the associated bs
* (the BlockReopenQueue takes ownership)
*
* flags contains the open flags for the associated bs
*
* returns a pointer to bs_queue, which is either the newly allocated
* bs_queue, or the existing bs_queue being used.
*
* bs must be drained between bdrv_reopen_queue() and bdrv_reopen_multiple().
*/
static BlockReopenQueue *bdrv_reopen_queue_child(BlockReopenQueue *bs_queue,
BlockDriverState *bs,
QDict *options,
const BdrvChildClass *klass,
BdrvChildRole role,
bool parent_is_format,
QDict *parent_options,
int parent_flags,
bool keep_old_opts)
{
assert(bs != NULL);
BlockReopenQueueEntry *bs_entry;
BdrvChild *child;
QDict *old_options, *explicit_options, *options_copy;
int flags;
QemuOpts *opts;
/* Make sure that the caller remembered to use a drained section. This is
* important to avoid graph changes between the recursive queuing here and
* bdrv_reopen_multiple(). */
assert(bs->quiesce_counter > 0);
GLOBAL_STATE_CODE();
if (bs_queue == NULL) {
bs_queue = g_new0(BlockReopenQueue, 1);
QTAILQ_INIT(bs_queue);
}
if (!options) {
options = qdict_new();
}
/* Check if this BlockDriverState is already in the queue */
QTAILQ_FOREACH(bs_entry, bs_queue, entry) {
if (bs == bs_entry->state.bs) {
break;
}
}
/*
* Precedence of options:
* 1. Explicitly passed in options (highest)
* 2. Retained from explicitly set options of bs
* 3. Inherited from parent node
* 4. Retained from effective options of bs
*/
/* Old explicitly set values (don't overwrite by inherited value) */
if (bs_entry || keep_old_opts) {
old_options = qdict_clone_shallow(bs_entry ?
bs_entry->state.explicit_options :
bs->explicit_options);
bdrv_join_options(bs, options, old_options);
qobject_unref(old_options);
}
explicit_options = qdict_clone_shallow(options);
/* Inherit from parent node */
if (parent_options) {
flags = 0;
klass->inherit_options(role, parent_is_format, &flags, options,
parent_flags, parent_options);
} else {
flags = bdrv_get_flags(bs);
}
if (keep_old_opts) {
/* Old values are used for options that aren't set yet */
old_options = qdict_clone_shallow(bs->options);
bdrv_join_options(bs, options, old_options);
qobject_unref(old_options);
}
/* We have the final set of options so let's update the flags */
options_copy = qdict_clone_shallow(options);
opts = qemu_opts_create(&bdrv_runtime_opts, NULL, 0, &error_abort);
qemu_opts_absorb_qdict(opts, options_copy, NULL);
update_flags_from_options(&flags, opts);
qemu_opts_del(opts);
qobject_unref(options_copy);
/* bdrv_open_inherit() sets and clears some additional flags internally */
flags &= ~BDRV_O_PROTOCOL;
if (flags & BDRV_O_RDWR) {
flags |= BDRV_O_ALLOW_RDWR;
}
if (!bs_entry) {
bs_entry = g_new0(BlockReopenQueueEntry, 1);
QTAILQ_INSERT_TAIL(bs_queue, bs_entry, entry);
} else {
qobject_unref(bs_entry->state.options);
qobject_unref(bs_entry->state.explicit_options);
}
bs_entry->state.bs = bs;
bs_entry->state.options = options;
bs_entry->state.explicit_options = explicit_options;
bs_entry->state.flags = flags;
block: Handle child references in bdrv_reopen_queue() Children in QMP are specified with BlockdevRef / BlockdevRefOrNull, which can contain a set of child options, a child reference, or NULL. In optional attributes like "backing" it can also be missing. Only the first case (set of child options) is being handled properly by bdrv_reopen_queue(). This patch deals with all the others. Here's how these cases should be handled when bdrv_reopen_queue() is deciding what to do with each child of a BlockDriverState: 1) Set of child options: if the child was implicitly created (i.e inherits_from points to the parent) then the options are removed from the parent's options QDict and are passed to the child with a recursive bdrv_reopen_queue() call. This case was already working fine. 2) Child reference: there's two possibilites here. 2a) Reference to the current child: if the child was implicitly created then it is put in the reopen queue, keeping its current set of options (since this was a child reference there was no way to specify a different set of options). If the child is not implicit then it keeps its current set of options but it is not reopened (and therefore does not inherit any new option from the parent). 2b) Reference to a different BDS: the current child is not put in the reopen queue at all. Passing a reference to a different BDS can be used to replace a child, although at the moment no driver implements this, so it results in an error. In any case, the current child is not going to be reopened (and might in fact disappear if it's replaced) 3) NULL: This is similar to (2b). Although no driver allows this yet it can be used to detach the current child so it should not be put in the reopen queue. 4) Missing option: at the moment "backing" is the only case where this can happen. With "blockdev-add", leaving "backing" out means that the default backing file is opened. We don't want to open a new image during reopen, so we require that "backing" is always present. We'll relax this requirement a bit in the next patch. If keep_old_opts is true and "backing" is missing then this behaves like 2a (the current child is reopened). Signed-off-by: Alberto Garcia <berto@igalia.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2019-03-12 17:48:45 +01:00
/*
* If keep_old_opts is false then it means that unspecified
* options must be reset to their original value. We don't allow
* resetting 'backing' but we need to know if the option is
* missing in order to decide if we have to return an error.
*/
if (!keep_old_opts) {
bs_entry->state.backing_missing =
!qdict_haskey(options, "backing") &&
!qdict_haskey(options, "backing.driver");
}
QLIST_FOREACH(child, &bs->children, next) {
block: Handle child references in bdrv_reopen_queue() Children in QMP are specified with BlockdevRef / BlockdevRefOrNull, which can contain a set of child options, a child reference, or NULL. In optional attributes like "backing" it can also be missing. Only the first case (set of child options) is being handled properly by bdrv_reopen_queue(). This patch deals with all the others. Here's how these cases should be handled when bdrv_reopen_queue() is deciding what to do with each child of a BlockDriverState: 1) Set of child options: if the child was implicitly created (i.e inherits_from points to the parent) then the options are removed from the parent's options QDict and are passed to the child with a recursive bdrv_reopen_queue() call. This case was already working fine. 2) Child reference: there's two possibilites here. 2a) Reference to the current child: if the child was implicitly created then it is put in the reopen queue, keeping its current set of options (since this was a child reference there was no way to specify a different set of options). If the child is not implicit then it keeps its current set of options but it is not reopened (and therefore does not inherit any new option from the parent). 2b) Reference to a different BDS: the current child is not put in the reopen queue at all. Passing a reference to a different BDS can be used to replace a child, although at the moment no driver implements this, so it results in an error. In any case, the current child is not going to be reopened (and might in fact disappear if it's replaced) 3) NULL: This is similar to (2b). Although no driver allows this yet it can be used to detach the current child so it should not be put in the reopen queue. 4) Missing option: at the moment "backing" is the only case where this can happen. With "blockdev-add", leaving "backing" out means that the default backing file is opened. We don't want to open a new image during reopen, so we require that "backing" is always present. We'll relax this requirement a bit in the next patch. If keep_old_opts is true and "backing" is missing then this behaves like 2a (the current child is reopened). Signed-off-by: Alberto Garcia <berto@igalia.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2019-03-12 17:48:45 +01:00
QDict *new_child_options = NULL;
bool child_keep_old = keep_old_opts;
/* reopen can only change the options of block devices that were
* implicitly created and inherited options. For other (referenced)
* block devices, a syntax like "backing.foo" results in an error. */
if (child->bs->inherits_from != bs) {
continue;
}
block: Handle child references in bdrv_reopen_queue() Children in QMP are specified with BlockdevRef / BlockdevRefOrNull, which can contain a set of child options, a child reference, or NULL. In optional attributes like "backing" it can also be missing. Only the first case (set of child options) is being handled properly by bdrv_reopen_queue(). This patch deals with all the others. Here's how these cases should be handled when bdrv_reopen_queue() is deciding what to do with each child of a BlockDriverState: 1) Set of child options: if the child was implicitly created (i.e inherits_from points to the parent) then the options are removed from the parent's options QDict and are passed to the child with a recursive bdrv_reopen_queue() call. This case was already working fine. 2) Child reference: there's two possibilites here. 2a) Reference to the current child: if the child was implicitly created then it is put in the reopen queue, keeping its current set of options (since this was a child reference there was no way to specify a different set of options). If the child is not implicit then it keeps its current set of options but it is not reopened (and therefore does not inherit any new option from the parent). 2b) Reference to a different BDS: the current child is not put in the reopen queue at all. Passing a reference to a different BDS can be used to replace a child, although at the moment no driver implements this, so it results in an error. In any case, the current child is not going to be reopened (and might in fact disappear if it's replaced) 3) NULL: This is similar to (2b). Although no driver allows this yet it can be used to detach the current child so it should not be put in the reopen queue. 4) Missing option: at the moment "backing" is the only case where this can happen. With "blockdev-add", leaving "backing" out means that the default backing file is opened. We don't want to open a new image during reopen, so we require that "backing" is always present. We'll relax this requirement a bit in the next patch. If keep_old_opts is true and "backing" is missing then this behaves like 2a (the current child is reopened). Signed-off-by: Alberto Garcia <berto@igalia.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2019-03-12 17:48:45 +01:00
/* Check if the options contain a child reference */
if (qdict_haskey(options, child->name)) {
const char *childref = qdict_get_try_str(options, child->name);
/*
* The current child must not be reopened if the child
* reference is null or points to a different node.
*/
if (g_strcmp0(childref, child->bs->node_name)) {
continue;
}
/*
* If the child reference points to the current child then
* reopen it with its existing set of options (note that
* it can still inherit new options from the parent).
*/
child_keep_old = true;
} else {
/* Extract child options ("child-name.*") */
char *child_key_dot = g_strdup_printf("%s.", child->name);
qdict_extract_subqdict(explicit_options, NULL, child_key_dot);
qdict_extract_subqdict(options, &new_child_options, child_key_dot);
g_free(child_key_dot);
}
bdrv_reopen_queue_child(bs_queue, child->bs, new_child_options,
child->klass, child->role, bs->drv->is_format,
options, flags, child_keep_old);
}
return bs_queue;
}
BlockReopenQueue *bdrv_reopen_queue(BlockReopenQueue *bs_queue,
BlockDriverState *bs,
QDict *options, bool keep_old_opts)
{
GLOBAL_STATE_CODE();
return bdrv_reopen_queue_child(bs_queue, bs, options, NULL, 0, false,
NULL, 0, keep_old_opts);
}
void bdrv_reopen_queue_free(BlockReopenQueue *bs_queue)
{
GLOBAL_STATE_CODE();
if (bs_queue) {
BlockReopenQueueEntry *bs_entry, *next;
QTAILQ_FOREACH_SAFE(bs_entry, bs_queue, entry, next) {
qobject_unref(bs_entry->state.explicit_options);
qobject_unref(bs_entry->state.options);
g_free(bs_entry);
}
g_free(bs_queue);
}
}
/*
* Reopen multiple BlockDriverStates atomically & transactionally.
*
* The queue passed in (bs_queue) must have been built up previous
* via bdrv_reopen_queue().
*
* Reopens all BDS specified in the queue, with the appropriate
* flags. All devices are prepared for reopen, and failure of any
* device will cause all device changes to be abandoned, and intermediate
* data cleaned up.
*
* If all devices prepare successfully, then the changes are committed
* to all devices.
*
* All affected nodes must be drained between bdrv_reopen_queue() and
* bdrv_reopen_multiple().
*
* To be called from the main thread, with all other AioContexts unlocked.
*/
int bdrv_reopen_multiple(BlockReopenQueue *bs_queue, Error **errp)
{
int ret = -1;
BlockReopenQueueEntry *bs_entry, *next;
AioContext *ctx;
Transaction *tran = tran_new();
g_autoptr(GHashTable) found = NULL;
g_autoptr(GSList) refresh_list = NULL;
assert(qemu_get_current_aio_context() == qemu_get_aio_context());
assert(bs_queue != NULL);
GLOBAL_STATE_CODE();
QTAILQ_FOREACH(bs_entry, bs_queue, entry) {
ctx = bdrv_get_aio_context(bs_entry->state.bs);
aio_context_acquire(ctx);
ret = bdrv_flush(bs_entry->state.bs);
aio_context_release(ctx);
if (ret < 0) {
error_setg_errno(errp, -ret, "Error flushing drive");
goto abort;
}
}
QTAILQ_FOREACH(bs_entry, bs_queue, entry) {
assert(bs_entry->state.bs->quiesce_counter > 0);
ctx = bdrv_get_aio_context(bs_entry->state.bs);
aio_context_acquire(ctx);
ret = bdrv_reopen_prepare(&bs_entry->state, bs_queue, tran, errp);
aio_context_release(ctx);
if (ret < 0) {
goto abort;
}
bs_entry->prepared = true;
}
found = g_hash_table_new(NULL, NULL);
QTAILQ_FOREACH(bs_entry, bs_queue, entry) {
BDRVReopenState *state = &bs_entry->state;
refresh_list = bdrv_topological_dfs(refresh_list, found, state->bs);
if (state->old_backing_bs) {
refresh_list = bdrv_topological_dfs(refresh_list, found,
state->old_backing_bs);
}
if (state->old_file_bs) {
refresh_list = bdrv_topological_dfs(refresh_list, found,
state->old_file_bs);
}
}
/*
* Note that file-posix driver rely on permission update done during reopen
* (even if no permission changed), because it wants "new" permissions for
* reconfiguring the fd and that's why it does it in raw_check_perm(), not
* in raw_reopen_prepare() which is called with "old" permissions.
*/
ret = bdrv_list_refresh_perms(refresh_list, bs_queue, tran, errp);
if (ret < 0) {
goto abort;
}
/*
* If we reach this point, we have success and just need to apply the
* changes.
*
* Reverse order is used to comfort qcow2 driver: on commit it need to write
* IN_USE flag to the image, to mark bitmaps in the image as invalid. But
* children are usually goes after parents in reopen-queue, so go from last
* to first element.
*/
QTAILQ_FOREACH_REVERSE(bs_entry, bs_queue, entry) {
ctx = bdrv_get_aio_context(bs_entry->state.bs);
aio_context_acquire(ctx);
bdrv_reopen_commit(&bs_entry->state);
aio_context_release(ctx);
}
tran_commit(tran);
QTAILQ_FOREACH_REVERSE(bs_entry, bs_queue, entry) {
BlockDriverState *bs = bs_entry->state.bs;
if (bs->drv->bdrv_reopen_commit_post) {
ctx = bdrv_get_aio_context(bs);
aio_context_acquire(ctx);
bs->drv->bdrv_reopen_commit_post(&bs_entry->state);
aio_context_release(ctx);
}
}
ret = 0;
goto cleanup;
abort:
tran_abort(tran);
QTAILQ_FOREACH_SAFE(bs_entry, bs_queue, entry, next) {
if (bs_entry->prepared) {
ctx = bdrv_get_aio_context(bs_entry->state.bs);
aio_context_acquire(ctx);
bdrv_reopen_abort(&bs_entry->state);
aio_context_release(ctx);
}
}
cleanup:
bdrv_reopen_queue_free(bs_queue);
return ret;
}
int bdrv_reopen(BlockDriverState *bs, QDict *opts, bool keep_old_opts,
Error **errp)
{
AioContext *ctx = bdrv_get_aio_context(bs);
BlockReopenQueue *queue;
int ret;
GLOBAL_STATE_CODE();
bdrv_subtree_drained_begin(bs);
if (ctx != qemu_get_aio_context()) {
aio_context_release(ctx);
}
queue = bdrv_reopen_queue(NULL, bs, opts, keep_old_opts);
ret = bdrv_reopen_multiple(queue, errp);
if (ctx != qemu_get_aio_context()) {
aio_context_acquire(ctx);
}
bdrv_subtree_drained_end(bs);
return ret;
}
int bdrv_reopen_set_read_only(BlockDriverState *bs, bool read_only,
Error **errp)
{
QDict *opts = qdict_new();
GLOBAL_STATE_CODE();
qdict_put_bool(opts, BDRV_OPT_READ_ONLY, read_only);
return bdrv_reopen(bs, opts, true, errp);
}
/*
* Take a BDRVReopenState and check if the value of 'backing' in the
* reopen_state->options QDict is valid or not.
*
* If 'backing' is missing from the QDict then return 0.
*
* If 'backing' contains the node name of the backing file of
* reopen_state->bs then return 0.
*
* If 'backing' contains a different node name (or is null) then check
* whether the current backing file can be replaced with the new one.
* If that's the case then reopen_state->replace_backing_bs is set to
* true and reopen_state->new_backing_bs contains a pointer to the new
* backing BlockDriverState (or NULL).
*
* Return 0 on success, otherwise return < 0 and set @errp.
*/
static int bdrv_reopen_parse_file_or_backing(BDRVReopenState *reopen_state,
bool is_backing, Transaction *tran,
Error **errp)
{
BlockDriverState *bs = reopen_state->bs;
BlockDriverState *new_child_bs;
BlockDriverState *old_child_bs = is_backing ? child_bs(bs->backing) :
child_bs(bs->file);
const char *child_name = is_backing ? "backing" : "file";
QObject *value;
const char *str;
GLOBAL_STATE_CODE();
value = qdict_get(reopen_state->options, child_name);
if (value == NULL) {
return 0;
}
switch (qobject_type(value)) {
case QTYPE_QNULL:
assert(is_backing); /* The 'file' option does not allow a null value */
new_child_bs = NULL;
break;
case QTYPE_QSTRING:
str = qstring_get_str(qobject_to(QString, value));
new_child_bs = bdrv_lookup_bs(NULL, str, errp);
if (new_child_bs == NULL) {
return -EINVAL;
} else if (bdrv_recurse_has_child(new_child_bs, bs)) {
error_setg(errp, "Making '%s' a %s child of '%s' would create a "
"cycle", str, child_name, bs->node_name);
return -EINVAL;
}
break;
default:
/*
* The options QDict has been flattened, so 'backing' and 'file'
* do not allow any other data type here.
*/
g_assert_not_reached();
}
if (old_child_bs == new_child_bs) {
return 0;
}
if (old_child_bs) {
if (bdrv_skip_implicit_filters(old_child_bs) == new_child_bs) {
return 0;
}
if (old_child_bs->implicit) {
error_setg(errp, "Cannot replace implicit %s child of %s",
child_name, bs->node_name);
return -EPERM;
}
}
if (bs->drv->is_filter && !old_child_bs) {
/*
* Filters always have a file or a backing child, so we are trying to
* change wrong child
*/
error_setg(errp, "'%s' is a %s filter node that does not support a "
"%s child", bs->node_name, bs->drv->format_name, child_name);
return -EINVAL;
}
if (is_backing) {
reopen_state->old_backing_bs = old_child_bs;
} else {
reopen_state->old_file_bs = old_child_bs;
}
return bdrv_set_file_or_backing_noperm(bs, new_child_bs, is_backing,
tran, errp);
}
/*
* Prepares a BlockDriverState for reopen. All changes are staged in the
* 'opaque' field of the BDRVReopenState, which is used and allocated by
* the block driver layer .bdrv_reopen_prepare()
*
* bs is the BlockDriverState to reopen
* flags are the new open flags
* queue is the reopen queue
*
* Returns 0 on success, non-zero on error. On error errp will be set
* as well.
*
* On failure, bdrv_reopen_abort() will be called to clean up any data.
* It is the responsibility of the caller to then call the abort() or
* commit() for any other BDS that have been left in a prepare() state
*
*/
static int bdrv_reopen_prepare(BDRVReopenState *reopen_state,
BlockReopenQueue *queue,
Transaction *change_child_tran, Error **errp)
{
int ret = -1;
int old_flags;
Error *local_err = NULL;
BlockDriver *drv;
QemuOpts *opts;
QDict *orig_reopen_opts;
char *discard = NULL;
bool read_only;
bool drv_prepared = false;
assert(reopen_state != NULL);
assert(reopen_state->bs->drv != NULL);
GLOBAL_STATE_CODE();
drv = reopen_state->bs->drv;
/* This function and each driver's bdrv_reopen_prepare() remove
* entries from reopen_state->options as they are processed, so
* we need to make a copy of the original QDict. */
orig_reopen_opts = qdict_clone_shallow(reopen_state->options);
/* Process generic block layer options */
opts = qemu_opts_create(&bdrv_runtime_opts, NULL, 0, &error_abort);
if (!qemu_opts_absorb_qdict(opts, reopen_state->options, errp)) {
ret = -EINVAL;
goto error;
}
/* This was already called in bdrv_reopen_queue_child() so the flags
* are up-to-date. This time we simply want to remove the options from
* QemuOpts in order to indicate that they have been processed. */
old_flags = reopen_state->flags;
update_flags_from_options(&reopen_state->flags, opts);
assert(old_flags == reopen_state->flags);
discard = qemu_opt_get_del(opts, BDRV_OPT_DISCARD);
if (discard != NULL) {
if (bdrv_parse_discard_flags(discard, &reopen_state->flags) != 0) {
error_setg(errp, "Invalid discard option");
ret = -EINVAL;
goto error;
}
}
reopen_state->detect_zeroes =
bdrv_parse_detect_zeroes(opts, reopen_state->flags, &local_err);
if (local_err) {
error_propagate(errp, local_err);
ret = -EINVAL;
goto error;
}
block: Forbid trying to change unsupported options during reopen The bdrv_reopen_prepare() function checks all options passed to each BlockDriverState (in the reopen_state->options QDict) and makes all necessary preparations to apply the option changes requested by the user. Options are removed from the QDict as they are processed, so at the end of bdrv_reopen_prepare() only the options that can't be changed are left. Then a loop goes over all remaining options and verifies that the old and new values are identical, returning an error if they're not. The problem is that at the moment there are options that are removed from the QDict although they can't be changed. The consequence of this is any modification to any of those options is silently ignored: (qemu) qemu-io virtio0 "reopen -o discard=on" This happens when all options from bdrv_runtime_opts are removed from the QDict but then only a few of them are processed. Since it's especially important that "node-name" and "driver" are not changed, the code puts them back into the QDict so they are checked at the end of the function. Instead of putting only those two options back into the QDict, this patch puts all unprocessed options using qemu_opts_to_qdict(). update_flags_from_options() also needs to be modified to prevent BDRV_OPT_CACHE_NO_FLUSH, BDRV_OPT_CACHE_DIRECT and BDRV_OPT_READ_ONLY from going back to the QDict. Signed-off-by: Alberto Garcia <berto@igalia.com> Reviewed-by: Max Reitz <mreitz@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2018-09-06 11:37:06 +02:00
/* All other options (including node-name and driver) must be unchanged.
* Put them back into the QDict, so that they are checked at the end
* of this function. */
qemu_opts_to_qdict(opts, reopen_state->options);
/* If we are to stay read-only, do not allow permission change
* to r/w. Attempting to set to r/w may fail if either BDRV_O_ALLOW_RDWR is
* not set, or if the BDS still has copy_on_read enabled */
read_only = !(reopen_state->flags & BDRV_O_RDWR);
ret = bdrv_can_set_read_only(reopen_state->bs, read_only, true, &local_err);
if (local_err) {
error_propagate(errp, local_err);
goto error;
}
if (drv->bdrv_reopen_prepare) {
/*
* If a driver-specific option is missing, it means that we
* should reset it to its default value.
* But not all options allow that, so we need to check it first.
*/
ret = bdrv_reset_options_allowed(reopen_state->bs,
reopen_state->options, errp);
if (ret) {
goto error;
}
ret = drv->bdrv_reopen_prepare(reopen_state, queue, &local_err);
if (ret) {
if (local_err != NULL) {
error_propagate(errp, local_err);
} else {
bdrv_refresh_filename(reopen_state->bs);
error_setg(errp, "failed while preparing to reopen image '%s'",
reopen_state->bs->filename);
}
goto error;
}
} else {
/* It is currently mandatory to have a bdrv_reopen_prepare()
* handler for each supported drv. */
error_setg(errp, "Block format '%s' used by node '%s' "
"does not support reopening files", drv->format_name,
bdrv_get_device_or_node_name(reopen_state->bs));
ret = -1;
goto error;
}
drv_prepared = true;
/*
* We must provide the 'backing' option if the BDS has a backing
* file or if the image file has a backing file name as part of
* its metadata. Otherwise the 'backing' option can be omitted.
*/
if (drv->supports_backing && reopen_state->backing_missing &&
(reopen_state->bs->backing || reopen_state->bs->backing_file[0])) {
block: Handle child references in bdrv_reopen_queue() Children in QMP are specified with BlockdevRef / BlockdevRefOrNull, which can contain a set of child options, a child reference, or NULL. In optional attributes like "backing" it can also be missing. Only the first case (set of child options) is being handled properly by bdrv_reopen_queue(). This patch deals with all the others. Here's how these cases should be handled when bdrv_reopen_queue() is deciding what to do with each child of a BlockDriverState: 1) Set of child options: if the child was implicitly created (i.e inherits_from points to the parent) then the options are removed from the parent's options QDict and are passed to the child with a recursive bdrv_reopen_queue() call. This case was already working fine. 2) Child reference: there's two possibilites here. 2a) Reference to the current child: if the child was implicitly created then it is put in the reopen queue, keeping its current set of options (since this was a child reference there was no way to specify a different set of options). If the child is not implicit then it keeps its current set of options but it is not reopened (and therefore does not inherit any new option from the parent). 2b) Reference to a different BDS: the current child is not put in the reopen queue at all. Passing a reference to a different BDS can be used to replace a child, although at the moment no driver implements this, so it results in an error. In any case, the current child is not going to be reopened (and might in fact disappear if it's replaced) 3) NULL: This is similar to (2b). Although no driver allows this yet it can be used to detach the current child so it should not be put in the reopen queue. 4) Missing option: at the moment "backing" is the only case where this can happen. With "blockdev-add", leaving "backing" out means that the default backing file is opened. We don't want to open a new image during reopen, so we require that "backing" is always present. We'll relax this requirement a bit in the next patch. If keep_old_opts is true and "backing" is missing then this behaves like 2a (the current child is reopened). Signed-off-by: Alberto Garcia <berto@igalia.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2019-03-12 17:48:45 +01:00
error_setg(errp, "backing is missing for '%s'",
reopen_state->bs->node_name);
ret = -EINVAL;
goto error;
}
/*
* Allow changing the 'backing' option. The new value can be
* either a reference to an existing node (using its node name)
* or NULL to simply detach the current backing file.
*/
ret = bdrv_reopen_parse_file_or_backing(reopen_state, true,
change_child_tran, errp);
if (ret < 0) {
goto error;
}
qdict_del(reopen_state->options, "backing");
/* Allow changing the 'file' option. In this case NULL is not allowed */
ret = bdrv_reopen_parse_file_or_backing(reopen_state, false,
change_child_tran, errp);
if (ret < 0) {
goto error;
}
qdict_del(reopen_state->options, "file");
/* Options that are not handled are only okay if they are unchanged
* compared to the old state. It is expected that some options are only
* used for the initial open, but not reopen (e.g. filename) */
if (qdict_size(reopen_state->options)) {
const QDictEntry *entry = qdict_first(reopen_state->options);
do {
QObject *new = entry->value;
QObject *old = qdict_get(reopen_state->bs->options, entry->key);
/* Allow child references (child_name=node_name) as long as they
* point to the current child (i.e. everything stays the same). */
if (qobject_type(new) == QTYPE_QSTRING) {
BdrvChild *child;
QLIST_FOREACH(child, &reopen_state->bs->children, next) {
if (!strcmp(child->name, entry->key)) {
break;
}
}
if (child) {
if (!strcmp(child->bs->node_name,
qstring_get_str(qobject_to(QString, new)))) {
continue; /* Found child with this name, skip option */
}
}
}
block: Document -drive problematic code and bugs -blockdev and blockdev_add convert their arguments via QObject to BlockdevOptions for qmp_blockdev_add(), which converts them back to QObject, then to a flattened QDict. The QDict's members are typed according to the QAPI schema. -drive converts its argument via QemuOpts to a (flat) QDict. This QDict's members are all QString. Thus, the QType of a flat QDict member depends on whether it comes from -drive or -blockdev/blockdev_add, except when the QAPI type maps to QString, which is the case for 'str' and enumeration types. The block layer core extracts generic configuration from the flat QDict, and the block driver extracts driver-specific configuration. Both commonly do so by converting (parts of) the flat QDict to QemuOpts, which turns all values into strings. Not exactly elegant, but correct. However, A few places access the flat QDict directly: * Most of them access members that are always QString. Correct. * bdrv_open_inherit() accesses a boolean, carefully. Correct. * nfs_config() uses a QObject input visitor. Correct only because the visited type contains nothing but QStrings. * nbd_config() and ssh_config() use a QObject input visitor, and the visited types contain non-QStrings: InetSocketAddress members @numeric, @to, @ipv4, @ipv6. -drive works as long as you don't try to use them (they're all optional). @to is ignored anyway. Reproducer: -drive driver=ssh,server.host=h,server.port=22,server.ipv4,path=p -drive driver=nbd,server.type=inet,server.data.host=h,server.data.port=22,server.data.ipv4 both fail with "Invalid parameter type for 'data.ipv4', expected: boolean" Add suitable comments to all these places. Mark the buggy ones FIXME. "Fortunately", -drive's driver-specific options are entirely undocumented. Signed-off-by: Markus Armbruster <armbru@redhat.com> Message-id: 1490895797-29094-5-git-send-email-armbru@redhat.com [mreitz: Fixed two typos] Reviewed-by: Eric Blake <eblake@redhat.com> Signed-off-by: Max Reitz <mreitz@redhat.com>
2017-03-30 19:43:12 +02:00
/*
* TODO: When using -drive to specify blockdev options, all values
* will be strings; however, when using -blockdev, blockdev-add or
* filenames using the json:{} pseudo-protocol, they will be
* correctly typed.
* In contrast, reopening options are (currently) always strings
* (because you can only specify them through qemu-io; all other
* callers do not specify any options).
* Therefore, when using anything other than -drive to create a BDS,
* this cannot detect non-string options as unchanged, because
* qobject_is_equal() always returns false for objects of different
* type. In the future, this should be remedied by correctly typing
* all options. For now, this is not too big of an issue because
* the user can simply omit options which cannot be changed anyway,
* so they will stay unchanged.
block: Document -drive problematic code and bugs -blockdev and blockdev_add convert their arguments via QObject to BlockdevOptions for qmp_blockdev_add(), which converts them back to QObject, then to a flattened QDict. The QDict's members are typed according to the QAPI schema. -drive converts its argument via QemuOpts to a (flat) QDict. This QDict's members are all QString. Thus, the QType of a flat QDict member depends on whether it comes from -drive or -blockdev/blockdev_add, except when the QAPI type maps to QString, which is the case for 'str' and enumeration types. The block layer core extracts generic configuration from the flat QDict, and the block driver extracts driver-specific configuration. Both commonly do so by converting (parts of) the flat QDict to QemuOpts, which turns all values into strings. Not exactly elegant, but correct. However, A few places access the flat QDict directly: * Most of them access members that are always QString. Correct. * bdrv_open_inherit() accesses a boolean, carefully. Correct. * nfs_config() uses a QObject input visitor. Correct only because the visited type contains nothing but QStrings. * nbd_config() and ssh_config() use a QObject input visitor, and the visited types contain non-QStrings: InetSocketAddress members @numeric, @to, @ipv4, @ipv6. -drive works as long as you don't try to use them (they're all optional). @to is ignored anyway. Reproducer: -drive driver=ssh,server.host=h,server.port=22,server.ipv4,path=p -drive driver=nbd,server.type=inet,server.data.host=h,server.data.port=22,server.data.ipv4 both fail with "Invalid parameter type for 'data.ipv4', expected: boolean" Add suitable comments to all these places. Mark the buggy ones FIXME. "Fortunately", -drive's driver-specific options are entirely undocumented. Signed-off-by: Markus Armbruster <armbru@redhat.com> Message-id: 1490895797-29094-5-git-send-email-armbru@redhat.com [mreitz: Fixed two typos] Reviewed-by: Eric Blake <eblake@redhat.com> Signed-off-by: Max Reitz <mreitz@redhat.com>
2017-03-30 19:43:12 +02:00
*/
if (!qobject_is_equal(new, old)) {
error_setg(errp, "Cannot change the option '%s'", entry->key);
ret = -EINVAL;
goto error;
}
} while ((entry = qdict_next(reopen_state->options, entry)));
}
ret = 0;
/* Restore the original reopen_state->options QDict */
qobject_unref(reopen_state->options);
reopen_state->options = qobject_ref(orig_reopen_opts);
error:
if (ret < 0 && drv_prepared) {
/* drv->bdrv_reopen_prepare() has succeeded, so we need to
* call drv->bdrv_reopen_abort() before signaling an error
* (bdrv_reopen_multiple() will not call bdrv_reopen_abort()
* when the respective bdrv_reopen_prepare() has failed) */
if (drv->bdrv_reopen_abort) {
drv->bdrv_reopen_abort(reopen_state);
}
}
qemu_opts_del(opts);
qobject_unref(orig_reopen_opts);
g_free(discard);
return ret;
}
/*
* Takes the staged changes for the reopen from bdrv_reopen_prepare(), and
* makes them final by swapping the staging BlockDriverState contents into
* the active BlockDriverState contents.
*/
static void bdrv_reopen_commit(BDRVReopenState *reopen_state)
{
BlockDriver *drv;
BlockDriverState *bs;
BdrvChild *child;
assert(reopen_state != NULL);
bs = reopen_state->bs;
drv = bs->drv;
assert(drv != NULL);
GLOBAL_STATE_CODE();
/* If there are any driver level actions to take */
if (drv->bdrv_reopen_commit) {
drv->bdrv_reopen_commit(reopen_state);
}
/* set BDS specific flags now */
qobject_unref(bs->explicit_options);
qobject_unref(bs->options);
qobject_ref(reopen_state->explicit_options);
qobject_ref(reopen_state->options);
bs->explicit_options = reopen_state->explicit_options;
bs->options = reopen_state->options;
bs->open_flags = reopen_state->flags;
bs->detect_zeroes = reopen_state->detect_zeroes;
/* Remove child references from bs->options and bs->explicit_options.
* Child options were already removed in bdrv_reopen_queue_child() */
QLIST_FOREACH(child, &bs->children, next) {
qdict_del(bs->explicit_options, child->name);
qdict_del(bs->options, child->name);
}
/* backing is probably removed, so it's not handled by previous loop */
qdict_del(bs->explicit_options, "backing");
qdict_del(bs->options, "backing");
bdrv_refresh_limits(bs, NULL, NULL);
}
/*
* Abort the reopen, and delete and free the staged changes in
* reopen_state
*/
static void bdrv_reopen_abort(BDRVReopenState *reopen_state)
{
BlockDriver *drv;
assert(reopen_state != NULL);
drv = reopen_state->bs->drv;
assert(drv != NULL);
GLOBAL_STATE_CODE();
if (drv->bdrv_reopen_abort) {
drv->bdrv_reopen_abort(reopen_state);
}
}
static void bdrv_close(BlockDriverState *bs)
{
BdrvAioNotifier *ban, *ban_next;
BdrvChild *child, *next;
GLOBAL_STATE_CODE();
assert(!bs->refcnt);
bdrv_drained_begin(bs); /* complete I/O */
bdrv_flush(bs);
bdrv_drain(bs); /* in case flush left pending I/O */
if (bs->drv) {
if (bs->drv->bdrv_close) {
/* Must unfreeze all children, so bdrv_unref_child() works */
bs->drv->bdrv_close(bs);
}
bs->drv = NULL;
}
QLIST_FOREACH_SAFE(child, &bs->children, next, next) {
bdrv_unref_child(bs, child);
}
bs->backing = NULL;
bs->file = NULL;
g_free(bs->opaque);
bs->opaque = NULL;
qatomic_set(&bs->copy_on_read, 0);
bs->backing_file[0] = '\0';
bs->backing_format[0] = '\0';
bs->total_sectors = 0;
bs->encrypted = false;
bs->sg = false;
qobject_unref(bs->options);
qobject_unref(bs->explicit_options);
bs->options = NULL;
bs->explicit_options = NULL;
qobject_unref(bs->full_open_options);
bs->full_open_options = NULL;
block: block-status cache for data regions As we have attempted before (https://lists.gnu.org/archive/html/qemu-devel/2019-01/msg06451.html, "file-posix: Cache lseek result for data regions"; https://lists.nongnu.org/archive/html/qemu-block/2021-02/msg00934.html, "file-posix: Cache next hole"), this patch seeks to reduce the number of SEEK_DATA/HOLE operations the file-posix driver has to perform. The main difference is that this time it is implemented as part of the general block layer code. The problem we face is that on some filesystems or in some circumstances, SEEK_DATA/HOLE is unreasonably slow. Given the implementation is outside of qemu, there is little we can do about its performance. We have already introduced the want_zero parameter to bdrv_co_block_status() to reduce the number of SEEK_DATA/HOLE calls unless we really want zero information; but sometimes we do want that information, because for files that consist largely of zero areas, special-casing those areas can give large performance boosts. So the real problem is with files that consist largely of data, so that inquiring the block status does not gain us much performance, but where such an inquiry itself takes a lot of time. To address this, we want to cache data regions. Most of the time, when bad performance is reported, it is in places where the image is iterated over from start to end (qemu-img convert or the mirror job), so a simple yet effective solution is to cache only the current data region. (Note that only caching data regions but not zero regions means that returning false information from the cache is not catastrophic: Treating zeroes as data is fine. While we try to invalidate the cache on zero writes and discards, such incongruences may still occur when there are other processes writing to the image.) We only use the cache for nodes without children (i.e. protocol nodes), because that is where the problem is: Drivers that rely on block-status implementations outside of qemu (e.g. SEEK_DATA/HOLE). Resolves: https://gitlab.com/qemu-project/qemu/-/issues/307 Signed-off-by: Hanna Reitz <hreitz@redhat.com> Message-Id: <20210812084148.14458-3-hreitz@redhat.com> Reviewed-by: Eric Blake <eblake@redhat.com> Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> [hreitz: Added `local_file == bs` assertion, as suggested by Vladimir] Signed-off-by: Hanna Reitz <hreitz@redhat.com>
2021-08-12 10:41:44 +02:00
g_free(bs->block_status_cache);
bs->block_status_cache = NULL;
bdrv_release_named_dirty_bitmaps(bs);
assert(QLIST_EMPTY(&bs->dirty_bitmaps));
QLIST_FOREACH_SAFE(ban, &bs->aio_notifiers, list, ban_next) {
g_free(ban);
}
QLIST_INIT(&bs->aio_notifiers);
bdrv_drained_end(bs);
block: End quiescent sections when a BDS is deleted If a BDS gets deleted during blk_drain_all(), it might miss a call to bdrv_do_drained_end(). This means missing a call to aio_enable_external() and the AIO context remains disabled for ever. This can cause a device to become irresponsive and to disrupt the guest execution, ie. hang, loop forever or worse. This scenario is quite easy to encounter with virtio-scsi on POWER when punching multiple blockdev-create QMP commands while the guest is booting and it is still running the SLOF firmware. This happens because SLOF disables/re-enables PCI devices multiple times via IO/MEM/MASTER bits of PCI_COMMAND register after the initial probe/feature negotiation, as it tends to work with a single device at a time at various stages like probing and running block/network bootloaders without doing a full reset in-between. This naturally generates many dataplane stops and starts, and thus many drain sections that can race with blockdev_create_run(). In the end, SLOF bails out. It is somehow reproducible on x86 but it requires to generate articial dataplane start/stop activity with stop/cont QMP commands. In this case, seabios ends up looping for ever, waiting for the virtio-scsi device to send a response to a command it never received. Add a helper that pairs all previously called bdrv_do_drained_begin() with a bdrv_do_drained_end() and call it from bdrv_close(). While at it, update the "/bdrv-drain/graph-change/drain_all" test in test-bdrv-drain so that it can catch the issue. BugId: https://bugzilla.redhat.com/show_bug.cgi?id=1874441 Signed-off-by: Greg Kurz <groug@kaod.org> Message-Id: <160346526998.272601.9045392804399803158.stgit@bahia.lan> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2020-10-23 17:01:10 +02:00
/*
* If we're still inside some bdrv_drain_all_begin()/end() sections, end
* them now since this BDS won't exist anymore when bdrv_drain_all_end()
* gets called.
*/
if (bs->quiesce_counter) {
bdrv_drain_all_end_quiesce(bs);
}
}
void bdrv_close_all(void)
{
assert(job_next(NULL) == NULL);
GLOBAL_STATE_CODE();
/* Drop references from requests still in flight, such as canceled block
* jobs whose AIO context has not been polled yet */
bdrv_drain_all();
blk_remove_all_bs();
blockdev_close_all_bdrv_states();
assert(QTAILQ_EMPTY(&all_bdrv_states));
}
static bool should_update_child(BdrvChild *c, BlockDriverState *to)
{
GQueue *queue;
GHashTable *found;
bool ret;
if (c->klass->stay_at_node) {
return false;
}
/* If the child @c belongs to the BDS @to, replacing the current
* c->bs by @to would mean to create a loop.
*
* Such a case occurs when appending a BDS to a backing chain.
* For instance, imagine the following chain:
*
* guest device -> node A -> further backing chain...
*
* Now we create a new BDS B which we want to put on top of this
* chain, so we first attach A as its backing node:
*
* node B
* |
* v
* guest device -> node A -> further backing chain...
*
* Finally we want to replace A by B. When doing that, we want to
* replace all pointers to A by pointers to B -- except for the
* pointer from B because (1) that would create a loop, and (2)
* that pointer should simply stay intact:
*
* guest device -> node B
* |
* v
* node A -> further backing chain...
*
* In general, when replacing a node A (c->bs) by a node B (@to),
* if A is a child of B, that means we cannot replace A by B there
* because that would create a loop. Silently detaching A from B
* is also not really an option. So overall just leaving A in
* place there is the most sensible choice.
*
* We would also create a loop in any cases where @c is only
* indirectly referenced by @to. Prevent this by returning false
* if @c is found (by breadth-first search) anywhere in the whole
* subtree of @to.
*/
ret = true;
found = g_hash_table_new(NULL, NULL);
g_hash_table_add(found, to);
queue = g_queue_new();
g_queue_push_tail(queue, to);
while (!g_queue_is_empty(queue)) {
BlockDriverState *v = g_queue_pop_head(queue);
BdrvChild *c2;
QLIST_FOREACH(c2, &v->children, next) {
if (c2 == c) {
ret = false;
break;
}
if (g_hash_table_contains(found, c2->bs)) {
continue;
}
g_queue_push_tail(queue, c2->bs);
g_hash_table_add(found, c2->bs);
}
}
g_queue_free(queue);
g_hash_table_destroy(found);
return ret;
}
typedef struct BdrvRemoveFilterOrCowChild {
BdrvChild *child;
block: Let replace_child_tran keep indirect pointer As of a future commit, bdrv_replace_child_noperm() will clear the indirect BdrvChild pointer passed to it if the new child BDS is NULL. bdrv_replace_child_tran() will want to let it do that, but revert this change in its abort handler. For that, we need to have it receive a BdrvChild ** pointer, too, and keep it stored in the BdrvReplaceChildState object that we attach to the transaction. Note that we do not need to store it in the BdrvReplaceChildState when new_bs is not NULL, because then there is nothing to revert. This is important so that bdrv_replace_node_noperm() can pass a pointer to a loop-local variable to bdrv_replace_child_tran() without worrying that this pointer will outlive one loop iteration. (Of course, for that to work, bdrv_replace_node_noperm() and in turn bdrv_replace_node() and its relatives may not be called with a NULL @to node. Luckily, they already are not, but now we should assert this.) bdrv_remove_file_or_backing_child() on the other hand needs to ensure that the indirect pointer it passes will stay valid for the duration of the transaction. Ensure this by keeping a strong reference to the BDS whose &bs->backing or &bs->file it passes to bdrv_replace_child_tran(), and giving up that reference only in the transaction .clean() handler. Signed-off-by: Hanna Reitz <hreitz@redhat.com> Message-Id: <20211111120829.81329-9-hreitz@redhat.com> Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> Message-Id: <20211115145409.176785-9-kwolf@redhat.com> Signed-off-by: Hanna Reitz <hreitz@redhat.com>
2021-11-15 15:54:04 +01:00
BlockDriverState *bs;
bool is_backing;
} BdrvRemoveFilterOrCowChild;
static void bdrv_remove_filter_or_cow_child_abort(void *opaque)
{
BdrvRemoveFilterOrCowChild *s = opaque;
BlockDriverState *parent_bs = s->child->opaque;
if (s->is_backing) {
parent_bs->backing = s->child;
} else {
parent_bs->file = s->child;
}
/*
* We don't have to restore child->bs here to undo bdrv_replace_child_tran()
* because that function is transactionable and it registered own completion
* entries in @tran, so .abort() for bdrv_replace_child_safe() will be
* called automatically.
*/
}
static void bdrv_remove_filter_or_cow_child_commit(void *opaque)
{
BdrvRemoveFilterOrCowChild *s = opaque;
GLOBAL_STATE_CODE();
bdrv_child_free(s->child);
}
block: Let replace_child_tran keep indirect pointer As of a future commit, bdrv_replace_child_noperm() will clear the indirect BdrvChild pointer passed to it if the new child BDS is NULL. bdrv_replace_child_tran() will want to let it do that, but revert this change in its abort handler. For that, we need to have it receive a BdrvChild ** pointer, too, and keep it stored in the BdrvReplaceChildState object that we attach to the transaction. Note that we do not need to store it in the BdrvReplaceChildState when new_bs is not NULL, because then there is nothing to revert. This is important so that bdrv_replace_node_noperm() can pass a pointer to a loop-local variable to bdrv_replace_child_tran() without worrying that this pointer will outlive one loop iteration. (Of course, for that to work, bdrv_replace_node_noperm() and in turn bdrv_replace_node() and its relatives may not be called with a NULL @to node. Luckily, they already are not, but now we should assert this.) bdrv_remove_file_or_backing_child() on the other hand needs to ensure that the indirect pointer it passes will stay valid for the duration of the transaction. Ensure this by keeping a strong reference to the BDS whose &bs->backing or &bs->file it passes to bdrv_replace_child_tran(), and giving up that reference only in the transaction .clean() handler. Signed-off-by: Hanna Reitz <hreitz@redhat.com> Message-Id: <20211111120829.81329-9-hreitz@redhat.com> Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> Message-Id: <20211115145409.176785-9-kwolf@redhat.com> Signed-off-by: Hanna Reitz <hreitz@redhat.com>
2021-11-15 15:54:04 +01:00
static void bdrv_remove_filter_or_cow_child_clean(void *opaque)
{
BdrvRemoveFilterOrCowChild *s = opaque;
/* Drop the bs reference after the transaction is done */
bdrv_unref(s->bs);
g_free(s);
}
static TransactionActionDrv bdrv_remove_filter_or_cow_child_drv = {
.abort = bdrv_remove_filter_or_cow_child_abort,
.commit = bdrv_remove_filter_or_cow_child_commit,
block: Let replace_child_tran keep indirect pointer As of a future commit, bdrv_replace_child_noperm() will clear the indirect BdrvChild pointer passed to it if the new child BDS is NULL. bdrv_replace_child_tran() will want to let it do that, but revert this change in its abort handler. For that, we need to have it receive a BdrvChild ** pointer, too, and keep it stored in the BdrvReplaceChildState object that we attach to the transaction. Note that we do not need to store it in the BdrvReplaceChildState when new_bs is not NULL, because then there is nothing to revert. This is important so that bdrv_replace_node_noperm() can pass a pointer to a loop-local variable to bdrv_replace_child_tran() without worrying that this pointer will outlive one loop iteration. (Of course, for that to work, bdrv_replace_node_noperm() and in turn bdrv_replace_node() and its relatives may not be called with a NULL @to node. Luckily, they already are not, but now we should assert this.) bdrv_remove_file_or_backing_child() on the other hand needs to ensure that the indirect pointer it passes will stay valid for the duration of the transaction. Ensure this by keeping a strong reference to the BDS whose &bs->backing or &bs->file it passes to bdrv_replace_child_tran(), and giving up that reference only in the transaction .clean() handler. Signed-off-by: Hanna Reitz <hreitz@redhat.com> Message-Id: <20211111120829.81329-9-hreitz@redhat.com> Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> Message-Id: <20211115145409.176785-9-kwolf@redhat.com> Signed-off-by: Hanna Reitz <hreitz@redhat.com>
2021-11-15 15:54:04 +01:00
.clean = bdrv_remove_filter_or_cow_child_clean,
};
/*
* A function to remove backing or file child of @bs.
* Function doesn't update permissions, caller is responsible for this.
*/
static void bdrv_remove_file_or_backing_child(BlockDriverState *bs,
BdrvChild *child,
Transaction *tran)
{
BdrvChild **childp;
BdrvRemoveFilterOrCowChild *s;
if (!child) {
return;
}
block: Let replace_child_tran keep indirect pointer As of a future commit, bdrv_replace_child_noperm() will clear the indirect BdrvChild pointer passed to it if the new child BDS is NULL. bdrv_replace_child_tran() will want to let it do that, but revert this change in its abort handler. For that, we need to have it receive a BdrvChild ** pointer, too, and keep it stored in the BdrvReplaceChildState object that we attach to the transaction. Note that we do not need to store it in the BdrvReplaceChildState when new_bs is not NULL, because then there is nothing to revert. This is important so that bdrv_replace_node_noperm() can pass a pointer to a loop-local variable to bdrv_replace_child_tran() without worrying that this pointer will outlive one loop iteration. (Of course, for that to work, bdrv_replace_node_noperm() and in turn bdrv_replace_node() and its relatives may not be called with a NULL @to node. Luckily, they already are not, but now we should assert this.) bdrv_remove_file_or_backing_child() on the other hand needs to ensure that the indirect pointer it passes will stay valid for the duration of the transaction. Ensure this by keeping a strong reference to the BDS whose &bs->backing or &bs->file it passes to bdrv_replace_child_tran(), and giving up that reference only in the transaction .clean() handler. Signed-off-by: Hanna Reitz <hreitz@redhat.com> Message-Id: <20211111120829.81329-9-hreitz@redhat.com> Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> Message-Id: <20211115145409.176785-9-kwolf@redhat.com> Signed-off-by: Hanna Reitz <hreitz@redhat.com>
2021-11-15 15:54:04 +01:00
/*
* Keep a reference to @bs so @childp will stay valid throughout the
* transaction (required by bdrv_replace_child_tran())
*/
bdrv_ref(bs);
if (child == bs->backing) {
childp = &bs->backing;
} else if (child == bs->file) {
childp = &bs->file;
} else {
g_assert_not_reached();
}
if (child->bs) {
/*
* Pass free_empty_child=false, we will free the child in
* bdrv_remove_filter_or_cow_child_commit()
*/
bdrv_replace_child_tran(childp, NULL, tran, false);
}
s = g_new(BdrvRemoveFilterOrCowChild, 1);
*s = (BdrvRemoveFilterOrCowChild) {
.child = child,
block: Let replace_child_tran keep indirect pointer As of a future commit, bdrv_replace_child_noperm() will clear the indirect BdrvChild pointer passed to it if the new child BDS is NULL. bdrv_replace_child_tran() will want to let it do that, but revert this change in its abort handler. For that, we need to have it receive a BdrvChild ** pointer, too, and keep it stored in the BdrvReplaceChildState object that we attach to the transaction. Note that we do not need to store it in the BdrvReplaceChildState when new_bs is not NULL, because then there is nothing to revert. This is important so that bdrv_replace_node_noperm() can pass a pointer to a loop-local variable to bdrv_replace_child_tran() without worrying that this pointer will outlive one loop iteration. (Of course, for that to work, bdrv_replace_node_noperm() and in turn bdrv_replace_node() and its relatives may not be called with a NULL @to node. Luckily, they already are not, but now we should assert this.) bdrv_remove_file_or_backing_child() on the other hand needs to ensure that the indirect pointer it passes will stay valid for the duration of the transaction. Ensure this by keeping a strong reference to the BDS whose &bs->backing or &bs->file it passes to bdrv_replace_child_tran(), and giving up that reference only in the transaction .clean() handler. Signed-off-by: Hanna Reitz <hreitz@redhat.com> Message-Id: <20211111120829.81329-9-hreitz@redhat.com> Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> Message-Id: <20211115145409.176785-9-kwolf@redhat.com> Signed-off-by: Hanna Reitz <hreitz@redhat.com>
2021-11-15 15:54:04 +01:00
.bs = bs,
.is_backing = (childp == &bs->backing),
};
tran_add(tran, &bdrv_remove_filter_or_cow_child_drv, s);
}
/*
* A function to remove backing-chain child of @bs if exists: cow child for
* format nodes (always .backing) and filter child for filters (may be .file or
* .backing)
*/
static void bdrv_remove_filter_or_cow_child(BlockDriverState *bs,
Transaction *tran)
{
bdrv_remove_file_or_backing_child(bs, bdrv_filter_or_cow_child(bs), tran);
}
static int bdrv_replace_node_noperm(BlockDriverState *from,
BlockDriverState *to,
bool auto_skip, Transaction *tran,
Error **errp)
{
BdrvChild *c, *next;
block: Let replace_child_tran keep indirect pointer As of a future commit, bdrv_replace_child_noperm() will clear the indirect BdrvChild pointer passed to it if the new child BDS is NULL. bdrv_replace_child_tran() will want to let it do that, but revert this change in its abort handler. For that, we need to have it receive a BdrvChild ** pointer, too, and keep it stored in the BdrvReplaceChildState object that we attach to the transaction. Note that we do not need to store it in the BdrvReplaceChildState when new_bs is not NULL, because then there is nothing to revert. This is important so that bdrv_replace_node_noperm() can pass a pointer to a loop-local variable to bdrv_replace_child_tran() without worrying that this pointer will outlive one loop iteration. (Of course, for that to work, bdrv_replace_node_noperm() and in turn bdrv_replace_node() and its relatives may not be called with a NULL @to node. Luckily, they already are not, but now we should assert this.) bdrv_remove_file_or_backing_child() on the other hand needs to ensure that the indirect pointer it passes will stay valid for the duration of the transaction. Ensure this by keeping a strong reference to the BDS whose &bs->backing or &bs->file it passes to bdrv_replace_child_tran(), and giving up that reference only in the transaction .clean() handler. Signed-off-by: Hanna Reitz <hreitz@redhat.com> Message-Id: <20211111120829.81329-9-hreitz@redhat.com> Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> Message-Id: <20211115145409.176785-9-kwolf@redhat.com> Signed-off-by: Hanna Reitz <hreitz@redhat.com>
2021-11-15 15:54:04 +01:00
assert(to != NULL);
GLOBAL_STATE_CODE();
block: Let replace_child_tran keep indirect pointer As of a future commit, bdrv_replace_child_noperm() will clear the indirect BdrvChild pointer passed to it if the new child BDS is NULL. bdrv_replace_child_tran() will want to let it do that, but revert this change in its abort handler. For that, we need to have it receive a BdrvChild ** pointer, too, and keep it stored in the BdrvReplaceChildState object that we attach to the transaction. Note that we do not need to store it in the BdrvReplaceChildState when new_bs is not NULL, because then there is nothing to revert. This is important so that bdrv_replace_node_noperm() can pass a pointer to a loop-local variable to bdrv_replace_child_tran() without worrying that this pointer will outlive one loop iteration. (Of course, for that to work, bdrv_replace_node_noperm() and in turn bdrv_replace_node() and its relatives may not be called with a NULL @to node. Luckily, they already are not, but now we should assert this.) bdrv_remove_file_or_backing_child() on the other hand needs to ensure that the indirect pointer it passes will stay valid for the duration of the transaction. Ensure this by keeping a strong reference to the BDS whose &bs->backing or &bs->file it passes to bdrv_replace_child_tran(), and giving up that reference only in the transaction .clean() handler. Signed-off-by: Hanna Reitz <hreitz@redhat.com> Message-Id: <20211111120829.81329-9-hreitz@redhat.com> Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> Message-Id: <20211115145409.176785-9-kwolf@redhat.com> Signed-off-by: Hanna Reitz <hreitz@redhat.com>
2021-11-15 15:54:04 +01:00
QLIST_FOREACH_SAFE(c, &from->parents, next_parent, next) {
assert(c->bs == from);
if (!should_update_child(c, to)) {
if (auto_skip) {
continue;
}
error_setg(errp, "Should not change '%s' link to '%s'",
c->name, from->node_name);
return -EINVAL;
}
if (c->frozen) {
error_setg(errp, "Cannot change '%s' link to '%s'",
c->name, from->node_name);
return -EPERM;
}
block: Let replace_child_tran keep indirect pointer As of a future commit, bdrv_replace_child_noperm() will clear the indirect BdrvChild pointer passed to it if the new child BDS is NULL. bdrv_replace_child_tran() will want to let it do that, but revert this change in its abort handler. For that, we need to have it receive a BdrvChild ** pointer, too, and keep it stored in the BdrvReplaceChildState object that we attach to the transaction. Note that we do not need to store it in the BdrvReplaceChildState when new_bs is not NULL, because then there is nothing to revert. This is important so that bdrv_replace_node_noperm() can pass a pointer to a loop-local variable to bdrv_replace_child_tran() without worrying that this pointer will outlive one loop iteration. (Of course, for that to work, bdrv_replace_node_noperm() and in turn bdrv_replace_node() and its relatives may not be called with a NULL @to node. Luckily, they already are not, but now we should assert this.) bdrv_remove_file_or_backing_child() on the other hand needs to ensure that the indirect pointer it passes will stay valid for the duration of the transaction. Ensure this by keeping a strong reference to the BDS whose &bs->backing or &bs->file it passes to bdrv_replace_child_tran(), and giving up that reference only in the transaction .clean() handler. Signed-off-by: Hanna Reitz <hreitz@redhat.com> Message-Id: <20211111120829.81329-9-hreitz@redhat.com> Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> Message-Id: <20211115145409.176785-9-kwolf@redhat.com> Signed-off-by: Hanna Reitz <hreitz@redhat.com>
2021-11-15 15:54:04 +01:00
/*
* Passing a pointer to the local variable @c is fine here, because
* @to is not NULL, and so &c will not be attached to the transaction.
*/
bdrv_replace_child_tran(&c, to, tran, true);
}
return 0;
}
/*
* With auto_skip=true bdrv_replace_node_common skips updating from parents
* if it creates a parent-child relation loop or if parent is block-job.
*
* With auto_skip=false the error is returned if from has a parent which should
* not be updated.
*
* With @detach_subchain=true @to must be in a backing chain of @from. In this
* case backing link of the cow-parent of @to is removed.
block: Let replace_child_tran keep indirect pointer As of a future commit, bdrv_replace_child_noperm() will clear the indirect BdrvChild pointer passed to it if the new child BDS is NULL. bdrv_replace_child_tran() will want to let it do that, but revert this change in its abort handler. For that, we need to have it receive a BdrvChild ** pointer, too, and keep it stored in the BdrvReplaceChildState object that we attach to the transaction. Note that we do not need to store it in the BdrvReplaceChildState when new_bs is not NULL, because then there is nothing to revert. This is important so that bdrv_replace_node_noperm() can pass a pointer to a loop-local variable to bdrv_replace_child_tran() without worrying that this pointer will outlive one loop iteration. (Of course, for that to work, bdrv_replace_node_noperm() and in turn bdrv_replace_node() and its relatives may not be called with a NULL @to node. Luckily, they already are not, but now we should assert this.) bdrv_remove_file_or_backing_child() on the other hand needs to ensure that the indirect pointer it passes will stay valid for the duration of the transaction. Ensure this by keeping a strong reference to the BDS whose &bs->backing or &bs->file it passes to bdrv_replace_child_tran(), and giving up that reference only in the transaction .clean() handler. Signed-off-by: Hanna Reitz <hreitz@redhat.com> Message-Id: <20211111120829.81329-9-hreitz@redhat.com> Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> Message-Id: <20211115145409.176785-9-kwolf@redhat.com> Signed-off-by: Hanna Reitz <hreitz@redhat.com>
2021-11-15 15:54:04 +01:00
*
* @to must not be NULL.
*/
static int bdrv_replace_node_common(BlockDriverState *from,
BlockDriverState *to,
bool auto_skip, bool detach_subchain,
Error **errp)
{
Transaction *tran = tran_new();
g_autoptr(GHashTable) found = NULL;
g_autoptr(GSList) refresh_list = NULL;
BlockDriverState *to_cow_parent = NULL;
int ret;
GLOBAL_STATE_CODE();
block: Let replace_child_tran keep indirect pointer As of a future commit, bdrv_replace_child_noperm() will clear the indirect BdrvChild pointer passed to it if the new child BDS is NULL. bdrv_replace_child_tran() will want to let it do that, but revert this change in its abort handler. For that, we need to have it receive a BdrvChild ** pointer, too, and keep it stored in the BdrvReplaceChildState object that we attach to the transaction. Note that we do not need to store it in the BdrvReplaceChildState when new_bs is not NULL, because then there is nothing to revert. This is important so that bdrv_replace_node_noperm() can pass a pointer to a loop-local variable to bdrv_replace_child_tran() without worrying that this pointer will outlive one loop iteration. (Of course, for that to work, bdrv_replace_node_noperm() and in turn bdrv_replace_node() and its relatives may not be called with a NULL @to node. Luckily, they already are not, but now we should assert this.) bdrv_remove_file_or_backing_child() on the other hand needs to ensure that the indirect pointer it passes will stay valid for the duration of the transaction. Ensure this by keeping a strong reference to the BDS whose &bs->backing or &bs->file it passes to bdrv_replace_child_tran(), and giving up that reference only in the transaction .clean() handler. Signed-off-by: Hanna Reitz <hreitz@redhat.com> Message-Id: <20211111120829.81329-9-hreitz@redhat.com> Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> Message-Id: <20211115145409.176785-9-kwolf@redhat.com> Signed-off-by: Hanna Reitz <hreitz@redhat.com>
2021-11-15 15:54:04 +01:00
assert(to != NULL);
if (detach_subchain) {
assert(bdrv_chain_contains(from, to));
assert(from != to);
for (to_cow_parent = from;
bdrv_filter_or_cow_bs(to_cow_parent) != to;
to_cow_parent = bdrv_filter_or_cow_bs(to_cow_parent))
{
;
}
}
/* Make sure that @from doesn't go away until we have successfully attached
* all of its parents to @to. */
bdrv_ref(from);
assert(qemu_get_current_aio_context() == qemu_get_aio_context());
assert(bdrv_get_aio_context(from) == bdrv_get_aio_context(to));
bdrv_drained_begin(from);
/*
* Do the replacement without permission update.
* Replacement may influence the permissions, we should calculate new
* permissions based on new graph. If we fail, we'll roll-back the
* replacement.
*/
ret = bdrv_replace_node_noperm(from, to, auto_skip, tran, errp);
if (ret < 0) {
goto out;
}
if (detach_subchain) {
bdrv_remove_filter_or_cow_child(to_cow_parent, tran);
}
found = g_hash_table_new(NULL, NULL);
refresh_list = bdrv_topological_dfs(refresh_list, found, to);
refresh_list = bdrv_topological_dfs(refresh_list, found, from);
ret = bdrv_list_refresh_perms(refresh_list, NULL, tran, errp);
if (ret < 0) {
goto out;
}
ret = 0;
out:
tran_finalize(tran, ret);
bdrv_drained_end(from);
bdrv_unref(from);
return ret;
}
block: Let replace_child_tran keep indirect pointer As of a future commit, bdrv_replace_child_noperm() will clear the indirect BdrvChild pointer passed to it if the new child BDS is NULL. bdrv_replace_child_tran() will want to let it do that, but revert this change in its abort handler. For that, we need to have it receive a BdrvChild ** pointer, too, and keep it stored in the BdrvReplaceChildState object that we attach to the transaction. Note that we do not need to store it in the BdrvReplaceChildState when new_bs is not NULL, because then there is nothing to revert. This is important so that bdrv_replace_node_noperm() can pass a pointer to a loop-local variable to bdrv_replace_child_tran() without worrying that this pointer will outlive one loop iteration. (Of course, for that to work, bdrv_replace_node_noperm() and in turn bdrv_replace_node() and its relatives may not be called with a NULL @to node. Luckily, they already are not, but now we should assert this.) bdrv_remove_file_or_backing_child() on the other hand needs to ensure that the indirect pointer it passes will stay valid for the duration of the transaction. Ensure this by keeping a strong reference to the BDS whose &bs->backing or &bs->file it passes to bdrv_replace_child_tran(), and giving up that reference only in the transaction .clean() handler. Signed-off-by: Hanna Reitz <hreitz@redhat.com> Message-Id: <20211111120829.81329-9-hreitz@redhat.com> Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> Message-Id: <20211115145409.176785-9-kwolf@redhat.com> Signed-off-by: Hanna Reitz <hreitz@redhat.com>
2021-11-15 15:54:04 +01:00
/**
* Replace node @from by @to (where neither may be NULL).
*/
int bdrv_replace_node(BlockDriverState *from, BlockDriverState *to,
Error **errp)
{
GLOBAL_STATE_CODE();
return bdrv_replace_node_common(from, to, true, false, errp);
}
int bdrv_drop_filter(BlockDriverState *bs, Error **errp)
{
GLOBAL_STATE_CODE();
return bdrv_replace_node_common(bs, bdrv_filter_or_cow_bs(bs), true, true,
errp);
}
/*
* Add new bs contents at the top of an image chain while the chain is
* live, while keeping required fields on the top layer.
*
* This will modify the BlockDriverState fields, and swap contents
* between bs_new and bs_top. Both bs_new and bs_top are modified.
*
* bs_new must not be attached to a BlockBackend and must not have backing
* child.
*
* This function does not create any image files.
*/
int bdrv_append(BlockDriverState *bs_new, BlockDriverState *bs_top,
Error **errp)
{
int ret;
Transaction *tran = tran_new();
GLOBAL_STATE_CODE();
assert(!bs_new->backing);
ret = bdrv_attach_child_noperm(bs_new, bs_top, "backing",
&child_of_bds, bdrv_backing_role(bs_new),
&bs_new->backing, tran, errp);
if (ret < 0) {
goto out;
}
ret = bdrv_replace_node_noperm(bs_top, bs_new, true, tran, errp);
if (ret < 0) {
goto out;
}
ret = bdrv_refresh_perms(bs_new, errp);
out:
tran_finalize(tran, ret);
bdrv_refresh_limits(bs_top, NULL, NULL);
return ret;
}
/* Not for empty child */
int bdrv_replace_child_bs(BdrvChild *child, BlockDriverState *new_bs,
Error **errp)
{
int ret;
Transaction *tran = tran_new();
g_autoptr(GHashTable) found = NULL;
g_autoptr(GSList) refresh_list = NULL;
BlockDriverState *old_bs = child->bs;
GLOBAL_STATE_CODE();
bdrv_ref(old_bs);
bdrv_drained_begin(old_bs);
bdrv_drained_begin(new_bs);
bdrv_replace_child_tran(&child, new_bs, tran, true);
/* @new_bs must have been non-NULL, so @child must not have been freed */
assert(child != NULL);
found = g_hash_table_new(NULL, NULL);
refresh_list = bdrv_topological_dfs(refresh_list, found, old_bs);
refresh_list = bdrv_topological_dfs(refresh_list, found, new_bs);
ret = bdrv_list_refresh_perms(refresh_list, NULL, tran, errp);
tran_finalize(tran, ret);
bdrv_drained_end(old_bs);
bdrv_drained_end(new_bs);
bdrv_unref(old_bs);
return ret;
}
static void bdrv_delete(BlockDriverState *bs)
{
assert(bdrv_op_blocker_is_empty(bs));
assert(!bs->refcnt);
GLOBAL_STATE_CODE();
/* remove from list, if necessary */
if (bs->node_name[0] != '\0') {
QTAILQ_REMOVE(&graph_bdrv_states, bs, node_list);
}
QTAILQ_REMOVE(&all_bdrv_states, bs, bs_list);
bdrv_close(bs);
g_free(bs);
}
/*
* Replace @bs by newly created block node.
*
* @options is a QDict of options to pass to the block drivers, or NULL for an
* empty set of options. The reference to the QDict belongs to the block layer
* after the call (even on failure), so if the caller intends to reuse the
* dictionary, it needs to use qobject_ref() before calling bdrv_open.
*/
BlockDriverState *bdrv_insert_node(BlockDriverState *bs, QDict *options,
int flags, Error **errp)
{
ERRP_GUARD();
int ret;
BlockDriverState *new_node_bs = NULL;
const char *drvname, *node_name;
BlockDriver *drv;
drvname = qdict_get_try_str(options, "driver");
if (!drvname) {
error_setg(errp, "driver is not specified");
goto fail;
}
drv = bdrv_find_format(drvname);
if (!drv) {
error_setg(errp, "Unknown driver: '%s'", drvname);
goto fail;
}
node_name = qdict_get_try_str(options, "node-name");
GLOBAL_STATE_CODE();
new_node_bs = bdrv_new_open_driver_opts(drv, node_name, options, flags,
errp);
options = NULL; /* bdrv_new_open_driver() eats options */
if (!new_node_bs) {
error_prepend(errp, "Could not create node: ");
goto fail;
}
bdrv_drained_begin(bs);
ret = bdrv_replace_node(bs, new_node_bs, errp);
bdrv_drained_end(bs);
if (ret < 0) {
error_prepend(errp, "Could not replace node: ");
goto fail;
}
return new_node_bs;
fail:
qobject_unref(options);
bdrv_unref(new_node_bs);
return NULL;
}
/*
* Run consistency checks on an image
*
* Returns 0 if the check could be completed (it doesn't mean that the image is
* free of errors) or -errno when an internal error occurred. The results of the
* check are stored in res.
*/
int coroutine_fn bdrv_co_check(BlockDriverState *bs,
BdrvCheckResult *res, BdrvCheckMode fix)
{
IO_CODE();
if (bs->drv == NULL) {
return -ENOMEDIUM;
}
if (bs->drv->bdrv_co_check == NULL) {
return -ENOTSUP;
}
memset(res, 0, sizeof(*res));
return bs->drv->bdrv_co_check(bs, res, fix);
}
/*
* Return values:
* 0 - success
* -EINVAL - backing format specified, but no file
* -ENOSPC - can't update the backing file because no space is left in the
* image file header
* -ENOTSUP - format driver doesn't support changing the backing file
*/
int bdrv_change_backing_file(BlockDriverState *bs, const char *backing_file,
const char *backing_fmt, bool require)
{
BlockDriver *drv = bs->drv;
int ret;
GLOBAL_STATE_CODE();
if (!drv) {
return -ENOMEDIUM;
}
/* Backing file format doesn't make sense without a backing file */
if (backing_fmt && !backing_file) {
return -EINVAL;
}
if (require && backing_file && !backing_fmt) {
return -EINVAL;
}
if (drv->bdrv_change_backing_file != NULL) {
ret = drv->bdrv_change_backing_file(bs, backing_file, backing_fmt);
} else {
ret = -ENOTSUP;
}
if (ret == 0) {
pstrcpy(bs->backing_file, sizeof(bs->backing_file), backing_file ?: "");
pstrcpy(bs->backing_format, sizeof(bs->backing_format), backing_fmt ?: "");
block: Add BDS.auto_backing_file If the backing file is overridden, this most probably does change the guest-visible data of a BDS. Therefore, we will need to consider this in bdrv_refresh_filename(). To see whether it has been overridden, we might want to compare bs->backing_file and bs->backing->bs->filename. However, bs->backing_file is changed by bdrv_set_backing_hd() (which is just used to change the backing child at runtime, without modifying the image header), so bs->backing_file most of the time simply contains a copy of bs->backing->bs->filename anyway, so it is useless for such a comparison. This patch adds an auto_backing_file BDS field which contains the backing file path as indicated by the image header, which is not changed by bdrv_set_backing_hd(). Because of bdrv_refresh_filename() magic, however, a BDS's filename may differ from what has been specified during bdrv_open(). Then, the comparison between bs->auto_backing_file and bs->backing->bs->filename may fail even though bs->backing was opened from bs->auto_backing_file. To mitigate this, we can copy the real BDS's filename (after the whole bdrv_open() and bdrv_refresh_filename() process) into bs->auto_backing_file, if we know the former has been opened based on the latter. This is only possible if no options modifying the backing file's behavior have been specified, though. To simplify things, this patch only copies the filename from the backing file if no options have been specified for it at all. Furthermore, there are cases where an overlay is created by qemu which already contains a BDS's filename (e.g. in blockdev-snapshot-sync). We do not need to worry about updating the overlay's bs->auto_backing_file there, because we actually wrote a post-bdrv_refresh_filename() filename into the image header. So all in all, there will be false negatives where (as of a future patch) bdrv_refresh_filename() will assume that the backing file differs from what was specified in the image header, even though it really does not. However, these cases should be limited to where (1) the user actually did override something in the backing chain (e.g. by specifying options for the backing file), or (2) the user executed a QMP command to change some node's backing file (e.g. change-backing-file or block-commit with @backing-file given) where the given filename does not happen to coincide with qemu's idea of the backing BDS's filename. Then again, (1) really is limited to -drive. With -blockdev or blockdev-add, you have to adhere to the schema, so a user cannot give partial "unimportant" options (e.g. by just setting backing.node-name and leaving the rest to the image header). Therefore, trying to fix this would mean trying to fix something for -drive only. To improve on (2), we would need a full infrastructure to "canonicalize" an arbitrary filename (+ options), so it can be compared against another. That seems a bit over the top, considering that filenames nowadays are there mostly for the user's entertainment. Signed-off-by: Max Reitz <mreitz@redhat.com> Reviewed-by: Eric Blake <eblake@redhat.com> Reviewed-by: Alberto Garcia <berto@igalia.com> Message-id: 20190201192935.18394-5-mreitz@redhat.com Signed-off-by: Max Reitz <mreitz@redhat.com>
2019-02-01 20:29:08 +01:00
pstrcpy(bs->auto_backing_file, sizeof(bs->auto_backing_file),
backing_file ?: "");
}
return ret;
}
/*
* Finds the first non-filter node above bs in the chain between
* active and bs. The returned node is either an immediate parent of
* bs, or there are only filter nodes between the two.
*
* Returns NULL if bs is not found in active's image chain,
* or if active == bs.
*
* Returns the bottommost base image if bs == NULL.
*/
BlockDriverState *bdrv_find_overlay(BlockDriverState *active,
BlockDriverState *bs)
{
GLOBAL_STATE_CODE();
bs = bdrv_skip_filters(bs);
active = bdrv_skip_filters(active);
while (active) {
BlockDriverState *next = bdrv_backing_chain_next(active);
if (bs == next) {
return active;
}
active = next;
}
return NULL;
}
/* Given a BDS, searches for the base layer. */
BlockDriverState *bdrv_find_base(BlockDriverState *bs)
{
GLOBAL_STATE_CODE();
return bdrv_find_overlay(bs, NULL);
}
/*
* Return true if at least one of the COW (backing) and filter links
* between @bs and @base is frozen. @errp is set if that's the case.
* @base must be reachable from @bs, or NULL.
*/
bool bdrv_is_backing_chain_frozen(BlockDriverState *bs, BlockDriverState *base,
Error **errp)
{
BlockDriverState *i;
BdrvChild *child;
GLOBAL_STATE_CODE();
for (i = bs; i != base; i = child_bs(child)) {
child = bdrv_filter_or_cow_child(i);
if (child && child->frozen) {
error_setg(errp, "Cannot change '%s' link from '%s' to '%s'",
child->name, i->node_name, child->bs->node_name);
return true;
}
}
return false;
}
/*
* Freeze all COW (backing) and filter links between @bs and @base.
* If any of the links is already frozen the operation is aborted and
* none of the links are modified.
* @base must be reachable from @bs, or NULL.
* Returns 0 on success. On failure returns < 0 and sets @errp.
*/
int bdrv_freeze_backing_chain(BlockDriverState *bs, BlockDriverState *base,
Error **errp)
{
BlockDriverState *i;
BdrvChild *child;
GLOBAL_STATE_CODE();
if (bdrv_is_backing_chain_frozen(bs, base, errp)) {
return -EPERM;
}
for (i = bs; i != base; i = child_bs(child)) {
child = bdrv_filter_or_cow_child(i);
if (child && child->bs->never_freeze) {
error_setg(errp, "Cannot freeze '%s' link to '%s'",
child->name, child->bs->node_name);
return -EPERM;
}
}
for (i = bs; i != base; i = child_bs(child)) {
child = bdrv_filter_or_cow_child(i);
if (child) {
child->frozen = true;
}
}
return 0;
}
/*
* Unfreeze all COW (backing) and filter links between @bs and @base.
* The caller must ensure that all links are frozen before using this
* function.
* @base must be reachable from @bs, or NULL.
*/
void bdrv_unfreeze_backing_chain(BlockDriverState *bs, BlockDriverState *base)
{
BlockDriverState *i;
BdrvChild *child;
GLOBAL_STATE_CODE();
for (i = bs; i != base; i = child_bs(child)) {
child = bdrv_filter_or_cow_child(i);
if (child) {
assert(child->frozen);
child->frozen = false;
}
}
}
/*
* Drops images above 'base' up to and including 'top', and sets the image
* above 'top' to have base as its backing file.
*
* Requires that the overlay to 'top' is opened r/w, so that the backing file
* information in 'bs' can be properly updated.
*
* E.g., this will convert the following chain:
* bottom <- base <- intermediate <- top <- active
*
* to
*
* bottom <- base <- active
*
* It is allowed for bottom==base, in which case it converts:
*
* base <- intermediate <- top <- active
*
* to
*
* base <- active
*
* If backing_file_str is non-NULL, it will be used when modifying top's
* overlay image metadata.
*
* Error conditions:
* if active == top, that is considered an error
*
*/
int bdrv_drop_intermediate(BlockDriverState *top, BlockDriverState *base,
const char *backing_file_str)
{
block: Update BlockDriverState.inherits_from on bdrv_drop_intermediate() The previous patch fixed the inherits_from pointer after block-stream, and this one does the same for block-commit. When block-commit finishes and the 'top' node is not the topmost one from the backing chain then all nodes above 'base' up to and including 'top' are removed from the chain. The bdrv_drop_intermediate() call converts a chain like this one: base <- intermediate <- top <- active into this one: base <- active In a simple scenario each backing file from the first chain has the inherits_from attribute pointing to its parent. This means that reopening 'active' will recursively reopen all its children, whose options can be changed in the process. However after the 'block-commit' call base.inherits_from is NULL and the chain is broken, so 'base' does not inherit from 'active' and will not be reopened automatically: $ qemu-img create -f qcow2 hd0.qcow2 1M $ qemu-img create -f qcow2 -b hd0.qcow2 hd1.qcow2 $ qemu-img create -f qcow2 -b hd1.qcow2 hd2.qcow2 $ $QEMU -drive if=none,file=hd2.qcow2 { 'execute': 'block-commit', 'arguments': { 'device': 'none0', 'top': 'hd1.qcow2' } } { 'execute': 'human-monitor-command', 'arguments': { 'command-line': 'qemu-io none0 "reopen -o backing.l2-cache-size=2M"' } } { "return": "Cannot change the option 'backing.l2-cache-size'\r\n"} This patch updates base.inherits_from in this scenario, and adds a test case. Signed-off-by: Alberto Garcia <berto@igalia.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2018-10-31 17:16:38 +01:00
BlockDriverState *explicit_top = top;
bool update_inherits_from;
BdrvChild *c;
Error *local_err = NULL;
int ret = -EIO;
g_autoptr(GSList) updated_children = NULL;
GSList *p;
GLOBAL_STATE_CODE();
bdrv_ref(top);
bdrv_subtree_drained_begin(top);
if (!top->drv || !base->drv) {
goto exit;
}
/* Make sure that base is in the backing chain of top */
if (!bdrv_chain_contains(top, base)) {
goto exit;
}
block: Update BlockDriverState.inherits_from on bdrv_drop_intermediate() The previous patch fixed the inherits_from pointer after block-stream, and this one does the same for block-commit. When block-commit finishes and the 'top' node is not the topmost one from the backing chain then all nodes above 'base' up to and including 'top' are removed from the chain. The bdrv_drop_intermediate() call converts a chain like this one: base <- intermediate <- top <- active into this one: base <- active In a simple scenario each backing file from the first chain has the inherits_from attribute pointing to its parent. This means that reopening 'active' will recursively reopen all its children, whose options can be changed in the process. However after the 'block-commit' call base.inherits_from is NULL and the chain is broken, so 'base' does not inherit from 'active' and will not be reopened automatically: $ qemu-img create -f qcow2 hd0.qcow2 1M $ qemu-img create -f qcow2 -b hd0.qcow2 hd1.qcow2 $ qemu-img create -f qcow2 -b hd1.qcow2 hd2.qcow2 $ $QEMU -drive if=none,file=hd2.qcow2 { 'execute': 'block-commit', 'arguments': { 'device': 'none0', 'top': 'hd1.qcow2' } } { 'execute': 'human-monitor-command', 'arguments': { 'command-line': 'qemu-io none0 "reopen -o backing.l2-cache-size=2M"' } } { "return": "Cannot change the option 'backing.l2-cache-size'\r\n"} This patch updates base.inherits_from in this scenario, and adds a test case. Signed-off-by: Alberto Garcia <berto@igalia.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2018-10-31 17:16:38 +01:00
/* If 'base' recursively inherits from 'top' then we should set
* base->inherits_from to top->inherits_from after 'top' and all
* other intermediate nodes have been dropped.
* If 'top' is an implicit node (e.g. "commit_top") we should skip
* it because no one inherits from it. We use explicit_top for that. */
explicit_top = bdrv_skip_implicit_filters(explicit_top);
block: Update BlockDriverState.inherits_from on bdrv_drop_intermediate() The previous patch fixed the inherits_from pointer after block-stream, and this one does the same for block-commit. When block-commit finishes and the 'top' node is not the topmost one from the backing chain then all nodes above 'base' up to and including 'top' are removed from the chain. The bdrv_drop_intermediate() call converts a chain like this one: base <- intermediate <- top <- active into this one: base <- active In a simple scenario each backing file from the first chain has the inherits_from attribute pointing to its parent. This means that reopening 'active' will recursively reopen all its children, whose options can be changed in the process. However after the 'block-commit' call base.inherits_from is NULL and the chain is broken, so 'base' does not inherit from 'active' and will not be reopened automatically: $ qemu-img create -f qcow2 hd0.qcow2 1M $ qemu-img create -f qcow2 -b hd0.qcow2 hd1.qcow2 $ qemu-img create -f qcow2 -b hd1.qcow2 hd2.qcow2 $ $QEMU -drive if=none,file=hd2.qcow2 { 'execute': 'block-commit', 'arguments': { 'device': 'none0', 'top': 'hd1.qcow2' } } { 'execute': 'human-monitor-command', 'arguments': { 'command-line': 'qemu-io none0 "reopen -o backing.l2-cache-size=2M"' } } { "return": "Cannot change the option 'backing.l2-cache-size'\r\n"} This patch updates base.inherits_from in this scenario, and adds a test case. Signed-off-by: Alberto Garcia <berto@igalia.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2018-10-31 17:16:38 +01:00
update_inherits_from = bdrv_inherits_from_recursive(base, explicit_top);
/* success - we can delete the intermediate states, and link top->base */
if (!backing_file_str) {
bdrv_refresh_filename(base);
backing_file_str = base->filename;
}
QLIST_FOREACH(c, &top->parents, next_parent) {
updated_children = g_slist_prepend(updated_children, c);
}
/*
* It seems correct to pass detach_subchain=true here, but it triggers
* one more yet not fixed bug, when due to nested aio_poll loop we switch to
* another drained section, which modify the graph (for example, removing
* the child, which we keep in updated_children list). So, it's a TODO.
*
* Note, bug triggered if pass detach_subchain=true here and run
* test-bdrv-drain. test_drop_intermediate_poll() test-case will crash.
* That's a FIXME.
*/
bdrv_replace_node_common(top, base, false, false, &local_err);
if (local_err) {
error_report_err(local_err);
goto exit;
}
for (p = updated_children; p; p = p->next) {
c = p->data;
if (c->klass->update_filename) {
ret = c->klass->update_filename(c, base, backing_file_str,
&local_err);
if (ret < 0) {
/*
* TODO: Actually, we want to rollback all previous iterations
* of this loop, and (which is almost impossible) previous
* bdrv_replace_node()...
*
* Note, that c->klass->update_filename may lead to permission
* update, so it's a bad idea to call it inside permission
* update transaction of bdrv_replace_node.
*/
error_report_err(local_err);
goto exit;
}
}
}
block: Update BlockDriverState.inherits_from on bdrv_drop_intermediate() The previous patch fixed the inherits_from pointer after block-stream, and this one does the same for block-commit. When block-commit finishes and the 'top' node is not the topmost one from the backing chain then all nodes above 'base' up to and including 'top' are removed from the chain. The bdrv_drop_intermediate() call converts a chain like this one: base <- intermediate <- top <- active into this one: base <- active In a simple scenario each backing file from the first chain has the inherits_from attribute pointing to its parent. This means that reopening 'active' will recursively reopen all its children, whose options can be changed in the process. However after the 'block-commit' call base.inherits_from is NULL and the chain is broken, so 'base' does not inherit from 'active' and will not be reopened automatically: $ qemu-img create -f qcow2 hd0.qcow2 1M $ qemu-img create -f qcow2 -b hd0.qcow2 hd1.qcow2 $ qemu-img create -f qcow2 -b hd1.qcow2 hd2.qcow2 $ $QEMU -drive if=none,file=hd2.qcow2 { 'execute': 'block-commit', 'arguments': { 'device': 'none0', 'top': 'hd1.qcow2' } } { 'execute': 'human-monitor-command', 'arguments': { 'command-line': 'qemu-io none0 "reopen -o backing.l2-cache-size=2M"' } } { "return": "Cannot change the option 'backing.l2-cache-size'\r\n"} This patch updates base.inherits_from in this scenario, and adds a test case. Signed-off-by: Alberto Garcia <berto@igalia.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2018-10-31 17:16:38 +01:00
if (update_inherits_from) {
base->inherits_from = explicit_top->inherits_from;
}
ret = 0;
exit:
bdrv_subtree_drained_end(top);
bdrv_unref(top);
return ret;
}
/**
* Implementation of BlockDriver.bdrv_get_allocated_file_size() that
* sums the size of all data-bearing children. (This excludes backing
* children.)
*/
static int64_t bdrv_sum_allocated_file_size(BlockDriverState *bs)
{
BdrvChild *child;
int64_t child_size, sum = 0;
QLIST_FOREACH(child, &bs->children, next) {
if (child->role & (BDRV_CHILD_DATA | BDRV_CHILD_METADATA |
BDRV_CHILD_FILTERED))
{
child_size = bdrv_get_allocated_file_size(child->bs);
if (child_size < 0) {
return child_size;
}
sum += child_size;
}
}
return sum;
}
/**
* Length of a allocated file in bytes. Sparse files are counted by actual
* allocated space. Return < 0 if error or unknown.
*/
int64_t bdrv_get_allocated_file_size(BlockDriverState *bs)
{
BlockDriver *drv = bs->drv;
IO_CODE();
if (!drv) {
return -ENOMEDIUM;
}
if (drv->bdrv_get_allocated_file_size) {
return drv->bdrv_get_allocated_file_size(bs);
}
if (drv->bdrv_file_open) {
/*
* Protocol drivers default to -ENOTSUP (most of their data is
* not stored in any of their children (if they even have any),
* so there is no generic way to figure it out).
*/
return -ENOTSUP;
} else if (drv->is_filter) {
/* Filter drivers default to the size of their filtered child */
return bdrv_get_allocated_file_size(bdrv_filter_bs(bs));
} else {
/* Other drivers default to summing their children's sizes */
return bdrv_sum_allocated_file_size(bs);
}
}
/*
* bdrv_measure:
* @drv: Format driver
* @opts: Creation options for new image
* @in_bs: Existing image containing data for new image (may be NULL)
* @errp: Error object
* Returns: A #BlockMeasureInfo (free using qapi_free_BlockMeasureInfo())
* or NULL on error
*
* Calculate file size required to create a new image.
*
* If @in_bs is given then space for allocated clusters and zero clusters
* from that image are included in the calculation. If @opts contains a
* backing file that is shared by @in_bs then backing clusters may be omitted
* from the calculation.
*
* If @in_bs is NULL then the calculation includes no allocated clusters
* unless a preallocation option is given in @opts.
*
* Note that @in_bs may use a different BlockDriver from @drv.
*
* If an error occurs the @errp pointer is set.
*/
BlockMeasureInfo *bdrv_measure(BlockDriver *drv, QemuOpts *opts,
BlockDriverState *in_bs, Error **errp)
{
IO_CODE();
if (!drv->bdrv_measure) {
error_setg(errp, "Block driver '%s' does not support size measurement",
drv->format_name);
return NULL;
}
return drv->bdrv_measure(opts, in_bs, errp);
}
/**
* Return number of sectors on success, -errno on error.
*/
int64_t bdrv_nb_sectors(BlockDriverState *bs)
{
BlockDriver *drv = bs->drv;
IO_CODE();
if (!drv)
return -ENOMEDIUM;
if (drv->has_variable_length) {
int ret = refresh_total_sectors(bs, bs->total_sectors);
if (ret < 0) {
return ret;
}
}
return bs->total_sectors;
}
/**
* Return length in bytes on success, -errno on error.
* The length is always a multiple of BDRV_SECTOR_SIZE.
*/
int64_t bdrv_getlength(BlockDriverState *bs)
{
int64_t ret = bdrv_nb_sectors(bs);
IO_CODE();
if (ret < 0) {
return ret;
}
if (ret > INT64_MAX / BDRV_SECTOR_SIZE) {
return -EFBIG;
}
return ret * BDRV_SECTOR_SIZE;
}
/* return 0 as number of sectors if no device present or error */
void bdrv_get_geometry(BlockDriverState *bs, uint64_t *nb_sectors_ptr)
{
int64_t nb_sectors = bdrv_nb_sectors(bs);
IO_CODE();
*nb_sectors_ptr = nb_sectors < 0 ? 0 : nb_sectors;
}
bool bdrv_is_sg(BlockDriverState *bs)
{
IO_CODE();
return bs->sg;
}
/**
* Return whether the given node supports compressed writes.
*/
bool bdrv_supports_compressed_writes(BlockDriverState *bs)
{
BlockDriverState *filtered;
IO_CODE();
if (!bs->drv || !block_driver_can_compress(bs->drv)) {
return false;
}
filtered = bdrv_filter_bs(bs);
if (filtered) {
/*
* Filters can only forward compressed writes, so we have to
* check the child.
*/
return bdrv_supports_compressed_writes(filtered);
}
return true;
}
const char *bdrv_get_format_name(BlockDriverState *bs)
{
IO_CODE();
return bs->drv ? bs->drv->format_name : NULL;
}
static int qsort_strcmp(const void *a, const void *b)
{
return strcmp(*(char *const *)a, *(char *const *)b);
}
void bdrv_iterate_format(void (*it)(void *opaque, const char *name),
void *opaque, bool read_only)
{
BlockDriver *drv;
int count = 0;
int i;
const char **formats = NULL;
GLOBAL_STATE_CODE();
QLIST_FOREACH(drv, &bdrv_drivers, list) {
if (drv->format_name) {
bool found = false;
int i = count;
if (use_bdrv_whitelist && !bdrv_is_whitelisted(drv, read_only)) {
continue;
}
while (formats && i && !found) {
found = !strcmp(formats[--i], drv->format_name);
}
if (!found) {
formats = g_renew(const char *, formats, count + 1);
formats[count++] = drv->format_name;
}
}
}
for (i = 0; i < (int)ARRAY_SIZE(block_driver_modules); i++) {
const char *format_name = block_driver_modules[i].format_name;
if (format_name) {
bool found = false;
int j = count;
if (use_bdrv_whitelist &&
!bdrv_format_is_whitelisted(format_name, read_only)) {
continue;
}
while (formats && j && !found) {
found = !strcmp(formats[--j], format_name);
}
if (!found) {
formats = g_renew(const char *, formats, count + 1);
formats[count++] = format_name;
}
}
}
qsort(formats, count, sizeof(formats[0]), qsort_strcmp);
for (i = 0; i < count; i++) {
it(opaque, formats[i]);
}
g_free(formats);
}
/* This function is to find a node in the bs graph */
BlockDriverState *bdrv_find_node(const char *node_name)
{
BlockDriverState *bs;
assert(node_name);
GLOBAL_STATE_CODE();
QTAILQ_FOREACH(bs, &graph_bdrv_states, node_list) {
if (!strcmp(node_name, bs->node_name)) {
return bs;
}
}
return NULL;
}
/* Put this QMP function here so it can access the static graph_bdrv_states. */
BlockDeviceInfoList *bdrv_named_nodes_list(bool flat,
Error **errp)
{
qapi: Add QAPI_LIST_PREPEND() macro block.c has a useful macro QAPI_LIST_ADD() for inserting at the front of any QAPI-generated list; move it from block.c to qapi/util.h so more places can use it, including one earlier place in block.c, and rename it to something more obvious (since we also have a lot of places that append, rather than prepend, to a list). There are many more places in the codebase that can benefit from using the macro, but converting them will be left to later patches. In theory, all QAPI list types are child classes of GenericList; but in practice, that relationship is not explicitly spelled out in the C type declarations (rather, it is something that happens implicitly due to C compatible layouts), and the macro does not actually depend on the GenericList type. We considered moving GenericList from visitor.h into util.h to group related code; however, such a move would be awkward if we do not also move GenericAlternate. Unfortunately, moving GenericAlternate would introduce its own problems of declaration circularity (qapi-builtin-types.h needs a complete definition of QEnumLookup from util.h, but GenericAlternate needs a complete definition of QType from qapi-builtin-types.h). Suggested-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Signed-off-by: Eric Blake <eblake@redhat.com> Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Reviewed-by: Markus Armbruster <armbru@redhat.com> Message-Id: <20201027050556.269064-3-eblake@redhat.com> [eblake: s/ADD/PREPEND/ per suggestion by Markus]
2020-10-27 06:05:47 +01:00
BlockDeviceInfoList *list;
BlockDriverState *bs;
GLOBAL_STATE_CODE();
list = NULL;
QTAILQ_FOREACH(bs, &graph_bdrv_states, node_list) {
BlockDeviceInfo *info = bdrv_block_device_info(NULL, bs, flat, errp);
if (!info) {
qapi_free_BlockDeviceInfoList(list);
return NULL;
}
qapi: Add QAPI_LIST_PREPEND() macro block.c has a useful macro QAPI_LIST_ADD() for inserting at the front of any QAPI-generated list; move it from block.c to qapi/util.h so more places can use it, including one earlier place in block.c, and rename it to something more obvious (since we also have a lot of places that append, rather than prepend, to a list). There are many more places in the codebase that can benefit from using the macro, but converting them will be left to later patches. In theory, all QAPI list types are child classes of GenericList; but in practice, that relationship is not explicitly spelled out in the C type declarations (rather, it is something that happens implicitly due to C compatible layouts), and the macro does not actually depend on the GenericList type. We considered moving GenericList from visitor.h into util.h to group related code; however, such a move would be awkward if we do not also move GenericAlternate. Unfortunately, moving GenericAlternate would introduce its own problems of declaration circularity (qapi-builtin-types.h needs a complete definition of QEnumLookup from util.h, but GenericAlternate needs a complete definition of QType from qapi-builtin-types.h). Suggested-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Signed-off-by: Eric Blake <eblake@redhat.com> Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Reviewed-by: Markus Armbruster <armbru@redhat.com> Message-Id: <20201027050556.269064-3-eblake@redhat.com> [eblake: s/ADD/PREPEND/ per suggestion by Markus]
2020-10-27 06:05:47 +01:00
QAPI_LIST_PREPEND(list, info);
}
return list;
}
typedef struct XDbgBlockGraphConstructor {
XDbgBlockGraph *graph;
GHashTable *graph_nodes;
} XDbgBlockGraphConstructor;
static XDbgBlockGraphConstructor *xdbg_graph_new(void)
{
XDbgBlockGraphConstructor *gr = g_new(XDbgBlockGraphConstructor, 1);
gr->graph = g_new0(XDbgBlockGraph, 1);
gr->graph_nodes = g_hash_table_new(NULL, NULL);
return gr;
}
static XDbgBlockGraph *xdbg_graph_finalize(XDbgBlockGraphConstructor *gr)
{
XDbgBlockGraph *graph = gr->graph;
g_hash_table_destroy(gr->graph_nodes);
g_free(gr);
return graph;
}
static uintptr_t xdbg_graph_node_num(XDbgBlockGraphConstructor *gr, void *node)
{
uintptr_t ret = (uintptr_t)g_hash_table_lookup(gr->graph_nodes, node);
if (ret != 0) {
return ret;
}
/*
* Start counting from 1, not 0, because 0 interferes with not-found (NULL)
* answer of g_hash_table_lookup.
*/
ret = g_hash_table_size(gr->graph_nodes) + 1;
g_hash_table_insert(gr->graph_nodes, node, (void *)ret);
return ret;
}
static void xdbg_graph_add_node(XDbgBlockGraphConstructor *gr, void *node,
XDbgBlockGraphNodeType type, const char *name)
{
XDbgBlockGraphNode *n;
n = g_new0(XDbgBlockGraphNode, 1);
n->id = xdbg_graph_node_num(gr, node);
n->type = type;
n->name = g_strdup(name);
qapi: Add QAPI_LIST_PREPEND() macro block.c has a useful macro QAPI_LIST_ADD() for inserting at the front of any QAPI-generated list; move it from block.c to qapi/util.h so more places can use it, including one earlier place in block.c, and rename it to something more obvious (since we also have a lot of places that append, rather than prepend, to a list). There are many more places in the codebase that can benefit from using the macro, but converting them will be left to later patches. In theory, all QAPI list types are child classes of GenericList; but in practice, that relationship is not explicitly spelled out in the C type declarations (rather, it is something that happens implicitly due to C compatible layouts), and the macro does not actually depend on the GenericList type. We considered moving GenericList from visitor.h into util.h to group related code; however, such a move would be awkward if we do not also move GenericAlternate. Unfortunately, moving GenericAlternate would introduce its own problems of declaration circularity (qapi-builtin-types.h needs a complete definition of QEnumLookup from util.h, but GenericAlternate needs a complete definition of QType from qapi-builtin-types.h). Suggested-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Signed-off-by: Eric Blake <eblake@redhat.com> Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Reviewed-by: Markus Armbruster <armbru@redhat.com> Message-Id: <20201027050556.269064-3-eblake@redhat.com> [eblake: s/ADD/PREPEND/ per suggestion by Markus]
2020-10-27 06:05:47 +01:00
QAPI_LIST_PREPEND(gr->graph->nodes, n);
}
static void xdbg_graph_add_edge(XDbgBlockGraphConstructor *gr, void *parent,
const BdrvChild *child)
{
BlockPermission qapi_perm;
XDbgBlockGraphEdge *edge;
GLOBAL_STATE_CODE();
edge = g_new0(XDbgBlockGraphEdge, 1);
edge->parent = xdbg_graph_node_num(gr, parent);
edge->child = xdbg_graph_node_num(gr, child->bs);
edge->name = g_strdup(child->name);
for (qapi_perm = 0; qapi_perm < BLOCK_PERMISSION__MAX; qapi_perm++) {
uint64_t flag = bdrv_qapi_perm_to_blk_perm(qapi_perm);
if (flag & child->perm) {
qapi: Add QAPI_LIST_PREPEND() macro block.c has a useful macro QAPI_LIST_ADD() for inserting at the front of any QAPI-generated list; move it from block.c to qapi/util.h so more places can use it, including one earlier place in block.c, and rename it to something more obvious (since we also have a lot of places that append, rather than prepend, to a list). There are many more places in the codebase that can benefit from using the macro, but converting them will be left to later patches. In theory, all QAPI list types are child classes of GenericList; but in practice, that relationship is not explicitly spelled out in the C type declarations (rather, it is something that happens implicitly due to C compatible layouts), and the macro does not actually depend on the GenericList type. We considered moving GenericList from visitor.h into util.h to group related code; however, such a move would be awkward if we do not also move GenericAlternate. Unfortunately, moving GenericAlternate would introduce its own problems of declaration circularity (qapi-builtin-types.h needs a complete definition of QEnumLookup from util.h, but GenericAlternate needs a complete definition of QType from qapi-builtin-types.h). Suggested-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Signed-off-by: Eric Blake <eblake@redhat.com> Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Reviewed-by: Markus Armbruster <armbru@redhat.com> Message-Id: <20201027050556.269064-3-eblake@redhat.com> [eblake: s/ADD/PREPEND/ per suggestion by Markus]
2020-10-27 06:05:47 +01:00
QAPI_LIST_PREPEND(edge->perm, qapi_perm);
}
if (flag & child->shared_perm) {
qapi: Add QAPI_LIST_PREPEND() macro block.c has a useful macro QAPI_LIST_ADD() for inserting at the front of any QAPI-generated list; move it from block.c to qapi/util.h so more places can use it, including one earlier place in block.c, and rename it to something more obvious (since we also have a lot of places that append, rather than prepend, to a list). There are many more places in the codebase that can benefit from using the macro, but converting them will be left to later patches. In theory, all QAPI list types are child classes of GenericList; but in practice, that relationship is not explicitly spelled out in the C type declarations (rather, it is something that happens implicitly due to C compatible layouts), and the macro does not actually depend on the GenericList type. We considered moving GenericList from visitor.h into util.h to group related code; however, such a move would be awkward if we do not also move GenericAlternate. Unfortunately, moving GenericAlternate would introduce its own problems of declaration circularity (qapi-builtin-types.h needs a complete definition of QEnumLookup from util.h, but GenericAlternate needs a complete definition of QType from qapi-builtin-types.h). Suggested-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Signed-off-by: Eric Blake <eblake@redhat.com> Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Reviewed-by: Markus Armbruster <armbru@redhat.com> Message-Id: <20201027050556.269064-3-eblake@redhat.com> [eblake: s/ADD/PREPEND/ per suggestion by Markus]
2020-10-27 06:05:47 +01:00
QAPI_LIST_PREPEND(edge->shared_perm, qapi_perm);
}
}
qapi: Add QAPI_LIST_PREPEND() macro block.c has a useful macro QAPI_LIST_ADD() for inserting at the front of any QAPI-generated list; move it from block.c to qapi/util.h so more places can use it, including one earlier place in block.c, and rename it to something more obvious (since we also have a lot of places that append, rather than prepend, to a list). There are many more places in the codebase that can benefit from using the macro, but converting them will be left to later patches. In theory, all QAPI list types are child classes of GenericList; but in practice, that relationship is not explicitly spelled out in the C type declarations (rather, it is something that happens implicitly due to C compatible layouts), and the macro does not actually depend on the GenericList type. We considered moving GenericList from visitor.h into util.h to group related code; however, such a move would be awkward if we do not also move GenericAlternate. Unfortunately, moving GenericAlternate would introduce its own problems of declaration circularity (qapi-builtin-types.h needs a complete definition of QEnumLookup from util.h, but GenericAlternate needs a complete definition of QType from qapi-builtin-types.h). Suggested-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Signed-off-by: Eric Blake <eblake@redhat.com> Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Reviewed-by: Markus Armbruster <armbru@redhat.com> Message-Id: <20201027050556.269064-3-eblake@redhat.com> [eblake: s/ADD/PREPEND/ per suggestion by Markus]
2020-10-27 06:05:47 +01:00
QAPI_LIST_PREPEND(gr->graph->edges, edge);
}
XDbgBlockGraph *bdrv_get_xdbg_block_graph(Error **errp)
{
BlockBackend *blk;
BlockJob *job;
BlockDriverState *bs;
BdrvChild *child;
XDbgBlockGraphConstructor *gr = xdbg_graph_new();
GLOBAL_STATE_CODE();
for (blk = blk_all_next(NULL); blk; blk = blk_all_next(blk)) {
char *allocated_name = NULL;
const char *name = blk_name(blk);
if (!*name) {
name = allocated_name = blk_get_attached_dev_id(blk);
}
xdbg_graph_add_node(gr, blk, X_DBG_BLOCK_GRAPH_NODE_TYPE_BLOCK_BACKEND,
name);
g_free(allocated_name);
if (blk_root(blk)) {
xdbg_graph_add_edge(gr, blk, blk_root(blk));
}
}
for (job = block_job_next(NULL); job; job = block_job_next(job)) {
GSList *el;
xdbg_graph_add_node(gr, job, X_DBG_BLOCK_GRAPH_NODE_TYPE_BLOCK_JOB,
job->job.id);
for (el = job->nodes; el; el = el->next) {
xdbg_graph_add_edge(gr, job, (BdrvChild *)el->data);
}
}
QTAILQ_FOREACH(bs, &graph_bdrv_states, node_list) {
xdbg_graph_add_node(gr, bs, X_DBG_BLOCK_GRAPH_NODE_TYPE_BLOCK_DRIVER,
bs->node_name);
QLIST_FOREACH(child, &bs->children, next) {
xdbg_graph_add_edge(gr, bs, child);
}
}
return xdbg_graph_finalize(gr);
}
BlockDriverState *bdrv_lookup_bs(const char *device,
const char *node_name,
Error **errp)
{
BlockBackend *blk;
BlockDriverState *bs;
GLOBAL_STATE_CODE();
if (device) {
blk = blk_by_name(device);
if (blk) {
bs = blk_bs(blk);
if (!bs) {
error_setg(errp, "Device '%s' has no medium", device);
}
return bs;
}
}
if (node_name) {
bs = bdrv_find_node(node_name);
if (bs) {
return bs;
}
}
error_setg(errp, "Cannot find device=\'%s\' nor node-name=\'%s\'",
device ? device : "",
node_name ? node_name : "");
return NULL;
}
/* If 'base' is in the same chain as 'top', return true. Otherwise,
* return false. If either argument is NULL, return false. */
bool bdrv_chain_contains(BlockDriverState *top, BlockDriverState *base)
{
GLOBAL_STATE_CODE();
while (top && top != base) {
top = bdrv_filter_or_cow_bs(top);
}
return top != NULL;
}
BlockDriverState *bdrv_next_node(BlockDriverState *bs)
{
GLOBAL_STATE_CODE();
if (!bs) {
return QTAILQ_FIRST(&graph_bdrv_states);
}
return QTAILQ_NEXT(bs, node_list);
}
BlockDriverState *bdrv_next_all_states(BlockDriverState *bs)
{
GLOBAL_STATE_CODE();
if (!bs) {
return QTAILQ_FIRST(&all_bdrv_states);
}
return QTAILQ_NEXT(bs, bs_list);
}
const char *bdrv_get_node_name(const BlockDriverState *bs)
{
IO_CODE();
return bs->node_name;
}
const char *bdrv_get_parent_name(const BlockDriverState *bs)
{
BdrvChild *c;
const char *name;
IO_CODE();
/* If multiple parents have a name, just pick the first one. */
QLIST_FOREACH(c, &bs->parents, next_parent) {
if (c->klass->get_name) {
name = c->klass->get_name(c);
if (name && *name) {
return name;
}
}
}
return NULL;
}
/* TODO check what callers really want: bs->node_name or blk_name() */
const char *bdrv_get_device_name(const BlockDriverState *bs)
{
IO_CODE();
return bdrv_get_parent_name(bs) ?: "";
}
/* This can be used to identify nodes that might not have a device
* name associated. Since node and device names live in the same
* namespace, the result is unambiguous. The exception is if both are
* absent, then this returns an empty (non-null) string. */
const char *bdrv_get_device_or_node_name(const BlockDriverState *bs)
{
IO_CODE();
return bdrv_get_parent_name(bs) ?: bs->node_name;
}
int bdrv_get_flags(BlockDriverState *bs)
{
IO_CODE();
return bs->open_flags;
}
int bdrv_has_zero_init_1(BlockDriverState *bs)
{
GLOBAL_STATE_CODE();
return 1;
}
int bdrv_has_zero_init(BlockDriverState *bs)
{
BlockDriverState *filtered;
GLOBAL_STATE_CODE();
if (!bs->drv) {
return 0;
}
/* If BS is a copy on write image, it is initialized to
the contents of the base image, which may not be zeroes. */
if (bdrv_cow_child(bs)) {
return 0;
}
if (bs->drv->bdrv_has_zero_init) {
return bs->drv->bdrv_has_zero_init(bs);
}
filtered = bdrv_filter_bs(bs);
if (filtered) {
return bdrv_has_zero_init(filtered);
}
/* safe default */
return 0;
}
bool bdrv_can_write_zeroes_with_unmap(BlockDriverState *bs)
{
IO_CODE();
if (!(bs->open_flags & BDRV_O_UNMAP)) {
return false;
}
block: Simplify bdrv_can_write_zeroes_with_unmap() We don't need the can_write_zeroes_with_unmap field in BlockDriverInfo, because it is redundant information with supported_zero_flags & BDRV_REQ_MAY_UNMAP. Note that BlockDriverInfo and supported_zero_flags are both per-device settings, rather than global state about the driver as a whole, which means one or both of these bits of information can already be conditional. Let's audit how they were set: crypto: always setting can_write_ to false is pointless (the struct starts life zero-initialized), no use of supported_ nbd: just recently fixed to set can_write_ if supported_ includes MAY_UNMAP (thus this commit effectively reverts bca80059e and solves the problem mentioned there in a more global way) file-posix, iscsi, qcow2: can_write_ is conditional, while supported_ was unconditional; but passing MAY_UNMAP would fail with ENOTSUP if the condition wasn't met qed: can_write_ is unconditional, but pwrite_zeroes lacks support for MAY_UNMAP and supported_ is not set. Perhaps support can be added later (since it would be similar to qcow2), but for now claiming false is no real loss all other drivers: can_write_ is not set, and supported_ is either unset or a passthrough Simplify the code by moving the conditional into supported_zero_flags for all drivers, then dropping the now-unused BDI field. For callers that relied on bdrv_can_write_zeroes_with_unmap(), we return the same per-device settings for drivers that had conditions (no observable change in behavior there); and can now return true (instead of false) for drivers that support passthrough (for example, the commit driver) which gives those drivers the same fix as nbd just got in bca80059e. For callers that relied on supported_zero_flags, we now have a few more places that can avoid a wasted call to pwrite_zeroes() that will just fail with ENOTSUP. Suggested-by: Paolo Bonzini <pbonzini@redhat.com> Signed-off-by: Eric Blake <eblake@redhat.com> Message-Id: <20180126193439.20219-1-eblake@redhat.com> Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
2018-01-26 20:34:39 +01:00
return bs->supported_zero_flags & BDRV_REQ_MAY_UNMAP;
}
void bdrv_get_backing_filename(BlockDriverState *bs,
char *filename, int filename_size)
{
IO_CODE();
pstrcpy(filename, filename_size, bs->backing_file);
}
int bdrv_get_info(BlockDriverState *bs, BlockDriverInfo *bdi)
{
block: introduce BDRV_MAX_LENGTH We are going to modify block layer to work with 64bit requests. And first step is moving to int64_t type for both offset and bytes arguments in all block request related functions. It's mostly safe (when widening signed or unsigned int to int64_t), but switching from uint64_t is questionable. So, let's first establish the set of requests we want to work with. First signed int64_t should be enough, as off_t is signed anyway. Then, obviously offset + bytes should not overflow. And most interesting: (offset + bytes) being aligned up should not overflow as well. Aligned to what alignment? First thing that comes in mind is bs->bl.request_alignment, as we align up request to this alignment. But there is another thing: look at bdrv_mark_request_serialising(). It aligns request up to some given alignment. And this parameter may be bdrv_get_cluster_size(), which is often a lot greater than bs->bl.request_alignment. Note also, that bdrv_mark_request_serialising() uses signed int64_t for calculations. So, actually, we already depend on some restrictions. Happily, bdrv_get_cluster_size() returns int and bs->bl.request_alignment has 32bit unsigned type, but defined to be a power of 2 less than INT_MAX. So, we may establish, that INT_MAX is absolute maximum for any kind of alignment that may occur with the request. Note, that bdrv_get_cluster_size() is not documented to return power of 2, still bdrv_mark_request_serialising() behaves like it is. Also, backup uses bdi.cluster_size and is not prepared to it not being power of 2. So, let's establish that Qemu supports only power-of-2 clusters and alignments. So, alignment can't be greater than 2^30. Finally to be safe with calculations, to not calculate different maximums for different nodes (depending on cluster size and request_alignment), let's simply set QEMU_ALIGN_DOWN(INT64_MAX, 2^30) as absolute maximum bytes length for Qemu. Actually, it's not much less than INT64_MAX. OK, then, let's apply it to block/io. Let's consider all block/io entry points of offset/bytes: 4 bytes/offset interface functions: bdrv_co_preadv_part(), bdrv_co_pwritev_part(), bdrv_co_copy_range_internal() and bdrv_co_pdiscard() and we check them all with bdrv_check_request(). We also have one entry point with only offset: bdrv_co_truncate(). Check the offset. And one public structure: BdrvTrackedRequest. Happily, it has only three external users: file-posix.c: adopted by this patch write-threshold.c: only read fields test-write-threshold.c: sets obviously small constant values Better is to make the structure private and add corresponding interfaces.. Still it's not obvious what kind of interface is needed for file-posix.c. Let's keep it public but add corresponding assertions. After this patch we'll convert functions in block/io.c to int64_t bytes and offset parameters. We can assume that offset/bytes pair always satisfy new restrictions, and make corresponding assertions where needed. If we reach some offset/bytes point in block/io.c missing bdrv_check_request() it is considered a bug. As well, if block/io.c modifies a offset/bytes request, expanding it more then aligning up to request_alignment, it's a bug too. For all io requests except for discard we keep for now old restriction of 32bit request length. iotest 206 output error message changed, as now test disk size is larger than new limit. Add one more test case with new maximum disk size to cover too-big-L1 case. Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Message-Id: <20201203222713.13507-5-vsementsov@virtuozzo.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2020-12-03 23:27:13 +01:00
int ret;
BlockDriver *drv = bs->drv;
IO_CODE();
/* if bs->drv == NULL, bs is closed, so there's nothing to do here */
if (!drv) {
return -ENOMEDIUM;
}
if (!drv->bdrv_get_info) {
BlockDriverState *filtered = bdrv_filter_bs(bs);
if (filtered) {
return bdrv_get_info(filtered, bdi);
}
return -ENOTSUP;
}
memset(bdi, 0, sizeof(*bdi));
block: introduce BDRV_MAX_LENGTH We are going to modify block layer to work with 64bit requests. And first step is moving to int64_t type for both offset and bytes arguments in all block request related functions. It's mostly safe (when widening signed or unsigned int to int64_t), but switching from uint64_t is questionable. So, let's first establish the set of requests we want to work with. First signed int64_t should be enough, as off_t is signed anyway. Then, obviously offset + bytes should not overflow. And most interesting: (offset + bytes) being aligned up should not overflow as well. Aligned to what alignment? First thing that comes in mind is bs->bl.request_alignment, as we align up request to this alignment. But there is another thing: look at bdrv_mark_request_serialising(). It aligns request up to some given alignment. And this parameter may be bdrv_get_cluster_size(), which is often a lot greater than bs->bl.request_alignment. Note also, that bdrv_mark_request_serialising() uses signed int64_t for calculations. So, actually, we already depend on some restrictions. Happily, bdrv_get_cluster_size() returns int and bs->bl.request_alignment has 32bit unsigned type, but defined to be a power of 2 less than INT_MAX. So, we may establish, that INT_MAX is absolute maximum for any kind of alignment that may occur with the request. Note, that bdrv_get_cluster_size() is not documented to return power of 2, still bdrv_mark_request_serialising() behaves like it is. Also, backup uses bdi.cluster_size and is not prepared to it not being power of 2. So, let's establish that Qemu supports only power-of-2 clusters and alignments. So, alignment can't be greater than 2^30. Finally to be safe with calculations, to not calculate different maximums for different nodes (depending on cluster size and request_alignment), let's simply set QEMU_ALIGN_DOWN(INT64_MAX, 2^30) as absolute maximum bytes length for Qemu. Actually, it's not much less than INT64_MAX. OK, then, let's apply it to block/io. Let's consider all block/io entry points of offset/bytes: 4 bytes/offset interface functions: bdrv_co_preadv_part(), bdrv_co_pwritev_part(), bdrv_co_copy_range_internal() and bdrv_co_pdiscard() and we check them all with bdrv_check_request(). We also have one entry point with only offset: bdrv_co_truncate(). Check the offset. And one public structure: BdrvTrackedRequest. Happily, it has only three external users: file-posix.c: adopted by this patch write-threshold.c: only read fields test-write-threshold.c: sets obviously small constant values Better is to make the structure private and add corresponding interfaces.. Still it's not obvious what kind of interface is needed for file-posix.c. Let's keep it public but add corresponding assertions. After this patch we'll convert functions in block/io.c to int64_t bytes and offset parameters. We can assume that offset/bytes pair always satisfy new restrictions, and make corresponding assertions where needed. If we reach some offset/bytes point in block/io.c missing bdrv_check_request() it is considered a bug. As well, if block/io.c modifies a offset/bytes request, expanding it more then aligning up to request_alignment, it's a bug too. For all io requests except for discard we keep for now old restriction of 32bit request length. iotest 206 output error message changed, as now test disk size is larger than new limit. Add one more test case with new maximum disk size to cover too-big-L1 case. Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Message-Id: <20201203222713.13507-5-vsementsov@virtuozzo.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2020-12-03 23:27:13 +01:00
ret = drv->bdrv_get_info(bs, bdi);
if (ret < 0) {
return ret;
}
if (bdi->cluster_size > BDRV_MAX_ALIGNMENT) {
return -EINVAL;
}
return 0;
}
ImageInfoSpecific *bdrv_get_specific_info(BlockDriverState *bs,
Error **errp)
{
BlockDriver *drv = bs->drv;
IO_CODE();
if (drv && drv->bdrv_get_specific_info) {
return drv->bdrv_get_specific_info(bs, errp);
}
return NULL;
}
BlockStatsSpecific *bdrv_get_specific_stats(BlockDriverState *bs)
{
BlockDriver *drv = bs->drv;
IO_CODE();
if (!drv || !drv->bdrv_get_specific_stats) {
return NULL;
}
return drv->bdrv_get_specific_stats(bs);
}
void bdrv_debug_event(BlockDriverState *bs, BlkdebugEvent event)
{
IO_CODE();
if (!bs || !bs->drv || !bs->drv->bdrv_debug_event) {
return;
}
bs->drv->bdrv_debug_event(bs, event);
}
static BlockDriverState *bdrv_find_debug_node(BlockDriverState *bs)
{
GLOBAL_STATE_CODE();
while (bs && bs->drv && !bs->drv->bdrv_debug_breakpoint) {
bs = bdrv_primary_bs(bs);
}
if (bs && bs->drv && bs->drv->bdrv_debug_breakpoint) {
assert(bs->drv->bdrv_debug_remove_breakpoint);
return bs;
}
return NULL;
}
int bdrv_debug_breakpoint(BlockDriverState *bs, const char *event,
const char *tag)
{
GLOBAL_STATE_CODE();
bs = bdrv_find_debug_node(bs);
if (bs) {
return bs->drv->bdrv_debug_breakpoint(bs, event, tag);
}
return -ENOTSUP;
}
int bdrv_debug_remove_breakpoint(BlockDriverState *bs, const char *tag)
{
GLOBAL_STATE_CODE();
bs = bdrv_find_debug_node(bs);
if (bs) {
return bs->drv->bdrv_debug_remove_breakpoint(bs, tag);
}
return -ENOTSUP;
}
int bdrv_debug_resume(BlockDriverState *bs, const char *tag)
{
GLOBAL_STATE_CODE();
while (bs && (!bs->drv || !bs->drv->bdrv_debug_resume)) {
bs = bdrv_primary_bs(bs);
}
if (bs && bs->drv && bs->drv->bdrv_debug_resume) {
return bs->drv->bdrv_debug_resume(bs, tag);
}
return -ENOTSUP;
}
bool bdrv_debug_is_suspended(BlockDriverState *bs, const char *tag)
{
GLOBAL_STATE_CODE();
while (bs && bs->drv && !bs->drv->bdrv_debug_is_suspended) {
bs = bdrv_primary_bs(bs);
}
if (bs && bs->drv && bs->drv->bdrv_debug_is_suspended) {
return bs->drv->bdrv_debug_is_suspended(bs, tag);
}
return false;
}
/* backing_file can either be relative, or absolute, or a protocol. If it is
* relative, it must be relative to the chain. So, passing in bs->filename
* from a BDS as backing_file should not be done, as that may be relative to
* the CWD rather than the chain. */
BlockDriverState *bdrv_find_backing_image(BlockDriverState *bs,
const char *backing_file)
{
char *filename_full = NULL;
char *backing_file_full = NULL;
char *filename_tmp = NULL;
int is_protocol = 0;
block: Leave BDS.backing_{file,format} constant Parts of the block layer treat BDS.backing_file as if it were whatever the image header says (i.e., if it is a relative path, it is relative to the overlay), other parts treat it like a cache for bs->backing->bs->filename (relative paths are relative to the CWD). Considering bs->backing->bs->filename exists, let us make it mean the former. Among other things, this now allows the user to specify a base when using qemu-img to commit an image file in a directory that is not the CWD (assuming, everything uses relative filenames). Before this patch: $ ./qemu-img create -f qcow2 foo/bot.qcow2 1M $ ./qemu-img create -f qcow2 -b bot.qcow2 foo/mid.qcow2 $ ./qemu-img create -f qcow2 -b mid.qcow2 foo/top.qcow2 $ ./qemu-img commit -b mid.qcow2 foo/top.qcow2 qemu-img: Did not find 'mid.qcow2' in the backing chain of 'foo/top.qcow2' $ ./qemu-img commit -b foo/mid.qcow2 foo/top.qcow2 qemu-img: Did not find 'foo/mid.qcow2' in the backing chain of 'foo/top.qcow2' $ ./qemu-img commit -b $PWD/foo/mid.qcow2 foo/top.qcow2 qemu-img: Did not find '[...]/foo/mid.qcow2' in the backing chain of 'foo/top.qcow2' After this patch: $ ./qemu-img commit -b mid.qcow2 foo/top.qcow2 Image committed. $ ./qemu-img commit -b foo/mid.qcow2 foo/top.qcow2 qemu-img: Did not find 'foo/mid.qcow2' in the backing chain of 'foo/top.qcow2' $ ./qemu-img commit -b $PWD/foo/mid.qcow2 foo/top.qcow2 Image committed. With this change, bdrv_find_backing_image() must look at whether the user has overridden a BDS's backing file. If so, it can no longer use bs->backing_file, but must instead compare the given filename against the backing node's filename directly. Note that this changes the QAPI output for a node's backing_file. We had very inconsistent output there (sometimes what the image header said, sometimes the actual filename of the backing image). This inconsistent output was effectively useless, so we have to decide one way or the other. Considering that bs->backing_file usually at runtime contained the path to the image relative to qemu's CWD (or absolute), this patch changes QAPI's backing_file to always report the bs->backing->bs->filename from now on. If you want to receive the image header information, you have to refer to full-backing-filename. This necessitates a change to iotest 228. The interesting information it really wanted is the image header, and it can get that now, but it has to use full-backing-filename instead of backing_file. Because of this patch's changes to bs->backing_file's behavior, we also need some reference output changes. Along with the changes to bs->backing_file, stop updating BDS.backing_format in bdrv_backing_attach() as well. This way, ImageInfo's backing-filename and backing-filename-format fields will represent what the image header says and nothing else. iotest 245 changes in behavior: With the backing node no longer overriding the parent node's backing_file string, you can now omit the @backing option when reopening a node with neither a default nor a current backing file even if it used to have a backing node at some point. 273 also changes: The base image is opened without a format layer, so ImageInfo.backing-filename-format used to report "file" for the base image's overlay after blockdev-snapshot. However, the image header never says "file" anywhere, so it now reports $IMGFMT. Signed-off-by: Max Reitz <mreitz@redhat.com>
2018-08-01 20:34:11 +02:00
bool filenames_refreshed = false;
BlockDriverState *curr_bs = NULL;
BlockDriverState *retval = NULL;
BlockDriverState *bs_below;
GLOBAL_STATE_CODE();
if (!bs || !bs->drv || !backing_file) {
return NULL;
}
filename_full = g_malloc(PATH_MAX);
backing_file_full = g_malloc(PATH_MAX);
is_protocol = path_has_protocol(backing_file);
/*
* Being largely a legacy function, skip any filters here
* (because filters do not have normal filenames, so they cannot
* match anyway; and allowing json:{} filenames is a bit out of
* scope).
*/
for (curr_bs = bdrv_skip_filters(bs);
bdrv_cow_child(curr_bs) != NULL;
curr_bs = bs_below)
{
bs_below = bdrv_backing_chain_next(curr_bs);
block: Leave BDS.backing_{file,format} constant Parts of the block layer treat BDS.backing_file as if it were whatever the image header says (i.e., if it is a relative path, it is relative to the overlay), other parts treat it like a cache for bs->backing->bs->filename (relative paths are relative to the CWD). Considering bs->backing->bs->filename exists, let us make it mean the former. Among other things, this now allows the user to specify a base when using qemu-img to commit an image file in a directory that is not the CWD (assuming, everything uses relative filenames). Before this patch: $ ./qemu-img create -f qcow2 foo/bot.qcow2 1M $ ./qemu-img create -f qcow2 -b bot.qcow2 foo/mid.qcow2 $ ./qemu-img create -f qcow2 -b mid.qcow2 foo/top.qcow2 $ ./qemu-img commit -b mid.qcow2 foo/top.qcow2 qemu-img: Did not find 'mid.qcow2' in the backing chain of 'foo/top.qcow2' $ ./qemu-img commit -b foo/mid.qcow2 foo/top.qcow2 qemu-img: Did not find 'foo/mid.qcow2' in the backing chain of 'foo/top.qcow2' $ ./qemu-img commit -b $PWD/foo/mid.qcow2 foo/top.qcow2 qemu-img: Did not find '[...]/foo/mid.qcow2' in the backing chain of 'foo/top.qcow2' After this patch: $ ./qemu-img commit -b mid.qcow2 foo/top.qcow2 Image committed. $ ./qemu-img commit -b foo/mid.qcow2 foo/top.qcow2 qemu-img: Did not find 'foo/mid.qcow2' in the backing chain of 'foo/top.qcow2' $ ./qemu-img commit -b $PWD/foo/mid.qcow2 foo/top.qcow2 Image committed. With this change, bdrv_find_backing_image() must look at whether the user has overridden a BDS's backing file. If so, it can no longer use bs->backing_file, but must instead compare the given filename against the backing node's filename directly. Note that this changes the QAPI output for a node's backing_file. We had very inconsistent output there (sometimes what the image header said, sometimes the actual filename of the backing image). This inconsistent output was effectively useless, so we have to decide one way or the other. Considering that bs->backing_file usually at runtime contained the path to the image relative to qemu's CWD (or absolute), this patch changes QAPI's backing_file to always report the bs->backing->bs->filename from now on. If you want to receive the image header information, you have to refer to full-backing-filename. This necessitates a change to iotest 228. The interesting information it really wanted is the image header, and it can get that now, but it has to use full-backing-filename instead of backing_file. Because of this patch's changes to bs->backing_file's behavior, we also need some reference output changes. Along with the changes to bs->backing_file, stop updating BDS.backing_format in bdrv_backing_attach() as well. This way, ImageInfo's backing-filename and backing-filename-format fields will represent what the image header says and nothing else. iotest 245 changes in behavior: With the backing node no longer overriding the parent node's backing_file string, you can now omit the @backing option when reopening a node with neither a default nor a current backing file even if it used to have a backing node at some point. 273 also changes: The base image is opened without a format layer, so ImageInfo.backing-filename-format used to report "file" for the base image's overlay after blockdev-snapshot. However, the image header never says "file" anywhere, so it now reports $IMGFMT. Signed-off-by: Max Reitz <mreitz@redhat.com>
2018-08-01 20:34:11 +02:00
if (bdrv_backing_overridden(curr_bs)) {
/*
* If the backing file was overridden, we can only compare
* directly against the backing node's filename.
*/
if (!filenames_refreshed) {
/*
* This will automatically refresh all of the
* filenames in the rest of the backing chain, so we
* only need to do this once.
*/
bdrv_refresh_filename(bs_below);
filenames_refreshed = true;
}
if (strcmp(backing_file, bs_below->filename) == 0) {
retval = bs_below;
break;
}
} else if (is_protocol || path_has_protocol(curr_bs->backing_file)) {
/*
* If either of the filename paths is actually a protocol, then
* compare unmodified paths; otherwise make paths relative.
*/
char *backing_file_full_ret;
if (strcmp(backing_file, curr_bs->backing_file) == 0) {
retval = bs_below;
break;
}
/* Also check against the full backing filename for the image */
backing_file_full_ret = bdrv_get_full_backing_filename(curr_bs,
NULL);
if (backing_file_full_ret) {
bool equal = strcmp(backing_file, backing_file_full_ret) == 0;
g_free(backing_file_full_ret);
if (equal) {
retval = bs_below;
break;
}
}
} else {
/* If not an absolute filename path, make it relative to the current
* image's filename path */
filename_tmp = bdrv_make_absolute_filename(curr_bs, backing_file,
NULL);
/* We are going to compare canonicalized absolute pathnames */
if (!filename_tmp || !realpath(filename_tmp, filename_full)) {
g_free(filename_tmp);
continue;
}
g_free(filename_tmp);
/* We need to make sure the backing filename we are comparing against
* is relative to the current image filename (or absolute) */
filename_tmp = bdrv_get_full_backing_filename(curr_bs, NULL);
if (!filename_tmp || !realpath(filename_tmp, backing_file_full)) {
g_free(filename_tmp);
continue;
}
g_free(filename_tmp);
if (strcmp(backing_file_full, filename_full) == 0) {
retval = bs_below;
break;
}
}
}
g_free(filename_full);
g_free(backing_file_full);
return retval;
}
void bdrv_init(void)
{
#ifdef CONFIG_BDRV_WHITELIST_TOOLS
use_bdrv_whitelist = 1;
#endif
module_call_init(MODULE_INIT_BLOCK);
}
void bdrv_init_with_whitelist(void)
{
use_bdrv_whitelist = 1;
bdrv_init();
}
int bdrv_activate(BlockDriverState *bs, Error **errp)
{
BdrvChild *child, *parent;
Error *local_err = NULL;
int ret;
dirty-bitmaps: clean-up bitmaps loading and migration logic This patch aims to bring the following behavior: 1. We don't load bitmaps, when started in inactive mode. It's the case of incoming migration. In this case we wait for bitmaps migration through migration channel (if 'dirty-bitmaps' capability is enabled) or for invalidation (to load bitmaps from the image). 2. We don't remove persistent bitmaps on inactivation. Instead, we only remove bitmaps after storing. This is the only way to restore bitmaps, if we decided to resume source after [failed] migration with 'dirty-bitmaps' capability enabled (which means, that bitmaps were not stored). 3. We load bitmaps on open and any invalidation, it's ok for all cases: - normal open - migration target invalidation with dirty-bitmaps capability (bitmaps are migrating through migration channel, the are not stored, so they should have IN_USE flag set and will be skipped when loading. However, it would fail if bitmaps are read-only[1]) - migration target invalidation without dirty-bitmaps capability (normal load of the bitmaps, if migrated with shared storage) - source invalidation with dirty-bitmaps capability (skip because IN_USE) - source invalidation without dirty-bitmaps capability (bitmaps were dropped, reload them) [1]: to accurately handle this, migration of read-only bitmaps is explicitly forbidden in this patch. New mechanism for not storing bitmaps when migrate with dirty-bitmaps capability is introduced: migration filed in BdrvDirtyBitmap. Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Signed-off-by: John Snow <jsnow@redhat.com>
2018-10-29 21:23:17 +01:00
BdrvDirtyBitmap *bm;
GLOBAL_STATE_CODE();
if (!bs->drv) {
return -ENOMEDIUM;
}
QLIST_FOREACH(child, &bs->children, next) {
bdrv_activate(child->bs, &local_err);
if (local_err) {
error_propagate(errp, local_err);
return -EINVAL;
}
}
/*
* Update permissions, they may differ for inactive nodes.
*
* Note that the required permissions of inactive images are always a
* subset of the permissions required after activating the image. This
* allows us to just get the permissions upfront without restricting
* bdrv_co_invalidate_cache().
*
* It also means that in error cases, we don't have to try and revert to
* the old permissions (which is an operation that could fail, too). We can
* just keep the extended permissions for the next time that an activation
* of the image is tried.
*/
if (bs->open_flags & BDRV_O_INACTIVE) {
bs->open_flags &= ~BDRV_O_INACTIVE;
ret = bdrv_refresh_perms(bs, errp);
if (ret < 0) {
bs->open_flags |= BDRV_O_INACTIVE;
return ret;
}
ret = bdrv_invalidate_cache(bs, errp);
if (ret < 0) {
bs->open_flags |= BDRV_O_INACTIVE;
return ret;
}
dirty-bitmaps: clean-up bitmaps loading and migration logic This patch aims to bring the following behavior: 1. We don't load bitmaps, when started in inactive mode. It's the case of incoming migration. In this case we wait for bitmaps migration through migration channel (if 'dirty-bitmaps' capability is enabled) or for invalidation (to load bitmaps from the image). 2. We don't remove persistent bitmaps on inactivation. Instead, we only remove bitmaps after storing. This is the only way to restore bitmaps, if we decided to resume source after [failed] migration with 'dirty-bitmaps' capability enabled (which means, that bitmaps were not stored). 3. We load bitmaps on open and any invalidation, it's ok for all cases: - normal open - migration target invalidation with dirty-bitmaps capability (bitmaps are migrating through migration channel, the are not stored, so they should have IN_USE flag set and will be skipped when loading. However, it would fail if bitmaps are read-only[1]) - migration target invalidation without dirty-bitmaps capability (normal load of the bitmaps, if migrated with shared storage) - source invalidation with dirty-bitmaps capability (skip because IN_USE) - source invalidation without dirty-bitmaps capability (bitmaps were dropped, reload them) [1]: to accurately handle this, migration of read-only bitmaps is explicitly forbidden in this patch. New mechanism for not storing bitmaps when migrate with dirty-bitmaps capability is introduced: migration filed in BdrvDirtyBitmap. Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Signed-off-by: John Snow <jsnow@redhat.com>
2018-10-29 21:23:17 +01:00
FOR_EACH_DIRTY_BITMAP(bs, bm) {
bdrv_dirty_bitmap_skip_store(bm, false);
}
ret = refresh_total_sectors(bs, bs->total_sectors);
if (ret < 0) {
bs->open_flags |= BDRV_O_INACTIVE;
error_setg_errno(errp, -ret, "Could not refresh total sector count");
return ret;
}
}
QLIST_FOREACH(parent, &bs->parents, next_parent) {
if (parent->klass->activate) {
parent->klass->activate(parent, &local_err);
if (local_err) {
block: Fix invalidate_cache error path for parent activation bdrv_co_invalidate_cache() clears the BDRV_O_INACTIVE flag before actually activating a node so that the correct permissions etc. are taken. In case of errors, the flag must be restored so that the next call to bdrv_co_invalidate_cache() retries activation. Restoring the flag was missing in the error path for a failed parent->role->activate() call. The consequence is that this attempt to activate all images correctly fails because we still set errp, however on the next attempt BDRV_O_INACTIVE is already clear, so we return success without actually retrying the failed action. An example where this is observable in practice is migration to a QEMU instance that has a raw format block node attached to a guest device with share-rw=off (the default) while another process holds BLK_PERM_WRITE for the same image. In this case, all activation steps before parent->role->activate() succeed because raw can tolerate other writers to the image. Only the parent callback (in particular blk_root_activate()) tries to implement the share-rw=on property and requests exclusive write permissions. This fails when the migration completes and correctly displays an error. However, a manual 'cont' will incorrectly resume the VM without calling blk_root_activate() again. This case is described in more detail in the following bug report: https://bugzilla.redhat.com/show_bug.cgi?id=1531888 Fix this by correctly restoring the BDRV_O_INACTIVE flag in the error path. Cc: qemu-stable@nongnu.org Signed-off-by: Kevin Wolf <kwolf@redhat.com> Tested-by: Markus Armbruster <armbru@redhat.com> Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
2019-01-31 15:16:10 +01:00
bs->open_flags |= BDRV_O_INACTIVE;
error_propagate(errp, local_err);
return -EINVAL;
}
}
}
return 0;
}
int coroutine_fn bdrv_co_invalidate_cache(BlockDriverState *bs, Error **errp)
{
Error *local_err = NULL;
IO_CODE();
assert(!(bs->open_flags & BDRV_O_INACTIVE));
if (bs->drv->bdrv_co_invalidate_cache) {
bs->drv->bdrv_co_invalidate_cache(bs, &local_err);
if (local_err) {
error_propagate(errp, local_err);
return -EINVAL;
}
}
return 0;
}
void bdrv_activate_all(Error **errp)
{
BlockDriverState *bs;
BdrvNextIterator it;
GLOBAL_STATE_CODE();
for (bs = bdrv_first(&it); bs; bs = bdrv_next(&it)) {
AioContext *aio_context = bdrv_get_aio_context(bs);
int ret;
aio_context_acquire(aio_context);
ret = bdrv_activate(bs, errp);
aio_context_release(aio_context);
if (ret < 0) {
bdrv_next_cleanup(&it);
return;
}
}
}
block: Don't inactivate children before parents bdrv_child_cb_inactivate() asserts that parents are already inactive when children get inactivated. This precondition is necessary because parents could still issue requests in their inactivation code. When block nodes are created individually with -blockdev, all of them are monitor owned and will be returned by bdrv_next() in an undefined order (in practice, in the order of their creation, which is usually children before parents), which obviously fails the assertion: qemu: block.c:899: bdrv_child_cb_inactivate: Assertion `bs->open_flags & BDRV_O_INACTIVE' failed. This patch fixes the ordering by skipping nodes with still active parents in bdrv_inactivate_recurse() because we know that they will be covered by recursion when the last active parent becomes inactive. With the correct parents-before-children ordering, we also got rid of the reason why commit aad0b7a0bfb introduced two passes, so we can go back to a single-pass recursion. This is necessary so we can rely on the BDRV_O_INACTIVE flag to skip nodes with active parents (the flag used to be set only in pass 2, so we would always skip non-root nodes in pass 1 because all parents would still be considered active; setting the flag in pass 1 would mean, that we never skip anything in pass 2 because all parents are already considered inactive). Because of the change to single pass, this patch is best reviewed with whitespace changes ignored. Signed-off-by: Kevin Wolf <kwolf@redhat.com> Reviewed-by: Max Reitz <mreitz@redhat.com>
2018-11-23 15:11:14 +01:00
static bool bdrv_has_bds_parent(BlockDriverState *bs, bool only_active)
{
BdrvChild *parent;
GLOBAL_STATE_CODE();
block: Don't inactivate children before parents bdrv_child_cb_inactivate() asserts that parents are already inactive when children get inactivated. This precondition is necessary because parents could still issue requests in their inactivation code. When block nodes are created individually with -blockdev, all of them are monitor owned and will be returned by bdrv_next() in an undefined order (in practice, in the order of their creation, which is usually children before parents), which obviously fails the assertion: qemu: block.c:899: bdrv_child_cb_inactivate: Assertion `bs->open_flags & BDRV_O_INACTIVE' failed. This patch fixes the ordering by skipping nodes with still active parents in bdrv_inactivate_recurse() because we know that they will be covered by recursion when the last active parent becomes inactive. With the correct parents-before-children ordering, we also got rid of the reason why commit aad0b7a0bfb introduced two passes, so we can go back to a single-pass recursion. This is necessary so we can rely on the BDRV_O_INACTIVE flag to skip nodes with active parents (the flag used to be set only in pass 2, so we would always skip non-root nodes in pass 1 because all parents would still be considered active; setting the flag in pass 1 would mean, that we never skip anything in pass 2 because all parents are already considered inactive). Because of the change to single pass, this patch is best reviewed with whitespace changes ignored. Signed-off-by: Kevin Wolf <kwolf@redhat.com> Reviewed-by: Max Reitz <mreitz@redhat.com>
2018-11-23 15:11:14 +01:00
QLIST_FOREACH(parent, &bs->parents, next_parent) {
if (parent->klass->parent_is_bds) {
block: Don't inactivate children before parents bdrv_child_cb_inactivate() asserts that parents are already inactive when children get inactivated. This precondition is necessary because parents could still issue requests in their inactivation code. When block nodes are created individually with -blockdev, all of them are monitor owned and will be returned by bdrv_next() in an undefined order (in practice, in the order of their creation, which is usually children before parents), which obviously fails the assertion: qemu: block.c:899: bdrv_child_cb_inactivate: Assertion `bs->open_flags & BDRV_O_INACTIVE' failed. This patch fixes the ordering by skipping nodes with still active parents in bdrv_inactivate_recurse() because we know that they will be covered by recursion when the last active parent becomes inactive. With the correct parents-before-children ordering, we also got rid of the reason why commit aad0b7a0bfb introduced two passes, so we can go back to a single-pass recursion. This is necessary so we can rely on the BDRV_O_INACTIVE flag to skip nodes with active parents (the flag used to be set only in pass 2, so we would always skip non-root nodes in pass 1 because all parents would still be considered active; setting the flag in pass 1 would mean, that we never skip anything in pass 2 because all parents are already considered inactive). Because of the change to single pass, this patch is best reviewed with whitespace changes ignored. Signed-off-by: Kevin Wolf <kwolf@redhat.com> Reviewed-by: Max Reitz <mreitz@redhat.com>
2018-11-23 15:11:14 +01:00
BlockDriverState *parent_bs = parent->opaque;
if (!only_active || !(parent_bs->open_flags & BDRV_O_INACTIVE)) {
return true;
}
}
}
return false;
}
static int bdrv_inactivate_recurse(BlockDriverState *bs)
block: Inactivate BDS when migration completes So far, live migration with shared storage meant that the image is in a not-really-ready don't-touch-me state on the destination while the source is still actively using it, but after completing the migration, the image was fully opened on both sides. This is bad. This patch adds a block driver callback to inactivate images on the source before completing the migration. Inactivation means that it goes to a state as if it was just live migrated to the qemu instance on the source (i.e. BDRV_O_INACTIVE is set). You're then supposed to continue either on the source or on the destination, which takes ownership of the image. A typical migration looks like this now with respect to disk images: 1. Destination qemu is started, the image is opened with BDRV_O_INACTIVE. The image is fully opened on the source. 2. Migration is about to complete. The source flushes the image and inactivates it. Now both sides have the image opened with BDRV_O_INACTIVE and are expecting the other side to still modify it. 3. One side (the destination on success) continues and calls bdrv_invalidate_all() in order to take ownership of the image again. This removes BDRV_O_INACTIVE on the resuming side; the flag remains set on the other side. This ensures that the same image isn't written to by both instances (unless both are resumed, but then you get what you deserve). This is important because .bdrv_close for non-BDRV_O_INACTIVE images could write to the image file, which is definitely forbidden while another host is using the image. Signed-off-by: Kevin Wolf <kwolf@redhat.com> Reviewed-by: Eric Blake <eblake@redhat.com> Reviewed-by: John Snow <jsnow@redhat.com>
2015-12-22 14:07:08 +01:00
{
BdrvChild *child, *parent;
block: Inactivate BDS when migration completes So far, live migration with shared storage meant that the image is in a not-really-ready don't-touch-me state on the destination while the source is still actively using it, but after completing the migration, the image was fully opened on both sides. This is bad. This patch adds a block driver callback to inactivate images on the source before completing the migration. Inactivation means that it goes to a state as if it was just live migrated to the qemu instance on the source (i.e. BDRV_O_INACTIVE is set). You're then supposed to continue either on the source or on the destination, which takes ownership of the image. A typical migration looks like this now with respect to disk images: 1. Destination qemu is started, the image is opened with BDRV_O_INACTIVE. The image is fully opened on the source. 2. Migration is about to complete. The source flushes the image and inactivates it. Now both sides have the image opened with BDRV_O_INACTIVE and are expecting the other side to still modify it. 3. One side (the destination on success) continues and calls bdrv_invalidate_all() in order to take ownership of the image again. This removes BDRV_O_INACTIVE on the resuming side; the flag remains set on the other side. This ensures that the same image isn't written to by both instances (unless both are resumed, but then you get what you deserve). This is important because .bdrv_close for non-BDRV_O_INACTIVE images could write to the image file, which is definitely forbidden while another host is using the image. Signed-off-by: Kevin Wolf <kwolf@redhat.com> Reviewed-by: Eric Blake <eblake@redhat.com> Reviewed-by: John Snow <jsnow@redhat.com>
2015-12-22 14:07:08 +01:00
int ret;
uint64_t cumulative_perms, cumulative_shared_perms;
block: Inactivate BDS when migration completes So far, live migration with shared storage meant that the image is in a not-really-ready don't-touch-me state on the destination while the source is still actively using it, but after completing the migration, the image was fully opened on both sides. This is bad. This patch adds a block driver callback to inactivate images on the source before completing the migration. Inactivation means that it goes to a state as if it was just live migrated to the qemu instance on the source (i.e. BDRV_O_INACTIVE is set). You're then supposed to continue either on the source or on the destination, which takes ownership of the image. A typical migration looks like this now with respect to disk images: 1. Destination qemu is started, the image is opened with BDRV_O_INACTIVE. The image is fully opened on the source. 2. Migration is about to complete. The source flushes the image and inactivates it. Now both sides have the image opened with BDRV_O_INACTIVE and are expecting the other side to still modify it. 3. One side (the destination on success) continues and calls bdrv_invalidate_all() in order to take ownership of the image again. This removes BDRV_O_INACTIVE on the resuming side; the flag remains set on the other side. This ensures that the same image isn't written to by both instances (unless both are resumed, but then you get what you deserve). This is important because .bdrv_close for non-BDRV_O_INACTIVE images could write to the image file, which is definitely forbidden while another host is using the image. Signed-off-by: Kevin Wolf <kwolf@redhat.com> Reviewed-by: Eric Blake <eblake@redhat.com> Reviewed-by: John Snow <jsnow@redhat.com>
2015-12-22 14:07:08 +01:00
GLOBAL_STATE_CODE();
if (!bs->drv) {
return -ENOMEDIUM;
}
block: Don't inactivate children before parents bdrv_child_cb_inactivate() asserts that parents are already inactive when children get inactivated. This precondition is necessary because parents could still issue requests in their inactivation code. When block nodes are created individually with -blockdev, all of them are monitor owned and will be returned by bdrv_next() in an undefined order (in practice, in the order of their creation, which is usually children before parents), which obviously fails the assertion: qemu: block.c:899: bdrv_child_cb_inactivate: Assertion `bs->open_flags & BDRV_O_INACTIVE' failed. This patch fixes the ordering by skipping nodes with still active parents in bdrv_inactivate_recurse() because we know that they will be covered by recursion when the last active parent becomes inactive. With the correct parents-before-children ordering, we also got rid of the reason why commit aad0b7a0bfb introduced two passes, so we can go back to a single-pass recursion. This is necessary so we can rely on the BDRV_O_INACTIVE flag to skip nodes with active parents (the flag used to be set only in pass 2, so we would always skip non-root nodes in pass 1 because all parents would still be considered active; setting the flag in pass 1 would mean, that we never skip anything in pass 2 because all parents are already considered inactive). Because of the change to single pass, this patch is best reviewed with whitespace changes ignored. Signed-off-by: Kevin Wolf <kwolf@redhat.com> Reviewed-by: Max Reitz <mreitz@redhat.com>
2018-11-23 15:11:14 +01:00
/* Make sure that we don't inactivate a child before its parent.
* It will be covered by recursion from the yet active parent. */
if (bdrv_has_bds_parent(bs, true)) {
return 0;
}
assert(!(bs->open_flags & BDRV_O_INACTIVE));
/* Inactivate this node */
if (bs->drv->bdrv_inactivate) {
block: Inactivate BDS when migration completes So far, live migration with shared storage meant that the image is in a not-really-ready don't-touch-me state on the destination while the source is still actively using it, but after completing the migration, the image was fully opened on both sides. This is bad. This patch adds a block driver callback to inactivate images on the source before completing the migration. Inactivation means that it goes to a state as if it was just live migrated to the qemu instance on the source (i.e. BDRV_O_INACTIVE is set). You're then supposed to continue either on the source or on the destination, which takes ownership of the image. A typical migration looks like this now with respect to disk images: 1. Destination qemu is started, the image is opened with BDRV_O_INACTIVE. The image is fully opened on the source. 2. Migration is about to complete. The source flushes the image and inactivates it. Now both sides have the image opened with BDRV_O_INACTIVE and are expecting the other side to still modify it. 3. One side (the destination on success) continues and calls bdrv_invalidate_all() in order to take ownership of the image again. This removes BDRV_O_INACTIVE on the resuming side; the flag remains set on the other side. This ensures that the same image isn't written to by both instances (unless both are resumed, but then you get what you deserve). This is important because .bdrv_close for non-BDRV_O_INACTIVE images could write to the image file, which is definitely forbidden while another host is using the image. Signed-off-by: Kevin Wolf <kwolf@redhat.com> Reviewed-by: Eric Blake <eblake@redhat.com> Reviewed-by: John Snow <jsnow@redhat.com>
2015-12-22 14:07:08 +01:00
ret = bs->drv->bdrv_inactivate(bs);
if (ret < 0) {
return ret;
}
}
block: Don't inactivate children before parents bdrv_child_cb_inactivate() asserts that parents are already inactive when children get inactivated. This precondition is necessary because parents could still issue requests in their inactivation code. When block nodes are created individually with -blockdev, all of them are monitor owned and will be returned by bdrv_next() in an undefined order (in practice, in the order of their creation, which is usually children before parents), which obviously fails the assertion: qemu: block.c:899: bdrv_child_cb_inactivate: Assertion `bs->open_flags & BDRV_O_INACTIVE' failed. This patch fixes the ordering by skipping nodes with still active parents in bdrv_inactivate_recurse() because we know that they will be covered by recursion when the last active parent becomes inactive. With the correct parents-before-children ordering, we also got rid of the reason why commit aad0b7a0bfb introduced two passes, so we can go back to a single-pass recursion. This is necessary so we can rely on the BDRV_O_INACTIVE flag to skip nodes with active parents (the flag used to be set only in pass 2, so we would always skip non-root nodes in pass 1 because all parents would still be considered active; setting the flag in pass 1 would mean, that we never skip anything in pass 2 because all parents are already considered inactive). Because of the change to single pass, this patch is best reviewed with whitespace changes ignored. Signed-off-by: Kevin Wolf <kwolf@redhat.com> Reviewed-by: Max Reitz <mreitz@redhat.com>
2018-11-23 15:11:14 +01:00
QLIST_FOREACH(parent, &bs->parents, next_parent) {
if (parent->klass->inactivate) {
ret = parent->klass->inactivate(parent);
block: Don't inactivate children before parents bdrv_child_cb_inactivate() asserts that parents are already inactive when children get inactivated. This precondition is necessary because parents could still issue requests in their inactivation code. When block nodes are created individually with -blockdev, all of them are monitor owned and will be returned by bdrv_next() in an undefined order (in practice, in the order of their creation, which is usually children before parents), which obviously fails the assertion: qemu: block.c:899: bdrv_child_cb_inactivate: Assertion `bs->open_flags & BDRV_O_INACTIVE' failed. This patch fixes the ordering by skipping nodes with still active parents in bdrv_inactivate_recurse() because we know that they will be covered by recursion when the last active parent becomes inactive. With the correct parents-before-children ordering, we also got rid of the reason why commit aad0b7a0bfb introduced two passes, so we can go back to a single-pass recursion. This is necessary so we can rely on the BDRV_O_INACTIVE flag to skip nodes with active parents (the flag used to be set only in pass 2, so we would always skip non-root nodes in pass 1 because all parents would still be considered active; setting the flag in pass 1 would mean, that we never skip anything in pass 2 because all parents are already considered inactive). Because of the change to single pass, this patch is best reviewed with whitespace changes ignored. Signed-off-by: Kevin Wolf <kwolf@redhat.com> Reviewed-by: Max Reitz <mreitz@redhat.com>
2018-11-23 15:11:14 +01:00
if (ret < 0) {
return ret;
}
}
block: Don't inactivate children before parents bdrv_child_cb_inactivate() asserts that parents are already inactive when children get inactivated. This precondition is necessary because parents could still issue requests in their inactivation code. When block nodes are created individually with -blockdev, all of them are monitor owned and will be returned by bdrv_next() in an undefined order (in practice, in the order of their creation, which is usually children before parents), which obviously fails the assertion: qemu: block.c:899: bdrv_child_cb_inactivate: Assertion `bs->open_flags & BDRV_O_INACTIVE' failed. This patch fixes the ordering by skipping nodes with still active parents in bdrv_inactivate_recurse() because we know that they will be covered by recursion when the last active parent becomes inactive. With the correct parents-before-children ordering, we also got rid of the reason why commit aad0b7a0bfb introduced two passes, so we can go back to a single-pass recursion. This is necessary so we can rely on the BDRV_O_INACTIVE flag to skip nodes with active parents (the flag used to be set only in pass 2, so we would always skip non-root nodes in pass 1 because all parents would still be considered active; setting the flag in pass 1 would mean, that we never skip anything in pass 2 because all parents are already considered inactive). Because of the change to single pass, this patch is best reviewed with whitespace changes ignored. Signed-off-by: Kevin Wolf <kwolf@redhat.com> Reviewed-by: Max Reitz <mreitz@redhat.com>
2018-11-23 15:11:14 +01:00
}
bdrv_get_cumulative_perm(bs, &cumulative_perms,
&cumulative_shared_perms);
if (cumulative_perms & (BLK_PERM_WRITE | BLK_PERM_WRITE_UNCHANGED)) {
/* Our inactive parents still need write access. Inactivation failed. */
return -EPERM;
}
block: Don't inactivate children before parents bdrv_child_cb_inactivate() asserts that parents are already inactive when children get inactivated. This precondition is necessary because parents could still issue requests in their inactivation code. When block nodes are created individually with -blockdev, all of them are monitor owned and will be returned by bdrv_next() in an undefined order (in practice, in the order of their creation, which is usually children before parents), which obviously fails the assertion: qemu: block.c:899: bdrv_child_cb_inactivate: Assertion `bs->open_flags & BDRV_O_INACTIVE' failed. This patch fixes the ordering by skipping nodes with still active parents in bdrv_inactivate_recurse() because we know that they will be covered by recursion when the last active parent becomes inactive. With the correct parents-before-children ordering, we also got rid of the reason why commit aad0b7a0bfb introduced two passes, so we can go back to a single-pass recursion. This is necessary so we can rely on the BDRV_O_INACTIVE flag to skip nodes with active parents (the flag used to be set only in pass 2, so we would always skip non-root nodes in pass 1 because all parents would still be considered active; setting the flag in pass 1 would mean, that we never skip anything in pass 2 because all parents are already considered inactive). Because of the change to single pass, this patch is best reviewed with whitespace changes ignored. Signed-off-by: Kevin Wolf <kwolf@redhat.com> Reviewed-by: Max Reitz <mreitz@redhat.com>
2018-11-23 15:11:14 +01:00
bs->open_flags |= BDRV_O_INACTIVE;
/*
* Update permissions, they may differ for inactive nodes.
* We only tried to loosen restrictions, so errors are not fatal, ignore
* them.
*/
bdrv_refresh_perms(bs, NULL);
block: Don't inactivate children before parents bdrv_child_cb_inactivate() asserts that parents are already inactive when children get inactivated. This precondition is necessary because parents could still issue requests in their inactivation code. When block nodes are created individually with -blockdev, all of them are monitor owned and will be returned by bdrv_next() in an undefined order (in practice, in the order of their creation, which is usually children before parents), which obviously fails the assertion: qemu: block.c:899: bdrv_child_cb_inactivate: Assertion `bs->open_flags & BDRV_O_INACTIVE' failed. This patch fixes the ordering by skipping nodes with still active parents in bdrv_inactivate_recurse() because we know that they will be covered by recursion when the last active parent becomes inactive. With the correct parents-before-children ordering, we also got rid of the reason why commit aad0b7a0bfb introduced two passes, so we can go back to a single-pass recursion. This is necessary so we can rely on the BDRV_O_INACTIVE flag to skip nodes with active parents (the flag used to be set only in pass 2, so we would always skip non-root nodes in pass 1 because all parents would still be considered active; setting the flag in pass 1 would mean, that we never skip anything in pass 2 because all parents are already considered inactive). Because of the change to single pass, this patch is best reviewed with whitespace changes ignored. Signed-off-by: Kevin Wolf <kwolf@redhat.com> Reviewed-by: Max Reitz <mreitz@redhat.com>
2018-11-23 15:11:14 +01:00
/* Recursively inactivate children */
QLIST_FOREACH(child, &bs->children, next) {
block: Don't inactivate children before parents bdrv_child_cb_inactivate() asserts that parents are already inactive when children get inactivated. This precondition is necessary because parents could still issue requests in their inactivation code. When block nodes are created individually with -blockdev, all of them are monitor owned and will be returned by bdrv_next() in an undefined order (in practice, in the order of their creation, which is usually children before parents), which obviously fails the assertion: qemu: block.c:899: bdrv_child_cb_inactivate: Assertion `bs->open_flags & BDRV_O_INACTIVE' failed. This patch fixes the ordering by skipping nodes with still active parents in bdrv_inactivate_recurse() because we know that they will be covered by recursion when the last active parent becomes inactive. With the correct parents-before-children ordering, we also got rid of the reason why commit aad0b7a0bfb introduced two passes, so we can go back to a single-pass recursion. This is necessary so we can rely on the BDRV_O_INACTIVE flag to skip nodes with active parents (the flag used to be set only in pass 2, so we would always skip non-root nodes in pass 1 because all parents would still be considered active; setting the flag in pass 1 would mean, that we never skip anything in pass 2 because all parents are already considered inactive). Because of the change to single pass, this patch is best reviewed with whitespace changes ignored. Signed-off-by: Kevin Wolf <kwolf@redhat.com> Reviewed-by: Max Reitz <mreitz@redhat.com>
2018-11-23 15:11:14 +01:00
ret = bdrv_inactivate_recurse(child->bs);
if (ret < 0) {
return ret;
}
}
block: Inactivate BDS when migration completes So far, live migration with shared storage meant that the image is in a not-really-ready don't-touch-me state on the destination while the source is still actively using it, but after completing the migration, the image was fully opened on both sides. This is bad. This patch adds a block driver callback to inactivate images on the source before completing the migration. Inactivation means that it goes to a state as if it was just live migrated to the qemu instance on the source (i.e. BDRV_O_INACTIVE is set). You're then supposed to continue either on the source or on the destination, which takes ownership of the image. A typical migration looks like this now with respect to disk images: 1. Destination qemu is started, the image is opened with BDRV_O_INACTIVE. The image is fully opened on the source. 2. Migration is about to complete. The source flushes the image and inactivates it. Now both sides have the image opened with BDRV_O_INACTIVE and are expecting the other side to still modify it. 3. One side (the destination on success) continues and calls bdrv_invalidate_all() in order to take ownership of the image again. This removes BDRV_O_INACTIVE on the resuming side; the flag remains set on the other side. This ensures that the same image isn't written to by both instances (unless both are resumed, but then you get what you deserve). This is important because .bdrv_close for non-BDRV_O_INACTIVE images could write to the image file, which is definitely forbidden while another host is using the image. Signed-off-by: Kevin Wolf <kwolf@redhat.com> Reviewed-by: Eric Blake <eblake@redhat.com> Reviewed-by: John Snow <jsnow@redhat.com>
2015-12-22 14:07:08 +01:00
return 0;
}
int bdrv_inactivate_all(void)
{
BlockDriverState *bs = NULL;
BdrvNextIterator it;
int ret = 0;
GSList *aio_ctxs = NULL, *ctx;
block: Inactivate BDS when migration completes So far, live migration with shared storage meant that the image is in a not-really-ready don't-touch-me state on the destination while the source is still actively using it, but after completing the migration, the image was fully opened on both sides. This is bad. This patch adds a block driver callback to inactivate images on the source before completing the migration. Inactivation means that it goes to a state as if it was just live migrated to the qemu instance on the source (i.e. BDRV_O_INACTIVE is set). You're then supposed to continue either on the source or on the destination, which takes ownership of the image. A typical migration looks like this now with respect to disk images: 1. Destination qemu is started, the image is opened with BDRV_O_INACTIVE. The image is fully opened on the source. 2. Migration is about to complete. The source flushes the image and inactivates it. Now both sides have the image opened with BDRV_O_INACTIVE and are expecting the other side to still modify it. 3. One side (the destination on success) continues and calls bdrv_invalidate_all() in order to take ownership of the image again. This removes BDRV_O_INACTIVE on the resuming side; the flag remains set on the other side. This ensures that the same image isn't written to by both instances (unless both are resumed, but then you get what you deserve). This is important because .bdrv_close for non-BDRV_O_INACTIVE images could write to the image file, which is definitely forbidden while another host is using the image. Signed-off-by: Kevin Wolf <kwolf@redhat.com> Reviewed-by: Eric Blake <eblake@redhat.com> Reviewed-by: John Snow <jsnow@redhat.com>
2015-12-22 14:07:08 +01:00
GLOBAL_STATE_CODE();
for (bs = bdrv_first(&it); bs; bs = bdrv_next(&it)) {
AioContext *aio_context = bdrv_get_aio_context(bs);
if (!g_slist_find(aio_ctxs, aio_context)) {
aio_ctxs = g_slist_prepend(aio_ctxs, aio_context);
aio_context_acquire(aio_context);
}
}
block: Inactivate BDS when migration completes So far, live migration with shared storage meant that the image is in a not-really-ready don't-touch-me state on the destination while the source is still actively using it, but after completing the migration, the image was fully opened on both sides. This is bad. This patch adds a block driver callback to inactivate images on the source before completing the migration. Inactivation means that it goes to a state as if it was just live migrated to the qemu instance on the source (i.e. BDRV_O_INACTIVE is set). You're then supposed to continue either on the source or on the destination, which takes ownership of the image. A typical migration looks like this now with respect to disk images: 1. Destination qemu is started, the image is opened with BDRV_O_INACTIVE. The image is fully opened on the source. 2. Migration is about to complete. The source flushes the image and inactivates it. Now both sides have the image opened with BDRV_O_INACTIVE and are expecting the other side to still modify it. 3. One side (the destination on success) continues and calls bdrv_invalidate_all() in order to take ownership of the image again. This removes BDRV_O_INACTIVE on the resuming side; the flag remains set on the other side. This ensures that the same image isn't written to by both instances (unless both are resumed, but then you get what you deserve). This is important because .bdrv_close for non-BDRV_O_INACTIVE images could write to the image file, which is definitely forbidden while another host is using the image. Signed-off-by: Kevin Wolf <kwolf@redhat.com> Reviewed-by: Eric Blake <eblake@redhat.com> Reviewed-by: John Snow <jsnow@redhat.com>
2015-12-22 14:07:08 +01:00
block: Don't inactivate children before parents bdrv_child_cb_inactivate() asserts that parents are already inactive when children get inactivated. This precondition is necessary because parents could still issue requests in their inactivation code. When block nodes are created individually with -blockdev, all of them are monitor owned and will be returned by bdrv_next() in an undefined order (in practice, in the order of their creation, which is usually children before parents), which obviously fails the assertion: qemu: block.c:899: bdrv_child_cb_inactivate: Assertion `bs->open_flags & BDRV_O_INACTIVE' failed. This patch fixes the ordering by skipping nodes with still active parents in bdrv_inactivate_recurse() because we know that they will be covered by recursion when the last active parent becomes inactive. With the correct parents-before-children ordering, we also got rid of the reason why commit aad0b7a0bfb introduced two passes, so we can go back to a single-pass recursion. This is necessary so we can rely on the BDRV_O_INACTIVE flag to skip nodes with active parents (the flag used to be set only in pass 2, so we would always skip non-root nodes in pass 1 because all parents would still be considered active; setting the flag in pass 1 would mean, that we never skip anything in pass 2 because all parents are already considered inactive). Because of the change to single pass, this patch is best reviewed with whitespace changes ignored. Signed-off-by: Kevin Wolf <kwolf@redhat.com> Reviewed-by: Max Reitz <mreitz@redhat.com>
2018-11-23 15:11:14 +01:00
for (bs = bdrv_first(&it); bs; bs = bdrv_next(&it)) {
/* Nodes with BDS parents are covered by recursion from the last
* parent that gets inactivated. Don't inactivate them a second
* time if that has already happened. */
if (bdrv_has_bds_parent(bs, false)) {
continue;
}
ret = bdrv_inactivate_recurse(bs);
if (ret < 0) {
bdrv_next_cleanup(&it);
goto out;
block: Inactivate BDS when migration completes So far, live migration with shared storage meant that the image is in a not-really-ready don't-touch-me state on the destination while the source is still actively using it, but after completing the migration, the image was fully opened on both sides. This is bad. This patch adds a block driver callback to inactivate images on the source before completing the migration. Inactivation means that it goes to a state as if it was just live migrated to the qemu instance on the source (i.e. BDRV_O_INACTIVE is set). You're then supposed to continue either on the source or on the destination, which takes ownership of the image. A typical migration looks like this now with respect to disk images: 1. Destination qemu is started, the image is opened with BDRV_O_INACTIVE. The image is fully opened on the source. 2. Migration is about to complete. The source flushes the image and inactivates it. Now both sides have the image opened with BDRV_O_INACTIVE and are expecting the other side to still modify it. 3. One side (the destination on success) continues and calls bdrv_invalidate_all() in order to take ownership of the image again. This removes BDRV_O_INACTIVE on the resuming side; the flag remains set on the other side. This ensures that the same image isn't written to by both instances (unless both are resumed, but then you get what you deserve). This is important because .bdrv_close for non-BDRV_O_INACTIVE images could write to the image file, which is definitely forbidden while another host is using the image. Signed-off-by: Kevin Wolf <kwolf@redhat.com> Reviewed-by: Eric Blake <eblake@redhat.com> Reviewed-by: John Snow <jsnow@redhat.com>
2015-12-22 14:07:08 +01:00
}
}
out:
for (ctx = aio_ctxs; ctx != NULL; ctx = ctx->next) {
AioContext *aio_context = ctx->data;
aio_context_release(aio_context);
}
g_slist_free(aio_ctxs);
return ret;
block: Inactivate BDS when migration completes So far, live migration with shared storage meant that the image is in a not-really-ready don't-touch-me state on the destination while the source is still actively using it, but after completing the migration, the image was fully opened on both sides. This is bad. This patch adds a block driver callback to inactivate images on the source before completing the migration. Inactivation means that it goes to a state as if it was just live migrated to the qemu instance on the source (i.e. BDRV_O_INACTIVE is set). You're then supposed to continue either on the source or on the destination, which takes ownership of the image. A typical migration looks like this now with respect to disk images: 1. Destination qemu is started, the image is opened with BDRV_O_INACTIVE. The image is fully opened on the source. 2. Migration is about to complete. The source flushes the image and inactivates it. Now both sides have the image opened with BDRV_O_INACTIVE and are expecting the other side to still modify it. 3. One side (the destination on success) continues and calls bdrv_invalidate_all() in order to take ownership of the image again. This removes BDRV_O_INACTIVE on the resuming side; the flag remains set on the other side. This ensures that the same image isn't written to by both instances (unless both are resumed, but then you get what you deserve). This is important because .bdrv_close for non-BDRV_O_INACTIVE images could write to the image file, which is definitely forbidden while another host is using the image. Signed-off-by: Kevin Wolf <kwolf@redhat.com> Reviewed-by: Eric Blake <eblake@redhat.com> Reviewed-by: John Snow <jsnow@redhat.com>
2015-12-22 14:07:08 +01:00
}
/**************************************************************/
/* removable device support */
/**
* Return TRUE if the media is present
*/
bool bdrv_is_inserted(BlockDriverState *bs)
{
BlockDriver *drv = bs->drv;
BdrvChild *child;
IO_CODE();
if (!drv) {
return false;
}
if (drv->bdrv_is_inserted) {
return drv->bdrv_is_inserted(bs);
}
QLIST_FOREACH(child, &bs->children, next) {
if (!bdrv_is_inserted(child->bs)) {
return false;
}
}
return true;
}
/**
* If eject_flag is TRUE, eject the media. Otherwise, close the tray
*/
void bdrv_eject(BlockDriverState *bs, bool eject_flag)
{
BlockDriver *drv = bs->drv;
IO_CODE();
if (drv && drv->bdrv_eject) {
drv->bdrv_eject(bs, eject_flag);
}
}
/**
* Lock or unlock the media (if it is locked, the user won't be able
* to eject it manually).
*/
void bdrv_lock_medium(BlockDriverState *bs, bool locked)
{
BlockDriver *drv = bs->drv;
IO_CODE();
trace_bdrv_lock_medium(bs, locked);
if (drv && drv->bdrv_lock_medium) {
drv->bdrv_lock_medium(bs, locked);
}
}
/* Get a reference to bs */
void bdrv_ref(BlockDriverState *bs)
{
GLOBAL_STATE_CODE();
bs->refcnt++;
}
/* Release a previously grabbed reference to bs.
* If after releasing, reference count is zero, the BlockDriverState is
* deleted. */
void bdrv_unref(BlockDriverState *bs)
{
GLOBAL_STATE_CODE();
if (!bs) {
return;
}
assert(bs->refcnt > 0);
if (--bs->refcnt == 0) {
bdrv_delete(bs);
}
}
struct BdrvOpBlocker {
Error *reason;
QLIST_ENTRY(BdrvOpBlocker) list;
};
bool bdrv_op_is_blocked(BlockDriverState *bs, BlockOpType op, Error **errp)
{
BdrvOpBlocker *blocker;
GLOBAL_STATE_CODE();
assert((int) op >= 0 && op < BLOCK_OP_TYPE_MAX);
if (!QLIST_EMPTY(&bs->op_blockers[op])) {
blocker = QLIST_FIRST(&bs->op_blockers[op]);
error_propagate_prepend(errp, error_copy(blocker->reason),
"Node '%s' is busy: ",
bdrv_get_device_or_node_name(bs));
return true;
}
return false;
}
void bdrv_op_block(BlockDriverState *bs, BlockOpType op, Error *reason)
{
BdrvOpBlocker *blocker;
GLOBAL_STATE_CODE();
assert((int) op >= 0 && op < BLOCK_OP_TYPE_MAX);
blocker = g_new0(BdrvOpBlocker, 1);
blocker->reason = reason;
QLIST_INSERT_HEAD(&bs->op_blockers[op], blocker, list);
}
void bdrv_op_unblock(BlockDriverState *bs, BlockOpType op, Error *reason)
{
BdrvOpBlocker *blocker, *next;
GLOBAL_STATE_CODE();
assert((int) op >= 0 && op < BLOCK_OP_TYPE_MAX);
QLIST_FOREACH_SAFE(blocker, &bs->op_blockers[op], list, next) {
if (blocker->reason == reason) {
QLIST_REMOVE(blocker, list);
g_free(blocker);
}
}
}
void bdrv_op_block_all(BlockDriverState *bs, Error *reason)
{
int i;
GLOBAL_STATE_CODE();
for (i = 0; i < BLOCK_OP_TYPE_MAX; i++) {
bdrv_op_block(bs, i, reason);
}
}
void bdrv_op_unblock_all(BlockDriverState *bs, Error *reason)
{
int i;
GLOBAL_STATE_CODE();
for (i = 0; i < BLOCK_OP_TYPE_MAX; i++) {
bdrv_op_unblock(bs, i, reason);
}
}
bool bdrv_op_blocker_is_empty(BlockDriverState *bs)
{
int i;
GLOBAL_STATE_CODE();
for (i = 0; i < BLOCK_OP_TYPE_MAX; i++) {
if (!QLIST_EMPTY(&bs->op_blockers[i])) {
return false;
}
}
return true;
}
void bdrv_img_create(const char *filename, const char *fmt,
const char *base_filename, const char *base_fmt,
char *options, uint64_t img_size, int flags, bool quiet,
Error **errp)
{
QemuOptsList *create_opts = NULL;
QemuOpts *opts = NULL;
const char *backing_fmt, *backing_file;
int64_t size;
BlockDriver *drv, *proto_drv;
Error *local_err = NULL;
int ret = 0;
GLOBAL_STATE_CODE();
/* Find driver and parse its options */
drv = bdrv_find_format(fmt);
if (!drv) {
error_setg(errp, "Unknown file format '%s'", fmt);
return;
}
proto_drv = bdrv_find_protocol(filename, true, errp);
if (!proto_drv) {
return;
}
if (!drv->create_opts) {
error_setg(errp, "Format driver '%s' does not support image creation",
drv->format_name);
return;
}
if (!proto_drv->create_opts) {
error_setg(errp, "Protocol driver '%s' does not support image creation",
proto_drv->format_name);
return;
}
/* Create parameter list */
create_opts = qemu_opts_append(create_opts, drv->create_opts);
create_opts = qemu_opts_append(create_opts, proto_drv->create_opts);
opts = qemu_opts_create(create_opts, NULL, 0, &error_abort);
/* Parse -o options */
if (options) {
if (!qemu_opts_do_parse(opts, options, NULL, errp)) {
goto out;
}
}
if (!qemu_opt_get(opts, BLOCK_OPT_SIZE)) {
qemu_opt_set_number(opts, BLOCK_OPT_SIZE, img_size, &error_abort);
} else if (img_size != UINT64_C(-1)) {
error_setg(errp, "The image size must be specified only once");
goto out;
}
if (base_filename) {
if (!qemu_opt_set(opts, BLOCK_OPT_BACKING_FILE, base_filename,
NULL)) {
error_setg(errp, "Backing file not supported for file format '%s'",
fmt);
goto out;
}
}
if (base_fmt) {
if (!qemu_opt_set(opts, BLOCK_OPT_BACKING_FMT, base_fmt, NULL)) {
error_setg(errp, "Backing file format not supported for file "
"format '%s'", fmt);
goto out;
}
}
backing_file = qemu_opt_get(opts, BLOCK_OPT_BACKING_FILE);
if (backing_file) {
if (!strcmp(filename, backing_file)) {
error_setg(errp, "Error: Trying to create an image with the "
"same filename as the backing file");
goto out;
}
if (backing_file[0] == '\0') {
error_setg(errp, "Expected backing file name, got empty string");
goto out;
}
}
backing_fmt = qemu_opt_get(opts, BLOCK_OPT_BACKING_FMT);
/* The size for the image must always be specified, unless we have a backing
* file and we have not been forbidden from opening it. */
size = qemu_opt_get_size(opts, BLOCK_OPT_SIZE, img_size);
if (backing_file && !(flags & BDRV_O_NO_BACKING)) {
BlockDriverState *bs;
char *full_backing;
int back_flags;
QDict *backing_options = NULL;
full_backing =
bdrv_get_full_backing_filename_from_filename(filename, backing_file,
&local_err);
if (local_err) {
goto out;
}
assert(full_backing);
/*
* No need to do I/O here, which allows us to open encrypted
* backing images without needing the secret
*/
back_flags = flags;
back_flags &= ~(BDRV_O_RDWR | BDRV_O_SNAPSHOT | BDRV_O_NO_BACKING);
back_flags |= BDRV_O_NO_IO;
backing_options = qdict_new();
if (backing_fmt) {
qdict_put_str(backing_options, "driver", backing_fmt);
}
qdict_put_bool(backing_options, BDRV_OPT_FORCE_SHARE, true);
bs = bdrv_open(full_backing, NULL, backing_options, back_flags,
&local_err);
g_free(full_backing);
if (!bs) {
error_append_hint(&local_err, "Could not open backing image.\n");
goto out;
} else {
qemu-img: Deprecate use of -b without -F Creating an image that requires format probing of the backing image is potentially unsafe (we've had several CVEs over the years based on probes leaking information to the guest on a subsequent boot, although these days tools like libvirt are aware of the issue enough to prevent the worst effects). For example, if our probing algorithm ever changes, or if other tools like libvirt determine a different probe result than we do, then subsequent use of that backing file under a different format will present corrupted data to the guest. Fortunately, the worst effects occur only when the backing image is originally raw, and we at least prevent commit into a probed raw backing file that would change its probed type. Still, it is worth starting a deprecation clock so that future qemu-img can refuse to create backing chains that would rely on probing, to encourage clients to avoid unsafe practices. Most warnings are intentionally emitted from bdrv_img_create() in the block layer, but qemu-img convert uses bdrv_create() which cannot emit its own warning without causing spurious warnings on other code paths. In the end, all command-line image creation or backing file rewriting now performs a check. Furthermore, if we probe a backing file as non-raw, then it is safe to explicitly record that result (rather than relying on future probes); only where we probe a raw image do we care about further warnings to the user when using such an image (for example, commits into a probed-raw backing file are prevented), to help them improve their tooling. But whether or not we make the probe results explicit, we still warn the user to remind them to upgrade their workflow to supply -F always. iotest 114 specifically wants to create an unsafe image for later amendment rather than defaulting to our new default of recording a probed format, so it needs an update. While touching it, expand it to cover all of the various warnings enabled by this patch. iotest 301 also shows a change to qcow messages. Signed-off-by: Eric Blake <eblake@redhat.com> Message-Id: <20200706203954.341758-11-eblake@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2020-07-06 22:39:54 +02:00
if (!backing_fmt) {
error_setg(&local_err,
"Backing file specified without backing format");
error_append_hint(&local_err, "Detected format of %s.",
bs->drv->format_name);
goto out;
qemu-img: Deprecate use of -b without -F Creating an image that requires format probing of the backing image is potentially unsafe (we've had several CVEs over the years based on probes leaking information to the guest on a subsequent boot, although these days tools like libvirt are aware of the issue enough to prevent the worst effects). For example, if our probing algorithm ever changes, or if other tools like libvirt determine a different probe result than we do, then subsequent use of that backing file under a different format will present corrupted data to the guest. Fortunately, the worst effects occur only when the backing image is originally raw, and we at least prevent commit into a probed raw backing file that would change its probed type. Still, it is worth starting a deprecation clock so that future qemu-img can refuse to create backing chains that would rely on probing, to encourage clients to avoid unsafe practices. Most warnings are intentionally emitted from bdrv_img_create() in the block layer, but qemu-img convert uses bdrv_create() which cannot emit its own warning without causing spurious warnings on other code paths. In the end, all command-line image creation or backing file rewriting now performs a check. Furthermore, if we probe a backing file as non-raw, then it is safe to explicitly record that result (rather than relying on future probes); only where we probe a raw image do we care about further warnings to the user when using such an image (for example, commits into a probed-raw backing file are prevented), to help them improve their tooling. But whether or not we make the probe results explicit, we still warn the user to remind them to upgrade their workflow to supply -F always. iotest 114 specifically wants to create an unsafe image for later amendment rather than defaulting to our new default of recording a probed format, so it needs an update. While touching it, expand it to cover all of the various warnings enabled by this patch. iotest 301 also shows a change to qcow messages. Signed-off-by: Eric Blake <eblake@redhat.com> Message-Id: <20200706203954.341758-11-eblake@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2020-07-06 22:39:54 +02:00
}
if (size == -1) {
/* Opened BS, have no size */
size = bdrv_getlength(bs);
if (size < 0) {
error_setg_errno(errp, -size, "Could not get size of '%s'",
backing_file);
bdrv_unref(bs);
goto out;
}
qemu_opt_set_number(opts, BLOCK_OPT_SIZE, size, &error_abort);
}
bdrv_unref(bs);
}
qemu-img: Deprecate use of -b without -F Creating an image that requires format probing of the backing image is potentially unsafe (we've had several CVEs over the years based on probes leaking information to the guest on a subsequent boot, although these days tools like libvirt are aware of the issue enough to prevent the worst effects). For example, if our probing algorithm ever changes, or if other tools like libvirt determine a different probe result than we do, then subsequent use of that backing file under a different format will present corrupted data to the guest. Fortunately, the worst effects occur only when the backing image is originally raw, and we at least prevent commit into a probed raw backing file that would change its probed type. Still, it is worth starting a deprecation clock so that future qemu-img can refuse to create backing chains that would rely on probing, to encourage clients to avoid unsafe practices. Most warnings are intentionally emitted from bdrv_img_create() in the block layer, but qemu-img convert uses bdrv_create() which cannot emit its own warning without causing spurious warnings on other code paths. In the end, all command-line image creation or backing file rewriting now performs a check. Furthermore, if we probe a backing file as non-raw, then it is safe to explicitly record that result (rather than relying on future probes); only where we probe a raw image do we care about further warnings to the user when using such an image (for example, commits into a probed-raw backing file are prevented), to help them improve their tooling. But whether or not we make the probe results explicit, we still warn the user to remind them to upgrade their workflow to supply -F always. iotest 114 specifically wants to create an unsafe image for later amendment rather than defaulting to our new default of recording a probed format, so it needs an update. While touching it, expand it to cover all of the various warnings enabled by this patch. iotest 301 also shows a change to qcow messages. Signed-off-by: Eric Blake <eblake@redhat.com> Message-Id: <20200706203954.341758-11-eblake@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2020-07-06 22:39:54 +02:00
/* (backing_file && !(flags & BDRV_O_NO_BACKING)) */
} else if (backing_file && !backing_fmt) {
error_setg(&local_err,
"Backing file specified without backing format");
goto out;
qemu-img: Deprecate use of -b without -F Creating an image that requires format probing of the backing image is potentially unsafe (we've had several CVEs over the years based on probes leaking information to the guest on a subsequent boot, although these days tools like libvirt are aware of the issue enough to prevent the worst effects). For example, if our probing algorithm ever changes, or if other tools like libvirt determine a different probe result than we do, then subsequent use of that backing file under a different format will present corrupted data to the guest. Fortunately, the worst effects occur only when the backing image is originally raw, and we at least prevent commit into a probed raw backing file that would change its probed type. Still, it is worth starting a deprecation clock so that future qemu-img can refuse to create backing chains that would rely on probing, to encourage clients to avoid unsafe practices. Most warnings are intentionally emitted from bdrv_img_create() in the block layer, but qemu-img convert uses bdrv_create() which cannot emit its own warning without causing spurious warnings on other code paths. In the end, all command-line image creation or backing file rewriting now performs a check. Furthermore, if we probe a backing file as non-raw, then it is safe to explicitly record that result (rather than relying on future probes); only where we probe a raw image do we care about further warnings to the user when using such an image (for example, commits into a probed-raw backing file are prevented), to help them improve their tooling. But whether or not we make the probe results explicit, we still warn the user to remind them to upgrade their workflow to supply -F always. iotest 114 specifically wants to create an unsafe image for later amendment rather than defaulting to our new default of recording a probed format, so it needs an update. While touching it, expand it to cover all of the various warnings enabled by this patch. iotest 301 also shows a change to qcow messages. Signed-off-by: Eric Blake <eblake@redhat.com> Message-Id: <20200706203954.341758-11-eblake@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2020-07-06 22:39:54 +02:00
}
if (size == -1) {
error_setg(errp, "Image creation needs a size parameter");
goto out;
}
if (!quiet) {
printf("Formatting '%s', fmt=%s ", filename, fmt);
qemu_opts_print(opts, " ");
puts("");
fflush(stdout);
}
ret = bdrv_create(drv, filename, opts, &local_err);
if (ret == -EFBIG) {
/* This is generally a better message than whatever the driver would
* deliver (especially because of the cluster_size_hint), since that
* is most probably not much different from "image too large". */
const char *cluster_size_hint = "";
if (qemu_opt_get_size(opts, BLOCK_OPT_CLUSTER_SIZE, 0)) {
cluster_size_hint = " (try using a larger cluster size)";
}
error_setg(errp, "The image size is too large for file format '%s'"
"%s", fmt, cluster_size_hint);
error_free(local_err);
local_err = NULL;
}
out:
qemu_opts_del(opts);
qemu_opts_free(create_opts);
error_propagate(errp, local_err);
}
AioContext *bdrv_get_aio_context(BlockDriverState *bs)
{
IO_CODE();
return bs ? bs->aio_context : qemu_get_aio_context();
}
AioContext *coroutine_fn bdrv_co_enter(BlockDriverState *bs)
{
Coroutine *self = qemu_coroutine_self();
AioContext *old_ctx = qemu_coroutine_get_aio_context(self);
AioContext *new_ctx;
IO_CODE();
/*
* Increase bs->in_flight to ensure that this operation is completed before
* moving the node to a different AioContext. Read new_ctx only afterwards.
*/
bdrv_inc_in_flight(bs);
new_ctx = bdrv_get_aio_context(bs);
aio_co_reschedule_self(new_ctx);
return old_ctx;
}
void coroutine_fn bdrv_co_leave(BlockDriverState *bs, AioContext *old_ctx)
{
IO_CODE();
aio_co_reschedule_self(old_ctx);
bdrv_dec_in_flight(bs);
}
void coroutine_fn bdrv_co_lock(BlockDriverState *bs)
{
AioContext *ctx = bdrv_get_aio_context(bs);
/* In the main thread, bs->aio_context won't change concurrently */
assert(qemu_get_current_aio_context() == qemu_get_aio_context());
/*
* We're in coroutine context, so we already hold the lock of the main
* loop AioContext. Don't lock it twice to avoid deadlocks.
*/
assert(qemu_in_coroutine());
if (ctx != qemu_get_aio_context()) {
aio_context_acquire(ctx);
}
}
void coroutine_fn bdrv_co_unlock(BlockDriverState *bs)
{
AioContext *ctx = bdrv_get_aio_context(bs);
assert(qemu_in_coroutine());
if (ctx != qemu_get_aio_context()) {
aio_context_release(ctx);
}
}
void bdrv_coroutine_enter(BlockDriverState *bs, Coroutine *co)
{
IO_CODE();
aio_co_enter(bdrv_get_aio_context(bs), co);
}
static void bdrv_do_remove_aio_context_notifier(BdrvAioNotifier *ban)
{
GLOBAL_STATE_CODE();
QLIST_REMOVE(ban, list);
g_free(ban);
}
static void bdrv_detach_aio_context(BlockDriverState *bs)
{
BdrvAioNotifier *baf, *baf_tmp;
assert(!bs->walking_aio_notifiers);
GLOBAL_STATE_CODE();
bs->walking_aio_notifiers = true;
QLIST_FOREACH_SAFE(baf, &bs->aio_notifiers, list, baf_tmp) {
if (baf->deleted) {
bdrv_do_remove_aio_context_notifier(baf);
} else {
baf->detach_aio_context(baf->opaque);
}
}
/* Never mind iterating again to check for ->deleted. bdrv_close() will
* remove remaining aio notifiers if we aren't called again.
*/
bs->walking_aio_notifiers = false;
block: Fix AioContext switch for bs->drv == NULL Even for block nodes with bs->drv == NULL, we can't just ignore a bdrv_set_aio_context() call. Leaving the node in its old context can mean that it's still in an iothread context in bdrv_close_all() during shutdown, resulting in an attempted unlock of the AioContext lock which we don't hold. This is an example stack trace of a related crash: #0 0x00007ffff59da57f in raise () at /lib64/libc.so.6 #1 0x00007ffff59c4895 in abort () at /lib64/libc.so.6 #2 0x0000555555b97b1e in error_exit (err=<optimized out>, msg=msg@entry=0x555555d386d0 <__func__.19059> "qemu_mutex_unlock_impl") at util/qemu-thread-posix.c:36 #3 0x0000555555b97f7f in qemu_mutex_unlock_impl (mutex=mutex@entry=0x5555568002f0, file=file@entry=0x555555d378df "util/async.c", line=line@entry=507) at util/qemu-thread-posix.c:97 #4 0x0000555555b92f55 in aio_context_release (ctx=ctx@entry=0x555556800290) at util/async.c:507 #5 0x0000555555b05cf8 in bdrv_prwv_co (child=child@entry=0x7fffc80012f0, offset=offset@entry=131072, qiov=qiov@entry=0x7fffffffd4f0, is_write=is_write@entry=true, flags=flags@entry=0) at block/io.c:833 #6 0x0000555555b060a9 in bdrv_pwritev (qiov=0x7fffffffd4f0, offset=131072, child=0x7fffc80012f0) at block/io.c:990 #7 0x0000555555b060a9 in bdrv_pwrite (child=0x7fffc80012f0, offset=131072, buf=<optimized out>, bytes=<optimized out>) at block/io.c:990 #8 0x0000555555ae172b in qcow2_cache_entry_flush (bs=bs@entry=0x555556810680, c=c@entry=0x5555568cc740, i=i@entry=0) at block/qcow2-cache.c:51 #9 0x0000555555ae18dd in qcow2_cache_write (bs=bs@entry=0x555556810680, c=0x5555568cc740) at block/qcow2-cache.c:248 #10 0x0000555555ae15de in qcow2_cache_flush (bs=0x555556810680, c=<optimized out>) at block/qcow2-cache.c:259 #11 0x0000555555ae16b1 in qcow2_cache_flush_dependency (c=0x5555568a1700, c=0x5555568a1700, bs=0x555556810680) at block/qcow2-cache.c:194 #12 0x0000555555ae16b1 in qcow2_cache_entry_flush (bs=bs@entry=0x555556810680, c=c@entry=0x5555568a1700, i=i@entry=0) at block/qcow2-cache.c:194 #13 0x0000555555ae18dd in qcow2_cache_write (bs=bs@entry=0x555556810680, c=0x5555568a1700) at block/qcow2-cache.c:248 #14 0x0000555555ae15de in qcow2_cache_flush (bs=bs@entry=0x555556810680, c=<optimized out>) at block/qcow2-cache.c:259 #15 0x0000555555ad242c in qcow2_inactivate (bs=bs@entry=0x555556810680) at block/qcow2.c:2124 #16 0x0000555555ad2590 in qcow2_close (bs=0x555556810680) at block/qcow2.c:2153 #17 0x0000555555ab0c62 in bdrv_close (bs=0x555556810680) at block.c:3358 #18 0x0000555555ab0c62 in bdrv_delete (bs=0x555556810680) at block.c:3542 #19 0x0000555555ab0c62 in bdrv_unref (bs=0x555556810680) at block.c:4598 #20 0x0000555555af4d72 in blk_remove_bs (blk=blk@entry=0x5555568103d0) at block/block-backend.c:785 #21 0x0000555555af4dbb in blk_remove_all_bs () at block/block-backend.c:483 #22 0x0000555555aae02f in bdrv_close_all () at block.c:3412 #23 0x00005555557f9796 in main (argc=<optimized out>, argv=<optimized out>, envp=<optimized out>) at vl.c:4776 The reproducer I used is a qcow2 image on gluster volume, where the virtual disk size (4 GB) is larger than the gluster volume size (64M), so we can easily trigger an ENOSPC. This backend is assigned to a virtio-blk device using an iothread, and then from the guest a 'dd if=/dev/zero of=/dev/vda bs=1G count=1' causes the VM to stop because of an I/O error. qemu_gluster_co_flush_to_disk() sets bs->drv = NULL on error, so when virtio-blk stops the dataplane, the block nodes stay in the iothread AioContext. A 'quit' monitor command issued from this paused state crashes the process. Fixes: https://bugzilla.redhat.com/show_bug.cgi?id=1631227 Cc: qemu-stable@nongnu.org Signed-off-by: Kevin Wolf <kwolf@redhat.com> Reviewed-by: Eric Blake <eblake@redhat.com> Reviewed-by: Max Reitz <mreitz@redhat.com> Reviewed-by: Stefano Garzarella <sgarzare@redhat.com>
2019-04-17 17:15:25 +02:00
if (bs->drv && bs->drv->bdrv_detach_aio_context) {
bs->drv->bdrv_detach_aio_context(bs);
}
if (bs->quiesce_counter) {
aio_enable_external(bs->aio_context);
}
bs->aio_context = NULL;
}
static void bdrv_attach_aio_context(BlockDriverState *bs,
AioContext *new_context)
{
BdrvAioNotifier *ban, *ban_tmp;
GLOBAL_STATE_CODE();
if (bs->quiesce_counter) {
aio_disable_external(new_context);
}
bs->aio_context = new_context;
block: Fix AioContext switch for bs->drv == NULL Even for block nodes with bs->drv == NULL, we can't just ignore a bdrv_set_aio_context() call. Leaving the node in its old context can mean that it's still in an iothread context in bdrv_close_all() during shutdown, resulting in an attempted unlock of the AioContext lock which we don't hold. This is an example stack trace of a related crash: #0 0x00007ffff59da57f in raise () at /lib64/libc.so.6 #1 0x00007ffff59c4895 in abort () at /lib64/libc.so.6 #2 0x0000555555b97b1e in error_exit (err=<optimized out>, msg=msg@entry=0x555555d386d0 <__func__.19059> "qemu_mutex_unlock_impl") at util/qemu-thread-posix.c:36 #3 0x0000555555b97f7f in qemu_mutex_unlock_impl (mutex=mutex@entry=0x5555568002f0, file=file@entry=0x555555d378df "util/async.c", line=line@entry=507) at util/qemu-thread-posix.c:97 #4 0x0000555555b92f55 in aio_context_release (ctx=ctx@entry=0x555556800290) at util/async.c:507 #5 0x0000555555b05cf8 in bdrv_prwv_co (child=child@entry=0x7fffc80012f0, offset=offset@entry=131072, qiov=qiov@entry=0x7fffffffd4f0, is_write=is_write@entry=true, flags=flags@entry=0) at block/io.c:833 #6 0x0000555555b060a9 in bdrv_pwritev (qiov=0x7fffffffd4f0, offset=131072, child=0x7fffc80012f0) at block/io.c:990 #7 0x0000555555b060a9 in bdrv_pwrite (child=0x7fffc80012f0, offset=131072, buf=<optimized out>, bytes=<optimized out>) at block/io.c:990 #8 0x0000555555ae172b in qcow2_cache_entry_flush (bs=bs@entry=0x555556810680, c=c@entry=0x5555568cc740, i=i@entry=0) at block/qcow2-cache.c:51 #9 0x0000555555ae18dd in qcow2_cache_write (bs=bs@entry=0x555556810680, c=0x5555568cc740) at block/qcow2-cache.c:248 #10 0x0000555555ae15de in qcow2_cache_flush (bs=0x555556810680, c=<optimized out>) at block/qcow2-cache.c:259 #11 0x0000555555ae16b1 in qcow2_cache_flush_dependency (c=0x5555568a1700, c=0x5555568a1700, bs=0x555556810680) at block/qcow2-cache.c:194 #12 0x0000555555ae16b1 in qcow2_cache_entry_flush (bs=bs@entry=0x555556810680, c=c@entry=0x5555568a1700, i=i@entry=0) at block/qcow2-cache.c:194 #13 0x0000555555ae18dd in qcow2_cache_write (bs=bs@entry=0x555556810680, c=0x5555568a1700) at block/qcow2-cache.c:248 #14 0x0000555555ae15de in qcow2_cache_flush (bs=bs@entry=0x555556810680, c=<optimized out>) at block/qcow2-cache.c:259 #15 0x0000555555ad242c in qcow2_inactivate (bs=bs@entry=0x555556810680) at block/qcow2.c:2124 #16 0x0000555555ad2590 in qcow2_close (bs=0x555556810680) at block/qcow2.c:2153 #17 0x0000555555ab0c62 in bdrv_close (bs=0x555556810680) at block.c:3358 #18 0x0000555555ab0c62 in bdrv_delete (bs=0x555556810680) at block.c:3542 #19 0x0000555555ab0c62 in bdrv_unref (bs=0x555556810680) at block.c:4598 #20 0x0000555555af4d72 in blk_remove_bs (blk=blk@entry=0x5555568103d0) at block/block-backend.c:785 #21 0x0000555555af4dbb in blk_remove_all_bs () at block/block-backend.c:483 #22 0x0000555555aae02f in bdrv_close_all () at block.c:3412 #23 0x00005555557f9796 in main (argc=<optimized out>, argv=<optimized out>, envp=<optimized out>) at vl.c:4776 The reproducer I used is a qcow2 image on gluster volume, where the virtual disk size (4 GB) is larger than the gluster volume size (64M), so we can easily trigger an ENOSPC. This backend is assigned to a virtio-blk device using an iothread, and then from the guest a 'dd if=/dev/zero of=/dev/vda bs=1G count=1' causes the VM to stop because of an I/O error. qemu_gluster_co_flush_to_disk() sets bs->drv = NULL on error, so when virtio-blk stops the dataplane, the block nodes stay in the iothread AioContext. A 'quit' monitor command issued from this paused state crashes the process. Fixes: https://bugzilla.redhat.com/show_bug.cgi?id=1631227 Cc: qemu-stable@nongnu.org Signed-off-by: Kevin Wolf <kwolf@redhat.com> Reviewed-by: Eric Blake <eblake@redhat.com> Reviewed-by: Max Reitz <mreitz@redhat.com> Reviewed-by: Stefano Garzarella <sgarzare@redhat.com>
2019-04-17 17:15:25 +02:00
if (bs->drv && bs->drv->bdrv_attach_aio_context) {
bs->drv->bdrv_attach_aio_context(bs, new_context);
}
assert(!bs->walking_aio_notifiers);
bs->walking_aio_notifiers = true;
QLIST_FOREACH_SAFE(ban, &bs->aio_notifiers, list, ban_tmp) {
if (ban->deleted) {
bdrv_do_remove_aio_context_notifier(ban);
} else {
ban->attached_aio_context(new_context, ban->opaque);
}
}
bs->walking_aio_notifiers = false;
}
/*
* Changes the AioContext used for fd handlers, timers, and BHs by this
* BlockDriverState and all its children and parents.
*
* Must be called from the main AioContext.
*
* The caller must own the AioContext lock for the old AioContext of bs, but it
* must not own the AioContext lock for new_context (unless new_context is the
* same as the current context of bs).
*
* @ignore will accumulate all visited BdrvChild object. The caller is
* responsible for freeing the list afterwards.
*/
void bdrv_set_aio_context_ignore(BlockDriverState *bs,
AioContext *new_context, GSList **ignore)
{
block: Do not poll in bdrv_do_drained_end() We should never poll anywhere in bdrv_do_drained_end() (including its recursive callees like bdrv_drain_invoke()), because it does not cope well with graph changes. In fact, it has been written based on the postulation that no graph changes will happen in it. Instead, the callers that want to poll must poll, i.e. all currently globally available wrappers: bdrv_drained_end(), bdrv_subtree_drained_end(), bdrv_unapply_subtree_drain(), and bdrv_drain_all_end(). Graph changes there do not matter. They can poll simply by passing a pointer to a drained_end_counter and wait until it reaches 0. This patch also adds a non-polling global wrapper for bdrv_do_drained_end() that takes a drained_end_counter pointer. We need such a variant because now no function called anywhere from bdrv_do_drained_end() must poll. This includes BdrvChildRole.drained_end(), which already must not poll according to its interface documentation, but bdrv_child_cb_drained_end() just violates that by invoking bdrv_drained_end() (which does poll). Therefore, BdrvChildRole.drained_end() must take a *drained_end_counter parameter, which bdrv_child_cb_drained_end() can pass on to the new bdrv_drained_end_no_poll() function. Note that we now have a pattern of all drained_end-related functions either polling or receiving a *drained_end_counter to let the caller poll based on that. A problem with a single poll loop is that when the drained section in bdrv_set_aio_context_ignore() ends, some nodes in the subgraph may be in the old contexts, while others are in the new context already. To let the collective poll in bdrv_drained_end() work correctly, we must not hold a lock to the old context, so that the old context can make progress in case it is different from the current context. (In the process, remove the comment saying that the current context is always the old context, because it is wrong.) In all other places, all nodes in a subtree must be in the same context, so we can just poll that. The exception of course is bdrv_drain_all_end(), but that always runs in the main context, so we can just poll NULL (like bdrv_drain_all_begin() does). Signed-off-by: Max Reitz <mreitz@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2019-07-19 11:26:14 +02:00
AioContext *old_context = bdrv_get_aio_context(bs);
GSList *children_to_process = NULL;
GSList *parents_to_process = NULL;
GSList *entry;
BdrvChild *child, *parent;
g_assert(qemu_get_current_aio_context() == qemu_get_aio_context());
GLOBAL_STATE_CODE();
block: Do not poll in bdrv_do_drained_end() We should never poll anywhere in bdrv_do_drained_end() (including its recursive callees like bdrv_drain_invoke()), because it does not cope well with graph changes. In fact, it has been written based on the postulation that no graph changes will happen in it. Instead, the callers that want to poll must poll, i.e. all currently globally available wrappers: bdrv_drained_end(), bdrv_subtree_drained_end(), bdrv_unapply_subtree_drain(), and bdrv_drain_all_end(). Graph changes there do not matter. They can poll simply by passing a pointer to a drained_end_counter and wait until it reaches 0. This patch also adds a non-polling global wrapper for bdrv_do_drained_end() that takes a drained_end_counter pointer. We need such a variant because now no function called anywhere from bdrv_do_drained_end() must poll. This includes BdrvChildRole.drained_end(), which already must not poll according to its interface documentation, but bdrv_child_cb_drained_end() just violates that by invoking bdrv_drained_end() (which does poll). Therefore, BdrvChildRole.drained_end() must take a *drained_end_counter parameter, which bdrv_child_cb_drained_end() can pass on to the new bdrv_drained_end_no_poll() function. Note that we now have a pattern of all drained_end-related functions either polling or receiving a *drained_end_counter to let the caller poll based on that. A problem with a single poll loop is that when the drained section in bdrv_set_aio_context_ignore() ends, some nodes in the subgraph may be in the old contexts, while others are in the new context already. To let the collective poll in bdrv_drained_end() work correctly, we must not hold a lock to the old context, so that the old context can make progress in case it is different from the current context. (In the process, remove the comment saying that the current context is always the old context, because it is wrong.) In all other places, all nodes in a subtree must be in the same context, so we can just poll that. The exception of course is bdrv_drain_all_end(), but that always runs in the main context, so we can just poll NULL (like bdrv_drain_all_begin() does). Signed-off-by: Max Reitz <mreitz@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2019-07-19 11:26:14 +02:00
if (old_context == new_context) {
block: don't set the same context Adds a fast path on aio context setting preventing unnecessary context setting routine. Also, it prevents issues with cyclic walk of child bds-es appeared because of registering aio walking notifiers: Call stack: 0 __GI_raise 1 __GI_abort 2 __assert_fail_base 3 __GI___assert_fail 4 bdrv_detach_aio_context (bs=0x55f54d65c000) <<< 5 bdrv_detach_aio_context (bs=0x55f54fc8a800) 6 bdrv_set_aio_context (bs=0x55f54fc8a800, ...) 7 block_job_attached_aio_context 8 bdrv_attach_aio_context (bs=0x55f54d65c000, ...) <<< 9 bdrv_set_aio_context (bs=0x55f54d65c000) 10 blk_set_aio_context 11 virtio_blk_data_plane_stop 12 virtio_bus_stop_ioeventfd 13 virtio_vmstate_change 14 vm_state_notify (running=0, state=RUN_STATE_SHUTDOWN) 15 do_vm_stop (state=RUN_STATE_SHUTDOWN, send_stop=true) 16 vm_stop (state=RUN_STATE_SHUTDOWN) 17 main_loop_should_exit 18 main_loop 19 main This can happen because of "new" context attachment to VM disk bds. When attaching a new context the corresponding aio context handler is called for each of aio_notifiers registered on the VM disk bds context. Among those handlers, there is the block_job_attached_aio_context handler which sets a new aio context for the block job bds. When doing so, the old context is detached from all the block job bds children and one of them is the VM disk bds, serving as backing store for the blockjob bds, although the VM disk bds is actually the initializer of that process. Since the VM disk bds is protected with walking_aio_notifiers flag from double processing in recursive calls, the assert fires. Signed-off-by: Denis Plotnikov <dplotnikov@virtuozzo.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2019-02-15 14:03:25 +01:00
return;
}
bdrv_drained_begin(bs);
QLIST_FOREACH(child, &bs->children, next) {
if (g_slist_find(*ignore, child)) {
continue;
}
*ignore = g_slist_prepend(*ignore, child);
children_to_process = g_slist_prepend(children_to_process, child);
}
QLIST_FOREACH(parent, &bs->parents, next_parent) {
if (g_slist_find(*ignore, parent)) {
continue;
}
*ignore = g_slist_prepend(*ignore, parent);
parents_to_process = g_slist_prepend(parents_to_process, parent);
}
for (entry = children_to_process;
entry != NULL;
entry = g_slist_next(entry)) {
child = entry->data;
bdrv_set_aio_context_ignore(child->bs, new_context, ignore);
}
g_slist_free(children_to_process);
for (entry = parents_to_process;
entry != NULL;
entry = g_slist_next(entry)) {
parent = entry->data;
assert(parent->klass->set_aio_ctx);
parent->klass->set_aio_ctx(parent, new_context, ignore);
}
g_slist_free(parents_to_process);
bdrv_detach_aio_context(bs);
block: Do not poll in bdrv_do_drained_end() We should never poll anywhere in bdrv_do_drained_end() (including its recursive callees like bdrv_drain_invoke()), because it does not cope well with graph changes. In fact, it has been written based on the postulation that no graph changes will happen in it. Instead, the callers that want to poll must poll, i.e. all currently globally available wrappers: bdrv_drained_end(), bdrv_subtree_drained_end(), bdrv_unapply_subtree_drain(), and bdrv_drain_all_end(). Graph changes there do not matter. They can poll simply by passing a pointer to a drained_end_counter and wait until it reaches 0. This patch also adds a non-polling global wrapper for bdrv_do_drained_end() that takes a drained_end_counter pointer. We need such a variant because now no function called anywhere from bdrv_do_drained_end() must poll. This includes BdrvChildRole.drained_end(), which already must not poll according to its interface documentation, but bdrv_child_cb_drained_end() just violates that by invoking bdrv_drained_end() (which does poll). Therefore, BdrvChildRole.drained_end() must take a *drained_end_counter parameter, which bdrv_child_cb_drained_end() can pass on to the new bdrv_drained_end_no_poll() function. Note that we now have a pattern of all drained_end-related functions either polling or receiving a *drained_end_counter to let the caller poll based on that. A problem with a single poll loop is that when the drained section in bdrv_set_aio_context_ignore() ends, some nodes in the subgraph may be in the old contexts, while others are in the new context already. To let the collective poll in bdrv_drained_end() work correctly, we must not hold a lock to the old context, so that the old context can make progress in case it is different from the current context. (In the process, remove the comment saying that the current context is always the old context, because it is wrong.) In all other places, all nodes in a subtree must be in the same context, so we can just poll that. The exception of course is bdrv_drain_all_end(), but that always runs in the main context, so we can just poll NULL (like bdrv_drain_all_begin() does). Signed-off-by: Max Reitz <mreitz@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2019-07-19 11:26:14 +02:00
/* Acquire the new context, if necessary */
if (qemu_get_aio_context() != new_context) {
block: Do not poll in bdrv_do_drained_end() We should never poll anywhere in bdrv_do_drained_end() (including its recursive callees like bdrv_drain_invoke()), because it does not cope well with graph changes. In fact, it has been written based on the postulation that no graph changes will happen in it. Instead, the callers that want to poll must poll, i.e. all currently globally available wrappers: bdrv_drained_end(), bdrv_subtree_drained_end(), bdrv_unapply_subtree_drain(), and bdrv_drain_all_end(). Graph changes there do not matter. They can poll simply by passing a pointer to a drained_end_counter and wait until it reaches 0. This patch also adds a non-polling global wrapper for bdrv_do_drained_end() that takes a drained_end_counter pointer. We need such a variant because now no function called anywhere from bdrv_do_drained_end() must poll. This includes BdrvChildRole.drained_end(), which already must not poll according to its interface documentation, but bdrv_child_cb_drained_end() just violates that by invoking bdrv_drained_end() (which does poll). Therefore, BdrvChildRole.drained_end() must take a *drained_end_counter parameter, which bdrv_child_cb_drained_end() can pass on to the new bdrv_drained_end_no_poll() function. Note that we now have a pattern of all drained_end-related functions either polling or receiving a *drained_end_counter to let the caller poll based on that. A problem with a single poll loop is that when the drained section in bdrv_set_aio_context_ignore() ends, some nodes in the subgraph may be in the old contexts, while others are in the new context already. To let the collective poll in bdrv_drained_end() work correctly, we must not hold a lock to the old context, so that the old context can make progress in case it is different from the current context. (In the process, remove the comment saying that the current context is always the old context, because it is wrong.) In all other places, all nodes in a subtree must be in the same context, so we can just poll that. The exception of course is bdrv_drain_all_end(), but that always runs in the main context, so we can just poll NULL (like bdrv_drain_all_begin() does). Signed-off-by: Max Reitz <mreitz@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2019-07-19 11:26:14 +02:00
aio_context_acquire(new_context);
}
bdrv_attach_aio_context(bs, new_context);
block: Do not poll in bdrv_do_drained_end() We should never poll anywhere in bdrv_do_drained_end() (including its recursive callees like bdrv_drain_invoke()), because it does not cope well with graph changes. In fact, it has been written based on the postulation that no graph changes will happen in it. Instead, the callers that want to poll must poll, i.e. all currently globally available wrappers: bdrv_drained_end(), bdrv_subtree_drained_end(), bdrv_unapply_subtree_drain(), and bdrv_drain_all_end(). Graph changes there do not matter. They can poll simply by passing a pointer to a drained_end_counter and wait until it reaches 0. This patch also adds a non-polling global wrapper for bdrv_do_drained_end() that takes a drained_end_counter pointer. We need such a variant because now no function called anywhere from bdrv_do_drained_end() must poll. This includes BdrvChildRole.drained_end(), which already must not poll according to its interface documentation, but bdrv_child_cb_drained_end() just violates that by invoking bdrv_drained_end() (which does poll). Therefore, BdrvChildRole.drained_end() must take a *drained_end_counter parameter, which bdrv_child_cb_drained_end() can pass on to the new bdrv_drained_end_no_poll() function. Note that we now have a pattern of all drained_end-related functions either polling or receiving a *drained_end_counter to let the caller poll based on that. A problem with a single poll loop is that when the drained section in bdrv_set_aio_context_ignore() ends, some nodes in the subgraph may be in the old contexts, while others are in the new context already. To let the collective poll in bdrv_drained_end() work correctly, we must not hold a lock to the old context, so that the old context can make progress in case it is different from the current context. (In the process, remove the comment saying that the current context is always the old context, because it is wrong.) In all other places, all nodes in a subtree must be in the same context, so we can just poll that. The exception of course is bdrv_drain_all_end(), but that always runs in the main context, so we can just poll NULL (like bdrv_drain_all_begin() does). Signed-off-by: Max Reitz <mreitz@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2019-07-19 11:26:14 +02:00
/*
* If this function was recursively called from
* bdrv_set_aio_context_ignore(), there may be nodes in the
* subtree that have not yet been moved to the new AioContext.
* Release the old one so bdrv_drained_end() can poll them.
*/
if (qemu_get_aio_context() != old_context) {
block: Do not poll in bdrv_do_drained_end() We should never poll anywhere in bdrv_do_drained_end() (including its recursive callees like bdrv_drain_invoke()), because it does not cope well with graph changes. In fact, it has been written based on the postulation that no graph changes will happen in it. Instead, the callers that want to poll must poll, i.e. all currently globally available wrappers: bdrv_drained_end(), bdrv_subtree_drained_end(), bdrv_unapply_subtree_drain(), and bdrv_drain_all_end(). Graph changes there do not matter. They can poll simply by passing a pointer to a drained_end_counter and wait until it reaches 0. This patch also adds a non-polling global wrapper for bdrv_do_drained_end() that takes a drained_end_counter pointer. We need such a variant because now no function called anywhere from bdrv_do_drained_end() must poll. This includes BdrvChildRole.drained_end(), which already must not poll according to its interface documentation, but bdrv_child_cb_drained_end() just violates that by invoking bdrv_drained_end() (which does poll). Therefore, BdrvChildRole.drained_end() must take a *drained_end_counter parameter, which bdrv_child_cb_drained_end() can pass on to the new bdrv_drained_end_no_poll() function. Note that we now have a pattern of all drained_end-related functions either polling or receiving a *drained_end_counter to let the caller poll based on that. A problem with a single poll loop is that when the drained section in bdrv_set_aio_context_ignore() ends, some nodes in the subgraph may be in the old contexts, while others are in the new context already. To let the collective poll in bdrv_drained_end() work correctly, we must not hold a lock to the old context, so that the old context can make progress in case it is different from the current context. (In the process, remove the comment saying that the current context is always the old context, because it is wrong.) In all other places, all nodes in a subtree must be in the same context, so we can just poll that. The exception of course is bdrv_drain_all_end(), but that always runs in the main context, so we can just poll NULL (like bdrv_drain_all_begin() does). Signed-off-by: Max Reitz <mreitz@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2019-07-19 11:26:14 +02:00
aio_context_release(old_context);
}
bdrv_drained_end(bs);
block: Do not poll in bdrv_do_drained_end() We should never poll anywhere in bdrv_do_drained_end() (including its recursive callees like bdrv_drain_invoke()), because it does not cope well with graph changes. In fact, it has been written based on the postulation that no graph changes will happen in it. Instead, the callers that want to poll must poll, i.e. all currently globally available wrappers: bdrv_drained_end(), bdrv_subtree_drained_end(), bdrv_unapply_subtree_drain(), and bdrv_drain_all_end(). Graph changes there do not matter. They can poll simply by passing a pointer to a drained_end_counter and wait until it reaches 0. This patch also adds a non-polling global wrapper for bdrv_do_drained_end() that takes a drained_end_counter pointer. We need such a variant because now no function called anywhere from bdrv_do_drained_end() must poll. This includes BdrvChildRole.drained_end(), which already must not poll according to its interface documentation, but bdrv_child_cb_drained_end() just violates that by invoking bdrv_drained_end() (which does poll). Therefore, BdrvChildRole.drained_end() must take a *drained_end_counter parameter, which bdrv_child_cb_drained_end() can pass on to the new bdrv_drained_end_no_poll() function. Note that we now have a pattern of all drained_end-related functions either polling or receiving a *drained_end_counter to let the caller poll based on that. A problem with a single poll loop is that when the drained section in bdrv_set_aio_context_ignore() ends, some nodes in the subgraph may be in the old contexts, while others are in the new context already. To let the collective poll in bdrv_drained_end() work correctly, we must not hold a lock to the old context, so that the old context can make progress in case it is different from the current context. (In the process, remove the comment saying that the current context is always the old context, because it is wrong.) In all other places, all nodes in a subtree must be in the same context, so we can just poll that. The exception of course is bdrv_drain_all_end(), but that always runs in the main context, so we can just poll NULL (like bdrv_drain_all_begin() does). Signed-off-by: Max Reitz <mreitz@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2019-07-19 11:26:14 +02:00
if (qemu_get_aio_context() != old_context) {
block: Do not poll in bdrv_do_drained_end() We should never poll anywhere in bdrv_do_drained_end() (including its recursive callees like bdrv_drain_invoke()), because it does not cope well with graph changes. In fact, it has been written based on the postulation that no graph changes will happen in it. Instead, the callers that want to poll must poll, i.e. all currently globally available wrappers: bdrv_drained_end(), bdrv_subtree_drained_end(), bdrv_unapply_subtree_drain(), and bdrv_drain_all_end(). Graph changes there do not matter. They can poll simply by passing a pointer to a drained_end_counter and wait until it reaches 0. This patch also adds a non-polling global wrapper for bdrv_do_drained_end() that takes a drained_end_counter pointer. We need such a variant because now no function called anywhere from bdrv_do_drained_end() must poll. This includes BdrvChildRole.drained_end(), which already must not poll according to its interface documentation, but bdrv_child_cb_drained_end() just violates that by invoking bdrv_drained_end() (which does poll). Therefore, BdrvChildRole.drained_end() must take a *drained_end_counter parameter, which bdrv_child_cb_drained_end() can pass on to the new bdrv_drained_end_no_poll() function. Note that we now have a pattern of all drained_end-related functions either polling or receiving a *drained_end_counter to let the caller poll based on that. A problem with a single poll loop is that when the drained section in bdrv_set_aio_context_ignore() ends, some nodes in the subgraph may be in the old contexts, while others are in the new context already. To let the collective poll in bdrv_drained_end() work correctly, we must not hold a lock to the old context, so that the old context can make progress in case it is different from the current context. (In the process, remove the comment saying that the current context is always the old context, because it is wrong.) In all other places, all nodes in a subtree must be in the same context, so we can just poll that. The exception of course is bdrv_drain_all_end(), but that always runs in the main context, so we can just poll NULL (like bdrv_drain_all_begin() does). Signed-off-by: Max Reitz <mreitz@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2019-07-19 11:26:14 +02:00
aio_context_acquire(old_context);
}
if (qemu_get_aio_context() != new_context) {
block: Do not poll in bdrv_do_drained_end() We should never poll anywhere in bdrv_do_drained_end() (including its recursive callees like bdrv_drain_invoke()), because it does not cope well with graph changes. In fact, it has been written based on the postulation that no graph changes will happen in it. Instead, the callers that want to poll must poll, i.e. all currently globally available wrappers: bdrv_drained_end(), bdrv_subtree_drained_end(), bdrv_unapply_subtree_drain(), and bdrv_drain_all_end(). Graph changes there do not matter. They can poll simply by passing a pointer to a drained_end_counter and wait until it reaches 0. This patch also adds a non-polling global wrapper for bdrv_do_drained_end() that takes a drained_end_counter pointer. We need such a variant because now no function called anywhere from bdrv_do_drained_end() must poll. This includes BdrvChildRole.drained_end(), which already must not poll according to its interface documentation, but bdrv_child_cb_drained_end() just violates that by invoking bdrv_drained_end() (which does poll). Therefore, BdrvChildRole.drained_end() must take a *drained_end_counter parameter, which bdrv_child_cb_drained_end() can pass on to the new bdrv_drained_end_no_poll() function. Note that we now have a pattern of all drained_end-related functions either polling or receiving a *drained_end_counter to let the caller poll based on that. A problem with a single poll loop is that when the drained section in bdrv_set_aio_context_ignore() ends, some nodes in the subgraph may be in the old contexts, while others are in the new context already. To let the collective poll in bdrv_drained_end() work correctly, we must not hold a lock to the old context, so that the old context can make progress in case it is different from the current context. (In the process, remove the comment saying that the current context is always the old context, because it is wrong.) In all other places, all nodes in a subtree must be in the same context, so we can just poll that. The exception of course is bdrv_drain_all_end(), but that always runs in the main context, so we can just poll NULL (like bdrv_drain_all_begin() does). Signed-off-by: Max Reitz <mreitz@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2019-07-19 11:26:14 +02:00
aio_context_release(new_context);
}
}
static bool bdrv_parent_can_set_aio_context(BdrvChild *c, AioContext *ctx,
GSList **ignore, Error **errp)
{
GLOBAL_STATE_CODE();
if (g_slist_find(*ignore, c)) {
return true;
}
*ignore = g_slist_prepend(*ignore, c);
/*
* A BdrvChildClass that doesn't handle AioContext changes cannot
* tolerate any AioContext changes
*/
if (!c->klass->can_set_aio_ctx) {
char *user = bdrv_child_user_desc(c);
error_setg(errp, "Changing iothreads is not supported by %s", user);
g_free(user);
return false;
}
if (!c->klass->can_set_aio_ctx(c, ctx, ignore, errp)) {
assert(!errp || *errp);
return false;
}
return true;
}
bool bdrv_child_can_set_aio_context(BdrvChild *c, AioContext *ctx,
GSList **ignore, Error **errp)
{
GLOBAL_STATE_CODE();
if (g_slist_find(*ignore, c)) {
return true;
}
*ignore = g_slist_prepend(*ignore, c);
return bdrv_can_set_aio_context(c->bs, ctx, ignore, errp);
}
/* @ignore will accumulate all visited BdrvChild object. The caller is
* responsible for freeing the list afterwards. */
bool bdrv_can_set_aio_context(BlockDriverState *bs, AioContext *ctx,
GSList **ignore, Error **errp)
{
BdrvChild *c;
if (bdrv_get_aio_context(bs) == ctx) {
return true;
}
GLOBAL_STATE_CODE();
QLIST_FOREACH(c, &bs->parents, next_parent) {
if (!bdrv_parent_can_set_aio_context(c, ctx, ignore, errp)) {
return false;
}
}
QLIST_FOREACH(c, &bs->children, next) {
if (!bdrv_child_can_set_aio_context(c, ctx, ignore, errp)) {
return false;
}
}
return true;
}
int bdrv_child_try_set_aio_context(BlockDriverState *bs, AioContext *ctx,
BdrvChild *ignore_child, Error **errp)
{
GSList *ignore;
bool ret;
GLOBAL_STATE_CODE();
ignore = ignore_child ? g_slist_prepend(NULL, ignore_child) : NULL;
ret = bdrv_can_set_aio_context(bs, ctx, &ignore, errp);
g_slist_free(ignore);
if (!ret) {
return -EPERM;
}
ignore = ignore_child ? g_slist_prepend(NULL, ignore_child) : NULL;
bdrv_set_aio_context_ignore(bs, ctx, &ignore);
g_slist_free(ignore);
return 0;
}
int bdrv_try_set_aio_context(BlockDriverState *bs, AioContext *ctx,
Error **errp)
{
GLOBAL_STATE_CODE();
return bdrv_child_try_set_aio_context(bs, ctx, NULL, errp);
}
void bdrv_add_aio_context_notifier(BlockDriverState *bs,
void (*attached_aio_context)(AioContext *new_context, void *opaque),
void (*detach_aio_context)(void *opaque), void *opaque)
{
BdrvAioNotifier *ban = g_new(BdrvAioNotifier, 1);
*ban = (BdrvAioNotifier){
.attached_aio_context = attached_aio_context,
.detach_aio_context = detach_aio_context,
.opaque = opaque
};
GLOBAL_STATE_CODE();
QLIST_INSERT_HEAD(&bs->aio_notifiers, ban, list);
}
void bdrv_remove_aio_context_notifier(BlockDriverState *bs,
void (*attached_aio_context)(AioContext *,
void *),
void (*detach_aio_context)(void *),
void *opaque)
{
BdrvAioNotifier *ban, *ban_next;
GLOBAL_STATE_CODE();
QLIST_FOREACH_SAFE(ban, &bs->aio_notifiers, list, ban_next) {
if (ban->attached_aio_context == attached_aio_context &&
ban->detach_aio_context == detach_aio_context &&
ban->opaque == opaque &&
ban->deleted == false)
{
if (bs->walking_aio_notifiers) {
ban->deleted = true;
} else {
bdrv_do_remove_aio_context_notifier(ban);
}
return;
}
}
abort();
}
int bdrv_amend_options(BlockDriverState *bs, QemuOpts *opts,
BlockDriverAmendStatusCB *status_cb, void *cb_opaque,
bool force,
Error **errp)
{
GLOBAL_STATE_CODE();
if (!bs->drv) {
error_setg(errp, "Node is ejected");
return -ENOMEDIUM;
}
if (!bs->drv->bdrv_amend_options) {
error_setg(errp, "Block driver '%s' does not support option amendment",
bs->drv->format_name);
return -ENOTSUP;
}
return bs->drv->bdrv_amend_options(bs, opts, status_cb,
cb_opaque, force, errp);
}
/*
* This function checks whether the given @to_replace is allowed to be
* replaced by a node that always shows the same data as @bs. This is
* used for example to verify whether the mirror job can replace
* @to_replace by the target mirrored from @bs.
* To be replaceable, @bs and @to_replace may either be guaranteed to
* always show the same data (because they are only connected through
* filters), or some driver may allow replacing one of its children
* because it can guarantee that this child's data is not visible at
* all (for example, for dissenting quorum children that have no other
* parents).
*/
bool bdrv_recurse_can_replace(BlockDriverState *bs,
BlockDriverState *to_replace)
{
BlockDriverState *filtered;
GLOBAL_STATE_CODE();
if (!bs || !bs->drv) {
return false;
}
if (bs == to_replace) {
return true;
}
/* See what the driver can do */
if (bs->drv->bdrv_recurse_can_replace) {
return bs->drv->bdrv_recurse_can_replace(bs, to_replace);
}
/* For filters without an own implementation, we can recurse on our own */
filtered = bdrv_filter_bs(bs);
if (filtered) {
return bdrv_recurse_can_replace(filtered, to_replace);
}
/* Safe default */
return false;
}
/*
* Check whether the given @node_name can be replaced by a node that
* has the same data as @parent_bs. If so, return @node_name's BDS;
* NULL otherwise.
*
* @node_name must be a (recursive) *child of @parent_bs (or this
* function will return NULL).
*
* The result (whether the node can be replaced or not) is only valid
* for as long as no graph or permission changes occur.
*/
BlockDriverState *check_to_replace_node(BlockDriverState *parent_bs,
const char *node_name, Error **errp)
{
BlockDriverState *to_replace_bs = bdrv_find_node(node_name);
AioContext *aio_context;
GLOBAL_STATE_CODE();
if (!to_replace_bs) {
error_setg(errp, "Failed to find node with node-name='%s'", node_name);
return NULL;
}
aio_context = bdrv_get_aio_context(to_replace_bs);
aio_context_acquire(aio_context);
if (bdrv_op_is_blocked(to_replace_bs, BLOCK_OP_TYPE_REPLACE, errp)) {
to_replace_bs = NULL;
goto out;
}
/* We don't want arbitrary node of the BDS chain to be replaced only the top
* most non filter in order to prevent data corruption.
* Another benefit is that this tests exclude backing files which are
* blocked by the backing blockers.
*/
if (!bdrv_recurse_can_replace(parent_bs, to_replace_bs)) {
error_setg(errp, "Cannot replace '%s' by a node mirrored from '%s', "
"because it cannot be guaranteed that doing so would not "
"lead to an abrupt change of visible data",
node_name, parent_bs->node_name);
to_replace_bs = NULL;
goto out;
}
out:
aio_context_release(aio_context);
return to_replace_bs;
}
/**
* Iterates through the list of runtime option keys that are said to
* be "strong" for a BDS. An option is called "strong" if it changes
* a BDS's data. For example, the null block driver's "size" and
* "read-zeroes" options are strong, but its "latency-ns" option is
* not.
*
* If a key returned by this function ends with a dot, all options
* starting with that prefix are strong.
*/
static const char *const *strong_options(BlockDriverState *bs,
const char *const *curopt)
{
static const char *const global_options[] = {
"driver", "filename", NULL
};
if (!curopt) {
return &global_options[0];
}
curopt++;
if (curopt == &global_options[ARRAY_SIZE(global_options) - 1] && bs->drv) {
curopt = bs->drv->strong_runtime_opts;
}
return (curopt && *curopt) ? curopt : NULL;
}
/**
* Copies all strong runtime options from bs->options to the given
* QDict. The set of strong option keys is determined by invoking
* strong_options().
*
* Returns true iff any strong option was present in bs->options (and
* thus copied to the target QDict) with the exception of "filename"
* and "driver". The caller is expected to use this value to decide
* whether the existence of strong options prevents the generation of
* a plain filename.
*/
static bool append_strong_runtime_options(QDict *d, BlockDriverState *bs)
{
bool found_any = false;
const char *const *option_name = NULL;
if (!bs->drv) {
return false;
}
while ((option_name = strong_options(bs, option_name))) {
bool option_given = false;
assert(strlen(*option_name) > 0);
if ((*option_name)[strlen(*option_name) - 1] != '.') {
QObject *entry = qdict_get(bs->options, *option_name);
if (!entry) {
continue;
}
qdict_put_obj(d, *option_name, qobject_ref(entry));
option_given = true;
} else {
const QDictEntry *entry;
for (entry = qdict_first(bs->options); entry;
entry = qdict_next(bs->options, entry))
{
if (strstart(qdict_entry_key(entry), *option_name, NULL)) {
qdict_put_obj(d, qdict_entry_key(entry),
qobject_ref(qdict_entry_value(entry)));
option_given = true;
}
}
}
/* While "driver" and "filename" need to be included in a JSON filename,
* their existence does not prohibit generation of a plain filename. */
if (!found_any && option_given &&
strcmp(*option_name, "driver") && strcmp(*option_name, "filename"))
{
found_any = true;
}
}
if (!qdict_haskey(d, "driver")) {
/* Drivers created with bdrv_new_open_driver() may not have a
* @driver option. Add it here. */
qdict_put_str(d, "driver", bs->drv->format_name);
}
return found_any;
}
/* Note: This function may return false positives; it may return true
* even if opening the backing file specified by bs's image header
* would result in exactly bs->backing. */
static bool bdrv_backing_overridden(BlockDriverState *bs)
{
GLOBAL_STATE_CODE();
if (bs->backing) {
return strcmp(bs->auto_backing_file,
bs->backing->bs->filename);
} else {
/* No backing BDS, so if the image header reports any backing
* file, it must have been suppressed */
return bs->auto_backing_file[0] != '\0';
}
}
/* Updates the following BDS fields:
* - exact_filename: A filename which may be used for opening a block device
* which (mostly) equals the given BDS (even without any
* other options; so reading and writing must return the same
* results, but caching etc. may be different)
* - full_open_options: Options which, when given when opening a block device
* (without a filename), result in a BDS (mostly)
* equalling the given one
* - filename: If exact_filename is set, it is copied here. Otherwise,
* full_open_options is converted to a JSON object, prefixed with
* "json:" (for use through the JSON pseudo protocol) and put here.
*/
void bdrv_refresh_filename(BlockDriverState *bs)
{
BlockDriver *drv = bs->drv;
BdrvChild *child;
BlockDriverState *primary_child_bs;
QDict *opts;
bool backing_overridden;
bool generate_json_filename; /* Whether our default implementation should
fill exact_filename (false) or not (true) */
GLOBAL_STATE_CODE();
if (!drv) {
return;
}
/* This BDS's file name may depend on any of its children's file names, so
* refresh those first */
QLIST_FOREACH(child, &bs->children, next) {
bdrv_refresh_filename(child->bs);
}
if (bs->implicit) {
/* For implicit nodes, just copy everything from the single child */
child = QLIST_FIRST(&bs->children);
assert(QLIST_NEXT(child, next) == NULL);
pstrcpy(bs->exact_filename, sizeof(bs->exact_filename),
child->bs->exact_filename);
pstrcpy(bs->filename, sizeof(bs->filename), child->bs->filename);
block: fix memleaks in bdrv_refresh_filename If we call the qmp 'query-block' while qemu is working on 'block-commit', it will cause memleaks, the memory leak stack is as follow: Indirect leak of 12360 byte(s) in 3 object(s) allocated from: #0 0x7f80f0b6d970 in __interceptor_calloc (/lib64/libasan.so.5+0xef970) #1 0x7f80ee86049d in g_malloc0 (/lib64/libglib-2.0.so.0+0x5249d) #2 0x55ea95b5bb67 in qdict_new /mnt/sdb/qemu-4.2.0-rc0/qobject/qdict.c:29 #3 0x55ea956cd043 in bdrv_refresh_filename /mnt/sdb/qemu-4.2.0-rc0/block.c:6427 #4 0x55ea956cc950 in bdrv_refresh_filename /mnt/sdb/qemu-4.2.0-rc0/block.c:6399 #5 0x55ea956cc950 in bdrv_refresh_filename /mnt/sdb/qemu-4.2.0-rc0/block.c:6399 #6 0x55ea956cc950 in bdrv_refresh_filename /mnt/sdb/qemu-4.2.0-rc0/block.c:6399 #7 0x55ea958818ea in bdrv_block_device_info /mnt/sdb/qemu-4.2.0-rc0/block/qapi.c:56 #8 0x55ea958879de in bdrv_query_info /mnt/sdb/qemu-4.2.0-rc0/block/qapi.c:392 #9 0x55ea9588b58f in qmp_query_block /mnt/sdb/qemu-4.2.0-rc0/block/qapi.c:578 #10 0x55ea95567392 in qmp_marshal_query_block qapi/qapi-commands-block-core.c:95 Indirect leak of 4120 byte(s) in 1 object(s) allocated from: #0 0x7f80f0b6d970 in __interceptor_calloc (/lib64/libasan.so.5+0xef970) #1 0x7f80ee86049d in g_malloc0 (/lib64/libglib-2.0.so.0+0x5249d) #2 0x55ea95b5bb67 in qdict_new /mnt/sdb/qemu-4.2.0-rc0/qobject/qdict.c:29 #3 0x55ea956cd043 in bdrv_refresh_filename /mnt/sdb/qemu-4.2.0-rc0/block.c:6427 #4 0x55ea956cc950 in bdrv_refresh_filename /mnt/sdb/qemu-4.2.0-rc0/block.c:6399 #5 0x55ea956cc950 in bdrv_refresh_filename /mnt/sdb/qemu-4.2.0-rc0/block.c:6399 #6 0x55ea9569f301 in bdrv_backing_attach /mnt/sdb/qemu-4.2.0-rc0/block.c:1064 #7 0x55ea956a99dd in bdrv_replace_child_noperm /mnt/sdb/qemu-4.2.0-rc0/block.c:2283 #8 0x55ea956b9b53 in bdrv_replace_node /mnt/sdb/qemu-4.2.0-rc0/block.c:4196 #9 0x55ea956b9e49 in bdrv_append /mnt/sdb/qemu-4.2.0-rc0/block.c:4236 #10 0x55ea958c3472 in commit_start /mnt/sdb/qemu-4.2.0-rc0/block/commit.c:306 #11 0x55ea94b68ab0 in qmp_block_commit /mnt/sdb/qemu-4.2.0-rc0/blockdev.c:3459 #12 0x55ea9556a7a7 in qmp_marshal_block_commit qapi/qapi-commands-block-core.c:407 Fixes: bb808d5f5c0978828a974d547e6032402c339555 Reported-by: Euler Robot <euler.robot@huawei.com> Signed-off-by: Pan Nengyuan <pannengyuan@huawei.com> Message-id: 20200116085600.24056-1-pannengyuan@huawei.com Signed-off-by: Max Reitz <mreitz@redhat.com>
2020-01-16 09:56:00 +01:00
qobject_unref(bs->full_open_options);
bs->full_open_options = qobject_ref(child->bs->full_open_options);
return;
}
backing_overridden = bdrv_backing_overridden(bs);
if (bs->open_flags & BDRV_O_NO_IO) {
/* Without I/O, the backing file does not change anything.
* Therefore, in such a case (primarily qemu-img), we can
* pretend the backing file has not been overridden even if
* it technically has been. */
backing_overridden = false;
}
/* Gather the options QDict */
opts = qdict_new();
generate_json_filename = append_strong_runtime_options(opts, bs);
generate_json_filename |= backing_overridden;
if (drv->bdrv_gather_child_options) {
/* Some block drivers may not want to present all of their children's
* options, or name them differently from BdrvChild.name */
drv->bdrv_gather_child_options(bs, opts, backing_overridden);
} else {
QLIST_FOREACH(child, &bs->children, next) {
if (child == bs->backing && !backing_overridden) {
/* We can skip the backing BDS if it has not been overridden */
continue;
}
qdict_put(opts, child->name,
qobject_ref(child->bs->full_open_options));
}
if (backing_overridden && !bs->backing) {
/* Force no backing file */
qdict_put_null(opts, "backing");
}
}
qobject_unref(bs->full_open_options);
bs->full_open_options = opts;
primary_child_bs = bdrv_primary_bs(bs);
if (drv->bdrv_refresh_filename) {
/* Obsolete information is of no use here, so drop the old file name
* information before refreshing it */
bs->exact_filename[0] = '\0';
drv->bdrv_refresh_filename(bs);
} else if (primary_child_bs) {
/*
* Try to reconstruct valid information from the underlying
* file -- this only works for format nodes (filter nodes
* cannot be probed and as such must be selected by the user
* either through an options dict, or through a special
* filename which the filter driver must construct in its
* .bdrv_refresh_filename() implementation).
*/
bs->exact_filename[0] = '\0';
/*
* We can use the underlying file's filename if:
* - it has a filename,
* - the current BDS is not a filter,
* - the file is a protocol BDS, and
* - opening that file (as this BDS's format) will automatically create
* the BDS tree we have right now, that is:
* - the user did not significantly change this BDS's behavior with
* some explicit (strong) options
* - no non-file child of this BDS has been overridden by the user
* Both of these conditions are represented by generate_json_filename.
*/
if (primary_child_bs->exact_filename[0] &&
primary_child_bs->drv->bdrv_file_open &&
!drv->is_filter && !generate_json_filename)
{
strcpy(bs->exact_filename, primary_child_bs->exact_filename);
}
}
if (bs->exact_filename[0]) {
pstrcpy(bs->filename, sizeof(bs->filename), bs->exact_filename);
} else {
GString *json = qobject_to_json(QOBJECT(bs->full_open_options));
if (snprintf(bs->filename, sizeof(bs->filename), "json:%s",
json->str) >= sizeof(bs->filename)) {
/* Give user a hint if we truncated things. */
strcpy(bs->filename + sizeof(bs->filename) - 4, "...");
}
g_string_free(json, true);
}
}
char *bdrv_dirname(BlockDriverState *bs, Error **errp)
{
BlockDriver *drv = bs->drv;
BlockDriverState *child_bs;
GLOBAL_STATE_CODE();
if (!drv) {
error_setg(errp, "Node '%s' is ejected", bs->node_name);
return NULL;
}
if (drv->bdrv_dirname) {
return drv->bdrv_dirname(bs, errp);
}
child_bs = bdrv_primary_bs(bs);
if (child_bs) {
return bdrv_dirname(child_bs, errp);
}
bdrv_refresh_filename(bs);
if (bs->exact_filename[0] != '\0') {
return path_combine(bs->exact_filename, "");
}
error_setg(errp, "Cannot generate a base directory for %s nodes",
drv->format_name);
return NULL;
}
/*
* Hot add/remove a BDS's child. So the user can take a child offline when
* it is broken and take a new child online
*/
void bdrv_add_child(BlockDriverState *parent_bs, BlockDriverState *child_bs,
Error **errp)
{
GLOBAL_STATE_CODE();
if (!parent_bs->drv || !parent_bs->drv->bdrv_add_child) {
error_setg(errp, "The node %s does not support adding a child",
bdrv_get_device_or_node_name(parent_bs));
return;
}
if (!QLIST_EMPTY(&child_bs->parents)) {
error_setg(errp, "The node %s already has a parent",
child_bs->node_name);
return;
}
parent_bs->drv->bdrv_add_child(parent_bs, child_bs, errp);
}
void bdrv_del_child(BlockDriverState *parent_bs, BdrvChild *child, Error **errp)
{
BdrvChild *tmp;
GLOBAL_STATE_CODE();
if (!parent_bs->drv || !parent_bs->drv->bdrv_del_child) {
error_setg(errp, "The node %s does not support removing a child",
bdrv_get_device_or_node_name(parent_bs));
return;
}
QLIST_FOREACH(tmp, &parent_bs->children, next) {
if (tmp == child) {
break;
}
}
if (!tmp) {
error_setg(errp, "The node %s does not have a child named %s",
bdrv_get_device_or_node_name(parent_bs),
bdrv_get_device_or_node_name(child->bs));
return;
}
parent_bs->drv->bdrv_del_child(parent_bs, child, errp);
}
int bdrv_make_empty(BdrvChild *c, Error **errp)
{
BlockDriver *drv = c->bs->drv;
int ret;
GLOBAL_STATE_CODE();
assert(c->perm & (BLK_PERM_WRITE | BLK_PERM_WRITE_UNCHANGED));
if (!drv->bdrv_make_empty) {
error_setg(errp, "%s does not support emptying nodes",
drv->format_name);
return -ENOTSUP;
}
ret = drv->bdrv_make_empty(c->bs);
if (ret < 0) {
error_setg_errno(errp, -ret, "Failed to empty %s",
c->bs->filename);
return ret;
}
return 0;
}
/*
* Return the child that @bs acts as an overlay for, and from which data may be
* copied in COW or COR operations. Usually this is the backing file.
*/
BdrvChild *bdrv_cow_child(BlockDriverState *bs)
{
IO_CODE();
if (!bs || !bs->drv) {
return NULL;
}
if (bs->drv->is_filter) {
return NULL;
}
if (!bs->backing) {
return NULL;
}
assert(bs->backing->role & BDRV_CHILD_COW);
return bs->backing;
}
/*
* If @bs acts as a filter for exactly one of its children, return
* that child.
*/
BdrvChild *bdrv_filter_child(BlockDriverState *bs)
{
BdrvChild *c;
IO_CODE();
if (!bs || !bs->drv) {
return NULL;
}
if (!bs->drv->is_filter) {
return NULL;
}
/* Only one of @backing or @file may be used */
assert(!(bs->backing && bs->file));
c = bs->backing ?: bs->file;
if (!c) {
return NULL;
}
assert(c->role & BDRV_CHILD_FILTERED);
return c;
}
/*
* Return either the result of bdrv_cow_child() or bdrv_filter_child(),
* whichever is non-NULL.
*
* Return NULL if both are NULL.
*/
BdrvChild *bdrv_filter_or_cow_child(BlockDriverState *bs)
{
BdrvChild *cow_child = bdrv_cow_child(bs);
BdrvChild *filter_child = bdrv_filter_child(bs);
IO_CODE();
/* Filter nodes cannot have COW backing files */
assert(!(cow_child && filter_child));
return cow_child ?: filter_child;
}
/*
* Return the primary child of this node: For filters, that is the
* filtered child. For other nodes, that is usually the child storing
* metadata.
* (A generally more helpful description is that this is (usually) the
* child that has the same filename as @bs.)
*
* Drivers do not necessarily have a primary child; for example quorum
* does not.
*/
BdrvChild *bdrv_primary_child(BlockDriverState *bs)
{
BdrvChild *c, *found = NULL;
IO_CODE();
QLIST_FOREACH(c, &bs->children, next) {
if (c->role & BDRV_CHILD_PRIMARY) {
assert(!found);
found = c;
}
}
return found;
}
static BlockDriverState *bdrv_do_skip_filters(BlockDriverState *bs,
bool stop_on_explicit_filter)
{
BdrvChild *c;
if (!bs) {
return NULL;
}
while (!(stop_on_explicit_filter && !bs->implicit)) {
c = bdrv_filter_child(bs);
if (!c) {
/*
* A filter that is embedded in a working block graph must
* have a child. Assert this here so this function does
* not return a filter node that is not expected by the
* caller.
*/
assert(!bs->drv || !bs->drv->is_filter);
break;
}
bs = c->bs;
}
/*
* Note that this treats nodes with bs->drv == NULL as not being
* filters (bs->drv == NULL should be replaced by something else
* anyway).
* The advantage of this behavior is that this function will thus
* always return a non-NULL value (given a non-NULL @bs).
*/
return bs;
}
/*
* Return the first BDS that has not been added implicitly or that
* does not have a filtered child down the chain starting from @bs
* (including @bs itself).
*/
BlockDriverState *bdrv_skip_implicit_filters(BlockDriverState *bs)
{
GLOBAL_STATE_CODE();
return bdrv_do_skip_filters(bs, true);
}
/*
* Return the first BDS that does not have a filtered child down the
* chain starting from @bs (including @bs itself).
*/
BlockDriverState *bdrv_skip_filters(BlockDriverState *bs)
{
IO_CODE();
return bdrv_do_skip_filters(bs, false);
}
/*
* For a backing chain, return the first non-filter backing image of
* the first non-filter image.
*/
BlockDriverState *bdrv_backing_chain_next(BlockDriverState *bs)
{
IO_CODE();
return bdrv_skip_filters(bdrv_cow_bs(bdrv_skip_filters(bs)));
}
block: block-status cache for data regions As we have attempted before (https://lists.gnu.org/archive/html/qemu-devel/2019-01/msg06451.html, "file-posix: Cache lseek result for data regions"; https://lists.nongnu.org/archive/html/qemu-block/2021-02/msg00934.html, "file-posix: Cache next hole"), this patch seeks to reduce the number of SEEK_DATA/HOLE operations the file-posix driver has to perform. The main difference is that this time it is implemented as part of the general block layer code. The problem we face is that on some filesystems or in some circumstances, SEEK_DATA/HOLE is unreasonably slow. Given the implementation is outside of qemu, there is little we can do about its performance. We have already introduced the want_zero parameter to bdrv_co_block_status() to reduce the number of SEEK_DATA/HOLE calls unless we really want zero information; but sometimes we do want that information, because for files that consist largely of zero areas, special-casing those areas can give large performance boosts. So the real problem is with files that consist largely of data, so that inquiring the block status does not gain us much performance, but where such an inquiry itself takes a lot of time. To address this, we want to cache data regions. Most of the time, when bad performance is reported, it is in places where the image is iterated over from start to end (qemu-img convert or the mirror job), so a simple yet effective solution is to cache only the current data region. (Note that only caching data regions but not zero regions means that returning false information from the cache is not catastrophic: Treating zeroes as data is fine. While we try to invalidate the cache on zero writes and discards, such incongruences may still occur when there are other processes writing to the image.) We only use the cache for nodes without children (i.e. protocol nodes), because that is where the problem is: Drivers that rely on block-status implementations outside of qemu (e.g. SEEK_DATA/HOLE). Resolves: https://gitlab.com/qemu-project/qemu/-/issues/307 Signed-off-by: Hanna Reitz <hreitz@redhat.com> Message-Id: <20210812084148.14458-3-hreitz@redhat.com> Reviewed-by: Eric Blake <eblake@redhat.com> Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> [hreitz: Added `local_file == bs` assertion, as suggested by Vladimir] Signed-off-by: Hanna Reitz <hreitz@redhat.com>
2021-08-12 10:41:44 +02:00
/**
* Check whether [offset, offset + bytes) overlaps with the cached
* block-status data region.
*
* If so, and @pnum is not NULL, set *pnum to `bsc.data_end - offset`,
* which is what bdrv_bsc_is_data()'s interface needs.
* Otherwise, *pnum is not touched.
*/
static bool bdrv_bsc_range_overlaps_locked(BlockDriverState *bs,
int64_t offset, int64_t bytes,
int64_t *pnum)
{
BdrvBlockStatusCache *bsc = qatomic_rcu_read(&bs->block_status_cache);
bool overlaps;
overlaps =
qatomic_read(&bsc->valid) &&
ranges_overlap(offset, bytes, bsc->data_start,
bsc->data_end - bsc->data_start);
if (overlaps && pnum) {
*pnum = bsc->data_end - offset;
}
return overlaps;
}
/**
* See block_int.h for this function's documentation.
*/
bool bdrv_bsc_is_data(BlockDriverState *bs, int64_t offset, int64_t *pnum)
{
IO_CODE();
block: block-status cache for data regions As we have attempted before (https://lists.gnu.org/archive/html/qemu-devel/2019-01/msg06451.html, "file-posix: Cache lseek result for data regions"; https://lists.nongnu.org/archive/html/qemu-block/2021-02/msg00934.html, "file-posix: Cache next hole"), this patch seeks to reduce the number of SEEK_DATA/HOLE operations the file-posix driver has to perform. The main difference is that this time it is implemented as part of the general block layer code. The problem we face is that on some filesystems or in some circumstances, SEEK_DATA/HOLE is unreasonably slow. Given the implementation is outside of qemu, there is little we can do about its performance. We have already introduced the want_zero parameter to bdrv_co_block_status() to reduce the number of SEEK_DATA/HOLE calls unless we really want zero information; but sometimes we do want that information, because for files that consist largely of zero areas, special-casing those areas can give large performance boosts. So the real problem is with files that consist largely of data, so that inquiring the block status does not gain us much performance, but where such an inquiry itself takes a lot of time. To address this, we want to cache data regions. Most of the time, when bad performance is reported, it is in places where the image is iterated over from start to end (qemu-img convert or the mirror job), so a simple yet effective solution is to cache only the current data region. (Note that only caching data regions but not zero regions means that returning false information from the cache is not catastrophic: Treating zeroes as data is fine. While we try to invalidate the cache on zero writes and discards, such incongruences may still occur when there are other processes writing to the image.) We only use the cache for nodes without children (i.e. protocol nodes), because that is where the problem is: Drivers that rely on block-status implementations outside of qemu (e.g. SEEK_DATA/HOLE). Resolves: https://gitlab.com/qemu-project/qemu/-/issues/307 Signed-off-by: Hanna Reitz <hreitz@redhat.com> Message-Id: <20210812084148.14458-3-hreitz@redhat.com> Reviewed-by: Eric Blake <eblake@redhat.com> Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> [hreitz: Added `local_file == bs` assertion, as suggested by Vladimir] Signed-off-by: Hanna Reitz <hreitz@redhat.com>
2021-08-12 10:41:44 +02:00
RCU_READ_LOCK_GUARD();
return bdrv_bsc_range_overlaps_locked(bs, offset, 1, pnum);
}
/**
* See block_int.h for this function's documentation.
*/
void bdrv_bsc_invalidate_range(BlockDriverState *bs,
int64_t offset, int64_t bytes)
{
IO_CODE();
block: block-status cache for data regions As we have attempted before (https://lists.gnu.org/archive/html/qemu-devel/2019-01/msg06451.html, "file-posix: Cache lseek result for data regions"; https://lists.nongnu.org/archive/html/qemu-block/2021-02/msg00934.html, "file-posix: Cache next hole"), this patch seeks to reduce the number of SEEK_DATA/HOLE operations the file-posix driver has to perform. The main difference is that this time it is implemented as part of the general block layer code. The problem we face is that on some filesystems or in some circumstances, SEEK_DATA/HOLE is unreasonably slow. Given the implementation is outside of qemu, there is little we can do about its performance. We have already introduced the want_zero parameter to bdrv_co_block_status() to reduce the number of SEEK_DATA/HOLE calls unless we really want zero information; but sometimes we do want that information, because for files that consist largely of zero areas, special-casing those areas can give large performance boosts. So the real problem is with files that consist largely of data, so that inquiring the block status does not gain us much performance, but where such an inquiry itself takes a lot of time. To address this, we want to cache data regions. Most of the time, when bad performance is reported, it is in places where the image is iterated over from start to end (qemu-img convert or the mirror job), so a simple yet effective solution is to cache only the current data region. (Note that only caching data regions but not zero regions means that returning false information from the cache is not catastrophic: Treating zeroes as data is fine. While we try to invalidate the cache on zero writes and discards, such incongruences may still occur when there are other processes writing to the image.) We only use the cache for nodes without children (i.e. protocol nodes), because that is where the problem is: Drivers that rely on block-status implementations outside of qemu (e.g. SEEK_DATA/HOLE). Resolves: https://gitlab.com/qemu-project/qemu/-/issues/307 Signed-off-by: Hanna Reitz <hreitz@redhat.com> Message-Id: <20210812084148.14458-3-hreitz@redhat.com> Reviewed-by: Eric Blake <eblake@redhat.com> Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> [hreitz: Added `local_file == bs` assertion, as suggested by Vladimir] Signed-off-by: Hanna Reitz <hreitz@redhat.com>
2021-08-12 10:41:44 +02:00
RCU_READ_LOCK_GUARD();
if (bdrv_bsc_range_overlaps_locked(bs, offset, bytes, NULL)) {
qatomic_set(&bs->block_status_cache->valid, false);
}
}
/**
* See block_int.h for this function's documentation.
*/
void bdrv_bsc_fill(BlockDriverState *bs, int64_t offset, int64_t bytes)
{
BdrvBlockStatusCache *new_bsc = g_new(BdrvBlockStatusCache, 1);
BdrvBlockStatusCache *old_bsc;
IO_CODE();
block: block-status cache for data regions As we have attempted before (https://lists.gnu.org/archive/html/qemu-devel/2019-01/msg06451.html, "file-posix: Cache lseek result for data regions"; https://lists.nongnu.org/archive/html/qemu-block/2021-02/msg00934.html, "file-posix: Cache next hole"), this patch seeks to reduce the number of SEEK_DATA/HOLE operations the file-posix driver has to perform. The main difference is that this time it is implemented as part of the general block layer code. The problem we face is that on some filesystems or in some circumstances, SEEK_DATA/HOLE is unreasonably slow. Given the implementation is outside of qemu, there is little we can do about its performance. We have already introduced the want_zero parameter to bdrv_co_block_status() to reduce the number of SEEK_DATA/HOLE calls unless we really want zero information; but sometimes we do want that information, because for files that consist largely of zero areas, special-casing those areas can give large performance boosts. So the real problem is with files that consist largely of data, so that inquiring the block status does not gain us much performance, but where such an inquiry itself takes a lot of time. To address this, we want to cache data regions. Most of the time, when bad performance is reported, it is in places where the image is iterated over from start to end (qemu-img convert or the mirror job), so a simple yet effective solution is to cache only the current data region. (Note that only caching data regions but not zero regions means that returning false information from the cache is not catastrophic: Treating zeroes as data is fine. While we try to invalidate the cache on zero writes and discards, such incongruences may still occur when there are other processes writing to the image.) We only use the cache for nodes without children (i.e. protocol nodes), because that is where the problem is: Drivers that rely on block-status implementations outside of qemu (e.g. SEEK_DATA/HOLE). Resolves: https://gitlab.com/qemu-project/qemu/-/issues/307 Signed-off-by: Hanna Reitz <hreitz@redhat.com> Message-Id: <20210812084148.14458-3-hreitz@redhat.com> Reviewed-by: Eric Blake <eblake@redhat.com> Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> [hreitz: Added `local_file == bs` assertion, as suggested by Vladimir] Signed-off-by: Hanna Reitz <hreitz@redhat.com>
2021-08-12 10:41:44 +02:00
*new_bsc = (BdrvBlockStatusCache) {
.valid = true,
.data_start = offset,
.data_end = offset + bytes,
};
QEMU_LOCK_GUARD(&bs->bsc_modify_lock);
old_bsc = qatomic_rcu_read(&bs->block_status_cache);
qatomic_rcu_set(&bs->block_status_cache, new_bsc);
if (old_bsc) {
g_free_rcu(old_bsc, rcu);
}
}