qemu-e2k/block/rbd.c

1684 lines
49 KiB
C
Raw Normal View History

/*
* QEMU Block driver for RADOS (Ceph)
*
* Copyright (C) 2010-2011 Christian Brunner <chb@muc.de>,
* Josh Durgin <josh.durgin@dreamhost.com>
*
* This work is licensed under the terms of the GNU GPL, version 2. See
* the COPYING file in the top-level directory.
*
* Contributions after 2012-01-13 are licensed under the terms of the
* GNU GPL, version 2 or (at your option) any later version.
*/
#include "qemu/osdep.h"
#include <rbd/librbd.h>
2016-03-14 09:01:28 +01:00
#include "qapi/error.h"
#include "qemu/error-report.h"
#include "qemu/module.h"
#include "qemu/option.h"
#include "block/block_int.h"
#include "block/qdict.h"
#include "crypto/secret.h"
#include "qemu/cutils.h"
#include "sysemu/replay.h"
#include "qapi/qmp/qstring.h"
#include "qapi/qmp/qdict.h"
rbd: Fix regression in legacy key/values containing escaped : Commit c7cacb3 accidentally broke legacy key-value parsing through pseudo-filename parsing of -drive file=rbd://..., for any key that contains an escaped ':'. Such a key is surprisingly common, thanks to mon_host specifying a 'host:port' string. The break happens because passing things from QDict through QemuOpts back to another QDict requires that we pack our parsed key/value pairs into a string, and then reparse that string, but the intermediate string that we created ("key1=value1:key2=value2") lost the \: escaping that was present in the original, so that we could no longer see which : were used as separators vs. those used as part of the original input. Fix it by collecting the key/value pairs through a QList, and sending that list on a round trip through a JSON QString (as in '["key1","value1","key2","value2"]') on its way through QemuOpts, rather than hand-rolling our own string. Since the string is only handled internally, this was faster than creating a full-blown struct of '[{"key1":"value1"},{"key2":"value2"}]', and safer at guaranteeing order compared to '{"key1":"value1","key2":"value2"}'. It would be nicer if we didn't have to round-trip through QemuOpts in the first place, but that's a much bigger task for later. Reproducer: ./x86_64-softmmu/qemu-system-x86_64 -nodefaults -nographic -qmp stdio \ -drive 'file=rbd:volumes/volume-ea141b5c-cdb3-4765-910d-e7008b209a70'\ ':id=compute:key=AQAVkvxXAAAAABAA9ZxWFYdRmV+DSwKr7BKKXg=='\ ':auth_supported=cephx\;none:mon_host=192.168.1.2\:6789'\ ',format=raw,if=none,id=drive-virtio-disk0,'\ 'serial=ea141b5c-cdb3-4765-910d-e7008b209a70,cache=writeback' Even without an RBD setup, this serves a test of whether we get the incorrect parser error of: qemu-system-x86_64: -drive file=rbd:...cache=writeback: conf option 6789 has no value or the correct behavior of hanging while trying to connect to the requested mon_host of 192.168.1.2:6789. Reported-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com> Signed-off-by: Eric Blake <eblake@redhat.com> Reviewed-by: Jeff Cody <jcody@redhat.com> Reviewed-by: Max Reitz <mreitz@redhat.com> Message-id: 20170331152730.12514-1-eblake@redhat.com Signed-off-by: Jeff Cody <jcody@redhat.com>
2017-03-31 17:27:30 +02:00
#include "qapi/qmp/qjson.h"
#include "qapi/qmp/qlist.h"
#include "qapi/qobject-input-visitor.h"
#include "qapi/qapi-visit-block-core.h"
/*
* When specifying the image filename use:
*
* rbd:poolname/devicename[@snapshotname][:option1=value1[:option2=value2...]]
*
* poolname must be the name of an existing rados pool.
*
* devicename is the name of the rbd image.
*
* Each option given is used to configure rados, and may be any valid
* Ceph option, "id", or "conf".
*
* The "id" option indicates what user we should authenticate as to
* the Ceph cluster. If it is excluded we will use the Ceph default
* (normally 'admin').
*
* The "conf" option specifies a Ceph configuration file to read. If
* it is not specified, we will read from the default Ceph locations
* (e.g., /etc/ceph/ceph.conf). To avoid reading _any_ configuration
* file, specify conf=/dev/null.
*
* Configuration values containing :, @, or = can be escaped with a
* leading "\".
*/
#define OBJ_MAX_SIZE (1UL << OBJ_DEFAULT_OBJ_ORDER)
#define RBD_MAX_SNAPS 100
#define RBD_ENCRYPTION_LUKS_HEADER_VERIFICATION_LEN 8
static const char rbd_luks_header_verification[
RBD_ENCRYPTION_LUKS_HEADER_VERIFICATION_LEN] = {
'L', 'U', 'K', 'S', 0xBA, 0xBE, 0, 1
};
static const char rbd_luks2_header_verification[
RBD_ENCRYPTION_LUKS_HEADER_VERIFICATION_LEN] = {
'L', 'U', 'K', 'S', 0xBA, 0xBE, 0, 2
};
typedef enum {
RBD_AIO_READ,
RBD_AIO_WRITE,
RBD_AIO_DISCARD,
block/rbd: add write zeroes support This patch wittingly sets BDRV_REQ_NO_FALLBACK and silently ignores BDRV_REQ_MAY_UNMAP for older librbd versions. The rationale for this is as follows (citing Ilya Dryomov current RBD maintainer): ---8<--- a) remove the BDRV_REQ_MAY_UNMAP check in qemu_rbd_co_pwrite_zeroes() and as a consequence always unmap if librbd is too old It's not clear what qemu's expectation is but in general Write Zeroes is allowed to unmap. The only guarantee is that subsequent reads return zeroes, everything else is a hint. This is how it is specified in the kernel and in the NVMe spec. In particular, block/nvme.c implements it as follows: if (flags & BDRV_REQ_MAY_UNMAP) { cdw12 |= (1 << 25); } This sets the Deallocate bit. But if it's not set, the device may still deallocate: """ If the Deallocate bit (CDW12.DEAC) is set to '1' in a Write Zeroes command, and the namespace supports clearing all bytes to 0h in the values read (e.g., bits 2:0 in the DLFEAT field are set to 001b) from a deallocated logical block and its metadata (excluding protection information), then for each specified logical block, the controller: - should deallocate that logical block; ... If the Deallocate bit is cleared to '0' in a Write Zeroes command, and the namespace supports clearing all bytes to 0h in the values read (e.g., bits 2:0 in the DLFEAT field are set to 001b) from a deallocated logical block and its metadata (excluding protection information), then, for each specified logical block, the controller: - may deallocate that logical block; """ https://nvmexpress.org/wp-content/uploads/NVM-Express-NVM-Command-Set-Specification-2021.06.02-Ratified-1.pdf b) set BDRV_REQ_NO_FALLBACK in supported_zero_flags Again, it's not clear what qemu expects here, but without it we end up in a ridiculous situation where specifying the "don't allow slow fallback" switch immediately fails all efficient zeroing requests on a device where Write Zeroes is always efficient: $ qemu-io -c 'help write' | grep -- '-[zun]' -n, -- with -z, don't allow slow fallback -u, -- with -z, allow unmapping -z, -- write zeroes using blk_co_pwrite_zeroes $ qemu-io -f rbd -c 'write -z -u -n 0 1M' rbd:foo/bar write failed: Operation not supported --->8--- Signed-off-by: Peter Lieven <pl@kamp.de> Reviewed-by: Ilya Dryomov <idryomov@gmail.com> Message-Id: <20210702172356.11574-6-idryomov@gmail.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2021-07-02 19:23:55 +02:00
RBD_AIO_FLUSH,
RBD_AIO_WRITE_ZEROES
} RBDAIOCmd;
typedef struct BDRVRBDState {
rados_t cluster;
rados_ioctx_t io_ctx;
rbd_image_t image;
char *image_name;
char *snap;
char *namespace;
uint64_t image_size;
uint64_t object_size;
} BDRVRBDState;
typedef struct RBDTask {
BlockDriverState *bs;
Coroutine *co;
bool complete;
int64_t ret;
} RBDTask;
typedef struct RBDDiffIterateReq {
uint64_t offs;
uint64_t bytes;
bool exists;
} RBDDiffIterateReq;
static int qemu_rbd_connect(rados_t *cluster, rados_ioctx_t *io_ctx,
BlockdevOptionsRbd *opts, bool cache,
const char *keypairs, const char *secretid,
Error **errp);
static char *qemu_rbd_strchr(char *src, char delim)
{
char *p;
for (p = src; *p; ++p) {
if (*p == delim) {
return p;
}
if (*p == '\\' && p[1] != '\0') {
++p;
}
}
return NULL;
}
static char *qemu_rbd_next_tok(char *src, char delim, char **p)
{
char *end;
*p = NULL;
end = qemu_rbd_strchr(src, delim);
if (end) {
*p = end + 1;
*end = '\0';
}
return src;
}
static void qemu_rbd_unescape(char *src)
{
char *p;
for (p = src; *src; ++src, ++p) {
if (*src == '\\' && src[1] != '\0') {
src++;
}
*p = *src;
}
*p = '\0';
}
static void qemu_rbd_parse_filename(const char *filename, QDict *options,
Error **errp)
{
const char *start;
rbd: Fix regression in legacy key/values containing escaped : Commit c7cacb3 accidentally broke legacy key-value parsing through pseudo-filename parsing of -drive file=rbd://..., for any key that contains an escaped ':'. Such a key is surprisingly common, thanks to mon_host specifying a 'host:port' string. The break happens because passing things from QDict through QemuOpts back to another QDict requires that we pack our parsed key/value pairs into a string, and then reparse that string, but the intermediate string that we created ("key1=value1:key2=value2") lost the \: escaping that was present in the original, so that we could no longer see which : were used as separators vs. those used as part of the original input. Fix it by collecting the key/value pairs through a QList, and sending that list on a round trip through a JSON QString (as in '["key1","value1","key2","value2"]') on its way through QemuOpts, rather than hand-rolling our own string. Since the string is only handled internally, this was faster than creating a full-blown struct of '[{"key1":"value1"},{"key2":"value2"}]', and safer at guaranteeing order compared to '{"key1":"value1","key2":"value2"}'. It would be nicer if we didn't have to round-trip through QemuOpts in the first place, but that's a much bigger task for later. Reproducer: ./x86_64-softmmu/qemu-system-x86_64 -nodefaults -nographic -qmp stdio \ -drive 'file=rbd:volumes/volume-ea141b5c-cdb3-4765-910d-e7008b209a70'\ ':id=compute:key=AQAVkvxXAAAAABAA9ZxWFYdRmV+DSwKr7BKKXg=='\ ':auth_supported=cephx\;none:mon_host=192.168.1.2\:6789'\ ',format=raw,if=none,id=drive-virtio-disk0,'\ 'serial=ea141b5c-cdb3-4765-910d-e7008b209a70,cache=writeback' Even without an RBD setup, this serves a test of whether we get the incorrect parser error of: qemu-system-x86_64: -drive file=rbd:...cache=writeback: conf option 6789 has no value or the correct behavior of hanging while trying to connect to the requested mon_host of 192.168.1.2:6789. Reported-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com> Signed-off-by: Eric Blake <eblake@redhat.com> Reviewed-by: Jeff Cody <jcody@redhat.com> Reviewed-by: Max Reitz <mreitz@redhat.com> Message-id: 20170331152730.12514-1-eblake@redhat.com Signed-off-by: Jeff Cody <jcody@redhat.com>
2017-03-31 17:27:30 +02:00
char *p, *buf;
QList *keypairs = NULL;
char *found_str, *image_name;
if (!strstart(filename, "rbd:", &start)) {
error_setg(errp, "File name must start with 'rbd:'");
return;
}
buf = g_strdup(start);
p = buf;
found_str = qemu_rbd_next_tok(p, '/', &p);
if (!p) {
error_setg(errp, "Pool name is required");
goto done;
}
qemu_rbd_unescape(found_str);
qdict_put_str(options, "pool", found_str);
if (qemu_rbd_strchr(p, '@')) {
image_name = qemu_rbd_next_tok(p, '@', &p);
found_str = qemu_rbd_next_tok(p, ':', &p);
qemu_rbd_unescape(found_str);
qdict_put_str(options, "snapshot", found_str);
} else {
image_name = qemu_rbd_next_tok(p, ':', &p);
}
/* Check for namespace in the image_name */
if (qemu_rbd_strchr(image_name, '/')) {
found_str = qemu_rbd_next_tok(image_name, '/', &image_name);
qemu_rbd_unescape(found_str);
qdict_put_str(options, "namespace", found_str);
} else {
qdict_put_str(options, "namespace", "");
}
qemu_rbd_unescape(image_name);
qdict_put_str(options, "image", image_name);
if (!p) {
goto done;
}
/* The following are essentially all key/value pairs, and we treat
* 'id' and 'conf' a bit special. Key/value pairs may be in any order. */
while (p) {
char *name, *value;
name = qemu_rbd_next_tok(p, '=', &p);
if (!p) {
error_setg(errp, "conf option %s has no value", name);
break;
}
qemu_rbd_unescape(name);
value = qemu_rbd_next_tok(p, ':', &p);
qemu_rbd_unescape(value);
if (!strcmp(name, "conf")) {
qdict_put_str(options, "conf", value);
} else if (!strcmp(name, "id")) {
qdict_put_str(options, "user", value);
} else {
rbd: Fix regression in legacy key/values containing escaped : Commit c7cacb3 accidentally broke legacy key-value parsing through pseudo-filename parsing of -drive file=rbd://..., for any key that contains an escaped ':'. Such a key is surprisingly common, thanks to mon_host specifying a 'host:port' string. The break happens because passing things from QDict through QemuOpts back to another QDict requires that we pack our parsed key/value pairs into a string, and then reparse that string, but the intermediate string that we created ("key1=value1:key2=value2") lost the \: escaping that was present in the original, so that we could no longer see which : were used as separators vs. those used as part of the original input. Fix it by collecting the key/value pairs through a QList, and sending that list on a round trip through a JSON QString (as in '["key1","value1","key2","value2"]') on its way through QemuOpts, rather than hand-rolling our own string. Since the string is only handled internally, this was faster than creating a full-blown struct of '[{"key1":"value1"},{"key2":"value2"}]', and safer at guaranteeing order compared to '{"key1":"value1","key2":"value2"}'. It would be nicer if we didn't have to round-trip through QemuOpts in the first place, but that's a much bigger task for later. Reproducer: ./x86_64-softmmu/qemu-system-x86_64 -nodefaults -nographic -qmp stdio \ -drive 'file=rbd:volumes/volume-ea141b5c-cdb3-4765-910d-e7008b209a70'\ ':id=compute:key=AQAVkvxXAAAAABAA9ZxWFYdRmV+DSwKr7BKKXg=='\ ':auth_supported=cephx\;none:mon_host=192.168.1.2\:6789'\ ',format=raw,if=none,id=drive-virtio-disk0,'\ 'serial=ea141b5c-cdb3-4765-910d-e7008b209a70,cache=writeback' Even without an RBD setup, this serves a test of whether we get the incorrect parser error of: qemu-system-x86_64: -drive file=rbd:...cache=writeback: conf option 6789 has no value or the correct behavior of hanging while trying to connect to the requested mon_host of 192.168.1.2:6789. Reported-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com> Signed-off-by: Eric Blake <eblake@redhat.com> Reviewed-by: Jeff Cody <jcody@redhat.com> Reviewed-by: Max Reitz <mreitz@redhat.com> Message-id: 20170331152730.12514-1-eblake@redhat.com Signed-off-by: Jeff Cody <jcody@redhat.com>
2017-03-31 17:27:30 +02:00
/*
* We pass these internally to qemu_rbd_set_keypairs(), so
* we can get away with the simpler list of [ "key1",
* "value1", "key2", "value2" ] rather than a raw dict
* { "key1": "value1", "key2": "value2" } where we can't
* guarantee order, or even a more correct but complex
* [ { "key1": "value1" }, { "key2": "value2" } ]
*/
if (!keypairs) {
keypairs = qlist_new();
}
qlist_append_str(keypairs, name);
qlist_append_str(keypairs, value);
}
}
rbd: Fix regression in legacy key/values containing escaped : Commit c7cacb3 accidentally broke legacy key-value parsing through pseudo-filename parsing of -drive file=rbd://..., for any key that contains an escaped ':'. Such a key is surprisingly common, thanks to mon_host specifying a 'host:port' string. The break happens because passing things from QDict through QemuOpts back to another QDict requires that we pack our parsed key/value pairs into a string, and then reparse that string, but the intermediate string that we created ("key1=value1:key2=value2") lost the \: escaping that was present in the original, so that we could no longer see which : were used as separators vs. those used as part of the original input. Fix it by collecting the key/value pairs through a QList, and sending that list on a round trip through a JSON QString (as in '["key1","value1","key2","value2"]') on its way through QemuOpts, rather than hand-rolling our own string. Since the string is only handled internally, this was faster than creating a full-blown struct of '[{"key1":"value1"},{"key2":"value2"}]', and safer at guaranteeing order compared to '{"key1":"value1","key2":"value2"}'. It would be nicer if we didn't have to round-trip through QemuOpts in the first place, but that's a much bigger task for later. Reproducer: ./x86_64-softmmu/qemu-system-x86_64 -nodefaults -nographic -qmp stdio \ -drive 'file=rbd:volumes/volume-ea141b5c-cdb3-4765-910d-e7008b209a70'\ ':id=compute:key=AQAVkvxXAAAAABAA9ZxWFYdRmV+DSwKr7BKKXg=='\ ':auth_supported=cephx\;none:mon_host=192.168.1.2\:6789'\ ',format=raw,if=none,id=drive-virtio-disk0,'\ 'serial=ea141b5c-cdb3-4765-910d-e7008b209a70,cache=writeback' Even without an RBD setup, this serves a test of whether we get the incorrect parser error of: qemu-system-x86_64: -drive file=rbd:...cache=writeback: conf option 6789 has no value or the correct behavior of hanging while trying to connect to the requested mon_host of 192.168.1.2:6789. Reported-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com> Signed-off-by: Eric Blake <eblake@redhat.com> Reviewed-by: Jeff Cody <jcody@redhat.com> Reviewed-by: Max Reitz <mreitz@redhat.com> Message-id: 20170331152730.12514-1-eblake@redhat.com Signed-off-by: Jeff Cody <jcody@redhat.com>
2017-03-31 17:27:30 +02:00
if (keypairs) {
qdict_put(options, "=keyvalue-pairs",
qstring_from_gstring(qobject_to_json(QOBJECT(keypairs))));
}
done:
g_free(buf);
qobject_unref(keypairs);
return;
}
static int qemu_rbd_set_auth(rados_t cluster, BlockdevOptionsRbd *opts,
Error **errp)
{
char *key, *acr;
rbd: New parameter auth-client-required Parameter auth-client-required lets you configure authentication methods. We tried to provide that in v2.9.0, but backed out due to interface design doubts (commit 464444fcc16). This commit is similar to what we backed out, but simpler: we use a list of enumeration values instead of a list of objects with a member of enumeration type. Let's review our reasons for backing out the first try, as stated in the commit message: * The implementation uses deprecated rados_conf_set() key "auth_supported". No biggie. Fixed: we use "auth-client-required". * The implementation makes -drive silently ignore invalid parameters "auth" and "auth-supported.*.X" where X isn't "auth". Fixable (in fact I'm going to fix similar bugs around parameter server), so again no biggie. That fix is commit 2836284db60. This commit doesn't bring the bugs back. * BlockdevOptionsRbd member @password-secret applies only to authentication method cephx. Should it be a variant member of RbdAuthMethod? We've had time to ponder, and we decided to stick to the way Ceph configuration works: the key configured separately, and silently ignored if the authentication method doesn't use it. * BlockdevOptionsRbd member @user could apply to both methods cephx and none, but I'm not sure it's actually used with none. If it isn't, should it be a variant member of RbdAuthMethod? Likewise. * The client offers a *set* of authentication methods, not a list. Should the methods be optional members of BlockdevOptionsRbd instead of members of list @auth-supported? The latter begs the question what multiple entries for the same method mean. Trivial question now that RbdAuthMethod contains nothing but @type, but less so when RbdAuthMethod acquires other members, such the ones discussed above. Again, we decided to stick to the way Ceph configuration works, except we make auth-client-required a list of enumeration values instead of a string containing keywords separated by delimiters. * How BlockdevOptionsRbd member @auth-supported interacts with settings from a configuration file specified with @conf is undocumented. I suspect it's untested, too. Not actually true, the documentation for @conf says "Values in the configuration file will be overridden by options specified via QAPI", and we've tested this. Signed-off-by: Markus Armbruster <armbru@redhat.com> Reviewed-by: Kevin Wolf <kwolf@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2018-06-14 21:14:42 +02:00
int r;
GString *accu;
RbdAuthModeList *auth;
if (opts->key_secret) {
key = qcrypto_secret_lookup_as_base64(opts->key_secret, errp);
if (!key) {
return -EIO;
}
r = rados_conf_set(cluster, "key", key);
g_free(key);
if (r < 0) {
error_setg_errno(errp, -r, "Could not set 'key'");
return r;
rbd: New parameter auth-client-required Parameter auth-client-required lets you configure authentication methods. We tried to provide that in v2.9.0, but backed out due to interface design doubts (commit 464444fcc16). This commit is similar to what we backed out, but simpler: we use a list of enumeration values instead of a list of objects with a member of enumeration type. Let's review our reasons for backing out the first try, as stated in the commit message: * The implementation uses deprecated rados_conf_set() key "auth_supported". No biggie. Fixed: we use "auth-client-required". * The implementation makes -drive silently ignore invalid parameters "auth" and "auth-supported.*.X" where X isn't "auth". Fixable (in fact I'm going to fix similar bugs around parameter server), so again no biggie. That fix is commit 2836284db60. This commit doesn't bring the bugs back. * BlockdevOptionsRbd member @password-secret applies only to authentication method cephx. Should it be a variant member of RbdAuthMethod? We've had time to ponder, and we decided to stick to the way Ceph configuration works: the key configured separately, and silently ignored if the authentication method doesn't use it. * BlockdevOptionsRbd member @user could apply to both methods cephx and none, but I'm not sure it's actually used with none. If it isn't, should it be a variant member of RbdAuthMethod? Likewise. * The client offers a *set* of authentication methods, not a list. Should the methods be optional members of BlockdevOptionsRbd instead of members of list @auth-supported? The latter begs the question what multiple entries for the same method mean. Trivial question now that RbdAuthMethod contains nothing but @type, but less so when RbdAuthMethod acquires other members, such the ones discussed above. Again, we decided to stick to the way Ceph configuration works, except we make auth-client-required a list of enumeration values instead of a string containing keywords separated by delimiters. * How BlockdevOptionsRbd member @auth-supported interacts with settings from a configuration file specified with @conf is undocumented. I suspect it's untested, too. Not actually true, the documentation for @conf says "Values in the configuration file will be overridden by options specified via QAPI", and we've tested this. Signed-off-by: Markus Armbruster <armbru@redhat.com> Reviewed-by: Kevin Wolf <kwolf@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2018-06-14 21:14:42 +02:00
}
}
rbd: New parameter auth-client-required Parameter auth-client-required lets you configure authentication methods. We tried to provide that in v2.9.0, but backed out due to interface design doubts (commit 464444fcc16). This commit is similar to what we backed out, but simpler: we use a list of enumeration values instead of a list of objects with a member of enumeration type. Let's review our reasons for backing out the first try, as stated in the commit message: * The implementation uses deprecated rados_conf_set() key "auth_supported". No biggie. Fixed: we use "auth-client-required". * The implementation makes -drive silently ignore invalid parameters "auth" and "auth-supported.*.X" where X isn't "auth". Fixable (in fact I'm going to fix similar bugs around parameter server), so again no biggie. That fix is commit 2836284db60. This commit doesn't bring the bugs back. * BlockdevOptionsRbd member @password-secret applies only to authentication method cephx. Should it be a variant member of RbdAuthMethod? We've had time to ponder, and we decided to stick to the way Ceph configuration works: the key configured separately, and silently ignored if the authentication method doesn't use it. * BlockdevOptionsRbd member @user could apply to both methods cephx and none, but I'm not sure it's actually used with none. If it isn't, should it be a variant member of RbdAuthMethod? Likewise. * The client offers a *set* of authentication methods, not a list. Should the methods be optional members of BlockdevOptionsRbd instead of members of list @auth-supported? The latter begs the question what multiple entries for the same method mean. Trivial question now that RbdAuthMethod contains nothing but @type, but less so when RbdAuthMethod acquires other members, such the ones discussed above. Again, we decided to stick to the way Ceph configuration works, except we make auth-client-required a list of enumeration values instead of a string containing keywords separated by delimiters. * How BlockdevOptionsRbd member @auth-supported interacts with settings from a configuration file specified with @conf is undocumented. I suspect it's untested, too. Not actually true, the documentation for @conf says "Values in the configuration file will be overridden by options specified via QAPI", and we've tested this. Signed-off-by: Markus Armbruster <armbru@redhat.com> Reviewed-by: Kevin Wolf <kwolf@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2018-06-14 21:14:42 +02:00
if (opts->has_auth_client_required) {
accu = g_string_new("");
for (auth = opts->auth_client_required; auth; auth = auth->next) {
if (accu->str[0]) {
g_string_append_c(accu, ';');
}
g_string_append(accu, RbdAuthMode_str(auth->value));
}
acr = g_string_free(accu, FALSE);
r = rados_conf_set(cluster, "auth_client_required", acr);
g_free(acr);
if (r < 0) {
error_setg_errno(errp, -r,
"Could not set 'auth_client_required'");
return r;
}
}
return 0;
}
rbd: Fix regression in legacy key/values containing escaped : Commit c7cacb3 accidentally broke legacy key-value parsing through pseudo-filename parsing of -drive file=rbd://..., for any key that contains an escaped ':'. Such a key is surprisingly common, thanks to mon_host specifying a 'host:port' string. The break happens because passing things from QDict through QemuOpts back to another QDict requires that we pack our parsed key/value pairs into a string, and then reparse that string, but the intermediate string that we created ("key1=value1:key2=value2") lost the \: escaping that was present in the original, so that we could no longer see which : were used as separators vs. those used as part of the original input. Fix it by collecting the key/value pairs through a QList, and sending that list on a round trip through a JSON QString (as in '["key1","value1","key2","value2"]') on its way through QemuOpts, rather than hand-rolling our own string. Since the string is only handled internally, this was faster than creating a full-blown struct of '[{"key1":"value1"},{"key2":"value2"}]', and safer at guaranteeing order compared to '{"key1":"value1","key2":"value2"}'. It would be nicer if we didn't have to round-trip through QemuOpts in the first place, but that's a much bigger task for later. Reproducer: ./x86_64-softmmu/qemu-system-x86_64 -nodefaults -nographic -qmp stdio \ -drive 'file=rbd:volumes/volume-ea141b5c-cdb3-4765-910d-e7008b209a70'\ ':id=compute:key=AQAVkvxXAAAAABAA9ZxWFYdRmV+DSwKr7BKKXg=='\ ':auth_supported=cephx\;none:mon_host=192.168.1.2\:6789'\ ',format=raw,if=none,id=drive-virtio-disk0,'\ 'serial=ea141b5c-cdb3-4765-910d-e7008b209a70,cache=writeback' Even without an RBD setup, this serves a test of whether we get the incorrect parser error of: qemu-system-x86_64: -drive file=rbd:...cache=writeback: conf option 6789 has no value or the correct behavior of hanging while trying to connect to the requested mon_host of 192.168.1.2:6789. Reported-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com> Signed-off-by: Eric Blake <eblake@redhat.com> Reviewed-by: Jeff Cody <jcody@redhat.com> Reviewed-by: Max Reitz <mreitz@redhat.com> Message-id: 20170331152730.12514-1-eblake@redhat.com Signed-off-by: Jeff Cody <jcody@redhat.com>
2017-03-31 17:27:30 +02:00
static int qemu_rbd_set_keypairs(rados_t cluster, const char *keypairs_json,
Error **errp)
{
rbd: Fix regression in legacy key/values containing escaped : Commit c7cacb3 accidentally broke legacy key-value parsing through pseudo-filename parsing of -drive file=rbd://..., for any key that contains an escaped ':'. Such a key is surprisingly common, thanks to mon_host specifying a 'host:port' string. The break happens because passing things from QDict through QemuOpts back to another QDict requires that we pack our parsed key/value pairs into a string, and then reparse that string, but the intermediate string that we created ("key1=value1:key2=value2") lost the \: escaping that was present in the original, so that we could no longer see which : were used as separators vs. those used as part of the original input. Fix it by collecting the key/value pairs through a QList, and sending that list on a round trip through a JSON QString (as in '["key1","value1","key2","value2"]') on its way through QemuOpts, rather than hand-rolling our own string. Since the string is only handled internally, this was faster than creating a full-blown struct of '[{"key1":"value1"},{"key2":"value2"}]', and safer at guaranteeing order compared to '{"key1":"value1","key2":"value2"}'. It would be nicer if we didn't have to round-trip through QemuOpts in the first place, but that's a much bigger task for later. Reproducer: ./x86_64-softmmu/qemu-system-x86_64 -nodefaults -nographic -qmp stdio \ -drive 'file=rbd:volumes/volume-ea141b5c-cdb3-4765-910d-e7008b209a70'\ ':id=compute:key=AQAVkvxXAAAAABAA9ZxWFYdRmV+DSwKr7BKKXg=='\ ':auth_supported=cephx\;none:mon_host=192.168.1.2\:6789'\ ',format=raw,if=none,id=drive-virtio-disk0,'\ 'serial=ea141b5c-cdb3-4765-910d-e7008b209a70,cache=writeback' Even without an RBD setup, this serves a test of whether we get the incorrect parser error of: qemu-system-x86_64: -drive file=rbd:...cache=writeback: conf option 6789 has no value or the correct behavior of hanging while trying to connect to the requested mon_host of 192.168.1.2:6789. Reported-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com> Signed-off-by: Eric Blake <eblake@redhat.com> Reviewed-by: Jeff Cody <jcody@redhat.com> Reviewed-by: Max Reitz <mreitz@redhat.com> Message-id: 20170331152730.12514-1-eblake@redhat.com Signed-off-by: Jeff Cody <jcody@redhat.com>
2017-03-31 17:27:30 +02:00
QList *keypairs;
QString *name;
QString *value;
const char *key;
size_t remaining;
int ret = 0;
rbd: Fix regression in legacy key/values containing escaped : Commit c7cacb3 accidentally broke legacy key-value parsing through pseudo-filename parsing of -drive file=rbd://..., for any key that contains an escaped ':'. Such a key is surprisingly common, thanks to mon_host specifying a 'host:port' string. The break happens because passing things from QDict through QemuOpts back to another QDict requires that we pack our parsed key/value pairs into a string, and then reparse that string, but the intermediate string that we created ("key1=value1:key2=value2") lost the \: escaping that was present in the original, so that we could no longer see which : were used as separators vs. those used as part of the original input. Fix it by collecting the key/value pairs through a QList, and sending that list on a round trip through a JSON QString (as in '["key1","value1","key2","value2"]') on its way through QemuOpts, rather than hand-rolling our own string. Since the string is only handled internally, this was faster than creating a full-blown struct of '[{"key1":"value1"},{"key2":"value2"}]', and safer at guaranteeing order compared to '{"key1":"value1","key2":"value2"}'. It would be nicer if we didn't have to round-trip through QemuOpts in the first place, but that's a much bigger task for later. Reproducer: ./x86_64-softmmu/qemu-system-x86_64 -nodefaults -nographic -qmp stdio \ -drive 'file=rbd:volumes/volume-ea141b5c-cdb3-4765-910d-e7008b209a70'\ ':id=compute:key=AQAVkvxXAAAAABAA9ZxWFYdRmV+DSwKr7BKKXg=='\ ':auth_supported=cephx\;none:mon_host=192.168.1.2\:6789'\ ',format=raw,if=none,id=drive-virtio-disk0,'\ 'serial=ea141b5c-cdb3-4765-910d-e7008b209a70,cache=writeback' Even without an RBD setup, this serves a test of whether we get the incorrect parser error of: qemu-system-x86_64: -drive file=rbd:...cache=writeback: conf option 6789 has no value or the correct behavior of hanging while trying to connect to the requested mon_host of 192.168.1.2:6789. Reported-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com> Signed-off-by: Eric Blake <eblake@redhat.com> Reviewed-by: Jeff Cody <jcody@redhat.com> Reviewed-by: Max Reitz <mreitz@redhat.com> Message-id: 20170331152730.12514-1-eblake@redhat.com Signed-off-by: Jeff Cody <jcody@redhat.com>
2017-03-31 17:27:30 +02:00
if (!keypairs_json) {
return ret;
}
keypairs = qobject_to(QList,
qobject_from_json(keypairs_json, &error_abort));
rbd: Fix regression in legacy key/values containing escaped : Commit c7cacb3 accidentally broke legacy key-value parsing through pseudo-filename parsing of -drive file=rbd://..., for any key that contains an escaped ':'. Such a key is surprisingly common, thanks to mon_host specifying a 'host:port' string. The break happens because passing things from QDict through QemuOpts back to another QDict requires that we pack our parsed key/value pairs into a string, and then reparse that string, but the intermediate string that we created ("key1=value1:key2=value2") lost the \: escaping that was present in the original, so that we could no longer see which : were used as separators vs. those used as part of the original input. Fix it by collecting the key/value pairs through a QList, and sending that list on a round trip through a JSON QString (as in '["key1","value1","key2","value2"]') on its way through QemuOpts, rather than hand-rolling our own string. Since the string is only handled internally, this was faster than creating a full-blown struct of '[{"key1":"value1"},{"key2":"value2"}]', and safer at guaranteeing order compared to '{"key1":"value1","key2":"value2"}'. It would be nicer if we didn't have to round-trip through QemuOpts in the first place, but that's a much bigger task for later. Reproducer: ./x86_64-softmmu/qemu-system-x86_64 -nodefaults -nographic -qmp stdio \ -drive 'file=rbd:volumes/volume-ea141b5c-cdb3-4765-910d-e7008b209a70'\ ':id=compute:key=AQAVkvxXAAAAABAA9ZxWFYdRmV+DSwKr7BKKXg=='\ ':auth_supported=cephx\;none:mon_host=192.168.1.2\:6789'\ ',format=raw,if=none,id=drive-virtio-disk0,'\ 'serial=ea141b5c-cdb3-4765-910d-e7008b209a70,cache=writeback' Even without an RBD setup, this serves a test of whether we get the incorrect parser error of: qemu-system-x86_64: -drive file=rbd:...cache=writeback: conf option 6789 has no value or the correct behavior of hanging while trying to connect to the requested mon_host of 192.168.1.2:6789. Reported-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com> Signed-off-by: Eric Blake <eblake@redhat.com> Reviewed-by: Jeff Cody <jcody@redhat.com> Reviewed-by: Max Reitz <mreitz@redhat.com> Message-id: 20170331152730.12514-1-eblake@redhat.com Signed-off-by: Jeff Cody <jcody@redhat.com>
2017-03-31 17:27:30 +02:00
remaining = qlist_size(keypairs) / 2;
assert(remaining);
while (remaining--) {
name = qobject_to(QString, qlist_pop(keypairs));
value = qobject_to(QString, qlist_pop(keypairs));
rbd: Fix regression in legacy key/values containing escaped : Commit c7cacb3 accidentally broke legacy key-value parsing through pseudo-filename parsing of -drive file=rbd://..., for any key that contains an escaped ':'. Such a key is surprisingly common, thanks to mon_host specifying a 'host:port' string. The break happens because passing things from QDict through QemuOpts back to another QDict requires that we pack our parsed key/value pairs into a string, and then reparse that string, but the intermediate string that we created ("key1=value1:key2=value2") lost the \: escaping that was present in the original, so that we could no longer see which : were used as separators vs. those used as part of the original input. Fix it by collecting the key/value pairs through a QList, and sending that list on a round trip through a JSON QString (as in '["key1","value1","key2","value2"]') on its way through QemuOpts, rather than hand-rolling our own string. Since the string is only handled internally, this was faster than creating a full-blown struct of '[{"key1":"value1"},{"key2":"value2"}]', and safer at guaranteeing order compared to '{"key1":"value1","key2":"value2"}'. It would be nicer if we didn't have to round-trip through QemuOpts in the first place, but that's a much bigger task for later. Reproducer: ./x86_64-softmmu/qemu-system-x86_64 -nodefaults -nographic -qmp stdio \ -drive 'file=rbd:volumes/volume-ea141b5c-cdb3-4765-910d-e7008b209a70'\ ':id=compute:key=AQAVkvxXAAAAABAA9ZxWFYdRmV+DSwKr7BKKXg=='\ ':auth_supported=cephx\;none:mon_host=192.168.1.2\:6789'\ ',format=raw,if=none,id=drive-virtio-disk0,'\ 'serial=ea141b5c-cdb3-4765-910d-e7008b209a70,cache=writeback' Even without an RBD setup, this serves a test of whether we get the incorrect parser error of: qemu-system-x86_64: -drive file=rbd:...cache=writeback: conf option 6789 has no value or the correct behavior of hanging while trying to connect to the requested mon_host of 192.168.1.2:6789. Reported-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com> Signed-off-by: Eric Blake <eblake@redhat.com> Reviewed-by: Jeff Cody <jcody@redhat.com> Reviewed-by: Max Reitz <mreitz@redhat.com> Message-id: 20170331152730.12514-1-eblake@redhat.com Signed-off-by: Jeff Cody <jcody@redhat.com>
2017-03-31 17:27:30 +02:00
assert(name && value);
key = qstring_get_str(name);
ret = rados_conf_set(cluster, key, qstring_get_str(value));
qobject_unref(value);
if (ret < 0) {
rbd: Fix regression in legacy key/values containing escaped : Commit c7cacb3 accidentally broke legacy key-value parsing through pseudo-filename parsing of -drive file=rbd://..., for any key that contains an escaped ':'. Such a key is surprisingly common, thanks to mon_host specifying a 'host:port' string. The break happens because passing things from QDict through QemuOpts back to another QDict requires that we pack our parsed key/value pairs into a string, and then reparse that string, but the intermediate string that we created ("key1=value1:key2=value2") lost the \: escaping that was present in the original, so that we could no longer see which : were used as separators vs. those used as part of the original input. Fix it by collecting the key/value pairs through a QList, and sending that list on a round trip through a JSON QString (as in '["key1","value1","key2","value2"]') on its way through QemuOpts, rather than hand-rolling our own string. Since the string is only handled internally, this was faster than creating a full-blown struct of '[{"key1":"value1"},{"key2":"value2"}]', and safer at guaranteeing order compared to '{"key1":"value1","key2":"value2"}'. It would be nicer if we didn't have to round-trip through QemuOpts in the first place, but that's a much bigger task for later. Reproducer: ./x86_64-softmmu/qemu-system-x86_64 -nodefaults -nographic -qmp stdio \ -drive 'file=rbd:volumes/volume-ea141b5c-cdb3-4765-910d-e7008b209a70'\ ':id=compute:key=AQAVkvxXAAAAABAA9ZxWFYdRmV+DSwKr7BKKXg=='\ ':auth_supported=cephx\;none:mon_host=192.168.1.2\:6789'\ ',format=raw,if=none,id=drive-virtio-disk0,'\ 'serial=ea141b5c-cdb3-4765-910d-e7008b209a70,cache=writeback' Even without an RBD setup, this serves a test of whether we get the incorrect parser error of: qemu-system-x86_64: -drive file=rbd:...cache=writeback: conf option 6789 has no value or the correct behavior of hanging while trying to connect to the requested mon_host of 192.168.1.2:6789. Reported-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com> Signed-off-by: Eric Blake <eblake@redhat.com> Reviewed-by: Jeff Cody <jcody@redhat.com> Reviewed-by: Max Reitz <mreitz@redhat.com> Message-id: 20170331152730.12514-1-eblake@redhat.com Signed-off-by: Jeff Cody <jcody@redhat.com>
2017-03-31 17:27:30 +02:00
error_setg_errno(errp, -ret, "invalid conf option %s", key);
qobject_unref(name);
ret = -EINVAL;
break;
}
qobject_unref(name);
}
qobject_unref(keypairs);
return ret;
}
#ifdef LIBRBD_SUPPORTS_ENCRYPTION
static int qemu_rbd_convert_luks_options(
RbdEncryptionOptionsLUKSBase *luks_opts,
char **passphrase,
size_t *passphrase_len,
Error **errp)
{
return qcrypto_secret_lookup(luks_opts->key_secret, (uint8_t **)passphrase,
passphrase_len, errp);
}
static int qemu_rbd_convert_luks_create_options(
RbdEncryptionCreateOptionsLUKSBase *luks_opts,
rbd_encryption_algorithm_t *alg,
char **passphrase,
size_t *passphrase_len,
Error **errp)
{
int r = 0;
r = qemu_rbd_convert_luks_options(
qapi_RbdEncryptionCreateOptionsLUKSBase_base(luks_opts),
passphrase, passphrase_len, errp);
if (r < 0) {
return r;
}
if (luks_opts->has_cipher_alg) {
switch (luks_opts->cipher_alg) {
case QCRYPTO_CIPHER_ALG_AES_128: {
*alg = RBD_ENCRYPTION_ALGORITHM_AES128;
break;
}
case QCRYPTO_CIPHER_ALG_AES_256: {
*alg = RBD_ENCRYPTION_ALGORITHM_AES256;
break;
}
default: {
r = -ENOTSUP;
error_setg_errno(errp, -r, "unknown encryption algorithm: %u",
luks_opts->cipher_alg);
return r;
}
}
} else {
/* default alg */
*alg = RBD_ENCRYPTION_ALGORITHM_AES256;
}
return 0;
}
static int qemu_rbd_encryption_format(rbd_image_t image,
RbdEncryptionCreateOptions *encrypt,
Error **errp)
{
int r = 0;
g_autofree char *passphrase = NULL;
size_t passphrase_len;
rbd_encryption_format_t format;
rbd_encryption_options_t opts;
rbd_encryption_luks1_format_options_t luks_opts;
rbd_encryption_luks2_format_options_t luks2_opts;
size_t opts_size;
uint64_t raw_size, effective_size;
r = rbd_get_size(image, &raw_size);
if (r < 0) {
error_setg_errno(errp, -r, "cannot get raw image size");
return r;
}
switch (encrypt->format) {
case RBD_IMAGE_ENCRYPTION_FORMAT_LUKS: {
memset(&luks_opts, 0, sizeof(luks_opts));
format = RBD_ENCRYPTION_FORMAT_LUKS1;
opts = &luks_opts;
opts_size = sizeof(luks_opts);
r = qemu_rbd_convert_luks_create_options(
qapi_RbdEncryptionCreateOptionsLUKS_base(&encrypt->u.luks),
&luks_opts.alg, &passphrase, &passphrase_len, errp);
if (r < 0) {
return r;
}
luks_opts.passphrase = passphrase;
luks_opts.passphrase_size = passphrase_len;
break;
}
case RBD_IMAGE_ENCRYPTION_FORMAT_LUKS2: {
memset(&luks2_opts, 0, sizeof(luks2_opts));
format = RBD_ENCRYPTION_FORMAT_LUKS2;
opts = &luks2_opts;
opts_size = sizeof(luks2_opts);
r = qemu_rbd_convert_luks_create_options(
qapi_RbdEncryptionCreateOptionsLUKS2_base(
&encrypt->u.luks2),
&luks2_opts.alg, &passphrase, &passphrase_len, errp);
if (r < 0) {
return r;
}
luks2_opts.passphrase = passphrase;
luks2_opts.passphrase_size = passphrase_len;
break;
}
default: {
r = -ENOTSUP;
error_setg_errno(
errp, -r, "unknown image encryption format: %u",
encrypt->format);
return r;
}
}
r = rbd_encryption_format(image, format, opts, opts_size);
if (r < 0) {
error_setg_errno(errp, -r, "encryption format fail");
return r;
}
r = rbd_get_size(image, &effective_size);
if (r < 0) {
error_setg_errno(errp, -r, "cannot get effective image size");
return r;
}
r = rbd_resize(image, raw_size + (raw_size - effective_size));
if (r < 0) {
error_setg_errno(errp, -r, "cannot resize image after format");
return r;
}
return 0;
}
static int qemu_rbd_encryption_load(rbd_image_t image,
RbdEncryptionOptions *encrypt,
Error **errp)
{
int r = 0;
g_autofree char *passphrase = NULL;
size_t passphrase_len;
rbd_encryption_luks1_format_options_t luks_opts;
rbd_encryption_luks2_format_options_t luks2_opts;
rbd_encryption_format_t format;
rbd_encryption_options_t opts;
size_t opts_size;
switch (encrypt->format) {
case RBD_IMAGE_ENCRYPTION_FORMAT_LUKS: {
memset(&luks_opts, 0, sizeof(luks_opts));
format = RBD_ENCRYPTION_FORMAT_LUKS1;
opts = &luks_opts;
opts_size = sizeof(luks_opts);
r = qemu_rbd_convert_luks_options(
qapi_RbdEncryptionOptionsLUKS_base(&encrypt->u.luks),
&passphrase, &passphrase_len, errp);
if (r < 0) {
return r;
}
luks_opts.passphrase = passphrase;
luks_opts.passphrase_size = passphrase_len;
break;
}
case RBD_IMAGE_ENCRYPTION_FORMAT_LUKS2: {
memset(&luks2_opts, 0, sizeof(luks2_opts));
format = RBD_ENCRYPTION_FORMAT_LUKS2;
opts = &luks2_opts;
opts_size = sizeof(luks2_opts);
r = qemu_rbd_convert_luks_options(
qapi_RbdEncryptionOptionsLUKS2_base(&encrypt->u.luks2),
&passphrase, &passphrase_len, errp);
if (r < 0) {
return r;
}
luks2_opts.passphrase = passphrase;
luks2_opts.passphrase_size = passphrase_len;
break;
}
default: {
r = -ENOTSUP;
error_setg_errno(
errp, -r, "unknown image encryption format: %u",
encrypt->format);
return r;
}
}
r = rbd_encryption_load(image, format, opts, opts_size);
if (r < 0) {
error_setg_errno(errp, -r, "encryption load fail");
return r;
}
return 0;
}
#endif
/* FIXME Deprecate and remove keypairs or make it available in QMP. */
static int qemu_rbd_do_create(BlockdevCreateOptions *options,
const char *keypairs, const char *password_secret,
Error **errp)
{
BlockdevCreateOptionsRbd *opts = &options->u.rbd;
rados_t cluster;
rados_ioctx_t io_ctx;
int obj_order = 0;
int ret;
assert(options->driver == BLOCKDEV_DRIVER_RBD);
if (opts->location->snapshot) {
error_setg(errp, "Can't use snapshot name for image creation");
return -EINVAL;
}
#ifndef LIBRBD_SUPPORTS_ENCRYPTION
if (opts->encrypt) {
error_setg(errp, "RBD library does not support image encryption");
return -ENOTSUP;
}
#endif
if (opts->has_cluster_size) {
int64_t objsize = opts->cluster_size;
if ((objsize - 1) & objsize) { /* not a power of 2? */
error_setg(errp, "obj size needs to be power of 2");
return -EINVAL;
}
if (objsize < 4096) {
error_setg(errp, "obj size too small");
return -EINVAL;
}
obj_order = ctz32(objsize);
}
ret = qemu_rbd_connect(&cluster, &io_ctx, opts->location, false, keypairs,
password_secret, errp);
if (ret < 0) {
return ret;
}
ret = rbd_create(io_ctx, opts->location->image, opts->size, &obj_order);
if (ret < 0) {
error_setg_errno(errp, -ret, "error rbd create");
goto out;
}
#ifdef LIBRBD_SUPPORTS_ENCRYPTION
if (opts->encrypt) {
rbd_image_t image;
ret = rbd_open(io_ctx, opts->location->image, &image, NULL);
if (ret < 0) {
error_setg_errno(errp, -ret,
"error opening image '%s' for encryption format",
opts->location->image);
goto out;
}
ret = qemu_rbd_encryption_format(image, opts->encrypt, errp);
rbd_close(image);
if (ret < 0) {
/* encryption format fail, try removing the image */
rbd_remove(io_ctx, opts->location->image);
goto out;
}
}
#endif
ret = 0;
out:
rados_ioctx_destroy(io_ctx);
rados_shutdown(cluster);
return ret;
}
static int qemu_rbd_co_create(BlockdevCreateOptions *options, Error **errp)
{
return qemu_rbd_do_create(options, NULL, NULL, errp);
}
static int qemu_rbd_extract_encryption_create_options(
QemuOpts *opts,
RbdEncryptionCreateOptions **spec,
Error **errp)
{
QDict *opts_qdict;
QDict *encrypt_qdict;
Visitor *v;
int ret = 0;
opts_qdict = qemu_opts_to_qdict(opts, NULL);
qdict_extract_subqdict(opts_qdict, &encrypt_qdict, "encrypt.");
qobject_unref(opts_qdict);
if (!qdict_size(encrypt_qdict)) {
*spec = NULL;
goto exit;
}
/* Convert options into a QAPI object */
v = qobject_input_visitor_new_flat_confused(encrypt_qdict, errp);
if (!v) {
ret = -EINVAL;
goto exit;
}
visit_type_RbdEncryptionCreateOptions(v, NULL, spec, errp);
visit_free(v);
if (!*spec) {
ret = -EINVAL;
goto exit;
}
exit:
qobject_unref(encrypt_qdict);
return ret;
}
static int coroutine_fn qemu_rbd_co_create_opts(BlockDriver *drv,
const char *filename,
QemuOpts *opts,
Error **errp)
{
BlockdevCreateOptions *create_options;
BlockdevCreateOptionsRbd *rbd_opts;
BlockdevOptionsRbd *loc;
RbdEncryptionCreateOptions *encrypt = NULL;
Error *local_err = NULL;
const char *keypairs, *password_secret;
QDict *options = NULL;
int ret = 0;
create_options = g_new0(BlockdevCreateOptions, 1);
create_options->driver = BLOCKDEV_DRIVER_RBD;
rbd_opts = &create_options->u.rbd;
rbd_opts->location = g_new0(BlockdevOptionsRbd, 1);
password_secret = qemu_opt_get(opts, "password-secret");
/* Read out options */
rbd_opts->size = ROUND_UP(qemu_opt_get_size_del(opts, BLOCK_OPT_SIZE, 0),
BDRV_SECTOR_SIZE);
rbd_opts->cluster_size = qemu_opt_get_size_del(opts,
BLOCK_OPT_CLUSTER_SIZE, 0);
rbd_opts->has_cluster_size = (rbd_opts->cluster_size != 0);
options = qdict_new();
qemu_rbd_parse_filename(filename, options, &local_err);
if (local_err) {
ret = -EINVAL;
error_propagate(errp, local_err);
goto exit;
}
ret = qemu_rbd_extract_encryption_create_options(opts, &encrypt, errp);
if (ret < 0) {
goto exit;
}
rbd_opts->encrypt = encrypt;
/*
* 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.
*/
loc = rbd_opts->location;
loc->pool = g_strdup(qdict_get_try_str(options, "pool"));
loc->conf = g_strdup(qdict_get_try_str(options, "conf"));
loc->user = g_strdup(qdict_get_try_str(options, "user"));
loc->q_namespace = g_strdup(qdict_get_try_str(options, "namespace"));
loc->image = g_strdup(qdict_get_try_str(options, "image"));
keypairs = qdict_get_try_str(options, "=keyvalue-pairs");
ret = qemu_rbd_do_create(create_options, keypairs, password_secret, errp);
if (ret < 0) {
goto exit;
}
exit:
qobject_unref(options);
qapi_free_BlockdevCreateOptions(create_options);
return ret;
}
static char *qemu_rbd_mon_host(BlockdevOptionsRbd *opts, Error **errp)
{
const char **vals;
const char *host, *port;
char *rados_str;
InetSocketAddressBaseList *p;
int i, cnt;
if (!opts->has_server) {
return NULL;
}
for (cnt = 0, p = opts->server; p; p = p->next) {
cnt++;
}
vals = g_new(const char *, cnt + 1);
for (i = 0, p = opts->server; p; p = p->next, i++) {
host = p->value->host;
port = p->value->port;
if (strchr(host, ':')) {
vals[i] = g_strdup_printf("[%s]:%s", host, port);
} else {
vals[i] = g_strdup_printf("%s:%s", host, port);
}
}
vals[i] = NULL;
rados_str = i ? g_strjoinv(";", (char **)vals) : NULL;
g_strfreev((char **)vals);
return rados_str;
}
static int qemu_rbd_connect(rados_t *cluster, rados_ioctx_t *io_ctx,
BlockdevOptionsRbd *opts, bool cache,
const char *keypairs, const char *secretid,
Error **errp)
{
char *mon_host = NULL;
Error *local_err = NULL;
int r;
if (secretid) {
if (opts->key_secret) {
error_setg(errp,
"Legacy 'password-secret' clashes with 'key-secret'");
return -EINVAL;
}
opts->key_secret = g_strdup(secretid);
}
mon_host = qemu_rbd_mon_host(opts, &local_err);
if (local_err) {
error_propagate(errp, local_err);
r = -EINVAL;
goto out;
}
r = rados_create(cluster, opts->user);
if (r < 0) {
error_setg_errno(errp, -r, "error initializing");
goto out;
}
/* try default location when conf=NULL, but ignore failure */
r = rados_conf_read_file(*cluster, opts->conf);
if (opts->conf && r < 0) {
error_setg_errno(errp, -r, "error reading conf file %s", opts->conf);
goto failed_shutdown;
}
r = qemu_rbd_set_keypairs(*cluster, keypairs, errp);
if (r < 0) {
goto failed_shutdown;
}
if (mon_host) {
r = rados_conf_set(*cluster, "mon_host", mon_host);
if (r < 0) {
goto failed_shutdown;
}
}
r = qemu_rbd_set_auth(*cluster, opts, errp);
if (r < 0) {
goto failed_shutdown;
}
/*
* Fallback to more conservative semantics if setting cache
* options fails. Ignore errors from setting rbd_cache because the
* only possible error is that the option does not exist, and
* librbd defaults to no caching. If write through caching cannot
* be set up, fall back to no caching.
*/
if (cache) {
rados_conf_set(*cluster, "rbd_cache", "true");
} else {
rados_conf_set(*cluster, "rbd_cache", "false");
}
r = rados_connect(*cluster);
if (r < 0) {
error_setg_errno(errp, -r, "error connecting");
goto failed_shutdown;
}
r = rados_ioctx_create(*cluster, opts->pool, io_ctx);
if (r < 0) {
error_setg_errno(errp, -r, "error opening pool %s", opts->pool);
goto failed_shutdown;
}
#ifdef HAVE_RBD_NAMESPACE_EXISTS
if (opts->q_namespace && strlen(opts->q_namespace) > 0) {
bool exists;
r = rbd_namespace_exists(*io_ctx, opts->q_namespace, &exists);
if (r < 0) {
error_setg_errno(errp, -r, "error checking namespace");
goto failed_ioctx_destroy;
}
if (!exists) {
error_setg(errp, "namespace '%s' does not exist",
opts->q_namespace);
r = -ENOENT;
goto failed_ioctx_destroy;
}
}
#endif
/*
* Set the namespace after opening the io context on the pool,
* if nspace == NULL or if nspace == "", it is just as we did nothing
*/
rados_ioctx_set_namespace(*io_ctx, opts->q_namespace);
r = 0;
goto out;
#ifdef HAVE_RBD_NAMESPACE_EXISTS
failed_ioctx_destroy:
rados_ioctx_destroy(*io_ctx);
#endif
failed_shutdown:
rados_shutdown(*cluster);
out:
g_free(mon_host);
return r;
}
static int qemu_rbd_convert_options(QDict *options, BlockdevOptionsRbd **opts,
Error **errp)
{
Visitor *v;
/* Convert the remaining options into a QAPI object */
v = qobject_input_visitor_new_flat_confused(options, errp);
if (!v) {
return -EINVAL;
}
visit_type_BlockdevOptionsRbd(v, NULL, opts, errp);
visit_free(v);
if (!opts) {
return -EINVAL;
}
return 0;
}
static int qemu_rbd_attempt_legacy_options(QDict *options,
BlockdevOptionsRbd **opts,
char **keypairs)
{
char *filename;
int r;
filename = g_strdup(qdict_get_try_str(options, "filename"));
if (!filename) {
return -EINVAL;
}
qdict_del(options, "filename");
qemu_rbd_parse_filename(filename, options, NULL);
/* keypairs freed by caller */
*keypairs = g_strdup(qdict_get_try_str(options, "=keyvalue-pairs"));
if (*keypairs) {
qdict_del(options, "=keyvalue-pairs");
}
r = qemu_rbd_convert_options(options, opts, NULL);
g_free(filename);
return r;
}
static int qemu_rbd_open(BlockDriverState *bs, QDict *options, int flags,
Error **errp)
{
BDRVRBDState *s = bs->opaque;
BlockdevOptionsRbd *opts = NULL;
const QDictEntry *e;
Error *local_err = NULL;
char *keypairs, *secretid;
rbd_image_info_t info;
int r;
keypairs = g_strdup(qdict_get_try_str(options, "=keyvalue-pairs"));
if (keypairs) {
qdict_del(options, "=keyvalue-pairs");
}
secretid = g_strdup(qdict_get_try_str(options, "password-secret"));
if (secretid) {
qdict_del(options, "password-secret");
}
r = qemu_rbd_convert_options(options, &opts, &local_err);
if (local_err) {
/* If keypairs are present, that means some options are present in
* the modern option format. Don't attempt to parse legacy option
* formats, as we won't support mixed usage. */
if (keypairs) {
error_propagate(errp, local_err);
goto out;
}
/* If the initial attempt to convert and process the options failed,
* we may be attempting to open an image file that has the rbd options
* specified in the older format consisting of all key/value pairs
* encoded in the filename. Go ahead and attempt to parse the
* filename, and see if we can pull out the required options. */
r = qemu_rbd_attempt_legacy_options(options, &opts, &keypairs);
if (r < 0) {
/* Propagate the original error, not the legacy parsing fallback
* error, as the latter was just a best-effort attempt. */
error_propagate(errp, local_err);
goto out;
}
/* Take care whenever deciding to actually deprecate; once this ability
* is removed, we will not be able to open any images with legacy-styled
* backing image strings. */
warn_report("RBD options encoded in the filename as keyvalue pairs "
"is deprecated");
}
/* Remove the processed options from the QDict (the visitor processes
* _all_ options in the QDict) */
while ((e = qdict_first(options))) {
qdict_del(options, e->key);
}
r = qemu_rbd_connect(&s->cluster, &s->io_ctx, opts,
!(flags & BDRV_O_NOCACHE), keypairs, secretid, errp);
if (r < 0) {
goto out;
}
s->snap = g_strdup(opts->snapshot);
s->image_name = g_strdup(opts->image);
/* rbd_open is always r/w */
r = rbd_open(s->io_ctx, s->image_name, &s->image, s->snap);
if (r < 0) {
error_setg_errno(errp, -r, "error reading header from %s",
s->image_name);
goto failed_open;
}
if (opts->encrypt) {
#ifdef LIBRBD_SUPPORTS_ENCRYPTION
r = qemu_rbd_encryption_load(s->image, opts->encrypt, errp);
if (r < 0) {
goto failed_post_open;
}
#else
r = -ENOTSUP;
error_setg(errp, "RBD library does not support image encryption");
goto failed_post_open;
#endif
}
r = rbd_stat(s->image, &info, sizeof(info));
if (r < 0) {
error_setg_errno(errp, -r, "error getting image info from %s",
s->image_name);
goto failed_post_open;
}
s->image_size = info.size;
s->object_size = info.obj_size;
/* If we are using an rbd snapshot, we must be r/o, otherwise
* leave as-is */
if (s->snap != NULL) {
r = bdrv_apply_auto_read_only(bs, "rbd snapshots are read-only", errp);
if (r < 0) {
goto failed_post_open;
}
}
block/rbd: add write zeroes support This patch wittingly sets BDRV_REQ_NO_FALLBACK and silently ignores BDRV_REQ_MAY_UNMAP for older librbd versions. The rationale for this is as follows (citing Ilya Dryomov current RBD maintainer): ---8<--- a) remove the BDRV_REQ_MAY_UNMAP check in qemu_rbd_co_pwrite_zeroes() and as a consequence always unmap if librbd is too old It's not clear what qemu's expectation is but in general Write Zeroes is allowed to unmap. The only guarantee is that subsequent reads return zeroes, everything else is a hint. This is how it is specified in the kernel and in the NVMe spec. In particular, block/nvme.c implements it as follows: if (flags & BDRV_REQ_MAY_UNMAP) { cdw12 |= (1 << 25); } This sets the Deallocate bit. But if it's not set, the device may still deallocate: """ If the Deallocate bit (CDW12.DEAC) is set to '1' in a Write Zeroes command, and the namespace supports clearing all bytes to 0h in the values read (e.g., bits 2:0 in the DLFEAT field are set to 001b) from a deallocated logical block and its metadata (excluding protection information), then for each specified logical block, the controller: - should deallocate that logical block; ... If the Deallocate bit is cleared to '0' in a Write Zeroes command, and the namespace supports clearing all bytes to 0h in the values read (e.g., bits 2:0 in the DLFEAT field are set to 001b) from a deallocated logical block and its metadata (excluding protection information), then, for each specified logical block, the controller: - may deallocate that logical block; """ https://nvmexpress.org/wp-content/uploads/NVM-Express-NVM-Command-Set-Specification-2021.06.02-Ratified-1.pdf b) set BDRV_REQ_NO_FALLBACK in supported_zero_flags Again, it's not clear what qemu expects here, but without it we end up in a ridiculous situation where specifying the "don't allow slow fallback" switch immediately fails all efficient zeroing requests on a device where Write Zeroes is always efficient: $ qemu-io -c 'help write' | grep -- '-[zun]' -n, -- with -z, don't allow slow fallback -u, -- with -z, allow unmapping -z, -- write zeroes using blk_co_pwrite_zeroes $ qemu-io -f rbd -c 'write -z -u -n 0 1M' rbd:foo/bar write failed: Operation not supported --->8--- Signed-off-by: Peter Lieven <pl@kamp.de> Reviewed-by: Ilya Dryomov <idryomov@gmail.com> Message-Id: <20210702172356.11574-6-idryomov@gmail.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2021-07-02 19:23:55 +02:00
#ifdef LIBRBD_SUPPORTS_WRITE_ZEROES
bs->supported_zero_flags = BDRV_REQ_MAY_UNMAP | BDRV_REQ_NO_FALLBACK;
#endif
/* When extending regular files, we get zeros from the OS */
bs->supported_truncate_flags = BDRV_REQ_ZERO_WRITE;
r = 0;
goto out;
failed_post_open:
rbd_close(s->image);
failed_open:
rados_ioctx_destroy(s->io_ctx);
g_free(s->snap);
g_free(s->image_name);
rados_shutdown(s->cluster);
out:
qapi_free_BlockdevOptionsRbd(opts);
g_free(keypairs);
g_free(secretid);
return r;
}
/* Since RBD is currently always opened R/W via the API,
* we just need to check if we are using a snapshot or not, in
* order to determine if we will allow it to be R/W */
static int qemu_rbd_reopen_prepare(BDRVReopenState *state,
BlockReopenQueue *queue, Error **errp)
{
BDRVRBDState *s = state->bs->opaque;
int ret = 0;
if (s->snap && state->flags & BDRV_O_RDWR) {
error_setg(errp,
"Cannot change node '%s' to r/w when using RBD snapshot",
bdrv_get_device_or_node_name(state->bs));
ret = -EINVAL;
}
return ret;
}
static void qemu_rbd_close(BlockDriverState *bs)
{
BDRVRBDState *s = bs->opaque;
rbd_close(s->image);
rados_ioctx_destroy(s->io_ctx);
g_free(s->snap);
g_free(s->image_name);
rados_shutdown(s->cluster);
}
/* Resize the RBD image and update the 'image_size' with the current size */
static int qemu_rbd_resize(BlockDriverState *bs, uint64_t size)
{
BDRVRBDState *s = bs->opaque;
int r;
r = rbd_resize(s->image, size);
if (r < 0) {
return r;
}
s->image_size = size;
return 0;
}
static void qemu_rbd_finish_bh(void *opaque)
{
RBDTask *task = opaque;
task->complete = true;
aio_co_wake(task->co);
}
/*
* This is the completion callback function for all rbd aio calls
* started from qemu_rbd_start_co().
*
* Note: this function is being called from a non qemu thread so
* we need to be careful about what we do here. Generally we only
* schedule a BH, and do the rest of the io completion handling
* from qemu_rbd_finish_bh() which runs in a qemu context.
*/
static void qemu_rbd_completion_cb(rbd_completion_t c, RBDTask *task)
{
task->ret = rbd_aio_get_return_value(c);
rbd_aio_release(c);
aio_bh_schedule_oneshot(bdrv_get_aio_context(task->bs),
qemu_rbd_finish_bh, task);
}
static int coroutine_fn qemu_rbd_start_co(BlockDriverState *bs,
uint64_t offset,
uint64_t bytes,
QEMUIOVector *qiov,
int flags,
RBDAIOCmd cmd)
{
BDRVRBDState *s = bs->opaque;
RBDTask task = { .bs = bs, .co = qemu_coroutine_self() };
rbd_completion_t c;
int r;
assert(!qiov || qiov->size == bytes);
if (cmd == RBD_AIO_WRITE || cmd == RBD_AIO_WRITE_ZEROES) {
/*
* RBD APIs don't allow us to write more than actual size, so in order
* to support growing images, we resize the image before write
* operations that exceed the current size.
*/
if (offset + bytes > s->image_size) {
int r = qemu_rbd_resize(bs, offset + bytes);
if (r < 0) {
return r;
}
}
}
r = rbd_aio_create_completion(&task,
(rbd_callback_t) qemu_rbd_completion_cb, &c);
if (r < 0) {
return r;
}
switch (cmd) {
case RBD_AIO_READ:
r = rbd_aio_readv(s->image, qiov->iov, qiov->niov, offset, c);
break;
case RBD_AIO_WRITE:
r = rbd_aio_writev(s->image, qiov->iov, qiov->niov, offset, c);
break;
case RBD_AIO_DISCARD:
r = rbd_aio_discard(s->image, offset, bytes, c);
break;
case RBD_AIO_FLUSH:
r = rbd_aio_flush(s->image, c);
break;
block/rbd: add write zeroes support This patch wittingly sets BDRV_REQ_NO_FALLBACK and silently ignores BDRV_REQ_MAY_UNMAP for older librbd versions. The rationale for this is as follows (citing Ilya Dryomov current RBD maintainer): ---8<--- a) remove the BDRV_REQ_MAY_UNMAP check in qemu_rbd_co_pwrite_zeroes() and as a consequence always unmap if librbd is too old It's not clear what qemu's expectation is but in general Write Zeroes is allowed to unmap. The only guarantee is that subsequent reads return zeroes, everything else is a hint. This is how it is specified in the kernel and in the NVMe spec. In particular, block/nvme.c implements it as follows: if (flags & BDRV_REQ_MAY_UNMAP) { cdw12 |= (1 << 25); } This sets the Deallocate bit. But if it's not set, the device may still deallocate: """ If the Deallocate bit (CDW12.DEAC) is set to '1' in a Write Zeroes command, and the namespace supports clearing all bytes to 0h in the values read (e.g., bits 2:0 in the DLFEAT field are set to 001b) from a deallocated logical block and its metadata (excluding protection information), then for each specified logical block, the controller: - should deallocate that logical block; ... If the Deallocate bit is cleared to '0' in a Write Zeroes command, and the namespace supports clearing all bytes to 0h in the values read (e.g., bits 2:0 in the DLFEAT field are set to 001b) from a deallocated logical block and its metadata (excluding protection information), then, for each specified logical block, the controller: - may deallocate that logical block; """ https://nvmexpress.org/wp-content/uploads/NVM-Express-NVM-Command-Set-Specification-2021.06.02-Ratified-1.pdf b) set BDRV_REQ_NO_FALLBACK in supported_zero_flags Again, it's not clear what qemu expects here, but without it we end up in a ridiculous situation where specifying the "don't allow slow fallback" switch immediately fails all efficient zeroing requests on a device where Write Zeroes is always efficient: $ qemu-io -c 'help write' | grep -- '-[zun]' -n, -- with -z, don't allow slow fallback -u, -- with -z, allow unmapping -z, -- write zeroes using blk_co_pwrite_zeroes $ qemu-io -f rbd -c 'write -z -u -n 0 1M' rbd:foo/bar write failed: Operation not supported --->8--- Signed-off-by: Peter Lieven <pl@kamp.de> Reviewed-by: Ilya Dryomov <idryomov@gmail.com> Message-Id: <20210702172356.11574-6-idryomov@gmail.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2021-07-02 19:23:55 +02:00
#ifdef LIBRBD_SUPPORTS_WRITE_ZEROES
case RBD_AIO_WRITE_ZEROES: {
int zero_flags = 0;
#ifdef RBD_WRITE_ZEROES_FLAG_THICK_PROVISION
if (!(flags & BDRV_REQ_MAY_UNMAP)) {
zero_flags = RBD_WRITE_ZEROES_FLAG_THICK_PROVISION;
}
#endif
r = rbd_aio_write_zeroes(s->image, offset, bytes, c, zero_flags, 0);
break;
}
#endif
default:
r = -EINVAL;
}
if (r < 0) {
error_report("rbd request failed early: cmd %d offset %" PRIu64
" bytes %" PRIu64 " flags %d r %d (%s)", cmd, offset,
bytes, flags, r, strerror(-r));
rbd_aio_release(c);
return r;
}
while (!task.complete) {
qemu_coroutine_yield();
}
if (task.ret < 0) {
error_report("rbd request failed: cmd %d offset %" PRIu64 " bytes %"
PRIu64 " flags %d task.ret %" PRIi64 " (%s)", cmd, offset,
bytes, flags, task.ret, strerror(-task.ret));
return task.ret;
}
/* zero pad short reads */
if (cmd == RBD_AIO_READ && task.ret < qiov->size) {
qemu_iovec_memset(qiov, task.ret, 0, qiov->size - task.ret);
}
return 0;
}
static int
block: use int64_t instead of uint64_t in driver read handlers We are generally moving to int64_t for both offset and bytes parameters on all io paths. Main motivation is realization of 64-bit write_zeroes operation for fast zeroing large disk chunks, up to the whole disk. We chose signed type, to be consistent with off_t (which is signed) and with possibility for signed return type (where negative value means error). So, convert driver read handlers parameters which are already 64bit to signed type. While being here, convert also flags parameter to be BdrvRequestFlags. Now let's consider all callers. Simple git grep '\->bdrv_\(aio\|co\)_preadv\(_part\)\?' shows that's there three callers of driver function: bdrv_driver_preadv() in block/io.c, passes int64_t, checked by bdrv_check_qiov_request() to be non-negative. qcow2_load_vmstate() does bdrv_check_qiov_request(). do_perform_cow_read() has uint64_t argument. And a lot of things in qcow2 driver are uint64_t, so converting it is big job. But we must not work with requests that don't satisfy bdrv_check_qiov_request(), so let's just assert it here. Still, the functions may be called directly, not only by drv->... Let's check: git grep '\.bdrv_\(aio\|co\)_preadv\(_part\)\?\s*=' | \ awk '{print $4}' | sed 's/,//' | sed 's/&//' | sort | uniq | \ while read func; do git grep "$func(" | \ grep -v "$func(BlockDriverState"; done The only one such caller: QEMUIOVector qiov = QEMU_IOVEC_INIT_BUF(qiov, &data, 1); ... ret = bdrv_replace_test_co_preadv(bs, 0, 1, &qiov, 0); in tests/unit/test-bdrv-drain.c, and it's OK obviously. Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Message-Id: <20210903102807.27127-4-vsementsov@virtuozzo.com> Reviewed-by: Eric Blake <eblake@redhat.com> [eblake: fix typos] Signed-off-by: Eric Blake <eblake@redhat.com>
2021-09-03 12:27:59 +02:00
coroutine_fn qemu_rbd_co_preadv(BlockDriverState *bs, int64_t offset,
int64_t bytes, QEMUIOVector *qiov,
BdrvRequestFlags flags)
{
return qemu_rbd_start_co(bs, offset, bytes, qiov, flags, RBD_AIO_READ);
}
static int
block: use int64_t instead of uint64_t in driver write handlers We are generally moving to int64_t for both offset and bytes parameters on all io paths. Main motivation is realization of 64-bit write_zeroes operation for fast zeroing large disk chunks, up to the whole disk. We chose signed type, to be consistent with off_t (which is signed) and with possibility for signed return type (where negative value means error). So, convert driver write handlers parameters which are already 64bit to signed type. While being here, convert also flags parameter to be BdrvRequestFlags. Now let's consider all callers. Simple git grep '\->bdrv_\(aio\|co\)_pwritev\(_part\)\?' shows that's there three callers of driver function: bdrv_driver_pwritev() and bdrv_driver_pwritev_compressed() in block/io.c, both pass int64_t, checked by bdrv_check_qiov_request() to be non-negative. qcow2_save_vmstate() does bdrv_check_qiov_request(). Still, the functions may be called directly, not only by drv->... Let's check: git grep '\.bdrv_\(aio\|co\)_pwritev\(_part\)\?\s*=' | \ awk '{print $4}' | sed 's/,//' | sed 's/&//' | sort | uniq | \ while read func; do git grep "$func(" | \ grep -v "$func(BlockDriverState"; done shows several callers: qcow2: qcow2_co_truncate() write at most up to @offset, which is checked in generic qcow2_co_truncate() by bdrv_check_request(). qcow2_co_pwritev_compressed_task() pass the request (or part of the request) that already went through normal write path, so it should be OK qcow: qcow_co_pwritev_compressed() pass int64_t, it's updated by this patch quorum: quorum_co_pwrite_zeroes() pass int64_t and int - OK throttle: throttle_co_pwritev_compressed() pass int64_t, it's updated by this patch vmdk: vmdk_co_pwritev_compressed() pass int64_t, it's updated by this patch Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Message-Id: <20210903102807.27127-5-vsementsov@virtuozzo.com> Reviewed-by: Eric Blake <eblake@redhat.com> Signed-off-by: Eric Blake <eblake@redhat.com>
2021-09-03 12:28:00 +02:00
coroutine_fn qemu_rbd_co_pwritev(BlockDriverState *bs, int64_t offset,
int64_t bytes, QEMUIOVector *qiov,
BdrvRequestFlags flags)
{
return qemu_rbd_start_co(bs, offset, bytes, qiov, flags, RBD_AIO_WRITE);
}
static int coroutine_fn qemu_rbd_co_flush(BlockDriverState *bs)
{
return qemu_rbd_start_co(bs, 0, 0, NULL, 0, RBD_AIO_FLUSH);
}
static int coroutine_fn qemu_rbd_co_pdiscard(BlockDriverState *bs,
block: use int64_t instead of int in driver discard handlers We are generally moving to int64_t for both offset and bytes parameters on all io paths. Main motivation is realization of 64-bit write_zeroes operation for fast zeroing large disk chunks, up to the whole disk. We chose signed type, to be consistent with off_t (which is signed) and with possibility for signed return type (where negative value means error). So, convert driver discard handlers bytes parameter to int64_t. The only caller of all updated function is bdrv_co_pdiscard in block/io.c. It is already prepared to work with 64bit requests, but pass at most max(bs->bl.max_pdiscard, INT_MAX) to the driver. Let's look at all updated functions: blkdebug: all calculations are still OK, thanks to bdrv_check_qiov_request(). both rule_check and bdrv_co_pdiscard are 64bit blklogwrites: pass to blk_loc_writes_co_log which is 64bit blkreplay, copy-on-read, filter-compress: pass to bdrv_co_pdiscard, OK copy-before-write: pass to bdrv_co_pdiscard which is 64bit and to cbw_do_copy_before_write which is 64bit file-posix: one handler calls raw_account_discard() is 64bit and both handlers calls raw_do_pdiscard(). Update raw_do_pdiscard, which pass to RawPosixAIOData::aio_nbytes, which is 64bit (and calls raw_account_discard()) gluster: somehow, third argument of glfs_discard_async is size_t. Let's set max_pdiscard accordingly. iscsi: iscsi_allocmap_set_invalid is 64bit, !is_byte_request_lun_aligned is 64bit. list.num is uint32_t. Let's clarify max_pdiscard and pdiscard_alignment. mirror_top: pass to bdrv_mirror_top_do_write() which is 64bit nbd: protocol limitation. max_pdiscard is alredy set strict enough, keep it as is for now. nvme: buf.nlb is uint32_t and we do shift. So, add corresponding limits to nvme_refresh_limits(). preallocate: pass to bdrv_co_pdiscard() which is 64bit. rbd: pass to qemu_rbd_start_co() which is 64bit. qcow2: calculations are still OK, thanks to bdrv_check_qiov_request(), qcow2_cluster_discard() is 64bit. raw-format: raw_adjust_offset() is 64bit, bdrv_co_pdiscard too. throttle: pass to bdrv_co_pdiscard() which is 64bit and to throttle_group_co_io_limits_intercept() which is 64bit as well. test-block-iothread: bytes argument is unused Great! Now all drivers are prepared to handle 64bit discard requests, or else have explicit max_pdiscard limits. Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Message-Id: <20210903102807.27127-11-vsementsov@virtuozzo.com> Reviewed-by: Eric Blake <eblake@redhat.com> Signed-off-by: Eric Blake <eblake@redhat.com>
2021-09-03 12:28:06 +02:00
int64_t offset, int64_t bytes)
{
block: use int64_t instead of int in driver discard handlers We are generally moving to int64_t for both offset and bytes parameters on all io paths. Main motivation is realization of 64-bit write_zeroes operation for fast zeroing large disk chunks, up to the whole disk. We chose signed type, to be consistent with off_t (which is signed) and with possibility for signed return type (where negative value means error). So, convert driver discard handlers bytes parameter to int64_t. The only caller of all updated function is bdrv_co_pdiscard in block/io.c. It is already prepared to work with 64bit requests, but pass at most max(bs->bl.max_pdiscard, INT_MAX) to the driver. Let's look at all updated functions: blkdebug: all calculations are still OK, thanks to bdrv_check_qiov_request(). both rule_check and bdrv_co_pdiscard are 64bit blklogwrites: pass to blk_loc_writes_co_log which is 64bit blkreplay, copy-on-read, filter-compress: pass to bdrv_co_pdiscard, OK copy-before-write: pass to bdrv_co_pdiscard which is 64bit and to cbw_do_copy_before_write which is 64bit file-posix: one handler calls raw_account_discard() is 64bit and both handlers calls raw_do_pdiscard(). Update raw_do_pdiscard, which pass to RawPosixAIOData::aio_nbytes, which is 64bit (and calls raw_account_discard()) gluster: somehow, third argument of glfs_discard_async is size_t. Let's set max_pdiscard accordingly. iscsi: iscsi_allocmap_set_invalid is 64bit, !is_byte_request_lun_aligned is 64bit. list.num is uint32_t. Let's clarify max_pdiscard and pdiscard_alignment. mirror_top: pass to bdrv_mirror_top_do_write() which is 64bit nbd: protocol limitation. max_pdiscard is alredy set strict enough, keep it as is for now. nvme: buf.nlb is uint32_t and we do shift. So, add corresponding limits to nvme_refresh_limits(). preallocate: pass to bdrv_co_pdiscard() which is 64bit. rbd: pass to qemu_rbd_start_co() which is 64bit. qcow2: calculations are still OK, thanks to bdrv_check_qiov_request(), qcow2_cluster_discard() is 64bit. raw-format: raw_adjust_offset() is 64bit, bdrv_co_pdiscard too. throttle: pass to bdrv_co_pdiscard() which is 64bit and to throttle_group_co_io_limits_intercept() which is 64bit as well. test-block-iothread: bytes argument is unused Great! Now all drivers are prepared to handle 64bit discard requests, or else have explicit max_pdiscard limits. Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Message-Id: <20210903102807.27127-11-vsementsov@virtuozzo.com> Reviewed-by: Eric Blake <eblake@redhat.com> Signed-off-by: Eric Blake <eblake@redhat.com>
2021-09-03 12:28:06 +02:00
return qemu_rbd_start_co(bs, offset, bytes, NULL, 0, RBD_AIO_DISCARD);
}
block/rbd: add write zeroes support This patch wittingly sets BDRV_REQ_NO_FALLBACK and silently ignores BDRV_REQ_MAY_UNMAP for older librbd versions. The rationale for this is as follows (citing Ilya Dryomov current RBD maintainer): ---8<--- a) remove the BDRV_REQ_MAY_UNMAP check in qemu_rbd_co_pwrite_zeroes() and as a consequence always unmap if librbd is too old It's not clear what qemu's expectation is but in general Write Zeroes is allowed to unmap. The only guarantee is that subsequent reads return zeroes, everything else is a hint. This is how it is specified in the kernel and in the NVMe spec. In particular, block/nvme.c implements it as follows: if (flags & BDRV_REQ_MAY_UNMAP) { cdw12 |= (1 << 25); } This sets the Deallocate bit. But if it's not set, the device may still deallocate: """ If the Deallocate bit (CDW12.DEAC) is set to '1' in a Write Zeroes command, and the namespace supports clearing all bytes to 0h in the values read (e.g., bits 2:0 in the DLFEAT field are set to 001b) from a deallocated logical block and its metadata (excluding protection information), then for each specified logical block, the controller: - should deallocate that logical block; ... If the Deallocate bit is cleared to '0' in a Write Zeroes command, and the namespace supports clearing all bytes to 0h in the values read (e.g., bits 2:0 in the DLFEAT field are set to 001b) from a deallocated logical block and its metadata (excluding protection information), then, for each specified logical block, the controller: - may deallocate that logical block; """ https://nvmexpress.org/wp-content/uploads/NVM-Express-NVM-Command-Set-Specification-2021.06.02-Ratified-1.pdf b) set BDRV_REQ_NO_FALLBACK in supported_zero_flags Again, it's not clear what qemu expects here, but without it we end up in a ridiculous situation where specifying the "don't allow slow fallback" switch immediately fails all efficient zeroing requests on a device where Write Zeroes is always efficient: $ qemu-io -c 'help write' | grep -- '-[zun]' -n, -- with -z, don't allow slow fallback -u, -- with -z, allow unmapping -z, -- write zeroes using blk_co_pwrite_zeroes $ qemu-io -f rbd -c 'write -z -u -n 0 1M' rbd:foo/bar write failed: Operation not supported --->8--- Signed-off-by: Peter Lieven <pl@kamp.de> Reviewed-by: Ilya Dryomov <idryomov@gmail.com> Message-Id: <20210702172356.11574-6-idryomov@gmail.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2021-07-02 19:23:55 +02:00
#ifdef LIBRBD_SUPPORTS_WRITE_ZEROES
static int
coroutine_fn qemu_rbd_co_pwrite_zeroes(BlockDriverState *bs, int64_t offset,
block: use int64_t instead of int in driver write_zeroes handlers We are generally moving to int64_t for both offset and bytes parameters on all io paths. Main motivation is realization of 64-bit write_zeroes operation for fast zeroing large disk chunks, up to the whole disk. We chose signed type, to be consistent with off_t (which is signed) and with possibility for signed return type (where negative value means error). So, convert driver write_zeroes handlers bytes parameter to int64_t. The only caller of all updated function is bdrv_co_do_pwrite_zeroes(). bdrv_co_do_pwrite_zeroes() itself is of course OK with widening of callee parameter type. Also, bdrv_co_do_pwrite_zeroes()'s max_write_zeroes is limited to INT_MAX. So, updated functions all are safe, they will not get "bytes" larger than before. Still, let's look through all updated functions, and add assertions to the ones which are actually unprepared to values larger than INT_MAX. For these drivers also set explicit max_pwrite_zeroes limit. Let's go: blkdebug: calculations can't overflow, thanks to bdrv_check_qiov_request() in generic layer. rule_check() and bdrv_co_pwrite_zeroes() both have 64bit argument. blklogwrites: pass to blk_log_writes_co_log() with 64bit argument. blkreplay, copy-on-read, filter-compress: pass to bdrv_co_pwrite_zeroes() which is OK copy-before-write: Calls cbw_do_copy_before_write() and bdrv_co_pwrite_zeroes, both have 64bit argument. file-posix: both handler calls raw_do_pwrite_zeroes, which is updated. In raw_do_pwrite_zeroes() calculations are OK due to bdrv_check_qiov_request(), bytes go to RawPosixAIOData::aio_nbytes which is uint64_t. Check also where that uint64_t gets handed: handle_aiocb_write_zeroes_block() passes a uint64_t[2] to ioctl(BLKZEROOUT), handle_aiocb_write_zeroes() calls do_fallocate() which takes off_t (and we compile to always have 64-bit off_t), as does handle_aiocb_write_zeroes_unmap. All look safe. gluster: bytes go to GlusterAIOCB::size which is int64_t and to glfs_zerofill_async works with off_t. iscsi: Aha, here we deal with iscsi_writesame16_task() that has uint32_t num_blocks argument and iscsi_writesame16_task() has uint16_t argument. Make comments, add assertions and clarify max_pwrite_zeroes calculation. iscsi_allocmap_() functions already has int64_t argument is_byte_request_lun_aligned is simple to update, do it. mirror_top: pass to bdrv_mirror_top_do_write which has uint64_t argument nbd: Aha, here we have protocol limitation, and NBDRequest::len is uint32_t. max_pwrite_zeroes is cleanly set to 32bit value, so we are OK for now. nvme: Again, protocol limitation. And no inherent limit for write-zeroes at all. But from code that calculates cdw12 it's obvious that we do have limit and alignment. Let's clarify it. Also, obviously the code is not prepared to handle bytes=0. Let's handle this case too. trace events already 64bit preallocate: pass to handle_write() and bdrv_co_pwrite_zeroes(), both 64bit. rbd: pass to qemu_rbd_start_co() which is 64bit. qcow2: offset + bytes and alignment still works good (thanks to bdrv_check_qiov_request()), so tail calculation is OK qcow2_subcluster_zeroize() has 64bit argument, should be OK trace events updated qed: qed_co_request wants int nb_sectors. Also in code we have size_t used for request length which may be 32bit. So, let's just keep INT_MAX as a limit (aligning it down to pwrite_zeroes_alignment) and don't care. raw-format: Is OK. raw_adjust_offset and bdrv_co_pwrite_zeroes are both 64bit. throttle: Both throttle_group_co_io_limits_intercept() and bdrv_co_pwrite_zeroes() are 64bit. vmdk: pass to vmdk_pwritev which is 64bit quorum: pass to quorum_co_pwritev() which is 64bit Hooray! At this point all block drivers are prepared to support 64bit write-zero requests, or have explicitly set max_pwrite_zeroes. Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Message-Id: <20210903102807.27127-8-vsementsov@virtuozzo.com> Reviewed-by: Eric Blake <eblake@redhat.com> [eblake: use <= rather than < in assertions relying on max_pwrite_zeroes] Signed-off-by: Eric Blake <eblake@redhat.com>
2021-09-03 12:28:03 +02:00
int64_t bytes, BdrvRequestFlags flags)
block/rbd: add write zeroes support This patch wittingly sets BDRV_REQ_NO_FALLBACK and silently ignores BDRV_REQ_MAY_UNMAP for older librbd versions. The rationale for this is as follows (citing Ilya Dryomov current RBD maintainer): ---8<--- a) remove the BDRV_REQ_MAY_UNMAP check in qemu_rbd_co_pwrite_zeroes() and as a consequence always unmap if librbd is too old It's not clear what qemu's expectation is but in general Write Zeroes is allowed to unmap. The only guarantee is that subsequent reads return zeroes, everything else is a hint. This is how it is specified in the kernel and in the NVMe spec. In particular, block/nvme.c implements it as follows: if (flags & BDRV_REQ_MAY_UNMAP) { cdw12 |= (1 << 25); } This sets the Deallocate bit. But if it's not set, the device may still deallocate: """ If the Deallocate bit (CDW12.DEAC) is set to '1' in a Write Zeroes command, and the namespace supports clearing all bytes to 0h in the values read (e.g., bits 2:0 in the DLFEAT field are set to 001b) from a deallocated logical block and its metadata (excluding protection information), then for each specified logical block, the controller: - should deallocate that logical block; ... If the Deallocate bit is cleared to '0' in a Write Zeroes command, and the namespace supports clearing all bytes to 0h in the values read (e.g., bits 2:0 in the DLFEAT field are set to 001b) from a deallocated logical block and its metadata (excluding protection information), then, for each specified logical block, the controller: - may deallocate that logical block; """ https://nvmexpress.org/wp-content/uploads/NVM-Express-NVM-Command-Set-Specification-2021.06.02-Ratified-1.pdf b) set BDRV_REQ_NO_FALLBACK in supported_zero_flags Again, it's not clear what qemu expects here, but without it we end up in a ridiculous situation where specifying the "don't allow slow fallback" switch immediately fails all efficient zeroing requests on a device where Write Zeroes is always efficient: $ qemu-io -c 'help write' | grep -- '-[zun]' -n, -- with -z, don't allow slow fallback -u, -- with -z, allow unmapping -z, -- write zeroes using blk_co_pwrite_zeroes $ qemu-io -f rbd -c 'write -z -u -n 0 1M' rbd:foo/bar write failed: Operation not supported --->8--- Signed-off-by: Peter Lieven <pl@kamp.de> Reviewed-by: Ilya Dryomov <idryomov@gmail.com> Message-Id: <20210702172356.11574-6-idryomov@gmail.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2021-07-02 19:23:55 +02:00
{
block: use int64_t instead of int in driver write_zeroes handlers We are generally moving to int64_t for both offset and bytes parameters on all io paths. Main motivation is realization of 64-bit write_zeroes operation for fast zeroing large disk chunks, up to the whole disk. We chose signed type, to be consistent with off_t (which is signed) and with possibility for signed return type (where negative value means error). So, convert driver write_zeroes handlers bytes parameter to int64_t. The only caller of all updated function is bdrv_co_do_pwrite_zeroes(). bdrv_co_do_pwrite_zeroes() itself is of course OK with widening of callee parameter type. Also, bdrv_co_do_pwrite_zeroes()'s max_write_zeroes is limited to INT_MAX. So, updated functions all are safe, they will not get "bytes" larger than before. Still, let's look through all updated functions, and add assertions to the ones which are actually unprepared to values larger than INT_MAX. For these drivers also set explicit max_pwrite_zeroes limit. Let's go: blkdebug: calculations can't overflow, thanks to bdrv_check_qiov_request() in generic layer. rule_check() and bdrv_co_pwrite_zeroes() both have 64bit argument. blklogwrites: pass to blk_log_writes_co_log() with 64bit argument. blkreplay, copy-on-read, filter-compress: pass to bdrv_co_pwrite_zeroes() which is OK copy-before-write: Calls cbw_do_copy_before_write() and bdrv_co_pwrite_zeroes, both have 64bit argument. file-posix: both handler calls raw_do_pwrite_zeroes, which is updated. In raw_do_pwrite_zeroes() calculations are OK due to bdrv_check_qiov_request(), bytes go to RawPosixAIOData::aio_nbytes which is uint64_t. Check also where that uint64_t gets handed: handle_aiocb_write_zeroes_block() passes a uint64_t[2] to ioctl(BLKZEROOUT), handle_aiocb_write_zeroes() calls do_fallocate() which takes off_t (and we compile to always have 64-bit off_t), as does handle_aiocb_write_zeroes_unmap. All look safe. gluster: bytes go to GlusterAIOCB::size which is int64_t and to glfs_zerofill_async works with off_t. iscsi: Aha, here we deal with iscsi_writesame16_task() that has uint32_t num_blocks argument and iscsi_writesame16_task() has uint16_t argument. Make comments, add assertions and clarify max_pwrite_zeroes calculation. iscsi_allocmap_() functions already has int64_t argument is_byte_request_lun_aligned is simple to update, do it. mirror_top: pass to bdrv_mirror_top_do_write which has uint64_t argument nbd: Aha, here we have protocol limitation, and NBDRequest::len is uint32_t. max_pwrite_zeroes is cleanly set to 32bit value, so we are OK for now. nvme: Again, protocol limitation. And no inherent limit for write-zeroes at all. But from code that calculates cdw12 it's obvious that we do have limit and alignment. Let's clarify it. Also, obviously the code is not prepared to handle bytes=0. Let's handle this case too. trace events already 64bit preallocate: pass to handle_write() and bdrv_co_pwrite_zeroes(), both 64bit. rbd: pass to qemu_rbd_start_co() which is 64bit. qcow2: offset + bytes and alignment still works good (thanks to bdrv_check_qiov_request()), so tail calculation is OK qcow2_subcluster_zeroize() has 64bit argument, should be OK trace events updated qed: qed_co_request wants int nb_sectors. Also in code we have size_t used for request length which may be 32bit. So, let's just keep INT_MAX as a limit (aligning it down to pwrite_zeroes_alignment) and don't care. raw-format: Is OK. raw_adjust_offset and bdrv_co_pwrite_zeroes are both 64bit. throttle: Both throttle_group_co_io_limits_intercept() and bdrv_co_pwrite_zeroes() are 64bit. vmdk: pass to vmdk_pwritev which is 64bit quorum: pass to quorum_co_pwritev() which is 64bit Hooray! At this point all block drivers are prepared to support 64bit write-zero requests, or have explicitly set max_pwrite_zeroes. Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Message-Id: <20210903102807.27127-8-vsementsov@virtuozzo.com> Reviewed-by: Eric Blake <eblake@redhat.com> [eblake: use <= rather than < in assertions relying on max_pwrite_zeroes] Signed-off-by: Eric Blake <eblake@redhat.com>
2021-09-03 12:28:03 +02:00
return qemu_rbd_start_co(bs, offset, bytes, NULL, flags,
block/rbd: add write zeroes support This patch wittingly sets BDRV_REQ_NO_FALLBACK and silently ignores BDRV_REQ_MAY_UNMAP for older librbd versions. The rationale for this is as follows (citing Ilya Dryomov current RBD maintainer): ---8<--- a) remove the BDRV_REQ_MAY_UNMAP check in qemu_rbd_co_pwrite_zeroes() and as a consequence always unmap if librbd is too old It's not clear what qemu's expectation is but in general Write Zeroes is allowed to unmap. The only guarantee is that subsequent reads return zeroes, everything else is a hint. This is how it is specified in the kernel and in the NVMe spec. In particular, block/nvme.c implements it as follows: if (flags & BDRV_REQ_MAY_UNMAP) { cdw12 |= (1 << 25); } This sets the Deallocate bit. But if it's not set, the device may still deallocate: """ If the Deallocate bit (CDW12.DEAC) is set to '1' in a Write Zeroes command, and the namespace supports clearing all bytes to 0h in the values read (e.g., bits 2:0 in the DLFEAT field are set to 001b) from a deallocated logical block and its metadata (excluding protection information), then for each specified logical block, the controller: - should deallocate that logical block; ... If the Deallocate bit is cleared to '0' in a Write Zeroes command, and the namespace supports clearing all bytes to 0h in the values read (e.g., bits 2:0 in the DLFEAT field are set to 001b) from a deallocated logical block and its metadata (excluding protection information), then, for each specified logical block, the controller: - may deallocate that logical block; """ https://nvmexpress.org/wp-content/uploads/NVM-Express-NVM-Command-Set-Specification-2021.06.02-Ratified-1.pdf b) set BDRV_REQ_NO_FALLBACK in supported_zero_flags Again, it's not clear what qemu expects here, but without it we end up in a ridiculous situation where specifying the "don't allow slow fallback" switch immediately fails all efficient zeroing requests on a device where Write Zeroes is always efficient: $ qemu-io -c 'help write' | grep -- '-[zun]' -n, -- with -z, don't allow slow fallback -u, -- with -z, allow unmapping -z, -- write zeroes using blk_co_pwrite_zeroes $ qemu-io -f rbd -c 'write -z -u -n 0 1M' rbd:foo/bar write failed: Operation not supported --->8--- Signed-off-by: Peter Lieven <pl@kamp.de> Reviewed-by: Ilya Dryomov <idryomov@gmail.com> Message-Id: <20210702172356.11574-6-idryomov@gmail.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2021-07-02 19:23:55 +02:00
RBD_AIO_WRITE_ZEROES);
}
#endif
static int qemu_rbd_getinfo(BlockDriverState *bs, BlockDriverInfo *bdi)
{
BDRVRBDState *s = bs->opaque;
bdi->cluster_size = s->object_size;
return 0;
}
static ImageInfoSpecific *qemu_rbd_get_specific_info(BlockDriverState *bs,
Error **errp)
{
BDRVRBDState *s = bs->opaque;
ImageInfoSpecific *spec_info;
char buf[RBD_ENCRYPTION_LUKS_HEADER_VERIFICATION_LEN] = {0};
int r;
if (s->image_size >= RBD_ENCRYPTION_LUKS_HEADER_VERIFICATION_LEN) {
r = rbd_read(s->image, 0,
RBD_ENCRYPTION_LUKS_HEADER_VERIFICATION_LEN, buf);
if (r < 0) {
error_setg_errno(errp, -r, "cannot read image start for probe");
return NULL;
}
}
spec_info = g_new(ImageInfoSpecific, 1);
*spec_info = (ImageInfoSpecific){
.type = IMAGE_INFO_SPECIFIC_KIND_RBD,
.u.rbd.data = g_new0(ImageInfoSpecificRbd, 1),
};
if (memcmp(buf, rbd_luks_header_verification,
RBD_ENCRYPTION_LUKS_HEADER_VERIFICATION_LEN) == 0) {
spec_info->u.rbd.data->encryption_format =
RBD_IMAGE_ENCRYPTION_FORMAT_LUKS;
spec_info->u.rbd.data->has_encryption_format = true;
} else if (memcmp(buf, rbd_luks2_header_verification,
RBD_ENCRYPTION_LUKS_HEADER_VERIFICATION_LEN) == 0) {
spec_info->u.rbd.data->encryption_format =
RBD_IMAGE_ENCRYPTION_FORMAT_LUKS2;
spec_info->u.rbd.data->has_encryption_format = true;
} else {
spec_info->u.rbd.data->has_encryption_format = false;
}
return spec_info;
}
/*
* rbd_diff_iterate2 allows to interrupt the exection by returning a negative
* value in the callback routine. Choose a value that does not conflict with
* an existing exitcode and return it if we want to prematurely stop the
* execution because we detected a change in the allocation status.
*/
#define QEMU_RBD_EXIT_DIFF_ITERATE2 -9000
static int qemu_rbd_diff_iterate_cb(uint64_t offs, size_t len,
int exists, void *opaque)
{
RBDDiffIterateReq *req = opaque;
assert(req->offs + req->bytes <= offs);
/* treat a hole like an unallocated area and bail out */
if (!exists) {
return 0;
}
if (!req->exists && offs > req->offs) {
/*
* we started in an unallocated area and hit the first allocated
* block. req->bytes must be set to the length of the unallocated area
* before the allocated area. stop further processing.
*/
req->bytes = offs - req->offs;
return QEMU_RBD_EXIT_DIFF_ITERATE2;
}
if (req->exists && offs > req->offs + req->bytes) {
/*
* we started in an allocated area and jumped over an unallocated area,
* req->bytes contains the length of the allocated area before the
* unallocated area. stop further processing.
*/
return QEMU_RBD_EXIT_DIFF_ITERATE2;
}
req->bytes += len;
req->exists = true;
return 0;
}
static int coroutine_fn qemu_rbd_co_block_status(BlockDriverState *bs,
bool want_zero, int64_t offset,
int64_t bytes, int64_t *pnum,
int64_t *map,
BlockDriverState **file)
{
BDRVRBDState *s = bs->opaque;
int status, r;
RBDDiffIterateReq req = { .offs = offset };
uint64_t features, flags;
uint64_t head = 0;
assert(offset + bytes <= s->image_size);
/* default to all sectors allocated */
status = BDRV_BLOCK_DATA | BDRV_BLOCK_OFFSET_VALID;
*map = offset;
*file = bs;
*pnum = bytes;
/* check if RBD image supports fast-diff */
r = rbd_get_features(s->image, &features);
if (r < 0) {
return status;
}
if (!(features & RBD_FEATURE_FAST_DIFF)) {
return status;
}
/* check if RBD fast-diff result is valid */
r = rbd_get_flags(s->image, &flags);
if (r < 0) {
return status;
}
if (flags & RBD_FLAG_FAST_DIFF_INVALID) {
return status;
}
#if LIBRBD_VERSION_CODE < LIBRBD_VERSION(1, 17, 0)
/*
* librbd had a bug until early 2022 that affected all versions of ceph that
* supported fast-diff. This bug results in reporting of incorrect offsets
* if the offset parameter to rbd_diff_iterate2 is not object aligned.
* Work around this bug by rounding down the offset to object boundaries.
* This is OK because we call rbd_diff_iterate2 with whole_object = true.
* However, this workaround only works for non cloned images with default
* striping.
*
* See: https://tracker.ceph.com/issues/53784
*/
/* check if RBD image has non-default striping enabled */
if (features & RBD_FEATURE_STRIPINGV2) {
return status;
}
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-Wdeprecated-declarations"
/*
* check if RBD image is a clone (= has a parent).
*
* rbd_get_parent_info is deprecated from Nautilus onwards, but the
* replacement rbd_get_parent is not present in Luminous and Mimic.
*/
if (rbd_get_parent_info(s->image, NULL, 0, NULL, 0, NULL, 0) != -ENOENT) {
return status;
}
#pragma GCC diagnostic pop
head = req.offs & (s->object_size - 1);
req.offs -= head;
bytes += head;
#endif
r = rbd_diff_iterate2(s->image, NULL, req.offs, bytes, true, true,
qemu_rbd_diff_iterate_cb, &req);
if (r < 0 && r != QEMU_RBD_EXIT_DIFF_ITERATE2) {
return status;
}
assert(req.bytes <= bytes);
if (!req.exists) {
if (r == 0) {
/*
* rbd_diff_iterate2 does not invoke callbacks for unallocated
* areas. This here catches the case where no callback was
* invoked at all (req.bytes == 0).
*/
assert(req.bytes == 0);
req.bytes = bytes;
}
status = BDRV_BLOCK_ZERO | BDRV_BLOCK_OFFSET_VALID;
}
assert(req.bytes > head);
*pnum = req.bytes - head;
return status;
}
static int64_t qemu_rbd_getlength(BlockDriverState *bs)
{
BDRVRBDState *s = bs->opaque;
int r;
r = rbd_get_size(s->image, &s->image_size);
if (r < 0) {
return r;
}
return s->image_size;
}
static int coroutine_fn qemu_rbd_co_truncate(BlockDriverState *bs,
int64_t offset,
bool exact,
PreallocMode prealloc,
BdrvRequestFlags flags,
Error **errp)
{
int r;
if (prealloc != PREALLOC_MODE_OFF) {
error_setg(errp, "Unsupported preallocation mode '%s'",
PreallocMode_str(prealloc));
return -ENOTSUP;
}
r = qemu_rbd_resize(bs, offset);
if (r < 0) {
error_setg_errno(errp, -r, "Failed to resize file");
return r;
}
return 0;
}
static int qemu_rbd_snap_create(BlockDriverState *bs,
QEMUSnapshotInfo *sn_info)
{
BDRVRBDState *s = bs->opaque;
int r;
if (sn_info->name[0] == '\0') {
return -EINVAL; /* we need a name for rbd snapshots */
}
/*
* rbd snapshots are using the name as the user controlled unique identifier
* we can't use the rbd snapid for that purpose, as it can't be set
*/
if (sn_info->id_str[0] != '\0' &&
strcmp(sn_info->id_str, sn_info->name) != 0) {
return -EINVAL;
}
if (strlen(sn_info->name) >= sizeof(sn_info->id_str)) {
return -ERANGE;
}
r = rbd_snap_create(s->image, sn_info->name);
if (r < 0) {
error_report("failed to create snap: %s", strerror(-r));
return r;
}
return 0;
}
static int qemu_rbd_snap_remove(BlockDriverState *bs,
const char *snapshot_id,
const char *snapshot_name,
Error **errp)
{
BDRVRBDState *s = bs->opaque;
int r;
if (!snapshot_name) {
error_setg(errp, "rbd need a valid snapshot name");
return -EINVAL;
}
/* If snapshot_id is specified, it must be equal to name, see
qemu_rbd_snap_list() */
if (snapshot_id && strcmp(snapshot_id, snapshot_name)) {
error_setg(errp,
"rbd do not support snapshot id, it should be NULL or "
"equal to snapshot name");
return -EINVAL;
}
r = rbd_snap_remove(s->image, snapshot_name);
if (r < 0) {
error_setg_errno(errp, -r, "Failed to remove the snapshot");
}
return r;
}
static int qemu_rbd_snap_rollback(BlockDriverState *bs,
const char *snapshot_name)
{
BDRVRBDState *s = bs->opaque;
return rbd_snap_rollback(s->image, snapshot_name);
}
static int qemu_rbd_snap_list(BlockDriverState *bs,
QEMUSnapshotInfo **psn_tab)
{
BDRVRBDState *s = bs->opaque;
QEMUSnapshotInfo *sn_info, *sn_tab = NULL;
int i, snap_count;
rbd_snap_info_t *snaps;
int max_snaps = RBD_MAX_SNAPS;
do {
snaps = g_new(rbd_snap_info_t, max_snaps);
snap_count = rbd_snap_list(s->image, snaps, &max_snaps);
if (snap_count <= 0) {
g_free(snaps);
}
} while (snap_count == -ERANGE);
if (snap_count <= 0) {
goto done;
}
sn_tab = g_new0(QEMUSnapshotInfo, snap_count);
for (i = 0; i < snap_count; i++) {
const char *snap_name = snaps[i].name;
sn_info = sn_tab + i;
pstrcpy(sn_info->id_str, sizeof(sn_info->id_str), snap_name);
pstrcpy(sn_info->name, sizeof(sn_info->name), snap_name);
sn_info->vm_state_size = snaps[i].size;
sn_info->date_sec = 0;
sn_info->date_nsec = 0;
sn_info->vm_clock_nsec = 0;
}
rbd_snap_list_end(snaps);
g_free(snaps);
done:
*psn_tab = sn_tab;
return snap_count;
}
static void coroutine_fn qemu_rbd_co_invalidate_cache(BlockDriverState *bs,
Error **errp)
{
BDRVRBDState *s = bs->opaque;
int r = rbd_invalidate_cache(s->image);
if (r < 0) {
error_setg_errno(errp, -r, "Failed to invalidate the cache");
}
}
static QemuOptsList qemu_rbd_create_opts = {
.name = "rbd-create-opts",
.head = QTAILQ_HEAD_INITIALIZER(qemu_rbd_create_opts.head),
.desc = {
{
.name = BLOCK_OPT_SIZE,
.type = QEMU_OPT_SIZE,
.help = "Virtual disk size"
},
{
.name = BLOCK_OPT_CLUSTER_SIZE,
.type = QEMU_OPT_SIZE,
.help = "RBD object size"
},
{
.name = "password-secret",
.type = QEMU_OPT_STRING,
.help = "ID of secret providing the password",
},
{
.name = "encrypt.format",
.type = QEMU_OPT_STRING,
.help = "Encrypt the image, format choices: 'luks', 'luks2'",
},
{
.name = "encrypt.cipher-alg",
.type = QEMU_OPT_STRING,
.help = "Name of encryption cipher algorithm"
" (allowed values: aes-128, aes-256)",
},
{
.name = "encrypt.key-secret",
.type = QEMU_OPT_STRING,
.help = "ID of secret providing LUKS passphrase",
},
{ /* end of list */ }
}
};
static const char *const qemu_rbd_strong_runtime_opts[] = {
"pool",
"namespace",
"image",
"conf",
"snapshot",
"user",
"server.",
"password-secret",
NULL
};
static BlockDriver bdrv_rbd = {
.format_name = "rbd",
.instance_size = sizeof(BDRVRBDState),
.bdrv_parse_filename = qemu_rbd_parse_filename,
.bdrv_file_open = qemu_rbd_open,
.bdrv_close = qemu_rbd_close,
.bdrv_reopen_prepare = qemu_rbd_reopen_prepare,
.bdrv_co_create = qemu_rbd_co_create,
.bdrv_co_create_opts = qemu_rbd_co_create_opts,
.bdrv_has_zero_init = bdrv_has_zero_init_1,
.bdrv_get_info = qemu_rbd_getinfo,
.bdrv_get_specific_info = qemu_rbd_get_specific_info,
.create_opts = &qemu_rbd_create_opts,
.bdrv_getlength = qemu_rbd_getlength,
.bdrv_co_truncate = qemu_rbd_co_truncate,
.protocol_name = "rbd",
.bdrv_co_preadv = qemu_rbd_co_preadv,
.bdrv_co_pwritev = qemu_rbd_co_pwritev,
.bdrv_co_flush_to_disk = qemu_rbd_co_flush,
.bdrv_co_pdiscard = qemu_rbd_co_pdiscard,
block/rbd: add write zeroes support This patch wittingly sets BDRV_REQ_NO_FALLBACK and silently ignores BDRV_REQ_MAY_UNMAP for older librbd versions. The rationale for this is as follows (citing Ilya Dryomov current RBD maintainer): ---8<--- a) remove the BDRV_REQ_MAY_UNMAP check in qemu_rbd_co_pwrite_zeroes() and as a consequence always unmap if librbd is too old It's not clear what qemu's expectation is but in general Write Zeroes is allowed to unmap. The only guarantee is that subsequent reads return zeroes, everything else is a hint. This is how it is specified in the kernel and in the NVMe spec. In particular, block/nvme.c implements it as follows: if (flags & BDRV_REQ_MAY_UNMAP) { cdw12 |= (1 << 25); } This sets the Deallocate bit. But if it's not set, the device may still deallocate: """ If the Deallocate bit (CDW12.DEAC) is set to '1' in a Write Zeroes command, and the namespace supports clearing all bytes to 0h in the values read (e.g., bits 2:0 in the DLFEAT field are set to 001b) from a deallocated logical block and its metadata (excluding protection information), then for each specified logical block, the controller: - should deallocate that logical block; ... If the Deallocate bit is cleared to '0' in a Write Zeroes command, and the namespace supports clearing all bytes to 0h in the values read (e.g., bits 2:0 in the DLFEAT field are set to 001b) from a deallocated logical block and its metadata (excluding protection information), then, for each specified logical block, the controller: - may deallocate that logical block; """ https://nvmexpress.org/wp-content/uploads/NVM-Express-NVM-Command-Set-Specification-2021.06.02-Ratified-1.pdf b) set BDRV_REQ_NO_FALLBACK in supported_zero_flags Again, it's not clear what qemu expects here, but without it we end up in a ridiculous situation where specifying the "don't allow slow fallback" switch immediately fails all efficient zeroing requests on a device where Write Zeroes is always efficient: $ qemu-io -c 'help write' | grep -- '-[zun]' -n, -- with -z, don't allow slow fallback -u, -- with -z, allow unmapping -z, -- write zeroes using blk_co_pwrite_zeroes $ qemu-io -f rbd -c 'write -z -u -n 0 1M' rbd:foo/bar write failed: Operation not supported --->8--- Signed-off-by: Peter Lieven <pl@kamp.de> Reviewed-by: Ilya Dryomov <idryomov@gmail.com> Message-Id: <20210702172356.11574-6-idryomov@gmail.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2021-07-02 19:23:55 +02:00
#ifdef LIBRBD_SUPPORTS_WRITE_ZEROES
.bdrv_co_pwrite_zeroes = qemu_rbd_co_pwrite_zeroes,
#endif
.bdrv_co_block_status = qemu_rbd_co_block_status,
.bdrv_snapshot_create = qemu_rbd_snap_create,
.bdrv_snapshot_delete = qemu_rbd_snap_remove,
.bdrv_snapshot_list = qemu_rbd_snap_list,
.bdrv_snapshot_goto = qemu_rbd_snap_rollback,
.bdrv_co_invalidate_cache = qemu_rbd_co_invalidate_cache,
.strong_runtime_opts = qemu_rbd_strong_runtime_opts,
};
static void bdrv_rbd_init(void)
{
bdrv_register(&bdrv_rbd);
}
block_init(bdrv_rbd_init);