linux/net/ceph/auth.c
Ilya Dryomov 6c1ea260f8 libceph: make authorizer destruction independent of ceph_auth_client
Starting the kernel client with cephx disabled and then enabling cephx
and restarting userspace daemons can result in a crash:

    [262671.478162] BUG: unable to handle kernel paging request at ffffebe000000000
    [262671.531460] IP: [<ffffffff811cd04a>] kfree+0x5a/0x130
    [262671.584334] PGD 0
    [262671.635847] Oops: 0000 [#1] SMP
    [262672.055841] CPU: 22 PID: 2961272 Comm: kworker/22:2 Not tainted 4.2.0-34-generic #39~14.04.1-Ubuntu
    [262672.162338] Hardware name: Dell Inc. PowerEdge R720/068CDY, BIOS 2.4.3 07/09/2014
    [262672.268937] Workqueue: ceph-msgr con_work [libceph]
    [262672.322290] task: ffff88081c2d0dc0 ti: ffff880149ae8000 task.ti: ffff880149ae8000
    [262672.428330] RIP: 0010:[<ffffffff811cd04a>]  [<ffffffff811cd04a>] kfree+0x5a/0x130
    [262672.535880] RSP: 0018:ffff880149aeba58  EFLAGS: 00010286
    [262672.589486] RAX: 000001e000000000 RBX: 0000000000000012 RCX: ffff8807e7461018
    [262672.695980] RDX: 000077ff80000000 RSI: ffff88081af2be04 RDI: 0000000000000012
    [262672.803668] RBP: ffff880149aeba78 R08: 0000000000000000 R09: 0000000000000000
    [262672.912299] R10: ffffebe000000000 R11: ffff880819a60e78 R12: ffff8800aec8df40
    [262673.021769] R13: ffffffffc035f70f R14: ffff8807e5b138e0 R15: ffff880da9785840
    [262673.131722] FS:  0000000000000000(0000) GS:ffff88081fac0000(0000) knlGS:0000000000000000
    [262673.245377] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
    [262673.303281] CR2: ffffebe000000000 CR3: 0000000001c0d000 CR4: 00000000001406e0
    [262673.417556] Stack:
    [262673.472943]  ffff880149aeba88 ffff88081af2be04 ffff8800aec8df40 ffff88081af2be04
    [262673.583767]  ffff880149aeba98 ffffffffc035f70f ffff880149aebac8 ffff8800aec8df00
    [262673.694546]  ffff880149aebac8 ffffffffc035c89e ffff8807e5b138e0 ffff8805b047f800
    [262673.805230] Call Trace:
    [262673.859116]  [<ffffffffc035f70f>] ceph_x_destroy_authorizer+0x1f/0x50 [libceph]
    [262673.968705]  [<ffffffffc035c89e>] ceph_auth_destroy_authorizer+0x3e/0x60 [libceph]
    [262674.078852]  [<ffffffffc0352805>] put_osd+0x45/0x80 [libceph]
    [262674.134249]  [<ffffffffc035290e>] remove_osd+0xae/0x140 [libceph]
    [262674.189124]  [<ffffffffc0352aa3>] __reset_osd+0x103/0x150 [libceph]
    [262674.243749]  [<ffffffffc0354703>] kick_requests+0x223/0x460 [libceph]
    [262674.297485]  [<ffffffffc03559e2>] ceph_osdc_handle_map+0x282/0x5e0 [libceph]
    [262674.350813]  [<ffffffffc035022e>] dispatch+0x4e/0x720 [libceph]
    [262674.403312]  [<ffffffffc034bd91>] try_read+0x3d1/0x1090 [libceph]
    [262674.454712]  [<ffffffff810ab7c2>] ? dequeue_entity+0x152/0x690
    [262674.505096]  [<ffffffffc034cb1b>] con_work+0xcb/0x1300 [libceph]
    [262674.555104]  [<ffffffff8108fb3e>] process_one_work+0x14e/0x3d0
    [262674.604072]  [<ffffffff810901ea>] worker_thread+0x11a/0x470
    [262674.652187]  [<ffffffff810900d0>] ? rescuer_thread+0x310/0x310
    [262674.699022]  [<ffffffff810957a2>] kthread+0xd2/0xf0
    [262674.744494]  [<ffffffff810956d0>] ? kthread_create_on_node+0x1c0/0x1c0
    [262674.789543]  [<ffffffff817bd81f>] ret_from_fork+0x3f/0x70
    [262674.834094]  [<ffffffff810956d0>] ? kthread_create_on_node+0x1c0/0x1c0

What happens is the following:

    (1) new MON session is established
    (2) old "none" ac is destroyed
    (3) new "cephx" ac is constructed
    ...
    (4) old OSD session (w/ "none" authorizer) is put
          ceph_auth_destroy_authorizer(ac, osd->o_auth.authorizer)

osd->o_auth.authorizer in the "none" case is just a bare pointer into
ac, which contains a single static copy for all services.  By the time
we get to (4), "none" ac, freed in (2), is long gone.  On top of that,
a new vtable installed in (3) points us at ceph_x_destroy_authorizer(),
so we end up trying to destroy a "none" authorizer with a "cephx"
destructor operating on invalid memory!

To fix this, decouple authorizer destruction from ac and do away with
a single static "none" authorizer by making a copy for each OSD or MDS
session.  Authorizers themselves are independent of ac and so there is
no reason for destroy_authorizer() to be an ac op.  Make it an op on
the authorizer itself by turning ceph_authorizer into a real struct.

Fixes: http://tracker.ceph.com/issues/15447

Reported-by: Alan Zhang <alan.zhang@linux.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Reviewed-by: Sage Weil <sage@redhat.com>
2016-04-25 20:54:13 +02:00

337 lines
7.4 KiB
C

#include <linux/ceph/ceph_debug.h>
#include <linux/module.h>
#include <linux/err.h>
#include <linux/slab.h>
#include <linux/ceph/types.h>
#include <linux/ceph/decode.h>
#include <linux/ceph/libceph.h>
#include <linux/ceph/messenger.h>
#include "auth_none.h"
#include "auth_x.h"
/*
* get protocol handler
*/
static u32 supported_protocols[] = {
CEPH_AUTH_NONE,
CEPH_AUTH_CEPHX
};
static int ceph_auth_init_protocol(struct ceph_auth_client *ac, int protocol)
{
switch (protocol) {
case CEPH_AUTH_NONE:
return ceph_auth_none_init(ac);
case CEPH_AUTH_CEPHX:
return ceph_x_init(ac);
default:
return -ENOENT;
}
}
/*
* setup, teardown.
*/
struct ceph_auth_client *ceph_auth_init(const char *name, const struct ceph_crypto_key *key)
{
struct ceph_auth_client *ac;
int ret;
dout("auth_init name '%s'\n", name);
ret = -ENOMEM;
ac = kzalloc(sizeof(*ac), GFP_NOFS);
if (!ac)
goto out;
mutex_init(&ac->mutex);
ac->negotiating = true;
if (name)
ac->name = name;
else
ac->name = CEPH_AUTH_NAME_DEFAULT;
dout("auth_init name %s\n", ac->name);
ac->key = key;
return ac;
out:
return ERR_PTR(ret);
}
void ceph_auth_destroy(struct ceph_auth_client *ac)
{
dout("auth_destroy %p\n", ac);
if (ac->ops)
ac->ops->destroy(ac);
kfree(ac);
}
/*
* Reset occurs when reconnecting to the monitor.
*/
void ceph_auth_reset(struct ceph_auth_client *ac)
{
mutex_lock(&ac->mutex);
dout("auth_reset %p\n", ac);
if (ac->ops && !ac->negotiating)
ac->ops->reset(ac);
ac->negotiating = true;
mutex_unlock(&ac->mutex);
}
int ceph_entity_name_encode(const char *name, void **p, void *end)
{
int len = strlen(name);
if (*p + 2*sizeof(u32) + len > end)
return -ERANGE;
ceph_encode_32(p, CEPH_ENTITY_TYPE_CLIENT);
ceph_encode_32(p, len);
ceph_encode_copy(p, name, len);
return 0;
}
/*
* Initiate protocol negotiation with monitor. Include entity name
* and list supported protocols.
*/
int ceph_auth_build_hello(struct ceph_auth_client *ac, void *buf, size_t len)
{
struct ceph_mon_request_header *monhdr = buf;
void *p = monhdr + 1, *end = buf + len, *lenp;
int i, num;
int ret;
mutex_lock(&ac->mutex);
dout("auth_build_hello\n");
monhdr->have_version = 0;
monhdr->session_mon = cpu_to_le16(-1);
monhdr->session_mon_tid = 0;
ceph_encode_32(&p, 0); /* no protocol, yet */
lenp = p;
p += sizeof(u32);
ceph_decode_need(&p, end, 1 + sizeof(u32), bad);
ceph_encode_8(&p, 1);
num = ARRAY_SIZE(supported_protocols);
ceph_encode_32(&p, num);
ceph_decode_need(&p, end, num * sizeof(u32), bad);
for (i = 0; i < num; i++)
ceph_encode_32(&p, supported_protocols[i]);
ret = ceph_entity_name_encode(ac->name, &p, end);
if (ret < 0)
goto out;
ceph_decode_need(&p, end, sizeof(u64), bad);
ceph_encode_64(&p, ac->global_id);
ceph_encode_32(&lenp, p - lenp - sizeof(u32));
ret = p - buf;
out:
mutex_unlock(&ac->mutex);
return ret;
bad:
ret = -ERANGE;
goto out;
}
static int ceph_build_auth_request(struct ceph_auth_client *ac,
void *msg_buf, size_t msg_len)
{
struct ceph_mon_request_header *monhdr = msg_buf;
void *p = monhdr + 1;
void *end = msg_buf + msg_len;
int ret;
monhdr->have_version = 0;
monhdr->session_mon = cpu_to_le16(-1);
monhdr->session_mon_tid = 0;
ceph_encode_32(&p, ac->protocol);
ret = ac->ops->build_request(ac, p + sizeof(u32), end);
if (ret < 0) {
pr_err("error %d building auth method %s request\n", ret,
ac->ops->name);
goto out;
}
dout(" built request %d bytes\n", ret);
ceph_encode_32(&p, ret);
ret = p + ret - msg_buf;
out:
return ret;
}
/*
* Handle auth message from monitor.
*/
int ceph_handle_auth_reply(struct ceph_auth_client *ac,
void *buf, size_t len,
void *reply_buf, size_t reply_len)
{
void *p = buf;
void *end = buf + len;
int protocol;
s32 result;
u64 global_id;
void *payload, *payload_end;
int payload_len;
char *result_msg;
int result_msg_len;
int ret = -EINVAL;
mutex_lock(&ac->mutex);
dout("handle_auth_reply %p %p\n", p, end);
ceph_decode_need(&p, end, sizeof(u32) * 3 + sizeof(u64), bad);
protocol = ceph_decode_32(&p);
result = ceph_decode_32(&p);
global_id = ceph_decode_64(&p);
payload_len = ceph_decode_32(&p);
payload = p;
p += payload_len;
ceph_decode_need(&p, end, sizeof(u32), bad);
result_msg_len = ceph_decode_32(&p);
result_msg = p;
p += result_msg_len;
if (p != end)
goto bad;
dout(" result %d '%.*s' gid %llu len %d\n", result, result_msg_len,
result_msg, global_id, payload_len);
payload_end = payload + payload_len;
if (global_id && ac->global_id != global_id) {
dout(" set global_id %lld -> %lld\n", ac->global_id, global_id);
ac->global_id = global_id;
}
if (ac->negotiating) {
/* server does not support our protocols? */
if (!protocol && result < 0) {
ret = result;
goto out;
}
/* set up (new) protocol handler? */
if (ac->protocol && ac->protocol != protocol) {
ac->ops->destroy(ac);
ac->protocol = 0;
ac->ops = NULL;
}
if (ac->protocol != protocol) {
ret = ceph_auth_init_protocol(ac, protocol);
if (ret) {
pr_err("error %d on auth protocol %d init\n",
ret, protocol);
goto out;
}
}
ac->negotiating = false;
}
ret = ac->ops->handle_reply(ac, result, payload, payload_end);
if (ret == -EAGAIN) {
ret = ceph_build_auth_request(ac, reply_buf, reply_len);
} else if (ret) {
pr_err("auth method '%s' error %d\n", ac->ops->name, ret);
}
out:
mutex_unlock(&ac->mutex);
return ret;
bad:
pr_err("failed to decode auth msg\n");
ret = -EINVAL;
goto out;
}
int ceph_build_auth(struct ceph_auth_client *ac,
void *msg_buf, size_t msg_len)
{
int ret = 0;
mutex_lock(&ac->mutex);
if (!ac->protocol)
ret = ceph_auth_build_hello(ac, msg_buf, msg_len);
else if (ac->ops->should_authenticate(ac))
ret = ceph_build_auth_request(ac, msg_buf, msg_len);
mutex_unlock(&ac->mutex);
return ret;
}
int ceph_auth_is_authenticated(struct ceph_auth_client *ac)
{
int ret = 0;
mutex_lock(&ac->mutex);
if (ac->ops)
ret = ac->ops->is_authenticated(ac);
mutex_unlock(&ac->mutex);
return ret;
}
EXPORT_SYMBOL(ceph_auth_is_authenticated);
int ceph_auth_create_authorizer(struct ceph_auth_client *ac,
int peer_type,
struct ceph_auth_handshake *auth)
{
int ret = 0;
mutex_lock(&ac->mutex);
if (ac->ops && ac->ops->create_authorizer)
ret = ac->ops->create_authorizer(ac, peer_type, auth);
mutex_unlock(&ac->mutex);
return ret;
}
EXPORT_SYMBOL(ceph_auth_create_authorizer);
void ceph_auth_destroy_authorizer(struct ceph_authorizer *a)
{
a->destroy(a);
}
EXPORT_SYMBOL(ceph_auth_destroy_authorizer);
int ceph_auth_update_authorizer(struct ceph_auth_client *ac,
int peer_type,
struct ceph_auth_handshake *a)
{
int ret = 0;
mutex_lock(&ac->mutex);
if (ac->ops && ac->ops->update_authorizer)
ret = ac->ops->update_authorizer(ac, peer_type, a);
mutex_unlock(&ac->mutex);
return ret;
}
EXPORT_SYMBOL(ceph_auth_update_authorizer);
int ceph_auth_verify_authorizer_reply(struct ceph_auth_client *ac,
struct ceph_authorizer *a, size_t len)
{
int ret = 0;
mutex_lock(&ac->mutex);
if (ac->ops && ac->ops->verify_authorizer_reply)
ret = ac->ops->verify_authorizer_reply(ac, a, len);
mutex_unlock(&ac->mutex);
return ret;
}
EXPORT_SYMBOL(ceph_auth_verify_authorizer_reply);
void ceph_auth_invalidate_authorizer(struct ceph_auth_client *ac, int peer_type)
{
mutex_lock(&ac->mutex);
if (ac->ops && ac->ops->invalidate_authorizer)
ac->ops->invalidate_authorizer(ac, peer_type);
mutex_unlock(&ac->mutex);
}
EXPORT_SYMBOL(ceph_auth_invalidate_authorizer);