KASAN: use-after-free Read in recv_work

[syzbot]

Hello,

syzbot found the following issue on:

HEAD commit: b01f250d Add linux-next specific files for 20210129
git tree: linux-next
console output: https://syzkaller.appspot.com/x/log.txt?x=14366378d00000
kernel config: https://syzkaller.appspot.com/x/.config?x=725bc96dc234fda7
dashboard link: https://syzkaller.appspot.com/bug?extid=f91dbbabacae745d334f
compiler: gcc (GCC) 10.1.0-syz 20200507

Unfortunately, I don't have any reproducer for this issue yet.

IMPORTANT: if you fix the issue, please add the following tag to the commit:
Reported-by: syzbot+f91dbb...@syzkaller.appspotmail.com

block nbd5: Receive control failed (result -107)
==================================================================
BUG: KASAN: use-after-free in recv_work+0x2a2/0x2c0 drivers/block/nbd.c:787
Read of size 8 at addr ffff88801181ed20 by task kworker/u5:2/8465

CPU: 1 PID: 8465 Comm: kworker/u5:2 Not tainted 5.11.0-rc5-next-20210129-syzkaller #0
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011
Workqueue: knbd5-recv recv_work
Call Trace:
__dump_stack lib/dump_stack.c:79 [inline]
dump_stack+0x107/0x163 lib/dump_stack.c:120
print_address_description.constprop.0.cold+0x5b/0x2f8 mm/kasan/report.c:232
__kasan_report mm/kasan/report.c:399 [inline]
kasan_report.cold+0x7c/0xd8 mm/kasan/report.c:416
recv_work+0x2a2/0x2c0 drivers/block/nbd.c:787
process_one_work+0x98d/0x15f0 kernel/workqueue.c:2275
worker_thread+0x64c/0x1120 kernel/workqueue.c:2421
kthread+0x3b1/0x4a0 kernel/kthread.c:292
ret_from_fork+0x1f/0x30 arch/x86/entry/entry_64.S:296

Allocated by task 17044:
kasan_save_stack+0x1b/0x40 mm/kasan/common.c:38
kasan_set_track mm/kasan/common.c:46 [inline]
set_alloc_info mm/kasan/common.c:403 [inline]
____kasan_kmalloc mm/kasan/common.c:434 [inline]
____kasan_kmalloc.constprop.0+0xa0/0xd0 mm/kasan/common.c:406
kasan_slab_alloc include/linux/kasan.h:208 [inline]
slab_post_alloc_hook mm/slab.h:516 [inline]
slab_alloc_node mm/slub.c:2907 [inline]
slab_alloc mm/slub.c:2915 [inline]
__kmalloc_track_caller+0x16c/0x2e0 mm/slub.c:4550
__do_krealloc mm/slab_common.c:1149 [inline]
krealloc+0x60/0xa0 mm/slab_common.c:1178
nbd_add_socket+0x263/0x6a0 drivers/block/nbd.c:1044
__nbd_ioctl drivers/block/nbd.c:1369 [inline]
nbd_ioctl+0x388/0xa50 drivers/block/nbd.c:1425
blkdev_ioctl+0x2a1/0x6d0 block/ioctl.c:576
block_ioctl+0xf9/0x140 fs/block_dev.c:1650
vfs_ioctl fs/ioctl.c:48 [inline]
__do_sys_ioctl fs/ioctl.c:753 [inline]
__se_sys_ioctl fs/ioctl.c:739 [inline]
__x64_sys_ioctl+0x193/0x200 fs/ioctl.c:739
do_syscall_64+0x2d/0x70 arch/x86/entry/common.c:46
entry_SYSCALL_64_after_hwframe+0x44/0xa9

Freed by task 17044:
kasan_save_stack+0x1b/0x40 mm/kasan/common.c:38
kasan_set_track+0x1c/0x30 mm/kasan/common.c:46
kasan_set_free_info+0x20/0x30 mm/kasan/generic.c:357
____kasan_slab_free.part.0+0xe1/0x110 mm/kasan/common.c:364
kasan_slab_free include/linux/kasan.h:191 [inline]
slab_free_hook mm/slub.c:1562 [inline]
slab_free_freelist_hook+0x82/0x1d0 mm/slub.c:1600
slab_free mm/slub.c:3161 [inline]
kfree+0xe5/0x7b0 mm/slub.c:4202
krealloc+0x45/0xa0 mm/slab_common.c:1180
nbd_add_socket+0x263/0x6a0 drivers/block/nbd.c:1044
__nbd_ioctl drivers/block/nbd.c:1369 [inline]
nbd_ioctl+0x388/0xa50 drivers/block/nbd.c:1425
blkdev_ioctl+0x2a1/0x6d0 block/ioctl.c:576
block_ioctl+0xf9/0x140 fs/block_dev.c:1650
vfs_ioctl fs/ioctl.c:48 [inline]
__do_sys_ioctl fs/ioctl.c:753 [inline]
__se_sys_ioctl fs/ioctl.c:739 [inline]
__x64_sys_ioctl+0x193/0x200 fs/ioctl.c:739
do_syscall_64+0x2d/0x70 arch/x86/entry/common.c:46
entry_SYSCALL_64_after_hwframe+0x44/0xa9

The buggy address belongs to the object at ffff88801181ed20
which belongs to the cache kmalloc-8 of size 8
The buggy address is located 0 bytes inside of
8-byte region [ffff88801181ed20, ffff88801181ed28)
The buggy address belongs to the page:
page:00000000e0a0f4b5 refcount:1 mapcount:0 mapping:0000000000000000 index:0x0 pfn:0x1181e
flags: 0xfff00000000200(slab)
raw: 00fff00000000200 ffffea000053d180 0000000500000005 ffff88800fc41280
raw: ffff88801181ef28 0000000080660062 00000001ffffffff 0000000000000000
page dumped because: kasan: bad access detected

Memory state around the buggy address:
ffff88801181ec00: fc 00 fc fc fc fc fa fc fc fc fc 00 fc fc fc fc
ffff88801181ec80: fa fc fc fc fc 00 fc fc fc fc 00 fc fc fc fc 00
>ffff88801181ed00: fc fc fc fc fa fc fc fc fc 00 fc fc fc fc 00 fc
^
ffff88801181ed80: fc fc fc 00 fc fc fc fc 00 fc fc fc fc fa fc fc
ffff88801181ee00: fc fc 00 fc fc fc fc fa fc fc fc fc 00 fc fc fc
==================================================================


---
This report is generated by a bot. It may contain errors.
See https://goo.gl/tpsmEJ for more information about syzbot.
syzbot engineers can be reached at syzk...@googlegroups.com.

syzbot will keep track of this issue. See:
https://goo.gl/tpsmEJ#status for how to communicate with syzbot.

[Hillf Danton]

Tue, 02 Feb 2021 02:28:20 -0800

> syzbot found the following issue on:
>
> HEAD commit: b01f250d Add linux-next specific files for 20210129
> git tree: linux-next
> console output: https://syzkaller.appspot.com/x/log.txt?x=14366378d00000
> kernel config: https://syzkaller.appspot.com/x/.config?x=725bc96dc234fda7
> dashboard link: https://syzkaller.appspot.com/bug?extid=f91dbbabacae745d334f
> compiler: gcc (GCC) 10.1.0-syz 20200507
>
> Unfortunately, I don't have any reproducer for this issue yet.
>
> IMPORTANT: if you fix the issue, please add the following tag to the commit:
> Reported-by: syzbot+f91dbb...@syzkaller.appspotmail.com
>
> block nbd5: Receive control failed (result -107)
> ==================================================================
> BUG: KASAN: use-after-free in recv_work+0x2a2/0x2c0 drivers/block/nbd.c:787
> Read of size 8 at addr ffff88801181ed20 by task kworker/u5:2/8465
>
> CPU: 1 PID: 8465 Comm: kworker/u5:2 Not tainted 5.11.0-rc5-next-20210129-syzkaller #0
> Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011
> Workqueue: knbd5-recv recv_work
> Call Trace:
> __dump_stack lib/dump_stack.c:79 [inline]
> dump_stack+0x107/0x163 lib/dump_stack.c:120
> print_address_description.constprop.0.cold+0x5b/0x2f8 mm/kasan/report.c:232
> __kasan_report mm/kasan/report.c:399 [inline]
> kasan_report.cold+0x7c/0xd8 mm/kasan/report.c:416
> recv_work+0x2a2/0x2c0 drivers/block/nbd.c:787

The uaf occurs in the worker context.

Change memory slice without parking worker.

What is reported here is different from the uaf fixed in b98e762e3d71,
and it is highly appreciated to add a Fixes tag.

Open code the park, parkme and unpark mechanism to serialise the ioctl
act and the kworker, based on two atomic counters, to fix the UAF.

Plus s/recv_work/nbd_recv_work/ to make it simpler to spot a Fixes tag.


--- a/drivers/block/nbd.c
+++ b/drivers/block/nbd.c
@@ -113,6 +113,7 @@ struct nbd_device {
struct mutex config_lock;
struct gendisk *disk;
struct workqueue_struct *recv_workq;
+ atomic_t nbd_workers, nbd_recfg;

struct list_head list;
struct task_struct *task_recv;
@@ -771,7 +772,7 @@ out:
return ret ? ERR_PTR(ret) : cmd;
}

-static void recv_work(struct work_struct *work)
+static void nbd_recv_work(struct work_struct *work)
{
struct recv_thread_args *args = container_of(work,
struct recv_thread_args,
@@ -780,8 +781,19 @@ static void recv_work(struct work_struct
struct nbd_config *config = nbd->config;
struct nbd_cmd *cmd;
struct request *rq;
+ int cut = 0;

+ atomic_inc(&nbd->nbd_workers);
while (1) {
+ if (atomic_read(&nbd->nbd_recfg)) {
+ if (!cut) {
+ cut = 1;
+ atomic_dec(&nbd->nbd_recfg);
+ }
+ schedule_timeout_interruptible(1);
+ continue;
+ }
+
cmd = nbd_read_stat(nbd, args->index);
if (IS_ERR(cmd)) {
struct nbd_sock *nsock = config->socks[args->index];
@@ -796,6 +808,7 @@ static void recv_work(struct work_struct
if (likely(!blk_should_fake_timeout(rq->q)))
blk_mq_complete_request(rq);
}
+ atomic_dec(&nbd->nbd_workers);
nbd_config_put(nbd);
atomic_dec(&config->recv_threads);
wake_up(&config->recv_wq);
@@ -1016,6 +1029,7 @@ static int nbd_add_socket(struct nbd_dev
struct socket *sock;
struct nbd_sock **socks;
struct nbd_sock *nsock;
+ int workers, acks;
int err;

sock = nbd_get_socket(nbd, arg, &err);
@@ -1047,6 +1061,18 @@ static int nbd_add_socket(struct nbd_dev
goto put_socket;
}

+ do {
+ workers = atomic_read(&nbd->nbd_workers);
+ acks = 1 + workers;
+ atomic_set(&nbd->nbd_recfg, acks);
+ } while (workers != atomic_read(&nbd->nbd_workers));
+ do {
+ workers = atomic_read(&nbd->nbd_workers);
+ if (workers + 1 == acks - atomic_read(&nbd->nbd_recfg))
+ break;
+ schedule_timeout_interruptible(1);
+ } while (1);
+
socks = krealloc(config->socks, (config->num_connections + 1) *
sizeof(struct nbd_sock *), GFP_KERNEL);
if (!socks) {
@@ -1066,6 +1092,7 @@ static int nbd_add_socket(struct nbd_dev
nsock->cookie = 0;
socks[config->num_connections++] = nsock;
atomic_inc(&config->live_connections);
+ atomic_set(&nbd->nbd_recfg, 0);
blk_mq_unfreeze_queue(nbd->disk->queue);

return 0;
@@ -1114,7 +1141,7 @@ static int nbd_reconnect_socket(struct n
nsock->fallback_index = -1;
nsock->sock = sock;
nsock->dead = false;
- INIT_WORK(&args->work, recv_work);
+ INIT_WORK(&args->work, nbd_recv_work);
args->index = i;
args->nbd = nbd;
nsock->cookie++;
@@ -1123,7 +1150,7 @@ static int nbd_reconnect_socket(struct n

clear_bit(NBD_RT_DISCONNECTED, &config->runtime_flags);

- /* We take the tx_mutex in an error path in the recv_work, so we
+ /* We take the tx_mutex in an error path in the nbd_recv_work, so we
* need to queue_work outside of the tx_mutex.
*/
queue_work(nbd->recv_workq, &args->work);
@@ -1267,6 +1294,8 @@ static int nbd_start_device(struct nbd_d
dev_err(disk_to_dev(nbd->disk), "Could not allocate knbd recv work queue.\n");
return -ENOMEM;
}
+ atomic_set(&nbd->nbd_workers, 0);
+ atomic_set(&nbd->nbd_recfg, 0);

blk_mq_update_nr_hw_queues(&nbd->tag_set, config->num_connections);
nbd->task_recv = current;
@@ -1305,7 +1334,7 @@ static int nbd_start_device(struct nbd_d
nbd->tag_set.timeout;
atomic_inc(&config->recv_threads);
refcount_inc(&nbd->config_refs);
- INIT_WORK(&args->work, recv_work);
+ INIT_WORK(&args->work, nbd_recv_work);
args->nbd = nbd;
args->index = i;
queue_work(nbd->recv_workq, &args->work);

[Hillf Danton]

Tue, 02 Feb 2021 02:28:20 -0800

> syzbot found the following issue on:
>
> HEAD commit: b01f250d Add linux-next specific files for 20210129
> git tree: linux-next
> console output: https://syzkaller.appspot.com/x/log.txt?x=14366378d00000
> kernel config: https://syzkaller.appspot.com/x/.config?x=725bc96dc234fda7
> dashboard link: https://syzkaller.appspot.com/bug?extid=f91dbbabacae745d334f
> compiler: gcc (GCC) 10.1.0-syz 20200507
>
> Unfortunately, I don't have any reproducer for this issue yet.
>
> IMPORTANT: if you fix the issue, please add the following tag to the commit:
> Reported-by: syzbot+f91dbb...@syzkaller.appspotmail.com
>
> block nbd5: Receive control failed (result -107)
> ==================================================================
> BUG: KASAN: use-after-free in recv_work+0x2a2/0x2c0 drivers/block/nbd.c:787
> Read of size 8 at addr ffff88801181ed20 by task kworker/u5:2/8465
>
> CPU: 1 PID: 8465 Comm: kworker/u5:2 Not tainted 5.11.0-rc5-next-20210129-syzkaller #0
> Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011
> Workqueue: knbd5-recv recv_work
> Call Trace:
> __dump_stack lib/dump_stack.c:79 [inline]
> dump_stack+0x107/0x163 lib/dump_stack.c:120
> print_address_description.constprop.0.cold+0x5b/0x2f8 mm/kasan/report.c:232
> __kasan_report mm/kasan/report.c:399 [inline]
> kasan_report.cold+0x7c/0xd8 mm/kasan/report.c:416
> recv_work+0x2a2/0x2c0 drivers/block/nbd.c:787

The uaf occurs in the worker context.

Change memory slice without parking worker.

What is reported here is different from the uaf fixed in b98e762e3d71,
and it is highly appreciated to add a Fixes tag.

Add a sem for nbd to serialise dereferencing sock in the ioctl and
kworker contexts in order to fix the UAF. This is the fix as simple as
thought of while turning the config_lock into a sem seems another good
option.

Plus s/recv_work/nbd_recv_work/ to make it simpler to spot a Fixes tag.

--- a/drivers/block/nbd.c
+++ b/drivers/block/nbd.c

@@ -36,6 +36,7 @@
#include <linux/types.h>
#include <linux/debugfs.h>
#include <linux/blk-mq.h>
+#include <linux/rwsem.h>

#include <linux/uaccess.h>
#include <asm/types.h>
@@ -113,6 +114,7 @@ struct nbd_device {

struct mutex config_lock;
struct gendisk *disk;
struct workqueue_struct *recv_workq;

+ struct rw_semaphore sk_realloc_sem;

struct list_head list;
struct task_struct *task_recv;

@@ -771,7 +773,7 @@ out:

return ret ? ERR_PTR(ret) : cmd;
}

-static void recv_work(struct work_struct *work)
+static void nbd_recv_work(struct work_struct *work)
{
struct recv_thread_args *args = container_of(work,
struct recv_thread_args,

@@ -784,11 +786,16 @@ static void recv_work(struct work_struct
while (1) {

cmd = nbd_read_stat(nbd, args->index);
if (IS_ERR(cmd)) {

- struct nbd_sock *nsock = config->socks[args->index];
+ struct nbd_sock *nsock;
+
+ down_read(&nbd->sk_realloc_sem);
+ nsock = config->socks[args->index];

mutex_lock(&nsock->tx_lock);
nbd_mark_nsock_dead(nbd, nsock, 1);
mutex_unlock(&nsock->tx_lock);
+
+ up_read(&nbd->sk_realloc_sem);
break;
}

@@ -1046,10 +1053,12 @@ static int nbd_add_socket(struct nbd_dev
err = -ENOMEM;
goto put_socket;
}
+ down_write(&nbd->sk_realloc_sem);

socks = krealloc(config->socks, (config->num_connections + 1) *
sizeof(struct nbd_sock *), GFP_KERNEL);
if (!socks) {

+ up_write(&nbd->sk_realloc_sem);
kfree(nsock);
err = -ENOMEM;
goto put_socket;
@@ -1066,6 +1075,7 @@ static int nbd_add_socket(struct nbd_dev

nsock->cookie = 0;
socks[config->num_connections++] = nsock;
atomic_inc(&config->live_connections);

+ up_write(&nbd->sk_realloc_sem);

blk_mq_unfreeze_queue(nbd->disk->queue);

return 0;

@@ -1114,7 +1124,7 @@ static int nbd_reconnect_socket(struct n

nsock->fallback_index = -1;
nsock->sock = sock;
nsock->dead = false;
- INIT_WORK(&args->work, recv_work);
+ INIT_WORK(&args->work, nbd_recv_work);
args->index = i;
args->nbd = nbd;
nsock->cookie++;

@@ -1123,7 +1133,7 @@ static int nbd_reconnect_socket(struct n

clear_bit(NBD_RT_DISCONNECTED, &config->runtime_flags);

- /* We take the tx_mutex in an error path in the recv_work, so we
+ /* We take the tx_mutex in an error path in the nbd_recv_work, so we
* need to queue_work outside of the tx_mutex.
*/
queue_work(nbd->recv_workq, &args->work);

@@ -1305,7 +1315,7 @@ static int nbd_start_device(struct nbd_d

nbd->tag_set.timeout;
atomic_inc(&config->recv_threads);
refcount_inc(&nbd->config_refs);
- INIT_WORK(&args->work, recv_work);
+ INIT_WORK(&args->work, nbd_recv_work);
args->nbd = nbd;
args->index = i;
queue_work(nbd->recv_workq, &args->work);

@@ -1728,6 +1738,7 @@ static int nbd_dev_add(int index)
disk->queue->limits.max_sectors = 256;

mutex_init(&nbd->config_lock);
+ init_rwsem(&nbd->sk_realloc_sem);
refcount_set(&nbd->config_refs, 0);
refcount_set(&nbd->refs, 1);
INIT_LIST_HEAD(&nbd->list);

[syzbot]

Auto-closing this bug as obsolete.
Crashes did not happen for a while, no reproducer and no activity.