Re: [PATCH 5.10.y] dm: fix NULL pointer dereference in __dm_suspend()

[Greg KH <gregkh@linuxfoundation.org>]

On Thu, Oct 23, 2025 at 03:07:28PM +0800, Li Lingfeng wrote:
> 
> 在 2025/10/23 13:20, Greg KH 写道:
> > On Thu, Oct 23, 2025 at 09:12:13AM +0800, Li Lingfeng wrote:
> > > 在 2025/10/22 17:20, Greg KH 写道:
> > > > On Wed, Oct 22, 2025 at 05:10:36PM +0800, Li Lingfeng wrote:
> > > > > 在 2025/10/22 16:45, Greg KH 写道:
> > > > > > On Wed, Oct 22, 2025 at 04:21:33PM +0800, Li Lingfeng wrote:
> > > > > > > Hi,
> > > > > > > 
> > > > > > > 在 2025/10/22 16:10, Greg KH 写道:
> > > > > > > > On Wed, Oct 22, 2025 at 04:01:04PM +0800, Li Lingfeng wrote:
> > > > > > > > > Hi,
> > > > > > > > > 
> > > > > > > > > 在 2025/10/14 11:03, Sasha Levin 写道:
> > > > > > > > > > From: Zheng Qixing <zhengqixing@huawei.com>
> > > > > > > > > > 
> > > > > > > > > > [ Upstream commit 8d33a030c566e1f105cd5bf27f37940b6367f3be ]
> > > > > > > > > > 
> > > > > > > > > > There is a race condition between dm device suspend and table load that
> > > > > > > > > > can lead to null pointer dereference. The issue occurs when suspend is
> > > > > > > > > > invoked before table load completes:
> > > > > > > > > > 
> > > > > > > > > > BUG: kernel NULL pointer dereference, address: 0000000000000054
> > > > > > > > > > Oops: 0000 [#1] PREEMPT SMP PTI
> > > > > > > > > > CPU: 6 PID: 6798 Comm: dmsetup Not tainted 6.6.0-g7e52f5f0ca9b #62
> > > > > > > > > > Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.16.1-2.fc37 04/01/2014
> > > > > > > > > > RIP: 0010:blk_mq_wait_quiesce_done+0x0/0x50
> > > > > > > > > > Call Trace:
> > > > > > > > > >        <TASK>
> > > > > > > > > >        blk_mq_quiesce_queue+0x2c/0x50
> > > > > > > > > >        dm_stop_queue+0xd/0x20
> > > > > > > > > >        __dm_suspend+0x130/0x330
> > > > > > > > > >        dm_suspend+0x11a/0x180
> > > > > > > > > >        dev_suspend+0x27e/0x560
> > > > > > > > > >        ctl_ioctl+0x4cf/0x850
> > > > > > > > > >        dm_ctl_ioctl+0xd/0x20
> > > > > > > > > >        vfs_ioctl+0x1d/0x50
> > > > > > > > > >        __se_sys_ioctl+0x9b/0xc0
> > > > > > > > > >        __x64_sys_ioctl+0x19/0x30
> > > > > > > > > >        x64_sys_call+0x2c4a/0x4620
> > > > > > > > > >        do_syscall_64+0x9e/0x1b0
> > > > > > > > > > 
> > > > > > > > > > The issue can be triggered as below:
> > > > > > > > > > 
> > > > > > > > > > T1 						T2
> > > > > > > > > > dm_suspend					table_load
> > > > > > > > > > __dm_suspend					dm_setup_md_queue
> > > > > > > > > > 						dm_mq_init_request_queue
> > > > > > > > > > 						blk_mq_init_allocated_queue
> > > > > > > > > > 						=> q->mq_ops = set->ops; (1)
> > > > > > > > > > dm_stop_queue / dm_wait_for_completion
> > > > > > > > > > => q->tag_set NULL pointer!	(2)
> > > > > > > > > > 						=> q->tag_set = set; (3)
> > > > > > > > > > 
> > > > > > > > > > Fix this by checking if a valid table (map) exists before performing
> > > > > > > > > > request-based suspend and waiting for target I/O. When map is NULL,
> > > > > > > > > > skip these table-dependent suspend steps.
> > > > > > > > > > 
> > > > > > > > > > Even when map is NULL, no I/O can reach any target because there is
> > > > > > > > > > no table loaded; I/O submitted in this state will fail early in the
> > > > > > > > > > DM layer. Skipping the table-dependent suspend logic in this case
> > > > > > > > > > is safe and avoids NULL pointer dereferences.
> > > > > > > > > > 
> > > > > > > > > > Fixes: c4576aed8d85 ("dm: fix request-based dm's use of dm_wait_for_completion")
> > > > > > > > > > Cc: stable@vger.kernel.org
> > > > > > > > > > Signed-off-by: Zheng Qixing <zhengqixing@huawei.com>
> > > > > > > > > > Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
> > > > > > > > > > [ omitted DMF_QUEUE_STOPPED flag setting and braces absent in 5.15 ]
> > > > > > > > > > Signed-off-by: Sasha Levin <sashal@kernel.org>
> > > > > > > > > > ---
> > > > > > > > > >       drivers/md/dm.c | 7 ++++---
> > > > > > > > > >       1 file changed, 4 insertions(+), 3 deletions(-)
> > > > > > > > > > 
> > > > > > > > > > diff --git a/drivers/md/dm.c b/drivers/md/dm.c
> > > > > > > > > > index 0868358a7a8d2..b51281ce15d4c 100644
> > > > > > > > > > --- a/drivers/md/dm.c
> > > > > > > > > > +++ b/drivers/md/dm.c
> > > > > > > > > > @@ -2457,7 +2457,7 @@ static int __dm_suspend(struct mapped_device *md, struct dm_table *map,
> > > > > > > > > >       {
> > > > > > > > > >       	bool do_lockfs = suspend_flags & DM_SUSPEND_LOCKFS_FLAG;
> > > > > > > > > >       	bool noflush = suspend_flags & DM_SUSPEND_NOFLUSH_FLAG;
> > > > > > > > > > -	int r;
> > > > > > > > > > +	int r = 0;
> > > > > > > > > >       	lockdep_assert_held(&md->suspend_lock);
> > > > > > > > > > @@ -2509,7 +2509,7 @@ static int __dm_suspend(struct mapped_device *md, struct dm_table *map,
> > > > > > > > > >       	 * Stop md->queue before flushing md->wq in case request-based
> > > > > > > > > >       	 * dm defers requests to md->wq from md->queue.
> > > > > > > > > >       	 */
> > > > > > > > > > -	if (dm_request_based(md))
> > > > > > > > > > +	if (map && dm_request_based(md))
> > > > > > > > > >       		dm_stop_queue(md->queue);
> > > > > > > > > It seems that before commit 80bd4a7aab4c ("blk-mq: move the srcu_struct
> > > > > > > > > used for quiescing to the tagset") was merged, blk_mq_wait_quiesce_done()
> > > > > > > > > would not attempt to access q->tag_set, so in dm_stop_queue() this kind
> > > > > > > > > of race condition would not cause a null pointer dereference. The change
> > > > > > > > > may not be necessary for 5.10, but adding it doesn’t appear to cause any
> > > > > > > > > issues either.
> > > > > > > > > >       	flush_workqueue(md->wq);
> > > > > > > > > > @@ -2519,7 +2519,8 @@ static int __dm_suspend(struct mapped_device *md, struct dm_table *map,
> > > > > > > > > >       	 * We call dm_wait_for_completion to wait for all existing requests
> > > > > > > > > >       	 * to finish.
> > > > > > > > > >       	 */
> > > > > > > > > > -	r = dm_wait_for_completion(md, task_state);
> > > > > > > > > > +	if (map)
> > > > > > > > > > +		r = dm_wait_for_completion(md, task_state);
> > > > > > > > > The fix tag c4576aed8d85 ("dm: fix request-based dm's use of
> > > > > > > > > dm_wait_for_completion") seems to correspond to the modification of
> > > > > > > > > dm_wait_for_completion().
> > > > > > > > > >       	if (!r)
> > > > > > > > > >       		set_bit(dmf_suspended_flag, &md->flags);
> > > > > > > > > Perhaps adding another fix tag would be more appropriate?
> > > > > > > > Those tags come directly from the original commit.
> > > > > > > Thanks for the clarification. Since the patch has already been merged into
> > > > > > > the mainline tree, we can't update the commit there. I was just wondering
> > > > > > > if it would be acceptable to add an additional “Fixes” tag when applying
> > > > > > > it to the stable branch, before merging.
> > > > > > I can, if needed, can you please provide that line?
> > > > > Fixes: 80bd4a7aab4c ("blk-mq: move the srcu_struct used for quiescing to the
> > > > > tagset")
> > > > This is odd, that is a 6.2 commit, yet this is being asked to be added
> > > > to 5.10.y.  So what help is this going to provide?
> > > > 
> > > > confused,
> > > > 
> > > > greg k-h
> > > Apologies, I didn't notice the version of this patch. Indeed, this patch
> > > was not included in 5.10, and 5.10 is not affected by this issue.
> > The original Fixes: tag points to a commit from 5.0, which is why this
> > was queued up for here.
> > 
> > > Perhaps we could add the new fix tag to the stable branch patches
> > > starting from 6.2?
> > I do not understand, sorry.
> > 
> > Let's start over, should this patch be included in the 5.10.y tree?  The
> > 5.4.y tree?  If not, why not?
> > 
> > And as this is already in the 5.15.y and 6.1.y releases, should it be
> > reverted from there?
> > 
> > thanks,
> > 
> > greg k-h
> The original patch made two changes:
> 1.
> @@ -2960,7 +2960,7 @@ static int __dm_suspend(struct mapped_device *md,
> struct dm_table *map,
>          * Stop md->queue before flushing md->wq in case request-based
>          * dm defers requests to md->wq from md->queue.
>          */
> -       if (dm_request_based(md)) {
> +       if (map && dm_request_based(md)) {
>                 dm_stop_queue(md->queue);
>                 set_bit(DMF_QUEUE_STOPPED, &md->flags);
>         }
> After commit 80bd4a7aab4c ("blk-mq: move the srcu_struct used for quiescing
> to the tagset") was merged in v6.2, dm_request_based() started to trigger
> access to tag_set, so a null check on map became necessary to avoid a
> potential null pointer dereference.
> The original patch merged into mainline was missing this Fixes tag.
> 
> 2.
> @@ -2972,7 +2972,8 @@ static int __dm_suspend(struct mapped_device *md,
> struct dm_table *map,
>          * We call dm_wait_for_completion to wait for all existing requests
>          * to finish.
>          */
> -       r = dm_wait_for_completion(md, task_state);
> +       if (map)
> +               r = dm_wait_for_completion(md, task_state);
>         if (!r)
>                 set_bit(dmf_suspended_flag, &md->flags);
> After commit c4576aed8d85 ("dm: fix request-based dm's use of
> dm_wait_for_completion")
> was merged in v5.0, dm_wait_for_completion() started to access tag_set, so
> the additional null check on map was also needed to prevent a null pointer
> dereference.
> 
> In summary:
> For kernels v6.2 and later, there should be two Fixes tags:
> Fixes: 80bd4a7aab4c ("blk-mq: move the srcu_struct used for quiescing to the
> tagset")
> Fixes: c4576aed8d85 ("dm: fix request-based dm's use of
> dm_wait_for_completion")
> 
> For kernels v5.0 to v6.2, only the second Fixes tag is needed:
> Fixes: c4576aed8d85 ("dm: fix request-based dm's use of
> dm_wait_for_completion")
> 
> Although the first change isn't actually required for kernels between v5.0
> and v6.2, it doesn't introduce any new issues, so there's no need to
> revert it.
> 
> My suggestion is:
> For stable kernels v6.2 and later, if this patch hasn't been applied yet,
> it would be better to include the extra Fixes tag for 80bd4a7aab4c when
> merging.

It has already been merged.

> If it has already been merged, there's no need to revert it, since the
> first change is harmless anyway.

Great, so there's nothing I need to do here, I like that :)

thanks for spelling it all out, makes more sense now.

greg k-h