From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from smtp.kernel.org (aws-us-west-2-korg-mail-1.web.codeaurora.org [10.30.226.201]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 9BB703019A1; Thu, 23 Oct 2025 08:44:01 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=10.30.226.201 ARC-Seal:i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1761209043; cv=none; b=XVOg19qCB5GNuGITnByqf1zdXSTWmTR8yOmRO1139PffYcb5O2oBDSqeI5dDyDvMnDpmxgX5i3NBnS8ThnZ81CLEaVS4lOasVe7xnjtZNkqgkdowX7BE1wsS8PJ05/0eUJQu4qN4aLoAIpgEouVsSAA5k0fU1/qlfs0iOBX8k/c= ARC-Message-Signature:i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1761209043; c=relaxed/simple; bh=kTu+yxrCVTXSPMrsilY136J3iqwH9zrgDwVA1u8FKHQ=; h=Date:From:To:Cc:Subject:Message-ID:References:MIME-Version: Content-Type:Content-Disposition:In-Reply-To; b=JxJ6TKv2kijhA6PVN0piVuEN7UyDg8IJRARpxcpstFFgoQEYYSj/gr/9ewRZjgOBqterk6/ERRgJoGczCMZ+XxIKEYGUJsa1ULaSv1rq36jOkxVobFv3Vi7xxKc59HhUcEaaa/rte8/NLOlXQ10HEmAzWlv/UmadI0y29KUWF/Y= ARC-Authentication-Results:i=1; smtp.subspace.kernel.org; dkim=pass (1024-bit key) header.d=linuxfoundation.org header.i=@linuxfoundation.org header.b=qx91cgvg; arc=none smtp.client-ip=10.30.226.201 Authentication-Results: smtp.subspace.kernel.org; dkim=pass (1024-bit key) header.d=linuxfoundation.org header.i=@linuxfoundation.org header.b="qx91cgvg" Received: by smtp.kernel.org (Postfix) with ESMTPSA id 7CD00C4CEE7; Thu, 23 Oct 2025 08:44:00 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=linuxfoundation.org; s=korg; t=1761209041; bh=kTu+yxrCVTXSPMrsilY136J3iqwH9zrgDwVA1u8FKHQ=; h=Date:From:To:Cc:Subject:References:In-Reply-To:From; b=qx91cgvgPerasgn9C1RxqsAHGfm91ikzwr6SHlgRF5QYr97PA5q10bMNWjgoIvOXE b51gFCNXEZ+xpcDHvAQqLn1RzPvqinJ+/MMjgL3UTZJf/9nL5za8XcpJvZbR7QLLsn iA1gycyELtUE7SMyMhc2h7KetNjldMQTO2YQS0Wo= Date: Thu, 23 Oct 2025 10:43:58 +0200 From: Greg KH To: Li Lingfeng Cc: Sasha Levin , stable@vger.kernel.org, Zheng Qixing , Mikulas Patocka , dm-devel@lists.linux.dev, yangerkun , "zhangyi (F)" , Hou Tao Subject: Re: [PATCH 5.10.y] dm: fix NULL pointer dereference in __dm_suspend() Message-ID: <2025102323-fiction-subprime-ecec@gregkh> References: <20251014030334.3868139-1-sashal@kernel.org> <2025102252-abrasive-glamorous-465b@gregkh> <1767bec2-b660-42b2-b828-b221e0896eba@huawei.com> <2025102212-pouring-embellish-95fc@gregkh> <28a7f62d-ae81-4629-864c-a0c1ecf98ee9@huawei.com> <2025102231-outlying-lapel-6159@gregkh> <2025102310-blinked-handsaw-ade3@gregkh> <45a1c253-da13-4390-b2bb-7fdc56341edb@huawei.com> Precedence: bulk X-Mailing-List: stable@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Disposition: inline Content-Transfer-Encoding: 8bit In-Reply-To: <45a1c253-da13-4390-b2bb-7fdc56341edb@huawei.com> 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 > > > > > > > > > > > > > > > > > > > > [ 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: > > > > > > > > > > > > > > > > > > > > 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 > > > > > > > > > > Signed-off-by: Mikulas Patocka > > > > > > > > > > [ omitted DMF_QUEUE_STOPPED flag setting and braces absent in 5.15 ] > > > > > > > > > > Signed-off-by: Sasha Levin > > > > > > > > > > --- > > > > > > > > > > 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