CVE-2024-56780

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> quota: flush quota_release_work upon quota writeback<br /> <br /> One of the paths quota writeback is called from is:<br /> <br /> freeze_super()<br /> sync_filesystem()<br /> ext4_sync_fs()<br /> dquot_writeback_dquots()<br /> <br /> Since we currently don&amp;#39;t always flush the quota_release_work queue in<br /> this path, we can end up with the following race:<br /> <br /> 1. dquot are added to releasing_dquots list during regular operations.<br /> 2. FS Freeze starts, however, this does not flush the quota_release_work queue.<br /> 3. Freeze completes.<br /> 4. Kernel eventually tries to flush the workqueue while FS is frozen which<br /> hits a WARN_ON since transaction gets started during frozen state:<br /> <br /> ext4_journal_check_start+0x28/0x110 [ext4] (unreliable)<br /> __ext4_journal_start_sb+0x64/0x1c0 [ext4]<br /> ext4_release_dquot+0x90/0x1d0 [ext4]<br /> quota_release_workfn+0x43c/0x4d0<br /> <br /> Which is the following line:<br /> <br /> WARN_ON(sb-&gt;s_writers.frozen == SB_FREEZE_COMPLETE);<br /> <br /> Which ultimately results in generic/390 failing due to dmesg<br /> noise. This was detected on powerpc machine 15 cores.<br /> <br /> To avoid this, make sure to flush the workqueue during<br /> dquot_writeback_dquots() so we dont have any pending workitems after<br /> freeze.