[lvc-project] [PATCH 5.10 0/2] mm/truncate: fix issue in ext4_set_page_dirty()

[Fedor Pchelkin]

Hi Roman,

On 24/01/11 02:37PM, Roman Smirnov wrote:
> Syzkaller reports warning in ext4_set_page_dirty() in 5.10 stable 
> releases. The problem can be fixed by the following patches 
> which can be cleanly applied to the 5.10 branch.

Please make sure that the patches you backport to 5.10 also exist in all
fresher stables (5.15, 6.1, ...). Otherwise, it is a "kernel regression"
when e.g. someone upgrades from 5.10 to 5.15 and the bug appears again
because it wasn't backported to 5.15. The whole process is described here
at [1].

In case of not obvious patches and changes, please first send patches to
lvc-patches at linuxtesting.org before sending to lkml. It's always better
when someone else cross-reviews the patches and points at potential
mistypes, missing details, etc.

> 
> Found by Linux Verification Center (linuxtesting.org) with Syzkaller.
> 
> Link: https://syzkaller.appspot.com/bug?extid=02f21431b65c214aa1d6
> 
> Matthew Wilcox (Oracle) (2):
>   mm/truncate: Inline invalidate_complete_page() into its one caller
>   mm/truncate: Replace page_mapped() call in invalidate_inode_page()
> 

Also, as Matthew pointed out in lkml, in this case it's not obvious why
these mm/truncate patches fix the ext4 problem reported by Syzkaller.
These patches weren't originally intended to fix the ext4 bug (there is
no bug report or similar in the commit descriptions). So some
explanation is needed why they should fix that.

[1]: https://gitlab.linuxtesting.ru/lvc/guides/-/blob/drafts/lvc_kernel/lvc_patch_dev.md