Re: [PATCH 0/6] ocfs2: fix stale xattr entry pointers and empty header reads

[Andrew Morton]

On Sat, 24 Oct 2026 23:10:06 +0800 Joseph Qi <joseph.qi@xxxxxxxxxxxxxxxxx> wrote:

> Five defects in the ocfs2 extended attribute code, plus one robustness
> measure that belongs with them, all on the paths that move entries
> between a block, a bucket and an index tree.  They fall into two groups.
> 

Thanks.

The series is a little awkward - [1/6] is "no functional change" hence
it's upstream-in-7.5-rc1 material.

The other 5 are cc:stable fixes, so they're upstream-in-7.4-rcX material.

That means splitting the series and it means upstreaming an untested
code combination - 2-5 haven't been tested without the presence of
[1/6].

I think I'll just cheat a bit.  I'll keep all 6 together and I'll put
them in mm-nonmm-unstable for now.  After Linus releases 7.4-rc1 I'll
move all 6 into mm-nonmm-hotfixes-unstable for upstreaming during the
7.4-rcX cycle.

I'll add 

Fixes: e39baff18b85 ("ocfs2: use the stored hash when checking xattr bucket collision")

to [1/6] just to make things look better and to reduce the risk
associated with -stable running that untested code combination.