Re: [PATCH 5/8] fstatat etc.: fix wrong machine code for calls

Paul Eggert <[email protected]> writes:

> On 2026-06-16 11:43, Bruno Haible wrote:
>> Paul Eggert wrote:
>>> ... assume that the calls’ arguments cannot be
>>> null pointers.  This assumption became false with the release of
>>> Linux kernel 6.11 (2024).
>> Wow. I'm impressed that you notice that. Is there a systematic way
>> for us to notice such changes?
>
> Only sort-of-systematic way I can think of offhand is to have LLMs
> review kernel and libc changes.
>
> I found the problem the hard way: some of my experimental code got
> miscompiled. Luckily I didn't need to spend much time diagnosing: the
> wrong machine code was a dead giveaway.

Agreed, I don't think the kernel has anything analogous to our NEWS
entries. There are ChangeLog's [1], but they are just the output of
'git diff'. Obviously, those are far too long to be useful.

Larger changes than this specific one may be picked up by lwn, phoronix,
etc. But that is mostly determined by whether the author is interested
in the topic and if they believe it will get clicks. That is how I found
out about an upcoming vmsplice change that I am a bit sad about [2] [3].

Collin

[1] https://cdn.kernel.org/pub/linux/kernel/v7.x/ChangeLog-7.1
[2] https://lwn.net/Articles/1075838/
[3] https://lists.gnu.org/archive/html/coreutils/2026-06/msg00024.html