Re: [PATCH 1/8] signal: Make SIGKILL during coredumps an explicit special case

On Mon, 2022-01-17 at 10:09 -0600, Eric W. Biederman wrote:

Olivier Langlois [off-list ref] writes:
From my perspective I am not at all convinced that io_uring is the
only
culprit.

Beyond that the purpose of a coredump is to snapshot the process as
it
is, before anything is shutdown so that someone can examine the
coredump
and figure out what failed.  Running around changing the state of the
process has a very real chance of hiding what is going wrong.

Further your change requires that there be a place for io_uring to
clean
things up.  Given that fundamentally that seems like the wrong thing
to
me I am not interested in making it easy to what looks like the wrong
thing.

All of this may be perfection being the enemy of the good (especially
as
your io_uring magic happens as a special case in do_coredump).  My
work
in this area is to remove hacks so I can be convinced the code works
100% of the time so unfortunately I am not interested in pick up a
change that is only good enough.  Someone else like Andrew Morton
might
be.

Fair enough.

You do bring good points but I am not so sure about the second one
considering that the coredump is meant to be a snapshot and if io_uring
still runs, the state may change as the dump is generated anyway.

I'll follow with interest what you finally come up with but my mindset
when I wrote the patch was that there does not seem to be any benefit
keeping io_uring active while coredumping and it has the potential to
create nasty issues.

I did stumble into core file truncation problem.

Pavel got that when modifying io_uring code:
https://lore.kernel.org/all/1b519092-2ebf-3800-306d-c354c24a9ad1@gmail.com/ (local)

and I find very likely that keeping io_uring active while coredumping
might create new nasty but subtle issues down the road...

Greetings,
Olivier