linux-kernel - Re: [PATCH v3 1/1] PM / Freezer: Skip zombie/dead processes to

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20250704092144.GH2001818@noisy.programming.kicks-ass.net>
Date: Fri, 4 Jul 2025 11:21:44 +0200
From: Peter Zijlstra <peterz@...radead.org>
To: Zihuan Zhang <zhangzihuan@...inos.cn>
Cc: "Rafael J. Wysocki" <rafael@...nel.org>, pavel@...nel.org,
	len.brown@...el.com, linux-pm@...r.kernel.org,
	linux-kernel@...r.kernel.org, Oleg Nesterov <oleg@...hat.com>
Subject: Re: [PATCH v3 1/1] PM / Freezer: Skip zombie/dead processes to

On Fri, Jul 04, 2025 at 04:48:20PM +0800, Zihuan Zhang wrote:
> Hi Peter,
> 
> Thanks for the feedback.
> 
> ??? 2025/7/4 16:19, Peter Zijlstra ??????:
> > ????????? ??? ??? ?????? ??? ??? ??? ??? ??? ??? ??? ??? ??? ??????
> > Depending on where they wait (I can't seem to find in a hurry) it might
> > make sense to make that wait FREEZABLE anyway.
> > 
> > For example, AFAICT it wouldn't hurt, and might even help some, to make
> > kernel/exit.c:do_wait() TASK_FREEZABLE.
> > 
> > So where do ZOMBIEs sleep? Don't they simply pass through do_task_dead()
> > and never get scheduled again? Notably, do_task_dead() already marks the
> > tasks as PF_NOFREEZE.
> > 
> > Anyway, yes, the condition it adds is relatively simple, but I really
> > don't see why we should complicate things *at*all*.
> 
> You???re absolutely right ??? zombie processes won???t be frozen in
> practice, since PF_NOFREEZE is already set in do_task_dead(). However, if we
> don???t explicitly skip them early in try_to_freeze_task(), they still go
> through the freezer logic path, including calls like freeze_task() ???
> freezing() before eventually returning without freezing.
> 
> This not only introduces unnecessary code path traversal, but also involves
> locking (e.g., spin_lock_irqsave/restore()), which could be avoided
> altogether if we bail out earlier.
> 
> Additionally, skipping zombies directly helps reduce the list traversal
> overhead in freeze_processes(), especially on systems with a large number of
> tasks, where zombies can account for a non-trivial fraction.
> 
> So while the practical effect might be small, the gain is low-risk and helps
> streamline the freezer logic a bit more.

You're missing the obvious. How about we 'fix' the PF_NOFREEZE handling
and help all cases that set that and not only zombies?