linux-kernel - Re: possible deadlock in smp_call_function_many

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] [day] [month] [year] [list]

Message-ID: <87o6taq6bk.ffs@tglx>
Date: Thu, 24 Jul 2025 11:35:59 +0200
From: Thomas Gleixner <tglx@...utronix.de>
To: 胡焜 <huk23@...udan.edu.cn>
Cc: 白烁冉 <baishuoran@...eu.edu.cn>, Peter Zijlstra
 <peterz@...radead.org>,
 "jjtan24@...udan.edu.cn" <jjtan24@...udan.edu.cn>,
 linux-kernel@...r.kernel.org
Subject: Re: possible deadlock in smp_call_function_many_cond

On Thu, Jul 24 2025 at 15:55, 胡焜 wrote:

Please do not top-post and trim your replies:

https://people.kernel.org/tglx/notes-about-netiquette

> My guess is that the crash point might be csd_lock_wait(csd) in
> ~/kernel/smp.c on line 885. Here the other CPUs are notified via IPI
> to execute flush_tlb_mm_range and enter spin wait. However, it has not
> yet been determined which CPU performed the operation that caused the
> IPI to return a timeout.

Guessing is not a proper engineering method :)

> I'll first provide the log that was reproduced in 6.16-rc4. we'll re-verify it on rc7.

> [    0.000000][    T0] Linux version 6.16.0-rc4 (qjj@...-Standard-PC-Q35-ICH9-2009) (gcc (Ubuntu 9.4.0-1ubuntu1~20.04.2) 9.4.0, GNU ld (GNU Binutils for Ubuntu) 2.34) #1 SMP PREEMPT_DYNAMIC Wed Jul  2 09:53:03 CST 2025
>
> [    0.000000][    T0] Command line: root=/dev/sda console=ttyS0 ftrace_dump_on_oops hrtimer_expire_entry hrtimer_expire_exit trace_event=ipi_entry,ipi_exit,irq_handler_entry,irq_handler_exit

Please remove the silly extra new-lines next time.
>
> [    0.427302][    T0] Unknown kernel command line parameters "hrtimer_expire_entry hrtimer_expire_exit spec_store_bypass_disable=prctl nbds_max=32", will be passed to user space.

                                                                 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
How is this helpful?