linux-kernel - Re: [PATCH sched_ext/for-6.13-fixes] sched_ext: Fix invalid irq restore in scx_ops

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 for Android: free password hash cracker in your pocket

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

Message-ID: <Z2MV001RfiG7DNqj@slm.duckdns.org>
Date: Wed, 18 Dec 2024 08:34:59 -1000
From: Tejun Heo <tj@...nel.org>
To: Ihor Solodrai <ihor.solodrai@...me>
Cc: David Vernet <void@...ifault.com>, sched-ext@...a.com,
	kernel-team@...a.com, linux-kernel@...r.kernel.org,
	bpf <bpf@...r.kernel.org>
Subject: Re: [PATCH sched_ext/for-6.13-fixes] sched_ext: Fix invalid irq
 restore in scx_ops_bypass()

Hello,

On Tue, Dec 17, 2024 at 11:44:08PM +0000, Ihor Solodrai wrote:
> I re-enabled selftests/sched_ext on BPF CI today. The kernel on CI
> includes this patch. Sometimes there is a failure on attempt to attach
> a dsp_local_on scheduler.
> 
> Examples of failed jobs:
> 
>   * https://github.com/kernel-patches/bpf/actions/runs/12379720791/job/34555104994
>   * https://github.com/kernel-patches/bpf/actions/runs/12382862660/job/34564648924
>   * https://github.com/kernel-patches/bpf/actions/runs/12381361846/job/34560047798
> 
> Here is a piece of log that is present in failed run, but not in
> a successful run:
> 
> 2024-12-17T19:30:12.9010943Z [    5.285022] sched_ext: BPF scheduler "dsp_local_on" enabled
> 2024-12-17T19:30:13.9022892Z ERR: dsp_local_on.c:37
> 2024-12-17T19:30:13.9025841Z Expected skel->data->uei.kind == EXIT_KIND(SCX_EXIT_ERROR) (0 == 1024)
> 2024-12-17T19:30:13.9256108Z ERR: exit.c:30
> 2024-12-17T19:30:13.9256641Z Failed to attach scheduler
> 2024-12-17T19:30:13.9611443Z [    6.345087] smpboot: CPU 1 is now offline
> 
> Could you please investigate?

The test prog is wrong in assuming all possible CPUs to be consecutive and
online but I'm not sure whether that's what's making the test flaky. Do you
have dmesg from a failed run?

Thanks.

-- 
tejun