Re: [BUG] general protection fault in cpuset_cpus_allowed

[Waiman Long]

On 8/28/26 1:43 AM, Farhad Alemi wrote:
> Hello,
>
> As part of the kernel research at ASU's SEFCOM
> lab, we hit the crash below. Crash reports can be found here:
>
>    https://github.com/farhad-alemi/public_bug_reports/tree/main/166-general-protection-fault-in-cpuset-cpus-allowed/
>
>    Oops: general protection fault, probably for non-canonical address
> 0xdffffc000000001d: 0000 [#1] SMP KASAN NOPTI
>    KASAN: null-ptr-deref in range [0x00000000000000e8-0x00000000000000ef]
>    RIP: 0010:bitmap_intersects include/linux/bitmap.h:440 [inline]
>    RIP: 0010:cpumask_intersects include/linux/cpumask.h:822 [inline]
>    RIP: 0010:guarantee_active_cpus kernel/cgroup/cpuset.c:481 [inline]
>    RIP: 0010:__cpuset_cpus_allowed_locked kernel/cgroup/cpuset.c:4022 [inline]
>    RIP: 0010:cpuset_cpus_allowed+0x14a/0x2f0 kernel/cgroup/cpuset.c:4071
>    Call Trace:
>     cpuset_cpus_allowed+0x14a/0x2f0 kernel/cgroup/cpuset.c:4071
>     __sched_setaffinity+0x98/0x3c0 kernel/sched/syscalls.c:1149
>     sched_setaffinity+0x341/0x4a0 kernel/sched/syscalls.c:1237
>     __x64_sys_sched_setaffinity+0x106/0x170 kernel/sched/syscalls.c:1262
>
> A second capture reaches the identical fault -- same address, same KASAN range,
> same RIP -- through cpuset_fork <- cgroup_post_fork <- copy_process <-
> kernel_clone <- clone3, with the crashing task being systemd at PID 1.
>
> Our reproducer.c is available upon request.
>
> Happy to test a patch if that would help.

This null pointer dereference should only happen if you are racing 
sched_setaffinity() with cpu hotplug operation. Right?

Anyway, I have sent out a patch to hopefully address this bug. Could you 
try out the patch to see if it fixes the problem?

Thanks,
Longman