APPLIED: [SRU][J][PATCH 0/1] CVE-2023-52621
Roxana Nicolescu
roxana.nicolescu at canonical.com
Mon Oct 7 07:39:19 UTC 2024
On 24/09/2024 17:29, Massimiliano Pellizzer wrote:
> [Impact]
>
> bpf: Check rcu_read_lock_trace_held() before calling bpf map helpers
>
> The three bpf_map_{lookup,update,delete}_elem() helper functions are also
> available for sleepable bpf programs, the patch adds the corresponding lock
> assertion for sleepable bpf programs.
>
> [Fix]
>
> Noble: Fixed
> Jammy: Backported the fix commit from linux-6.1.y
> Focal: Not affected
> Bionic: Not affected
> Xenial: Not affected
>
> [Test Case]
>
> Compile tested.
>
> [Where problems could occur]
>
> The fix for CVE-2023-52621 affetcs the BPF subsystem, in particular the
> handling of BPF map operations in sleepable BPF programs. A bug in the
> patch could lead to kernel warnings or crashes when using BPF maps,
> especially when BPF JIT is disabled.
>
> Hou Tao (1):
> bpf: Check rcu_read_lock_trace_held() before calling bpf map helpers
>
> kernel/bpf/helpers.c | 14 +++++++++-----
> 1 file changed, 9 insertions(+), 5 deletions(-)
>
Applied to jammy:linux master-next branch. Thanks!
More information about the kernel-team
mailing list