On Fri, Jul 28, 2023 at 08:44 AM +02, tglozar@xxxxxxxxxx wrote:
> From: Tomas Glozar <tglozar@xxxxxxxxxx>
>
> Disabling preemption in sock_map_sk_acquire conflicts with GFP_ATOMIC
> allocation later in sk_psock_init_link on PREEMPT_RT kernels, since
> GFP_ATOMIC might sleep on RT (see bpf: Make BPF and PREEMPT_RT co-exist
> patchset notes for details).
>
> This causes calling bpf_map_update_elem on BPF_MAP_TYPE_SOCKMAP maps to
> BUG (sleeping function called from invalid context) on RT kernels.
>
> preempt_disable was introduced together with lock_sk and rcu_read_lock
> in commit 99ba2b5aba24e ("bpf: sockhash, disallow bpf_tcp_close and update
> in parallel"), probably to match disabled migration of BPF programs, and
> is no longer necessary.
>
> Remove preempt_disable to fix BUG in sock_map_update_common on RT.
>
> Signed-off-by: Tomas Glozar <tglozar@xxxxxxxxxx>
> ---
We disable softirq and hold a spin lock when modifying the map/hash in
sock_{map,hash}_update_common so this LGTM:
Reviewed-by: Jakub Sitnicki <jakub@xxxxxxxxxxxxxx>
You might want some extra tags:
Link: https://lore.kernel.org/all/20200224140131.461979697@xxxxxxxxxxxxx/
Fixes: 99ba2b5aba24 ("bpf: sockhash, disallow bpf_tcp_close and update in parallel")
