diff options
author | Alexei Starovoitov <ast@kernel.org> | 2024-04-24 19:00:23 +0300 |
---|---|---|
committer | Alexei Starovoitov <ast@kernel.org> | 2024-04-24 19:06:50 +0300 |
commit | dc92febf7b93da5049fe177804e6b1961fcc6bd7 (patch) | |
tree | cd49ada6055ba7c3933b03d82998efb303ec19a1 /kernel/bpf/cgroup.c | |
parent | 6e10b6350a67d398c795ac0b93a7bb7103633fe4 (diff) | |
download | linux-dc92febf7b93da5049fe177804e6b1961fcc6bd7.tar.xz |
bpf: Don't check for recursion in bpf_wq_work.
__bpf_prog_enter_sleepable_recur does recursion check which is not applicable
to wq callback. The callback function is part of bpf program and bpf prog might
be running on the same cpu. So recursion check would incorrectly prevent
callback from running. The code can call __bpf_prog_enter_sleepable(), but
run_ctx would be fake, hence use explicit rcu_read_lock_trace();
migrate_disable(); to address this problem. Another reason to open code is
__bpf_prog_enter* are not available in !JIT configs.
Reported-by: kernel test robot <lkp@intel.com>
Closes: https://lore.kernel.org/oe-kbuild-all/202404241719.IIGdpAku-lkp@intel.com/
Closes: https://lore.kernel.org/oe-kbuild-all/202404241811.FFV4Bku3-lkp@intel.com/
Fixes: eb48f6cd41a0 ("bpf: wq: add bpf_wq_init")
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Diffstat (limited to 'kernel/bpf/cgroup.c')
0 files changed, 0 insertions, 0 deletions