On Fri, Jan 17, 2025 at 06:41:07AM -0800, Breno Leitao wrote:
The perf_iterate_ctx() function performs RCU list traversal but currently lacks RCU read lock protection. This causes lockdep warnings when running perf probe with unshare(1) under CONFIG_PROVE_RCU_LIST=y:
WARNING: suspicious RCU usage kernel/events/core.c:8168 RCU-list traversed in non-reader section!!
Call Trace: lockdep_rcu_suspicious ? perf_event_addr_filters_apply perf_iterate_ctx perf_event_exec begin_new_exec ? load_elf_phdrs load_elf_binary ? lock_acquire ? find_held_lock ? bprm_execve bprm_execve do_execveat_common.isra.0 __x64_sys_execve do_syscall_64 entry_SYSCALL_64_after_hwframeThis protection was previously present but was removed in commit bd2756811766 ("perf: Rewrite core context handling"). Add back the necessary rcu_read_lock()/rcu_read_unlock() pair around perf_iterate_ctx() call in perf_event_exec().
Hurm, I think it got ripped out because we no longer need to refer that perf_event_ctxp[].
Anyway, please write it like so:
diff --git a/kernel/events/core.c b/kernel/events/core.c index 0f8c55990783..b77f95089d62 100644 --- a/kernel/events/core.c +++ b/kernel/events/core.c @@ -8320,7 +8320,8 @@ void perf_event_exec(void)
perf_event_enable_on_exec(ctx); perf_event_remove_on_exec(ctx); - perf_iterate_ctx(ctx, perf_event_addr_filters_exec, NULL, true); + scoped_guard(rcu) + perf_iterate_ctx(ctx, perf_event_addr_filters_exec, NULL, true);
perf_unpin_context(ctx); put_ctx(ctx);