The function match_records() may take a while due to a large number of string comparisons, so when in PREEMPT_VOLUNTARY kernels we could face RCU stalls due to that. Add a cond_resched() to prevent that. Suggested-by: Steven Rostedt <rostedt@xxxxxxxxxxx> Acked-by: Paul E. McKenney <paulmck@xxxxxxxxxx> # from RCU CPU stall warning perspective Cc: Masami Hiramatsu <mhiramat@xxxxxxxxxx> Cc: Mark Rutland <mark.rutland@xxxxxxx> Signed-off-by: Guilherme G. Piccoli <gpiccoli@xxxxxxxxxx> --- Hi Steve / Paul, thanks for the discussions on the first thread [0], much appreciated! Here is the "official" version. Steve: lemme know if it's good for you, and in case you prefer to send it yourself (since you proposed it on IRC), fine by me! Paul: kept your ACK (thanks for that BTW) even though I changed the place of cond_resched() to align with Steve's preference. Lemme know in case you want to drop this ACK. Cheers, Guilherme [0] https://lore.kernel.org/lkml/1ef5fe19-a82f-835e-fda5-455e9c2b94b4@xxxxxxxxxx/ kernel/trace/ftrace.c | 1 + 1 file changed, 1 insertion(+) diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c index 7dc023641bf1..80639bdb85f6 100644 --- a/kernel/trace/ftrace.c +++ b/kernel/trace/ftrace.c @@ -4192,6 +4192,7 @@ match_records(struct ftrace_hash *hash, char *func, int len, char *mod) } found = 1; } + cond_resched(); } while_for_each_ftrace_rec(); out_unlock: mutex_unlock(&ftrace_lock); -- 2.38.0
