Skip to contentSkip to conversation

dump_stack: avoid the livelock of the dump_lock

PATCH In review [email protected] 1 participant 1 message Nov 6, 2019

Subsystems (3)
lore
Share & tools
lore

[email protected]

author maintainer
#0 lore raw

PATCH 12/17 dump_stack: avoid the livelock of the dump_lock

From: Kevin Hao <[email protected]>
Subject: dump_stack: avoid the livelock of the dump_lock

In the current code, we use the atomic_cmpxchg() to serialize the output of the dump_stack(), but this implementation suffers the thundering herd problem. We have observed such kind of livelock on a Marvell cn96xx board(24 cpus) when heavily using the dump_stack() in a kprobe handler. Actually we can let the competitors to wait for the releasing of the lock before jumping to atomic_cmpxchg(). This will definitely mitigate the thundering herd problem. Thanks Linus for the suggestion.

[[email protected]: fix comment]
Fixes b58d977432c8 ("dump_stack: serialize the output from dump_stack()")

---

 lib/dump_stack.c |    7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)
lib/dump_stack.c renamed from lib/dump_stack.c~dump_stack-avoid-the-livelock-of-the-dump_lock +6 -1
@@ -106,7 +106,12 @@ retry:
 		was_locked = 1;
 	} else {
 		local_irq_restore(flags);
-		cpu_relax();
+		/*
+		 * Wait for the lock to release before jumping to
+		 * atomic_cmpxchg() in order to mitigate the thundering herd
+		 * problem.
+		 */
+		do { cpu_relax(); } while (atomic_read(&dump_lock) != -1);
 		goto retry;
 	}
 

_