libevent: fix race in the common case.

libevent: fix race in the common case.

There's a small race in the libevent code: it tests whether the monotonic clock
works and, if so, uses it and skips time-warp detection. However, the flag is a
static and thus causes a race.

We don't want to lose the ability to skip time-warp detection when the system
has a monotonic clock because we expect the vast majority of systems to have
it. But I'm not quite confident enough to remove support for systems without it
yet.

Thus this patch moves the race so that it only triggers on old systems that
don't have a monotonic clock. It's still there because we don't have easy
access to a "once" object this far down and we don't want to add a static init
function. But it will stop triggering the detectors and only arise on very old
systems.

BUG=360449
R=jln@chromium.org

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=262853

[jln (very slow on Chromium), 2014-04-09 20:13:08]

lgtm, but not sure why you inverted the polarity on use_monotonic.

For integers, no static initializer code will have to run to initialize to 1.

https://codereview.chromium.org/231493002/diff/1/third_party/libevent/event.c
File third_party/libevent/event.c (right):

https://codereview.chromium.org/231493002/diff/1/third_party/libevent/event.c...
third_party/libevent/event.c:110: static int no_monotonic_clock = 0;
Isn't it more clear to keep this as "use_monotonic" and statically initialize it
to 1?

"no__monotonic_clock == 0" is harder to read, as a double negation.

[agl, 2014-04-09 20:25:20]

https://codereview.chromium.org/231493002/diff/1/third_party/libevent/event.c
File third_party/libevent/event.c (right):

https://codereview.chromium.org/231493002/diff/1/third_party/libevent/event.c...
third_party/libevent/event.c:110: static int no_monotonic_clock = 0;
On 2014/04/09 20:13:09, jln wrote:
> Isn't it more clear to keep this as "use_monotonic" and statically initialize
it
> to 1?
> 
> "no__monotonic_clock == 0" is harder to read, as a double negation.

Yea, fair point. I was just aiming for it to be BSS, but you're right that it's
not worth the extra thought when reading the code.

Done.

[agl, 2014-04-09 20:25:44]

The CQ bit was checked by agl@chromium.org

[commit-bot: I haz the power, 2014-04-09 20:26:13]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/agl@chromium.org/231493002/20001

[commit-bot: I haz the power, 2014-04-09 21:06:00]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2014-04-09 21:06:00]

Retried try job too often on chromium_presubmit for step(s) presubmit
http://build.chromium.org/p/tryserver.chromium/buildstatus?builder=chromium_p...

[agl, 2014-04-09 23:21:10]

Committed patchset #2 manually as r262853.