Wait on the monotonic clock to guard against changes to the system clock.

Wait on the monotonic clock to guard against changes to the system clock.

Committed: https://code.google.com/p/dart/source/detail?r=1522

[cshapiro, 2011-11-12 03:31:52]

The existing thread test seems kind of weak.  I am wondering if there might be a
better way to test this change.

[siva, 2011-11-14 18:32:28]

LGTM

http://codereview.chromium.org/8463032/diff/1/runtime/vm/thread_linux.cc
File runtime/vm/thread_linux.cc (right):

http://codereview.chromium.org/8463032/diff/1/runtime/vm/thread_linux.cc#newc...
runtime/vm/thread_linux.cc:21: int64_t secs = millis / 1000;
Maybe we should have 3 constants to make this more readable:
static const int kMillisPerSecond = 1000;
static const int kNanosPerMillis = 1000 * 1000;
static const int kNanosPerSecond = kNanosPerMillis * kMiliisPerSecond;

http://codereview.chromium.org/8463032/diff/1/runtime/vm/thread_linux.cc#newc...
runtime/vm/thread_linux.cc:163: {
The scoping feels weird why not just name the variables different:
pthread_mutexattr_t mutex_attr;
pthread_condaddr_t cond_attr;

[cshapiro, 2011-11-14 22:24:25]

http://codereview.chromium.org/8463032/diff/1/runtime/vm/thread_linux.cc
File runtime/vm/thread_linux.cc (right):

http://codereview.chromium.org/8463032/diff/1/runtime/vm/thread_linux.cc#newc...
runtime/vm/thread_linux.cc:21: int64_t secs = millis / 1000;
Good idea.  I have added to globals.h time constants borrowed from Chrome's Time
class.

http://codereview.chromium.org/8463032/diff/1/runtime/vm/thread_linux.cc#newc...
runtime/vm/thread_linux.cc:163: {
Yeah, in fact, I wrote it that way originally.  Reverted.  Done.