Simplify implementation of Mutex.

Simplify implementation of Mutex.

Also moves Mutex to its own file mutex.{cc,h}.

R=svenpanne@chromium.org

Committed: https://code.google.com/p/v8/source/detail?r=15964

[Benedikt Meurer, 2013-07-30 13:50:43]

PTAL

[Sven Panne, 2013-07-30 15:09:56]

LGTM if the scope change mentioned below is OK.

https://codereview.chromium.org/21087012/diff/8001/src/isolate.cc
File src/isolate.cc (right):

https://codereview.chromium.org/21087012/diff/8001/src/isolate.cc#newcode373
src/isolate.cc:373: per_thread = AllocatePerIsolateThreadData(thread_id);
Why is it OK to move this outside the ScopedLock?

https://codereview.chromium.org/21087012/diff/8001/src/mutex.h
File src/mutex.h (right):

https://codereview.chromium.org/21087012/diff/8001/src/mutex.h#newcode106
src/mutex.h:106: class ScopedLock {
It would be cool if we could restrict the usage of Mutexes to this class, i.e.
without public Lock/Unlock methods, but I guess we are not really there yet...

[Benedikt Meurer, 2013-07-30 16:50:52]

https://codereview.chromium.org/21087012/diff/8001/src/isolate.cc
File src/isolate.cc (right):

https://codereview.chromium.org/21087012/diff/8001/src/isolate.cc#newcode373
src/isolate.cc:373: per_thread = AllocatePerIsolateThreadData(thread_id);
On 2013/07/30 15:09:56, Sven Panne wrote:
> Why is it OK to move this outside the ScopedLock?

Because AllocatePerIsolateThreadData() acquires the process_wide_mutex_ prior to
inserting the new data into the thread_data_table_.

https://codereview.chromium.org/21087012/diff/8001/src/mutex.h
File src/mutex.h (right):

https://codereview.chromium.org/21087012/diff/8001/src/mutex.h#newcode106
src/mutex.h:106: class ScopedLock {
On 2013/07/30 15:09:56, Sven Panne wrote:
> It would be cool if we could restrict the usage of Mutexes to this class, i.e.
> without public Lock/Unlock methods, but I guess we are not really there yet...

Not yet, unfortunately...

[Benedikt Meurer, 2013-07-30 17:13:09]

Committed patchset #5 manually as r15964 (presubmit successful).