Reland r65996. Disallows Singletons on non-joinable thread.

Reland r65996.  Disallows Singletons on non-joinable thread.

Test breakages caused by this change have been fixed here or in other changelists.

BUG=61753
TEST=none

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=66719

[willchan no longer on Chromium, 2010-11-16 23:36:01]

wtc: This address your comments.

[wtc, 2010-11-17 01:56:18]

LGTM.

http://codereview.chromium.org/5024003/diff/11001/base/platform_thread_win.cc
File base/platform_thread_win.cc (right):

http://codereview.chromium.org/5024003/diff/11001/base/platform_thread_win.cc...
base/platform_thread_win.cc:77: bool PlatformThread::Create(size_t stack_size,
Delegate* delegate,
You should document that passing out_thread_handle=NULL to
Create() to create a non-joinable thread is for internal
use only.  The public way to create a non-joinable thread is
the CreateNonJoinable method.

[willchan no longer on Chromium, 2010-11-17 19:03:26]

http://codereview.chromium.org/5024003/diff/11001/base/platform_thread_win.cc
File base/platform_thread_win.cc (right):

http://codereview.chromium.org/5024003/diff/11001/base/platform_thread_win.cc...
base/platform_thread_win.cc:77: bool PlatformThread::Create(size_t stack_size,
Delegate* delegate,
On 2010/11/17 01:56:18, wtc wrote:
> You should document that passing out_thread_handle=NULL to
> Create() to create a non-joinable thread is for internal
> use only.  The public way to create a non-joinable thread is
> the CreateNonJoinable method.

Thanks for pointing this out.  I guess I missed this from the previous review :(
 Sorry!  I don't want to change the API.  So I refactored the body into a
CreateThreadInternal() function and added a DCHECK to make sure thread_handle !=
NULL in the public API.  Please take another look.

[wtc, 2010-11-17 19:09:54]

LGTM.

[wtc, 2010-11-18 22:31:49]

willchan: the new change to chrome/browser/browser_thread.cc
in Patch Set 7 looks good to me.