Provide windows notification of thread termination

Provide windows notification of thread termination

[Relanding revert 114954 with fix to be compatible
wit new re-linking strategy.  Better care was needed
in bracketing the callbacks that were used, to be
sure the new linker left all items located together.]

This code should only have much of an impact on the XP
platform, where thread teardown notification are not
currently made.

Provide an automatic fallback scan of the linker
list of notification callbacks via DllMain.

This fallback is ONLY used if there are no notifications
coming via the linker list. This allows all existing
code to continue working as it does today, and provides
thread teardown notification on XP even though we use
the LoadLibrary call.

It is a minimal change, suitable for pushing to other
channels (beta? stable?).

r=siggi
bug=107974, 103209
Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=115622

[Sigurður Ásgeirsson, 2011-12-22 20:04:51]

nit: notifictaion in CL description should be notification.

otherwise LGTM

http://codereview.chromium.org/9024015/diff/2002/base/base.gyp
File base/base.gyp (right):

http://codereview.chromium.org/9024015/diff/2002/base/base.gyp#newcode130
base/base.gyp:130: 'win/dllmain.cc',
nit: this file is out of order here. Is it your intent to bake it into
base_unittests?

[jar (doing other things), 2011-12-22 22:04:41]

Changes made per comments by siggi (includin ga bunch of typos in the
description of this CL).

Note that the first rev of this CL was exactly what was reverted.

http://codereview.chromium.org/9024015/diff/2002/base/base.gyp
File base/base.gyp (right):

http://codereview.chromium.org/9024015/diff/2002/base/base.gyp#newcode130
base/base.gyp:130: 'win/dllmain.cc',
It was in the wrong place.... and I don't know how the original CL evolved to
this :-(

I changeed it to be alphabetized.

I think (if I recall correctly) this can't be built into the base library, but
can be built into the base test library.  I *think* the reason to list it here
is to ensure that it is built when we build the base unit tests (we don't do
much run-time testing, but at least it compiles with base).

Other than the compilation, I can't recall why it was valuable to have it
here.... so I could remove it... but it was meant to be a re-land of the
revert... and compiling it regularly (with base) seems like a good thing.

If you prefer (or argue), I'll remove it.


On 2011/12/22 20:04:51, Ruðrugis wrote:
> nit: this file is out of order here. Is it your intent to bake it into
> base_unittests?

[Sigurður Ásgeirsson, 2011-12-22 22:07:54]

LGTM