[PNaCl Toolchain] Use static libc++ from Chrome toolchain on Mac

[PNaCl Toolchain] Use static libc++ from Chrome toolchain on Mac

https://codereview.chromium.org/1572753002/ stopped building our custom libc++
but failed to add the linker path to the static libc++ library, resulting in
a dependence on the system libc++ dylib. This fails for OSX 10.6 systems, which
don't have the dylib.

BUG=None
R=mcgrathr@chromium.org

Committed: https://chromium.googlesource.com/native_client/src/native_client/+/2d5bba1b48c0d3091df391c81ed7cd3236c0a127

[Roland McGrath, 2016-01-14 19:41:49]

typo in log msg: s/doen't/don't/

The libc++-static binaries should get into the hashables list along with the
clang++ binary.

[Derek Schuff, 2016-01-14 19:50:39]

Description was changed from

==========
[PNaCl Toolchain] Use static libc++ from Chrome toolchain on Mac

https://codereview.chromium.org/1572753002/ stopped building our custom libc++
but failed to add the linker path to the static libc++ library, resulting in
a dependence on the system libc++ dylib. This fails for OSX 10.6 systems, which
doen't have the dylib.

BUG=None
R=mcgrathr@chromium.org
==========

to

==========
[PNaCl Toolchain] Use static libc++ from Chrome toolchain on Mac

https://codereview.chromium.org/1572753002/ stopped building our custom libc++
but failed to add the linker path to the static libc++ library, resulting in
a dependence on the system libc++ dylib. This fails for OSX 10.6 systems, which
don't have the dylib.

BUG=None
R=mcgrathr@chromium.org
==========

[Derek Schuff, 2016-01-14 19:50:55]

On 2016/01/14 19:41:49, Roland McGrath wrote:
> typo in log msg: s/doen't/don't/
> 
> The libc++-static binaries should get into the hashables list along with the
> clang++ binary.

Done and done.

[Roland McGrath, 2016-01-14 19:56:00]

lgtm

[Derek Schuff, 2016-01-16 00:26:39]

Patchset #3 (id:40001) has been deleted

[Mark Seaborn, 2016-01-16 00:30:31]

Derek, can you commit this so that we can unblock the DEPS roll?  (Thanks for
fixing this, by the way.)

[Mark Seaborn, 2016-01-19 21:59:58]

The CQ bit was checked by mseaborn@chromium.org

[commit-bot: I haz the power, 2016-01-19 22:00:10]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1586243003/20001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1586243003/20001

[Derek Schuff, 2016-01-19 22:32:41]

On 2016/01/16 00:30:31, Mark Seaborn wrote:
> Derek, can you commit this so that we can unblock the DEPS roll?  (Thanks for
> fixing this, by the way.)

Sorry for the late reply; the short answer is no, it doesn't quite work. See the
LLVM regression tests stage in
https://build.chromium.org/p/tryserver.nacl/builders/nacl-toolchain-mac-pnacl...
(which is unfortunately hidden behind a try job which was "successful" because
it pulled the cached build, and skipped the regression tests).

The reason is that when linked with the static libc++, there seems to be a
subtle disagreement around the std:error_code and/or llvm::errc enumerations.
(it's the cause of all the failures seen in the log, but there are different
manifestations in the unit tests vs the llvm-ar binary). The numbers seem to be
correct, but it may be related to the way the types are handled; I'm still
looking into it.

[commit-bot: I haz the power, 2016-01-19 22:41:55]

Description was changed from

==========
[PNaCl Toolchain] Use static libc++ from Chrome toolchain on Mac

https://codereview.chromium.org/1572753002/ stopped building our custom libc++
but failed to add the linker path to the static libc++ library, resulting in
a dependence on the system libc++ dylib. This fails for OSX 10.6 systems, which
don't have the dylib.

BUG=None
R=mcgrathr@chromium.org
==========

to

==========
[PNaCl Toolchain] Use static libc++ from Chrome toolchain on Mac

https://codereview.chromium.org/1572753002/ stopped building our custom libc++
but failed to add the linker path to the static libc++ library, resulting in
a dependence on the system libc++ dylib. This fails for OSX 10.6 systems, which
don't have the dylib.

BUG=None
R=mcgrathr@chromium.org

Committed:
https://chromium.googlesource.com/native_client/src/native_client/+/2d5bba1b4...
==========

[commit-bot: I haz the power, 2016-01-19 22:41:56]

Committed patchset #2 (id:20001) as
https://chromium.googlesource.com/native_client/src/native_client/+/2d5bba1b4...

[Derek Schuff, 2016-01-20 17:43:06]

A revert of this CL (patchset #2 id:20001) has been created in
https://codereview.chromium.org/1609323002/ by dschuff@chromium.org.

The reason for reverting is: This change breaks the LLVM regression tests.