Workaround for issue breaking NACL on ChromeOS.

Workaround for issue breaking NACL on ChromeOS.

ChromeOS uses AFDO to improve Chrome performance. However, after the GN
migration, NACL broke. We triaged this to using AFDO on
tcmalloc. tcmalloc has some dependencies on specific stack traces and
section allocations that may break once AFDO is used. So, we need to
disable AFDO for the file debugallocation_shim.cc (object bisecting
pointed to it). We need a quick workaround, so we are doing this with a
GCC pragma. The real solution will come in a few days by a change in GN
to avoid passing -fauto-profile to tcmalloc files.

BUG=chromium:629593
TEST=verified by hand problem does not reproduce after this change.

Committed: https://crrev.com/a6f64234d71eeb6cd0571bd963301c8a617889f4
Cr-Commit-Position: refs/heads/master@{#409081}

[llozano, 2016-08-01 20:03:09]

Description was changed from

==========
Workaround for issue breaking NACL on ChromeOS.

ChromeOS uses AFDO to improve Chrome performance. However, after the GN
migration, NACL broke. We triaged this to using AFDO on
tcmalloc. tcmalloc has some dependencies on specific stack traces and
section allocations that may break once AFDO is used. So, we need to
disable AFDO for the file debugallocation_shim.cc (object bisecting
pointed to it). We need a quick workaround, so we are doing this with a
GCC pragma. The real solution will come in a few days by a change in GN
to avoid passing -fauto-profile to tcmalloc files.

BUG=chromium:629593
TEST=verified by hand problem does not reproduce after this change.
==========

to

==========
Workaround for issue breaking NACL on ChromeOS.

ChromeOS uses AFDO to improve Chrome performance. However, after the GN
migration, NACL broke. We triaged this to using AFDO on
tcmalloc. tcmalloc has some dependencies on specific stack traces and
section allocations that may break once AFDO is used. So, we need to
disable AFDO for the file debugallocation_shim.cc (object bisecting
pointed to it). We need a quick workaround, so we are doing this with a
GCC pragma. The real solution will come in a few days by a change in GN
to avoid passing -fauto-profile to tcmalloc files.

BUG=chromium:629593
TEST=verified by hand problem does not reproduce after this change.
==========

[llozano, 2016-08-01 20:03:09]

llozano@chromium.org changed reviewers:
+ dpranke@chromium.org, thakis@chromium.org

[llozano, 2016-08-01 20:03:29]

The CQ bit was checked by llozano@chromium.org

[commit-bot: I haz the power, 2016-08-01 20:03:54]

CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2016-08-01 20:03:55]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2016-08-01 20:03:56]

No L-G-T-M from a valid reviewer yet. 
CQ run can only be started by full committers or once the patch has
received an L-G-T-M from a full committer.
Even if an L-G-T-M may have been provided, it was from a non-committer,
_not_ a full super star committer.
See http://www.chromium.org/getting-involved/become-a-committer
Note that this has nothing to do with OWNERS files.

[Dirk Pranke, 2016-08-01 20:08:06]

lgtm, but thakis@ needs to review also (for owner review).

[Nico, 2016-08-01 20:10:06]

lgtm, looking forward to the more permanent fix.

Out of interest, why wasn't this a problem in the gyp build?

[llozano, 2016-08-01 20:32:53]

On 2016/08/01 20:10:06, Nico wrote:
> lgtm, looking forward to the more permanent fix.
> 
> Out of interest, why wasn't this a problem in the gyp build?

not sure.

This is a problem only for nacl_helper. On GYP, nacl_helper was built
diferently. So, I assume nacl_helper was not using AFDO in GYP.

[llozano, 2016-08-01 20:33:24]

The CQ bit was checked by llozano@chromium.org

[Nico, 2016-08-01 20:33:52]

hm, maybe the extra cros build flags should only be applied if !is_nacl then?
then we'd keep matching the gyp build and this wouldn't be needed

[commit-bot: I haz the power, 2016-08-01 20:33:54]

CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[Dirk Pranke, 2016-08-01 20:38:30]

No, nacl_helper was a regular linux executable even in the GYP build. My guess
is that the layout of the executable is just different enough to not be
triggering things.

[commit-bot: I haz the power, 2016-08-01 20:50:04]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2016-08-01 20:50:05]

Try jobs failed on following builders:
  cast_shell_linux on master.tryserver.chromium.linux (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.linux/builders/cast_shell_linu...)

[Dirk Pranke, 2016-08-01 20:54:57]

On 2016/08/01 20:50:05, commit-bot: I haz the power wrote:
> Try jobs failed on following builders:
>   cast_shell_linux on master.tryserver.chromium.linux (JOB_FAILED,
>
http://build.chromium.org/p/tryserver.chromium.linux/builders/cast_shell_linu...)

I guess this syntax doesn't actually work w/ clang ...

[llozano, 2016-08-01 21:43:35]

On 2016/08/01 20:54:57, Dirk Pranke wrote:
> On 2016/08/01 20:50:05, commit-bot: I haz the power wrote:
> > Try jobs failed on following builders:
> >   cast_shell_linux on master.tryserver.chromium.linux (JOB_FAILED,
> >
>
http://build.chromium.org/p/tryserver.chromium.linux/builders/cast_shell_linu...)
> 
> I guess this syntax doesn't actually work w/ clang ...

yes, I was not expecting that. please review again.

[llozano, 2016-08-01 21:45:19]

The CQ bit was checked by llozano@chromium.org

[llozano, 2016-08-01 21:45:19]

The patchset sent to the CQ was uploaded after l-g-t-m from
dpranke@chromium.org, thakis@chromium.org
Link to the patchset: https://codereview.chromium.org/2198253002/#ps20001
(title: "Workaround for issue breaking NACL on ChromeOS.")

[commit-bot: I haz the power, 2016-08-01 21:45:48]

CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[Nico, 2016-08-01 21:47:29]

https://codereview.chromium.org/2198253002/diff/20001/base/allocator/debugall...
File base/allocator/debugallocation_shim.cc (right):

https://codereview.chromium.org/2198253002/diff/20001/base/allocator/debugall...
base/allocator/debugallocation_shim.cc:10: #if ((__GNUC__ == 4 && __GNUC_MINOR__
>= 9) || __GNUC__ > 4)
I think you can just do `#if !defined(__clang__)`

[llozano, 2016-08-01 21:53:49]

On 2016/08/01 21:47:29, Nico wrote:
>
https://codereview.chromium.org/2198253002/diff/20001/base/allocator/debugall...
> File base/allocator/debugallocation_shim.cc (right):
> 
>
https://codereview.chromium.org/2198253002/diff/20001/base/allocator/debugall...
> base/allocator/debugallocation_shim.cc:10: #if ((__GNUC__ == 4 &&
__GNUC_MINOR__
> >= 9) || __GNUC__ > 4)
> I think you can just do `#if !defined(__clang__)`

done.

[llozano, 2016-08-01 21:54:00]

The CQ bit was checked by llozano@chromium.org

[llozano, 2016-08-01 21:54:00]

The patchset sent to the CQ was uploaded after l-g-t-m from
dpranke@chromium.org, thakis@chromium.org
Link to the patchset: https://codereview.chromium.org/2198253002/#ps40001
(title: "Workaround for issue breaking NACL on ChromeOS.")

[commit-bot: I haz the power, 2016-08-01 21:54:30]

CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[llozano, 2016-08-01 22:31:58]

On 2016/08/01 20:33:52, Nico wrote:
> hm, maybe the extra cros build flags should only be applied if !is_nacl then?
> then we'd keep matching the gyp build and this wouldn't be needed

not sure we can do that?
on ChromeOS nacl is always on? So, I would not be able to build chrome with
Nacl?

[commit-bot: I haz the power, 2016-08-01 22:40:14]

Description was changed from

==========
Workaround for issue breaking NACL on ChromeOS.

ChromeOS uses AFDO to improve Chrome performance. However, after the GN
migration, NACL broke. We triaged this to using AFDO on
tcmalloc. tcmalloc has some dependencies on specific stack traces and
section allocations that may break once AFDO is used. So, we need to
disable AFDO for the file debugallocation_shim.cc (object bisecting
pointed to it). We need a quick workaround, so we are doing this with a
GCC pragma. The real solution will come in a few days by a change in GN
to avoid passing -fauto-profile to tcmalloc files.

BUG=chromium:629593
TEST=verified by hand problem does not reproduce after this change.
==========

to

==========
Workaround for issue breaking NACL on ChromeOS.

ChromeOS uses AFDO to improve Chrome performance. However, after the GN
migration, NACL broke. We triaged this to using AFDO on
tcmalloc. tcmalloc has some dependencies on specific stack traces and
section allocations that may break once AFDO is used. So, we need to
disable AFDO for the file debugallocation_shim.cc (object bisecting
pointed to it). We need a quick workaround, so we are doing this with a
GCC pragma. The real solution will come in a few days by a change in GN
to avoid passing -fauto-profile to tcmalloc files.

BUG=chromium:629593
TEST=verified by hand problem does not reproduce after this change.
==========

[commit-bot: I haz the power, 2016-08-01 22:40:16]

Committed patchset #3 (id:40001)

[commit-bot: I haz the power, 2016-08-01 22:43:37]

Description was changed from

==========
Workaround for issue breaking NACL on ChromeOS.

ChromeOS uses AFDO to improve Chrome performance. However, after the GN
migration, NACL broke. We triaged this to using AFDO on
tcmalloc. tcmalloc has some dependencies on specific stack traces and
section allocations that may break once AFDO is used. So, we need to
disable AFDO for the file debugallocation_shim.cc (object bisecting
pointed to it). We need a quick workaround, so we are doing this with a
GCC pragma. The real solution will come in a few days by a change in GN
to avoid passing -fauto-profile to tcmalloc files.

BUG=chromium:629593
TEST=verified by hand problem does not reproduce after this change.
==========

to

==========
Workaround for issue breaking NACL on ChromeOS.

ChromeOS uses AFDO to improve Chrome performance. However, after the GN
migration, NACL broke. We triaged this to using AFDO on
tcmalloc. tcmalloc has some dependencies on specific stack traces and
section allocations that may break once AFDO is used. So, we need to
disable AFDO for the file debugallocation_shim.cc (object bisecting
pointed to it). We need a quick workaround, so we are doing this with a
GCC pragma. The real solution will come in a few days by a change in GN
to avoid passing -fauto-profile to tcmalloc files.

BUG=chromium:629593
TEST=verified by hand problem does not reproduce after this change.

Committed: https://crrev.com/a6f64234d71eeb6cd0571bd963301c8a617889f4
Cr-Commit-Position: refs/heads/master@{#409081}
==========

[commit-bot: I haz the power, 2016-08-01 22:43:38]

Patchset 3 (id:??) landed as
https://crrev.com/a6f64234d71eeb6cd0571bd963301c8a617889f4
Cr-Commit-Position: refs/heads/master@{#409081}

[Will Harris, 2016-08-16 16:30:57]

please use lowest level OWNERS if there is one.