Fix ProcessUtilTest.GetTerminationStatusCrash on Android L+

Fix ProcessUtilTest.GetTerminationStatusCrash on Android L+

On Android L+, signal and sigaction symbols are provided by libsigchain
that override the system's versions. There is a bug in these functions
where they essentially ignore requests to install SIG_DFL. This causes
ProcessUtilTest.GetTerminationStatusCrash to fail (as
CrashingChildProcess goes into infinite loop instead of crashing).

Workaround this issue by explicitly performing a syscall to
__NR_rt_sigaction to install SIG_DFL on Android, as breakpad does
(see https://breakpad.appspot.com/1804002/).

BUG=512255
TEST=ProcessUtilTest.GetTerminationStatusCrash

Committed: https://crrev.com/31510fe3e2d3b1f3123f391db52372b2506c46ca
Cr-Commit-Position: refs/heads/master@{#339584}

[Ian Cullinan, 2015-07-20 23:36:15]

cullinan@amazon.com changed reviewers:
+ thakis@chromium.org

[Ian Cullinan, 2015-07-20 23:37:38]

PTAL at this fix for http://crbug.com/512255

[Nico, 2015-07-20 23:44:07]

Looks terrible, please commit :-) (after updating this so that the comment stays
next to the line it was next to before)

Does base need to grow a dep on lss in some gyp / gn file too?

https://codereview.chromium.org/1241333002/diff/1/base/process/process_util_u...
File base/process/process_util_unittest.cc (right):

https://codereview.chromium.org/1241333002/diff/1/base/process/process_util_u...
base/process/process_util_unittest.cc:233: // instead of an abnormal termination
through the crash dump handler.
keep this with the ::signal call in the elif posix below

https://codereview.chromium.org/1241333002/diff/1/base/process/process_util_u...
base/process/process_util_unittest.cc:239: // To work around this, directly call
the system's sigaction.
whaaaaaaaat that is horrible

[Ian Cullinan, 2015-07-21 00:01:07]

On 2015/07/20 23:44:07, Nico wrote:
> Looks terrible, please commit :-) (after updating this so that the comment
stays
> next to the line it was next to before)
> 
> Does base need to grow a dep on lss in some gyp / gn file too?

I don't think so - lss is just that one header file, and I can't find any gyp or
gn files that mention it.

>
https://codereview.chromium.org/1241333002/diff/1/base/process/process_util_u...
> File base/process/process_util_unittest.cc (right):
> 
>
https://codereview.chromium.org/1241333002/diff/1/base/process/process_util_u...
> base/process/process_util_unittest.cc:233: // instead of an abnormal
termination
> through the crash dump handler.
> keep this with the ::signal call in the elif posix below

Done

>
https://codereview.chromium.org/1241333002/diff/1/base/process/process_util_u...
> base/process/process_util_unittest.cc:239: // To work around this, directly
call
> the system's sigaction.
> whaaaaaaaat that is horrible

Yeah :(

[Nico, 2015-07-21 00:03:18]

"lgtm"

[Ian Cullinan, 2015-07-21 00:04:39]

The CQ bit was checked by cullinan@amazon.com

[commit-bot: I haz the power, 2015-07-21 00:05:05]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1241333002/20001

[commit-bot: I haz the power, 2015-07-21 01:00:53]

Committed patchset #2 (id:20001)

[commit-bot: I haz the power, 2015-07-21 01:01:49]

Patchset 2 (id:??) landed as
https://crrev.com/31510fe3e2d3b1f3123f391db52372b2506c46ca
Cr-Commit-Position: refs/heads/master@{#339584}

[Primiano Tucci (use gerrit), 2015-07-21 09:04:11]

primiano@chromium.org changed reviewers:
+ primiano@chromium.org

[Primiano Tucci (use gerrit), 2015-07-21 09:04:11]

https://codereview.chromium.org/1241333002/diff/1/base/process/process_util_u...
File base/process/process_util_unittest.cc (right):

https://codereview.chromium.org/1241333002/diff/1/base/process/process_util_u...
base/process/process_util_unittest.cc:239: // To work around this, directly call
the system's sigaction.
On 2015/07/20 23:44:07, Nico wrote:
> whaaaaaaaat that is horrible

Yes, it is, but to the best of my knowledge this is the "best" workaround around
that.
Also, this is... ahem... **not the most horrifying** thing we've done to work
around these kinds of issues (typedefing + #include_next to fix typos in the
NDK, that's my favourite).

[michaelbai, 2015-07-21 16:30:36]

On 2015/07/21 09:04:11, Primiano Tucci wrote:
>
https://codereview.chromium.org/1241333002/diff/1/base/process/process_util_u...
> File base/process/process_util_unittest.cc (right):
> 
>
https://codereview.chromium.org/1241333002/diff/1/base/process/process_util_u...
> base/process/process_util_unittest.cc:239: // To work around this, directly
call
> the system's sigaction.
> On 2015/07/20 23:44:07, Nico wrote:
> > whaaaaaaaat that is horrible
> 
> Yes, it is, but to the best of my knowledge this is the "best" workaround
around
> that.
> Also, this is... ahem... **not the most horrifying** thing we've done to work
> around these kinds of issues (typedefing + #include_next to fix typos in the
> NDK, that's my favourite).

This patch broke the mips builder for Android

../../third_party/lss/linux_syscall_support.h:121:21: fatal error: sgidefs.h: No
such file or directory
 #include <sgidefs.h>
                     ^

[Nico, 2015-07-21 16:34:38]

On 2015/07/21 16:30:36, michaelbai wrote:
> On 2015/07/21 09:04:11, Primiano Tucci wrote:
> >
>
https://codereview.chromium.org/1241333002/diff/1/base/process/process_util_u...
> > File base/process/process_util_unittest.cc (right):
> > 
> >
>
https://codereview.chromium.org/1241333002/diff/1/base/process/process_util_u...
> > base/process/process_util_unittest.cc:239: // To work around this, directly
> call
> > the system's sigaction.
> > On 2015/07/20 23:44:07, Nico wrote:
> > > whaaaaaaaat that is horrible
> > 
> > Yes, it is, but to the best of my knowledge this is the "best" workaround
> around
> > that.
> > Also, this is... ahem... **not the most horrifying** thing we've done to
work
> > around these kinds of issues (typedefing + #include_next to fix typos in the
> > NDK, that's my favourite).
> 
> This patch broke the mips builder for Android
> 
> ../../third_party/lss/linux_syscall_support.h:121:21: fatal error: sgidefs.h:
No
> such file or directory
>  #include <sgidefs.h>
>                      ^

Yup, tracked at crbug.com/512384

[michaelbai, 2015-07-21 16:43:04]

On 2015/07/21 16:34:38, Nico wrote:
> On 2015/07/21 16:30:36, michaelbai wrote:
> > On 2015/07/21 09:04:11, Primiano Tucci wrote:
> > >
> >
>
https://codereview.chromium.org/1241333002/diff/1/base/process/process_util_u...
> > > File base/process/process_util_unittest.cc (right):
> > > 
> > >
> >
>
https://codereview.chromium.org/1241333002/diff/1/base/process/process_util_u...
> > > base/process/process_util_unittest.cc:239: // To work around this,
directly
> > call
> > > the system's sigaction.
> > > On 2015/07/20 23:44:07, Nico wrote:
> > > > whaaaaaaaat that is horrible
> > > 
> > > Yes, it is, but to the best of my knowledge this is the "best" workaround
> > around
> > > that.
> > > Also, this is... ahem... **not the most horrifying** thing we've done to
> work
> > > around these kinds of issues (typedefing + #include_next to fix typos in
the
> > > NDK, that's my favourite).
> > 
> > This patch broke the mips builder for Android
> > 
> > ../../third_party/lss/linux_syscall_support.h:121:21: fatal error:
sgidefs.h:
> No
> > such file or directory
> >  #include <sgidefs.h>
> >                      ^
> 
> Yup, tracked at crbug.com/512384

It seemed that you can't fix this quickly, I need to revert this patch as it
broke the buildbot :(

[michaelbai, 2015-07-21 16:48:52]

On 2015/07/21 16:43:04, michaelbai wrote:
> On 2015/07/21 16:34:38, Nico wrote:
> > On 2015/07/21 16:30:36, michaelbai wrote:
> > > On 2015/07/21 09:04:11, Primiano Tucci wrote:
> > > >
> > >
> >
>
https://codereview.chromium.org/1241333002/diff/1/base/process/process_util_u...
> > > > File base/process/process_util_unittest.cc (right):
> > > > 
> > > >
> > >
> >
>
https://codereview.chromium.org/1241333002/diff/1/base/process/process_util_u...
> > > > base/process/process_util_unittest.cc:239: // To work around this,
> directly
> > > call
> > > > the system's sigaction.
> > > > On 2015/07/20 23:44:07, Nico wrote:
> > > > > whaaaaaaaat that is horrible
> > > > 
> > > > Yes, it is, but to the best of my knowledge this is the "best"
workaround
> > > around
> > > > that.
> > > > Also, this is... ahem... **not the most horrifying** thing we've done to
> > work
> > > > around these kinds of issues (typedefing + #include_next to fix typos in
> the
> > > > NDK, that's my favourite).
> > > 
> > > This patch broke the mips builder for Android
> > > 
> > > ../../third_party/lss/linux_syscall_support.h:121:21: fatal error:
> sgidefs.h:
> > No
> > > such file or directory
> > >  #include <sgidefs.h>
> > >                      ^
> > 
> > Yup, tracked at crbug.com/512384
> 
> It seemed that you can't fix this quickly, I need to revert this patch as it
> broke the buildbot :(

I didn't mean you, Nico :)

[Nico, 2015-07-21 16:50:22]

On 2015/07/21 16:48:52, michaelbai wrote:
> On 2015/07/21 16:43:04, michaelbai wrote:
> > On 2015/07/21 16:34:38, Nico wrote:
> > > On 2015/07/21 16:30:36, michaelbai wrote:
> > > > On 2015/07/21 09:04:11, Primiano Tucci wrote:
> > > > >
> > > >
> > >
> >
>
https://codereview.chromium.org/1241333002/diff/1/base/process/process_util_u...
> > > > > File base/process/process_util_unittest.cc (right):
> > > > > 
> > > > >
> > > >
> > >
> >
>
https://codereview.chromium.org/1241333002/diff/1/base/process/process_util_u...
> > > > > base/process/process_util_unittest.cc:239: // To work around this,
> > directly
> > > > call
> > > > > the system's sigaction.
> > > > > On 2015/07/20 23:44:07, Nico wrote:
> > > > > > whaaaaaaaat that is horrible
> > > > > 
> > > > > Yes, it is, but to the best of my knowledge this is the "best"
> workaround
> > > > around
> > > > > that.
> > > > > Also, this is... ahem... **not the most horrifying** thing we've done
to
> > > work
> > > > > around these kinds of issues (typedefing + #include_next to fix typos
in
> > the
> > > > > NDK, that's my favourite).
> > > > 
> > > > This patch broke the mips builder for Android
> > > > 
> > > > ../../third_party/lss/linux_syscall_support.h:121:21: fatal error:
> > sgidefs.h:
> > > No
> > > > such file or directory
> > > >  #include <sgidefs.h>
> > > >                      ^
> > > 
> > > Yup, tracked at crbug.com/512384
> > 
> > It seemed that you can't fix this quickly, I need to revert this patch as it
> > broke the buildbot :(
> 
> I didn't mean you, Nico :)

That bot doesn't have a trybot and not many devs build mips, so I think it's OK
to wait a bit with the revert maybe?

[michaelbai, 2015-07-21 16:55:11]

On 2015/07/21 16:50:22, Nico wrote:
> On 2015/07/21 16:48:52, michaelbai wrote:
> > On 2015/07/21 16:43:04, michaelbai wrote:
> > > On 2015/07/21 16:34:38, Nico wrote:
> > > > On 2015/07/21 16:30:36, michaelbai wrote:
> > > > > On 2015/07/21 09:04:11, Primiano Tucci wrote:
> > > > > >
> > > > >
> > > >
> > >
> >
>
https://codereview.chromium.org/1241333002/diff/1/base/process/process_util_u...
> > > > > > File base/process/process_util_unittest.cc (right):
> > > > > > 
> > > > > >
> > > > >
> > > >
> > >
> >
>
https://codereview.chromium.org/1241333002/diff/1/base/process/process_util_u...
> > > > > > base/process/process_util_unittest.cc:239: // To work around this,
> > > directly
> > > > > call
> > > > > > the system's sigaction.
> > > > > > On 2015/07/20 23:44:07, Nico wrote:
> > > > > > > whaaaaaaaat that is horrible
> > > > > > 
> > > > > > Yes, it is, but to the best of my knowledge this is the "best"
> > workaround
> > > > > around
> > > > > > that.
> > > > > > Also, this is... ahem... **not the most horrifying** thing we've
done
> to
> > > > work
> > > > > > around these kinds of issues (typedefing + #include_next to fix
typos
> in
> > > the
> > > > > > NDK, that's my favourite).
> > > > > 
> > > > > This patch broke the mips builder for Android
> > > > > 
> > > > > ../../third_party/lss/linux_syscall_support.h:121:21: fatal error:
> > > sgidefs.h:
> > > > No
> > > > > such file or directory
> > > > >  #include <sgidefs.h>
> > > > >                      ^
> > > > 
> > > > Yup, tracked at crbug.com/512384
> > > 
> > > It seemed that you can't fix this quickly, I need to revert this patch as
it
> > > broke the buildbot :(
> > 
> > I didn't mean you, Nico :)
> 
> That bot doesn't have a trybot and not many devs build mips, so I think it's
OK
> to wait a bit with the revert maybe?


This is unit tests and seemed not a urgent patch, and if I don't revert this, it
could hide other breakage in mips build.

[michaelbai, 2015-07-21 16:57:25]

A revert of this CL (patchset #2 id:20001) has been created in
https://codereview.chromium.org/1247023002/ by michaelbai@chromium.org.

The reason for reverting is: This patch broke the mips builder for Android

../../third_party/lss/linux_syscall_support.h:121:21: fatal error: sgidefs.h: No
such file or directory
 #include <sgidefs.h>.