base/process: Use /proc/self/stat to read startup time.

base/process: Use /proc/self/stat to read startup time.

After a new renderer task is cloned, CurrentProcessInfo::CreationTime
is called to compute the process startup time.

CurrentProcessInfo::CreationTime tries to read this information from
/proc/$PID/stat, where $PID is the current pid as self reported
by the new child process.

If the new task was cloned with CLONE_NEWPID, $PID will be in a different
namespace than the pid namespace represented in /proc, causing
CurrentProcessInfo::CreationTime to fail.

This CL changes CurrentProcessInfo::CreationTime so that it now
reads from /proc/self/stat instead of /proc/$PID/stat.

BUG=
TEST=Chrome now starts on intel cros devices even with DCHECKs enabled.

Review-Url: https://codereview.chromium.org/2715583002
Cr-Commit-Position: refs/heads/master@{#452588}
Committed: https://chromium.googlesource.com/chromium/src/+/4c25edcf92c0d4acccdc02f6f3a4a5264115e05c

[Daniele Castagna, 2017-02-22 21:07:40]

The CQ bit was checked by dcastagna@chromium.org to run a CQ dry run

[Daniele Castagna, 2017-02-22 21:08:11]

dcastagna@chromium.org changed reviewers:
+ jln@chromium.org

[commit-bot: I haz the power, 2017-02-22 21:08:14]

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

[jln (very slow on Chromium), 2017-02-22 22:37:15]

https://codereview.chromium.org/2715583002/diff/1/base/process/internal_linux.cc
File base/process/internal_linux.cc (right):

https://codereview.chromium.org/2715583002/diff/1/base/process/internal_linux...
base/process/internal_linux.cc:142: std::vector<std::string> proc_stats;
A little too much code duplication here. Perhaps refactor a tad if it's worth
it?  Up to you.

https://codereview.chromium.org/2715583002/diff/1/base/process/internal_linux.h
File base/process/internal_linux.h (right):

https://codereview.chromium.org/2715583002/diff/1/base/process/internal_linux...
base/process/internal_linux.h:77: int64_t ReadProcStatsAndGetFieldAsInt64(const
std::string& pid,
Nit: I think we still don't allow overloading like this. Also it seems weird to
take pid as a string.

Let's just have a ReadProcSelfStatsAndGet...() instead ?

[commit-bot: I haz the power, 2017-02-22 22:38:16]

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

[commit-bot: I haz the power, 2017-02-22 22:38:17]

Dry run: This issue passed the CQ dry run.

[Daniele Castagna, 2017-02-23 18:03:42]

The CQ bit was checked by dcastagna@chromium.org to run a CQ dry run

[Daniele Castagna, 2017-02-23 18:04:10]

https://codereview.chromium.org/2715583002/diff/1/base/process/internal_linux.cc
File base/process/internal_linux.cc (right):

https://codereview.chromium.org/2715583002/diff/1/base/process/internal_linux...
base/process/internal_linux.cc:142: std::vector<std::string> proc_stats;
On 2017/02/22 at 22:37:15, jln (very slow on Chromium) wrote:
> A little too much code duplication here. Perhaps refactor a tad if it's worth
it?  Up to you.

Done.

https://codereview.chromium.org/2715583002/diff/1/base/process/internal_linux.h
File base/process/internal_linux.h (right):

https://codereview.chromium.org/2715583002/diff/1/base/process/internal_linux...
base/process/internal_linux.h:77: int64_t ReadProcStatsAndGetFieldAsInt64(const
std::string& pid,
On 2017/02/22 at 22:37:15, jln (very slow on Chromium) wrote:
> Nit: I think we still don't allow overloading like this. Also it seems weird
to take pid as a string.
> 

Overloading should be fine as long as it's not confusing from the caller POV.

> Let's just have a ReadProcSelfStatsAndGet...() instead ?

I like this suggestion though. Done.

[commit-bot: I haz the power, 2017-02-23 18:04:18]

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

[jln (very slow on Chromium), 2017-02-23 18:48:07]

lgtm

[Daniele Castagna, 2017-02-23 18:54:52]

dcastagna@chromium.org changed reviewers:
+ thakis@chromium.org

[Daniele Castagna, 2017-02-23 18:54:53]

+thakis for ownership.

[Nico, 2017-02-23 19:01:24]

lgtm

[commit-bot: I haz the power, 2017-02-23 19:20:24]

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

[commit-bot: I haz the power, 2017-02-23 19:20:25]

Dry run: This issue passed the CQ dry run.

[Daniele Castagna, 2017-02-23 19:27:22]

The CQ bit was checked by dcastagna@chromium.org

[commit-bot: I haz the power, 2017-02-23 19:28:06]

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, 2017-02-23 19:35:55]

CQ is committing da patch.

Bot data: {"patchset_id": 20001, "attempt_start_ts": 1487878042266100,
"parent_rev": "39b6a2fef0949d471f9c7b0de93909943e070d41", "commit_rev":
"4c25edcf92c0d4acccdc02f6f3a4a5264115e05c"}

[commit-bot: I haz the power, 2017-02-23 19:36:27]

Description was changed from

==========
base/process: Use /proc/self/stat to read startup time.

After a new renderer task is cloned, CurrentProcessInfo::CreationTime
is called to compute the process startup time.

CurrentProcessInfo::CreationTime tries to read this information from
/proc/$PID/stat, where $PID is the current pid as self reported
by the new child process.

If the new task was cloned with CLONE_NEWPID, $PID will be in a different
namespace than the pid namespace represented in /proc, causing
CurrentProcessInfo::CreationTime to fail.

This CL changes CurrentProcessInfo::CreationTime so that it now
reads from /proc/self/stat instead of /proc/$PID/stat.

BUG=
TEST=Chrome now starts on intel cros devices even with DCHECKs enabled.
==========

to

==========
base/process: Use /proc/self/stat to read startup time.

After a new renderer task is cloned, CurrentProcessInfo::CreationTime
is called to compute the process startup time.

CurrentProcessInfo::CreationTime tries to read this information from
/proc/$PID/stat, where $PID is the current pid as self reported
by the new child process.

If the new task was cloned with CLONE_NEWPID, $PID will be in a different
namespace than the pid namespace represented in /proc, causing
CurrentProcessInfo::CreationTime to fail.

This CL changes CurrentProcessInfo::CreationTime so that it now
reads from /proc/self/stat instead of /proc/$PID/stat.

BUG=
TEST=Chrome now starts on intel cros devices even with DCHECKs enabled.

Review-Url: https://codereview.chromium.org/2715583002
Cr-Commit-Position: refs/heads/master@{#452588}
Committed:
https://chromium.googlesource.com/chromium/src/+/4c25edcf92c0d4acccdc02f6f3a4...
==========

[commit-bot: I haz the power, 2017-02-23 19:36:29]

Committed patchset #2 (id:20001) as
https://chromium.googlesource.com/chromium/src/+/4c25edcf92c0d4acccdc02f6f3a4...