AIX: Fix to get more accurate Thread's CPU time

AIX: Fix to get more accurate Thread's CPU time

On AIX clock_gettime provides CPU time with a resolution of
10ms, which causes the ThreadTicks testcase to fail since at
the 2 instances the CPU time of the thread outputs to 0.
Using AIX's API thread_cputime instead which provides CPU
time with a resolution of 1ns.
The testcase was added as part of https://codereview.chromium.org/1976603005

R=jochen@chromium.org, lpy@chromium.org, joransiu@ca.ibm.com, jyan@ca.ibm.com, michael_dawson@ca.ibm.com, mbrandy@us.ibm.com

BUG=
LOG=N

Committed: https://crrev.com/3361b7fd44f06dc1dc0569a326fb0ba3e0b41588
Cr-Commit-Position: refs/heads/master@{#38029}

[JaideepBajwa, 2016-07-22 20:46:37]

PTAL

[john.yan, 2016-07-22 20:47:52]

lgtm

[lpy, 2016-07-22 22:47:30]

https://codereview.chromium.org/2174003002/diff/1/src/base/platform/time.cc
File src/base/platform/time.cc (right):

https://codereview.chromium.org/2174003002/diff/1/src/base/platform/time.cc#n...
src/base/platform/time.cc:70: struct timespec ts;
hmmmm, what happen when clk_id is CLOCK_THREAD_CPUTIME_ID and thread_cputime(-1,
&tc) == 0? Why do you still need to call clock_gettime under this condition?

https://codereview.chromium.org/2174003002/diff/1/src/base/platform/time.cc#n...
src/base/platform/time.cc:81: result += (ts.tv_nsec /
v8::base::Time::kNanosecondsPerMicrosecond);
wrong indentation.

[JaideepBajwa, 2016-07-22 23:14:15]

On 2016/07/22 22:47:30, lpy wrote:
> https://codereview.chromium.org/2174003002/diff/1/src/base/platform/time.cc
> File src/base/platform/time.cc (right):
> 
>
https://codereview.chromium.org/2174003002/diff/1/src/base/platform/time.cc#n...
> src/base/platform/time.cc:70: struct timespec ts;
> hmmmm, what happen when clk_id is CLOCK_THREAD_CPUTIME_ID and
thread_cputime(-1,
> &tc) == 0? Why do you still need to call clock_gettime under this condition?
I only want to run thread_cputime when checking the Thread's CPU time. -1
indicates the
current thread. I also run clock_gettime to get the time in sec since
thread_cputime
only gives time in nsec. 
For MONOTONIC_CLOCK clock_gettime works well on AIX and gives correct sec and
nsec.
>
https://codereview.chromium.org/2174003002/diff/1/src/base/platform/time.cc#n...
> src/base/platform/time.cc:81: result += (ts.tv_nsec /
> v8::base::Time::kNanosecondsPerMicrosecond);
> wrong indentation.
I'll fix this. Thanks.

[lpy, 2016-07-22 23:20:58]

On 2016/07/22 23:14:15, JaideepBajwa wrote:
> On 2016/07/22 22:47:30, lpy wrote:
> > https://codereview.chromium.org/2174003002/diff/1/src/base/platform/time.cc
> > File src/base/platform/time.cc (right):
> > 
> >
>
https://codereview.chromium.org/2174003002/diff/1/src/base/platform/time.cc#n...
> > src/base/platform/time.cc:70: struct timespec ts;
> > hmmmm, what happen when clk_id is CLOCK_THREAD_CPUTIME_ID and
> thread_cputime(-1,
> > &tc) == 0? Why do you still need to call clock_gettime under this condition?
> I only want to run thread_cputime when checking the Thread's CPU time. -1
> indicates the
> current thread. I also run clock_gettime to get the time in sec since
> thread_cputime
> only gives time in nsec. 
> For MONOTONIC_CLOCK clock_gettime works well on AIX and gives correct sec and
> nsec.


Thanks for the clarification. LGTM from my side.

[jochen (gone - plz use gerrit), 2016-07-25 16:18:44]

lgtm

[JaideepBajwa, 2016-07-25 16:24:17]

The CQ bit was checked by bjaideep@ca.ibm.com

[JaideepBajwa, 2016-07-25 16:24:17]

The patchset sent to the CQ was uploaded after l-g-t-m from jyan@ca.ibm.com
Link to the patchset: https://codereview.chromium.org/2174003002/#ps20001
(title: "fixed incorrect indentation")

[commit-bot: I haz the power, 2016-07-25 16:24:28]

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

[JaideepBajwa, 2016-07-25 16:24:53]

On 2016/07/25 16:18:44, jochen wrote:
> lgtm

Thanks.

[commit-bot: I haz the power, 2016-07-25 16:45:01]

Committed patchset #2 (id:20001)

[commit-bot: I haz the power, 2016-07-25 16:46:38]

Description was changed from

==========
AIX: Fix to get more accurate Thread's CPU time

On AIX clock_gettime provides CPU time with a resolution of
10ms, which causes the ThreadTicks testcase to fail since at
the 2 instances the CPU time of the thread outputs to 0.
Using AIX's API thread_cputime instead which provides CPU
time with a resolution of 1ns.
The testcase was added as part of https://codereview.chromium.org/1976603005

R=jochen@chromium.org, lpy@chromium.org, joransiu@ca.ibm.com, jyan@ca.ibm.com,
michael_dawson@ca.ibm.com, mbrandy@us.ibm.com

BUG=
LOG=N
==========

to

==========
AIX: Fix to get more accurate Thread's CPU time

On AIX clock_gettime provides CPU time with a resolution of
10ms, which causes the ThreadTicks testcase to fail since at
the 2 instances the CPU time of the thread outputs to 0.
Using AIX's API thread_cputime instead which provides CPU
time with a resolution of 1ns.
The testcase was added as part of https://codereview.chromium.org/1976603005

R=jochen@chromium.org, lpy@chromium.org, joransiu@ca.ibm.com, jyan@ca.ibm.com,
michael_dawson@ca.ibm.com, mbrandy@us.ibm.com

BUG=
LOG=N

Committed: https://crrev.com/3361b7fd44f06dc1dc0569a326fb0ba3e0b41588
Cr-Commit-Position: refs/heads/master@{#38029}
==========

[commit-bot: I haz the power, 2016-07-25 16:46:39]

Patchset 2 (id:??) landed as
https://crrev.com/3361b7fd44f06dc1dc0569a326fb0ba3e0b41588
Cr-Commit-Position: refs/heads/master@{#38029}