[Telemetry] Improve _GetProcJiffies operation with using grep.

[Telemetry] Improve _GetProcJiffies operation with using grep.

Previously, telemetry read the whole '/proc/timer_list' file just to parse the
first occurence of "jiffies". In case the file is big and telemetry is run
against remote device, this can create a big delay for transferring the file
to host device.

This patch modify it so that telemetry just use
'grep jiffies /proc/timer_list' to get the only lines that contain 'jiffies'.

Context: https://groups.google.com/a/chromium.org/forum/#!topic/telemetry/SblQqF0tHbc

Try job runs:
./tools/perf/run_benchmark --browser=trybot-all-android page_cycler.top_10_mobile --also-run-disabled-tests
https://codereview.chromium.org/1100233003

./tools/perf/run_benchmark --browser=trybot-all-mac page_cycler.intl_ja_zh
https://codereview.chromium.org/1106833004

./tools/perf/run_benchmark --browser=trybot-all-linux page_cycler.intl_ja_zh
https://codereview.chromium.org/1102673005

Committed: https://crrev.com/c3d7606e8efbb5ac063f2e98a14e09e48aff5022
Cr-Commit-Position: refs/heads/master@{#326961}

[nednguyen, 2015-04-24 22:13:49]

nednguyen@google.com changed reviewers:
+ aiolos@chromium.org, jbudorick@chromium.org

[nednguyen, 2015-04-24 23:14:20]

Patchset #1 (id:1) has been deleted

[aiolos (Not reviewing), 2015-04-24 23:55:15]

https://codereview.chromium.org/1106043003/diff/20001/tools/telemetry/telemet...
File tools/telemetry/telemetry/core/platform/linux_based_platform_backend.py
(right):

https://codereview.chromium.org/1106043003/diff/20001/tools/telemetry/telemet...
tools/telemetry/telemetry/core/platform/linux_based_platform_backend.py:154:
['grep', '-m', '1', 'jiffies:','/proc/timer_list'])
I'm a little wary of this change because it makes the code more fragile due to
dependence on the formatting of the files. If the format changed so that
'last-jiffies' or 'next-jiffies' had the colon appear right after it instead of
with while space between them, this would break, and we probably would never
know since we would still be getting a reasonable value. 1st Q: How much would
we throw off our tests if that happened? 2nd Q:  Are we willing to take that
risk?

[jbudorick, 2015-04-24 23:56:56]

https://codereview.chromium.org/1106043003/diff/20001/tools/telemetry/telemet...
File tools/telemetry/telemetry/core/platform/linux_based_platform_backend.py
(right):

https://codereview.chromium.org/1106043003/diff/20001/tools/telemetry/telemet...
tools/telemetry/telemetry/core/platform/linux_based_platform_backend.py:154:
['grep', '-m', '1', 'jiffies:','/proc/timer_list'])
On 2015/04/24 23:55:15, aiolos wrote:
> I'm a little wary of this change because it makes the code more fragile due to
> dependence on the formatting of the files. If the format changed so that
> 'last-jiffies' or 'next-jiffies' had the colon appear right after it instead
of
> with while space between them, this would break, and we probably would never
> know since we would still be getting a reasonable value. 1st Q: How much would
> we throw off our tests if that happened? 2nd Q:  Are we willing to take that
> risk?

Agreed, this seems like too much of an optimization. What about just grepping
for "jiffies" in /proc/timer_list and then doing some amount of processing with
the output?

[aiolos (Not reviewing), 2015-04-25 00:53:55]

https://codereview.chromium.org/1106043003/diff/20001/tools/telemetry/telemet...
File tools/telemetry/telemetry/core/platform/linux_based_platform_backend.py
(right):

https://codereview.chromium.org/1106043003/diff/20001/tools/telemetry/telemet...
tools/telemetry/telemetry/core/platform/linux_based_platform_backend.py:154:
['grep', '-m', '1', 'jiffies:','/proc/timer_list'])
On 2015/04/24 23:56:56, jbudorick wrote:
> On 2015/04/24 23:55:15, aiolos wrote:
> > I'm a little wary of this change because it makes the code more fragile due
to
> > dependence on the formatting of the files. If the format changed so that
> > 'last-jiffies' or 'next-jiffies' had the colon appear right after it instead
> of
> > with while space between them, this would break, and we probably would never
> > know since we would still be getting a reasonable value. 1st Q: How much
would
> > we throw off our tests if that happened? 2nd Q:  Are we willing to take that
> > risk?
> 
> Agreed, this seems like too much of an optimization. What about just grepping
> for "jiffies" in /proc/timer_list and then doing some amount of processing
with
> the output?

I'm much happier with that idea.

[nednguyen, 2015-04-25 02:02:07]

https://codereview.chromium.org/1106043003/diff/20001/tools/telemetry/telemet...
File tools/telemetry/telemetry/core/platform/linux_based_platform_backend.py
(right):

https://codereview.chromium.org/1106043003/diff/20001/tools/telemetry/telemet...
tools/telemetry/telemetry/core/platform/linux_based_platform_backend.py:154:
['grep', '-m', '1', 'jiffies:','/proc/timer_list'])
On 2015/04/25 00:53:55, aiolos wrote:
> On 2015/04/24 23:56:56, jbudorick wrote:
> > On 2015/04/24 23:55:15, aiolos wrote:
> > > I'm a little wary of this change because it makes the code more fragile
due
> to
> > > dependence on the formatting of the files. If the format changed so that
> > > 'last-jiffies' or 'next-jiffies' had the colon appear right after it
instead
> > of
> > > with while space between them, this would break, and we probably would
never
> > > know since we would still be getting a reasonable value. 1st Q: How much
> would
> > > we throw off our tests if that happened? 2nd Q:  Are we willing to take
that
> > > risk?
> > 
> > Agreed, this seems like too much of an optimization. What about just
grepping
> > for "jiffies" in /proc/timer_list and then doing some amount of processing
> with
> > the output?
> 
> I'm much happier with that idea.

Good call, done.  I use regex so the parsing is more resilient than the original
version/

[aiolos (Not reviewing), 2015-04-25 03:03:16]

lgtm

https://codereview.chromium.org/1106043003/diff/40001/tools/telemetry/telemet...
File
tools/telemetry/telemetry/core/platform/linux_based_platform_backend_unittest.py
(right):

https://codereview.chromium.org/1106043003/diff/40001/tools/telemetry/telemet...
tools/telemetry/telemetry/core/platform/linux_based_platform_backend_unittest.py:57:
jiffies: 10505463300
nit: Could you add a line for '.last_jiffies' like it had in the origional file
as well?

[nednguyen, 2015-04-25 04:10:24]

https://codereview.chromium.org/1106043003/diff/40001/tools/telemetry/telemet...
File
tools/telemetry/telemetry/core/platform/linux_based_platform_backend_unittest.py
(right):

https://codereview.chromium.org/1106043003/diff/40001/tools/telemetry/telemet...
tools/telemetry/telemetry/core/platform/linux_based_platform_backend_unittest.py:57:
jiffies: 10505463300
On 2015/04/25 03:03:16, aiolos wrote:
> nit: Could you add a line for '.last_jiffies' like it had in the origional
file
> as well?

Done.

[nednguyen, 2015-04-25 04:49:18]

The CQ bit was checked by nednguyen@google.com

[nednguyen, 2015-04-25 04:49:19]

The patchset sent to the CQ was uploaded after l-g-t-m from aiolos@chromium.org
Link to the patchset: https://codereview.chromium.org/1106043003/#ps60001
(title: "Update unittest to better match actual /proc/timer_list file")

[commit-bot: I haz the power, 2015-04-25 04:49:45]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1106043003/60001

[commit-bot: I haz the power, 2015-04-25 06:12:35]

Committed patchset #3 (id:60001)

[commit-bot: I haz the power, 2015-04-25 06:13:35]

Patchset 3 (id:??) landed as
https://crrev.com/c3d7606e8efbb5ac063f2e98a14e09e48aff5022
Cr-Commit-Position: refs/heads/master@{#326961}