Roll src/tools/gyp to 6ee91ad8659871916f9aa840d42e1513befdf638.

Roll src/tools/gyp to 6ee91ad8659871916f9aa840d42e1513befdf638.

This is mainly to avoid excessive linker parallelism:
- Reduce DefaultConcurrentLinks from phys/4GB to phys/5GB.

There are three gyp changes pulled in by this CL:
https://chromium.googlesource.com/external/gyp/+log/5122240c..6ee91ad8

R=thakis@chromium.org
BUG=440500
Committed: https://crrev.com/945d5e45f7a7a24c86e44b6259cd9658d5ce2189
Cr-Commit-Position: refs/heads/master@{#343251}

[brucedawson, 2015-07-29 00:10:36]

Here's a roll CL for the 4GB to 5 GB gyp change. There are three changes
included - since you reviewed them all I assume you can assess the risks.

I used roll_dep.py but I didn't like its commit message so I added to it with
the range URL syntax, etc.

There is a separate change to update DefaultConcurrentLinks for gn
(https://codereview.chromium.org/1261073004). It would be nice to merge the two
but it looks like that is considered bad form? Feel free to review that one as
well if you want.

[brucedawson, 2015-07-29 18:16:40]

I got some more data last night which may make me want to suggest an even more
conservative calculation, so maybe we should sit on this for now.

Basically, linker parallelism is more random than I'd expected. Even 12 linkers
is rarely hit, but when it is probably too much even with VS 2013. So
restricting to 10 may be smarter.

I'll gather more data over the weekend.

[Nico, 2015-07-29 18:27:10]

lgtm

I have no opinion on doing the gn change in this same CL or not. Either way has
its pros and cons, but they seem fairly minor.

[brucedawson, 2015-07-29 18:42:25]

Here's a sample of some of the variation that I saw across 14 builds:

VS 2013:
Max commit: 39.0 GB, max linker WS: 21.858 GB with 4 linkers
Max commit: 48.2 GB, max linker WS: 35.740 GB with 12 linkers

VS 2015:
Max commit: 45.7 GB, max linker WS: 30.246 GB with 5 linkers
Max commit: 57.9 GB, max linker WS: 43.176 GB with 12 linkers

Perversely enough the worst-case memory consumption happened when
GYP_LINK_CONCURRENCY was set to 12. It's not clear whether that was coincidence
or if clamping linker concurrency to twelve then perturbed the timing such that
12 big links happened simultaneously.

[brucedawson, 2015-08-11 20:41:13]

Further tests showed that reducing the links did not reliably
correlated with making memory worse and should reduce the worst-case.
Switching to VS 2015 may require reducing parallel links further -
going as low as 8 parallel links on a Z620 does not harm build times
in the full rebuild case.

[brucedawson, 2015-08-11 20:41:16]

The CQ bit was checked by brucedawson@chromium.org

[commit-bot: I haz the power, 2015-08-11 20:41:53]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1262693003/1
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1262693003/1

[commit-bot: I haz the power, 2015-08-11 21:29:58]

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

[commit-bot: I haz the power, 2015-08-11 21:29:58]

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

[brucedawson, 2015-08-12 22:06:33]

This change is blocked on https://codereview.chromium.org/1284423002/ and will
be landed after that change lands (to fix the android_compile_dbg and
linux_android_rel_ng failures).

[brucedawson, 2015-08-13 00:25:00]

After the blocking change was landed linux_android_rel_ng failed
WebRtcBrowserTest.CallWithSctpDataAndMedia both with and without the patch...

[brucedawson, 2015-08-13 17:53:02]

The CQ bit was checked by brucedawson@chromium.org

[brucedawson, 2015-08-13 17:53:03]

The patchset sent to the CQ was uploaded after l-g-t-m from thakis@chromium.org
Link to the patchset: https://codereview.chromium.org/1262693003/#ps80001
(title: "Resetting to the original change.")

[commit-bot: I haz the power, 2015-08-13 17:53:15]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1262693003/80001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1262693003/80001

[commit-bot: I haz the power, 2015-08-13 19:16:36]

Committed patchset #5 (id:80001)

[commit-bot: I haz the power, 2015-08-13 19:17:19]

Patchset 5 (id:??) landed as
https://crrev.com/945d5e45f7a7a24c86e44b6259cd9658d5ce2189
Cr-Commit-Position: refs/heads/master@{#343251}