android: make exclude_unwind_tables depend only on is_official_build

android: make exclude_unwind_tables depend only on is_official_build

Context: unwind tables are stripped in order to save precious binary
size on Android (order of 1-2 MB on arm32, ~10 MB on arm64 and x86).
The price to pay is that, in the case of a crash, the device is NOT
able to unwind the stack (using conventional libunwind). The only
thing that is able to unwind the stack is breakpad (offline) using
CFI information.

The rationale of this change is: buildtype=Official should match
what we ship, espeically in terms of perf and size. As the name
suggests, "branding" should not cause behavioral changes.
Unwind tables are not a "branding" feature.
Unfortunately this change makes crashes of local official builds harder
to debug (Requiring all the breakpad {micro,mini}dump_stackwalk
machinery), but that matches what happens on the build that are
ultimately shipped to the public.

BUG=610994

[Primiano Tucci (use gerrit), 2016-05-11 13:16:20]

primiano@chromium.org changed reviewers:
+ agrieve@chromium.org, yfriedman@chromium.org

[Primiano Tucci (use gerrit), 2016-05-11 13:21:53]

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

[commit-bot: I haz the power, 2016-05-11 13:22:07]

Dry run: CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1967103002/1
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1967103002/1

[Yaron, 2016-05-11 13:29:45]

On 2016/05/11 13:22:07, commit-bot: I haz the power wrote:
> Dry run: CQ is trying da patch. Follow status at
>  https://chromium-cq-status.appspot.com/patch-status/1967103002/1
> View timeline at
>  https://chromium-cq-status.appspot.com/patch-timeline/1967103002/1

oh. I had one concern - will this break any of our bots/testing? I imagine some
perf bots could build "official" to try and get as close to the shipping product
as possible but if the symbols aren't archived (a random bot won't upload
symbols to breakpad) and symbolization doesn't happen on the device for some
reason, then we'd be left with an inactionable crash. I suspect we can overcome
this by ensuring that bots symbolize any crashes they hit (and we typically do)
but wasn't sure about all the perf bots so wanted to double check with you

[Primiano Tucci (use gerrit), 2016-05-11 13:34:07]

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

[Primiano Tucci (use gerrit), 2016-05-11 13:34:07]

+jbudorick the question here is: are you aware of any bots that today build with
official (but NOT branding) which do NOT have a {micro,mini}dump_stackwalk step
to decode crashes?

On 2016/05/11 13:29:45, Yaron wrote:
> oh. I had one concern - will this break any of our bots/testing? I imagine
some
> perf bots could build "official" to try and get as close to the shipping
product
> as possible but if the symbols aren't archived (a random bot won't upload
> symbols to breakpad) and symbolization doesn't happen on the device for some
> reason, then we'd be left with an inactionable crash. I suspect we can
overcome
> this by ensuring that bots symbolize any crashes they hit (and we typically
do)
> but wasn't sure about all the perf bots so wanted to double check with you

I was worried about the same.
AFAIK perf bots use both official + branding, so are already in this situation:
https://build.chromium.org/p/chromium.perf/builders/Android%20Builder/builds/...

Not sure if there are other cases. I thought that we have a microdump_stackwalk
step on all bots these days.

[commit-bot: I haz the power, 2016-05-11 13:59:49]

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

[commit-bot: I haz the power, 2016-05-11 13:59:50]

Dry run: Try jobs failed on following builders:
  win_chromium_rel_ng on tryserver.chromium.win (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.win/builders/win_chromium_rel_...)

[Primiano Tucci (use gerrit), 2016-05-12 12:48:21]

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

[commit-bot: I haz the power, 2016-05-12 12:48:41]

Dry run: CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1967103002/1
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1967103002/1

[commit-bot: I haz the power, 2016-05-12 13:25:35]

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

[commit-bot: I haz the power, 2016-05-12 13:25:36]

Dry run: This issue passed the CQ dry run.

[jbudorick, 2016-05-17 00:09:49]

On 2016/05/11 13:34:07, Primiano Tucci wrote:
> +jbudorick the question here is: are you aware of any bots that today build
with
> official (but NOT branding) which do NOT have a {micro,mini}dump_stackwalk
step
> to decode crashes?

Sorry for the delayed response. I don't know of _any_ bot with a
microdump_stackwalk step yet. Tracking work on that over in
https://bugs.chromium.org/p/chromium/issues/detail?id=593106

> 
> On 2016/05/11 13:29:45, Yaron wrote:
> > oh. I had one concern - will this break any of our bots/testing? I imagine
> some
> > perf bots could build "official" to try and get as close to the shipping
> product
> > as possible but if the symbols aren't archived (a random bot won't upload
> > symbols to breakpad) and symbolization doesn't happen on the device for some
> > reason, then we'd be left with an inactionable crash. I suspect we can
> overcome
> > this by ensuring that bots symbolize any crashes they hit (and we typically
> do)
> > but wasn't sure about all the perf bots so wanted to double check with you
> 
> I was worried about the same.
> AFAIK perf bots use both official + branding, so are already in this
situation:
>
https://build.chromium.org/p/chromium.perf/builders/Android%20Builder/builds/...
> 
> Not sure if there are other cases. I thought that we have a
microdump_stackwalk
> step on all bots these days.