V8 Buildbot: Disable lsan on bots where it's not supported.

V8 Buildbot: Disable lsan on bots where it's not supported.

BUG=509995
TBR=thakis@chromium.org, jochen@chromium.org

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=296015

[Michael Achenbach, 2015-07-14 13:45:34]

machenbach@chromium.org changed reviewers:
+ jochen@chromium.org, thakis@chromium.org

[Michael Achenbach, 2015-07-14 13:45:41]

PTAL

[Nico, 2015-07-14 13:58:18]

Rather than setting and then unsetting this again, is it possible to not add
lsan in the first place if it does nothing? Do you use chromium/config.py? If
so, could you change def asan() in there to only add lsan if the target arch is
intel?

[Michael Achenbach, 2015-07-14 14:09:32]

On 2015/07/14 13:58:18, Nico wrote:
> Rather than setting and then unsetting this again, is it possible to not add
> lsan in the first place if it does nothing? Do you use chromium/config.py? If
> so, could you change def asan() in there to only add lsan if the target arch
is
> intel?

But the target_arch is intel (ia32). The v8_target_arch is arm or mipsel on the
bots above.

[Michael Achenbach, 2015-07-14 14:11:19]

On 2015/07/14 14:09:32, Michael Achenbach wrote:
> On 2015/07/14 13:58:18, Nico wrote:
> > Rather than setting and then unsetting this again, is it possible to not add
> > lsan in the first place if it does nothing? Do you use chromium/config.py?
If
> > so, could you change def asan() in there to only add lsan if the target arch
> is
> > intel?
> 
> But the target_arch is intel (ia32). The v8_target_arch is arm or mipsel on
the
> bots above.

And the other bots that have a similar configuration (on chromium.lkgr) are all
not recipes yet.

[Nico, 2015-07-14 14:11:45]

On 2015/07/14 14:09:32, Michael Achenbach wrote:
> On 2015/07/14 13:58:18, Nico wrote:
> > Rather than setting and then unsetting this again, is it possible to not add
> > lsan in the first place if it does nothing? Do you use chromium/config.py?
If
> > so, could you change def asan() in there to only add lsan if the target arch
> is
> > intel?
> 
> But the target_arch is intel (ia32). The v8_target_arch is arm or mipsel on
the
> bots above.

Oh. Ooooh. Then rs-lgtm as long as this build config doesn't infect bots on the
chromium waterfalls :-) /me runs away

[Nico, 2015-07-14 14:12:24]

On 2015/07/14 14:11:19, Michael Achenbach wrote:
> On 2015/07/14 14:09:32, Michael Achenbach wrote:
> > On 2015/07/14 13:58:18, Nico wrote:
> > > Rather than setting and then unsetting this again, is it possible to not
add
> > > lsan in the first place if it does nothing? Do you use chromium/config.py?
> If
> > > so, could you change def asan() in there to only add lsan if the target
arch
> > is
> > > intel?
> > 
> > But the target_arch is intel (ia32). The v8_target_arch is arm or mipsel on
> the
> > bots above.
> 
> And the other bots that have a similar configuration (on chromium.lkgr) are
all
> not recipes yet.

Wait, we have bots on lkgr that build for intel but build v8 for some other
config?

[Michael Achenbach, 2015-07-14 14:19:53]

On 2015/07/14 14:12:24, Nico wrote:
> On 2015/07/14 14:11:19, Michael Achenbach wrote:
> > On 2015/07/14 14:09:32, Michael Achenbach wrote:
> > > On 2015/07/14 13:58:18, Nico wrote:
> > > > Rather than setting and then unsetting this again, is it possible to not
> add
> > > > lsan in the first place if it does nothing? Do you use
chromium/config.py?
> > If
> > > > so, could you change def asan() in there to only add lsan if the target
> arch
> > > is
> > > > intel?
> > > 
> > > But the target_arch is intel (ia32). The v8_target_arch is arm or mipsel
on
> > the
> > > bots above.
> > 
> > And the other bots that have a similar configuration (on chromium.lkgr) are
> all
> > not recipes yet.
> 
> Wait, we have bots on lkgr that build for intel but build v8 for some other
> config?

They build for ia32 with a v8 using the arm simulator generating arm code. I
thought from a clang compiler point-of-view this should be considered intel 32
bits.

[Michael Achenbach, 2015-07-14 14:21:17]

On 2015/07/14 14:19:53, Michael Achenbach wrote:
> On 2015/07/14 14:12:24, Nico wrote:
> > On 2015/07/14 14:11:19, Michael Achenbach wrote:
> > > On 2015/07/14 14:09:32, Michael Achenbach wrote:
> > > > On 2015/07/14 13:58:18, Nico wrote:
> > > > > Rather than setting and then unsetting this again, is it possible to
not
> > add
> > > > > lsan in the first place if it does nothing? Do you use
> chromium/config.py?
> > > If
> > > > > so, could you change def asan() in there to only add lsan if the
target
> > arch
> > > > is
> > > > > intel?
> > > > 
> > > > But the target_arch is intel (ia32). The v8_target_arch is arm or mipsel
> on
> > > the
> > > > bots above.
> > > 
> > > And the other bots that have a similar configuration (on chromium.lkgr)
are
> > all
> > > not recipes yet.
> > 
> > Wait, we have bots on lkgr that build for intel but build v8 for some other
> > config?
> 
> They build for ia32 with a v8 using the arm simulator generating arm code. I
> thought from a clang compiler point-of-view this should be considered intel 32
> bits.

Commented on the bug.

[jochen (gone - plz use gerrit), 2015-07-14 14:37:25]

lgtm once thakis is happy for realz

[Michael Achenbach, 2015-07-14 20:11:33]

Patch 2 disables lsan for 32 bits according to
https://code.google.com/p/chromium/issues/detail?id=509995#c9

[Nico, 2015-07-14 20:15:12]

lgtm

[Michael Achenbach, 2015-07-15 06:44:26]

The CQ bit was checked by machenbach@chromium.org

[Michael Achenbach, 2015-07-15 06:44:26]

The patchset sent to the CQ was uploaded after l-g-t-m from jochen@chromium.org
Link to the patchset: https://codereview.chromium.org/1241643002/#ps20001
(title: "Disable for 32 bits.")

[commit-bot: I haz the power, 2015-07-15 06:44:34]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1241643002/20001

[commit-bot: I haz the power, 2015-07-15 06:50:26]

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

[commit-bot: I haz the power, 2015-07-15 06:50:27]

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

[Michael Achenbach, 2015-07-15 07:16:36]

machenbach@chromium.org changed reviewers:
+ glider@chromium.org, kjellander@chromium.org, phoglund@chromium.org

[Michael Achenbach, 2015-07-15 07:16:36]

machenbach@chromium.org changed required reviewers:
+ glider@chromium.org

[Michael Achenbach, 2015-07-15 07:16:37]

Patch 3 adds forgotten expectations for disabling lsan on some 32 bit webrtc
bots. CC'ing more folks (all OOO unfortunately).

https://codereview.chromium.org/1241643002/diff/40001/scripts/slave/recipe_mo...
File scripts/slave/recipe_modules/chromium/config.py (left):

https://codereview.chromium.org/1241643002/diff/40001/scripts/slave/recipe_mo...
scripts/slave/recipe_modules/chromium/config.py:519: def
chromium_win_clang_asan(c):
This is now covered by excluding 32 bits above.

[Michael Achenbach, 2015-07-15 07:20:29]

FYI webrtc folks: I think also your 32 bit asan bots will fail on the current
clang roll like ours did. This CL should fix that.

[Alexander Potapenko, 2015-07-15 09:44:36]

lgtm

[Michael Achenbach, 2015-07-15 09:55:46]

The CQ bit was checked by machenbach@chromium.org

[Michael Achenbach, 2015-07-15 09:55:46]

The patchset sent to the CQ was uploaded after l-g-t-m from thakis@chromium.org,
jochen@chromium.org
Link to the patchset: https://codereview.chromium.org/1241643002/#ps40001
(title: "Moar expectations")

[commit-bot: I haz the power, 2015-07-15 09:55:54]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1241643002/40001

[commit-bot: I haz the power, 2015-07-15 09:59:06]

Committed patchset #3 (id:40001) as
http://src.chromium.org/viewvc/chrome?view=rev&revision=296015