TSANv2: disable non sandbox-related threading checks under TSAN.

TSANv2: disable non sandbox-related threading checks under TSAN.

Allow Chromium to start under TSAN by removing the threading checks
before fork() in that environment.

Note that TSANv2 doesn't reliably support non-spawn fork() for now:
https://code.google.com/p/thread-sanitizer/issues/detail?id=44

But at least this allows using it on Chrome despite risks of running
into deadlocks.

NOTRY=true

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=247708

[jln (very slow on Chromium), 2014-01-28 22:28:00]

Dmitry, Alexander, should we do this or would you rather wait for TSAN to
support fork() better?

[Dmitry Vyukov, 2014-01-29 06:27:15]

On 2014/01/28 22:28:00, jln wrote:
> Dmitry, Alexander, should we do this or would you rather wait for TSAN to
> support fork() better?

Even if TSAN supports fork better, these CHECKs will still fail.

LGTM

[jln (very slow on Chromium), 2014-01-29 08:15:24]

On Tue, Jan 28, 2014 at 10:27 PM, <dvyukov@google.com> wrote:

> On 2014/01/28 22:28:00, jln wrote:
>
>> Dmitry, Alexander, should we do this or would you rather wait for TSAN to
>> support fork() better?
>>
>
> Even if TSAN supports fork better, these CHECKs will still fail.
>

Yes. My question is more: do we prevent people from using Chrome with TSAN
until then (to shelter them from hard to debug deadlocks) or do we let them?

It looks like we both favor the later.

To unsubscribe from this group and stop receiving emails from it, send an email
to chromium-reviews+unsubscribe@chromium.org.

[commit-bot: I haz the power, 2014-01-29 08:16:25]

No LGTM from a valid reviewer yet. Only full committers are accepted.
Even if an LGTM may have been provided, it was from a non-committer or
a lowly provisional committer, _not_ a full super star committer.
See http://www.chromium.org/getting-involved/become-a-committer
Note that this has nothing to do with OWNERS files.

[jln (very slow on Chromium), 2014-01-29 08:21:40]

jochen: could you please owner stamp
content/browser/renderer_host/render_sandbox_host_linux.cc?

[dvyukov, 2014-01-29 08:30:31]

LGTM from chromium account if that helps

[dvyukov, 2014-01-29 08:35:18]

On 2014/01/29 08:15:24, jln wrote:
> On Tue, Jan 28, 2014 at 10:27 PM, <mailto:dvyukov@google.com> wrote:
> 
> > On 2014/01/28 22:28:00, jln wrote:
> >
> >> Dmitry, Alexander, should we do this or would you rather wait for TSAN to
> >> support fork() better?
> >>
> >
> > Even if TSAN supports fork better, these CHECKs will still fail.
> >
> 
> Yes. My question is more: do we prevent people from using Chrome with TSAN
> until then (to shelter them from hard to debug deadlocks) or do we let them?
> 
> It looks like we both favor the later.

Main users today are Alexander with buildbots (who is in the position to debug
it) and ClusterFuzz (who will contact us in case of any problems). We do not
want to delay them.

Btw, last week I've made some changes to tsan fork handling which will hopefully
resolve most of the problems (they probably do not reach chromium yet).
We've decided to fully support:
1. forking in a single-threaded process (e.g. chromium with aura).
2. forking in a multi-threaded process and then going single-threaded (race
detection is mostly irrelevant in this case).
3. forking in a multi-threaded process and then exec'ing
If you fork in a multi-threaded process and then create more threads, then tsan
will print a bold warning and die. Or you can pass a special flag, then tsan
will continue, but you are warned already.

[jochen (gone - plz use gerrit), 2014-01-29 08:53:00]

lgtm

[Alexander Potapenko, 2014-01-29 09:03:18]

On 2014/01/29 08:53:00, jochen wrote:
> lgtm

LGTM
I've uploaded a similar CL, but apparently forgot to send it out:
https://codereview.chromium.org/143403003

Note that the TSan background thread itself should not lead to deadlocks (and
Dmitry has fixed issue 44 you're referring to)

[commit-bot: I haz the power, 2014-01-29 09:03:42]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/jln@chromium.org/130423014/20001

[commit-bot: I haz the power, 2014-01-29 10:23:26]

Retried try job too often on android_dbg for step(s) slave_steps
http://build.chromium.org/p/tryserver.chromium/buildstatus?builder=android_db...

[commit-bot: I haz the power, 2014-01-29 17:04:08]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/jln@chromium.org/130423014/20001

[commit-bot: I haz the power, 2014-01-29 17:06:17]

Change committed as 247708