[Mac] Log stack trace for CHECK in bug 97285.

[Mac] Log stack trace for CHECK in bug 97285.

The hypothesis in comment #55 on the bug is that while setting up a
new channel to an already-connected plugin,
NPChannelBase::RemoveRoute() can be called on the earlier channel, so
the renderer and plugin channel mappings get out of sync.  This will
add a Breakpad key encoding the top stack frames when RemoveRoute() is
called within the scoped of WebPluginDelegateProxy::Initialize().

objc_zombies.mm already implemented the stackframe-encoding code, so I
pulled that to a central location.  That code can be left in place
when the other code is removed after some traces are generated.

BUG=97285
TEST=crash server.


Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=137896

[Scott Hess - ex-Googler, 2012-05-16 22:55:43]

Brett, sorry for roping you in, but you look like optimal OWNER.

[Scott Hess - ex-Googler, 2012-05-16 22:59:52]

OK, adding a couple, that was unfair:

thakis: moving the stack thing from objc_zombies and sharing
piman: understands the potential race
brettw: OWNERS overload for rubber-stamp at some point.

[Nico, 2012-05-16 23:04:56]

mac bits lgtm

https://chromiumcodereview.appspot.com/10408004/diff/1/base/mac/crash_logging.mm
File base/mac/crash_logging.mm (right):

https://chromiumcodereview.appspot.com/10408004/diff/1/base/mac/crash_logging...
base/mac/crash_logging.mm:40: static const size_t kBreakpadValueMax = 255;
Remove static, also below

https://chromiumcodereview.appspot.com/10408004/diff/1/base/mac/crash_logging...
base/mac/crash_logging.mm:44: static const size_t kMaxCount = kBreakpadValueMax
/ 11;
s/11/strlen("0xaabbccdd ")?

[piman, 2012-05-16 23:19:27]

https://chromiumcodereview.appspot.com/10408004/diff/1/content/renderer/webpl...
File content/renderer/webplugin_delegate_proxy.cc (right):

https://chromiumcodereview.appspot.com/10408004/diff/1/content/renderer/webpl...
content/renderer/webplugin_delegate_proxy.cc:316: }
Is it ok not to reset here? You might get RemoveRoute from the couple of sync
message calls below.

[Scott Hess - ex-Googler, 2012-05-17 22:08:11]

PTAL?

https://chromiumcodereview.appspot.com/10408004/diff/1/base/mac/crash_logging.mm
File base/mac/crash_logging.mm (right):

https://chromiumcodereview.appspot.com/10408004/diff/1/base/mac/crash_logging...
base/mac/crash_logging.mm:40: static const size_t kBreakpadValueMax = 255;
On 2012/05/16 23:04:56, Nico wrote:
> Remove static, also below

Done.

https://chromiumcodereview.appspot.com/10408004/diff/1/base/mac/crash_logging...
base/mac/crash_logging.mm:44: static const size_t kMaxCount = kBreakpadValueMax
/ 11;
On 2012/05/16 23:04:56, Nico wrote:
> s/11/strlen("0xaabbccdd ")?

Rather than pile things higher and deeper, I revised the snippet to just track
the overall length and stop adding elements when it would become too long.

https://chromiumcodereview.appspot.com/10408004/diff/1/content/renderer/webpl...
File content/renderer/webplugin_delegate_proxy.cc (right):

https://chromiumcodereview.appspot.com/10408004/diff/1/content/renderer/webpl...
content/renderer/webplugin_delegate_proxy.cc:316: }
On 2012/05/16 23:19:27, piman wrote:
> Is it ok not to reset here? You might get RemoveRoute from the couple of sync
> message calls below.

Hmm.  auto_reset<> doesn't have a Reset(), but I can wrap it in a scoped_ptr<>,
which is ugly as heck, but this is already ugly as sin, so ...

[piman, 2012-05-17 22:10:54]

lgtm

[brettw, 2012-05-18 05:51:46]

lgtm

[commit-bot: I haz the power, 2012-05-18 16:39:14]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/shess@chromium.org/10408004/6003

[commit-bot: I haz the power, 2012-05-18 18:07:15]

Change committed as 137896