Avoid uninitialized memory when decomposing a matrix fails.

Avoid uninitialized memory when decomposing a matrix fails.

It is possible for TransformationMatrix::decompose to fail, and call
sites should check this.

BUG=469756
Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=192434

[jbroman, 2015-03-24 02:09:27]

jbroman@chromium.org changed reviewers:
+ fmalita@chromium.org, pdr@chromium.org, schenney@chromium.org

[jbroman, 2015-03-24 02:09:28]

R=fmalita,schenney,pdr for platform/ OWNERS
CC=dstockwell,alancutter because I want to check that this behaviour is
reasonable for animation under these circumstances (it's what
TransformationMatrix::blend does when decomposition fails, so I hope so).

Tried to write a unit test for this, but while it read from uninitialized memory
without this patch, I couldn't get it to be detected in an MSAN build of the
unit test binary.

[shans, 2015-03-24 02:40:08]

On 2015/03/24 02:09:28, jbroman wrote:
> R=fmalita,schenney,pdr for platform/ OWNERS
> CC=dstockwell,alancutter because I want to check that this behaviour is
> reasonable for animation under these circumstances (it's what
> TransformationMatrix::blend does when decomposition fails, so I hope so).
> 
> Tried to write a unit test for this, but while it read from uninitialized
memory
> without this patch, I couldn't get it to be detected in an MSAN build of the
> unit test binary.

Regardless of whether the memory is initialized, I think the behavior you've
implemented is more correct than what's currently there.

LGTM.

[pdr., 2015-03-24 03:36:10]

Change LGTM

Can you add the test from the bug too?

[alancutter (OOO until 2018), 2015-03-24 08:25:38]

Does this change fix crbug.com/469247 as well?

[f(malita), 2015-03-24 13:37:14]

lgtm++

[jbroman, 2015-03-24 14:29:53]

On 2015/03/24 03:36:10, pdr wrote:
> Can you add the test from the bug too?

I can get it to fail reliably in content_shell not in layout test mode, even
with tweaks. :(

On 2015/03/24 08:25:38, alancutter wrote:
> Does this change fix crbug.com/469247 as well?

No clue; I don't have access to that bug.

[jbroman, 2015-03-24 14:29:59]

The CQ bit was checked by jbroman@chromium.org

[commit-bot: I haz the power, 2015-03-24 14:30:06]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1027513003/1

[commit-bot: I haz the power, 2015-03-24 14:33:25]

Committed patchset #1 (id:1) as
https://src.chromium.org/viewvc/blink?view=rev&revision=192434