Initialize the pad memory in GrBufferAllocPool.

Initialize the pad memory in GrBufferAllocPool.

Msan bot detects Use-of-uninitialized-value on the pad.

BUG=445745
Committed: https://skia.googlesource.com/skia/+/8f25c667ea975c92d87ad6cb96963946f9845542

[dshwang, 2015-01-22 15:28:56]

dongseong.hwang@intel.com changed reviewers:
+ bsalomon@chromium.org, reed@chromium.org

[dshwang, 2015-01-22 15:28:56]

could you review?

It relieves Use-of-uninitialized-value error a lot, but doesn't fix it
completely. I guess there is another reason.

[dshwang, 2015-01-22 15:32:17]

before this patch, I can reproduce Use-of-uninitialized-value once over about 10
times
after this patch, it's very difficult to reproduce it, but it's rarely
reproduced.

./out_msancros/Release/content_browsertests
--gtest_filter=*WebRtcGetUserMediaBrowserTest.RenderSameTrackMediastreamAndStop
--brave-new-test-launcher --single-process --use-fake-ui-for-media-stream=true

[dshwang, 2015-01-22 15:37:40]

On 2015/01/22 15:32:17, dshwang wrote:
> after this patch, it's very difficult to reproduce it, but it's rarely
> reproduced.

it might be fixed. I couldn't reproduce again. I might be confused.

[bsalomon, 2015-01-22 15:49:24]

bsalomon@google.com changed reviewers:
+ bsalomon@google.com

[bsalomon, 2015-01-22 15:49:25]

It seems ok, but I'm wondering if we can limit this to Chromium since it isn't
necessary for other clients.

[dshwang, 2015-01-22 16:19:06]

On 2015/01/22 15:49:25, bsalomon wrote:
> It seems ok, but I'm wondering if we can limit this to Chromium since it isn't
> necessary for other clients.

IMO it's good to apply it to other clients also.
If a hacker dump gpu memory, he/she can get uninitialized value.

[bsalomon, 2015-01-22 18:00:49]

On 2015/01/22 16:19:06, dshwang wrote:
> On 2015/01/22 15:49:25, bsalomon wrote:
> > It seems ok, but I'm wondering if we can limit this to Chromium since it
isn't
> > necessary for other clients.
> 
> IMO it's good to apply it to other clients also.
> If a hacker dump gpu memory, he/she can get uninitialized value.

I suppose that's true. lgtm.

[dshwang, 2015-01-22 18:29:30]

On 2015/01/22 18:00:49, bsalomon wrote:
> On 2015/01/22 16:19:06, dshwang wrote:
> > On 2015/01/22 15:49:25, bsalomon wrote:
> > > It seems ok, but I'm wondering if we can limit this to Chromium since it
> isn't
> > > necessary for other clients.
> > 
> > IMO it's good to apply it to other clients also.
> > If a hacker dump gpu memory, he/she can get uninitialized value.
> 
> I suppose that's true. lgtm.

thank you for reviewing!

[dshwang, 2015-01-22 18:29:33]

The CQ bit was checked by dongseong.hwang@intel.com

[commit-bot: I haz the power, 2015-01-22 18:29:44]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/865913002/1

[commit-bot: I haz the power, 2015-01-22 18:40:23]

Committed patchset #1 (id:1) as
https://skia.googlesource.com/skia/+/8f25c667ea975c92d87ad6cb96963946f9845542