New issue
Advanced search Search tips
Note: Color blocks (like or ) mean that a user may not be available. Tooltip shows the reason.

Issue 726226 link

Starred by 13 users

Issue metadata

Status: Verified
Owner:
Last visit > 30 days ago
Closed: May 2017
Cc:
Components:
EstimatedDays: ----
NextAction: ----
OS: Android
Pri: 1
Type: Bug-Regression



Sign in to add a comment

Text on web site is going corrupted and missing on scrolling the page

Project Member Reported by dknandiraju@chromium.org, May 25 2017

Issue description

Application Version (from "Chrome Settings > About Chrome"): 60.0.3110.0
Android Build Number (from "Android Settings > About Phone/Tablet"): 7.0.0
Device: Moto g4 Plus

Steps to reproduce: 
1) Launch chrome and complete FRE
2) Open any website with text article (eg., m.timesofindia.com)
3) Browse through any article and try to scroll slowly

Observed behavior: Text is corrupted

Expected behavior: Text should not get corrupted

Frequency: 5/5
<number of times you were able to reproduce> 

Additional comments: 

 
Issue 726240 has been merged into this issue.
Labels: Type-Bug-Regression
Note:
1) Good Build: 60.0.3109.0
   Bad Build:  60.0.3110.0
2) Bisect range: https://chromium.googlesource.com/chromium/src/+log/60.0.3109.0..60.0.3110.0?pretty=fuller&n=10000
3) Please find logs and video @ http://go/chrome-androidlogs1/6/726226
4) Unable to find the exact culprit as an exception "We don't have enough builds to bisect" is getting thrown when tried to bisect builds
5) This issue is observed only on N and O devices not observed on other devices
Cc: tedc...@chromium.org
Labels: -Pri-2 ReleaseBlock-Dev Pri-1
Status: Assigned (was: Untriaged)
Cc: boliu@chromium.org
Owner: aelias@chromium.org
Cc: jainabhi...@chromium.org
 Issue 726379  has been merged into this issue.
Labels: Hotlist-ConOps

Comment 7 by aelias@chromium.org, May 25 2017

Cc: aelias@chromium.org
 Issue 726368  has been merged into this issue.

Comment 8 by boliu@chromium.org, May 25 2017

Components: Internals>Compositing>Rasterization
I *think* this is only happens when gpu raster is enabled. Someone can double check by setting --disable-gpu-rasterization

Comment 9 by aelias@chromium.org, May 25 2017

Confirmed, only reproes with GPU raster.
Not see this issue after disable gpu-rasterization
Still repro on ToT r474709

Comment 12 by boliu@chromium.org, May 25 2017

Components: Internals>Skia
Owner: bsalomon@chromium.org
Bisected to this skia roll: https://chromium-review.googlesource.com/513374
Then this skia CL: https://skia-review.googlesource.com/17824

Comment 13 by boliu@chromium.org, May 25 2017

still reverts cleanly and fixes this. revert first, ask questions later?

Comment 14 by boliu@chromium.org, May 25 2017

oh already reverted
Yep. I will need to merge the revert to the Skia branch for M60, which has already been created.
Project Member

Comment 16 by bugdroid1@chromium.org, May 25 2017

The following revision refers to this bug:
  https://skia.googlesource.com/skia/+/1c214313248a4b5a69af14539608c54fb67c2bf8

commit 1c214313248a4b5a69af14539608c54fb67c2bf8
Author: Brian Salomon <bsalomon@google.com>
Date: Thu May 25 20:03:07 2017

Revert "Clear atlas textures at creation in Chrome"

This reverts commit 8785df15fe5a57651597d138d3c5aea0ffe3f661.

Reason for revert: Chrome bug
Bug:  chromium:726226 

Original change's description:
> Clear atlas textures at creation in Chrome
> 
> Bug: chromium:656320
> Change-Id: Ia65274aa733f199be188579821e745920493aefc
> Reviewed-on: https://skia-review.googlesource.com/17824
> Reviewed-by: Robert Phillips <robertphillips@google.com>
> Commit-Queue: Brian Salomon <bsalomon@google.com>
> 

TBR=bsalomon@google.com,robertphillips@google.com
# Not skipping CQ checks because original CL landed > 1 day ago.
Bug: chromium:656320

Change-Id: Ibc4dd7f184866b58481f2dc7e7e88da9169e7adc
Reviewed-on: https://skia-review.googlesource.com/17988
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>

[modify] https://crrev.com/1c214313248a4b5a69af14539608c54fb67c2bf8/include/gpu/GrCaps.h
[modify] https://crrev.com/1c214313248a4b5a69af14539608c54fb67c2bf8/src/gpu/GrDrawOpAtlas.cpp
[modify] https://crrev.com/1c214313248a4b5a69af14539608c54fb67c2bf8/src/gpu/gl/GrGLCaps.cpp
[modify] https://crrev.com/1c214313248a4b5a69af14539608c54fb67c2bf8/src/gpu/GrCaps.cpp

Project Member

Comment 17 by bugdroid1@chromium.org, May 25 2017

Labels: merge-merged-m60
The following revision refers to this bug:
  https://skia.googlesource.com/skia/+/89ca3a3b0210355a40bc0bc0ffd0104f91b67997

commit 89ca3a3b0210355a40bc0bc0ffd0104f91b67997
Author: Brian Salomon <bsalomon@google.com>
Date: Thu May 25 20:28:48 2017

Revert "Clear atlas textures at creation in Chrome"

This reverts commit 8785df15fe5a57651597d138d3c5aea0ffe3f661.

Reason for revert: Chrome bug
Bug:  chromium:726226 

Original change's description:
> Clear atlas textures at creation in Chrome
>
> Bug: chromium:656320
> Change-Id: Ia65274aa733f199be188579821e745920493aefc
> Reviewed-on: https://skia-review.googlesource.com/17824
> Reviewed-by: Robert Phillips <robertphillips@google.com>
> Commit-Queue: Brian Salomon <bsalomon@google.com>
>

TBR=bsalomon@google.com,robertphillips@google.com
Bug: chromium:656320

No-Tree-Checks: true
No-Try: true
No-Presubmit: true
Change-Id: Ibc4dd7f184866b58481f2dc7e7e88da9169e7adc
Reviewed-On: https://skia-review.googlesource.com/17988
Reviewed-By: Brian Salomon <bsalomon@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
Reviewed-on: https://skia-review.googlesource.com/17991

[modify] https://crrev.com/89ca3a3b0210355a40bc0bc0ffd0104f91b67997/include/gpu/GrCaps.h
[modify] https://crrev.com/89ca3a3b0210355a40bc0bc0ffd0104f91b67997/src/gpu/GrDrawOpAtlas.cpp
[modify] https://crrev.com/89ca3a3b0210355a40bc0bc0ffd0104f91b67997/src/gpu/gl/GrGLCaps.cpp
[modify] https://crrev.com/89ca3a3b0210355a40bc0bc0ffd0104f91b67997/src/gpu/GrCaps.cpp

Status: Started (was: Assigned)
This is reverted on Skia master and Skia M60. However, the Skia autoroller is paused until Chrome branches for M60. At that point the M60 branch of Chrome will pick up the M60 branch of Skia and get the fix. Also, the autoroller will restart and Chrome master will get the fix.

I've been able to reproduce this locally on several phones with different Adreno GPUs but not on my linux desktop or phones with GPUs from different vendors. I'm starting to wonder if this is a driver issue.
Cc: candr...@chromium.org krav...@chromium.org piman@chromium.org amineer@chromium.org
 Issue 726587  has been merged into this issue.
 Issue 726357  has been merged into this issue.
This appears to be an Adreno driver bug when we clear the single channel texture used for the grayscale glyph atlas.

I've made the following discoveries:

Inserting a glFlush() after the clear fixes it.

Oddly this still reproduces even when glColorMask is set to all GL_FALSE before glClear(GL_COLOR_BUFFER_BIT) is called.

I don't seem to be able to reproduce this in standalone Skia testing, just in Chrome. Perhaps there is something done by the command buffer or gpu process that is necessary for the repro.


 Issue 726756  has been merged into this issue.
Branch 3112 is the M60 candidate (we'll still need explicit confirmation but I expect it shortly).  I pinned that branch to the m60 Skia branch and kicked a build, should deploy in ~2-3 hours.
I checked all the duplicate bug reports and they're on various Adreno 4xx and Adreno 5xx.  I'd suggest adding a GPU-service-side workaround targeted to those GPU ranges before the reland.
Components: Internals>GPU>VendorSpecific

Comment 26 by boliu@chromium.org, May 26 2017

Cc: ppolise...@chromium.org
 Issue 726790  has been merged into this issue.

Comment 27 by boliu@chromium.org, May 26 2017

Labels: -Restrict-View-Google
Status: Fixed (was: Started)
60.0.3112.3 is rolling out to canary now and will fix this.
Labels: Hotlist-ConOps-Source-PlayStore Hotlist-ConOps-Channel-Canary
Status: Verified (was: Fixed)
This issue is not reproducible on all devices on latest 60.0.3112.6, hence verified
 Issue 726802  has been merged into this issue.
 Issue 727027  has been merged into this issue.
Cc: kbr@chromium.org zmo@chromium.org ynovikov@chromium.org kainino@chromium.org sunn...@chromium.org
 Issue 726528  has been merged into this issue.

Sign in to add a comment