New issue
Advanced search Search tips

Issue 866644 link

Starred by 1 user

Issue metadata

Status: Fixed
Owner:
Closed: Jul 30
Cc:
Components:
EstimatedDays: ----
NextAction: ----
OS: Linux , Chrome
Pri: 1
Type: Bug



Sign in to add a comment

glBufferData fails with "Error: 2 for Command kBufferData"

Project Member Reported by dcasta...@chromium.org, Jul 23

Issue description

While investigating  crbug.com/866643  we noticed that Google Keep on ChromeOS would often stop working.
The LOG contained this error message:
"Error: 2 for Command kBufferData"

Investigating a little bit more we noticed that HandleBufferData tries to get a transfer buffer that has been previously deleted.

Reverting crrev.com/d1b4ae2892b05cf622b3f649cb38b995ed2d1733 (GPU: Don't flush when destroying transfer buffers.) seems to fix the issue.

Assigning to the original author of that CL.
 
M69 Beta promotion is coming VERY soon. Your bug is labelled as Beta  ReleaseBlock, pls make sure to land the fix and request a merge into the release branch latest by 1:00 PM PT this Friday, 07/27. Thank you.

Owner: vmi...@chromium.org
Victor, jdarpinian seems to be OOO for this week, do you know if anyone else can take a look at this regression?
M69 Beta promotion is coming VERY soon. Your bug is labelled as Beta  ReleaseBlock, pls make sure to land the fix and request a merge into the release branch latest by 1:00 PM PT this Friday, 07/27. Thank you.
Cc: piman@chromium.org
Labels: -Pri-3 Pri-2
Cc: -piman@chromium.org vmi...@chromium.org
Components: -Internals>GPU Internals>GPU>Internals
Labels: -Pri-2 Pri-1
Owner: piman@chromium.org
I'll take a look.
https://chromium-review.googlesource.com/c/chromium/src/+/1152149 should fix this. I verified locally by injecting a transfer buffer free after BufferData (fails before, works after).
Project Member

Comment 7 by bugdroid1@chromium.org, Jul 27

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/4b15cfbec4cb34d660c4582935977dc2e5281a87

commit 4b15cfbec4cb34d660c4582935977dc2e5281a87
Author: Antoine Labour <piman@chromium.org>
Date: Fri Jul 27 01:43:38 2018

PPAPI: flush if pending before DestroyTransferBuffer

http://crrev.com/c/1105466 changed the precondition for
CommandBuffer::DestroyTransferBuffer to only require an OrderingBarrier
rather than a full flush, but the implementation in
PpapiCommandBufferProxy wasn't fixed to handle that.
This change ensures we issue the flush before the DestroyTransferBuffer
IPC if one is pending.

Bug:  866644 
Change-Id: I6d6a3b6a752c4bfcd32f66153c84829140ed2b5f
Reviewed-on: https://chromium-review.googlesource.com/1152149
Reviewed-by: Sunny Sachanandani <sunnyps@chromium.org>
Commit-Queue: Antoine Labour <piman@chromium.org>
Cr-Commit-Position: refs/heads/master@{#578511}
[modify] https://crrev.com/4b15cfbec4cb34d660c4582935977dc2e5281a87/ppapi/proxy/ppapi_command_buffer_proxy.cc

Labels: Merge-Request-69
Project Member

Comment 9 by sheriffbot@chromium.org, Jul 28

Labels: -Merge-Request-69 Hotlist-Merge-Approved Merge-Approved-69
Your change meets the bar and is auto-approved for M69. Please go ahead and merge the CL to branch 3497 manually. Please contact milestone owner if you have questions.
Owners: amineer@(Android), kariahda@(iOS), cindyb@(ChromeOS), govind@(Desktop)

For more details visit https://www.chromium.org/issue-tracking/autotriage - Your friendly Sheriffbot

Please merge your change to M69 branch 3497 by 2:00 PM PT Monday, 07/30, so we can pick it up for next week last M69 Dev release. Thank you.

Project Member

Comment 11 by bugdroid1@chromium.org, Jul 30

Labels: -merge-approved-69 merge-merged-3497
The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/29664e245be7dc5a81a1a089c09050815a8ee2a2

commit 29664e245be7dc5a81a1a089c09050815a8ee2a2
Author: Antoine Labour <piman@chromium.org>
Date: Mon Jul 30 16:42:18 2018

PPAPI: flush if pending before DestroyTransferBuffer

http://crrev.com/c/1105466 changed the precondition for
CommandBuffer::DestroyTransferBuffer to only require an OrderingBarrier
rather than a full flush, but the implementation in
PpapiCommandBufferProxy wasn't fixed to handle that.
This change ensures we issue the flush before the DestroyTransferBuffer
IPC if one is pending.

Bug:  866644 
Change-Id: I6d6a3b6a752c4bfcd32f66153c84829140ed2b5f
Reviewed-on: https://chromium-review.googlesource.com/1152149
Reviewed-by: Sunny Sachanandani <sunnyps@chromium.org>
Commit-Queue: Antoine Labour <piman@chromium.org>
Cr-Original-Commit-Position: refs/heads/master@{#578511}(cherry picked from commit 4b15cfbec4cb34d660c4582935977dc2e5281a87)
Reviewed-on: https://chromium-review.googlesource.com/1155049
Reviewed-by: Antoine Labour <piman@chromium.org>
Cr-Commit-Position: refs/branch-heads/3497@{#212}
Cr-Branched-From: 271eaf50594eb818c9295dc78d364aea18c82ea8-refs/heads/master@{#576753}
[modify] https://crrev.com/29664e245be7dc5a81a1a089c09050815a8ee2a2/ppapi/proxy/ppapi_command_buffer_proxy.cc

Status: Fixed (was: Assigned)

Sign in to add a comment