| Summary: | ANGLE OpenGL: Cherry-pick commit GL: Fix bug with: Skip redundant flushes. | ||||||
|---|---|---|---|---|---|---|---|
| Product: | WebKit | Reporter: | Kimmo Kinnunen <kkinnunen> | ||||
| Component: | ANGLE | Assignee: | Kimmo Kinnunen <kkinnunen> | ||||
| Status: | RESOLVED CONFIGURATION CHANGED | ||||||
| Severity: | Normal | CC: | dino, geofflang, kbr, kkinnunen, kpiddington, webkit-bug-importer | ||||
| Priority: | P2 | Keywords: | InRadar | ||||
| Version: | WebKit Local Build | ||||||
| Hardware: | Unspecified | ||||||
| OS: | Unspecified | ||||||
| Bug Depends on: | 232476 | ||||||
| Bug Blocks: | 232472 | ||||||
| Attachments: |
|
||||||
|
Description
Kimmo Kinnunen
2021-10-29 00:34:25 PDT
Created attachment 442800 [details]
Patch
Are you sure you want to cherry-pick this? I'm not sure it increased stability in Chrome, and will only make WebKit's and upstream ANGLE diverge further. (In reply to Kenneth Russell from comment #2) > Are you sure you want to cherry-pick this? I'm not sure it increased > stability in Chrome, and will only make WebKit's and upstream ANGLE diverge > further. If I understand correctly, bug 232476 is part of mitigation for the crashes. That patch contains a bug that is fixed by this patch? I see. Yes, these were all CLs associated with the parent bug (https://crbug.com/1181068). I'm not sure though how much the flush-related work improved stability. Maybe geofflang@ can comment. Chunking TexSubImage2D calls into 120 KB max chunks was the primary workaround for the kernel panics. This patch definitely trended the total flush crashes down too. The largest fix was the final one which fixed some invalid GL usage in Chrome. Won't this roll into WebKit when ANGLE synchronizes? Thanks Geoff. It will come into WebKit during the next ANGLE roll into WebKit. It was unclear whether the WebKit team needed this sooner - we're aiming for a matter of a week or two for that roll. This was merged as part of ANGLE update. |