Bug 217342

Summary:

[GPU Process] Implement GPU rendering resource cache

Product:

WebKit

Reporter:

Said Abou-Hallawa <sabouhallawa>

Component:

Canvas

Assignee:

Said Abou-Hallawa <sabouhallawa>

Status:

RESOLVED FIXED

Severity:

Normal

CC:

annulen, calvaris, cdumez, cgarcia, changseok, dino, eric.carlson, esprehn+autocc, ews-watchlist, fmalita, glenn, gustavo, gyuyoung.kim, japhet, jer.noble, jonlee, menard, mmaxfield, pdr, philipj, pnormand, ryuan.choi, schenney, sergio, simon.fraser, thorton, vjaquez, webkit-bug-importer, wenson_hsieh

Priority:

Keywords:

InRadar

Version:

WebKit Nightly Build

Hardware:

Unspecified

OS:

Unspecified

Bug Depends on:

217166, 217550, 217554, 217558, 217566, 217573, 217596, 217809, 218472, 218529, 218839, 218843, 218865, 218871, 218934

Bug Blocks:

Attachments:

Description	Flags
Patch	none
Patch	none
Patch	none
Patch	none
Patch	none
Patch	ews-feeder: commit-queue-
Patch	ews-feeder: commit-queue-

Description Said Abou-Hallawa 2020-10-05 15:13:29 PDT

All the rendering resources, such as ImageFrames, Fonts and ImageBuffers will be cached in the GPU side. Managing this cache will be done by RemoteRenderingBackend and RemoteRenderingBackendProxy. Both of these objects will have a separate objects for maintaining the cached objects. The goal is to have these objects cached once and referenced by their ids multiple times.

Comment 1 Said Abou-Hallawa 2020-10-05 15:33:06 PDT

Created attachment 410571 [details]
Patch

Comment 2 Said Abou-Hallawa 2020-10-05 18:09:14 PDT

Created attachment 410597 [details]
Patch

Comment 3 Said Abou-Hallawa 2020-10-05 22:25:23 PDT

Created attachment 410613 [details]
Patch

Comment 4 Said Abou-Hallawa 2020-10-05 23:40:34 PDT

Created attachment 410617 [details]
Patch

Comment 5 Myles C. Maxfield 2020-10-06 22:27:19 PDT

Comment on attachment 410617 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=410617&action=review

It seems like there will be 3 or 4 patches in here once this is broken up. I'll review those in more detail (just because this one is pretty big).

> Source/WebCore/platform/graphics/GraphicsContext.cpp:795
> +    if (m_impl && image.isPDFDocumentImage())

Why is this relevant?

> Source/WebCore/platform/graphics/cg/GraphicsContextCG.cpp:-292
> -    if (m_impl) {
> -        m_impl->drawNativeImage(image, imageSize, destRect, srcRect, options);
> -        return;
> -    }
> -

This is surprising.

Comment 6 Said Abou-Hallawa 2020-10-07 02:00:19 PDT

Created attachment 410738 [details]
Patch

Comment 7 Said Abou-Hallawa 2020-10-07 02:17:19 PDT

Created attachment 410739 [details]
Patch

Comment 8 Said Abou-Hallawa 2020-10-07 02:25:44 PDT

Created attachment 410740 [details]
Patch

Comment 9 Radar WebKit Bug Importer 2020-10-08 19:46:24 PDT

<rdar://problem/70122760>

Comment 10 Simon Fraser (smfr) 2021-01-06 09:16:21 PST

Is this patch still useful?

Comment 11 Said Abou-Hallawa 2021-03-05 10:50:02 PST

The attached patch was split and landed separately in the blocking bugs.