Bug 238618 - RemoteRenderingBackendProxy fails to maintain correct state when gpu process crashes and upon deletion
Summary: RemoteRenderingBackendProxy fails to maintain correct state when gpu process ...
Status: RESOLVED FIXED
Alias: None
Product: WebKit
Classification: Unclassified
Component: Canvas (show other bugs)
Version: WebKit Nightly Build
Hardware: Unspecified Unspecified
: P2 Normal
Assignee: Kimmo Kinnunen
URL:
Keywords: InRadar
: 237295 (view as bug list)
Depends on:
Blocks: 237674 238516
  Show dependency tree / graph
 
Reported: 2022-03-31 11:06 PDT by Kimmo Kinnunen
Modified: 2022-04-26 05:41 PDT (History)
4 users (show)

See Also:

Attachments
Patch (7.43 KB, patch)
2022-03-31 11:18 PDT, Kimmo Kinnunen 		no flags Details | Formatted Diff | Diff
Patch (6.42 KB, patch)
2022-03-31 12:17 PDT, Kimmo Kinnunen 		no flags Details | Formatted Diff | Diff
For landing (6.38 KB, patch)
2022-04-04 04:42 PDT, Kimmo Kinnunen 		no flags Details | Formatted Diff | Diff
Show Obsolete (2) View All Add an attachment (proposed patch, testcase, etc.)

Note You need to log in before you can comment on or make changes to this bug.
Description Kimmo Kinnunen 2022-03-31 11:06:57 PDT 
RemoteRenderingBackendProxy fails to maintain correct state when gpu process crashes and upon deletion
Comment 1 Kimmo Kinnunen 2022-03-31 11:18:23 PDT 
Created attachment 456257 [details]
Patch
Comment 2 Simon Fraser (smfr) 2022-03-31 12:12:04 PDT 
Comment on attachment 456257 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=456257&action=review

> Source/WebKit/ChangeLog:3
> +        RemoteRenderingBackendProxy fails to maintain correct state when gpu process crashes and upon deletion

But this doesn't fix the issue where the GPU Process crashes while we're blocked on the m_wakeUpSemaphore, right?

> Source/WebKit/Platform/IPC/MessageReceiveQueueMap.cpp:64
> +            ALWAYS_LOG_WITH_STREAM(stream << "MessageReceiveQueueMap::addImpl - adding duplicate any id receiver " << static_cast<uint8_t>(receiverName));

When debugging I wanted a TextStream& operator<<(TextStream&, ReceiverName). Maybe we should add one, or do some kind of IPC::description() thing that makes them easy to print.
Comment 3 Kimmo Kinnunen 2022-03-31 12:17:06 PDT 
Created attachment 456268 [details]
Patch
Comment 4 Kimmo Kinnunen 2022-03-31 12:23:36 PDT 
> But this doesn't fix the issue where the GPU Process crashes while we're blocked on the m_wakeUpSemaphore, right?

No, this doesn't address it. Will work on it next week.

>> Source/WebKit/Platform/IPC/MessageReceiveQueueMap.cpp:64
>> +            ALWAYS_LOG_WITH_STREAM(stream << "MessageReceiveQueueMap::addImpl - adding duplicate any id receiver " << static_cast<uint8_t>(receiverName));

In fact this was a botched hunk that belonged to previous patch.

> When debugging I wanted a TextStream& operator<<(TextStream&, ReceiverName). Maybe we should add one, or do some kind of IPC::description() thing that makes them easy to print.

Added bug 238624, can do that tomorrow.
Comment 5 Kimmo Kinnunen 2022-04-04 04:42:32 PDT 
Created attachment 456558 [details]
For landing
Comment 6 EWS 2022-04-04 08:46:48 PDT 
Committed r292289 (249187@main): <https://commits.webkit.org/249187@main>

All reviewed patches have been landed. Closing bug and clearing flags on attachment 456558 [details].
Comment 7 Radar WebKit Bug Importer 2022-04-04 08:47:17 PDT 
<rdar://problem/91241701>
Comment 8 Said Abou-Hallawa 2022-04-15 18:22:05 PDT 
*** Bug 237295 has been marked as a duplicate of this bug. ***