Summary: | REGRESSION (r121395): fast/workers/worker-context-gc.html failing on GTK Linux 64-bit Debug, Release | ||
---|---|---|---|
Product: | WebKit | Reporter: | Zan Dobersek <zan> |
Component: | Tools / Tests | Assignee: | Nobody <webkit-unassigned> |
Status: | RESOLVED FIXED | ||
Severity: | Normal | CC: | abarth, charles.wei, jochen, levin |
Priority: | P2 | Keywords: | LayoutTestFailure, MakingBotsRed, Regression |
Version: | 528+ (Nightly build) | ||
Hardware: | Unspecified | ||
OS: | Unspecified | ||
URL: | http://build.webkit.org/results/GTK%20Linux%2064-bit%20Debug/r121407%20(34422)/results.html |
Description
Zan Dobersek
2012-06-27 23:59:12 PDT
I would roll out the change. btw, I just noticed that this line in trunk/Source/WebCore/workers/WorkerThread.cpp: m_groupSettings->setIndexedDBDatabasePath(settings->indexedDBDatabasePath()); is incorrect. It should be m_groupSettings->setIndexedDBDatabasePath(settings->indexedDBDatabasePath().isolatedCopy()); but I doubt that issue is causing this problem. I suppose the questions to ask are: 1. Why use m_workerObject in WorkerMessagingProxy::startWorkerContext when it already has m_scriptExecutionContext? 2. How do you know that every item in the line "m_workerObject->scriptExecutionContext())->page()->group().groupSettings()" is alive by the time that this method is called? I should have noticed this stuff in the original code review. By the way, we're considering moving the IDB state from PageGroup to just Page to make IDB work more like some of the other storage systems. Rolled out in bug #90143, bug #88338 reopened. Removing the crash expectation shortly. |