| Summary: | [ BigSur+ Debug wk2 EWS ] imported/w3c/web-platform-tests/html/webappapis/dynamic-markup-insertion/opening-the-input-stream/quirks.window.html is a flaky crash | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| Product: | WebKit | Reporter: | Robert Jenner <jenner> | ||||||
| Component: | Tools / Tests | Assignee: | Robert Jenner <jenner> | ||||||
| Status: | RESOLVED FIXED | ||||||||
| Severity: | Normal | CC: | cdumez, sihui_liu, webkit-bot-watchers-bugzilla, webkit-bug-importer | ||||||
| Priority: | P2 | Keywords: | InRadar | ||||||
| Version: | WebKit Nightly Build | ||||||||
| Hardware: | Unspecified | ||||||||
| OS: | Unspecified | ||||||||
| Attachments: |
|
||||||||
|
Description
Robert Jenner
2022-02-24 15:35:32 PST
Since this was having a negative impact on EWS, I have marked the expectations here as [ Pass Crash ] while I am investigating this: https://trac.webkit.org/changeset/290472/webkit The crash is pretty flaky. It happens more on Big Sur than it does on Monterey, but I have seen it in Monterey's history. I have attempted to reproduce it on both Big Sur and Monterey, as well as on Intel and Apple Silicon platforms, and I have been unable to reproduce it, even with --force (since I have set an expectation for the test) and up-to 10000 iterations. Created attachment 455259 [details]
Patch
Comment on attachment 455259 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=455259&action=review > Source/WebKit/NetworkProcess/storage/NetworkStorageManager.cpp:786 > + if (!m_localOriginStorageManagers.contains(origin)) double hash map look up here.. > Source/WebKit/NetworkProcess/storage/NetworkStorageManager.cpp:790 > + auto manager = m_localOriginStorageManagers.get(origin); .. and here. This is unnecessarily inefficient. > Source/WebKit/WebProcess/WebStorage/StorageAreaMap.cpp:327 > +StorageType StorageAreaMap::computeStorageType() can this be const? > Source/WebKit/WebProcess/WebStorage/StorageAreaMap.cpp:336 > +WebCore::ClientOrigin StorageAreaMap::clientOrigin() Can this be const? Comment on attachment 455259 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=455259&action=review r=me with changes. > Source/WebKit/WebProcess/WebStorage/StorageAreaMap.cpp:360 > auto completionHandler = [this, weakThis = WeakPtr { *this }, weakConnection = WeakPtr { ipcConnection }](auto remoteAreaIdentifier, auto items, auto messageIdentifier) mutable { Seems to be capturing the ipcConnection unnecessarily now. Comment on attachment 455259 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=455259&action=review >> Source/WebKit/NetworkProcess/storage/NetworkStorageManager.cpp:786 >> + if (!m_localOriginStorageManagers.contains(origin)) > > double hash map look up here.. Will use find() instead. >> Source/WebKit/WebProcess/WebStorage/StorageAreaMap.cpp:327 >> +StorageType StorageAreaMap::computeStorageType() > > can this be const? Sure. >> Source/WebKit/WebProcess/WebStorage/StorageAreaMap.cpp:336 >> +WebCore::ClientOrigin StorageAreaMap::clientOrigin() > > Can this be const? Sure. >> Source/WebKit/WebProcess/WebStorage/StorageAreaMap.cpp:360 >> auto completionHandler = [this, weakThis = WeakPtr { *this }, weakConnection = WeakPtr { ipcConnection }](auto remoteAreaIdentifier, auto items, auto messageIdentifier) mutable { > > Seems to be capturing the ipcConnection unnecessarily now. Will remove. Created attachment 455576 [details]
Patch for landing
Committed r291778 (248806@main): <https://commits.webkit.org/248806@main> All reviewed patches have been landed. Closing bug and clearing flags on attachment 455576 [details]. Re-opening for pull request https://github.com/WebKit/WebKit/pull/872 Test gardening commit r294603 (250829@main): <https://commits.webkit.org/250829@main> Reviewed commits have been landed. Closing PR #872 and removing active labels. |