This test appears to have been crashing on release bots since it was added with https://commits.webkit.org/251478@main TestWebKitAPI.WKContentRuleListStoreTest.CrossOriginCookieBlocking 2022-06-15 13:06:59.429 com.apple.WebKit.Networking.Development[18874:37711573] DiskCookieStorage changing policy from 2 to 0, cookie file: file:///Users/buildbot/Library/Developer/CoreSimulator/Devices/98B5DF45-8DBE-439A-A3A4-E96932F8FE06/data/Library/Cookies/Cookies.binarycookies Child process terminated with signal 5: Trace/BPT trap [18873] pas panic: deallocation did fail at 0x2: Large heap did not find object https://results.webkit.org/?suite=api-tests&test=TestWebKitAPI.WKContentRuleListStoreTest.CrossOriginCookieBlocking
<rdar://problem/95247117>
Actually, this is arm64 specific.
Strange. This has something to do with this line: return HTTPResponse({ { "Set-Cookie"_s, "testCookie=42; Path=/; SameSite=None; Secure"_s } }, "<script>alert('hi')</script>"_s); If I write it out by making a HashMap<String, String>, add to it, then move it to the HTTPResponse constructor it works fine. So this isn't a big deal. I'd like to understand a bit more about this then fix it.
Created attachment 460300 [details] Patch
Comment on attachment 460300 [details] Patch r=me
Committed r295642 (251647@main): <https://commits.webkit.org/251647@main> All reviewed patches have been landed. Closing bug and clearing flags on attachment 460300 [details].
I have disabled this test while this issue is investigated.
Still consistently crashing on Release History: https://results.webkit.org/?suite=api-tests&test=TestWebKitAPI.WKContentRuleListStoreTest.CrossOriginCookieBlocking&limit=50000
Reopen while under investigation.
Test gardening commit 251889@main (469551ebd9f7): <https://commits.webkit.org/251889@main> Reviewed commits have been landed. Closing PR #1833 and removing active labels.
Created attachment 460724 [details] Patch
Comment on attachment 460724 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=460724&action=review > Tools/TestWebKitAPI/Tests/WebKitCocoa/WKContentExtensionStore.mm:217 > + // FIXME: This is exactly equivalent code, but the code below crashes on release builds on arm64e. Should the comment say arm64 now?
r=me too
Comment on attachment 460724 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=460724&action=review >> Tools/TestWebKitAPI/Tests/WebKitCocoa/WKContentExtensionStore.mm:217 >> + // FIXME: This is exactly equivalent code, but the code below crashes on release builds on arm64e. > > Should the comment say arm64 now? I think this is an arm64e bug, but to be conservative I'm enabling the workaround on all arm64 CPUs.
Committed 252199@main (0d69815a681d): <https://commits.webkit.org/252199@main> All reviewed patches have been landed. Closing bug and clearing flags on attachment 460724 [details].
Reopening to attach new patch.
Created attachment 460735 [details] Patch
Committed 252239@main (051738d08e49): <https://commits.webkit.org/252239@main> All reviewed patches have been landed. Closing bug and clearing flags on attachment 460735 [details].