As I mentioned to darin today, a couple of times in the past few weeks, while running the page loading test on win98, a GPF dialog and talkback came up while the test was run. The bizarre thing is that the test continued to run, and completed with no apparent degradation in loading times, despite the fact that this crash had occurred on one of the threads. The UI was still usable, and content was downloaded and displayed correctly the whole time. Reproducing this will likely be difficult, but perhaps there is a clue in the stack trace from talkback. This happened on win98 for Build ID 2001052313 and Build ID 2001060606. (Note: another similar crash that paw and I were mentioning to darin (meta refresh overnight to home.netscape.com or www.nytimes.com) turns out to have a different stack, and is a thread crash on the nsSocketTransport; separate bug to follow). MSVCRT.DLL + 0x1648 (0x78001648) CopyChars1To1 [d:\builds\seamonkey\mozilla\string\obsolete\bufferRoutines.h, line 245] nsStr::StrAppend [d:\builds\seamonkey\mozilla\string\obsolete\nsStr.cpp, line 185] nsStr::GrowCapacity [d:\builds\seamonkey\mozilla\string\obsolete\nsStr.cpp, line 137] nsCString::SetCapacity [d:\builds\seamonkey\mozilla\string\obsolete\nsString.cpp, line 214] nsString::SetLength [d:\builds\seamonkey\mozilla\string\obsolete\nsString2.cpp, line 198] nsACString::do_AssignFromReadable [d:\builds\seamonkey\mozilla\string\src\nsAString.cpp, line 713] nsAString::AssignFromReadable [d:\builds\seamonkey\mozilla\string\src\nsAString.cpp, line 658] nsLocalFile::ResolveAndStat [d:\builds\seamonkey\mozilla\xpcom\io\nsLocalFileWin.cpp, line 522] nsLocalFile::Exists [d:\builds\seamonkey\mozilla\xpcom\io\nsLocalFileWin.cpp, line 1543] nsFileIO::Open [d:\builds\seamonkey\mozilla\netwerk\base\src\nsFileStreams.cpp, line 145] nsFileTransport::Process [d:\builds\seamonkey\mozilla\netwerk\base\src\nsFileTransport.cpp, line 643]
This looks like a race condition that I have already fixed. I check the fix into the TRUNK yesterday. Please reopen if you see this problem after todays build.
Sounds good. If I see it again, I will reopen.
VERIFIED: mozilla 0.9.4 and no reports of this. Looked like bug 80013, but I'm not expert here...