![]() ![]() ![]() Firefox does not crash at startup with this build, which surprises me, because I expect that it is a *non-PGO* build *with* my fix. So, I tried the builds and I'm really puzzled at this is a *non-PGO* build *without* my fix. This makes me very worried about any version-specific blocklist perhaps causing this failure mode. Now this is still a little broken, but I basically see in the assembly that in some cases GetFileVersionInfoSizeW itself calls LoadLibraryEx. > xul.dll!patched_LdrLoadDll(wchar_t * filePath, unsigned long * flags, _UNICODE_STRING * moduleFileName, void * * handle) + 0x42486 + 0x32 + 0x31 bytes Xul.dll!patched_LdrLoadDll(wchar_t * filePath, unsigned long * flags, _UNICODE_STRING * moduleFileName, void * * handle) + 0x42486 + 0x32 + 0x31 bytes Without fsecure the stack loop is pretty straightforward: Xul.dll!patched_LdrLoadDll(wchar_t * filePath, unsigned long * flags, _UNICODE_STRING * moduleFileName, void * * handle) + 0x42486 bytesĜ++ Do we know if this crash only happens with F-secure installed also?īelow here the stack is unreliable, but it may involve a loop of these + 0xc + 0xc + 0x31 bytes I just got off 12 hours of air travel and 4 days of jetlag, and the thought of me in the code review path of a release is terrifying right now so let's go through this list again and be extra sure, please.įirefox appears to be crashing with a stack overflow exception:įirst-chance exception at 0x7c90e8e5 (ntdll.dll) in firefox.exe: 0xC00000FD: Stack + 0xaf + 0x31624 + 0x9b48 + 0x93 + 0x1c + 0x27 + 0x1f + 0x17d + 0x1cb + 0x18 + 0x22e bytesįshook32 is part of F-Secure. Still, someone should double check my math, BECAUSE: h file) on FF8, but Kev's patch on FF9 and up. By my reckoning that makes Nick right, and we'd want his patch (against the. cpp in bug 648581, which landed September 19. TIL that Ehsan moved the blocklist from a. We don't seem quite clear yet on whether our blocklist.xml tests have affirmatively ruled out that option, there's back and forth in this bug about how to test it - we should be sure I'd normally insert a thing about how we don't typically DLL blocklist extensions since they can be blocked with the addon blocklist which is more user-friendly, but from the sounds of it RoboForm might be hacking around that Marcia uses "7.5.2.0" in comment 2 and this patch uses 7.6.2.0 - is that deliberate? (Judging from the rest of comment 2 I'd assume she just typo'd, but let's be sure) I haven't tested it so that would need to happen before we landed/shipped it Roboform dll block for toolkit/xre/nsWindowsDllBlocklist.cpp Robohelp and Firefox should be working fine now.Comment on attachment 574474 Download the roboform-firefox.xpi file.Next, I checked in the Add/Remove Programs list in the Control Panel and there was nothing there either.) (Note that I checked my extensions in Firefox, but there was no RoboForm extension. Read the instructions for RoboForm for Firefox, the second set of instructions from the top.Open Internet Explorer (don't worry it won't kill you for this one time) and go to the Firefox page.The trouble is that I couldn't switch to the RoboForm Web site to download the latest fix in Firefox, so I was forced to improvise by opening Internet Explorer, downloading the file and installing it in Firefox. The folks at RoboForm must have heard about this issue and posted a fix on their Web site. It saves you the trouble of entering this data yourself each time you come to a Web form, which for me is frequently, and unlike some of these programs, the data is stored locally and RoboForm does not attempt to phone home. ![]() RoboForm is a nifty little program that stores your passwords (something that I know is not necessary with Firefox, but is still handy, especially with multiple users) and other forms information such as name, address, phone number, credit card, etc. I installed the latest version of the Firefox browser yesterday, 1.06, and when I restarted my browser, I couldn't move beyond my Home page without triggering a RoboForm error, which then forced Firefox to shut down. ![]()
0 Comments
Leave a Reply. |