I have an odd problem with 2 animated gifs. While testing them, the browser crashed. I am only putting the platform as PC, but its probably all. To reproduce the problem go to http://jazz/users/pnunn/publish/ and click on evilpeng.gif and trousers.gif and go back and forth between the 2 images. The problem with the images is the igf gathered_data field gets larger than the maximum read ahead allowed (now set at 60K). here's the call stack at crash: ImageConsumer::SetKeepPumpingData(nsIChannel * 0x02ffab90, nsISupports * 0x00000000) line 128 + 15 bytes ImageConsumer::OnStopRequest(ImageConsumer * const 0x02f62680, nsIChannel * 0x02ffab90, nsISupports * 0x00000000, unsigned int 0, const unsigned short * 0x0012fe6c) line 387 ImageConsumer::KeepPumpingStream(nsITimer * 0x02fcf030, void * 0x02f62680) line 355 TimerImpl::Fire(unsigned long 2543627) line 308 + 17 bytes TimerImpl::ProcessTimeouts(unsigned long 2543627) line 187 FireTimeout(HWND__ * 0x00000000, unsigned int 275, unsigned int 16514, unsigned long 2543627) line 101 + 9 bytes USER32! 77e7185c() main(int 1, char * * 0x00ae1b90) line 133 + 11 bytes mainCRTStartup() line 338 + 17 bytes
Pam: You should probably own this. It sounds squarely in the imagelib camp and doesn't really have anything to do with necko. Also, Gagan's swamped.
A fix for bug#6525 looks like it fixed this bug. but...this still worries me. An error condition on decoding might not be handled properly when there is a timer involved. I'm leaving this bug open with a target of m15 to remind myself to study this more.
fixed by previous checkins.
Bulk move of all Necko (to be deleted component) bugs to new Networking component.
verified: NT 2000012520