Go to: http://www.space.com/spaceimagined/tv/fifth_trek_000405.html Press <down> or use the "scroll down thing" to go to the botton of the page. Expected Results: To see the end of the page. Actual Results: Mozillas *gone*, it doesn't crash, it simply *goes* away Reproducibility: Allways Additional Information: It doesn't appen on www.slashdot.org, nor on www.cnn.com nor on www.sapce.com. Build ID: 2000040908

On M14/Windows 98, I have no problems. But if I scroll down on 2000040815/Windows 95, it crashes and takes down the entire system. BSOD with fatal exception OE at address 0237:BFF9A25B, and then it brings up one of those grey-bordered white boxes, reading, GPF(I think?) in module <unknown> at address Bff7:00000013, with one button: CANCEL. Then I have to reboot. I've done this twice already.

*** Bug 35278 has been marked as a duplicate of this bug. ***

confirmed using m14 (build ID: 2000022820) on win98.

Rather than Mozilla just "disappearing" for me, I got a GPF when scrolling down near the bottom. Details of the GPF: MOZILLA caused an invalid page fault in module KERNEL32.DLL at 0167:bff87ede. Registers: EAX=c00309c4 CS=0167 EIP=bff87ede EFLGS=00010212 EBX=0068f3a8 SS=016f ESP=0058ff94 EBP=00590000 ECX=bc015900 DS=016f ESI=81783a90 FS=0f97 EDX=bff76855 ES=016f EDI=005901e4 GS=0000 Bytes at CS:EIP: 53 56 57 8b 30 83 7d 10 01 8b 4e 38 89 4d f8 75 Stack dump: This might be a problem with the source or something that's being displayed, since it only happens on that page at a certain point. Having trouble coming up with a better component to attribute this to. Changing OS and Platform to all.

I tried it again using the mousewheel and going much slower. The error seems to kick in right as the bottom of that cyan-colored table comes into view. I also got a different GPF this time, hope this helps somewhat: MOZILLA executed an invalid instruction in module <unknown> at 0000:00000017. Registers: EAX=0068f638 CS=0167 EIP=00000017 EFLGS=00010207 EBX=0068f6c8 SS=016f ESP=00590098 EBP=005900c0 ECX=44015900 DS=016f ESI=8178bb44 FS=12f7 EDX=bff76855 ES=016f EDI=0059016c GS=0000 Bytes at CS:EIP: f0 df 94 00 f0 53 ff 00 f0 00 00 63 05 28 00 56 Stack dump: 0059009c 0000016f bff76849 0059016c 0068f6c8 00590188 00590144 00590280 bff76855 0068f6c8 00590154 bff87fd5 0059016c 0068f6c8 00590188 00590144

Confirmed with 2000-04-09-08-M15 on WinNT; Mozilla crashes so quickly that neither Talkback nor Dr. Watson notice. Memory used by Mozilla is reported as available by Task Manager after the crash. This could be un-neighbourly in Win9x-land, where apps run in a shared memory space -- Mozilla's memory could not have been properly freed. The comments from danielpeng@bigfoot.com 2000-04-09 regarding Win95 seem to bear this out. Over several trials, the crash always happened after pageing down the third or fourth time.

The crash does come as the bottom edge of the cyan table is just about to become visible. Inspecting that table's HTML, the only thing that looked odd was the values for width, cellpadding, and size attributes, which had an extraneous space inside the quote marks: " 191". Fixing that made no difference, though.

Tried it under Linux using build ID 2000041316, didn't crash the browser, it appears to be a Windows only bug. I'll take a look again next time I've got a 98 or NT machine available.

Crashed Windows 98 using Mozilla nightly 2000041308 MOZILLA caused an invalid page fault in module KERNEL32.DLL at 0187:bff87ede. Registers: EAX=c00308b0 CS=0187 EIP=bff87ede EFLGS=00010212 EBX=0068f314 SS=018f ESP=0058ffdc EBP=00590048 ECX=005901fc DS=018f ESI=8198f348 FS=8bbf EDX=bff76855 ES=018f EDI=00590224 GS=0000 Bytes at CS:EIP: 53 56 57 8b 30 83 7d 10 01 8b 4e 38 89 4d f8 75 Stack dump:

are any of you still seeing this in current nighty builds?

No crash with 2000-04-22-08-M16 on WinNT; page displays fine.

WORKSFORME on 20000428 W95. jmbelo@bes.pt - are you still seeing this? Gerv

jmbelo@bes.pt - are you still seeing this? yes

I am unalbe to reproduce the crash with 050808 build under NT. The page does scroll unusually slow for me though. There is all kinds of <layer> stuff going on at that page I think. That may be the problem. Over to Layout for a look.

It does not crash with 5/9/2000 build on WINNT, but scroll performance is slow. I think this is another case of long tables painting too much when scrolling. I tried both gfx and native scrollbars (Both perform about the same.)

It doesn;t happen with build 2000041805 (M15 milestone) Not with keydown, pagedown or the scroll thingy.. I'll be downloading the nightly build then (right after I automate downloading it..hmmm :)

Adding nsbeta2 keyword since table scrolling performance is a known performance bottleneck.

[nsbeta2-]

Removing nsbeta2, crash keywords, adding qawanted.

THis still worksforme with mozilla build 071108 on NT and 98.

Adding crash keyword

Adding nsbeta2 keyword to bugs with nsbeta2 triage value in status field so the queries don't get screwed up

I tried this with build 2000081008 in w2k and all was fixed.

I didn't see it in build 2000081008 running windows 2K.

Seem to be just fine now, build 20000816 on NT

OK, I've not seen this bug recently on NT 4.0 (latest build tested 2000081508). Also no crash on N6 PR2 and M17. It's time to squash the bug. Marking FIXED as I've seen the bug before but can't reproduce it now so something somewhere must have changed and fixed it!

Marking verified pere last comments.