One line of html code to crash Winblows

[Aloone_Jonez]

Firstly all drivers need to run in ring 0 as NT, Linux, BSD or any kernel that's in the slightest bit decent doesn't allow anything other than ring 0 direct access to the hardware.

Secondly it can't be a graphics driver problem because it affects a verity of graphics cards I for example have a S3 pro savage and have exactly the same symptoms when I click on the link.

It must be a major flaw in the NT kernel or whatever NT uses as instead of the X Window system, there is simply no other way this could affect so many different hardware configurations.

I hope Microsoft do a decent job of fixing it.

Fuck it, I hope everyone gets pissed off with Windows and finds a suitable alternative like Linux, BSD or even BeOS for fuck sake and the industry start to support that OS, but that'll never happen anyway/

[toadlife]

Fuck it, I hope everyone gets pissed off with Windows and finds a suitable alternative like Linux, BSD or even BeOS for fuck sake and the industry start to support that OS, but that'll never happen anyway/

Probbaly won't happen, since the vast majority of users will never even know this flaw existed.

[Aloone_Jonez]

I know but it was wishful thinking.

[Jenda]

Hehe, crashed mine... well... my uncle's...

[WMD]

It must be a major flaw in the NT kernel or whatever NT uses as instead of the X Window system,

Those things you mention are one and the same.  The NT kernel is the "X Window system," in that it draws the base graphics.  You still need explorer.exe to do anything useful...but still, this flaw must be in the kernel.

[mobrien_12]

Win 98, Mozilla 1.7.8, ATI Rage Pro Turbo, nothing happens.

[Orethrius]

Strictly out of curiosity: did anyone else think to close everything else first before clicking that link?  Here are my traceback questions to those of you who experienced a crash:

(1)  Which Operating System (full version here, plus SP if Windows) did you test?
(2)  Were the patches up-to-date?
(3)  Which browser were you using?
(4)  What kind of graphics card do you have?
(5)  What were you doing when the crash occured?

I believe this is a kernel issue as well, but I may be wrong so it's a good idea to trace it back as far as possible.

[MrX]

well, i don't think its ever going to load in FF for beos. it's probly just not  'stable enough' i guess but nothing crashes, nothing slows, no memory gets eaten up, but the scroll bars adjust as they should, even tho it still shows white after 2 minutes.

note- i was able to make my FF crash on BeOS by making a html file with 10000 pop ups on it. i would like to see what happens on winshit.

[Lead Head]

My PC doens't even blue screen just freezes. But my Dell 5100 laptop with a mobility radeon 7500 blue screens and said there was a problem with the ATI driver

[muzzy]

Doesn't crash my w2k3 box, and i don't have w2k test box available at the moment. It does however lag down the w2k3 significantly, a lot more than it should.

It doesn't really eat memory, just cpu. Huge loads of cpu. The whole thing makes me think this is an issue with GDI. Stuff just doesn't make sense here.

It's a serious issue anyway, I'll look into it a little.

[Lead Head]

I am going to test it on my old 650 MHz Athlon with win 98. Muzzy what kind of CPU are you using?

[muzzy]

Oh, I have a rather modern system. Anyway, I finally took time to trace through it, and it seems a call to StrechBlt is done everytime the window is drawn, and that lags the system down on my w2k3.

So, the image itself is 400x300, but since the page requests it as huge, the system will ask GDI to resize it to 9999999x9999999. This happens everytime the window is redrawn. For reasonable resizes, this would be fine, however here it is not.

I'd have to look at the actual implementation of the thing, but right now it looks like it's issue with GDI, and perhaps with driver implementation of the stretching support as well.

[MrX]

you are so smart muzzy.

[Jenda]

Let's all bow down to Muzzy!!!

[toadlife]

Thanks for the insight Muzzy,

Could it be that a bug in GDI is causing an infinte loop situation, which "tricks" the Windows kernel into thinking the video driver is in an infinite loop?