Windows crashes caused by bad drivers

[Patrick]

Now I can rest a little easier:
http://www.wired.com/news/technology/bu ... _tophead_3

[Gomer_X]

So the real problem with Windows is device drivers make the system crash? Per the article Microsoft is working on a tool to analyze driver source code to make sure the driver isn't buggy.

If we can't get source to drivers for Linux, how's Microsoft going to get the source? I guess they could give their scanning tool to the hardwre manufactureres. It's ironic that open source is the only thing that can make Windows work.

I don't have a lot of faith. The only time I ever have problems with my Windows system crashing is when I install the Microsoft certified nVidia drivers for my motherboard.

[dann]

To write drivers for Windows OS don't you need to use the Windows Development tools? And in using the Windows Development tools isn't there something in the license that says MS has the right to view your code? Maybe I am wrong here. I thought I read that somewhere in the EULA.

Maybe I am wrong.

Anyway, I actually read the article and it sounds like they came up with some new test, but is it really revolutionary? What are other OS's using to test drivers?

Reflecting on this I find it interesting that, in the Linux and BSD world there is a lot of emphasis placed on the stability of drivers and make sure "stable" code is shipped with ample testing (to some extent). It seems development of drivers is kept closer to the heart of the OS. Yes, I know you can find drivers out there that are not incorporated into the kernel that any joe can write, but these are clearly understood to be use at your own risk.

Now Windows on the other hand has let the entire world of HW manufcaturers run wild with creating drivers. In the past, and I suspect it still exists, you have issues with drivers overwriting dll's for another piece of software and causing conflict and all that. Now MS has tried to resolve some of these issues with driver signing, but has it made much of a difference? Apparently not according to this article.

Interesting indeed. But again, is it revolutionary?

[Gomer_X]

I know there's work to keep closed drivers out of the Linux kernel for the same reason. Bad drivers cause problems, and without access to the code, kernel hackers can't do anything about it. The driver is the problem, but that doesn't keep people from blaming the kernel.

[Judland]

Maybe I'm wrong here, but wouldn't writing drivers for Microsoft's crap be easier if the code making up the OS was available?

If OS stability is directly the responsibility of the device makers (like MS is claiming... not due to their sloppy programmers of course), then this would explain one of the reasons why Linux will always be a better, more stable OS.... 'cause it's source code is available for the device manufacturers.

[Patrick]

one of the reasons why Linux will always be a better, more stable OS.... 'cause it's source code is available for the device manufacturers.

Unfortunately not all of the manufacturers release drivers for linux. It's up to the linux community to reverse engineer it. Wireless support was a major headache until ndiswrapper was developed. Everytime a vendor came out with a new version of a card (with a new chipset) it wouldn't work in linux. It's slowly getting better.

[treehead]

<snip>Per the article Microsoft is working on a tool to analyze driver source code to make sure the driver isn't buggy.<snip>

;think about that for a minute.

;treehead

[treehead]

;you all hit it right on the head.

... Now MS has tried to resolve some of these issues with driver signing, but has it made much of a difference? Apparently not according to this article.

Interesting indeed. But again, is it revolutionary?

;exactly. isn't this what MS tried to accomplish through windows logo testing? even if it works--and we'd all agree that i won't, it can't, but even if it does--what makes MS think that anyone is more likely to submit their code to the scrutiny of their buggy software than they were to their logo program for the windows HCL?

"...we're building tools that can do actual proof about the software and how it works in order to guarantee the reliability."

;any programmer worth his salt will tell you this is impossible. it's why linux is so rock solid.

Maybe I'm wrong here, but wouldn't writing drivers for Microsoft's crap be easier if the code making up the OS was available?

;you're not wrong. you're dead on. it (yet again) comes down to the fact that free software is free and anything else isn't. it's chapter one of the stallman story--literally. you simply cannot attempt to paint a room blindfolded. you're going to screw up somewhere, and even if you don't, how do you know it's the right color when you're done?

;treehead