If you write Windows kernel mode drivers you really should watch this video. You will be impressed with the work that has gone into the Kernel Mode Driver Framework. This framework abstracts some of the pain points away for driver developers giving them the freedom to concentrate on their algorithms related to device usability.
Kernel Mode Driver Framework Video
About The Author
Eugenia Loli
Ex-programmer, ex-editor in chief at OSNews.com, now a visual artist/filmmaker.
Follow me on Twitter @EugeniaLoli
Any changes to the kernel drivers model is a welcome in windows as the current model corrupts the drivers like crazy as to not be able to trust both ATI and nvidia drivers anymore.
On fresh windows XP, and windows server 2003 installations nvidia drivers crashes the OS and brings the BSOD in a matter of a month, and the cause is nv4_mini.sys file which windows was not able to uninstall or repair as well as nvidia software was not able to keep it clean. On ATI systems crashes are as normal as water.
Of course during the one month period I connected the computer to the internet and played 8 games.
On contrary, Linux complex OpenGL-based games are working with me for a year without any single crash (FC5).
whereas in windows an archade game might trash your graphics subsystem. By the way the part of the graphics driver that stimulate the crash was direct3d not directdraw, all discovered from dxdiag.exe tool.
I liked vista handling of graphics because on XP the system “sniperelite” game crashed immediately after installation and sound was lost, whereas in vista I was able to play the whole game without a heck which really shocked me, because it’s for the first time I saw different NT windows are really different in stability. vista test system was v 5472.
MS try your best and try it quickly or people are gonna look elsewhere soon… Good Luck for all of us.
Sounds like you’ve got bad hardware.
The number of XP/S2K3 systems that I’ve owned, own, and used that maintained/maintain stability for months at a time with an ATI/NVIDIA card is too many to count.
Erm, depends, there have been some *really* crappy drivers from ATI/Nvidia, who seemed to be more concerned with break neck speed rather than stability.
Then again, when I was running Windows XP, I had a Matrox video card; which was rock solid.
Get WHQL’ed drivers, and it’s almost always guaranteed that they’ll be stable. I have yet to have stability problems with 9/10 ATI/NVIDIA drivers that I try.
All the drivers I have used were WHCL certified; All of them were crashable after a while buy the games you play. Of course no games no unstability.
My hardware are all working reliably because the same hardware never gave me a problem with the 2nd OS Fedora Core 5. So no problems from hardware at all.
I’m a hardware reviewer. I deal with video cards on a daily basis. I benchmark them. I test new games. I deal with everything from the highest-end hardware to the lowest-end.
I have yet to find a combination of hardware that *consistently* crashes in the driver when playing any game. Trust me on this when I say that something is up with your hardware. The fact that things didn’t crash in Linux is not a certifiable test that your hardware is not at fault.
If ATI/NVIDIA released drivers that crashed almost consistently when playing any game, they wouldn’t be in the video card business long.
“If ATI/NVIDIA released drivers that crashed almost consistently when playing any game, they wouldn’t be in the video card business long.”
Actually, they will be in the business no matter what because there is no competition to PC gaming from Apple or Linux.
And the system doesn’t crash consistently, but only from some games (that exposes the weakness of the driver model of current windows) and when you contact both manufacturers nvidia and ATI they will do two things:
1. Blame you for installing inferior game or beta game.
2. Issue a fix in the next driver version (read the what’s new file in a newly issued driver from nvidia, it would say as some of the features it is a bug fix for Doom3 or a HL2 or….
3. Dodge from the problem and tell you that your system hardware or windows is malfuntioning, and then go in the process of proofing proofing the opposite. (over the phone supporter: But, Sir are you an expert? Me: Yes supporter: Yeh, right! give me a break!
I pray for this situation to end, and yes I believe you are experienced but maybe in the hardware and software but not with the pain of going through these from the eyes of a normal unexperienced user.
thanks .. and keep posting
List your hardware, and what games crash on it.
Heh, for how *long* do you run these tests?
I had a bad SCSI host adapter and SCSI CD writer once, guess what, XP never ever ‘told’ me anything about errors, linux did. Now which do you prever?
Of course, it’s likely that games will stress the system more so xp will likely turn up some problems, then again linux is all about openess and information
Edited 2006-08-16 09:16
It depends on the test. For stability testing we usually run 3DMark06’s two HDR tests overnight in a loop.
Regarding your Linux experience, what I find funny is that the xorg drivers, which are based on reverse engineering, perform certain tasks better than the binary drivers and are more stable – hence my call for the drivers, or atleast the core 2D and 3D to be opensourced (leave the DRM, TV out etc. part of the closed source driver).
Give out stable and reliable the current ones are, without all the fancy driver information, imagine how go they will be if all the information was made available – and not only would us *NIX (Linux/*BSD/Solaris/etc) users would benefit but end users on Windows too; their bugs would be promptly addressed rather than placed in the queue, and fixed based on whether the company actually gives a stuff or not.
Either you have made this up or your computer is broken.
XP systems simply do not crash all the time.
And I have used both ATI and nVidia.
>Either you have made this up or your computer is broken.
XP systems simply do not crash all the time.
And I have used both ATI and nVidia.<
Well, I hate to say it, but that statement is not true, I have installed Windows XP Pro SP2 on newer hardware whth many different drivers for my various gfx cards (mostly ati and nvidia) and I get random systom freezes often, maybe twice a week. I have no viruses as this computer is virtually never connected to the net and this is a legit paid for copy of xp sp2.
I even contacted microsoft and told them my media might have been bad, they sent me a new copy of xp pro sp2 and i was greeted with the same results.
I am less then impressed.
Prior to messing with contacting microsoft, i updated to the newest drivers on virtually every component i could get updates for, and when that didin’t work i tried reverting to old drivers that i had never used……. still didn’t work.
my hardware works with virtually every other OS that i try it with…….. you do the math.
“Either you have made this up or your computer is broken.”
My computer is not broken I am sure of that, checked memory system with MemTest86 and Windows Memory diagnostics, no problems at 2 complete cycles.
CPU heating is OK at 55 C up to 58 with full saturation, and the heatsink of CPU and Fan is one of the best from Zalman costing 80$.
Mobo temperature is OK too at 35 C
Graphics Card is 1 year old 6600 GT and its fan is working and there is another fan on PCI lower to it to cool it even more ( a turbo fan that occupy 2 PCI slots)
Case Temperature is very low because the case is open and I have 9 fans working for it.
The PSU is 650 Watt and from Antec which is the best money can buy.
My hardware is fine for sure, back to software, I have discovered that the file nv4_mini.sys is the one the BSOD shows and It said FAULT_PAGE…. error 0x00000050.
Now by searching for the file, I found it to be at
1. C:windowsServicePackFilesi386 carring a version of 5673
2. c:windowssystem32drivers carrying a version of 91.31
3. C:windowssystem32ReInstallBackups 009DriverFiles carrying a version of 8421.
Now to carry the job of Microsoft and nvidia for repair, the trick that I did and solved the problem was to
1. Uninstall the nvidia drivers
2. reboot
3. Delete all the previously mentioned files + nv4_vga.sys from c:windowssystem32 (By the way nvidia and MS uninstaller failed to remove them! how bad)
4. reinstall the driver again
5. reboot
6. run Direct3D from dxdiag.exe diagnostic tool to test the stability of the driver
7. Everything is working fine now
All the problems above happened from installing an archade game that was not good (actually it was game #9 to be installed on the system).
But 3 days more after trying MS flightsimulatorX beta the screen cursor froze in the middle of the screen and another cursor appeard after enabling cursor trailing from “control mouse”. But this is not a big deal because it will disappear if I restart the system.
The OSs that caused this problem on 2 systems were windows server 2003 and windows XP both patched with the latest SPs.
So to protect my time I decided to image the system without nvidia drivers installed then install the drivers after imaging so if such things happen afterwards I can restore it to its healthy time, otherwise windows server 2003 didn’t make for me much headaches like windows XP.
My computer is not broken I am sure of that, checked memory system with MemTest86 and Windows Memory diagnostics, no problems at 2 complete cycles.
did you check if windows Xp that came with you computer is not pirated that can be causing the problems.or may be
a Microsoft’s critical update is conflicting with something in xp. I had a time where right after i installed a Microsoft’s critical update, can not remember which one, windows would start to run slow,lock up for a few seconds. apps would take along time to open up.
Edited 2006-08-16 11:06
“did you check if windows Xp that came with you computer is not pirated that can be causing the problems”
No, it is not pirated version and all system files were digitally signed by Microsoft. And no errors on setuperr.log
Edited 2006-08-16 11:34
You might want to testing your memory for errors (Memtest86) and also it could be your power supply not delivering a strong enough or steady current. I’ve had less than 5 unexpected restarts in the past few years running XP, usually during games, because of my power supply. After replacing it it was solid as a rock.
I also take precautions to minimize possible virus/spyware infections by running SpyBot daily, NOD32 antivirus running in the background (Norton is evil!!), and Windows Defender. I haven’t had a single occurance of spyware or a virus in over a year.
Good advice, I was thinking odds were on ram or power too. Would give them a firm chip re-seating, maybe even a gentle contact cleaning with a soft eraser if not bright and shiny and shuffle the sockets around. A real good
power supply like the antec neo HE would be my next step if the ram tested out.
A RAPID dev kit for drivers sounds like scary prospect. If there is anywhere I want programmers engrossed in what would normally be trivial low level detail and to take their time, it is with drivers. Good hardware + Bad drivers = Bad hardware.
I think this is not the case. Dealing with power management and PnP in Windows is a “bear” as Doron says. It is largely due to the fully asynchronous nature of the Windows kernel, in which you could get IO requests at just about any time, including while your device is in the process of shutting down.
This code is hard to write and really adds no value to the drivers. No one should have to become an expert in the Windows kernel Power subsystem and PnP manager just to write a driver for some simple device. This is one area in which Open Source has an advantage, because right now Windows people have to read tons of (hard to find) docs and do some experiments to find out what IRPs they can get at which stage. In the absence of open inspection into the windows kernel source, Microsoft HAS to make something like this.
I’m curious: how does linux handle PnP and power management? Last I remember it was using user-space daemons to do the device enumeration and setup. And I got the impression that power management was largely handled outside of the kernel. I’d actually love some pointers to text on the structure and design of the Linux I/O subsystem (if it even has such a structure as a building-block).
Just help yourself.
http://www.kernel.org/git/?p=linux/kernel/git/stable/linux-2.6.17.y…
http://www.kernel.org/git/?p=linux/kernel/git/stable/linux-2.6.17.y…
Thanks! Exactly what I wanted!
Well consistent crashing is foreign to me, i have yet to see my WindowsXP BSOD or lock up, even in gaming, with my NVIDIA 9600 things are sweet. And im using middle of the range hardware … no extravagant cooling, nothing …
The only time iv had problem with hardlocks in my life was when i had an early ADM64, and there was a fault in the CPU memory controller, so with one DIMM used, it would pass memtest86 or whatever its called, but with two DIMMs used, even memtest86 locked up, that was major foobar.
Edited 2006-08-16 06:55
All those “on my system”-remarks are somehow pointless. It may always be a random bug. That doesn’t mean that it shouldn’t be fixed, but judging the whole system because of one case is well, not too informative.
If it happens on most systems, this is interesting.
Sorry Virginia, there is no such thing as a random bug. There may however, be a bug in random.
I support alot of windows machines, and i have to say built correctly. they are stable, i have Win NT servers still servering up hundreds of users there files without problems. and NT workstations that users have been using for years without problems.
sure some people seem to have random and odd issues but thats not the norm. and can useally be tracked down to something not OS born.
i likewise, have win2k and winXP machines users have been using for years with no issues. no random crashing, not other sillyness.
if your computer with ANY OS windows, Linux, freeBSD whatever crashs or locksup randomly you have issues.
that could be hardware OR software.
-Nex6