More Linux Kernel Patches To Mimic Windows

[Sidicas]

You know that there are servers being sold with Coreboot, right? I have never heard of somebody complaining Linux costs more than windows.

No, I didn't know... But I do get bombarded by Microsoft's sales department regularly saying about how much cheaper Windows is than Linux for X,Y,Z reasons.. Then they go and start offering me free copies of Windows Server to just "try it out".. You know how it goes.. It's not as bad as it used to be, but Microsoft is still pushing their Windows Server software as the best solution to everything under the sun..

[Drago]

No, I didn't know... But I do get bombarded by Microsoft's sales department regularly saying about how much cheaper Windows is than Linux for X,Y,Z reasons.. Then they go and start offering me free copies of Windows Server to just "try it out".. You know how it goes.. It's not as bad as it used to be, but Microsoft is still pushing their Windows Server software as the best solution to everything under the sun..

Well, that's called FUD. Yes, ms are so much better at marketing.

[AlbertP]

I AM surprised the e1000 driver is being blacklisted. I believe that is an intel gigabit chip. Usually they are so good about these things...

There is more Intel hardware out that doesn't properly support ASPM. The Intel 3945 wifi driver also contains a work-around for an ASPM-related hardware bug.
On most Intel hardware the presence of ASPM is correctly detected, at least when ASPM is set up properly in the BIOS. When you use pcie_aspm=force because of the famous 2.6.38+ ASPM problems, you will run into problems without a patch to disable ASPM completely.
By the way, I think that he means e1000e instead of e1000, as e1000e is a driver for PCIe hardware.

[DanL]

There's indeed hardware out there that does not fully comply with the PCIe specs of which ASPM is a part, so blacklisting them is a start. Likewise hardware that does play nice with ASPM but doesn't advertise supporting it should be able to work with ASPM
Shame HW makers toe the Microsoft line when designing their devices. Why can't they make hardware that will work with ALL OS's not just Windows?

I think you have it backwards. Why can't MS actually follow the hardware standard as defined so that if it works with Windows, it should work with other OS's?
A) They don't know how to follow and/or don't care about the standard
B) They purposefully don't follow the standard so that other OS's break and look bad on hardware only tested with Windows
C) A combo of the first two.

[liam]

There is more Intel hardware out that doesn't properly support ASPM. The Intel 3945 wifi driver also contains a work-around for an ASPM-related hardware bug.
On most Intel hardware the presence of ASPM is correctly detected, at least when ASPM is set up properly in the BIOS. When you use pcie_aspm=force because of the famous 2.6.38+ ASPM problems, you will run into problems without a patch to disable ASPM completely.
By the way, I think that he means e1000e instead of e1000, as e1000e is a driver for PCIe hardware.

Wow.
I had no idea. I imagined that since Intel is a member of the PCISIG, and a major producer of system components that they would be on top of this.
I imagine the cause of these problems was this http://www.pcisig.com/specifications...2009-08-20.pdf. Once they made it optional they opened the door to hell (via the hot air coming from everyones pc).

[locovaca]

I think you have it backwards. Why can't MS actually follow the hardware standard as defined so that if it works with Windows, it should work with other OS's?
A) They don't know how to follow and/or don't care about the standard
B) They purposefully don't follow the standard so that other OS's break and look bad on hardware only tested with Windows
C) A combo of the first two.

D) Microsoft doesn't write a majority of the drivers, the hardware manufacturers.

[DanL]

D) Microsoft doesn't write a majority of the drivers, the hardware manufacturers.

You don't think MS writes the Windows "kernel" code to deal with ACPI and ASPM?

[DeepDayze]

I think you have it backwards. Why can't MS actually follow the hardware standard as defined so that if it works with Windows, it should work with other OS's?
A) They don't know how to follow and/or don't care about the standard
B) They purposefully don't follow the standard so that other OS's break and look bad on hardware only tested with Windows
C) A combo of the first two.

Yes that's right my bad..mornings aren't my thing

MS doesn't like any standards but their own...and didn't they help draft the PCIe specs to begin with? The hardware standards are out there for all OS's to work with and not just Windows

[DeepDayze]

You don't think MS writes the Windows "kernel" code to deal with ACPI and ASPM?

They write the API/ABI that driver writers must use when interacting with the Windows kernel

[agd5f]

Two example, for my new laptop ThinkPad x120e, the hardware is fine, but BIOS causes Linux to hang at boot. This only happens if I entered the BIOS setup, or booting windows, and then restart to Linux. I have to turn it off, and then back on to boot in Linux. I mainly use it on my desk with external display,keyboard and mouse. The freaking thing doesn't have Wake on keyboard press, although it has one powered while off usb port. The lid have to stay opened all the time.
These two examples I believe can be easily fixed, if one had the BIOS sources.

These could just as easily be OS bugs rather than BIOS bugs.

Better yet why not provide documentation to the Linux developers who'll then gladly write a driver for the device and maintain it.

Sadly this isn't the case. It takes a very skilled developer to write device drivers especially for complex hardware. In most cases where documentation is available no one steps up to write a driver unless the manufacturer does. Even when the manufacturer does provide open source drivers, in many cases the driver isn't accepted due to coding style, etc.

But I do like the idea of keeping a list of manufs. that have been naughty or nice.. Just about everything works under Linux whether the manuf. supports it or not, the real question is about how well it works...

Everything can be fine and dandy for months and then a company comes up with a stupid idea like nVidia Optimus, and then it just doesn't work right and probably won't for years and people need to know before they go buying hardware.. I've been using Linux for almost a decade now and I *almost* made the mistake of buying an nVidia Optimus laptop, one of the ones without the hardware switch... Simply because I wasn't following laptop hardware that closely at the time..

There's nothing secret about Optimus. The only reason it's not supported on Linux is that it requires piles and piles of work in the X server to make it work which no manufacturer can justify considering the size of the Linux desktop market share. Anyone with a decent knowledge of X and GPUs could write the code, but so far few people have stepped up. Microsoft wrote a lot of the graphics infrastructure in windows that makes this kind of thing possible.

You don't think MS writes the Windows "kernel" code to deal with ACPI and ASPM?

Microsoft didn't support ASPM at a high level until Windows 7 IIRC. The bug only exists because Linux was an early adopter of ASPM. The ASPM "regression" fix was just to leave the ASPM registers as set by the bios which is what windows probably did in versions prior to those that supported ASPM. A lot of bios vendors enabled ASPM so it would work as long as the OS didn't mess with it. It hardly seems like the hw manufacturers' fault.