Foxconn Mobos deliberately sabotaging Linux ACPI - Blatant false advertising

[WanderingKnight]

http://ubuntuforums.org/showthread.php?t=869249

Short summary: A Linux user tried to find out why his ACPI-enabled Linux kernel would not work correctly on his new Foxconn motherboard. He disassembled the BIOS, only to find specific code targeted at Linux that pointed to a badly written DSDT table, which caused all sorts of weird errors, warnings and freezes. Not only that, but when making Linux identify itself as Windows, the Windows code section contained bogus warnings that Windows ignores but make Linux crash. The motherboard, on the other hand, is advertised as fully ACPI compliant. The user provided the company with a fix, but they didn't accept it.

I'm posting this on a thread of its own to get the notice out. Whether you're a Windows or Linux user, everyone should agree that this is a shoddy (and illegal) market practice. The most worrying part is that Foxconn is one of the largest manufacturers in the world, even if they're not focused on their own brand of mobos.

[Vexx]

This is one of those "wtf???" moments in watching the hi-tech business. It sounds as if they practiced poor encapsulation coding as well.

I think a lot of people forget that businesses often don't make decisions based on rational analysis. Way too often they're actually made by someone who is all about waving their cohoneys around, who played golf with whom, or who got a special gift.

[Claies]

Such is business. >_>

I, for one, trust only ASUS and Gigabyte.

[Kovensky]

Foxconn just entered my black list...

It appears that ASUS is the only rational mobo manufacturer...

[hobbes_fan]

I wonder if dell is aware of this. Foxconn and ECS I think still are the main OEM supplier for them. Then again this may not be an issue as Dell have a custom BIOS. Now I wouldn't bother with a fix, it's too much of a pain worst case scenario = bricking the motherboard and it's just not worth the hassle (anyone who's messed up a Bios update will understand).

Now granted the poster is pissed off there's no point abusing customer service reps. 90%of the time they have NFI. It's pretty clear that he/she is not someone up to speed and just regurgitating basic details. Take the advice - return it, spread the word. Now Foxconn has failed here big time and will pay dearly, but considering it is using an older chipset it just maybe its a POS board, there are a lot of foxconn boards out there this is the only board I know of with this issue ( though you would think that they'd spend a bit more time with bios updates etc etc.) What other Foxconn boards are affected? Now it's a bit quick to complain of a massive antitrust conspiracy when we have 1 incident. If it is a pattern of behaviour, then that's a different story. It could just be a poorly written bios. From a business perspective I wouldn't just be accepting random code off the internet, I'd be verifying the issue and conducting my own investigation. Also Microsoft certification is once again proving its not worth a cup of spit. ACPI is an Intel standard too not just MS.

But you would think corporations would learn how to deal with knowledgeable people after the Creative 3rd party driver fiasco.

Asus? Their AM2 to AM2+ bios support is by far the worst of the big three, (gigabyte, MSI and Asus.) somewhat soured on them. Honestly they're all pretty much the same in terms of corporate level assholery.

[Vexx]

Lately, I've taken to using as much ASUS as possible in my home-kit-builds: mobo, video, etc. But I'll be the first to admit the ASUS bios interface is messy and has poor knobbage on the AMD side .... I've not been using Intel cpu in a few years so can't comment on the bios on those mobos.

yeah, the fix is nice if one is already stuck with a Foxconn board..... but Foxconn actually has a fairly well documented history of being dilweeds when it comes to Linux. The real issue is that they're claiming to meet ACPI specs ... and in the immortal words, "It's a trap."

[Epyon9283]

I don't want to legitimize what Foxconn did but they specifically state that they don't support Linux. Right then and there I would have tried to obtain a refund. Nuts to them.

If they passed a Microsoft ACPI compliance test that doesn't necessarily mean that their board actually complies with the published specs. MS plays fast and loose with standards. Look at IE.

None of the computers I've ever owned (all with asus, abit, or gigabyte mobos) had properly working sleep in Linux. Not one. Either the machines don't sleep at all, they sleep for a couple seconds, come back up and then hang, or they never wake up. Windows has always worked properly. I have a feeling that most of these manufacturers only test with Windows. Why invest the time and money testing for Linux when only a small percentage of your customers are running it?

Also, going to a low paid tech support drone with an issue like this and then giving them attitude because they stated that Linux isn't supported isn't the way to get stuff fixed. Try to find someone higher up the chain of command. If I was sitting in a call center making $12/hr I couldn't give a rat's ass if our product doesn't work with an OS we don't support. What OSes to support is a product management decision, not a technical support decision. Telling them they suck gets you nothing but someone even less willing to you. I've done support. If someone told me they were using an unsupported configuration it would be an easy ticket to close. Solution? Use a supported configuration.

[Vexx]

I think the issue was:
1) there was specific code in the bios to address Linux (one that unnecessarily led to bad tables when the good tables lived right next door but only Windows was invited). Kind of weird to say one doesn't support Linux when one has code in the bios to "support Linux" ... but yeah, it mostly means they never *tested* it with Linux.
2) They claimed to meet ACPI specs.... last I heard, "certification by Microsoft" doesn't equal "ACPI compliant". Bad adverting at the least.

This is useful information for IT people who don't want to have to worry about mobo which or what if they run a heterogeneous network, though I think the discoverer took some karmic damage in the way he beat up on weary call support drones.

[Ending]

Never even heard of them, but the global brands shouldn't have such issues. I have been using MSI's motherboards for quite a while and they have always scored well in both tests and practise.

[hobbes_fan]

Foxconn - also oem for MSI mainly their budget lines only thing that changes is they get a nice red PCB and the BIOS chip is MSI code. There's a lot of rebadging going on in the PC hardware industry. (intel boards used to manufactured by Asus but now the majority I hear are also foxconn). So in all reality they are more global than what you'd think. The issue here though is not the hardware-it's the about 4-5mb of code that pretty much is uncharted territory for average home users-the BIOS.

Personally I don't think its a case of "sabotaging" Linux. Hell, if they make any server boards they must know a significant number are going to get paired with Linux server o/s. If they really wanted to sabotage Linux they just wouldn't put a Linux ACPI table in the BIOS. Rather having poorly written code in there rather signifies it was put in as an afterthought. Again I have to reiterate that this is a very very basic chipset and this particular chipset in my experience is somewhat of a rarity. So IMO they're either a) incompetent b) lazy c) lack the technical skills or a combination of all of the above rather than full on saboteurs. Lets face it Linux makes up maybe 10% of the desktop market (if that) personally I think they just don't really give a toss (tech support responses pretty much state this) and the majority of that 10% won't even be looking at Foxconn.

Thoughts on Linux and hardware in general

Spoiler:

I think for linux distros to make more headway in the home PC market, they have to make an effort to get certified hardware - not just a case of having workarounds (ndiswrapper for example) just a defined set of chipsets and hardware where all features are available. Sure there are occasionally lists of chipsets but how easy is it to find chipset info on manufacturer pages ie wireless modems, tv tuners etc etc

As a sidenote Asus seems to be the most Linux friendly (probably common knowledge) http://www.phoronix.com/scan.php?pag...item&px=NjEwNw

[Epyon9283]

I would guess that the Linux specific code was probably left over from the company that wrote the BIOS (most companies get their BIOS code from American Megatrends or Phoenix Technologies). Foxconn probably made modifications for Windows on their boards but left the Linux stuff untouched.

It may also be possible that at one point during development they were supporting Linux but they dropped it due to time or budget constraints. They may have left it in because they forgot about it or because they didn't feel like having to run regression tests against a modified BIOS without the Linux specific code.

I just have a hard time believing that they did this with some sort of malice.

[WanderingKnight]

Why didn't they accept the user's fix, then?

[Ending]

Why didn't they accept the user's fix, then?

Thousands of messages, one employee...

And if the employee noticed the message, his boss might say it's not important enough.

And even after that the boss' bosses might think it too much trouble to bother with. If the message wasn't simply lost somewhere in the chain.

[Vexx]

yeah, I'd have to figure "inertial incompetence" wins over malice most of the time. I've had some amazing arguments with Tier 1-3-engineering over whether their DNS cache was poisoned (and lo, they want me to check my power first....) or oh great, ya'll changed all the IPs and network classes and failed to notify your static IP customers....

So I can see how the simple inertia of "don't send this to tier_the_next had as much to do with the problem as malice.

[Ledgem]

Originally Posted by Vexx

yeah, I'd have to figure "inertial incompetence" wins over malice most of the time.

Most likely, but sometimes it's easy to get the two mixed up. I can remember how I deduced that a cellular tower was malfunctioning near my girlfriend's parents' house (easy deducation - I had echoing and regularly dropped calls in that area, but moving to a location a mile or two away resulted in flawless service). I called up AT&T (then Cingular) and explained the situation. The tech was a nice lady, but after hearing that I'd just gotten off the plane from Los Angeles and was now on a Hawaiian island, she gave me a theory about what was really wrong. My cellphone must have been used to transmitting around buildings and concrete, she surmised, and now it wasn't used to not having to transmit around those. The recommended advice was to wait a few days, and maybe by then my cellphone would have acclimated. I mean, look - I can't even make this stuff up, she really said that. I was shocked into silence for a second or two, but before I could ask her if she really believed what she'd just said the call was dropped. I'm sure the tech was really just clued out about technology and was trying as best she could to be helpful, but if I suspected that she knew anything about how technology works I'd really have to wonder if that explanation wasn't just an attempt to make me sit there and accept that my cellphone problems were due to forces beyond anyone's control, rather than demanding that they send a tech out to fix the damn tower. (I later called and got a guy who agreed with my initial thoughts, and said they'd send someone out after taking down the address I'd noted the problem from.)

As to your question WanderingKnight, I'm not surprised that they wouldn't accept patches from random users. Even though people in IT claim that certification is largely useless in terms of identifying talent, certifications do create a trail of responsibility. What if the BIOS fixes contained their own bugs and created major problems for some company that bought computers with these motherboards? If Foxconn just accepted patches from almost anyone then they could probably successfully be sued for negligence. If the patches were provided by some other company, Foxconn could at least blame and sue that contracted company.

Why doesn't Foxconn accept the patches and perfom in-house testing and any modifications that they want? That'd be the ideal...

[hobbes_fan]

I wouldn't be accepting 3rd party code either. It opens up too much in terms of litigation. One it hasn't been thoroughly tested - one guy saying "hey it works" isn't sufficient - where's the testing methodology? If it bricks the motherboard under a different set of conditions well foxconn is liable by supporting that fix. Now by saying they don't support Linux covers their ass without any effort or expense. Whether its good business practice is debatable, but they are under no obligation to support every o/s ( eg my 780g board can't run win98 or anything earlier than ME)

[WanderingKnight]

Well, ACPI compliance does not mean Windows-only compatibility, at least on legal paper.

[Epyon9283]

There is no legal paper here. Foxconn isn't even listed as an adopter company on the ACPI site. Even if they were, the adopter's agreement states nothing about actually conforming to the spec. The group responsible for ACPI provides no compliance tests for hardware.

Microsoft was one of a few companies that originally drafted the ACPI specs. As far as Microsoft is concerned, compatibility with Windows is the only thing that matters.

[Claies]

Let's close the story with this...

Foxconn released information stating that American Megatrends shipped a defective BIOS to the following boards:
- MSI P965 Platinum
- Asus P5K-E
- Asus P5E WS
- Asus P5E WS PRO

There's no ETA of an official fix, but at least we know it's a screwup in the BIOS and not sabotage in the chipset.

http://blogs.zdnet.com/hardware/?p=2307

https://bugs.launchpad.net/ubuntu/+s...ux/+bug/251338

[felix]

My experience with FOXCONN from having to deal with it for friends is that their crap particularly the bios is nothing less then self-destructing. I'm sure this is not sabotage but simply more of their stupidity and incopatence.