BSOD Makes Appearance at Olympic Opening Ceremonies 521
Whiteox writes "A BSOD was projected onto the roof of the National Stadium during the grand finale to the four-hour spectacular at the Olympics. Lenovo chairman Yang Yuanqing chose to go with XP instead of Vista because of the complexity of the IT functions at the Games. His comment on Vista? 'If it's not stable, it could have some problems,' he said. Evidently Bill Gates attended the opening ceremony, so he must have witnessed it."
Might as well get used to it (Score:5, Interesting)
Visible computer glitches pop up in the most unexpected places these days. I went to a 25th anniversay screening of Wargames at a local theater recently. I wasn't even aware that I was in a digital theater until about halfway through the movie their server lost connection to the host and the movie theater screen suddenly turned into a giant Windows desktop. It was a little unnerving (I had thought I was looking at an actual film).
I think it's something we will just get used to seeing in this increasingly digital age. I just hope I'm not driving down the street one day and see a "lost connection to server" message flashing on a stoplight.
Faked (Score:5, Interesting)
Re:well (Score:3, Interesting)
Re:In fairness to software engineering (Score:3, Interesting)
Back when Windows NT was being developed, I heard that device drivers had to communicate with hardware through the hardware abstraction layer (HAL), and this made Windows NT very stable. Then I heard that they decided to allow hardware drivers to connect directly to hardware because sometimes going through HAL had a performance hit. I can't find much information on the history, but these lecture notes [kent.edu] seem to confirm that drivers can now bypass HAL. Is this why bad drivers can still crash Windows?
Microsoft at least provides tools to verify that drivers work properly [microsoft.com].
Re:Might as well get used to it (Score:1, Interesting)
I recall wondering past an ATM one night and it was happily (perhaps) displaying the Win98 boot screen, then nothing, then the Win98 boot screen over and over.
Although deep down you know these things (desktop OSs) run fairly ordinary objects in our life, it's a little weird when you see signs of it at the surface.
Re:In fairness to software engineering (Score:1, Interesting)
Oh really? Then I guess Linux is not ready for production [miguelcarrasco.net]. And neither is Mac. [miguelcarrasco.net] Stop acting like a dumbass just to get mod points. Looks like you've succeeded well in baiting some zealot mods though. (Posting as AC since I used some mod points on this article).
Re:In fairness to software engineering (Score:5, Interesting)
As somebody who has written a bad device driver for Mac OSX I can confirm that a bad driver can and frequently has crashed my OS X kernel.
OS X is based on a microkernel, but in practice it is as monolithic as Linux or BSD.
Re:In fairness to software engineering (Score:4, Interesting)
100% true. NT 3.5 and 3.51 had the video outside the kernel. NT 4.0 moved it to kernel level. This was a big to do at the time, with everyone claiming that NT 4 was going to become unstable that way. Ironically, XP probably wouldn't have been used for projecting graphic images on a ceiling if that change had not been made 2 generations back. Damned if you do, damned if you don't...
Re:well (Score:5, Interesting)
>They are covered up well, quickly fixed, or not noticed, but they are there
I learned this when I saw a circus fire and noticed that the clowns put the fire out while making it look like part of the act. It was both comforting and frightening at the same time.
Re:Might as well get used to it (Score:3, Interesting)
The displays in the Red Line/Purple Line subway of LA's Metro Rail system are all on some embedded version of Windows. I've seen bluescreens and other errors, plus one time when you actually saw the desktop. With Internet Exploder and Windows Media Player among the icons on the desktop. It's not as crucial as a stoplight or an ATM but it's disheartening to see. Another place Windows runs in the LA transit system is on the monitors on almost every bus showing entertainment and ad programming to the captive audience on the bus. Seen lots of Windows Fail on that system.
Re:In fairness to software engineering (Score:2, Interesting)
Not quite, doing development stuff I'd be downloading tools, libraries, and software packages to help out, and just about every time I'd get about 6 months after initial install of windows and it would go into BSOD cycle, eventually it'd be bad enough to reinstall windows and the cycle would start over again.
So yah, not normal usage... but I'm doing roughly the same thing with my MB Pro and I haven't run into the same problem.
Re:In fairness to software engineering (Score:3, Interesting)
Re:In fairness to software engineering (Score:3, Interesting)
Bad drivers can crash any system using a monolithic kernel.
They can crash microkernel based systems severley too--microkernel systems like Microsoft Windows NT/2k/XP/Vista. The quality of the system architecture overall is far more important than the kernel architecture chosen.
Only low level problems can cause a Windows BSOD
Not true. Driver issues are the main reason, but user-level software can behave badly too. You cite anti-virus and firewall software, which aren't exactly "low level". Developer tools are usually the worst user-level offenders.
XP these days is stable (it only took 7 years but they made it) and you won't see a blue screen using signed drivers and hardware that isn't malfunctioning.
The thing is it can be difficult to find signed drivers for your system. If you want them for Vista you're SOL unless you have a very recent system (by and large, upgrading to Vista is a really dumb idea--everyone should stay with XP until they are willing to get a completely new machine), and if you have XP or 2K the opposite is true (the latest hardware ain't ever gonna have drivers, and its all closed software so no option to backport). I still run into a LOT of hardware with bad, unsigned drivers that is essential to some application yet behaves badly.
I'd guess the picture is in any case a fake.
Given the PRC's track record and the fact that they used "performance enhancing" technology on the official opening ceremonies broadcast I do not regard ANY media out of these olympics as trustworthy. To me, it is just as likely that the official Chinese broadcast digitally erased BSODs from the feed as it is that someone doctored a photo to ADD the BSOD and embarrass the Chinese olympic organisers.
Re:In fairness to software engineering (Score:3, Interesting)
Have you looked at the efforts of the Minix 3 operating system? It's a true microkernel where most drivers run outside of ring0 with limited access to hardware and/or the kernel.
Not just that, but it has stuff in place to severely limit the impact of a rogue driver and can restart dead or dying drivers, not to mention it embraces message passing with interrupts being passed to the driver as low-latency messages.
Other operating systems like QNX implement things in a similar way, although QNX also has guranteed near realtime scheduling and resource allocation allowing the whole system to be partitioned from the development stage.
Re:Of course... (Score:3, Interesting)
...assuming the error is not completely unrecoverable:
Then Linux can restart X and chug happily along.
It still won't look pretty if you are using it as
a projector in front of a billion people but you also
won't have a door stop.
Re:In fairness to software engineering (Score:3, Interesting)
Even still, this is an interesting situation. Assuming that virtually all BSODs and spontaneous resets on Windows are caused by faulty hardware drivers, apparently, these drivers, produced by professionals, even those certified by Microsoft, even those _shipped_ by Microsoft, seem to cause crashes a whole lot more often than those produced by a horde of hobbyists on the open-source side of the OS world.
Re:Of course... (Score:3, Interesting)
In Vista, the only driver crashes I've seen cause only a brief screen flicker.
Of course, also in Vista, no one can hear you scream.
Re:In fairness to software engineering (Score:3, Interesting)
But also they answer for the drivers, so a "bad driver" issue is actually a kernel issue.
I don't understand why "bad drivers" are not supposed to be the responsibility of MS. It's possible to design a system resilient to that kind of failure.
Well, this is very true.
However, you've got to look at the context. Firstly, Microsoft are more concerned about the system being stable on the sort of hardware bought by businesses - half-decent quality PCs and servers - and these tend to use relatively conservative hardware which has decent drivers.
Secondly, a bit of history - while the idea of true microkernels with practically every driver being a true userland process is not new, the performance penalty they introduce (which is less of an issue on modern hardware) was considered unacceptable when the NT kernel was first designed - and the kind of overhaul that would be necessary to change this is something Microsoft have historically shied away from.
Re:In fairness to software engineering (Score:2, Interesting)
In all but TWO cases (one case being a total hard drive failure where the system continued to run without HD access until a page swap was required
I've got a story that I like to tell about a Linux webserver that had a hard drive failure after running long enough to have the entire web root cached into RAM. We hdparm -Yed it to sleep and unplugged the drive, then let the machine run diskless for a few months until we got around to replacing it.
This hosted some very well known and active open source projects at the time too.
Re:well (Score:2, Interesting)
...so because nobody one the FOSS side has implemented fsck.ntfs-3g, it's Microsoft's problem?
Re:well (Score:3, Interesting)
IMHO , It also perfectly explains why IBM decided to get rid of tiny computers and CPUs running tiny computers (except consoles). I think they got sick of them, really.
These guys manufacture mainframes which theoretically run forever. Imagine you see that screen in that company culture while a bank calls for a CPU upgrade of a mainframe which runs for 10 years non stop which will be still done without turning it off :)
Re:well (Score:3, Interesting)
FYI: the scheduler is part of the OS kernel which decides which process/thread to run next.
Sorry...the first post wasn't specific enough and I read it as a scheduler for processes i.e. cron.
A better analogy: suppose Microsoft implemented ext2 in Windows, but not fsck. Is it Linux's fault that you can't use volumes from a hard drive that Linux did not mount properly?
The point I'm trying to get across is that, for example, there is a big difference between Linux wanting to use NTFS and Windows wanting to use ext2. The NTFS spec is a trade secret, and all work that has been done with ntfs-3g has been essentially hacking around a black box. The reason why there is not a chkdsk implementation in Linux is because that box is so secretive in the way it handles internal corruption (which is the nature of file systems, I know) and fixes itself that we can only chip away at it, hoping to get lucky and consistent. If Microsoft wanted to use ext2, then it's much easier: the spec is open, the code is open. Anyone could port fsck at will to Windows in that case if Microsoft didn't do it for them (and if Microsoft decided to close the source for its implementation of ext2, it definitely wouldn't be Linux's fault).
I know this is Linux's problem, but certainly not the developers' fault. If you really want an analogy, think of it as trying to crack the trade-secret recipe for Coca-Cola. You can get close, but without knowing the exact ingredients and processes...you get the idea. If Microsoft opened up NTFS your arguments would hold, but until then...