Linux instability is *NOT* a figment of peoples imagination

Linux instability is *NOT* a figment of peoples imagination

Post by Erik Funkenbusc » Wed, 23 Jan 2002 18:26:53



http://www.veryComputer.com/

According to this statement on gentoo's web page:

"The bad news is that a major Athlon CPU bug has been discovered, and it
affects Linux 2.4. Note that this is a bug in the actual CPU itself, and is
not a Linux bug. However, it becomes our problem because there are very many
semi-broken Athlon/Duron/Athlon MP CPUs out there."

It appears that Athlon and Duron processors for nearly 18 months have had a
serious bug which effect Linux boxes AND NOBODY KNEW IT.

Linux boxes all around the world running Athlon processors with AGP graphics
have been crashing and been unstable, yet linux advocates here claim that
never happens.  Anyone running a Linux on such a system would have been
plagued by irrational and completely unpredicatble crashes if those systems
were compiled for anything more than 486 compatibility.

Sure, this isn't a bug with Linux, but like the pentium F00F and other bugs,
it's something Linux has to deal with to be a stable OS.

How is it that such a gave bug can go unnoticed for 18 months?  Maybe linux
users were afraid to talk about their crashes for fear of being ridiculed by
their compatriates for not knowing how to configure their systems.  Or maybe
there aren't enough people using linux on the desktop to have run across the
problem.

It's not like this is a rare configuration.  10's of millions of computers are
out there with Duron's and Athlons running AGP video.

In any event, this strikes me as a very serious credibility gap if this bug
can exist for 18 months without anyone mentioning it.  Could this be
indicative of a massive coverup by Linux users to pretend that their systems
are more stable than they really are?

Just a little bit of * theory to brighten your day :-)

 
 
 

Linux instability is *NOT* a figment of peoples imagination

Post by Darren Winspe » Wed, 23 Jan 2002 19:16:38



Quote:> http://www.gentoo.org/

> According to this statement on gentoo's web page:

> "The bad news is that a major Athlon CPU bug has been discovered, and it
> affects Linux 2.4. Note that this is a bug in the actual CPU itself, and
> is not a Linux bug. However, it becomes our problem because there are
> very many semi-broken Athlon/Duron/Athlon MP CPUs out there."

> It appears that Athlon and Duron processors for nearly 18 months have
> had a serious bug which effect Linux boxes AND NOBODY KNEW IT.

> Linux boxes all around the world running Athlon processors with AGP
> graphics have been crashing and been unstable, yet linux advocates here
> claim that never happens.  Anyone running a Linux on such a system would
> have been plagued by irrational and completely unpredicatble crashes if
> those systems were compiled for anything more than 486 compatibility.

My box had very unpredictable crashes and lock ups for a while.  When I
pulled out one of my DIMMS to lend to a friend who was trying to diagnose
why his system was unusable, the crashes stopped.

I don't really do much with OpenGL on my system, although I've just
installed the Return to Castle Wolfenstein multiplayer on here.  Anyway,
RtCW is very unstable, it will core dump if I so much as look at it the
wrong way.  However, the same is true of the Windows version.

--
Darren Winsper (El Capitano)

Our government claims to be fighting terrorism to "defend our freedoms",
yet it is busy taking those freedoms away in order to "fight terrorism".

 
 
 

Linux instability is *NOT* a figment of peoples imagination

Post by Jauco Noordz » Wed, 23 Jan 2002 19:42:42


On Tue, 22 Jan 2002 10:16:38 +0000, Darren Winsper


>My box had very unpredictable crashes and lock ups for a while.  When I
>pulled out one of my DIMMS to lend to a friend who was trying to diagnose
>why his system was unusable, the crashes stopped.

Actually my Duron/agp box never crashed, but when compiling a linux
from scratch system I never succeeded in booting from my nwe kernel
(with Duron support) Maybe it's the bug. I don't know
 
 
 

Linux instability is *NOT* a figment of peoples imagination

Post by Stephen Cornel » Wed, 23 Jan 2002 21:08:44



> In any event, this strikes me as a very serious credibility gap if this bug
> can exist for 18 months without anyone mentioning it.  Could this be
> indicative of a massive coverup by Linux users to pretend that their systems
> are more stable than they really are?

Many people who had the problem were probably blaming their video
drivers.  I had a couple of freezes on my Athlon MP system which I put
down to NVidia drivers.  Disabling AGP, as suggested in the NVidia
documentation, cured the problem; it turns out that this catches the
bug.
--

University of Cambridge, Zoology Department, Downing Street, CAMBRIDGE CB2 3EJ
 
 
 

Linux instability is *NOT* a figment of peoples imagination

Post by Osugi Saka » Wed, 23 Jan 2002 21:07:57



> http://www.gentoo.org/

> According to this statement on gentoo's web page:

> "The bad news is that a major Athlon CPU bug has been discovered, and it
> affects Linux 2.4. Note that this is a bug in the actual CPU itself, and is
> not a Linux bug. However, it becomes our problem because there are very many
> semi-broken Athlon/Duron/Athlon MP CPUs out there."

> It appears that Athlon and Duron processors for nearly 18 months have had a
> serious bug which effect Linux boxes AND NOBODY KNEW IT.

> Linux boxes all around the world running Athlon processors with AGP graphics
> have been crashing and been unstable, yet linux advocates here claim that
> never happens.  Anyone running a Linux on such a system would have been
> plagued by irrational and completely unpredicatble crashes if those systems
> were compiled for anything more than 486 compatibility.

NOBODY KNEW IT is a bit strong, I think. I knew i had some stability
problems - but it only showed up with tribes two and nvidia
drivers. Specifically, whenever tribes 2 was updated, or when nvidia
released new drivers, i would have a crash the next time i played
tribes 2, practically a given.

I did some looking and found something on the nvidia site about some
nvidia settings in the xfree config file that could cause
instability. Changed the settings and viola, no more instability.

I have an Athlon cpu, run Mandrake, and use source rpms whenever
practical, so I should (i assume) have seen this bug a lot.

Point is that i am a fairly normal computer user -- certainly not a
programmer -- and it never occured to me that the source of the
instability might be in my cpu itself. I certainly never considered it
to be a flaw in linux.

The fact is that, in my case at least, everything else was rock
solid. X has never crashed (except for tribes 2 obviously). I had no,
and continue to have no, `irrational and completely unpredictable'
crashes. If i hadn't play tribes 2, i never would have noticed any
instability at all. So it is fair to say, in my case at least, that
linux was ultra-stable for me, despite this bug.

So perhaps we should not assume that every single user of a buggy AMD
cpu was or is affected?

You have a good point though, --- how many people had some instability
but never mentioned it? Maybe they thought ``but linux is stable so it
must be my fault'' or maybe they thought, like you say, that they
should not admit a problem in public. Too bad you go overboard with
the rhetoric.

Of course, COLA is hardly the place to look for people having trouble
with their systems. Had you checked the loki tribes newsgroup, you
would have found plenty of people complaining about freezes and
crashes.

--
Osugi Sakae

I am not a number, I am a free man

 
 
 

Linux instability is *NOT* a figment of peoples imagination

Post by Shawn Gran » Wed, 23 Jan 2002 22:03:52


Quote:> How is it that such a gave bug can go unnoticed for 18 months?  Maybe
> linux users were afraid to talk about their crashes for fear of being
> ridiculed by
> their compatriates for not knowing how to configure their systems.

A Linux cult member will never admit thier systems crashing. All they do is
lie and lie and lie.
 
 
 

Linux instability is *NOT* a figment of peoples imagination

Post by Charlie Eber » Wed, 23 Jan 2002 22:16:31


Quote:> http://www.veryComputer.com/

> According to this statement on gentoo's web page:

> "The bad news is that a major Athlon CPU bug has been discovered, and it
> affects Linux 2.4. Note that this is a bug in the actual CPU itself, and is
> not a Linux bug. However, it becomes our problem because there are very many
> semi-broken Athlon/Duron/Athlon MP CPUs out there."

> It appears that Athlon and Duron processors for nearly 18 months have had a
> serious bug which effect Linux boxes AND NOBODY KNEW IT.

> Linux boxes all around the world running Athlon processors with AGP graphics
> have been crashing and been unstable, yet linux advocates here claim that
> never happens.  Anyone running a Linux on such a system would have been
> plagued by irrational and completely unpredicatble crashes if those systems
> were compiled for anything more than 486 compatibility.

> Sure, this isn't a bug with Linux, but like the pentium F00F and other bugs,
> it's something Linux has to deal with to be a stable OS.

> How is it that such a gave bug can go unnoticed for 18 months?  Maybe linux
> users were afraid to talk about their crashes for fear of being ridiculed by
> their compatriates for not knowing how to configure their systems.  Or maybe
> there aren't enough people using linux on the desktop to have run across the
> problem.

> It's not like this is a rare configuration.  10's of millions of computers are
> out there with Duron's and Athlons running AGP video.

> In any event, this strikes me as a very serious credibility gap if this bug
> can exist for 18 months without anyone mentioning it.  Could this be
> indicative of a massive coverup by Linux users to pretend that their systems
> are more stable than they really are?

> Just a little bit of * theory to brighten your day :-)

Ah no.

I read that and I have alot of freinds using these CPU's and I've
never heard anybody * about their box locking up.

There must have been some rare circumstances to get it to do that.

Anyway, they patched it the day they found it.

And that's something that MS would never do.

Remember their WORM attack warning they sat on for 6 months???

Then when they got the patch out, 3 months later the worms hit
and their customer base didn't patch their systems because MS
didn't provide an easy patching tool like Linux OS's do.

And the internet was *ed and port 80 was thus banned for
a while.

And then another worm came thru and we were banned and *ed
again for the same reasons.

--

Charlie

 
 
 

Linux instability is *NOT* a figment of peoples imagination

Post by Ian Pege » Wed, 23 Jan 2002 23:05:20



Quote:> http://www.gentoo.org/

> According to this statement on gentoo's web page:

> "The bad news is that a major Athlon CPU bug has been discovered, and it
> affects Linux 2.4. Note that this is a bug in the actual CPU itself, and
> is not a Linux bug. However, it becomes our problem because there are
> very many semi-broken Athlon/Duron/Athlon MP CPUs out there."

Very interesting. And why, if it is a CPU bug, does it not affect
Windows?

--

Ian - looking through a glass onion

 
 
 

Linux instability is *NOT* a figment of peoples imagination

Post by Kano A » Wed, 23 Jan 2002 23:18:39



> On Tue, 22 Jan 2002 10:16:38 +0000, Darren Winsper

> >My box had very unpredictable crashes and lock ups for a while.  When I
> >pulled out one of my DIMMS to lend to a friend who was trying to diagnose
> >why his system was unusable, the crashes stopped.

> Actually my Duron/agp box never crashed, but when compiling a linux
> from scratch system I never succeeded in booting from my nwe kernel
> (with Duron support) Maybe it's the bug. I don't know

I would not suspect that.  

The bug has to do with memory paging.  There was a change in the
paging and that is where the bug is.  The article I read about it
stated that when you do a kernel compile there is a question that you
can answer in the litany questions that will nullify the bug (use old
paging? IIRC).  In addition current kernel developers are going to
code a scheme to automatically detect and work around the bug.

I believe that it was over at http://www.linuxtoday.com/ that I found
the link to the details behind this bug.

Anyway
KA

 
 
 

Linux instability is *NOT* a figment of peoples imagination

Post by maddm » Wed, 23 Jan 2002 23:25:27



> http://www.veryComputer.com/

> According to this statement on gentoo's web page:

> "The bad news is that a major Athlon CPU bug has been discovered, and it
> affects Linux 2.4. Note that this is a bug in the actual CPU itself, and is
> not a Linux bug. However, it becomes our problem because there are very many
> semi-broken Athlon/Duron/Athlon MP CPUs out there."

> It appears that Athlon and Duron processors for nearly 18 months have had a
> serious bug which effect Linux boxes AND NOBODY KNEW IT.

> Linux boxes all around the world running Athlon processors with AGP graphics
> have been crashing and been unstable, yet linux advocates here claim that
> never happens.  Anyone running a Linux on such a system would have been
> plagued by irrational and completely unpredicatble crashes if those systems
> were compiled for anything more than 486 compatibility.

No one claims that a Linux system is perfectly stable, as faulty
hardware or badly written programs can cause crashes.  But, the kernel
itself isn't a hacked, buggy, POS like a certain commercial OS we all
know and hate.  I've found Linux to be more tolerant of faulty
hardware than Windows.  For instance, I once had a NIC go bad on me,
way bad.  On booting Linux, kudzu identified it as bad, didn't
activate it, and continued loading.  Windows on the same machine froze
on boot, making me go into safe mode and remove the driver before the
OS would boot.

As far as software, while a buggy program can crash, it doesn't tend
to pull the whole OS down like a house of cards.

Quote:

> Sure, this isn't a bug with Linux, but like the pentium F00F and other bugs,
> it's something Linux has to deal with to be a stable OS.

> How is it that such a gave bug can go unnoticed for 18 months?  Maybe linux
> users were afraid to talk about their crashes for fear of being ridiculed by
> their compatriates for not knowing how to configure their systems.  Or maybe
> there aren't enough people using linux on the desktop to have run across the
> problem.

Ever occur to you that a bug in the CPU is paticularly hard to
pinpoint?  Wasn't it you (or some other wintroll) that was just
commenting on the number of posts in linux help groups?  People have
problems, its just that your options are more extensive than "wait
until the next version, spend more money, and hope like hell it works
better."

Quote:

> It's not like this is a rare configuration.  10's of millions of computers are
> out there with Duron's and Athlons running AGP video.

> In any event, this strikes me as a very serious credibility gap if this bug
> can exist for 18 months without anyone mentioning it.  Could this be
> indicative of a massive coverup by Linux users to pretend that their systems
> are more stable than they really are?

> Just a little bit of * theory to brighten your day :-)

Bah - I'd be willing to bet that even with this bug, Linux+AMD+AGP is
more stable and secure than windows.  In any case, serious servers are
rarely built with AMD, most people still going pentium.  (though that
may be changing now that Athlon supports multiprocessor)

You've been busy this morning, Erik.  Going for a raise?

 
 
 

Linux instability is *NOT* a figment of peoples imagination

Post by philicord » Wed, 23 Jan 2002 23:49:29


Quote:> Very interesting. And why, if it is a CPU bug, does it not affect
> Windows?

It does. AMD posted a fix for MSWindows as well.
 
 
 

Linux instability is *NOT* a figment of peoples imagination

Post by Jack Trippe » Wed, 23 Jan 2002 23:00:13



> How is it that such a gave bug can go unnoticed for 18 months?  Maybe
> linux users were afraid to talk about their crashes for fear of being
> ridiculed by their compatriates for not knowing how to configure their
> systems.  Or maybe there aren't enough people using linux on the desktop
> to have run across the problem.

> It's not like this is a rare configuration.  10's of millions of
> computers are out there with Duron's and Athlons running AGP video.

> In any event, this strikes me as a very serious credibility gap if this
> bug can exist for 18 months without anyone mentioning it.  Could this be
> indicative of a massive coverup by Linux users to pretend that their
> systems are more stable than they really are?

Yeah, uhmmm....no? The idea that Linux users suffer crashes and "keep it
a secret!" is ludicrous. Even imagining your view of an "average" Linux
user as a blind-sighted advocate settling for less, there's still a huge
amount of users like myself, who don't know that much about computers and
are using Linux because it _works_. If Linux didn't work, if it was
instable code and buggy as all hell, I'd grab my Win 98SE discs in a New
York minute. Apologies to the ideologists of this group, but I'm a short
sighted end-user who just wants his computer to work =). And it does,
with Linux.
--
http://www.veryComputer.com/
http://www.veryComputer.com/
http://www.veryComputer.com/
http://www.veryComputer.com/

Registered Linux user...something or other.
"Shut the * up" - my wife.

 
 
 

Linux instability is *NOT* a figment of peoples imagination

Post by Jack Trippe » Wed, 23 Jan 2002 23:08:08



>> How is it that such a gave bug can go unnoticed for 18 months?  Maybe
>> linux users were afraid to talk about their crashes for fear of being
>> ridiculed by
>> their compatriates for not knowing how to configure their systems.

> A Linux cult member will never admit thier systems crashing. All they do
> is lie and lie and lie.

What a nice, broad blanket statement! I've always admitted myself, on
this very NG that Linux occasionally does something weird. However, it's
nothing I can't CRTL-ALT-F1 out of, log in as root, and kill.And even
then, it's probably a punishment for using beta apps with bleeding edge
Mandrake stuff...my brother has no problems with Debian, I'll probably
have to take the plunge into that sooner or later, eh?

--
http://www.veryComputer.com/
http://www.veryComputer.com/
http://www.veryComputer.com/
http://www.veryComputer.com/

Registered Linux user...something or other.
"Shut the * up" - my wife.

 
 
 

Linux instability is *NOT* a figment of peoples imagination

Post by Christopher Lansdow » Thu, 24 Jan 2002 00:17:13


The reason it goes so long without being noticed is that it takes edge
conditions to actually hit the bug, so for many of us we haven't noticed
these problems at all.

Like me.  I haven't run accross AGP instability with my current setup.  I
was with my old motherboard, though.  Maybe this was actually the
problem, most people attributed it to the chipset, though, as it seemed
to only affect certain revisions.

Not to mention that until pretty recently, harware-accelerated (non-3dfx)
rendering was beta software and so you'd expect development software to
be buggy.

Noone ever claimed that the development versions of program were stable
in linux, though surprisingly enough they sometimes even are.

-Chris

--

 
 
 

Linux instability is *NOT* a figment of peoples imagination

Post by yt.. » Thu, 24 Jan 2002 00:37:56



> http://www.gentoo.org/
> According to this statement on gentoo's web page:
> "The bad news is that a major Athlon CPU bug has been discovered, and it
> affects Linux 2.4. Note that this is a bug in the actual CPU itself, and is
> not a Linux bug. However, it becomes our problem because there are very many
> semi-broken Athlon/Duron/Athlon MP CPUs out there."

Actually, if youd researched this even a little bit, you would know that it has
very much to do with nvidia drivers in particular.

Ive bee working with people on this bug for some time, and it isnt an athlon
bug.  It appears reliably, and only when playing tuxracer on a dual PII system,
a PIII system, and both my athlon systems.

Blah, erik is an idiot again.

-----.

--
Theres a hole in the world like a great black pit and
its filled with people who are filled with shit and the
vermin of the world inhabit it