Hello,
To my dismay when arriving home i came to the conclusion that Tyan has
switched off the parallel port by _default_ inside several BIOS releases
in quite a number of her Opteron based motherboards.
This happens to be also the case with my Tyan Thunder K8W motherboard.
It also seems to a rather important issue when running a Linux
distribution based on kernel 2.6.x on such a Tyan Opteron board.
Tyan switching off the Parallel Port is even also reported at the public
forum at amd.com :
http://forums.amd.com/lofiversion/index.php/t43260.html
Recently however when updating to the newest BIOS release for Tyan
Opteron boards the parallel port seems to be switched on again inside
the factory default settings. Exactly as we have seen with your Tyan
Tomcat single Opteron CPU board.
The fact however that _all_ the various Linux distributions I have
tested so far, both 32bit and 64bit AMD64, seem to have severe problems
with the parallel port implementation of Tyan Opteron based boards, is
so far _nowhere_ mentioned.
All problems of Tyan Opteron based machines silently locking up during
installation and/or during normal operation of running Linux, both
32bit and 64bit, without any display of kernel panic of any other
logging method, seem to be solved when switching off the Parallel Port
inside its BIOS.
So here my official complaint to Tyan and the Linux kernel developers
to either make public notice of these problems, or preferably create a
solution/workaround for this problem either in software or hardware or
maybe inside a BIOS update.
Regards,
Robert M. Stockmann
--
Robert M. Stockmann - RHCE
Network Engineer - UNIX/Linux Specialist
crashrecovery.org [email protected]
On Fri, 20 May 2005, Robert M. Stockmann wrote:
>
> All problems of Tyan Opteron based machines silently locking up during
> installation and/or during normal operation of running Linux, both
> 32bit and 64bit, without any display of kernel panic of any other
> logging method, seem to be solved when switching off the Parallel Port
> inside its BIOS.
Can you do an install with the thing turned off, and then
- compile the kernel with CONFIG_PCI_DEBUG
- boot with the parallel port enabled, and send as much of the bootup
output (and /proc/iomem and /proc/ioport) as possible
- boot with the parallel port disabled, and send the same output for that
working case.
I have no clue why the parallel port should matter, but it could change
some resource allocation issues.
Linus
Linus Torvalds <[email protected]> writes:
> On Fri, 20 May 2005, Robert M. Stockmann wrote:
>>
>> All problems of Tyan Opteron based machines silently locking up during
>> installation and/or during normal operation of running Linux, both
>> 32bit and 64bit, without any display of kernel panic of any other
>> logging method, seem to be solved when switching off the Parallel Port
>> inside its BIOS.
The common Tyan problem case is when the machine has more than 3GB
of RAM and "memory remapping" is enabled to recover the memory
below the PCI memory hole. SOmething in that setup leads
to problems and random memory corruption. I suspect a BIOS bug
here.
Workaround is to not enable that option in the BIOS setup.
Then another older Tyan board (it might have been the K8W)
was *extremly* picky in what DIMMs it accepted and in what
slots because someone apparently didnt follow the AMD specification for
the memory controller trace lines fully. That also caused common problems.
> Can you do an install with the thing turned off, and then
> - compile the kernel with CONFIG_PCI_DEBUG
> - boot with the parallel port enabled, and send as much of the bootup
> output (and /proc/iomem and /proc/ioport) as possible
> - boot with the parallel port disabled, and send the same output for that
> working case.
>
> I have no clue why the parallel port should matter, but it could change
> some resource allocation issues.
It is the first time I heard about such a issue so it cannot be too
wide spread anyways.
-Andi
Don't always blame BIOS, if you like you could use LinuxBIOS instead....
YH
> -----Original Message-----
> From: Andi Kleen [mailto:[email protected]]
> Sent: Wednesday, May 25, 2005 9:50 AM
> To: Linus Torvalds
> Cc: [email protected]; Tech Support;
> [email protected]; Andrew Morton; [email protected];
> [email protected]; [email protected]; [email protected]
> Subject: Re: Tyan Opteron boards and problems with parallel ports
>
> Linus Torvalds <[email protected]> writes:
>
> > On Fri, 20 May 2005, Robert M. Stockmann wrote:
> >>
> >> All problems of Tyan Opteron based machines silently locking up
> >> during installation and/or during normal operation of
> running Linux,
> >> both 32bit and 64bit, without any display of kernel panic of any
> >> other logging method, seem to be solved when switching off the
> >> Parallel Port inside its BIOS.
>
> The common Tyan problem case is when the machine has more
> than 3GB of RAM and "memory remapping" is enabled to recover
> the memory below the PCI memory hole. SOmething in that setup
> leads to problems and random memory corruption. I suspect a
> BIOS bug here.
>
> Workaround is to not enable that option in the BIOS setup.
>
> Then another older Tyan board (it might have been the K8W)
> was *extremly* picky in what DIMMs it accepted and in what
> slots because someone apparently didnt follow the AMD
> specification for the memory controller trace lines fully.
> That also caused common problems.
>
>
> > Can you do an install with the thing turned off, and then
> > - compile the kernel with CONFIG_PCI_DEBUG
> > - boot with the parallel port enabled, and send as much of
> the bootup
> > output (and /proc/iomem and /proc/ioport) as possible
> > - boot with the parallel port disabled, and send the same
> output for that
> > working case.
> >
> > I have no clue why the parallel port should matter, but it could
> > change some resource allocation issues.
>
> It is the first time I heard about such a issue so it cannot
> be too wide spread anyways.
>
> -Andi
> -
> To unsubscribe from this list: send the line "unsubscribe
> linux-kernel" in the body of a message to
> [email protected] More majordomo info at
> http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at http://www.tux.org/lkml/
>
YhLu> Don't always blame BIOS, if you like you could
YhLu> use LinuxBIOS instead...
Just curious, but why isn't this project (LinuxBIOS) mentioned on the
Tyan web site, or is it and I just missed it?
You do work for Tyan, right?
Pete
Peter J. Stieber wrote:
> YhLu> Don't always blame BIOS, if you like you could
> YhLu> use LinuxBIOS instead...
>
> Just curious, but why isn't this project (LinuxBIOS) mentioned on the
> Tyan web site, or is it and I just missed it?
>
> You do work for Tyan, right?
What has that to do with anything? I doubt that suggestions about boot
options are on the website or come from the Tyan website, either.
Note: I'm not endorsing LinuxBIOS for Opteron, I haven't personally
tried it. But the value of the suggestion depends on how it works, not
who makes it. There appear to be a lot of reports of problems with
Opteron lately, if the BIOS isn't buggy then the documentation may have
lost in translation.
--
-bill davidsen ([email protected])
"The secret to procrastination is to put things off until the
last possible moment - but no longer" -me
YhLu>>> Don't always blame BIOS, if you like you could
YhLu>>> use LinuxBIOS instead...
PJS>> Just curious, but why isn't this project (LinuxBIOS)
PJS>> mentioned on the Tyan web site, or is it and I
PJS>> just missed it?
PJS>>
PJS>> You do work for Tyan, right?
BD = Bill Davidsen
BD> What has that to do with anything? I doubt that suggestions
BD> about boot options are on the website or come from the
BD> Tyan website, either.
BD>
BD> Note: I'm not endorsing LinuxBIOS for Opteron, I haven't
BD> personally tried it. But the value of the suggestion depends
BD> on how it works, not who makes it. There appear to be a
BD> lot of reports of problems with Opteron lately, if the BIOS
BD> isn't buggy then the documentation may have lost in
BD> translation.
I have been having the "memory.c bad pmds" with a Tyan S2885
motherboard.
https://www.redhat.com/archives/fedora-list/2005-May/msg01690.html
http://www.lib.uaa.alaska.edu/linux-kernel/archive/2005-Week-19/1397.html
When Yhlu brought up the topic of LinuxBIOS, I thought he might be
suggesting that this would prove there are no problems with the BIOS
(i.e. the same problems would occur with LinuxBIOS as with the
motherboards built-in BIOS). Looking at the LinuxBIOS web site, I got
the impression (I may be wrong) that this was a Tyan supported effort.
If that was the case I was wondering why Tyan didn't mention LinuxBIOS
on their web site. It would make me more comfortable if this were a Tyan
supported effort. That's why I asked the question.
I'm just trying anything I can to get rid of the bad pmd messages.
Pete
Can you give any information on the badpmd problem? There is a thread
about the badpmd problem and a couple of patches you might want to use
to get some debug information. I also suspect that it might have
something to do with the actual motherboard but people have reported
badpmds on other mobos.
I've put it down for now but the more info the better :-) I've
experienced problems with badpmds under load io, mem etc. Sometimes it
appears with little load.
On Thu, 2005-05-26 at 13:40 -0700, Peter J. Stieber wrote:
> YhLu>>> Don't always blame BIOS, if you like you could
> YhLu>>> use LinuxBIOS instead...
>
> PJS>> Just curious, but why isn't this project (LinuxBIOS)
> PJS>> mentioned on the Tyan web site, or is it and I
> PJS>> just missed it?
> PJS>>
> PJS>> You do work for Tyan, right?
>
> BD = Bill Davidsen
> BD> What has that to do with anything? I doubt that suggestions
> BD> about boot options are on the website or come from the
> BD> Tyan website, either.
> BD>
> BD> Note: I'm not endorsing LinuxBIOS for Opteron, I haven't
> BD> personally tried it. But the value of the suggestion depends
> BD> on how it works, not who makes it. There appear to be a
> BD> lot of reports of problems with Opteron lately, if the BIOS
> BD> isn't buggy then the documentation may have lost in
> BD> translation.
>
> I have been having the "memory.c bad pmds" with a Tyan S2885
> motherboard.
>
> https://www.redhat.com/archives/fedora-list/2005-May/msg01690.html
> http://www.lib.uaa.alaska.edu/linux-kernel/archive/2005-Week-19/1397.html
>
> When Yhlu brought up the topic of LinuxBIOS, I thought he might be
> suggesting that this would prove there are no problems with the BIOS
> (i.e. the same problems would occur with LinuxBIOS as with the
> motherboards built-in BIOS). Looking at the LinuxBIOS web site, I got
> the impression (I may be wrong) that this was a Tyan supported effort.
> If that was the case I was wondering why Tyan didn't mention LinuxBIOS
> on their web site. It would make me more comfortable if this were a Tyan
> supported effort. That's why I asked the question.
>
> I'm just trying anything I can to get rid of the bad pmd messages.
>
> Pete
>
>
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to [email protected]
> More majordomo info at http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at http://www.tux.org/lkml/
Just read the other existing thread linked. Is everyone running the same
model opteron on these Tyan boards (246)? To update even further.
Besides the parallel port problem I've sent back about 5 of these tyan
motherboards. A couple of then simply didn't have the gigabit network
adapters available via bios. In some cases linux loaded the drivers for
the adapters regardless of their setting in BIOS. In other cases they
simply were not available.
Also, are the processes you run swapping out/in memory repeatedly? Or
using a large amount of threading? Thread pools? I'm trying to get rid
of this problem myself. I've seen it extensively in 2.6.11.5 and not as
much in 2.6.11.8 but I haven't tested lately past .8
-Christopher Warner
On Thu, 2005-05-26 at 13:40 -0700, Peter J. Stieber wrote:
> YhLu>>> Don't always blame BIOS, if you like you could
> YhLu>>> use LinuxBIOS instead...
>
> PJS>> Just curious, but why isn't this project (LinuxBIOS)
> PJS>> mentioned on the Tyan web site, or is it and I
> PJS>> just missed it?
> PJS>>
> PJS>> You do work for Tyan, right?
>
> BD = Bill Davidsen
> BD> What has that to do with anything? I doubt that suggestions
> BD> about boot options are on the website or come from the
> BD> Tyan website, either.
> BD>
> BD> Note: I'm not endorsing LinuxBIOS for Opteron, I haven't
> BD> personally tried it. But the value of the suggestion depends
> BD> on how it works, not who makes it. There appear to be a
> BD> lot of reports of problems with Opteron lately, if the BIOS
> BD> isn't buggy then the documentation may have lost in
> BD> translation.
>
> I have been having the "memory.c bad pmds" with a Tyan S2885
> motherboard.
>
> https://www.redhat.com/archives/fedora-list/2005-May/msg01690.html
> http://www.lib.uaa.alaska.edu/linux-kernel/archive/2005-Week-19/1397.html
>
> When Yhlu brought up the topic of LinuxBIOS, I thought he might be
> suggesting that this would prove there are no problems with the BIOS
> (i.e. the same problems would occur with LinuxBIOS as with the
> motherboards built-in BIOS). Looking at the LinuxBIOS web site, I got
> the impression (I may be wrong) that this was a Tyan supported effort.
> If that was the case I was wondering why Tyan didn't mention LinuxBIOS
> on their web site. It would make me more comfortable if this were a Tyan
> supported effort. That's why I asked the question.
>
> I'm just trying anything I can to get rid of the bad pmd messages.
>
> Pete
>
>
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to [email protected]
> More majordomo info at http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at http://www.tux.org/lkml/
On Thu, 26 May 2005, Christopher Warner wrote:
> Just read the other existing thread linked. Is everyone running the same
> model opteron on these Tyan boards (246)? To update even further.
> Besides the parallel port problem I've sent back about 5 of these tyan
> motherboards. A couple of then simply didn't have the gigabit network
> adapters available via bios. In some cases linux loaded the drivers for
> the adapters regardless of their setting in BIOS. In other cases they
> simply were not available.
I have 4 tyan s2882's. 2 with 244's and 2GB of ram and 2 with 246 and 4GB
of ram. I don't appear to have bad parallel ports or bad ethernet
interfaces. They're running fedora core 3's 2.6.11 x86_64 kernel
(2.6.11-1.14_FC3smp)... I cna't tell you what bios version off the top of
my head but i can reboot one tomorrow if that helps.
> Also, are the processes you run swapping out/in memory repeatedly? Or
> using a large amount of threading? Thread pools? I'm trying to get rid
> of this problem myself. I've seen it extensively in 2.6.11.5 and not as
> much in 2.6.11.8 but I haven't tested lately past .8
>
> -Christopher Warner
>
>
> On Thu, 2005-05-26 at 13:40 -0700, Peter J. Stieber wrote:
>> YhLu>>> Don't always blame BIOS, if you like you could
>> YhLu>>> use LinuxBIOS instead...
>>
>> PJS>> Just curious, but why isn't this project (LinuxBIOS)
>> PJS>> mentioned on the Tyan web site, or is it and I
>> PJS>> just missed it?
>> PJS>>
>> PJS>> You do work for Tyan, right?
>>
>> BD = Bill Davidsen
>> BD> What has that to do with anything? I doubt that suggestions
>> BD> about boot options are on the website or come from the
>> BD> Tyan website, either.
>> BD>
>> BD> Note: I'm not endorsing LinuxBIOS for Opteron, I haven't
>> BD> personally tried it. But the value of the suggestion depends
>> BD> on how it works, not who makes it. There appear to be a
>> BD> lot of reports of problems with Opteron lately, if the BIOS
>> BD> isn't buggy then the documentation may have lost in
>> BD> translation.
>>
>> I have been having the "memory.c bad pmds" with a Tyan S2885
>> motherboard.
>>
>> https://www.redhat.com/archives/fedora-list/2005-May/msg01690.html
>> http://www.lib.uaa.alaska.edu/linux-kernel/archive/2005-Week-19/1397.html
>>
>> When Yhlu brought up the topic of LinuxBIOS, I thought he might be
>> suggesting that this would prove there are no problems with the BIOS
>> (i.e. the same problems would occur with LinuxBIOS as with the
>> motherboards built-in BIOS). Looking at the LinuxBIOS web site, I got
>> the impression (I may be wrong) that this was a Tyan supported effort.
>> If that was the case I was wondering why Tyan didn't mention LinuxBIOS
>> on their web site. It would make me more comfortable if this were a Tyan
>> supported effort. That's why I asked the question.
>>
>> I'm just trying anything I can to get rid of the bad pmd messages.
>>
>> Pete
>>
>>
>> -
>> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
>> the body of a message to [email protected]
>> More majordomo info at http://vger.kernel.org/majordomo-info.html
>> Please read the FAQ at http://www.tux.org/lkml/
>
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to [email protected]
> More majordomo info at http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at http://www.tux.org/lkml/
>
--
--------------------------------------------------------------------------
Joel Jaeggli Unix Consulting [email protected]
GPG Key Fingerprint: 5C6E 0104 BAF0 40B0 5BD3 C38B F000 35AB B67F 56B2
On Thu, 2005-05-26 at 21:41 -0700, Joel Jaeggli wrote:
> On Thu, 26 May 2005, Christopher Warner wrote:
>
> > Just read the other existing thread linked. Is everyone running the same
> > model opteron on these Tyan boards (246)? To update even further.
> > Besides the parallel port problem I've sent back about 5 of these tyan
> > motherboards. A couple of then simply didn't have the gigabit network
> > adapters available via bios. In some cases linux loaded the drivers for
> > the adapters regardless of their setting in BIOS. In other cases they
> > simply were not available.
>
> I have 4 tyan s2882's. 2 with 244's and 2GB of ram and 2 with 246 and 4GB
> of ram. I don't appear to have bad parallel ports or bad ethernet
> interfaces. They're running fedora core 3's 2.6.11 x86_64 kernel
> (2.6.11-1.14_FC3smp)... I cna't tell you what bios version off the top of
> my head but i can reboot one tomorrow if that helps.
Not to interested in the BIOS versions anymore. As the tyan boards in
question have been rma'd I'm not exactly sure whats going on with them
now. The tyan boards I have in here right now appear to work properly
except for the pmd issue, i've noticed nothing wrong with them besides
that.
What we need is to try and single out the variables that are causing the
badpmd's. Also the more people who report badpmd's with Andi Kleen's
intial patch the better. Especially on different archs; would also be
good. So far from lkml i'm only seeing Tyan S28* boards as of recent.
-Christopher Warner