I need some advice how to mittigate root cause of the problem that causes my system to crash during the POST memory test following the shutdown or restart from SuSE 11.2 (11.1 had the same problem).
I have installed multiple O/S-es on my PC, XP Pro SP3, W7RC trial, Ubuntu 9.04 and SuSE 11.2
I frequently use other O/S-es but hardware/firmware problems occur only after I exit from SuSE.
Following the problem, the only way to restart my PC is to turn off the power switch and wait for few moments. Following this, when I enter the Setup in BIOS, I often (but not always) get the following error:
“The system has experienced boot failures because of overclocking or changes of voltage.
Last settings in this page may not coincide with current H/W states.”
Sysinfo:
Mobo: GA-EP45T-UD3P
RAM: Corsair DD3 4GB Interlaced
CPU: Intel Core Duo E8500 3.16GHz
Video: GeForce 9500 GT
H/D: 1 x Western Digital WD10EACS
1 x Samsung HD501LJ
DVD: 2 x LG GH22NS30
BIOS: Award (vintage 2008, F7 , last available BIOS update)
To eliminate possibility of faulty memory or any compatibility issues, I have extensively tested the memory with Memtest and Windiag with no erros found. I have also used two different brands of DDR3 RAM for testing.
I do not overclock my MoBo and always load the Failsafe defaults following the problems after exiting from SuSE.
11.1 is still supported and 11.2 gets good attention from the developers.
If you are confident that only openSUSE has this problem, and that this problem is not a recent hardware glitch nor BIOS change (and hence your previous experience with multiple O/S-es on your PC, XP Pro SP3, W7RC trial, Ubuntu 9.04 are still valid) then IMHO, if no one posts a solution nor investigative method in this thread, you should write a bug report on both 11.1 and 11.2. There is guidance here for raising bug reports: Submitting Bug Reports - openSUSE
Note on this forum we are volunteers who support openSUSE. We are not the packagers nor the developers. So its possible one of our forum volunteers may have an idea as to what is happening (and maybe know a BIOS setting to change), IMHO the odds are the packagers/developers who look at the bug submission may come to the crux of the problem quicker.
MrZorg adjusted his/her AFDB on Wednesday 19 Aug 2009 13:56 to write:
>
> I need some advice how to mittigate root cause of the problem that
> causes my system to crash during the POST memory test following the
> shutdown or restart from SuSE 11.2 (11.1 had the same problem).
>
> I have installed multiple O/S-es on my PC, XP Pro SP3, W7RC trial,
> Ubuntu 9.04 and SuSE 11.2
>
> I frequently use other O/S-es but hardware/firmware problems occur only
> after I exit from SuSE.
>
> Following the problem, the only way to restart my PC is to turn off the
> power switch and wait for few moments. Following this, when I enter the
> Setup in BIOS, I often (but not always) get the following error:
>
> “The system has experienced boot failures because of overclocking or
> changes of voltage.
> Last settings in this page may not coincide with current H/W states.”
>
> Sysinfo:
>
> Mobo: GA-EP45T-UD3P
> RAM: Corsair DD3 4GB Interlaced
> CPU: Intel Core Duo E8500 3.16GHz
> Video: GeForce 9500 GT
> H/D: 1 x Western Digital WD10EACS
> 1 x Samsung HD501LJ
> DVD: 2 x LG GH22NS30
> BIOS: Award (vintage 2008, F7 , last available BIOS update)
>
> To eliminate possibility of faulty memory or any compatibility issues,
> I have extensively tested the memory with Memtest and Windiag with no
> erros found. I have also used two different brands of DDR3 RAM for
> testing.
>
> I do not overclock my MoBo and always load the Failsafe defaults
> following the problems after exiting from SuSE.
>
> Thanks in advance.
>
>
The only thing I can think of off the top of my head is have a look if SuSE
is loading anything to do with “update microcode” in your boot.msg ( you can
either red it straight from /var/log/boot.msg or use Yast>misc to view.
Maybe SuSE is updating the BIOS code and not clearing it at shutdown???
Possibly leaving it resident, I cannot see how but stranger things have
happened.
That is just a guess mind you and about the only thing I can possibly think
of now.
You can disable the loading from Yast>System>runlevel editor ( if it is even
there ) just to see if it makes any difference.
HTH
Mark
Caveat emptor
Nullus in verba
Nil illegitimi carborundum
Thanks mate. I’ll first try what is suggested in the post below.
Regardless of the result, I’ll prep the bug report.
I can ensure you that I have much more confidence in the enthusiasts who are volunteering on sites like this one, than in the professional mob.
I have played a fair bit with BIOS, but this Gigabyte’s product can not lock up access to CMOS, which is a bit dissapointing.
Neither the disabling of all green and overclocking features did help. I even tried to replace all [Auto] settings with the actual sensed voltages etc.
If the mobo had CMOS update protection, there would be a good chance to pinpoint part of the shutdown sequence which corrupts my CMOS settings. The system is stable for hours when I log in, which proves that the shutdown doesn’t like my gear (or vice versa). Also, turning the system off while logged in, seems to keep CMOS unchanged. Still need to try this several times to prove the point.
I built this system back in March and had problems with SuSE almost from day one, as I installed O/S-es in less then a week. The only thing I haven’t tried was to install SuSE v10, which was running on my old Gigabyte mobo. v10 can’t recognise SATA DVD drives, so I didn’t even bother to install it from USB drive.
Eventually I gave up and posted this yesterday.
All other O/S-es work fine, with no glitches whatsoever. The irony is that I’m writing this from Fedora 11, which I installed on the same PC few hours ago. >:(
Thanks mate. Will try. The only potential problem with the proposed scenario is that the system is stable while I’m logged in. The subsequent start or restart of PC starts crashing POST memory test. On the other hand it would make sense that SuSE successfully updates CMOS, but fails to reset it properly on exit.
Will keep you posted.
MrZorg adjusted his/her AFDB on Thursday 20 Aug 2009 14:06 to write:
> Thanks mate. Will try. The only potential problem with the proposed
> scenario is that the system is stable while I’m logged in. The
> subsequent start or restart of PC starts crashing POST memory test. On
> the other hand it would make sense that SuSE successfully updates CMOS,
> but fails to reset it properly on exit.
> Will keep you posted.
>
> Cheers…
>
>
No SuSE does not update the cmos or even touch it at all, that would be a
big big no no…
What happens is “IF” you have a buggy CPU microcode ( Intel did have
problems with some CPU`s) then during boot SuSE will load ( only into memory
not onto the actual BIOS/CMOS ) a fixed image, this only stays in memory
till the machine is shut down, you can disable this loading by going to
Yast>System>runlevel and turning off the microcade.ctl but honestly I do not
think it is that.
Same goes for the DSDT code for buggy acpi stuff.
I cannot see anyway of these causing any problem but as stated stranger
things have happened, I was just waffling off the top of my head about what
“might” cause it, theoretically.
No OS should have the ability to alter any code in the BIOS/CMOS unless you
specifically allow it, such as in Windows there are the BIOS flash programs
that can be run from the OS.
As your prob seems to go after you power off I was guessing that it was
something left in volatile memory and not the chip itself.
I do not do Intel ( well not very often ) but have never seen this error
before and especially not after using SuSE or any of the other distros.
The only thing close to this is the way sometimes Windows can swap a bit in
say a network chip so that it cannot be initialised by other OS, and there
have been times when Linux has done the same but not very often but that is
not the BIOS/CMOS.
I have had a quick look around and cannot see anywhere in the system where
anything could be left “in memory” ( think TSR from the old days ) that
might cause your error, however I am willing to be proved wrong and would
love to see a good explanation of why and who is responsible.
If this was a wide spread problem then I would have thought that there would
be a lot more with this error on these groups, you are the first and 11.1
has been out for a long time now.
When this occurs can you se in the BIOS if anything has been changed?
Make a list of the settings and when it happens go back into the BIOS and
check to see which is the culprit against said list.
Another thought is go back to a previous BIOS and see if that does it, it
could be a BIOS problem, I have Gigabyte boards and one will not boot with
the latest BIOS but if I drop it back one revision the board runs like a
dream 24/7 @100% CPU, this is not the boards fault it is down to the
hardware I have inside it.
HTH
–
Mark
Caveat emptor
Nullus in verba
Nil illegitimi carborundum
Sorry, my bad. Used wrong expression. As I’m using multi-boot I tend to make reference to different boot sessions by their O/S names.
Fully agree, could be a buggy driver, or a service. My mobo is relatively new, but then…
This is worth investigating. However, I hope this is not the misleading fact, but I was using Ubuntu for several months. I am currently logged into Fedora 11 session while I’m writing this.
I have no problems exiting from these two. The only safe way to leave SuSE session at the moment is to power off the PC. But then, until I check, I can’t tell if these two have microcode service up and running or not.
This is possible, too.
Again, I agree. I have Gigabyte tool that runs in Windows and is used for tweaking of memory and CPU. Another one enables “green” function where CPU clock goes as low as 2 GHz if PC is not busy. BTW, I disabled relevant switches in BIOS, hoping that would prevent changes, but it didn’t, which makes me suspect something else. If a driver or a script was intentionally updating CMOS, the switches would work. I suspect something is writing random data into a wrong location. Furthermore…
I was suspecting that something in mobo design uses volatile RAM or CPU cache during POST. I’ll see if I can find some forum where I can find how mobos with dual BIOS boot. There must be more to that than what I can see from the User Manual. But then if this is the case, any leftover garbage should be wiped off when POST starts.
I’ve been building and upgrading my own PCs for more than 10 years. I am predominantly using Windows, but started using Linux from SuSE 7.2 while it was still German owned. SuSE 10 and later 10.3 was running on previous Gigabyte mobo which was also tweakable, with no problems whatsoever. I may try to install SuSE 11 on another PC, which is also built around Gigabyte and Intel.
The only microcode problem that I ever came across was with ATI video card, when card started displaying coloured garbage following driver installation and reboot. After replacing video card and having the same problem again, I started investigating. Thanks golie for the internet already being available back then. I found a site where I was able to download floppy ISO. This used to re-flash microcode and was able to see the picture again. Once I had this going, I managed to recreate problem several times and identify the culprit. It was not the driver. Having also a video capture features, the card came with DVD player software. During the software installation, zone ID was written guess where? The idea was good, but the destination address was bad. I was able to prove it by a gadget which allowed me to display the microcode content of a video card. I have published my findings on the ATI support site. Their support was less then useful when I asked them for advice. One of the forum folks pointed me out to an unofficial ATI site where I got my tools from. This is why I said before that I more appreciate help from the enthusiasts and don’t rely much on the official support.
Yeah, I had the same problem once. I think one of the older versions of Mandrake used doing this to me. It would turn off the NIC. The MSI mobo did not have an option to re enable it as it was not integrated. I had to re-enter Linux and push the power button. Kind like my current situation.
Mate, why do you think I’m playing with this problem, when I can use other distros that work?
I only wish I had more time to play with it.
I am disadvantaged by the fact that I was not doing any serious programming for something like 15 years. Never programmed anything in Linux kernel, but worked back then for a company where we tweaked, believe it or not, good ol’ VMS. Had subscription to VMS source and used the non-paged pool in much smarter ways then DEC did. Did lot of MACRO32 programming back then.
My only Unix experience is some scripting I have done and few C and Pro C programs I wrote while converting few VMS based applications to Unix.
Agree again. Mind you, I build my own PC’s. Most people buy brands as it is simpler and cheaper nowadays. I do pay very much attention to compatibility and conduct the research about the components before making the final selection.
PC vendors do lot of matching and testing before they put their bits together.
So far GA-EP45T-UD3P was only reported to have some compatibility issues with some brands and models of DDR3 RAM. There are a number of references to this mobo on various overclockers’ sites, but nothing really negative. Besides, I do not overclock my hardware.
To be quite honest, if this was possible, we wouldn’t be having this discussion right now.
After powering off the PC, absolutely every value in BIOS Setup is identical to the values I recorded on the paper.
I was thinking about doing it at one point of time. I’ll give it a try as it worked for you and will let you know how did it go.
MrZorg adjusted his/her AFDB on Friday 21 Aug 2009 15:06 to write:
> Sorry, my bad. Used wrong expression. As I’m using multi-boot I tend to
> make reference to different boot sessions by their O/S names.
> Fully agree, could be a buggy driver, or a service. My mobo is
> relatively new, but then…
>
No probs,
Could be something like that I have 3 Gigabyte boards here all ATI/AMD chips
and have no probs, that is why I was leaning towards the microcode as the
AMD`s do not need it.
However AFAIK this is for older Intel CPU`s, you can actually d/load
different firmware from Intel IIRC and place them in the right place and
this will be used instead of the default.
If nothing else seems to work and it looks like it “could” be a problem It
may be wise to have a look on their site to see if anything has been
reported there.
Note,
I still do not think it is this but who knows.
Another thought that I have just posted in another thread is:
Did you have to do anything special during install to get the machine to
boot?
Like disable ACPI or APIC or install any 3rd party drivers to get the
install going?
I have known that these settings can be amended to the boot kernel after the
install and cause problems.
HTH
Mark
Caveat emptor
Nullus in verba
Nil illegitimi carborundum
