Skip to contentSkip to conversation

nforce2 random lockups - still no solution ?

DISCUSSION Samuel Kvasnica 8 participants 12 messages Oct 21, 2003 – Nov 8, 2003 2 messages not in the archive

lore
Share & tools
lore
#0 lore raw

Hello,

Last few weeks I spent quite much time trying to get working two nforce2 chipset based motherboards with latest linux 2.4.22 - ASUS A7N8 Deluxe 2.0 and MSI K7N2 Delta. Althought the latest kernel version already detects correctly all the nforce2 stuff and APIC seems to run work as well, some random lockups still remain. At the very beginning I believed I could get rid of lockups by setting lower udma mode on IDE (hdparm -X udma3). In the fact this only reduced very much the probability of lockups - with udma5 I could freeze the system within few minutes e.g. during kernel compilation. with udma3 system is usually stable for several days, but sometimes it locks. When it happens, it happens mostly during first 30 mins after boot. I'm booting with noapic, nolapic and acpi=off, but it doesn't seem to have really any effect on lockups.
Unfortunatelly, I can't get any debug info. I've redirected syslog to flash-card but didn't get even a bit more of info, it seems to blow-up the chipset completelly and immediatelly.

Now, in the system with MSI K7N2 motherboard I have a framegrabber (Hauppauge PVR-250) installed, using ivtv driver.
I'm able to lock-up the system when streaming uncompressed video (e.g. cat /dev/yuv0 >/dev/null) and the lockups are also hard, w/o debug info. The ivtv driver is using DMA very heavily but seems to work on other chipsets. So these lock-up problems might be rather DMA then APIC related. Interesting is that I've never had such a lock-up when running WinXP on same computer ( :-) seems impossible), even under load and with the framegrabber. I've run memtest on both machines for 24 hours, it shouldn't be bad memory. On both machines I'm using just onboard hardware, except for the AGP graphic card (matrox and nvidia) and the framegrabber. Hard disk is on parallel IDE, SATA controller is disabled.

So my conclusion is that the nforce2 lock-ups are still unsolved at the moment. Any nforce experts around ?

Please, reply directly using cc:, I'm not on the list.

thanks,

Sam

#1 lore raw

Turn off apic in the bios.. it should then work well ..

Quoted 52 lines · Samuel Kvasnica

Hello,

Last few weeks I spent quite much time trying to get working two nforce2 chipset based motherboards with latest linux 2.4.22 - ASUS A7N8 Deluxe 2.0 and
MSI K7N2 Delta. Althought the latest kernel version already detects correctly all the nforce2 stuff and APIC seems to run work as well, some random lockups still remain. At the very beginning I believed I could get rid of lockups by setting lower udma mode on IDE (hdparm -X udma3).
In the fact this only reduced very much the probability of lockups - with udma5 I could freeze the system within few minutes e.g. during kernel compilation.
with udma3 system is usually stable for several days, but sometimes it locks. When it happens, it happens mostly during first 30 mins after boot. I'm booting with noapic, nolapic and acpi=off, but it doesn't seem to have really any effect on lockups.
Unfortunatelly, I can't get any debug info. I've redirected syslog to flash-card but didn't get even a bit more of info, it seems to blow-up the chipset completelly and immediatelly.

Now, in the system with MSI K7N2 motherboard I have a framegrabber (Hauppauge PVR-250) installed, using ivtv driver.
I'm able to lock-up the system when streaming uncompressed video (e.g. cat /dev/yuv0 >/dev/null) and the lockups are also hard, w/o debug info. The ivtv driver is using DMA very heavily but seems to work on other chipsets. So these lock-up problems might be rather DMA then APIC related. Interesting is that I've never had such a lock-up when running WinXP on same computer ( :-) seems impossible), even under load and with the framegrabber.
I've run memtest on both machines for 24 hours, it shouldn't be bad memory. On both machines I'm using just onboard hardware, except for the AGP graphic card
(matrox and nvidia) and the framegrabber. Hard disk is on parallel IDE, SATA controller is disabled.

So my conclusion is that the nforce2 lock-ups are still unsolved at the moment. Any nforce experts around ?

Please, reply directly using cc:, I'm not on the list.

thanks,

Sam

-
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/

#2 lore raw
Quoted 10 lines · Matt H.

Turn off apic in the bios.. it should then work well ..

Quoted 6 lines · Samuel Kvasnica

Hello,

Last few weeks I spent quite much time trying to get working two nforce2 chipset based motherboards with latest linux 2.4.22 - ASUS A7N8 Deluxe 2.0 and
MSI K7N2 Delta. Althought the latest kernel version already detects

..

If turning off APIC doesn't solve the problem, try a kernel with LOCAL APIC support turned off.

Gabor

#3 lore raw

Gabor,

I'm booting the kernel with acpi=off, noapic and nolapic options and according to syslog it is really disabled. Lockups are very rare (they come either during 30mins after boot or never) but I cat get them very quickly when using the framegrabber with ivtv driver. The ivtv driver itself works rock solid on older PIII system (btw, that PIII runs SMP kernel !). So I think there are definitely DMA problems with nforce2 boards under linux.

Sam

Quoted 6 lines · Vitez Gabor

If turning off APIC doesn't solve the problem, try a kernel with LOCAL APIC support turned off.

Gabor

A message that is not in the archive (sent off-list or not yet received): [email protected]

A message that is not in the archive (sent off-list or not yet received): [email protected]

#6 lore raw

Dear Gabor,

thantks a lot, indeed you are right ! I've been confused by some nforce FAQs where 'nolapic' option was recommended. In fact I did never check whether this option really exists. Now, after recompiling the kernel the framegrabber works with uncompressed stream for almost 24h and it is rock-solid.

So, a workaround recommendation for all using ivtv driver on nforce2 chipsets and kernels up to 2.4.22:

*** RECOMPILE YOUR KERNEL WITH LOCAL APIC DISABLED ***,

otherwise you'll experience very rare random lockups while watching the compressed stream and lockups within 10 minutes when watching the uncompressed yuv stream.

What I'd like to know is whether this bug is AMD processor or chipset related. Is there a patch available ? I tried 2.6.0-test8 and it wasn't stable. I'd prefer to use APIC instead of XT-PIC because some drivers e.g. kernel DRM don't support shared interrupts and I can't get own interrupt for the video card.

Sam

Quoted 19 lines · Vitez Gabor

Hi,

Quoted 3 lines · Samuel Kvasnica

I'm booting the kernel with acpi=off, noapic and nolapic options and

nolapic? That's not a valid kernel command line parameter I'm afraid.

fgrep -ri nolapic $KERNEL_SOURCE_DIR

gives nothing. ???

Gabor

#7 lore raw
Quoted 13 lines · Samuel Kvasnica

Dear Gabor,

thantks a lot, indeed you are right ! I've been confused by some nforce FAQs where 'nolapic' option was recommended. In fact I did never check whether this option
really exists. Now, after recompiling the kernel the framegrabber works with uncompressed stream for almost 24h and it is rock-solid.

So, a workaround recommendation for all using ivtv driver on nforce2 chipsets and kernels up to 2.4.22:

*** RECOMPILE YOUR KERNEL WITH LOCAL APIC DISABLED ***,

Quoted 7 lines

otherwise you'll experience very rare random lockups while watching the compressed stream and lockups within 10 minutes when watching the uncompressed
yuv stream.

What I'd like to know is whether this bug is AMD processor or chipset related.

Chipset and/or BIOS. AMD processors are known to work in other mobos.

You may try disabling just I/O-APIC or ACPI.

#8 lore raw
Quoted 9 lines
Quoted 3 lines

What I'd like to know is whether this bug is AMD processor or chipset related.

Chipset and/or BIOS. AMD processors are known to work in other mobos.

You may try disabling just I/O-APIC or ACPI.

No, only disabling local apic really helps. On both ASUS and MSI nforce2 motherboards. I can even enable apic and acpi with disabled local apic (anyway with XT-PIC only) and have no troubles. The lockups seem to appear when drivers heavily employing DMA are used (as udma5 hdd or framegrabber).

#9 lore raw
In reply to Samuel Kvasnica
Quoted 12 lines · Samuel Kvasnica

Dear Gabor,

thantks a lot, indeed you are right ! I've been confused by some nforce FAQs where 'nolapic' option was recommended. In fact I did never check whether this option really exists. Now, after recompiling the kernel the framegrabber works with uncompressed stream for almost 24h and it is rock-solid.

So, a workaround recommendation for all using ivtv driver on nforce2 chipsets and kernels up to 2.4.22:

*** RECOMPILE YOUR KERNEL WITH LOCAL APIC DISABLED ***,

otherwise you'll experience very rare random lockups while watching the compressed stream and lockups within 10 minutes when watching the uncompressed yuv stream.

Quoted 23 lines

What I'd like to know is whether this bug is AMD processor or chipset related. Is there a patch available ? I tried 2.6.0-test8 and it wasn't stable. I'd prefer to use APIC instead of XT-PIC because some drivers e.g. kernel DRM don't support shared interrupts and I can't get own interrupt for the video card.

Sam

Quoted 16 lines · Vitez Gabor

Hi,

Quoted 1 line · Samuel Kvasnica

I'm booting the kernel with acpi=off, noapic and nolapic options and

nolapic? That's not a valid kernel command line parameter I'm afraid. fgrep -ri nolapic $KERNEL_SOURCE_DIR

gives nothing. ???

Gabor

Extract fron linux-2.6test8 Documentation/kernel-parameters.txt
nolapic [IA-32,APIC] Do not enable or use the local APIC.

So, It is supported in the kernel 2.6.x

Cheers
James

#10 lore raw
In reply to Samuel Kvasnica
Quoted 9 lines · Samuel Kvasnica

Now, in the system with MSI K7N2 motherboard I have a framegrabber (Hauppauge PVR-250) installed, using ivtv driver.
I'm able to lock-up the system when streaming uncompressed video (e.g. cat /dev/yuv0 >/dev/null) and the lockups are also hard, w/o debug info. The ivtv driver is using DMA very heavily but seems to work on other chipsets. So these lock-up problems might be rather DMA then APIC related. Interesting is that I've never had such a lock-up when running WinXP on same computer ( :-) seems impossible), even under load and with the framegrabber.

It can be a too-aggressive chipset configuration triggering chipset big.

Maybe do lspci -vvvxxx and it's WinXP equivalent (I think such tool for Win ought to exist somewhere, although I had no need for it yet) and compare the result.
--
vda

#11 lore raw
Quoted 14 lines
Quoted 13 lines

Now, in the system with MSI K7N2 motherboard I have a framegrabber (Hauppauge PVR-250) installed, using ivtv driver.
I'm able to lock-up the system when streaming uncompressed video (e.g. cat /dev/yuv0 >/dev/null) and the lockups are also hard, w/o debug info. The ivtv driver is using DMA very heavily but seems to work on other chipsets. So these lock-up problems might be rather DMA then APIC related. Interesting is that I've never had such a lock-up when running WinXP on same computer ( :-) seems impossible), even under load and with the framegrabber.

Quoted 8 lines

It can be a too-aggressive chipset configuration triggering chipset big.

Maybe do lspci -vvvxxx and it's WinXP equivalent (I think such tool for Win ought to exist somewhere, although I had no need for it yet) and compare the result.
--
vda

It was local APIC ! After recompiling 2.4.22 without local apic everything works smoothly since several weeks. I wonder when there'll be a kernel patch that really solves these nforce2/amd issues.
Sam

#12 lore raw

Quoting Samuel Kvasnica <[email protected]>:

Quoted 5 lines

It was local APIC ! After recompiling 2.4.22 without local apic everything works smoothly since several weeks. I wonder when there'll be a kernel
patch that really solves these nforce2/amd issues.
Sam

Disabling local APIC on 2.6.0-test9-mm2 also fixes this (I haven't tried on earlier kernels).

-sandalle

Signature · 17 lines
PGP Key Fingerprint:  FCFF 26A1 BE21 08F4 BB91  FAED 1D7B 7D74 A8EF DD61
http://search.keyserver.net:11371/pks/lookup?op=get&search=0xA8EFDD61

-----BEGIN GEEK CODE BLOCK-----
Version: 3.12
GCS/E/IT$ d-- s++:+>: a-- C++(+++) BL++++VIS>$ P+(++) L+++ E-(---) W++ N+@ o?
K? w++++>-- O M-@ V-- PS+(+++) PE(-) Y++(+) PGP++(+) t+() 5++ X(+) R+(++)
tv(--)b++(+++) DI+@ D++(+++) G>+++ e>+++ h---(++) r++ y+
------END GEEK CODE BLOCK------

Eric Sandall                     |  Source Mage GNU/Linux Developer
[email protected]                  |  http://www.sourcemage.org/
http://eric.sandall.us/          |  SysAdmin @ Inst. Shock Physics @ WSU
http://counter.li.org/  #196285  |  http://www.shock.wsu.edu/

----------------------------------------------------------------
This message was sent using IMP, the Internet Messaging Program.
#13 lore raw
Quoted 7 lines · Eric Sandall

Quoting Samuel Kvasnica <[email protected]>:

Quoted 5 lines

It was local APIC ! After recompiling 2.4.22 without local apic everything works smoothly since several weeks. I wonder when there'll be a kernel
patch that really solves these nforce2/amd issues.
Sam

Quoted 5 lines

Disabling local APIC on 2.6.0-test9-mm2 also fixes this (I haven't tried on earlier kernels).

-sandalle

I am seeing the same thing here. Disk activity and APIC don't seem to do well. What kind of performance hit if any are we taking by disabling APIC?

Signature · 6 lines
Josh McKinney		     |	Webmaster: http://joshandangie.org
--------------------------------------------------------------------------
                             | They that can give up essential liberty
Linux, the choice       -o)  | to obtain a little temporary safety deserve 
of the GNU generation    /\  | neither liberty or safety. 
                        _\_v |                          -Benjamin Franklin