kernel 2.1.116+ lock up

kernel 2.1.116+ lock up

Post by Riku Anttil » Wed, 28 Oct 1998 04:00:00




> With kernels 2.1.115 and below the sound worked just fine. and since there
> doesn't seem to have been some sound changes in 115-116 it must be something
> different.

I compiled my first 2.1.x kernel (126) today and noticed the same problem.
Remembering seeing this post I checked out using the sound driver sources
from 115 (didn't check if they differed), which compiled fine but produced
the same fault.

Guess I'm stuck with 115 then:( Has this bug been reported?

-Rixa

 
 
 

kernel 2.1.116+ lock up

Post by Michael Gilbe » Thu, 29 Oct 1998 04:00:00


Quote:>> With kernels 2.1.115 and below the sound worked just fine. and since there
>> doesn't seem to have been some sound changes in 115-116 it must be something different.

>I compiled my first 2.1.x kernel (126) today and noticed the same problem.
>Remembering seeing this post I checked out using the sound driver sources
>from 115 (didn't check if they differed), which compiled fine but produced
>the same fault.

On the AlphaStation 200, of the type that came from ONSALE in large quantities
recently, I had the same problem. Sound died between 2.1.115 and 2.1.116. I
went crazy trying to figure it out.

It is NOT the sound drivers themselves at fault. It is the PCI code. Should
you indeed have the same box, or any 'avanti' based system, the code just
needs a little patch. The following fix worked for me:

--- linux/arch/alpha/kernel/sys_sio.c.orig      Sat Oct 24 15:24:56 1998

 avanti_pci_fixup(void)
 {
        layout_all_busses(DEFAULT_IO_BASE, APECS_AND_LCA_DEFAULT_MEM_BASE);
-       sio_pci_fixup(noname_map_irq, 0x0b0a090f);
+       sio_pci_fixup(noname_map_irq, 0x0b0a0e0f);
        sio_fixup_irq_levels(sio_collect_irq_levels());
         enable_ide(0x26e);
 }


Technology Development and Special Projects
Office of Information Technology (OIT)
University of Massachusetts at Amherst
Voice: 413-545-3124   Fax: 413-545-3203
A115 Lederle Graduate Research Center
UMASS Amherst, MA 01003 USA

 
 
 

kernel 2.1.116+ lock up

Post by Riku Anttil » Thu, 29 Oct 1998 04:00:00



> It is NOT the sound drivers themselves at fault. It is the PCI code. Should
> you indeed have the same box, or any 'avanti' based system, the code just
> needs a little patch.

Thanks!
I have a Multia, but imitating you patch did lead to success. Here's the
change I did:

*** linux/arch/alpha/kernel/sys_sio.c.orig      Wed Oct 28 10:52:40 1998
--- linux/arch/alpha/kernel/sys_sio.c   Wed Oct 28 18:19:43 1998
*************** noname_pci_fixup(void)
*** 232 ****
!       sio_pci_fixup(noname_map_irq, 0x0b0a0f09);
--- 232 ----
!       sio_pci_fixup(noname_map_irq, 0x0b0a0f0e);

 
 
 

1. SMP in 2.1.116 Still Locking Up

Sadly I have been able to get the SMP on 2.1.116 on a dual
PPro 200 system to lock up when memory has become somewhat
tight.  I guess the fix in this latest kernel doesn't
work for my hardware.

HP Netserver LDPRO, 160MB RAM, RAID0 over 2 partitions
(14GB total).

I'm testing the RAID part by copying /usr to /mnt/raid0.
This fills up memory.  The box began to swap when suddenly
it just up and died.

I have 2 * 100MB swap partitions configured too.  The RAID
partitions are on the same disks as the swap partitions.
Is this a problem ?

/dev/sda1       2GB
/dev/sdb1       100MB Swap
/dev/sdb2       1GB /opt
/dev/sdb3       ~7.8GB RAID0-1
/dev/sdc1       100MB Swap
/dev/sdc2       1GB /var
/dev/sdc3       ~7.8GB RAID0-2

The last thing XOSVIEW reported before the system locked
like a safe was I was using 544KB of swap.  The interesting
thing I noted was that using fsck or copying files to
or from the raid partition caused huge blocks of memory to
be allocated to cache and buffer, that never appeared to be
free'd up.  I suppose this is the "new" behavior.

Hope this helps Linus figure it out.

|____________________________|______________________________________|

 My views do not necessarily represent those of the Hewlett Packard
 Company and should be taken with a large dose of salt or whatever
 passes for sodium in your neck of the woods/universe/continuum/etc...
 ___________________________________________________________________

2. Need app that returns 2 or more equal strings in a file for bash script

3. kernel 2.1.116+ and udb sound ???

4. which processor runs a process

5. kernel: nfs_stat_to_errno: bad nfs status return value 116

6. Intel EtherExpress PRO/100

7. Term 116 at sunsite.unc.edu

8. \\-- NETWORK PROBLEM

9. modules problem on 2.1.116/5

10. umask 117, 116

11. HELP! statfs error 116

12. bad nfs status return value: 116

13. Versatec 116 card driver - where