Bug 5216

I can also reproduce this on an hp compaq NC6120.

Most recent working kernel: 2.6.12.6
Distribution: Debian (Etch)
Hardware Environment: HP Compaq NC6120 (Pentium M)
Software Environment: Grub loading Initrd

Also, the bug seems to be solved in 2.6.13-mm1.

If additional info is required please advice.

The bug seems to be solved in 2.6.13.1 for me.

It might be the 2.6.13's PCI regression. Closing!

Sorry for the delay, but I did not find time to test 2.6.13.1 earlier.

It is clearly not fixed in 2.6.13.1, I uploaded the new pictures to http://
schabi.de/temp/kernelbug.

It now tells me:
PCI: Using ACPI for IRQ routing
PCI: If a device doesn't work, try "pci=routeirq". If it helps, post a report
PCI: Cannot allocate ressource region 7 of bridge 0000:00:1c.1
PCI: Cannot allocate ressource region 8 of bridge 0000:00:1c.1
PCI: Cannot allocate ressource region 9 of bridge 0000:00:1c.1

and hangs.

When specifying pci=routeirq as boot parameter, it dumps me a list of Interrupt 
routings (see http://schabi.de/temp/kernelbug/100_1061.jpg ) and then the same 
three "cannot allocate ressource" messages.

Thanks,
Markus

I did some additional tests, all of 2.6.13-rc2, 2.6.13-rc4 and 2.6.13-rc5 fail 
with the same bug.

1.6.13-mm1 also hangs, but gives additional information (see 
http://schabi.de/temp/kernelbug/100_1067.jpg ), the start of the message is:

BUG: soft lockup detected on CPU#0!

Pid: 1, comm:    swapper
EIP: 0060:[<c023fb44>] CPU: 0
EIP is at strstr+0x20/0x30
 EFLAGS: 00000283   Not tainted (2.6.13-mm1)

Could you please give me the acpidump output from the system?It sounds not 
like an interrupt issue.
You can get the tool(acpidmp) from:
http://www.kernel.org/pub/linux/kernel/people/lenb/acpi/utils/

Created attachment 6043 [details]
acpidump output

[resend it, since bugzilla is crashed and my input is lost]
The issue is strange. I'm tring to reproduce it here. Could you please attach 
your config file? I tried your 2.6.13.1 config, but it's not good (it enable 
CONFIG_HIGHMEM64G=y, but pentium-m doesn't support PAE, so this option will 
make system unbootable).

This is in fact the config file I used. I enabled the CONFIG_HIGHMEM because a 
friend of mine told me that the new Pentium M processors support and will also 
enable the Stack Execution Protection when using the Linux kernel with PAE.

Btw, /proc/cpuinfo lists pae under the flags, and 2.6.12.6 works with the same 
HIGHMEM settings:

schabi@lunix:~$ cat /proc/cpuinfo
processor       : 0
vendor_id       : GenuineIntel
cpu family      : 6
model           : 13
model name      : Intel(R) Pentium(R) M processor 2.00GHz
stepping        : 8
cpu MHz         : 798.088
cache size      : 2048 KB
fdiv_bug        : no
hlt_bug         : no
f00f_bug        : no
coma_bug        : no
fpu             : yes
fpu_exception   : yes
cpuid level     : 2
wp              : yes
flags           : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov 
pat clflush dts acpi mmx fxsr sse sse2 ss tm pbe est tm2
bogomips        : 1576.14

schabi@lunix:~$ zcat /proc/config.gz | grep HIGHMEM
# CONFIG_NOHIGHMEM is not set
# CONFIG_HIGHMEM4G is not set
CONFIG_HIGHMEM64G=y
CONFIG_HIGHMEM=y
# CONFIG_DEBUG_HIGHMEM is not set
schabi@lunix:~$ uname -a
Linux lunix 2.6.12.6 #1 Thu Sep 8 13:32:38 CEST 2005 i686 GNU/Linux
schabi@lunix:~$ dmesg | grep MEM
127MB HIGHMEM available.
896MB LOWMEM available.

Created attachment 6082 [details]
my 2.6.14-rc1 config

I just tried 2.6.14-rc2 with NO, 4G and 64G Highmem, it's the same.

2.6.13.2 also does not fix it.

Sorry, I missed your CPU has NX support. I'll try your config soon.

Did you have any progress lately?

I can also run some debug/test software or try patches if you need it, although 
I might show a high latency.

Btw, do you think looking for a BIOS update could be worth trying?

No, I can't reproduce it using your config file. Can you still boot 
a '2.6.12.2' kernel in the system? Updating BIOS might be not valuable for 
this bug to me. Could you please narrow which version kernel is the last one 
working? Otherwise, I have no other idea.

I just rechecked: 2.6.12.2 still boots, as well as 2.6.12.6, both with the 
identical config files I put up on http://schabi.de/temp/kernelbug/.

2.6.12.6 is the last known kernel to work, all tested 2.6.13.X and 2.6.14.X 
fail, as I posted earlier. I compiled all the kernels using make-kpkg, they are:

kernel-image-2.6.13_10.00.Custom_i386.deb
kernel-image-2.6.13.1_10.00.Custom_i386.deb
kernel-image-2.6.13.2_10.00.Custom_i386.deb
kernel-image-2.6.13-mm1_10.00.Custom_i386.deb
kernel-image-2.6.13-rc2_10.00.Custom_i386.deb
kernel-image-2.6.13-rc3_10.00.Custom_i386.deb
kernel-image-2.6.13-rc4_10.00.Custom_i386.deb
kernel-image-2.6.13-rc5_10.00.Custom_i386.deb
kernel-image-2.6.13-rc6_10.00.Custom_i386.deb
kernel-image-2.6.13-rc7_10.00.Custom_i386.deb
kernel-image-2.6.14-rc1_10.00.Custom_i386.deb
kernel-image-2.6.14-rc2_10.00.Custom_i386.deb
kernel-image-2.6.14-rc2hm4g_10.00.Custom_i386.deb
kernel-image-2.6.14-rc2hm64g_10.00.Custom_i386.deb

Do you have any ideas which kernel trees / pre releases I could test?

Should I use git and do a binary search on all changes between 2.6.12 and 2.6.
13?

Are there any diagnostics (ACPI debug etc.) I can turn on?

Another Idea might be a GCC difference, AFAIR all my kernels are compiled with 
gcc 4.0.1 (Debian 4.0.1-2). If you have a different compiler, could you give me 
your compiled image so I can test it?

In the meanwhile, I'll try to create a minimal kernel config (leave out or 
modulize all unneded stuff) on 2.6.14-rc2 and try whether I get this one to 
boot.

Thanks,
Markus

Somebody mentioned in acpi-dev mail list this issue could be fixed by 
pci=noacpi. Does it work for you?

I'll try, thanks for the tip.

Btw, as I'm currently a little confused about NX: I read that successfull NX 
initialization will show a message in dmesg, but I could not find any trace 
there. Does that mean that my processor does not support NX? As I have 1 Gig of 
RAM, will I need Highmem4G then?

Thanks,
Markus

pci=noacpi works for 2.6.14-rc2 - the kernel boots up fine, but it does not work 
on 2.6.13.2.

I still have some problems with the ipw2200 module under 2.6.14-rc2, but that's 
a different issue.

>As I have 1 Gig of RAM, will I need Highmem4G then?
No, just enable 'CONFIG_HIGHMEM64G'.
Can you give me the dmesg with pci=noacpi? Also does acpi=noirq work in the 
system?

acpi=noirq does work under 2.4.14-rc2, I did not test 2.6.13.X yet.

Created attachment 6177 [details]
dmesg with acpi=noirq

Created attachment 6178 [details]
dmesg with pci=noacpi

*** This bug has been marked as a duplicate of 5221 ***