[parisc-linux] evms_vgscan crash hppa kernel

[parisc-linux] evms_vgscan crash hppa kernel

[joel.soete]

Hi all,

I just apply evms-0.2.3 patches against hppa kernel 2.4.9-pa78.
This kernel compiles, boots and runs also fine on my b180.

But when I launch evms_vgscan the kernel crash with following message:

...
Kernel Fault: Code=15 regs=124fc6c0 (Addr=faf00918)

     YZrvWESTHLNXBCVMcbcbcbcbOGFRQPDI
PSW: 00000000000001001111111100001111
r0-3     00000000 10117000 1018c314 00000000
r4-7     10085e40 00000001 00000000 faf00908
r8-11    17b57c80 124bf300 00023674 faf00788
r12-15   00000000 ffffffff 000c2310 00000000
r16-19   00000000 000255c8 00014000 00023d60
r20-23   00000200 00000000 00000200 102ac400
r24-27   00000200 124fc808 00000000 1029c010
r28-31   00000000 00000000 124fc6c0 102ac420
sr0-3    00000000 00000000 00000000 000002d5
sr4-7    00000000 00000000 00000000 00000000

IASQ: 00000000 00000000 IAOQ: 1018c350 1018c354
 IIR: 48fa0020    ISR: 00000000  IOR: faf00918
 CPU:        0   CR30: 124fc000 CR31: 10348000
 ORIG_R28: 124fc508

a "grep 1018c3 /boot/System.map"
1018c394 t evms_ioctl_cmd_get_minor

I would like to debug this by obtaining a core dump of the kernel after the 
reboot but I do not know how? (I did not find back the mail explaining how to do 
it in detail)

Thanks in advance for help,
        Joel




-------------------------------------------------
This mail sent through Tiscalinet Webmail (http://webmail.tiscali.be)

Re: [parisc-linux] evms_vgscan crash hppa kernel

[Matthew Wilcox]

On Wed, Nov 07, 2001 at 03:56:16PM +0100, joel.soete@freebel.net wrote:
> IASQ: 00000000 00000000 IAOQ: 1018c350 1018c354
>  IIR: 48fa0020    ISR: 00000000  IOR: faf00918
>  CPU:        0   CR30: 124fc000 CR31: 10348000
>  ORIG_R28: 124fc508
> 
> a "grep 1018c3 /boot/System.map"
> 1018c394 t evms_ioctl_cmd_get_minor
> 
> I would like to debug this by obtaining a core dump of the kernel after the 
> reboot but I do not know how? (I did not find back the mail explaining how to do 
> it in detail)

1018c350 is before 1018c394, so it's actually in the previous function.

-- 
Revolutions do not require corporate support.

Re: [parisc-linux] evms_vgscan crash hppa kernel

[joel.soete]

Quoting Matthew Wilcox <willy@debian.org>:

> On Wed, Nov 07, 2001 at 03:56:16PM +0100, joel.soete@freebel.net wrote:
> > IASQ: 00000000 00000000 IAOQ: 1018c350 1018c354
> >  IIR: 48fa0020    ISR: 00000000  IOR: faf00918
> >  CPU:        0   CR30: 124fc000 CR31: 10348000
> >  ORIG_R28: 124fc508
> > 
> > a "grep 1018c3 /boot/System.map"
> > 1018c394 t evms_ioctl_cmd_get_minor
> > 
> > I would like to debug this by obtaining a core dump of the kernel
> after the 
> > reboot but I do not know how? (I did not find back the mail explaining
> how to do 
> > it in detail)
> 
> 1018c350 is before 1018c394, so it's actually in the previous function.

Well ... in System.map I found:

1018bfc4 t evms_ioctl_cmd_get_logical_disk
1018c088 t evms_ioctl_cmd_get_logical_disk_info
1018c1c8 t evms_ioctl_cmd_sector_io
1018c394 t evms_ioctl_cmd_get_minor
1018c494 t evms_ioctl_cmd_get_volume_data

So it should be in 'evms_ioctl_cmd_sector_io', is it right?

Thank Matthew,
    Joel

> 
> -- 
> Revolutions do not require corporate support.
> 
> 


-------------------------------------------------
This mail sent through Tiscalinet Webmail (http://webmail.tiscali.be)

Re: [parisc-linux] evms_vgscan crash hppa kernel

[Matthew Wilcox]

On Wed, Nov 07, 2001 at 04:13:38PM +0100, joel.soete@freebel.net wrote:
> So it should be in 'evms_ioctl_cmd_sector_io', is it right?

Correct.

-- 
Revolutions do not require corporate support.

Re: [parisc-linux] evms_vgscan crash hppa kernel ... continue

[joel.soete]

Hi all,

I retry now with kernel 2.4.16-pa16 and evms cvs dated 2011-12-11 and
unfortunately it still crashes but at crash time I've got he following
interesting message to me:

--evms: Sector IO ioctl: Allocating sector buffer.
WARNING! Stack pointer and cr30 do not correspond!

.... a dump ...


Can some one explain me what it means and what can be the reason.
I post here the question because I see that it cames from
arch/parisc/kernel/traps.c but have no knowledge in the register naming for this
processor.

Thanks in advance for help,
        Joel


Quoting joel.soete@freebel.net:

> Quoting Matthew Wilcox <willy@debian.org>:
> 
> > On Wed, Nov 07, 2001 at 03:56:16PM +0100, joel.soete@freebel.net
> wrote:
> > > IASQ: 00000000 00000000 IAOQ: 1018c350 1018c354
> > >  IIR: 48fa0020    ISR: 00000000  IOR: faf00918
> > >  CPU:        0   CR30: 124fc000 CR31: 10348000
> > >  ORIG_R28: 124fc508
> > > 
> > > a "grep 1018c3 /boot/System.map"
> > > 1018c394 t evms_ioctl_cmd_get_minor
> > > 
> > > I would like to debug this by obtaining a core dump of the kernel
> > after the 
> > > reboot but I do not know how? (I did not find back the mail
> explaining
> > how to do 
> > > it in detail)
> > 
> > 1018c350 is before 1018c394, so it's actually in the previous
> function.
> 
> Well ... in System.map I found:
> 
> 1018bfc4 t evms_ioctl_cmd_get_logical_disk
> 1018c088 t evms_ioctl_cmd_get_logical_disk_info
> 1018c1c8 t evms_ioctl_cmd_sector_io
> 1018c394 t evms_ioctl_cmd_get_minor
> 1018c494 t evms_ioctl_cmd_get_volume_data
> 
> So it should be in 'evms_ioctl_cmd_sector_io', is it right?
> 
> Thank Matthew,
>     Joel
> 
> > 
> > -- 
> > Revolutions do not require corporate support.
> > 
> > 
> 
> 
> -------------------------------------------------
> This mail sent through Tiscalinet Webmail (http://webmail.tiscali.be)
> 


-------------------------------------------------
This mail sent through Tiscalinet Webmail (http://webmail.tiscali.be)

Re: [parisc-linux] evms_vgscan crash hppa kernel

[Grant Grundler]

joel.soete@freebel.net wrote:
> I would like to debug this by obtaining a core dump of the kernel after the 
> reboot but I do not know how? (I did not find back the mail explaining how
> to do it in detail)

Linux (from Linus' tree) does not support core dumps.
I'm only aware of coredump support patch from SGI.

The folks that manage HPUX core dump tools expresed interest
in porting the same tools to parisc-linux...assuming
we would dump core in the same format.

If a spec published for that, would anyone commit to implementing it?

FWIW, I think much of the work is portable to ia64-linux in case
that would help justify an HP project to pick up/fund this work.

grant

Grant Grundler
parisc-linux {PCI|IOMMU|SMP} hacker
+1.408.447.7253