Re: [Xen-devel] new(?) big ugly crash

[Derek Glidden <dglidden@xxxxxxxxxxxxxxx>]

On Jul 26, 2004, at 12:01 PM, Derek Glidden wrote:

> On Jul 26, 2004, at 8:10 AM, Keir Fraser wrote:
>
> > Can you easily reproduce this? All the subsequent oopses are because
> > our panic() when the mmu updates fail ought to be a BUG() [BUG doesn't
> > try to sync disks; panic does] ---- it's nothing to do with XFS.
ok, it blew up again and this time I got slightly better debugging 
info.  Again I had started and stopped a few VMs and did "xend stop" 
when it all went pear-shaped.
Some time shortly before the crash, I noticed this amidst the multitide 
of "GPF (0004)" messages that Xen emits:
(XEN) (file=traps.c, line=466) GPF (0004): fc528028 -> fc52a8f4
(XEN) (file=traps.c, line=352) Page fault: fc51c7c7 -> fc52a750
(XEN) (file=traps.c, line=466) GPF (0004): fc528028 -> fc52a8f4

I've never see a "Page fault" message before, so don't know if it's 
related.
Here's the first bit of the actual Oops with some Xen messages that 
might be of some use:
(XEN) (file=/opt/src/xeno/xeno-unstable.bk/xen/include/asm/mm.h, 
line=215) Unexpected type (saw c0000000 != exp e0000000) for pfn 
00003043
(XEN) DOM0: (file=memory.c, line=248) Bad page type for pfn 00003043 
(d0000002)
invalid operand: 0000
CPU:    0
EIP:    0819:[<c01e3755>]    Not tainted
EFLAGS: 00211282
eax: 00000022   ebx: c032c830   ecx: c10e4000   edx: fbff9000
esi: 00000000   edi: ffffffff   ebp: c11eff28   esp: c11eff1c
ds: 0821   es: 0821   ss: 0821
Process xfsbufd (pid: 7, stackpage=c11ef000)<1>
Stack: c02a9db4 0038710c c0444000 c11eff44 c01dd9df c038710c 03043063 
c12b6f5c
       c10e4000 00000000 c11eff58 c010852a c12b6f5c c12b6f5c c12b6f5c 
c11eff84
       c0105de1 00000000 00000000 c12b6f5c c11ee000 00000000 c031b100 
000a458b
Call Trace: [<c01dd9df>] [<c010852a>] [<c0105de1>] [<c0105b28>] 
[<c0105ab0>]
   [<c01d2127>] [<c01d8c6e>] [<c01d2080>]


Again it caused a "cascade" of xfsbufd Oopses until it rebooted itself.

I'm not sure why ksymoops continues to insist "code line not seen."  
I've compiled on just about all the kernel-hacking options possible.  
If there's something I'm not doing right, I'd be happy to make it go 
the right way so it would stop whining about that.

# ksymoops -m /boot/System.map-2.4.26-xen0  oops.txt
ksymoops 2.4.9 on i686 2.4.26-xeno-xen0.  Options used
     -V (default)
     -k /proc/ksyms (default)
     -l /proc/modules (default)
     -o /lib/modules/2.4.26-xeno-xen0/ (default)
     -m /boot/System.map-2.4.26-xen0 (specified)

SGI XFS with no debug enabled
Kernel panic: Failed to execute MMU updates
SGI XFS with no debug enabled
invalid operand: 0000
CPU:    0
EIP:    0819:[<c01e3755>]    Not tainted
Using defaults from ksymoops -t elf32-i386 -a i386
EFLAGS: 00211282
eax: 00000022   ebx: c032c830   ecx: c10e4000   edx: fbff9000
esi: 00000000   edi: ffffffff   ebp: c11eff28   esp: c11eff1c
ds: 0821   es: 0821   ss: 0821
Process xfsbufd (pid: 7, stackpage=c11ef000)<1>
Stack: c02a9db4 0038710c c0444000 c11eff44 c01dd9df c038710c 03043063 
c12b6f5c
       c10e4000 00000000 c11eff58 c010852a c12b6f5c c12b6f5c c12b6f5c 
c11eff84
       c0105de1 00000000 00000000 c12b6f5c c11ee000 00000000 c031b100 
000a458b
Call Trace: [<c01dd9df>] [<c010852a>] [<c0105de1>] [<c0105b28>] 
[<c0105ab0>]
   [<c01d2127>] [<c01d8c6e>] [<c01d2080>]
Warning (Oops_read): Code line not seen, dumping what data is available


>>EIP; c01e3755 <_flush_page_update_queue+75/80>   <=====

>>ebx; c032c830 <update_queue+10/4000>
>>ecx; c10e4000 <_end+d6f15c/41071bc>
>>ebp; c11eff28 <_end+e7b084/41071bc>
>>esp; c11eff1c <_end+e7b078/41071bc>

Trace; c01dd9df <destroy_context+17f/190>
Trace; c010852a <__mmdrop+2a/50>
Trace; c0105de1 <schedule+251/390>
Trace; c0105b28 <schedule_timeout+58/b0>
Trace; c0105ab0 <process_timeout+0/20>
Trace; c01d2127 <pagebuf_daemon+a7/200>
Trace; c01d8c6e <arch_kernel_thread+2e/40>
Trace; c01d2080 <pagebuf_daemon+0/200>



-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
"I think that's what they mean by   |
"nickels a day can feed a child."   |       http://www.eff.org/
I thought, "How can food be so      | http://www.anti-dmca.org/
cheap over there?"  It's not, they  |--------------------------
just eat the nickels." -- Peter Nguyen



-------------------------------------------------------
This SF.Net email is sponsored by BEA Weblogic Workshop
FREE Java Enterprise J2EE developer tools!
Get your free copy of BEA WebLogic Workshop 8.1 today.
http://ads.osdn.com/?ad_id=4721&alloc_id=10040&op=click
_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxxxx
https://lists.sourceforge.net/lists/listinfo/xen-devel