Mailing List Archive: frontend and backend crash cause by kernel paging?

frontend and backend crash cause by kernel paging?

Aug 6, 2003, 3:23 PM

Post #1 of 3 (980 views)

The frontend and backend crashed simultaneously on me today. This was
unusual because it was the first time that it has crashed on me while
viewing a recording (of course it may have been recording another show
at the time, not sure)

var/log/messages had this to say about the incident:

Aug 6 17:00:02 void kernel: Unable to handle kernel paging request at
virtual a
ddress 2e323139
Aug 6 17:00:02 void kernel: printing eip:
Aug 6 17:00:02 void kernel: c013217d
Aug 6 17:00:02 void kernel: *pde = 00000000
Aug 6 17:00:02 void kernel: Oops: 0002
Aug 6 17:00:02 void kernel: CPU: 0
Aug 6 17:00:02 void kernel: EIP: 0010:[<c013217d>] Not tainted
Aug 6 17:00:02 void kernel: EFLAGS: 00010046
Aug 6 17:00:02 void kernel: eax: 2e323135 ebx: cffb6dd4 ecx:
cf44e000 edx
: c50ab000
Aug 6 17:00:02 void kernel: esi: 0000001c edi: cf44eec0 ebp:
c12729c0 esp
: c1309efc
Aug 6 17:00:02 void kernel: ds: 0018 es: 0018 ss: 0018
Aug 6 17:00:02 void kernel: Process kswapd (pid: 5, stackpage=c1309000)
Aug 6 17:00:02 void kernel: Stack: 00000296 cf44eec0 cf44eec0 c0131a7e
cffb6dd4
cf44eec0 cf44eec0 c013c777
Aug 6 17:00:02 void kernel: cffb6dd4 cf44eec0 c013e9fd cf44eec0
00000000
c12729c0 0000232c c0291bb0
Aug 6 17:00:02 void kernel: c01326e5 c12729c0 000001d0 c1308000
000001db
000001d0 00000003 0000001f
Aug 6 17:00:02 void kernel: Call Trace: [<c0131a7e>] [<c013c777>]
[<c013e9fd
>] [<c01326e5>] [<c01328e1>]
Aug 6 17:00:02 void kernel: [<c0132966>] [<c0132a94>] [<c0132af9>]
[<c0132c2d
>] [<c0105000>] [<c01072be>]
Aug 6 17:00:02 void kernel: [<c0132b90>]
Aug 6 17:00:02 void kernel:
Aug 6 17:00:02 void kernel: Code: 89 50 04 c7 01 00 00 00 00 8b 43 08
8d 53 08
89 48 04 89 01

And 2 more entries almost indentical

Any ideas on how to prevent this from happening again in the future?

Re: frontend and backend crash cause by kernel paging? [ In reply to ]

matt at solecize

Aug 6, 2003, 3:31 PM

Post #2 of 3 (956 views)

Permalink

One other note, the other log entries were attributed to mythfrontend
and mythbackend processes (as opposed to kswapd in this example)

Matt wrote:

> The frontend and backend crashed simultaneously on me today. This was
> unusual because it was the first time that it has crashed on me while
> viewing a recording (of course it may have been recording another show
> at the time, not sure)
>
> var/log/messages had this to say about the incident:
>
> Aug 6 17:00:02 void kernel: Unable to handle kernel paging request at
> virtual a
> ddress 2e323139
> Aug 6 17:00:02 void kernel: printing eip:
> Aug 6 17:00:02 void kernel: c013217d
> Aug 6 17:00:02 void kernel: *pde = 00000000
> Aug 6 17:00:02 void kernel: Oops: 0002
> Aug 6 17:00:02 void kernel: CPU: 0
> Aug 6 17:00:02 void kernel: EIP: 0010:[<c013217d>] Not tainted
> Aug 6 17:00:02 void kernel: EFLAGS: 00010046
> Aug 6 17:00:02 void kernel: eax: 2e323135 ebx: cffb6dd4 ecx:
> cf44e000 edx
> : c50ab000
> Aug 6 17:00:02 void kernel: esi: 0000001c edi: cf44eec0 ebp:
> c12729c0 esp
> : c1309efc
> Aug 6 17:00:02 void kernel: ds: 0018 es: 0018 ss: 0018
> Aug 6 17:00:02 void kernel: Process kswapd (pid: 5, stackpage=c1309000)
> Aug 6 17:00:02 void kernel: Stack: 00000296 cf44eec0 cf44eec0
> c0131a7e cffb6dd4
> cf44eec0 cf44eec0 c013c777
> Aug 6 17:00:02 void kernel: cffb6dd4 cf44eec0 c013e9fd
> cf44eec0 00000000
> c12729c0 0000232c c0291bb0
> Aug 6 17:00:02 void kernel: c01326e5 c12729c0 000001d0
> c1308000 000001db
> 000001d0 00000003 0000001f
> Aug 6 17:00:02 void kernel: Call Trace: [<c0131a7e>] [<c013c777>]
> [<c013e9fd
> >] [<c01326e5>] [<c01328e1>]
> Aug 6 17:00:02 void kernel: [<c0132966>] [<c0132a94>] [<c0132af9>]
> [<c0132c2d
> >] [<c0105000>] [<c01072be>]
> Aug 6 17:00:02 void kernel: [<c0132b90>]
> Aug 6 17:00:02 void kernel:
> Aug 6 17:00:02 void kernel: Code: 89 50 04 c7 01 00 00 00 00 8b 43 08
> 8d 53 08
> 89 48 04 89 01
>
>
> And 2 more entries almost indentical
>
> Any ideas on how to prevent this from happening again in the future?
>
>

Re: frontend and backend crash cause by kernel paging? [ In reply to ]

ray at comarre

Aug 6, 2003, 3:37 PM

Post #3 of 3 (980 views)

Permalink

At 05:23 PM 8/6/2003 -0500, Matt wrote:
>The frontend and backend crashed simultaneously on me today. This was
>unusual because it was the first time that it has crashed on me while
>viewing a recording (of course it may have been recording another show at
>the time, not sure)
>
>var/log/messages had this to say about the incident:
>
>Aug 6 17:00:02 void kernel: Unable to handle kernel paging request at
>virtual a
>ddress 2e323139
[rest deleted]

I can't be certain from the info you posted, but this looks like a kernel
problem with swap. If you are running a swap partition, make sure it is
good (readable and writable). If you are not running a swap partition, try
adding one (it can be quite small) and see if it eliminates the problem.

On one (only one) of the Linux hosts I run, I saw a similar problem for
quite some time ... not with Myth, but with other processes that filled
available RAM, like big ftp transfers ... and I fixed it by adding a small
swap partition. I expect it is some quirky combination of kernel settings,
RAM size, and who knows what that makes this occur only on one host, but
you may have a similar problem.

Even if I have the details wrong, this is a kernel crash, not a Myth crash.