Weird 2.0.38 instability

Raphael Manfredi (Raphael_Manfredi@pobox.com)
Fri, 19 Nov 1999 20:10:01 +0100


[I'm not a member of this list, please Cc replies to me -- RAM]

I've been running a UP machine with 2.0.36, then 2.0.38 without any
problems, as a workstation at home. Now that I have an SMP machine,
I moved it as a file server, DNS, etc... (but not a firewall, I have
an old 486 for that).

Well, since 3 weeks, I've been having messages looking like the
following (I've added "uname -a" output):

Linux lyon 2.0.38 #1 Sun Oct 3 15:59:30 MEST 1999 i586 unknown
------------------------------------------------------------------
Unable to handle kernel paging request at virtual address fb2ca9a9
current->tss.cr3 = 06ddf000, %cr3 = 06ddf000
*pde = 00000000
Oops: 0000
CPU: 0
EIP: 0010:[memcpy_toiovec+82/132]
EFLAGS: 00010212
eax: 0000001c ebx: 0000001c ecx: 00000007 edx: 080b10ac
esi: 3b2ca9a9 edi: 080b10ac ebp: 03ce3f70 esp: 03ce3ec0
ds: 0018 es: 002b fs: 002b gs: 002b ss: 0018
Process wwwoffled (pid: 6214, process nr: 71, stackpage=03ce3000)
Stack: 00000018 00000000 0000001c 03ce3f78 012a5414 0000001c 0014528c 03ce3f70
3b2ca9a9 0000001c 012a5414 03ce3f7c 00000000 00000000 31a2a971 012a5438
00000000 00000024 05063018 08f78948 21a3a19b 0014fb8e 012a5414 03ce3f78
Call Trace: [tcp_recvmsg+716/1036] [inet_recvmsg+114/136] [sock_read+171/192] [sys_read+192/232] [system_call+85/124]
Code: f3 a5 83 e3 03 89 d9 f3 a4 07 8b 5c 24 10 01 5c 24 1c 29 5c
------------------------------------------------------------------

The machine stays up, but I'm uncomfortable with this.
Since the machine is a server, it must remain up and running 24/7,
as well as reliable in its behaviour...

My questions are simple:

1) Is this problem known?
2) Should I downgrade to 2.0.36?
3) Should I upgrade to 2.2.13?

My SMP machine runs 2.2.13 but experiences "network lockup"
(i.e. it looses its connection on the LAN, but stays otherwise up),
but it could be an SMP-only problem.

Thank you for your advice, and, if possible, for pointers to
more information about this behaviour.

Raphael

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/