kernel/network problem
Hi,
I've got a 2.2.4 kernel, a 3c905BTX2 network card (bus matering
enabled). I am using potato. My 3com card is compiled a a module.
My problem is that quite often my network hangs. Usually, a "ifconfig
eth0 down; sh /etc/init.d/network" is enough to restart it. But, why do
I need to do that ? Another example, if I use tcpdump to try to find
what is the problem, my network hangs in less than 2 minutes. Last time
I had a problem, I had this output on the console:
Mar 28 19:56:30 groumph kernel: NET4: AppleTalk 0.18 for Linux NET4.0
Mar 28 19:56:35 groumph atalkd[26784]: restart (1.4b2+asun2.1.1)
Mar 28 19:56:36 groumph kernel: Oops: 0000
Mar 28 19:56:36 groumph kernel: CPU: 0
Mar 28 19:56:36 groumph kernel: EIP: 0010:[<c48478fe>]
Mar 28 19:56:36 groumph kernel: EFLAGS: 00010287
Mar 28 19:56:36 groumph kernel: eax: 00000000 ebx: 00000001 ecx:
c0d07e78 edx: 00000000
Mar 28 19:56:36 groumph kernel: esi: c30e4a80 edi: c30e4a80 ebp:
00000000 esp: c0d07e9c
Mar 28 19:56:36 groumph kernel: ds: 0018 es: 0018 ss: 0018
Mar 28 19:56:36 groumph kernel: Process atalkd (pid: 26784, process nr:
109, stackpage=c0d07000)
Mar 28 19:56:36 groumph kernel: Stack: c30e4a80 00000000 c48453b9
c30e4a80 c0d07f5c c30e4a80 c0d07f54 c0d06000
Mar 28 19:56:36 groumph kernel: 000000ec 0000ffff c4845b2d
c30e4a80 00008916 00000000 bffffb68 c324dd00
Mar 28 19:56:36 groumph kernel: 00000002 c0d06000 c0d07f44
00000001 c027ba40 40014000 c0d07f8c 00000000
Mar 28 19:56:36 groumph kernel: Call Trace: [<c48453b9>] [<c4845b2d>]
[sk_alloc+18/60] [d_alloc+24/336] [<c4847087>] [sock_ioctl+29/36]
[sys_ioctl+303/328]
Mar 28 19:56:36 groumph kernel: [system_call+52/64]
Mar 28 19:56:36 groumph kernel: Code: 66 8b 40 58 66 3d 05 03 74 06 66
3d 00 02 75 0e 56 e8 64 ff
Mar 28 19:56:36 groumph afpd[26800]: main: atp_open: Cannot assign
requested address
to restart the network, I had also to do a "rmmod 3c59x " before the
"ifconfig eth0 down; sh /etc/init.d/network".
Here is the output I get when my network card is loaded:
Mar 28 20:01:40 groumph kernel: 3c59x.c:v0.99H 11/17/98 Donald Becker
http://cesdis.gsfc.nasa.gov/linux/drivers/vortex.html
Mar 28 20:01:40 groumph kernel: eth0: 3Com 3c905B Cyclone 100baseTx at
0x6800, 00:a0:24:5b:2a:38, IRQ 10
Mar 28 20:01:40 groumph kernel: 8K byte-wide RAM 5:3 Rx:Tx split,
autoselect/Autonegotiate interface.
Mar 28 20:01:40 groumph kernel: MII transceiver found at address 24,
status 786d.
Mar 28 20:01:40 groumph kernel: MII transceiver found at address 0,
status 786d.
Mar 28 20:01:40 groumph kernel: Enabling bus-master transmits and
whole-frame receives.
Could DMA and/or bus mastering be the problem ? Because I get this
message when booting:
Mar 28 17:31:43 groumph kernel: PCI: Using configuration type 1
Mar 28 17:31:43 groumph kernel: PCI: Probing PCI hardware
Mar 28 17:31:43 groumph kernel: PCI: 00:38 [1106/0586]: Work around ISA
DMA hangs (00)
Mar 28 17:31:43 groumph kernel: Activating ISA DMA hang workarounds.
Last thing: I get this problem on any distribution/kernel (slink/2.0.36
for example). Can someone help me ?
Thanks in advance
--
\\|// VDB g TRY Atlantis BBS. Free BBS under GNU licence
(O O) ********************************************
~~~~OOO~(_)~oOOO~~~ <:o) Available on telnet://bbs.resus.univ-mrs.fr
mailto:vdb@pipo.com :-xxx Infos on http://bbs.resus.univ-mrs.fr/
~~~~ooO~~~~~Ooo~~~~ ~~:-) ********************************************
Reply to: