> The oops isn't very useful unless it's been fed to ksymoops for address
> resolution. Ksymoops.cc is in linux/scripts btw :) and just in case you
> don't know how to use it..
> ksymoops path_to_System.map < your_oops > save_output
> Any way to eliminate the irq sharing between net/scsi? I've seen comments
> from guru types suggesting this before.
I did this when I got in to work this morning and it still
oopsed...:-< I've run the oops through ksymoops as recommended and I get
the following:
>>EIP: c0171926 <ip_rule_match+36/250>
Trace: c016ed62 <fib_validate_source+86/164>
Trace: c01720f0 <ip_fw_check+1ec/584>
Trace: c017347c <ipfw_input_check+48/50>
Trace: c014c0e7 <call_in_firewall+2f/4c>
Trace: c015a75f <ip_rcv+1b7/238>
Trace: c014eac4 <net_bh+158/1bc>
Trace: c011b771 <do_bottom_half+85/a8>
Trace: c010c7e2 <do_IRQ+4a/54>
Trace: c010aed4 <ret_from_intr>
Trace: c0106000 <init_task_union>
Trace: c010937b <cpu_idle+37/78>
Trace: c0107f70 <this_must_match_init_task+1f70/2000>
Code: c0171926 <ip_rule_match+36/250>
Code: c0171926 <ip_rule_match+36/250> 23 53 08 andl 0x8(%ebx),%edx
Code: c0171929 <ip_rule_match+39/250> 39 13 cmpl %edx,(%ebx)
Code: c017192b <ip_rule_match+3b/250> 0f 95 c1 setne %cl
Code: c017192e <ip_rule_match+3e/250> 0f b6 d1 movzbl %cl,%edx
Code: c0171931 <ip_rule_match+41/250> f6 43 18 01 testb
$0x1,0x18(%ebx)
Code: c0171935 <ip_rule_match+45/250> 74 09 je c0171940
<ip_rule_match+50/250>
Code: c0171937 <ip_rule_match+47/250> 85 d2 testl %edx,%edx
Code: c0171939 <ip_rule_match+49/250> 74 00 je c017193b
<ip_rule_match+4b/250>
Code: c017193b <ip_rule_match+4b/250> 90 nop
Code: c017193c <ip_rule_match+4c/250> 90 nop
Code: c017193d <ip_rule_match+4d/250> 90 nop
This isn't quite what I expected to see. Which part of this stack
is the closest to currently running? I was expecting something IRQ
related since this has been a sore spot for SMP it seems. (Especially
with 4 procs, that is why I've only been using 2 for quite some time
now.) If there is any more information I can provide at all, I'd be more
than happy to do so. I would really like to see this resolved before 2.2
and I'm sure that that is getting fairly near:->
Another strange thing happened to me just a bit ago. I rebooted
on the 2.1.115 kernel and it saw both of my tulip cards on irq 5. When I
boot to 2.0.34 or 2.0.35 one shows up on irq 5 and the other on irq 10.
It looks like strange things are afoot in irq land. I may try some more
things later, but I think I'll wait for some more suggestions first so
I'm not just stabbing in the dark. (BTW: I tried to run the mptrace.c
program, but it dumped core after outputting about one screen of data.
Am I doing something wrong?)
Thanks Much,
Daniel Thom
-
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.altern.org/andrebalsa/doc/lkml-faq.html