crazy ethernet interrupt

Hi group!

Since yesterday, i have only ethernet-transmission,
as long as i'm moving the mouse pointer somewhere
around on the screen. Stopping the mouse, the
transfer stops at once, moving again, transfer is
continued.

Does anyone have an idea?

System is OpenSuse 10.2 with all updates on
ASUS-Board with AMD.

Gunter

On Fri, 30 Mar 2007, in the Usenet newsgroup comp.os.linux.networking, in
article <euj057$mth$02$(E-Mail Removed)>, Gunter Lindemann wrote:

>Since yesterday, i have only ethernet-transmission,
>as long as i'm moving the mouse pointer somewhere
>around on the screen. Stopping the mouse, the
>transfer stops at once, moving again, transfer is
>continued.

What did you change? Look in /var/log/messages - and perhaps
/var/log/boot.msg. When the system booted, the mouse was using
some IRQ, and the Ethernet was using something else. You changed
_something_ and now the IRQs are not the same. I don't know why,
as I don't know what you did. Look also at the output of the
commands /sbin/ifconfig -a (which should show what IRQ the NIC is
using) and 'cat /proc/interrupts'

Old guy

Moe Trin wrote:

> What did you change?

I'm not aware that I have done anything but booting,
do my daily work and shutdown. I didn't install or
remove anything intentionally the last days.

> Look in /var/log/messages - and perhaps
> /var/log/boot.msg. When the system booted, the mouse was using
> some IRQ, and the Ethernet was using something else. You changed
> _something_ and now the IRQs are not the same. I don't know why,
> as I don't know what you did. Look also at the output of the
> commands /sbin/ifconfig -a (which should show what IRQ the NIC is
> using) and 'cat /proc/interrupts'

Thanks a lot! I had a hunch, that it has something to do with
any IRQ, but didn't know, where to look for it. I will do, as
you told me!

Greetings!
Gunter

Moe Trin wrote:

> What did you change?

Got an idea, what i actually did:
The USB-Mouse connector was loosened and i
put it back while the system was running.
I guess, since then i have that described
behavior.

> using) and 'cat /proc/interrupts'

brings me that:

CPU0
0: 3208816 XT-PIC timer
1: 7141 XT-PIC i8042
2: 0 XT-PIC cascade
6: 4 XT-PIC floppy
7: 1 XT-PIC parport0
8: 2 XT-PIC rtc
9: 1 XT-PIC acpi
11: 56079 XT-PIC aic7xxx
12: 22631 XT-PIC uhci_hcd:usb1, uhci_hcd:usb2, eth0
14: 110587 XT-PIC ide0
NMI: 0
LOC: 0
ERR: 0
MIS: 0

But i have no idea, how to change the
interrupts again.

Gunter

On Mon, 02 Apr 2007, in the Usenet newsgroup comp.os.linux.networking, in
article <euqq7f$t77$02$(E-Mail Removed)>, Gunter Lindemann wrote:

>Moe Trin wrote:
>
>> What did you change?
>
>Got an idea, what i actually did:
>The USB-Mouse connector was loosened and i
>put it back while the system was running.

Oh how I HATE these programs that "help" you manage the hardware in your
computer.

>I guess, since then i have that described behavior.

Your interrupt list shows that the kernel is expecting the USB and
Ethernet on the same IRQ, though in fact they may or may not all be
doing so.

>brings me that:
>
> CPU0
> 0: 3208816 XT-PIC timer
> 1: 7141 XT-PIC i8042
> 2: 0 XT-PIC cascade
> 6: 4 XT-PIC floppy
> 7: 1 XT-PIC parport0
> 8: 2 XT-PIC rtc
> 9: 1 XT-PIC acpi

Reasonable

> 11: 56079 XT-PIC aic7xxx

SCSI host adaptor

> 12: 22631 XT-PIC uhci_hcd:usb1, uhci_hcd:usb2, eth0

USB and NIC - what does the NIC show in /var/log/messages during boot?

> 14: 110587 XT-PIC ide0

IDE controller - normal.

>But i have no idea, how to change the
>interrupts again.

Normally, the IRQ is set using a BIOS setup of some kind. This may be set
to AUTO (or similar) which allows the O/S to kick things about. It's a
pain in the butt, especially when you have available "low" IRQs.

For a PCI interface - whether a plug-in card, or a motherboard mounted
device, there will be access to four IRQ wires - generally these are
ganged across all cards and motherboard devices - known as INTA through
INTD. These are connected to the Southbridge chip of the chipset. In the
original IBM design of the 1980s, the IRQs were connected to a priority
encoder, and the four or five output leads of that chip pulled the chain
on the CPU and identified which of 256 possible interrupt routines to
run. The interrupts are no longer "hard wired" a specific interrupt
vector, and this relationship is controlled in software.

What appears to have happened in your case is that the loose connection
confused some setup program. What I would be doing is poking about in the
BIOS setup and seeing if I can set the IRQ for the PCI cards to something
other than what the USB is using, then on re-booting, see that the kernel
can now see it as a separate device.

Old guy

Moe Trin wrote:

> USB and NIC - what does the NIC show in /var/log/messages during boot?

/var/log # grep NIC messages

nothing!

/var/log # grep nic messages

many lines containing the german word "nicht" (not)

similar on /var/log/boot.msg

Gunter

On Tue, 03 Apr 2007, in the Usenet newsgroup comp.os.linux.networking, in
article <eut0o8$hj2$02$(E-Mail Removed)>, Gunter Lindemann wrote:

>Moe Trin wrote:

>> USB and NIC - what does the NIC show in /var/log/messages during boot?
>
>/var/log # grep NIC messages
>
>nothing!

Not surprising - how about 'grep -2 eth /var/log/messages' (you may also
need to look at older logs, which might be named messages.0, messages.1
and so on which might be compressed - i.e. use zgrep or bzgrep if needed).
You may also want to look in those same logs to see how the USB stuff is
being configured.

>/var/log # grep nic messages
>
>many lines containing the german word "nicht" (not)

Not relevant to the problem, but use 'grep -w' to avoid that.

Old guy

Moe Trin wrote:

> Not surprising - how about 'grep -2 eth /var/log/messages'

I think, i found the entry, that logged the moment, when
i just booted, determined the loosened USB-Mouse and
plugged it again (Mar 28 14:30:56).

I saved the messages of this process in a textfile.
May i send it to you by email? It is about 200 lines
(16 kB).

Gunter