de4x5 hangs aic7xxx, tulip misses port on multiport card

Brian Ristuccia (brianr@osiris.978.org)
Tue, 20 Apr 1999 23:54:42 -0400

The de4x5 driver when loaded to support my 4 port DEC DC21140 based ethernet
card hoses my second SCSI controller. After loading the driver and bringing
up the interfaces, the SCSI card goes into a permenantly hosed condition
where every attempt to access devices connected to it results in timeouts
like this one: "scsi : aborting command due to timeout : pid 21989, scsi1,
channel 0, id 0, lun 0 Read (10) 00 00 92 2d 1b 00 00 08 00" Even after
unloading the de4x5 driver, the SCSI controller is still hosed. Another
aic7xxx controller shares the same IRQ and is not affected.

CPU0 CPU1
0: 47642 0 XT-PIC timer
1: 1346 1359 IO-APIC-edge keyboard
2: 0 0 XT-PIC cascade
7: 12058 0 IO-APIC-edge Crystal audio controller
12: 1042 1128 IO-APIC-edge PS/2 Mouse
13: 1 0 XT-PIC fpu
14: 6 1 IO-APIC-edge ide0
16: 13122 13133 IO-APIC-level bttv, DC21140 (eth2)
17: 3072461 3072630 IO-APIC-level aic7xxx, aic7xxx, DC21140 (eth1)
18: 977 994 IO-APIC-level Intel EtherExpress Pro 10/100 Ethernet, DC21140 (eth4)
19: 12 8 IO-APIC-level DC21140 (eth3)
NMI: 0
ERR: 0

The card _must_ share at least one IRQ with at least one of the aic7xxx
cards because there's only 4 physical IRQ pins on this Intel pr440fx board
and they're shared between the PCI slots and the onboard components. I've
upgraded to the latest BIOS, but aside from making my second cpu #12 instead
of #4, this had no noticable effect on anything.

The 4 port ethernet board uses a DEC DC21052 PCI bridge and 4 DEC DC21140
PCI 10/100 ethernet chips.

The tulip driver does strange things and refuses to recognise the first one
of the four ports on my card. It shows a bogus MAC address in the log file
and says things like "EEPROM not present" and "eth1: Old style EEPROM -- no
media selection information." I can not use ifconfig to bring up the
interface the error messages and bogus MAC refer to. The affected port also
happens to share IRQ 17 with my aic7xxx cards.

After loading and unloading the de4x5 driver, the tulip driver works
correctly, but my second SCSI card is still hosed. Only a reboot will fix
the affected SCSI card. After the reboot, the tulip driver will refuses to
use the first port ethernet again.

Any clues?

-- 
Brian Ristuccia
brianr@osiris.978.org
bristucc@baynetworks.com
bristucc@cs.uml.edu


-
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/