[tulip] Tulip and PowerPC endian issue?

Jim Rowe rowejames@acmsystems.com
Thu Jan 30 13:56:02 2003


As  you suggested I used the tulip driver from Scyld in the
netdrivers-3.3.tgz package and I still get the same byte swapped MAC
address.

I built the module with:

make KERN_INCLUDE=/usr/src/linux/include MODULEDIR=/tmp/pcmcia-test
PCMCIA=/usr/src/linux/include/pcmcia

Then:
cd /lib/modules/2.4.19-rc1-acms1/kernel/drivers/pcmcia/
insmod pcmcia_core.o
insmod yenta_socket.o
insmod ds.o
cd /usr/src/netdrivers-3.3
insmod pci-scan.o
insmod tulip.o

The first 32-bit word is still byte swapped, however the second now
shows zeros instead of the 79:F0 as last time.

Again my MAC address should be 00:10:7A:69:F0:79.

Here is my log file:
 Jan 30 10:40:22 eth0 kernel: Linux Kernel Card Services 3.1.22 
Jan 30 10:40:22 eth0 kernel:   options:  [pci] [cardbus] 
Jan 30 10:40:23 eth0 kernel: Yenta IRQ list 0000, PCI irq30 
Jan 30 10:40:23 eth0 kernel: Socket status: 30000006 
Jan 30 10:40:23 eth0 kernel: Yenta IRQ list 0000, PCI irq30 
Jan 30 10:40:23 eth0 kernel: Socket status: 30000820 
Jan 30 10:40:25 eth0 kernel: cs: cb_alloc(bus 2): vendor 0x1317, device
0x1985 
Jan 30 10:40:25 eth0 kernel: PCI: moved device 02:00.0 resource 0 (101)
to 8000 
Jan 30 10:40:25 eth0 kernel: PCI: moved device 02:00.0 resource 1 (200)
to 80400000 
Jan 30 10:40:25 eth0 kernel: PCI: moved device 02:00.0 resource 6 (7201)
to bf800000 
Jan 30 10:40:25 eth0 kernel: PCI: Enabling device 02:00.0 (0000 -> 0003)
Jan 30 10:40:26 eth0 kernel: pci-scan.c:v1.11 8/31/2002  Donald Becker
<becker@scyld.com> http://www.scyld.com/linux/drivers.html 
Jan 30 10:40:27 eth0 kernel: tulip.c:v0.95f 11/17/2002  Written by
Donald Becker <becker@scyld.com> 
Jan 30 10:40:27 eth0 kernel:   http://www.scyld.com/network/tulip.html 
Jan 30 10:40:27 eth0 kernel:   The PCI BIOS has not enabled the device
at 2/0!  Updating PCI command 0003->0007. 
Jan 30 10:40:27 eth0 kernel: eth1: ADMtek Centaur-C rev 17 at
0xc902f000, 69:7A:10:00:00:00, IRQ 30. 
Jan 30 10:40:27 eth0 kernel: eth1:  MII transceiver #1 config 1100
status 7849 advertising 05e1. 
Jan 30 10:40:27 eth0 kernel:   PCI latency timer (CFLT) is unreasonably
low at 0.  Setting to 32 clocks. 

-- Jim



On Wed, 2003-01-29 at 16:58, Donald Becker wrote:
> On 29 Jan 2003, Jim Rowe wrote:
> 
> > I'm using a Netgear FA511 CardBus PC Card with a Ricoh RL5c476 PCI to
> > Cardbus bridge on a PPC platform.
> > 
> > The MAC address of my card is 00:10:7A:69:F0:79, however ifconfig and
> > /var/log/messages report it to be 69:7A:10:00:79:F0. Has anyone else run
> > into this problem?
> > I'm using the tulip.o driver from the 2.4.19 kernel, along with the
> > kernel based card services.
> 
> The driver must endian-correct for station address.
> The driver in the kernel did not copy the code from my driver correctly,
> resulting in this bug.  The driver from Scyld _does_ work correctly.
> 
> > Nov 30 00:28:56 eth0 kernel: Linux Tulip driver version 0.9.15-pre11
> > (May 11, 2002) 
> > Nov 30 00:28:56 eth0 kernel: eth1: ADMtek Comet rev 17 at 0x8000,
> > 69:7A:10:00:79:F0, IRQ 30
-- 
Jim Rowe
Advanced CounterMeasure Systems
Email: jrowe@acmsystems.com