3c905B and version 0.99F of driver (again)

Paul S. Graham grahamp@ee.byu.edu
Fri Aug 14 20:21:03 1998


Sorry for taking so long to provide more information, but I have been
out of the lab for a few days.  Anyway, to be more specific, here is
the conf.modules file I use with RedHat 5.1 and some snippets
illustrating messages the drivers produced for various failures and
successes. 

##############################################################################
### conf.modules
##############################################################################
alias eth0 3c59x
options 3c59x debug=2
alias net-pf-4 off
alias net-pf-5 off

##############################################################################
# extracts from /var/log/messages
##############################################################################

Success with kernel version 2.0.34 and 3c59x.c v0.99E

Aug 14 10:06:35 shmack kernel: 3c59x.c:v0.99E 5/12/98 Donald Becker http://cesdis.gsfc.nasa.gov/linux/drivers/vortex.html
Aug 14 10:06:35 shmack kernel: loading device 'eth0'...
Aug 14 10:06:35 shmack kernel: eth0: 3Com 3c905B Cyclone 100baseTx at 0x1000, 00:10:4b:d0:ef:9b, IRQ 11
Aug 14 10:06:35 shmack kernel:   8K byte-wide RAM 5:3 Rx:Tx split, NWay Autonegotiation interface.
Aug 14 10:06:35 shmack kernel:   Enabling bus-master transmits and whole-frame receives.

Failure with kernel version 2.0.34 and 3c59x.c v0.99F

Aug 14 11:43:06 shmack kernel: 3c59x.c:v0.99F 8/7/98 Donald Becker http://cesdis.gsfc.nasa.gov/linux/drivers/vortex.html
Aug 14 11:43:06 shmack kernel:   The PCI BIOS has not enabled this device!  Updating PCI command 0000->0005.
Aug 14 11:43:06 shmack kernel: loading device 'eth0'...
Aug 14 11:43:06 shmack kernel: eth0: 3Com 3c905B Cyclone 100baseTx at 0x1000, 00:10:4b:d0:ef:9b, IRQ 11
Aug 14 11:43:06 shmack kernel:   8K byte-wide RAM 5:3 Rx:Tx split, Autonegotiate interface.
Aug 14 11:43:06 shmack kernel:   MII transceiver found at address 0, status 7869.
Aug 14 11:43:06 shmack kernel:   MII transceiver found at address 1, status 786d.
Aug 14 11:43:06 shmack kernel:   Enabling bus-master transmits and whole-frame receives.
Aug 14 11:43:06 shmack kernel: eth0: MII #0 status e1b7, link partner capability e1b7, setting full-duplex.

Success with kernel version 2.0.35 and 3c59x.c v0.99E

Aug 14 14:36:14 shmack kernel: 3c59x.c:v0.99E 5/12/98 Donald Becker http://cesdis.gsfc.nasa.gov/linux/drivers/vortex.html
Aug 14 14:36:14 shmack kernel: eth0: 3Com 3c905B Cyclone 100baseTx at 0x1000, 00:10:4b:d0:ef:9b, IRQ 11
Aug 14 14:36:14 shmack kernel:   8K byte-wide RAM 5:3 Rx:Tx split, NWay Autonegotiation interface.
Aug 14 14:36:14 shmack kernel:   Enabling bus-master transmits and whole-frame receives.

Failure with kernel version 2.0.35 and 3c59x.c v0.99F

Aug 14 17:25:05 shmack kernel: 3c59x.c:v0.99F 8/7/98 Donald Becker http://cesdis.gsfc.nasa.gov/linux/drivers/vortex.html
Aug 14 17:25:05 shmack kernel:   The PCI BIOS has not enabled this device!  Updating PCI command 0000->0005.
Aug 14 17:25:05 shmack kernel: eth0: 3Com 3c905B Cyclone 100baseTx at 0x1000, 00:10:4b:d0:ef:9b, IRQ 11
Aug 14 17:25:05 shmack kernel:   8K byte-wide RAM 5:3 Rx:Tx split, Autonegotiate interface.
Aug 14 17:25:05 shmack kernel:   MII transceiver found at address 0, status 7869.
Aug 14 17:25:05 shmack kernel:   MII transceiver found at address 1, status 786d.
Aug 14 17:25:05 shmack kernel:   Enabling bus-master transmits and whole-frame receives.
Aug 14 17:25:05 shmack kernel: eth0: MII #0 status e1b7, link partner capability e1b7, setting full-duplex.

Success with kernel 2.0.35 and 3c59x.c v0.99E (with debug=3)

Aug 14 17:45:40 shmack kernel: 3c59x.c:v0.99E 5/12/98 Donald Becker http://cesdis.gsfc.nasa.gov/linux/drivers/vortex.html
Aug 14 17:45:40 shmack kernel: eth0: 3Com 3c905B Cyclone 100baseTx at 0x1000, 00:10:4b:d0:ef:9b, IRQ 11
Aug 14 17:45:40 shmack kernel:   8K byte-wide RAM 5:3 Rx:Tx split, NWay Autonegotiation interface.
Aug 14 17:45:40 shmack kernel:   Enabling bus-master transmits and whole-frame receives.

##############################################################################

Notice that I don't get the same debug information from the 0.99E
version of the driver at the same debug levels.  Another thing to
mention is that the cards should be working in full-duplex on a 10BT
(not a 100BT) network.  It is not clear to me what the drivers
negotiated.

>>>>> "Donald" == Donald Becker <becker@cesdis1.gsfc.nasa.gov> writes:

    Donald> On Wed, 12 Aug 1998, Paul S. Graham wrote:
    >> I have installed RedHat 5.1 (version 2.0.34 of the kernel) on
    >> two Dell XPS R400 (Pentium II 400s) machines which came with
    >> the 3c905B cards.  I installed version 0.99E of the driver
    >> (3c59x.c) onto the machines and it seems to work quite well.
    >> 
    >> Today, I noticed that a new version of the driver (0.99F) was
    >> available on the WWW site, so I compiled and installed it.  On
    >> both machines I get errors such as under debug level 2:
    >> 
    >> The PCI BIOS has not enabled this device! Updating PCI command
    >> ...

    Donald> This is normal, and the same message that you should
    Donald> gotten from the previous driver versions.  I'll be
    Donald> removing this message in the next version now that most
    Donald> BIOSes are doing this.  (It's to support Win95 running old
    Donald> device drivers in "DOS Boxes".)

What debug level should I run the 0.99E driver in to get the same
information? As shown above, I even tried debug=3 and did not receive
the same information.  

Furthermore, someone (Andrew Chan) evidently has been able to get the
driver to work under RedHat 5.1 using the new kernel (2.0.35) on the
same hardware I am using (a Dell XPSR400 with BIOS version A03, an
3c905B card, and dual boot capability between Linux and NT).  One
noticeable difference was that he was using a Cisco 100BaseT switch,
where I am using 10BT at full-duplex (I think we are using 3Com
equipment).

I installed 2.0.35 from the RH updates to see if this would fix the
problem, but as I noted above, I still was met with the same failure
as before with the v0.99F driver.  I wonder if the problem has to do
with autonegotiating the the interface type or 100BT vs. 10BT somehow.

One last question: Is there supposed to be two MII tranceivers
detected with a single card? I suspect that this is normal.

Anyway, I figured the problem was worth a little more probing.

--Paul
=====================================
Paul Graham (grahamp@ee.byu.edu)
459 CB, Electrical Engineering Dept.
Brigham Young University
Provo, Utah 84602
phone: (801)378-7206  fax: (801)378-6586