[3c509] 3c515 issue

Kris Van Hees kvanhees@sinenomine.net
Tue Oct 1 14:20:01 2002


On Tue, Oct 01, 2002 at 08:56:57AM -0400, Donald Becker wrote:
> On Tue, 1 Oct 2002, Kris Van Hees wrote:
> > Hm, I shuld have realized that from using the driver in the 2.4.19 kernel.
> > Unfortunately, the Scyld driver at ftp://ftp.scyld.com/pub/network/isa/3c515.c
> > does not compile with the 2.4.19 kernel.  Here is what I get:
> 
> Yes, that is correct.
> There is a cost to the users that results from the political
> machinations a few years ago that resulted in my drivers updates being
> rejected, and other people's bogus changes being accepted instead.

As far as the compilation issue: is there any hope you or someone else at
Scyld might put an updated driver for download somewhere?  Or do you have
an updated driver that can be purchased through Scyld or as part of the
Beowulf stuff Scyld sells?

Concerning the issue I seem to be running in: the driver seems to be doing a
media selection that is never really going to switch to anything other than
MII given that the 3c515-diag util gives the following information:
3c515-diag.c:v1.00 7/24/2000 Donald Becker http://scyld.com/diag/index.html
Looking for card at 280.

A ISA Fast EtherLink III board was detected at I/O 0x280.
Window 0: 0000 0000 0000 0000 0000 0095 0000 0000.
Window 1: 6790 0000 0000 2002 8000 00ff 7ffc 2002.
Window 2: 6000 0408 dd87 0000 0000 0000 00da 4002.
Window 3: 001b 0162 0000 0000 f040 7fff 7fff 6002.
Window 4: 0000 06d0 2000 0c80 0002 8c02 0000 8002.
Window 5: 1ffc 1ffc 00da 1ffc 0007 0298 00da a002.
Window 6: 0000 0000 0000 0800 0000 0225 021e c002.
Window 7: 0000 0000 0000 0000 8000 00ff 0000 e002.
Operation registers: 0000 0000 0000 e002 8000 00ff 7ffc e002.
Boomerang registers: ffff ffff ffff ffff ffff ffff ffff ffff
  (at 0x680)       : ffff ffff ffff ffff ffff ffff ffff ffff.
Corkscrew registers: 0483 008a 0000 0000 0001 01bf 0000 ffff.
EEPROM contents:
 0060 0804 87dd 010a c333 0036 474b 6d50 0487 5150 0060 0804 87dd 3f10 0000 0000
 11c7 0000 001b 0061 0000 0003 0000 00fe 6d50 5150 87dd 0804 0a76 1010 1782 3300
 6f43 206d 6146 7473 4520 6874 7265 694c 6b6e 4920 4153 5015 506d 0251 a822 2a9e
 06e8 0047 0280 03e0 2020 4079 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000  64K word-wide RAM 1:1 Rx:Tx split, autoselect/MII interface.
   Done card at 0x280.

Your driver uses the value at bytes 8-9 in Window 3 (0xf040).  Given the media
masks listed in the driver, that implies that only the mask for the MII entry
would ever match.  Is that intentional?  Or is the 'media' actually the
transceiver options (like BNC connector vs RJ-45 connector)?

	Kris