Re: Test 2.2.10 kernel for Mac IIci
> > Bad. Could you send me the Nubus probe output to check the card is
> > actually supported?
> Yes, the card works. It works in 2.2.10 and worked back in the 2.1.x
> series. Dmesg output is attached below.
Ok, I'll take a look.
> > Has anybody _ever_ gotten ethernet driver modules to work on Mac? I mean,
> > the cards are usually probed (and registered with the net layer) way
> > before the usual time (from the Nubus bus probe, to be precise). Maybe
> > what's missing in the module code is another Nubus scan ??
> I've never used _any_ modules on the Mac.
I've used modules a lot but mostly for slip/ppp or filesystems.
> Linux version 2.2.10 (firstname.lastname@example.org) (gcc version egcs-2.91.66
Debian GNU/Linux (egcs-1.1.2 release)) #1 Fri Oct 1 04:07:50 EDT 1999
That's well before the tarball I started from.
> NuBus: Scanning NuBus slots.
> Slot D:
> Board resource:
> type: [cat 0x1 type 0x0 hw 0x0 sw 0x0]
> name: MacCon Ethernet
> board id: 0x8
Nice and detailed ...
> Macintosh SCSI: resetting the SCSI bus... done
> mac_insert_irq: Warning: dev_id of MacSCSI-5380 is zero
Huh - needs cleaning up (but the dev_id argument isn't used by that
driver, and doesn't need to, or is there a Mac with more than one 5380
chip in it?).
> Checking for internal Macintosh ethernet (SONIC).. none.
> daynaport.c: v0.02 1999-05-17 Alan Cox (Alan.Cox@linux.org) and others
> eth0: asante: testing board: 16K memory - OK
> eth0: MacCon Ethernet in slot D (type asante)
> MAC 00:00:94:03:b7:a2 IRQ 60, shared memory at 0xfd0d0000-0xfd0d3fff.
Fine, so 2.2.10 separated the Nubus device probe from the card driver
init. Modules _should_ work that way. And mac8390.c has been modularized.
But after a close look at the code, I've come to conclude that the new
2.2.10 kernel doesn't use daynaport.c anymore (which is what your working
kernel used). So what's up here - problems with mac8390.c or with module
code? Your test with today's kernel should tell :-)