Great! and Damn! :-)
>If you are booting LINUX from ARC via MILO, you could change the kernel's
>setup of the IRQ's assigned to the PCI slots in arch/alpha/kernel/bios32.c
>fixup code (ie in the code that inits the PIRQ registers).
I'm actually using aboot 0.3... for some strange reason, I'm a little
reluctant to blow away flashware, and SRM w/aboot is more familiar
SunOS/BSD territory for me. But your pointer to bios32.c was very
helpful.
>The PCI "slot" that the TGA is in may currently be setup in the PIRQs with
>IRQ 15, for example; simply switch whatever the TGA slot's IRQ for IRQ 9 in
>the PIRQ setup code, thus giving the DE435 the non-conflicting IRQ, as the
>TGA doesn't use its IRQ anyway.
Just for kicks, I tried physically swapping the TGA and the DE435 PCI
cards, but IRQ 9 moved with the DE435. It weirded me out for a second,
but seemed to make it clear that maybe what I really wanted to do was
tell bios32.c either that interrupt 9 wasn't available for allocation
(since it isn't, eh?), or at least sort it behind my only remaining
unused PCI interrupt - IRQ 10. Having done the latter, I'm still having
the same trouble with hearing traffic, but the card did take IRQ 10 (and
a lot of <sc ...> console errors went away), so I think I'm on the right
track.
I notice that an ifconfig shows the errant Ethernet as not having a base
address. Could be related :-)
>BTW, the AS250 is a *nice* box, quite fast from all I've heard... :-)
'Tis true! Small and ripping fast. Nice array of standard ports, too...
I needed some extra disk space, so I just grabbed a Sun SCSI disk, threw
it on the external micro-SCSI connector, and went to town.
>From the config file, MSS_DMA was set to 0; if this card is at all like the
>"card" built into the MULTIA/UDB, it will use DMA 3. And note that the MULTIA
>"card" is hardwired to the adress/IRQ/DMA, ie there's no registers for setting
>them as on a normal MSS. Now, this may be incorrect info for the card in the
>AS250, so your mileage may vary. If your card *has* those registers, it should
>be settable to avoid IRQ 9, by choosing some other IRQ that's free.
The card has jumpers for rehoming the I/O base address, but apparently
not the IRQ. The AlphaStation 250 User Information manual says that the
PCXBJ device sits on slot 2, at IOBASE 530, 388 (eh?), with DMAmode and
chan 0, 1 (eh?), and IRQ 9. SRM's "iconfig" utility tells me that SRM's
add_sound macro chooses I/O 530, IRQ 9, and DMA 0. That's why I tried
using DMA 0, at least. Having worked around the IRQ issue, I'm leaving
the sound device unconfigured for now, though I'd like to get it working
at some point.
>Do you have any machine and/or soundcard specs for this beastie? If not, I
>may know someone who is at least familiar with it enough to confirm/deny our
>suspicions...
Painfully few. I'm going on scraps of info that the 250 User Info manual
has (which only sporadically includes sound card info in its various
tables) and output from SRM's iconfig (which I tend to trust a little
more).
dp
Copyright © 1995-1997 Red Hat Software. Legal notices