Subject: Re: More pyxis/500a info..
To: Matthew Jacob <firstname.lastname@example.org>
From: Jeff Roberson <email@example.com>
Date: 01/03/2000 23:41:26
I tried forcing the rev to 2, and commenting out that one line which it
hangs on. In both cases the kernel boots and detects all devices
succesfully, but in both cases I get weird errors accessing devices on the
PCI bus. For instance, I can configure my nic, but I can not send any
data through it. Does any one have any pointers to documentation for this
chipset? And for these problems specifically? I'm just poking around in
the dark here. I'm going to look through the freebsd source tomorrow to
see if I can find anything particularly enlightening.
On Mon, 3 Jan 2000, Matthew Jacob wrote:
> Date: Mon, 3 Jan 2000 22:44:27 -0800 (PST)
> From: Matthew Jacob <firstname.lastname@example.org>
> To: Jeff Roberson <email@example.com>
> Cc: firstname.lastname@example.org
> Subject: Re: More pyxis/500a info..
> Ummm- the same trick is used in a number of places- it may be that we all
> die here because we're fooling with CIA setup that the prom is using, but
> I'm a bit rusty at this end of the code for NetBSD to know for sure.
> There've been a lot of unexplained issues/problems with a lot of this
> goop- I'll be looking at the 8200 breakage this week or next (finally) to
> try and fix that end of it.
> Have you tried commenting this code out (i.e., forcing the rev to > 1)?
> On Mon, 3 Jan 2000, Jeff Roberson wrote:
> > So I found out the exact line that my machine hangs on. It's in
> > cia_dma_init (cia_dma.c) at aproximately line 215. The line reads:
> > REGVAL(CIA_PCI_W2BASE) = CIA_PYXIS_BUG_BASE |
> > CIA_PCI_WnBASE_SG_EN | CIA_PCI_WnBASE_W_EN;
> > I found this simply by adding printf's here and there around suspicious
> > portions of code. Interestingly enough, if I comment this out the kernel
> > boots, but I don't know how reliable it is. I'll try installing from this
> > kernel in a bit. Anyway, would anyone care to guess at why this causes my
> > machine to hang?
> > Thanks,
> > Jeff