2.4: PCIBIOS_MIN_MEM correct?

2.4: PCIBIOS_MIN_MEM correct?

[Joseph P. Garcia]

should PCIBIOS_MIN_MEM be 0x10000000?  I caught this when i noticed it is
used as the minimum address in the pci assign resource function, used in such
situations as the OF/ATY-overlap re-allocate in wallstreets.  My system would
panic with the lcd melting effect as some may recall.

I brute forced in the following lines in
drivers/pci/setup-res.c:pci_assign_resource()
     min = (res->flags & IORESOURCE_IO) ? PCIBIOS_MIN_IO : PCIBIOS_MIN_MEM;
+    if(res->start==0)
+         min = 0x83000000;

resulting in the same mapping used by Michael Schmitz earlier 2.2 patch.
(its the only re-alloc on my system, so it works :)  My kernel isn't
panicing.. yet.

Anyway, looking in /proc/iomem, one might get the idea that PCIBIOS_MIN_MEM
should be defined as 0x80000000.  This might resolve the problem I had in a
less brute way.  I haven't tried it myself yet.  waiting for a second opinion.

The macro appears to be new in 2.4, so its understandable why it looks like
most archs seem to have similar values.

Any one have any insight on this?

--
Joseph P. Garcia
http://www.execpc.com/~jpgarcia

** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: 2.4: PCIBIOS_MIN_MEM correct?

[Geert Uytterhoeven]

On Sun, 4 Feb 2001, Joseph P. Garcia wrote:
> should PCIBIOS_MIN_MEM be 0x10000000?  I caught this when i noticed it is

That value is suited for `PC' systems.

> used as the minimum address in the pci assign resource function, used in such
> situations as the OF/ATY-overlap re-allocate in wallstreets.  My system would
> panic with the lcd melting effect as some may recall.
>
> I brute forced in the following lines in
> drivers/pci/setup-res.c:pci_assign_resource()
>      min = (res->flags & IORESOURCE_IO) ? PCIBIOS_MIN_IO : PCIBIOS_MIN_MEM;
> +    if(res->start==0)
> +         min = 0x83000000;
>
> resulting in the same mapping used by Michael Schmitz earlier 2.2 patch.
> (its the only re-alloc on my system, so it works :)  My kernel isn't
> panicing.. yet.
>
> Anyway, looking in /proc/iomem, one might get the idea that PCIBIOS_MIN_MEM
> should be defined as 0x80000000.  This might resolve the problem I had in a
> less brute way.  I haven't tried it myself yet.  waiting for a second opinion.

The correct fix is to add bus resources that indicate the allowed region.
pci_assign_resource() will assign a region from the allowed region.

Gr{oetje,eeting}s,

						Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
							    -- Linus Torvalds


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/