Skip site navigation (1)Skip section navigation (2)
Date:      Fri, 18 Feb 2005 10:23:10 -0700 (MST)
From:      Warner Losh <imp@bsdimp.com>
To:        amd64@freebsd.org
Subject:   PC Card subpart to R3000 thread
Message-ID:  <20050218.102310.74705720.imp@bsdimp.com>

next in thread | raw e-mail | index | archive | help
# I'll note that I really like to be cc'd on changes that impact the
# pccard part of the system :-)

: Yeah the recipient of the fix just emailed me about this, I am
: guessing that the #if 0 is the uncommitable part. Is there any way
: that this can be done by the kernel (the PCI reg write, that is)? Is
: there any reason that it can't be done there?

I'd be extremely reluctant to commit the #if 0 part of the fix.  The
problem is that we don't quite do bus numbering/renumbering correctly
for pci busses in general.  There's some kludges in cardbus bridge to
cope with this, but I don't like them much at all.

So, yes, something can be done about this in the kernel, but that
something is rather more complicated than this overly simplistic
kludge.

--- src/sys/dev/pci/pci_pci.c.orig	Thu Jul  1 03:46:28 2004
+++ src/sys/dev/pci/pci_pci.c		Sat Jan 22 01:21:50 2005
@@ -319,6 +319,8 @@
				start = sc->iobase;
			if (end > sc->iolimit)
				end = sc->iolimit;
+			if (start <= end)
+					ok = 1;


This looks OK to me.  At first I thought it was crazy and unnecessary
to set this, but I think it might be OK.  It does seem very odd that
we've not seen this before now.  Oh, wait, the <= should be <.  ==
would imply that the range is 0.

Warner



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20050218.102310.74705720.imp>