Even after the recent changes [1][2] to make amdtemp(4) attach to pchb(4)
(rather than directly to pci), it still does not work.
There are, as far as I can tell, two separate problems.
1. The current amdtemp.ioconf incorrectly sets pchb as the pseudo-root and
attempts to configure "amdtemp* at pchb?". The resulting ioconf.c ends up
with no valid configuration data for the device. This one is relatively easy
to resolve:
Index: src/sys/modules/amdtemp/amdtemp.ioconf
===================================================================
RCS file: /cvsroot/src/sys/modules/amdtemp/amdtemp.ioconf,v
retrieving revision 1.2
diff -u -p -r1.2 amdtemp.ioconf
--- src/sys/modules/amdtemp/amdtemp.ioconf 18 Aug 2011 20:57:11 -0000
1.2
+++ src/sys/modules/amdtemp/amdtemp.ioconf 20 Aug 2011 00:21:25 -0000
@@ -6,6 +6,6 @@ include "conf/files"
include "dev/pci/files.pci"
include "arch/x86/pci/files.pci"
-pseudo-root pchb*
+pseudo-root amdtempbus*
-amdtemp* at pchb?
+amdtemp* at amdtempbus?
2. The more difficult problem is that pchb(4) needs to have a rescan routine.
Much of pchbattach() should probably move into pchbrescan(), but I think it
would need some modifications to make sure that things don't get doubly
attached. (I ran into this same problem when iic(4) grew its own rescan a
few weeks ago, as part of modularization of the spdmem(4) and sdtemp(4)
drivers.)
I can fix #1, but I'm not at all familiar enough with pcibus and pchb to
tackle #2
[1] http://mail-index.netbsd.org/source-changes/2011/08/18/msg026092.html
[2] http://mail-index.netbsd.org/source-changes/2011/08/18/msg026093.html
-------------------------------------------------------------------------
| Paul Goyette | PGP Key fingerprint: | E-mail addresses: |
| Customer Service | FA29 0E3B 35AF E8AE 6651 | paul at whooppee.com |
| Network Engineer | 0786 F758 55DE 53BA 7731 | pgoyette at juniper.net |
| Kernel Developer | | pgoyette at netbsd.org |
-------------------------------------------------------------------------