Subject: Joys with the new adaptec driver
To: None <port-macppc@netbsd.org>
From: Bill Studenmund <wrstuden@nas.nasa.gov>
List: port-macppc
Date: 03/10/2000 18:31:05
As mentioned on current-users, Frank van der Linden <frank@wins.uva.nl> is
working on a new adaptec driver, which should support the U2W Adaptec
cards.

I've made a little progress, and wanted to share my results to hopefully
spur someone else into working on it. :-) Frank can work on a lot of
things, but he doesn't have a powermac. :-)

I first grabbed Frank's driver (see messages on current-users), and then
had to make the following changes:

I added "#define AHC_ALLOW_MEMIO 1" to the top of the file. One area of
memory access will need to use memory mapped i/o on macppc, and this
enables it.

I changed the PCI_MAPREG_MEM_TYPE_32BIT around line 639 (after adding
above) to PCI_MAPREG_MEM_TYPE_64BIT .

I made the following change to sys/dev/pci/pci_map.c:

diff -u -r1.6 pci_map.c
--- pci_map.c   2000/01/25 22:30:04     1.6
+++ pci_map.c   2000/03/11 02:29:05
@@ -160,7 +160,10 @@
                break;
        case PCI_MAPREG_MEM_TYPE_64BIT:
                printf("pci_mem_find: 64-bit memory mapping register\n");
+               break;
+#if 0
                return (1);
+#endif
        default:
                printf("pci_mem_find: reserved mapping register type\n");
                return (1);

(which I will discuss on tech-kern).

Now the controller panics the kernel with a sequencer parity error. Frank
suggested playing with some of the parity enables, but that hasn't worked
for me yet. :-(

Take care,

Bill