On 2002.11.23 18:11 Matt Thomas wrote:

> What's in SPR_SDR1?  Can you print out the contents of the
> SR registers?
SDR1:   0x7ff0000
SR[]:   0x00fffff0,   0x00fffff0,   0x00fffff0,   0x00fffff0,  =20
        0x00fffff0,   0x00fffff0,   0x00000006,   0x00000007,  =20
        0x00000008,   0x00000009,   0x0000000a,   0x0000000b,  =20
        0x0000000c,   0x0000000d,   0x0000000e,   0x0000000f,  =20
iBAT[]: 0x00001ffe 0x00000012, 0x00000000 0x00000000,=20
        0x00000000 0x00000000, 0x00000000 0x00000000,=20
dBAT[]: 0x00001ffe 0x00000012, 0x00000000 0x00000000,=20
        0x00000000 0x00000000, 0x00000000 0x00000000,=20
Dumped with pmap.c:pmap_print_mmuregs() in front of the mtsrin
instruction.
The pagetable is allways the last 64 kB of available RAM, 128 MB in this
case. (SDR1 is set to 0xfff0000 on the 256 MB machine.)

> Well, something is using 0x6000.0000 for some purpose.  I wonder if
> OFW is using it as a direct store SR?  I've never seen that but it
> is possible.
Hmmm. But why does it fail at SR0 when FIRMWORKSBUGS is enabled?
Changing SR0 should make no difference as there is the BAT mapping. I
should have a closer look at the OFW interaction.=20
--=20



tsch=FC=DF,
         Jochen

Homepage: http://www.unixag-kl.fh-kl.de/~jkunz/