7448/2.6.20 problems (double boots or hangs)

My taiga board is now dead -- on our development board (with a taiga
kernel
and a custom modified dts) we see either


1) it hangs at the same point ("PID has table entries")
   or
2) it boots but it seems to restart the boot process after the "PID hash
table entries line"

This was never a problem with 2.6.10 or with 2.6.22.

I modified a driver, saw this oddness and changed to the original driver
(the oddness went away).  I did this about 3 times so I can say "ah
ha!!"

We're using 2.6.20.11...but I recall later to 2.6.20s were no better...

Any recommendations/hypothesis?

marty

arch: exit                   =20
[    0.000000] Zone PFN ranges:
[    0.000000]   DMA             0 ->   131072
[    0.000000]   Normal     131072 ->   131072
[    0.000000] early_node_map[1] active PFN ranges
[    0.000000]     0:        0 ->   131072       =20
[    0.000000] Built 1 zonelists.  Total pages: 130048
[    0.000000] Kernel command line:
nfsroot=3D13.121.21.120:/usr/local/tmp/ganymede ip=3Ddhcp
console=3DttyS0,115200
[    0.000000] mpic: Setting up MPIC "Tsi108_PIC" version 1.2 at
c0007400, max 4 CPUs                        =20
[    0.000000] mpic: ISU size: 24, shift: 5, mask: 1f

[    0.000000] mpic: Initializing for 24 sources    =20
[    0.000000] PID hash table entries: 2048 (order: 11, 8192 bytes)
                                                                  =20
[    0.000000] Using MPC7448 HPC2 machine description

12 more lines

-------- Original-Nachricht --------

Datum: Wed, 19 Mar 2008 17:52:16 -0400
Von: "Leisner, Martin" [off-list ref]
An: linuxppc-dev@ozlabs.org
CC: "Lund, Nathan \\(<redacted>)" <redacted>
Betreff: 7448/2.6.20 problems (double boots or hangs)

My taiga board is now dead -- on our development board (with a taiga
kernel
and a custom modified dts) we see either


1) it hangs at the same point ("PID has table entries")
   or
2) it boots but it seems to restart the boot process after the "PID hash
table entries line"

I observed the same behavior on my AmigaOne, if the PAGE_COHERENT flag was
set for page mappings by CPU_FTR_NEED_COHERENT. The 7448 CPU doesn't need
CPU_FTR_NEED_COHERENT to be set (no CPU bug that needs to be worked
around), but there is no define for the 7448 in arch/powerpc/cputable.h in
kernel v2.6.20 (but it's in v2.6.22!). Thus I guess it is detected as a
744[0|5|7], which have this flag set.
On the other side my hardware is _very_ buggy, so it's unlikely that this
could be the problem on your hardware.

Gerhard
-- 
Psssst! Schon vom neuen GMX MultiMessenger gehört?
Der kann`s mit allen: http://www.gmx.net/de/go/multimessenger