AW: AW: Linux 2.4.24 on PQ2FADS

Stefano,

-----Ursprüngliche Nachricht-----
Von: Stefano Gaiotto [mailto:Stefano.Gaiotto@marconi.com]
Gesendet: Dienstag, 9. März 2004 17:05
Betreff: Re: AW: Linux 2.4.24 on PQ2FADS

I had not an entry for Hip7 such that you send me! Thak you very much
for your help.

I'm not sure if MPC8275 is actually a HIP7 core. We used a MPC8280 on
our boards. It is important that the CPU ID is correct in the cputable
file. In our case, with the missing CPU ID the board always hang in the
init program.

I'm going to chek my SDRAM initialization because, now, I see
corrupted stamps, like these:

## Booting image at 00100000 ...
   Image Name:   2.2.4 Kernel for PQ2FADS
   Image Type:   PowerPC Linux Kernel Image (gzip compressed)
   Data Size:    571353 Bytes = 558 kB
   Load Address: 00000000
   Entry Point:  00000000
   Verifying Checksum ... OK
   Uncompressing Kernel Image ... OK
Memory BAT mapping: BAT2=32Mb, BAT3=0Mb, residual: 0Mb
                                                      ÀLinux
version 2.4.24
(gai
otto@gexs22) (gcc version 3.3.2) #5 Tue Mar 9 17:00:09 CET 2004
                                                               ÀADS setup
arch
ÀO
n node 0 totalpages: 8192
                         Àzone(0): 8192 pages.
                                              Àzone(1): 0 pages.

Àzone(2): 0
page
s.
  ÀKernel command line: root=/dev/nfs rw
nfsroot=172.16.246.111:/home/gaiotto/fa
dsroot nobats ip=172.16.119.3
 ÀÀÀÀ$                 ÀADS init IRQ. NR_IRÀÀÀÀÀ

Maybe that your load address (0x100000) is too low and is overwritten
when the kernel is uncompressed....

Best regards,

Thomas

Thomas Schäfer [off-list ref] on 09/03/2004 14:24:39

just a wild guess:

check the cpu_specs[] array in your arch/ppc/kernel/cputable.c file.
It should contain something like this entry for your 8275 CPU:

    { /* 82xx (Hip7 603e cores) */
      0xffff0000, 0x80820000, "82xx (Hip7)",
      CPU_FTR_SPLIT_ID_CACHE | CPU_FTR_CAN_DOZE | CPU_FTR_USE_TB |
      CPU_FTR_CAN_NAP,
      COMMON_PPC,
      32, 32,
      __setup_cpu_603
    }

The CPU ID (0x80820000 in this case) must match your 8275 CPU ID.
Otherwise, generic PPC will be assumed which leads to strange
behaviour as soon as the  MMU will be enabled by the kernel.

second guess:

Because of the random errors you described I would check the SDRAM
settings of the board:
- check ALL settings of the memory controller against the settings
  described in the data sheet of your SDRAM.
- check memory controller initialization sequence against the sequence
  described in the data sheet of the SDRAM and the memory controller
  of the CPU
- if both are correct: check signal integrity on your board

quoted

-----Ursprüngliche Nachricht-----
Von: Stefano Gaiotto
Gesendet: Dienstag, 9. März 2004 13:53
Betreff: Linux 2.4.24 on PQ2FADS

I'm trying to have Linux 2.4.24 working on my evaluation board
PQ2FADS-VR with ppc8275, but I'm suffering of a lot of instability
problems. The Boot-loader I use is Uboot-1.0.0 witch initialise
correctly the board at speed 66,133,199 and correctly fills bd_info
struct. But when I issue the command "=> bootm 100000" I observe
every time a different behaviour: sometimes I don't see anything,
sometimes the board stops in "Calibrating delay loop", sometimes
after having written the line:

Memory: 30936k available (984k kernel code, 344k data, 52k init, 0k
highmem)

Or the line

POSIX conformance testing by UNIFIX

Only occasionally it reaches the init program (stopping here).

The same behaviour I observed with the images delivered by the
Arabella distribution!

Do you have any suggestion? MMU? Cache? Hardware?

** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/