Linux doesn't boot on Vertex4 based board.

[Leonid]

Hi:

I'm trying to port Linux on Vertex4 board, similar to Xilinx ML403
(PPC405 hard core). The main differences between my board and ML403 are:

- I'm using UARTLITE, not UART16550 core.
- I have only 32M of PLB_SDRAM (ML403 got 64M of DDR).
- Flash is not in the end of the address space (see boot snapshot
below).
- I have TEMAC instead of EMAC.
- On my board PPC405 is running on 100MHz vs ML403 300MHz.

I can be somewhat confused regarding ML403 characteristics since I don't
have actual ML403 yet - will be getting it in several days.

I successfully ported ML403 u-boot to my board, but couldn't get kernel
to boot. I tried linux-xilinx-26 from Monta Vista server, 2.6.19.2 from
DENX and last 2.6.20 from kernel.org. Latest one already has uartlite
ported, for previous two I tried to port myself. Below please find
typical u-boot output (I enabled DEBUG in cmd_boot file):

U-Boot> run akanfs
TFTP from server 192.168.0.141; our IP address is 192.168.0.213
Filename 'LM200/rel/1.0.1-20/uImage'.
Load address: 0x1000000
Loading:
#################################################################
 
#################################################################
         #######################################
done
Bytes transferred = 862398 (d28be hex)
## Booting image at 01000000 ...
   Image Name:   Linux-2.6.20
   Image Type:   PowerPC Linux Kernel Image (gzip compressed)
   Data Size:    862334 Bytes = 842.1 kB
   Load Address: 00000000
   Entry Point:  00000000
   Verifying Checksum ... OK
   Uncompressing Kernel Image ... OK
## Current stack ends at 0x01F966A0 => set upper limit to 0x00800000
## cmdline at 0x007FFF00 ... 0x007FFF9C
memstart    = 0x00000000
memsize     = 0x02000000
flashstart  = 0x22000000
flashsize   = 0x00400000
flashoffset = 0x00000000
sramstart   = 0x00000000
sramsize    = 0x00000000
bootflags   = 0x00000000
procfreq    =    100 MHz
plb_busfreq =    100 MHz
ethaddr     = 00:01:02:03:FD:71
IP addr     = 192.168.0.213
baudrate    = 115200 bps
No initrd
## Transferring control to Linux (at address 00000000) ...

I realize that can be 1001 reasons for such behavior, yet I'd appreciate
any ideas you can share with me. I tried use early serial output... no
success.

Thanks,

Leonid.