Re: mpc8xx-2.2.13: Now booting the kernel

LiuTao wrote:

Hi Brendan:

mpc8xx-2.2.13 still doesn't run on my board. I use emulator(EST tool)
to debug the system through JTAG port. I download the zImage to the
board, then run it. Something displayed on serial port:
loaded at:     00800000 0080B1C8
board data at: 00800158 00800180
relocated to:  007F0100 007F0128
zimage at:     00830263 0088A2B1
avail ram:     0088B000 01000000

Linux/PPC load:
Uncompressing Linux...done.
Now booting the kernel

Now, the program has unziped the image section(vmlinux) to 0x0,
then jump to 0x0 to enter the linux kernel. Then the program
call identify_machine() in head.S. Here, the problem happened.
Because the content in location of identify_machine() is not
correct. I don't know the reason. Maybe unzip has problem?
I sent help main to the maillist, but nobody give me any
suggestions.
My board is customed too. What kind of debug tool are you using
to debug linux on your board?

I'm not using anything.  I have a Wiggler that comes with the SDS SingleStep
deubbuger.  This is OK to run the bootloader program that uncompresses the
kernel into memory but it will not work properly when the kernel is actually
running.  I think it is to do with the MMU and the debugger is not setup to
handle this.  Your emulator may experience similar problems.
I generally use printk in the areas of interest to see what is going on.

What was the problem in identify_machine() ?  I'll have a look at it and see
if I can see anything.

Brendan Simon.

Brendan J Simon wrote:

quoted

Hi Liu.
I have had embedded-2.2.5 linux running on my custom mpc860 board and
have now upgraded to mpc8xx-2.2.13.  I am having problems similar to the
problems you had.
I am getting "Now booting the kernel" and then nothing else.
You said below that you understood and resolved the problem.
What was the problem and how did you resolve it ?

I thought this was a cache problem but I have disabled the caches in
arch/ppc/kernel/head.S and arch/ppc/mbxboot/head.S but the problem is
still there.

Thanks,
Brendan Simon.

-------- ORIGINAL MESSAGE --------

Hi all:

I understood the problem and resolved it.

LiuTao

LiuTao wrote:

quoted

Hi all:

I am porting Linux to a board which has MPC860MH50, 16M RAM and 2M
Flash.
I am using visionPROBE(emulator of EST) to debug my program.
I download the zImage to the board by emulator, then run it.
I can see the following:

loaded at:     00800000 0080B1C8
board data at: 00800158 00800180
relocated to:  007F0100 007F0128
zimage at:     00830263 0088A2B1
avail ram:     0088B000 01000000

Linux/PPC load:
Uncompressing Linux...done.
Now booting the kernel

Now, I find that the program stoped at 0x111C.

So I set a breakpoint at "blr" in head.S at which the program will

jump

quoted

to the kernel. I ran the program again, it stoped at the breakpoint.
Then I ran it step by step, I found that it jumped to 0x0000000C, I
think
that's the correct address. Then, I ran it continuously, it stopped at

quoted

0x111C again.

That's very strange. If I don't set breakpoint at "blr" but set a
breakpoint
at 0x000C, the program can't stop at the breakpoint but stop at

0x111C.

quoted

If I
set breakpoint at "blr" and run it step by step, the program can get
0x000C.

Do you have any suggestions?
Thanks!

LiuTao

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