I'm trying to debug my kernel using PPCBoot to load the kernel image via
tftpboot and then stop it in start_kernel.
I do the following.
1. Power up the to PPCBoot prompt
2. Load the vmlinux.img image via tftpboot created using the mkimage tool.
3. Load vmlinux into gdb that's connected to the bdi2000 at addr 0xc0000000
4. set a break at start_kernel (breaking on _start didn't work at all).
5. Telling PPCBoot to bootm
6. The debugger stops in start_kernel
Now, when I try to step or next I get the following error from gdb:
Breakpoint 2, start_kernel () at init/main.c:530
(gdb) display/i $pc
1: x/i $pc 0xc0142660 <start_kernel+16>: lis r9,-16364
(gdb) s
Cannot access memory at address 0x1100
I was under the impressions that the MMU has already been initilized at this
point and we should be running with vaddrs, hence the 0xc0000000
start_kernel function. But, it appears that for some reason gdb is confused
about v to p.
What am I doing wrong here?
Confused.
-bmb
** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/
I was looking through some stuff from late 1998.
At this point the 405 was believed to have an on-chip USB as well (405x3).
Does anybody know why it disappeared? Meaningless to say that I could use it
now :-)
Sincerely,
Dr. Karsten Jeppesen
VP of Technology
Total Impact
[off-list ref]
----------------------------------------------------------------------
Lie, n.:
A very poor substitute for the truth, but the only one
discovered to date.
----------------------------------------------------------------------
** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/
"Boerner, Brian" wrote:
I'm trying to debug my kernel using PPCBoot to load the kernel image via
tftpboot and then stop it in start_kernel.
Ensure you are using the latest 2_4_devel kernel and Abatron firmware.
We have been developing something that will work better for Linux,
with less overhead and more flexibility on both sides. It was required
due to 2.4.x page table changes, and I finally got the Linux part of
it checked in.
4. set a break at start_kernel (breaking on _start didn't work at all).
That's because _start isn't a virtual address as indicated in the
vmlinux
symbol table. Set a hard breakpoint at zero if you must stop there.
-- Dan
** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/