--- Wolfgang Denk <wd at denx.de> wrote: > Single-stepping over an RFI instruction? I don't > think you'll get > this working.
thanks, I discovered that the 'hard' way!. However I found the problem but I don't have any clue what is happenning. On the start_here after turn_on_mmu there are some memory access instructions: start_here: .... c000361c addi r11,r11,_endl at l c0003620 lis r8, __bss_start at h c0003624 addi r8,r8,__bss_start at l c0003628 subf r11,r8,r11 .... c0003644 stwu r0,4(r8) the problem is that for some reason the when the c000361c instruction is executed the program counter jumps to the c0003644 instruction. R8 don't have the value it should be and the current value is way out of the 16MB declare initialy on initial_bats. This cause an SIGSTOP signal cause the MMU can't translate the address requested. I think this should be a debuging or compiler issue cause I don't have any explanation about what is happenning. I repeated the procedure several times and it is not on the same instruction but is the same cause, and in the same procedure (start_here). Previous instructions were not executed, and a memory access is tried with anunexpected value on the register. Any suggestions? Debugging Environment: BDI2000 -> gdb 5.0 MPC750 (6xx,7xx,7400) using a PReP Type Thanks, Freddy Lugo ** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/
