[ECOS] Redboot-GDB

[Mark Salter]

>>>>> Michael Anburaj writes:

> Hi Nick/Mark,
> Let me summarize my understanding about RedbootÂ’s application invocation 
> through GDB.

...

> If the above is (more or less) correct, then the possibilities for the 
> application code not to run after continue is issues at the gdb prompt are:
> 1. ARM9 var issue: Since, I used an older version of ecos source, the cache 
> macros could have issues. I have synched up to the current ecos CVS 
> repository & will test it on Monday.
> 2. The other possibly: improper SDRAM configuration, screwing the stack 
> space and/or code. Which may not be the case, since the same code is running 
> in a product without any issues at the same CPU speed.
> 3. MMU page table inconsistency.

You can test the basic GDB stub entry/exit handling by typing:

  $k#6b

at the RedBoot prompt. This is the GDB 'kill' packet. You won't
see it echoed when typing, but after typing 'b' you should see
a '+' and you should be back at the RedBoot prompt.

Also, after loading your program, you could try the gdb 'stepi'
command to step the program one machine instruction at a time.

--Mark



-- 
Before posting, please read the FAQ: http://sources.redhat.com/fom/ecos
and search the list archive: http://sources.redhat.com/ml/ecos-discuss