Re: Using gdb to debug FIQ on arm9 (imx233), and running into "Cannot find bounds of current function"

[Petr HluzÃn <petr dot hluzin at gmail dot com>]

Hi

(The regulars did not reply, at least you'll get a low quality reply from me.)

On 17 September 2013 16:09, Juha Lumme <juha.lumme@gmail.com> wrote:
> ...
> Now I continue executing the code, and initiate the condition to
> trigger the FIQ, the gdb properly breaks:
> Breakpoint 1, 0xffff001c in ?? ()

This line means that GDB does not know what function is being
executed. This is because set_fiq_handler() copies your FIQ routine at
the end of interrupt table - I personally would expect to take address
of your interrupt routine and write a mere jump to it, however I am
not familiar with ARM.

Making GDB work in this situation would be difficult, I guess you
could abuse GDB's overlay support or use JIT debug symbol adding,
clearly an overkill. Instead I recommend to use set_fiq_handler() to
copy a single absolute-jump instruction that would execute your FIQ
routine.

>
> If I try "stepi", the whole thing just locks up, and I can't do
> anything anymore on that session.. I can't seem to move past this
> after this, and not sure what to try next..
> Any ideas what I might be doing wrong ?

I think "stepi" should work even in this situation. Does GDB hang by
waiting on a remote target (openocd)?

Anyway, the code of set_fiq_handler() does not check the `length'
parameter. If I understand the meaning correctly then there is a
(small) limit for the parameter and an assertion would be appropriate.

(This seems to be related to
http://stackoverflow.com/questions/18823285/gdb-arm-debugger-unable-to-trigger-into-fiq-in-linux-driver.)

--
Petr Hluzin