Re: [PATCH, v2] Share ptrace options discovery/linux native code between GDB and gdbserver

[Luis Machado <lgustavo@codesourcery.com>]

On 08/20/2013 12:38 PM, Tom Tromey wrote:
>>>>>> "Luis" == Luis Machado <lgustavo@codesourcery.com> writes:
>
> Luis> This is the second iteration of the patch to do some sharing of ptrace
> Luis> and native linux stuff between GDB and gdbserver.
>
> Luis> I've included fixes based on reviews, so hopefully this is in better
> Luis> shape now.
>
> Thanks.  I like it.  I have a few nits though.
>
> Luis>           * common/linux-ptrace.c: Include nat/linux-waitpid.h.
>
> I suppose at some point this file ought to be moved.

Right. Should i move it now as part of this cleanup or should i move it 
in a follow up?

>
> Luis>           * config.in (PTRACE_TYPE_ARG4): Undefine.
>
> Usually for config.in I just write:
>
> 	* config.in: Rebuild.
>
> ... since presumably you just used autoheader.
>

Oops. I should probably rebuild it then.

> Luis>           * configure.ac (AC_CACHE_CHECK): Add void * to the list of
> Luis>           ptrace's 4th argument's types.
> Luis>           Check the type of PTRACE_TYPE_ARG4.
>
> AC_CACHE_CHECK isn't the right function name.  Perhaps no name is best
> here.

Got it. Let's go with no name.

>
> Also, doesn't this check need to be done in gdbserver as well?

For gdbserver we explicitly define the correct types for each argument.

Since this is a runtime check, we may run into issues if we attempt to 
run a binary on a build machine if the target machine is from a 
different architecture. Does it make sense?

Luis