Re: [RFA 2/3] gdbarch-ification of ravenscar-thread support.

[Pedro Alves <palves@redhat.com>]

On 12/14/2012 03:02 PM, Joel Brobecker wrote:
> Hello,
> 
> This one is the biggie, moving ravenscar-thread to the gdbarch age.
> 
> The changes in the ravenscar modules themselves are quite
> straightforward, but the changes to Makefile.in (adding the
> right files to the right lists), and configure.tgt (adding
> ravenscar-sparc-thread.o everywhere sparc-tdep.o is listed)
> where slightly trickier. Also, I have a question, hence the
> RFA.
> 
> The question: Right now, I put ravenscar-thread.o in the list of
> objects to always be built, regardless of whether or not GDB is
> configured with a target that can take advantage of it.  I seems
> slightly wasteful, but somehow consistent with have solib-target
> always available, for instance.

It's a little bit different, because solib-target is driven entirely
from data the target pushes to GDB.  Any random embedded target can
enable that with no changes to GDB.  Ravenscar support always
needs code changes in GDB, OTOH.

> It's really easy to change: we
> just have to remove ravenscar-thread.o from Makefile.in, and
> add it to the gdb_target_obs instead. WDYT?

Don't see why not.  You already have to add ravenscar-ARCH-thread.c
there anyway.  But I'm not going to object.

BTW, ARCH-ravenscar-thread.c would be a naming scheme more in line
with other files.  Can we change that before adding more ports?

> 
> gdb/ChangeLog:
> 
>         * gdbarch.sh: Add "struct ravenscar_arch_ops" advance
>         declaration.

s/advance/forward/

Otherwise looks fine to me.

-- 
Pedro Alves