Re: [PATCH] Provide useful completer for "info registers"

[Andreas Arnez <arnez at linux dot vnet dot ibm dot com>]

On Thu, Dec 04 2014, Pedro Alves wrote:

> Thanks Andreas,
>
> I think a register completer is a great idea.

Thanks, I thought so, too ;-)

>
> On 11/26/2014 08:54 PM, Sergio Durigan Junior wrote:
>> I'd say this patch also needs a testcase :-).  I know that this is
>> architecture specific, so I'd personally be happy with something very
>> simple, maybe testing only one or two architectures would be enough.
>
> I think $pc, $sp, $fp (the user regs) should work everywhere.
>
> See user-regs.c and std-regs.c.
>
> Actually, looks like the patch misses considering those for completion?
>
> See infcmd.c:registers_info:
>
>       /* A register name?  */
>       {
>         int regnum = user_reg_map_name_to_regnum (gdbarch, start, end - start);
>
>         if (regnum >= 0)
>           {
>             /* User registers lie completely outside of the range of
>                normal registers.  Catch them early so that the target
>                never sees them.  */
>             if (regnum >= gdbarch_num_regs (gdbarch)
>                           + gdbarch_num_pseudo_regs (gdbarch))
>               {

Yes, the patch misses them...  Note that this is consistent with the
list of registers shown by "mt print registers" or "info registers all".
But you're right, the "info registers" command accepts user registers,
so the completer should offer them as well.  Added this in v3.

It seems that user registers can not be listed by any means in GDB.
Most architectures only have the built-in set of user registers ($pc,
$sp, $fp, and $ps), but some define additional ones.  This creates a
difficulty with the test case, which tries to determine the full list of
registers and reggroups the completer is supposed to yield.  Thus I just
added such a command in v3 as well.