This is the mail archive of the
gdb-patches@sources.redhat.com
mailing list for the GDB project.
Re: [PATCH] Print vector registers in natural format, not hex
Daniel Jacobowitz writes:
> On Fri, Oct 04, 2002 at 05:11:27PM -0400, Klee Dienes wrote:
> > On Friday, October 4, 2002, at 04:53 PM, Daniel Jacobowitz wrote:
> >
> > >Just got one nit.
> > >
> > >First of all, in characters
> > > v16_int8 = {'`', 'a', 'b', 'c', 'd', 'e', 'f', 'g', '@', 'I',
> > >'\017', '°', 'Z', '@', '-', '°', 'Z', 'T'}
> > >
> > >is more natural to me. It's a vector of chars, not a string.
> > >
> > >Secondly, this is an 'int8' type. Are characters really the way to go?
> >
> > Although I agree with you, that's not really a result of this patch,
> > but rather long-standing GDB behavior. The patch is only causing $v1
> > to be printed according to its type declaration, which in this case is:
> >
> > type = union __gdb_builtin_type_vec128 {
> > int128_t uint128;
> > float v4_float[4];
> > int32_t v4_int32[4];
> > int16_t v8_int16[8];
> > int8_t v16_int8[16];
> > }
> >
> > Then the following code in c-valprint.c causes it to be printed as a
> > string:
> >
> > /* For an array of chars, print with string syntax. */
> > if (eltlen == 1 &&
> > ((TYPE_CODE (elttype) == TYPE_CODE_INT)
> > || ((current_language->la_language == language_m2)
> > && (TYPE_CODE (elttype) == TYPE_CODE_CHAR)))
> > && (format == 0 || format == 's'))
> >
> > You could probably make a good case that arrays of 'int' should be
> > printed as you describe, but that's a debate I'll leave to someone else.
>
> Ah, you're right. Thanks for the detail.
>
And you haven't noticed this comment in the testfile, I take it :-) I
discovered this while writing the vector support, but didn't fix it. I
don't know that it actually needs to be fixed. It's a feature, not a
bug, one could argue.
# Note: in LE case, the char array is printed WITHOUT the last character.
# Gdb treats the terminating null char in the array like the terminating
# null char in a string and doesn't print it. This is not a failure, but
# the way gdb works.
Elena
> Otherwise, I like Klee's patch... anyone else?
>
> --
> Daniel Jacobowitz
> MontaVista Software Debian GNU/Linux Developer