Hi!
On Thu, Sep 27, 2018 at 04:17:57PM -0700, Carl Love wrote:
> + if (icode == CODE_FOR_rs6000_mffsl
> + && rs6000_isa_flags_explicit & OPTION_MASK_SOFT_FLOAT)
> +fatal_error (input_location,
> + "__builtin_mffsl() not supported with -msoft-float");
Please use plain "error
Segher:
I have addressed the various formatting and other minor issues.
I checked to see if the builtins worked with -msoft-float, which they
didn't. I added checks for the -msoft-float into the functions:
rs6000_expand_set_fpscr_rn_builtin(),
rs6000_expand_set_fpscr_drn_builtin(),
rs6000_ex
Hi Carl,
Sorry for the late review.
On Mon, Sep 17, 2018 at 03:03:28PM -0700, Carl Love wrote:
> * config/rs6000.c (rs6000_expand_mtfsb0_mtfsb1_builtin): Add.
config/rs6000/rs6000.c
> +RS6000_BUILTIN_X (RS6000_BUILTIN_MTFSB0, "__builtin_mtfsb0",
> + RS6000_BTM_ALWAYS,
> +
Segher:
This is an updated patch to add the following builtins:
__builtin_mffsl, __builtin_set_fpscr_rn, __builtin_set_fpscr_rn,
__builtin_mtfsb0, __builtin_mtfsb1.
I have addressed you comments with regards to the change log entries.
I have also addressed the various comments about the code, fu
Hi Carl,
On Fri, Aug 17, 2018 at 11:46:06AM -0700, Carl Love wrote:
> > In addition to listing
> > the builtin, I added a C style comment to describe the builtin a
> > little. I don't see any of the other builtins documented like this.
> > But I felt some explanation of the builtins were
> > hel
GCC maintainers:
> In addition to listing
> the builtin, I added a C style comment to describe the builtin a
> little. I don't see any of the other builtins documented like this.
> But I felt some explanation of the builtins were
> helpful. Suggestions
> on a better way to add the comments on t
GCC maintainers:
The following patch adds builtins to change the value of the FPSCR.
Specifically, __builtin_set_fpscr_rn, __builtin_set_fpscr_drn,
__builtin_mffsl, __builtin_mtfsb0, __builtin_mtfsb1. I added
documentation on the builtins in extend.texi. In addition to listing
the builtin, I ad