Steve Ellcey <sell...@cavium.com> writes:
> @@ -1029,13 +1029,18 @@ aarch64_gen_far_branch (rtx * operands, int
> pos_label, const char * dest,
> }
>
> void
> -aarch64_err_no_fpadvsimd (machine_mode mode, const char *msg)
> +aarch64_err_no_fpadvsimd (machine_mode mode)
> {
> - const char *mc = FLOAT_MODE_P (mode) ? "floating-point" : "vector";
> if (TARGET_GENERAL_REGS_ONLY)
> - error ("%qs is incompatible with %s %s", "-mgeneral-regs-only", mc, msg);
> + if (FLOAT_MODE_P (mode))
> + error ("-mgeneral-regs-only is incompatible with the use of floating
> point types");
> + else
> + error ("-mgeneral-regs-only is incompatible with the use of vector
> types");
> else
> - error ("%qs feature modifier is incompatible with %s %s", "+nofp", mc,
> msg);
> + if (FLOAT_MODE_P (mode))
> + error ("+nofp feature modifier is incompatible with the use of
> floating point types");
> + else
> + error ("+nofp feature modifier is incompatible with the use of vector
> types");
This regresses a couple of things:
- before the patch, the option would be properly quoted, whereas now
it's unquoted. Either keeping %qs or using %<...%> would fix that.
Using %qs is probably nicer since we can reuse the translation for any
other options or features that end up being incompatible.
- "floating-point" is preferred for modifiers over "floating point".
- some lines are now longer than 80 chars.
Patch LGTM otherwise, but someone else will need to approve.
Thanks,
Richard
