On Mon, 6 May 2013, DJ Delorie wrote:
> Note that I had to make a few changes (fixes?) in the MI portions of
> gcc to avoid problems I encountered, I don't know if these changes are
> "correct" or if there are better ways to avoid those cases. Those
In any case, they should best be posted in separate messages, each one
with its own rationale.
> +@item -mlarge
> +@opindex mlarge
> +Use large-model addressing (20-bit pointers, 32-bit size_t).
> +
> +@item -msmall
> +@opindex msmall
> +Use small-model addressing (16-bit pointers, 16-bit size_t).
@code{size_t}.
> Index: gcc/config/msp430/predicates.md
> ===================================================================
> --- gcc/config/msp430/predicates.md (revision 0)
> +++ gcc/config/msp430/predicates.md (revision 0)
> @@ -0,0 +1,60 @@
This file should have copyright / license notices.
> +
> +void
> +msp430_expand_helper (rtx *operands, const char *helper_name, bool
> const_variants)
In general each function should have a comment explaining parameters or
return values, or saying it implements a particular standard hook (I think
the approach elsewhere in this file of defining the hook macro immediately
above the function, instead of having a comment, is fine, but various
functions still need comments).
> + if (const_variants
> + && CONST_INT_P (operands[2])
> + && INTVAL (operands[2]) >= 1
> + && INTVAL (operands[2]) <= 15)
> + {
> + helper_const = (char *) xmalloc (strlen (helper_name) + 4);
> + sprintf (helper_const, "%s_%ld", helper_name, INTVAL (operands[2]));
I think snprintf should be used in preference to snprintf. And INTVAL
returns HOST_WIDE_INT, not "long", so use HOST_WIDE_INT_PRINT_DEC instead
of %ld (there are a couple of places in msp430_print_operand that are also
using %ld with INTVAL).
> + if (!rv)
> + {
> + fprintf (stderr, "Can't subreg!!!\n");
> + debug_rtx (r);
> + gcc_unreachable ();
fprintf to stderr isn't a good way of reporting internal errors.
fatal_insn seems more like what's wanted, although this isn't actually an
insn you have here.
> +#define TARGET_CPU_CPP_BUILTINS() \
> + do \
> + { \
> + builtin_define ("NO_TRAMPOLINES"); \
Don't define NO_TRAMPOLINES; that's in the user's namespace. (For
testcases testing NO_TRAMPOLINES, lib/gcc.exp defines it if the board file
defines gcc,no_trampolines.)
--
Joseph S. Myers
[email protected]
