| Message ID | 1407345815-14551-46-git-send-email-dmalcolm@redhat.com |
|---|---|
| State | New |
| Headers | show |
On 08/06/14 11:20, David Malcolm wrote: > (define_bypass) clauses in .md files can specify the name of a guard > function as their final operand. Currently these functions are called > with a pair of rtx. This patch strengthens insn-automata.c so that such > guard functions are passed a pair of rtx_insn *, allowing these guard > functions to be similarly strengthened in the per-target phase of this > patch kit. > > gcc/ > * genautomata.c (output_internal_insn_latency_func): When writing > the function "internal_insn_latency" to insn-automata.c, > strengthen params "insn" and "insn2" from rtx to rtx_insn *, thus > allowing the optional guard function of (define_bypass) clauses to > expect a pair of rtx_insn *, rather than a pair of rtx. > (output_insn_latency_func): When writing the function > "insn_latency", add an "uncast_" prefix to params "insn" and > "insn2", reintroducing "insn" and "insn2" as rtx_insn * locals > using checked casts from the params, thus enabling the above > change to the generated "internal_insn_latency" function. OK. Jeff
On Wed, 2014-08-13 at 12:07 -0600, Jeff Law wrote: > On 08/06/14 11:20, David Malcolm wrote: > > (define_bypass) clauses in .md files can specify the name of a guard > > function as their final operand. Currently these functions are called > > with a pair of rtx. This patch strengthens insn-automata.c so that such > > guard functions are passed a pair of rtx_insn *, allowing these guard > > functions to be similarly strengthened in the per-target phase of this > > patch kit. > > > > gcc/ > > * genautomata.c (output_internal_insn_latency_func): When writing > > the function "internal_insn_latency" to insn-automata.c, > > strengthen params "insn" and "insn2" from rtx to rtx_insn *, thus > > allowing the optional guard function of (define_bypass) clauses to > > expect a pair of rtx_insn *, rather than a pair of rtx. > > (output_insn_latency_func): When writing the function > > "insn_latency", add an "uncast_" prefix to params "insn" and > > "insn2", reintroducing "insn" and "insn2" as rtx_insn * locals > > using checked casts from the params, thus enabling the above > > change to the generated "internal_insn_latency" function. > OK. Thanks; committed to trunk as r214273, with the trivial fixup of as_a_nullable to safe_as_a.
diff --git a/gcc/genautomata.c b/gcc/genautomata.c index 0c61278..3017e20 100644 --- a/gcc/genautomata.c +++ b/gcc/genautomata.c @@ -8360,7 +8360,7 @@ output_internal_insn_latency_func (void) decl_t decl; struct bypass_decl *bypass; - fprintf (output_file, "static int\n%s (int %s ATTRIBUTE_UNUSED,\n\tint %s ATTRIBUTE_UNUSED,\n\trtx %s ATTRIBUTE_UNUSED,\n\trtx %s ATTRIBUTE_UNUSED)\n", + fprintf (output_file, "static int\n%s (int %s ATTRIBUTE_UNUSED,\n\tint %s ATTRIBUTE_UNUSED,\n\trtx_insn *%s ATTRIBUTE_UNUSED,\n\trtx_insn *%s ATTRIBUTE_UNUSED)\n", INTERNAL_INSN_LATENCY_FUNC_NAME, INTERNAL_INSN_CODE_NAME, INTERNAL_INSN2_CODE_NAME, INSN_PARAMETER_NAME, INSN2_PARAMETER_NAME); @@ -8477,10 +8477,16 @@ output_internal_maximal_insn_latency_func (void) static void output_insn_latency_func (void) { - fprintf (output_file, "int\n%s (rtx %s, rtx %s)\n", + fprintf (output_file, "int\n%s (rtx uncast_%s, rtx uncast_%s)\n", INSN_LATENCY_FUNC_NAME, INSN_PARAMETER_NAME, INSN2_PARAMETER_NAME); fprintf (output_file, "{\n int %s, %s;\n", INTERNAL_INSN_CODE_NAME, INTERNAL_INSN2_CODE_NAME); + fprintf (output_file, + " rtx_insn *%s = as_a_nullable <rtx_insn *> (uncast_%s);\n", + INSN_PARAMETER_NAME, INSN_PARAMETER_NAME); + fprintf (output_file, + " rtx_insn *%s = as_a_nullable <rtx_insn *> (uncast_%s);\n", + INSN2_PARAMETER_NAME, INSN2_PARAMETER_NAME); output_internal_insn_code_evaluation (INSN_PARAMETER_NAME, INTERNAL_INSN_CODE_NAME, 0); output_internal_insn_code_evaluation (INSN2_PARAMETER_NAME,