Changes to allow generation of mmx code from sse

diff --git a/orc/orcprogram-sse.c b/orc/orcprogram-sse.c
index eb97fbe..8252c69 100644 (file)
--- a/orc/orcprogram-sse.c
+++ b/orc/orcprogram-sse.c
@@ -13,6 +13,7 @@
 #include <orc/orcutils.h>
 #include <orc/orcdebug.h>
 
+#undef MMX
 #define SIZE 65536
 
 #define ORC_SSE_ALIGNED_DEST_CUTOFF 64
@@ -60,9 +61,15 @@ orc_sse_init (void)
 #endif
 
 #if defined(HAVE_I386)
+#ifndef MMX
   if (!(orc_sse_get_cpu_flags () & ORC_TARGET_SSE_SSE2)) {
     sse_target.executable = FALSE;
   }
+#else
+  if (!(orc_mmx_get_cpu_flags () & ORC_TARGET_MMX_MMX)) {
+    mmx_target.executable = FALSE;
+  }
+#endif
 #endif
 
   orc_target_register (&sse_target);
@@ -85,9 +92,14 @@ orc_compiler_sse_get_default_flags (void)
 #if defined(HAVE_AMD64) || defined(HAVE_I386)
   flags |= orc_sse_get_cpu_flags ();
 #else
+#ifndef MMX
   flags |= ORC_TARGET_SSE_SSE2;
   flags |= ORC_TARGET_SSE_SSE3;
   flags |= ORC_TARGET_SSE_SSSE3;
+#else
+  flags |= ORC_TARGET_MMX_MMX;
+  flags |= ORC_TARGET_MMX_3DNOW;
+#endif
 #endif
 
   return flags;
@@ -97,8 +109,13 @@ static const char *
 sse_get_flag_name (int shift)
 {
   static const char *flags[] = {
+#ifndef MMX
     "sse2", "sse3", "ssse3", "sse41", "sse42", "sse4a", "sse5",
     "frame_pointer", "short_jumps", "64bit"
+#else
+    "mmx", "mmxext", "3dnow", "3dnowext", "ssse3", "sse41", "",
+    "frame_pointer", "short_jumps", "64bit"
+#endif
   };
 
   if (shift >= 0 && shift < sizeof(flags)/sizeof(flags[0])) {
@@ -130,9 +147,15 @@ orc_compiler_sse_init (OrcCompiler *compiler)
     }
     compiler->valid_regs[X86_EDI] = 0;
     compiler->valid_regs[X86_ESP] = 0;
+#ifndef MMX
     for(i=X86_XMM0;i<X86_XMM0+16;i++){
       compiler->valid_regs[i] = 1;
     }
+#else
+    for(i=X86_XMM0;i<X86_XMM0+8;i++){
+      compiler->valid_regs[i] = 1;
+    }
+#endif
     compiler->save_regs[X86_EBX] = 1;
     compiler->save_regs[X86_EBP] = 1;
     compiler->save_regs[X86_R12] = 1;
@@ -176,24 +199,11 @@ orc_compiler_sse_init (OrcCompiler *compiler)
   }
   compiler->valid_regs[compiler->exec_reg] = 0;
 
-  switch (orc_program_get_max_var_size (compiler->program)) {
-    case 1:
-      compiler->loop_shift = 4;
-      break;
-    case 2:
-      compiler->loop_shift = 3;
-      break;
-    case 4:
-      compiler->loop_shift = 2;
-      break;
-    case 8:
-      compiler->loop_shift = 1;
-      break;
-    default:
-      ORC_ERROR("unhandled max var size %d",
-          orc_program_get_max_var_size (compiler->program));
-      break;
-  }
+#ifndef MMX
+  compiler->loop_shift = 5 - orc_program_get_max_var_size (compiler->program);
+#else
+  compiler->loop_shift = 4 - orc_program_get_max_var_size (compiler->program);
+#endif
 
   compiler->unroll_shift = 1;
   compiler->alloc_loop_counter = TRUE;
@@ -214,7 +224,11 @@ sse_save_accumulators (OrcCompiler *compiler)
       case ORC_VAR_TYPE_ACCUMULATOR:
         src = compiler->vars[i].alloc;
 
+#ifndef MMX
         orc_sse_emit_pshufd (compiler, ORC_SSE_SHUF(3,2,3,2), src, compiler->tmpreg);
+#else
+        orc_mmx_emit_pshufw (compiler, ORC_MMX_SHUF(3,2,3,2), src, compiler->tmpreg);
+#endif
 
         if (compiler->vars[i].size == 2) {
           orc_sse_emit_660f (compiler, "paddw", 0xfd, compiler->tmpreg, src);
@@ -222,6 +236,7 @@ sse_save_accumulators (OrcCompiler *compiler)
           orc_sse_emit_660f (compiler, "paddd", 0xfe, compiler->tmpreg, src);
         }
 
+#ifndef MMX
         orc_sse_emit_pshufd (compiler, ORC_SSE_SHUF(1,1,1,1), src, compiler->tmpreg);
 
         if (compiler->vars[i].size == 2) {
@@ -229,9 +244,14 @@ sse_save_accumulators (OrcCompiler *compiler)
         } else {
           orc_sse_emit_660f (compiler, "paddd", 0xfe, compiler->tmpreg, src);
         }
+#endif
 
         if (compiler->vars[i].size == 2) {
+#ifndef MMX
           orc_sse_emit_pshuflw (compiler, ORC_SSE_SHUF(1,1,1,1), src, compiler->tmpreg);
+#else
+          orc_mmx_emit_pshufw (compiler, ORC_MMX_SHUF(1,1,1,1), src, compiler->tmpreg);
+#endif
 
           orc_sse_emit_660f (compiler, "paddw", 0xfd, compiler->tmpreg, src);
         }
@@ -274,7 +294,11 @@ sse_load_constant (OrcCompiler *compiler, int reg, int size, int value)
   } else {
     orc_x86_emit_mov_imm_reg (compiler, 4, value, compiler->gp_tmpreg);
     orc_x86_emit_mov_reg_sse (compiler, compiler->gp_tmpreg, reg);
+#ifndef MMX
     orc_sse_emit_pshufd (compiler, 0, reg, reg);
+#else
+    orc_mmx_emit_pshufw (compiler, 0, reg, reg);
+#endif
   }
 
 }
@@ -468,6 +492,7 @@ orc_emit_split_n_regions (OrcCompiler *compiler)
   orc_x86_emit_label (compiler, 7);
 }
 
+#ifndef MMX
 static int
 orc_program_has_float (OrcCompiler *compiler)
 {
@@ -479,6 +504,7 @@ orc_program_has_float (OrcCompiler *compiler)
   }
   return FALSE;
 }
+#endif
 
 #define LABEL_REGION1_SKIP 1
 #define LABEL_INNER_LOOP_START 2
@@ -492,7 +518,9 @@ orc_program_has_float (OrcCompiler *compiler)
 void
 orc_compiler_sse_assemble (OrcCompiler *compiler)
 {
+#ifndef MMX
   int set_mxcsr = FALSE;
+#endif
   int align_var;
 
   if (0 && orc_x86_assemble_copy_check (compiler)) {
@@ -507,10 +535,12 @@ orc_compiler_sse_assemble (OrcCompiler *compiler)
 
   orc_x86_emit_prologue (compiler);
 
+#ifndef MMX
   if (orc_program_has_float (compiler)) {
     set_mxcsr = TRUE;
     orc_sse_set_mxcsr (compiler);
   }
+#endif
 
   sse_load_constants_outer (compiler);
 
@@ -668,9 +698,11 @@ orc_compiler_sse_assemble (OrcCompiler *compiler)
 
   sse_save_accumulators (compiler);
 
+#ifndef MMX
   if (set_mxcsr) {
     orc_sse_restore_mxcsr (compiler);
   }
+#endif
   orc_x86_emit_epilogue (compiler);
 
   orc_x86_do_fixups (compiler);

diff --git a/orc/orcrules-sse.c b/orc/orcrules-sse.c
index 14c9290..ad049cb 100644 (file)
--- a/orc/orcrules-sse.c
+++ b/orc/orcrules-sse.c
@@ -11,6 +11,7 @@
 #include <orc/orcdebug.h>
 #include <orc/orcsse.h>
 
+#undef MMX
 #define SIZE 65536
 
 /* sse rules */
@@ -32,9 +33,15 @@ sse_rule_loadpX (OrcCompiler *compiler, void *user, OrcInstruction *insn)
       orc_sse_emit_punpcklbw (compiler, reg, reg);
     }
     if (src->size <= 2) {
+#ifdef MMX
+      orc_mmx_emit_pshufw (compiler, 0, reg, reg);
+#else
       orc_sse_emit_pshuflw (compiler, 0, reg, reg);
+#endif
     }
+#ifndef MMX
     orc_sse_emit_pshufd (compiler, 0, reg, reg);
+#endif
   } else if (src->vartype == ORC_VAR_TYPE_CONST) {
     int value = src->value;
 
@@ -55,7 +62,9 @@ sse_rule_loadpX (OrcCompiler *compiler, void *user, OrcInstruction *insn)
     } else {
       orc_x86_emit_mov_imm_reg (compiler, 4, value, compiler->gp_tmpreg);
       orc_x86_emit_mov_reg_sse (compiler, compiler->gp_tmpreg, reg);
+#ifndef MMX
       orc_sse_emit_pshufd (compiler, 0, reg, reg);
+#endif
     }
   }
 }
@@ -275,9 +284,11 @@ sse_rule_accl (OrcCompiler *p, void *user, OrcInstruction *insn)
   int src = p->vars[insn->src_args[0]].alloc;
   int dest = p->vars[insn->dest_args[0]].alloc;
 
+#ifndef MMX
   if (p->loop_shift == 0) {
     orc_sse_emit_pslldq (p, 12, src);
   }
+#endif
   orc_sse_emit_paddd (p, src, dest);
 }
 
@@ -288,6 +299,7 @@ sse_rule_accsadubl (OrcCompiler *p, void *user, OrcInstruction *insn)
   int src2 = p->vars[insn->src_args[1]].alloc;
   int dest = p->vars[insn->dest_args[0]].alloc;
   int tmp = p->tmpreg;
+#ifndef MMX
   int tmp2 = X86_XMM7;
 
   if (p->loop_shift == 0) {
@@ -314,6 +326,10 @@ sse_rule_accsadubl (OrcCompiler *p, void *user, OrcInstruction *insn)
     orc_sse_emit_movdqa (p, src1, tmp);
     orc_sse_emit_psadbw (p, src2, tmp);
   }
+#else
+  orc_sse_emit_movdqa (p, src1, tmp);
+  orc_sse_emit_psadbw (p, src2, tmp);
+#endif
   orc_sse_emit_paddd (p, tmp, dest);
 }
 
@@ -344,20 +360,15 @@ sse_rule_signw_slow (OrcCompiler *p, void *user, OrcInstruction *insn)
   int src = p->vars[insn->src_args[0]].alloc;
   int dest = p->vars[insn->dest_args[0]].alloc;
   int tmp = p->tmpreg;
-  int gptmp = p->gp_tmpreg;
 
   if (src != dest) {
     orc_sse_emit_movdqa (p, src, dest);
   }
 
-  orc_x86_emit_mov_imm_reg (p, 4, 0x00010001, gptmp);
-  orc_x86_emit_mov_reg_sse (p, gptmp, tmp);
-  orc_sse_emit_pshufd (p, 0, tmp, tmp);
+  tmp = orc_compiler_get_constant (p, 2, 0x0001);
   orc_sse_emit_pminsw (p, tmp, dest);
 
-  orc_x86_emit_mov_imm_reg (p, 4, 0xffffffff, gptmp);
-  orc_x86_emit_mov_reg_sse (p, gptmp, tmp);
-  orc_sse_emit_pshufd (p, 0, tmp, tmp);
+  tmp = orc_compiler_get_constant (p, 2, 0xffff);
   orc_sse_emit_pmaxsw (p, tmp, dest);
 }
 
@@ -935,11 +946,8 @@ sse_rule_maxuw_slow (OrcCompiler *p, void *user, OrcInstruction *insn)
   int src = p->vars[insn->src_args[1]].alloc;
   int dest = p->vars[insn->dest_args[0]].alloc;
   int tmp = p->tmpreg;
-  int gptmp = p->gp_tmpreg;
 
-  orc_x86_emit_mov_imm_reg (p, 4, 0x80008000, gptmp);
-  orc_x86_emit_mov_reg_sse (p, gptmp, tmp);
-  orc_sse_emit_pshufd (p, 0, tmp, tmp);
+  tmp = orc_compiler_get_constant (p, 2, 0x8000);
   orc_sse_emit_pxor(p, tmp, src);
   orc_sse_emit_pxor(p, tmp, dest);
   orc_sse_emit_pmaxsw (p, src, dest);
@@ -1129,6 +1137,7 @@ sse_rule_avgul (OrcCompiler *p, void *user, OrcInstruction *insn)
   orc_sse_emit_psubd(p, tmp, dest);
 }
 
+#ifndef MMX
 /* float ops */
 
 #define UNARY_F(opcode,insn_name,code) \
@@ -1243,6 +1252,7 @@ sse_rule_convlf (OrcCompiler *p, void *user, OrcInstruction *insn)
       p->vars[insn->src_args[0]].alloc,
       p->vars[insn->dest_args[0]].alloc);
 }
+#endif
 
 void
 orc_compiler_sse_register_rules (OrcTarget *target)
@@ -1347,6 +1357,7 @@ orc_compiler_sse_register_rules (OrcTarget *target)
   orc_rule_register (rule_set, "accl", sse_rule_accl, NULL);
   orc_rule_register (rule_set, "accsadubl", sse_rule_accsadubl, NULL);
 
+#ifndef MMX
   orc_rule_register (rule_set, "addf", sse_rule_addf, NULL);
   orc_rule_register (rule_set, "subf", sse_rule_subf, NULL);
   orc_rule_register (rule_set, "mulf", sse_rule_mulf, NULL);
@@ -1359,6 +1370,7 @@ orc_compiler_sse_register_rules (OrcTarget *target)
   orc_rule_register (rule_set, "cmplef", sse_rule_cmplef, NULL);
   orc_rule_register (rule_set, "convfl", sse_rule_convfl, NULL);
   orc_rule_register (rule_set, "convlf", sse_rule_convlf, NULL);
+#endif
 
   /* slow rules */
   orc_rule_register (rule_set, "maxuw", sse_rule_maxuw_slow, NULL);