re PR target/18668 (use prescott's fisttp)

	PR target/18668
	* config/i386/i386.h (x86_fisttp): New.
	(TARGET_FISTTP): New macro.
	* config/i386/i386.c (x86_fisttp): Set for NOCONA.
	(output_fix_trunc): Add fisttp parameter.  Generate fisttp x87
	instruction when fisttp flag is set.
	* config/i386/i386-protos.h (output_fix_trunc): Change declaration.

	* config/i386/i386.md (type attribute): Add fisttp.
	(unit attribute): Set to i387 for fisttp type.
	(X87MODEF, X87MODEI, SSEMODEF, SSEMODEI24): New mode macros.
	(fix_truncxfdi2, fix_truncxfsi2): Generate fisttp patterns for
	TARGET_FISTTP.
	(fix_truncdfdi2, fix_truncsfdi2, fix_truncdfsi2, fix_truncsfsi2):
	Generate fisttp patterns for TARGET_FISTTP. Implement using mode
	macros.
	(fix_truncxfhi2, fix_truncdfhi2, fix_truncsfhi2): Generate fisttp
	patterns for TARGET_FISTTP.  Enable patterns for
	(TARGET_FISTTP && !TARGET_SSE_MATH). Implement using mode macros.
	(fix_trunc<mode>_i387_fisttp_1, fix_trunc<mode>_i387_fisttp,
	fix_trunc<mode>_i387_fisttp_with_temp): New instruction patterns to
	implement fisttp x87 insn.
	(fix_trunc*_i387_fisttp splitters): New patterns.
	(*fix_truncdi_i387, *fix_truncsi_i387, *fix_trunchi_i387):
	Rename to *fix_trunc<mode>_i387_1.  Implement using mode macros.
	Disable patterns for TARGET_FISTTP.  Add comment about FLAGS_REG
	clobber.
	(fix_truncdi_memory, fix_truncdi_nomemory, fix_trunchi_nomemory):
	Rename to fix_trunc<mode>_i387 and fix_trunc<mode>_i387_with_temp.
	Implement using mode macros. Disable patterns for TARGET_FISTTP.
	(fix_truncsi_memory, fix_truncsi_nomemory, fix_trunchi_memory,
	fix_trunchi_nomemory): Rename to fix_trunc<mode>_i387 and
	fix_trunc<mode>_i387_with_temp. Implement using mode macros.
	Disable patterns for TARGET_FISTTP.
	(fix_trunc*_i387 splitters): Implement usign mode macros.
	(fix_truncdfdi_sse, fix_truncsfdi_sse, fix_truncdfsi_sse,
	fix_truncsfsi_sse): Disable for (TARGET_FISTTP && !TARGET_SSE_MATH).
	(fix_trunx*_sse peephole2s): Implement using mode macros.

From-SVN: r96477

diff --git a/gcc/ChangeLog b/gcc/ChangeLog
index 48e144d72f79b9e83182a021ea75492fb6e398bf..b4489a480135cfca70e6a91cc90a7349617f4ac2 100644 (file)
--- a/gcc/ChangeLog
+++ b/gcc/ChangeLog
@@ -1,3 +1,44 @@
+2005-03-15  Uros Bizjak  <uros@kss-loka.si>
+
+       PR target/18668
+       * config/i386/i386.h (x86_fisttp): New.
+       (TARGET_FISTTP): New macro.
+       * config/i386/i386.c (x86_fisttp): Set for NOCONA.
+       (output_fix_trunc): Add fisttp parameter.  Generate fisttp x87
+       instruction when fisttp flag is set.
+       * config/i386/i386-protos.h (output_fix_trunc): Change declaration.
+
+       * config/i386/i386.md (type attribute): Add fisttp.
+       (unit attribute): Set to i387 for fisttp type.
+       (X87MODEF, X87MODEI, SSEMODEF, SSEMODEI24): New mode macros.
+       (fix_truncxfdi2, fix_truncxfsi2): Generate fisttp patterns for
+       TARGET_FISTTP.
+       (fix_truncdfdi2, fix_truncsfdi2, fix_truncdfsi2, fix_truncsfsi2):
+       Generate fisttp patterns for TARGET_FISTTP. Implement using mode
+       macros.
+       (fix_truncxfhi2, fix_truncdfhi2, fix_truncsfhi2): Generate fisttp
+       patterns for TARGET_FISTTP.  Enable patterns for
+       (TARGET_FISTTP && !TARGET_SSE_MATH). Implement using mode macros.
+       (fix_trunc<mode>_i387_fisttp_1, fix_trunc<mode>_i387_fisttp,
+       fix_trunc<mode>_i387_fisttp_with_temp): New instruction patterns to
+       implement fisttp x87 insn.
+       (fix_trunc*_i387_fisttp splitters): New patterns.
+       (*fix_truncdi_i387, *fix_truncsi_i387, *fix_trunchi_i387):
+       Rename to *fix_trunc<mode>_i387_1.  Implement using mode macros.
+       Disable patterns for TARGET_FISTTP.  Add comment about FLAGS_REG
+       clobber.
+       (fix_truncdi_memory, fix_truncdi_nomemory, fix_trunchi_nomemory):
+       Rename to fix_trunc<mode>_i387 and fix_trunc<mode>_i387_with_temp.
+       Implement using mode macros. Disable patterns for TARGET_FISTTP.
+       (fix_truncsi_memory, fix_truncsi_nomemory, fix_trunchi_memory,
+       fix_trunchi_nomemory): Rename to fix_trunc<mode>_i387 and
+       fix_trunc<mode>_i387_with_temp. Implement using mode macros.
+       Disable patterns for TARGET_FISTTP.
+       (fix_trunc*_i387 splitters): Implement usign mode macros.
+       (fix_truncdfdi_sse, fix_truncsfdi_sse, fix_truncdfsi_sse,
+       fix_truncsfsi_sse): Disable for (TARGET_FISTTP && !TARGET_SSE_MATH).
+       (fix_trunx*_sse peephole2s): Implement using mode macros.
+
 2005-03-15  J"orn Rennecke <joern.rennecke@st.com>
 
        PR rtl-optimization/20291

diff --git a/gcc/config/i386/i386-protos.h b/gcc/config/i386/i386-protos.h
index 53edfd161045a62f63c610550369354ef1605a1c..bf693860da71b8f732ab06e7d629d1266611b23f 100644 (file)
--- a/gcc/config/i386/i386-protos.h
+++ b/gcc/config/i386/i386-protos.h
@@ -117,7 +117,7 @@ extern void split_ti (rtx[], int, rtx[], rtx[]);
 extern const char *output_set_got (rtx);
 extern const char *output_387_binary_op (rtx, rtx*);
 extern const char *output_387_reg_move (rtx, rtx*);
-extern const char *output_fix_trunc (rtx, rtx*);
+extern const char *output_fix_trunc (rtx, rtx*, int);
 extern const char *output_fp_compare (rtx, rtx*, int, int);
 
 extern void i386_output_dwarf_dtprel (FILE*, int, rtx);

diff --git a/gcc/config/i386/i386.c b/gcc/config/i386/i386.c
index bc311aaf23ab9a7ce71b53dace9a21d111a981f1..f76812f335fa29d9e79aea72db12c7b31d5d9a42 100644 (file)
--- a/gcc/config/i386/i386.c
+++ b/gcc/config/i386/i386.c
@@ -525,6 +525,7 @@ const int x86_double_with_add = ~m_386;
 const int x86_use_bit_test = m_386;
 const int x86_unroll_strlen = m_486 | m_PENT | m_PPRO | m_ATHLON_K8 | m_K6;
 const int x86_cmove = m_PPRO | m_ATHLON_K8 | m_PENT4 | m_NOCONA;
+const int x86_fisttp = m_NOCONA;
 const int x86_3dnow_a = m_ATHLON_K8;
 const int x86_deep_branch = m_PPRO | m_K6 | m_ATHLON_K8 | m_PENT4 | m_NOCONA;
 /* Branch hints were put in P4 based on simulation result. But
@@ -7282,7 +7283,7 @@ emit_i387_cw_initialization (rtx current_mode, rtx new_mode, int mode)
    operand may be [SDX]Fmode.  */
 
 const char *
-output_fix_trunc (rtx insn, rtx *operands)
+output_fix_trunc (rtx insn, rtx *operands, int fisttp)
 {
   int stack_top_dies = find_regno_note (insn, REG_DEAD, FIRST_STACK_REG) != 0;
   int dimode_p = GET_MODE (operands[0]) == DImode;
@@ -7290,7 +7291,7 @@ output_fix_trunc (rtx insn, rtx *operands)
   /* Jump through a hoop or two for DImode, since the hardware has no
      non-popping instruction.  We used to do this a different way, but
      that was somewhat fragile and broke with post-reload splitters.  */
-  if (dimode_p && !stack_top_dies)
+  if ((dimode_p || fisttp) && !stack_top_dies)
     output_asm_insn ("fld\t%y1", operands);
 
   if (!STACK_TOP_P (operands[1]))
@@ -7299,12 +7300,17 @@ output_fix_trunc (rtx insn, rtx *operands)
   if (GET_CODE (operands[0]) != MEM)
     abort ();
 
-  output_asm_insn ("fldcw\t%3", operands);
-  if (stack_top_dies || dimode_p)
-    output_asm_insn ("fistp%z0\t%0", operands);
+  if (fisttp)
+      output_asm_insn ("fisttp%z0\t%0", operands);
   else
-    output_asm_insn ("fist%z0\t%0", operands);
-  output_asm_insn ("fldcw\t%2", operands);
+    {
+      output_asm_insn ("fldcw\t%3", operands);
+      if (stack_top_dies || dimode_p)
+       output_asm_insn ("fistp%z0\t%0", operands);
+      else
+       output_asm_insn ("fist%z0\t%0", operands);
+      output_asm_insn ("fldcw\t%2", operands);
+    }
 
   return "";
 }

diff --git a/gcc/config/i386/i386.h b/gcc/config/i386/i386.h
index d5b8c82b9c9ac5383a210a568a95d3b852084f87..765eff8b9fdf7f5184ff8638740ba52a1cf8537f 100644 (file)
--- a/gcc/config/i386/i386.h
+++ b/gcc/config/i386/i386.h
@@ -229,7 +229,7 @@ extern int target_flags;
 
 #define TUNEMASK (1 << ix86_tune)
 extern const int x86_use_leave, x86_push_memory, x86_zero_extend_with_and;
-extern const int x86_use_bit_test, x86_cmove, x86_deep_branch;
+extern const int x86_use_bit_test, x86_cmove, x86_fisttp, x86_deep_branch;
 extern const int x86_branch_hints, x86_unroll_strlen;
 extern const int x86_double_with_add, x86_partial_reg_stall, x86_movx;
 extern const int x86_use_loop, x86_use_himode_fiop, x86_use_simode_fiop;
@@ -258,6 +258,7 @@ extern int x86_prefetch_sse;
 /* For sane SSE instruction set generation we need fcomi instruction.  It is
    safe to enable all CMOVE instructions.  */
 #define TARGET_CMOVE ((x86_cmove & (1 << ix86_arch)) || TARGET_SSE)
+#define TARGET_FISTTP (x86_fisttp & (1 << ix86_arch))
 #define TARGET_DEEP_BRANCH_PREDICTION (x86_deep_branch & TUNEMASK)
 #define TARGET_BRANCH_PREDICTION_HINTS (x86_branch_hints & TUNEMASK)
 #define TARGET_DOUBLE_WITH_ADD (x86_double_with_add & TUNEMASK)

diff --git a/gcc/config/i386/i386.md b/gcc/config/i386/i386.md
index 61904db4b586db0c1326cc880647d67119ccef5b..e168d112933e713cdf23b47eb5499c8499aaaf96 100644 (file)
--- a/gcc/config/i386/i386.md
+++ b/gcc/config/i386/i386.md
@@ -194,7 +194,7 @@
    icmp,test,ibr,setcc,icmov,
    push,pop,call,callv,leave,
    str,cld,
-   fmov,fop,fsgn,fmul,fdiv,fpspc,fcmov,fcmp,fxch,fistp,frndint,
+   fmov,fop,fsgn,fmul,fdiv,fpspc,fcmov,fcmp,fxch,fistp,fisttp,frndint,
    sselog,sselog1,sseiadd,sseishft,sseimul,
    sse,ssemov,sseadd,ssemul,ssecmp,ssecomi,ssecvt,sseicvt,ssediv,
    mmx,mmxmov,mmxadd,mmxmul,mmxcmp,mmxcvt,mmxshft"
@@ -207,7 +207,7 @@
 
 ;; The CPU unit operations uses.
 (define_attr "unit" "integer,i387,sse,mmx,unknown"
-  (cond [(eq_attr "type" "fmov,fop,fsgn,fmul,fdiv,fpspc,fcmov,fcmp,fxch,fistp,frndint")
+  (cond [(eq_attr "type" "fmov,fop,fsgn,fmul,fdiv,fpspc,fcmov,fcmp,fxch,fistp,fisttp,frndint")
           (const_string "i387")
         (eq_attr "type" "sselog,sselog1,sseiadd,sseishft,sseimul,
                          sse,ssemov,sseadd,ssemul,ssecmp,ssecomi,ssecvt,sseicvt,ssediv")
@@ -442,8 +442,21 @@
   [(set_attr "length" "128")
    (set_attr "type" "multi")])
 
+;; All x87 floating point modes
+(define_mode_macro X87MODEF [SF DF XF])
+ 
+;; All integer modes handled by x87 fisttp operator.
+(define_mode_macro X87MODEI [HI SI DI])
+
 ;; All integer modes handled by integer x87 operators.
 (define_mode_macro X87MODEI12 [HI SI])
+
+;; All SSE floating point modes
+(define_mode_macro SSEMODEF [SF DF])
+ 
+;; All integer modes handled by SSE cvtts?2si* operators.
+(define_mode_macro SSEMODEI24 [SI DI])
+
 \f
 ;; Scheduling descriptions
 
@@ -3998,8 +4011,6 @@
   [(set (match_dup 0) (float_truncate:DF (match_dup 1)))]
   "")
 \f
-;; %%% Break up all these bad boys.
-
 ;; Signed conversion to DImode.
 
 (define_expand "fix_truncxfdi2"
@@ -4007,361 +4018,234 @@
                    (fix:DI (match_operand:XF 1 "register_operand" "")))
              (clobber (reg:CC FLAGS_REG))])]
   "TARGET_80387"
-  "")
-
-(define_expand "fix_truncdfdi2"
-  [(parallel [(set (match_operand:DI 0 "nonimmediate_operand" "")
-                   (fix:DI (match_operand:DF 1 "register_operand" "")))
-              (clobber (reg:CC FLAGS_REG))])]
-  "TARGET_80387 || (TARGET_64BIT && TARGET_SSE2)"
 {
-  if (TARGET_64BIT && TARGET_SSE2)
+  if (TARGET_FISTTP)
    {
-     rtx out = REG_P (operands[0]) ? operands[0] : gen_reg_rtx (DImode);
-     emit_insn (gen_fix_truncdfdi_sse (out, operands[1]));
-     if (out != operands[0])
-       emit_move_insn (operands[0], out);
+     emit_insn (gen_fix_truncdi_fisttp_i387_1 (operands[0], operands[1]));
      DONE;
    }
 })
 
-(define_expand "fix_truncsfdi2"
+(define_expand "fix_trunc<mode>di2"
   [(parallel [(set (match_operand:DI 0 "nonimmediate_operand" "")
-                  (fix:DI (match_operand:SF 1 "register_operand" "")))
-              (clobber (reg:CC FLAGS_REG))])] 
-  "TARGET_80387 || (TARGET_64BIT && TARGET_SSE)"
+                   (fix:DI (match_operand:SSEMODEF 1 "register_operand" "")))
+              (clobber (reg:CC FLAGS_REG))])]
+  "TARGET_80387 || (TARGET_64BIT && SSE_FLOAT_MODE_P (<MODE>mode))"
 {
-  if (TARGET_64BIT && TARGET_SSE)
+  if (TARGET_FISTTP
+      && !(TARGET_64BIT && SSE_FLOAT_MODE_P (<MODE>mode) && TARGET_SSE_MATH))
+   {
+     emit_insn (gen_fix_truncdi_fisttp_i387_1 (operands[0], operands[1]));
+     DONE;
+   }
+  if (TARGET_64BIT && SSE_FLOAT_MODE_P (<MODE>mode))
    {
      rtx out = REG_P (operands[0]) ? operands[0] : gen_reg_rtx (DImode);
-     emit_insn (gen_fix_truncsfdi_sse (out, operands[1]));
+     emit_insn (gen_fix_trunc<mode>di_sse (out, operands[1]));
      if (out != operands[0])
        emit_move_insn (operands[0], out);
      DONE;
    }
 })
 
-;; See the comments in i386.h near OPTIMIZE_MODE_SWITCHING for the description
-;; of the machinery.
-(define_insn_and_split "*fix_truncdi_i387"
-  [(set (match_operand:DI 0 "nonimmediate_operand" "=m,?r")
-       (fix:DI (match_operand 1 "register_operand" "f,f")))
-   (clobber (reg:CC FLAGS_REG))]
-  "TARGET_80387 && FLOAT_MODE_P (GET_MODE (operands[1]))
-   && !reload_completed && !reload_in_progress
-   && (!SSE_FLOAT_MODE_P (GET_MODE (operands[1])) || !TARGET_64BIT)"
-  "#"
-  "&& 1"
-  [(const_int 0)]
-{
-  ix86_optimize_mode_switching = 1;
-  operands[2] = assign_386_stack_local (HImode, 1);
-  operands[3] = assign_386_stack_local (HImode, 2);
-  if (memory_operand (operands[0], VOIDmode))
-    emit_insn (gen_fix_truncdi_memory (operands[0], operands[1],
-                                      operands[2], operands[3]));
-  else
-    {
-      operands[4] = assign_386_stack_local (DImode, 0);
-      emit_insn (gen_fix_truncdi_nomemory (operands[0], operands[1],
-                                          operands[2], operands[3],
-                                          operands[4]));
-    }
-  DONE;
-}
-  [(set_attr "type" "fistp")
-   (set_attr "i387_cw" "trunc")
-   (set_attr "mode" "DI")])
-
-(define_insn "fix_truncdi_nomemory"
-  [(set (match_operand:DI 0 "nonimmediate_operand" "=m,?r")
-       (fix:DI (match_operand 1 "register_operand" "f,f")))
-   (use (match_operand:HI 2 "memory_operand" "m,m"))
-   (use (match_operand:HI 3 "memory_operand" "m,m"))
-   (clobber (match_operand:DI 4 "memory_operand" "=m,m"))
-   (clobber (match_scratch:DF 5 "=&1f,&1f"))]
-  "TARGET_80387 && FLOAT_MODE_P (GET_MODE (operands[1]))
-   && (!SSE_FLOAT_MODE_P (GET_MODE (operands[1])) || !TARGET_64BIT)"
-  "#"
-  [(set_attr "type" "fistp")
-   (set_attr "i387_cw" "trunc")
-   (set_attr "mode" "DI")])
-
-(define_insn "fix_truncdi_memory"
-  [(set (match_operand:DI 0 "memory_operand" "=m")
-       (fix:DI (match_operand 1 "register_operand" "f")))
-   (use (match_operand:HI 2 "memory_operand" "m"))
-   (use (match_operand:HI 3 "memory_operand" "m"))
-   (clobber (match_scratch:DF 4 "=&1f"))]
-  "TARGET_80387 && FLOAT_MODE_P (GET_MODE (operands[1]))
-   && (!SSE_FLOAT_MODE_P (GET_MODE (operands[1])) || !TARGET_64BIT)"
-  "* return output_fix_trunc (insn, operands);"
-  [(set_attr "type" "fistp")
-   (set_attr "i387_cw" "trunc")
-   (set_attr "mode" "DI")])
-
-(define_split 
-  [(set (match_operand:DI 0 "register_operand" "")
-       (fix:DI (match_operand 1 "register_operand" "")))
-   (use (match_operand:HI 2 "memory_operand" ""))
-   (use (match_operand:HI 3 "memory_operand" ""))
-   (clobber (match_operand:DI 4 "memory_operand" ""))
-   (clobber (match_scratch 5 ""))]
-  "reload_completed"
-  [(parallel [(set (match_dup 4) (fix:DI (match_dup 1)))
-             (use (match_dup 2))
-             (use (match_dup 3))
-             (clobber (match_dup 5))])
-   (set (match_dup 0) (match_dup 4))]
-  "")
-
-(define_split 
-  [(set (match_operand:DI 0 "memory_operand" "")
-       (fix:DI (match_operand 1 "register_operand" "")))
-   (use (match_operand:HI 2 "memory_operand" ""))
-   (use (match_operand:HI 3 "memory_operand" ""))
-   (clobber (match_operand:DI 4 "memory_operand" ""))
-   (clobber (match_scratch 5 ""))]
-  "reload_completed"
-  [(parallel [(set (match_dup 0) (fix:DI (match_dup 1)))
-             (use (match_dup 2))
-             (use (match_dup 3))
-             (clobber (match_dup 5))])]
-  "")
-
-;; When SSE available, it is always faster to use it!
-(define_insn "fix_truncsfdi_sse"
-  [(set (match_operand:DI 0 "register_operand" "=r,r")
-       (fix:DI (match_operand:SF 1 "nonimmediate_operand" "x,xm")))]
-  "TARGET_64BIT && TARGET_SSE"
-  "cvttss2si{q}\t{%1, %0|%0, %1}"
-  [(set_attr "type" "sseicvt")
-   (set_attr "mode" "SF")
-   (set_attr "athlon_decode" "double,vector")])
-
-;; Avoid vector decoded form of the instruction.
-(define_peephole2
-  [(match_scratch:SF 2 "x")
-   (set (match_operand:DI 0 "register_operand" "")
-       (fix:DI (match_operand:SF 1 "memory_operand" "")))]
-  "TARGET_K8 && !optimize_size"
-  [(set (match_dup 2) (match_dup 1))
-   (set (match_dup 0) (fix:DI (match_dup 2)))]
-  "")
-
-(define_insn "fix_truncdfdi_sse"
-  [(set (match_operand:DI 0 "register_operand" "=r,r")
-       (fix:DI (match_operand:DF 1 "nonimmediate_operand" "Y,Ym")))]
-  "TARGET_64BIT && TARGET_SSE2"
-  "cvttsd2si{q}\t{%1, %0|%0, %1}"
-  [(set_attr "type" "sseicvt,sseicvt")
-   (set_attr "mode" "DF")
-   (set_attr "athlon_decode" "double,vector")])
-
-;; Avoid vector decoded form of the instruction.
-(define_peephole2
-  [(match_scratch:DF 2 "Y")
-   (set (match_operand:DI 0 "register_operand" "")
-       (fix:DI (match_operand:DF 1 "memory_operand" "")))]
-  "TARGET_K8 && !optimize_size"
-  [(set (match_dup 2) (match_dup 1))
-   (set (match_dup 0) (fix:DI (match_dup 2)))]
-  "")
-
 ;; Signed conversion to SImode.
 
 (define_expand "fix_truncxfsi2"
   [(parallel [(set (match_operand:SI 0 "nonimmediate_operand" "")
-                  (fix:SI (match_operand:XF 1 "register_operand" "")))
+                   (fix:SI (match_operand:XF 1 "register_operand" "")))
              (clobber (reg:CC FLAGS_REG))])]
   "TARGET_80387"
-  "")
-
-(define_expand "fix_truncdfsi2"
-  [(parallel [(set (match_operand:SI 0 "nonimmediate_operand" "")
-                  (fix:SI (match_operand:DF 1 "register_operand" "")))
-             (clobber (reg:CC FLAGS_REG))])]
-  "TARGET_80387 || TARGET_SSE2"
 {
-  if (TARGET_SSE2)
+  if (TARGET_FISTTP)
    {
-     rtx out = REG_P (operands[0]) ? operands[0] : gen_reg_rtx (SImode);
-     emit_insn (gen_fix_truncdfsi_sse (out, operands[1]));
-     if (out != operands[0])
-       emit_move_insn (operands[0], out);
+     emit_insn (gen_fix_truncsi_fisttp_i387_1 (operands[0], operands[1]));
      DONE;
    }
 })
 
-(define_expand "fix_truncsfsi2"
+(define_expand "fix_trunc<mode>si2"
   [(parallel [(set (match_operand:SI 0 "nonimmediate_operand" "")
-                  (fix:SI (match_operand:SF 1 "register_operand" "")))
-             (clobber (reg:CC FLAGS_REG))])] 
-  "TARGET_80387 || TARGET_SSE"
+                  (fix:SI (match_operand:SSEMODEF 1 "register_operand" "")))
+             (clobber (reg:CC FLAGS_REG))])]
+  "TARGET_80387 || (SSE_FLOAT_MODE_P (<MODE>mode))"
 {
-  if (TARGET_SSE)
+  if (TARGET_FISTTP
+      && !(SSE_FLOAT_MODE_P (<MODE>mode) && TARGET_SSE_MATH))
+   {
+     emit_insn (gen_fix_truncsi_fisttp_i387_1 (operands[0], operands[1]));
+     DONE;
+   }
+  if (SSE_FLOAT_MODE_P (<MODE>mode))
    {
      rtx out = REG_P (operands[0]) ? operands[0] : gen_reg_rtx (SImode);
-     emit_insn (gen_fix_truncsfsi_sse (out, operands[1]));
+     emit_insn (gen_fix_trunc<mode>si_sse (out, operands[1]));
      if (out != operands[0])
        emit_move_insn (operands[0], out);
      DONE;
    }
 })
 
-;; See the comments in i386.h near OPTIMIZE_MODE_SWITCHING for the description
-;; of the machinery.
-(define_insn_and_split "*fix_truncsi_i387"
-  [(set (match_operand:SI 0 "nonimmediate_operand" "=m,?r")
-       (fix:SI (match_operand 1 "register_operand" "f,f")))
-   (clobber (reg:CC FLAGS_REG))]
-  "TARGET_80387 && FLOAT_MODE_P (GET_MODE (operands[1]))
-   && !reload_completed && !reload_in_progress
-   && !SSE_FLOAT_MODE_P (GET_MODE (operands[1]))"
-  "#"
-  "&& 1"
-  [(const_int 0)]
+;; Signed conversion to HImode.
+
+(define_expand "fix_trunc<mode>hi2"
+  [(parallel [(set (match_operand:HI 0 "nonimmediate_operand" "")
+                  (fix:HI (match_operand:X87MODEF 1 "register_operand" "")))
+              (clobber (reg:CC FLAGS_REG))])]
+  "TARGET_80387
+   && !(SSE_FLOAT_MODE_P (<MODE>mode) && (!TARGET_FISTTP || TARGET_SSE_MATH))"
 {
-  ix86_optimize_mode_switching = 1;
-  operands[2] = assign_386_stack_local (HImode, 1);
-  operands[3] = assign_386_stack_local (HImode, 2);
-  if (memory_operand (operands[0], VOIDmode))
-    emit_insn (gen_fix_truncsi_memory (operands[0], operands[1],
-                                      operands[2], operands[3]));
-  else
-    {
-      operands[4] = assign_386_stack_local (SImode, 0);
-      emit_insn (gen_fix_truncsi_nomemory (operands[0], operands[1],
-                                          operands[2], operands[3],
-                                          operands[4]));
-    }
-  DONE;
-}
-  [(set_attr "type" "fistp")
-   (set_attr "i387_cw" "trunc")
-   (set_attr "mode" "SI")])
+  if (TARGET_FISTTP)
+   {
+     emit_insn (gen_fix_trunchi_fisttp_i387_1 (operands[0], operands[1]));
+     DONE;
+   }
+})
 
-(define_insn "fix_truncsi_nomemory"
-  [(set (match_operand:SI 0 "nonimmediate_operand" "=m,?r")
-       (fix:SI (match_operand 1 "register_operand" "f,f")))
-   (use (match_operand:HI 2 "memory_operand" "m,m"))
-   (use (match_operand:HI 3 "memory_operand" "m,m"))
-   (clobber (match_operand:SI 4 "memory_operand" "=m,m"))]
-  "TARGET_80387 && FLOAT_MODE_P (GET_MODE (operands[1]))
-   && !SSE_FLOAT_MODE_P (GET_MODE (operands[1]))"
-  "#"
-  [(set_attr "type" "fistp")
-   (set_attr "i387_cw" "trunc")
-   (set_attr "mode" "SI")])
+;; When SSE is available, it is always faster to use it!
+(define_insn "fix_truncsfdi_sse"
+  [(set (match_operand:DI 0 "register_operand" "=r,r")
+       (fix:DI (match_operand:SF 1 "nonimmediate_operand" "x,xm")))]
+  "TARGET_64BIT && TARGET_SSE && (!TARGET_FISTTP || TARGET_SSE_MATH)"
+  "cvttss2si{q}\t{%1, %0|%0, %1}"
+  [(set_attr "type" "sseicvt")
+   (set_attr "mode" "SF")
+   (set_attr "athlon_decode" "double,vector")])
 
-(define_insn "fix_truncsi_memory"
-  [(set (match_operand:SI 0 "memory_operand" "=m")
-       (fix:SI (match_operand 1 "register_operand" "f")))
-   (use (match_operand:HI 2 "memory_operand" "m"))
-   (use (match_operand:HI 3 "memory_operand" "m"))]
-  "TARGET_80387 && FLOAT_MODE_P (GET_MODE (operands[1]))
-   && !SSE_FLOAT_MODE_P (GET_MODE (operands[1]))"
-  "* return output_fix_trunc (insn, operands);"
-  [(set_attr "type" "fistp")
-   (set_attr "i387_cw" "trunc")
-   (set_attr "mode" "SI")])
+(define_insn "fix_truncdfdi_sse"
+  [(set (match_operand:DI 0 "register_operand" "=r,r")
+       (fix:DI (match_operand:DF 1 "nonimmediate_operand" "Y,Ym")))]
+  "TARGET_64BIT && TARGET_SSE2 && (!TARGET_FISTTP || TARGET_SSE_MATH)"
+  "cvttsd2si{q}\t{%1, %0|%0, %1}"
+  [(set_attr "type" "sseicvt")
+   (set_attr "mode" "DF")
+   (set_attr "athlon_decode" "double,vector")])
 
-;; When SSE available, it is always faster to use it!
 (define_insn "fix_truncsfsi_sse"
   [(set (match_operand:SI 0 "register_operand" "=r,r")
        (fix:SI (match_operand:SF 1 "nonimmediate_operand" "x,xm")))]
-  "TARGET_SSE"
+  "TARGET_SSE && (!TARGET_FISTTP || TARGET_SSE_MATH)"
   "cvttss2si\t{%1, %0|%0, %1}"
   [(set_attr "type" "sseicvt")
    (set_attr "mode" "DF")
    (set_attr "athlon_decode" "double,vector")])
 
-;; Avoid vector decoded form of the instruction.
-(define_peephole2
-  [(match_scratch:SF 2 "x")
-   (set (match_operand:SI 0 "register_operand" "")
-       (fix:SI (match_operand:SF 1 "memory_operand" "")))]
-  "TARGET_K8 && !optimize_size"
-  [(set (match_dup 2) (match_dup 1))
-   (set (match_dup 0) (fix:SI (match_dup 2)))]
-  "")
-
 (define_insn "fix_truncdfsi_sse"
   [(set (match_operand:SI 0 "register_operand" "=r,r")
        (fix:SI (match_operand:DF 1 "nonimmediate_operand" "Y,Ym")))]
-  "TARGET_SSE2"
+  "TARGET_SSE2 && (!TARGET_FISTTP || TARGET_SSE_MATH)"
   "cvttsd2si\t{%1, %0|%0, %1}"
   [(set_attr "type" "sseicvt")
    (set_attr "mode" "DF")
    (set_attr "athlon_decode" "double,vector")])
 
-;; Avoid vector decoded form of the instruction.
+;; Avoid vector decoded forms of the instruction.
 (define_peephole2
   [(match_scratch:DF 2 "Y")
-   (set (match_operand:SI 0 "register_operand" "")
-       (fix:SI (match_operand:DF 1 "memory_operand" "")))]
+   (set (match_operand:SSEMODEI24 0 "register_operand" "")
+       (fix:SSEMODEI24 (match_operand:DF 1 "memory_operand" "")))]
   "TARGET_K8 && !optimize_size"
   [(set (match_dup 2) (match_dup 1))
-   (set (match_dup 0) (fix:SI (match_dup 2)))]
+   (set (match_dup 0) (fix:SSEMODEI24 (match_dup 2)))]
   "")
 
-(define_split 
-  [(set (match_operand:SI 0 "register_operand" "")
-       (fix:SI (match_operand 1 "register_operand" "")))
-   (use (match_operand:HI 2 "memory_operand" ""))
-   (use (match_operand:HI 3 "memory_operand" ""))
-   (clobber (match_operand:SI 4 "memory_operand" ""))]
-  "reload_completed"
-  [(parallel [(set (match_dup 4) (fix:SI (match_dup 1)))
-             (use (match_dup 2))
-             (use (match_dup 3))])
-   (set (match_dup 0) (match_dup 4))]
+(define_peephole2
+  [(match_scratch:SF 2 "x")
+   (set (match_operand:SSEMODEI24 0 "register_operand" "")
+       (fix:SSEMODEI24 (match_operand:SF 1 "memory_operand" "")))]
+  "TARGET_K8 && !optimize_size"
+  [(set (match_dup 2) (match_dup 1))
+   (set (match_dup 0) (fix:SSEMODEI24 (match_dup 2)))]
   "")
 
-(define_split 
-  [(set (match_operand:SI 0 "memory_operand" "")
-       (fix:SI (match_operand 1 "register_operand" "")))
-   (use (match_operand:HI 2 "memory_operand" ""))
-   (use (match_operand:HI 3 "memory_operand" ""))
-   (clobber (match_operand:SI 4 "memory_operand" ""))]
-  "reload_completed"
-  [(parallel [(set (match_dup 0) (fix:SI (match_dup 1)))
-             (use (match_dup 2))
-             (use (match_dup 3))])]
-  "")
+(define_insn_and_split "fix_trunc<mode>_fisttp_i387_1"
+  [(set (match_operand:X87MODEI 0 "nonimmediate_operand" "=m,?r")
+       (fix:X87MODEI (match_operand 1 "register_operand" "f,f")))]
+  "TARGET_80387 && TARGET_FISTTP
+   && FLOAT_MODE_P (GET_MODE (operands[1]))
+   && !((SSE_FLOAT_MODE_P (GET_MODE (operands[1]))
+        && (TARGET_64BIT || <MODE>mode != DImode))
+       && TARGET_SSE_MATH)
+   && !(reload_completed || reload_in_progress)"
+  "#"
+  "&& 1"
+  [(const_int 0)]
+{
+  if (memory_operand (operands[0], VOIDmode))
+    emit_insn (gen_fix_trunc<mode>_i387_fisttp (operands[0], operands[1]));
+  else
+    {
+      operands[2] = assign_386_stack_local (<MODE>mode, 0);
+      emit_insn (gen_fix_trunc<mode>_i387_fisttp_with_temp (operands[0],
+                                                           operands[1],
+                                                           operands[2]));
+    }
+  DONE;
+}
+  [(set_attr "type" "fisttp")
+   (set_attr "mode" "<MODE>")])
 
-;; Signed conversion to HImode.
+(define_insn "fix_trunc<mode>_i387_fisttp"
+  [(set (match_operand:X87MODEI 0 "memory_operand" "=m")
+       (fix:X87MODEI (match_operand 1 "register_operand" "f")))
+   (clobber (match_scratch:XF 2 "=&1f"))]
+  "TARGET_80387 && TARGET_FISTTP
+   && FLOAT_MODE_P (GET_MODE (operands[1]))
+   && !((SSE_FLOAT_MODE_P (GET_MODE (operands[1]))
+        && (TARGET_64BIT || <MODE>mode != DImode))
+       && TARGET_SSE_MATH)"
+  "* return output_fix_trunc (insn, operands, 1);"
+  [(set_attr "type" "fisttp")
+   (set_attr "mode" "<MODE>")])
 
-(define_expand "fix_truncxfhi2"
-  [(parallel [(set (match_operand:HI 0 "nonimmediate_operand" "")
-                   (fix:HI (match_operand:XF 1 "register_operand" "")))
-              (clobber (reg:CC FLAGS_REG))])] 
-  "TARGET_80387"
-  "")
+(define_insn "fix_trunc<mode>_i387_fisttp_with_temp"
+  [(set (match_operand:X87MODEI 0 "nonimmediate_operand" "=m,?r")
+       (fix:X87MODEI (match_operand 1 "register_operand" "f,f")))
+   (clobber (match_operand:X87MODEI 2 "memory_operand" "=m,m"))
+   (clobber (match_scratch:XF 3 "=&1f,&1f"))]
+  "TARGET_80387 && TARGET_FISTTP
+   && FLOAT_MODE_P (GET_MODE (operands[1]))
+   && !((SSE_FLOAT_MODE_P (GET_MODE (operands[1]))
+       && (TARGET_64BIT || <MODE>mode != DImode))
+       && TARGET_SSE_MATH)"
+  "#"
+  [(set_attr "type" "fisttp")
+   (set_attr "mode" "<MODE>")])
 
-(define_expand "fix_truncdfhi2"
-  [(parallel [(set (match_operand:HI 0 "nonimmediate_operand" "")
-                  (fix:HI (match_operand:DF 1 "register_operand" "")))
-              (clobber (reg:CC FLAGS_REG))])]
-  "TARGET_80387 && !TARGET_SSE2"
+(define_split
+  [(set (match_operand:X87MODEI 0 "register_operand" "")
+       (fix:X87MODEI (match_operand 1 "register_operand" "")))
+   (clobber (match_operand:X87MODEI 2 "memory_operand" ""))
+   (clobber (match_scratch 3 ""))]
+  "reload_completed"
+  [(parallel [(set (match_dup 2) (fix:X87MODEI (match_dup 1)))
+             (clobber (match_dup 3))])
+   (set (match_dup 0) (match_dup 2))]
   "")
 
-(define_expand "fix_truncsfhi2"
-  [(parallel [(set (match_operand:HI 0 "nonimmediate_operand" "")
-                  (fix:HI (match_operand:SF 1 "register_operand" "")))
-               (clobber (reg:CC FLAGS_REG))])]
-  "TARGET_80387 && !TARGET_SSE"
+(define_split
+  [(set (match_operand:X87MODEI 0 "memory_operand" "")
+       (fix:X87MODEI (match_operand 1 "register_operand" "")))
+   (clobber (match_operand:X87MODEI 2 "memory_operand" ""))
+   (clobber (match_scratch 3 ""))]
+  "reload_completed"
+  [(parallel [(set (match_dup 0) (fix:X87MODEI (match_dup 1)))
+             (clobber (match_dup 3))])]
   "")
 
 ;; See the comments in i386.h near OPTIMIZE_MODE_SWITCHING for the description
-;; of the machinery.
-(define_insn_and_split "*fix_trunchi_i387"
-  [(set (match_operand:HI 0 "nonimmediate_operand" "=m,?r")
-       (fix:HI (match_operand 1 "register_operand" "f,f")))
-   (clobber (reg:CC FLAGS_REG))]
-  "TARGET_80387 && FLOAT_MODE_P (GET_MODE (operands[1]))
-   && !reload_completed && !reload_in_progress
-   && !SSE_FLOAT_MODE_P (GET_MODE (operands[1]))"
+;; of the machinery. Please note the clobber of FLAGS_REG. In i387 control
+;; word calculation (inserted by LCM in mode switching pass) a FLAGS_REG
+;; clobbering insns can be used. Look at emit_i387_cw_initialization ()
+;; function in i386.c.
+(define_insn_and_split "*fix_trunc<mode>_i387_1"
+  [(set (match_operand:X87MODEI 0 "nonimmediate_operand" "=m,?r")
+       (fix:X87MODEI (match_operand 1 "register_operand" "f,f")))
+   (clobber (reg:CC FLAGS_REG))]
+  "TARGET_80387 && !TARGET_FISTTP
+   && FLOAT_MODE_P (GET_MODE (operands[1]))
+   && !(SSE_FLOAT_MODE_P (GET_MODE (operands[1]))
+        && (TARGET_64BIT || <MODE>mode != DImode))
+   && !(reload_completed || reload_in_progress)"
   "#"
   "&& 1"
   [(const_int 0)]
@@ -4370,70 +4254,129 @@
   operands[2] = assign_386_stack_local (HImode, 1);
   operands[3] = assign_386_stack_local (HImode, 2);
   if (memory_operand (operands[0], VOIDmode))
-    emit_insn (gen_fix_trunchi_memory (operands[0], operands[1],
-                                      operands[2], operands[3]));
+    emit_insn (gen_fix_trunc<mode>_i387 (operands[0], operands[1],
+                                        operands[2], operands[3]));
   else
     {
-      operands[4] = assign_386_stack_local (HImode, 0);
-      emit_insn (gen_fix_trunchi_nomemory (operands[0], operands[1],
-                                          operands[2], operands[3],
-                                          operands[4]));
+      operands[4] = assign_386_stack_local (<MODE>mode, 0);
+      emit_insn (gen_fix_trunc<mode>_i387_with_temp (operands[0], operands[1],
+                                                    operands[2], operands[3],
+                                                    operands[4]));
     }
   DONE;
 }
   [(set_attr "type" "fistp")
    (set_attr "i387_cw" "trunc")
-   (set_attr "mode" "HI")])
+   (set_attr "mode" "<MODE>")])
+
+(define_insn "fix_truncdi_i387"
+  [(set (match_operand:DI 0 "memory_operand" "=m")
+       (fix:DI (match_operand 1 "register_operand" "f")))
+   (use (match_operand:HI 2 "memory_operand" "m"))
+   (use (match_operand:HI 3 "memory_operand" "m"))
+   (clobber (match_scratch:XF 4 "=&1f"))]
+  "TARGET_80387 && !TARGET_FISTTP
+   && FLOAT_MODE_P (GET_MODE (operands[1]))
+   && !(TARGET_64BIT && SSE_FLOAT_MODE_P (GET_MODE (operands[1])))"
+  "* return output_fix_trunc (insn, operands, 0);"
+  [(set_attr "type" "fistp")
+   (set_attr "i387_cw" "trunc")
+   (set_attr "mode" "DI")])
 
-(define_insn "fix_trunchi_nomemory"
-  [(set (match_operand:HI 0 "nonimmediate_operand" "=m,?r")
-       (fix:HI (match_operand 1 "register_operand" "f,f")))
+(define_insn "fix_truncdi_i387_with_temp"
+  [(set (match_operand:DI 0 "nonimmediate_operand" "=m,?r")
+       (fix:DI (match_operand 1 "register_operand" "f,f")))
    (use (match_operand:HI 2 "memory_operand" "m,m"))
    (use (match_operand:HI 3 "memory_operand" "m,m"))
-   (clobber (match_operand:HI 4 "memory_operand" "=m,m"))]
-  "TARGET_80387 && FLOAT_MODE_P (GET_MODE (operands[1]))
-   && !SSE_FLOAT_MODE_P (GET_MODE (operands[1]))"
+   (clobber (match_operand:DI 4 "memory_operand" "=m,m"))
+   (clobber (match_scratch:XF 5 "=&1f,&1f"))]
+  "TARGET_80387 && !TARGET_FISTTP
+   && FLOAT_MODE_P (GET_MODE (operands[1]))
+   && !(TARGET_64BIT && SSE_FLOAT_MODE_P (GET_MODE (operands[1])))"
   "#"
   [(set_attr "type" "fistp")
    (set_attr "i387_cw" "trunc")
-   (set_attr "mode" "HI")])
+   (set_attr "mode" "DI")])
 
-(define_insn "fix_trunchi_memory"
-  [(set (match_operand:HI 0 "memory_operand" "=m")
-       (fix:HI (match_operand 1 "register_operand" "f")))
+(define_split 
+  [(set (match_operand:DI 0 "register_operand" "")
+       (fix:DI (match_operand 1 "register_operand" "")))
+   (use (match_operand:HI 2 "memory_operand" ""))
+   (use (match_operand:HI 3 "memory_operand" ""))
+   (clobber (match_operand:DI 4 "memory_operand" ""))
+   (clobber (match_scratch 5 ""))]
+  "reload_completed"
+  [(parallel [(set (match_dup 4) (fix:DI (match_dup 1)))
+             (use (match_dup 2))
+             (use (match_dup 3))
+             (clobber (match_dup 5))])
+   (set (match_dup 0) (match_dup 4))]
+  "")
+
+(define_split 
+  [(set (match_operand:DI 0 "memory_operand" "")
+       (fix:DI (match_operand 1 "register_operand" "")))
+   (use (match_operand:HI 2 "memory_operand" ""))
+   (use (match_operand:HI 3 "memory_operand" ""))
+   (clobber (match_operand:DI 4 "memory_operand" ""))
+   (clobber (match_scratch 5 ""))]
+  "reload_completed"
+  [(parallel [(set (match_dup 0) (fix:DI (match_dup 1)))
+             (use (match_dup 2))
+             (use (match_dup 3))
+             (clobber (match_dup 5))])]
+  "")
+
+(define_insn "fix_trunc<mode>_i387"
+  [(set (match_operand:X87MODEI12 0 "memory_operand" "=m")
+       (fix:X87MODEI12 (match_operand 1 "register_operand" "f")))
    (use (match_operand:HI 2 "memory_operand" "m"))
    (use (match_operand:HI 3 "memory_operand" "m"))]
-  "TARGET_80387 && FLOAT_MODE_P (GET_MODE (operands[1]))
+  "TARGET_80387 && !TARGET_FISTTP
+   && FLOAT_MODE_P (GET_MODE (operands[1]))
    && !SSE_FLOAT_MODE_P (GET_MODE (operands[1]))"
-  "* return output_fix_trunc (insn, operands);"
+  "* return output_fix_trunc (insn, operands, 0);"
   [(set_attr "type" "fistp")
    (set_attr "i387_cw" "trunc")
-   (set_attr "mode" "HI")])
+   (set_attr "mode" "<MODE>")])
+
+(define_insn "fix_trunc<mode>_i387_with_temp"
+  [(set (match_operand:X87MODEI12 0 "nonimmediate_operand" "=m,?r")
+       (fix:X87MODEI12 (match_operand 1 "register_operand" "f,f")))
+   (use (match_operand:HI 2 "memory_operand" "m,m"))
+   (use (match_operand:HI 3 "memory_operand" "m,m"))
+   (clobber (match_operand:X87MODEI12 4 "memory_operand" "=m,m"))]
+  "TARGET_80387 && !TARGET_FISTTP
+   && FLOAT_MODE_P (GET_MODE (operands[1]))
+   && !SSE_FLOAT_MODE_P (GET_MODE (operands[1]))"
+  "#"
+  [(set_attr "type" "fistp")
+   (set_attr "i387_cw" "trunc")
+   (set_attr "mode" "<MODE>")])
 
 (define_split 
-  [(set (match_operand:HI 0 "memory_operand" "")
-       (fix:HI (match_operand 1 "register_operand" "")))
+  [(set (match_operand:X87MODEI12 0 "register_operand" "")
+       (fix:X87MODEI12 (match_operand 1 "register_operand" "")))
    (use (match_operand:HI 2 "memory_operand" ""))
    (use (match_operand:HI 3 "memory_operand" ""))
-   (clobber (match_operand:HI 4 "memory_operand" ""))]
+   (clobber (match_operand:X87MODEI12 4 "memory_operand" ""))]
   "reload_completed"
-  [(parallel [(set (match_dup 0) (fix:HI (match_dup 1)))
+  [(parallel [(set (match_dup 4) (fix:X87MODEI12 (match_dup 1)))
              (use (match_dup 2))
-             (use (match_dup 3))])]
+             (use (match_dup 3))])
+   (set (match_dup 0) (match_dup 4))]
   "")
 
 (define_split 
-  [(set (match_operand:HI 0 "register_operand" "")
-       (fix:HI (match_operand 1 "register_operand" "")))
+  [(set (match_operand:X87MODEI12 0 "memory_operand" "")
+       (fix:X87MODEI12 (match_operand 1 "register_operand" "")))
    (use (match_operand:HI 2 "memory_operand" ""))
    (use (match_operand:HI 3 "memory_operand" ""))
-   (clobber (match_operand:HI 4 "memory_operand" ""))]
+   (clobber (match_operand:X87MODEI12 4 "memory_operand" ""))]
   "reload_completed"
-  [(parallel [(set (match_dup 4) (fix:HI (match_dup 1)))
+  [(parallel [(set (match_dup 0) (fix:X87MODEI12 (match_dup 1)))
              (use (match_dup 2))
-             (use (match_dup 3))
-             (clobber (match_dup 4))])
-   (set (match_dup 0) (match_dup 4))]
+             (use (match_dup 3))])]
   "")
 
 (define_insn "x86_fnstcw_1"