re PR target/20754 (ACATS cxg1005 fails at runtime on hppa-linux)

	PR target/20754
	* pa.md: Create separate 32 and 64-bit move patterns for SI, DI, SF
	and DF modes.  Add alternatives to copy between general and floating
	point registers to the 32-bit patterns.
	* pa-64.h (SECONDARY_MEMORY_NEEDED_RTX): Delete undefine.
	* pa.h (SECONDARY_MEMORY_NEEDED_RTX): Delete define.
	(SECONDARY_MEMORY_NEEDED): Secondary memory is only needed when
	generating 64-bit code.
	* pa.c (output_move_double): Handle copies between general and
	floating registers.

From-SVN: r109557

5 files changed, 116 insertions, 30 deletions

diff --git a/gcc/ChangeLog b/gcc/ChangeLog
index b74c09f..e4a8ce86 100644
--- a/gcc/ChangeLog
+++ b/gcc/ChangeLog
@@ -1,3 +1,16 @@
+2006-01-10  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
+
+	PR target/20754
+	* pa.md: Create separate 32 and 64-bit move patterns for SI, DI, SF
+	and DF modes.  Add alternatives to copy between general and floating
+	point registers to the 32-bit patterns.
+	* pa-64.h (SECONDARY_MEMORY_NEEDED_RTX): Delete undefine.
+	* pa.h (SECONDARY_MEMORY_NEEDED_RTX): Delete define.
+	(SECONDARY_MEMORY_NEEDED): Secondary memory is only needed when
+	generating 64-bit code.
+	* pa.c (output_move_double): Handle copies between general and
+	floating registers.
+
 2006-01-10  Stuart Hastings  <stuart@apple.com>
 
 	* gcc/config/i386/i386.md (set_got): Update.
diff --git a/gcc/config/pa/pa-64.h b/gcc/config/pa/pa-64.h
index a59b4e0..750b13c 100644
--- a/gcc/config/pa/pa-64.h
+++ b/gcc/config/pa/pa-64.h
@@ -70,10 +70,6 @@ Boston, MA 02110-1301, USA.  */
    relocs which appear in stabs.  */
 #undef DBX_DEBUGGING_INFO
 
-/* We want the compiler to select a suitable secondary memory location.
-   ?!? This may not work reliably.  Keep an eye out for problems.  */
-#undef SECONDARY_MEMORY_NEEDED_RTX
-
 /* ?!? This needs to be made compile-time selectable.
 
    The PA64 runtime model has arguments that grow to higher addresses
diff --git a/gcc/config/pa/pa.c b/gcc/config/pa/pa.c
index d4800fe..0d8ec24 100644
--- a/gcc/config/pa/pa.c
+++ b/gcc/config/pa/pa.c
@@ -2209,6 +2209,25 @@ output_move_double (rtx *operands)
      supposed to allow to happen.  */
   gcc_assert (optype0 == REGOP || optype1 == REGOP);
 
+  /* Handle copies between general and floating registers.  */
+
+  if (optype0 == REGOP && optype1 == REGOP
+      && FP_REG_P (operands[0]) ^ FP_REG_P (operands[1]))
+    {
+      if (FP_REG_P (operands[0]))
+	{
+	  output_asm_insn ("{stws|stw} %1,-16(%%sp)", operands);
+	  output_asm_insn ("{stws|stw} %R1,-12(%%sp)", operands);
+	  return "{fldds|fldd} -16(%%sp),%0";
+	}
+      else
+	{
+	  output_asm_insn ("{fstds|fstd} %1,-16(%%sp)", operands);
+	  output_asm_insn ("{ldws|ldw} -16(%%sp),%0", operands);
+	  return "{ldws|ldw} -12(%%sp),%R0";
+	}
+    }
+
    /* Handle auto decrementing and incrementing loads and stores
      specifically, since the structure of the function doesn't work
      for them without major modification.  Do it better when we learn
diff --git a/gcc/config/pa/pa.h b/gcc/config/pa/pa.h
index a98d1cd..c29a876 100644
--- a/gcc/config/pa/pa.h
+++ b/gcc/config/pa/pa.h
@@ -531,14 +531,15 @@ extern struct rtx_def *hppa_pic_save_rtx (void);
   reg_classes_intersect_p ((CLASS), FP_REGS)
 
 /* On the PA it is not possible to directly move data between
-   GENERAL_REGS and FP_REGS.  */
-#define SECONDARY_MEMORY_NEEDED(CLASS1, CLASS2, MODE)		\
-  (MAYBE_FP_REG_CLASS_P (CLASS1) != FP_REG_CLASS_P (CLASS2)	\
-   || MAYBE_FP_REG_CLASS_P (CLASS2) != FP_REG_CLASS_P (CLASS1))
-
-/* Return the stack location to use for secondary memory needed reloads.  */
-#define SECONDARY_MEMORY_NEEDED_RTX(MODE) \
-  gen_rtx_MEM (MODE, gen_rtx_PLUS (Pmode, stack_pointer_rtx, GEN_INT (-16)))
+   GENERAL_REGS and FP_REGS.  On the 32-bit port, we use the
+   location at SP-16.  We don't expose this location in the RTL to
+   avoid scheduling related problems.  For example, the store and
+   load could be separated by a call to a pure or const function
+   which has no frame and uses SP-16.  */
+#define SECONDARY_MEMORY_NEEDED(CLASS1, CLASS2, MODE)			\
+  (TARGET_64BIT								\
+   && (MAYBE_FP_REG_CLASS_P (CLASS1) != FP_REG_CLASS_P (CLASS2)		\
+       || MAYBE_FP_REG_CLASS_P (CLASS2) != FP_REG_CLASS_P (CLASS1)))
 
 
 /* Stack layout; function entry, exit and calling.  */
diff --git a/gcc/config/pa/pa.md b/gcc/config/pa/pa.md
index 556c941..b51feb0 100644
--- a/gcc/config/pa/pa.md
+++ b/gcc/config/pa/pa.md
@@ -2307,12 +2307,41 @@
 
 (define_insn ""
   [(set (match_operand:SI 0 "move_dest_operand"
+			  "=r,r,r,r,r,r,Q,!*q,!r,!*f,*f,T,r,f")
+	(match_operand:SI 1 "move_src_operand"
+			  "A,r,J,N,K,RQ,rM,!rM,!*q,!*fM,RT,*f,f,r"))]
+  "(register_operand (operands[0], SImode)
+    || reg_or_0_operand (operands[1], SImode))
+   && !TARGET_SOFT_FLOAT
+   && !TARGET_64BIT"
+  "@
+   ldw RT'%A1,%0
+   copy %1,%0
+   ldi %1,%0
+   ldil L'%1,%0
+   {zdepi|depwi,z} %Z1,%0
+   ldw%M1 %1,%0
+   stw%M0 %r1,%0
+   mtsar %r1
+   {mfctl|mfctl,w} %%sar,%0
+   fcpy,sgl %f1,%0
+   fldw%F1 %1,%0
+   fstw%F0 %1,%0
+   {fstws|fstw} %1,-16(%%sp)\n\t{ldws|ldw} -16(%%sp),%0
+   {stws|stw} %1,-16(%%sp)\n\t{fldws|fldw} -16(%%sp),%0"
+  [(set_attr "type" "load,move,move,move,shift,load,store,move,move,fpalu,fpload,fpstore,move,move")
+   (set_attr "pa_combine_type" "addmove")
+   (set_attr "length" "4,4,4,4,4,4,4,4,4,4,4,4,8,8")])
+
+(define_insn ""
+  [(set (match_operand:SI 0 "move_dest_operand"
 			  "=r,r,r,r,r,r,Q,!*q,!r,!*f,*f,T")
 	(match_operand:SI 1 "move_src_operand"
 			  "A,r,J,N,K,RQ,rM,!rM,!*q,!*fM,RT,*f"))]
   "(register_operand (operands[0], SImode)
     || reg_or_0_operand (operands[1], SImode))
-   && !TARGET_SOFT_FLOAT"
+   && !TARGET_SOFT_FLOAT
+   && TARGET_64BIT"
   "@
    ldw RT'%A1,%0
    copy %1,%0
@@ -3840,9 +3869,9 @@
 
 (define_insn ""
   [(set (match_operand:DF 0 "move_dest_operand"
-			  "=f,*r,Q,?o,?Q,f,*r,*r")
+			  "=f,*r,Q,?o,?Q,f,*r,*r,r,f")
 	(match_operand:DF 1 "reg_or_0_or_nonsymb_mem_operand"
-			  "fG,*rG,f,*r,*r,RQ,o,RQ"))]
+			  "fG,*rG,f,*r,*r,RQ,o,RQ,f,r"))]
   "(register_operand (operands[0], DFmode)
     || reg_or_0_operand (operands[1], DFmode))
    && !(GET_CODE (operands[1]) == CONST_DOUBLE
@@ -3851,13 +3880,15 @@
    && !TARGET_SOFT_FLOAT"
   "*
 {
-  if (FP_REG_P (operands[0]) || FP_REG_P (operands[1])
-      || operands[1] == CONST0_RTX (DFmode))
+  if ((FP_REG_P (operands[0]) || FP_REG_P (operands[1])
+       || operands[1] == CONST0_RTX (DFmode))
+      && !(REG_P (operands[0]) && REG_P (operands[1])
+	   && FP_REG_P (operands[0]) ^ FP_REG_P (operands[1])))
     return output_fp_move_double (operands);
   return output_move_double (operands);
 }"
-  [(set_attr "type" "fpalu,move,fpstore,store,store,fpload,load,load")
-   (set_attr "length" "4,8,4,8,16,4,8,16")])
+  [(set_attr "type" "fpalu,move,fpstore,store,store,fpload,load,load,move,move")
+   (set_attr "length" "4,8,4,8,16,4,8,16,12,12")])
 
 (define_insn ""
   [(set (match_operand:DF 0 "indexed_memory_operand" "=R")
@@ -4012,9 +4043,9 @@
 
 (define_insn ""
   [(set (match_operand:DF 0 "move_dest_operand"
-			  "=r,?o,?Q,r,r")
+			  "=r,?o,?Q,r,r,r,f")
 	(match_operand:DF 1 "reg_or_0_or_nonsymb_mem_operand"
-			  "rG,r,r,o,RQ"))]
+			  "rG,r,r,o,RQ,f,r"))]
   "(register_operand (operands[0], DFmode)
     || reg_or_0_operand (operands[1], DFmode))
    && !TARGET_64BIT
@@ -4023,8 +4054,8 @@
 {
   return output_move_double (operands);
 }"
-  [(set_attr "type" "move,store,store,load,load")
-   (set_attr "length" "8,8,16,8,16")])
+  [(set_attr "type" "move,store,store,load,load,move,move")
+   (set_attr "length" "8,8,16,8,16,12,12")])
 
 (define_insn ""
   [(set (match_operand:DF 0 "move_dest_operand"
@@ -4154,22 +4185,25 @@
 
 (define_insn ""
   [(set (match_operand:DI 0 "move_dest_operand"
-			  "=r,o,Q,r,r,r,*f,*f,T")
+			  "=r,o,Q,r,r,r,*f,*f,T,r,f")
 	(match_operand:DI 1 "general_operand"
-			  "rM,r,r,o*R,Q,i,*fM,RT,*f"))]
+			  "rM,r,r,o*R,Q,i,*fM,RT,*f,f,r"))]
   "(register_operand (operands[0], DImode)
     || reg_or_0_operand (operands[1], DImode))
    && !TARGET_64BIT
    && !TARGET_SOFT_FLOAT"
   "*
 {
-  if (FP_REG_P (operands[0]) || FP_REG_P (operands[1])
-      || (operands[1] == CONST0_RTX (DImode)))
+  if ((FP_REG_P (operands[0]) || FP_REG_P (operands[1])
+       || operands[1] == CONST0_RTX (DFmode))
+      && !(REG_P (operands[0]) && REG_P (operands[1])
+	   && FP_REG_P (operands[0]) ^ FP_REG_P (operands[1])))
     return output_fp_move_double (operands);
   return output_move_double (operands);
 }"
-  [(set_attr "type" "move,store,store,load,load,multi,fpalu,fpload,fpstore")
-   (set_attr "length" "8,8,16,8,16,16,4,4,4")])
+  [(set_attr "type"
+    "move,store,store,load,load,multi,fpalu,fpload,fpstore,move,move")
+   (set_attr "length" "8,8,16,8,16,16,4,4,4,12,12")])
 
 (define_insn ""
   [(set (match_operand:DI 0 "move_dest_operand"
@@ -4380,12 +4414,35 @@
 
 (define_insn ""
   [(set (match_operand:SF 0 "move_dest_operand"
+			  "=f,!*r,f,*r,Q,Q,r,f")
+	(match_operand:SF 1 "reg_or_0_or_nonsymb_mem_operand"
+			  "fG,!*rG,RQ,RQ,f,*rG,f,r"))]
+  "(register_operand (operands[0], SFmode)
+    || reg_or_0_operand (operands[1], SFmode))
+   && !TARGET_SOFT_FLOAT
+   && !TARGET_64BIT"
+  "@
+   fcpy,sgl %f1,%0
+   copy %r1,%0
+   fldw%F1 %1,%0
+   ldw%M1 %1,%0
+   fstw%F0 %1,%0
+   stw%M0 %r1,%0
+   {fstws|fstw} %1,-16(%%sp)\n\t{ldws|ldw} -16(%%sp),%0
+   {stws|stw} %1,-16(%%sp)\n\t{fldws|fldw} -16(%%sp),%0"
+  [(set_attr "type" "fpalu,move,fpload,load,fpstore,store,move,move")
+   (set_attr "pa_combine_type" "addmove")
+   (set_attr "length" "4,4,4,4,4,4,8,8")])
+
+(define_insn ""
+  [(set (match_operand:SF 0 "move_dest_operand"
 			  "=f,!*r,f,*r,Q,Q")
 	(match_operand:SF 1 "reg_or_0_or_nonsymb_mem_operand"
 			  "fG,!*rG,RQ,RQ,f,*rG"))]
   "(register_operand (operands[0], SFmode)
     || reg_or_0_operand (operands[1], SFmode))
-   && !TARGET_SOFT_FLOAT"
+   && !TARGET_SOFT_FLOAT
+   && TARGET_64BIT"
   "@
    fcpy,sgl %f1,%0
    copy %r1,%0