re PR target/50358 (AVR: Implement [u]maddqihi4 [u]msubqihi4 patterns on the enhanced core)

	PR target/50358
	* config/avr/avr.md (*ashiftqihi2.signx.1): New insn.
	(*maddqi4, *maddqi4.const): New insns.
	(*msubqi4, *msubqi4.const): New insns.
	* config/avr/avr.c (avr_rtx_costs): Record costs of above in cases
	PLUS:QI and MINUS:QI.  Increase costs of multiply-add/-sub for
	HImode by 1 in the case of multiplying with a CONST_INT.
	Record cost of *ashiftqihi2.signx.1 in case ASHIFT:QI.

From-SVN: r178912

2 files changed, 116 insertions, 1 deletions

diff --git a/gcc/config/avr/avr.c b/gcc/config/avr/avr.c
index 773aec2..b69efa9 100644
--- a/gcc/config/avr/avr.c
+++ b/gcc/config/avr/avr.c
@@ -1528,7 +1528,7 @@ notice_update_cc (rtx body ATTRIBUTE_UNUSED, rtx insn)
       /* Insn doesn't leave CC in a usable state.  */
       CC_STATUS_INIT;
 
-      /* Correct CC for the ashrqi3 with the shift count as CONST_INT != 6 */
+      /* Correct CC for the ashrqi3 with the shift count as CONST_INT < 6 */
       set = single_set (insn);
       if (set)
 	{
@@ -5570,6 +5570,17 @@ avr_rtx_costs (rtx x, int codearg, int outer_code ATTRIBUTE_UNUSED,
       switch (mode)
 	{
 	case QImode:
+          if (AVR_HAVE_MUL
+              && MULT == GET_CODE (XEXP (x, 0))
+              && register_operand (XEXP (x, 1), QImode))
+            {
+              /* multiply-add */
+              *total = COSTS_N_INSNS (speed ? 4 : 3);
+              /* multiply-add with constant: will be split and load constant. */
+              if (CONST_INT_P (XEXP (XEXP (x, 0), 1)))
+                *total = COSTS_N_INSNS (1) + *total;
+              return true;
+            }
 	  *total = COSTS_N_INSNS (1);
 	  if (GET_CODE (XEXP (x, 1)) != CONST_INT)
 	    *total += avr_operand_rtx_cost (XEXP (x, 1), mode, code, 1, speed);
@@ -5583,7 +5594,11 @@ avr_rtx_costs (rtx x, int codearg, int outer_code ATTRIBUTE_UNUSED,
               && (ZERO_EXTEND == GET_CODE (XEXP (XEXP (x, 0), 0))
                   || SIGN_EXTEND == GET_CODE (XEXP (XEXP (x, 0), 0))))
             {
+              /* multiply-add */
               *total = COSTS_N_INSNS (speed ? 5 : 4);
+              /* multiply-add with constant: will be split and load constant. */
+              if (CONST_INT_P (XEXP (XEXP (x, 0), 1)))
+                *total = COSTS_N_INSNS (1) + *total;
               return true;
             }
 	  if (GET_CODE (XEXP (x, 1)) != CONST_INT)
@@ -5619,6 +5634,18 @@ avr_rtx_costs (rtx x, int codearg, int outer_code ATTRIBUTE_UNUSED,
 
     case MINUS:
       if (AVR_HAVE_MUL
+          && QImode == mode
+          && register_operand (XEXP (x, 0), QImode)
+          && MULT == GET_CODE (XEXP (x, 1)))
+        {
+          /* multiply-sub */
+          *total = COSTS_N_INSNS (speed ? 4 : 3);
+          /* multiply-sub with constant: will be split and load constant. */
+          if (CONST_INT_P (XEXP (XEXP (x, 1), 1)))
+            *total = COSTS_N_INSNS (1) + *total;
+          return true;
+        }
+      if (AVR_HAVE_MUL
           && HImode == mode
           && register_operand (XEXP (x, 0), HImode)
           && (MULT == GET_CODE (XEXP (x, 1))
@@ -5626,7 +5653,11 @@ avr_rtx_costs (rtx x, int codearg, int outer_code ATTRIBUTE_UNUSED,
           && (ZERO_EXTEND == GET_CODE (XEXP (XEXP (x, 1), 0))
               || SIGN_EXTEND == GET_CODE (XEXP (XEXP (x, 1), 0))))
         {
+          /* multiply-sub */
           *total = COSTS_N_INSNS (speed ? 5 : 4);
+          /* multiply-sub with constant: will be split and load constant. */
+          if (CONST_INT_P (XEXP (XEXP (x, 1), 1)))
+            *total = COSTS_N_INSNS (1) + *total;
           return true;
         }
     case AND:
@@ -5815,6 +5846,13 @@ avr_rtx_costs (rtx x, int codearg, int outer_code ATTRIBUTE_UNUSED,
                 }
             }
           
+          if (const1_rtx == (XEXP (x, 1))
+              && SIGN_EXTEND == GET_CODE (XEXP (x, 0)))
+            {
+              *total = COSTS_N_INSNS (2);
+              return true;
+            }
+          
 	  if (GET_CODE (XEXP (x, 1)) != CONST_INT)
 	    {
 	      *total = COSTS_N_INSNS (!speed ? 5 : 41);
diff --git a/gcc/config/avr/avr.md b/gcc/config/avr/avr.md
index 8721314..c161504 100644
--- a/gcc/config/avr/avr.md
+++ b/gcc/config/avr/avr.md
@@ -1138,6 +1138,72 @@
    (set_attr "cc" "clobber")])
 
 ;******************************************************************************
+; multiply-add/sub QI: $0 = $3 +/- $1*$2
+;******************************************************************************
+
+(define_insn "*maddqi4"
+  [(set (match_operand:QI 0 "register_operand"                  "=r")
+        (plus:QI (mult:QI (match_operand:QI 1 "register_operand" "r")
+                          (match_operand:QI 2 "register_operand" "r"))
+                 (match_operand:QI 3 "register_operand"          "0")))]
+  
+  "AVR_HAVE_MUL"
+  "mul %1,%2
+	add %A0,r0
+	clr __zero_reg__"
+  [(set_attr "length" "4")
+   (set_attr "cc" "clobber")])
+
+(define_insn "*msubqi4"
+  [(set (match_operand:QI 0 "register_operand"                   "=r")
+        (minus:QI (match_operand:QI 3 "register_operand"          "0")
+                  (mult:QI (match_operand:QI 1 "register_operand" "r")
+                           (match_operand:QI 2 "register_operand" "r"))))]
+  "AVR_HAVE_MUL"
+  "mul %1,%2
+	sub %A0,r0
+	clr __zero_reg__"
+  [(set_attr "length" "4")
+   (set_attr "cc" "clobber")])
+
+(define_insn_and_split "*maddqi4.const"
+  [(set (match_operand:QI 0 "register_operand"                   "=r")
+        (plus:QI (mult:QI (match_operand:QI 1 "register_operand"  "r")
+                          (match_operand:QI 2 "const_int_operand" "n"))
+                 (match_operand:QI 3 "register_operand"           "0")))
+   (clobber (match_scratch:QI 4                                 "=&d"))]
+  "AVR_HAVE_MUL"
+  "#"
+  "&& reload_completed"
+  [(set (match_dup 4)
+        (match_dup 2))
+   ; *maddqi4
+   (set (match_dup 0)
+        (plus:QI (mult:QI (match_dup 1)
+                          (match_dup 4))
+                 (match_dup 3)))]
+  "")
+
+(define_insn_and_split "*msubqi4.const"
+  [(set (match_operand:QI 0 "register_operand"                    "=r")
+        (minus:QI (match_operand:QI 3 "register_operand"           "0")
+                  (mult:QI (match_operand:QI 1 "register_operand"  "r")
+                           (match_operand:QI 2 "const_int_operand" "n"))))
+   (clobber (match_scratch:QI 4                                  "=&d"))]
+  "AVR_HAVE_MUL"
+  "#"
+  "&& reload_completed"
+  [(set (match_dup 4)
+        (match_dup 2))
+   ; *msubqi4
+   (set (match_dup 0)
+        (minus:QI (match_dup 3)
+                  (mult:QI (match_dup 1)
+                           (match_dup 4))))]
+  "")
+
+
+;******************************************************************************
 ; multiply-add/sub HI: $0 = $3 +/- $1*$2  with 8-bit values $1, $2
 ;******************************************************************************
 
@@ -1497,6 +1563,17 @@
 ;; expand decides to use ASHIFT instead of MUL because ASHIFT costs are cheaper
 ;; at that time.  Fix that.
 
+(define_insn "*ashiftqihi2.signx.1"
+  [(set (match_operand:HI 0 "register_operand"                           "=r,*r")
+        (ashift:HI (sign_extend:HI (match_operand:QI 1 "register_operand" "0,r"))
+                   (const_int 1)))]
+  ""
+  "@
+	lsl %A0\;sbc %B0,%B0
+	mov %A0,%1\;lsl %A0\;sbc %B0,%B0"
+  [(set_attr "length" "2,3")
+   (set_attr "cc" "clobber")])
+
 (define_insn_and_split "*ashifthi3.signx.const"
   [(set (match_operand:HI 0 "register_operand"                           "=r")
         (ashift:HI (sign_extend:HI (match_operand:QI 1 "register_operand" "d"))