From ac2b0710c88240b27999b19e2bd56d86c4651e46 Mon Sep 17 00:00:00 2001
From: George Koehler <xkernigh@netscape.net>
Date: Tue, 17 Oct 2017 17:53:03 -0400
Subject: [PATCH] Add more rules for single-precision reg_float.

The result of single-precision fadds, fsubs, and such can go into a
register variable, like we already do with double precision.  This
avoids an extra fmr from a temporary register to the regvar.
---
 mach/powerpc/ncg/table | 31 +++++++++++++++++++++++--------
 1 file changed, 23 insertions(+), 8 deletions(-)

diff --git a/mach/powerpc/ncg/table b/mach/powerpc/ncg/table
index b6a7bfaa8..daf52b385 100644
--- a/mach/powerpc/ncg/table
+++ b/mach/powerpc/ncg/table
@@ -260,21 +260,21 @@ INSTRUCTIONS
   extsb           GPR:wo, GPR:ro.
   extsh           GPR:wo, GPR:ro.
   fadd            FREG+DLOCAL:wo, FREG:ro, FREG:ro cost(4, 5).
-  fadds           FSREG:wo, FSREG:ro, FSREG:ro cost(4, 5).
+  fadds           FSREG+LOCAL:wo, FSREG:ro, FSREG:ro cost(4, 5).
   fcmpo           CR:wo, FREG:ro, FREG:ro cost(4, 5).
   fcmpo           CR:wo, FSREG:ro, FSREG:ro cost(4, 5).
   fctiwz          FREG:wo, FREG:ro.
   fdiv            FREG+DLOCAL:wo, FREG:ro, FREG:ro cost(4, 35).
-  fdivs           FSREG:wo, FSREG:ro, FSREG:ro cost(4, 21).
+  fdivs           FSREG+LOCAL:wo, FSREG:ro, FSREG:ro cost(4, 21).
   fmr             FPR:wo, FPR:ro cost(4, 5).
   fmr             FSREG:wo, FSREG:ro cost(4, 5).
   fmul            FREG+DLOCAL:wo, FREG:ro, FREG:ro cost(4, 5).
-  fmuls           FSREG:wo, FSREG:ro, FSREG:ro cost(4, 5).
+  fmuls           FSREG+LOCAL:wo, FSREG:ro, FSREG:ro cost(4, 5).
   fneg            FREG+DLOCAL:wo, FREG:ro cost(4, 5).
-  fneg            FSREG:wo, FSREG:ro cost(4, 5).
+  fneg            FSREG+LOCAL:wo, FSREG:ro cost(4, 5).
   frsp            FSREG:wo, FREG:ro cost(4, 5).
   fsub            FREG+DLOCAL:wo, FREG:ro, FREG:ro cost(4, 5).
-  fsubs           FSREG:wo, FSREG:ro, FSREG:ro cost(4, 5).
+  fsubs           FSREG+LOCAL:wo, FSREG:ro, FSREG:ro cost(4, 5).
   lbz             GPR:wo, IND_RC_B+IND_RL_B:ro cost(4, 3).
   lbzx            GPR:wo, GPR:ro, GPR:ro cost(4, 3).
   lfd             FPR+DLOCAL:wo, IND_RC_D+IND_RL_D:ro cost(4, 5).
@@ -2082,26 +2082,35 @@ PATTERNS
 		leaving
 			loe ".fs_00000000"
 
-	pat adf $1==INT32                  /* Add single */
+	pat adf $1==4                      /* Add single */
 		with FSREG FSREG
 			uses reusing %1, FSREG
 			gen
 				fadds %a, %2, %1
 			yields %a
+	pat adf stl $1==4 && inreg($2)==reg_float
+		with FSREG FSREG
+			gen fadds {LOCAL, $2}, %2, %1
 
-	pat sbf $1==INT32                  /* Subtract single */
+	pat sbf $1==4                      /* Subtract single */
 		with FSREG FSREG
 			uses reusing %1, FSREG
 			gen
 				fsubs %a, %2, %1
 			yields %a
+	pat sbf stl $1==4 && inreg($2)==reg_float
+		with FSREG FSREG
+			gen fsubs {LOCAL, $2}, %2, %1
 
-	pat mlf $1==INT32                  /* Multiply single */
+	pat mlf $1==4                      /* Multiply single */
 		with FSREG FSREG
 			uses reusing %1, FSREG
 			gen
 				fmuls %a, %2, %1
 			yields %a
+	pat mlf stl $1==4 && inreg($2)==reg_float
+		with FSREG FSREG
+			gen fmuls {LOCAL, $2}, %2, %1
 
 	pat dvf $1==INT32                  /* Divide single */
 		with FSREG FSREG
@@ -2109,6 +2118,9 @@ PATTERNS
 			gen
 				fdivs %a, %2, %1
 			yields %a
+	pat dvf stl $1==4 && inreg($2)==reg_float
+		with FSREG FSREG
+			gen fdivs {LOCAL, $2}, %2, %1
 
 	pat ngf $1==INT32                  /* Negate single */
 		with FSREG
@@ -2116,6 +2128,9 @@ PATTERNS
 			gen
 				fneg %a, %1
 			yields %a
+	pat ngf stl $1==4 && inreg($2)==reg_float
+		with FSREG
+			gen fneg {LOCAL, $2}, %1
 
 	pat cmf $1==INT32                  /* Compare single */
 		with FSREG FSREG
-- 
2.34.1