[dev.link] Merge branch 'master' into dev.link

2 conflicts, that make sense.
	src/cmd/internal/obj/objfile.go
	src/cmd/link/internal/loader/loader.go

Change-Id: Ib224e2d248cb568fa1e888af79dd908b2f5e05ff

1 files changed, 248 insertions, 248 deletions

diff --git a/src/cmd/compile/internal/ssa/gen/ARM64.rules b/src/cmd/compile/internal/ssa/gen/ARM64.rules
index 311067e87a..c4a3532632 100644
--- a/src/cmd/compile/internal/ssa/gen/ARM64.rules
+++ b/src/cmd/compile/internal/ssa/gen/ARM64.rules
@@ -1116,280 +1116,280 @@
 
 // if a register move has only 1 use, just use the same register without emitting instruction
 // MOVDnop doesn't emit instruction, only for ensuring the type.
-(MOVDreg x) && x.Uses == 1 -> (MOVDnop x)
+(MOVDreg x) && x.Uses == 1 => (MOVDnop x)
 
 // fold constant into arithmatic ops
-(ADD x (MOVDconst [c])) -> (ADDconst [c] x)
-(SUB x (MOVDconst [c])) -> (SUBconst [c] x)
-(AND x (MOVDconst [c])) -> (ANDconst [c] x)
-(OR  x (MOVDconst [c])) -> (ORconst  [c] x)
-(XOR x (MOVDconst [c])) -> (XORconst [c] x)
-(TST x (MOVDconst [c])) -> (TSTconst [c] x)
-(TSTW x (MOVDconst [c])) -> (TSTWconst [c] x)
-(CMN x (MOVDconst [c])) -> (CMNconst [c] x)
-(CMNW x (MOVDconst [c])) -> (CMNWconst [c] x)
-(BIC x (MOVDconst [c])) -> (ANDconst [^c] x)
-(EON x (MOVDconst [c])) -> (XORconst [^c] x)
-(ORN x (MOVDconst [c])) -> (ORconst  [^c] x)
-
-(SLL x (MOVDconst [c])) -> (SLLconst x [c&63]) // Note: I don't think we ever generate bad constant shifts (i.e. c>=64)
-(SRL x (MOVDconst [c])) -> (SRLconst x [c&63])
-(SRA x (MOVDconst [c])) -> (SRAconst x [c&63])
-
-(CMP x (MOVDconst [c])) -> (CMPconst [c] x)
-(CMP (MOVDconst [c]) x) -> (InvertFlags (CMPconst [c] x))
+(ADD x (MOVDconst [c])) => (ADDconst [c] x)
+(SUB x (MOVDconst [c])) => (SUBconst [c] x)
+(AND x (MOVDconst [c])) => (ANDconst [c] x)
+(OR  x (MOVDconst [c])) => (ORconst  [c] x)
+(XOR x (MOVDconst [c])) => (XORconst [c] x)
+(TST x (MOVDconst [c])) => (TSTconst [c] x)
+(TSTW x (MOVDconst [c])) => (TSTWconst [int32(c)] x)
+(CMN x (MOVDconst [c])) => (CMNconst [c] x)
+(CMNW x (MOVDconst [c])) => (CMNWconst [int32(c)] x)
+(BIC x (MOVDconst [c])) => (ANDconst [^c] x)
+(EON x (MOVDconst [c])) => (XORconst [^c] x)
+(ORN x (MOVDconst [c])) => (ORconst  [^c] x)
+
+(SLL x (MOVDconst [c])) => (SLLconst x [c&63]) // Note: I don't think we ever generate bad constant shifts (i.e. c>=64)
+(SRL x (MOVDconst [c])) => (SRLconst x [c&63])
+(SRA x (MOVDconst [c])) => (SRAconst x [c&63])
+
+(CMP x (MOVDconst [c])) => (CMPconst [c] x)
+(CMP (MOVDconst [c]) x) => (InvertFlags (CMPconst [c] x))
 (CMPW x (MOVDconst [c])) => (CMPWconst [int32(c)] x)
 (CMPW (MOVDconst [c]) x) => (InvertFlags (CMPWconst [int32(c)] x))
 
 // Canonicalize the order of arguments to comparisons - helps with CSE.
-((CMP|CMPW) x y) && x.ID > y.ID -> (InvertFlags ((CMP|CMPW) y x))
+((CMP|CMPW) x y) && x.ID > y.ID => (InvertFlags ((CMP|CMPW) y x))
 
-// mul-neg -> mneg
-(NEG (MUL x y)) -> (MNEG x y)
-(NEG (MULW x y)) -> (MNEGW x y)
-(MUL (NEG x) y) -> (MNEG x y)
-(MULW (NEG x) y) -> (MNEGW x y)
+// mul-neg => mneg
+(NEG (MUL x y)) => (MNEG x y)
+(NEG (MULW x y)) => (MNEGW x y)
+(MUL (NEG x) y) => (MNEG x y)
+(MULW (NEG x) y) => (MNEGW x y)
 
 // madd/msub
-(ADD a l:(MUL  x y)) && l.Uses==1 && clobber(l) -> (MADD a x y)
-(SUB a l:(MUL  x y)) && l.Uses==1 && clobber(l) -> (MSUB a x y)
-(ADD a l:(MNEG x y)) && l.Uses==1 && clobber(l) -> (MSUB a x y)
-(SUB a l:(MNEG x y)) && l.Uses==1 && clobber(l) -> (MADD a x y)
+(ADD a l:(MUL  x y)) && l.Uses==1 && clobber(l) => (MADD a x y)
+(SUB a l:(MUL  x y)) && l.Uses==1 && clobber(l) => (MSUB a x y)
+(ADD a l:(MNEG x y)) && l.Uses==1 && clobber(l) => (MSUB a x y)
+(SUB a l:(MNEG x y)) && l.Uses==1 && clobber(l) => (MADD a x y)
 
-(ADD a l:(MULW  x y)) && a.Type.Size() != 8 && l.Uses==1 && clobber(l) -> (MADDW a x y)
-(SUB a l:(MULW  x y)) && a.Type.Size() != 8 && l.Uses==1 && clobber(l) -> (MSUBW a x y)
-(ADD a l:(MNEGW x y)) && a.Type.Size() != 8 && l.Uses==1 && clobber(l) -> (MSUBW a x y)
-(SUB a l:(MNEGW x y)) && a.Type.Size() != 8 && l.Uses==1 && clobber(l) -> (MADDW a x y)
+(ADD a l:(MULW  x y)) && a.Type.Size() != 8 && l.Uses==1 && clobber(l) => (MADDW a x y)
+(SUB a l:(MULW  x y)) && a.Type.Size() != 8 && l.Uses==1 && clobber(l) => (MSUBW a x y)
+(ADD a l:(MNEGW x y)) && a.Type.Size() != 8 && l.Uses==1 && clobber(l) => (MSUBW a x y)
+(SUB a l:(MNEGW x y)) && a.Type.Size() != 8 && l.Uses==1 && clobber(l) => (MADDW a x y)
 
 // optimize ADCSflags, SBCSflags and friends
-(ADCSflags x y (Select1 <types.TypeFlags> (ADDSconstflags [-1] (ADCzerocarry <typ.UInt64> c)))) -> (ADCSflags x y c)
-(ADCSflags x y (Select1 <types.TypeFlags> (ADDSconstflags [-1] (MOVDconst [0])))) -> (ADDSflags x y)
-(SBCSflags x y (Select1 <types.TypeFlags> (NEGSflags (NEG <typ.UInt64> (NGCzerocarry <typ.UInt64> bo))))) -> (SBCSflags x y bo)
-(SBCSflags x y (Select1 <types.TypeFlags> (NEGSflags (MOVDconst [0])))) -> (SUBSflags x y)
+(ADCSflags x y (Select1 <types.TypeFlags> (ADDSconstflags [-1] (ADCzerocarry <typ.UInt64> c)))) => (ADCSflags x y c)
+(ADCSflags x y (Select1 <types.TypeFlags> (ADDSconstflags [-1] (MOVDconst [0])))) => (ADDSflags x y)
+(SBCSflags x y (Select1 <types.TypeFlags> (NEGSflags (NEG <typ.UInt64> (NGCzerocarry <typ.UInt64> bo))))) => (SBCSflags x y bo)
+(SBCSflags x y (Select1 <types.TypeFlags> (NEGSflags (MOVDconst [0])))) => (SUBSflags x y)
 
 // mul by constant
-(MUL x (MOVDconst [-1])) -> (NEG x)
-(MUL _ (MOVDconst [0])) -> (MOVDconst [0])
-(MUL x (MOVDconst [1])) -> x
-(MUL x (MOVDconst [c])) && isPowerOfTwo(c) -> (SLLconst [log2(c)] x)
-(MUL x (MOVDconst [c])) && isPowerOfTwo(c-1) && c >= 3 -> (ADDshiftLL x x [log2(c-1)])
-(MUL x (MOVDconst [c])) && isPowerOfTwo(c+1) && c >= 7 -> (ADDshiftLL (NEG <x.Type> x) x [log2(c+1)])
-(MUL x (MOVDconst [c])) && c%3 == 0 && isPowerOfTwo(c/3) -> (SLLconst [log2(c/3)] (ADDshiftLL <x.Type> x x [1]))
-(MUL x (MOVDconst [c])) && c%5 == 0 && isPowerOfTwo(c/5) -> (SLLconst [log2(c/5)] (ADDshiftLL <x.Type> x x [2]))
-(MUL x (MOVDconst [c])) && c%7 == 0 && isPowerOfTwo(c/7) -> (SLLconst [log2(c/7)] (ADDshiftLL <x.Type> (NEG <x.Type> x) x [3]))
-(MUL x (MOVDconst [c])) && c%9 == 0 && isPowerOfTwo(c/9) -> (SLLconst [log2(c/9)] (ADDshiftLL <x.Type> x x [3]))
-
-(MULW x (MOVDconst [c])) && int32(c)==-1 -> (NEG x)
-(MULW _ (MOVDconst [c])) && int32(c)==0 -> (MOVDconst [0])
-(MULW x (MOVDconst [c])) && int32(c)==1 -> x
-(MULW x (MOVDconst [c])) && isPowerOfTwo(c) -> (SLLconst [log2(c)] x)
-(MULW x (MOVDconst [c])) && isPowerOfTwo(c-1) && int32(c) >= 3 -> (ADDshiftLL x x [log2(c-1)])
-(MULW x (MOVDconst [c])) && isPowerOfTwo(c+1) && int32(c) >= 7 -> (ADDshiftLL (NEG <x.Type> x) x [log2(c+1)])
-(MULW x (MOVDconst [c])) && c%3 == 0 && isPowerOfTwo(c/3) && is32Bit(c) -> (SLLconst [log2(c/3)] (ADDshiftLL <x.Type> x x [1]))
-(MULW x (MOVDconst [c])) && c%5 == 0 && isPowerOfTwo(c/5) && is32Bit(c) -> (SLLconst [log2(c/5)] (ADDshiftLL <x.Type> x x [2]))
-(MULW x (MOVDconst [c])) && c%7 == 0 && isPowerOfTwo(c/7) && is32Bit(c) -> (SLLconst [log2(c/7)] (ADDshiftLL <x.Type> (NEG <x.Type> x) x [3]))
-(MULW x (MOVDconst [c])) && c%9 == 0 && isPowerOfTwo(c/9) && is32Bit(c) -> (SLLconst [log2(c/9)] (ADDshiftLL <x.Type> x x [3]))
+(MUL x (MOVDconst [-1])) => (NEG x)
+(MUL _ (MOVDconst [0])) => (MOVDconst [0])
+(MUL x (MOVDconst [1])) => x
+(MUL x (MOVDconst [c])) && isPowerOfTwo(c) => (SLLconst [log2(c)] x)
+(MUL x (MOVDconst [c])) && isPowerOfTwo(c-1) && c >= 3 => (ADDshiftLL x x [log2(c-1)])
+(MUL x (MOVDconst [c])) && isPowerOfTwo(c+1) && c >= 7 => (ADDshiftLL (NEG <x.Type> x) x [log2(c+1)])
+(MUL x (MOVDconst [c])) && c%3 == 0 && isPowerOfTwo(c/3) => (SLLconst [log2(c/3)] (ADDshiftLL <x.Type> x x [1]))
+(MUL x (MOVDconst [c])) && c%5 == 0 && isPowerOfTwo(c/5) => (SLLconst [log2(c/5)] (ADDshiftLL <x.Type> x x [2]))
+(MUL x (MOVDconst [c])) && c%7 == 0 && isPowerOfTwo(c/7) => (SLLconst [log2(c/7)] (ADDshiftLL <x.Type> (NEG <x.Type> x) x [3]))
+(MUL x (MOVDconst [c])) && c%9 == 0 && isPowerOfTwo(c/9) => (SLLconst [log2(c/9)] (ADDshiftLL <x.Type> x x [3]))
+
+(MULW x (MOVDconst [c])) && int32(c)==-1 => (NEG x)
+(MULW _ (MOVDconst [c])) && int32(c)==0 => (MOVDconst [0])
+(MULW x (MOVDconst [c])) && int32(c)==1 => x
+(MULW x (MOVDconst [c])) && isPowerOfTwo(c) => (SLLconst [log2(c)] x)
+(MULW x (MOVDconst [c])) && isPowerOfTwo(c-1) && int32(c) >= 3 => (ADDshiftLL x x [log2(c-1)])
+(MULW x (MOVDconst [c])) && isPowerOfTwo(c+1) && int32(c) >= 7 => (ADDshiftLL (NEG <x.Type> x) x [log2(c+1)])
+(MULW x (MOVDconst [c])) && c%3 == 0 && isPowerOfTwo(c/3) && is32Bit(c) => (SLLconst [log2(c/3)] (ADDshiftLL <x.Type> x x [1]))
+(MULW x (MOVDconst [c])) && c%5 == 0 && isPowerOfTwo(c/5) && is32Bit(c) => (SLLconst [log2(c/5)] (ADDshiftLL <x.Type> x x [2]))
+(MULW x (MOVDconst [c])) && c%7 == 0 && isPowerOfTwo(c/7) && is32Bit(c) => (SLLconst [log2(c/7)] (ADDshiftLL <x.Type> (NEG <x.Type> x) x [3]))
+(MULW x (MOVDconst [c])) && c%9 == 0 && isPowerOfTwo(c/9) && is32Bit(c) => (SLLconst [log2(c/9)] (ADDshiftLL <x.Type> x x [3]))
 
 // mneg by constant
-(MNEG x (MOVDconst [-1])) -> x
-(MNEG _ (MOVDconst [0])) -> (MOVDconst [0])
-(MNEG x (MOVDconst [1])) -> (NEG x)
-(MNEG x (MOVDconst [c])) && isPowerOfTwo(c) -> (NEG (SLLconst <x.Type> [log2(c)] x))
-(MNEG x (MOVDconst [c])) && isPowerOfTwo(c-1) && c >= 3 -> (NEG (ADDshiftLL <x.Type> x x [log2(c-1)]))
-(MNEG x (MOVDconst [c])) && isPowerOfTwo(c+1) && c >= 7 -> (NEG (ADDshiftLL <x.Type> (NEG <x.Type> x) x [log2(c+1)]))
-(MNEG x (MOVDconst [c])) && c%3 == 0 && isPowerOfTwo(c/3) -> (SLLconst <x.Type> [log2(c/3)] (SUBshiftLL <x.Type> x x [2]))
-(MNEG x (MOVDconst [c])) && c%5 == 0 && isPowerOfTwo(c/5) -> (NEG (SLLconst <x.Type> [log2(c/5)] (ADDshiftLL <x.Type> x x [2])))
-(MNEG x (MOVDconst [c])) && c%7 == 0 && isPowerOfTwo(c/7) -> (SLLconst <x.Type> [log2(c/7)] (SUBshiftLL <x.Type> x x [3]))
-(MNEG x (MOVDconst [c])) && c%9 == 0 && isPowerOfTwo(c/9) -> (NEG (SLLconst <x.Type> [log2(c/9)] (ADDshiftLL <x.Type> x x [3])))
-
-(MNEGW x (MOVDconst [c])) && int32(c)==-1 -> x
-(MNEGW _ (MOVDconst [c])) && int32(c)==0 -> (MOVDconst [0])
-(MNEGW x (MOVDconst [c])) && int32(c)==1 -> (NEG x)
-(MNEGW x (MOVDconst [c])) && isPowerOfTwo(c) -> (NEG (SLLconst <x.Type> [log2(c)] x))
-(MNEGW x (MOVDconst [c])) && isPowerOfTwo(c-1) && int32(c) >= 3 -> (NEG (ADDshiftLL <x.Type> x x [log2(c-1)]))
-(MNEGW x (MOVDconst [c])) && isPowerOfTwo(c+1) && int32(c) >= 7 -> (NEG (ADDshiftLL <x.Type> (NEG <x.Type> x) x [log2(c+1)]))
-(MNEGW x (MOVDconst [c])) && c%3 == 0 && isPowerOfTwo(c/3) && is32Bit(c) -> (SLLconst <x.Type> [log2(c/3)] (SUBshiftLL <x.Type> x x [2]))
-(MNEGW x (MOVDconst [c])) && c%5 == 0 && isPowerOfTwo(c/5) && is32Bit(c) -> (NEG (SLLconst <x.Type> [log2(c/5)] (ADDshiftLL <x.Type> x x [2])))
-(MNEGW x (MOVDconst [c])) && c%7 == 0 && isPowerOfTwo(c/7) && is32Bit(c) -> (SLLconst <x.Type> [log2(c/7)] (SUBshiftLL <x.Type> x x [3]))
-(MNEGW x (MOVDconst [c])) && c%9 == 0 && isPowerOfTwo(c/9) && is32Bit(c) -> (NEG (SLLconst <x.Type> [log2(c/9)] (ADDshiftLL <x.Type> x x [3])))
-
-(MADD a x (MOVDconst [-1])) -> (SUB a x)
-(MADD a _ (MOVDconst [0])) -> a
-(MADD a x (MOVDconst [1])) -> (ADD a x)
-(MADD a x (MOVDconst [c])) && isPowerOfTwo(c) -> (ADDshiftLL a x [log2(c)])
-(MADD a x (MOVDconst [c])) && isPowerOfTwo(c-1) && c>=3 -> (ADD a (ADDshiftLL <x.Type> x x [log2(c-1)]))
-(MADD a x (MOVDconst [c])) && isPowerOfTwo(c+1) && c>=7 -> (SUB a (SUBshiftLL <x.Type> x x [log2(c+1)]))
-(MADD a x (MOVDconst [c])) && c%3 == 0 && isPowerOfTwo(c/3) -> (SUBshiftLL a (SUBshiftLL <x.Type> x x [2]) [log2(c/3)])
-(MADD a x (MOVDconst [c])) && c%5 == 0 && isPowerOfTwo(c/5) -> (ADDshiftLL a (ADDshiftLL <x.Type> x x [2]) [log2(c/5)])
-(MADD a x (MOVDconst [c])) && c%7 == 0 && isPowerOfTwo(c/7) -> (SUBshiftLL a (SUBshiftLL <x.Type> x x [3]) [log2(c/7)])
-(MADD a x (MOVDconst [c])) && c%9 == 0 && isPowerOfTwo(c/9) -> (ADDshiftLL a (ADDshiftLL <x.Type> x x [3]) [log2(c/9)])
-
-(MADD a (MOVDconst [-1]) x) -> (SUB a x)
-(MADD a (MOVDconst [0]) _) -> a
-(MADD a (MOVDconst [1]) x) -> (ADD a x)
-(MADD a (MOVDconst [c]) x) && isPowerOfTwo(c) -> (ADDshiftLL a x [log2(c)])
-(MADD a (MOVDconst [c]) x) && isPowerOfTwo(c-1) && c>=3 -> (ADD a (ADDshiftLL <x.Type> x x [log2(c-1)]))
-(MADD a (MOVDconst [c]) x) && isPowerOfTwo(c+1) && c>=7 -> (SUB a (SUBshiftLL <x.Type> x x [log2(c+1)]))
-(MADD a (MOVDconst [c]) x) && c%3 == 0 && isPowerOfTwo(c/3) -> (SUBshiftLL a (SUBshiftLL <x.Type> x x [2]) [log2(c/3)])
-(MADD a (MOVDconst [c]) x) && c%5 == 0 && isPowerOfTwo(c/5) -> (ADDshiftLL a (ADDshiftLL <x.Type> x x [2]) [log2(c/5)])
-(MADD a (MOVDconst [c]) x) && c%7 == 0 && isPowerOfTwo(c/7) -> (SUBshiftLL a (SUBshiftLL <x.Type> x x [3]) [log2(c/7)])
-(MADD a (MOVDconst [c]) x) && c%9 == 0 && isPowerOfTwo(c/9) -> (ADDshiftLL a (ADDshiftLL <x.Type> x x [3]) [log2(c/9)])
-
-(MADDW a x (MOVDconst [c])) && int32(c)==-1 -> (SUB a x)
-(MADDW a _ (MOVDconst [c])) && int32(c)==0 -> a
-(MADDW a x (MOVDconst [c])) && int32(c)==1 -> (ADD a x)
-(MADDW a x (MOVDconst [c])) && isPowerOfTwo(c) -> (ADDshiftLL a x [log2(c)])
-(MADDW a x (MOVDconst [c])) && isPowerOfTwo(c-1) && int32(c)>=3 -> (ADD a (ADDshiftLL <x.Type> x x [log2(c-1)]))
-(MADDW a x (MOVDconst [c])) && isPowerOfTwo(c+1) && int32(c)>=7 -> (SUB a (SUBshiftLL <x.Type> x x [log2(c+1)]))
-(MADDW a x (MOVDconst [c])) && c%3 == 0 && isPowerOfTwo(c/3) && is32Bit(c) -> (SUBshiftLL a (SUBshiftLL <x.Type> x x [2]) [log2(c/3)])
-(MADDW a x (MOVDconst [c])) && c%5 == 0 && isPowerOfTwo(c/5) && is32Bit(c) -> (ADDshiftLL a (ADDshiftLL <x.Type> x x [2]) [log2(c/5)])
-(MADDW a x (MOVDconst [c])) && c%7 == 0 && isPowerOfTwo(c/7) && is32Bit(c) -> (SUBshiftLL a (SUBshiftLL <x.Type> x x [3]) [log2(c/7)])
-(MADDW a x (MOVDconst [c])) && c%9 == 0 && isPowerOfTwo(c/9) && is32Bit(c) -> (ADDshiftLL a (ADDshiftLL <x.Type> x x [3]) [log2(c/9)])
-
-(MADDW a (MOVDconst [c]) x) && int32(c)==-1 -> (SUB a x)
-(MADDW a (MOVDconst [c]) _) && int32(c)==0 -> a
-(MADDW a (MOVDconst [c]) x) && int32(c)==1 -> (ADD a x)
-(MADDW a (MOVDconst [c]) x) && isPowerOfTwo(c) -> (ADDshiftLL a x [log2(c)])
-(MADDW a (MOVDconst [c]) x) && isPowerOfTwo(c-1) && int32(c)>=3 -> (ADD a (ADDshiftLL <x.Type> x x [log2(c-1)]))
-(MADDW a (MOVDconst [c]) x) && isPowerOfTwo(c+1) && int32(c)>=7 -> (SUB a (SUBshiftLL <x.Type> x x [log2(c+1)]))
-(MADDW a (MOVDconst [c]) x) && c%3 == 0 && isPowerOfTwo(c/3) && is32Bit(c) -> (SUBshiftLL a (SUBshiftLL <x.Type> x x [2]) [log2(c/3)])
-(MADDW a (MOVDconst [c]) x) && c%5 == 0 && isPowerOfTwo(c/5) && is32Bit(c) -> (ADDshiftLL a (ADDshiftLL <x.Type> x x [2]) [log2(c/5)])
-(MADDW a (MOVDconst [c]) x) && c%7 == 0 && isPowerOfTwo(c/7) && is32Bit(c) -> (SUBshiftLL a (SUBshiftLL <x.Type> x x [3]) [log2(c/7)])
-(MADDW a (MOVDconst [c]) x) && c%9 == 0 && isPowerOfTwo(c/9) && is32Bit(c) -> (ADDshiftLL a (ADDshiftLL <x.Type> x x [3]) [log2(c/9)])
-
-(MSUB a x (MOVDconst [-1])) -> (ADD a x)
-(MSUB a _ (MOVDconst [0])) -> a
-(MSUB a x (MOVDconst [1])) -> (SUB a x)
-(MSUB a x (MOVDconst [c])) && isPowerOfTwo(c) -> (SUBshiftLL a x [log2(c)])
-(MSUB a x (MOVDconst [c])) && isPowerOfTwo(c-1) && c>=3 -> (SUB a (ADDshiftLL <x.Type> x x [log2(c-1)]))
-(MSUB a x (MOVDconst [c])) && isPowerOfTwo(c+1) && c>=7 -> (ADD a (SUBshiftLL <x.Type> x x [log2(c+1)]))
-(MSUB a x (MOVDconst [c])) && c%3 == 0 && isPowerOfTwo(c/3) -> (ADDshiftLL a (SUBshiftLL <x.Type> x x [2]) [log2(c/3)])
-(MSUB a x (MOVDconst [c])) && c%5 == 0 && isPowerOfTwo(c/5) -> (SUBshiftLL a (ADDshiftLL <x.Type> x x [2]) [log2(c/5)])
-(MSUB a x (MOVDconst [c])) && c%7 == 0 && isPowerOfTwo(c/7) -> (ADDshiftLL a (SUBshiftLL <x.Type> x x [3]) [log2(c/7)])
-(MSUB a x (MOVDconst [c])) && c%9 == 0 && isPowerOfTwo(c/9) -> (SUBshiftLL a (ADDshiftLL <x.Type> x x [3]) [log2(c/9)])
-
-(MSUB a (MOVDconst [-1]) x) -> (ADD a x)
-(MSUB a (MOVDconst [0]) _) -> a
-(MSUB a (MOVDconst [1]) x) -> (SUB a x)
-(MSUB a (MOVDconst [c]) x) && isPowerOfTwo(c) -> (SUBshiftLL a x [log2(c)])
-(MSUB a (MOVDconst [c]) x) && isPowerOfTwo(c-1) && c>=3 -> (SUB a (ADDshiftLL <x.Type> x x [log2(c-1)]))
-(MSUB a (MOVDconst [c]) x) && isPowerOfTwo(c+1) && c>=7 -> (ADD a (SUBshiftLL <x.Type> x x [log2(c+1)]))
-(MSUB a (MOVDconst [c]) x) && c%3 == 0 && isPowerOfTwo(c/3) -> (ADDshiftLL a (SUBshiftLL <x.Type> x x [2]) [log2(c/3)])
-(MSUB a (MOVDconst [c]) x) && c%5 == 0 && isPowerOfTwo(c/5) -> (SUBshiftLL a (ADDshiftLL <x.Type> x x [2]) [log2(c/5)])
-(MSUB a (MOVDconst [c]) x) && c%7 == 0 && isPowerOfTwo(c/7) -> (ADDshiftLL a (SUBshiftLL <x.Type> x x [3]) [log2(c/7)])
-(MSUB a (MOVDconst [c]) x) && c%9 == 0 && isPowerOfTwo(c/9) -> (SUBshiftLL a (ADDshiftLL <x.Type> x x [3]) [log2(c/9)])
-
-(MSUBW a x (MOVDconst [c])) && int32(c)==-1 -> (ADD a x)
-(MSUBW a _ (MOVDconst [c])) && int32(c)==0 -> a
-(MSUBW a x (MOVDconst [c])) && int32(c)==1 -> (SUB a x)
-(MSUBW a x (MOVDconst [c])) && isPowerOfTwo(c) -> (SUBshiftLL a x [log2(c)])
-(MSUBW a x (MOVDconst [c])) && isPowerOfTwo(c-1) && int32(c)>=3 -> (SUB a (ADDshiftLL <x.Type> x x [log2(c-1)]))
-(MSUBW a x (MOVDconst [c])) && isPowerOfTwo(c+1) && int32(c)>=7 -> (ADD a (SUBshiftLL <x.Type> x x [log2(c+1)]))
-(MSUBW a x (MOVDconst [c])) && c%3 == 0 && isPowerOfTwo(c/3) && is32Bit(c) -> (ADDshiftLL a (SUBshiftLL <x.Type> x x [2]) [log2(c/3)])
-(MSUBW a x (MOVDconst [c])) && c%5 == 0 && isPowerOfTwo(c/5) && is32Bit(c) -> (SUBshiftLL a (ADDshiftLL <x.Type> x x [2]) [log2(c/5)])
-(MSUBW a x (MOVDconst [c])) && c%7 == 0 && isPowerOfTwo(c/7) && is32Bit(c) -> (ADDshiftLL a (SUBshiftLL <x.Type> x x [3]) [log2(c/7)])
-(MSUBW a x (MOVDconst [c])) && c%9 == 0 && isPowerOfTwo(c/9) && is32Bit(c) -> (SUBshiftLL a (ADDshiftLL <x.Type> x x [3]) [log2(c/9)])
-
-(MSUBW a (MOVDconst [c]) x) && int32(c)==-1 -> (ADD a x)
-(MSUBW a (MOVDconst [c]) _) && int32(c)==0 -> a
-(MSUBW a (MOVDconst [c]) x) && int32(c)==1 -> (SUB a x)
-(MSUBW a (MOVDconst [c]) x) && isPowerOfTwo(c) -> (SUBshiftLL a x [log2(c)])
-(MSUBW a (MOVDconst [c]) x) && isPowerOfTwo(c-1) && int32(c)>=3 -> (SUB a (ADDshiftLL <x.Type> x x [log2(c-1)]))
-(MSUBW a (MOVDconst [c]) x) && isPowerOfTwo(c+1) && int32(c)>=7 -> (ADD a (SUBshiftLL <x.Type> x x [log2(c+1)]))
-(MSUBW a (MOVDconst [c]) x) && c%3 == 0 && isPowerOfTwo(c/3) && is32Bit(c) -> (ADDshiftLL a (SUBshiftLL <x.Type> x x [2]) [log2(c/3)])
-(MSUBW a (MOVDconst [c]) x) && c%5 == 0 && isPowerOfTwo(c/5) && is32Bit(c) -> (SUBshiftLL a (ADDshiftLL <x.Type> x x [2]) [log2(c/5)])
-(MSUBW a (MOVDconst [c]) x) && c%7 == 0 && isPowerOfTwo(c/7) && is32Bit(c) -> (ADDshiftLL a (SUBshiftLL <x.Type> x x [3]) [log2(c/7)])
-(MSUBW a (MOVDconst [c]) x) && c%9 == 0 && isPowerOfTwo(c/9) && is32Bit(c) -> (SUBshiftLL a (ADDshiftLL <x.Type> x x [3]) [log2(c/9)])
+(MNEG x (MOVDconst [-1])) => x
+(MNEG _ (MOVDconst [0])) => (MOVDconst [0])
+(MNEG x (MOVDconst [1])) => (NEG x)
+(MNEG x (MOVDconst [c])) && isPowerOfTwo(c) => (NEG (SLLconst <x.Type> [log2(c)] x))
+(MNEG x (MOVDconst [c])) && isPowerOfTwo(c-1) && c >= 3 => (NEG (ADDshiftLL <x.Type> x x [log2(c-1)]))
+(MNEG x (MOVDconst [c])) && isPowerOfTwo(c+1) && c >= 7 => (NEG (ADDshiftLL <x.Type> (NEG <x.Type> x) x [log2(c+1)]))
+(MNEG x (MOVDconst [c])) && c%3 == 0 && isPowerOfTwo(c/3) => (SLLconst <x.Type> [log2(c/3)] (SUBshiftLL <x.Type> x x [2]))
+(MNEG x (MOVDconst [c])) && c%5 == 0 && isPowerOfTwo(c/5) => (NEG (SLLconst <x.Type> [log2(c/5)] (ADDshiftLL <x.Type> x x [2])))
+(MNEG x (MOVDconst [c])) && c%7 == 0 && isPowerOfTwo(c/7) => (SLLconst <x.Type> [log2(c/7)] (SUBshiftLL <x.Type> x x [3]))
+(MNEG x (MOVDconst [c])) && c%9 == 0 && isPowerOfTwo(c/9) => (NEG (SLLconst <x.Type> [log2(c/9)] (ADDshiftLL <x.Type> x x [3])))
+
+(MNEGW x (MOVDconst [c])) && int32(c)==-1 => x
+(MNEGW _ (MOVDconst [c])) && int32(c)==0 => (MOVDconst [0])
+(MNEGW x (MOVDconst [c])) && int32(c)==1 => (NEG x)
+(MNEGW x (MOVDconst [c])) && isPowerOfTwo(c) => (NEG (SLLconst <x.Type> [log2(c)] x))
+(MNEGW x (MOVDconst [c])) && isPowerOfTwo(c-1) && int32(c) >= 3 => (NEG (ADDshiftLL <x.Type> x x [log2(c-1)]))
+(MNEGW x (MOVDconst [c])) && isPowerOfTwo(c+1) && int32(c) >= 7 => (NEG (ADDshiftLL <x.Type> (NEG <x.Type> x) x [log2(c+1)]))
+(MNEGW x (MOVDconst [c])) && c%3 == 0 && isPowerOfTwo(c/3) && is32Bit(c) => (SLLconst <x.Type> [log2(c/3)] (SUBshiftLL <x.Type> x x [2]))
+(MNEGW x (MOVDconst [c])) && c%5 == 0 && isPowerOfTwo(c/5) && is32Bit(c) => (NEG (SLLconst <x.Type> [log2(c/5)] (ADDshiftLL <x.Type> x x [2])))
+(MNEGW x (MOVDconst [c])) && c%7 == 0 && isPowerOfTwo(c/7) && is32Bit(c) => (SLLconst <x.Type> [log2(c/7)] (SUBshiftLL <x.Type> x x [3]))
+(MNEGW x (MOVDconst [c])) && c%9 == 0 && isPowerOfTwo(c/9) && is32Bit(c) => (NEG (SLLconst <x.Type> [log2(c/9)] (ADDshiftLL <x.Type> x x [3])))
+
+(MADD a x (MOVDconst [-1])) => (SUB a x)
+(MADD a _ (MOVDconst [0])) => a
+(MADD a x (MOVDconst [1])) => (ADD a x)
+(MADD a x (MOVDconst [c])) && isPowerOfTwo(c) => (ADDshiftLL a x [log2(c)])
+(MADD a x (MOVDconst [c])) && isPowerOfTwo(c-1) && c>=3 => (ADD a (ADDshiftLL <x.Type> x x [log2(c-1)]))
+(MADD a x (MOVDconst [c])) && isPowerOfTwo(c+1) && c>=7 => (SUB a (SUBshiftLL <x.Type> x x [log2(c+1)]))
+(MADD a x (MOVDconst [c])) && c%3 == 0 && isPowerOfTwo(c/3) => (SUBshiftLL a (SUBshiftLL <x.Type> x x [2]) [log2(c/3)])
+(MADD a x (MOVDconst [c])) && c%5 == 0 && isPowerOfTwo(c/5) => (ADDshiftLL a (ADDshiftLL <x.Type> x x [2]) [log2(c/5)])
+(MADD a x (MOVDconst [c])) && c%7 == 0 && isPowerOfTwo(c/7) => (SUBshiftLL a (SUBshiftLL <x.Type> x x [3]) [log2(c/7)])
+(MADD a x (MOVDconst [c])) && c%9 == 0 && isPowerOfTwo(c/9) => (ADDshiftLL a (ADDshiftLL <x.Type> x x [3]) [log2(c/9)])
+
+(MADD a (MOVDconst [-1]) x) => (SUB a x)
+(MADD a (MOVDconst [0]) _) => a
+(MADD a (MOVDconst [1]) x) => (ADD a x)
+(MADD a (MOVDconst [c]) x) && isPowerOfTwo(c) => (ADDshiftLL a x [log2(c)])
+(MADD a (MOVDconst [c]) x) && isPowerOfTwo(c-1) && c>=3 => (ADD a (ADDshiftLL <x.Type> x x [log2(c-1)]))
+(MADD a (MOVDconst [c]) x) && isPowerOfTwo(c+1) && c>=7 => (SUB a (SUBshiftLL <x.Type> x x [log2(c+1)]))
+(MADD a (MOVDconst [c]) x) && c%3 == 0 && isPowerOfTwo(c/3) => (SUBshiftLL a (SUBshiftLL <x.Type> x x [2]) [log2(c/3)])
+(MADD a (MOVDconst [c]) x) && c%5 == 0 && isPowerOfTwo(c/5) => (ADDshiftLL a (ADDshiftLL <x.Type> x x [2]) [log2(c/5)])
+(MADD a (MOVDconst [c]) x) && c%7 == 0 && isPowerOfTwo(c/7) => (SUBshiftLL a (SUBshiftLL <x.Type> x x [3]) [log2(c/7)])
+(MADD a (MOVDconst [c]) x) && c%9 == 0 && isPowerOfTwo(c/9) => (ADDshiftLL a (ADDshiftLL <x.Type> x x [3]) [log2(c/9)])
+
+(MADDW a x (MOVDconst [c])) && int32(c)==-1 => (SUB a x)
+(MADDW a _ (MOVDconst [c])) && int32(c)==0 => a
+(MADDW a x (MOVDconst [c])) && int32(c)==1 => (ADD a x)
+(MADDW a x (MOVDconst [c])) && isPowerOfTwo(c) => (ADDshiftLL a x [log2(c)])
+(MADDW a x (MOVDconst [c])) && isPowerOfTwo(c-1) && int32(c)>=3 => (ADD a (ADDshiftLL <x.Type> x x [log2(c-1)]))
+(MADDW a x (MOVDconst [c])) && isPowerOfTwo(c+1) && int32(c)>=7 => (SUB a (SUBshiftLL <x.Type> x x [log2(c+1)]))
+(MADDW a x (MOVDconst [c])) && c%3 == 0 && isPowerOfTwo(c/3) && is32Bit(c) => (SUBshiftLL a (SUBshiftLL <x.Type> x x [2]) [log2(c/3)])
+(MADDW a x (MOVDconst [c])) && c%5 == 0 && isPowerOfTwo(c/5) && is32Bit(c) => (ADDshiftLL a (ADDshiftLL <x.Type> x x [2]) [log2(c/5)])
+(MADDW a x (MOVDconst [c])) && c%7 == 0 && isPowerOfTwo(c/7) && is32Bit(c) => (SUBshiftLL a (SUBshiftLL <x.Type> x x [3]) [log2(c/7)])
+(MADDW a x (MOVDconst [c])) && c%9 == 0 && isPowerOfTwo(c/9) && is32Bit(c) => (ADDshiftLL a (ADDshiftLL <x.Type> x x [3]) [log2(c/9)])
+
+(MADDW a (MOVDconst [c]) x) && int32(c)==-1 => (SUB a x)
+(MADDW a (MOVDconst [c]) _) && int32(c)==0 => a
+(MADDW a (MOVDconst [c]) x) && int32(c)==1 => (ADD a x)
+(MADDW a (MOVDconst [c]) x) && isPowerOfTwo(c) => (ADDshiftLL a x [log2(c)])
+(MADDW a (MOVDconst [c]) x) && isPowerOfTwo(c-1) && int32(c)>=3 => (ADD a (ADDshiftLL <x.Type> x x [log2(c-1)]))
+(MADDW a (MOVDconst [c]) x) && isPowerOfTwo(c+1) && int32(c)>=7 => (SUB a (SUBshiftLL <x.Type> x x [log2(c+1)]))
+(MADDW a (MOVDconst [c]) x) && c%3 == 0 && isPowerOfTwo(c/3) && is32Bit(c) => (SUBshiftLL a (SUBshiftLL <x.Type> x x [2]) [log2(c/3)])
+(MADDW a (MOVDconst [c]) x) && c%5 == 0 && isPowerOfTwo(c/5) && is32Bit(c) => (ADDshiftLL a (ADDshiftLL <x.Type> x x [2]) [log2(c/5)])
+(MADDW a (MOVDconst [c]) x) && c%7 == 0 && isPowerOfTwo(c/7) && is32Bit(c) => (SUBshiftLL a (SUBshiftLL <x.Type> x x [3]) [log2(c/7)])
+(MADDW a (MOVDconst [c]) x) && c%9 == 0 && isPowerOfTwo(c/9) && is32Bit(c) => (ADDshiftLL a (ADDshiftLL <x.Type> x x [3]) [log2(c/9)])
+
+(MSUB a x (MOVDconst [-1])) => (ADD a x)
+(MSUB a _ (MOVDconst [0])) => a
+(MSUB a x (MOVDconst [1])) => (SUB a x)
+(MSUB a x (MOVDconst [c])) && isPowerOfTwo(c) => (SUBshiftLL a x [log2(c)])
+(MSUB a x (MOVDconst [c])) && isPowerOfTwo(c-1) && c>=3 => (SUB a (ADDshiftLL <x.Type> x x [log2(c-1)]))
+(MSUB a x (MOVDconst [c])) && isPowerOfTwo(c+1) && c>=7 => (ADD a (SUBshiftLL <x.Type> x x [log2(c+1)]))
+(MSUB a x (MOVDconst [c])) && c%3 == 0 && isPowerOfTwo(c/3) => (ADDshiftLL a (SUBshiftLL <x.Type> x x [2]) [log2(c/3)])
+(MSUB a x (MOVDconst [c])) && c%5 == 0 && isPowerOfTwo(c/5) => (SUBshiftLL a (ADDshiftLL <x.Type> x x [2]) [log2(c/5)])
+(MSUB a x (MOVDconst [c])) && c%7 == 0 && isPowerOfTwo(c/7) => (ADDshiftLL a (SUBshiftLL <x.Type> x x [3]) [log2(c/7)])
+(MSUB a x (MOVDconst [c])) && c%9 == 0 && isPowerOfTwo(c/9) => (SUBshiftLL a (ADDshiftLL <x.Type> x x [3]) [log2(c/9)])
+
+(MSUB a (MOVDconst [-1]) x) => (ADD a x)
+(MSUB a (MOVDconst [0]) _) => a
+(MSUB a (MOVDconst [1]) x) => (SUB a x)
+(MSUB a (MOVDconst [c]) x) && isPowerOfTwo(c) => (SUBshiftLL a x [log2(c)])
+(MSUB a (MOVDconst [c]) x) && isPowerOfTwo(c-1) && c>=3 => (SUB a (ADDshiftLL <x.Type> x x [log2(c-1)]))
+(MSUB a (MOVDconst [c]) x) && isPowerOfTwo(c+1) && c>=7 => (ADD a (SUBshiftLL <x.Type> x x [log2(c+1)]))
+(MSUB a (MOVDconst [c]) x) && c%3 == 0 && isPowerOfTwo(c/3) => (ADDshiftLL a (SUBshiftLL <x.Type> x x [2]) [log2(c/3)])
+(MSUB a (MOVDconst [c]) x) && c%5 == 0 && isPowerOfTwo(c/5) => (SUBshiftLL a (ADDshiftLL <x.Type> x x [2]) [log2(c/5)])
+(MSUB a (MOVDconst [c]) x) && c%7 == 0 && isPowerOfTwo(c/7) => (ADDshiftLL a (SUBshiftLL <x.Type> x x [3]) [log2(c/7)])
+(MSUB a (MOVDconst [c]) x) && c%9 == 0 && isPowerOfTwo(c/9) => (SUBshiftLL a (ADDshiftLL <x.Type> x x [3]) [log2(c/9)])
+
+(MSUBW a x (MOVDconst [c])) && int32(c)==-1 => (ADD a x)
+(MSUBW a _ (MOVDconst [c])) && int32(c)==0 => a
+(MSUBW a x (MOVDconst [c])) && int32(c)==1 => (SUB a x)
+(MSUBW a x (MOVDconst [c])) && isPowerOfTwo(c) => (SUBshiftLL a x [log2(c)])
+(MSUBW a x (MOVDconst [c])) && isPowerOfTwo(c-1) && int32(c)>=3 => (SUB a (ADDshiftLL <x.Type> x x [log2(c-1)]))
+(MSUBW a x (MOVDconst [c])) && isPowerOfTwo(c+1) && int32(c)>=7 => (ADD a (SUBshiftLL <x.Type> x x [log2(c+1)]))
+(MSUBW a x (MOVDconst [c])) && c%3 == 0 && isPowerOfTwo(c/3) && is32Bit(c) => (ADDshiftLL a (SUBshiftLL <x.Type> x x [2]) [log2(c/3)])
+(MSUBW a x (MOVDconst [c])) && c%5 == 0 && isPowerOfTwo(c/5) && is32Bit(c) => (SUBshiftLL a (ADDshiftLL <x.Type> x x [2]) [log2(c/5)])
+(MSUBW a x (MOVDconst [c])) && c%7 == 0 && isPowerOfTwo(c/7) && is32Bit(c) => (ADDshiftLL a (SUBshiftLL <x.Type> x x [3]) [log2(c/7)])
+(MSUBW a x (MOVDconst [c])) && c%9 == 0 && isPowerOfTwo(c/9) && is32Bit(c) => (SUBshiftLL a (ADDshiftLL <x.Type> x x [3]) [log2(c/9)])
+
+(MSUBW a (MOVDconst [c]) x) && int32(c)==-1 => (ADD a x)
+(MSUBW a (MOVDconst [c]) _) && int32(c)==0 => a
+(MSUBW a (MOVDconst [c]) x) && int32(c)==1 => (SUB a x)
+(MSUBW a (MOVDconst [c]) x) && isPowerOfTwo(c) => (SUBshiftLL a x [log2(c)])
+(MSUBW a (MOVDconst [c]) x) && isPowerOfTwo(c-1) && int32(c)>=3 => (SUB a (ADDshiftLL <x.Type> x x [log2(c-1)]))
+(MSUBW a (MOVDconst [c]) x) && isPowerOfTwo(c+1) && int32(c)>=7 => (ADD a (SUBshiftLL <x.Type> x x [log2(c+1)]))
+(MSUBW a (MOVDconst [c]) x) && c%3 == 0 && isPowerOfTwo(c/3) && is32Bit(c) => (ADDshiftLL a (SUBshiftLL <x.Type> x x [2]) [log2(c/3)])
+(MSUBW a (MOVDconst [c]) x) && c%5 == 0 && isPowerOfTwo(c/5) && is32Bit(c) => (SUBshiftLL a (ADDshiftLL <x.Type> x x [2]) [log2(c/5)])
+(MSUBW a (MOVDconst [c]) x) && c%7 == 0 && isPowerOfTwo(c/7) && is32Bit(c) => (ADDshiftLL a (SUBshiftLL <x.Type> x x [3]) [log2(c/7)])
+(MSUBW a (MOVDconst [c]) x) && c%9 == 0 && isPowerOfTwo(c/9) && is32Bit(c) => (SUBshiftLL a (ADDshiftLL <x.Type> x x [3]) [log2(c/9)])
 
 // div by constant
-(UDIV x (MOVDconst [1])) -> x
-(UDIV x (MOVDconst [c])) && isPowerOfTwo(c) -> (SRLconst [log2(c)] x)
-(UDIVW x (MOVDconst [c])) && uint32(c)==1 -> x
-(UDIVW x (MOVDconst [c])) && isPowerOfTwo(c) && is32Bit(c) -> (SRLconst [log2(c)] x)
-(UMOD _ (MOVDconst [1])) -> (MOVDconst [0])
-(UMOD x (MOVDconst [c])) && isPowerOfTwo(c) -> (ANDconst [c-1] x)
-(UMODW _ (MOVDconst [c])) && uint32(c)==1 -> (MOVDconst [0])
-(UMODW x (MOVDconst [c])) && isPowerOfTwo(c) && is32Bit(c) -> (ANDconst [c-1] x)
+(UDIV x (MOVDconst [1])) => x
+(UDIV x (MOVDconst [c])) && isPowerOfTwo(c) => (SRLconst [log2(c)] x)
+(UDIVW x (MOVDconst [c])) && uint32(c)==1 => x
+(UDIVW x (MOVDconst [c])) && isPowerOfTwo(c) && is32Bit(c) => (SRLconst [log2(c)] x)
+(UMOD _ (MOVDconst [1])) => (MOVDconst [0])
+(UMOD x (MOVDconst [c])) && isPowerOfTwo(c) => (ANDconst [c-1] x)
+(UMODW _ (MOVDconst [c])) && uint32(c)==1 => (MOVDconst [0])
+(UMODW x (MOVDconst [c])) && isPowerOfTwo(c) && is32Bit(c) => (ANDconst [c-1] x)
 
 // generic simplifications
-(ADD x (NEG y)) -> (SUB x y)
-(SUB x x) -> (MOVDconst [0])
-(AND x x) -> x
-(OR  x x) -> x
-(XOR x x) -> (MOVDconst [0])
-(BIC x x) -> (MOVDconst [0])
-(EON x x) -> (MOVDconst [-1])
-(ORN x x) -> (MOVDconst [-1])
-(AND x (MVN y)) -> (BIC x y)
-(XOR x (MVN y)) -> (EON x y)
-(OR  x (MVN y)) -> (ORN x y)
-(MVN (XOR x y)) -> (EON x y)
+(ADD x (NEG y)) => (SUB x y)
+(SUB x x) => (MOVDconst [0])
+(AND x x) => x
+(OR  x x) => x
+(XOR x x) => (MOVDconst [0])
+(BIC x x) => (MOVDconst [0])
+(EON x x) => (MOVDconst [-1])
+(ORN x x) => (MOVDconst [-1])
+(AND x (MVN y)) => (BIC x y)
+(XOR x (MVN y)) => (EON x y)
+(OR  x (MVN y)) => (ORN x y)
+(MVN (XOR x y)) => (EON x y)
 (CSEL [cc] x (MOVDconst [0]) flag) => (CSEL0 [cc] x flag)
 (CSEL [cc] (MOVDconst [0]) y flag) => (CSEL0 [arm64Negate(cc)] y flag)
-(SUB x (SUB y z)) -> (SUB (ADD <v.Type> x z) y)
-(SUB (SUB x y) z) -> (SUB x (ADD <y.Type> y z))
+(SUB x (SUB y z)) => (SUB (ADD <v.Type> x z) y)
+(SUB (SUB x y) z) => (SUB x (ADD <y.Type> y z))
 
 // remove redundant *const ops
-(ADDconst [0]  x) -> x
-(SUBconst [0]  x) -> x
-(ANDconst [0]  _) -> (MOVDconst [0])
-(ANDconst [-1] x) -> x
-(ORconst  [0]  x) -> x
-(ORconst  [-1] _) -> (MOVDconst [-1])
-(XORconst [0]  x) -> x
-(XORconst [-1] x) -> (MVN x)
+(ADDconst [0]  x) => x
+(SUBconst [0]  x) => x
+(ANDconst [0]  _) => (MOVDconst [0])
+(ANDconst [-1] x) => x
+(ORconst  [0]  x) => x
+(ORconst  [-1] _) => (MOVDconst [-1])
+(XORconst [0]  x) => x
+(XORconst [-1] x) => (MVN x)
 
 // generic constant folding
-(ADDconst [c] (MOVDconst [d]))  -> (MOVDconst [c+d])
-(ADDconst [c] (ADDconst [d] x)) -> (ADDconst [c+d] x)
-(ADDconst [c] (SUBconst [d] x)) -> (ADDconst [c-d] x)
-(SUBconst [c] (MOVDconst [d]))  -> (MOVDconst [d-c])
-(SUBconst [c] (SUBconst [d] x)) -> (ADDconst [-c-d] x)
-(SUBconst [c] (ADDconst [d] x)) -> (ADDconst [-c+d] x)
-(SLLconst [c] (MOVDconst [d]))  -> (MOVDconst [d<<uint64(c)])
-(SRLconst [c] (MOVDconst [d]))  -> (MOVDconst [int64(uint64(d)>>uint64(c))])
-(SRAconst [c] (MOVDconst [d]))  -> (MOVDconst [d>>uint64(c)])
-(MUL   (MOVDconst [c]) (MOVDconst [d])) -> (MOVDconst [c*d])
-(MULW  (MOVDconst [c]) (MOVDconst [d])) -> (MOVDconst [int64(int32(c)*int32(d))])
-(MNEG  (MOVDconst [c]) (MOVDconst [d])) -> (MOVDconst [-c*d])
-(MNEGW (MOVDconst [c]) (MOVDconst [d])) -> (MOVDconst [-int64(int32(c)*int32(d))])
-(MADD  (MOVDconst [c]) x y) -> (ADDconst [c] (MUL   <x.Type> x y))
-(MADDW (MOVDconst [c]) x y) -> (ADDconst [c] (MULW  <x.Type> x y))
-(MSUB  (MOVDconst [c]) x y) -> (ADDconst [c] (MNEG  <x.Type> x y))
-(MSUBW (MOVDconst [c]) x y) -> (ADDconst [c] (MNEGW <x.Type> x y))
-(MADD  a (MOVDconst [c]) (MOVDconst [d])) -> (ADDconst [c*d] a)
-(MADDW a (MOVDconst [c]) (MOVDconst [d])) -> (ADDconst [int64(int32(c)*int32(d))] a)
-(MSUB  a (MOVDconst [c]) (MOVDconst [d])) -> (SUBconst [c*d] a)
-(MSUBW a (MOVDconst [c]) (MOVDconst [d])) -> (SUBconst [int64(int32(c)*int32(d))] a)
-(DIV   (MOVDconst [c]) (MOVDconst [d])) -> (MOVDconst [c/d])
-(UDIV  (MOVDconst [c]) (MOVDconst [d])) -> (MOVDconst [int64(uint64(c)/uint64(d))])
-(DIVW  (MOVDconst [c]) (MOVDconst [d])) -> (MOVDconst [int64(int32(c)/int32(d))])
-(UDIVW (MOVDconst [c]) (MOVDconst [d])) -> (MOVDconst [int64(uint32(c)/uint32(d))])
-(MOD   (MOVDconst [c]) (MOVDconst [d])) -> (MOVDconst [c%d])
-(UMOD  (MOVDconst [c]) (MOVDconst [d])) -> (MOVDconst [int64(uint64(c)%uint64(d))])
-(MODW  (MOVDconst [c]) (MOVDconst [d])) -> (MOVDconst [int64(int32(c)%int32(d))])
-(UMODW (MOVDconst [c]) (MOVDconst [d])) -> (MOVDconst [int64(uint32(c)%uint32(d))])
-(ANDconst [c] (MOVDconst [d]))  -> (MOVDconst [c&d])
-(ANDconst [c] (ANDconst [d] x)) -> (ANDconst [c&d] x)
-(ANDconst [c] (MOVWUreg x)) -> (ANDconst [c&(1<<32-1)] x)
-(ANDconst [c] (MOVHUreg x)) -> (ANDconst [c&(1<<16-1)] x)
-(ANDconst [c] (MOVBUreg x)) -> (ANDconst [c&(1<<8-1)] x)
-(MOVWUreg (ANDconst [c] x)) -> (ANDconst [c&(1<<32-1)] x)
-(MOVHUreg (ANDconst [c] x)) -> (ANDconst [c&(1<<16-1)] x)
-(MOVBUreg (ANDconst [c] x)) -> (ANDconst [c&(1<<8-1)] x)
-(ORconst  [c] (MOVDconst [d]))  -> (MOVDconst [c|d])
-(ORconst  [c] (ORconst [d] x))  -> (ORconst [c|d] x)
-(XORconst [c] (MOVDconst [d]))  -> (MOVDconst [c^d])
-(XORconst [c] (XORconst [d] x)) -> (XORconst [c^d] x)
-(MVN (MOVDconst [c])) -> (MOVDconst [^c])
-(NEG (MOVDconst [c])) -> (MOVDconst [-c])
-(MOVBreg  (MOVDconst [c])) -> (MOVDconst [int64(int8(c))])
-(MOVBUreg (MOVDconst [c])) -> (MOVDconst [int64(uint8(c))])
-(MOVHreg  (MOVDconst [c])) -> (MOVDconst [int64(int16(c))])
-(MOVHUreg (MOVDconst [c])) -> (MOVDconst [int64(uint16(c))])
-(MOVWreg  (MOVDconst [c])) -> (MOVDconst [int64(int32(c))])
-(MOVWUreg (MOVDconst [c])) -> (MOVDconst [int64(uint32(c))])
-(MOVDreg  (MOVDconst [c])) -> (MOVDconst [c])
+(ADDconst [c] (MOVDconst [d]))  => (MOVDconst [c+d])
+(ADDconst [c] (ADDconst [d] x)) => (ADDconst [c+d] x)
+(ADDconst [c] (SUBconst [d] x)) => (ADDconst [c-d] x)
+(SUBconst [c] (MOVDconst [d]))  => (MOVDconst [d-c])
+(SUBconst [c] (SUBconst [d] x)) => (ADDconst [-c-d] x)
+(SUBconst [c] (ADDconst [d] x)) => (ADDconst [-c+d] x)
+(SLLconst [c] (MOVDconst [d]))  => (MOVDconst [d<<uint64(c)])
+(SRLconst [c] (MOVDconst [d]))  => (MOVDconst [int64(uint64(d)>>uint64(c))])
+(SRAconst [c] (MOVDconst [d]))  => (MOVDconst [d>>uint64(c)])
+(MUL   (MOVDconst [c]) (MOVDconst [d])) => (MOVDconst [c*d])
+(MULW  (MOVDconst [c]) (MOVDconst [d])) => (MOVDconst [int64(int32(c)*int32(d))])
+(MNEG  (MOVDconst [c]) (MOVDconst [d])) => (MOVDconst [-c*d])
+(MNEGW (MOVDconst [c]) (MOVDconst [d])) => (MOVDconst [-int64(int32(c)*int32(d))])
+(MADD  (MOVDconst [c]) x y) => (ADDconst [c] (MUL   <x.Type> x y))
+(MADDW (MOVDconst [c]) x y) => (ADDconst [c] (MULW  <x.Type> x y))
+(MSUB  (MOVDconst [c]) x y) => (ADDconst [c] (MNEG  <x.Type> x y))
+(MSUBW (MOVDconst [c]) x y) => (ADDconst [c] (MNEGW <x.Type> x y))
+(MADD  a (MOVDconst [c]) (MOVDconst [d])) => (ADDconst [c*d] a)
+(MADDW a (MOVDconst [c]) (MOVDconst [d])) => (ADDconst [int64(int32(c)*int32(d))] a)
+(MSUB  a (MOVDconst [c]) (MOVDconst [d])) => (SUBconst [c*d] a)
+(MSUBW a (MOVDconst [c]) (MOVDconst [d])) => (SUBconst [int64(int32(c)*int32(d))] a)
+(DIV   (MOVDconst [c]) (MOVDconst [d])) => (MOVDconst [c/d])
+(UDIV  (MOVDconst [c]) (MOVDconst [d])) => (MOVDconst [int64(uint64(c)/uint64(d))])
+(DIVW  (MOVDconst [c]) (MOVDconst [d])) => (MOVDconst [int64(int32(c)/int32(d))])
+(UDIVW (MOVDconst [c]) (MOVDconst [d])) => (MOVDconst [int64(uint32(c)/uint32(d))])
+(MOD   (MOVDconst [c]) (MOVDconst [d])) => (MOVDconst [c%d])
+(UMOD  (MOVDconst [c]) (MOVDconst [d])) => (MOVDconst [int64(uint64(c)%uint64(d))])
+(MODW  (MOVDconst [c]) (MOVDconst [d])) => (MOVDconst [int64(int32(c)%int32(d))])
+(UMODW (MOVDconst [c]) (MOVDconst [d])) => (MOVDconst [int64(uint32(c)%uint32(d))])
+(ANDconst [c] (MOVDconst [d]))  => (MOVDconst [c&d])
+(ANDconst [c] (ANDconst [d] x)) => (ANDconst [c&d] x)
+(ANDconst [c] (MOVWUreg x)) => (ANDconst [c&(1<<32-1)] x)
+(ANDconst [c] (MOVHUreg x)) => (ANDconst [c&(1<<16-1)] x)
+(ANDconst [c] (MOVBUreg x)) => (ANDconst [c&(1<<8-1)] x)
+(MOVWUreg (ANDconst [c] x)) => (ANDconst [c&(1<<32-1)] x)
+(MOVHUreg (ANDconst [c] x)) => (ANDconst [c&(1<<16-1)] x)
+(MOVBUreg (ANDconst [c] x)) => (ANDconst [c&(1<<8-1)] x)
+(ORconst  [c] (MOVDconst [d]))  => (MOVDconst [c|d])
+(ORconst  [c] (ORconst [d] x))  => (ORconst [c|d] x)
+(XORconst [c] (MOVDconst [d]))  => (MOVDconst [c^d])
+(XORconst [c] (XORconst [d] x)) => (XORconst [c^d] x)
+(MVN (MOVDconst [c])) => (MOVDconst [^c])
+(NEG (MOVDconst [c])) => (MOVDconst [-c])
+(MOVBreg  (MOVDconst [c])) => (MOVDconst [int64(int8(c))])
+(MOVBUreg (MOVDconst [c])) => (MOVDconst [int64(uint8(c))])
+(MOVHreg  (MOVDconst [c])) => (MOVDconst [int64(int16(c))])
+(MOVHUreg (MOVDconst [c])) => (MOVDconst [int64(uint16(c))])
+(MOVWreg  (MOVDconst [c])) => (MOVDconst [int64(int32(c))])
+(MOVWUreg (MOVDconst [c])) => (MOVDconst [int64(uint32(c))])
+(MOVDreg  (MOVDconst [c])) => (MOVDconst [c])
 
 // constant comparisons
 (CMPconst  (MOVDconst [x]) [y]) => (FlagConstant [subFlags64(x,y)])