cmd/compile: fix incorrect rewriting to if condition

Some ARM64 rewriting rules convert 'comparing to zero' conditions of if
statements to a simplified version utilizing CMN and CMP instructions to
branch over condition flags, in order to save one Add or Sub caculation.

Such optimizations lead to wrong branching in case an overflow/underflow
occurs when executing CMN or CMP.

Fix the issue by introducing new block opcodes that don't honor the
overflow/underflow flag, in the following categories:

  Block-Op        Meaning                   ARM condition codes
  1. LTnoov        less than                 MI
  2. GEnoov        greater than or equal     PL
  3. LEnoov        less than or equal        MI || EQ
  4. GTnoov        greater than              NEQ & PL

The backend generates two consecutive branch instructions for 'LEnoov'
and 'GTnoov' to model their expected behavior. A slight change to 'gc'
and amd64/386 backends is made to unify the code generation.

Add a test 'TestCondRewrite' as justification, it covers 32 incorrect rules
identified on arm64, more might be needed on other arches, like 32-bit arm.

Add two benchmarks profiling the aforementioned category 1&2 and category
3&4 separetely, we expect the first two categories will show performance
improvement and the second will not result in visible regression compared with
the non-optimized version.

This change also updates TestFormats to support using %#x.

Examples exhibiting where does the issue come from:
  1: 'if x + 3 < 0' might be converted to:
  before:
    CMN $3, R0
    BGE <else branch> // wrong branch is taken if 'x+3' overflows
  after:
    CMN $3, R0
    BPL <else branch>

  2: 'if y - 3 > 0' might be converted to:
  before:
    CMP $3, R0
    BLE <else branch> // wrong branch is taken if 'y-3' underflows
  after:
    CMP $3, R0
    BMI <else branch>
    BEQ <else branch>

Benchmark data from different kinds of arm64 servers, 'old' is the non-optimized
version (not the parent commit), generally the optimization version outperforms.

S1:
name                    old time/op  new time/op  delta
CondRewrite/SoloJump  13.6ns ± 0%  12.9ns ± 0%  -5.15%  (p=0.000 n=10+10)
CondRewrite/CombJump  13.8ns ± 1%  12.9ns ± 0%  -6.32%  (p=0.000 n=10+10)

S2:
name                     old time/op  new time/op  delta
CondRewrite/SoloJump  11.6ns ± 0%  10.9ns ± 0%  -6.03%  (p=0.000 n=10+10)
CondRewrite/CombJump  11.4ns ± 0%  10.8ns ± 1%  -5.53%  (p=0.000 n=10+10)

S3:
name                     old time/op  new time/op  delta
CondRewrite/SoloJump  7.36ns ± 0%  7.50ns ± 0%  +1.79%  (p=0.000 n=9+10)
CondRewrite/CombJump  7.35ns ± 0%  7.75ns ± 0%  +5.51%  (p=0.000 n=8+9)

S4:
name                      old time/op  new time/op  delta
CondRewrite/SoloJump-224  11.5ns ± 1%  10.9ns ± 0%  -4.97%  (p=0.000 n=10+10)
CondRewrite/CombJump-224  11.9ns ± 0%  11.5ns ± 0%  -2.95%  (p=0.000 n=10+10)

S5:
name                     old time/op  new time/op  delta
CondRewrite/SoloJump  10.0ns ± 0%  10.0ns ± 0%  -0.45%  (p=0.000 n=9+10)
CondRewrite/CombJump  9.93ns ± 0%  9.77ns ± 0%  -1.53%  (p=0.000 n=10+9)

Go1 perf. data:

name                     old time/op    new time/op    delta
BinaryTree17              6.29s ± 1%     6.30s ± 1%    ~     (p=1.000 n=5+5)
Fannkuch11                5.40s ± 0%     5.40s ± 0%    ~     (p=0.841 n=5+5)
FmtFprintfEmpty          97.9ns ± 0%    98.9ns ± 3%    ~     (p=0.937 n=4+5)
FmtFprintfString          171ns ± 3%     171ns ± 2%    ~     (p=0.754 n=5+5)
FmtFprintfInt             212ns ± 0%     217ns ± 6%  +2.55%  (p=0.008 n=5+5)
FmtFprintfIntInt          296ns ± 1%     297ns ± 2%    ~     (p=0.516 n=5+5)
FmtFprintfPrefixedInt     371ns ± 2%     374ns ± 7%    ~     (p=1.000 n=5+5)
FmtFprintfFloat           435ns ± 1%     439ns ± 2%    ~     (p=0.056 n=5+5)
FmtManyArgs              1.37µs ± 1%    1.36µs ± 1%    ~     (p=0.730 n=5+5)
GobDecode                14.6ms ± 4%    14.4ms ± 4%    ~     (p=0.690 n=5+5)
GobEncode                11.8ms ±20%    11.6ms ±15%    ~     (p=1.000 n=5+5)
Gzip                      507ms ± 0%     491ms ± 0%  -3.22%  (p=0.008 n=5+5)
Gunzip                   73.8ms ± 0%    73.9ms ± 0%    ~     (p=0.690 n=5+5)
HTTPClientServer          116µs ± 0%     116µs ± 0%    ~     (p=0.686 n=4+4)
JSONEncode               21.8ms ± 1%    21.6ms ± 2%    ~     (p=0.151 n=5+5)
JSONDecode                104ms ± 1%     103ms ± 1%  -1.08%  (p=0.016 n=5+5)
Mandelbrot200            9.53ms ± 0%    9.53ms ± 0%    ~     (p=0.421 n=5+5)
GoParse                  7.55ms ± 1%    7.51ms ± 1%    ~     (p=0.151 n=5+5)
RegexpMatchEasy0_32       158ns ± 0%     158ns ± 0%    ~     (all equal)
RegexpMatchEasy0_1K       606ns ± 1%     608ns ± 3%    ~     (p=0.937 n=5+5)
RegexpMatchEasy1_32       143ns ± 0%     144ns ± 1%    ~     (p=0.095 n=5+4)
RegexpMatchEasy1_1K       927ns ± 2%     944ns ± 2%    ~     (p=0.056 n=5+5)
RegexpMatchMedium_32     16.0ns ± 0%    16.0ns ± 0%    ~     (all equal)
RegexpMatchMedium_1K     69.3µs ± 2%    69.7µs ± 0%    ~     (p=0.690 n=5+5)
RegexpMatchHard_32       3.73µs ± 0%    3.73µs ± 1%    ~     (p=0.984 n=5+5)
RegexpMatchHard_1K        111µs ± 1%     110µs ± 0%    ~     (p=0.151 n=5+5)
Revcomp                   1.91s ±47%     1.77s ±68%    ~     (p=1.000 n=5+5)
Template                  138ms ± 1%     138ms ± 1%    ~     (p=1.000 n=5+5)
TimeParse                 787ns ± 2%     785ns ± 1%    ~     (p=0.540 n=5+5)
TimeFormat                729ns ± 1%     726ns ± 1%    ~     (p=0.151 n=5+5)

Updates #38740
Change-Id: I06c604874acdc1e63e66452dadee5df053045222
Reviewed-on: https://go-review.googlesource.com/c/go/+/233097
Reviewed-by: Keith Randall <khr@golang.org>
Run-TryBot: Keith Randall <khr@golang.org>

11 files changed, 891 insertions, 141 deletions

diff --git a/src/cmd/compile/fmtmap_test.go b/src/cmd/compile/fmtmap_test.go
index 5a24296c5e..6f69abf4fb 100644
--- a/src/cmd/compile/fmtmap_test.go
+++ b/src/cmd/compile/fmtmap_test.go
@@ -151,9 +151,11 @@ var knownFormats = map[string]string{
 	"int %x":                                          "",
 	"int16 %d":                                        "",
 	"int16 %x":                                        "",
+	"int32 %#x":                                       "",
 	"int32 %d":                                        "",
 	"int32 %v":                                        "",
 	"int32 %x":                                        "",
+	"int64 %#x":                                       "",
 	"int64 %+d":                                       "",
 	"int64 %-10d":                                     "",
 	"int64 %.5d":                                      "",
diff --git a/src/cmd/compile/internal/amd64/ssa.go b/src/cmd/compile/internal/amd64/ssa.go
index b58696d650..47cb422ab1 100644
--- a/src/cmd/compile/internal/amd64/ssa.go
+++ b/src/cmd/compile/internal/amd64/ssa.go
@@ -1252,11 +1252,11 @@ var blockJump = [...]struct {
 	ssa.BlockAMD64NAN: {x86.AJPS, x86.AJPC},
 }
 
-var eqfJumps = [2][2]gc.FloatingEQNEJump{
+var eqfJumps = [2][2]gc.IndexJump{
 	{{Jump: x86.AJNE, Index: 1}, {Jump: x86.AJPS, Index: 1}}, // next == b.Succs[0]
 	{{Jump: x86.AJNE, Index: 1}, {Jump: x86.AJPC, Index: 0}}, // next == b.Succs[1]
 }
-var nefJumps = [2][2]gc.FloatingEQNEJump{
+var nefJumps = [2][2]gc.IndexJump{
 	{{Jump: x86.AJNE, Index: 0}, {Jump: x86.AJPC, Index: 1}}, // next == b.Succs[0]
 	{{Jump: x86.AJNE, Index: 0}, {Jump: x86.AJPS, Index: 0}}, // next == b.Succs[1]
 }
@@ -1296,10 +1296,10 @@ func ssaGenBlock(s *gc.SSAGenState, b, next *ssa.Block) {
 		p.To.Sym = b.Aux.(*obj.LSym)
 
 	case ssa.BlockAMD64EQF:
-		s.FPJump(b, next, &eqfJumps)
+		s.CombJump(b, next, &eqfJumps)
 
 	case ssa.BlockAMD64NEF:
-		s.FPJump(b, next, &nefJumps)
+		s.CombJump(b, next, &nefJumps)
 
 	case ssa.BlockAMD64EQ, ssa.BlockAMD64NE,
 		ssa.BlockAMD64LT, ssa.BlockAMD64GE,
diff --git a/src/cmd/compile/internal/arm64/ssa.go b/src/cmd/compile/internal/arm64/ssa.go
index 06c520d76a..709253b48d 100644
--- a/src/cmd/compile/internal/arm64/ssa.go
+++ b/src/cmd/compile/internal/arm64/ssa.go
@@ -998,6 +998,20 @@ var blockJump = map[ssa.BlockKind]struct {
 	ssa.BlockARM64FGE:  {arm64.ABGE, arm64.ABLT},
 	ssa.BlockARM64FLE:  {arm64.ABLS, arm64.ABHI},
 	ssa.BlockARM64FGT:  {arm64.ABGT, arm64.ABLE},
+	ssa.BlockARM64LTnoov:{arm64.ABMI, arm64.ABPL},
+	ssa.BlockARM64GEnoov:{arm64.ABPL, arm64.ABMI},
+}
+
+// To model a 'LEnoov' ('<=' without overflow checking) branching
+var leJumps = [2][2]gc.IndexJump{
+	{{Jump: arm64.ABEQ, Index: 0}, {Jump: arm64.ABPL, Index: 1}}, // next == b.Succs[0]
+	{{Jump: arm64.ABMI, Index: 0}, {Jump: arm64.ABEQ, Index: 0}}, // next == b.Succs[1]
+}
+
+// To model a 'GTnoov' ('>' without overflow checking) branching
+var gtJumps = [2][2]gc.IndexJump{
+	{{Jump: arm64.ABMI, Index: 1}, {Jump: arm64.ABEQ, Index: 1}}, // next == b.Succs[0]
+	{{Jump: arm64.ABEQ, Index: 1}, {Jump: arm64.ABPL, Index: 0}}, // next == b.Succs[1]
 }
 
 func ssaGenBlock(s *gc.SSAGenState, b, next *ssa.Block) {
@@ -1045,7 +1059,8 @@ func ssaGenBlock(s *gc.SSAGenState, b, next *ssa.Block) {
 		ssa.BlockARM64Z, ssa.BlockARM64NZ,
 		ssa.BlockARM64ZW, ssa.BlockARM64NZW,
 		ssa.BlockARM64FLT, ssa.BlockARM64FGE,
-		ssa.BlockARM64FLE, ssa.BlockARM64FGT:
+		ssa.BlockARM64FLE, ssa.BlockARM64FGT,
+		ssa.BlockARM64LTnoov, ssa.BlockARM64GEnoov:
 		jmp := blockJump[b.Kind]
 		var p *obj.Prog
 		switch next {
@@ -1087,6 +1102,10 @@ func ssaGenBlock(s *gc.SSAGenState, b, next *ssa.Block) {
 		p.From.Type = obj.TYPE_CONST
 		p.Reg = b.Controls[0].Reg()
 
+	case ssa.BlockARM64LEnoov:
+		s.CombJump(b, next, &leJumps)
+	case ssa.BlockARM64GTnoov:
+		s.CombJump(b, next, &gtJumps)
 	default:
 		b.Fatalf("branch not implemented: %s", b.LongString())
 	}
diff --git a/src/cmd/compile/internal/gc/ssa.go b/src/cmd/compile/internal/gc/ssa.go
index 70f6dd6e18..c0902cdea6 100644
--- a/src/cmd/compile/internal/gc/ssa.go
+++ b/src/cmd/compile/internal/gc/ssa.go
@@ -6313,34 +6313,39 @@ func defframe(s *SSAGenState, e *ssafn) {
 	thearch.ZeroRange(pp, p, frame+lo, hi-lo, &state)
 }
 
-type FloatingEQNEJump struct {
+// For generating consecutive jump instructions to model a specific branching
+type IndexJump struct {
 	Jump  obj.As
 	Index int
 }
 
-func (s *SSAGenState) oneFPJump(b *ssa.Block, jumps *FloatingEQNEJump) {
-	p := s.Prog(jumps.Jump)
-	p.To.Type = obj.TYPE_BRANCH
+func (s *SSAGenState) oneJump(b *ssa.Block, jump *IndexJump) {
+	p := s.Br(jump.Jump, b.Succs[jump.Index].Block())
 	p.Pos = b.Pos
-	to := jumps.Index
-	s.Branches = append(s.Branches, Branch{p, b.Succs[to].Block()})
 }
 
-func (s *SSAGenState) FPJump(b, next *ssa.Block, jumps *[2][2]FloatingEQNEJump) {
+// CombJump generates combinational instructions (2 at present) for a block jump,
+// thereby the behaviour of non-standard condition codes could be simulated
+func (s *SSAGenState) CombJump(b, next *ssa.Block, jumps *[2][2]IndexJump) {
 	switch next {
 	case b.Succs[0].Block():
-		s.oneFPJump(b, &jumps[0][0])
-		s.oneFPJump(b, &jumps[0][1])
+		s.oneJump(b, &jumps[0][0])
+		s.oneJump(b, &jumps[0][1])
 	case b.Succs[1].Block():
-		s.oneFPJump(b, &jumps[1][0])
-		s.oneFPJump(b, &jumps[1][1])
+		s.oneJump(b, &jumps[1][0])
+		s.oneJump(b, &jumps[1][1])
 	default:
-		s.oneFPJump(b, &jumps[1][0])
-		s.oneFPJump(b, &jumps[1][1])
-		q := s.Prog(obj.AJMP)
+		var q *obj.Prog
+		if b.Likely != ssa.BranchUnlikely {
+			s.oneJump(b, &jumps[1][0])
+			s.oneJump(b, &jumps[1][1])
+			q = s.Br(obj.AJMP, b.Succs[1].Block())
+		} else {
+			s.oneJump(b, &jumps[0][0])
+			s.oneJump(b, &jumps[0][1])
+			q = s.Br(obj.AJMP, b.Succs[0].Block())
+		}
 		q.Pos = b.Pos
-		q.To.Type = obj.TYPE_BRANCH
-		s.Branches = append(s.Branches, Branch{q, b.Succs[1].Block()})
 	}
 }
 
diff --git a/src/cmd/compile/internal/ssa/gen/ARM64.rules b/src/cmd/compile/internal/ssa/gen/ARM64.rules
index 03202414d6..47f221495a 100644
--- a/src/cmd/compile/internal/ssa/gen/ARM64.rules
+++ b/src/cmd/compile/internal/ssa/gen/ARM64.rules
@@ -598,31 +598,31 @@
 
 (EQ (CMPconst [0] x:(ADDconst [c] y)) yes no) && x.Uses == 1 => (EQ (CMNconst [c] y) yes no)
 (NE (CMPconst [0] x:(ADDconst [c] y)) yes no) && x.Uses == 1 => (NE (CMNconst [c] y) yes no)
-(LT (CMPconst [0] x:(ADDconst [c] y)) yes no) && x.Uses == 1 => (LT (CMNconst [c] y) yes no)
-(LE (CMPconst [0] x:(ADDconst [c] y)) yes no) && x.Uses == 1 => (LE (CMNconst [c] y) yes no)
-(GT (CMPconst [0] x:(ADDconst [c] y)) yes no) && x.Uses == 1 => (GT (CMNconst [c] y) yes no)
-(GE (CMPconst [0] x:(ADDconst [c] y)) yes no) && x.Uses == 1 => (GE (CMNconst [c] y) yes no)
+(LT (CMPconst [0] x:(ADDconst [c] y)) yes no) && x.Uses == 1 => (LTnoov (CMNconst [c] y) yes no)
+(LE (CMPconst [0] x:(ADDconst [c] y)) yes no) && x.Uses == 1 => (LEnoov (CMNconst [c] y) yes no)
+(GT (CMPconst [0] x:(ADDconst [c] y)) yes no) && x.Uses == 1 => (GTnoov (CMNconst [c] y) yes no)
+(GE (CMPconst [0] x:(ADDconst [c] y)) yes no) && x.Uses == 1 => (GEnoov (CMNconst [c] y) yes no)
 
 (EQ (CMPWconst [0] x:(ADDconst [c] y)) yes no) && x.Uses == 1 => (EQ (CMNWconst [int32(c)] y) yes no)
 (NE (CMPWconst [0] x:(ADDconst [c] y)) yes no) && x.Uses == 1 => (NE (CMNWconst [int32(c)] y) yes no)
-(LT (CMPWconst [0] x:(ADDconst [c] y)) yes no) && x.Uses == 1 => (LT (CMNWconst [int32(c)] y) yes no)
-(LE (CMPWconst [0] x:(ADDconst [c] y)) yes no) && x.Uses == 1 => (LE (CMNWconst [int32(c)] y) yes no)
-(GT (CMPWconst [0] x:(ADDconst [c] y)) yes no) && x.Uses == 1 => (GT (CMNWconst [int32(c)] y) yes no)
-(GE (CMPWconst [0] x:(ADDconst [c] y)) yes no) && x.Uses == 1 => (GE (CMNWconst [int32(c)] y) yes no)
+(LT (CMPWconst [0] x:(ADDconst [c] y)) yes no) && x.Uses == 1 => (LTnoov (CMNWconst [int32(c)] y) yes no)
+(LE (CMPWconst [0] x:(ADDconst [c] y)) yes no) && x.Uses == 1 => (LEnoov (CMNWconst [int32(c)] y) yes no)
+(GT (CMPWconst [0] x:(ADDconst [c] y)) yes no) && x.Uses == 1 => (GTnoov (CMNWconst [int32(c)] y) yes no)
+(GE (CMPWconst [0] x:(ADDconst [c] y)) yes no) && x.Uses == 1 => (GEnoov (CMNWconst [int32(c)] y) yes no)
 
 (EQ (CMPconst [0] z:(ADD x y)) yes no) && z.Uses == 1 => (EQ (CMN x y) yes no)
 (NE (CMPconst [0] z:(ADD x y)) yes no) && z.Uses == 1 => (NE (CMN x y) yes no)
-(LT (CMPconst [0] z:(ADD x y)) yes no) && z.Uses == 1 => (LT (CMN x y) yes no)
-(LE (CMPconst [0] z:(ADD x y)) yes no) && z.Uses == 1 => (LE (CMN x y) yes no)
-(GT (CMPconst [0] z:(ADD x y)) yes no) && z.Uses == 1 => (GT (CMN x y) yes no)
-(GE (CMPconst [0] z:(ADD x y)) yes no) && z.Uses == 1 => (GE (CMN x y) yes no)
+(LT (CMPconst [0] z:(ADD x y)) yes no) && z.Uses == 1 => (LTnoov (CMN x y) yes no)
+(LE (CMPconst [0] z:(ADD x y)) yes no) && z.Uses == 1 => (LEnoov (CMN x y) yes no)
+(GT (CMPconst [0] z:(ADD x y)) yes no) && z.Uses == 1 => (GTnoov (CMN x y) yes no)
+(GE (CMPconst [0] z:(ADD x y)) yes no) && z.Uses == 1 => (GEnoov (CMN x y) yes no)
 
 (EQ (CMPWconst [0] z:(ADD x y)) yes no) && z.Uses == 1 => (EQ (CMNW x y) yes no)
 (NE (CMPWconst [0] z:(ADD x y)) yes no) && z.Uses == 1 => (NE (CMNW x y) yes no)
-(LT (CMPWconst [0] z:(ADD x y)) yes no) && z.Uses == 1 => (LT (CMNW x y) yes no)
-(LE (CMPWconst [0] z:(ADD x y)) yes no) && z.Uses == 1 => (LE (CMNW x y) yes no)
-(GT (CMPWconst [0] z:(ADD x y)) yes no) && z.Uses == 1 => (GT (CMNW x y) yes no)
-(GE (CMPWconst [0] z:(ADD x y)) yes no) && z.Uses == 1 => (GE (CMNW x y) yes no)
+(LT (CMPWconst [0] z:(ADD x y)) yes no) && z.Uses == 1 => (LTnoov (CMNW x y) yes no)
+(LE (CMPWconst [0] z:(ADD x y)) yes no) && z.Uses == 1 => (LEnoov (CMNW x y) yes no)
+(GT (CMPWconst [0] z:(ADD x y)) yes no) && z.Uses == 1 => (GTnoov (CMNW x y) yes no)
+(GE (CMPWconst [0] z:(ADD x y)) yes no) && z.Uses == 1 => (GEnoov (CMNW x y) yes no)
 
 (EQ (CMP x z:(NEG y)) yes no) && z.Uses == 1 => (EQ (CMN x y) yes no)
 (NE (CMP x z:(NEG y)) yes no) && z.Uses == 1 => (NE (CMN x y) yes no)
@@ -645,31 +645,31 @@
 
 (EQ (CMPconst [0]  z:(MADD a x y)) yes no) && z.Uses==1 => (EQ (CMN a (MUL <x.Type> x y)) yes no)
 (NE (CMPconst [0]  z:(MADD a x y)) yes no) && z.Uses==1 => (NE (CMN a (MUL <x.Type> x y)) yes no)
-(LT (CMPconst [0]  z:(MADD a x y)) yes no) && z.Uses==1 => (LT (CMN a (MUL <x.Type> x y)) yes no)
-(LE (CMPconst [0]  z:(MADD a x y)) yes no) && z.Uses==1 => (LE (CMN a (MUL <x.Type> x y)) yes no)
-(GT (CMPconst [0]  z:(MADD a x y)) yes no) && z.Uses==1 => (GT (CMN a (MUL <x.Type> x y)) yes no)
-(GE (CMPconst [0]  z:(MADD a x y)) yes no) && z.Uses==1 => (GE (CMN a (MUL <x.Type> x y)) yes no)
+(LT (CMPconst [0]  z:(MADD a x y)) yes no) && z.Uses==1 => (LTnoov (CMN a (MUL <x.Type> x y)) yes no)
+(LE (CMPconst [0]  z:(MADD a x y)) yes no) && z.Uses==1 => (LEnoov (CMN a (MUL <x.Type> x y)) yes no)
+(GT (CMPconst [0]  z:(MADD a x y)) yes no) && z.Uses==1 => (GTnoov (CMN a (MUL <x.Type> x y)) yes no)
+(GE (CMPconst [0]  z:(MADD a x y)) yes no) && z.Uses==1 => (GEnoov (CMN a (MUL <x.Type> x y)) yes no)
 
 (EQ (CMPconst [0]  z:(MSUB a x y)) yes no) && z.Uses==1 => (EQ (CMP a (MUL <x.Type> x y)) yes no)
 (NE (CMPconst [0]  z:(MSUB a x y)) yes no) && z.Uses==1 => (NE (CMP a (MUL <x.Type> x y)) yes no)
-(LE (CMPconst [0]  z:(MSUB a x y)) yes no) && z.Uses==1 => (LE (CMP a (MUL <x.Type> x y)) yes no)
-(LT (CMPconst [0]  z:(MSUB a x y)) yes no) && z.Uses==1 => (LT (CMP a (MUL <x.Type> x y)) yes no)
-(GE (CMPconst [0]  z:(MSUB a x y)) yes no) && z.Uses==1 => (GE (CMP a (MUL <x.Type> x y)) yes no)
-(GT (CMPconst [0]  z:(MSUB a x y)) yes no) && z.Uses==1 => (GT (CMP a (MUL <x.Type> x y)) yes no)
+(LE (CMPconst [0]  z:(MSUB a x y)) yes no) && z.Uses==1 => (LEnoov (CMP a (MUL <x.Type> x y)) yes no)
+(LT (CMPconst [0]  z:(MSUB a x y)) yes no) && z.Uses==1 => (LTnoov (CMP a (MUL <x.Type> x y)) yes no)
+(GE (CMPconst [0]  z:(MSUB a x y)) yes no) && z.Uses==1 => (GEnoov (CMP a (MUL <x.Type> x y)) yes no)
+(GT (CMPconst [0]  z:(MSUB a x y)) yes no) && z.Uses==1 => (GTnoov (CMP a (MUL <x.Type> x y)) yes no)
 
 (EQ (CMPWconst [0] z:(MADDW a x y)) yes no) && z.Uses==1 => (EQ (CMNW a (MULW <x.Type> x y)) yes no)
 (NE (CMPWconst [0] z:(MADDW a x y)) yes no) && z.Uses==1 => (NE (CMNW a (MULW <x.Type> x y)) yes no)
-(LE (CMPWconst [0] z:(MADDW a x y)) yes no) && z.Uses==1 => (LE (CMNW a (MULW <x.Type> x y)) yes no)
-(LT (CMPWconst [0] z:(MADDW a x y)) yes no) && z.Uses==1 => (LT (CMNW a (MULW <x.Type> x y)) yes no)
-(GE (CMPWconst [0] z:(MADDW a x y)) yes no) && z.Uses==1 => (GE (CMNW a (MULW <x.Type> x y)) yes no)
-(GT (CMPWconst [0] z:(MADDW a x y)) yes no) && z.Uses==1 => (GT (CMNW a (MULW <x.Type> x y)) yes no)
+(LE (CMPWconst [0] z:(MADDW a x y)) yes no) && z.Uses==1 => (LEnoov (CMNW a (MULW <x.Type> x y)) yes no)
+(LT (CMPWconst [0] z:(MADDW a x y)) yes no) && z.Uses==1 => (LTnoov (CMNW a (MULW <x.Type> x y)) yes no)
+(GE (CMPWconst [0] z:(MADDW a x y)) yes no) && z.Uses==1 => (GEnoov (CMNW a (MULW <x.Type> x y)) yes no)
+(GT (CMPWconst [0] z:(MADDW a x y)) yes no) && z.Uses==1 => (GTnoov (CMNW a (MULW <x.Type> x y)) yes no)
 
 (EQ (CMPWconst [0] z:(MSUBW a x y)) yes no) && z.Uses==1 => (EQ (CMPW a (MULW <x.Type> x y)) yes no)
 (NE (CMPWconst [0] z:(MSUBW a x y)) yes no) && z.Uses==1 => (NE (CMPW a (MULW <x.Type> x y)) yes no)
-(LE (CMPWconst [0] z:(MSUBW a x y)) yes no) && z.Uses==1 => (LE (CMPW a (MULW <x.Type> x y)) yes no)
-(LT (CMPWconst [0] z:(MSUBW a x y)) yes no) && z.Uses==1 => (LT (CMPW a (MULW <x.Type> x y)) yes no)
-(GE (CMPWconst [0] z:(MSUBW a x y)) yes no) && z.Uses==1 => (GE (CMPW a (MULW <x.Type> x y)) yes no)
-(GT (CMPWconst [0] z:(MSUBW a x y)) yes no) && z.Uses==1 => (GT (CMPW a (MULW <x.Type> x y)) yes no)
+(LE (CMPWconst [0] z:(MSUBW a x y)) yes no) && z.Uses==1 => (LEnoov (CMPW a (MULW <x.Type> x y)) yes no)
+(LT (CMPWconst [0] z:(MSUBW a x y)) yes no) && z.Uses==1 => (LTnoov (CMPW a (MULW <x.Type> x y)) yes no)
+(GE (CMPWconst [0] z:(MSUBW a x y)) yes no) && z.Uses==1 => (GEnoov (CMPW a (MULW <x.Type> x y)) yes no)
+(GT (CMPWconst [0] z:(MSUBW a x y)) yes no) && z.Uses==1 => (GTnoov (CMPW a (MULW <x.Type> x y)) yes no)
 
 // Absorb bit-tests into block
 (Z  (ANDconst [c] x) yes no) && oneBit(c) => (TBZ  [int64(ntz64(c))] x yes no)
@@ -1503,6 +1503,10 @@
 (FGT (InvertFlags cmp) yes no) -> (FLT cmp yes no)
 (FLE (InvertFlags cmp) yes no) -> (FGE cmp yes no)
 (FGE (InvertFlags cmp) yes no) -> (FLE cmp yes no)
+(LTnoov (InvertFlags cmp) yes no) => (GTnoov cmp yes no)
+(GEnoov (InvertFlags cmp) yes no) => (LEnoov cmp yes no)
+(LEnoov (InvertFlags cmp) yes no) => (GEnoov cmp yes no)
+(GTnoov (InvertFlags cmp) yes no) => (LTnoov cmp yes no)
 
 // absorb InvertFlags into CSEL(0)
 (CSEL {cc} x y (InvertFlags cmp)) -> (CSEL {arm64Invert(cc.(Op))} x y cmp)
diff --git a/src/cmd/compile/internal/ssa/gen/ARM64Ops.go b/src/cmd/compile/internal/ssa/gen/ARM64Ops.go
index 73e18bc56a..63faab2909 100644
--- a/src/cmd/compile/internal/ssa/gen/ARM64Ops.go
+++ b/src/cmd/compile/internal/ssa/gen/ARM64Ops.go
@@ -701,6 +701,10 @@ func init() {
 		{name: "FLE", controls: 1},
 		{name: "FGT", controls: 1},
 		{name: "FGE", controls: 1},
+		{name: "LTnoov", controls: 1}, // 'LT' but without honoring overflow
+		{name: "LEnoov", controls: 1}, // 'LE' but without honoring overflow
+		{name: "GTnoov", controls: 1}, // 'GT' but without honoring overflow
+		{name: "GEnoov", controls: 1}, // 'GE' but without honoring overflow
 	}
 
 	archs = append(archs, arch{
diff --git a/src/cmd/compile/internal/ssa/opGen.go b/src/cmd/compile/internal/ssa/opGen.go
index d619f36cf5..4a83a46be2 100644
--- a/src/cmd/compile/internal/ssa/opGen.go
+++ b/src/cmd/compile/internal/ssa/opGen.go
@@ -82,6 +82,10 @@ const (
 	BlockARM64FLE
 	BlockARM64FGT
 	BlockARM64FGE
+	BlockARM64LTnoov
+	BlockARM64LEnoov
+	BlockARM64GTnoov
+	BlockARM64GEnoov
 
 	BlockMIPSEQ
 	BlockMIPSNE
@@ -192,26 +196,30 @@ var blockString = [...]string{
 	BlockARMUGT: "UGT",
 	BlockARMUGE: "UGE",
 
-	BlockARM64EQ:   "EQ",
-	BlockARM64NE:   "NE",
-	BlockARM64LT:   "LT",
-	BlockARM64LE:   "LE",
-	BlockARM64GT:   "GT",
-	BlockARM64GE:   "GE",
-	BlockARM64ULT:  "ULT",
-	BlockARM64ULE:  "ULE",
-	BlockARM64UGT:  "UGT",
-	BlockARM64UGE:  "UGE",
-	BlockARM64Z:    "Z",
-	BlockARM64NZ:   "NZ",
-	BlockARM64ZW:   "ZW",
-	BlockARM64NZW:  "NZW",
-	BlockARM64TBZ:  "TBZ",
-	BlockARM64TBNZ: "TBNZ",
-	BlockARM64FLT:  "FLT",
-	BlockARM64FLE:  "FLE",
-	BlockARM64FGT:  "FGT",
-	BlockARM64FGE:  "FGE",
+	BlockARM64EQ:     "EQ",
+	BlockARM64NE:     "NE",
+	BlockARM64LT:     "LT",
+	BlockARM64LE:     "LE",
+	BlockARM64GT:     "GT",
+	BlockARM64GE:     "GE",
+	BlockARM64ULT:    "ULT",
+	BlockARM64ULE:    "ULE",
+	BlockARM64UGT:    "UGT",
+	BlockARM64UGE:    "UGE",
+	BlockARM64Z:      "Z",
+	BlockARM64NZ:     "NZ",
+	BlockARM64ZW:     "ZW",
+	BlockARM64NZW:    "NZW",
+	BlockARM64TBZ:    "TBZ",
+	BlockARM64TBNZ:   "TBNZ",
+	BlockARM64FLT:    "FLT",
+	BlockARM64FLE:    "FLE",
+	BlockARM64FGT:    "FGT",
+	BlockARM64FGE:    "FGE",
+	BlockARM64LTnoov: "LTnoov",
+	BlockARM64LEnoov: "LEnoov",
+	BlockARM64GTnoov: "GTnoov",
+	BlockARM64GEnoov: "GEnoov",
 
 	BlockMIPSEQ:  "EQ",
 	BlockMIPSNE:  "NE",
diff --git a/src/cmd/compile/internal/ssa/rewriteARM64.go b/src/cmd/compile/internal/ssa/rewriteARM64.go
index d243ea9407..4b8ef43303 100644
--- a/src/cmd/compile/internal/ssa/rewriteARM64.go
+++ b/src/cmd/compile/internal/ssa/rewriteARM64.go
@@ -26436,7 +26436,7 @@ func rewriteBlockARM64(b *Block) bool {
 		}
 		// match: (GE (CMPconst [0] x:(ADDconst [c] y)) yes no)
 		// cond: x.Uses == 1
-		// result: (GE (CMNconst [c] y) yes no)
+		// result: (GEnoov (CMNconst [c] y) yes no)
 		for b.Controls[0].Op == OpARM64CMPconst {
 			v_0 := b.Controls[0]
 			if auxIntToInt64(v_0.AuxInt) != 0 {
@@ -26454,12 +26454,12 @@ func rewriteBlockARM64(b *Block) bool {
 			v0 := b.NewValue0(v_0.Pos, OpARM64CMNconst, types.TypeFlags)
 			v0.AuxInt = int64ToAuxInt(c)
 			v0.AddArg(y)
-			b.resetWithControl(BlockARM64GE, v0)
+			b.resetWithControl(BlockARM64GEnoov, v0)
 			return true
 		}
 		// match: (GE (CMPWconst [0] x:(ADDconst [c] y)) yes no)
 		// cond: x.Uses == 1
-		// result: (GE (CMNWconst [int32(c)] y) yes no)
+		// result: (GEnoov (CMNWconst [int32(c)] y) yes no)
 		for b.Controls[0].Op == OpARM64CMPWconst {
 			v_0 := b.Controls[0]
 			if auxIntToInt32(v_0.AuxInt) != 0 {
@@ -26477,12 +26477,12 @@ func rewriteBlockARM64(b *Block) bool {
 			v0 := b.NewValue0(v_0.Pos, OpARM64CMNWconst, types.TypeFlags)
 			v0.AuxInt = int32ToAuxInt(int32(c))
 			v0.AddArg(y)
-			b.resetWithControl(BlockARM64GE, v0)
+			b.resetWithControl(BlockARM64GEnoov, v0)
 			return true
 		}
 		// match: (GE (CMPconst [0] z:(ADD x y)) yes no)
 		// cond: z.Uses == 1
-		// result: (GE (CMN x y) yes no)
+		// result: (GEnoov (CMN x y) yes no)
 		for b.Controls[0].Op == OpARM64CMPconst {
 			v_0 := b.Controls[0]
 			if auxIntToInt64(v_0.AuxInt) != 0 {
@@ -26503,14 +26503,14 @@ func rewriteBlockARM64(b *Block) bool {
 				}
 				v0 := b.NewValue0(v_0.Pos, OpARM64CMN, types.TypeFlags)
 				v0.AddArg2(x, y)
-				b.resetWithControl(BlockARM64GE, v0)
+				b.resetWithControl(BlockARM64GEnoov, v0)
 				return true
 			}
 			break
 		}
 		// match: (GE (CMPWconst [0] z:(ADD x y)) yes no)
 		// cond: z.Uses == 1
-		// result: (GE (CMNW x y) yes no)
+		// result: (GEnoov (CMNW x y) yes no)
 		for b.Controls[0].Op == OpARM64CMPWconst {
 			v_0 := b.Controls[0]
 			if auxIntToInt32(v_0.AuxInt) != 0 {
@@ -26531,7 +26531,7 @@ func rewriteBlockARM64(b *Block) bool {
 				}
 				v0 := b.NewValue0(v_0.Pos, OpARM64CMNW, types.TypeFlags)
 				v0.AddArg2(x, y)
-				b.resetWithControl(BlockARM64GE, v0)
+				b.resetWithControl(BlockARM64GEnoov, v0)
 				return true
 			}
 			break
@@ -26578,7 +26578,7 @@ func rewriteBlockARM64(b *Block) bool {
 		}
 		// match: (GE (CMPconst [0] z:(MADD a x y)) yes no)
 		// cond: z.Uses==1
-		// result: (GE (CMN a (MUL <x.Type> x y)) yes no)
+		// result: (GEnoov (CMN a (MUL <x.Type> x y)) yes no)
 		for b.Controls[0].Op == OpARM64CMPconst {
 			v_0 := b.Controls[0]
 			if auxIntToInt64(v_0.AuxInt) != 0 {
@@ -26598,12 +26598,12 @@ func rewriteBlockARM64(b *Block) bool {
 			v1 := b.NewValue0(v_0.Pos, OpARM64MUL, x.Type)
 			v1.AddArg2(x, y)
 			v0.AddArg2(a, v1)
-			b.resetWithControl(BlockARM64GE, v0)
+			b.resetWithControl(BlockARM64GEnoov, v0)
 			return true
 		}
 		// match: (GE (CMPconst [0] z:(MSUB a x y)) yes no)
 		// cond: z.Uses==1
-		// result: (GE (CMP a (MUL <x.Type> x y)) yes no)
+		// result: (GEnoov (CMP a (MUL <x.Type> x y)) yes no)
 		for b.Controls[0].Op == OpARM64CMPconst {
 			v_0 := b.Controls[0]
 			if auxIntToInt64(v_0.AuxInt) != 0 {
@@ -26623,12 +26623,12 @@ func rewriteBlockARM64(b *Block) bool {
 			v1 := b.NewValue0(v_0.Pos, OpARM64MUL, x.Type)
 			v1.AddArg2(x, y)
 			v0.AddArg2(a, v1)
-			b.resetWithControl(BlockARM64GE, v0)
+			b.resetWithControl(BlockARM64GEnoov, v0)
 			return true
 		}
 		// match: (GE (CMPWconst [0] z:(MADDW a x y)) yes no)
 		// cond: z.Uses==1
-		// result: (GE (CMNW a (MULW <x.Type> x y)) yes no)
+		// result: (GEnoov (CMNW a (MULW <x.Type> x y)) yes no)
 		for b.Controls[0].Op == OpARM64CMPWconst {
 			v_0 := b.Controls[0]
 			if auxIntToInt32(v_0.AuxInt) != 0 {
@@ -26648,12 +26648,12 @@ func rewriteBlockARM64(b *Block) bool {
 			v1 := b.NewValue0(v_0.Pos, OpARM64MULW, x.Type)
 			v1.AddArg2(x, y)
 			v0.AddArg2(a, v1)
-			b.resetWithControl(BlockARM64GE, v0)
+			b.resetWithControl(BlockARM64GEnoov, v0)
 			return true
 		}
 		// match: (GE (CMPWconst [0] z:(MSUBW a x y)) yes no)
 		// cond: z.Uses==1
-		// result: (GE (CMPW a (MULW <x.Type> x y)) yes no)
+		// result: (GEnoov (CMPW a (MULW <x.Type> x y)) yes no)
 		for b.Controls[0].Op == OpARM64CMPWconst {
 			v_0 := b.Controls[0]
 			if auxIntToInt32(v_0.AuxInt) != 0 {
@@ -26673,7 +26673,7 @@ func rewriteBlockARM64(b *Block) bool {
 			v1 := b.NewValue0(v_0.Pos, OpARM64MULW, x.Type)
 			v1.AddArg2(x, y)
 			v0.AddArg2(a, v1)
-			b.resetWithControl(BlockARM64GE, v0)
+			b.resetWithControl(BlockARM64GEnoov, v0)
 			return true
 		}
 		// match: (GE (CMPWconst [0] x) yes no)
@@ -26740,6 +26740,15 @@ func rewriteBlockARM64(b *Block) bool {
 			b.resetWithControl(BlockARM64LE, cmp)
 			return true
 		}
+	case BlockARM64GEnoov:
+		// match: (GEnoov (InvertFlags cmp) yes no)
+		// result: (LEnoov cmp yes no)
+		for b.Controls[0].Op == OpARM64InvertFlags {
+			v_0 := b.Controls[0]
+			cmp := v_0.Args[0]
+			b.resetWithControl(BlockARM64LEnoov, cmp)
+			return true
+		}
 	case BlockARM64GT:
 		// match: (GT (CMPWconst [0] x:(ANDconst [c] y)) yes no)
 		// cond: x.Uses == 1
@@ -26845,7 +26854,7 @@ func rewriteBlockARM64(b *Block) bool {
 		}
 		// match: (GT (CMPconst [0] x:(ADDconst [c] y)) yes no)
 		// cond: x.Uses == 1
-		// result: (GT (CMNconst [c] y) yes no)
+		// result: (GTnoov (CMNconst [c] y) yes no)
 		for b.Controls[0].Op == OpARM64CMPconst {
 			v_0 := b.Controls[0]
 			if auxIntToInt64(v_0.AuxInt) != 0 {
@@ -26863,12 +26872,12 @@ func rewriteBlockARM64(b *Block) bool {
 			v0 := b.NewValue0(v_0.Pos, OpARM64CMNconst, types.TypeFlags)
 			v0.AuxInt = int64ToAuxInt(c)
 			v0.AddArg(y)
-			b.resetWithControl(BlockARM64GT, v0)
+			b.resetWithControl(BlockARM64GTnoov, v0)
 			return true
 		}
 		// match: (GT (CMPWconst [0] x:(ADDconst [c] y)) yes no)
 		// cond: x.Uses == 1
-		// result: (GT (CMNWconst [int32(c)] y) yes no)
+		// result: (GTnoov (CMNWconst [int32(c)] y) yes no)
 		for b.Controls[0].Op == OpARM64CMPWconst {
 			v_0 := b.Controls[0]
 			if auxIntToInt32(v_0.AuxInt) != 0 {
@@ -26886,12 +26895,12 @@ func rewriteBlockARM64(b *Block) bool {
 			v0 := b.NewValue0(v_0.Pos, OpARM64CMNWconst, types.TypeFlags)
 			v0.AuxInt = int32ToAuxInt(int32(c))
 			v0.AddArg(y)
-			b.resetWithControl(BlockARM64GT, v0)
+			b.resetWithControl(BlockARM64GTnoov, v0)
 			return true
 		}
 		// match: (GT (CMPconst [0] z:(ADD x y)) yes no)
 		// cond: z.Uses == 1
-		// result: (GT (CMN x y) yes no)
+		// result: (GTnoov (CMN x y) yes no)
 		for b.Controls[0].Op == OpARM64CMPconst {
 			v_0 := b.Controls[0]
 			if auxIntToInt64(v_0.AuxInt) != 0 {
@@ -26912,14 +26921,14 @@ func rewriteBlockARM64(b *Block) bool {
 				}
 				v0 := b.NewValue0(v_0.Pos, OpARM64CMN, types.TypeFlags)
 				v0.AddArg2(x, y)
-				b.resetWithControl(BlockARM64GT, v0)
+				b.resetWithControl(BlockARM64GTnoov, v0)
 				return true
 			}
 			break
 		}
 		// match: (GT (CMPWconst [0] z:(ADD x y)) yes no)
 		// cond: z.Uses == 1
-		// result: (GT (CMNW x y) yes no)
+		// result: (GTnoov (CMNW x y) yes no)
 		for b.Controls[0].Op == OpARM64CMPWconst {
 			v_0 := b.Controls[0]
 			if auxIntToInt32(v_0.AuxInt) != 0 {
@@ -26940,7 +26949,7 @@ func rewriteBlockARM64(b *Block) bool {
 				}
 				v0 := b.NewValue0(v_0.Pos, OpARM64CMNW, types.TypeFlags)
 				v0.AddArg2(x, y)
-				b.resetWithControl(BlockARM64GT, v0)
+				b.resetWithControl(BlockARM64GTnoov, v0)
 				return true
 			}
 			break
@@ -26987,7 +26996,7 @@ func rewriteBlockARM64(b *Block) bool {
 		}
 		// match: (GT (CMPconst [0] z:(MADD a x y)) yes no)
 		// cond: z.Uses==1
-		// result: (GT (CMN a (MUL <x.Type> x y)) yes no)
+		// result: (GTnoov (CMN a (MUL <x.Type> x y)) yes no)
 		for b.Controls[0].Op == OpARM64CMPconst {
 			v_0 := b.Controls[0]
 			if auxIntToInt64(v_0.AuxInt) != 0 {
@@ -27007,12 +27016,12 @@ func rewriteBlockARM64(b *Block) bool {
 			v1 := b.NewValue0(v_0.Pos, OpARM64MUL, x.Type)
 			v1.AddArg2(x, y)
 			v0.AddArg2(a, v1)
-			b.resetWithControl(BlockARM64GT, v0)
+			b.resetWithControl(BlockARM64GTnoov, v0)
 			return true
 		}
 		// match: (GT (CMPconst [0] z:(MSUB a x y)) yes no)
 		// cond: z.Uses==1
-		// result: (GT (CMP a (MUL <x.Type> x y)) yes no)
+		// result: (GTnoov (CMP a (MUL <x.Type> x y)) yes no)
 		for b.Controls[0].Op == OpARM64CMPconst {
 			v_0 := b.Controls[0]
 			if auxIntToInt64(v_0.AuxInt) != 0 {
@@ -27032,12 +27041,12 @@ func rewriteBlockARM64(b *Block) bool {
 			v1 := b.NewValue0(v_0.Pos, OpARM64MUL, x.Type)
 			v1.AddArg2(x, y)
 			v0.AddArg2(a, v1)
-			b.resetWithControl(BlockARM64GT, v0)
+			b.resetWithControl(BlockARM64GTnoov, v0)
 			return true
 		}
 		// match: (GT (CMPWconst [0] z:(MADDW a x y)) yes no)
 		// cond: z.Uses==1
-		// result: (GT (CMNW a (MULW <x.Type> x y)) yes no)
+		// result: (GTnoov (CMNW a (MULW <x.Type> x y)) yes no)
 		for b.Controls[0].Op == OpARM64CMPWconst {
 			v_0 := b.Controls[0]
 			if auxIntToInt32(v_0.AuxInt) != 0 {
@@ -27057,12 +27066,12 @@ func rewriteBlockARM64(b *Block) bool {
 			v1 := b.NewValue0(v_0.Pos, OpARM64MULW, x.Type)
 			v1.AddArg2(x, y)
 			v0.AddArg2(a, v1)
-			b.resetWithControl(BlockARM64GT, v0)
+			b.resetWithControl(BlockARM64GTnoov, v0)
 			return true
 		}
 		// match: (GT (CMPWconst [0] z:(MSUBW a x y)) yes no)
 		// cond: z.Uses==1
-		// result: (GT (CMPW a (MULW <x.Type> x y)) yes no)
+		// result: (GTnoov (CMPW a (MULW <x.Type> x y)) yes no)
 		for b.Controls[0].Op == OpARM64CMPWconst {
 			v_0 := b.Controls[0]
 			if auxIntToInt32(v_0.AuxInt) != 0 {
@@ -27082,7 +27091,7 @@ func rewriteBlockARM64(b *Block) bool {
 			v1 := b.NewValue0(v_0.Pos, OpARM64MULW, x.Type)
 			v1.AddArg2(x, y)
 			v0.AddArg2(a, v1)
-			b.resetWithControl(BlockARM64GT, v0)
+			b.resetWithControl(BlockARM64GTnoov, v0)
 			return true
 		}
 		// match: (GT (FlagEQ) yes no)
@@ -27126,6 +27135,15 @@ func rewriteBlockARM64(b *Block) bool {
 			b.resetWithControl(BlockARM64LT, cmp)
 			return true
 		}
+	case BlockARM64GTnoov:
+		// match: (GTnoov (InvertFlags cmp) yes no)
+		// result: (LTnoov cmp yes no)
+		for b.Controls[0].Op == OpARM64InvertFlags {
+			v_0 := b.Controls[0]
+			cmp := v_0.Args[0]
+			b.resetWithControl(BlockARM64LTnoov, cmp)
+			return true
+		}
 	case BlockIf:
 		// match: (If (Equal cc) yes no)
 		// result: (EQ cc yes no)
@@ -27351,7 +27369,7 @@ func rewriteBlockARM64(b *Block) bool {
 		}
 		// match: (LE (CMPconst [0] x:(ADDconst [c] y)) yes no)
 		// cond: x.Uses == 1
-		// result: (LE (CMNconst [c] y) yes no)
+		// result: (LEnoov (CMNconst [c] y) yes no)
 		for b.Controls[0].Op == OpARM64CMPconst {
 			v_0 := b.Controls[0]
 			if auxIntToInt64(v_0.AuxInt) != 0 {
@@ -27369,12 +27387,12 @@ func rewriteBlockARM64(b *Block) bool {
 			v0 := b.NewValue0(v_0.Pos, OpARM64CMNconst, types.TypeFlags)
 			v0.AuxInt = int64ToAuxInt(c)
 			v0.AddArg(y)
-			b.resetWithControl(BlockARM64LE, v0)
+			b.resetWithControl(BlockARM64LEnoov, v0)
 			return true
 		}
 		// match: (LE (CMPWconst [0] x:(ADDconst [c] y)) yes no)
 		// cond: x.Uses == 1
-		// result: (LE (CMNWconst [int32(c)] y) yes no)
+		// result: (LEnoov (CMNWconst [int32(c)] y) yes no)
 		for b.Controls[0].Op == OpARM64CMPWconst {
 			v_0 := b.Controls[0]
 			if auxIntToInt32(v_0.AuxInt) != 0 {
@@ -27392,12 +27410,12 @@ func rewriteBlockARM64(b *Block) bool {
 			v0 := b.NewValue0(v_0.Pos, OpARM64CMNWconst, types.TypeFlags)
 			v0.AuxInt = int32ToAuxInt(int32(c))
 			v0.AddArg(y)
-			b.resetWithControl(BlockARM64LE, v0)
+			b.resetWithControl(BlockARM64LEnoov, v0)
 			return true
 		}
 		// match: (LE (CMPconst [0] z:(ADD x y)) yes no)
 		// cond: z.Uses == 1
-		// result: (LE (CMN x y) yes no)
+		// result: (LEnoov (CMN x y) yes no)
 		for b.Controls[0].Op == OpARM64CMPconst {
 			v_0 := b.Controls[0]
 			if auxIntToInt64(v_0.AuxInt) != 0 {
@@ -27418,14 +27436,14 @@ func rewriteBlockARM64(b *Block) bool {
 				}
 				v0 := b.NewValue0(v_0.Pos, OpARM64CMN, types.TypeFlags)
 				v0.AddArg2(x, y)
-				b.resetWithControl(BlockARM64LE, v0)
+				b.resetWithControl(BlockARM64LEnoov, v0)
 				return true
 			}
 			break
 		}
 		// match: (LE (CMPWconst [0] z:(ADD x y)) yes no)
 		// cond: z.Uses == 1
-		// result: (LE (CMNW x y) yes no)
+		// result: (LEnoov (CMNW x y) yes no)
 		for b.Controls[0].Op == OpARM64CMPWconst {
 			v_0 := b.Controls[0]
 			if auxIntToInt32(v_0.AuxInt) != 0 {
@@ -27446,7 +27464,7 @@ func rewriteBlockARM64(b *Block) bool {
 				}
 				v0 := b.NewValue0(v_0.Pos, OpARM64CMNW, types.TypeFlags)
 				v0.AddArg2(x, y)
-				b.resetWithControl(BlockARM64LE, v0)
+				b.resetWithControl(BlockARM64LEnoov, v0)
 				return true
 			}
 			break
@@ -27493,7 +27511,7 @@ func rewriteBlockARM64(b *Block) bool {
 		}
 		// match: (LE (CMPconst [0] z:(MADD a x y)) yes no)
 		// cond: z.Uses==1
-		// result: (LE (CMN a (MUL <x.Type> x y)) yes no)
+		// result: (LEnoov (CMN a (MUL <x.Type> x y)) yes no)
 		for b.Controls[0].Op == OpARM64CMPconst {
 			v_0 := b.Controls[0]
 			if auxIntToInt64(v_0.AuxInt) != 0 {
@@ -27513,12 +27531,12 @@ func rewriteBlockARM64(b *Block) bool {
 			v1 := b.NewValue0(v_0.Pos, OpARM64MUL, x.Type)
 			v1.AddArg2(x, y)
 			v0.AddArg2(a, v1)
-			b.resetWithControl(BlockARM64LE, v0)
+			b.resetWithControl(BlockARM64LEnoov, v0)
 			return true
 		}
 		// match: (LE (CMPconst [0] z:(MSUB a x y)) yes no)
 		// cond: z.Uses==1
-		// result: (LE (CMP a (MUL <x.Type> x y)) yes no)
+		// result: (LEnoov (CMP a (MUL <x.Type> x y)) yes no)
 		for b.Controls[0].Op == OpARM64CMPconst {
 			v_0 := b.Controls[0]
 			if auxIntToInt64(v_0.AuxInt) != 0 {
@@ -27538,12 +27556,12 @@ func rewriteBlockARM64(b *Block) bool {
 			v1 := b.NewValue0(v_0.Pos, OpARM64MUL, x.Type)
 			v1.AddArg2(x, y)
 			v0.AddArg2(a, v1)
-			b.resetWithControl(BlockARM64LE, v0)
+			b.resetWithControl(BlockARM64LEnoov, v0)
 			return true
 		}
 		// match: (LE (CMPWconst [0] z:(MADDW a x y)) yes no)
 		// cond: z.Uses==1
-		// result: (LE (CMNW a (MULW <x.Type> x y)) yes no)
+		// result: (LEnoov (CMNW a (MULW <x.Type> x y)) yes no)
 		for b.Controls[0].Op == OpARM64CMPWconst {
 			v_0 := b.Controls[0]
 			if auxIntToInt32(v_0.AuxInt) != 0 {
@@ -27563,12 +27581,12 @@ func rewriteBlockARM64(b *Block) bool {
 			v1 := b.NewValue0(v_0.Pos, OpARM64MULW, x.Type)
 			v1.AddArg2(x, y)
 			v0.AddArg2(a, v1)
-			b.resetWithControl(BlockARM64LE, v0)
+			b.resetWithControl(BlockARM64LEnoov, v0)
 			return true
 		}
 		// match: (LE (CMPWconst [0] z:(MSUBW a x y)) yes no)
 		// cond: z.Uses==1
-		// result: (LE (CMPW a (MULW <x.Type> x y)) yes no)
+		// result: (LEnoov (CMPW a (MULW <x.Type> x y)) yes no)
 		for b.Controls[0].Op == OpARM64CMPWconst {
 			v_0 := b.Controls[0]
 			if auxIntToInt32(v_0.AuxInt) != 0 {
@@ -27588,7 +27606,7 @@ func rewriteBlockARM64(b *Block) bool {
 			v1 := b.NewValue0(v_0.Pos, OpARM64MULW, x.Type)
 			v1.AddArg2(x, y)
 			v0.AddArg2(a, v1)
-			b.resetWithControl(BlockARM64LE, v0)
+			b.resetWithControl(BlockARM64LEnoov, v0)
 			return true
 		}
 		// match: (LE (FlagEQ) yes no)
@@ -27631,6 +27649,15 @@ func rewriteBlockARM64(b *Block) bool {
 			b.resetWithControl(BlockARM64GE, cmp)
 			return true
 		}
+	case BlockARM64LEnoov:
+		// match: (LEnoov (InvertFlags cmp) yes no)
+		// result: (GEnoov cmp yes no)
+		for b.Controls[0].Op == OpARM64InvertFlags {
+			v_0 := b.Controls[0]
+			cmp := v_0.Args[0]
+			b.resetWithControl(BlockARM64GEnoov, cmp)
+			return true
+		}
 	case BlockARM64LT:
 		// match: (LT (CMPWconst [0] x:(ANDconst [c] y)) yes no)
 		// cond: x.Uses == 1
@@ -27736,7 +27763,7 @@ func rewriteBlockARM64(b *Block) bool {
 		}
 		// match: (LT (CMPconst [0] x:(ADDconst [c] y)) yes no)
 		// cond: x.Uses == 1
-		// result: (LT (CMNconst [c] y) yes no)
+		// result: (LTnoov (CMNconst [c] y) yes no)
 		for b.Controls[0].Op == OpARM64CMPconst {
 			v_0 := b.Controls[0]
 			if auxIntToInt64(v_0.AuxInt) != 0 {
@@ -27754,12 +27781,12 @@ func rewriteBlockARM64(b *Block) bool {
 			v0 := b.NewValue0(v_0.Pos, OpARM64CMNconst, types.TypeFlags)
 			v0.AuxInt = int64ToAuxInt(c)
 			v0.AddArg(y)
-			b.resetWithControl(BlockARM64LT, v0)
+			b.resetWithControl(BlockARM64LTnoov, v0)
 			return true
 		}
 		// match: (LT (CMPWconst [0] x:(ADDconst [c] y)) yes no)
 		// cond: x.Uses == 1
-		// result: (LT (CMNWconst [int32(c)] y) yes no)
+		// result: (LTnoov (CMNWconst [int32(c)] y) yes no)
 		for b.Controls[0].Op == OpARM64CMPWconst {
 			v_0 := b.Controls[0]
 			if auxIntToInt32(v_0.AuxInt) != 0 {
@@ -27777,12 +27804,12 @@ func rewriteBlockARM64(b *Block) bool {
 			v0 := b.NewValue0(v_0.Pos, OpARM64CMNWconst, types.TypeFlags)
 			v0.AuxInt = int32ToAuxInt(int32(c))
 			v0.AddArg(y)
-			b.resetWithControl(BlockARM64LT, v0)
+			b.resetWithControl(BlockARM64LTnoov, v0)
 			return true
 		}
 		// match: (LT (CMPconst [0] z:(ADD x y)) yes no)
 		// cond: z.Uses == 1
-		// result: (LT (CMN x y) yes no)
+		// result: (LTnoov (CMN x y) yes no)
 		for b.Controls[0].Op == OpARM64CMPconst {
 			v_0 := b.Controls[0]
 			if auxIntToInt64(v_0.AuxInt) != 0 {
@@ -27803,14 +27830,14 @@ func rewriteBlockARM64(b *Block) bool {
 				}
 				v0 := b.NewValue0(v_0.Pos, OpARM64CMN, types.TypeFlags)
 				v0.AddArg2(x, y)
-				b.resetWithControl(BlockARM64LT, v0)
+				b.resetWithControl(BlockARM64LTnoov, v0)
 				return true
 			}
 			break
 		}
 		// match: (LT (CMPWconst [0] z:(ADD x y)) yes no)
 		// cond: z.Uses == 1
-		// result: (LT (CMNW x y) yes no)
+		// result: (LTnoov (CMNW x y) yes no)
 		for b.Controls[0].Op == OpARM64CMPWconst {
 			v_0 := b.Controls[0]
 			if auxIntToInt32(v_0.AuxInt) != 0 {
@@ -27831,7 +27858,7 @@ func rewriteBlockARM64(b *Block) bool {
 				}
 				v0 := b.NewValue0(v_0.Pos, OpARM64CMNW, types.TypeFlags)
 				v0.AddArg2(x, y)
-				b.resetWithControl(BlockARM64LT, v0)
+				b.resetWithControl(BlockARM64LTnoov, v0)
 				return true
 			}
 			break
@@ -27878,7 +27905,7 @@ func rewriteBlockARM64(b *Block) bool {
 		}
 		// match: (LT (CMPconst [0] z:(MADD a x y)) yes no)
 		// cond: z.Uses==1
-		// result: (LT (CMN a (MUL <x.Type> x y)) yes no)
+		// result: (LTnoov (CMN a (MUL <x.Type> x y)) yes no)
 		for b.Controls[0].Op == OpARM64CMPconst {
 			v_0 := b.Controls[0]
 			if auxIntToInt64(v_0.AuxInt) != 0 {
@@ -27898,12 +27925,12 @@ func rewriteBlockARM64(b *Block) bool {
 			v1 := b.NewValue0(v_0.Pos, OpARM64MUL, x.Type)
 			v1.AddArg2(x, y)
 			v0.AddArg2(a, v1)
-			b.resetWithControl(BlockARM64LT, v0)
+			b.resetWithControl(BlockARM64LTnoov, v0)
 			return true
 		}
 		// match: (LT (CMPconst [0] z:(MSUB a x y)) yes no)
 		// cond: z.Uses==1
-		// result: (LT (CMP a (MUL <x.Type> x y)) yes no)
+		// result: (LTnoov (CMP a (MUL <x.Type> x y)) yes no)
 		for b.Controls[0].Op == OpARM64CMPconst {
 			v_0 := b.Controls[0]
 			if auxIntToInt64(v_0.AuxInt) != 0 {
@@ -27923,12 +27950,12 @@ func rewriteBlockARM64(b *Block) bool {
 			v1 := b.NewValue0(v_0.Pos, OpARM64MUL, x.Type)
 			v1.AddArg2(x, y)
 			v0.AddArg2(a, v1)
-			b.resetWithControl(BlockARM64LT, v0)
+			b.resetWithControl(BlockARM64LTnoov, v0)
 			return true
 		}
 		// match: (LT (CMPWconst [0] z:(MADDW a x y)) yes no)
 		// cond: z.Uses==1
-		// result: (LT (CMNW a (MULW <x.Type> x y)) yes no)
+		// result: (LTnoov (CMNW a (MULW <x.Type> x y)) yes no)
 		for b.Controls[0].Op == OpARM64CMPWconst {
 			v_0 := b.Controls[0]
 			if auxIntToInt32(v_0.AuxInt) != 0 {
@@ -27948,12 +27975,12 @@ func rewriteBlockARM64(b *Block) bool {
 			v1 := b.NewValue0(v_0.Pos, OpARM64MULW, x.Type)
 			v1.AddArg2(x, y)
 			v0.AddArg2(a, v1)
-			b.resetWithControl(BlockARM64LT, v0)
+			b.resetWithControl(BlockARM64LTnoov, v0)
 			return true
 		}
 		// match: (LT (CMPWconst [0] z:(MSUBW a x y)) yes no)
 		// cond: z.Uses==1
-		// result: (LT (CMPW a (MULW <x.Type> x y)) yes no)
+		// result: (LTnoov (CMPW a (MULW <x.Type> x y)) yes no)
 		for b.Controls[0].Op == OpARM64CMPWconst {
 			v_0 := b.Controls[0]
 			if auxIntToInt32(v_0.AuxInt) != 0 {
@@ -27973,7 +28000,7 @@ func rewriteBlockARM64(b *Block) bool {
 			v1 := b.NewValue0(v_0.Pos, OpARM64MULW, x.Type)
 			v1.AddArg2(x, y)
 			v0.AddArg2(a, v1)
-			b.resetWithControl(BlockARM64LT, v0)
+			b.resetWithControl(BlockARM64LTnoov, v0)
 			return true
 		}
 		// match: (LT (CMPWconst [0] x) yes no)
@@ -28041,6 +28068,15 @@ func rewriteBlockARM64(b *Block) bool {
 			b.resetWithControl(BlockARM64GT, cmp)
 			return true
 		}
+	case BlockARM64LTnoov:
+		// match: (LTnoov (InvertFlags cmp) yes no)
+		// result: (GTnoov cmp yes no)
+		for b.Controls[0].Op == OpARM64InvertFlags {
+			v_0 := b.Controls[0]
+			cmp := v_0.Args[0]
+			b.resetWithControl(BlockARM64GTnoov, cmp)
+			return true
+		}
 	case BlockARM64NE:
 		// match: (NE (CMPWconst [0] x:(ANDconst [c] y)) yes no)
 		// cond: x.Uses == 1
diff --git a/src/cmd/compile/internal/ssa/rewriteCond_test.go b/src/cmd/compile/internal/ssa/rewriteCond_test.go
new file mode 100644
index 0000000000..b8feff77e5
--- /dev/null
+++ b/src/cmd/compile/internal/ssa/rewriteCond_test.go
@@ -0,0 +1,536 @@
+// Copyright 2020 The Go Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+
+package ssa
+
+import (
+	"math"
+	"math/rand"
+	"testing"
+	"runtime"
+)
+
+var (
+	x64   int64 = math.MaxInt64 - 2
+	x64b  int64 = math.MaxInt64 - 2
+	x64c  int64 = math.MaxInt64 - 2
+	y64   int64 = math.MinInt64 + 1
+	x32   int32 = math.MaxInt32 - 2
+	x32b  int32 = math.MaxInt32 - 2
+	y32   int32 = math.MinInt32 + 1
+	one64 int64 = 1
+	one32 int32 = 1
+	v64   int64 = 11 // ensure it's not 2**n +/- 1
+	v64_n int64 = -11
+	v32   int32 = 11
+	v32_n int32 = -11
+)
+
+var crTests = []struct {
+	name string
+	tf   func(t *testing.T)
+}{
+	{"AddConst64", testAddConst64},
+	{"AddConst32", testAddConst32},
+	{"AddVar64", testAddVar64},
+	{"AddVar32", testAddVar32},
+	{"MAddVar64", testMAddVar64},
+	{"MAddVar32", testMAddVar32},
+	{"MSubVar64", testMSubVar64},
+	{"MSubVar32", testMSubVar32},
+}
+
+var crBenches = []struct {
+	name string
+	bf   func(b *testing.B)
+}{
+	{"SoloJump", benchSoloJump},
+	{"CombJump", benchCombJump},
+}
+
+// Test int32/int64's add/sub/madd/msub operations with boundary values to
+// ensure the optimization to 'comparing to zero' expressions of if-statements
+// yield expected results.
+// 32 rewriting rules are covered. At least two scenarios for "Canonicalize
+// the order of arguments to comparisons", which helps with CSE, are covered.
+// The tedious if-else structures are necessary to ensure all concerned rules
+// and machine code sequences are covered.
+// It's for arm64 initially, please see https://github.com/golang/go/issues/38740
+func TestCondRewrite(t *testing.T) {
+	if runtime.GOARCH == "arm" {
+		t.Skip("fix on arm expected!")
+	}
+	for _, test := range crTests {
+		t.Run(test.name, test.tf)
+	}
+}
+
+// Profile the aforementioned optimization from two angles:
+//   SoloJump: generated branching code has one 'jump', for '<' and '>='
+//   CombJump: generated branching code has two consecutive 'jump', for '<=' and '>'
+// We expect that 'CombJump' is generally on par with the non-optimized code, and
+// 'SoloJump' demonstrates some improvement.
+// It's for arm64 initially, please see https://github.com/golang/go/issues/38740
+func BenchmarkCondRewrite(b *testing.B) {
+	for _, bench := range crBenches {
+		b.Run(bench.name, bench.bf)
+	}
+}
+
+// var +/- const
+func testAddConst64(t *testing.T) {
+	if x64+11 < 0 {
+	} else {
+		t.Errorf("'%#x + 11 < 0' failed", x64)
+	}
+
+	if x64+13 <= 0 {
+	} else {
+		t.Errorf("'%#x + 13 <= 0' failed", x64)
+	}
+
+	if y64-11 > 0 {
+	} else {
+		t.Errorf("'%#x - 11 > 0' failed", y64)
+	}
+
+	if y64-13 >= 0 {
+	} else {
+		t.Errorf("'%#x - 13 >= 0' failed", y64)
+	}
+
+	if x64+19 > 0 {
+		t.Errorf("'%#x + 19 > 0' failed", x64)
+	}
+
+	if x64+23 >= 0 {
+		t.Errorf("'%#x + 23 >= 0' failed", x64)
+	}
+
+	if y64-19 < 0 {
+		t.Errorf("'%#x - 19 < 0' failed", y64)
+	}
+
+	if y64-23 <= 0 {
+		t.Errorf("'%#x - 23 <= 0' failed", y64)
+	}
+}
+
+// 32-bit var +/- const
+func testAddConst32(t *testing.T) {
+	if x32+11 < 0 {
+	} else {
+		t.Errorf("'%#x + 11 < 0' failed", x32)
+	}
+
+	if x32+13 <= 0 {
+	} else {
+		t.Errorf("'%#x + 13 <= 0' failed", x32)
+	}
+
+	if y32-11 > 0 {
+	} else {
+		t.Errorf("'%#x - 11 > 0' failed", y32)
+	}
+
+	if y32-13 >= 0 {
+	} else {
+		t.Errorf("'%#x - 13 >= 0' failed", y32)
+	}
+
+	if x32+19 > 0 {
+		t.Errorf("'%#x + 19 > 0' failed", x32)
+	}
+
+	if x32+23 >= 0 {
+		t.Errorf("'%#x + 23 >= 0' failed", x32)
+	}
+
+	if y32-19 < 0 {
+		t.Errorf("'%#x - 19 < 0' failed", y32)
+	}
+
+	if y32-23 <= 0 {
+		t.Errorf("'%#x - 23 <= 0' failed", y32)
+	}
+}
+
+// var + var
+func testAddVar64(t *testing.T) {
+	if x64+v64 < 0 {
+	} else {
+		t.Errorf("'%#x + %#x < 0' failed", x64, v64)
+	}
+
+	if x64+v64 <= 0 {
+	} else {
+		t.Errorf("'%#x + %#x <= 0' failed", x64, v64)
+	}
+
+	if y64+v64_n > 0 {
+	} else {
+		t.Errorf("'%#x + %#x > 0' failed", y64, v64_n)
+	}
+
+	if y64+v64_n >= 0 {
+	} else {
+		t.Errorf("'%#x + %#x >= 0' failed", y64, v64_n)
+	}
+
+	if x64+v64 > 0 {
+		t.Errorf("'%#x + %#x > 0' failed", x64, v64)
+	}
+
+	if x64+v64 >= 0 {
+		t.Errorf("'%#x + %#x >= 0' failed", x64, v64)
+	}
+
+	if y64+v64_n < 0 {
+		t.Errorf("'%#x + %#x < 0' failed", y64, v64_n)
+	}
+
+	if y64+v64_n <= 0 {
+		t.Errorf("'%#x + %#x <= 0' failed", y64, v64_n)
+	}
+}
+
+// 32-bit var+var
+func testAddVar32(t *testing.T) {
+	if x32+v32 < 0 {
+	} else {
+		t.Errorf("'%#x + %#x < 0' failed", x32, v32)
+	}
+
+	if x32+v32 <= 0 {
+	} else {
+		t.Errorf("'%#x + %#x <= 0' failed", x32, v32)
+	}
+
+	if y32+v32_n > 0 {
+	} else {
+		t.Errorf("'%#x + %#x > 0' failed", y32, v32_n)
+	}
+
+	if y32+v32_n >= 0 {
+	} else {
+		t.Errorf("'%#x + %#x >= 0' failed", y32, v32_n)
+	}
+
+	if x32+v32 > 0 {
+		t.Errorf("'%#x + %#x > 0' failed", x32, v32)
+	}
+
+	if x32+v32 >= 0 {
+		t.Errorf("'%#x + %#x >= 0' failed", x32, v32)
+	}
+
+	if y32+v32_n < 0 {
+		t.Errorf("'%#x + %#x < 0' failed", y32, v32_n)
+	}
+
+	if y32+v32_n <= 0 {
+		t.Errorf("'%#x + %#x <= 0' failed", y32, v32_n)
+	}
+}
+
+// multiply-add
+func testMAddVar64(t *testing.T) {
+	if x64+v64*one64 < 0 {
+	} else {
+		t.Errorf("'%#x + %#x*1 < 0' failed", x64, v64)
+	}
+
+	if x64+v64*one64 <= 0 {
+	} else {
+		t.Errorf("'%#x + %#x*1 <= 0' failed", x64, v64)
+	}
+
+	if y64+v64_n*one64 > 0 {
+	} else {
+		t.Errorf("'%#x + %#x*1 > 0' failed", y64, v64_n)
+	}
+
+	if y64+v64_n*one64 >= 0 {
+	} else {
+		t.Errorf("'%#x + %#x*1 >= 0' failed", y64, v64_n)
+	}
+
+	if x64+v64*one64 > 0 {
+		t.Errorf("'%#x + %#x*1 > 0' failed", x64, v64)
+	}
+
+	if x64+v64*one64 >= 0 {
+		t.Errorf("'%#x + %#x*1 >= 0' failed", x64, v64)
+	}
+
+	if y64+v64_n*one64 < 0 {
+		t.Errorf("'%#x + %#x*1 < 0' failed", y64, v64_n)
+	}
+
+	if y64+v64_n*one64 <= 0 {
+		t.Errorf("'%#x + %#x*1 <= 0' failed", y64, v64_n)
+	}
+}
+
+// 32-bit multiply-add
+func testMAddVar32(t *testing.T) {
+	if x32+v32*one32 < 0 {
+	} else {
+		t.Errorf("'%#x + %#x*1 < 0' failed", x32, v32)
+	}
+
+	if x32+v32*one32 <= 0 {
+	} else {
+		t.Errorf("'%#x + %#x*1 <= 0' failed", x32, v32)
+	}
+
+	if y32+v32_n*one32 > 0 {
+	} else {
+		t.Errorf("'%#x + %#x*1 > 0' failed", y32, v32_n)
+	}
+
+	if y32+v32_n*one32 >= 0 {
+	} else {
+		t.Errorf("'%#x + %#x*1 >= 0' failed", y32, v32_n)
+	}
+
+	if x32+v32*one32 > 0 {
+		t.Errorf("'%#x + %#x*1 > 0' failed", x32, v32)
+	}
+
+	if x32+v32*one32 >= 0 {
+		t.Errorf("'%#x + %#x*1 >= 0' failed", x32, v32)
+	}
+
+	if y32+v32_n*one32 < 0 {
+		t.Errorf("'%#x + %#x*1 < 0' failed", y32, v32_n)
+	}
+
+	if y32+v32_n*one32 <= 0 {
+		t.Errorf("'%#x + %#x*1 <= 0' failed", y32, v32_n)
+	}
+}
+
+// multiply-sub
+func testMSubVar64(t *testing.T) {
+	if x64-v64_n*one64 < 0 {
+	} else {
+		t.Errorf("'%#x - %#x*1 < 0' failed", x64, v64_n)
+	}
+
+	if x64-v64_n*one64 <= 0 {
+	} else {
+		t.Errorf("'%#x - %#x*1 <= 0' failed", x64, v64_n)
+	}
+
+	if y64-v64*one64 > 0 {
+	} else {
+		t.Errorf("'%#x - %#x*1 > 0' failed", y64, v64)
+	}
+
+	if y64-v64*one64 >= 0 {
+	} else {
+		t.Errorf("'%#x - %#x*1 >= 0' failed", y64, v64)
+	}
+
+	if x64-v64_n*one64 > 0 {
+		t.Errorf("'%#x - %#x*1 > 0' failed", x64, v64_n)
+	}
+
+	if x64-v64_n*one64 >= 0 {
+		t.Errorf("'%#x - %#x*1 >= 0' failed", x64, v64_n)
+	}
+
+	if y64-v64*one64 < 0 {
+		t.Errorf("'%#x - %#x*1 < 0' failed", y64, v64)
+	}
+
+	if y64-v64*one64 <= 0 {
+		t.Errorf("'%#x - %#x*1 <= 0' failed", y64, v64)
+	}
+
+	if x64-x64b*one64 < 0 {
+		t.Errorf("'%#x - %#x*1 < 0' failed", x64, x64b)
+	}
+
+	if x64-x64b*one64 >= 0 {
+	} else {
+		t.Errorf("'%#x - %#x*1 >= 0' failed", x64, x64b)
+	}
+}
+
+// 32-bit multiply-sub
+func testMSubVar32(t *testing.T) {
+	if x32-v32_n*one32 < 0 {
+	} else {
+		t.Errorf("'%#x - %#x*1 < 0' failed", x32, v32_n)
+	}
+
+	if x32-v32_n*one32 <= 0 {
+	} else {
+		t.Errorf("'%#x - %#x*1 <= 0' failed", x32, v32_n)
+	}
+
+	if y32-v32*one32 > 0 {
+	} else {
+		t.Errorf("'%#x - %#x*1 > 0' failed", y32, v32)
+	}
+
+	if y32-v32*one32 >= 0 {
+	} else {
+		t.Errorf("'%#x - %#x*1 >= 0' failed", y32, v32)
+	}
+
+	if x32-v32_n*one32 > 0 {
+		t.Errorf("'%#x - %#x*1 > 0' failed", x32, v32_n)
+	}
+
+	if x32-v32_n*one32 >= 0 {
+		t.Errorf("'%#x - %#x*1 >= 0' failed", x32, v32_n)
+	}
+
+	if y32-v32*one32 < 0 {
+		t.Errorf("'%#x - %#x*1 < 0' failed", y32, v32)
+	}
+
+	if y32-v32*one32 <= 0 {
+		t.Errorf("'%#x - %#x*1 <= 0' failed", y32, v32)
+	}
+
+	if x32-x32b*one32 < 0 {
+		t.Errorf("'%#x - %#x*1 < 0' failed", x32, x32b)
+	}
+
+	if x32-x32b*one32 >= 0 {
+	} else {
+		t.Errorf("'%#x - %#x*1 >= 0' failed", x32, x32b)
+	}
+}
+
+var rnd = rand.New(rand.NewSource(0))
+var sink int64
+
+func benchSoloJump(b *testing.B) {
+	r1 := x64
+	r2 := x64b
+	r3 := x64c
+	r4 := y64
+	d := rnd.Int63n(10)
+
+	// 6 out 10 conditions evaluate to true
+	for i := 0; i < b.N; i++ {
+		if r1+r2 < 0 {
+			d *= 2
+			d /= 2
+		}
+
+		if r1+r3 >= 0 {
+			d *= 2
+			d /= 2
+		}
+
+		if r1+r2*one64 < 0 {
+			d *= 2
+			d /= 2
+		}
+
+		if r2+r3*one64 >= 0 {
+			d *= 2
+			d /= 2
+		}
+
+		if r1-r2*v64 >= 0 {
+			d *= 2
+			d /= 2
+		}
+
+		if r3-r4*v64 < 0 {
+			d *= 2
+			d /= 2
+		}
+
+		if r1+11 < 0 {
+			d *= 2
+			d /= 2
+		}
+
+		if r1+13 >= 0 {
+			d *= 2
+			d /= 2
+		}
+
+		if r4-17 < 0 {
+			d *= 2
+			d /= 2
+		}
+
+		if r4-19 >= 0 {
+			d *= 2
+			d /= 2
+		}
+	}
+	sink = d
+}
+
+func benchCombJump(b *testing.B) {
+	r1 := x64
+	r2 := x64b
+	r3 := x64c
+	r4 := y64
+	d := rnd.Int63n(10)
+
+	// 6 out 10 conditions evaluate to true
+	for i := 0; i < b.N; i++ {
+		if r1+r2 <= 0 {
+			d *= 2
+			d /= 2
+		}
+
+		if r1+r3 > 0 {
+			d *= 2
+			d /= 2
+		}
+
+		if r1+r2*one64 <= 0 {
+			d *= 2
+			d /= 2
+		}
+
+		if r2+r3*one64 > 0 {
+			d *= 2
+			d /= 2
+		}
+
+		if r1-r2*v64 > 0 {
+			d *= 2
+			d /= 2
+		}
+
+		if r3-r4*v64 <= 0 {
+			d *= 2
+			d /= 2
+		}
+
+		if r1+11 <= 0 {
+			d *= 2
+			d /= 2
+		}
+
+		if r1+13 > 0 {
+			d *= 2
+			d /= 2
+		}
+
+		if r4-17 <= 0 {
+			d *= 2
+			d /= 2
+		}
+
+		if r4-19 > 0 {
+			d *= 2
+			d /= 2
+		}
+	}
+	sink = d
+}
diff --git a/src/cmd/compile/internal/x86/ssa.go b/src/cmd/compile/internal/x86/ssa.go
index 0c7e5bdb97..2de978c28a 100644
--- a/src/cmd/compile/internal/x86/ssa.go
+++ b/src/cmd/compile/internal/x86/ssa.go
@@ -885,11 +885,11 @@ var blockJump = [...]struct {
 	ssa.Block386NAN: {x86.AJPS, x86.AJPC},
 }
 
-var eqfJumps = [2][2]gc.FloatingEQNEJump{
+var eqfJumps = [2][2]gc.IndexJump{
 	{{Jump: x86.AJNE, Index: 1}, {Jump: x86.AJPS, Index: 1}}, // next == b.Succs[0]
 	{{Jump: x86.AJNE, Index: 1}, {Jump: x86.AJPC, Index: 0}}, // next == b.Succs[1]
 }
-var nefJumps = [2][2]gc.FloatingEQNEJump{
+var nefJumps = [2][2]gc.IndexJump{
 	{{Jump: x86.AJNE, Index: 0}, {Jump: x86.AJPC, Index: 1}}, // next == b.Succs[0]
 	{{Jump: x86.AJNE, Index: 0}, {Jump: x86.AJPS, Index: 0}}, // next == b.Succs[1]
 }
@@ -929,10 +929,10 @@ func ssaGenBlock(s *gc.SSAGenState, b, next *ssa.Block) {
 		p.To.Sym = b.Aux.(*obj.LSym)
 
 	case ssa.Block386EQF:
-		s.FPJump(b, next, &eqfJumps)
+		s.CombJump(b, next, &eqfJumps)
 
 	case ssa.Block386NEF:
-		s.FPJump(b, next, &nefJumps)
+		s.CombJump(b, next, &nefJumps)
 
 	case ssa.Block386EQ, ssa.Block386NE,
 		ssa.Block386LT, ssa.Block386GE,
diff --git a/test/codegen/comparisons.go b/test/codegen/comparisons.go
index c020ea8eb7..eb2f3317c9 100644
--- a/test/codegen/comparisons.go
+++ b/test/codegen/comparisons.go
@@ -248,3 +248,139 @@ func CmpLogicalToZero(a, b, c uint32, d, e uint64) uint64 {
 	}
 	return 0
 }
+
+// The following CmpToZero_ex* check that cmp|cmn with bmi|bpl are generated for
+// 'comparing to zero' expressions
+
+// var + const
+func CmpToZero_ex1(a int64, e int32) int {
+	// arm64:`CMN`,-`ADD`,`(BMI|BPL)`
+	if a+3 < 0 {
+		return 1
+	}
+
+	// arm64:`CMN`,-`ADD`,`BEQ`,`(BMI|BPL)`
+	if a+5 <= 0 {
+		return 1
+	}
+
+	// arm64:`CMN`,-`ADD`,`(BMI|BPL)`
+	if a+13 >= 0 {
+		return 2
+	}
+
+	// arm64:`CMP`,-`SUB`,`(BMI|BPL)`
+	if a-7 < 0 {
+		return 3
+	}
+
+	// arm64:`CMP`,-`SUB`,`(BMI|BPL)`
+	if a-11 >= 0 {
+		return 4
+	}
+
+	// arm64:`CMP`,-`SUB`,`BEQ`,`(BMI|BPL)`
+	if a-19 > 0 {
+		return 4
+	}
+
+	// arm64:`CMNW`,-`ADDW`,`(BMI|BPL)`
+	if e+3 < 0 {
+		return 5
+	}
+
+	// arm64:`CMNW`,-`ADDW`,`(BMI|BPL)`
+	if e+13 >= 0 {
+		return 6
+	}
+
+	// arm64:`CMPW`,-`SUBW`,`(BMI|BPL)`
+	if e-7 < 0 {
+		return 7
+	}
+
+	// arm64:`CMPW`,-`SUBW`,`(BMI|BPL)`
+	if e-11 >= 0 {
+		return 8
+	}
+
+	return 0
+}
+
+// var + var
+// TODO: optimize 'var - var'
+func CmpToZero_ex2(a, b, c int64, e, f, g int32) int {
+	// arm64:`CMN`,-`ADD`,`(BMI|BPL)`
+	if a+b < 0 {
+		return 1
+	}
+
+	// arm64:`CMN`,-`ADD`,`BEQ`,`(BMI|BPL)`
+	if a+c <= 0 {
+		return 1
+	}
+
+	// arm64:`CMN`,-`ADD`,`(BMI|BPL)`
+	if b+c >= 0 {
+		return 2
+	}
+
+	// arm64:`CMNW`,-`ADDW`,`(BMI|BPL)`
+	if e+f < 0 {
+		return 5
+	}
+
+	// arm64:`CMNW`,-`ADDW`,`(BMI|BPL)`
+	if f+g >= 0 {
+		return 6
+	}
+	return 0
+}
+
+// var + var*var
+func CmpToZero_ex3(a, b, c, d int64, e, f, g, h int32) int {
+	// arm64:`CMN`,-`MADD`,`MUL`,`(BMI|BPL)`
+	if a+b*c < 0 {
+		return 1
+	}
+
+	// arm64:`CMN`,-`MADD`,`MUL`,`(BMI|BPL)`
+	if b+c*d >= 0 {
+		return 2
+	}
+
+	// arm64:`CMNW`,-`MADDW`,`MULW`,`BEQ`,`(BMI|BPL)`
+	if e+f*g > 0 {
+		return 5
+	}
+
+	// arm64:`CMNW`,-`MADDW`,`MULW`,`BEQ`,`(BMI|BPL)`
+	if f+g*h <= 0 {
+		return 6
+	}
+	return 0
+}
+
+// var - var*var
+func CmpToZero_ex4(a, b, c, d int64, e, f, g, h int32) int {
+	// arm64:`CMP`,-`MSUB`,`MUL`,`BEQ`,`(BMI|BPL)`
+	if a-b*c > 0 {
+		return 1
+	}
+
+	// arm64:`CMP`,-`MSUB`,`MUL`,`(BMI|BPL)`
+	if b-c*d >= 0 {
+		return 2
+	}
+
+	// arm64:`CMPW`,-`MSUBW`,`MULW`,`(BMI|BPL)`
+	if e-f*g < 0 {
+		return 5
+	}
+
+	// arm64:`CMPW`,-`MSUBW`,`MULW`,`(BMI|BPL)`
+	if f-g*h >= 0 {
+		return 6
+	}
+	return 0
+}