cmd/compile: optimize AMD64's bit wise operation

Currently "arr[idx] |= 0x80" is compiled to MOVLload->BTSL->MOVLstore.
And this CL optimizes it to a single BTSLconstmodify. Other bit wise
operations with a direct memory operand are also implemented.

1. The size of the executable bin/go decreases about 4KB, and the total size
of pkg/linux_amd64 (excluding cmd/compile) decreases about 0.6KB.

2. There a little improvement in the go1 benchmark test (excluding noise).
name                     old time/op    new time/op    delta
BinaryTree17-4              2.66s ± 4%     2.66s ± 3%    ~     (p=0.596 n=49+49)
Fannkuch11-4                2.38s ± 2%     2.32s ± 2%  -2.69%  (p=0.000 n=50+50)
FmtFprintfEmpty-4          42.7ns ± 4%    43.2ns ± 7%  +1.31%  (p=0.009 n=50+50)
FmtFprintfString-4         71.0ns ± 5%    72.0ns ± 3%  +1.33%  (p=0.000 n=50+50)
FmtFprintfInt-4            80.7ns ± 4%    80.6ns ± 3%    ~     (p=0.931 n=50+50)
FmtFprintfIntInt-4          125ns ± 3%     126ns ± 4%    ~     (p=0.051 n=50+50)
FmtFprintfPrefixedInt-4     158ns ± 1%     142ns ± 3%  -9.84%  (p=0.000 n=36+50)
FmtFprintfFloat-4           215ns ± 4%     212ns ± 4%  -1.23%  (p=0.002 n=50+50)
FmtManyArgs-4               519ns ± 3%     510ns ± 3%  -1.77%  (p=0.000 n=50+50)
GobDecode-4                6.49ms ± 6%    6.52ms ± 5%    ~     (p=0.866 n=50+50)
GobEncode-4                5.93ms ± 8%    6.01ms ± 7%    ~     (p=0.076 n=50+50)
Gzip-4                      222ms ± 4%     224ms ± 8%  +0.80%  (p=0.001 n=50+50)
Gunzip-4                   36.6ms ± 5%    36.4ms ± 4%    ~     (p=0.093 n=50+50)
HTTPClientServer-4         59.1µs ± 1%    58.9µs ± 2%  -0.24%  (p=0.039 n=49+48)
JSONEncode-4               9.23ms ± 4%    9.21ms ± 5%    ~     (p=0.244 n=50+50)
JSONDecode-4               48.8ms ± 4%    48.7ms ± 4%    ~     (p=0.653 n=50+50)
Mandelbrot200-4            3.81ms ± 4%    3.80ms ± 3%    ~     (p=0.834 n=50+50)
GoParse-4                  3.20ms ± 5%    3.19ms ± 5%    ~     (p=0.494 n=50+50)
RegexpMatchEasy0_32-4      78.1ns ± 2%    77.4ns ± 3%  -0.86%  (p=0.005 n=50+50)
RegexpMatchEasy0_1K-4       233ns ± 3%     233ns ± 3%    ~     (p=0.074 n=50+50)
RegexpMatchEasy1_32-4      74.2ns ± 3%    73.4ns ± 3%  -1.06%  (p=0.000 n=50+50)
RegexpMatchEasy1_1K-4       369ns ± 2%     364ns ± 4%  -1.41%  (p=0.000 n=36+50)
RegexpMatchMedium_32-4      109ns ± 4%     107ns ± 3%  -2.06%  (p=0.001 n=50+50)
RegexpMatchMedium_1K-4     31.5µs ± 3%    30.8µs ± 3%  -2.20%  (p=0.000 n=50+50)
RegexpMatchHard_32-4       1.57µs ± 3%    1.56µs ± 2%  -0.57%  (p=0.016 n=50+50)
RegexpMatchHard_1K-4       47.4µs ± 4%    47.0µs ± 3%  -0.82%  (p=0.008 n=50+50)
Revcomp-4                   414ms ± 7%     412ms ± 7%    ~     (p=0.285 n=50+50)
Template-4                 64.3ms ± 4%    62.7ms ± 3%  -2.44%  (p=0.000 n=50+50)
TimeParse-4                 316ns ± 3%     313ns ± 3%    ~     (p=0.122 n=50+50)
TimeFormat-4                291ns ± 3%     293ns ± 3%  +0.80%  (p=0.001 n=50+50)
[Geo mean]                 46.5µs         46.2µs       -0.81%

name                     old speed      new speed      delta
GobDecode-4               118MB/s ± 6%   118MB/s ± 5%    ~     (p=0.863 n=50+50)
GobEncode-4               130MB/s ± 9%   128MB/s ± 8%    ~     (p=0.076 n=50+50)
Gzip-4                   87.4MB/s ± 4%  86.8MB/s ± 7%  -0.78%  (p=0.002 n=50+50)
Gunzip-4                  531MB/s ± 5%   533MB/s ± 4%    ~     (p=0.093 n=50+50)
JSONEncode-4              210MB/s ± 4%   211MB/s ± 5%    ~     (p=0.247 n=50+50)
JSONDecode-4             39.8MB/s ± 4%  39.9MB/s ± 4%    ~     (p=0.654 n=50+50)
GoParse-4                18.1MB/s ± 5%  18.2MB/s ± 5%    ~     (p=0.493 n=50+50)
RegexpMatchEasy0_32-4     410MB/s ± 2%   413MB/s ± 3%  +0.86%  (p=0.004 n=50+50)
RegexpMatchEasy0_1K-4    4.39GB/s ± 3%  4.38GB/s ± 3%    ~     (p=0.063 n=50+50)
RegexpMatchEasy1_32-4     432MB/s ± 3%   436MB/s ± 3%  +1.07%  (p=0.000 n=50+50)
RegexpMatchEasy1_1K-4    2.77GB/s ± 2%  2.81GB/s ± 4%  +1.46%  (p=0.000 n=36+50)
RegexpMatchMedium_32-4   9.16MB/s ± 3%  9.35MB/s ± 4%  +2.09%  (p=0.001 n=50+50)
RegexpMatchMedium_1K-4   32.5MB/s ± 3%  33.2MB/s ± 3%  +2.25%  (p=0.000 n=50+50)
RegexpMatchHard_32-4     20.4MB/s ± 3%  20.5MB/s ± 2%  +0.56%  (p=0.017 n=50+50)
RegexpMatchHard_1K-4     21.6MB/s ± 4%  21.8MB/s ± 3%  +0.83%  (p=0.008 n=50+50)
Revcomp-4                 613MB/s ± 4%   618MB/s ± 7%    ~     (p=0.152 n=48+50)
Template-4               30.2MB/s ± 4%  30.9MB/s ± 3%  +2.49%  (p=0.000 n=50+50)
[Geo mean]                127MB/s        128MB/s       +0.64%

Change-Id: If405198283855d75697f66cf894b2bef458f620e
Reviewed-on: https://go-review.googlesource.com/135422
Reviewed-by: Keith Randall <khr@golang.org>

6 files changed, 1852 insertions, 47 deletions

diff --git a/src/cmd/compile/internal/amd64/ssa.go b/src/cmd/compile/internal/amd64/ssa.go
index 818bc35941..b4c4b1f4cd 100644
--- a/src/cmd/compile/internal/amd64/ssa.go
+++ b/src/cmd/compile/internal/amd64/ssa.go
@@ -695,6 +695,7 @@ func ssaGenValue(s *gc.SSAGenState, v *ssa.Value) {
 		p.To.Type = obj.TYPE_REG
 		p.To.Reg = v.Reg()
 	case ssa.OpAMD64MOVQstore, ssa.OpAMD64MOVSSstore, ssa.OpAMD64MOVSDstore, ssa.OpAMD64MOVLstore, ssa.OpAMD64MOVWstore, ssa.OpAMD64MOVBstore, ssa.OpAMD64MOVOstore,
+		ssa.OpAMD64BTCQmodify, ssa.OpAMD64BTCLmodify, ssa.OpAMD64BTRQmodify, ssa.OpAMD64BTRLmodify, ssa.OpAMD64BTSQmodify, ssa.OpAMD64BTSLmodify,
 		ssa.OpAMD64ADDQmodify, ssa.OpAMD64SUBQmodify, ssa.OpAMD64ANDQmodify, ssa.OpAMD64ORQmodify, ssa.OpAMD64XORQmodify,
 		ssa.OpAMD64ADDLmodify, ssa.OpAMD64SUBLmodify, ssa.OpAMD64ANDLmodify, ssa.OpAMD64ORLmodify, ssa.OpAMD64XORLmodify:
 		p := s.Prog(v.Op.Asm())
@@ -763,7 +764,8 @@ func ssaGenValue(s *gc.SSAGenState, v *ssa.Value) {
 		}
 		fallthrough
 	case ssa.OpAMD64ANDQconstmodify, ssa.OpAMD64ANDLconstmodify, ssa.OpAMD64ORQconstmodify, ssa.OpAMD64ORLconstmodify,
-		ssa.OpAMD64XORQconstmodify, ssa.OpAMD64XORLconstmodify:
+		ssa.OpAMD64BTCQconstmodify, ssa.OpAMD64BTCLconstmodify, ssa.OpAMD64BTSQconstmodify, ssa.OpAMD64BTSLconstmodify,
+		ssa.OpAMD64BTRQconstmodify, ssa.OpAMD64BTRLconstmodify, ssa.OpAMD64XORQconstmodify, ssa.OpAMD64XORLconstmodify:
 		sc := v.AuxValAndOff()
 		off := sc.Off()
 		val := sc.Val()
diff --git a/src/cmd/compile/internal/ssa/gen/AMD64.rules b/src/cmd/compile/internal/ssa/gen/AMD64.rules
index 803b8896b0..76a4fc9ab7 100644
--- a/src/cmd/compile/internal/ssa/gen/AMD64.rules
+++ b/src/cmd/compile/internal/ssa/gen/AMD64.rules
@@ -709,7 +709,17 @@
 (ANDL x (MOVLconst [c])) -> (ANDLconst [c] x)
 
 (AND(L|Q)const [c] (AND(L|Q)const [d] x)) -> (AND(L|Q)const [c & d] x)
+(BTR(L|Q)const [c] (AND(L|Q)const [d] x)) -> (AND(L|Q)const [d &^ 1<<uint32(c)] x)
+(AND(L|Q)const [c] (BTR(L|Q)const [d] x)) -> (AND(L|Q)const [c &^ 1<<uint32(d)] x)
+(BTR(L|Q)const [c] (BTR(L|Q)const [d] x)) -> (AND(L|Q)const [^(1<<uint32(c) | 1<<uint32(d))] x)
 (XOR(L|Q)const [c] (XOR(L|Q)const [d] x)) -> (XOR(L|Q)const [c ^ d] x)
+(BTC(L|Q)const [c] (XOR(L|Q)const [d] x)) -> (XOR(L|Q)const [d ^ 1<<uint32(c)] x)
+(XOR(L|Q)const [c] (BTC(L|Q)const [d] x)) -> (XOR(L|Q)const [c ^ 1<<uint32(d)] x)
+(BTC(L|Q)const [c] (BTC(L|Q)const [d] x)) -> (XOR(L|Q)const [1<<uint32(c) ^ 1<<uint32(d)] x)
+(OR(L|Q)const [c] (OR(L|Q)const [d] x)) -> (OR(L|Q)const [c | d] x)
+(OR(L|Q)const [c] (BTS(L|Q)const [d] x)) -> (OR(L|Q)const [c | 1<<uint32(d)] x)
+(BTS(L|Q)const [c] (OR(L|Q)const [d] x)) -> (OR(L|Q)const [d | 1<<uint32(c)] x)
+(BTS(L|Q)const [c] (BTS(L|Q)const [d] x)) -> (OR(L|Q)const [1<<uint32(d) | 1<<uint32(c)] x)
 
 (MULLconst [c] (MULLconst [d] x)) -> (MULLconst [int64(int32(c * d))] x)
 (MULQconst [c] (MULQconst [d] x)) && is32Bit(c*d) -> (MULQconst [c * d] x)
@@ -1051,14 +1061,14 @@
 	((ADD|SUB|MUL|DIV)SSload [off1+off2] {sym} val base mem)
 ((ADD|SUB|MUL|DIV)SDload [off1] {sym} val (ADDQconst [off2] base) mem) && is32Bit(off1+off2) ->
 	((ADD|SUB|MUL|DIV)SDload [off1+off2] {sym} val base mem)
-((ADD|AND|OR|XOR)Qconstmodify [valoff1] {sym} (ADDQconst [off2] base) mem) && ValAndOff(valoff1).canAdd(off2) ->
-	((ADD|AND|OR|XOR)Qconstmodify [ValAndOff(valoff1).add(off2)] {sym} base mem)
-((ADD|AND|OR|XOR)Lconstmodify [valoff1] {sym} (ADDQconst [off2] base) mem) && ValAndOff(valoff1).canAdd(off2) ->
-	((ADD|AND|OR|XOR)Lconstmodify [ValAndOff(valoff1).add(off2)] {sym} base mem)
-((ADD|SUB|AND|OR|XOR)Qmodify [off1] {sym} (ADDQconst [off2] base) val mem) && is32Bit(off1+off2) ->
-	((ADD|SUB|AND|OR|XOR)Qmodify [off1+off2] {sym} base val mem)
-((ADD|SUB|AND|OR|XOR)Lmodify [off1] {sym} (ADDQconst [off2] base) val mem) && is32Bit(off1+off2) ->
-	((ADD|SUB|AND|OR|XOR)Lmodify [off1+off2] {sym} base val mem)
+((ADD|AND|OR|XOR|BTC|BTR|BTS)Qconstmodify [valoff1] {sym} (ADDQconst [off2] base) mem) && ValAndOff(valoff1).canAdd(off2) ->
+	((ADD|AND|OR|XOR|BTC|BTR|BTS)Qconstmodify [ValAndOff(valoff1).add(off2)] {sym} base mem)
+((ADD|AND|OR|XOR|BTC|BTR|BTS)Lconstmodify [valoff1] {sym} (ADDQconst [off2] base) mem) && ValAndOff(valoff1).canAdd(off2) ->
+	((ADD|AND|OR|XOR|BTC|BTR|BTS)Lconstmodify [ValAndOff(valoff1).add(off2)] {sym} base mem)
+((ADD|SUB|AND|OR|XOR|BTC|BTR|BTS)Qmodify [off1] {sym} (ADDQconst [off2] base) val mem) && is32Bit(off1+off2) ->
+	((ADD|SUB|AND|OR|XOR|BTC|BTR|BTS)Qmodify [off1+off2] {sym} base val mem)
+((ADD|SUB|AND|OR|XOR|BTC|BTR|BTS)Lmodify [off1] {sym} (ADDQconst [off2] base) val mem) && is32Bit(off1+off2) ->
+	((ADD|SUB|AND|OR|XOR|BTC|BTR|BTS)Lmodify [off1+off2] {sym} base val mem)
 
 // Fold constants into stores.
 (MOVQstore [off] {sym} ptr (MOVQconst [c]) mem) && validValAndOff(c,off) ->
@@ -1106,18 +1116,18 @@
 ((ADD|SUB|MUL|DIV)SDload [off1] {sym1} val (LEAQ [off2] {sym2} base) mem)
 	&& is32Bit(off1+off2) && canMergeSym(sym1, sym2) ->
 	((ADD|SUB|MUL|DIV)SDload [off1+off2] {mergeSym(sym1,sym2)} val base mem)
-((ADD|AND|OR|XOR)Qconstmodify [valoff1] {sym1} (LEAQ [off2] {sym2} base) mem)
+((ADD|AND|OR|XOR|BTC|BTR|BTS)Qconstmodify [valoff1] {sym1} (LEAQ [off2] {sym2} base) mem)
 	&& ValAndOff(valoff1).canAdd(off2) && canMergeSym(sym1, sym2) ->
-	((ADD|AND|OR|XOR)Qconstmodify [ValAndOff(valoff1).add(off2)] {mergeSym(sym1,sym2)} base mem)
-((ADD|AND|OR|XOR)Lconstmodify [valoff1] {sym1} (LEAQ [off2] {sym2} base) mem)
+	((ADD|AND|OR|XOR|BTC|BTR|BTS)Qconstmodify [ValAndOff(valoff1).add(off2)] {mergeSym(sym1,sym2)} base mem)
+((ADD|AND|OR|XOR|BTC|BTR|BTS)Lconstmodify [valoff1] {sym1} (LEAQ [off2] {sym2} base) mem)
 	&& ValAndOff(valoff1).canAdd(off2) && canMergeSym(sym1, sym2) ->
-	((ADD|AND|OR|XOR)Lconstmodify [ValAndOff(valoff1).add(off2)] {mergeSym(sym1,sym2)} base mem)
-((ADD|SUB|AND|OR|XOR)Qmodify [off1] {sym1} (LEAQ [off2] {sym2} base) val mem)
+	((ADD|AND|OR|XOR|BTC|BTR|BTS)Lconstmodify [ValAndOff(valoff1).add(off2)] {mergeSym(sym1,sym2)} base mem)
+((ADD|SUB|AND|OR|XOR|BTC|BTR|BTS)Qmodify [off1] {sym1} (LEAQ [off2] {sym2} base) val mem)
 	&& is32Bit(off1+off2) && canMergeSym(sym1, sym2) ->
-	((ADD|SUB|AND|OR|XOR)Qmodify [off1+off2] {mergeSym(sym1,sym2)} base val mem)
-((ADD|SUB|AND|OR|XOR)Lmodify [off1] {sym1} (LEAQ [off2] {sym2} base) val mem)
+	((ADD|SUB|AND|OR|XOR|BTC|BTR|BTS)Qmodify [off1+off2] {mergeSym(sym1,sym2)} base val mem)
+((ADD|SUB|AND|OR|XOR|BTC|BTR|BTS)Lmodify [off1] {sym1} (LEAQ [off2] {sym2} base) val mem)
 	&& is32Bit(off1+off2) && canMergeSym(sym1, sym2) ->
-	((ADD|SUB|AND|OR|XOR)Lmodify [off1+off2] {mergeSym(sym1,sym2)} base val mem)
+	((ADD|SUB|AND|OR|XOR|BTC|BTR|BTS)Lmodify [off1+off2] {mergeSym(sym1,sym2)} base val mem)
 
 // generating indexed loads and stores
 (MOV(B|W|L|Q|SS|SD)load [off1] {sym1} (LEAQ1 [off2] {sym2} ptr idx) mem) && is32Bit(off1+off2) && canMergeSym(sym1, sym2) ->
@@ -1424,6 +1434,12 @@
 (XORLconst [c] (MOVLconst [d])) -> (MOVLconst [c^d])
 (NOTQ (MOVQconst [c])) -> (MOVQconst [^c])
 (NOTL (MOVLconst [c])) -> (MOVLconst [^c])
+(BTSQconst [c] (MOVQconst [d])) -> (MOVQconst [d|(1<<uint32(c))])
+(BTSLconst [c] (MOVLconst [d])) -> (MOVLconst [d|(1<<uint32(c))])
+(BTRQconst [c] (MOVQconst [d])) -> (MOVQconst [d&^(1<<uint32(c))])
+(BTRLconst [c] (MOVLconst [d])) -> (MOVLconst [d&^(1<<uint32(c))])
+(BTCQconst [c] (MOVQconst [d])) -> (MOVQconst [d^(1<<uint32(c))])
+(BTCLconst [c] (MOVLconst [d])) -> (MOVLconst [d^(1<<uint32(c))])
 
 // generic simplifications
 // TODO: more of this
@@ -2304,11 +2320,11 @@
 ((ADD|SUB|MUL|DIV)SD x l:(MOVSDload [off] {sym} ptr mem)) && canMergeLoad(v, l, x) && clobber(l) -> ((ADD|SUB|MUL|DIV)SDload x [off] {sym} ptr mem)
 ((ADD|SUB|MUL|DIV)SS x l:(MOVSSload [off] {sym} ptr mem)) && canMergeLoad(v, l, x) && clobber(l) -> ((ADD|SUB|MUL|DIV)SSload x [off] {sym} ptr mem)
 (MOVLstore {sym} [off] ptr y:((ADD|AND|OR|XOR)Lload x [off] {sym} ptr mem) mem) && y.Uses==1 && clobber(y) -> ((ADD|AND|OR|XOR)Lmodify [off] {sym} ptr x mem)
-(MOVLstore {sym} [off] ptr y:((ADD|SUB|AND|OR|XOR)L l:(MOVLload [off] {sym} ptr mem) x) mem) && y.Uses==1 && l.Uses==1 && clobber(y) && clobber(l) ->
-	((ADD|SUB|AND|OR|XOR)Lmodify [off] {sym} ptr x mem)
+(MOVLstore {sym} [off] ptr y:((ADD|SUB|AND|OR|XOR|BTC|BTR|BTS)L l:(MOVLload [off] {sym} ptr mem) x) mem) && y.Uses==1 && l.Uses==1 && clobber(y) && clobber(l) ->
+	((ADD|SUB|AND|OR|XOR|BTC|BTR|BTS)Lmodify [off] {sym} ptr x mem)
 (MOVQstore {sym} [off] ptr y:((ADD|AND|OR|XOR)Qload x [off] {sym} ptr mem) mem) && y.Uses==1 && clobber(y) -> ((ADD|AND|OR|XOR)Qmodify [off] {sym} ptr x mem)
-(MOVQstore {sym} [off] ptr y:((ADD|SUB|AND|OR|XOR)Q l:(MOVQload [off] {sym} ptr mem) x) mem) && y.Uses==1 && l.Uses==1 && clobber(y) && clobber(l) ->
-	((ADD|SUB|AND|OR|XOR)Qmodify [off] {sym} ptr x mem)
+(MOVQstore {sym} [off] ptr y:((ADD|SUB|AND|OR|XOR|BTC|BTR|BTS)Q l:(MOVQload [off] {sym} ptr mem) x) mem) && y.Uses==1 && l.Uses==1 && clobber(y) && clobber(l) ->
+	((ADD|SUB|AND|OR|XOR|BTC|BTR|BTS)Qmodify [off] {sym} ptr x mem)
 
 // Merge ADDQconst and LEAQ into atomic loads.
 (MOVQatomicload [off1] {sym} (ADDQconst [off2] ptr) mem) && is32Bit(off1+off2) ->
@@ -2392,12 +2408,12 @@
 (MOVWQZX (MOVBQZX x)) -> (MOVBQZX x)
 (MOVBQZX (MOVBQZX x)) -> (MOVBQZX x)
 
-(MOVQstore [off] {sym} ptr a:((ADD|AND|OR|XOR)Qconst [c] l:(MOVQload [off] {sym} ptr2 mem)) mem)
-	&& isSamePtr(ptr, ptr2) && a.Uses == 1 && l.Uses == 1 && validValAndOff(c,off) ->
-	((ADD|AND|OR|XOR)Qconstmodify {sym} [makeValAndOff(c,off)] ptr mem)
-(MOVLstore [off] {sym} ptr a:((ADD|AND|OR|XOR)Lconst [c] l:(MOVLload [off] {sym} ptr2 mem)) mem)
-	&& isSamePtr(ptr, ptr2) && a.Uses == 1 && l.Uses == 1 && validValAndOff(c,off) ->
-	((ADD|AND|OR|XOR)Lconstmodify {sym} [makeValAndOff(c,off)] ptr mem)
+(MOVQstore [off] {sym} ptr a:((ADD|AND|OR|XOR|BTC|BTR|BTS)Qconst [c] l:(MOVQload [off] {sym} ptr2 mem)) mem)
+	&& isSamePtr(ptr, ptr2) && a.Uses == 1 && l.Uses == 1 && validValAndOff(c,off) && clobber(l) && clobber(a) ->
+	((ADD|AND|OR|XOR|BTC|BTR|BTS)Qconstmodify {sym} [makeValAndOff(c,off)] ptr mem)
+(MOVLstore [off] {sym} ptr a:((ADD|AND|OR|XOR|BTC|BTR|BTS)Lconst [c] l:(MOVLload [off] {sym} ptr2 mem)) mem)
+	&& isSamePtr(ptr, ptr2) && a.Uses == 1 && l.Uses == 1 && validValAndOff(c,off) && clobber(l) && clobber(a) ->
+	((ADD|AND|OR|XOR|BTC|BTR|BTS)Lconstmodify {sym} [makeValAndOff(c,off)] ptr mem)
 
 // float <-> int register moves, with no conversion.
 // These come up when compiling math.{Float{32,64}bits,Float{32,64}frombits}.
diff --git a/src/cmd/compile/internal/ssa/gen/AMD64Ops.go b/src/cmd/compile/internal/ssa/gen/AMD64Ops.go
index 68facebe47..017c07071d 100644
--- a/src/cmd/compile/internal/ssa/gen/AMD64Ops.go
+++ b/src/cmd/compile/internal/ssa/gen/AMD64Ops.go
@@ -289,6 +289,20 @@ func init() {
 		{name: "BTSLconst", argLength: 1, reg: gp11, asm: "BTSL", resultInArg0: true, clobberFlags: true, aux: "Int8"}, // set bit auxint in arg0, 0 <= auxint < 32
 		{name: "BTSQconst", argLength: 1, reg: gp11, asm: "BTSQ", resultInArg0: true, clobberFlags: true, aux: "Int8"}, // set bit auxint in arg0, 0 <= auxint < 64
 
+		// direct bit operation on memory operand
+		{name: "BTCQmodify", argLength: 3, reg: gpstore, asm: "BTCQ", aux: "SymOff", typ: "Mem", clobberFlags: true, faultOnNilArg0: true, symEffect: "Read,Write"},     // complement bit arg1 in 64-bit arg0+auxint+aux, arg2=mem
+		{name: "BTCLmodify", argLength: 3, reg: gpstore, asm: "BTCL", aux: "SymOff", typ: "Mem", clobberFlags: true, faultOnNilArg0: true, symEffect: "Read,Write"},     // complement bit arg1 in 32-bit arg0+auxint+aux, arg2=mem
+		{name: "BTSQmodify", argLength: 3, reg: gpstore, asm: "BTSQ", aux: "SymOff", typ: "Mem", clobberFlags: true, faultOnNilArg0: true, symEffect: "Read,Write"},     // set bit arg1 in 64-bit arg0+auxint+aux, arg2=mem
+		{name: "BTSLmodify", argLength: 3, reg: gpstore, asm: "BTSL", aux: "SymOff", typ: "Mem", clobberFlags: true, faultOnNilArg0: true, symEffect: "Read,Write"},     // set bit arg1 in 32-bit arg0+auxint+aux, arg2=mem
+		{name: "BTRQmodify", argLength: 3, reg: gpstore, asm: "BTRQ", aux: "SymOff", typ: "Mem", clobberFlags: true, faultOnNilArg0: true, symEffect: "Read,Write"},     // reset bit arg1 in 64-bit arg0+auxint+aux, arg2=mem
+		{name: "BTRLmodify", argLength: 3, reg: gpstore, asm: "BTRL", aux: "SymOff", typ: "Mem", clobberFlags: true, faultOnNilArg0: true, symEffect: "Read,Write"},     // reset bit arg1 in 32-bit arg0+auxint+aux, arg2=mem
+		{name: "BTCQconstmodify", argLength: 2, reg: gpstoreconst, asm: "BTCQ", aux: "SymValAndOff", clobberFlags: true, faultOnNilArg0: true, symEffect: "Read,Write"}, // complement bit ValAndOff(AuxInt).Val() in 64-bit arg0+ValAndOff(AuxInt).Off()+aux, arg1=mem
+		{name: "BTCLconstmodify", argLength: 2, reg: gpstoreconst, asm: "BTCL", aux: "SymValAndOff", clobberFlags: true, faultOnNilArg0: true, symEffect: "Read,Write"}, // complement bit ValAndOff(AuxInt).Val() in 32-bit arg0+ValAndOff(AuxInt).Off()+aux, arg1=mem
+		{name: "BTSQconstmodify", argLength: 2, reg: gpstoreconst, asm: "BTSQ", aux: "SymValAndOff", clobberFlags: true, faultOnNilArg0: true, symEffect: "Read,Write"}, // set bit ValAndOff(AuxInt).Val() in 64-bit arg0+ValAndOff(AuxInt).Off()+aux, arg1=mem
+		{name: "BTSLconstmodify", argLength: 2, reg: gpstoreconst, asm: "BTSL", aux: "SymValAndOff", clobberFlags: true, faultOnNilArg0: true, symEffect: "Read,Write"}, // set bit ValAndOff(AuxInt).Val() in 32-bit arg0+ValAndOff(AuxInt).Off()+aux, arg1=mem
+		{name: "BTRQconstmodify", argLength: 2, reg: gpstoreconst, asm: "BTRQ", aux: "SymValAndOff", clobberFlags: true, faultOnNilArg0: true, symEffect: "Read,Write"}, // reset bit ValAndOff(AuxInt).Val() in 64-bit arg0+ValAndOff(AuxInt).Off()+aux, arg1=mem
+		{name: "BTRLconstmodify", argLength: 2, reg: gpstoreconst, asm: "BTRL", aux: "SymValAndOff", clobberFlags: true, faultOnNilArg0: true, symEffect: "Read,Write"}, // reset bit ValAndOff(AuxInt).Val() in 32-bit arg0+ValAndOff(AuxInt).Off()+aux, arg1=mem
+
 		{name: "TESTQ", argLength: 2, reg: gp2flags, commutative: true, asm: "TESTQ", typ: "Flags"}, // (arg0 & arg1) compare to 0
 		{name: "TESTL", argLength: 2, reg: gp2flags, commutative: true, asm: "TESTL", typ: "Flags"}, // (arg0 & arg1) compare to 0
 		{name: "TESTW", argLength: 2, reg: gp2flags, commutative: true, asm: "TESTW", typ: "Flags"}, // (arg0 & arg1) compare to 0
diff --git a/src/cmd/compile/internal/ssa/opGen.go b/src/cmd/compile/internal/ssa/opGen.go
index 77b9875fd6..fe63633750 100644
--- a/src/cmd/compile/internal/ssa/opGen.go
+++ b/src/cmd/compile/internal/ssa/opGen.go
@@ -550,6 +550,18 @@ const (
 	OpAMD64BTRQconst
 	OpAMD64BTSLconst
 	OpAMD64BTSQconst
+	OpAMD64BTCQmodify
+	OpAMD64BTCLmodify
+	OpAMD64BTSQmodify
+	OpAMD64BTSLmodify
+	OpAMD64BTRQmodify
+	OpAMD64BTRLmodify
+	OpAMD64BTCQconstmodify
+	OpAMD64BTCLconstmodify
+	OpAMD64BTSQconstmodify
+	OpAMD64BTSLconstmodify
+	OpAMD64BTRQconstmodify
+	OpAMD64BTRLconstmodify
 	OpAMD64TESTQ
 	OpAMD64TESTL
 	OpAMD64TESTW
@@ -6902,6 +6914,180 @@ var opcodeTable = [...]opInfo{
 		},
 	},
 	{
+		name:           "BTCQmodify",
+		auxType:        auxSymOff,
+		argLen:         3,
+		clobberFlags:   true,
+		faultOnNilArg0: true,
+		symEffect:      SymRead | SymWrite,
+		asm:            x86.ABTCQ,
+		reg: regInfo{
+			inputs: []inputInfo{
+				{1, 65535},      // AX CX DX BX SP BP SI DI R8 R9 R10 R11 R12 R13 R14 R15
+				{0, 4295032831}, // AX CX DX BX SP BP SI DI R8 R9 R10 R11 R12 R13 R14 R15 SB
+			},
+		},
+	},
+	{
+		name:           "BTCLmodify",
+		auxType:        auxSymOff,
+		argLen:         3,
+		clobberFlags:   true,
+		faultOnNilArg0: true,
+		symEffect:      SymRead | SymWrite,
+		asm:            x86.ABTCL,
+		reg: regInfo{
+			inputs: []inputInfo{
+				{1, 65535},      // AX CX DX BX SP BP SI DI R8 R9 R10 R11 R12 R13 R14 R15
+				{0, 4295032831}, // AX CX DX BX SP BP SI DI R8 R9 R10 R11 R12 R13 R14 R15 SB
+			},
+		},
+	},
+	{
+		name:           "BTSQmodify",
+		auxType:        auxSymOff,
+		argLen:         3,
+		clobberFlags:   true,
+		faultOnNilArg0: true,
+		symEffect:      SymRead | SymWrite,
+		asm:            x86.ABTSQ,
+		reg: regInfo{
+			inputs: []inputInfo{
+				{1, 65535},      // AX CX DX BX SP BP SI DI R8 R9 R10 R11 R12 R13 R14 R15
+				{0, 4295032831}, // AX CX DX BX SP BP SI DI R8 R9 R10 R11 R12 R13 R14 R15 SB
+			},
+		},
+	},
+	{
+		name:           "BTSLmodify",
+		auxType:        auxSymOff,
+		argLen:         3,
+		clobberFlags:   true,
+		faultOnNilArg0: true,
+		symEffect:      SymRead | SymWrite,
+		asm:            x86.ABTSL,
+		reg: regInfo{
+			inputs: []inputInfo{
+				{1, 65535},      // AX CX DX BX SP BP SI DI R8 R9 R10 R11 R12 R13 R14 R15
+				{0, 4295032831}, // AX CX DX BX SP BP SI DI R8 R9 R10 R11 R12 R13 R14 R15 SB
+			},
+		},
+	},
+	{
+		name:           "BTRQmodify",
+		auxType:        auxSymOff,
+		argLen:         3,
+		clobberFlags:   true,
+		faultOnNilArg0: true,
+		symEffect:      SymRead | SymWrite,
+		asm:            x86.ABTRQ,
+		reg: regInfo{
+			inputs: []inputInfo{
+				{1, 65535},      // AX CX DX BX SP BP SI DI R8 R9 R10 R11 R12 R13 R14 R15
+				{0, 4295032831}, // AX CX DX BX SP BP SI DI R8 R9 R10 R11 R12 R13 R14 R15 SB
+			},
+		},
+	},
+	{
+		name:           "BTRLmodify",
+		auxType:        auxSymOff,
+		argLen:         3,
+		clobberFlags:   true,
+		faultOnNilArg0: true,
+		symEffect:      SymRead | SymWrite,
+		asm:            x86.ABTRL,
+		reg: regInfo{
+			inputs: []inputInfo{
+				{1, 65535},      // AX CX DX BX SP BP SI DI R8 R9 R10 R11 R12 R13 R14 R15
+				{0, 4295032831}, // AX CX DX BX SP BP SI DI R8 R9 R10 R11 R12 R13 R14 R15 SB
+			},
+		},
+	},
+	{
+		name:           "BTCQconstmodify",
+		auxType:        auxSymValAndOff,
+		argLen:         2,
+		clobberFlags:   true,
+		faultOnNilArg0: true,
+		symEffect:      SymRead | SymWrite,
+		asm:            x86.ABTCQ,
+		reg: regInfo{
+			inputs: []inputInfo{
+				{0, 4295032831}, // AX CX DX BX SP BP SI DI R8 R9 R10 R11 R12 R13 R14 R15 SB
+			},
+		},
+	},
+	{
+		name:           "BTCLconstmodify",
+		auxType:        auxSymValAndOff,
+		argLen:         2,
+		clobberFlags:   true,
+		faultOnNilArg0: true,
+		symEffect:      SymRead | SymWrite,
+		asm:            x86.ABTCL,
+		reg: regInfo{
+			inputs: []inputInfo{
+				{0, 4295032831}, // AX CX DX BX SP BP SI DI R8 R9 R10 R11 R12 R13 R14 R15 SB
+			},
+		},
+	},
+	{
+		name:           "BTSQconstmodify",
+		auxType:        auxSymValAndOff,
+		argLen:         2,
+		clobberFlags:   true,
+		faultOnNilArg0: true,
+		symEffect:      SymRead | SymWrite,
+		asm:            x86.ABTSQ,
+		reg: regInfo{
+			inputs: []inputInfo{
+				{0, 4295032831}, // AX CX DX BX SP BP SI DI R8 R9 R10 R11 R12 R13 R14 R15 SB
+			},
+		},
+	},
+	{
+		name:           "BTSLconstmodify",
+		auxType:        auxSymValAndOff,
+		argLen:         2,
+		clobberFlags:   true,
+		faultOnNilArg0: true,
+		symEffect:      SymRead | SymWrite,
+		asm:            x86.ABTSL,
+		reg: regInfo{
+			inputs: []inputInfo{
+				{0, 4295032831}, // AX CX DX BX SP BP SI DI R8 R9 R10 R11 R12 R13 R14 R15 SB
+			},
+		},
+	},
+	{
+		name:           "BTRQconstmodify",
+		auxType:        auxSymValAndOff,
+		argLen:         2,
+		clobberFlags:   true,
+		faultOnNilArg0: true,
+		symEffect:      SymRead | SymWrite,
+		asm:            x86.ABTRQ,
+		reg: regInfo{
+			inputs: []inputInfo{
+				{0, 4295032831}, // AX CX DX BX SP BP SI DI R8 R9 R10 R11 R12 R13 R14 R15 SB
+			},
+		},
+	},
+	{
+		name:           "BTRLconstmodify",
+		auxType:        auxSymValAndOff,
+		argLen:         2,
+		clobberFlags:   true,
+		faultOnNilArg0: true,
+		symEffect:      SymRead | SymWrite,
+		asm:            x86.ABTRL,
+		reg: regInfo{
+			inputs: []inputInfo{
+				{0, 4295032831}, // AX CX DX BX SP BP SI DI R8 R9 R10 R11 R12 R13 R14 R15 SB
+			},
+		},
+	},
+	{
 		name:        "TESTQ",
 		argLen:      2,
 		commutative: true,
diff --git a/src/cmd/compile/internal/ssa/rewriteAMD64.go b/src/cmd/compile/internal/ssa/rewriteAMD64.go
index 98b36a96a0..cd82a5642c 100644
--- a/src/cmd/compile/internal/ssa/rewriteAMD64.go
+++ b/src/cmd/compile/internal/ssa/rewriteAMD64.go
@@ -65,18 +65,46 @@ func rewriteValueAMD64(v *Value) bool {
 		return rewriteValueAMD64_OpAMD64ANDQmodify_0(v)
 	case OpAMD64BSFQ:
 		return rewriteValueAMD64_OpAMD64BSFQ_0(v)
+	case OpAMD64BTCLconst:
+		return rewriteValueAMD64_OpAMD64BTCLconst_0(v)
+	case OpAMD64BTCLconstmodify:
+		return rewriteValueAMD64_OpAMD64BTCLconstmodify_0(v)
+	case OpAMD64BTCLmodify:
+		return rewriteValueAMD64_OpAMD64BTCLmodify_0(v)
+	case OpAMD64BTCQconst:
+		return rewriteValueAMD64_OpAMD64BTCQconst_0(v)
+	case OpAMD64BTCQconstmodify:
+		return rewriteValueAMD64_OpAMD64BTCQconstmodify_0(v)
+	case OpAMD64BTCQmodify:
+		return rewriteValueAMD64_OpAMD64BTCQmodify_0(v)
 	case OpAMD64BTLconst:
 		return rewriteValueAMD64_OpAMD64BTLconst_0(v)
 	case OpAMD64BTQconst:
 		return rewriteValueAMD64_OpAMD64BTQconst_0(v)
 	case OpAMD64BTRLconst:
 		return rewriteValueAMD64_OpAMD64BTRLconst_0(v)
+	case OpAMD64BTRLconstmodify:
+		return rewriteValueAMD64_OpAMD64BTRLconstmodify_0(v)
+	case OpAMD64BTRLmodify:
+		return rewriteValueAMD64_OpAMD64BTRLmodify_0(v)
 	case OpAMD64BTRQconst:
 		return rewriteValueAMD64_OpAMD64BTRQconst_0(v)
+	case OpAMD64BTRQconstmodify:
+		return rewriteValueAMD64_OpAMD64BTRQconstmodify_0(v)
+	case OpAMD64BTRQmodify:
+		return rewriteValueAMD64_OpAMD64BTRQmodify_0(v)
 	case OpAMD64BTSLconst:
 		return rewriteValueAMD64_OpAMD64BTSLconst_0(v)
+	case OpAMD64BTSLconstmodify:
+		return rewriteValueAMD64_OpAMD64BTSLconstmodify_0(v)
+	case OpAMD64BTSLmodify:
+		return rewriteValueAMD64_OpAMD64BTSLmodify_0(v)
 	case OpAMD64BTSQconst:
 		return rewriteValueAMD64_OpAMD64BTSQconst_0(v)
+	case OpAMD64BTSQconstmodify:
+		return rewriteValueAMD64_OpAMD64BTSQconstmodify_0(v)
+	case OpAMD64BTSQmodify:
+		return rewriteValueAMD64_OpAMD64BTSQmodify_0(v)
 	case OpAMD64CMOVLCC:
 		return rewriteValueAMD64_OpAMD64CMOVLCC_0(v)
 	case OpAMD64CMOVLCS:
@@ -278,7 +306,7 @@ func rewriteValueAMD64(v *Value) bool {
 	case OpAMD64MOVQloadidx8:
 		return rewriteValueAMD64_OpAMD64MOVQloadidx8_0(v)
 	case OpAMD64MOVQstore:
-		return rewriteValueAMD64_OpAMD64MOVQstore_0(v) || rewriteValueAMD64_OpAMD64MOVQstore_10(v) || rewriteValueAMD64_OpAMD64MOVQstore_20(v)
+		return rewriteValueAMD64_OpAMD64MOVQstore_0(v) || rewriteValueAMD64_OpAMD64MOVQstore_10(v) || rewriteValueAMD64_OpAMD64MOVQstore_20(v) || rewriteValueAMD64_OpAMD64MOVQstore_30(v)
 	case OpAMD64MOVQstoreconst:
 		return rewriteValueAMD64_OpAMD64MOVQstoreconst_0(v)
 	case OpAMD64MOVQstoreconstidx1:
@@ -3590,6 +3618,22 @@ func rewriteValueAMD64_OpAMD64ANDLconst_0(v *Value) bool {
 		v.AddArg(x)
 		return true
 	}
+	// match: (ANDLconst [c] (BTRLconst [d] x))
+	// cond:
+	// result: (ANDLconst [c &^ 1<<uint32(d)] x)
+	for {
+		c := v.AuxInt
+		v_0 := v.Args[0]
+		if v_0.Op != OpAMD64BTRLconst {
+			break
+		}
+		d := v_0.AuxInt
+		x := v_0.Args[0]
+		v.reset(OpAMD64ANDLconst)
+		v.AuxInt = c &^ 1 << uint32(d)
+		v.AddArg(x)
+		return true
+	}
 	// match: (ANDLconst [ 0xFF] x)
 	// cond:
 	// result: (MOVBQZX x)
@@ -4101,6 +4145,22 @@ func rewriteValueAMD64_OpAMD64ANDQconst_0(v *Value) bool {
 		v.AddArg(x)
 		return true
 	}
+	// match: (ANDQconst [c] (BTRQconst [d] x))
+	// cond:
+	// result: (ANDQconst [c &^ 1<<uint32(d)] x)
+	for {
+		c := v.AuxInt
+		v_0 := v.Args[0]
+		if v_0.Op != OpAMD64BTRQconst {
+			break
+		}
+		d := v_0.AuxInt
+		x := v_0.Args[0]
+		v.reset(OpAMD64ANDQconst)
+		v.AuxInt = c &^ 1 << uint32(d)
+		v.AddArg(x)
+		return true
+	}
 	// match: (ANDQconst [ 0xFF] x)
 	// cond:
 	// result: (MOVBQZX x)
@@ -4429,6 +4489,320 @@ func rewriteValueAMD64_OpAMD64BSFQ_0(v *Value) bool {
 	}
 	return false
 }
+func rewriteValueAMD64_OpAMD64BTCLconst_0(v *Value) bool {
+	// match: (BTCLconst [c] (XORLconst [d] x))
+	// cond:
+	// result: (XORLconst [d ^ 1<<uint32(c)] x)
+	for {
+		c := v.AuxInt
+		v_0 := v.Args[0]
+		if v_0.Op != OpAMD64XORLconst {
+			break
+		}
+		d := v_0.AuxInt
+		x := v_0.Args[0]
+		v.reset(OpAMD64XORLconst)
+		v.AuxInt = d ^ 1<<uint32(c)
+		v.AddArg(x)
+		return true
+	}
+	// match: (BTCLconst [c] (BTCLconst [d] x))
+	// cond:
+	// result: (XORLconst [1<<uint32(c) ^ 1<<uint32(d)] x)
+	for {
+		c := v.AuxInt
+		v_0 := v.Args[0]
+		if v_0.Op != OpAMD64BTCLconst {
+			break
+		}
+		d := v_0.AuxInt
+		x := v_0.Args[0]
+		v.reset(OpAMD64XORLconst)
+		v.AuxInt = 1<<uint32(c) ^ 1<<uint32(d)
+		v.AddArg(x)
+		return true
+	}
+	// match: (BTCLconst [c] (MOVLconst [d]))
+	// cond:
+	// result: (MOVLconst [d^(1<<uint32(c))])
+	for {
+		c := v.AuxInt
+		v_0 := v.Args[0]
+		if v_0.Op != OpAMD64MOVLconst {
+			break
+		}
+		d := v_0.AuxInt
+		v.reset(OpAMD64MOVLconst)
+		v.AuxInt = d ^ (1 << uint32(c))
+		return true
+	}
+	return false
+}
+func rewriteValueAMD64_OpAMD64BTCLconstmodify_0(v *Value) bool {
+	// match: (BTCLconstmodify [valoff1] {sym} (ADDQconst [off2] base) mem)
+	// cond: ValAndOff(valoff1).canAdd(off2)
+	// result: (BTCLconstmodify [ValAndOff(valoff1).add(off2)] {sym} base mem)
+	for {
+		valoff1 := v.AuxInt
+		sym := v.Aux
+		_ = v.Args[1]
+		v_0 := v.Args[0]
+		if v_0.Op != OpAMD64ADDQconst {
+			break
+		}
+		off2 := v_0.AuxInt
+		base := v_0.Args[0]
+		mem := v.Args[1]
+		if !(ValAndOff(valoff1).canAdd(off2)) {
+			break
+		}
+		v.reset(OpAMD64BTCLconstmodify)
+		v.AuxInt = ValAndOff(valoff1).add(off2)
+		v.Aux = sym
+		v.AddArg(base)
+		v.AddArg(mem)
+		return true
+	}
+	// match: (BTCLconstmodify [valoff1] {sym1} (LEAQ [off2] {sym2} base) mem)
+	// cond: ValAndOff(valoff1).canAdd(off2) && canMergeSym(sym1, sym2)
+	// result: (BTCLconstmodify [ValAndOff(valoff1).add(off2)] {mergeSym(sym1,sym2)} base mem)
+	for {
+		valoff1 := v.AuxInt
+		sym1 := v.Aux
+		_ = v.Args[1]
+		v_0 := v.Args[0]
+		if v_0.Op != OpAMD64LEAQ {
+			break
+		}
+		off2 := v_0.AuxInt
+		sym2 := v_0.Aux
+		base := v_0.Args[0]
+		mem := v.Args[1]
+		if !(ValAndOff(valoff1).canAdd(off2) && canMergeSym(sym1, sym2)) {
+			break
+		}
+		v.reset(OpAMD64BTCLconstmodify)
+		v.AuxInt = ValAndOff(valoff1).add(off2)
+		v.Aux = mergeSym(sym1, sym2)
+		v.AddArg(base)
+		v.AddArg(mem)
+		return true
+	}
+	return false
+}
+func rewriteValueAMD64_OpAMD64BTCLmodify_0(v *Value) bool {
+	// match: (BTCLmodify [off1] {sym} (ADDQconst [off2] base) val mem)
+	// cond: is32Bit(off1+off2)
+	// result: (BTCLmodify [off1+off2] {sym} base val mem)
+	for {
+		off1 := v.AuxInt
+		sym := v.Aux
+		_ = v.Args[2]
+		v_0 := v.Args[0]
+		if v_0.Op != OpAMD64ADDQconst {
+			break
+		}
+		off2 := v_0.AuxInt
+		base := v_0.Args[0]
+		val := v.Args[1]
+		mem := v.Args[2]
+		if !(is32Bit(off1 + off2)) {
+			break
+		}
+		v.reset(OpAMD64BTCLmodify)
+		v.AuxInt = off1 + off2
+		v.Aux = sym
+		v.AddArg(base)
+		v.AddArg(val)
+		v.AddArg(mem)
+		return true
+	}
+	// match: (BTCLmodify [off1] {sym1} (LEAQ [off2] {sym2} base) val mem)
+	// cond: is32Bit(off1+off2) && canMergeSym(sym1, sym2)
+	// result: (BTCLmodify [off1+off2] {mergeSym(sym1,sym2)} base val mem)
+	for {
+		off1 := v.AuxInt
+		sym1 := v.Aux
+		_ = v.Args[2]
+		v_0 := v.Args[0]
+		if v_0.Op != OpAMD64LEAQ {
+			break
+		}
+		off2 := v_0.AuxInt
+		sym2 := v_0.Aux
+		base := v_0.Args[0]
+		val := v.Args[1]
+		mem := v.Args[2]
+		if !(is32Bit(off1+off2) && canMergeSym(sym1, sym2)) {
+			break
+		}
+		v.reset(OpAMD64BTCLmodify)
+		v.AuxInt = off1 + off2
+		v.Aux = mergeSym(sym1, sym2)
+		v.AddArg(base)
+		v.AddArg(val)
+		v.AddArg(mem)
+		return true
+	}
+	return false
+}
+func rewriteValueAMD64_OpAMD64BTCQconst_0(v *Value) bool {
+	// match: (BTCQconst [c] (XORQconst [d] x))
+	// cond:
+	// result: (XORQconst [d ^ 1<<uint32(c)] x)
+	for {
+		c := v.AuxInt
+		v_0 := v.Args[0]
+		if v_0.Op != OpAMD64XORQconst {
+			break
+		}
+		d := v_0.AuxInt
+		x := v_0.Args[0]
+		v.reset(OpAMD64XORQconst)
+		v.AuxInt = d ^ 1<<uint32(c)
+		v.AddArg(x)
+		return true
+	}
+	// match: (BTCQconst [c] (BTCQconst [d] x))
+	// cond:
+	// result: (XORQconst [1<<uint32(c) ^ 1<<uint32(d)] x)
+	for {
+		c := v.AuxInt
+		v_0 := v.Args[0]
+		if v_0.Op != OpAMD64BTCQconst {
+			break
+		}
+		d := v_0.AuxInt
+		x := v_0.Args[0]
+		v.reset(OpAMD64XORQconst)
+		v.AuxInt = 1<<uint32(c) ^ 1<<uint32(d)
+		v.AddArg(x)
+		return true
+	}
+	// match: (BTCQconst [c] (MOVQconst [d]))
+	// cond:
+	// result: (MOVQconst [d^(1<<uint32(c))])
+	for {
+		c := v.AuxInt
+		v_0 := v.Args[0]
+		if v_0.Op != OpAMD64MOVQconst {
+			break
+		}
+		d := v_0.AuxInt
+		v.reset(OpAMD64MOVQconst)
+		v.AuxInt = d ^ (1 << uint32(c))
+		return true
+	}
+	return false
+}
+func rewriteValueAMD64_OpAMD64BTCQconstmodify_0(v *Value) bool {
+	// match: (BTCQconstmodify [valoff1] {sym} (ADDQconst [off2] base) mem)
+	// cond: ValAndOff(valoff1).canAdd(off2)
+	// result: (BTCQconstmodify [ValAndOff(valoff1).add(off2)] {sym} base mem)
+	for {
+		valoff1 := v.AuxInt
+		sym := v.Aux
+		_ = v.Args[1]
+		v_0 := v.Args[0]
+		if v_0.Op != OpAMD64ADDQconst {
+			break
+		}
+		off2 := v_0.AuxInt
+		base := v_0.Args[0]
+		mem := v.Args[1]
+		if !(ValAndOff(valoff1).canAdd(off2)) {
+			break
+		}
+		v.reset(OpAMD64BTCQconstmodify)
+		v.AuxInt = ValAndOff(valoff1).add(off2)
+		v.Aux = sym
+		v.AddArg(base)
+		v.AddArg(mem)
+		return true
+	}
+	// match: (BTCQconstmodify [valoff1] {sym1} (LEAQ [off2] {sym2} base) mem)
+	// cond: ValAndOff(valoff1).canAdd(off2) && canMergeSym(sym1, sym2)
+	// result: (BTCQconstmodify [ValAndOff(valoff1).add(off2)] {mergeSym(sym1,sym2)} base mem)
+	for {
+		valoff1 := v.AuxInt
+		sym1 := v.Aux
+		_ = v.Args[1]
+		v_0 := v.Args[0]
+		if v_0.Op != OpAMD64LEAQ {
+			break
+		}
+		off2 := v_0.AuxInt
+		sym2 := v_0.Aux
+		base := v_0.Args[0]
+		mem := v.Args[1]
+		if !(ValAndOff(valoff1).canAdd(off2) && canMergeSym(sym1, sym2)) {
+			break
+		}
+		v.reset(OpAMD64BTCQconstmodify)
+		v.AuxInt = ValAndOff(valoff1).add(off2)
+		v.Aux = mergeSym(sym1, sym2)
+		v.AddArg(base)
+		v.AddArg(mem)
+		return true
+	}
+	return false
+}
+func rewriteValueAMD64_OpAMD64BTCQmodify_0(v *Value) bool {
+	// match: (BTCQmodify [off1] {sym} (ADDQconst [off2] base) val mem)
+	// cond: is32Bit(off1+off2)
+	// result: (BTCQmodify [off1+off2] {sym} base val mem)
+	for {
+		off1 := v.AuxInt
+		sym := v.Aux
+		_ = v.Args[2]
+		v_0 := v.Args[0]
+		if v_0.Op != OpAMD64ADDQconst {
+			break
+		}
+		off2 := v_0.AuxInt
+		base := v_0.Args[0]
+		val := v.Args[1]
+		mem := v.Args[2]
+		if !(is32Bit(off1 + off2)) {
+			break
+		}
+		v.reset(OpAMD64BTCQmodify)
+		v.AuxInt = off1 + off2
+		v.Aux = sym
+		v.AddArg(base)
+		v.AddArg(val)
+		v.AddArg(mem)
+		return true
+	}
+	// match: (BTCQmodify [off1] {sym1} (LEAQ [off2] {sym2} base) val mem)
+	// cond: is32Bit(off1+off2) && canMergeSym(sym1, sym2)
+	// result: (BTCQmodify [off1+off2] {mergeSym(sym1,sym2)} base val mem)
+	for {
+		off1 := v.AuxInt
+		sym1 := v.Aux
+		_ = v.Args[2]
+		v_0 := v.Args[0]
+		if v_0.Op != OpAMD64LEAQ {
+			break
+		}
+		off2 := v_0.AuxInt
+		sym2 := v_0.Aux
+		base := v_0.Args[0]
+		val := v.Args[1]
+		mem := v.Args[2]
+		if !(is32Bit(off1+off2) && canMergeSym(sym1, sym2)) {
+			break
+		}
+		v.reset(OpAMD64BTCQmodify)
+		v.AuxInt = off1 + off2
+		v.Aux = mergeSym(sym1, sym2)
+		v.AddArg(base)
+		v.AddArg(val)
+		v.AddArg(mem)
+		return true
+	}
+	return false
+}
 func rewriteValueAMD64_OpAMD64BTLconst_0(v *Value) bool {
 	// match: (BTLconst [c] (SHRQconst [d] x))
 	// cond: (c+d)<64
@@ -4643,6 +5017,160 @@ func rewriteValueAMD64_OpAMD64BTRLconst_0(v *Value) bool {
 		v.AddArg(x)
 		return true
 	}
+	// match: (BTRLconst [c] (ANDLconst [d] x))
+	// cond:
+	// result: (ANDLconst [d &^ 1<<uint32(c)] x)
+	for {
+		c := v.AuxInt
+		v_0 := v.Args[0]
+		if v_0.Op != OpAMD64ANDLconst {
+			break
+		}
+		d := v_0.AuxInt
+		x := v_0.Args[0]
+		v.reset(OpAMD64ANDLconst)
+		v.AuxInt = d &^ 1 << uint32(c)
+		v.AddArg(x)
+		return true
+	}
+	// match: (BTRLconst [c] (BTRLconst [d] x))
+	// cond:
+	// result: (ANDLconst [^(1<<uint32(c) | 1<<uint32(d))] x)
+	for {
+		c := v.AuxInt
+		v_0 := v.Args[0]
+		if v_0.Op != OpAMD64BTRLconst {
+			break
+		}
+		d := v_0.AuxInt
+		x := v_0.Args[0]
+		v.reset(OpAMD64ANDLconst)
+		v.AuxInt = ^(1<<uint32(c) | 1<<uint32(d))
+		v.AddArg(x)
+		return true
+	}
+	// match: (BTRLconst [c] (MOVLconst [d]))
+	// cond:
+	// result: (MOVLconst [d&^(1<<uint32(c))])
+	for {
+		c := v.AuxInt
+		v_0 := v.Args[0]
+		if v_0.Op != OpAMD64MOVLconst {
+			break
+		}
+		d := v_0.AuxInt
+		v.reset(OpAMD64MOVLconst)
+		v.AuxInt = d &^ (1 << uint32(c))
+		return true
+	}
+	return false
+}
+func rewriteValueAMD64_OpAMD64BTRLconstmodify_0(v *Value) bool {
+	// match: (BTRLconstmodify [valoff1] {sym} (ADDQconst [off2] base) mem)
+	// cond: ValAndOff(valoff1).canAdd(off2)
+	// result: (BTRLconstmodify [ValAndOff(valoff1).add(off2)] {sym} base mem)
+	for {
+		valoff1 := v.AuxInt
+		sym := v.Aux
+		_ = v.Args[1]
+		v_0 := v.Args[0]
+		if v_0.Op != OpAMD64ADDQconst {
+			break
+		}
+		off2 := v_0.AuxInt
+		base := v_0.Args[0]
+		mem := v.Args[1]
+		if !(ValAndOff(valoff1).canAdd(off2)) {
+			break
+		}
+		v.reset(OpAMD64BTRLconstmodify)
+		v.AuxInt = ValAndOff(valoff1).add(off2)
+		v.Aux = sym
+		v.AddArg(base)
+		v.AddArg(mem)
+		return true
+	}
+	// match: (BTRLconstmodify [valoff1] {sym1} (LEAQ [off2] {sym2} base) mem)
+	// cond: ValAndOff(valoff1).canAdd(off2) && canMergeSym(sym1, sym2)
+	// result: (BTRLconstmodify [ValAndOff(valoff1).add(off2)] {mergeSym(sym1,sym2)} base mem)
+	for {
+		valoff1 := v.AuxInt
+		sym1 := v.Aux
+		_ = v.Args[1]
+		v_0 := v.Args[0]
+		if v_0.Op != OpAMD64LEAQ {
+			break
+		}
+		off2 := v_0.AuxInt
+		sym2 := v_0.Aux
+		base := v_0.Args[0]
+		mem := v.Args[1]
+		if !(ValAndOff(valoff1).canAdd(off2) && canMergeSym(sym1, sym2)) {
+			break
+		}
+		v.reset(OpAMD64BTRLconstmodify)
+		v.AuxInt = ValAndOff(valoff1).add(off2)
+		v.Aux = mergeSym(sym1, sym2)
+		v.AddArg(base)
+		v.AddArg(mem)
+		return true
+	}
+	return false
+}
+func rewriteValueAMD64_OpAMD64BTRLmodify_0(v *Value) bool {
+	// match: (BTRLmodify [off1] {sym} (ADDQconst [off2] base) val mem)
+	// cond: is32Bit(off1+off2)
+	// result: (BTRLmodify [off1+off2] {sym} base val mem)
+	for {
+		off1 := v.AuxInt
+		sym := v.Aux
+		_ = v.Args[2]
+		v_0 := v.Args[0]
+		if v_0.Op != OpAMD64ADDQconst {
+			break
+		}
+		off2 := v_0.AuxInt
+		base := v_0.Args[0]
+		val := v.Args[1]
+		mem := v.Args[2]
+		if !(is32Bit(off1 + off2)) {
+			break
+		}
+		v.reset(OpAMD64BTRLmodify)
+		v.AuxInt = off1 + off2
+		v.Aux = sym
+		v.AddArg(base)
+		v.AddArg(val)
+		v.AddArg(mem)
+		return true
+	}
+	// match: (BTRLmodify [off1] {sym1} (LEAQ [off2] {sym2} base) val mem)
+	// cond: is32Bit(off1+off2) && canMergeSym(sym1, sym2)
+	// result: (BTRLmodify [off1+off2] {mergeSym(sym1,sym2)} base val mem)
+	for {
+		off1 := v.AuxInt
+		sym1 := v.Aux
+		_ = v.Args[2]
+		v_0 := v.Args[0]
+		if v_0.Op != OpAMD64LEAQ {
+			break
+		}
+		off2 := v_0.AuxInt
+		sym2 := v_0.Aux
+		base := v_0.Args[0]
+		val := v.Args[1]
+		mem := v.Args[2]
+		if !(is32Bit(off1+off2) && canMergeSym(sym1, sym2)) {
+			break
+		}
+		v.reset(OpAMD64BTRLmodify)
+		v.AuxInt = off1 + off2
+		v.Aux = mergeSym(sym1, sym2)
+		v.AddArg(base)
+		v.AddArg(val)
+		v.AddArg(mem)
+		return true
+	}
 	return false
 }
 func rewriteValueAMD64_OpAMD64BTRQconst_0(v *Value) bool {
@@ -4682,6 +5210,160 @@ func rewriteValueAMD64_OpAMD64BTRQconst_0(v *Value) bool {
 		v.AddArg(x)
 		return true
 	}
+	// match: (BTRQconst [c] (ANDQconst [d] x))
+	// cond:
+	// result: (ANDQconst [d &^ 1<<uint32(c)] x)
+	for {
+		c := v.AuxInt
+		v_0 := v.Args[0]
+		if v_0.Op != OpAMD64ANDQconst {
+			break
+		}
+		d := v_0.AuxInt
+		x := v_0.Args[0]
+		v.reset(OpAMD64ANDQconst)
+		v.AuxInt = d &^ 1 << uint32(c)
+		v.AddArg(x)
+		return true
+	}
+	// match: (BTRQconst [c] (BTRQconst [d] x))
+	// cond:
+	// result: (ANDQconst [^(1<<uint32(c) | 1<<uint32(d))] x)
+	for {
+		c := v.AuxInt
+		v_0 := v.Args[0]
+		if v_0.Op != OpAMD64BTRQconst {
+			break
+		}
+		d := v_0.AuxInt
+		x := v_0.Args[0]
+		v.reset(OpAMD64ANDQconst)
+		v.AuxInt = ^(1<<uint32(c) | 1<<uint32(d))
+		v.AddArg(x)
+		return true
+	}
+	// match: (BTRQconst [c] (MOVQconst [d]))
+	// cond:
+	// result: (MOVQconst [d&^(1<<uint32(c))])
+	for {
+		c := v.AuxInt
+		v_0 := v.Args[0]
+		if v_0.Op != OpAMD64MOVQconst {
+			break
+		}
+		d := v_0.AuxInt
+		v.reset(OpAMD64MOVQconst)
+		v.AuxInt = d &^ (1 << uint32(c))
+		return true
+	}
+	return false
+}
+func rewriteValueAMD64_OpAMD64BTRQconstmodify_0(v *Value) bool {
+	// match: (BTRQconstmodify [valoff1] {sym} (ADDQconst [off2] base) mem)
+	// cond: ValAndOff(valoff1).canAdd(off2)
+	// result: (BTRQconstmodify [ValAndOff(valoff1).add(off2)] {sym} base mem)
+	for {
+		valoff1 := v.AuxInt
+		sym := v.Aux
+		_ = v.Args[1]
+		v_0 := v.Args[0]
+		if v_0.Op != OpAMD64ADDQconst {
+			break
+		}
+		off2 := v_0.AuxInt
+		base := v_0.Args[0]
+		mem := v.Args[1]
+		if !(ValAndOff(valoff1).canAdd(off2)) {
+			break
+		}
+		v.reset(OpAMD64BTRQconstmodify)
+		v.AuxInt = ValAndOff(valoff1).add(off2)
+		v.Aux = sym
+		v.AddArg(base)
+		v.AddArg(mem)
+		return true
+	}
+	// match: (BTRQconstmodify [valoff1] {sym1} (LEAQ [off2] {sym2} base) mem)
+	// cond: ValAndOff(valoff1).canAdd(off2) && canMergeSym(sym1, sym2)
+	// result: (BTRQconstmodify [ValAndOff(valoff1).add(off2)] {mergeSym(sym1,sym2)} base mem)
+	for {
+		valoff1 := v.AuxInt
+		sym1 := v.Aux
+		_ = v.Args[1]
+		v_0 := v.Args[0]
+		if v_0.Op != OpAMD64LEAQ {
+			break
+		}
+		off2 := v_0.AuxInt
+		sym2 := v_0.Aux
+		base := v_0.Args[0]
+		mem := v.Args[1]
+		if !(ValAndOff(valoff1).canAdd(off2) && canMergeSym(sym1, sym2)) {
+			break
+		}
+		v.reset(OpAMD64BTRQconstmodify)
+		v.AuxInt = ValAndOff(valoff1).add(off2)
+		v.Aux = mergeSym(sym1, sym2)
+		v.AddArg(base)
+		v.AddArg(mem)
+		return true
+	}
+	return false
+}
+func rewriteValueAMD64_OpAMD64BTRQmodify_0(v *Value) bool {
+	// match: (BTRQmodify [off1] {sym} (ADDQconst [off2] base) val mem)
+	// cond: is32Bit(off1+off2)
+	// result: (BTRQmodify [off1+off2] {sym} base val mem)
+	for {
+		off1 := v.AuxInt
+		sym := v.Aux
+		_ = v.Args[2]
+		v_0 := v.Args[0]
+		if v_0.Op != OpAMD64ADDQconst {
+			break
+		}
+		off2 := v_0.AuxInt
+		base := v_0.Args[0]
+		val := v.Args[1]
+		mem := v.Args[2]
+		if !(is32Bit(off1 + off2)) {
+			break
+		}
+		v.reset(OpAMD64BTRQmodify)
+		v.AuxInt = off1 + off2
+		v.Aux = sym
+		v.AddArg(base)
+		v.AddArg(val)
+		v.AddArg(mem)
+		return true
+	}
+	// match: (BTRQmodify [off1] {sym1} (LEAQ [off2] {sym2} base) val mem)
+	// cond: is32Bit(off1+off2) && canMergeSym(sym1, sym2)
+	// result: (BTRQmodify [off1+off2] {mergeSym(sym1,sym2)} base val mem)
+	for {
+		off1 := v.AuxInt
+		sym1 := v.Aux
+		_ = v.Args[2]
+		v_0 := v.Args[0]
+		if v_0.Op != OpAMD64LEAQ {
+			break
+		}
+		off2 := v_0.AuxInt
+		sym2 := v_0.Aux
+		base := v_0.Args[0]
+		val := v.Args[1]
+		mem := v.Args[2]
+		if !(is32Bit(off1+off2) && canMergeSym(sym1, sym2)) {
+			break
+		}
+		v.reset(OpAMD64BTRQmodify)
+		v.AuxInt = off1 + off2
+		v.Aux = mergeSym(sym1, sym2)
+		v.AddArg(base)
+		v.AddArg(val)
+		v.AddArg(mem)
+		return true
+	}
 	return false
 }
 func rewriteValueAMD64_OpAMD64BTSLconst_0(v *Value) bool {
@@ -4721,6 +5403,160 @@ func rewriteValueAMD64_OpAMD64BTSLconst_0(v *Value) bool {
 		v.AddArg(x)
 		return true
 	}
+	// match: (BTSLconst [c] (ORLconst [d] x))
+	// cond:
+	// result: (ORLconst [d | 1<<uint32(c)] x)
+	for {
+		c := v.AuxInt
+		v_0 := v.Args[0]
+		if v_0.Op != OpAMD64ORLconst {
+			break
+		}
+		d := v_0.AuxInt
+		x := v_0.Args[0]
+		v.reset(OpAMD64ORLconst)
+		v.AuxInt = d | 1<<uint32(c)
+		v.AddArg(x)
+		return true
+	}
+	// match: (BTSLconst [c] (BTSLconst [d] x))
+	// cond:
+	// result: (ORLconst [1<<uint32(d) | 1<<uint32(c)] x)
+	for {
+		c := v.AuxInt
+		v_0 := v.Args[0]
+		if v_0.Op != OpAMD64BTSLconst {
+			break
+		}
+		d := v_0.AuxInt
+		x := v_0.Args[0]
+		v.reset(OpAMD64ORLconst)
+		v.AuxInt = 1<<uint32(d) | 1<<uint32(c)
+		v.AddArg(x)
+		return true
+	}
+	// match: (BTSLconst [c] (MOVLconst [d]))
+	// cond:
+	// result: (MOVLconst [d|(1<<uint32(c))])
+	for {
+		c := v.AuxInt
+		v_0 := v.Args[0]
+		if v_0.Op != OpAMD64MOVLconst {
+			break
+		}
+		d := v_0.AuxInt
+		v.reset(OpAMD64MOVLconst)
+		v.AuxInt = d | (1 << uint32(c))
+		return true
+	}
+	return false
+}
+func rewriteValueAMD64_OpAMD64BTSLconstmodify_0(v *Value) bool {
+	// match: (BTSLconstmodify [valoff1] {sym} (ADDQconst [off2] base) mem)
+	// cond: ValAndOff(valoff1).canAdd(off2)
+	// result: (BTSLconstmodify [ValAndOff(valoff1).add(off2)] {sym} base mem)
+	for {
+		valoff1 := v.AuxInt
+		sym := v.Aux
+		_ = v.Args[1]
+		v_0 := v.Args[0]
+		if v_0.Op != OpAMD64ADDQconst {
+			break
+		}
+		off2 := v_0.AuxInt
+		base := v_0.Args[0]
+		mem := v.Args[1]
+		if !(ValAndOff(valoff1).canAdd(off2)) {
+			break
+		}
+		v.reset(OpAMD64BTSLconstmodify)
+		v.AuxInt = ValAndOff(valoff1).add(off2)
+		v.Aux = sym
+		v.AddArg(base)
+		v.AddArg(mem)
+		return true
+	}
+	// match: (BTSLconstmodify [valoff1] {sym1} (LEAQ [off2] {sym2} base) mem)
+	// cond: ValAndOff(valoff1).canAdd(off2) && canMergeSym(sym1, sym2)
+	// result: (BTSLconstmodify [ValAndOff(valoff1).add(off2)] {mergeSym(sym1,sym2)} base mem)
+	for {
+		valoff1 := v.AuxInt
+		sym1 := v.Aux
+		_ = v.Args[1]
+		v_0 := v.Args[0]
+		if v_0.Op != OpAMD64LEAQ {
+			break
+		}
+		off2 := v_0.AuxInt
+		sym2 := v_0.Aux
+		base := v_0.Args[0]
+		mem := v.Args[1]
+		if !(ValAndOff(valoff1).canAdd(off2) && canMergeSym(sym1, sym2)) {
+			break
+		}
+		v.reset(OpAMD64BTSLconstmodify)
+		v.AuxInt = ValAndOff(valoff1).add(off2)
+		v.Aux = mergeSym(sym1, sym2)
+		v.AddArg(base)
+		v.AddArg(mem)
+		return true
+	}
+	return false
+}
+func rewriteValueAMD64_OpAMD64BTSLmodify_0(v *Value) bool {
+	// match: (BTSLmodify [off1] {sym} (ADDQconst [off2] base) val mem)
+	// cond: is32Bit(off1+off2)
+	// result: (BTSLmodify [off1+off2] {sym} base val mem)
+	for {
+		off1 := v.AuxInt
+		sym := v.Aux
+		_ = v.Args[2]
+		v_0 := v.Args[0]
+		if v_0.Op != OpAMD64ADDQconst {
+			break
+		}
+		off2 := v_0.AuxInt
+		base := v_0.Args[0]
+		val := v.Args[1]
+		mem := v.Args[2]
+		if !(is32Bit(off1 + off2)) {
+			break
+		}
+		v.reset(OpAMD64BTSLmodify)
+		v.AuxInt = off1 + off2
+		v.Aux = sym
+		v.AddArg(base)
+		v.AddArg(val)
+		v.AddArg(mem)
+		return true
+	}
+	// match: (BTSLmodify [off1] {sym1} (LEAQ [off2] {sym2} base) val mem)
+	// cond: is32Bit(off1+off2) && canMergeSym(sym1, sym2)
+	// result: (BTSLmodify [off1+off2] {mergeSym(sym1,sym2)} base val mem)
+	for {
+		off1 := v.AuxInt
+		sym1 := v.Aux
+		_ = v.Args[2]
+		v_0 := v.Args[0]
+		if v_0.Op != OpAMD64LEAQ {
+			break
+		}
+		off2 := v_0.AuxInt
+		sym2 := v_0.Aux
+		base := v_0.Args[0]
+		val := v.Args[1]
+		mem := v.Args[2]
+		if !(is32Bit(off1+off2) && canMergeSym(sym1, sym2)) {
+			break
+		}
+		v.reset(OpAMD64BTSLmodify)
+		v.AuxInt = off1 + off2
+		v.Aux = mergeSym(sym1, sym2)
+		v.AddArg(base)
+		v.AddArg(val)
+		v.AddArg(mem)
+		return true
+	}
 	return false
 }
 func rewriteValueAMD64_OpAMD64BTSQconst_0(v *Value) bool {
@@ -4760,6 +5596,160 @@ func rewriteValueAMD64_OpAMD64BTSQconst_0(v *Value) bool {
 		v.AddArg(x)
 		return true
 	}
+	// match: (BTSQconst [c] (ORQconst [d] x))
+	// cond:
+	// result: (ORQconst [d | 1<<uint32(c)] x)
+	for {
+		c := v.AuxInt
+		v_0 := v.Args[0]
+		if v_0.Op != OpAMD64ORQconst {
+			break
+		}
+		d := v_0.AuxInt
+		x := v_0.Args[0]
+		v.reset(OpAMD64ORQconst)
+		v.AuxInt = d | 1<<uint32(c)
+		v.AddArg(x)
+		return true
+	}
+	// match: (BTSQconst [c] (BTSQconst [d] x))
+	// cond:
+	// result: (ORQconst [1<<uint32(d) | 1<<uint32(c)] x)
+	for {
+		c := v.AuxInt
+		v_0 := v.Args[0]
+		if v_0.Op != OpAMD64BTSQconst {
+			break
+		}
+		d := v_0.AuxInt
+		x := v_0.Args[0]
+		v.reset(OpAMD64ORQconst)
+		v.AuxInt = 1<<uint32(d) | 1<<uint32(c)
+		v.AddArg(x)
+		return true
+	}
+	// match: (BTSQconst [c] (MOVQconst [d]))
+	// cond:
+	// result: (MOVQconst [d|(1<<uint32(c))])
+	for {
+		c := v.AuxInt
+		v_0 := v.Args[0]
+		if v_0.Op != OpAMD64MOVQconst {
+			break
+		}
+		d := v_0.AuxInt
+		v.reset(OpAMD64MOVQconst)
+		v.AuxInt = d | (1 << uint32(c))
+		return true
+	}
+	return false
+}
+func rewriteValueAMD64_OpAMD64BTSQconstmodify_0(v *Value) bool {
+	// match: (BTSQconstmodify [valoff1] {sym} (ADDQconst [off2] base) mem)
+	// cond: ValAndOff(valoff1).canAdd(off2)
+	// result: (BTSQconstmodify [ValAndOff(valoff1).add(off2)] {sym} base mem)
+	for {
+		valoff1 := v.AuxInt
+		sym := v.Aux
+		_ = v.Args[1]
+		v_0 := v.Args[0]
+		if v_0.Op != OpAMD64ADDQconst {
+			break
+		}
+		off2 := v_0.AuxInt
+		base := v_0.Args[0]
+		mem := v.Args[1]
+		if !(ValAndOff(valoff1).canAdd(off2)) {
+			break
+		}
+		v.reset(OpAMD64BTSQconstmodify)
+		v.AuxInt = ValAndOff(valoff1).add(off2)
+		v.Aux = sym
+		v.AddArg(base)
+		v.AddArg(mem)
+		return true
+	}
+	// match: (BTSQconstmodify [valoff1] {sym1} (LEAQ [off2] {sym2} base) mem)
+	// cond: ValAndOff(valoff1).canAdd(off2) && canMergeSym(sym1, sym2)
+	// result: (BTSQconstmodify [ValAndOff(valoff1).add(off2)] {mergeSym(sym1,sym2)} base mem)
+	for {
+		valoff1 := v.AuxInt
+		sym1 := v.Aux
+		_ = v.Args[1]
+		v_0 := v.Args[0]
+		if v_0.Op != OpAMD64LEAQ {
+			break
+		}
+		off2 := v_0.AuxInt
+		sym2 := v_0.Aux
+		base := v_0.Args[0]
+		mem := v.Args[1]
+		if !(ValAndOff(valoff1).canAdd(off2) && canMergeSym(sym1, sym2)) {
+			break
+		}
+		v.reset(OpAMD64BTSQconstmodify)
+		v.AuxInt = ValAndOff(valoff1).add(off2)
+		v.Aux = mergeSym(sym1, sym2)
+		v.AddArg(base)
+		v.AddArg(mem)
+		return true
+	}
+	return false
+}
+func rewriteValueAMD64_OpAMD64BTSQmodify_0(v *Value) bool {
+	// match: (BTSQmodify [off1] {sym} (ADDQconst [off2] base) val mem)
+	// cond: is32Bit(off1+off2)
+	// result: (BTSQmodify [off1+off2] {sym} base val mem)
+	for {
+		off1 := v.AuxInt
+		sym := v.Aux
+		_ = v.Args[2]
+		v_0 := v.Args[0]
+		if v_0.Op != OpAMD64ADDQconst {
+			break
+		}
+		off2 := v_0.AuxInt
+		base := v_0.Args[0]
+		val := v.Args[1]
+		mem := v.Args[2]
+		if !(is32Bit(off1 + off2)) {
+			break
+		}
+		v.reset(OpAMD64BTSQmodify)
+		v.AuxInt = off1 + off2
+		v.Aux = sym
+		v.AddArg(base)
+		v.AddArg(val)
+		v.AddArg(mem)
+		return true
+	}
+	// match: (BTSQmodify [off1] {sym1} (LEAQ [off2] {sym2} base) val mem)
+	// cond: is32Bit(off1+off2) && canMergeSym(sym1, sym2)
+	// result: (BTSQmodify [off1+off2] {mergeSym(sym1,sym2)} base val mem)
+	for {
+		off1 := v.AuxInt
+		sym1 := v.Aux
+		_ = v.Args[2]
+		v_0 := v.Args[0]
+		if v_0.Op != OpAMD64LEAQ {
+			break
+		}
+		off2 := v_0.AuxInt
+		sym2 := v_0.Aux
+		base := v_0.Args[0]
+		val := v.Args[1]
+		mem := v.Args[2]
+		if !(is32Bit(off1+off2) && canMergeSym(sym1, sym2)) {
+			break
+		}
+		v.reset(OpAMD64BTSQmodify)
+		v.AuxInt = off1 + off2
+		v.Aux = mergeSym(sym1, sym2)
+		v.AddArg(base)
+		v.AddArg(val)
+		v.AddArg(mem)
+		return true
+	}
 	return false
 }
 func rewriteValueAMD64_OpAMD64CMOVLCC_0(v *Value) bool {
@@ -15876,8 +16866,140 @@ func rewriteValueAMD64_OpAMD64MOVLstore_20(v *Value) bool {
 		v.AddArg(mem)
 		return true
 	}
+	// match: (MOVLstore {sym} [off] ptr y:(BTCL l:(MOVLload [off] {sym} ptr mem) x) mem)
+	// cond: y.Uses==1 && l.Uses==1 && clobber(y) && clobber(l)
+	// result: (BTCLmodify [off] {sym} ptr x mem)
+	for {
+		off := v.AuxInt
+		sym := v.Aux
+		_ = v.Args[2]
+		ptr := v.Args[0]
+		y := v.Args[1]
+		if y.Op != OpAMD64BTCL {
+			break
+		}
+		_ = y.Args[1]
+		l := y.Args[0]
+		if l.Op != OpAMD64MOVLload {
+			break
+		}
+		if l.AuxInt != off {
+			break
+		}
+		if l.Aux != sym {
+			break
+		}
+		_ = l.Args[1]
+		if ptr != l.Args[0] {
+			break
+		}
+		mem := l.Args[1]
+		x := y.Args[1]
+		if mem != v.Args[2] {
+			break
+		}
+		if !(y.Uses == 1 && l.Uses == 1 && clobber(y) && clobber(l)) {
+			break
+		}
+		v.reset(OpAMD64BTCLmodify)
+		v.AuxInt = off
+		v.Aux = sym
+		v.AddArg(ptr)
+		v.AddArg(x)
+		v.AddArg(mem)
+		return true
+	}
+	// match: (MOVLstore {sym} [off] ptr y:(BTRL l:(MOVLload [off] {sym} ptr mem) x) mem)
+	// cond: y.Uses==1 && l.Uses==1 && clobber(y) && clobber(l)
+	// result: (BTRLmodify [off] {sym} ptr x mem)
+	for {
+		off := v.AuxInt
+		sym := v.Aux
+		_ = v.Args[2]
+		ptr := v.Args[0]
+		y := v.Args[1]
+		if y.Op != OpAMD64BTRL {
+			break
+		}
+		_ = y.Args[1]
+		l := y.Args[0]
+		if l.Op != OpAMD64MOVLload {
+			break
+		}
+		if l.AuxInt != off {
+			break
+		}
+		if l.Aux != sym {
+			break
+		}
+		_ = l.Args[1]
+		if ptr != l.Args[0] {
+			break
+		}
+		mem := l.Args[1]
+		x := y.Args[1]
+		if mem != v.Args[2] {
+			break
+		}
+		if !(y.Uses == 1 && l.Uses == 1 && clobber(y) && clobber(l)) {
+			break
+		}
+		v.reset(OpAMD64BTRLmodify)
+		v.AuxInt = off
+		v.Aux = sym
+		v.AddArg(ptr)
+		v.AddArg(x)
+		v.AddArg(mem)
+		return true
+	}
+	// match: (MOVLstore {sym} [off] ptr y:(BTSL l:(MOVLload [off] {sym} ptr mem) x) mem)
+	// cond: y.Uses==1 && l.Uses==1 && clobber(y) && clobber(l)
+	// result: (BTSLmodify [off] {sym} ptr x mem)
+	for {
+		off := v.AuxInt
+		sym := v.Aux
+		_ = v.Args[2]
+		ptr := v.Args[0]
+		y := v.Args[1]
+		if y.Op != OpAMD64BTSL {
+			break
+		}
+		_ = y.Args[1]
+		l := y.Args[0]
+		if l.Op != OpAMD64MOVLload {
+			break
+		}
+		if l.AuxInt != off {
+			break
+		}
+		if l.Aux != sym {
+			break
+		}
+		_ = l.Args[1]
+		if ptr != l.Args[0] {
+			break
+		}
+		mem := l.Args[1]
+		x := y.Args[1]
+		if mem != v.Args[2] {
+			break
+		}
+		if !(y.Uses == 1 && l.Uses == 1 && clobber(y) && clobber(l)) {
+			break
+		}
+		v.reset(OpAMD64BTSLmodify)
+		v.AuxInt = off
+		v.Aux = sym
+		v.AddArg(ptr)
+		v.AddArg(x)
+		v.AddArg(mem)
+		return true
+	}
+	return false
+}
+func rewriteValueAMD64_OpAMD64MOVLstore_30(v *Value) bool {
 	// match: (MOVLstore [off] {sym} ptr a:(ADDLconst [c] l:(MOVLload [off] {sym} ptr2 mem)) mem)
-	// cond: isSamePtr(ptr, ptr2) && a.Uses == 1 && l.Uses == 1 && validValAndOff(c,off)
+	// cond: isSamePtr(ptr, ptr2) && a.Uses == 1 && l.Uses == 1 && validValAndOff(c,off) && clobber(l) && clobber(a)
 	// result: (ADDLconstmodify {sym} [makeValAndOff(c,off)] ptr mem)
 	for {
 		off := v.AuxInt
@@ -15905,7 +17027,7 @@ func rewriteValueAMD64_OpAMD64MOVLstore_20(v *Value) bool {
 		if mem != v.Args[2] {
 			break
 		}
-		if !(isSamePtr(ptr, ptr2) && a.Uses == 1 && l.Uses == 1 && validValAndOff(c, off)) {
+		if !(isSamePtr(ptr, ptr2) && a.Uses == 1 && l.Uses == 1 && validValAndOff(c, off) && clobber(l) && clobber(a)) {
 			break
 		}
 		v.reset(OpAMD64ADDLconstmodify)
@@ -15916,7 +17038,7 @@ func rewriteValueAMD64_OpAMD64MOVLstore_20(v *Value) bool {
 		return true
 	}
 	// match: (MOVLstore [off] {sym} ptr a:(ANDLconst [c] l:(MOVLload [off] {sym} ptr2 mem)) mem)
-	// cond: isSamePtr(ptr, ptr2) && a.Uses == 1 && l.Uses == 1 && validValAndOff(c,off)
+	// cond: isSamePtr(ptr, ptr2) && a.Uses == 1 && l.Uses == 1 && validValAndOff(c,off) && clobber(l) && clobber(a)
 	// result: (ANDLconstmodify {sym} [makeValAndOff(c,off)] ptr mem)
 	for {
 		off := v.AuxInt
@@ -15944,7 +17066,7 @@ func rewriteValueAMD64_OpAMD64MOVLstore_20(v *Value) bool {
 		if mem != v.Args[2] {
 			break
 		}
-		if !(isSamePtr(ptr, ptr2) && a.Uses == 1 && l.Uses == 1 && validValAndOff(c, off)) {
+		if !(isSamePtr(ptr, ptr2) && a.Uses == 1 && l.Uses == 1 && validValAndOff(c, off) && clobber(l) && clobber(a)) {
 			break
 		}
 		v.reset(OpAMD64ANDLconstmodify)
@@ -15955,7 +17077,7 @@ func rewriteValueAMD64_OpAMD64MOVLstore_20(v *Value) bool {
 		return true
 	}
 	// match: (MOVLstore [off] {sym} ptr a:(ORLconst [c] l:(MOVLload [off] {sym} ptr2 mem)) mem)
-	// cond: isSamePtr(ptr, ptr2) && a.Uses == 1 && l.Uses == 1 && validValAndOff(c,off)
+	// cond: isSamePtr(ptr, ptr2) && a.Uses == 1 && l.Uses == 1 && validValAndOff(c,off) && clobber(l) && clobber(a)
 	// result: (ORLconstmodify {sym} [makeValAndOff(c,off)] ptr mem)
 	for {
 		off := v.AuxInt
@@ -15983,7 +17105,7 @@ func rewriteValueAMD64_OpAMD64MOVLstore_20(v *Value) bool {
 		if mem != v.Args[2] {
 			break
 		}
-		if !(isSamePtr(ptr, ptr2) && a.Uses == 1 && l.Uses == 1 && validValAndOff(c, off)) {
+		if !(isSamePtr(ptr, ptr2) && a.Uses == 1 && l.Uses == 1 && validValAndOff(c, off) && clobber(l) && clobber(a)) {
 			break
 		}
 		v.reset(OpAMD64ORLconstmodify)
@@ -15993,11 +17115,8 @@ func rewriteValueAMD64_OpAMD64MOVLstore_20(v *Value) bool {
 		v.AddArg(mem)
 		return true
 	}
-	return false
-}
-func rewriteValueAMD64_OpAMD64MOVLstore_30(v *Value) bool {
 	// match: (MOVLstore [off] {sym} ptr a:(XORLconst [c] l:(MOVLload [off] {sym} ptr2 mem)) mem)
-	// cond: isSamePtr(ptr, ptr2) && a.Uses == 1 && l.Uses == 1 && validValAndOff(c,off)
+	// cond: isSamePtr(ptr, ptr2) && a.Uses == 1 && l.Uses == 1 && validValAndOff(c,off) && clobber(l) && clobber(a)
 	// result: (XORLconstmodify {sym} [makeValAndOff(c,off)] ptr mem)
 	for {
 		off := v.AuxInt
@@ -16025,7 +17144,7 @@ func rewriteValueAMD64_OpAMD64MOVLstore_30(v *Value) bool {
 		if mem != v.Args[2] {
 			break
 		}
-		if !(isSamePtr(ptr, ptr2) && a.Uses == 1 && l.Uses == 1 && validValAndOff(c, off)) {
+		if !(isSamePtr(ptr, ptr2) && a.Uses == 1 && l.Uses == 1 && validValAndOff(c, off) && clobber(l) && clobber(a)) {
 			break
 		}
 		v.reset(OpAMD64XORLconstmodify)
@@ -16035,6 +17154,123 @@ func rewriteValueAMD64_OpAMD64MOVLstore_30(v *Value) bool {
 		v.AddArg(mem)
 		return true
 	}
+	// match: (MOVLstore [off] {sym} ptr a:(BTCLconst [c] l:(MOVLload [off] {sym} ptr2 mem)) mem)
+	// cond: isSamePtr(ptr, ptr2) && a.Uses == 1 && l.Uses == 1 && validValAndOff(c,off) && clobber(l) && clobber(a)
+	// result: (BTCLconstmodify {sym} [makeValAndOff(c,off)] ptr mem)
+	for {
+		off := v.AuxInt
+		sym := v.Aux
+		_ = v.Args[2]
+		ptr := v.Args[0]
+		a := v.Args[1]
+		if a.Op != OpAMD64BTCLconst {
+			break
+		}
+		c := a.AuxInt
+		l := a.Args[0]
+		if l.Op != OpAMD64MOVLload {
+			break
+		}
+		if l.AuxInt != off {
+			break
+		}
+		if l.Aux != sym {
+			break
+		}
+		_ = l.Args[1]
+		ptr2 := l.Args[0]
+		mem := l.Args[1]
+		if mem != v.Args[2] {
+			break
+		}
+		if !(isSamePtr(ptr, ptr2) && a.Uses == 1 && l.Uses == 1 && validValAndOff(c, off) && clobber(l) && clobber(a)) {
+			break
+		}
+		v.reset(OpAMD64BTCLconstmodify)
+		v.AuxInt = makeValAndOff(c, off)
+		v.Aux = sym
+		v.AddArg(ptr)
+		v.AddArg(mem)
+		return true
+	}
+	// match: (MOVLstore [off] {sym} ptr a:(BTRLconst [c] l:(MOVLload [off] {sym} ptr2 mem)) mem)
+	// cond: isSamePtr(ptr, ptr2) && a.Uses == 1 && l.Uses == 1 && validValAndOff(c,off) && clobber(l) && clobber(a)
+	// result: (BTRLconstmodify {sym} [makeValAndOff(c,off)] ptr mem)
+	for {
+		off := v.AuxInt
+		sym := v.Aux
+		_ = v.Args[2]
+		ptr := v.Args[0]
+		a := v.Args[1]
+		if a.Op != OpAMD64BTRLconst {
+			break
+		}
+		c := a.AuxInt
+		l := a.Args[0]
+		if l.Op != OpAMD64MOVLload {
+			break
+		}
+		if l.AuxInt != off {
+			break
+		}
+		if l.Aux != sym {
+			break
+		}
+		_ = l.Args[1]
+		ptr2 := l.Args[0]
+		mem := l.Args[1]
+		if mem != v.Args[2] {
+			break
+		}
+		if !(isSamePtr(ptr, ptr2) && a.Uses == 1 && l.Uses == 1 && validValAndOff(c, off) && clobber(l) && clobber(a)) {
+			break
+		}
+		v.reset(OpAMD64BTRLconstmodify)
+		v.AuxInt = makeValAndOff(c, off)
+		v.Aux = sym
+		v.AddArg(ptr)
+		v.AddArg(mem)
+		return true
+	}
+	// match: (MOVLstore [off] {sym} ptr a:(BTSLconst [c] l:(MOVLload [off] {sym} ptr2 mem)) mem)
+	// cond: isSamePtr(ptr, ptr2) && a.Uses == 1 && l.Uses == 1 && validValAndOff(c,off) && clobber(l) && clobber(a)
+	// result: (BTSLconstmodify {sym} [makeValAndOff(c,off)] ptr mem)
+	for {
+		off := v.AuxInt
+		sym := v.Aux
+		_ = v.Args[2]
+		ptr := v.Args[0]
+		a := v.Args[1]
+		if a.Op != OpAMD64BTSLconst {
+			break
+		}
+		c := a.AuxInt
+		l := a.Args[0]
+		if l.Op != OpAMD64MOVLload {
+			break
+		}
+		if l.AuxInt != off {
+			break
+		}
+		if l.Aux != sym {
+			break
+		}
+		_ = l.Args[1]
+		ptr2 := l.Args[0]
+		mem := l.Args[1]
+		if mem != v.Args[2] {
+			break
+		}
+		if !(isSamePtr(ptr, ptr2) && a.Uses == 1 && l.Uses == 1 && validValAndOff(c, off) && clobber(l) && clobber(a)) {
+			break
+		}
+		v.reset(OpAMD64BTSLconstmodify)
+		v.AuxInt = makeValAndOff(c, off)
+		v.Aux = sym
+		v.AddArg(ptr)
+		v.AddArg(mem)
+		return true
+	}
 	// match: (MOVLstore [off] {sym} ptr (MOVLf2i val) mem)
 	// cond:
 	// result: (MOVSSstore [off] {sym} ptr val mem)
@@ -18527,8 +19763,137 @@ func rewriteValueAMD64_OpAMD64MOVQstore_20(v *Value) bool {
 		v.AddArg(mem)
 		return true
 	}
+	// match: (MOVQstore {sym} [off] ptr y:(BTCQ l:(MOVQload [off] {sym} ptr mem) x) mem)
+	// cond: y.Uses==1 && l.Uses==1 && clobber(y) && clobber(l)
+	// result: (BTCQmodify [off] {sym} ptr x mem)
+	for {
+		off := v.AuxInt
+		sym := v.Aux
+		_ = v.Args[2]
+		ptr := v.Args[0]
+		y := v.Args[1]
+		if y.Op != OpAMD64BTCQ {
+			break
+		}
+		_ = y.Args[1]
+		l := y.Args[0]
+		if l.Op != OpAMD64MOVQload {
+			break
+		}
+		if l.AuxInt != off {
+			break
+		}
+		if l.Aux != sym {
+			break
+		}
+		_ = l.Args[1]
+		if ptr != l.Args[0] {
+			break
+		}
+		mem := l.Args[1]
+		x := y.Args[1]
+		if mem != v.Args[2] {
+			break
+		}
+		if !(y.Uses == 1 && l.Uses == 1 && clobber(y) && clobber(l)) {
+			break
+		}
+		v.reset(OpAMD64BTCQmodify)
+		v.AuxInt = off
+		v.Aux = sym
+		v.AddArg(ptr)
+		v.AddArg(x)
+		v.AddArg(mem)
+		return true
+	}
+	// match: (MOVQstore {sym} [off] ptr y:(BTRQ l:(MOVQload [off] {sym} ptr mem) x) mem)
+	// cond: y.Uses==1 && l.Uses==1 && clobber(y) && clobber(l)
+	// result: (BTRQmodify [off] {sym} ptr x mem)
+	for {
+		off := v.AuxInt
+		sym := v.Aux
+		_ = v.Args[2]
+		ptr := v.Args[0]
+		y := v.Args[1]
+		if y.Op != OpAMD64BTRQ {
+			break
+		}
+		_ = y.Args[1]
+		l := y.Args[0]
+		if l.Op != OpAMD64MOVQload {
+			break
+		}
+		if l.AuxInt != off {
+			break
+		}
+		if l.Aux != sym {
+			break
+		}
+		_ = l.Args[1]
+		if ptr != l.Args[0] {
+			break
+		}
+		mem := l.Args[1]
+		x := y.Args[1]
+		if mem != v.Args[2] {
+			break
+		}
+		if !(y.Uses == 1 && l.Uses == 1 && clobber(y) && clobber(l)) {
+			break
+		}
+		v.reset(OpAMD64BTRQmodify)
+		v.AuxInt = off
+		v.Aux = sym
+		v.AddArg(ptr)
+		v.AddArg(x)
+		v.AddArg(mem)
+		return true
+	}
+	// match: (MOVQstore {sym} [off] ptr y:(BTSQ l:(MOVQload [off] {sym} ptr mem) x) mem)
+	// cond: y.Uses==1 && l.Uses==1 && clobber(y) && clobber(l)
+	// result: (BTSQmodify [off] {sym} ptr x mem)
+	for {
+		off := v.AuxInt
+		sym := v.Aux
+		_ = v.Args[2]
+		ptr := v.Args[0]
+		y := v.Args[1]
+		if y.Op != OpAMD64BTSQ {
+			break
+		}
+		_ = y.Args[1]
+		l := y.Args[0]
+		if l.Op != OpAMD64MOVQload {
+			break
+		}
+		if l.AuxInt != off {
+			break
+		}
+		if l.Aux != sym {
+			break
+		}
+		_ = l.Args[1]
+		if ptr != l.Args[0] {
+			break
+		}
+		mem := l.Args[1]
+		x := y.Args[1]
+		if mem != v.Args[2] {
+			break
+		}
+		if !(y.Uses == 1 && l.Uses == 1 && clobber(y) && clobber(l)) {
+			break
+		}
+		v.reset(OpAMD64BTSQmodify)
+		v.AuxInt = off
+		v.Aux = sym
+		v.AddArg(ptr)
+		v.AddArg(x)
+		v.AddArg(mem)
+		return true
+	}
 	// match: (MOVQstore [off] {sym} ptr a:(ADDQconst [c] l:(MOVQload [off] {sym} ptr2 mem)) mem)
-	// cond: isSamePtr(ptr, ptr2) && a.Uses == 1 && l.Uses == 1 && validValAndOff(c,off)
+	// cond: isSamePtr(ptr, ptr2) && a.Uses == 1 && l.Uses == 1 && validValAndOff(c,off) && clobber(l) && clobber(a)
 	// result: (ADDQconstmodify {sym} [makeValAndOff(c,off)] ptr mem)
 	for {
 		off := v.AuxInt
@@ -18556,7 +19921,7 @@ func rewriteValueAMD64_OpAMD64MOVQstore_20(v *Value) bool {
 		if mem != v.Args[2] {
 			break
 		}
-		if !(isSamePtr(ptr, ptr2) && a.Uses == 1 && l.Uses == 1 && validValAndOff(c, off)) {
+		if !(isSamePtr(ptr, ptr2) && a.Uses == 1 && l.Uses == 1 && validValAndOff(c, off) && clobber(l) && clobber(a)) {
 			break
 		}
 		v.reset(OpAMD64ADDQconstmodify)
@@ -18567,7 +19932,7 @@ func rewriteValueAMD64_OpAMD64MOVQstore_20(v *Value) bool {
 		return true
 	}
 	// match: (MOVQstore [off] {sym} ptr a:(ANDQconst [c] l:(MOVQload [off] {sym} ptr2 mem)) mem)
-	// cond: isSamePtr(ptr, ptr2) && a.Uses == 1 && l.Uses == 1 && validValAndOff(c,off)
+	// cond: isSamePtr(ptr, ptr2) && a.Uses == 1 && l.Uses == 1 && validValAndOff(c,off) && clobber(l) && clobber(a)
 	// result: (ANDQconstmodify {sym} [makeValAndOff(c,off)] ptr mem)
 	for {
 		off := v.AuxInt
@@ -18595,7 +19960,7 @@ func rewriteValueAMD64_OpAMD64MOVQstore_20(v *Value) bool {
 		if mem != v.Args[2] {
 			break
 		}
-		if !(isSamePtr(ptr, ptr2) && a.Uses == 1 && l.Uses == 1 && validValAndOff(c, off)) {
+		if !(isSamePtr(ptr, ptr2) && a.Uses == 1 && l.Uses == 1 && validValAndOff(c, off) && clobber(l) && clobber(a)) {
 			break
 		}
 		v.reset(OpAMD64ANDQconstmodify)
@@ -18606,7 +19971,7 @@ func rewriteValueAMD64_OpAMD64MOVQstore_20(v *Value) bool {
 		return true
 	}
 	// match: (MOVQstore [off] {sym} ptr a:(ORQconst [c] l:(MOVQload [off] {sym} ptr2 mem)) mem)
-	// cond: isSamePtr(ptr, ptr2) && a.Uses == 1 && l.Uses == 1 && validValAndOff(c,off)
+	// cond: isSamePtr(ptr, ptr2) && a.Uses == 1 && l.Uses == 1 && validValAndOff(c,off) && clobber(l) && clobber(a)
 	// result: (ORQconstmodify {sym} [makeValAndOff(c,off)] ptr mem)
 	for {
 		off := v.AuxInt
@@ -18634,7 +19999,7 @@ func rewriteValueAMD64_OpAMD64MOVQstore_20(v *Value) bool {
 		if mem != v.Args[2] {
 			break
 		}
-		if !(isSamePtr(ptr, ptr2) && a.Uses == 1 && l.Uses == 1 && validValAndOff(c, off)) {
+		if !(isSamePtr(ptr, ptr2) && a.Uses == 1 && l.Uses == 1 && validValAndOff(c, off) && clobber(l) && clobber(a)) {
 			break
 		}
 		v.reset(OpAMD64ORQconstmodify)
@@ -18645,7 +20010,7 @@ func rewriteValueAMD64_OpAMD64MOVQstore_20(v *Value) bool {
 		return true
 	}
 	// match: (MOVQstore [off] {sym} ptr a:(XORQconst [c] l:(MOVQload [off] {sym} ptr2 mem)) mem)
-	// cond: isSamePtr(ptr, ptr2) && a.Uses == 1 && l.Uses == 1 && validValAndOff(c,off)
+	// cond: isSamePtr(ptr, ptr2) && a.Uses == 1 && l.Uses == 1 && validValAndOff(c,off) && clobber(l) && clobber(a)
 	// result: (XORQconstmodify {sym} [makeValAndOff(c,off)] ptr mem)
 	for {
 		off := v.AuxInt
@@ -18673,7 +20038,7 @@ func rewriteValueAMD64_OpAMD64MOVQstore_20(v *Value) bool {
 		if mem != v.Args[2] {
 			break
 		}
-		if !(isSamePtr(ptr, ptr2) && a.Uses == 1 && l.Uses == 1 && validValAndOff(c, off)) {
+		if !(isSamePtr(ptr, ptr2) && a.Uses == 1 && l.Uses == 1 && validValAndOff(c, off) && clobber(l) && clobber(a)) {
 			break
 		}
 		v.reset(OpAMD64XORQconstmodify)
@@ -18683,6 +20048,126 @@ func rewriteValueAMD64_OpAMD64MOVQstore_20(v *Value) bool {
 		v.AddArg(mem)
 		return true
 	}
+	// match: (MOVQstore [off] {sym} ptr a:(BTCQconst [c] l:(MOVQload [off] {sym} ptr2 mem)) mem)
+	// cond: isSamePtr(ptr, ptr2) && a.Uses == 1 && l.Uses == 1 && validValAndOff(c,off) && clobber(l) && clobber(a)
+	// result: (BTCQconstmodify {sym} [makeValAndOff(c,off)] ptr mem)
+	for {
+		off := v.AuxInt
+		sym := v.Aux
+		_ = v.Args[2]
+		ptr := v.Args[0]
+		a := v.Args[1]
+		if a.Op != OpAMD64BTCQconst {
+			break
+		}
+		c := a.AuxInt
+		l := a.Args[0]
+		if l.Op != OpAMD64MOVQload {
+			break
+		}
+		if l.AuxInt != off {
+			break
+		}
+		if l.Aux != sym {
+			break
+		}
+		_ = l.Args[1]
+		ptr2 := l.Args[0]
+		mem := l.Args[1]
+		if mem != v.Args[2] {
+			break
+		}
+		if !(isSamePtr(ptr, ptr2) && a.Uses == 1 && l.Uses == 1 && validValAndOff(c, off) && clobber(l) && clobber(a)) {
+			break
+		}
+		v.reset(OpAMD64BTCQconstmodify)
+		v.AuxInt = makeValAndOff(c, off)
+		v.Aux = sym
+		v.AddArg(ptr)
+		v.AddArg(mem)
+		return true
+	}
+	// match: (MOVQstore [off] {sym} ptr a:(BTRQconst [c] l:(MOVQload [off] {sym} ptr2 mem)) mem)
+	// cond: isSamePtr(ptr, ptr2) && a.Uses == 1 && l.Uses == 1 && validValAndOff(c,off) && clobber(l) && clobber(a)
+	// result: (BTRQconstmodify {sym} [makeValAndOff(c,off)] ptr mem)
+	for {
+		off := v.AuxInt
+		sym := v.Aux
+		_ = v.Args[2]
+		ptr := v.Args[0]
+		a := v.Args[1]
+		if a.Op != OpAMD64BTRQconst {
+			break
+		}
+		c := a.AuxInt
+		l := a.Args[0]
+		if l.Op != OpAMD64MOVQload {
+			break
+		}
+		if l.AuxInt != off {
+			break
+		}
+		if l.Aux != sym {
+			break
+		}
+		_ = l.Args[1]
+		ptr2 := l.Args[0]
+		mem := l.Args[1]
+		if mem != v.Args[2] {
+			break
+		}
+		if !(isSamePtr(ptr, ptr2) && a.Uses == 1 && l.Uses == 1 && validValAndOff(c, off) && clobber(l) && clobber(a)) {
+			break
+		}
+		v.reset(OpAMD64BTRQconstmodify)
+		v.AuxInt = makeValAndOff(c, off)
+		v.Aux = sym
+		v.AddArg(ptr)
+		v.AddArg(mem)
+		return true
+	}
+	return false
+}
+func rewriteValueAMD64_OpAMD64MOVQstore_30(v *Value) bool {
+	// match: (MOVQstore [off] {sym} ptr a:(BTSQconst [c] l:(MOVQload [off] {sym} ptr2 mem)) mem)
+	// cond: isSamePtr(ptr, ptr2) && a.Uses == 1 && l.Uses == 1 && validValAndOff(c,off) && clobber(l) && clobber(a)
+	// result: (BTSQconstmodify {sym} [makeValAndOff(c,off)] ptr mem)
+	for {
+		off := v.AuxInt
+		sym := v.Aux
+		_ = v.Args[2]
+		ptr := v.Args[0]
+		a := v.Args[1]
+		if a.Op != OpAMD64BTSQconst {
+			break
+		}
+		c := a.AuxInt
+		l := a.Args[0]
+		if l.Op != OpAMD64MOVQload {
+			break
+		}
+		if l.AuxInt != off {
+			break
+		}
+		if l.Aux != sym {
+			break
+		}
+		_ = l.Args[1]
+		ptr2 := l.Args[0]
+		mem := l.Args[1]
+		if mem != v.Args[2] {
+			break
+		}
+		if !(isSamePtr(ptr, ptr2) && a.Uses == 1 && l.Uses == 1 && validValAndOff(c, off) && clobber(l) && clobber(a)) {
+			break
+		}
+		v.reset(OpAMD64BTSQconstmodify)
+		v.AuxInt = makeValAndOff(c, off)
+		v.Aux = sym
+		v.AddArg(ptr)
+		v.AddArg(mem)
+		return true
+	}
 	// match: (MOVQstore [off] {sym} ptr (MOVQf2i val) mem)
 	// cond:
 	// result: (MOVSDstore [off] {sym} ptr val mem)
@@ -33141,6 +34626,38 @@ func rewriteValueAMD64_OpAMD64ORLconst_0(v *Value) bool {
 		v.AddArg(x)
 		return true
 	}
+	// match: (ORLconst [c] (ORLconst [d] x))
+	// cond:
+	// result: (ORLconst [c | d] x)
+	for {
+		c := v.AuxInt
+		v_0 := v.Args[0]
+		if v_0.Op != OpAMD64ORLconst {
+			break
+		}
+		d := v_0.AuxInt
+		x := v_0.Args[0]
+		v.reset(OpAMD64ORLconst)
+		v.AuxInt = c | d
+		v.AddArg(x)
+		return true
+	}
+	// match: (ORLconst [c] (BTSLconst [d] x))
+	// cond:
+	// result: (ORLconst [c | 1<<uint32(d)] x)
+	for {
+		c := v.AuxInt
+		v_0 := v.Args[0]
+		if v_0.Op != OpAMD64BTSLconst {
+			break
+		}
+		d := v_0.AuxInt
+		x := v_0.Args[0]
+		v.reset(OpAMD64ORLconst)
+		v.AuxInt = c | 1<<uint32(d)
+		v.AddArg(x)
+		return true
+	}
 	// match: (ORLconst [c] x)
 	// cond: int32(c)==0
 	// result: x
@@ -44160,6 +45677,38 @@ func rewriteValueAMD64_OpAMD64ORQconst_0(v *Value) bool {
 		v.AddArg(x)
 		return true
 	}
+	// match: (ORQconst [c] (ORQconst [d] x))
+	// cond:
+	// result: (ORQconst [c | d] x)
+	for {
+		c := v.AuxInt
+		v_0 := v.Args[0]
+		if v_0.Op != OpAMD64ORQconst {
+			break
+		}
+		d := v_0.AuxInt
+		x := v_0.Args[0]
+		v.reset(OpAMD64ORQconst)
+		v.AuxInt = c | d
+		v.AddArg(x)
+		return true
+	}
+	// match: (ORQconst [c] (BTSQconst [d] x))
+	// cond:
+	// result: (ORQconst [c | 1<<uint32(d)] x)
+	for {
+		c := v.AuxInt
+		v_0 := v.Args[0]
+		if v_0.Op != OpAMD64BTSQconst {
+			break
+		}
+		d := v_0.AuxInt
+		x := v_0.Args[0]
+		v.reset(OpAMD64ORQconst)
+		v.AuxInt = c | 1<<uint32(d)
+		v.AddArg(x)
+		return true
+	}
 	// match: (ORQconst [0] x)
 	// cond:
 	// result: x
@@ -54886,6 +56435,22 @@ func rewriteValueAMD64_OpAMD64XORLconst_10(v *Value) bool {
 		v.AddArg(x)
 		return true
 	}
+	// match: (XORLconst [c] (BTCLconst [d] x))
+	// cond:
+	// result: (XORLconst [c ^ 1<<uint32(d)] x)
+	for {
+		c := v.AuxInt
+		v_0 := v.Args[0]
+		if v_0.Op != OpAMD64BTCLconst {
+			break
+		}
+		d := v_0.AuxInt
+		x := v_0.Args[0]
+		v.reset(OpAMD64XORLconst)
+		v.AuxInt = c ^ 1<<uint32(d)
+		v.AddArg(x)
+		return true
+	}
 	// match: (XORLconst [c] x)
 	// cond: int32(c)==0
 	// result: x
@@ -55409,6 +56974,22 @@ func rewriteValueAMD64_OpAMD64XORQconst_0(v *Value) bool {
 		v.AddArg(x)
 		return true
 	}
+	// match: (XORQconst [c] (BTCQconst [d] x))
+	// cond:
+	// result: (XORQconst [c ^ 1<<uint32(d)] x)
+	for {
+		c := v.AuxInt
+		v_0 := v.Args[0]
+		if v_0.Op != OpAMD64BTCQconst {
+			break
+		}
+		d := v_0.AuxInt
+		x := v_0.Args[0]
+		v.reset(OpAMD64XORQconst)
+		v.AuxInt = c ^ 1<<uint32(d)
+		v.AddArg(x)
+		return true
+	}
 	// match: (XORQconst [0] x)
 	// cond:
 	// result: x
diff --git a/test/codegen/bits.go b/test/codegen/bits.go
index e95e3f64cd..65d57c8f9f 100644
--- a/test/codegen/bits.go
+++ b/test/codegen/bits.go
@@ -270,6 +270,12 @@ func bitOpOnMem(a []uint32) {
 	a[1] |= 220
 	// amd64:`XORL\s[$]240,\s8\([A-Z]+\)`
 	a[2] ^= 240
+	// amd64:`BTRL\s[$]15,\s12\([A-Z]+\)`,-`ANDL`
+	a[3] &= 0xffff7fff
+	// amd64:`BTSL\s[$]14,\s16\([A-Z]+\)`,-`ORL`
+	a[4] |= 0x4000
+	// amd64:`BTCL\s[$]13,\s20\([A-Z]+\)`,-`XORL`
+	a[5] ^= 0x2000
 }
 
 // Check AND masking on arm64 (Issue #19857)