From 891ba1526a02771bb60477ca495a724ec2a7937b Mon Sep 17 00:00:00 2001
From: AZero13 <gfunni234@gmail.com>
Date: Thu, 11 Sep 2025 09:15:23 -0400
Subject: [PATCH 1/2] Pre-commit test (NFC)

---
 .../test/CodeGen/AArch64/and-mask-variable.ll | 30 +++++++++++++++++++
 1 file changed, 30 insertions(+)
 create mode 100644 llvm/test/CodeGen/AArch64/and-mask-variable.ll

diff --git a/llvm/test/CodeGen/AArch64/and-mask-variable.ll b/llvm/test/CodeGen/AArch64/and-mask-variable.ll
new file mode 100644
index 0000000000000..a92f3cf5ec092
--- /dev/null
+++ b/llvm/test/CodeGen/AArch64/and-mask-variable.ll
@@ -0,0 +1,30 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -mtriple=aarch64-none-elf -verify-machineinstrs %s -o - | FileCheck %s --check-prefixes=CHECK,CHECK-SD
+; RUN: llc -mtriple=aarch64-none-elf -global-isel -verify-machineinstrs %s -o - | FileCheck %s --check-prefixes=CHECK,CHECK-GI
+
+define i32 @mask_pair(i32 %x, i32 %y) {
+; CHECK-LABEL: mask_pair:
+; CHECK:       // %bb.0:
+; CHECK-NEXT:    mov w8, #-1 // =0xffffffff
+; CHECK-NEXT:    lsl w8, w8, w1
+; CHECK-NEXT:    and w0, w8, w0
+; CHECK-NEXT:    ret
+  %shl = shl nsw i32 -1, %y
+  %and = and i32 %shl, %x
+  ret i32 %and
+}
+
+define i64 @mask_pair_64(i64 %x, i64 %y) {
+; CHECK-LABEL: mask_pair_64:
+; CHECK:       // %bb.0:
+; CHECK-NEXT:    mov x8, #-1 // =0xffffffffffffffff
+; CHECK-NEXT:    lsl x8, x8, x1
+; CHECK-NEXT:    and x0, x8, x0
+; CHECK-NEXT:    ret
+  %shl = shl nsw i64 -1, %y
+  %and = and i64 %shl, %x
+  ret i64 %and
+}
+;; NOTE: These prefixes are unused and the list is autogenerated. Do not add tests below this line:
+; CHECK-GI: {{.*}}
+; CHECK-SD: {{.*}}

From 90723996e8df05e9399e6f3b62f3be7293ff622a Mon Sep 17 00:00:00 2001
From: AZero13 <gfunni234@gmail.com>
Date: Thu, 11 Sep 2025 09:16:29 -0400
Subject: [PATCH 2/2] [AArch64] shouldFoldMaskToVariableShiftPair should be
 true for scalars up to the biggest legal type

For AArch64, we want to do this for scalars up to the biggest legal type.
---
 llvm/lib/Target/AArch64/AArch64ISelLowering.h | 10 ++
 .../test/CodeGen/AArch64/and-mask-variable.ll | 78 ++++++++++++---
 llvm/test/CodeGen/AArch64/extract-bits.ll     | 98 ++++++++-----------
 llvm/test/CodeGen/AArch64/extract-lowbits.ll  | 66 ++++++-------
 4 files changed, 145 insertions(+), 107 deletions(-)

diff --git a/llvm/lib/Target/AArch64/AArch64ISelLowering.h b/llvm/lib/Target/AArch64/AArch64ISelLowering.h
index 46738365080f9..9665bc871b6b9 100644
--- a/llvm/lib/Target/AArch64/AArch64ISelLowering.h
+++ b/llvm/lib/Target/AArch64/AArch64ISelLowering.h
@@ -300,6 +300,16 @@ class AArch64TargetLowering : public TargetLowering {
   bool shouldFoldConstantShiftPairToMask(const SDNode *N,
                                          CombineLevel Level) const override;
 
+  /// Return true if it is profitable to fold a pair of shifts into a mask.
+  bool shouldFoldMaskToVariableShiftPair(SDValue Y) const override {
+    EVT VT = Y.getValueType();
+
+    if (VT.isVector())
+      return false;
+
+    return VT.getScalarSizeInBits() <= 64;
+  }
+
   bool shouldFoldSelectWithIdentityConstant(unsigned BinOpcode, EVT VT,
                                             unsigned SelectOpcode, SDValue X,
                                             SDValue Y) const override;
diff --git a/llvm/test/CodeGen/AArch64/and-mask-variable.ll b/llvm/test/CodeGen/AArch64/and-mask-variable.ll
index a92f3cf5ec092..f41cdc6dd241b 100644
--- a/llvm/test/CodeGen/AArch64/and-mask-variable.ll
+++ b/llvm/test/CodeGen/AArch64/and-mask-variable.ll
@@ -3,28 +3,78 @@
 ; RUN: llc -mtriple=aarch64-none-elf -global-isel -verify-machineinstrs %s -o - | FileCheck %s --check-prefixes=CHECK,CHECK-GI
 
 define i32 @mask_pair(i32 %x, i32 %y) {
-; CHECK-LABEL: mask_pair:
-; CHECK:       // %bb.0:
-; CHECK-NEXT:    mov w8, #-1 // =0xffffffff
-; CHECK-NEXT:    lsl w8, w8, w1
-; CHECK-NEXT:    and w0, w8, w0
-; CHECK-NEXT:    ret
+; CHECK-SD-LABEL: mask_pair:
+; CHECK-SD:       // %bb.0:
+; CHECK-SD-NEXT:    lsr w8, w0, w1
+; CHECK-SD-NEXT:    lsl w0, w8, w1
+; CHECK-SD-NEXT:    ret
+;
+; CHECK-GI-LABEL: mask_pair:
+; CHECK-GI:       // %bb.0:
+; CHECK-GI-NEXT:    mov w8, #-1 // =0xffffffff
+; CHECK-GI-NEXT:    lsl w8, w8, w1
+; CHECK-GI-NEXT:    and w0, w8, w0
+; CHECK-GI-NEXT:    ret
   %shl = shl nsw i32 -1, %y
   %and = and i32 %shl, %x
   ret i32 %and
 }
 
 define i64 @mask_pair_64(i64 %x, i64 %y) {
-; CHECK-LABEL: mask_pair_64:
-; CHECK:       // %bb.0:
-; CHECK-NEXT:    mov x8, #-1 // =0xffffffffffffffff
-; CHECK-NEXT:    lsl x8, x8, x1
-; CHECK-NEXT:    and x0, x8, x0
-; CHECK-NEXT:    ret
+; CHECK-SD-LABEL: mask_pair_64:
+; CHECK-SD:       // %bb.0:
+; CHECK-SD-NEXT:    lsr x8, x0, x1
+; CHECK-SD-NEXT:    lsl x0, x8, x1
+; CHECK-SD-NEXT:    ret
+;
+; CHECK-GI-LABEL: mask_pair_64:
+; CHECK-GI:       // %bb.0:
+; CHECK-GI-NEXT:    mov x8, #-1 // =0xffffffffffffffff
+; CHECK-GI-NEXT:    lsl x8, x8, x1
+; CHECK-GI-NEXT:    and x0, x8, x0
+; CHECK-GI-NEXT:    ret
   %shl = shl nsw i64 -1, %y
   %and = and i64 %shl, %x
   ret i64 %and
 }
+
+define i128 @mask_pair_128(i128 %x, i128 %y) {
+; CHECK-SD-LABEL: mask_pair_128:
+; CHECK-SD:       // %bb.0:
+; CHECK-SD-NEXT:    mov x8, #-1 // =0xffffffffffffffff
+; CHECK-SD-NEXT:    mvn w9, w2
+; CHECK-SD-NEXT:    mov x10, #9223372036854775807 // =0x7fffffffffffffff
+; CHECK-SD-NEXT:    lsl x8, x8, x2
+; CHECK-SD-NEXT:    lsr x9, x10, x9
+; CHECK-SD-NEXT:    tst x2, #0x40
+; CHECK-SD-NEXT:    orr x9, x8, x9
+; CHECK-SD-NEXT:    csel x9, x8, x9, ne
+; CHECK-SD-NEXT:    csel x8, xzr, x8, ne
+; CHECK-SD-NEXT:    and x0, x8, x0
+; CHECK-SD-NEXT:    and x1, x9, x1
+; CHECK-SD-NEXT:    ret
+;
+; CHECK-GI-LABEL: mask_pair_128:
+; CHECK-GI:       // %bb.0:
+; CHECK-GI-NEXT:    mov w8, #64 // =0x40
+; CHECK-GI-NEXT:    mov x9, #-1 // =0xffffffffffffffff
+; CHECK-GI-NEXT:    sub x10, x2, #64
+; CHECK-GI-NEXT:    sub x8, x8, x2
+; CHECK-GI-NEXT:    lsl x11, x9, x2
+; CHECK-GI-NEXT:    cmp x2, #64
+; CHECK-GI-NEXT:    lsr x8, x9, x8
+; CHECK-GI-NEXT:    lsl x9, x9, x10
+; CHECK-GI-NEXT:    csel x10, x11, xzr, lo
+; CHECK-GI-NEXT:    orr x8, x8, x11
+; CHECK-GI-NEXT:    and x0, x10, x0
+; CHECK-GI-NEXT:    csel x8, x8, x9, lo
+; CHECK-GI-NEXT:    cmp x2, #0
+; CHECK-GI-NEXT:    csinv x8, x8, xzr, ne
+; CHECK-GI-NEXT:    and x1, x8, x1
+; CHECK-GI-NEXT:    ret
+  %shl = shl nsw i128 -1, %y
+  %and = and i128 %shl, %x
+  ret i128 %and
+}
 ;; NOTE: These prefixes are unused and the list is autogenerated. Do not add tests below this line:
-; CHECK-GI: {{.*}}
-; CHECK-SD: {{.*}}
+; CHECK: {{.*}}
diff --git a/llvm/test/CodeGen/AArch64/extract-bits.ll b/llvm/test/CodeGen/AArch64/extract-bits.ll
index 8e822d19a19b9..5a96116142b51 100644
--- a/llvm/test/CodeGen/AArch64/extract-bits.ll
+++ b/llvm/test/CodeGen/AArch64/extract-bits.ll
@@ -532,11 +532,10 @@ define i32 @bextr64_32_b2(i64 %val, i64 %numskipbits, i8 %numlowbits) nounwind {
 define i32 @bextr32_c0(i32 %val, i32 %numskipbits, i32 %numlowbits) nounwind {
 ; CHECK-LABEL: bextr32_c0:
 ; CHECK:       // %bb.0:
-; CHECK-NEXT:    neg w8, w2
-; CHECK-NEXT:    mov w9, #-1 // =0xffffffff
-; CHECK-NEXT:    lsr w10, w0, w1
-; CHECK-NEXT:    lsr w8, w9, w8
-; CHECK-NEXT:    and w0, w8, w10
+; CHECK-NEXT:    lsr w8, w0, w1
+; CHECK-NEXT:    neg w9, w2
+; CHECK-NEXT:    lsl w8, w8, w9
+; CHECK-NEXT:    lsr w0, w8, w9
 ; CHECK-NEXT:    ret
   %shifted = lshr i32 %val, %numskipbits
   %numhighbits = sub i32 32, %numlowbits
@@ -548,12 +547,11 @@ define i32 @bextr32_c0(i32 %val, i32 %numskipbits, i32 %numlowbits) nounwind {
 define i32 @bextr32_c1_indexzext(i32 %val, i8 %numskipbits, i8 %numlowbits) nounwind {
 ; CHECK-LABEL: bextr32_c1_indexzext:
 ; CHECK:       // %bb.0:
-; CHECK-NEXT:    mov w8, #32 // =0x20
-; CHECK-NEXT:    mov w9, #-1 // =0xffffffff
-; CHECK-NEXT:    lsr w10, w0, w1
-; CHECK-NEXT:    sub w8, w8, w2
-; CHECK-NEXT:    lsr w8, w9, w8
-; CHECK-NEXT:    and w0, w8, w10
+; CHECK-NEXT:    lsr w8, w0, w1
+; CHECK-NEXT:    mov w9, #32 // =0x20
+; CHECK-NEXT:    sub w9, w9, w2
+; CHECK-NEXT:    lsl w8, w8, w9
+; CHECK-NEXT:    lsr w0, w8, w9
 ; CHECK-NEXT:    ret
   %skip = zext i8 %numskipbits to i32
   %shifted = lshr i32 %val, %skip
@@ -569,10 +567,9 @@ define i32 @bextr32_c2_load(ptr %w, i32 %numskipbits, i32 %numlowbits) nounwind
 ; CHECK:       // %bb.0:
 ; CHECK-NEXT:    ldr w8, [x0]
 ; CHECK-NEXT:    neg w9, w2
-; CHECK-NEXT:    mov w10, #-1 // =0xffffffff
-; CHECK-NEXT:    lsr w9, w10, w9
 ; CHECK-NEXT:    lsr w8, w8, w1
-; CHECK-NEXT:    and w0, w9, w8
+; CHECK-NEXT:    lsl w8, w8, w9
+; CHECK-NEXT:    lsr w0, w8, w9
 ; CHECK-NEXT:    ret
   %val = load i32, ptr %w
   %shifted = lshr i32 %val, %numskipbits
@@ -587,11 +584,10 @@ define i32 @bextr32_c3_load_indexzext(ptr %w, i8 %numskipbits, i8 %numlowbits) n
 ; CHECK:       // %bb.0:
 ; CHECK-NEXT:    ldr w8, [x0]
 ; CHECK-NEXT:    mov w9, #32 // =0x20
-; CHECK-NEXT:    mov w10, #-1 // =0xffffffff
 ; CHECK-NEXT:    sub w9, w9, w2
 ; CHECK-NEXT:    lsr w8, w8, w1
-; CHECK-NEXT:    lsr w9, w10, w9
-; CHECK-NEXT:    and w0, w9, w8
+; CHECK-NEXT:    lsl w8, w8, w9
+; CHECK-NEXT:    lsr w0, w8, w9
 ; CHECK-NEXT:    ret
   %val = load i32, ptr %w
   %skip = zext i8 %numskipbits to i32
@@ -606,11 +602,10 @@ define i32 @bextr32_c3_load_indexzext(ptr %w, i8 %numskipbits, i8 %numlowbits) n
 define i32 @bextr32_c4_commutative(i32 %val, i32 %numskipbits, i32 %numlowbits) nounwind {
 ; CHECK-LABEL: bextr32_c4_commutative:
 ; CHECK:       // %bb.0:
-; CHECK-NEXT:    neg w8, w2
-; CHECK-NEXT:    mov w9, #-1 // =0xffffffff
-; CHECK-NEXT:    lsr w10, w0, w1
-; CHECK-NEXT:    lsr w8, w9, w8
-; CHECK-NEXT:    and w0, w10, w8
+; CHECK-NEXT:    lsr w8, w0, w1
+; CHECK-NEXT:    neg w9, w2
+; CHECK-NEXT:    lsl w8, w8, w9
+; CHECK-NEXT:    lsr w0, w8, w9
 ; CHECK-NEXT:    ret
   %shifted = lshr i32 %val, %numskipbits
   %numhighbits = sub i32 32, %numlowbits
@@ -624,11 +619,10 @@ define i32 @bextr32_c4_commutative(i32 %val, i32 %numskipbits, i32 %numlowbits)
 define i64 @bextr64_c0(i64 %val, i64 %numskipbits, i64 %numlowbits) nounwind {
 ; CHECK-LABEL: bextr64_c0:
 ; CHECK:       // %bb.0:
-; CHECK-NEXT:    neg x8, x2
-; CHECK-NEXT:    mov x9, #-1 // =0xffffffffffffffff
-; CHECK-NEXT:    lsr x10, x0, x1
-; CHECK-NEXT:    lsr x8, x9, x8
-; CHECK-NEXT:    and x0, x8, x10
+; CHECK-NEXT:    lsr x8, x0, x1
+; CHECK-NEXT:    neg x9, x2
+; CHECK-NEXT:    lsl x8, x8, x9
+; CHECK-NEXT:    lsr x0, x8, x9
 ; CHECK-NEXT:    ret
   %shifted = lshr i64 %val, %numskipbits
   %numhighbits = sub i64 64, %numlowbits
@@ -640,13 +634,12 @@ define i64 @bextr64_c0(i64 %val, i64 %numskipbits, i64 %numlowbits) nounwind {
 define i64 @bextr64_c1_indexzext(i64 %val, i8 %numskipbits, i8 %numlowbits) nounwind {
 ; CHECK-LABEL: bextr64_c1_indexzext:
 ; CHECK:       // %bb.0:
-; CHECK-NEXT:    mov w8, #64 // =0x40
-; CHECK-NEXT:    mov x9, #-1 // =0xffffffffffffffff
 ; CHECK-NEXT:    // kill: def $w1 killed $w1 def $x1
-; CHECK-NEXT:    lsr x10, x0, x1
-; CHECK-NEXT:    sub w8, w8, w2
-; CHECK-NEXT:    lsr x8, x9, x8
-; CHECK-NEXT:    and x0, x8, x10
+; CHECK-NEXT:    lsr x8, x0, x1
+; CHECK-NEXT:    mov w9, #64 // =0x40
+; CHECK-NEXT:    sub w9, w9, w2
+; CHECK-NEXT:    lsl x8, x8, x9
+; CHECK-NEXT:    lsr x0, x8, x9
 ; CHECK-NEXT:    ret
   %skip = zext i8 %numskipbits to i64
   %shifted = lshr i64 %val, %skip
@@ -662,10 +655,9 @@ define i64 @bextr64_c2_load(ptr %w, i64 %numskipbits, i64 %numlowbits) nounwind
 ; CHECK:       // %bb.0:
 ; CHECK-NEXT:    ldr x8, [x0]
 ; CHECK-NEXT:    neg x9, x2
-; CHECK-NEXT:    mov x10, #-1 // =0xffffffffffffffff
-; CHECK-NEXT:    lsr x9, x10, x9
 ; CHECK-NEXT:    lsr x8, x8, x1
-; CHECK-NEXT:    and x0, x9, x8
+; CHECK-NEXT:    lsl x8, x8, x9
+; CHECK-NEXT:    lsr x0, x8, x9
 ; CHECK-NEXT:    ret
   %val = load i64, ptr %w
   %shifted = lshr i64 %val, %numskipbits
@@ -679,13 +671,12 @@ define i64 @bextr64_c3_load_indexzext(ptr %w, i8 %numskipbits, i8 %numlowbits) n
 ; CHECK-LABEL: bextr64_c3_load_indexzext:
 ; CHECK:       // %bb.0:
 ; CHECK-NEXT:    ldr x8, [x0]
+; CHECK-NEXT:    // kill: def $w1 killed $w1 def $x1
 ; CHECK-NEXT:    mov w9, #64 // =0x40
-; CHECK-NEXT:    mov x10, #-1 // =0xffffffffffffffff
 ; CHECK-NEXT:    sub w9, w9, w2
-; CHECK-NEXT:    // kill: def $w1 killed $w1 def $x1
 ; CHECK-NEXT:    lsr x8, x8, x1
-; CHECK-NEXT:    lsr x9, x10, x9
-; CHECK-NEXT:    and x0, x9, x8
+; CHECK-NEXT:    lsl x8, x8, x9
+; CHECK-NEXT:    lsr x0, x8, x9
 ; CHECK-NEXT:    ret
   %val = load i64, ptr %w
   %skip = zext i8 %numskipbits to i64
@@ -700,11 +691,10 @@ define i64 @bextr64_c3_load_indexzext(ptr %w, i8 %numskipbits, i8 %numlowbits) n
 define i64 @bextr64_c4_commutative(i64 %val, i64 %numskipbits, i64 %numlowbits) nounwind {
 ; CHECK-LABEL: bextr64_c4_commutative:
 ; CHECK:       // %bb.0:
-; CHECK-NEXT:    neg x8, x2
-; CHECK-NEXT:    mov x9, #-1 // =0xffffffffffffffff
-; CHECK-NEXT:    lsr x10, x0, x1
-; CHECK-NEXT:    lsr x8, x9, x8
-; CHECK-NEXT:    and x0, x10, x8
+; CHECK-NEXT:    lsr x8, x0, x1
+; CHECK-NEXT:    neg x9, x2
+; CHECK-NEXT:    lsl x8, x8, x9
+; CHECK-NEXT:    lsr x0, x8, x9
 ; CHECK-NEXT:    ret
   %shifted = lshr i64 %val, %numskipbits
   %numhighbits = sub i64 64, %numlowbits
@@ -737,11 +727,10 @@ define i32 @bextr64_32_c0(i64 %val, i64 %numskipbits, i64 %numlowbits) nounwind
 define i32 @bextr64_32_c1(i64 %val, i64 %numskipbits, i32 %numlowbits) nounwind {
 ; CHECK-LABEL: bextr64_32_c1:
 ; CHECK:       // %bb.0:
-; CHECK-NEXT:    neg w8, w2
-; CHECK-NEXT:    mov w9, #-1 // =0xffffffff
-; CHECK-NEXT:    lsr x10, x0, x1
-; CHECK-NEXT:    lsr w8, w9, w8
-; CHECK-NEXT:    and w0, w8, w10
+; CHECK-NEXT:    lsr x8, x0, x1
+; CHECK-NEXT:    neg w9, w2
+; CHECK-NEXT:    lsl w8, w8, w9
+; CHECK-NEXT:    lsr w0, w8, w9
 ; CHECK-NEXT:    ret
   %shifted = lshr i64 %val, %numskipbits
   %truncshifted = trunc i64 %shifted to i32
@@ -756,11 +745,10 @@ define i32 @bextr64_32_c1(i64 %val, i64 %numskipbits, i32 %numlowbits) nounwind
 define i32 @bextr64_32_c2(i64 %val, i64 %numskipbits, i32 %numlowbits) nounwind {
 ; CHECK-LABEL: bextr64_32_c2:
 ; CHECK:       // %bb.0:
-; CHECK-NEXT:    neg w8, w2
-; CHECK-NEXT:    mov w9, #-1 // =0xffffffff
-; CHECK-NEXT:    lsr x10, x0, x1
-; CHECK-NEXT:    lsr w8, w9, w8
-; CHECK-NEXT:    and w0, w8, w10
+; CHECK-NEXT:    lsr x8, x0, x1
+; CHECK-NEXT:    neg w9, w2
+; CHECK-NEXT:    lsl w8, w8, w9
+; CHECK-NEXT:    lsr w0, w8, w9
 ; CHECK-NEXT:    ret
   %shifted = lshr i64 %val, %numskipbits
   %numhighbits = sub i32 32, %numlowbits
diff --git a/llvm/test/CodeGen/AArch64/extract-lowbits.ll b/llvm/test/CodeGen/AArch64/extract-lowbits.ll
index 4b8f3e86b5fef..368440c65df84 100644
--- a/llvm/test/CodeGen/AArch64/extract-lowbits.ll
+++ b/llvm/test/CodeGen/AArch64/extract-lowbits.ll
@@ -347,10 +347,9 @@ define i64 @bzhi64_b4_commutative(i64 %val, i64 %numlowbits) nounwind {
 define i32 @bzhi32_c0(i32 %val, i32 %numlowbits) nounwind {
 ; CHECK-LABEL: bzhi32_c0:
 ; CHECK:       // %bb.0:
-; CHECK-NEXT:    mov w8, #-1 // =0xffffffff
-; CHECK-NEXT:    neg w9, w1
-; CHECK-NEXT:    lsr w8, w8, w9
-; CHECK-NEXT:    and w0, w8, w0
+; CHECK-NEXT:    neg w8, w1
+; CHECK-NEXT:    lsl w9, w0, w8
+; CHECK-NEXT:    lsr w0, w9, w8
 ; CHECK-NEXT:    ret
   %numhighbits = sub i32 32, %numlowbits
   %mask = lshr i32 -1, %numhighbits
@@ -362,10 +361,9 @@ define i32 @bzhi32_c1_indexzext(i32 %val, i8 %numlowbits) nounwind {
 ; CHECK-LABEL: bzhi32_c1_indexzext:
 ; CHECK:       // %bb.0:
 ; CHECK-NEXT:    mov w8, #32 // =0x20
-; CHECK-NEXT:    mov w9, #-1 // =0xffffffff
 ; CHECK-NEXT:    sub w8, w8, w1
-; CHECK-NEXT:    lsr w8, w9, w8
-; CHECK-NEXT:    and w0, w8, w0
+; CHECK-NEXT:    lsl w9, w0, w8
+; CHECK-NEXT:    lsr w0, w9, w8
 ; CHECK-NEXT:    ret
   %numhighbits = sub i8 32, %numlowbits
   %sh_prom = zext i8 %numhighbits to i32
@@ -377,11 +375,10 @@ define i32 @bzhi32_c1_indexzext(i32 %val, i8 %numlowbits) nounwind {
 define i32 @bzhi32_c2_load(ptr %w, i32 %numlowbits) nounwind {
 ; CHECK-LABEL: bzhi32_c2_load:
 ; CHECK:       // %bb.0:
-; CHECK-NEXT:    mov w8, #-1 // =0xffffffff
+; CHECK-NEXT:    ldr w8, [x0]
 ; CHECK-NEXT:    neg w9, w1
-; CHECK-NEXT:    ldr w10, [x0]
-; CHECK-NEXT:    lsr w8, w8, w9
-; CHECK-NEXT:    and w0, w8, w10
+; CHECK-NEXT:    lsl w8, w8, w9
+; CHECK-NEXT:    lsr w0, w8, w9
 ; CHECK-NEXT:    ret
   %val = load i32, ptr %w
   %numhighbits = sub i32 32, %numlowbits
@@ -394,11 +391,10 @@ define i32 @bzhi32_c3_load_indexzext(ptr %w, i8 %numlowbits) nounwind {
 ; CHECK-LABEL: bzhi32_c3_load_indexzext:
 ; CHECK:       // %bb.0:
 ; CHECK-NEXT:    mov w8, #32 // =0x20
-; CHECK-NEXT:    mov w9, #-1 // =0xffffffff
-; CHECK-NEXT:    ldr w10, [x0]
+; CHECK-NEXT:    ldr w9, [x0]
 ; CHECK-NEXT:    sub w8, w8, w1
-; CHECK-NEXT:    lsr w8, w9, w8
-; CHECK-NEXT:    and w0, w8, w10
+; CHECK-NEXT:    lsl w9, w9, w8
+; CHECK-NEXT:    lsr w0, w9, w8
 ; CHECK-NEXT:    ret
   %val = load i32, ptr %w
   %numhighbits = sub i8 32, %numlowbits
@@ -411,10 +407,9 @@ define i32 @bzhi32_c3_load_indexzext(ptr %w, i8 %numlowbits) nounwind {
 define i32 @bzhi32_c4_commutative(i32 %val, i32 %numlowbits) nounwind {
 ; CHECK-LABEL: bzhi32_c4_commutative:
 ; CHECK:       // %bb.0:
-; CHECK-NEXT:    mov w8, #-1 // =0xffffffff
-; CHECK-NEXT:    neg w9, w1
-; CHECK-NEXT:    lsr w8, w8, w9
-; CHECK-NEXT:    and w0, w0, w8
+; CHECK-NEXT:    neg w8, w1
+; CHECK-NEXT:    lsl w9, w0, w8
+; CHECK-NEXT:    lsr w0, w9, w8
 ; CHECK-NEXT:    ret
   %numhighbits = sub i32 32, %numlowbits
   %mask = lshr i32 -1, %numhighbits
@@ -427,10 +422,9 @@ define i32 @bzhi32_c4_commutative(i32 %val, i32 %numlowbits) nounwind {
 define i64 @bzhi64_c0(i64 %val, i64 %numlowbits) nounwind {
 ; CHECK-LABEL: bzhi64_c0:
 ; CHECK:       // %bb.0:
-; CHECK-NEXT:    mov x8, #-1 // =0xffffffffffffffff
-; CHECK-NEXT:    neg x9, x1
-; CHECK-NEXT:    lsr x8, x8, x9
-; CHECK-NEXT:    and x0, x8, x0
+; CHECK-NEXT:    neg x8, x1
+; CHECK-NEXT:    lsl x9, x0, x8
+; CHECK-NEXT:    lsr x0, x9, x8
 ; CHECK-NEXT:    ret
   %numhighbits = sub i64 64, %numlowbits
   %mask = lshr i64 -1, %numhighbits
@@ -442,10 +436,9 @@ define i64 @bzhi64_c1_indexzext(i64 %val, i8 %numlowbits) nounwind {
 ; CHECK-LABEL: bzhi64_c1_indexzext:
 ; CHECK:       // %bb.0:
 ; CHECK-NEXT:    mov w8, #64 // =0x40
-; CHECK-NEXT:    mov x9, #-1 // =0xffffffffffffffff
 ; CHECK-NEXT:    sub w8, w8, w1
-; CHECK-NEXT:    lsr x8, x9, x8
-; CHECK-NEXT:    and x0, x8, x0
+; CHECK-NEXT:    lsl x9, x0, x8
+; CHECK-NEXT:    lsr x0, x9, x8
 ; CHECK-NEXT:    ret
   %numhighbits = sub i8 64, %numlowbits
   %sh_prom = zext i8 %numhighbits to i64
@@ -457,11 +450,10 @@ define i64 @bzhi64_c1_indexzext(i64 %val, i8 %numlowbits) nounwind {
 define i64 @bzhi64_c2_load(ptr %w, i64 %numlowbits) nounwind {
 ; CHECK-LABEL: bzhi64_c2_load:
 ; CHECK:       // %bb.0:
-; CHECK-NEXT:    mov x8, #-1 // =0xffffffffffffffff
+; CHECK-NEXT:    ldr x8, [x0]
 ; CHECK-NEXT:    neg x9, x1
-; CHECK-NEXT:    ldr x10, [x0]
-; CHECK-NEXT:    lsr x8, x8, x9
-; CHECK-NEXT:    and x0, x8, x10
+; CHECK-NEXT:    lsl x8, x8, x9
+; CHECK-NEXT:    lsr x0, x8, x9
 ; CHECK-NEXT:    ret
   %val = load i64, ptr %w
   %numhighbits = sub i64 64, %numlowbits
@@ -474,11 +466,10 @@ define i64 @bzhi64_c3_load_indexzext(ptr %w, i8 %numlowbits) nounwind {
 ; CHECK-LABEL: bzhi64_c3_load_indexzext:
 ; CHECK:       // %bb.0:
 ; CHECK-NEXT:    mov w8, #64 // =0x40
-; CHECK-NEXT:    mov x9, #-1 // =0xffffffffffffffff
-; CHECK-NEXT:    ldr x10, [x0]
+; CHECK-NEXT:    ldr x9, [x0]
 ; CHECK-NEXT:    sub w8, w8, w1
-; CHECK-NEXT:    lsr x8, x9, x8
-; CHECK-NEXT:    and x0, x8, x10
+; CHECK-NEXT:    lsl x9, x9, x8
+; CHECK-NEXT:    lsr x0, x9, x8
 ; CHECK-NEXT:    ret
   %val = load i64, ptr %w
   %numhighbits = sub i8 64, %numlowbits
@@ -491,10 +482,9 @@ define i64 @bzhi64_c3_load_indexzext(ptr %w, i8 %numlowbits) nounwind {
 define i64 @bzhi64_c4_commutative(i64 %val, i64 %numlowbits) nounwind {
 ; CHECK-LABEL: bzhi64_c4_commutative:
 ; CHECK:       // %bb.0:
-; CHECK-NEXT:    mov x8, #-1 // =0xffffffffffffffff
-; CHECK-NEXT:    neg x9, x1
-; CHECK-NEXT:    lsr x8, x8, x9
-; CHECK-NEXT:    and x0, x0, x8
+; CHECK-NEXT:    neg x8, x1
+; CHECK-NEXT:    lsl x9, x0, x8
+; CHECK-NEXT:    lsr x0, x9, x8
 ; CHECK-NEXT:    ret
   %numhighbits = sub i64 64, %numlowbits
   %mask = lshr i64 -1, %numhighbits