Index: test/CodeGen/X86/clear-highbits.ll
===================================================================
--- /dev/null
+++ test/CodeGen/X86/clear-highbits.ll
@@ -0,0 +1,669 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -mtriple=i686-unknown-linux-gnu -mattr=-bmi,-tbm,-bmi2 < %s | FileCheck %s --check-prefixes=CHECK,X86,NOTRANSFORM,X86-NOTRANSFORM,FALLBACK0,X86-FALLBACK0
+; RUN: llc -mtriple=i686-unknown-linux-gnu -mattr=+bmi,-tbm,-bmi2 < %s | FileCheck %s --check-prefixes=CHECK,X86,NOTRANSFORM,X86-NOTRANSFORM,FALLBACK1,X86-FALLBACK1
+; RUN: llc -mtriple=i686-unknown-linux-gnu -mattr=+bmi,+tbm,-bmi2 < %s | FileCheck %s --check-prefixes=CHECK,X86,NOTRANSFORM,X86-NOTRANSFORM,FALLBACK2,X86-FALLBACK2
+; RUN: llc -mtriple=i686-unknown-linux-gnu -mattr=+bmi,+tbm,+bmi2 < %s | FileCheck %s --check-prefixes=CHECK,X86,TRANSFORM,X86-TRANSFORM,FALLBACK3,X86-FALLBACK3
+; RUN: llc -mtriple=i686-unknown-linux-gnu -mattr=+bmi,-tbm,+bmi2 < %s | FileCheck %s --check-prefixes=CHECK,X86,TRANSFORM,X86-TRANSFORM,FALLBACK4,X86-FALLBACK4
+; RUN: llc -mtriple=x86_64-unknown-linux-gnu -mattr=-bmi,-tbm,-bmi2 < %s | FileCheck %s --check-prefixes=CHECK,X64,NOTRANSFORM,X64-NOTRANSFORM,FALLBACK0,X64-FALLBACK0
+; RUN: llc -mtriple=x86_64-unknown-linux-gnu -mattr=+bmi,-tbm,-bmi2 < %s | FileCheck %s --check-prefixes=CHECK,X64,NOTRANSFORM,X64-NOTRANSFORM,FALLBACK1,X64-FALLBACK1
+; RUN: llc -mtriple=x86_64-unknown-linux-gnu -mattr=+bmi,+tbm,-bmi2 < %s | FileCheck %s --check-prefixes=CHECK,X64,NOTRANSFORM,X64-NOTRANSFORM,FALLBACK2,X64-FALLBACK2
+; RUN: llc -mtriple=x86_64-unknown-linux-gnu -mattr=+bmi,+tbm,+bmi2 < %s | FileCheck %s --check-prefixes=CHECK,X64,TRANSFORM,X64-TRANSFORM,FALLBACK3,X64-FALLBACK3
+; RUN: llc -mtriple=x86_64-unknown-linux-gnu -mattr=+bmi,-tbm,+bmi2 < %s | FileCheck %s --check-prefixes=CHECK,X64,TRANSFORM,X64-TRANSFORM,FALLBACK4,X64-FALLBACK4
+
+; Patterns:
+;   c) x &  (-1 >> y)
+;   d) x << y >> y
+; are equivalent, but we prefer the second variant if we have BMI2.
+
+; ---------------------------------------------------------------------------- ;
+; Pattern c. 32-bit
+; ---------------------------------------------------------------------------- ;
+
+define i32 @clear_highbits32_c0(i32 %val, i32 %numhighbits) nounwind {
+; X86-NOTRANSFORM-LABEL: clear_highbits32_c0:
+; X86-NOTRANSFORM:       # %bb.0:
+; X86-NOTRANSFORM-NEXT:    movb {{[0-9]+}}(%esp), %cl
+; X86-NOTRANSFORM-NEXT:    movl $-1, %eax
+; X86-NOTRANSFORM-NEXT:    shrl %cl, %eax
+; X86-NOTRANSFORM-NEXT:    andl {{[0-9]+}}(%esp), %eax
+; X86-NOTRANSFORM-NEXT:    retl
+;
+; X86-TRANSFORM-LABEL: clear_highbits32_c0:
+; X86-TRANSFORM:       # %bb.0:
+; X86-TRANSFORM-NEXT:    movb {{[0-9]+}}(%esp), %al
+; X86-TRANSFORM-NEXT:    movl $-1, %ecx
+; X86-TRANSFORM-NEXT:    shrxl %eax, %ecx, %eax
+; X86-TRANSFORM-NEXT:    andl {{[0-9]+}}(%esp), %eax
+; X86-TRANSFORM-NEXT:    retl
+;
+; X64-NOTRANSFORM-LABEL: clear_highbits32_c0:
+; X64-NOTRANSFORM:       # %bb.0:
+; X64-NOTRANSFORM-NEXT:    movl $-1, %eax
+; X64-NOTRANSFORM-NEXT:    movl %esi, %ecx
+; X64-NOTRANSFORM-NEXT:    shrl %cl, %eax
+; X64-NOTRANSFORM-NEXT:    andl %edi, %eax
+; X64-NOTRANSFORM-NEXT:    retq
+;
+; X64-TRANSFORM-LABEL: clear_highbits32_c0:
+; X64-TRANSFORM:       # %bb.0:
+; X64-TRANSFORM-NEXT:    movl $-1, %eax
+; X64-TRANSFORM-NEXT:    shrxl %esi, %eax, %eax
+; X64-TRANSFORM-NEXT:    andl %edi, %eax
+; X64-TRANSFORM-NEXT:    retq
+  %mask = lshr i32 -1, %numhighbits
+  %masked = and i32 %mask, %val
+  ret i32 %masked
+}
+
+define i32 @clear_highbits32_c1_indexzext(i32 %val, i8 %numhighbits) nounwind {
+; X86-NOTRANSFORM-LABEL: clear_highbits32_c1_indexzext:
+; X86-NOTRANSFORM:       # %bb.0:
+; X86-NOTRANSFORM-NEXT:    movb {{[0-9]+}}(%esp), %cl
+; X86-NOTRANSFORM-NEXT:    movl $-1, %eax
+; X86-NOTRANSFORM-NEXT:    shrl %cl, %eax
+; X86-NOTRANSFORM-NEXT:    andl {{[0-9]+}}(%esp), %eax
+; X86-NOTRANSFORM-NEXT:    retl
+;
+; X86-TRANSFORM-LABEL: clear_highbits32_c1_indexzext:
+; X86-TRANSFORM:       # %bb.0:
+; X86-TRANSFORM-NEXT:    movb {{[0-9]+}}(%esp), %al
+; X86-TRANSFORM-NEXT:    movl $-1, %ecx
+; X86-TRANSFORM-NEXT:    shrxl %eax, %ecx, %eax
+; X86-TRANSFORM-NEXT:    andl {{[0-9]+}}(%esp), %eax
+; X86-TRANSFORM-NEXT:    retl
+;
+; X64-NOTRANSFORM-LABEL: clear_highbits32_c1_indexzext:
+; X64-NOTRANSFORM:       # %bb.0:
+; X64-NOTRANSFORM-NEXT:    movl $-1, %eax
+; X64-NOTRANSFORM-NEXT:    movl %esi, %ecx
+; X64-NOTRANSFORM-NEXT:    shrl %cl, %eax
+; X64-NOTRANSFORM-NEXT:    andl %edi, %eax
+; X64-NOTRANSFORM-NEXT:    retq
+;
+; X64-TRANSFORM-LABEL: clear_highbits32_c1_indexzext:
+; X64-TRANSFORM:       # %bb.0:
+; X64-TRANSFORM-NEXT:    movl $-1, %eax
+; X64-TRANSFORM-NEXT:    shrxl %esi, %eax, %eax
+; X64-TRANSFORM-NEXT:    andl %edi, %eax
+; X64-TRANSFORM-NEXT:    retq
+  %sh_prom = zext i8 %numhighbits to i32
+  %mask = lshr i32 -1, %sh_prom
+  %masked = and i32 %mask, %val
+  ret i32 %masked
+}
+
+define i32 @clear_highbits32_c2_load(i32* %w, i32 %numhighbits) nounwind {
+; X86-NOTRANSFORM-LABEL: clear_highbits32_c2_load:
+; X86-NOTRANSFORM:       # %bb.0:
+; X86-NOTRANSFORM-NEXT:    movl {{[0-9]+}}(%esp), %edx
+; X86-NOTRANSFORM-NEXT:    movb {{[0-9]+}}(%esp), %cl
+; X86-NOTRANSFORM-NEXT:    movl $-1, %eax
+; X86-NOTRANSFORM-NEXT:    shrl %cl, %eax
+; X86-NOTRANSFORM-NEXT:    andl (%edx), %eax
+; X86-NOTRANSFORM-NEXT:    retl
+;
+; X86-TRANSFORM-LABEL: clear_highbits32_c2_load:
+; X86-TRANSFORM:       # %bb.0:
+; X86-TRANSFORM-NEXT:    movl {{[0-9]+}}(%esp), %ecx
+; X86-TRANSFORM-NEXT:    movb {{[0-9]+}}(%esp), %al
+; X86-TRANSFORM-NEXT:    movl $-1, %edx
+; X86-TRANSFORM-NEXT:    shrxl %eax, %edx, %eax
+; X86-TRANSFORM-NEXT:    andl (%ecx), %eax
+; X86-TRANSFORM-NEXT:    retl
+;
+; X64-NOTRANSFORM-LABEL: clear_highbits32_c2_load:
+; X64-NOTRANSFORM:       # %bb.0:
+; X64-NOTRANSFORM-NEXT:    movl $-1, %eax
+; X64-NOTRANSFORM-NEXT:    movl %esi, %ecx
+; X64-NOTRANSFORM-NEXT:    shrl %cl, %eax
+; X64-NOTRANSFORM-NEXT:    andl (%rdi), %eax
+; X64-NOTRANSFORM-NEXT:    retq
+;
+; X64-TRANSFORM-LABEL: clear_highbits32_c2_load:
+; X64-TRANSFORM:       # %bb.0:
+; X64-TRANSFORM-NEXT:    movl $-1, %eax
+; X64-TRANSFORM-NEXT:    shrxl %esi, %eax, %eax
+; X64-TRANSFORM-NEXT:    andl (%rdi), %eax
+; X64-TRANSFORM-NEXT:    retq
+  %val = load i32, i32* %w
+  %mask = lshr i32 -1, %numhighbits
+  %masked = and i32 %mask, %val
+  ret i32 %masked
+}
+
+define i32 @clear_highbits32_c3_load_indexzext(i32* %w, i8 %numhighbits) nounwind {
+; X86-NOTRANSFORM-LABEL: clear_highbits32_c3_load_indexzext:
+; X86-NOTRANSFORM:       # %bb.0:
+; X86-NOTRANSFORM-NEXT:    movl {{[0-9]+}}(%esp), %edx
+; X86-NOTRANSFORM-NEXT:    movb {{[0-9]+}}(%esp), %cl
+; X86-NOTRANSFORM-NEXT:    movl $-1, %eax
+; X86-NOTRANSFORM-NEXT:    shrl %cl, %eax
+; X86-NOTRANSFORM-NEXT:    andl (%edx), %eax
+; X86-NOTRANSFORM-NEXT:    retl
+;
+; X86-TRANSFORM-LABEL: clear_highbits32_c3_load_indexzext:
+; X86-TRANSFORM:       # %bb.0:
+; X86-TRANSFORM-NEXT:    movl {{[0-9]+}}(%esp), %ecx
+; X86-TRANSFORM-NEXT:    movb {{[0-9]+}}(%esp), %al
+; X86-TRANSFORM-NEXT:    movl $-1, %edx
+; X86-TRANSFORM-NEXT:    shrxl %eax, %edx, %eax
+; X86-TRANSFORM-NEXT:    andl (%ecx), %eax
+; X86-TRANSFORM-NEXT:    retl
+;
+; X64-NOTRANSFORM-LABEL: clear_highbits32_c3_load_indexzext:
+; X64-NOTRANSFORM:       # %bb.0:
+; X64-NOTRANSFORM-NEXT:    movl $-1, %eax
+; X64-NOTRANSFORM-NEXT:    movl %esi, %ecx
+; X64-NOTRANSFORM-NEXT:    shrl %cl, %eax
+; X64-NOTRANSFORM-NEXT:    andl (%rdi), %eax
+; X64-NOTRANSFORM-NEXT:    retq
+;
+; X64-TRANSFORM-LABEL: clear_highbits32_c3_load_indexzext:
+; X64-TRANSFORM:       # %bb.0:
+; X64-TRANSFORM-NEXT:    movl $-1, %eax
+; X64-TRANSFORM-NEXT:    shrxl %esi, %eax, %eax
+; X64-TRANSFORM-NEXT:    andl (%rdi), %eax
+; X64-TRANSFORM-NEXT:    retq
+  %val = load i32, i32* %w
+  %sh_prom = zext i8 %numhighbits to i32
+  %mask = lshr i32 -1, %sh_prom
+  %masked = and i32 %mask, %val
+  ret i32 %masked
+}
+
+define i32 @clear_highbits32_c4_commutative(i32 %val, i32 %numhighbits) nounwind {
+; X86-NOTRANSFORM-LABEL: clear_highbits32_c4_commutative:
+; X86-NOTRANSFORM:       # %bb.0:
+; X86-NOTRANSFORM-NEXT:    movb {{[0-9]+}}(%esp), %cl
+; X86-NOTRANSFORM-NEXT:    movl $-1, %eax
+; X86-NOTRANSFORM-NEXT:    shrl %cl, %eax
+; X86-NOTRANSFORM-NEXT:    andl {{[0-9]+}}(%esp), %eax
+; X86-NOTRANSFORM-NEXT:    retl
+;
+; X86-TRANSFORM-LABEL: clear_highbits32_c4_commutative:
+; X86-TRANSFORM:       # %bb.0:
+; X86-TRANSFORM-NEXT:    movb {{[0-9]+}}(%esp), %al
+; X86-TRANSFORM-NEXT:    movl $-1, %ecx
+; X86-TRANSFORM-NEXT:    shrxl %eax, %ecx, %eax
+; X86-TRANSFORM-NEXT:    andl {{[0-9]+}}(%esp), %eax
+; X86-TRANSFORM-NEXT:    retl
+;
+; X64-NOTRANSFORM-LABEL: clear_highbits32_c4_commutative:
+; X64-NOTRANSFORM:       # %bb.0:
+; X64-NOTRANSFORM-NEXT:    movl $-1, %eax
+; X64-NOTRANSFORM-NEXT:    movl %esi, %ecx
+; X64-NOTRANSFORM-NEXT:    shrl %cl, %eax
+; X64-NOTRANSFORM-NEXT:    andl %edi, %eax
+; X64-NOTRANSFORM-NEXT:    retq
+;
+; X64-TRANSFORM-LABEL: clear_highbits32_c4_commutative:
+; X64-TRANSFORM:       # %bb.0:
+; X64-TRANSFORM-NEXT:    movl $-1, %eax
+; X64-TRANSFORM-NEXT:    shrxl %esi, %eax, %eax
+; X64-TRANSFORM-NEXT:    andl %edi, %eax
+; X64-TRANSFORM-NEXT:    retq
+  %mask = lshr i32 -1, %numhighbits
+  %masked = and i32 %val, %mask ; swapped order
+  ret i32 %masked
+}
+
+; 64-bit
+
+define i64 @clear_highbits64_c0(i64 %val, i64 %numhighbits) nounwind {
+; X86-NOTRANSFORM-LABEL: clear_highbits64_c0:
+; X86-NOTRANSFORM:       # %bb.0:
+; X86-NOTRANSFORM-NEXT:    movb {{[0-9]+}}(%esp), %cl
+; X86-NOTRANSFORM-NEXT:    movl $-1, %eax
+; X86-NOTRANSFORM-NEXT:    movl $-1, %edx
+; X86-NOTRANSFORM-NEXT:    shrl %cl, %edx
+; X86-NOTRANSFORM-NEXT:    shrdl %cl, %eax, %eax
+; X86-NOTRANSFORM-NEXT:    testb $32, %cl
+; X86-NOTRANSFORM-NEXT:    je .LBB5_2
+; X86-NOTRANSFORM-NEXT:  # %bb.1:
+; X86-NOTRANSFORM-NEXT:    movl %edx, %eax
+; X86-NOTRANSFORM-NEXT:    xorl %edx, %edx
+; X86-NOTRANSFORM-NEXT:  .LBB5_2:
+; X86-NOTRANSFORM-NEXT:    andl {{[0-9]+}}(%esp), %eax
+; X86-NOTRANSFORM-NEXT:    andl {{[0-9]+}}(%esp), %edx
+; X86-NOTRANSFORM-NEXT:    retl
+;
+; X86-TRANSFORM-LABEL: clear_highbits64_c0:
+; X86-TRANSFORM:       # %bb.0:
+; X86-TRANSFORM-NEXT:    movb {{[0-9]+}}(%esp), %cl
+; X86-TRANSFORM-NEXT:    movl $-1, %eax
+; X86-TRANSFORM-NEXT:    shrxl %ecx, %eax, %edx
+; X86-TRANSFORM-NEXT:    shrdl %cl, %eax, %eax
+; X86-TRANSFORM-NEXT:    testb $32, %cl
+; X86-TRANSFORM-NEXT:    je .LBB5_2
+; X86-TRANSFORM-NEXT:  # %bb.1:
+; X86-TRANSFORM-NEXT:    movl %edx, %eax
+; X86-TRANSFORM-NEXT:    xorl %edx, %edx
+; X86-TRANSFORM-NEXT:  .LBB5_2:
+; X86-TRANSFORM-NEXT:    andl {{[0-9]+}}(%esp), %eax
+; X86-TRANSFORM-NEXT:    andl {{[0-9]+}}(%esp), %edx
+; X86-TRANSFORM-NEXT:    retl
+;
+; X64-NOTRANSFORM-LABEL: clear_highbits64_c0:
+; X64-NOTRANSFORM:       # %bb.0:
+; X64-NOTRANSFORM-NEXT:    movq $-1, %rax
+; X64-NOTRANSFORM-NEXT:    movl %esi, %ecx
+; X64-NOTRANSFORM-NEXT:    shrq %cl, %rax
+; X64-NOTRANSFORM-NEXT:    andq %rdi, %rax
+; X64-NOTRANSFORM-NEXT:    retq
+;
+; X64-TRANSFORM-LABEL: clear_highbits64_c0:
+; X64-TRANSFORM:       # %bb.0:
+; X64-TRANSFORM-NEXT:    movq $-1, %rax
+; X64-TRANSFORM-NEXT:    shrxq %rsi, %rax, %rax
+; X64-TRANSFORM-NEXT:    andq %rdi, %rax
+; X64-TRANSFORM-NEXT:    retq
+  %mask = lshr i64 -1, %numhighbits
+  %masked = and i64 %mask, %val
+  ret i64 %masked
+}
+
+define i64 @clear_highbits64_c1_indexzext(i64 %val, i8 %numhighbits) nounwind {
+; X86-NOTRANSFORM-LABEL: clear_highbits64_c1_indexzext:
+; X86-NOTRANSFORM:       # %bb.0:
+; X86-NOTRANSFORM-NEXT:    movb {{[0-9]+}}(%esp), %cl
+; X86-NOTRANSFORM-NEXT:    movl $-1, %eax
+; X86-NOTRANSFORM-NEXT:    movl $-1, %edx
+; X86-NOTRANSFORM-NEXT:    shrl %cl, %edx
+; X86-NOTRANSFORM-NEXT:    shrdl %cl, %eax, %eax
+; X86-NOTRANSFORM-NEXT:    testb $32, %cl
+; X86-NOTRANSFORM-NEXT:    je .LBB6_2
+; X86-NOTRANSFORM-NEXT:  # %bb.1:
+; X86-NOTRANSFORM-NEXT:    movl %edx, %eax
+; X86-NOTRANSFORM-NEXT:    xorl %edx, %edx
+; X86-NOTRANSFORM-NEXT:  .LBB6_2:
+; X86-NOTRANSFORM-NEXT:    andl {{[0-9]+}}(%esp), %eax
+; X86-NOTRANSFORM-NEXT:    andl {{[0-9]+}}(%esp), %edx
+; X86-NOTRANSFORM-NEXT:    retl
+;
+; X86-TRANSFORM-LABEL: clear_highbits64_c1_indexzext:
+; X86-TRANSFORM:       # %bb.0:
+; X86-TRANSFORM-NEXT:    movb {{[0-9]+}}(%esp), %cl
+; X86-TRANSFORM-NEXT:    movl $-1, %eax
+; X86-TRANSFORM-NEXT:    shrxl %ecx, %eax, %edx
+; X86-TRANSFORM-NEXT:    shrdl %cl, %eax, %eax
+; X86-TRANSFORM-NEXT:    testb $32, %cl
+; X86-TRANSFORM-NEXT:    je .LBB6_2
+; X86-TRANSFORM-NEXT:  # %bb.1:
+; X86-TRANSFORM-NEXT:    movl %edx, %eax
+; X86-TRANSFORM-NEXT:    xorl %edx, %edx
+; X86-TRANSFORM-NEXT:  .LBB6_2:
+; X86-TRANSFORM-NEXT:    andl {{[0-9]+}}(%esp), %eax
+; X86-TRANSFORM-NEXT:    andl {{[0-9]+}}(%esp), %edx
+; X86-TRANSFORM-NEXT:    retl
+;
+; X64-NOTRANSFORM-LABEL: clear_highbits64_c1_indexzext:
+; X64-NOTRANSFORM:       # %bb.0:
+; X64-NOTRANSFORM-NEXT:    movq $-1, %rax
+; X64-NOTRANSFORM-NEXT:    movl %esi, %ecx
+; X64-NOTRANSFORM-NEXT:    shrq %cl, %rax
+; X64-NOTRANSFORM-NEXT:    andq %rdi, %rax
+; X64-NOTRANSFORM-NEXT:    retq
+;
+; X64-TRANSFORM-LABEL: clear_highbits64_c1_indexzext:
+; X64-TRANSFORM:       # %bb.0:
+; X64-TRANSFORM-NEXT:    # kill: def $esi killed $esi def $rsi
+; X64-TRANSFORM-NEXT:    movq $-1, %rax
+; X64-TRANSFORM-NEXT:    shrxq %rsi, %rax, %rax
+; X64-TRANSFORM-NEXT:    andq %rdi, %rax
+; X64-TRANSFORM-NEXT:    retq
+  %sh_prom = zext i8 %numhighbits to i64
+  %mask = lshr i64 -1, %sh_prom
+  %masked = and i64 %mask, %val
+  ret i64 %masked
+}
+
+define i64 @clear_highbits64_c2_load(i64* %w, i64 %numhighbits) nounwind {
+; X86-NOTRANSFORM-LABEL: clear_highbits64_c2_load:
+; X86-NOTRANSFORM:       # %bb.0:
+; X86-NOTRANSFORM-NEXT:    pushl %esi
+; X86-NOTRANSFORM-NEXT:    movl {{[0-9]+}}(%esp), %esi
+; X86-NOTRANSFORM-NEXT:    movb {{[0-9]+}}(%esp), %cl
+; X86-NOTRANSFORM-NEXT:    movl $-1, %eax
+; X86-NOTRANSFORM-NEXT:    movl $-1, %edx
+; X86-NOTRANSFORM-NEXT:    shrl %cl, %edx
+; X86-NOTRANSFORM-NEXT:    shrdl %cl, %eax, %eax
+; X86-NOTRANSFORM-NEXT:    testb $32, %cl
+; X86-NOTRANSFORM-NEXT:    je .LBB7_2
+; X86-NOTRANSFORM-NEXT:  # %bb.1:
+; X86-NOTRANSFORM-NEXT:    movl %edx, %eax
+; X86-NOTRANSFORM-NEXT:    xorl %edx, %edx
+; X86-NOTRANSFORM-NEXT:  .LBB7_2:
+; X86-NOTRANSFORM-NEXT:    andl (%esi), %eax
+; X86-NOTRANSFORM-NEXT:    andl 4(%esi), %edx
+; X86-NOTRANSFORM-NEXT:    popl %esi
+; X86-NOTRANSFORM-NEXT:    retl
+;
+; X86-TRANSFORM-LABEL: clear_highbits64_c2_load:
+; X86-TRANSFORM:       # %bb.0:
+; X86-TRANSFORM-NEXT:    pushl %esi
+; X86-TRANSFORM-NEXT:    movl {{[0-9]+}}(%esp), %esi
+; X86-TRANSFORM-NEXT:    movb {{[0-9]+}}(%esp), %cl
+; X86-TRANSFORM-NEXT:    movl $-1, %eax
+; X86-TRANSFORM-NEXT:    shrxl %ecx, %eax, %edx
+; X86-TRANSFORM-NEXT:    shrdl %cl, %eax, %eax
+; X86-TRANSFORM-NEXT:    testb $32, %cl
+; X86-TRANSFORM-NEXT:    je .LBB7_2
+; X86-TRANSFORM-NEXT:  # %bb.1:
+; X86-TRANSFORM-NEXT:    movl %edx, %eax
+; X86-TRANSFORM-NEXT:    xorl %edx, %edx
+; X86-TRANSFORM-NEXT:  .LBB7_2:
+; X86-TRANSFORM-NEXT:    andl (%esi), %eax
+; X86-TRANSFORM-NEXT:    andl 4(%esi), %edx
+; X86-TRANSFORM-NEXT:    popl %esi
+; X86-TRANSFORM-NEXT:    retl
+;
+; X64-NOTRANSFORM-LABEL: clear_highbits64_c2_load:
+; X64-NOTRANSFORM:       # %bb.0:
+; X64-NOTRANSFORM-NEXT:    movq $-1, %rax
+; X64-NOTRANSFORM-NEXT:    movl %esi, %ecx
+; X64-NOTRANSFORM-NEXT:    shrq %cl, %rax
+; X64-NOTRANSFORM-NEXT:    andq (%rdi), %rax
+; X64-NOTRANSFORM-NEXT:    retq
+;
+; X64-TRANSFORM-LABEL: clear_highbits64_c2_load:
+; X64-TRANSFORM:       # %bb.0:
+; X64-TRANSFORM-NEXT:    movq $-1, %rax
+; X64-TRANSFORM-NEXT:    shrxq %rsi, %rax, %rax
+; X64-TRANSFORM-NEXT:    andq (%rdi), %rax
+; X64-TRANSFORM-NEXT:    retq
+  %val = load i64, i64* %w
+  %mask = lshr i64 -1, %numhighbits
+  %masked = and i64 %mask, %val
+  ret i64 %masked
+}
+
+define i64 @clear_highbits64_c3_load_indexzext(i64* %w, i8 %numhighbits) nounwind {
+; X86-NOTRANSFORM-LABEL: clear_highbits64_c3_load_indexzext:
+; X86-NOTRANSFORM:       # %bb.0:
+; X86-NOTRANSFORM-NEXT:    pushl %esi
+; X86-NOTRANSFORM-NEXT:    movl {{[0-9]+}}(%esp), %esi
+; X86-NOTRANSFORM-NEXT:    movb {{[0-9]+}}(%esp), %cl
+; X86-NOTRANSFORM-NEXT:    movl $-1, %eax
+; X86-NOTRANSFORM-NEXT:    movl $-1, %edx
+; X86-NOTRANSFORM-NEXT:    shrl %cl, %edx
+; X86-NOTRANSFORM-NEXT:    shrdl %cl, %eax, %eax
+; X86-NOTRANSFORM-NEXT:    testb $32, %cl
+; X86-NOTRANSFORM-NEXT:    je .LBB8_2
+; X86-NOTRANSFORM-NEXT:  # %bb.1:
+; X86-NOTRANSFORM-NEXT:    movl %edx, %eax
+; X86-NOTRANSFORM-NEXT:    xorl %edx, %edx
+; X86-NOTRANSFORM-NEXT:  .LBB8_2:
+; X86-NOTRANSFORM-NEXT:    andl (%esi), %eax
+; X86-NOTRANSFORM-NEXT:    andl 4(%esi), %edx
+; X86-NOTRANSFORM-NEXT:    popl %esi
+; X86-NOTRANSFORM-NEXT:    retl
+;
+; X86-TRANSFORM-LABEL: clear_highbits64_c3_load_indexzext:
+; X86-TRANSFORM:       # %bb.0:
+; X86-TRANSFORM-NEXT:    pushl %esi
+; X86-TRANSFORM-NEXT:    movl {{[0-9]+}}(%esp), %esi
+; X86-TRANSFORM-NEXT:    movb {{[0-9]+}}(%esp), %cl
+; X86-TRANSFORM-NEXT:    movl $-1, %eax
+; X86-TRANSFORM-NEXT:    shrxl %ecx, %eax, %edx
+; X86-TRANSFORM-NEXT:    shrdl %cl, %eax, %eax
+; X86-TRANSFORM-NEXT:    testb $32, %cl
+; X86-TRANSFORM-NEXT:    je .LBB8_2
+; X86-TRANSFORM-NEXT:  # %bb.1:
+; X86-TRANSFORM-NEXT:    movl %edx, %eax
+; X86-TRANSFORM-NEXT:    xorl %edx, %edx
+; X86-TRANSFORM-NEXT:  .LBB8_2:
+; X86-TRANSFORM-NEXT:    andl (%esi), %eax
+; X86-TRANSFORM-NEXT:    andl 4(%esi), %edx
+; X86-TRANSFORM-NEXT:    popl %esi
+; X86-TRANSFORM-NEXT:    retl
+;
+; X64-NOTRANSFORM-LABEL: clear_highbits64_c3_load_indexzext:
+; X64-NOTRANSFORM:       # %bb.0:
+; X64-NOTRANSFORM-NEXT:    movq $-1, %rax
+; X64-NOTRANSFORM-NEXT:    movl %esi, %ecx
+; X64-NOTRANSFORM-NEXT:    shrq %cl, %rax
+; X64-NOTRANSFORM-NEXT:    andq (%rdi), %rax
+; X64-NOTRANSFORM-NEXT:    retq
+;
+; X64-TRANSFORM-LABEL: clear_highbits64_c3_load_indexzext:
+; X64-TRANSFORM:       # %bb.0:
+; X64-TRANSFORM-NEXT:    # kill: def $esi killed $esi def $rsi
+; X64-TRANSFORM-NEXT:    movq $-1, %rax
+; X64-TRANSFORM-NEXT:    shrxq %rsi, %rax, %rax
+; X64-TRANSFORM-NEXT:    andq (%rdi), %rax
+; X64-TRANSFORM-NEXT:    retq
+  %val = load i64, i64* %w
+  %sh_prom = zext i8 %numhighbits to i64
+  %mask = lshr i64 -1, %sh_prom
+  %masked = and i64 %mask, %val
+  ret i64 %masked
+}
+
+define i64 @clear_highbits64_c4_commutative(i64 %val, i64 %numhighbits) nounwind {
+; X86-NOTRANSFORM-LABEL: clear_highbits64_c4_commutative:
+; X86-NOTRANSFORM:       # %bb.0:
+; X86-NOTRANSFORM-NEXT:    movb {{[0-9]+}}(%esp), %cl
+; X86-NOTRANSFORM-NEXT:    movl $-1, %eax
+; X86-NOTRANSFORM-NEXT:    movl $-1, %edx
+; X86-NOTRANSFORM-NEXT:    shrl %cl, %edx
+; X86-NOTRANSFORM-NEXT:    shrdl %cl, %eax, %eax
+; X86-NOTRANSFORM-NEXT:    testb $32, %cl
+; X86-NOTRANSFORM-NEXT:    je .LBB9_2
+; X86-NOTRANSFORM-NEXT:  # %bb.1:
+; X86-NOTRANSFORM-NEXT:    movl %edx, %eax
+; X86-NOTRANSFORM-NEXT:    xorl %edx, %edx
+; X86-NOTRANSFORM-NEXT:  .LBB9_2:
+; X86-NOTRANSFORM-NEXT:    andl {{[0-9]+}}(%esp), %eax
+; X86-NOTRANSFORM-NEXT:    andl {{[0-9]+}}(%esp), %edx
+; X86-NOTRANSFORM-NEXT:    retl
+;
+; X86-TRANSFORM-LABEL: clear_highbits64_c4_commutative:
+; X86-TRANSFORM:       # %bb.0:
+; X86-TRANSFORM-NEXT:    movb {{[0-9]+}}(%esp), %cl
+; X86-TRANSFORM-NEXT:    movl $-1, %eax
+; X86-TRANSFORM-NEXT:    shrxl %ecx, %eax, %edx
+; X86-TRANSFORM-NEXT:    shrdl %cl, %eax, %eax
+; X86-TRANSFORM-NEXT:    testb $32, %cl
+; X86-TRANSFORM-NEXT:    je .LBB9_2
+; X86-TRANSFORM-NEXT:  # %bb.1:
+; X86-TRANSFORM-NEXT:    movl %edx, %eax
+; X86-TRANSFORM-NEXT:    xorl %edx, %edx
+; X86-TRANSFORM-NEXT:  .LBB9_2:
+; X86-TRANSFORM-NEXT:    andl {{[0-9]+}}(%esp), %eax
+; X86-TRANSFORM-NEXT:    andl {{[0-9]+}}(%esp), %edx
+; X86-TRANSFORM-NEXT:    retl
+;
+; X64-NOTRANSFORM-LABEL: clear_highbits64_c4_commutative:
+; X64-NOTRANSFORM:       # %bb.0:
+; X64-NOTRANSFORM-NEXT:    movq $-1, %rax
+; X64-NOTRANSFORM-NEXT:    movl %esi, %ecx
+; X64-NOTRANSFORM-NEXT:    shrq %cl, %rax
+; X64-NOTRANSFORM-NEXT:    andq %rdi, %rax
+; X64-NOTRANSFORM-NEXT:    retq
+;
+; X64-TRANSFORM-LABEL: clear_highbits64_c4_commutative:
+; X64-TRANSFORM:       # %bb.0:
+; X64-TRANSFORM-NEXT:    movq $-1, %rax
+; X64-TRANSFORM-NEXT:    shrxq %rsi, %rax, %rax
+; X64-TRANSFORM-NEXT:    andq %rdi, %rax
+; X64-TRANSFORM-NEXT:    retq
+  %mask = lshr i64 -1, %numhighbits
+  %masked = and i64 %val, %mask ; swapped order
+  ret i64 %masked
+}
+
+; Multi-use tests
+
+declare void @use32(i32)
+declare void @use64(i64)
+
+define i32 @oneuse32(i32 %val, i32 %numhighbits) nounwind {
+; X86-NOTRANSFORM-LABEL: oneuse32:
+; X86-NOTRANSFORM:       # %bb.0:
+; X86-NOTRANSFORM-NEXT:    pushl %esi
+; X86-NOTRANSFORM-NEXT:    subl $8, %esp
+; X86-NOTRANSFORM-NEXT:    movb {{[0-9]+}}(%esp), %cl
+; X86-NOTRANSFORM-NEXT:    movl $-1, %esi
+; X86-NOTRANSFORM-NEXT:    shrl %cl, %esi
+; X86-NOTRANSFORM-NEXT:    movl %esi, (%esp)
+; X86-NOTRANSFORM-NEXT:    calll use32
+; X86-NOTRANSFORM-NEXT:    andl {{[0-9]+}}(%esp), %esi
+; X86-NOTRANSFORM-NEXT:    movl %esi, %eax
+; X86-NOTRANSFORM-NEXT:    addl $8, %esp
+; X86-NOTRANSFORM-NEXT:    popl %esi
+; X86-NOTRANSFORM-NEXT:    retl
+;
+; X86-TRANSFORM-LABEL: oneuse32:
+; X86-TRANSFORM:       # %bb.0:
+; X86-TRANSFORM-NEXT:    pushl %esi
+; X86-TRANSFORM-NEXT:    subl $8, %esp
+; X86-TRANSFORM-NEXT:    movb {{[0-9]+}}(%esp), %al
+; X86-TRANSFORM-NEXT:    movl $-1, %ecx
+; X86-TRANSFORM-NEXT:    shrxl %eax, %ecx, %esi
+; X86-TRANSFORM-NEXT:    movl %esi, (%esp)
+; X86-TRANSFORM-NEXT:    calll use32
+; X86-TRANSFORM-NEXT:    andl {{[0-9]+}}(%esp), %esi
+; X86-TRANSFORM-NEXT:    movl %esi, %eax
+; X86-TRANSFORM-NEXT:    addl $8, %esp
+; X86-TRANSFORM-NEXT:    popl %esi
+; X86-TRANSFORM-NEXT:    retl
+;
+; X64-NOTRANSFORM-LABEL: oneuse32:
+; X64-NOTRANSFORM:       # %bb.0:
+; X64-NOTRANSFORM-NEXT:    pushq %rbp
+; X64-NOTRANSFORM-NEXT:    pushq %rbx
+; X64-NOTRANSFORM-NEXT:    pushq %rax
+; X64-NOTRANSFORM-NEXT:    movl %edi, %ebx
+; X64-NOTRANSFORM-NEXT:    movl $-1, %ebp
+; X64-NOTRANSFORM-NEXT:    movl %esi, %ecx
+; X64-NOTRANSFORM-NEXT:    shrl %cl, %ebp
+; X64-NOTRANSFORM-NEXT:    movl %ebp, %edi
+; X64-NOTRANSFORM-NEXT:    callq use32
+; X64-NOTRANSFORM-NEXT:    andl %ebx, %ebp
+; X64-NOTRANSFORM-NEXT:    movl %ebp, %eax
+; X64-NOTRANSFORM-NEXT:    addq $8, %rsp
+; X64-NOTRANSFORM-NEXT:    popq %rbx
+; X64-NOTRANSFORM-NEXT:    popq %rbp
+; X64-NOTRANSFORM-NEXT:    retq
+;
+; X64-TRANSFORM-LABEL: oneuse32:
+; X64-TRANSFORM:       # %bb.0:
+; X64-TRANSFORM-NEXT:    pushq %rbp
+; X64-TRANSFORM-NEXT:    pushq %rbx
+; X64-TRANSFORM-NEXT:    pushq %rax
+; X64-TRANSFORM-NEXT:    movl %edi, %ebx
+; X64-TRANSFORM-NEXT:    movl $-1, %eax
+; X64-TRANSFORM-NEXT:    shrxl %esi, %eax, %ebp
+; X64-TRANSFORM-NEXT:    movl %ebp, %edi
+; X64-TRANSFORM-NEXT:    callq use32
+; X64-TRANSFORM-NEXT:    andl %ebx, %ebp
+; X64-TRANSFORM-NEXT:    movl %ebp, %eax
+; X64-TRANSFORM-NEXT:    addq $8, %rsp
+; X64-TRANSFORM-NEXT:    popq %rbx
+; X64-TRANSFORM-NEXT:    popq %rbp
+; X64-TRANSFORM-NEXT:    retq
+  %mask = lshr i32 -1, %numhighbits
+  call void @use32(i32 %mask)
+  %masked = and i32 %mask, %val
+  ret i32 %masked
+}
+
+define i64 @oneuse64(i64 %val, i64 %numhighbits) nounwind {
+; X86-NOTRANSFORM-LABEL: oneuse64:
+; X86-NOTRANSFORM:       # %bb.0:
+; X86-NOTRANSFORM-NEXT:    pushl %edi
+; X86-NOTRANSFORM-NEXT:    pushl %esi
+; X86-NOTRANSFORM-NEXT:    pushl %eax
+; X86-NOTRANSFORM-NEXT:    movb {{[0-9]+}}(%esp), %cl
+; X86-NOTRANSFORM-NEXT:    movl $-1, %esi
+; X86-NOTRANSFORM-NEXT:    movl $-1, %edi
+; X86-NOTRANSFORM-NEXT:    shrl %cl, %edi
+; X86-NOTRANSFORM-NEXT:    shrdl %cl, %esi, %esi
+; X86-NOTRANSFORM-NEXT:    testb $32, %cl
+; X86-NOTRANSFORM-NEXT:    je .LBB11_2
+; X86-NOTRANSFORM-NEXT:  # %bb.1:
+; X86-NOTRANSFORM-NEXT:    movl %edi, %esi
+; X86-NOTRANSFORM-NEXT:    xorl %edi, %edi
+; X86-NOTRANSFORM-NEXT:  .LBB11_2:
+; X86-NOTRANSFORM-NEXT:    subl $8, %esp
+; X86-NOTRANSFORM-NEXT:    pushl %edi
+; X86-NOTRANSFORM-NEXT:    pushl %esi
+; X86-NOTRANSFORM-NEXT:    calll use64
+; X86-NOTRANSFORM-NEXT:    addl $16, %esp
+; X86-NOTRANSFORM-NEXT:    andl {{[0-9]+}}(%esp), %esi
+; X86-NOTRANSFORM-NEXT:    andl {{[0-9]+}}(%esp), %edi
+; X86-NOTRANSFORM-NEXT:    movl %esi, %eax
+; X86-NOTRANSFORM-NEXT:    movl %edi, %edx
+; X86-NOTRANSFORM-NEXT:    addl $4, %esp
+; X86-NOTRANSFORM-NEXT:    popl %esi
+; X86-NOTRANSFORM-NEXT:    popl %edi
+; X86-NOTRANSFORM-NEXT:    retl
+;
+; X86-TRANSFORM-LABEL: oneuse64:
+; X86-TRANSFORM:       # %bb.0:
+; X86-TRANSFORM-NEXT:    pushl %edi
+; X86-TRANSFORM-NEXT:    pushl %esi
+; X86-TRANSFORM-NEXT:    pushl %eax
+; X86-TRANSFORM-NEXT:    movb {{[0-9]+}}(%esp), %cl
+; X86-TRANSFORM-NEXT:    movl $-1, %esi
+; X86-TRANSFORM-NEXT:    shrxl %ecx, %esi, %edi
+; X86-TRANSFORM-NEXT:    shrdl %cl, %esi, %esi
+; X86-TRANSFORM-NEXT:    testb $32, %cl
+; X86-TRANSFORM-NEXT:    je .LBB11_2
+; X86-TRANSFORM-NEXT:  # %bb.1:
+; X86-TRANSFORM-NEXT:    movl %edi, %esi
+; X86-TRANSFORM-NEXT:    xorl %edi, %edi
+; X86-TRANSFORM-NEXT:  .LBB11_2:
+; X86-TRANSFORM-NEXT:    subl $8, %esp
+; X86-TRANSFORM-NEXT:    pushl %edi
+; X86-TRANSFORM-NEXT:    pushl %esi
+; X86-TRANSFORM-NEXT:    calll use64
+; X86-TRANSFORM-NEXT:    addl $16, %esp
+; X86-TRANSFORM-NEXT:    andl {{[0-9]+}}(%esp), %esi
+; X86-TRANSFORM-NEXT:    andl {{[0-9]+}}(%esp), %edi
+; X86-TRANSFORM-NEXT:    movl %esi, %eax
+; X86-TRANSFORM-NEXT:    movl %edi, %edx
+; X86-TRANSFORM-NEXT:    addl $4, %esp
+; X86-TRANSFORM-NEXT:    popl %esi
+; X86-TRANSFORM-NEXT:    popl %edi
+; X86-TRANSFORM-NEXT:    retl
+;
+; X64-NOTRANSFORM-LABEL: oneuse64:
+; X64-NOTRANSFORM:       # %bb.0:
+; X64-NOTRANSFORM-NEXT:    pushq %r14
+; X64-NOTRANSFORM-NEXT:    pushq %rbx
+; X64-NOTRANSFORM-NEXT:    pushq %rax
+; X64-NOTRANSFORM-NEXT:    movq %rdi, %r14
+; X64-NOTRANSFORM-NEXT:    movq $-1, %rbx
+; X64-NOTRANSFORM-NEXT:    movl %esi, %ecx
+; X64-NOTRANSFORM-NEXT:    shrq %cl, %rbx
+; X64-NOTRANSFORM-NEXT:    movq %rbx, %rdi
+; X64-NOTRANSFORM-NEXT:    callq use64
+; X64-NOTRANSFORM-NEXT:    andq %r14, %rbx
+; X64-NOTRANSFORM-NEXT:    movq %rbx, %rax
+; X64-NOTRANSFORM-NEXT:    addq $8, %rsp
+; X64-NOTRANSFORM-NEXT:    popq %rbx
+; X64-NOTRANSFORM-NEXT:    popq %r14
+; X64-NOTRANSFORM-NEXT:    retq
+;
+; X64-TRANSFORM-LABEL: oneuse64:
+; X64-TRANSFORM:       # %bb.0:
+; X64-TRANSFORM-NEXT:    pushq %r14
+; X64-TRANSFORM-NEXT:    pushq %rbx
+; X64-TRANSFORM-NEXT:    pushq %rax
+; X64-TRANSFORM-NEXT:    movq %rdi, %r14
+; X64-TRANSFORM-NEXT:    movq $-1, %rax
+; X64-TRANSFORM-NEXT:    shrxq %rsi, %rax, %rbx
+; X64-TRANSFORM-NEXT:    movq %rbx, %rdi
+; X64-TRANSFORM-NEXT:    callq use64
+; X64-TRANSFORM-NEXT:    andq %r14, %rbx
+; X64-TRANSFORM-NEXT:    movq %rbx, %rax
+; X64-TRANSFORM-NEXT:    addq $8, %rsp
+; X64-TRANSFORM-NEXT:    popq %rbx
+; X64-TRANSFORM-NEXT:    popq %r14
+; X64-TRANSFORM-NEXT:    retq
+  %mask = lshr i64 -1, %numhighbits
+  call void @use64(i64 %mask)
+  %masked = and i64 %mask, %val
+  ret i64 %masked
+}
Index: test/CodeGen/X86/clear-lowbits.ll
===================================================================
--- /dev/null
+++ test/CodeGen/X86/clear-lowbits.ll
@@ -0,0 +1,669 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -mtriple=i686-unknown-linux-gnu -mattr=-bmi,-tbm,-bmi2 < %s | FileCheck %s --check-prefixes=CHECK,X86,NOTRANSFORM,X86-NOTRANSFORM,FALLBACK0,X86-FALLBACK0
+; RUN: llc -mtriple=i686-unknown-linux-gnu -mattr=+bmi,-tbm,-bmi2 < %s | FileCheck %s --check-prefixes=CHECK,X86,NOTRANSFORM,X86-NOTRANSFORM,FALLBACK1,X86-FALLBACK1
+; RUN: llc -mtriple=i686-unknown-linux-gnu -mattr=+bmi,+tbm,-bmi2 < %s | FileCheck %s --check-prefixes=CHECK,X86,NOTRANSFORM,X86-NOTRANSFORM,FALLBACK2,X86-FALLBACK2
+; RUN: llc -mtriple=i686-unknown-linux-gnu -mattr=+bmi,+tbm,+bmi2 < %s | FileCheck %s --check-prefixes=CHECK,X86,TRANSFORM,X86-TRANSFORM,FALLBACK3,X86-FALLBACK3
+; RUN: llc -mtriple=i686-unknown-linux-gnu -mattr=+bmi,-tbm,+bmi2 < %s | FileCheck %s --check-prefixes=CHECK,X86,TRANSFORM,X86-TRANSFORM,FALLBACK4,X86-FALLBACK4
+; RUN: llc -mtriple=x86_64-unknown-linux-gnu -mattr=-bmi,-tbm,-bmi2 < %s | FileCheck %s --check-prefixes=CHECK,X64,NOTRANSFORM,X64-NOTRANSFORM,FALLBACK0,X64-FALLBACK0
+; RUN: llc -mtriple=x86_64-unknown-linux-gnu -mattr=+bmi,-tbm,-bmi2 < %s | FileCheck %s --check-prefixes=CHECK,X64,NOTRANSFORM,X64-NOTRANSFORM,FALLBACK1,X64-FALLBACK1
+; RUN: llc -mtriple=x86_64-unknown-linux-gnu -mattr=+bmi,+tbm,-bmi2 < %s | FileCheck %s --check-prefixes=CHECK,X64,NOTRANSFORM,X64-NOTRANSFORM,FALLBACK2,X64-FALLBACK2
+; RUN: llc -mtriple=x86_64-unknown-linux-gnu -mattr=+bmi,+tbm,+bmi2 < %s | FileCheck %s --check-prefixes=CHECK,X64,TRANSFORM,X64-TRANSFORM,FALLBACK3,X64-FALLBACK3
+; RUN: llc -mtriple=x86_64-unknown-linux-gnu -mattr=+bmi,-tbm,+bmi2 < %s | FileCheck %s --check-prefixes=CHECK,X64,TRANSFORM,X64-TRANSFORM,FALLBACK4,X64-FALLBACK4
+
+; Patterns:
+;   c) x &  (-1 << y)
+;   d) x >> y << y
+; are equivalent, but we prefer the second variant if we have BMI2.
+
+; ---------------------------------------------------------------------------- ;
+; Pattern c. 32-bit
+; ---------------------------------------------------------------------------- ;
+
+define i32 @clear_lowbits32_c0(i32 %val, i32 %numlowbits) nounwind {
+; X86-NOTRANSFORM-LABEL: clear_lowbits32_c0:
+; X86-NOTRANSFORM:       # %bb.0:
+; X86-NOTRANSFORM-NEXT:    movb {{[0-9]+}}(%esp), %cl
+; X86-NOTRANSFORM-NEXT:    movl $-1, %eax
+; X86-NOTRANSFORM-NEXT:    shll %cl, %eax
+; X86-NOTRANSFORM-NEXT:    andl {{[0-9]+}}(%esp), %eax
+; X86-NOTRANSFORM-NEXT:    retl
+;
+; X86-TRANSFORM-LABEL: clear_lowbits32_c0:
+; X86-TRANSFORM:       # %bb.0:
+; X86-TRANSFORM-NEXT:    movb {{[0-9]+}}(%esp), %al
+; X86-TRANSFORM-NEXT:    movl $-1, %ecx
+; X86-TRANSFORM-NEXT:    shlxl %eax, %ecx, %eax
+; X86-TRANSFORM-NEXT:    andl {{[0-9]+}}(%esp), %eax
+; X86-TRANSFORM-NEXT:    retl
+;
+; X64-NOTRANSFORM-LABEL: clear_lowbits32_c0:
+; X64-NOTRANSFORM:       # %bb.0:
+; X64-NOTRANSFORM-NEXT:    movl $-1, %eax
+; X64-NOTRANSFORM-NEXT:    movl %esi, %ecx
+; X64-NOTRANSFORM-NEXT:    shll %cl, %eax
+; X64-NOTRANSFORM-NEXT:    andl %edi, %eax
+; X64-NOTRANSFORM-NEXT:    retq
+;
+; X64-TRANSFORM-LABEL: clear_lowbits32_c0:
+; X64-TRANSFORM:       # %bb.0:
+; X64-TRANSFORM-NEXT:    movl $-1, %eax
+; X64-TRANSFORM-NEXT:    shlxl %esi, %eax, %eax
+; X64-TRANSFORM-NEXT:    andl %edi, %eax
+; X64-TRANSFORM-NEXT:    retq
+  %mask = shl i32 -1, %numlowbits
+  %masked = and i32 %mask, %val
+  ret i32 %masked
+}
+
+define i32 @clear_lowbits32_c1_indexzext(i32 %val, i8 %numlowbits) nounwind {
+; X86-NOTRANSFORM-LABEL: clear_lowbits32_c1_indexzext:
+; X86-NOTRANSFORM:       # %bb.0:
+; X86-NOTRANSFORM-NEXT:    movb {{[0-9]+}}(%esp), %cl
+; X86-NOTRANSFORM-NEXT:    movl $-1, %eax
+; X86-NOTRANSFORM-NEXT:    shll %cl, %eax
+; X86-NOTRANSFORM-NEXT:    andl {{[0-9]+}}(%esp), %eax
+; X86-NOTRANSFORM-NEXT:    retl
+;
+; X86-TRANSFORM-LABEL: clear_lowbits32_c1_indexzext:
+; X86-TRANSFORM:       # %bb.0:
+; X86-TRANSFORM-NEXT:    movb {{[0-9]+}}(%esp), %al
+; X86-TRANSFORM-NEXT:    movl $-1, %ecx
+; X86-TRANSFORM-NEXT:    shlxl %eax, %ecx, %eax
+; X86-TRANSFORM-NEXT:    andl {{[0-9]+}}(%esp), %eax
+; X86-TRANSFORM-NEXT:    retl
+;
+; X64-NOTRANSFORM-LABEL: clear_lowbits32_c1_indexzext:
+; X64-NOTRANSFORM:       # %bb.0:
+; X64-NOTRANSFORM-NEXT:    movl $-1, %eax
+; X64-NOTRANSFORM-NEXT:    movl %esi, %ecx
+; X64-NOTRANSFORM-NEXT:    shll %cl, %eax
+; X64-NOTRANSFORM-NEXT:    andl %edi, %eax
+; X64-NOTRANSFORM-NEXT:    retq
+;
+; X64-TRANSFORM-LABEL: clear_lowbits32_c1_indexzext:
+; X64-TRANSFORM:       # %bb.0:
+; X64-TRANSFORM-NEXT:    movl $-1, %eax
+; X64-TRANSFORM-NEXT:    shlxl %esi, %eax, %eax
+; X64-TRANSFORM-NEXT:    andl %edi, %eax
+; X64-TRANSFORM-NEXT:    retq
+  %sh_prom = zext i8 %numlowbits to i32
+  %mask = shl i32 -1, %sh_prom
+  %masked = and i32 %mask, %val
+  ret i32 %masked
+}
+
+define i32 @clear_lowbits32_c2_load(i32* %w, i32 %numlowbits) nounwind {
+; X86-NOTRANSFORM-LABEL: clear_lowbits32_c2_load:
+; X86-NOTRANSFORM:       # %bb.0:
+; X86-NOTRANSFORM-NEXT:    movl {{[0-9]+}}(%esp), %edx
+; X86-NOTRANSFORM-NEXT:    movb {{[0-9]+}}(%esp), %cl
+; X86-NOTRANSFORM-NEXT:    movl $-1, %eax
+; X86-NOTRANSFORM-NEXT:    shll %cl, %eax
+; X86-NOTRANSFORM-NEXT:    andl (%edx), %eax
+; X86-NOTRANSFORM-NEXT:    retl
+;
+; X86-TRANSFORM-LABEL: clear_lowbits32_c2_load:
+; X86-TRANSFORM:       # %bb.0:
+; X86-TRANSFORM-NEXT:    movl {{[0-9]+}}(%esp), %ecx
+; X86-TRANSFORM-NEXT:    movb {{[0-9]+}}(%esp), %al
+; X86-TRANSFORM-NEXT:    movl $-1, %edx
+; X86-TRANSFORM-NEXT:    shlxl %eax, %edx, %eax
+; X86-TRANSFORM-NEXT:    andl (%ecx), %eax
+; X86-TRANSFORM-NEXT:    retl
+;
+; X64-NOTRANSFORM-LABEL: clear_lowbits32_c2_load:
+; X64-NOTRANSFORM:       # %bb.0:
+; X64-NOTRANSFORM-NEXT:    movl $-1, %eax
+; X64-NOTRANSFORM-NEXT:    movl %esi, %ecx
+; X64-NOTRANSFORM-NEXT:    shll %cl, %eax
+; X64-NOTRANSFORM-NEXT:    andl (%rdi), %eax
+; X64-NOTRANSFORM-NEXT:    retq
+;
+; X64-TRANSFORM-LABEL: clear_lowbits32_c2_load:
+; X64-TRANSFORM:       # %bb.0:
+; X64-TRANSFORM-NEXT:    movl $-1, %eax
+; X64-TRANSFORM-NEXT:    shlxl %esi, %eax, %eax
+; X64-TRANSFORM-NEXT:    andl (%rdi), %eax
+; X64-TRANSFORM-NEXT:    retq
+  %val = load i32, i32* %w
+  %mask = shl i32 -1, %numlowbits
+  %masked = and i32 %mask, %val
+  ret i32 %masked
+}
+
+define i32 @clear_lowbits32_c3_load_indexzext(i32* %w, i8 %numlowbits) nounwind {
+; X86-NOTRANSFORM-LABEL: clear_lowbits32_c3_load_indexzext:
+; X86-NOTRANSFORM:       # %bb.0:
+; X86-NOTRANSFORM-NEXT:    movl {{[0-9]+}}(%esp), %edx
+; X86-NOTRANSFORM-NEXT:    movb {{[0-9]+}}(%esp), %cl
+; X86-NOTRANSFORM-NEXT:    movl $-1, %eax
+; X86-NOTRANSFORM-NEXT:    shll %cl, %eax
+; X86-NOTRANSFORM-NEXT:    andl (%edx), %eax
+; X86-NOTRANSFORM-NEXT:    retl
+;
+; X86-TRANSFORM-LABEL: clear_lowbits32_c3_load_indexzext:
+; X86-TRANSFORM:       # %bb.0:
+; X86-TRANSFORM-NEXT:    movl {{[0-9]+}}(%esp), %ecx
+; X86-TRANSFORM-NEXT:    movb {{[0-9]+}}(%esp), %al
+; X86-TRANSFORM-NEXT:    movl $-1, %edx
+; X86-TRANSFORM-NEXT:    shlxl %eax, %edx, %eax
+; X86-TRANSFORM-NEXT:    andl (%ecx), %eax
+; X86-TRANSFORM-NEXT:    retl
+;
+; X64-NOTRANSFORM-LABEL: clear_lowbits32_c3_load_indexzext:
+; X64-NOTRANSFORM:       # %bb.0:
+; X64-NOTRANSFORM-NEXT:    movl $-1, %eax
+; X64-NOTRANSFORM-NEXT:    movl %esi, %ecx
+; X64-NOTRANSFORM-NEXT:    shll %cl, %eax
+; X64-NOTRANSFORM-NEXT:    andl (%rdi), %eax
+; X64-NOTRANSFORM-NEXT:    retq
+;
+; X64-TRANSFORM-LABEL: clear_lowbits32_c3_load_indexzext:
+; X64-TRANSFORM:       # %bb.0:
+; X64-TRANSFORM-NEXT:    movl $-1, %eax
+; X64-TRANSFORM-NEXT:    shlxl %esi, %eax, %eax
+; X64-TRANSFORM-NEXT:    andl (%rdi), %eax
+; X64-TRANSFORM-NEXT:    retq
+  %val = load i32, i32* %w
+  %sh_prom = zext i8 %numlowbits to i32
+  %mask = shl i32 -1, %sh_prom
+  %masked = and i32 %mask, %val
+  ret i32 %masked
+}
+
+define i32 @clear_lowbits32_c4_commutative(i32 %val, i32 %numlowbits) nounwind {
+; X86-NOTRANSFORM-LABEL: clear_lowbits32_c4_commutative:
+; X86-NOTRANSFORM:       # %bb.0:
+; X86-NOTRANSFORM-NEXT:    movb {{[0-9]+}}(%esp), %cl
+; X86-NOTRANSFORM-NEXT:    movl $-1, %eax
+; X86-NOTRANSFORM-NEXT:    shll %cl, %eax
+; X86-NOTRANSFORM-NEXT:    andl {{[0-9]+}}(%esp), %eax
+; X86-NOTRANSFORM-NEXT:    retl
+;
+; X86-TRANSFORM-LABEL: clear_lowbits32_c4_commutative:
+; X86-TRANSFORM:       # %bb.0:
+; X86-TRANSFORM-NEXT:    movb {{[0-9]+}}(%esp), %al
+; X86-TRANSFORM-NEXT:    movl $-1, %ecx
+; X86-TRANSFORM-NEXT:    shlxl %eax, %ecx, %eax
+; X86-TRANSFORM-NEXT:    andl {{[0-9]+}}(%esp), %eax
+; X86-TRANSFORM-NEXT:    retl
+;
+; X64-NOTRANSFORM-LABEL: clear_lowbits32_c4_commutative:
+; X64-NOTRANSFORM:       # %bb.0:
+; X64-NOTRANSFORM-NEXT:    movl $-1, %eax
+; X64-NOTRANSFORM-NEXT:    movl %esi, %ecx
+; X64-NOTRANSFORM-NEXT:    shll %cl, %eax
+; X64-NOTRANSFORM-NEXT:    andl %edi, %eax
+; X64-NOTRANSFORM-NEXT:    retq
+;
+; X64-TRANSFORM-LABEL: clear_lowbits32_c4_commutative:
+; X64-TRANSFORM:       # %bb.0:
+; X64-TRANSFORM-NEXT:    movl $-1, %eax
+; X64-TRANSFORM-NEXT:    shlxl %esi, %eax, %eax
+; X64-TRANSFORM-NEXT:    andl %edi, %eax
+; X64-TRANSFORM-NEXT:    retq
+  %mask = shl i32 -1, %numlowbits
+  %masked = and i32 %val, %mask ; swapped order
+  ret i32 %masked
+}
+
+; 64-bit
+
+define i64 @clear_lowbits64_c0(i64 %val, i64 %numlowbits) nounwind {
+; X86-NOTRANSFORM-LABEL: clear_lowbits64_c0:
+; X86-NOTRANSFORM:       # %bb.0:
+; X86-NOTRANSFORM-NEXT:    movb {{[0-9]+}}(%esp), %cl
+; X86-NOTRANSFORM-NEXT:    movl $-1, %edx
+; X86-NOTRANSFORM-NEXT:    movl $-1, %eax
+; X86-NOTRANSFORM-NEXT:    shll %cl, %eax
+; X86-NOTRANSFORM-NEXT:    shldl %cl, %edx, %edx
+; X86-NOTRANSFORM-NEXT:    testb $32, %cl
+; X86-NOTRANSFORM-NEXT:    je .LBB5_2
+; X86-NOTRANSFORM-NEXT:  # %bb.1:
+; X86-NOTRANSFORM-NEXT:    movl %eax, %edx
+; X86-NOTRANSFORM-NEXT:    xorl %eax, %eax
+; X86-NOTRANSFORM-NEXT:  .LBB5_2:
+; X86-NOTRANSFORM-NEXT:    andl {{[0-9]+}}(%esp), %edx
+; X86-NOTRANSFORM-NEXT:    andl {{[0-9]+}}(%esp), %eax
+; X86-NOTRANSFORM-NEXT:    retl
+;
+; X86-TRANSFORM-LABEL: clear_lowbits64_c0:
+; X86-TRANSFORM:       # %bb.0:
+; X86-TRANSFORM-NEXT:    movb {{[0-9]+}}(%esp), %cl
+; X86-TRANSFORM-NEXT:    movl $-1, %edx
+; X86-TRANSFORM-NEXT:    shlxl %ecx, %edx, %eax
+; X86-TRANSFORM-NEXT:    shldl %cl, %edx, %edx
+; X86-TRANSFORM-NEXT:    testb $32, %cl
+; X86-TRANSFORM-NEXT:    je .LBB5_2
+; X86-TRANSFORM-NEXT:  # %bb.1:
+; X86-TRANSFORM-NEXT:    movl %eax, %edx
+; X86-TRANSFORM-NEXT:    xorl %eax, %eax
+; X86-TRANSFORM-NEXT:  .LBB5_2:
+; X86-TRANSFORM-NEXT:    andl {{[0-9]+}}(%esp), %edx
+; X86-TRANSFORM-NEXT:    andl {{[0-9]+}}(%esp), %eax
+; X86-TRANSFORM-NEXT:    retl
+;
+; X64-NOTRANSFORM-LABEL: clear_lowbits64_c0:
+; X64-NOTRANSFORM:       # %bb.0:
+; X64-NOTRANSFORM-NEXT:    movq $-1, %rax
+; X64-NOTRANSFORM-NEXT:    movl %esi, %ecx
+; X64-NOTRANSFORM-NEXT:    shlq %cl, %rax
+; X64-NOTRANSFORM-NEXT:    andq %rdi, %rax
+; X64-NOTRANSFORM-NEXT:    retq
+;
+; X64-TRANSFORM-LABEL: clear_lowbits64_c0:
+; X64-TRANSFORM:       # %bb.0:
+; X64-TRANSFORM-NEXT:    movq $-1, %rax
+; X64-TRANSFORM-NEXT:    shlxq %rsi, %rax, %rax
+; X64-TRANSFORM-NEXT:    andq %rdi, %rax
+; X64-TRANSFORM-NEXT:    retq
+  %mask = shl i64 -1, %numlowbits
+  %masked = and i64 %mask, %val
+  ret i64 %masked
+}
+
+define i64 @clear_lowbits64_c1_indexzext(i64 %val, i8 %numlowbits) nounwind {
+; X86-NOTRANSFORM-LABEL: clear_lowbits64_c1_indexzext:
+; X86-NOTRANSFORM:       # %bb.0:
+; X86-NOTRANSFORM-NEXT:    movb {{[0-9]+}}(%esp), %cl
+; X86-NOTRANSFORM-NEXT:    movl $-1, %edx
+; X86-NOTRANSFORM-NEXT:    movl $-1, %eax
+; X86-NOTRANSFORM-NEXT:    shll %cl, %eax
+; X86-NOTRANSFORM-NEXT:    shldl %cl, %edx, %edx
+; X86-NOTRANSFORM-NEXT:    testb $32, %cl
+; X86-NOTRANSFORM-NEXT:    je .LBB6_2
+; X86-NOTRANSFORM-NEXT:  # %bb.1:
+; X86-NOTRANSFORM-NEXT:    movl %eax, %edx
+; X86-NOTRANSFORM-NEXT:    xorl %eax, %eax
+; X86-NOTRANSFORM-NEXT:  .LBB6_2:
+; X86-NOTRANSFORM-NEXT:    andl {{[0-9]+}}(%esp), %edx
+; X86-NOTRANSFORM-NEXT:    andl {{[0-9]+}}(%esp), %eax
+; X86-NOTRANSFORM-NEXT:    retl
+;
+; X86-TRANSFORM-LABEL: clear_lowbits64_c1_indexzext:
+; X86-TRANSFORM:       # %bb.0:
+; X86-TRANSFORM-NEXT:    movb {{[0-9]+}}(%esp), %cl
+; X86-TRANSFORM-NEXT:    movl $-1, %edx
+; X86-TRANSFORM-NEXT:    shlxl %ecx, %edx, %eax
+; X86-TRANSFORM-NEXT:    shldl %cl, %edx, %edx
+; X86-TRANSFORM-NEXT:    testb $32, %cl
+; X86-TRANSFORM-NEXT:    je .LBB6_2
+; X86-TRANSFORM-NEXT:  # %bb.1:
+; X86-TRANSFORM-NEXT:    movl %eax, %edx
+; X86-TRANSFORM-NEXT:    xorl %eax, %eax
+; X86-TRANSFORM-NEXT:  .LBB6_2:
+; X86-TRANSFORM-NEXT:    andl {{[0-9]+}}(%esp), %edx
+; X86-TRANSFORM-NEXT:    andl {{[0-9]+}}(%esp), %eax
+; X86-TRANSFORM-NEXT:    retl
+;
+; X64-NOTRANSFORM-LABEL: clear_lowbits64_c1_indexzext:
+; X64-NOTRANSFORM:       # %bb.0:
+; X64-NOTRANSFORM-NEXT:    movq $-1, %rax
+; X64-NOTRANSFORM-NEXT:    movl %esi, %ecx
+; X64-NOTRANSFORM-NEXT:    shlq %cl, %rax
+; X64-NOTRANSFORM-NEXT:    andq %rdi, %rax
+; X64-NOTRANSFORM-NEXT:    retq
+;
+; X64-TRANSFORM-LABEL: clear_lowbits64_c1_indexzext:
+; X64-TRANSFORM:       # %bb.0:
+; X64-TRANSFORM-NEXT:    # kill: def $esi killed $esi def $rsi
+; X64-TRANSFORM-NEXT:    movq $-1, %rax
+; X64-TRANSFORM-NEXT:    shlxq %rsi, %rax, %rax
+; X64-TRANSFORM-NEXT:    andq %rdi, %rax
+; X64-TRANSFORM-NEXT:    retq
+  %sh_prom = zext i8 %numlowbits to i64
+  %mask = shl i64 -1, %sh_prom
+  %masked = and i64 %mask, %val
+  ret i64 %masked
+}
+
+define i64 @clear_lowbits64_c2_load(i64* %w, i64 %numlowbits) nounwind {
+; X86-NOTRANSFORM-LABEL: clear_lowbits64_c2_load:
+; X86-NOTRANSFORM:       # %bb.0:
+; X86-NOTRANSFORM-NEXT:    pushl %esi
+; X86-NOTRANSFORM-NEXT:    movl {{[0-9]+}}(%esp), %esi
+; X86-NOTRANSFORM-NEXT:    movb {{[0-9]+}}(%esp), %cl
+; X86-NOTRANSFORM-NEXT:    movl $-1, %edx
+; X86-NOTRANSFORM-NEXT:    movl $-1, %eax
+; X86-NOTRANSFORM-NEXT:    shll %cl, %eax
+; X86-NOTRANSFORM-NEXT:    shldl %cl, %edx, %edx
+; X86-NOTRANSFORM-NEXT:    testb $32, %cl
+; X86-NOTRANSFORM-NEXT:    je .LBB7_2
+; X86-NOTRANSFORM-NEXT:  # %bb.1:
+; X86-NOTRANSFORM-NEXT:    movl %eax, %edx
+; X86-NOTRANSFORM-NEXT:    xorl %eax, %eax
+; X86-NOTRANSFORM-NEXT:  .LBB7_2:
+; X86-NOTRANSFORM-NEXT:    andl 4(%esi), %edx
+; X86-NOTRANSFORM-NEXT:    andl (%esi), %eax
+; X86-NOTRANSFORM-NEXT:    popl %esi
+; X86-NOTRANSFORM-NEXT:    retl
+;
+; X86-TRANSFORM-LABEL: clear_lowbits64_c2_load:
+; X86-TRANSFORM:       # %bb.0:
+; X86-TRANSFORM-NEXT:    pushl %esi
+; X86-TRANSFORM-NEXT:    movl {{[0-9]+}}(%esp), %esi
+; X86-TRANSFORM-NEXT:    movb {{[0-9]+}}(%esp), %cl
+; X86-TRANSFORM-NEXT:    movl $-1, %edx
+; X86-TRANSFORM-NEXT:    shlxl %ecx, %edx, %eax
+; X86-TRANSFORM-NEXT:    shldl %cl, %edx, %edx
+; X86-TRANSFORM-NEXT:    testb $32, %cl
+; X86-TRANSFORM-NEXT:    je .LBB7_2
+; X86-TRANSFORM-NEXT:  # %bb.1:
+; X86-TRANSFORM-NEXT:    movl %eax, %edx
+; X86-TRANSFORM-NEXT:    xorl %eax, %eax
+; X86-TRANSFORM-NEXT:  .LBB7_2:
+; X86-TRANSFORM-NEXT:    andl 4(%esi), %edx
+; X86-TRANSFORM-NEXT:    andl (%esi), %eax
+; X86-TRANSFORM-NEXT:    popl %esi
+; X86-TRANSFORM-NEXT:    retl
+;
+; X64-NOTRANSFORM-LABEL: clear_lowbits64_c2_load:
+; X64-NOTRANSFORM:       # %bb.0:
+; X64-NOTRANSFORM-NEXT:    movq $-1, %rax
+; X64-NOTRANSFORM-NEXT:    movl %esi, %ecx
+; X64-NOTRANSFORM-NEXT:    shlq %cl, %rax
+; X64-NOTRANSFORM-NEXT:    andq (%rdi), %rax
+; X64-NOTRANSFORM-NEXT:    retq
+;
+; X64-TRANSFORM-LABEL: clear_lowbits64_c2_load:
+; X64-TRANSFORM:       # %bb.0:
+; X64-TRANSFORM-NEXT:    movq $-1, %rax
+; X64-TRANSFORM-NEXT:    shlxq %rsi, %rax, %rax
+; X64-TRANSFORM-NEXT:    andq (%rdi), %rax
+; X64-TRANSFORM-NEXT:    retq
+  %val = load i64, i64* %w
+  %mask = shl i64 -1, %numlowbits
+  %masked = and i64 %mask, %val
+  ret i64 %masked
+}
+
+define i64 @clear_lowbits64_c3_load_indexzext(i64* %w, i8 %numlowbits) nounwind {
+; X86-NOTRANSFORM-LABEL: clear_lowbits64_c3_load_indexzext:
+; X86-NOTRANSFORM:       # %bb.0:
+; X86-NOTRANSFORM-NEXT:    pushl %esi
+; X86-NOTRANSFORM-NEXT:    movl {{[0-9]+}}(%esp), %esi
+; X86-NOTRANSFORM-NEXT:    movb {{[0-9]+}}(%esp), %cl
+; X86-NOTRANSFORM-NEXT:    movl $-1, %edx
+; X86-NOTRANSFORM-NEXT:    movl $-1, %eax
+; X86-NOTRANSFORM-NEXT:    shll %cl, %eax
+; X86-NOTRANSFORM-NEXT:    shldl %cl, %edx, %edx
+; X86-NOTRANSFORM-NEXT:    testb $32, %cl
+; X86-NOTRANSFORM-NEXT:    je .LBB8_2
+; X86-NOTRANSFORM-NEXT:  # %bb.1:
+; X86-NOTRANSFORM-NEXT:    movl %eax, %edx
+; X86-NOTRANSFORM-NEXT:    xorl %eax, %eax
+; X86-NOTRANSFORM-NEXT:  .LBB8_2:
+; X86-NOTRANSFORM-NEXT:    andl 4(%esi), %edx
+; X86-NOTRANSFORM-NEXT:    andl (%esi), %eax
+; X86-NOTRANSFORM-NEXT:    popl %esi
+; X86-NOTRANSFORM-NEXT:    retl
+;
+; X86-TRANSFORM-LABEL: clear_lowbits64_c3_load_indexzext:
+; X86-TRANSFORM:       # %bb.0:
+; X86-TRANSFORM-NEXT:    pushl %esi
+; X86-TRANSFORM-NEXT:    movl {{[0-9]+}}(%esp), %esi
+; X86-TRANSFORM-NEXT:    movb {{[0-9]+}}(%esp), %cl
+; X86-TRANSFORM-NEXT:    movl $-1, %edx
+; X86-TRANSFORM-NEXT:    shlxl %ecx, %edx, %eax
+; X86-TRANSFORM-NEXT:    shldl %cl, %edx, %edx
+; X86-TRANSFORM-NEXT:    testb $32, %cl
+; X86-TRANSFORM-NEXT:    je .LBB8_2
+; X86-TRANSFORM-NEXT:  # %bb.1:
+; X86-TRANSFORM-NEXT:    movl %eax, %edx
+; X86-TRANSFORM-NEXT:    xorl %eax, %eax
+; X86-TRANSFORM-NEXT:  .LBB8_2:
+; X86-TRANSFORM-NEXT:    andl 4(%esi), %edx
+; X86-TRANSFORM-NEXT:    andl (%esi), %eax
+; X86-TRANSFORM-NEXT:    popl %esi
+; X86-TRANSFORM-NEXT:    retl
+;
+; X64-NOTRANSFORM-LABEL: clear_lowbits64_c3_load_indexzext:
+; X64-NOTRANSFORM:       # %bb.0:
+; X64-NOTRANSFORM-NEXT:    movq $-1, %rax
+; X64-NOTRANSFORM-NEXT:    movl %esi, %ecx
+; X64-NOTRANSFORM-NEXT:    shlq %cl, %rax
+; X64-NOTRANSFORM-NEXT:    andq (%rdi), %rax
+; X64-NOTRANSFORM-NEXT:    retq
+;
+; X64-TRANSFORM-LABEL: clear_lowbits64_c3_load_indexzext:
+; X64-TRANSFORM:       # %bb.0:
+; X64-TRANSFORM-NEXT:    # kill: def $esi killed $esi def $rsi
+; X64-TRANSFORM-NEXT:    movq $-1, %rax
+; X64-TRANSFORM-NEXT:    shlxq %rsi, %rax, %rax
+; X64-TRANSFORM-NEXT:    andq (%rdi), %rax
+; X64-TRANSFORM-NEXT:    retq
+  %val = load i64, i64* %w
+  %sh_prom = zext i8 %numlowbits to i64
+  %mask = shl i64 -1, %sh_prom
+  %masked = and i64 %mask, %val
+  ret i64 %masked
+}
+
+define i64 @clear_lowbits64_c4_commutative(i64 %val, i64 %numlowbits) nounwind {
+; X86-NOTRANSFORM-LABEL: clear_lowbits64_c4_commutative:
+; X86-NOTRANSFORM:       # %bb.0:
+; X86-NOTRANSFORM-NEXT:    movb {{[0-9]+}}(%esp), %cl
+; X86-NOTRANSFORM-NEXT:    movl $-1, %edx
+; X86-NOTRANSFORM-NEXT:    movl $-1, %eax
+; X86-NOTRANSFORM-NEXT:    shll %cl, %eax
+; X86-NOTRANSFORM-NEXT:    shldl %cl, %edx, %edx
+; X86-NOTRANSFORM-NEXT:    testb $32, %cl
+; X86-NOTRANSFORM-NEXT:    je .LBB9_2
+; X86-NOTRANSFORM-NEXT:  # %bb.1:
+; X86-NOTRANSFORM-NEXT:    movl %eax, %edx
+; X86-NOTRANSFORM-NEXT:    xorl %eax, %eax
+; X86-NOTRANSFORM-NEXT:  .LBB9_2:
+; X86-NOTRANSFORM-NEXT:    andl {{[0-9]+}}(%esp), %edx
+; X86-NOTRANSFORM-NEXT:    andl {{[0-9]+}}(%esp), %eax
+; X86-NOTRANSFORM-NEXT:    retl
+;
+; X86-TRANSFORM-LABEL: clear_lowbits64_c4_commutative:
+; X86-TRANSFORM:       # %bb.0:
+; X86-TRANSFORM-NEXT:    movb {{[0-9]+}}(%esp), %cl
+; X86-TRANSFORM-NEXT:    movl $-1, %edx
+; X86-TRANSFORM-NEXT:    shlxl %ecx, %edx, %eax
+; X86-TRANSFORM-NEXT:    shldl %cl, %edx, %edx
+; X86-TRANSFORM-NEXT:    testb $32, %cl
+; X86-TRANSFORM-NEXT:    je .LBB9_2
+; X86-TRANSFORM-NEXT:  # %bb.1:
+; X86-TRANSFORM-NEXT:    movl %eax, %edx
+; X86-TRANSFORM-NEXT:    xorl %eax, %eax
+; X86-TRANSFORM-NEXT:  .LBB9_2:
+; X86-TRANSFORM-NEXT:    andl {{[0-9]+}}(%esp), %edx
+; X86-TRANSFORM-NEXT:    andl {{[0-9]+}}(%esp), %eax
+; X86-TRANSFORM-NEXT:    retl
+;
+; X64-NOTRANSFORM-LABEL: clear_lowbits64_c4_commutative:
+; X64-NOTRANSFORM:       # %bb.0:
+; X64-NOTRANSFORM-NEXT:    movq $-1, %rax
+; X64-NOTRANSFORM-NEXT:    movl %esi, %ecx
+; X64-NOTRANSFORM-NEXT:    shlq %cl, %rax
+; X64-NOTRANSFORM-NEXT:    andq %rdi, %rax
+; X64-NOTRANSFORM-NEXT:    retq
+;
+; X64-TRANSFORM-LABEL: clear_lowbits64_c4_commutative:
+; X64-TRANSFORM:       # %bb.0:
+; X64-TRANSFORM-NEXT:    movq $-1, %rax
+; X64-TRANSFORM-NEXT:    shlxq %rsi, %rax, %rax
+; X64-TRANSFORM-NEXT:    andq %rdi, %rax
+; X64-TRANSFORM-NEXT:    retq
+  %mask = shl i64 -1, %numlowbits
+  %masked = and i64 %val, %mask ; swapped order
+  ret i64 %masked
+}
+
+; Multi-use tests
+
+declare void @use32(i32)
+declare void @use64(i64)
+
+define i32 @oneuse32(i32 %val, i32 %numlowbits) nounwind {
+; X86-NOTRANSFORM-LABEL: oneuse32:
+; X86-NOTRANSFORM:       # %bb.0:
+; X86-NOTRANSFORM-NEXT:    pushl %esi
+; X86-NOTRANSFORM-NEXT:    subl $8, %esp
+; X86-NOTRANSFORM-NEXT:    movb {{[0-9]+}}(%esp), %cl
+; X86-NOTRANSFORM-NEXT:    movl $-1, %esi
+; X86-NOTRANSFORM-NEXT:    shll %cl, %esi
+; X86-NOTRANSFORM-NEXT:    movl %esi, (%esp)
+; X86-NOTRANSFORM-NEXT:    calll use32
+; X86-NOTRANSFORM-NEXT:    andl {{[0-9]+}}(%esp), %esi
+; X86-NOTRANSFORM-NEXT:    movl %esi, %eax
+; X86-NOTRANSFORM-NEXT:    addl $8, %esp
+; X86-NOTRANSFORM-NEXT:    popl %esi
+; X86-NOTRANSFORM-NEXT:    retl
+;
+; X86-TRANSFORM-LABEL: oneuse32:
+; X86-TRANSFORM:       # %bb.0:
+; X86-TRANSFORM-NEXT:    pushl %esi
+; X86-TRANSFORM-NEXT:    subl $8, %esp
+; X86-TRANSFORM-NEXT:    movb {{[0-9]+}}(%esp), %al
+; X86-TRANSFORM-NEXT:    movl $-1, %ecx
+; X86-TRANSFORM-NEXT:    shlxl %eax, %ecx, %esi
+; X86-TRANSFORM-NEXT:    movl %esi, (%esp)
+; X86-TRANSFORM-NEXT:    calll use32
+; X86-TRANSFORM-NEXT:    andl {{[0-9]+}}(%esp), %esi
+; X86-TRANSFORM-NEXT:    movl %esi, %eax
+; X86-TRANSFORM-NEXT:    addl $8, %esp
+; X86-TRANSFORM-NEXT:    popl %esi
+; X86-TRANSFORM-NEXT:    retl
+;
+; X64-NOTRANSFORM-LABEL: oneuse32:
+; X64-NOTRANSFORM:       # %bb.0:
+; X64-NOTRANSFORM-NEXT:    pushq %rbp
+; X64-NOTRANSFORM-NEXT:    pushq %rbx
+; X64-NOTRANSFORM-NEXT:    pushq %rax
+; X64-NOTRANSFORM-NEXT:    movl %edi, %ebx
+; X64-NOTRANSFORM-NEXT:    movl $-1, %ebp
+; X64-NOTRANSFORM-NEXT:    movl %esi, %ecx
+; X64-NOTRANSFORM-NEXT:    shll %cl, %ebp
+; X64-NOTRANSFORM-NEXT:    movl %ebp, %edi
+; X64-NOTRANSFORM-NEXT:    callq use32
+; X64-NOTRANSFORM-NEXT:    andl %ebx, %ebp
+; X64-NOTRANSFORM-NEXT:    movl %ebp, %eax
+; X64-NOTRANSFORM-NEXT:    addq $8, %rsp
+; X64-NOTRANSFORM-NEXT:    popq %rbx
+; X64-NOTRANSFORM-NEXT:    popq %rbp
+; X64-NOTRANSFORM-NEXT:    retq
+;
+; X64-TRANSFORM-LABEL: oneuse32:
+; X64-TRANSFORM:       # %bb.0:
+; X64-TRANSFORM-NEXT:    pushq %rbp
+; X64-TRANSFORM-NEXT:    pushq %rbx
+; X64-TRANSFORM-NEXT:    pushq %rax
+; X64-TRANSFORM-NEXT:    movl %edi, %ebx
+; X64-TRANSFORM-NEXT:    movl $-1, %eax
+; X64-TRANSFORM-NEXT:    shlxl %esi, %eax, %ebp
+; X64-TRANSFORM-NEXT:    movl %ebp, %edi
+; X64-TRANSFORM-NEXT:    callq use32
+; X64-TRANSFORM-NEXT:    andl %ebx, %ebp
+; X64-TRANSFORM-NEXT:    movl %ebp, %eax
+; X64-TRANSFORM-NEXT:    addq $8, %rsp
+; X64-TRANSFORM-NEXT:    popq %rbx
+; X64-TRANSFORM-NEXT:    popq %rbp
+; X64-TRANSFORM-NEXT:    retq
+  %mask = shl i32 -1, %numlowbits
+  call void @use32(i32 %mask)
+  %masked = and i32 %mask, %val
+  ret i32 %masked
+}
+
+define i64 @oneuse64(i64 %val, i64 %numlowbits) nounwind {
+; X86-NOTRANSFORM-LABEL: oneuse64:
+; X86-NOTRANSFORM:       # %bb.0:
+; X86-NOTRANSFORM-NEXT:    pushl %edi
+; X86-NOTRANSFORM-NEXT:    pushl %esi
+; X86-NOTRANSFORM-NEXT:    pushl %eax
+; X86-NOTRANSFORM-NEXT:    movb {{[0-9]+}}(%esp), %cl
+; X86-NOTRANSFORM-NEXT:    movl $-1, %esi
+; X86-NOTRANSFORM-NEXT:    movl $-1, %edi
+; X86-NOTRANSFORM-NEXT:    shll %cl, %edi
+; X86-NOTRANSFORM-NEXT:    shldl %cl, %esi, %esi
+; X86-NOTRANSFORM-NEXT:    testb $32, %cl
+; X86-NOTRANSFORM-NEXT:    je .LBB11_2
+; X86-NOTRANSFORM-NEXT:  # %bb.1:
+; X86-NOTRANSFORM-NEXT:    movl %edi, %esi
+; X86-NOTRANSFORM-NEXT:    xorl %edi, %edi
+; X86-NOTRANSFORM-NEXT:  .LBB11_2:
+; X86-NOTRANSFORM-NEXT:    subl $8, %esp
+; X86-NOTRANSFORM-NEXT:    pushl %esi
+; X86-NOTRANSFORM-NEXT:    pushl %edi
+; X86-NOTRANSFORM-NEXT:    calll use64
+; X86-NOTRANSFORM-NEXT:    addl $16, %esp
+; X86-NOTRANSFORM-NEXT:    andl {{[0-9]+}}(%esp), %esi
+; X86-NOTRANSFORM-NEXT:    andl {{[0-9]+}}(%esp), %edi
+; X86-NOTRANSFORM-NEXT:    movl %edi, %eax
+; X86-NOTRANSFORM-NEXT:    movl %esi, %edx
+; X86-NOTRANSFORM-NEXT:    addl $4, %esp
+; X86-NOTRANSFORM-NEXT:    popl %esi
+; X86-NOTRANSFORM-NEXT:    popl %edi
+; X86-NOTRANSFORM-NEXT:    retl
+;
+; X86-TRANSFORM-LABEL: oneuse64:
+; X86-TRANSFORM:       # %bb.0:
+; X86-TRANSFORM-NEXT:    pushl %edi
+; X86-TRANSFORM-NEXT:    pushl %esi
+; X86-TRANSFORM-NEXT:    pushl %eax
+; X86-TRANSFORM-NEXT:    movb {{[0-9]+}}(%esp), %cl
+; X86-TRANSFORM-NEXT:    movl $-1, %esi
+; X86-TRANSFORM-NEXT:    shlxl %ecx, %esi, %edi
+; X86-TRANSFORM-NEXT:    shldl %cl, %esi, %esi
+; X86-TRANSFORM-NEXT:    testb $32, %cl
+; X86-TRANSFORM-NEXT:    je .LBB11_2
+; X86-TRANSFORM-NEXT:  # %bb.1:
+; X86-TRANSFORM-NEXT:    movl %edi, %esi
+; X86-TRANSFORM-NEXT:    xorl %edi, %edi
+; X86-TRANSFORM-NEXT:  .LBB11_2:
+; X86-TRANSFORM-NEXT:    subl $8, %esp
+; X86-TRANSFORM-NEXT:    pushl %esi
+; X86-TRANSFORM-NEXT:    pushl %edi
+; X86-TRANSFORM-NEXT:    calll use64
+; X86-TRANSFORM-NEXT:    addl $16, %esp
+; X86-TRANSFORM-NEXT:    andl {{[0-9]+}}(%esp), %esi
+; X86-TRANSFORM-NEXT:    andl {{[0-9]+}}(%esp), %edi
+; X86-TRANSFORM-NEXT:    movl %edi, %eax
+; X86-TRANSFORM-NEXT:    movl %esi, %edx
+; X86-TRANSFORM-NEXT:    addl $4, %esp
+; X86-TRANSFORM-NEXT:    popl %esi
+; X86-TRANSFORM-NEXT:    popl %edi
+; X86-TRANSFORM-NEXT:    retl
+;
+; X64-NOTRANSFORM-LABEL: oneuse64:
+; X64-NOTRANSFORM:       # %bb.0:
+; X64-NOTRANSFORM-NEXT:    pushq %r14
+; X64-NOTRANSFORM-NEXT:    pushq %rbx
+; X64-NOTRANSFORM-NEXT:    pushq %rax
+; X64-NOTRANSFORM-NEXT:    movq %rdi, %r14
+; X64-NOTRANSFORM-NEXT:    movq $-1, %rbx
+; X64-NOTRANSFORM-NEXT:    movl %esi, %ecx
+; X64-NOTRANSFORM-NEXT:    shlq %cl, %rbx
+; X64-NOTRANSFORM-NEXT:    movq %rbx, %rdi
+; X64-NOTRANSFORM-NEXT:    callq use64
+; X64-NOTRANSFORM-NEXT:    andq %r14, %rbx
+; X64-NOTRANSFORM-NEXT:    movq %rbx, %rax
+; X64-NOTRANSFORM-NEXT:    addq $8, %rsp
+; X64-NOTRANSFORM-NEXT:    popq %rbx
+; X64-NOTRANSFORM-NEXT:    popq %r14
+; X64-NOTRANSFORM-NEXT:    retq
+;
+; X64-TRANSFORM-LABEL: oneuse64:
+; X64-TRANSFORM:       # %bb.0:
+; X64-TRANSFORM-NEXT:    pushq %r14
+; X64-TRANSFORM-NEXT:    pushq %rbx
+; X64-TRANSFORM-NEXT:    pushq %rax
+; X64-TRANSFORM-NEXT:    movq %rdi, %r14
+; X64-TRANSFORM-NEXT:    movq $-1, %rax
+; X64-TRANSFORM-NEXT:    shlxq %rsi, %rax, %rbx
+; X64-TRANSFORM-NEXT:    movq %rbx, %rdi
+; X64-TRANSFORM-NEXT:    callq use64
+; X64-TRANSFORM-NEXT:    andq %r14, %rbx
+; X64-TRANSFORM-NEXT:    movq %rbx, %rax
+; X64-TRANSFORM-NEXT:    addq $8, %rsp
+; X64-TRANSFORM-NEXT:    popq %rbx
+; X64-TRANSFORM-NEXT:    popq %r14
+; X64-TRANSFORM-NEXT:    retq
+  %mask = shl i64 -1, %numlowbits
+  call void @use64(i64 %mask)
+  %masked = and i64 %mask, %val
+  ret i64 %masked
+}