Index: lib/Target/X86/X86ISelLowering.cpp =================================================================== --- lib/Target/X86/X86ISelLowering.cpp +++ lib/Target/X86/X86ISelLowering.cpp @@ -32969,6 +32969,14 @@ // transform should reduce code size. It may also enable secondary transforms // from improved known-bits analysis or instruction selection. APInt MaskVal = AndC->getAPIntValue(); + + // If this can be matched by a zero extend, don't optimize. + if (MaskVal.isMask()) { + unsigned TO = MaskVal.countTrailingOnes(); + if (TO >= 8 && isPowerOf2_32(TO)) + return SDValue(); + } + APInt NewMaskVal = MaskVal.lshr(ShiftC->getAPIntValue()); unsigned OldMaskSize = MaskVal.getMinSignedBits(); unsigned NewMaskSize = NewMaskVal.getMinSignedBits(); Index: test/CodeGen/X86/avx512bw-intrinsics-fast-isel.ll =================================================================== --- test/CodeGen/X86/avx512bw-intrinsics-fast-isel.ll +++ test/CodeGen/X86/avx512bw-intrinsics-fast-isel.ll @@ -112,21 +112,21 @@ ; X32-NEXT: .cfi_def_cfa_offset 12 ; X32-NEXT: .cfi_offset %esi, -12 ; X32-NEXT: .cfi_offset %ebx, -8 -; X32-NEXT: movl {{[0-9]+}}(%esp), %ecx -; X32-NEXT: kmovd %ecx, %k0 +; X32-NEXT: movl {{[0-9]+}}(%esp), %eax +; X32-NEXT: kmovd %eax, %k0 ; X32-NEXT: kshiftrq $1, %k0, %k1 -; X32-NEXT: movl %ecx, %eax -; X32-NEXT: andb $2, %al -; X32-NEXT: shrb %al -; X32-NEXT: kmovd %eax, %k2 +; X32-NEXT: movl %eax, %ecx +; X32-NEXT: andb $2, %cl +; X32-NEXT: shrb %cl +; X32-NEXT: kmovd %ecx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $62, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $2, %k0, %k1 -; X32-NEXT: movl %ecx, %eax -; X32-NEXT: andb $15, %al -; X32-NEXT: movl %eax, %edx +; X32-NEXT: movl %eax, %ecx +; X32-NEXT: andb $15, %cl +; X32-NEXT: movl %ecx, %edx ; X32-NEXT: shrb $2, %dl ; X32-NEXT: kmovd %edx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 @@ -134,127 +134,124 @@ ; X32-NEXT: kshiftrq $61, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $3, %k0, %k1 -; X32-NEXT: shrb $3, %al -; X32-NEXT: kmovd %eax, %k2 +; X32-NEXT: shrb $3, %cl +; X32-NEXT: kmovd %ecx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $60, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $4, %k0, %k1 -; X32-NEXT: movl %ecx, %eax -; X32-NEXT: shrb $4, %al -; X32-NEXT: kmovd %eax, %k2 +; X32-NEXT: movl %eax, %ecx +; X32-NEXT: shrb $4, %cl +; X32-NEXT: kmovd %ecx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $59, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $5, %k0, %k1 -; X32-NEXT: movl %ecx, %eax -; X32-NEXT: shrb $5, %al -; X32-NEXT: andb $1, %al -; X32-NEXT: kmovd %eax, %k2 +; X32-NEXT: movl %eax, %ecx +; X32-NEXT: shrb $5, %cl +; X32-NEXT: andb $1, %cl +; X32-NEXT: kmovd %ecx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $58, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $6, %k0, %k1 -; X32-NEXT: movl %ecx, %eax -; X32-NEXT: shrb $6, %al -; X32-NEXT: kmovd %eax, %k2 +; X32-NEXT: movl %eax, %ecx +; X32-NEXT: shrb $6, %cl +; X32-NEXT: kmovd %ecx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $57, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $7, %k0, %k1 -; X32-NEXT: movl %ecx, %eax -; X32-NEXT: shrb $7, %al -; X32-NEXT: kmovd %eax, %k2 +; X32-NEXT: movl %eax, %ecx +; X32-NEXT: shrb $7, %cl +; X32-NEXT: kmovd %ecx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $56, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $8, %k0, %k1 -; X32-NEXT: movb %ch, %al -; X32-NEXT: kmovd %eax, %k2 +; X32-NEXT: movb %ah, %cl +; X32-NEXT: kmovd %ecx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $55, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $9, %k0, %k1 -; X32-NEXT: andb $2, %al -; X32-NEXT: shrb %al -; X32-NEXT: kmovd %eax, %k2 +; X32-NEXT: andb $2, %cl +; X32-NEXT: shrb %cl +; X32-NEXT: kmovd %ecx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $54, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $10, %k0, %k1 -; X32-NEXT: movb %ch, %al -; X32-NEXT: andb $15, %al -; X32-NEXT: movl %eax, %edx +; X32-NEXT: movb %ah, %cl +; X32-NEXT: movl %eax, %esi +; X32-NEXT: andb $15, %cl +; X32-NEXT: movl %ecx, %edx ; X32-NEXT: shrb $2, %dl ; X32-NEXT: kmovd %edx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 +; X32-NEXT: shrb $3, %cl +; X32-NEXT: kmovd %ecx, %k2 +; X32-NEXT: movzwl %si, %edx +; X32-NEXT: movl %edx, %ecx +; X32-NEXT: shrl $12, %ecx +; X32-NEXT: kmovd %ecx, %k3 +; X32-NEXT: movl %esi, %ecx +; X32-NEXT: shrl $13, %ecx +; X32-NEXT: andb $1, %cl +; X32-NEXT: kmovd %ecx, %k4 +; X32-NEXT: movl %edx, %ecx +; X32-NEXT: shrl $14, %ecx +; X32-NEXT: kmovd %ecx, %k5 +; X32-NEXT: shrl $16, %eax +; X32-NEXT: movl %eax, %ebx +; X32-NEXT: andb $2, %bl +; X32-NEXT: shrb %bl +; X32-NEXT: kmovd %ebx, %k6 +; X32-NEXT: movl %eax, %ecx +; X32-NEXT: andb $15, %cl +; X32-NEXT: movl %ecx, %ebx +; X32-NEXT: shrb $2, %bl +; X32-NEXT: kmovd %ebx, %k7 +; X32-NEXT: movl {{[0-9]+}}(%esp), %ebx ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $53, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $11, %k0, %k1 -; X32-NEXT: shrb $3, %al -; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 -; X32-NEXT: movl %ecx, %eax -; X32-NEXT: shrl $12, %eax -; X32-NEXT: andl $15, %eax -; X32-NEXT: kmovd %eax, %k2 -; X32-NEXT: movl %ecx, %eax -; X32-NEXT: shrl $13, %eax -; X32-NEXT: andb $1, %al -; X32-NEXT: kmovd %eax, %k3 -; X32-NEXT: movl %ecx, %eax -; X32-NEXT: shrl $14, %eax -; X32-NEXT: andl $3, %eax -; X32-NEXT: kmovd %eax, %k4 -; X32-NEXT: movl %ecx, %eax -; X32-NEXT: shrl $15, %eax -; X32-NEXT: andl $1, %eax -; X32-NEXT: kmovd %eax, %k5 -; X32-NEXT: movl %ecx, %edx -; X32-NEXT: shrl $16, %edx -; X32-NEXT: movl %edx, %eax -; X32-NEXT: andb $2, %al -; X32-NEXT: shrb %al -; X32-NEXT: kmovd %eax, %k6 -; X32-NEXT: movl %edx, %ebx -; X32-NEXT: andb $15, %bl -; X32-NEXT: movl %ebx, %eax -; X32-NEXT: shrb $2, %al -; X32-NEXT: kmovd %eax, %k7 -; X32-NEXT: movl {{[0-9]+}}(%esp), %eax ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $52, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $12, %k0, %k1 -; X32-NEXT: kxorq %k2, %k1, %k1 +; X32-NEXT: kxorq %k3, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $51, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $13, %k0, %k1 -; X32-NEXT: kxorq %k3, %k1, %k1 +; X32-NEXT: kxorq %k4, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $50, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $14, %k0, %k1 -; X32-NEXT: kxorq %k4, %k1, %k1 +; X32-NEXT: kxorq %k5, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $49, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $15, %k0, %k1 -; X32-NEXT: kxorq %k5, %k1, %k1 +; X32-NEXT: shrl $15, %edx +; X32-NEXT: kmovd %edx, %k2 +; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $48, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $16, %k0, %k1 -; X32-NEXT: kmovd %edx, %k2 +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $47, %k1, %k1 @@ -270,46 +267,46 @@ ; X32-NEXT: kshiftrq $45, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $19, %k0, %k1 -; X32-NEXT: shrb $3, %bl -; X32-NEXT: kmovd %ebx, %k2 +; X32-NEXT: shrb $3, %cl +; X32-NEXT: kmovd %ecx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $44, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $20, %k0, %k1 -; X32-NEXT: movl %edx, %ebx -; X32-NEXT: shrb $4, %bl -; X32-NEXT: kmovd %ebx, %k2 +; X32-NEXT: movl %eax, %ecx +; X32-NEXT: shrb $4, %cl +; X32-NEXT: kmovd %ecx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $43, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $21, %k0, %k1 -; X32-NEXT: movl %edx, %ebx -; X32-NEXT: shrb $5, %bl -; X32-NEXT: andb $1, %bl -; X32-NEXT: kmovd %ebx, %k2 +; X32-NEXT: movl %eax, %ecx +; X32-NEXT: shrb $5, %cl +; X32-NEXT: andb $1, %cl +; X32-NEXT: kmovd %ecx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $42, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $22, %k0, %k1 -; X32-NEXT: movl %edx, %ebx -; X32-NEXT: shrb $6, %bl -; X32-NEXT: kmovd %ebx, %k2 +; X32-NEXT: movl %eax, %ecx +; X32-NEXT: shrb $6, %cl +; X32-NEXT: kmovd %ecx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $41, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $23, %k0, %k1 -; X32-NEXT: shrb $7, %dl -; X32-NEXT: kmovd %edx, %k2 +; X32-NEXT: shrb $7, %al +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $40, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $24, %k0, %k1 -; X32-NEXT: movl %ecx, %edx +; X32-NEXT: movl %esi, %edx ; X32-NEXT: shrl $24, %edx ; X32-NEXT: kmovd %edx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 @@ -317,19 +314,19 @@ ; X32-NEXT: kshiftrq $39, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $25, %k0, %k1 -; X32-NEXT: movl %edx, %ebx -; X32-NEXT: andb $2, %bl -; X32-NEXT: shrb %bl -; X32-NEXT: kmovd %ebx, %k2 +; X32-NEXT: movl %edx, %eax +; X32-NEXT: andb $2, %al +; X32-NEXT: shrb %al +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $38, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $26, %k0, %k1 ; X32-NEXT: andb $15, %dl -; X32-NEXT: movl %edx, %ebx -; X32-NEXT: shrb $2, %bl -; X32-NEXT: kmovd %ebx, %k2 +; X32-NEXT: movl %edx, %eax +; X32-NEXT: shrb $2, %al +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $37, %k1, %k1 @@ -342,26 +339,27 @@ ; X32-NEXT: kshiftrq $36, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $28, %k0, %k1 -; X32-NEXT: movl %ecx, %edx -; X32-NEXT: shrl $28, %edx -; X32-NEXT: kmovd %edx, %k2 +; X32-NEXT: movl %esi, %ecx +; X32-NEXT: movl %ecx, %eax +; X32-NEXT: shrl $28, %eax +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $35, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $29, %k0, %k1 -; X32-NEXT: movl %ecx, %edx -; X32-NEXT: shrl $29, %edx -; X32-NEXT: andb $1, %dl -; X32-NEXT: kmovd %edx, %k2 +; X32-NEXT: movl %ecx, %eax +; X32-NEXT: shrl $29, %eax +; X32-NEXT: andb $1, %al +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $34, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $30, %k0, %k1 -; X32-NEXT: movl %ecx, %edx -; X32-NEXT: shrl $30, %edx -; X32-NEXT: kmovd %edx, %k2 +; X32-NEXT: movl %ecx, %eax +; X32-NEXT: shrl $30, %eax +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $33, %k1, %k1 @@ -374,99 +372,99 @@ ; X32-NEXT: kshiftrq $32, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $32, %k0, %k1 -; X32-NEXT: kmovd %eax, %k2 +; X32-NEXT: kmovd %ebx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $31, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $33, %k0, %k1 -; X32-NEXT: movl %eax, %ecx -; X32-NEXT: andb $2, %cl -; X32-NEXT: shrb %cl -; X32-NEXT: kmovd %ecx, %k2 +; X32-NEXT: movl %ebx, %eax +; X32-NEXT: andb $2, %al +; X32-NEXT: shrb %al +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $30, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $34, %k0, %k1 +; X32-NEXT: movl %ebx, %eax +; X32-NEXT: andb $15, %al ; X32-NEXT: movl %eax, %ecx -; X32-NEXT: andb $15, %cl -; X32-NEXT: movl %ecx, %edx -; X32-NEXT: shrb $2, %dl -; X32-NEXT: kmovd %edx, %k2 +; X32-NEXT: shrb $2, %cl +; X32-NEXT: kmovd %ecx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $29, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $35, %k0, %k1 -; X32-NEXT: shrb $3, %cl -; X32-NEXT: kmovd %ecx, %k2 +; X32-NEXT: shrb $3, %al +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $28, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $36, %k0, %k1 -; X32-NEXT: movl %eax, %ecx -; X32-NEXT: shrb $4, %cl -; X32-NEXT: kmovd %ecx, %k2 +; X32-NEXT: movl %ebx, %eax +; X32-NEXT: shrb $4, %al +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $27, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $37, %k0, %k1 -; X32-NEXT: movl %eax, %ecx -; X32-NEXT: shrb $5, %cl -; X32-NEXT: andb $1, %cl -; X32-NEXT: kmovd %ecx, %k2 +; X32-NEXT: movl %ebx, %eax +; X32-NEXT: shrb $5, %al +; X32-NEXT: andb $1, %al +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $26, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $38, %k0, %k1 -; X32-NEXT: movl %eax, %ecx -; X32-NEXT: shrb $6, %cl -; X32-NEXT: kmovd %ecx, %k2 +; X32-NEXT: movl %ebx, %eax +; X32-NEXT: shrb $6, %al +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $25, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $39, %k0, %k1 -; X32-NEXT: movl %eax, %ecx -; X32-NEXT: shrb $7, %cl -; X32-NEXT: kmovd %ecx, %k2 +; X32-NEXT: movl %ebx, %eax +; X32-NEXT: shrb $7, %al +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $24, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $40, %k0, %k1 -; X32-NEXT: movb %ah, %cl -; X32-NEXT: kmovd %ecx, %k2 +; X32-NEXT: movb %bh, %al +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 -; X32-NEXT: andb $2, %cl -; X32-NEXT: shrb %cl -; X32-NEXT: kmovd %ecx, %k2 -; X32-NEXT: movb %ah, %cl -; X32-NEXT: andb $15, %cl -; X32-NEXT: movl %ecx, %edx -; X32-NEXT: shrb $2, %dl -; X32-NEXT: kmovd %edx, %k3 -; X32-NEXT: shrb $3, %cl -; X32-NEXT: kmovd %ecx, %k4 -; X32-NEXT: movl %eax, %ecx -; X32-NEXT: shrl $13, %ecx -; X32-NEXT: andb $1, %cl -; X32-NEXT: kmovd %ecx, %k5 +; X32-NEXT: andb $2, %al +; X32-NEXT: shrb %al +; X32-NEXT: kmovd %eax, %k2 +; X32-NEXT: movb %bh, %al +; X32-NEXT: andb $15, %al ; X32-NEXT: movl %eax, %ecx +; X32-NEXT: shrb $2, %cl +; X32-NEXT: kmovd %ecx, %k3 +; X32-NEXT: shrb $3, %al +; X32-NEXT: kmovd %eax, %k4 +; X32-NEXT: movl %ebx, %eax +; X32-NEXT: shrl $13, %eax +; X32-NEXT: andb $1, %al +; X32-NEXT: kmovd %eax, %k5 +; X32-NEXT: movl %ebx, %ecx ; X32-NEXT: shrl $16, %ecx -; X32-NEXT: movl %ecx, %edx -; X32-NEXT: andb $2, %dl -; X32-NEXT: shrb %dl -; X32-NEXT: kmovd %edx, %k6 +; X32-NEXT: movl %ecx, %eax +; X32-NEXT: andb $2, %al +; X32-NEXT: shrb %al +; X32-NEXT: kmovd %eax, %k6 ; X32-NEXT: movl %ecx, %edx ; X32-NEXT: andb $15, %dl -; X32-NEXT: movl %edx, %ebx -; X32-NEXT: shrb $2, %bl -; X32-NEXT: kmovd %ebx, %k7 +; X32-NEXT: movl %edx, %eax +; X32-NEXT: shrb $2, %al +; X32-NEXT: kmovd %eax, %k7 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $23, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 @@ -477,23 +475,20 @@ ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $42, %k0, %k1 ; X32-NEXT: kxorq %k3, %k1, %k1 -; X32-NEXT: kshiftlq $63, %k1, %k1 -; X32-NEXT: kshiftrq $21, %k1, %k1 -; X32-NEXT: kxorq %k0, %k1, %k0 -; X32-NEXT: kshiftrq $43, %k0, %k1 -; X32-NEXT: kxorq %k4, %k1, %k1 +; X32-NEXT: movzwl %bx, %eax ; X32-NEXT: movl %eax, %esi ; X32-NEXT: shrl $12, %esi -; X32-NEXT: andl $15, %esi ; X32-NEXT: kmovd %esi, %k2 ; X32-NEXT: movl %eax, %esi ; X32-NEXT: shrl $14, %esi -; X32-NEXT: andl $3, %esi ; X32-NEXT: kmovd %esi, %k3 -; X32-NEXT: movl %eax, %esi -; X32-NEXT: shrl $15, %esi -; X32-NEXT: andl $1, %esi -; X32-NEXT: kmovd %esi, %k4 +; X32-NEXT: kshiftlq $63, %k1, %k1 +; X32-NEXT: kshiftrq $21, %k1, %k1 +; X32-NEXT: kxorq %k0, %k1, %k0 +; X32-NEXT: kshiftrq $43, %k0, %k1 +; X32-NEXT: kxorq %k4, %k1, %k1 +; X32-NEXT: shrl $15, %eax +; X32-NEXT: kmovd %eax, %k4 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $20, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 @@ -541,30 +536,30 @@ ; X32-NEXT: kshiftrq $12, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k4 ; X32-NEXT: kshiftrq $52, %k4, %k0 -; X32-NEXT: movl %ecx, %edx -; X32-NEXT: shrb $4, %dl -; X32-NEXT: kmovd %edx, %k1 +; X32-NEXT: movl %ecx, %eax +; X32-NEXT: shrb $4, %al +; X32-NEXT: kmovd %eax, %k1 ; X32-NEXT: kxorq %k1, %k0, %k5 -; X32-NEXT: movl %ecx, %edx -; X32-NEXT: shrb $5, %dl -; X32-NEXT: andb $1, %dl -; X32-NEXT: kmovd %edx, %k6 -; X32-NEXT: movl %ecx, %edx -; X32-NEXT: shrb $6, %dl -; X32-NEXT: kmovd %edx, %k7 +; X32-NEXT: movl %ecx, %eax +; X32-NEXT: shrb $5, %al +; X32-NEXT: andb $1, %al +; X32-NEXT: kmovd %eax, %k6 +; X32-NEXT: movl %ecx, %eax +; X32-NEXT: shrb $6, %al +; X32-NEXT: kmovd %eax, %k7 ; X32-NEXT: shrb $7, %cl ; X32-NEXT: kmovd %ecx, %k0 -; X32-NEXT: movl %eax, %ecx +; X32-NEXT: movl %ebx, %ecx ; X32-NEXT: shrl $24, %ecx ; X32-NEXT: kmovd %ecx, %k1 -; X32-NEXT: movl %ecx, %edx -; X32-NEXT: andb $2, %dl -; X32-NEXT: shrb %dl -; X32-NEXT: kmovd %edx, %k2 +; X32-NEXT: movl %ecx, %eax +; X32-NEXT: andb $2, %al +; X32-NEXT: shrb %al +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: andb $15, %cl -; X32-NEXT: movl %ecx, %edx -; X32-NEXT: shrb $2, %dl -; X32-NEXT: kmovd %edx, %k3 +; X32-NEXT: movl %ecx, %eax +; X32-NEXT: shrb $2, %al +; X32-NEXT: kmovd %eax, %k3 ; X32-NEXT: kshiftlq $63, %k5, %k5 ; X32-NEXT: kshiftrq $11, %k5, %k5 ; X32-NEXT: kxorq %k4, %k5, %k4 @@ -577,10 +572,10 @@ ; X32-NEXT: kxorq %k7, %k4, %k6 ; X32-NEXT: shrb $3, %cl ; X32-NEXT: kmovd %ecx, %k4 -; X32-NEXT: movl %eax, %ecx -; X32-NEXT: shrl $29, %ecx -; X32-NEXT: andb $1, %cl -; X32-NEXT: kmovd %ecx, %k7 +; X32-NEXT: movl %ebx, %eax +; X32-NEXT: shrl $29, %eax +; X32-NEXT: andb $1, %al +; X32-NEXT: kmovd %eax, %k7 ; X32-NEXT: kshiftlq $63, %k6, %k6 ; X32-NEXT: kshiftrq $9, %k6, %k6 ; X32-NEXT: kxorq %k5, %k6, %k5 @@ -591,12 +586,12 @@ ; X32-NEXT: kxorq %k5, %k0, %k0 ; X32-NEXT: kshiftrq $56, %k0, %k5 ; X32-NEXT: kxorq %k1, %k5, %k1 -; X32-NEXT: movl %eax, %ecx -; X32-NEXT: shrl $28, %ecx -; X32-NEXT: kmovd %ecx, %k5 -; X32-NEXT: movl %eax, %ecx -; X32-NEXT: shrl $30, %ecx -; X32-NEXT: kmovd %ecx, %k6 +; X32-NEXT: movl %ebx, %eax +; X32-NEXT: shrl $28, %eax +; X32-NEXT: kmovd %eax, %k5 +; X32-NEXT: movl %ebx, %eax +; X32-NEXT: shrl $30, %eax +; X32-NEXT: kmovd %eax, %k6 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $7, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 @@ -627,8 +622,8 @@ ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $62, %k0, %k1 ; X32-NEXT: kxorq %k6, %k1, %k1 -; X32-NEXT: shrl $31, %eax -; X32-NEXT: kmovd %eax, %k2 +; X32-NEXT: shrl $31, %ebx +; X32-NEXT: kmovd %ebx, %k2 ; X32-NEXT: movb {{[0-9]+}}(%esp), %al ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $1, %k1, %k1 @@ -666,21 +661,21 @@ ; X32-NEXT: .cfi_def_cfa_offset 12 ; X32-NEXT: .cfi_offset %esi, -12 ; X32-NEXT: .cfi_offset %ebx, -8 -; X32-NEXT: movl {{[0-9]+}}(%esp), %ecx -; X32-NEXT: kmovd %ecx, %k0 +; X32-NEXT: movl {{[0-9]+}}(%esp), %eax +; X32-NEXT: kmovd %eax, %k0 ; X32-NEXT: kshiftrq $1, %k0, %k1 -; X32-NEXT: movl %ecx, %eax -; X32-NEXT: andb $2, %al -; X32-NEXT: shrb %al -; X32-NEXT: kmovd %eax, %k2 +; X32-NEXT: movl %eax, %ecx +; X32-NEXT: andb $2, %cl +; X32-NEXT: shrb %cl +; X32-NEXT: kmovd %ecx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $62, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $2, %k0, %k1 -; X32-NEXT: movl %ecx, %eax -; X32-NEXT: andb $15, %al -; X32-NEXT: movl %eax, %edx +; X32-NEXT: movl %eax, %ecx +; X32-NEXT: andb $15, %cl +; X32-NEXT: movl %ecx, %edx ; X32-NEXT: shrb $2, %dl ; X32-NEXT: kmovd %edx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 @@ -688,127 +683,124 @@ ; X32-NEXT: kshiftrq $61, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $3, %k0, %k1 -; X32-NEXT: shrb $3, %al -; X32-NEXT: kmovd %eax, %k2 +; X32-NEXT: shrb $3, %cl +; X32-NEXT: kmovd %ecx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $60, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $4, %k0, %k1 -; X32-NEXT: movl %ecx, %eax -; X32-NEXT: shrb $4, %al -; X32-NEXT: kmovd %eax, %k2 +; X32-NEXT: movl %eax, %ecx +; X32-NEXT: shrb $4, %cl +; X32-NEXT: kmovd %ecx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $59, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $5, %k0, %k1 -; X32-NEXT: movl %ecx, %eax -; X32-NEXT: shrb $5, %al -; X32-NEXT: andb $1, %al -; X32-NEXT: kmovd %eax, %k2 +; X32-NEXT: movl %eax, %ecx +; X32-NEXT: shrb $5, %cl +; X32-NEXT: andb $1, %cl +; X32-NEXT: kmovd %ecx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $58, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $6, %k0, %k1 -; X32-NEXT: movl %ecx, %eax -; X32-NEXT: shrb $6, %al -; X32-NEXT: kmovd %eax, %k2 +; X32-NEXT: movl %eax, %ecx +; X32-NEXT: shrb $6, %cl +; X32-NEXT: kmovd %ecx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $57, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $7, %k0, %k1 -; X32-NEXT: movl %ecx, %eax -; X32-NEXT: shrb $7, %al -; X32-NEXT: kmovd %eax, %k2 +; X32-NEXT: movl %eax, %ecx +; X32-NEXT: shrb $7, %cl +; X32-NEXT: kmovd %ecx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $56, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $8, %k0, %k1 -; X32-NEXT: movb %ch, %al -; X32-NEXT: kmovd %eax, %k2 +; X32-NEXT: movb %ah, %cl +; X32-NEXT: kmovd %ecx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $55, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $9, %k0, %k1 -; X32-NEXT: andb $2, %al -; X32-NEXT: shrb %al -; X32-NEXT: kmovd %eax, %k2 +; X32-NEXT: andb $2, %cl +; X32-NEXT: shrb %cl +; X32-NEXT: kmovd %ecx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $54, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $10, %k0, %k1 -; X32-NEXT: movb %ch, %al -; X32-NEXT: andb $15, %al -; X32-NEXT: movl %eax, %edx +; X32-NEXT: movb %ah, %cl +; X32-NEXT: movl %eax, %esi +; X32-NEXT: andb $15, %cl +; X32-NEXT: movl %ecx, %edx ; X32-NEXT: shrb $2, %dl ; X32-NEXT: kmovd %edx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 +; X32-NEXT: shrb $3, %cl +; X32-NEXT: kmovd %ecx, %k2 +; X32-NEXT: movzwl %si, %edx +; X32-NEXT: movl %edx, %ecx +; X32-NEXT: shrl $12, %ecx +; X32-NEXT: kmovd %ecx, %k3 +; X32-NEXT: movl %esi, %ecx +; X32-NEXT: shrl $13, %ecx +; X32-NEXT: andb $1, %cl +; X32-NEXT: kmovd %ecx, %k4 +; X32-NEXT: movl %edx, %ecx +; X32-NEXT: shrl $14, %ecx +; X32-NEXT: kmovd %ecx, %k5 +; X32-NEXT: shrl $16, %eax +; X32-NEXT: movl %eax, %ebx +; X32-NEXT: andb $2, %bl +; X32-NEXT: shrb %bl +; X32-NEXT: kmovd %ebx, %k6 +; X32-NEXT: movl %eax, %ecx +; X32-NEXT: andb $15, %cl +; X32-NEXT: movl %ecx, %ebx +; X32-NEXT: shrb $2, %bl +; X32-NEXT: kmovd %ebx, %k7 +; X32-NEXT: movl {{[0-9]+}}(%esp), %ebx ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $53, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 -; X32-NEXT: kshiftrq $11, %k0, %k1 -; X32-NEXT: shrb $3, %al -; X32-NEXT: kmovd %eax, %k2 +; X32-NEXT: kshiftrq $11, %k0, %k1 ; X32-NEXT: kxorq %k2, %k1, %k1 -; X32-NEXT: movl %ecx, %eax -; X32-NEXT: shrl $12, %eax -; X32-NEXT: andl $15, %eax -; X32-NEXT: kmovd %eax, %k2 -; X32-NEXT: movl %ecx, %eax -; X32-NEXT: shrl $13, %eax -; X32-NEXT: andb $1, %al -; X32-NEXT: kmovd %eax, %k3 -; X32-NEXT: movl %ecx, %eax -; X32-NEXT: shrl $14, %eax -; X32-NEXT: andl $3, %eax -; X32-NEXT: kmovd %eax, %k4 -; X32-NEXT: movl %ecx, %eax -; X32-NEXT: shrl $15, %eax -; X32-NEXT: andl $1, %eax -; X32-NEXT: kmovd %eax, %k5 -; X32-NEXT: movl %ecx, %edx -; X32-NEXT: shrl $16, %edx -; X32-NEXT: movl %edx, %eax -; X32-NEXT: andb $2, %al -; X32-NEXT: shrb %al -; X32-NEXT: kmovd %eax, %k6 -; X32-NEXT: movl %edx, %ebx -; X32-NEXT: andb $15, %bl -; X32-NEXT: movl %ebx, %eax -; X32-NEXT: shrb $2, %al -; X32-NEXT: kmovd %eax, %k7 -; X32-NEXT: movl {{[0-9]+}}(%esp), %eax ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $52, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $12, %k0, %k1 -; X32-NEXT: kxorq %k2, %k1, %k1 +; X32-NEXT: kxorq %k3, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $51, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $13, %k0, %k1 -; X32-NEXT: kxorq %k3, %k1, %k1 +; X32-NEXT: kxorq %k4, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $50, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $14, %k0, %k1 -; X32-NEXT: kxorq %k4, %k1, %k1 +; X32-NEXT: kxorq %k5, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $49, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $15, %k0, %k1 -; X32-NEXT: kxorq %k5, %k1, %k1 +; X32-NEXT: shrl $15, %edx +; X32-NEXT: kmovd %edx, %k2 +; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $48, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $16, %k0, %k1 -; X32-NEXT: kmovd %edx, %k2 +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $47, %k1, %k1 @@ -824,46 +816,46 @@ ; X32-NEXT: kshiftrq $45, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $19, %k0, %k1 -; X32-NEXT: shrb $3, %bl -; X32-NEXT: kmovd %ebx, %k2 +; X32-NEXT: shrb $3, %cl +; X32-NEXT: kmovd %ecx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $44, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $20, %k0, %k1 -; X32-NEXT: movl %edx, %ebx -; X32-NEXT: shrb $4, %bl -; X32-NEXT: kmovd %ebx, %k2 +; X32-NEXT: movl %eax, %ecx +; X32-NEXT: shrb $4, %cl +; X32-NEXT: kmovd %ecx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $43, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $21, %k0, %k1 -; X32-NEXT: movl %edx, %ebx -; X32-NEXT: shrb $5, %bl -; X32-NEXT: andb $1, %bl -; X32-NEXT: kmovd %ebx, %k2 +; X32-NEXT: movl %eax, %ecx +; X32-NEXT: shrb $5, %cl +; X32-NEXT: andb $1, %cl +; X32-NEXT: kmovd %ecx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $42, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $22, %k0, %k1 -; X32-NEXT: movl %edx, %ebx -; X32-NEXT: shrb $6, %bl -; X32-NEXT: kmovd %ebx, %k2 +; X32-NEXT: movl %eax, %ecx +; X32-NEXT: shrb $6, %cl +; X32-NEXT: kmovd %ecx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $41, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $23, %k0, %k1 -; X32-NEXT: shrb $7, %dl -; X32-NEXT: kmovd %edx, %k2 +; X32-NEXT: shrb $7, %al +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $40, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $24, %k0, %k1 -; X32-NEXT: movl %ecx, %edx +; X32-NEXT: movl %esi, %edx ; X32-NEXT: shrl $24, %edx ; X32-NEXT: kmovd %edx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 @@ -871,19 +863,19 @@ ; X32-NEXT: kshiftrq $39, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $25, %k0, %k1 -; X32-NEXT: movl %edx, %ebx -; X32-NEXT: andb $2, %bl -; X32-NEXT: shrb %bl -; X32-NEXT: kmovd %ebx, %k2 +; X32-NEXT: movl %edx, %eax +; X32-NEXT: andb $2, %al +; X32-NEXT: shrb %al +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $38, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $26, %k0, %k1 ; X32-NEXT: andb $15, %dl -; X32-NEXT: movl %edx, %ebx -; X32-NEXT: shrb $2, %bl -; X32-NEXT: kmovd %ebx, %k2 +; X32-NEXT: movl %edx, %eax +; X32-NEXT: shrb $2, %al +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $37, %k1, %k1 @@ -896,26 +888,27 @@ ; X32-NEXT: kshiftrq $36, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $28, %k0, %k1 -; X32-NEXT: movl %ecx, %edx -; X32-NEXT: shrl $28, %edx -; X32-NEXT: kmovd %edx, %k2 +; X32-NEXT: movl %esi, %ecx +; X32-NEXT: movl %ecx, %eax +; X32-NEXT: shrl $28, %eax +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $35, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $29, %k0, %k1 -; X32-NEXT: movl %ecx, %edx -; X32-NEXT: shrl $29, %edx -; X32-NEXT: andb $1, %dl -; X32-NEXT: kmovd %edx, %k2 +; X32-NEXT: movl %ecx, %eax +; X32-NEXT: shrl $29, %eax +; X32-NEXT: andb $1, %al +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $34, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $30, %k0, %k1 -; X32-NEXT: movl %ecx, %edx -; X32-NEXT: shrl $30, %edx -; X32-NEXT: kmovd %edx, %k2 +; X32-NEXT: movl %ecx, %eax +; X32-NEXT: shrl $30, %eax +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $33, %k1, %k1 @@ -928,99 +921,99 @@ ; X32-NEXT: kshiftrq $32, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $32, %k0, %k1 -; X32-NEXT: kmovd %eax, %k2 +; X32-NEXT: kmovd %ebx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $31, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $33, %k0, %k1 -; X32-NEXT: movl %eax, %ecx -; X32-NEXT: andb $2, %cl -; X32-NEXT: shrb %cl -; X32-NEXT: kmovd %ecx, %k2 +; X32-NEXT: movl %ebx, %eax +; X32-NEXT: andb $2, %al +; X32-NEXT: shrb %al +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $30, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $34, %k0, %k1 +; X32-NEXT: movl %ebx, %eax +; X32-NEXT: andb $15, %al ; X32-NEXT: movl %eax, %ecx -; X32-NEXT: andb $15, %cl -; X32-NEXT: movl %ecx, %edx -; X32-NEXT: shrb $2, %dl -; X32-NEXT: kmovd %edx, %k2 +; X32-NEXT: shrb $2, %cl +; X32-NEXT: kmovd %ecx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $29, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $35, %k0, %k1 -; X32-NEXT: shrb $3, %cl -; X32-NEXT: kmovd %ecx, %k2 +; X32-NEXT: shrb $3, %al +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $28, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $36, %k0, %k1 -; X32-NEXT: movl %eax, %ecx -; X32-NEXT: shrb $4, %cl -; X32-NEXT: kmovd %ecx, %k2 +; X32-NEXT: movl %ebx, %eax +; X32-NEXT: shrb $4, %al +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $27, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $37, %k0, %k1 -; X32-NEXT: movl %eax, %ecx -; X32-NEXT: shrb $5, %cl -; X32-NEXT: andb $1, %cl -; X32-NEXT: kmovd %ecx, %k2 +; X32-NEXT: movl %ebx, %eax +; X32-NEXT: shrb $5, %al +; X32-NEXT: andb $1, %al +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $26, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $38, %k0, %k1 -; X32-NEXT: movl %eax, %ecx -; X32-NEXT: shrb $6, %cl -; X32-NEXT: kmovd %ecx, %k2 +; X32-NEXT: movl %ebx, %eax +; X32-NEXT: shrb $6, %al +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $25, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $39, %k0, %k1 -; X32-NEXT: movl %eax, %ecx -; X32-NEXT: shrb $7, %cl -; X32-NEXT: kmovd %ecx, %k2 +; X32-NEXT: movl %ebx, %eax +; X32-NEXT: shrb $7, %al +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $24, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $40, %k0, %k1 -; X32-NEXT: movb %ah, %cl -; X32-NEXT: kmovd %ecx, %k2 +; X32-NEXT: movb %bh, %al +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 -; X32-NEXT: andb $2, %cl -; X32-NEXT: shrb %cl -; X32-NEXT: kmovd %ecx, %k2 -; X32-NEXT: movb %ah, %cl -; X32-NEXT: andb $15, %cl -; X32-NEXT: movl %ecx, %edx -; X32-NEXT: shrb $2, %dl -; X32-NEXT: kmovd %edx, %k3 -; X32-NEXT: shrb $3, %cl -; X32-NEXT: kmovd %ecx, %k4 -; X32-NEXT: movl %eax, %ecx -; X32-NEXT: shrl $13, %ecx -; X32-NEXT: andb $1, %cl -; X32-NEXT: kmovd %ecx, %k5 +; X32-NEXT: andb $2, %al +; X32-NEXT: shrb %al +; X32-NEXT: kmovd %eax, %k2 +; X32-NEXT: movb %bh, %al +; X32-NEXT: andb $15, %al ; X32-NEXT: movl %eax, %ecx +; X32-NEXT: shrb $2, %cl +; X32-NEXT: kmovd %ecx, %k3 +; X32-NEXT: shrb $3, %al +; X32-NEXT: kmovd %eax, %k4 +; X32-NEXT: movl %ebx, %eax +; X32-NEXT: shrl $13, %eax +; X32-NEXT: andb $1, %al +; X32-NEXT: kmovd %eax, %k5 +; X32-NEXT: movl %ebx, %ecx ; X32-NEXT: shrl $16, %ecx -; X32-NEXT: movl %ecx, %edx -; X32-NEXT: andb $2, %dl -; X32-NEXT: shrb %dl -; X32-NEXT: kmovd %edx, %k6 +; X32-NEXT: movl %ecx, %eax +; X32-NEXT: andb $2, %al +; X32-NEXT: shrb %al +; X32-NEXT: kmovd %eax, %k6 ; X32-NEXT: movl %ecx, %edx ; X32-NEXT: andb $15, %dl -; X32-NEXT: movl %edx, %ebx -; X32-NEXT: shrb $2, %bl -; X32-NEXT: kmovd %ebx, %k7 +; X32-NEXT: movl %edx, %eax +; X32-NEXT: shrb $2, %al +; X32-NEXT: kmovd %eax, %k7 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $23, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 @@ -1031,23 +1024,20 @@ ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $42, %k0, %k1 ; X32-NEXT: kxorq %k3, %k1, %k1 -; X32-NEXT: kshiftlq $63, %k1, %k1 -; X32-NEXT: kshiftrq $21, %k1, %k1 -; X32-NEXT: kxorq %k0, %k1, %k0 -; X32-NEXT: kshiftrq $43, %k0, %k1 -; X32-NEXT: kxorq %k4, %k1, %k1 +; X32-NEXT: movzwl %bx, %eax ; X32-NEXT: movl %eax, %esi ; X32-NEXT: shrl $12, %esi -; X32-NEXT: andl $15, %esi ; X32-NEXT: kmovd %esi, %k2 ; X32-NEXT: movl %eax, %esi ; X32-NEXT: shrl $14, %esi -; X32-NEXT: andl $3, %esi ; X32-NEXT: kmovd %esi, %k3 -; X32-NEXT: movl %eax, %esi -; X32-NEXT: shrl $15, %esi -; X32-NEXT: andl $1, %esi -; X32-NEXT: kmovd %esi, %k4 +; X32-NEXT: kshiftlq $63, %k1, %k1 +; X32-NEXT: kshiftrq $21, %k1, %k1 +; X32-NEXT: kxorq %k0, %k1, %k0 +; X32-NEXT: kshiftrq $43, %k0, %k1 +; X32-NEXT: kxorq %k4, %k1, %k1 +; X32-NEXT: shrl $15, %eax +; X32-NEXT: kmovd %eax, %k4 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $20, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 @@ -1095,30 +1085,30 @@ ; X32-NEXT: kshiftrq $12, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k4 ; X32-NEXT: kshiftrq $52, %k4, %k0 -; X32-NEXT: movl %ecx, %edx -; X32-NEXT: shrb $4, %dl -; X32-NEXT: kmovd %edx, %k1 +; X32-NEXT: movl %ecx, %eax +; X32-NEXT: shrb $4, %al +; X32-NEXT: kmovd %eax, %k1 ; X32-NEXT: kxorq %k1, %k0, %k5 -; X32-NEXT: movl %ecx, %edx -; X32-NEXT: shrb $5, %dl -; X32-NEXT: andb $1, %dl -; X32-NEXT: kmovd %edx, %k6 -; X32-NEXT: movl %ecx, %edx -; X32-NEXT: shrb $6, %dl -; X32-NEXT: kmovd %edx, %k7 +; X32-NEXT: movl %ecx, %eax +; X32-NEXT: shrb $5, %al +; X32-NEXT: andb $1, %al +; X32-NEXT: kmovd %eax, %k6 +; X32-NEXT: movl %ecx, %eax +; X32-NEXT: shrb $6, %al +; X32-NEXT: kmovd %eax, %k7 ; X32-NEXT: shrb $7, %cl ; X32-NEXT: kmovd %ecx, %k0 -; X32-NEXT: movl %eax, %ecx +; X32-NEXT: movl %ebx, %ecx ; X32-NEXT: shrl $24, %ecx ; X32-NEXT: kmovd %ecx, %k1 -; X32-NEXT: movl %ecx, %edx -; X32-NEXT: andb $2, %dl -; X32-NEXT: shrb %dl -; X32-NEXT: kmovd %edx, %k2 +; X32-NEXT: movl %ecx, %eax +; X32-NEXT: andb $2, %al +; X32-NEXT: shrb %al +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: andb $15, %cl -; X32-NEXT: movl %ecx, %edx -; X32-NEXT: shrb $2, %dl -; X32-NEXT: kmovd %edx, %k3 +; X32-NEXT: movl %ecx, %eax +; X32-NEXT: shrb $2, %al +; X32-NEXT: kmovd %eax, %k3 ; X32-NEXT: kshiftlq $63, %k5, %k5 ; X32-NEXT: kshiftrq $11, %k5, %k5 ; X32-NEXT: kxorq %k4, %k5, %k4 @@ -1131,10 +1121,10 @@ ; X32-NEXT: kxorq %k7, %k4, %k6 ; X32-NEXT: shrb $3, %cl ; X32-NEXT: kmovd %ecx, %k4 -; X32-NEXT: movl %eax, %ecx -; X32-NEXT: shrl $29, %ecx -; X32-NEXT: andb $1, %cl -; X32-NEXT: kmovd %ecx, %k7 +; X32-NEXT: movl %ebx, %eax +; X32-NEXT: shrl $29, %eax +; X32-NEXT: andb $1, %al +; X32-NEXT: kmovd %eax, %k7 ; X32-NEXT: kshiftlq $63, %k6, %k6 ; X32-NEXT: kshiftrq $9, %k6, %k6 ; X32-NEXT: kxorq %k5, %k6, %k5 @@ -1145,12 +1135,12 @@ ; X32-NEXT: kxorq %k5, %k0, %k0 ; X32-NEXT: kshiftrq $56, %k0, %k5 ; X32-NEXT: kxorq %k1, %k5, %k1 -; X32-NEXT: movl %eax, %ecx -; X32-NEXT: shrl $28, %ecx -; X32-NEXT: kmovd %ecx, %k5 -; X32-NEXT: movl %eax, %ecx -; X32-NEXT: shrl $30, %ecx -; X32-NEXT: kmovd %ecx, %k6 +; X32-NEXT: movl %ebx, %eax +; X32-NEXT: shrl $28, %eax +; X32-NEXT: kmovd %eax, %k5 +; X32-NEXT: movl %ebx, %eax +; X32-NEXT: shrl $30, %eax +; X32-NEXT: kmovd %eax, %k6 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $7, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 @@ -1181,8 +1171,8 @@ ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $62, %k0, %k1 ; X32-NEXT: kxorq %k6, %k1, %k1 -; X32-NEXT: shrl $31, %eax -; X32-NEXT: kmovd %eax, %k2 +; X32-NEXT: shrl $31, %ebx +; X32-NEXT: kmovd %ebx, %k2 ; X32-NEXT: movb {{[0-9]+}}(%esp), %al ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $1, %k1, %k1 @@ -1701,21 +1691,21 @@ ; X32-NEXT: subl $8, %esp ; X32-NEXT: .cfi_offset %esi, -16 ; X32-NEXT: .cfi_offset %ebx, -12 -; X32-NEXT: movl 8(%ebp), %ecx -; X32-NEXT: kmovd %ecx, %k0 +; X32-NEXT: movl 8(%ebp), %eax +; X32-NEXT: kmovd %eax, %k0 ; X32-NEXT: kshiftrq $1, %k0, %k1 -; X32-NEXT: movl %ecx, %eax -; X32-NEXT: andb $2, %al -; X32-NEXT: shrb %al -; X32-NEXT: kmovd %eax, %k2 +; X32-NEXT: movl %eax, %ecx +; X32-NEXT: andb $2, %cl +; X32-NEXT: shrb %cl +; X32-NEXT: kmovd %ecx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $62, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $2, %k0, %k1 -; X32-NEXT: movl %ecx, %eax -; X32-NEXT: andb $15, %al -; X32-NEXT: movl %eax, %edx +; X32-NEXT: movl %eax, %ecx +; X32-NEXT: andb $15, %cl +; X32-NEXT: movl %ecx, %edx ; X32-NEXT: shrb $2, %dl ; X32-NEXT: kmovd %edx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 @@ -1723,127 +1713,124 @@ ; X32-NEXT: kshiftrq $61, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $3, %k0, %k1 -; X32-NEXT: shrb $3, %al -; X32-NEXT: kmovd %eax, %k2 +; X32-NEXT: shrb $3, %cl +; X32-NEXT: kmovd %ecx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $60, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $4, %k0, %k1 -; X32-NEXT: movl %ecx, %eax -; X32-NEXT: shrb $4, %al -; X32-NEXT: kmovd %eax, %k2 +; X32-NEXT: movl %eax, %ecx +; X32-NEXT: shrb $4, %cl +; X32-NEXT: kmovd %ecx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $59, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $5, %k0, %k1 -; X32-NEXT: movl %ecx, %eax -; X32-NEXT: shrb $5, %al -; X32-NEXT: andb $1, %al -; X32-NEXT: kmovd %eax, %k2 +; X32-NEXT: movl %eax, %ecx +; X32-NEXT: shrb $5, %cl +; X32-NEXT: andb $1, %cl +; X32-NEXT: kmovd %ecx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $58, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $6, %k0, %k1 -; X32-NEXT: movl %ecx, %eax -; X32-NEXT: shrb $6, %al -; X32-NEXT: kmovd %eax, %k2 +; X32-NEXT: movl %eax, %ecx +; X32-NEXT: shrb $6, %cl +; X32-NEXT: kmovd %ecx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $57, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $7, %k0, %k1 -; X32-NEXT: movl %ecx, %eax -; X32-NEXT: shrb $7, %al -; X32-NEXT: kmovd %eax, %k2 +; X32-NEXT: movl %eax, %ecx +; X32-NEXT: shrb $7, %cl +; X32-NEXT: kmovd %ecx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $56, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $8, %k0, %k1 -; X32-NEXT: movb %ch, %al -; X32-NEXT: kmovd %eax, %k2 +; X32-NEXT: movb %ah, %cl +; X32-NEXT: kmovd %ecx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 -; X32-NEXT: andb $2, %al -; X32-NEXT: shrb %al -; X32-NEXT: kmovd %eax, %k2 -; X32-NEXT: movb %ch, %al -; X32-NEXT: andb $15, %al -; X32-NEXT: movl %eax, %edx -; X32-NEXT: shrb $2, %dl -; X32-NEXT: kmovd %edx, %k3 -; X32-NEXT: shrb $3, %al -; X32-NEXT: kmovd %eax, %k4 -; X32-NEXT: movl %ecx, %eax -; X32-NEXT: shrl $13, %eax -; X32-NEXT: andb $1, %al -; X32-NEXT: kmovd %eax, %k5 -; X32-NEXT: movl %ecx, %edx -; X32-NEXT: shrl $16, %edx -; X32-NEXT: movl %edx, %eax -; X32-NEXT: andb $2, %al -; X32-NEXT: shrb %al -; X32-NEXT: kmovd %eax, %k6 -; X32-NEXT: movl %edx, %ebx -; X32-NEXT: andb $15, %bl -; X32-NEXT: movl %ebx, %eax -; X32-NEXT: shrb $2, %al -; X32-NEXT: kmovd %eax, %k7 -; X32-NEXT: movl 12(%ebp), %eax ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $55, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $9, %k0, %k1 +; X32-NEXT: andb $2, %cl +; X32-NEXT: shrb %cl +; X32-NEXT: kmovd %ecx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $54, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $10, %k0, %k1 -; X32-NEXT: kxorq %k3, %k1, %k1 +; X32-NEXT: movb %ah, %cl +; X32-NEXT: movl %eax, %esi +; X32-NEXT: andb $15, %cl +; X32-NEXT: movl %ecx, %edx +; X32-NEXT: shrb $2, %dl +; X32-NEXT: kmovd %edx, %k2 +; X32-NEXT: kxorq %k2, %k1, %k1 +; X32-NEXT: shrb $3, %cl +; X32-NEXT: kmovd %ecx, %k2 +; X32-NEXT: movzwl %si, %edx +; X32-NEXT: movl %edx, %ecx +; X32-NEXT: shrl $12, %ecx +; X32-NEXT: kmovd %ecx, %k3 +; X32-NEXT: movl %esi, %ecx +; X32-NEXT: shrl $13, %ecx +; X32-NEXT: andb $1, %cl +; X32-NEXT: kmovd %ecx, %k4 +; X32-NEXT: movl %edx, %ecx +; X32-NEXT: shrl $14, %ecx +; X32-NEXT: kmovd %ecx, %k5 +; X32-NEXT: shrl $16, %eax +; X32-NEXT: movl %eax, %ebx +; X32-NEXT: andb $2, %bl +; X32-NEXT: shrb %bl +; X32-NEXT: kmovd %ebx, %k6 +; X32-NEXT: movl %eax, %ecx +; X32-NEXT: andb $15, %cl +; X32-NEXT: movl %ecx, %ebx +; X32-NEXT: shrb $2, %bl +; X32-NEXT: kmovd %ebx, %k7 +; X32-NEXT: movl 12(%ebp), %ebx ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $53, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $11, %k0, %k1 -; X32-NEXT: kxorq %k4, %k1, %k1 +; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $52, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $12, %k0, %k1 -; X32-NEXT: movl %ecx, %esi -; X32-NEXT: shrl $12, %esi -; X32-NEXT: andl $15, %esi -; X32-NEXT: kmovd %esi, %k2 -; X32-NEXT: kxorq %k2, %k1, %k1 +; X32-NEXT: kxorq %k3, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $51, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $13, %k0, %k1 -; X32-NEXT: kxorq %k5, %k1, %k1 +; X32-NEXT: kxorq %k4, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $50, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $14, %k0, %k1 -; X32-NEXT: movl %ecx, %esi -; X32-NEXT: shrl $14, %esi -; X32-NEXT: andl $3, %esi -; X32-NEXT: kmovd %esi, %k2 -; X32-NEXT: kxorq %k2, %k1, %k1 +; X32-NEXT: kxorq %k5, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $49, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $15, %k0, %k1 -; X32-NEXT: movl %ecx, %esi -; X32-NEXT: shrl $15, %esi -; X32-NEXT: andl $1, %esi -; X32-NEXT: kmovd %esi, %k2 +; X32-NEXT: shrl $15, %edx +; X32-NEXT: kmovd %edx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $48, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $16, %k0, %k1 -; X32-NEXT: kmovd %edx, %k2 +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $47, %k1, %k1 @@ -1859,46 +1846,46 @@ ; X32-NEXT: kshiftrq $45, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $19, %k0, %k1 -; X32-NEXT: shrb $3, %bl -; X32-NEXT: kmovd %ebx, %k2 +; X32-NEXT: shrb $3, %cl +; X32-NEXT: kmovd %ecx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $44, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $20, %k0, %k1 -; X32-NEXT: movl %edx, %ebx -; X32-NEXT: shrb $4, %bl -; X32-NEXT: kmovd %ebx, %k2 +; X32-NEXT: movl %eax, %ecx +; X32-NEXT: shrb $4, %cl +; X32-NEXT: kmovd %ecx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $43, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $21, %k0, %k1 -; X32-NEXT: movl %edx, %ebx -; X32-NEXT: shrb $5, %bl -; X32-NEXT: andb $1, %bl -; X32-NEXT: kmovd %ebx, %k2 +; X32-NEXT: movl %eax, %ecx +; X32-NEXT: shrb $5, %cl +; X32-NEXT: andb $1, %cl +; X32-NEXT: kmovd %ecx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $42, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $22, %k0, %k1 -; X32-NEXT: movl %edx, %ebx -; X32-NEXT: shrb $6, %bl -; X32-NEXT: kmovd %ebx, %k2 +; X32-NEXT: movl %eax, %ecx +; X32-NEXT: shrb $6, %cl +; X32-NEXT: kmovd %ecx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $41, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $23, %k0, %k1 -; X32-NEXT: shrb $7, %dl -; X32-NEXT: kmovd %edx, %k2 +; X32-NEXT: shrb $7, %al +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $40, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $24, %k0, %k1 -; X32-NEXT: movl %ecx, %edx +; X32-NEXT: movl %esi, %edx ; X32-NEXT: shrl $24, %edx ; X32-NEXT: kmovd %edx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 @@ -1906,19 +1893,19 @@ ; X32-NEXT: kshiftrq $39, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $25, %k0, %k1 -; X32-NEXT: movl %edx, %ebx -; X32-NEXT: andb $2, %bl -; X32-NEXT: shrb %bl -; X32-NEXT: kmovd %ebx, %k2 +; X32-NEXT: movl %edx, %eax +; X32-NEXT: andb $2, %al +; X32-NEXT: shrb %al +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $38, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $26, %k0, %k1 ; X32-NEXT: andb $15, %dl -; X32-NEXT: movl %edx, %ebx -; X32-NEXT: shrb $2, %bl -; X32-NEXT: kmovd %ebx, %k2 +; X32-NEXT: movl %edx, %eax +; X32-NEXT: shrb $2, %al +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $37, %k1, %k1 @@ -1931,26 +1918,27 @@ ; X32-NEXT: kshiftrq $36, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $28, %k0, %k1 -; X32-NEXT: movl %ecx, %edx -; X32-NEXT: shrl $28, %edx -; X32-NEXT: kmovd %edx, %k2 +; X32-NEXT: movl %esi, %ecx +; X32-NEXT: movl %ecx, %eax +; X32-NEXT: shrl $28, %eax +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $35, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $29, %k0, %k1 -; X32-NEXT: movl %ecx, %edx -; X32-NEXT: shrl $29, %edx -; X32-NEXT: andb $1, %dl -; X32-NEXT: kmovd %edx, %k2 +; X32-NEXT: movl %ecx, %eax +; X32-NEXT: shrl $29, %eax +; X32-NEXT: andb $1, %al +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $34, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $30, %k0, %k1 -; X32-NEXT: movl %ecx, %edx -; X32-NEXT: shrl $30, %edx -; X32-NEXT: kmovd %edx, %k2 +; X32-NEXT: movl %ecx, %eax +; X32-NEXT: shrl $30, %eax +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $33, %k1, %k1 @@ -1963,99 +1951,99 @@ ; X32-NEXT: kshiftrq $32, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $32, %k0, %k1 -; X32-NEXT: kmovd %eax, %k2 +; X32-NEXT: kmovd %ebx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $31, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $33, %k0, %k1 -; X32-NEXT: movl %eax, %ecx -; X32-NEXT: andb $2, %cl -; X32-NEXT: shrb %cl -; X32-NEXT: kmovd %ecx, %k2 +; X32-NEXT: movl %ebx, %eax +; X32-NEXT: andb $2, %al +; X32-NEXT: shrb %al +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $30, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $34, %k0, %k1 +; X32-NEXT: movl %ebx, %eax +; X32-NEXT: andb $15, %al ; X32-NEXT: movl %eax, %ecx -; X32-NEXT: andb $15, %cl -; X32-NEXT: movl %ecx, %edx -; X32-NEXT: shrb $2, %dl -; X32-NEXT: kmovd %edx, %k2 +; X32-NEXT: shrb $2, %cl +; X32-NEXT: kmovd %ecx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $29, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $35, %k0, %k1 -; X32-NEXT: shrb $3, %cl -; X32-NEXT: kmovd %ecx, %k2 +; X32-NEXT: shrb $3, %al +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $28, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $36, %k0, %k1 -; X32-NEXT: movl %eax, %ecx -; X32-NEXT: shrb $4, %cl -; X32-NEXT: kmovd %ecx, %k2 +; X32-NEXT: movl %ebx, %eax +; X32-NEXT: shrb $4, %al +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $27, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $37, %k0, %k1 -; X32-NEXT: movl %eax, %ecx -; X32-NEXT: shrb $5, %cl -; X32-NEXT: andb $1, %cl -; X32-NEXT: kmovd %ecx, %k2 +; X32-NEXT: movl %ebx, %eax +; X32-NEXT: shrb $5, %al +; X32-NEXT: andb $1, %al +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $26, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $38, %k0, %k1 -; X32-NEXT: movl %eax, %ecx -; X32-NEXT: shrb $6, %cl -; X32-NEXT: kmovd %ecx, %k2 +; X32-NEXT: movl %ebx, %eax +; X32-NEXT: shrb $6, %al +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $25, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $39, %k0, %k1 -; X32-NEXT: movl %eax, %ecx -; X32-NEXT: shrb $7, %cl -; X32-NEXT: kmovd %ecx, %k2 +; X32-NEXT: movl %ebx, %eax +; X32-NEXT: shrb $7, %al +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $24, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $40, %k0, %k1 -; X32-NEXT: movb %ah, %cl -; X32-NEXT: kmovd %ecx, %k2 +; X32-NEXT: movb %bh, %al +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 -; X32-NEXT: andb $2, %cl -; X32-NEXT: shrb %cl -; X32-NEXT: kmovd %ecx, %k2 -; X32-NEXT: movb %ah, %cl -; X32-NEXT: andb $15, %cl -; X32-NEXT: movl %ecx, %edx -; X32-NEXT: shrb $2, %dl -; X32-NEXT: kmovd %edx, %k3 -; X32-NEXT: shrb $3, %cl -; X32-NEXT: kmovd %ecx, %k4 -; X32-NEXT: movl %eax, %ecx -; X32-NEXT: shrl $13, %ecx -; X32-NEXT: andb $1, %cl -; X32-NEXT: kmovd %ecx, %k5 +; X32-NEXT: andb $2, %al +; X32-NEXT: shrb %al +; X32-NEXT: kmovd %eax, %k2 +; X32-NEXT: movb %bh, %al +; X32-NEXT: andb $15, %al ; X32-NEXT: movl %eax, %ecx +; X32-NEXT: shrb $2, %cl +; X32-NEXT: kmovd %ecx, %k3 +; X32-NEXT: shrb $3, %al +; X32-NEXT: kmovd %eax, %k4 +; X32-NEXT: movl %ebx, %eax +; X32-NEXT: shrl $13, %eax +; X32-NEXT: andb $1, %al +; X32-NEXT: kmovd %eax, %k5 +; X32-NEXT: movl %ebx, %ecx ; X32-NEXT: shrl $16, %ecx -; X32-NEXT: movl %ecx, %edx -; X32-NEXT: andb $2, %dl -; X32-NEXT: shrb %dl -; X32-NEXT: kmovd %edx, %k6 +; X32-NEXT: movl %ecx, %eax +; X32-NEXT: andb $2, %al +; X32-NEXT: shrb %al +; X32-NEXT: kmovd %eax, %k6 ; X32-NEXT: movl %ecx, %edx ; X32-NEXT: andb $15, %dl -; X32-NEXT: movl %edx, %ebx -; X32-NEXT: shrb $2, %bl -; X32-NEXT: kmovd %ebx, %k7 +; X32-NEXT: movl %edx, %eax +; X32-NEXT: shrb $2, %al +; X32-NEXT: kmovd %eax, %k7 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $23, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 @@ -2075,9 +2063,9 @@ ; X32-NEXT: kshiftrq $20, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $44, %k0, %k1 +; X32-NEXT: movzwl %bx, %eax ; X32-NEXT: movl %eax, %esi ; X32-NEXT: shrl $12, %esi -; X32-NEXT: andl $15, %esi ; X32-NEXT: kmovd %esi, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 @@ -2091,17 +2079,14 @@ ; X32-NEXT: kshiftrq $46, %k0, %k1 ; X32-NEXT: movl %eax, %esi ; X32-NEXT: shrl $14, %esi -; X32-NEXT: andl $3, %esi ; X32-NEXT: kmovd %esi, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $17, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $47, %k0, %k1 -; X32-NEXT: movl %eax, %esi -; X32-NEXT: shrl $15, %esi -; X32-NEXT: andl $1, %esi -; X32-NEXT: kmovd %esi, %k2 +; X32-NEXT: shrl $15, %eax +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $16, %k1, %k1 @@ -2130,30 +2115,30 @@ ; X32-NEXT: kshiftrq $12, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $52, %k0, %k1 -; X32-NEXT: movl %ecx, %edx -; X32-NEXT: shrb $4, %dl -; X32-NEXT: kmovd %edx, %k2 +; X32-NEXT: movl %ecx, %eax +; X32-NEXT: shrb $4, %al +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 -; X32-NEXT: movl %ecx, %edx -; X32-NEXT: shrb $5, %dl -; X32-NEXT: andb $1, %dl -; X32-NEXT: kmovd %edx, %k2 -; X32-NEXT: movl %ecx, %edx -; X32-NEXT: shrb $6, %dl -; X32-NEXT: kmovd %edx, %k3 +; X32-NEXT: movl %ecx, %eax +; X32-NEXT: shrb $5, %al +; X32-NEXT: andb $1, %al +; X32-NEXT: kmovd %eax, %k2 +; X32-NEXT: movl %ecx, %eax +; X32-NEXT: shrb $6, %al +; X32-NEXT: kmovd %eax, %k3 ; X32-NEXT: shrb $7, %cl ; X32-NEXT: kmovd %ecx, %k4 -; X32-NEXT: movl %eax, %ecx +; X32-NEXT: movl %ebx, %ecx ; X32-NEXT: shrl $24, %ecx ; X32-NEXT: kmovd %ecx, %k5 -; X32-NEXT: movl %ecx, %edx -; X32-NEXT: andb $2, %dl -; X32-NEXT: shrb %dl -; X32-NEXT: kmovd %edx, %k6 +; X32-NEXT: movl %ecx, %eax +; X32-NEXT: andb $2, %al +; X32-NEXT: shrb %al +; X32-NEXT: kmovd %eax, %k6 ; X32-NEXT: andb $15, %cl -; X32-NEXT: movl %ecx, %edx -; X32-NEXT: shrb $2, %dl -; X32-NEXT: kmovd %edx, %k7 +; X32-NEXT: movl %ecx, %eax +; X32-NEXT: shrb $2, %al +; X32-NEXT: kmovd %eax, %k7 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $11, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 @@ -2195,34 +2180,34 @@ ; X32-NEXT: kshiftrq $4, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $60, %k0, %k1 -; X32-NEXT: movl %eax, %ecx -; X32-NEXT: shrl $28, %ecx -; X32-NEXT: kmovd %ecx, %k2 +; X32-NEXT: movl %ebx, %eax +; X32-NEXT: shrl $28, %eax +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $3, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $61, %k0, %k1 -; X32-NEXT: movl %eax, %ecx -; X32-NEXT: shrl $29, %ecx -; X32-NEXT: andb $1, %cl -; X32-NEXT: kmovd %ecx, %k2 +; X32-NEXT: movl %ebx, %eax +; X32-NEXT: shrl $29, %eax +; X32-NEXT: andb $1, %al +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $2, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $62, %k0, %k1 -; X32-NEXT: movl %eax, %ecx -; X32-NEXT: shrl $30, %ecx -; X32-NEXT: kmovd %ecx, %k2 +; X32-NEXT: movl %ebx, %eax +; X32-NEXT: shrl $30, %eax +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $1, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftlq $1, %k0, %k0 ; X32-NEXT: kshiftrq $1, %k0, %k0 -; X32-NEXT: shrl $31, %eax -; X32-NEXT: kmovd %eax, %k1 +; X32-NEXT: shrl $31, %ebx +; X32-NEXT: kmovd %ebx, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: korq %k1, %k0, %k1 ; X32-NEXT: vptestmb %zmm0, %zmm1, %k0 {%k1} @@ -2348,21 +2333,21 @@ ; X32-NEXT: subl $8, %esp ; X32-NEXT: .cfi_offset %esi, -16 ; X32-NEXT: .cfi_offset %ebx, -12 -; X32-NEXT: movl 8(%ebp), %ecx -; X32-NEXT: kmovd %ecx, %k0 +; X32-NEXT: movl 8(%ebp), %eax +; X32-NEXT: kmovd %eax, %k0 ; X32-NEXT: kshiftrq $1, %k0, %k1 -; X32-NEXT: movl %ecx, %eax -; X32-NEXT: andb $2, %al -; X32-NEXT: shrb %al -; X32-NEXT: kmovd %eax, %k2 +; X32-NEXT: movl %eax, %ecx +; X32-NEXT: andb $2, %cl +; X32-NEXT: shrb %cl +; X32-NEXT: kmovd %ecx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $62, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $2, %k0, %k1 -; X32-NEXT: movl %ecx, %eax -; X32-NEXT: andb $15, %al -; X32-NEXT: movl %eax, %edx +; X32-NEXT: movl %eax, %ecx +; X32-NEXT: andb $15, %cl +; X32-NEXT: movl %ecx, %edx ; X32-NEXT: shrb $2, %dl ; X32-NEXT: kmovd %edx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 @@ -2370,127 +2355,124 @@ ; X32-NEXT: kshiftrq $61, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $3, %k0, %k1 -; X32-NEXT: shrb $3, %al -; X32-NEXT: kmovd %eax, %k2 +; X32-NEXT: shrb $3, %cl +; X32-NEXT: kmovd %ecx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $60, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $4, %k0, %k1 -; X32-NEXT: movl %ecx, %eax -; X32-NEXT: shrb $4, %al -; X32-NEXT: kmovd %eax, %k2 +; X32-NEXT: movl %eax, %ecx +; X32-NEXT: shrb $4, %cl +; X32-NEXT: kmovd %ecx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $59, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $5, %k0, %k1 -; X32-NEXT: movl %ecx, %eax -; X32-NEXT: shrb $5, %al -; X32-NEXT: andb $1, %al -; X32-NEXT: kmovd %eax, %k2 +; X32-NEXT: movl %eax, %ecx +; X32-NEXT: shrb $5, %cl +; X32-NEXT: andb $1, %cl +; X32-NEXT: kmovd %ecx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $58, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $6, %k0, %k1 -; X32-NEXT: movl %ecx, %eax -; X32-NEXT: shrb $6, %al -; X32-NEXT: kmovd %eax, %k2 +; X32-NEXT: movl %eax, %ecx +; X32-NEXT: shrb $6, %cl +; X32-NEXT: kmovd %ecx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $57, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $7, %k0, %k1 -; X32-NEXT: movl %ecx, %eax -; X32-NEXT: shrb $7, %al -; X32-NEXT: kmovd %eax, %k2 +; X32-NEXT: movl %eax, %ecx +; X32-NEXT: shrb $7, %cl +; X32-NEXT: kmovd %ecx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $56, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $8, %k0, %k1 -; X32-NEXT: movb %ch, %al -; X32-NEXT: kmovd %eax, %k2 +; X32-NEXT: movb %ah, %cl +; X32-NEXT: kmovd %ecx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 -; X32-NEXT: andb $2, %al -; X32-NEXT: shrb %al -; X32-NEXT: kmovd %eax, %k2 -; X32-NEXT: movb %ch, %al -; X32-NEXT: andb $15, %al -; X32-NEXT: movl %eax, %edx -; X32-NEXT: shrb $2, %dl -; X32-NEXT: kmovd %edx, %k3 -; X32-NEXT: shrb $3, %al -; X32-NEXT: kmovd %eax, %k4 -; X32-NEXT: movl %ecx, %eax -; X32-NEXT: shrl $13, %eax -; X32-NEXT: andb $1, %al -; X32-NEXT: kmovd %eax, %k5 -; X32-NEXT: movl %ecx, %edx -; X32-NEXT: shrl $16, %edx -; X32-NEXT: movl %edx, %eax -; X32-NEXT: andb $2, %al -; X32-NEXT: shrb %al -; X32-NEXT: kmovd %eax, %k6 -; X32-NEXT: movl %edx, %ebx -; X32-NEXT: andb $15, %bl -; X32-NEXT: movl %ebx, %eax -; X32-NEXT: shrb $2, %al -; X32-NEXT: kmovd %eax, %k7 -; X32-NEXT: movl 12(%ebp), %eax ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $55, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $9, %k0, %k1 +; X32-NEXT: andb $2, %cl +; X32-NEXT: shrb %cl +; X32-NEXT: kmovd %ecx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $54, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $10, %k0, %k1 -; X32-NEXT: kxorq %k3, %k1, %k1 +; X32-NEXT: movb %ah, %cl +; X32-NEXT: movl %eax, %esi +; X32-NEXT: andb $15, %cl +; X32-NEXT: movl %ecx, %edx +; X32-NEXT: shrb $2, %dl +; X32-NEXT: kmovd %edx, %k2 +; X32-NEXT: kxorq %k2, %k1, %k1 +; X32-NEXT: shrb $3, %cl +; X32-NEXT: kmovd %ecx, %k2 +; X32-NEXT: movzwl %si, %edx +; X32-NEXT: movl %edx, %ecx +; X32-NEXT: shrl $12, %ecx +; X32-NEXT: kmovd %ecx, %k3 +; X32-NEXT: movl %esi, %ecx +; X32-NEXT: shrl $13, %ecx +; X32-NEXT: andb $1, %cl +; X32-NEXT: kmovd %ecx, %k4 +; X32-NEXT: movl %edx, %ecx +; X32-NEXT: shrl $14, %ecx +; X32-NEXT: kmovd %ecx, %k5 +; X32-NEXT: shrl $16, %eax +; X32-NEXT: movl %eax, %ebx +; X32-NEXT: andb $2, %bl +; X32-NEXT: shrb %bl +; X32-NEXT: kmovd %ebx, %k6 +; X32-NEXT: movl %eax, %ecx +; X32-NEXT: andb $15, %cl +; X32-NEXT: movl %ecx, %ebx +; X32-NEXT: shrb $2, %bl +; X32-NEXT: kmovd %ebx, %k7 +; X32-NEXT: movl 12(%ebp), %ebx ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $53, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $11, %k0, %k1 -; X32-NEXT: kxorq %k4, %k1, %k1 +; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $52, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $12, %k0, %k1 -; X32-NEXT: movl %ecx, %esi -; X32-NEXT: shrl $12, %esi -; X32-NEXT: andl $15, %esi -; X32-NEXT: kmovd %esi, %k2 -; X32-NEXT: kxorq %k2, %k1, %k1 +; X32-NEXT: kxorq %k3, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $51, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $13, %k0, %k1 -; X32-NEXT: kxorq %k5, %k1, %k1 +; X32-NEXT: kxorq %k4, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $50, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $14, %k0, %k1 -; X32-NEXT: movl %ecx, %esi -; X32-NEXT: shrl $14, %esi -; X32-NEXT: andl $3, %esi -; X32-NEXT: kmovd %esi, %k2 -; X32-NEXT: kxorq %k2, %k1, %k1 +; X32-NEXT: kxorq %k5, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $49, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $15, %k0, %k1 -; X32-NEXT: movl %ecx, %esi -; X32-NEXT: shrl $15, %esi -; X32-NEXT: andl $1, %esi -; X32-NEXT: kmovd %esi, %k2 +; X32-NEXT: shrl $15, %edx +; X32-NEXT: kmovd %edx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $48, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $16, %k0, %k1 -; X32-NEXT: kmovd %edx, %k2 +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $47, %k1, %k1 @@ -2506,46 +2488,46 @@ ; X32-NEXT: kshiftrq $45, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $19, %k0, %k1 -; X32-NEXT: shrb $3, %bl -; X32-NEXT: kmovd %ebx, %k2 +; X32-NEXT: shrb $3, %cl +; X32-NEXT: kmovd %ecx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $44, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $20, %k0, %k1 -; X32-NEXT: movl %edx, %ebx -; X32-NEXT: shrb $4, %bl -; X32-NEXT: kmovd %ebx, %k2 +; X32-NEXT: movl %eax, %ecx +; X32-NEXT: shrb $4, %cl +; X32-NEXT: kmovd %ecx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $43, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $21, %k0, %k1 -; X32-NEXT: movl %edx, %ebx -; X32-NEXT: shrb $5, %bl -; X32-NEXT: andb $1, %bl -; X32-NEXT: kmovd %ebx, %k2 +; X32-NEXT: movl %eax, %ecx +; X32-NEXT: shrb $5, %cl +; X32-NEXT: andb $1, %cl +; X32-NEXT: kmovd %ecx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $42, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $22, %k0, %k1 -; X32-NEXT: movl %edx, %ebx -; X32-NEXT: shrb $6, %bl -; X32-NEXT: kmovd %ebx, %k2 +; X32-NEXT: movl %eax, %ecx +; X32-NEXT: shrb $6, %cl +; X32-NEXT: kmovd %ecx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $41, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $23, %k0, %k1 -; X32-NEXT: shrb $7, %dl -; X32-NEXT: kmovd %edx, %k2 +; X32-NEXT: shrb $7, %al +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $40, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $24, %k0, %k1 -; X32-NEXT: movl %ecx, %edx +; X32-NEXT: movl %esi, %edx ; X32-NEXT: shrl $24, %edx ; X32-NEXT: kmovd %edx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 @@ -2553,19 +2535,19 @@ ; X32-NEXT: kshiftrq $39, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $25, %k0, %k1 -; X32-NEXT: movl %edx, %ebx -; X32-NEXT: andb $2, %bl -; X32-NEXT: shrb %bl -; X32-NEXT: kmovd %ebx, %k2 +; X32-NEXT: movl %edx, %eax +; X32-NEXT: andb $2, %al +; X32-NEXT: shrb %al +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $38, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $26, %k0, %k1 ; X32-NEXT: andb $15, %dl -; X32-NEXT: movl %edx, %ebx -; X32-NEXT: shrb $2, %bl -; X32-NEXT: kmovd %ebx, %k2 +; X32-NEXT: movl %edx, %eax +; X32-NEXT: shrb $2, %al +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $37, %k1, %k1 @@ -2578,26 +2560,27 @@ ; X32-NEXT: kshiftrq $36, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $28, %k0, %k1 -; X32-NEXT: movl %ecx, %edx -; X32-NEXT: shrl $28, %edx -; X32-NEXT: kmovd %edx, %k2 +; X32-NEXT: movl %esi, %ecx +; X32-NEXT: movl %ecx, %eax +; X32-NEXT: shrl $28, %eax +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $35, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $29, %k0, %k1 -; X32-NEXT: movl %ecx, %edx -; X32-NEXT: shrl $29, %edx -; X32-NEXT: andb $1, %dl -; X32-NEXT: kmovd %edx, %k2 +; X32-NEXT: movl %ecx, %eax +; X32-NEXT: shrl $29, %eax +; X32-NEXT: andb $1, %al +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $34, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $30, %k0, %k1 -; X32-NEXT: movl %ecx, %edx -; X32-NEXT: shrl $30, %edx -; X32-NEXT: kmovd %edx, %k2 +; X32-NEXT: movl %ecx, %eax +; X32-NEXT: shrl $30, %eax +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $33, %k1, %k1 @@ -2610,99 +2593,99 @@ ; X32-NEXT: kshiftrq $32, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $32, %k0, %k1 -; X32-NEXT: kmovd %eax, %k2 +; X32-NEXT: kmovd %ebx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $31, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $33, %k0, %k1 -; X32-NEXT: movl %eax, %ecx -; X32-NEXT: andb $2, %cl -; X32-NEXT: shrb %cl -; X32-NEXT: kmovd %ecx, %k2 +; X32-NEXT: movl %ebx, %eax +; X32-NEXT: andb $2, %al +; X32-NEXT: shrb %al +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $30, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $34, %k0, %k1 +; X32-NEXT: movl %ebx, %eax +; X32-NEXT: andb $15, %al ; X32-NEXT: movl %eax, %ecx -; X32-NEXT: andb $15, %cl -; X32-NEXT: movl %ecx, %edx -; X32-NEXT: shrb $2, %dl -; X32-NEXT: kmovd %edx, %k2 +; X32-NEXT: shrb $2, %cl +; X32-NEXT: kmovd %ecx, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $29, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $35, %k0, %k1 -; X32-NEXT: shrb $3, %cl -; X32-NEXT: kmovd %ecx, %k2 +; X32-NEXT: shrb $3, %al +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $28, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $36, %k0, %k1 -; X32-NEXT: movl %eax, %ecx -; X32-NEXT: shrb $4, %cl -; X32-NEXT: kmovd %ecx, %k2 +; X32-NEXT: movl %ebx, %eax +; X32-NEXT: shrb $4, %al +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $27, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $37, %k0, %k1 -; X32-NEXT: movl %eax, %ecx -; X32-NEXT: shrb $5, %cl -; X32-NEXT: andb $1, %cl -; X32-NEXT: kmovd %ecx, %k2 +; X32-NEXT: movl %ebx, %eax +; X32-NEXT: shrb $5, %al +; X32-NEXT: andb $1, %al +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $26, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $38, %k0, %k1 -; X32-NEXT: movl %eax, %ecx -; X32-NEXT: shrb $6, %cl -; X32-NEXT: kmovd %ecx, %k2 +; X32-NEXT: movl %ebx, %eax +; X32-NEXT: shrb $6, %al +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $25, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $39, %k0, %k1 -; X32-NEXT: movl %eax, %ecx -; X32-NEXT: shrb $7, %cl -; X32-NEXT: kmovd %ecx, %k2 +; X32-NEXT: movl %ebx, %eax +; X32-NEXT: shrb $7, %al +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $24, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $40, %k0, %k1 -; X32-NEXT: movb %ah, %cl -; X32-NEXT: kmovd %ecx, %k2 +; X32-NEXT: movb %bh, %al +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 -; X32-NEXT: andb $2, %cl -; X32-NEXT: shrb %cl -; X32-NEXT: kmovd %ecx, %k2 -; X32-NEXT: movb %ah, %cl -; X32-NEXT: andb $15, %cl -; X32-NEXT: movl %ecx, %edx -; X32-NEXT: shrb $2, %dl -; X32-NEXT: kmovd %edx, %k3 -; X32-NEXT: shrb $3, %cl -; X32-NEXT: kmovd %ecx, %k4 -; X32-NEXT: movl %eax, %ecx -; X32-NEXT: shrl $13, %ecx -; X32-NEXT: andb $1, %cl -; X32-NEXT: kmovd %ecx, %k5 +; X32-NEXT: andb $2, %al +; X32-NEXT: shrb %al +; X32-NEXT: kmovd %eax, %k2 +; X32-NEXT: movb %bh, %al +; X32-NEXT: andb $15, %al ; X32-NEXT: movl %eax, %ecx +; X32-NEXT: shrb $2, %cl +; X32-NEXT: kmovd %ecx, %k3 +; X32-NEXT: shrb $3, %al +; X32-NEXT: kmovd %eax, %k4 +; X32-NEXT: movl %ebx, %eax +; X32-NEXT: shrl $13, %eax +; X32-NEXT: andb $1, %al +; X32-NEXT: kmovd %eax, %k5 +; X32-NEXT: movl %ebx, %ecx ; X32-NEXT: shrl $16, %ecx -; X32-NEXT: movl %ecx, %edx -; X32-NEXT: andb $2, %dl -; X32-NEXT: shrb %dl -; X32-NEXT: kmovd %edx, %k6 +; X32-NEXT: movl %ecx, %eax +; X32-NEXT: andb $2, %al +; X32-NEXT: shrb %al +; X32-NEXT: kmovd %eax, %k6 ; X32-NEXT: movl %ecx, %edx ; X32-NEXT: andb $15, %dl -; X32-NEXT: movl %edx, %ebx -; X32-NEXT: shrb $2, %bl -; X32-NEXT: kmovd %ebx, %k7 +; X32-NEXT: movl %edx, %eax +; X32-NEXT: shrb $2, %al +; X32-NEXT: kmovd %eax, %k7 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $23, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 @@ -2722,9 +2705,9 @@ ; X32-NEXT: kshiftrq $20, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $44, %k0, %k1 +; X32-NEXT: movzwl %bx, %eax ; X32-NEXT: movl %eax, %esi ; X32-NEXT: shrl $12, %esi -; X32-NEXT: andl $15, %esi ; X32-NEXT: kmovd %esi, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 @@ -2738,17 +2721,14 @@ ; X32-NEXT: kshiftrq $46, %k0, %k1 ; X32-NEXT: movl %eax, %esi ; X32-NEXT: shrl $14, %esi -; X32-NEXT: andl $3, %esi ; X32-NEXT: kmovd %esi, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $17, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $47, %k0, %k1 -; X32-NEXT: movl %eax, %esi -; X32-NEXT: shrl $15, %esi -; X32-NEXT: andl $1, %esi -; X32-NEXT: kmovd %esi, %k2 +; X32-NEXT: shrl $15, %eax +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $16, %k1, %k1 @@ -2777,30 +2757,30 @@ ; X32-NEXT: kshiftrq $12, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $52, %k0, %k1 -; X32-NEXT: movl %ecx, %edx -; X32-NEXT: shrb $4, %dl -; X32-NEXT: kmovd %edx, %k2 +; X32-NEXT: movl %ecx, %eax +; X32-NEXT: shrb $4, %al +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 -; X32-NEXT: movl %ecx, %edx -; X32-NEXT: shrb $5, %dl -; X32-NEXT: andb $1, %dl -; X32-NEXT: kmovd %edx, %k2 -; X32-NEXT: movl %ecx, %edx -; X32-NEXT: shrb $6, %dl -; X32-NEXT: kmovd %edx, %k3 +; X32-NEXT: movl %ecx, %eax +; X32-NEXT: shrb $5, %al +; X32-NEXT: andb $1, %al +; X32-NEXT: kmovd %eax, %k2 +; X32-NEXT: movl %ecx, %eax +; X32-NEXT: shrb $6, %al +; X32-NEXT: kmovd %eax, %k3 ; X32-NEXT: shrb $7, %cl ; X32-NEXT: kmovd %ecx, %k4 -; X32-NEXT: movl %eax, %ecx +; X32-NEXT: movl %ebx, %ecx ; X32-NEXT: shrl $24, %ecx ; X32-NEXT: kmovd %ecx, %k5 -; X32-NEXT: movl %ecx, %edx -; X32-NEXT: andb $2, %dl -; X32-NEXT: shrb %dl -; X32-NEXT: kmovd %edx, %k6 +; X32-NEXT: movl %ecx, %eax +; X32-NEXT: andb $2, %al +; X32-NEXT: shrb %al +; X32-NEXT: kmovd %eax, %k6 ; X32-NEXT: andb $15, %cl -; X32-NEXT: movl %ecx, %edx -; X32-NEXT: shrb $2, %dl -; X32-NEXT: kmovd %edx, %k7 +; X32-NEXT: movl %ecx, %eax +; X32-NEXT: shrb $2, %al +; X32-NEXT: kmovd %eax, %k7 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $11, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 @@ -2842,34 +2822,34 @@ ; X32-NEXT: kshiftrq $4, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $60, %k0, %k1 -; X32-NEXT: movl %eax, %ecx -; X32-NEXT: shrl $28, %ecx -; X32-NEXT: kmovd %ecx, %k2 +; X32-NEXT: movl %ebx, %eax +; X32-NEXT: shrl $28, %eax +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $3, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $61, %k0, %k1 -; X32-NEXT: movl %eax, %ecx -; X32-NEXT: shrl $29, %ecx -; X32-NEXT: andb $1, %cl -; X32-NEXT: kmovd %ecx, %k2 +; X32-NEXT: movl %ebx, %eax +; X32-NEXT: shrl $29, %eax +; X32-NEXT: andb $1, %al +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $2, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftrq $62, %k0, %k1 -; X32-NEXT: movl %eax, %ecx -; X32-NEXT: shrl $30, %ecx -; X32-NEXT: kmovd %ecx, %k2 +; X32-NEXT: movl %ebx, %eax +; X32-NEXT: shrl $30, %eax +; X32-NEXT: kmovd %eax, %k2 ; X32-NEXT: kxorq %k2, %k1, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: kshiftrq $1, %k1, %k1 ; X32-NEXT: kxorq %k0, %k1, %k0 ; X32-NEXT: kshiftlq $1, %k0, %k0 ; X32-NEXT: kshiftrq $1, %k0, %k0 -; X32-NEXT: shrl $31, %eax -; X32-NEXT: kmovd %eax, %k1 +; X32-NEXT: shrl $31, %ebx +; X32-NEXT: kmovd %ebx, %k1 ; X32-NEXT: kshiftlq $63, %k1, %k1 ; X32-NEXT: korq %k1, %k0, %k1 ; X32-NEXT: vptestnmb %zmm0, %zmm1, %k0 {%k1} Index: test/CodeGen/X86/avx512bw-intrinsics-upgrade.ll =================================================================== --- test/CodeGen/X86/avx512bw-intrinsics-upgrade.ll +++ test/CodeGen/X86/avx512bw-intrinsics-upgrade.ll @@ -1797,14 +1797,20 @@ ; ; AVX512F-32-LABEL: test_mask_cmp_b_512: ; AVX512F-32: # %bb.0: -; AVX512F-32-NEXT: pushl %ebx +; AVX512F-32-NEXT: pushl %ebp ; AVX512F-32-NEXT: .cfi_def_cfa_offset 8 -; AVX512F-32-NEXT: pushl %esi +; AVX512F-32-NEXT: pushl %ebx ; AVX512F-32-NEXT: .cfi_def_cfa_offset 12 +; AVX512F-32-NEXT: pushl %edi +; AVX512F-32-NEXT: .cfi_def_cfa_offset 16 +; AVX512F-32-NEXT: pushl %esi +; AVX512F-32-NEXT: .cfi_def_cfa_offset 20 ; AVX512F-32-NEXT: subl $60, %esp -; AVX512F-32-NEXT: .cfi_def_cfa_offset 72 -; AVX512F-32-NEXT: .cfi_offset %esi, -12 -; AVX512F-32-NEXT: .cfi_offset %ebx, -8 +; AVX512F-32-NEXT: .cfi_def_cfa_offset 80 +; AVX512F-32-NEXT: .cfi_offset %esi, -20 +; AVX512F-32-NEXT: .cfi_offset %edi, -16 +; AVX512F-32-NEXT: .cfi_offset %ebx, -12 +; AVX512F-32-NEXT: .cfi_offset %ebp, -8 ; AVX512F-32-NEXT: movl {{[0-9]+}}(%esp), %ebx ; AVX512F-32-NEXT: movl %ebx, %eax ; AVX512F-32-NEXT: shrl $16, %eax @@ -1844,7 +1850,7 @@ ; AVX512F-32-NEXT: kmovd %ecx, %k5 ; AVX512F-32-NEXT: movb %bh, %cl ; AVX512F-32-NEXT: kmovd %ecx, %k1 -; AVX512F-32-NEXT: movl %ebx, %esi +; AVX512F-32-NEXT: movl %ebx, %ebp ; AVX512F-32-NEXT: andb $2, %cl ; AVX512F-32-NEXT: shrb %cl ; AVX512F-32-NEXT: kshiftlq $63, %k2, %k2 @@ -1870,7 +1876,7 @@ ; AVX512F-32-NEXT: kshiftrq $5, %k7, %k0 ; AVX512F-32-NEXT: kxorq %k4, %k0, %k4 ; AVX512F-32-NEXT: kmovd %ecx, %k0 -; AVX512F-32-NEXT: movl %esi, %ecx +; AVX512F-32-NEXT: movl %ebp, %ecx ; AVX512F-32-NEXT: shrl $13, %ecx ; AVX512F-32-NEXT: andb $1, %cl ; AVX512F-32-NEXT: kshiftlq $63, %k4, %k4 @@ -1915,9 +1921,10 @@ ; AVX512F-32-NEXT: movl %eax, %ecx ; AVX512F-32-NEXT: shrb $5, %cl ; AVX512F-32-NEXT: andb $1, %cl -; AVX512F-32-NEXT: movl %esi, %edx +; AVX512F-32-NEXT: movzwl %bp, %edx +; AVX512F-32-NEXT: movl %edx, %esi +; AVX512F-32-NEXT: movl %edx, %edi ; AVX512F-32-NEXT: shrl $12, %edx -; AVX512F-32-NEXT: andl $15, %edx ; AVX512F-32-NEXT: kshiftlq $63, %k3, %k3 ; AVX512F-32-NEXT: kshiftrq $53, %k3, %k3 ; AVX512F-32-NEXT: kxorq %k7, %k3, %k3 @@ -1932,28 +1939,24 @@ ; AVX512F-32-NEXT: kmovd %ecx, %k0 ; AVX512F-32-NEXT: movl %eax, %edx ; AVX512F-32-NEXT: shrb $6, %dl -; AVX512F-32-NEXT: movl %esi, %ecx -; AVX512F-32-NEXT: shrl $14, %ecx -; AVX512F-32-NEXT: andl $3, %ecx +; AVX512F-32-NEXT: shrl $15, %esi +; AVX512F-32-NEXT: shrl $14, %edi ; AVX512F-32-NEXT: kshiftlq $63, %k7, %k7 ; AVX512F-32-NEXT: kshiftrq $51, %k7, %k7 ; AVX512F-32-NEXT: kxorq %k3, %k7, %k3 ; AVX512F-32-NEXT: kshiftrq $13, %k3, %k7 ; AVX512F-32-NEXT: kxorq %k4, %k7, %k4 -; AVX512F-32-NEXT: kmovd %ecx, %k7 -; AVX512F-32-NEXT: movl %esi, %ecx -; AVX512F-32-NEXT: shrl $15, %ecx -; AVX512F-32-NEXT: andl $1, %ecx ; AVX512F-32-NEXT: kshiftlq $63, %k4, %k4 ; AVX512F-32-NEXT: kshiftrq $50, %k4, %k4 ; AVX512F-32-NEXT: kxorq %k3, %k4, %k3 ; AVX512F-32-NEXT: kshiftrq $14, %k3, %k4 +; AVX512F-32-NEXT: kmovd %edi, %k7 ; AVX512F-32-NEXT: kxorq %k7, %k4, %k4 ; AVX512F-32-NEXT: kshiftlq $63, %k4, %k4 ; AVX512F-32-NEXT: kshiftrq $49, %k4, %k4 ; AVX512F-32-NEXT: kxorq %k3, %k4, %k3 ; AVX512F-32-NEXT: kshiftrq $15, %k3, %k4 -; AVX512F-32-NEXT: kmovd %ecx, %k7 +; AVX512F-32-NEXT: kmovd %esi, %k7 ; AVX512F-32-NEXT: kxorq %k7, %k4, %k4 ; AVX512F-32-NEXT: kshiftlq $63, %k4, %k4 ; AVX512F-32-NEXT: kshiftrq $48, %k4, %k4 @@ -1962,7 +1965,7 @@ ; AVX512F-32-NEXT: kmovd %eax, %k7 ; AVX512F-32-NEXT: kxorq %k7, %k4, %k4 ; AVX512F-32-NEXT: kmovd %edx, %k7 -; AVX512F-32-NEXT: movl %esi, %edx +; AVX512F-32-NEXT: movl %ebp, %edx ; AVX512F-32-NEXT: shrl $24, %edx ; AVX512F-32-NEXT: # kill: def %al killed %al killed %eax def %eax ; AVX512F-32-NEXT: shrb $7, %al @@ -2002,19 +2005,19 @@ ; AVX512F-32-NEXT: shrb $3, %cl ; AVX512F-32-NEXT: kshiftlq $63, %k6, %k6 ; AVX512F-32-NEXT: kshiftrq $43, %k6, %k6 -; AVX512F-32-NEXT: kxorq %k1, %k6, %k1 -; AVX512F-32-NEXT: kshiftrq $21, %k1, %k6 -; AVX512F-32-NEXT: kxorq %k0, %k6, %k6 -; AVX512F-32-NEXT: kmovd %ecx, %k0 -; AVX512F-32-NEXT: movl %esi, %ecx -; AVX512F-32-NEXT: shrl $29, %ecx -; AVX512F-32-NEXT: andb $1, %cl -; AVX512F-32-NEXT: kshiftlq $63, %k6, %k6 -; AVX512F-32-NEXT: kshiftrq $42, %k6, %k6 ; AVX512F-32-NEXT: kxorq %k1, %k6, %k6 -; AVX512F-32-NEXT: kshiftrq $22, %k6, %k1 -; AVX512F-32-NEXT: kxorq %k7, %k1, %k7 +; AVX512F-32-NEXT: kshiftrq $21, %k6, %k1 +; AVX512F-32-NEXT: kxorq %k0, %k1, %k0 ; AVX512F-32-NEXT: kmovd %ecx, %k1 +; AVX512F-32-NEXT: movl %ebp, %ecx +; AVX512F-32-NEXT: shrl $29, %ecx +; AVX512F-32-NEXT: andb $1, %cl +; AVX512F-32-NEXT: kshiftlq $63, %k0, %k0 +; AVX512F-32-NEXT: kshiftrq $42, %k0, %k0 +; AVX512F-32-NEXT: kxorq %k6, %k0, %k6 +; AVX512F-32-NEXT: kshiftrq $22, %k6, %k0 +; AVX512F-32-NEXT: kxorq %k7, %k0, %k7 +; AVX512F-32-NEXT: kmovd %ecx, %k0 ; AVX512F-32-NEXT: movl %ebx, %edx ; AVX512F-32-NEXT: andb $2, %dl ; AVX512F-32-NEXT: shrb %dl @@ -2045,68 +2048,68 @@ ; AVX512F-32-NEXT: shrb $4, %cl ; AVX512F-32-NEXT: kshiftlq $63, %k4, %k4 ; AVX512F-32-NEXT: kshiftrq $38, %k4, %k4 -; AVX512F-32-NEXT: kxorq %k7, %k4, %k7 -; AVX512F-32-NEXT: kshiftrq $26, %k7, %k4 -; AVX512F-32-NEXT: kxorq %k2, %k4, %k2 -; AVX512F-32-NEXT: kmovd %ecx, %k4 +; AVX512F-32-NEXT: kxorq %k7, %k4, %k4 +; AVX512F-32-NEXT: kshiftrq $26, %k4, %k7 +; AVX512F-32-NEXT: kxorq %k2, %k7, %k7 +; AVX512F-32-NEXT: kmovd %ecx, %k2 ; AVX512F-32-NEXT: movl %ebx, %ecx ; AVX512F-32-NEXT: shrb $5, %cl ; AVX512F-32-NEXT: andb $1, %cl -; AVX512F-32-NEXT: movl %esi, %edx +; AVX512F-32-NEXT: movl %ebp, %edx ; AVX512F-32-NEXT: shrl $28, %edx -; AVX512F-32-NEXT: kshiftlq $63, %k2, %k2 -; AVX512F-32-NEXT: kshiftrq $37, %k2, %k2 -; AVX512F-32-NEXT: kxorq %k7, %k2, %k2 -; AVX512F-32-NEXT: kshiftrq $27, %k2, %k7 -; AVX512F-32-NEXT: kxorq %k0, %k7, %k0 -; AVX512F-32-NEXT: kshiftlq $63, %k0, %k0 -; AVX512F-32-NEXT: kshiftrq $36, %k0, %k0 -; AVX512F-32-NEXT: kxorq %k2, %k0, %k2 -; AVX512F-32-NEXT: kshiftrq $28, %k2, %k0 +; AVX512F-32-NEXT: kshiftlq $63, %k7, %k7 +; AVX512F-32-NEXT: kshiftrq $37, %k7, %k7 +; AVX512F-32-NEXT: kxorq %k4, %k7, %k4 +; AVX512F-32-NEXT: kshiftrq $27, %k4, %k7 +; AVX512F-32-NEXT: kxorq %k1, %k7, %k1 +; AVX512F-32-NEXT: kshiftlq $63, %k1, %k1 +; AVX512F-32-NEXT: kshiftrq $36, %k1, %k1 +; AVX512F-32-NEXT: kxorq %k4, %k1, %k1 +; AVX512F-32-NEXT: kshiftrq $28, %k1, %k4 ; AVX512F-32-NEXT: kmovd %edx, %k7 -; AVX512F-32-NEXT: kxorq %k7, %k0, %k7 -; AVX512F-32-NEXT: kmovd %ecx, %k0 +; AVX512F-32-NEXT: kxorq %k7, %k4, %k7 +; AVX512F-32-NEXT: kmovd %ecx, %k4 ; AVX512F-32-NEXT: movl %ebx, %edx ; AVX512F-32-NEXT: shrb $6, %dl -; AVX512F-32-NEXT: movl %esi, %ecx -; AVX512F-32-NEXT: shrl $30, %ecx +; AVX512F-32-NEXT: movl %ebp, %ecx +; AVX512F-32-NEXT: shrl $31, %ecx +; AVX512F-32-NEXT: movl %ebp, %esi +; AVX512F-32-NEXT: shrl $30, %esi ; AVX512F-32-NEXT: kshiftlq $63, %k7, %k7 ; AVX512F-32-NEXT: kshiftrq $35, %k7, %k7 -; AVX512F-32-NEXT: kxorq %k2, %k7, %k2 -; AVX512F-32-NEXT: kshiftrq $29, %k2, %k7 ; AVX512F-32-NEXT: kxorq %k1, %k7, %k1 -; AVX512F-32-NEXT: kmovd %ecx, %k7 -; AVX512F-32-NEXT: movl %esi, %ecx -; AVX512F-32-NEXT: shrl $31, %ecx +; AVX512F-32-NEXT: kshiftrq $29, %k1, %k7 +; AVX512F-32-NEXT: kxorq %k0, %k7, %k0 +; AVX512F-32-NEXT: kshiftlq $63, %k0, %k0 +; AVX512F-32-NEXT: kshiftrq $34, %k0, %k0 +; AVX512F-32-NEXT: kxorq %k1, %k0, %k0 +; AVX512F-32-NEXT: kshiftrq $30, %k0, %k1 +; AVX512F-32-NEXT: kmovd %esi, %k7 +; AVX512F-32-NEXT: kxorq %k7, %k1, %k1 ; AVX512F-32-NEXT: kshiftlq $63, %k1, %k1 -; AVX512F-32-NEXT: kshiftrq $34, %k1, %k1 -; AVX512F-32-NEXT: kxorq %k2, %k1, %k1 -; AVX512F-32-NEXT: kshiftrq $30, %k1, %k2 -; AVX512F-32-NEXT: kxorq %k7, %k2, %k2 -; AVX512F-32-NEXT: kshiftlq $63, %k2, %k2 -; AVX512F-32-NEXT: kshiftrq $33, %k2, %k2 -; AVX512F-32-NEXT: kxorq %k1, %k2, %k1 -; AVX512F-32-NEXT: kshiftrq $31, %k1, %k2 +; AVX512F-32-NEXT: kshiftrq $33, %k1, %k1 +; AVX512F-32-NEXT: kxorq %k0, %k1, %k0 +; AVX512F-32-NEXT: kshiftrq $31, %k0, %k1 ; AVX512F-32-NEXT: kmovd %ecx, %k7 -; AVX512F-32-NEXT: kxorq %k7, %k2, %k2 -; AVX512F-32-NEXT: kshiftlq $63, %k2, %k2 -; AVX512F-32-NEXT: kshiftrq $32, %k2, %k2 -; AVX512F-32-NEXT: kxorq %k1, %k2, %k1 -; AVX512F-32-NEXT: kshiftrq $32, %k1, %k2 +; AVX512F-32-NEXT: kxorq %k7, %k1, %k1 +; AVX512F-32-NEXT: kshiftlq $63, %k1, %k1 +; AVX512F-32-NEXT: kshiftrq $32, %k1, %k1 +; AVX512F-32-NEXT: kxorq %k0, %k1, %k0 +; AVX512F-32-NEXT: kshiftrq $32, %k0, %k1 ; AVX512F-32-NEXT: kmovd %ebx, %k7 -; AVX512F-32-NEXT: kxorq %k7, %k2, %k2 +; AVX512F-32-NEXT: kxorq %k7, %k1, %k1 ; AVX512F-32-NEXT: kmovd %edx, %k7 ; AVX512F-32-NEXT: movl %ebx, %ecx ; AVX512F-32-NEXT: shrb $7, %cl -; AVX512F-32-NEXT: kshiftlq $63, %k2, %k2 -; AVX512F-32-NEXT: kshiftrq $31, %k2, %k2 -; AVX512F-32-NEXT: kxorq %k1, %k2, %k1 -; AVX512F-32-NEXT: kshiftrq $33, %k1, %k2 -; AVX512F-32-NEXT: kxorq %k5, %k2, %k2 -; AVX512F-32-NEXT: kshiftlq $63, %k2, %k2 -; AVX512F-32-NEXT: kshiftrq $30, %k2, %k2 -; AVX512F-32-NEXT: kxorq %k1, %k2, %k2 -; AVX512F-32-NEXT: kshiftrq $34, %k2, %k1 +; AVX512F-32-NEXT: kshiftlq $63, %k1, %k1 +; AVX512F-32-NEXT: kshiftrq $31, %k1, %k1 +; AVX512F-32-NEXT: kxorq %k0, %k1, %k0 +; AVX512F-32-NEXT: kshiftrq $33, %k0, %k1 +; AVX512F-32-NEXT: kxorq %k5, %k1, %k1 +; AVX512F-32-NEXT: kshiftlq $63, %k1, %k1 +; AVX512F-32-NEXT: kshiftrq $30, %k1, %k1 +; AVX512F-32-NEXT: kxorq %k0, %k1, %k0 +; AVX512F-32-NEXT: kshiftrq $34, %k0, %k1 ; AVX512F-32-NEXT: kxorq %k6, %k1, %k5 ; AVX512F-32-NEXT: kmovd %ecx, %k6 ; AVX512F-32-NEXT: movb %bh, %cl @@ -2115,34 +2118,34 @@ ; AVX512F-32-NEXT: shrb %cl ; AVX512F-32-NEXT: kshiftlq $63, %k5, %k5 ; AVX512F-32-NEXT: kshiftrq $29, %k5, %k5 -; AVX512F-32-NEXT: kxorq %k2, %k5, %k5 -; AVX512F-32-NEXT: kshiftrq $35, %k5, %k2 -; AVX512F-32-NEXT: kxorq %k3, %k2, %k3 -; AVX512F-32-NEXT: kmovd %ecx, %k2 +; AVX512F-32-NEXT: kxorq %k0, %k5, %k5 +; AVX512F-32-NEXT: kshiftrq $35, %k5, %k0 +; AVX512F-32-NEXT: kxorq %k3, %k0, %k3 +; AVX512F-32-NEXT: kmovd %ecx, %k0 ; AVX512F-32-NEXT: movl %eax, %ecx ; AVX512F-32-NEXT: shrb $2, %al ; AVX512F-32-NEXT: kshiftlq $63, %k3, %k3 ; AVX512F-32-NEXT: kshiftrq $28, %k3, %k3 ; AVX512F-32-NEXT: kxorq %k5, %k3, %k5 ; AVX512F-32-NEXT: kshiftrq $36, %k5, %k3 -; AVX512F-32-NEXT: kxorq %k4, %k3, %k4 +; AVX512F-32-NEXT: kxorq %k2, %k3, %k2 ; AVX512F-32-NEXT: kmovd %eax, %k3 ; AVX512F-32-NEXT: movl %ebx, %eax ; AVX512F-32-NEXT: shrl $16, %eax ; AVX512F-32-NEXT: shrb $3, %cl -; AVX512F-32-NEXT: kshiftlq $63, %k4, %k4 -; AVX512F-32-NEXT: kshiftrq $27, %k4, %k4 -; AVX512F-32-NEXT: kxorq %k5, %k4, %k5 -; AVX512F-32-NEXT: kshiftrq $37, %k5, %k4 -; AVX512F-32-NEXT: kxorq %k0, %k4, %k0 +; AVX512F-32-NEXT: kshiftlq $63, %k2, %k2 +; AVX512F-32-NEXT: kshiftrq $27, %k2, %k2 +; AVX512F-32-NEXT: kxorq %k5, %k2, %k2 +; AVX512F-32-NEXT: kshiftrq $37, %k2, %k5 +; AVX512F-32-NEXT: kxorq %k4, %k5, %k5 ; AVX512F-32-NEXT: kmovd %ecx, %k4 ; AVX512F-32-NEXT: movl %ebx, %ecx ; AVX512F-32-NEXT: shrl $13, %ecx ; AVX512F-32-NEXT: andb $1, %cl -; AVX512F-32-NEXT: kshiftlq $63, %k0, %k0 -; AVX512F-32-NEXT: kshiftrq $26, %k0, %k0 -; AVX512F-32-NEXT: kxorq %k5, %k0, %k0 -; AVX512F-32-NEXT: kshiftrq $38, %k0, %k5 +; AVX512F-32-NEXT: kshiftlq $63, %k5, %k5 +; AVX512F-32-NEXT: kshiftrq $26, %k5, %k5 +; AVX512F-32-NEXT: kxorq %k2, %k5, %k2 +; AVX512F-32-NEXT: kshiftrq $38, %k2, %k5 ; AVX512F-32-NEXT: kxorq %k7, %k5, %k7 ; AVX512F-32-NEXT: kmovd %ecx, %k5 ; AVX512F-32-NEXT: movl %eax, %edx @@ -2150,11 +2153,10 @@ ; AVX512F-32-NEXT: shrb %dl ; AVX512F-32-NEXT: kshiftlq $63, %k7, %k7 ; AVX512F-32-NEXT: kshiftrq $25, %k7, %k7 -; AVX512F-32-NEXT: kxorq %k0, %k7, %k7 -; AVX512F-32-NEXT: kshiftrq $39, %k7, %k0 -; AVX512F-32-NEXT: kxorq %k6, %k0, %k6 -; AVX512F-32-NEXT: kmovd %edx, %k0 -; AVX512F-32-NEXT: kmovq %k0, {{[0-9]+}}(%esp) # 8-byte Spill +; AVX512F-32-NEXT: kxorq %k2, %k7, %k7 +; AVX512F-32-NEXT: kshiftrq $39, %k7, %k2 +; AVX512F-32-NEXT: kxorq %k6, %k2, %k6 +; AVX512F-32-NEXT: kmovd %edx, %k2 ; AVX512F-32-NEXT: movl %eax, %ecx ; AVX512F-32-NEXT: andb $15, %cl ; AVX512F-32-NEXT: movl %ecx, %edx @@ -2165,46 +2167,44 @@ ; AVX512F-32-NEXT: kshiftrq $40, %k6, %k7 ; AVX512F-32-NEXT: kxorq %k1, %k7, %k7 ; AVX512F-32-NEXT: kmovd %ecx, %k1 -; AVX512F-32-NEXT: movl %ebx, %ecx +; AVX512F-32-NEXT: kmovq %k1, {{[0-9]+}}(%esp) # 8-byte Spill +; AVX512F-32-NEXT: movzwl %bx, %ecx +; AVX512F-32-NEXT: movl %ecx, %esi +; AVX512F-32-NEXT: movl %ecx, %edi ; AVX512F-32-NEXT: shrl $12, %ecx -; AVX512F-32-NEXT: andl $15, %ecx ; AVX512F-32-NEXT: kshiftlq $63, %k7, %k7 ; AVX512F-32-NEXT: kshiftrq $23, %k7, %k7 ; AVX512F-32-NEXT: kxorq %k6, %k7, %k6 ; AVX512F-32-NEXT: kshiftrq $41, %k6, %k7 -; AVX512F-32-NEXT: kxorq %k2, %k7, %k2 -; AVX512F-32-NEXT: kmovd %ecx, %k0 -; AVX512F-32-NEXT: movl %ebx, %ecx -; AVX512F-32-NEXT: shrl $14, %ecx -; AVX512F-32-NEXT: andl $3, %ecx -; AVX512F-32-NEXT: kshiftlq $63, %k2, %k2 -; AVX512F-32-NEXT: kshiftrq $22, %k2, %k2 -; AVX512F-32-NEXT: kxorq %k6, %k2, %k2 -; AVX512F-32-NEXT: kshiftrq $42, %k2, %k6 +; AVX512F-32-NEXT: kxorq %k0, %k7, %k0 +; AVX512F-32-NEXT: kmovd %ecx, %k1 +; AVX512F-32-NEXT: shrl $14, %edi +; AVX512F-32-NEXT: kshiftlq $63, %k0, %k0 +; AVX512F-32-NEXT: kshiftrq $22, %k0, %k0 +; AVX512F-32-NEXT: kxorq %k6, %k0, %k0 +; AVX512F-32-NEXT: kshiftrq $42, %k0, %k6 ; AVX512F-32-NEXT: kxorq %k3, %k6, %k3 -; AVX512F-32-NEXT: kmovd %ecx, %k7 -; AVX512F-32-NEXT: movl %ebx, %ecx -; AVX512F-32-NEXT: shrl $15, %ecx -; AVX512F-32-NEXT: andl $1, %ecx +; AVX512F-32-NEXT: kmovd %edi, %k7 +; AVX512F-32-NEXT: shrl $15, %esi ; AVX512F-32-NEXT: kshiftlq $63, %k3, %k3 ; AVX512F-32-NEXT: kshiftrq $21, %k3, %k3 -; AVX512F-32-NEXT: kxorq %k2, %k3, %k2 -; AVX512F-32-NEXT: kshiftrq $43, %k2, %k3 +; AVX512F-32-NEXT: kxorq %k0, %k3, %k0 +; AVX512F-32-NEXT: kshiftrq $43, %k0, %k3 ; AVX512F-32-NEXT: kxorq %k4, %k3, %k3 -; AVX512F-32-NEXT: kmovd %ecx, %k6 +; AVX512F-32-NEXT: kmovd %esi, %k6 ; AVX512F-32-NEXT: shrb $3, %dl ; AVX512F-32-NEXT: kshiftlq $63, %k3, %k3 ; AVX512F-32-NEXT: kshiftrq $20, %k3, %k3 -; AVX512F-32-NEXT: kxorq %k2, %k3, %k3 -; AVX512F-32-NEXT: kshiftrq $44, %k3, %k2 -; AVX512F-32-NEXT: kxorq %k0, %k2, %k0 -; AVX512F-32-NEXT: kmovd %edx, %k2 +; AVX512F-32-NEXT: kxorq %k0, %k3, %k3 +; AVX512F-32-NEXT: kshiftrq $44, %k3, %k0 +; AVX512F-32-NEXT: kxorq %k1, %k0, %k1 +; AVX512F-32-NEXT: kmovd %edx, %k0 ; AVX512F-32-NEXT: movl %eax, %ecx ; AVX512F-32-NEXT: shrb $4, %cl -; AVX512F-32-NEXT: kshiftlq $63, %k0, %k0 -; AVX512F-32-NEXT: kshiftrq $19, %k0, %k0 -; AVX512F-32-NEXT: kxorq %k3, %k0, %k0 -; AVX512F-32-NEXT: kshiftrq $45, %k0, %k3 +; AVX512F-32-NEXT: kshiftlq $63, %k1, %k1 +; AVX512F-32-NEXT: kshiftrq $19, %k1, %k1 +; AVX512F-32-NEXT: kxorq %k3, %k1, %k1 +; AVX512F-32-NEXT: kshiftrq $45, %k1, %k3 ; AVX512F-32-NEXT: kxorq %k5, %k3, %k4 ; AVX512F-32-NEXT: kmovd %ecx, %k3 ; AVX512F-32-NEXT: movl %eax, %ecx @@ -2212,21 +2212,21 @@ ; AVX512F-32-NEXT: andb $1, %cl ; AVX512F-32-NEXT: kshiftlq $63, %k4, %k4 ; AVX512F-32-NEXT: kshiftrq $18, %k4, %k4 -; AVX512F-32-NEXT: kxorq %k0, %k4, %k0 -; AVX512F-32-NEXT: kshiftrq $46, %k0, %k4 +; AVX512F-32-NEXT: kxorq %k1, %k4, %k1 +; AVX512F-32-NEXT: kshiftrq $46, %k1, %k4 ; AVX512F-32-NEXT: kxorq %k7, %k4, %k5 ; AVX512F-32-NEXT: kmovd %ecx, %k4 ; AVX512F-32-NEXT: movl %eax, %ecx ; AVX512F-32-NEXT: shrb $6, %cl ; AVX512F-32-NEXT: kshiftlq $63, %k5, %k5 ; AVX512F-32-NEXT: kshiftrq $17, %k5, %k5 -; AVX512F-32-NEXT: kxorq %k0, %k5, %k0 -; AVX512F-32-NEXT: kshiftrq $47, %k0, %k5 +; AVX512F-32-NEXT: kxorq %k1, %k5, %k1 +; AVX512F-32-NEXT: kshiftrq $47, %k1, %k5 ; AVX512F-32-NEXT: kxorq %k6, %k5, %k5 ; AVX512F-32-NEXT: kshiftlq $63, %k5, %k5 ; AVX512F-32-NEXT: kshiftrq $16, %k5, %k5 -; AVX512F-32-NEXT: kxorq %k0, %k5, %k0 -; AVX512F-32-NEXT: kshiftrq $48, %k0, %k5 +; AVX512F-32-NEXT: kxorq %k1, %k5, %k1 +; AVX512F-32-NEXT: kshiftrq $48, %k1, %k5 ; AVX512F-32-NEXT: kmovd %eax, %k6 ; AVX512F-32-NEXT: kxorq %k6, %k5, %k6 ; AVX512F-32-NEXT: kmovd %ecx, %k5 @@ -2236,17 +2236,17 @@ ; AVX512F-32-NEXT: shrb $7, %al ; AVX512F-32-NEXT: kshiftlq $63, %k6, %k6 ; AVX512F-32-NEXT: kshiftrq $15, %k6, %k6 -; AVX512F-32-NEXT: kxorq %k0, %k6, %k6 -; AVX512F-32-NEXT: kshiftrq $49, %k6, %k0 -; AVX512F-32-NEXT: kmovq {{[0-9]+}}(%esp), %k7 # 8-byte Reload -; AVX512F-32-NEXT: kxorq %k7, %k0, %k7 -; AVX512F-32-NEXT: kmovd %eax, %k0 +; AVX512F-32-NEXT: kxorq %k1, %k6, %k1 +; AVX512F-32-NEXT: kshiftrq $49, %k1, %k6 +; AVX512F-32-NEXT: kxorq %k2, %k6, %k6 +; AVX512F-32-NEXT: kmovd %eax, %k2 ; AVX512F-32-NEXT: movl %edx, %eax -; AVX512F-32-NEXT: kshiftlq $63, %k7, %k7 -; AVX512F-32-NEXT: kshiftrq $14, %k7, %k7 -; AVX512F-32-NEXT: kxorq %k6, %k7, %k6 -; AVX512F-32-NEXT: kshiftrq $50, %k6, %k7 -; AVX512F-32-NEXT: kxorq %k1, %k7, %k7 +; AVX512F-32-NEXT: kshiftlq $63, %k6, %k6 +; AVX512F-32-NEXT: kshiftrq $14, %k6, %k6 +; AVX512F-32-NEXT: kxorq %k1, %k6, %k6 +; AVX512F-32-NEXT: kshiftrq $50, %k6, %k1 +; AVX512F-32-NEXT: kmovq {{[0-9]+}}(%esp), %k7 # 8-byte Reload +; AVX512F-32-NEXT: kxorq %k7, %k1, %k7 ; AVX512F-32-NEXT: kmovd %edx, %k1 ; AVX512F-32-NEXT: # kill: def %dl killed %dl killed %edx def %edx ; AVX512F-32-NEXT: andb $15, %dl @@ -2256,8 +2256,8 @@ ; AVX512F-32-NEXT: kshiftrq $13, %k7, %k7 ; AVX512F-32-NEXT: kxorq %k6, %k7, %k6 ; AVX512F-32-NEXT: kshiftrq $51, %k6, %k7 -; AVX512F-32-NEXT: kxorq %k2, %k7, %k7 -; AVX512F-32-NEXT: kmovd %eax, %k2 +; AVX512F-32-NEXT: kxorq %k0, %k7, %k7 +; AVX512F-32-NEXT: kmovd %eax, %k0 ; AVX512F-32-NEXT: movl %edx, %eax ; AVX512F-32-NEXT: shrb $2, %dl ; AVX512F-32-NEXT: kshiftlq $63, %k7, %k7 @@ -2286,20 +2286,20 @@ ; AVX512F-32-NEXT: kshiftrq $9, %k7, %k7 ; AVX512F-32-NEXT: kxorq %k6, %k7, %k6 ; AVX512F-32-NEXT: kshiftrq $55, %k6, %k7 -; AVX512F-32-NEXT: kxorq %k0, %k7, %k0 -; AVX512F-32-NEXT: kshiftlq $63, %k0, %k0 -; AVX512F-32-NEXT: kshiftrq $8, %k0, %k0 -; AVX512F-32-NEXT: kxorq %k6, %k0, %k0 -; AVX512F-32-NEXT: kshiftrq $56, %k0, %k6 +; AVX512F-32-NEXT: kxorq %k2, %k7, %k2 +; AVX512F-32-NEXT: kshiftlq $63, %k2, %k2 +; AVX512F-32-NEXT: kshiftrq $8, %k2, %k2 +; AVX512F-32-NEXT: kxorq %k6, %k2, %k2 +; AVX512F-32-NEXT: kshiftrq $56, %k2, %k6 ; AVX512F-32-NEXT: kxorq %k1, %k6, %k1 ; AVX512F-32-NEXT: kshiftlq $63, %k1, %k1 ; AVX512F-32-NEXT: kshiftrq $7, %k1, %k1 -; AVX512F-32-NEXT: kxorq %k0, %k1, %k0 -; AVX512F-32-NEXT: kshiftrq $57, %k0, %k1 ; AVX512F-32-NEXT: kxorq %k2, %k1, %k1 -; AVX512F-32-NEXT: kshiftlq $63, %k1, %k1 -; AVX512F-32-NEXT: kshiftrq $6, %k1, %k1 -; AVX512F-32-NEXT: kxorq %k0, %k1, %k0 +; AVX512F-32-NEXT: kshiftrq $57, %k1, %k2 +; AVX512F-32-NEXT: kxorq %k0, %k2, %k0 +; AVX512F-32-NEXT: kshiftlq $63, %k0, %k0 +; AVX512F-32-NEXT: kshiftrq $6, %k0, %k0 +; AVX512F-32-NEXT: kxorq %k1, %k0, %k0 ; AVX512F-32-NEXT: kshiftrq $58, %k0, %k1 ; AVX512F-32-NEXT: kxorq %k3, %k1, %k1 ; AVX512F-32-NEXT: movl %ebx, %eax @@ -2362,11 +2362,13 @@ ; AVX512F-32-NEXT: kmovq %k0, {{[0-9]+}}(%esp) ; AVX512F-32-NEXT: addl {{[0-9]+}}(%esp), %eax ; AVX512F-32-NEXT: adcl {{[0-9]+}}(%esp), %edx -; AVX512F-32-NEXT: addl %esi, %eax +; AVX512F-32-NEXT: addl %ebp, %eax ; AVX512F-32-NEXT: adcl %ebx, %edx ; AVX512F-32-NEXT: addl $60, %esp ; AVX512F-32-NEXT: popl %esi +; AVX512F-32-NEXT: popl %edi ; AVX512F-32-NEXT: popl %ebx +; AVX512F-32-NEXT: popl %ebp ; AVX512F-32-NEXT: vzeroupper ; AVX512F-32-NEXT: retl %res0 = call i64 @llvm.x86.avx512.mask.cmp.b.512(<64 x i8> %a0, <64 x i8> %a1, i32 0, i64 %mask) @@ -2495,14 +2497,20 @@ ; ; AVX512F-32-LABEL: test_mask_x86_avx512_ucmp_b_512: ; AVX512F-32: # %bb.0: -; AVX512F-32-NEXT: pushl %ebx +; AVX512F-32-NEXT: pushl %ebp ; AVX512F-32-NEXT: .cfi_def_cfa_offset 8 -; AVX512F-32-NEXT: pushl %esi +; AVX512F-32-NEXT: pushl %ebx ; AVX512F-32-NEXT: .cfi_def_cfa_offset 12 +; AVX512F-32-NEXT: pushl %edi +; AVX512F-32-NEXT: .cfi_def_cfa_offset 16 +; AVX512F-32-NEXT: pushl %esi +; AVX512F-32-NEXT: .cfi_def_cfa_offset 20 ; AVX512F-32-NEXT: subl $60, %esp -; AVX512F-32-NEXT: .cfi_def_cfa_offset 72 -; AVX512F-32-NEXT: .cfi_offset %esi, -12 -; AVX512F-32-NEXT: .cfi_offset %ebx, -8 +; AVX512F-32-NEXT: .cfi_def_cfa_offset 80 +; AVX512F-32-NEXT: .cfi_offset %esi, -20 +; AVX512F-32-NEXT: .cfi_offset %edi, -16 +; AVX512F-32-NEXT: .cfi_offset %ebx, -12 +; AVX512F-32-NEXT: .cfi_offset %ebp, -8 ; AVX512F-32-NEXT: movl {{[0-9]+}}(%esp), %ebx ; AVX512F-32-NEXT: movl %ebx, %eax ; AVX512F-32-NEXT: shrl $16, %eax @@ -2542,7 +2550,7 @@ ; AVX512F-32-NEXT: kmovd %ecx, %k5 ; AVX512F-32-NEXT: movb %bh, %cl ; AVX512F-32-NEXT: kmovd %ecx, %k1 -; AVX512F-32-NEXT: movl %ebx, %esi +; AVX512F-32-NEXT: movl %ebx, %ebp ; AVX512F-32-NEXT: andb $2, %cl ; AVX512F-32-NEXT: shrb %cl ; AVX512F-32-NEXT: kshiftlq $63, %k2, %k2 @@ -2568,7 +2576,7 @@ ; AVX512F-32-NEXT: kshiftrq $5, %k7, %k0 ; AVX512F-32-NEXT: kxorq %k4, %k0, %k4 ; AVX512F-32-NEXT: kmovd %ecx, %k0 -; AVX512F-32-NEXT: movl %esi, %ecx +; AVX512F-32-NEXT: movl %ebp, %ecx ; AVX512F-32-NEXT: shrl $13, %ecx ; AVX512F-32-NEXT: andb $1, %cl ; AVX512F-32-NEXT: kshiftlq $63, %k4, %k4 @@ -2613,9 +2621,10 @@ ; AVX512F-32-NEXT: movl %eax, %ecx ; AVX512F-32-NEXT: shrb $5, %cl ; AVX512F-32-NEXT: andb $1, %cl -; AVX512F-32-NEXT: movl %esi, %edx +; AVX512F-32-NEXT: movzwl %bp, %edx +; AVX512F-32-NEXT: movl %edx, %esi +; AVX512F-32-NEXT: movl %edx, %edi ; AVX512F-32-NEXT: shrl $12, %edx -; AVX512F-32-NEXT: andl $15, %edx ; AVX512F-32-NEXT: kshiftlq $63, %k3, %k3 ; AVX512F-32-NEXT: kshiftrq $53, %k3, %k3 ; AVX512F-32-NEXT: kxorq %k7, %k3, %k3 @@ -2630,28 +2639,24 @@ ; AVX512F-32-NEXT: kmovd %ecx, %k0 ; AVX512F-32-NEXT: movl %eax, %edx ; AVX512F-32-NEXT: shrb $6, %dl -; AVX512F-32-NEXT: movl %esi, %ecx -; AVX512F-32-NEXT: shrl $14, %ecx -; AVX512F-32-NEXT: andl $3, %ecx +; AVX512F-32-NEXT: shrl $15, %esi +; AVX512F-32-NEXT: shrl $14, %edi ; AVX512F-32-NEXT: kshiftlq $63, %k7, %k7 ; AVX512F-32-NEXT: kshiftrq $51, %k7, %k7 ; AVX512F-32-NEXT: kxorq %k3, %k7, %k3 ; AVX512F-32-NEXT: kshiftrq $13, %k3, %k7 ; AVX512F-32-NEXT: kxorq %k4, %k7, %k4 -; AVX512F-32-NEXT: kmovd %ecx, %k7 -; AVX512F-32-NEXT: movl %esi, %ecx -; AVX512F-32-NEXT: shrl $15, %ecx -; AVX512F-32-NEXT: andl $1, %ecx ; AVX512F-32-NEXT: kshiftlq $63, %k4, %k4 ; AVX512F-32-NEXT: kshiftrq $50, %k4, %k4 ; AVX512F-32-NEXT: kxorq %k3, %k4, %k3 ; AVX512F-32-NEXT: kshiftrq $14, %k3, %k4 +; AVX512F-32-NEXT: kmovd %edi, %k7 ; AVX512F-32-NEXT: kxorq %k7, %k4, %k4 ; AVX512F-32-NEXT: kshiftlq $63, %k4, %k4 ; AVX512F-32-NEXT: kshiftrq $49, %k4, %k4 ; AVX512F-32-NEXT: kxorq %k3, %k4, %k3 ; AVX512F-32-NEXT: kshiftrq $15, %k3, %k4 -; AVX512F-32-NEXT: kmovd %ecx, %k7 +; AVX512F-32-NEXT: kmovd %esi, %k7 ; AVX512F-32-NEXT: kxorq %k7, %k4, %k4 ; AVX512F-32-NEXT: kshiftlq $63, %k4, %k4 ; AVX512F-32-NEXT: kshiftrq $48, %k4, %k4 @@ -2660,7 +2665,7 @@ ; AVX512F-32-NEXT: kmovd %eax, %k7 ; AVX512F-32-NEXT: kxorq %k7, %k4, %k4 ; AVX512F-32-NEXT: kmovd %edx, %k7 -; AVX512F-32-NEXT: movl %esi, %edx +; AVX512F-32-NEXT: movl %ebp, %edx ; AVX512F-32-NEXT: shrl $24, %edx ; AVX512F-32-NEXT: # kill: def %al killed %al killed %eax def %eax ; AVX512F-32-NEXT: shrb $7, %al @@ -2700,19 +2705,19 @@ ; AVX512F-32-NEXT: shrb $3, %cl ; AVX512F-32-NEXT: kshiftlq $63, %k6, %k6 ; AVX512F-32-NEXT: kshiftrq $43, %k6, %k6 -; AVX512F-32-NEXT: kxorq %k1, %k6, %k1 -; AVX512F-32-NEXT: kshiftrq $21, %k1, %k6 -; AVX512F-32-NEXT: kxorq %k0, %k6, %k6 -; AVX512F-32-NEXT: kmovd %ecx, %k0 -; AVX512F-32-NEXT: movl %esi, %ecx -; AVX512F-32-NEXT: shrl $29, %ecx -; AVX512F-32-NEXT: andb $1, %cl -; AVX512F-32-NEXT: kshiftlq $63, %k6, %k6 -; AVX512F-32-NEXT: kshiftrq $42, %k6, %k6 ; AVX512F-32-NEXT: kxorq %k1, %k6, %k6 -; AVX512F-32-NEXT: kshiftrq $22, %k6, %k1 -; AVX512F-32-NEXT: kxorq %k7, %k1, %k7 +; AVX512F-32-NEXT: kshiftrq $21, %k6, %k1 +; AVX512F-32-NEXT: kxorq %k0, %k1, %k0 ; AVX512F-32-NEXT: kmovd %ecx, %k1 +; AVX512F-32-NEXT: movl %ebp, %ecx +; AVX512F-32-NEXT: shrl $29, %ecx +; AVX512F-32-NEXT: andb $1, %cl +; AVX512F-32-NEXT: kshiftlq $63, %k0, %k0 +; AVX512F-32-NEXT: kshiftrq $42, %k0, %k0 +; AVX512F-32-NEXT: kxorq %k6, %k0, %k6 +; AVX512F-32-NEXT: kshiftrq $22, %k6, %k0 +; AVX512F-32-NEXT: kxorq %k7, %k0, %k7 +; AVX512F-32-NEXT: kmovd %ecx, %k0 ; AVX512F-32-NEXT: movl %ebx, %edx ; AVX512F-32-NEXT: andb $2, %dl ; AVX512F-32-NEXT: shrb %dl @@ -2743,68 +2748,68 @@ ; AVX512F-32-NEXT: shrb $4, %cl ; AVX512F-32-NEXT: kshiftlq $63, %k4, %k4 ; AVX512F-32-NEXT: kshiftrq $38, %k4, %k4 -; AVX512F-32-NEXT: kxorq %k7, %k4, %k7 -; AVX512F-32-NEXT: kshiftrq $26, %k7, %k4 -; AVX512F-32-NEXT: kxorq %k2, %k4, %k2 -; AVX512F-32-NEXT: kmovd %ecx, %k4 +; AVX512F-32-NEXT: kxorq %k7, %k4, %k4 +; AVX512F-32-NEXT: kshiftrq $26, %k4, %k7 +; AVX512F-32-NEXT: kxorq %k2, %k7, %k7 +; AVX512F-32-NEXT: kmovd %ecx, %k2 ; AVX512F-32-NEXT: movl %ebx, %ecx ; AVX512F-32-NEXT: shrb $5, %cl ; AVX512F-32-NEXT: andb $1, %cl -; AVX512F-32-NEXT: movl %esi, %edx +; AVX512F-32-NEXT: movl %ebp, %edx ; AVX512F-32-NEXT: shrl $28, %edx -; AVX512F-32-NEXT: kshiftlq $63, %k2, %k2 -; AVX512F-32-NEXT: kshiftrq $37, %k2, %k2 -; AVX512F-32-NEXT: kxorq %k7, %k2, %k2 -; AVX512F-32-NEXT: kshiftrq $27, %k2, %k7 -; AVX512F-32-NEXT: kxorq %k0, %k7, %k0 -; AVX512F-32-NEXT: kshiftlq $63, %k0, %k0 -; AVX512F-32-NEXT: kshiftrq $36, %k0, %k0 -; AVX512F-32-NEXT: kxorq %k2, %k0, %k2 -; AVX512F-32-NEXT: kshiftrq $28, %k2, %k0 +; AVX512F-32-NEXT: kshiftlq $63, %k7, %k7 +; AVX512F-32-NEXT: kshiftrq $37, %k7, %k7 +; AVX512F-32-NEXT: kxorq %k4, %k7, %k4 +; AVX512F-32-NEXT: kshiftrq $27, %k4, %k7 +; AVX512F-32-NEXT: kxorq %k1, %k7, %k1 +; AVX512F-32-NEXT: kshiftlq $63, %k1, %k1 +; AVX512F-32-NEXT: kshiftrq $36, %k1, %k1 +; AVX512F-32-NEXT: kxorq %k4, %k1, %k1 +; AVX512F-32-NEXT: kshiftrq $28, %k1, %k4 ; AVX512F-32-NEXT: kmovd %edx, %k7 -; AVX512F-32-NEXT: kxorq %k7, %k0, %k7 -; AVX512F-32-NEXT: kmovd %ecx, %k0 +; AVX512F-32-NEXT: kxorq %k7, %k4, %k7 +; AVX512F-32-NEXT: kmovd %ecx, %k4 ; AVX512F-32-NEXT: movl %ebx, %edx ; AVX512F-32-NEXT: shrb $6, %dl -; AVX512F-32-NEXT: movl %esi, %ecx -; AVX512F-32-NEXT: shrl $30, %ecx +; AVX512F-32-NEXT: movl %ebp, %ecx +; AVX512F-32-NEXT: shrl $31, %ecx +; AVX512F-32-NEXT: movl %ebp, %esi +; AVX512F-32-NEXT: shrl $30, %esi ; AVX512F-32-NEXT: kshiftlq $63, %k7, %k7 ; AVX512F-32-NEXT: kshiftrq $35, %k7, %k7 -; AVX512F-32-NEXT: kxorq %k2, %k7, %k2 -; AVX512F-32-NEXT: kshiftrq $29, %k2, %k7 ; AVX512F-32-NEXT: kxorq %k1, %k7, %k1 -; AVX512F-32-NEXT: kmovd %ecx, %k7 -; AVX512F-32-NEXT: movl %esi, %ecx -; AVX512F-32-NEXT: shrl $31, %ecx +; AVX512F-32-NEXT: kshiftrq $29, %k1, %k7 +; AVX512F-32-NEXT: kxorq %k0, %k7, %k0 +; AVX512F-32-NEXT: kshiftlq $63, %k0, %k0 +; AVX512F-32-NEXT: kshiftrq $34, %k0, %k0 +; AVX512F-32-NEXT: kxorq %k1, %k0, %k0 +; AVX512F-32-NEXT: kshiftrq $30, %k0, %k1 +; AVX512F-32-NEXT: kmovd %esi, %k7 +; AVX512F-32-NEXT: kxorq %k7, %k1, %k1 ; AVX512F-32-NEXT: kshiftlq $63, %k1, %k1 -; AVX512F-32-NEXT: kshiftrq $34, %k1, %k1 -; AVX512F-32-NEXT: kxorq %k2, %k1, %k1 -; AVX512F-32-NEXT: kshiftrq $30, %k1, %k2 -; AVX512F-32-NEXT: kxorq %k7, %k2, %k2 -; AVX512F-32-NEXT: kshiftlq $63, %k2, %k2 -; AVX512F-32-NEXT: kshiftrq $33, %k2, %k2 -; AVX512F-32-NEXT: kxorq %k1, %k2, %k1 -; AVX512F-32-NEXT: kshiftrq $31, %k1, %k2 +; AVX512F-32-NEXT: kshiftrq $33, %k1, %k1 +; AVX512F-32-NEXT: kxorq %k0, %k1, %k0 +; AVX512F-32-NEXT: kshiftrq $31, %k0, %k1 ; AVX512F-32-NEXT: kmovd %ecx, %k7 -; AVX512F-32-NEXT: kxorq %k7, %k2, %k2 -; AVX512F-32-NEXT: kshiftlq $63, %k2, %k2 -; AVX512F-32-NEXT: kshiftrq $32, %k2, %k2 -; AVX512F-32-NEXT: kxorq %k1, %k2, %k1 -; AVX512F-32-NEXT: kshiftrq $32, %k1, %k2 +; AVX512F-32-NEXT: kxorq %k7, %k1, %k1 +; AVX512F-32-NEXT: kshiftlq $63, %k1, %k1 +; AVX512F-32-NEXT: kshiftrq $32, %k1, %k1 +; AVX512F-32-NEXT: kxorq %k0, %k1, %k0 +; AVX512F-32-NEXT: kshiftrq $32, %k0, %k1 ; AVX512F-32-NEXT: kmovd %ebx, %k7 -; AVX512F-32-NEXT: kxorq %k7, %k2, %k2 +; AVX512F-32-NEXT: kxorq %k7, %k1, %k1 ; AVX512F-32-NEXT: kmovd %edx, %k7 ; AVX512F-32-NEXT: movl %ebx, %ecx ; AVX512F-32-NEXT: shrb $7, %cl -; AVX512F-32-NEXT: kshiftlq $63, %k2, %k2 -; AVX512F-32-NEXT: kshiftrq $31, %k2, %k2 -; AVX512F-32-NEXT: kxorq %k1, %k2, %k1 -; AVX512F-32-NEXT: kshiftrq $33, %k1, %k2 -; AVX512F-32-NEXT: kxorq %k5, %k2, %k2 -; AVX512F-32-NEXT: kshiftlq $63, %k2, %k2 -; AVX512F-32-NEXT: kshiftrq $30, %k2, %k2 -; AVX512F-32-NEXT: kxorq %k1, %k2, %k2 -; AVX512F-32-NEXT: kshiftrq $34, %k2, %k1 +; AVX512F-32-NEXT: kshiftlq $63, %k1, %k1 +; AVX512F-32-NEXT: kshiftrq $31, %k1, %k1 +; AVX512F-32-NEXT: kxorq %k0, %k1, %k0 +; AVX512F-32-NEXT: kshiftrq $33, %k0, %k1 +; AVX512F-32-NEXT: kxorq %k5, %k1, %k1 +; AVX512F-32-NEXT: kshiftlq $63, %k1, %k1 +; AVX512F-32-NEXT: kshiftrq $30, %k1, %k1 +; AVX512F-32-NEXT: kxorq %k0, %k1, %k0 +; AVX512F-32-NEXT: kshiftrq $34, %k0, %k1 ; AVX512F-32-NEXT: kxorq %k6, %k1, %k5 ; AVX512F-32-NEXT: kmovd %ecx, %k6 ; AVX512F-32-NEXT: movb %bh, %cl @@ -2813,34 +2818,34 @@ ; AVX512F-32-NEXT: shrb %cl ; AVX512F-32-NEXT: kshiftlq $63, %k5, %k5 ; AVX512F-32-NEXT: kshiftrq $29, %k5, %k5 -; AVX512F-32-NEXT: kxorq %k2, %k5, %k5 -; AVX512F-32-NEXT: kshiftrq $35, %k5, %k2 -; AVX512F-32-NEXT: kxorq %k3, %k2, %k3 -; AVX512F-32-NEXT: kmovd %ecx, %k2 +; AVX512F-32-NEXT: kxorq %k0, %k5, %k5 +; AVX512F-32-NEXT: kshiftrq $35, %k5, %k0 +; AVX512F-32-NEXT: kxorq %k3, %k0, %k3 +; AVX512F-32-NEXT: kmovd %ecx, %k0 ; AVX512F-32-NEXT: movl %eax, %ecx ; AVX512F-32-NEXT: shrb $2, %al ; AVX512F-32-NEXT: kshiftlq $63, %k3, %k3 ; AVX512F-32-NEXT: kshiftrq $28, %k3, %k3 ; AVX512F-32-NEXT: kxorq %k5, %k3, %k5 ; AVX512F-32-NEXT: kshiftrq $36, %k5, %k3 -; AVX512F-32-NEXT: kxorq %k4, %k3, %k4 +; AVX512F-32-NEXT: kxorq %k2, %k3, %k2 ; AVX512F-32-NEXT: kmovd %eax, %k3 ; AVX512F-32-NEXT: movl %ebx, %eax ; AVX512F-32-NEXT: shrl $16, %eax ; AVX512F-32-NEXT: shrb $3, %cl -; AVX512F-32-NEXT: kshiftlq $63, %k4, %k4 -; AVX512F-32-NEXT: kshiftrq $27, %k4, %k4 -; AVX512F-32-NEXT: kxorq %k5, %k4, %k5 -; AVX512F-32-NEXT: kshiftrq $37, %k5, %k4 -; AVX512F-32-NEXT: kxorq %k0, %k4, %k0 +; AVX512F-32-NEXT: kshiftlq $63, %k2, %k2 +; AVX512F-32-NEXT: kshiftrq $27, %k2, %k2 +; AVX512F-32-NEXT: kxorq %k5, %k2, %k2 +; AVX512F-32-NEXT: kshiftrq $37, %k2, %k5 +; AVX512F-32-NEXT: kxorq %k4, %k5, %k5 ; AVX512F-32-NEXT: kmovd %ecx, %k4 ; AVX512F-32-NEXT: movl %ebx, %ecx ; AVX512F-32-NEXT: shrl $13, %ecx ; AVX512F-32-NEXT: andb $1, %cl -; AVX512F-32-NEXT: kshiftlq $63, %k0, %k0 -; AVX512F-32-NEXT: kshiftrq $26, %k0, %k0 -; AVX512F-32-NEXT: kxorq %k5, %k0, %k0 -; AVX512F-32-NEXT: kshiftrq $38, %k0, %k5 +; AVX512F-32-NEXT: kshiftlq $63, %k5, %k5 +; AVX512F-32-NEXT: kshiftrq $26, %k5, %k5 +; AVX512F-32-NEXT: kxorq %k2, %k5, %k2 +; AVX512F-32-NEXT: kshiftrq $38, %k2, %k5 ; AVX512F-32-NEXT: kxorq %k7, %k5, %k7 ; AVX512F-32-NEXT: kmovd %ecx, %k5 ; AVX512F-32-NEXT: movl %eax, %edx @@ -2848,11 +2853,10 @@ ; AVX512F-32-NEXT: shrb %dl ; AVX512F-32-NEXT: kshiftlq $63, %k7, %k7 ; AVX512F-32-NEXT: kshiftrq $25, %k7, %k7 -; AVX512F-32-NEXT: kxorq %k0, %k7, %k7 -; AVX512F-32-NEXT: kshiftrq $39, %k7, %k0 -; AVX512F-32-NEXT: kxorq %k6, %k0, %k6 -; AVX512F-32-NEXT: kmovd %edx, %k0 -; AVX512F-32-NEXT: kmovq %k0, {{[0-9]+}}(%esp) # 8-byte Spill +; AVX512F-32-NEXT: kxorq %k2, %k7, %k7 +; AVX512F-32-NEXT: kshiftrq $39, %k7, %k2 +; AVX512F-32-NEXT: kxorq %k6, %k2, %k6 +; AVX512F-32-NEXT: kmovd %edx, %k2 ; AVX512F-32-NEXT: movl %eax, %ecx ; AVX512F-32-NEXT: andb $15, %cl ; AVX512F-32-NEXT: movl %ecx, %edx @@ -2863,46 +2867,44 @@ ; AVX512F-32-NEXT: kshiftrq $40, %k6, %k7 ; AVX512F-32-NEXT: kxorq %k1, %k7, %k7 ; AVX512F-32-NEXT: kmovd %ecx, %k1 -; AVX512F-32-NEXT: movl %ebx, %ecx +; AVX512F-32-NEXT: kmovq %k1, {{[0-9]+}}(%esp) # 8-byte Spill +; AVX512F-32-NEXT: movzwl %bx, %ecx +; AVX512F-32-NEXT: movl %ecx, %esi +; AVX512F-32-NEXT: movl %ecx, %edi ; AVX512F-32-NEXT: shrl $12, %ecx -; AVX512F-32-NEXT: andl $15, %ecx ; AVX512F-32-NEXT: kshiftlq $63, %k7, %k7 ; AVX512F-32-NEXT: kshiftrq $23, %k7, %k7 ; AVX512F-32-NEXT: kxorq %k6, %k7, %k6 ; AVX512F-32-NEXT: kshiftrq $41, %k6, %k7 -; AVX512F-32-NEXT: kxorq %k2, %k7, %k2 -; AVX512F-32-NEXT: kmovd %ecx, %k0 -; AVX512F-32-NEXT: movl %ebx, %ecx -; AVX512F-32-NEXT: shrl $14, %ecx -; AVX512F-32-NEXT: andl $3, %ecx -; AVX512F-32-NEXT: kshiftlq $63, %k2, %k2 -; AVX512F-32-NEXT: kshiftrq $22, %k2, %k2 -; AVX512F-32-NEXT: kxorq %k6, %k2, %k2 -; AVX512F-32-NEXT: kshiftrq $42, %k2, %k6 +; AVX512F-32-NEXT: kxorq %k0, %k7, %k0 +; AVX512F-32-NEXT: kmovd %ecx, %k1 +; AVX512F-32-NEXT: shrl $14, %edi +; AVX512F-32-NEXT: kshiftlq $63, %k0, %k0 +; AVX512F-32-NEXT: kshiftrq $22, %k0, %k0 +; AVX512F-32-NEXT: kxorq %k6, %k0, %k0 +; AVX512F-32-NEXT: kshiftrq $42, %k0, %k6 ; AVX512F-32-NEXT: kxorq %k3, %k6, %k3 -; AVX512F-32-NEXT: kmovd %ecx, %k7 -; AVX512F-32-NEXT: movl %ebx, %ecx -; AVX512F-32-NEXT: shrl $15, %ecx -; AVX512F-32-NEXT: andl $1, %ecx +; AVX512F-32-NEXT: kmovd %edi, %k7 +; AVX512F-32-NEXT: shrl $15, %esi ; AVX512F-32-NEXT: kshiftlq $63, %k3, %k3 ; AVX512F-32-NEXT: kshiftrq $21, %k3, %k3 -; AVX512F-32-NEXT: kxorq %k2, %k3, %k2 -; AVX512F-32-NEXT: kshiftrq $43, %k2, %k3 +; AVX512F-32-NEXT: kxorq %k0, %k3, %k0 +; AVX512F-32-NEXT: kshiftrq $43, %k0, %k3 ; AVX512F-32-NEXT: kxorq %k4, %k3, %k3 -; AVX512F-32-NEXT: kmovd %ecx, %k6 +; AVX512F-32-NEXT: kmovd %esi, %k6 ; AVX512F-32-NEXT: shrb $3, %dl ; AVX512F-32-NEXT: kshiftlq $63, %k3, %k3 ; AVX512F-32-NEXT: kshiftrq $20, %k3, %k3 -; AVX512F-32-NEXT: kxorq %k2, %k3, %k3 -; AVX512F-32-NEXT: kshiftrq $44, %k3, %k2 -; AVX512F-32-NEXT: kxorq %k0, %k2, %k0 -; AVX512F-32-NEXT: kmovd %edx, %k2 +; AVX512F-32-NEXT: kxorq %k0, %k3, %k3 +; AVX512F-32-NEXT: kshiftrq $44, %k3, %k0 +; AVX512F-32-NEXT: kxorq %k1, %k0, %k1 +; AVX512F-32-NEXT: kmovd %edx, %k0 ; AVX512F-32-NEXT: movl %eax, %ecx ; AVX512F-32-NEXT: shrb $4, %cl -; AVX512F-32-NEXT: kshiftlq $63, %k0, %k0 -; AVX512F-32-NEXT: kshiftrq $19, %k0, %k0 -; AVX512F-32-NEXT: kxorq %k3, %k0, %k0 -; AVX512F-32-NEXT: kshiftrq $45, %k0, %k3 +; AVX512F-32-NEXT: kshiftlq $63, %k1, %k1 +; AVX512F-32-NEXT: kshiftrq $19, %k1, %k1 +; AVX512F-32-NEXT: kxorq %k3, %k1, %k1 +; AVX512F-32-NEXT: kshiftrq $45, %k1, %k3 ; AVX512F-32-NEXT: kxorq %k5, %k3, %k4 ; AVX512F-32-NEXT: kmovd %ecx, %k3 ; AVX512F-32-NEXT: movl %eax, %ecx @@ -2910,21 +2912,21 @@ ; AVX512F-32-NEXT: andb $1, %cl ; AVX512F-32-NEXT: kshiftlq $63, %k4, %k4 ; AVX512F-32-NEXT: kshiftrq $18, %k4, %k4 -; AVX512F-32-NEXT: kxorq %k0, %k4, %k0 -; AVX512F-32-NEXT: kshiftrq $46, %k0, %k4 +; AVX512F-32-NEXT: kxorq %k1, %k4, %k1 +; AVX512F-32-NEXT: kshiftrq $46, %k1, %k4 ; AVX512F-32-NEXT: kxorq %k7, %k4, %k5 ; AVX512F-32-NEXT: kmovd %ecx, %k4 ; AVX512F-32-NEXT: movl %eax, %ecx ; AVX512F-32-NEXT: shrb $6, %cl ; AVX512F-32-NEXT: kshiftlq $63, %k5, %k5 ; AVX512F-32-NEXT: kshiftrq $17, %k5, %k5 -; AVX512F-32-NEXT: kxorq %k0, %k5, %k0 -; AVX512F-32-NEXT: kshiftrq $47, %k0, %k5 +; AVX512F-32-NEXT: kxorq %k1, %k5, %k1 +; AVX512F-32-NEXT: kshiftrq $47, %k1, %k5 ; AVX512F-32-NEXT: kxorq %k6, %k5, %k5 ; AVX512F-32-NEXT: kshiftlq $63, %k5, %k5 ; AVX512F-32-NEXT: kshiftrq $16, %k5, %k5 -; AVX512F-32-NEXT: kxorq %k0, %k5, %k0 -; AVX512F-32-NEXT: kshiftrq $48, %k0, %k5 +; AVX512F-32-NEXT: kxorq %k1, %k5, %k1 +; AVX512F-32-NEXT: kshiftrq $48, %k1, %k5 ; AVX512F-32-NEXT: kmovd %eax, %k6 ; AVX512F-32-NEXT: kxorq %k6, %k5, %k6 ; AVX512F-32-NEXT: kmovd %ecx, %k5 @@ -2934,17 +2936,17 @@ ; AVX512F-32-NEXT: shrb $7, %al ; AVX512F-32-NEXT: kshiftlq $63, %k6, %k6 ; AVX512F-32-NEXT: kshiftrq $15, %k6, %k6 -; AVX512F-32-NEXT: kxorq %k0, %k6, %k6 -; AVX512F-32-NEXT: kshiftrq $49, %k6, %k0 -; AVX512F-32-NEXT: kmovq {{[0-9]+}}(%esp), %k7 # 8-byte Reload -; AVX512F-32-NEXT: kxorq %k7, %k0, %k7 -; AVX512F-32-NEXT: kmovd %eax, %k0 +; AVX512F-32-NEXT: kxorq %k1, %k6, %k1 +; AVX512F-32-NEXT: kshiftrq $49, %k1, %k6 +; AVX512F-32-NEXT: kxorq %k2, %k6, %k6 +; AVX512F-32-NEXT: kmovd %eax, %k2 ; AVX512F-32-NEXT: movl %edx, %eax -; AVX512F-32-NEXT: kshiftlq $63, %k7, %k7 -; AVX512F-32-NEXT: kshiftrq $14, %k7, %k7 -; AVX512F-32-NEXT: kxorq %k6, %k7, %k6 -; AVX512F-32-NEXT: kshiftrq $50, %k6, %k7 -; AVX512F-32-NEXT: kxorq %k1, %k7, %k7 +; AVX512F-32-NEXT: kshiftlq $63, %k6, %k6 +; AVX512F-32-NEXT: kshiftrq $14, %k6, %k6 +; AVX512F-32-NEXT: kxorq %k1, %k6, %k6 +; AVX512F-32-NEXT: kshiftrq $50, %k6, %k1 +; AVX512F-32-NEXT: kmovq {{[0-9]+}}(%esp), %k7 # 8-byte Reload +; AVX512F-32-NEXT: kxorq %k7, %k1, %k7 ; AVX512F-32-NEXT: kmovd %edx, %k1 ; AVX512F-32-NEXT: # kill: def %dl killed %dl killed %edx def %edx ; AVX512F-32-NEXT: andb $15, %dl @@ -2954,8 +2956,8 @@ ; AVX512F-32-NEXT: kshiftrq $13, %k7, %k7 ; AVX512F-32-NEXT: kxorq %k6, %k7, %k6 ; AVX512F-32-NEXT: kshiftrq $51, %k6, %k7 -; AVX512F-32-NEXT: kxorq %k2, %k7, %k7 -; AVX512F-32-NEXT: kmovd %eax, %k2 +; AVX512F-32-NEXT: kxorq %k0, %k7, %k7 +; AVX512F-32-NEXT: kmovd %eax, %k0 ; AVX512F-32-NEXT: movl %edx, %eax ; AVX512F-32-NEXT: shrb $2, %dl ; AVX512F-32-NEXT: kshiftlq $63, %k7, %k7 @@ -2984,20 +2986,20 @@ ; AVX512F-32-NEXT: kshiftrq $9, %k7, %k7 ; AVX512F-32-NEXT: kxorq %k6, %k7, %k6 ; AVX512F-32-NEXT: kshiftrq $55, %k6, %k7 -; AVX512F-32-NEXT: kxorq %k0, %k7, %k0 -; AVX512F-32-NEXT: kshiftlq $63, %k0, %k0 -; AVX512F-32-NEXT: kshiftrq $8, %k0, %k0 -; AVX512F-32-NEXT: kxorq %k6, %k0, %k0 -; AVX512F-32-NEXT: kshiftrq $56, %k0, %k6 +; AVX512F-32-NEXT: kxorq %k2, %k7, %k2 +; AVX512F-32-NEXT: kshiftlq $63, %k2, %k2 +; AVX512F-32-NEXT: kshiftrq $8, %k2, %k2 +; AVX512F-32-NEXT: kxorq %k6, %k2, %k2 +; AVX512F-32-NEXT: kshiftrq $56, %k2, %k6 ; AVX512F-32-NEXT: kxorq %k1, %k6, %k1 ; AVX512F-32-NEXT: kshiftlq $63, %k1, %k1 ; AVX512F-32-NEXT: kshiftrq $7, %k1, %k1 -; AVX512F-32-NEXT: kxorq %k0, %k1, %k0 -; AVX512F-32-NEXT: kshiftrq $57, %k0, %k1 ; AVX512F-32-NEXT: kxorq %k2, %k1, %k1 -; AVX512F-32-NEXT: kshiftlq $63, %k1, %k1 -; AVX512F-32-NEXT: kshiftrq $6, %k1, %k1 -; AVX512F-32-NEXT: kxorq %k0, %k1, %k0 +; AVX512F-32-NEXT: kshiftrq $57, %k1, %k2 +; AVX512F-32-NEXT: kxorq %k0, %k2, %k0 +; AVX512F-32-NEXT: kshiftlq $63, %k0, %k0 +; AVX512F-32-NEXT: kshiftrq $6, %k0, %k0 +; AVX512F-32-NEXT: kxorq %k1, %k0, %k0 ; AVX512F-32-NEXT: kshiftrq $58, %k0, %k1 ; AVX512F-32-NEXT: kxorq %k3, %k1, %k1 ; AVX512F-32-NEXT: movl %ebx, %eax @@ -3060,11 +3062,13 @@ ; AVX512F-32-NEXT: kmovq %k0, {{[0-9]+}}(%esp) ; AVX512F-32-NEXT: addl {{[0-9]+}}(%esp), %eax ; AVX512F-32-NEXT: adcl {{[0-9]+}}(%esp), %edx -; AVX512F-32-NEXT: addl %esi, %eax +; AVX512F-32-NEXT: addl %ebp, %eax ; AVX512F-32-NEXT: adcl %ebx, %edx ; AVX512F-32-NEXT: addl $60, %esp ; AVX512F-32-NEXT: popl %esi +; AVX512F-32-NEXT: popl %edi ; AVX512F-32-NEXT: popl %ebx +; AVX512F-32-NEXT: popl %ebp ; AVX512F-32-NEXT: vzeroupper ; AVX512F-32-NEXT: retl %res0 = call i64 @llvm.x86.avx512.mask.ucmp.b.512(<64 x i8> %a0, <64 x i8> %a1, i32 0, i64 %mask) Index: test/CodeGen/X86/zext-demanded.ll =================================================================== --- test/CodeGen/X86/zext-demanded.ll +++ test/CodeGen/X86/zext-demanded.ll @@ -27,51 +27,45 @@ ret i32 %z } -; TODO: We need to stop moving the and across the shift to get a movzx define i32 @test3(i32 %x) { ; CHECK-LABEL: test3: ; CHECK: # %bb.0: -; CHECK-NEXT: shrl %edi -; CHECK-NEXT: andl $127, %edi -; CHECK-NEXT: movl %edi, %eax +; CHECK-NEXT: movzbl %dil, %eax +; CHECK-NEXT: shrl %eax ; CHECK-NEXT: retq %y = and i32 %x, 255 %z = lshr i32 %y, 1 ret i32 %z } -; TODO: We need to stop moving the and across the shift to get a movzx define i16 @test4(i16 %x) { ; CHECK-LABEL: test4: ; CHECK: # %bb.0: -; CHECK-NEXT: shrl %edi -; CHECK-NEXT: andl $127, %edi -; CHECK-NEXT: movl %edi, %eax +; CHECK-NEXT: movzbl %dil, %eax +; CHECK-NEXT: shrl %eax +; CHECK-NEXT: # kill: def %ax killed %ax killed %eax ; CHECK-NEXT: retq %y = and i16 %x, 255 %z = lshr i16 %y, 1 ret i16 %z } -; TODO: We need to stop moving the and across the shift to get a movzx define i16 @test5(i16 %x) { ; CHECK-LABEL: test5: ; CHECK: # %bb.0: -; CHECK-NEXT: shrl $9, %edi -; CHECK-NEXT: andl $127, %edi -; CHECK-NEXT: movl %edi, %eax +; CHECK-NEXT: movzwl %di, %eax +; CHECK-NEXT: shrl $9, %eax +; CHECK-NEXT: # kill: def %ax killed %ax killed %eax ; CHECK-NEXT: retq %y = lshr i16 %x, 9 ret i16 %y } -; TODO: We need to stop moving the and across the shift to get a movzx define i32 @test6(i32 %x) { ; CHECK-LABEL: test6: ; CHECK: # %bb.0: -; CHECK-NEXT: shrl $9, %edi -; CHECK-NEXT: andl $127, %edi -; CHECK-NEXT: movl %edi, %eax +; CHECK-NEXT: movzwl %di, %eax +; CHECK-NEXT: shrl $9, %eax ; CHECK-NEXT: retq %y = and i32 %x, 65535 %z = lshr i32 %y, 9