5ecd363295
This reverts commit 122efef8ee9be57055d204d52c38700fe933c033. - Patch fixed to not reuse definitions from predecessors in EH landing pads. - Late review suggestions (by MaskRay) have been addressed. - M68k/pipeline.ll test updated. - Init captures added in processBlock() to avoid capturing structured bindings. - RISCV has this disabled for now. Original commit message: A new pass MachineLateInstrsCleanup is added to be run after PEI. This is a simple pass that removes redundant and identical instructions whenever found by scanning the MF once while keeping track of register definitions in a map. These instructions are typically immediate loads resulting from rematerialization, and address loads emitted by target in eliminateFrameInde(). This is enabled by default, but a target could easily disable it by means of 'disablePass(&MachineLateInstrsCleanupID);'. This late cleanup is naturally not "optimal" in removing instructions as it is done by looking at phys-regs, but still quite effective. It would be desirable to improve other parts of CodeGen and avoid these redundant instructions in the first place, but there are no ideas for this yet. Differential Revision: https://reviews.llvm.org/D123394 Reviewed By: RKSimon, foad, craig.topper, arsenm, asb
676 lines
22 KiB
LLVM
676 lines
22 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
|
|
; RUN: llc < %s -mtriple=i686-unknown-unknown | FileCheck %s --check-prefixes=X86,X86-FAST
|
|
; RUN: llc < %s -mtriple=i686-unknown-unknown -mattr=+slow-shld | FileCheck %s --check-prefixes=X86,X86-SLOW
|
|
; RUN: llc < %s -mtriple=x86_64-unknown-unknown | FileCheck %s --check-prefixes=X64,X64-FAST
|
|
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+slow-shld | FileCheck %s --check-prefixes=X64,X64-SLOW
|
|
|
|
declare i8 @llvm.fshl.i8(i8, i8, i8) nounwind readnone
|
|
declare i16 @llvm.fshl.i16(i16, i16, i16) nounwind readnone
|
|
declare i32 @llvm.fshl.i32(i32, i32, i32) nounwind readnone
|
|
declare i64 @llvm.fshl.i64(i64, i64, i64) nounwind readnone
|
|
declare i128 @llvm.fshl.i128(i128, i128, i128) nounwind readnone
|
|
|
|
;
|
|
; Variable Funnel Shift
|
|
;
|
|
|
|
define i8 @var_shift_i8(i8 %x, i8 %y, i8 %z) nounwind {
|
|
; X86-LABEL: var_shift_i8:
|
|
; X86: # %bb.0:
|
|
; X86-NEXT: movzbl {{[0-9]+}}(%esp), %ecx
|
|
; X86-NEXT: movzbl {{[0-9]+}}(%esp), %edx
|
|
; X86-NEXT: movl {{[0-9]+}}(%esp), %eax
|
|
; X86-NEXT: shll $8, %eax
|
|
; X86-NEXT: orl %edx, %eax
|
|
; X86-NEXT: andb $7, %cl
|
|
; X86-NEXT: shll %cl, %eax
|
|
; X86-NEXT: movb %ah, %al
|
|
; X86-NEXT: retl
|
|
;
|
|
; X64-LABEL: var_shift_i8:
|
|
; X64: # %bb.0:
|
|
; X64-NEXT: movl %edx, %ecx
|
|
; X64-NEXT: shll $8, %edi
|
|
; X64-NEXT: movzbl %sil, %eax
|
|
; X64-NEXT: orl %edi, %eax
|
|
; X64-NEXT: andb $7, %cl
|
|
; X64-NEXT: # kill: def $cl killed $cl killed $ecx
|
|
; X64-NEXT: shll %cl, %eax
|
|
; X64-NEXT: shrl $8, %eax
|
|
; X64-NEXT: # kill: def $al killed $al killed $eax
|
|
; X64-NEXT: retq
|
|
%tmp = tail call i8 @llvm.fshl.i8(i8 %x, i8 %y, i8 %z)
|
|
ret i8 %tmp
|
|
}
|
|
|
|
define i16 @var_shift_i16(i16 %x, i16 %y, i16 %z) nounwind {
|
|
; X86-FAST-LABEL: var_shift_i16:
|
|
; X86-FAST: # %bb.0:
|
|
; X86-FAST-NEXT: movzwl {{[0-9]+}}(%esp), %edx
|
|
; X86-FAST-NEXT: movzwl {{[0-9]+}}(%esp), %eax
|
|
; X86-FAST-NEXT: movzbl {{[0-9]+}}(%esp), %ecx
|
|
; X86-FAST-NEXT: andb $15, %cl
|
|
; X86-FAST-NEXT: shldw %cl, %dx, %ax
|
|
; X86-FAST-NEXT: retl
|
|
;
|
|
; X86-SLOW-LABEL: var_shift_i16:
|
|
; X86-SLOW: # %bb.0:
|
|
; X86-SLOW-NEXT: movzbl {{[0-9]+}}(%esp), %ecx
|
|
; X86-SLOW-NEXT: movzwl {{[0-9]+}}(%esp), %edx
|
|
; X86-SLOW-NEXT: movl {{[0-9]+}}(%esp), %eax
|
|
; X86-SLOW-NEXT: shll $16, %eax
|
|
; X86-SLOW-NEXT: orl %edx, %eax
|
|
; X86-SLOW-NEXT: andb $15, %cl
|
|
; X86-SLOW-NEXT: shll %cl, %eax
|
|
; X86-SLOW-NEXT: shrl $16, %eax
|
|
; X86-SLOW-NEXT: # kill: def $ax killed $ax killed $eax
|
|
; X86-SLOW-NEXT: retl
|
|
;
|
|
; X64-FAST-LABEL: var_shift_i16:
|
|
; X64-FAST: # %bb.0:
|
|
; X64-FAST-NEXT: movl %edx, %ecx
|
|
; X64-FAST-NEXT: movl %edi, %eax
|
|
; X64-FAST-NEXT: andb $15, %cl
|
|
; X64-FAST-NEXT: # kill: def $cl killed $cl killed $ecx
|
|
; X64-FAST-NEXT: shldw %cl, %si, %ax
|
|
; X64-FAST-NEXT: # kill: def $ax killed $ax killed $eax
|
|
; X64-FAST-NEXT: retq
|
|
;
|
|
; X64-SLOW-LABEL: var_shift_i16:
|
|
; X64-SLOW: # %bb.0:
|
|
; X64-SLOW-NEXT: movl %edx, %ecx
|
|
; X64-SLOW-NEXT: shll $16, %edi
|
|
; X64-SLOW-NEXT: movzwl %si, %eax
|
|
; X64-SLOW-NEXT: orl %edi, %eax
|
|
; X64-SLOW-NEXT: andb $15, %cl
|
|
; X64-SLOW-NEXT: # kill: def $cl killed $cl killed $ecx
|
|
; X64-SLOW-NEXT: shll %cl, %eax
|
|
; X64-SLOW-NEXT: shrl $16, %eax
|
|
; X64-SLOW-NEXT: # kill: def $ax killed $ax killed $eax
|
|
; X64-SLOW-NEXT: retq
|
|
%tmp = tail call i16 @llvm.fshl.i16(i16 %x, i16 %y, i16 %z)
|
|
ret i16 %tmp
|
|
}
|
|
|
|
define i32 @var_shift_i32(i32 %x, i32 %y, i32 %z) nounwind {
|
|
; X86-FAST-LABEL: var_shift_i32:
|
|
; X86-FAST: # %bb.0:
|
|
; X86-FAST-NEXT: movzbl {{[0-9]+}}(%esp), %ecx
|
|
; X86-FAST-NEXT: movl {{[0-9]+}}(%esp), %edx
|
|
; X86-FAST-NEXT: movl {{[0-9]+}}(%esp), %eax
|
|
; X86-FAST-NEXT: shldl %cl, %edx, %eax
|
|
; X86-FAST-NEXT: retl
|
|
;
|
|
; X86-SLOW-LABEL: var_shift_i32:
|
|
; X86-SLOW: # %bb.0:
|
|
; X86-SLOW-NEXT: movl {{[0-9]+}}(%esp), %eax
|
|
; X86-SLOW-NEXT: movzbl {{[0-9]+}}(%esp), %ecx
|
|
; X86-SLOW-NEXT: movl {{[0-9]+}}(%esp), %edx
|
|
; X86-SLOW-NEXT: shll %cl, %edx
|
|
; X86-SLOW-NEXT: notb %cl
|
|
; X86-SLOW-NEXT: shrl %eax
|
|
; X86-SLOW-NEXT: shrl %cl, %eax
|
|
; X86-SLOW-NEXT: orl %edx, %eax
|
|
; X86-SLOW-NEXT: retl
|
|
;
|
|
; X64-FAST-LABEL: var_shift_i32:
|
|
; X64-FAST: # %bb.0:
|
|
; X64-FAST-NEXT: movl %edx, %ecx
|
|
; X64-FAST-NEXT: movl %edi, %eax
|
|
; X64-FAST-NEXT: # kill: def $cl killed $cl killed $ecx
|
|
; X64-FAST-NEXT: shldl %cl, %esi, %eax
|
|
; X64-FAST-NEXT: retq
|
|
;
|
|
; X64-SLOW-LABEL: var_shift_i32:
|
|
; X64-SLOW: # %bb.0:
|
|
; X64-SLOW-NEXT: movl %edx, %ecx
|
|
; X64-SLOW-NEXT: movl %esi, %eax
|
|
; X64-SLOW-NEXT: shll %cl, %edi
|
|
; X64-SLOW-NEXT: shrl %eax
|
|
; X64-SLOW-NEXT: notb %cl
|
|
; X64-SLOW-NEXT: # kill: def $cl killed $cl killed $ecx
|
|
; X64-SLOW-NEXT: shrl %cl, %eax
|
|
; X64-SLOW-NEXT: orl %edi, %eax
|
|
; X64-SLOW-NEXT: retq
|
|
%tmp = tail call i32 @llvm.fshl.i32(i32 %x, i32 %y, i32 %z)
|
|
ret i32 %tmp
|
|
}
|
|
|
|
define i32 @var_shift_i32_optsize(i32 %x, i32 %y, i32 %z) nounwind optsize {
|
|
; X86-LABEL: var_shift_i32_optsize:
|
|
; X86: # %bb.0:
|
|
; X86-NEXT: movb {{[0-9]+}}(%esp), %cl
|
|
; X86-NEXT: movl {{[0-9]+}}(%esp), %edx
|
|
; X86-NEXT: movl {{[0-9]+}}(%esp), %eax
|
|
; X86-NEXT: shldl %cl, %edx, %eax
|
|
; X86-NEXT: retl
|
|
;
|
|
; X64-LABEL: var_shift_i32_optsize:
|
|
; X64: # %bb.0:
|
|
; X64-NEXT: movl %edx, %ecx
|
|
; X64-NEXT: movl %edi, %eax
|
|
; X64-NEXT: # kill: def $cl killed $cl killed $ecx
|
|
; X64-NEXT: shldl %cl, %esi, %eax
|
|
; X64-NEXT: retq
|
|
%tmp = tail call i32 @llvm.fshl.i32(i32 %x, i32 %y, i32 %z)
|
|
ret i32 %tmp
|
|
}
|
|
|
|
define i32 @var_shift_i32_pgso(i32 %x, i32 %y, i32 %z) nounwind !prof !14 {
|
|
; X86-LABEL: var_shift_i32_pgso:
|
|
; X86: # %bb.0:
|
|
; X86-NEXT: movb {{[0-9]+}}(%esp), %cl
|
|
; X86-NEXT: movl {{[0-9]+}}(%esp), %edx
|
|
; X86-NEXT: movl {{[0-9]+}}(%esp), %eax
|
|
; X86-NEXT: shldl %cl, %edx, %eax
|
|
; X86-NEXT: retl
|
|
;
|
|
; X64-LABEL: var_shift_i32_pgso:
|
|
; X64: # %bb.0:
|
|
; X64-NEXT: movl %edx, %ecx
|
|
; X64-NEXT: movl %edi, %eax
|
|
; X64-NEXT: # kill: def $cl killed $cl killed $ecx
|
|
; X64-NEXT: shldl %cl, %esi, %eax
|
|
; X64-NEXT: retq
|
|
%tmp = tail call i32 @llvm.fshl.i32(i32 %x, i32 %y, i32 %z)
|
|
ret i32 %tmp
|
|
}
|
|
|
|
define i64 @var_shift_i64(i64 %x, i64 %y, i64 %z) nounwind {
|
|
; X86-FAST-LABEL: var_shift_i64:
|
|
; X86-FAST: # %bb.0:
|
|
; X86-FAST-NEXT: pushl %edi
|
|
; X86-FAST-NEXT: pushl %esi
|
|
; X86-FAST-NEXT: movl {{[0-9]+}}(%esp), %esi
|
|
; X86-FAST-NEXT: movl {{[0-9]+}}(%esp), %edx
|
|
; X86-FAST-NEXT: movl {{[0-9]+}}(%esp), %ecx
|
|
; X86-FAST-NEXT: testb $32, %cl
|
|
; X86-FAST-NEXT: jne .LBB5_1
|
|
; X86-FAST-NEXT: # %bb.2:
|
|
; X86-FAST-NEXT: movl %edx, %edi
|
|
; X86-FAST-NEXT: movl {{[0-9]+}}(%esp), %edx
|
|
; X86-FAST-NEXT: jmp .LBB5_3
|
|
; X86-FAST-NEXT: .LBB5_1:
|
|
; X86-FAST-NEXT: movl %esi, %edi
|
|
; X86-FAST-NEXT: movl {{[0-9]+}}(%esp), %esi
|
|
; X86-FAST-NEXT: .LBB5_3:
|
|
; X86-FAST-NEXT: movl %edi, %eax
|
|
; X86-FAST-NEXT: shldl %cl, %esi, %eax
|
|
; X86-FAST-NEXT: # kill: def $cl killed $cl killed $ecx
|
|
; X86-FAST-NEXT: shldl %cl, %edi, %edx
|
|
; X86-FAST-NEXT: popl %esi
|
|
; X86-FAST-NEXT: popl %edi
|
|
; X86-FAST-NEXT: retl
|
|
;
|
|
; X86-SLOW-LABEL: var_shift_i64:
|
|
; X86-SLOW: # %bb.0:
|
|
; X86-SLOW-NEXT: pushl %ebx
|
|
; X86-SLOW-NEXT: pushl %edi
|
|
; X86-SLOW-NEXT: pushl %esi
|
|
; X86-SLOW-NEXT: movl {{[0-9]+}}(%esp), %eax
|
|
; X86-SLOW-NEXT: movl {{[0-9]+}}(%esp), %edx
|
|
; X86-SLOW-NEXT: movl {{[0-9]+}}(%esp), %ebx
|
|
; X86-SLOW-NEXT: testb $32, %bl
|
|
; X86-SLOW-NEXT: jne .LBB5_1
|
|
; X86-SLOW-NEXT: # %bb.2:
|
|
; X86-SLOW-NEXT: movl %edx, %esi
|
|
; X86-SLOW-NEXT: movl {{[0-9]+}}(%esp), %edx
|
|
; X86-SLOW-NEXT: jmp .LBB5_3
|
|
; X86-SLOW-NEXT: .LBB5_1:
|
|
; X86-SLOW-NEXT: movl %eax, %esi
|
|
; X86-SLOW-NEXT: movl {{[0-9]+}}(%esp), %eax
|
|
; X86-SLOW-NEXT: .LBB5_3:
|
|
; X86-SLOW-NEXT: movl %esi, %edi
|
|
; X86-SLOW-NEXT: movl %ebx, %ecx
|
|
; X86-SLOW-NEXT: shll %cl, %edi
|
|
; X86-SLOW-NEXT: shrl %eax
|
|
; X86-SLOW-NEXT: notb %cl
|
|
; X86-SLOW-NEXT: shrl %cl, %eax
|
|
; X86-SLOW-NEXT: orl %edi, %eax
|
|
; X86-SLOW-NEXT: shrl %esi
|
|
; X86-SLOW-NEXT: shrl %cl, %esi
|
|
; X86-SLOW-NEXT: movl %ebx, %ecx
|
|
; X86-SLOW-NEXT: shll %cl, %edx
|
|
; X86-SLOW-NEXT: orl %esi, %edx
|
|
; X86-SLOW-NEXT: popl %esi
|
|
; X86-SLOW-NEXT: popl %edi
|
|
; X86-SLOW-NEXT: popl %ebx
|
|
; X86-SLOW-NEXT: retl
|
|
;
|
|
; X64-FAST-LABEL: var_shift_i64:
|
|
; X64-FAST: # %bb.0:
|
|
; X64-FAST-NEXT: movq %rdx, %rcx
|
|
; X64-FAST-NEXT: movq %rdi, %rax
|
|
; X64-FAST-NEXT: # kill: def $cl killed $cl killed $rcx
|
|
; X64-FAST-NEXT: shldq %cl, %rsi, %rax
|
|
; X64-FAST-NEXT: retq
|
|
;
|
|
; X64-SLOW-LABEL: var_shift_i64:
|
|
; X64-SLOW: # %bb.0:
|
|
; X64-SLOW-NEXT: movq %rdx, %rcx
|
|
; X64-SLOW-NEXT: movq %rsi, %rax
|
|
; X64-SLOW-NEXT: shlq %cl, %rdi
|
|
; X64-SLOW-NEXT: shrq %rax
|
|
; X64-SLOW-NEXT: notb %cl
|
|
; X64-SLOW-NEXT: # kill: def $cl killed $cl killed $rcx
|
|
; X64-SLOW-NEXT: shrq %cl, %rax
|
|
; X64-SLOW-NEXT: orq %rdi, %rax
|
|
; X64-SLOW-NEXT: retq
|
|
%tmp = tail call i64 @llvm.fshl.i64(i64 %x, i64 %y, i64 %z)
|
|
ret i64 %tmp
|
|
}
|
|
|
|
define i128 @var_shift_i128(i128 %x, i128 %y, i128 %z) nounwind {
|
|
; X86-FAST-LABEL: var_shift_i128:
|
|
; X86-FAST: # %bb.0:
|
|
; X86-FAST-NEXT: pushl %ebp
|
|
; X86-FAST-NEXT: pushl %ebx
|
|
; X86-FAST-NEXT: pushl %edi
|
|
; X86-FAST-NEXT: pushl %esi
|
|
; X86-FAST-NEXT: movl {{[0-9]+}}(%esp), %edx
|
|
; X86-FAST-NEXT: movl {{[0-9]+}}(%esp), %esi
|
|
; X86-FAST-NEXT: movl {{[0-9]+}}(%esp), %ebx
|
|
; X86-FAST-NEXT: movl {{[0-9]+}}(%esp), %edi
|
|
; X86-FAST-NEXT: movl {{[0-9]+}}(%esp), %ecx
|
|
; X86-FAST-NEXT: testb $64, %cl
|
|
; X86-FAST-NEXT: jne .LBB6_1
|
|
; X86-FAST-NEXT: # %bb.2:
|
|
; X86-FAST-NEXT: movl %edi, %eax
|
|
; X86-FAST-NEXT: movl %esi, %edi
|
|
; X86-FAST-NEXT: movl {{[0-9]+}}(%esp), %esi
|
|
; X86-FAST-NEXT: movl %ebx, %ebp
|
|
; X86-FAST-NEXT: movl %edx, %ebx
|
|
; X86-FAST-NEXT: movl {{[0-9]+}}(%esp), %edx
|
|
; X86-FAST-NEXT: testb $32, %cl
|
|
; X86-FAST-NEXT: je .LBB6_5
|
|
; X86-FAST-NEXT: .LBB6_4:
|
|
; X86-FAST-NEXT: movl %edx, %esi
|
|
; X86-FAST-NEXT: movl %edi, %edx
|
|
; X86-FAST-NEXT: movl %ebx, %edi
|
|
; X86-FAST-NEXT: movl %eax, %ebx
|
|
; X86-FAST-NEXT: jmp .LBB6_6
|
|
; X86-FAST-NEXT: .LBB6_1:
|
|
; X86-FAST-NEXT: movl {{[0-9]+}}(%esp), %ebp
|
|
; X86-FAST-NEXT: movl {{[0-9]+}}(%esp), %eax
|
|
; X86-FAST-NEXT: testb $32, %cl
|
|
; X86-FAST-NEXT: jne .LBB6_4
|
|
; X86-FAST-NEXT: .LBB6_5:
|
|
; X86-FAST-NEXT: movl %eax, %ebp
|
|
; X86-FAST-NEXT: .LBB6_6:
|
|
; X86-FAST-NEXT: movl %ebx, %eax
|
|
; X86-FAST-NEXT: shldl %cl, %ebp, %eax
|
|
; X86-FAST-NEXT: movl %edi, %ebp
|
|
; X86-FAST-NEXT: shldl %cl, %ebx, %ebp
|
|
; X86-FAST-NEXT: movl %edx, %ebx
|
|
; X86-FAST-NEXT: shldl %cl, %edi, %ebx
|
|
; X86-FAST-NEXT: # kill: def $cl killed $cl killed $ecx
|
|
; X86-FAST-NEXT: shldl %cl, %edx, %esi
|
|
; X86-FAST-NEXT: movl {{[0-9]+}}(%esp), %ecx
|
|
; X86-FAST-NEXT: movl %esi, 12(%ecx)
|
|
; X86-FAST-NEXT: movl %ebx, 8(%ecx)
|
|
; X86-FAST-NEXT: movl %ebp, 4(%ecx)
|
|
; X86-FAST-NEXT: movl %eax, (%ecx)
|
|
; X86-FAST-NEXT: movl %ecx, %eax
|
|
; X86-FAST-NEXT: popl %esi
|
|
; X86-FAST-NEXT: popl %edi
|
|
; X86-FAST-NEXT: popl %ebx
|
|
; X86-FAST-NEXT: popl %ebp
|
|
; X86-FAST-NEXT: retl $4
|
|
;
|
|
; X86-SLOW-LABEL: var_shift_i128:
|
|
; X86-SLOW: # %bb.0:
|
|
; X86-SLOW-NEXT: pushl %ebp
|
|
; X86-SLOW-NEXT: pushl %ebx
|
|
; X86-SLOW-NEXT: pushl %edi
|
|
; X86-SLOW-NEXT: pushl %esi
|
|
; X86-SLOW-NEXT: pushl %eax
|
|
; X86-SLOW-NEXT: movl {{[0-9]+}}(%esp), %esi
|
|
; X86-SLOW-NEXT: movl {{[0-9]+}}(%esp), %edi
|
|
; X86-SLOW-NEXT: movl {{[0-9]+}}(%esp), %edx
|
|
; X86-SLOW-NEXT: movl {{[0-9]+}}(%esp), %ebp
|
|
; X86-SLOW-NEXT: movl {{[0-9]+}}(%esp), %eax
|
|
; X86-SLOW-NEXT: testb $64, %al
|
|
; X86-SLOW-NEXT: jne .LBB6_1
|
|
; X86-SLOW-NEXT: # %bb.2:
|
|
; X86-SLOW-NEXT: movl %ebp, %ecx
|
|
; X86-SLOW-NEXT: movl %edi, %ebp
|
|
; X86-SLOW-NEXT: movl {{[0-9]+}}(%esp), %edi
|
|
; X86-SLOW-NEXT: movl %edx, %ebx
|
|
; X86-SLOW-NEXT: movl %esi, %edx
|
|
; X86-SLOW-NEXT: movl {{[0-9]+}}(%esp), %esi
|
|
; X86-SLOW-NEXT: testb $32, %al
|
|
; X86-SLOW-NEXT: je .LBB6_5
|
|
; X86-SLOW-NEXT: .LBB6_4:
|
|
; X86-SLOW-NEXT: movl %esi, (%esp) # 4-byte Spill
|
|
; X86-SLOW-NEXT: movl %ebp, %esi
|
|
; X86-SLOW-NEXT: movl %edx, %ebp
|
|
; X86-SLOW-NEXT: movl %ecx, %edx
|
|
; X86-SLOW-NEXT: jmp .LBB6_6
|
|
; X86-SLOW-NEXT: .LBB6_1:
|
|
; X86-SLOW-NEXT: movl {{[0-9]+}}(%esp), %ebx
|
|
; X86-SLOW-NEXT: movl {{[0-9]+}}(%esp), %ecx
|
|
; X86-SLOW-NEXT: testb $32, %al
|
|
; X86-SLOW-NEXT: jne .LBB6_4
|
|
; X86-SLOW-NEXT: .LBB6_5:
|
|
; X86-SLOW-NEXT: movl %ecx, %ebx
|
|
; X86-SLOW-NEXT: movl %edi, (%esp) # 4-byte Spill
|
|
; X86-SLOW-NEXT: .LBB6_6:
|
|
; X86-SLOW-NEXT: movl %edx, %edi
|
|
; X86-SLOW-NEXT: movl %eax, %ecx
|
|
; X86-SLOW-NEXT: shll %cl, %edi
|
|
; X86-SLOW-NEXT: shrl %ebx
|
|
; X86-SLOW-NEXT: movb %al, %ch
|
|
; X86-SLOW-NEXT: notb %ch
|
|
; X86-SLOW-NEXT: movb %ch, %cl
|
|
; X86-SLOW-NEXT: shrl %cl, %ebx
|
|
; X86-SLOW-NEXT: orl %edi, %ebx
|
|
; X86-SLOW-NEXT: movl %ebp, %edi
|
|
; X86-SLOW-NEXT: movb %al, %cl
|
|
; X86-SLOW-NEXT: shll %cl, %edi
|
|
; X86-SLOW-NEXT: shrl %edx
|
|
; X86-SLOW-NEXT: movb %ch, %cl
|
|
; X86-SLOW-NEXT: shrl %cl, %edx
|
|
; X86-SLOW-NEXT: orl %edi, %edx
|
|
; X86-SLOW-NEXT: movl %esi, %edi
|
|
; X86-SLOW-NEXT: movb %al, %cl
|
|
; X86-SLOW-NEXT: shll %cl, %edi
|
|
; X86-SLOW-NEXT: shrl %ebp
|
|
; X86-SLOW-NEXT: movb %ch, %cl
|
|
; X86-SLOW-NEXT: shrl %cl, %ebp
|
|
; X86-SLOW-NEXT: orl %edi, %ebp
|
|
; X86-SLOW-NEXT: movb %al, %cl
|
|
; X86-SLOW-NEXT: movl (%esp), %eax # 4-byte Reload
|
|
; X86-SLOW-NEXT: shll %cl, %eax
|
|
; X86-SLOW-NEXT: shrl %esi
|
|
; X86-SLOW-NEXT: movb %ch, %cl
|
|
; X86-SLOW-NEXT: shrl %cl, %esi
|
|
; X86-SLOW-NEXT: orl %eax, %esi
|
|
; X86-SLOW-NEXT: movl {{[0-9]+}}(%esp), %eax
|
|
; X86-SLOW-NEXT: movl %esi, 12(%eax)
|
|
; X86-SLOW-NEXT: movl %ebp, 8(%eax)
|
|
; X86-SLOW-NEXT: movl %edx, 4(%eax)
|
|
; X86-SLOW-NEXT: movl %ebx, (%eax)
|
|
; X86-SLOW-NEXT: addl $4, %esp
|
|
; X86-SLOW-NEXT: popl %esi
|
|
; X86-SLOW-NEXT: popl %edi
|
|
; X86-SLOW-NEXT: popl %ebx
|
|
; X86-SLOW-NEXT: popl %ebp
|
|
; X86-SLOW-NEXT: retl $4
|
|
;
|
|
; X64-FAST-LABEL: var_shift_i128:
|
|
; X64-FAST: # %bb.0:
|
|
; X64-FAST-NEXT: testb $64, %r8b
|
|
; X64-FAST-NEXT: cmovneq %rdi, %rsi
|
|
; X64-FAST-NEXT: cmoveq %rcx, %rdx
|
|
; X64-FAST-NEXT: cmovneq %rcx, %rdi
|
|
; X64-FAST-NEXT: movq %rdi, %rax
|
|
; X64-FAST-NEXT: movl %r8d, %ecx
|
|
; X64-FAST-NEXT: shldq %cl, %rdx, %rax
|
|
; X64-FAST-NEXT: shldq %cl, %rdi, %rsi
|
|
; X64-FAST-NEXT: movq %rsi, %rdx
|
|
; X64-FAST-NEXT: retq
|
|
;
|
|
; X64-SLOW-LABEL: var_shift_i128:
|
|
; X64-SLOW: # %bb.0:
|
|
; X64-SLOW-NEXT: testb $64, %r8b
|
|
; X64-SLOW-NEXT: cmovneq %rdi, %rsi
|
|
; X64-SLOW-NEXT: cmoveq %rcx, %rdx
|
|
; X64-SLOW-NEXT: cmovneq %rcx, %rdi
|
|
; X64-SLOW-NEXT: movq %rdi, %rax
|
|
; X64-SLOW-NEXT: movl %r8d, %ecx
|
|
; X64-SLOW-NEXT: shlq %cl, %rax
|
|
; X64-SLOW-NEXT: shrq %rdx
|
|
; X64-SLOW-NEXT: movl %r8d, %r9d
|
|
; X64-SLOW-NEXT: notb %r9b
|
|
; X64-SLOW-NEXT: movl %r9d, %ecx
|
|
; X64-SLOW-NEXT: shrq %cl, %rdx
|
|
; X64-SLOW-NEXT: orq %rdx, %rax
|
|
; X64-SLOW-NEXT: movl %r8d, %ecx
|
|
; X64-SLOW-NEXT: shlq %cl, %rsi
|
|
; X64-SLOW-NEXT: shrq %rdi
|
|
; X64-SLOW-NEXT: movl %r9d, %ecx
|
|
; X64-SLOW-NEXT: shrq %cl, %rdi
|
|
; X64-SLOW-NEXT: orq %rsi, %rdi
|
|
; X64-SLOW-NEXT: movq %rdi, %rdx
|
|
; X64-SLOW-NEXT: retq
|
|
%tmp = tail call i128 @llvm.fshl.i128(i128 %x, i128 %y, i128 %z)
|
|
ret i128 %tmp
|
|
}
|
|
|
|
;
|
|
; Const Funnel Shift
|
|
;
|
|
|
|
define i8 @const_shift_i8(i8 %x, i8 %y) nounwind {
|
|
; X86-LABEL: const_shift_i8:
|
|
; X86: # %bb.0:
|
|
; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax
|
|
; X86-NEXT: movzbl {{[0-9]+}}(%esp), %ecx
|
|
; X86-NEXT: shrb %cl
|
|
; X86-NEXT: shlb $7, %al
|
|
; X86-NEXT: orb %cl, %al
|
|
; X86-NEXT: retl
|
|
;
|
|
; X64-LABEL: const_shift_i8:
|
|
; X64: # %bb.0:
|
|
; X64-NEXT: # kill: def $esi killed $esi def $rsi
|
|
; X64-NEXT: # kill: def $edi killed $edi def $rdi
|
|
; X64-NEXT: shrb %sil
|
|
; X64-NEXT: shlb $7, %dil
|
|
; X64-NEXT: leal (%rdi,%rsi), %eax
|
|
; X64-NEXT: # kill: def $al killed $al killed $eax
|
|
; X64-NEXT: retq
|
|
%tmp = tail call i8 @llvm.fshl.i8(i8 %x, i8 %y, i8 7)
|
|
ret i8 %tmp
|
|
}
|
|
|
|
define i16 @const_shift_i16(i16 %x, i16 %y) nounwind {
|
|
; X86-FAST-LABEL: const_shift_i16:
|
|
; X86-FAST: # %bb.0:
|
|
; X86-FAST-NEXT: movzwl {{[0-9]+}}(%esp), %ecx
|
|
; X86-FAST-NEXT: movzwl {{[0-9]+}}(%esp), %eax
|
|
; X86-FAST-NEXT: shldw $7, %cx, %ax
|
|
; X86-FAST-NEXT: retl
|
|
;
|
|
; X86-SLOW-LABEL: const_shift_i16:
|
|
; X86-SLOW: # %bb.0:
|
|
; X86-SLOW-NEXT: movl {{[0-9]+}}(%esp), %eax
|
|
; X86-SLOW-NEXT: movzwl {{[0-9]+}}(%esp), %ecx
|
|
; X86-SLOW-NEXT: shrl $9, %ecx
|
|
; X86-SLOW-NEXT: shll $7, %eax
|
|
; X86-SLOW-NEXT: orl %ecx, %eax
|
|
; X86-SLOW-NEXT: # kill: def $ax killed $ax killed $eax
|
|
; X86-SLOW-NEXT: retl
|
|
;
|
|
; X64-FAST-LABEL: const_shift_i16:
|
|
; X64-FAST: # %bb.0:
|
|
; X64-FAST-NEXT: movl %edi, %eax
|
|
; X64-FAST-NEXT: shldw $7, %si, %ax
|
|
; X64-FAST-NEXT: # kill: def $ax killed $ax killed $eax
|
|
; X64-FAST-NEXT: retq
|
|
;
|
|
; X64-SLOW-LABEL: const_shift_i16:
|
|
; X64-SLOW: # %bb.0:
|
|
; X64-SLOW-NEXT: movzwl %si, %eax
|
|
; X64-SLOW-NEXT: shll $7, %edi
|
|
; X64-SLOW-NEXT: shrl $9, %eax
|
|
; X64-SLOW-NEXT: orl %edi, %eax
|
|
; X64-SLOW-NEXT: # kill: def $ax killed $ax killed $eax
|
|
; X64-SLOW-NEXT: retq
|
|
%tmp = tail call i16 @llvm.fshl.i16(i16 %x, i16 %y, i16 7)
|
|
ret i16 %tmp
|
|
}
|
|
|
|
define i32 @const_shift_i32(i32 %x, i32 %y) nounwind {
|
|
; X86-FAST-LABEL: const_shift_i32:
|
|
; X86-FAST: # %bb.0:
|
|
; X86-FAST-NEXT: movl {{[0-9]+}}(%esp), %ecx
|
|
; X86-FAST-NEXT: movl {{[0-9]+}}(%esp), %eax
|
|
; X86-FAST-NEXT: shldl $7, %ecx, %eax
|
|
; X86-FAST-NEXT: retl
|
|
;
|
|
; X86-SLOW-LABEL: const_shift_i32:
|
|
; X86-SLOW: # %bb.0:
|
|
; X86-SLOW-NEXT: movl {{[0-9]+}}(%esp), %eax
|
|
; X86-SLOW-NEXT: movl {{[0-9]+}}(%esp), %ecx
|
|
; X86-SLOW-NEXT: shrl $25, %ecx
|
|
; X86-SLOW-NEXT: shll $7, %eax
|
|
; X86-SLOW-NEXT: orl %ecx, %eax
|
|
; X86-SLOW-NEXT: retl
|
|
;
|
|
; X64-FAST-LABEL: const_shift_i32:
|
|
; X64-FAST: # %bb.0:
|
|
; X64-FAST-NEXT: movl %edi, %eax
|
|
; X64-FAST-NEXT: shldl $7, %esi, %eax
|
|
; X64-FAST-NEXT: retq
|
|
;
|
|
; X64-SLOW-LABEL: const_shift_i32:
|
|
; X64-SLOW: # %bb.0:
|
|
; X64-SLOW-NEXT: # kill: def $esi killed $esi def $rsi
|
|
; X64-SLOW-NEXT: # kill: def $edi killed $edi def $rdi
|
|
; X64-SLOW-NEXT: shrl $25, %esi
|
|
; X64-SLOW-NEXT: shll $7, %edi
|
|
; X64-SLOW-NEXT: leal (%rdi,%rsi), %eax
|
|
; X64-SLOW-NEXT: retq
|
|
%tmp = tail call i32 @llvm.fshl.i32(i32 %x, i32 %y, i32 7)
|
|
ret i32 %tmp
|
|
}
|
|
|
|
define i64 @const_shift_i64(i64 %x, i64 %y) nounwind {
|
|
; X86-FAST-LABEL: const_shift_i64:
|
|
; X86-FAST: # %bb.0:
|
|
; X86-FAST-NEXT: movl {{[0-9]+}}(%esp), %eax
|
|
; X86-FAST-NEXT: movl {{[0-9]+}}(%esp), %ecx
|
|
; X86-FAST-NEXT: movl {{[0-9]+}}(%esp), %edx
|
|
; X86-FAST-NEXT: shrdl $25, %ecx, %eax
|
|
; X86-FAST-NEXT: shldl $7, %ecx, %edx
|
|
; X86-FAST-NEXT: retl
|
|
;
|
|
; X86-SLOW-LABEL: const_shift_i64:
|
|
; X86-SLOW: # %bb.0:
|
|
; X86-SLOW-NEXT: pushl %esi
|
|
; X86-SLOW-NEXT: movl {{[0-9]+}}(%esp), %ecx
|
|
; X86-SLOW-NEXT: movl {{[0-9]+}}(%esp), %edx
|
|
; X86-SLOW-NEXT: movl {{[0-9]+}}(%esp), %esi
|
|
; X86-SLOW-NEXT: shrl $25, %esi
|
|
; X86-SLOW-NEXT: movl %ecx, %eax
|
|
; X86-SLOW-NEXT: shll $7, %eax
|
|
; X86-SLOW-NEXT: orl %esi, %eax
|
|
; X86-SLOW-NEXT: shrl $25, %ecx
|
|
; X86-SLOW-NEXT: shll $7, %edx
|
|
; X86-SLOW-NEXT: orl %ecx, %edx
|
|
; X86-SLOW-NEXT: popl %esi
|
|
; X86-SLOW-NEXT: retl
|
|
;
|
|
; X64-FAST-LABEL: const_shift_i64:
|
|
; X64-FAST: # %bb.0:
|
|
; X64-FAST-NEXT: movq %rdi, %rax
|
|
; X64-FAST-NEXT: shldq $7, %rsi, %rax
|
|
; X64-FAST-NEXT: retq
|
|
;
|
|
; X64-SLOW-LABEL: const_shift_i64:
|
|
; X64-SLOW: # %bb.0:
|
|
; X64-SLOW-NEXT: shrq $57, %rsi
|
|
; X64-SLOW-NEXT: shlq $7, %rdi
|
|
; X64-SLOW-NEXT: leaq (%rdi,%rsi), %rax
|
|
; X64-SLOW-NEXT: retq
|
|
%tmp = tail call i64 @llvm.fshl.i64(i64 %x, i64 %y, i64 7)
|
|
ret i64 %tmp
|
|
}
|
|
|
|
;
|
|
; Combine Consecutive Loads
|
|
;
|
|
|
|
define i8 @combine_fshl_load_i8(ptr %p) nounwind {
|
|
; X86-LABEL: combine_fshl_load_i8:
|
|
; X86: # %bb.0:
|
|
; X86-NEXT: movl {{[0-9]+}}(%esp), %eax
|
|
; X86-NEXT: movzbl 1(%eax), %eax
|
|
; X86-NEXT: retl
|
|
;
|
|
; X64-LABEL: combine_fshl_load_i8:
|
|
; X64: # %bb.0:
|
|
; X64-NEXT: movzbl 1(%rdi), %eax
|
|
; X64-NEXT: retq
|
|
%p1 = getelementptr i8, ptr %p, i32 1
|
|
%ld0 = load i8, ptr%p
|
|
%ld1 = load i8, ptr%p1
|
|
%res = call i8 @llvm.fshl.i8(i8 %ld1, i8 %ld0, i8 8)
|
|
ret i8 %res
|
|
}
|
|
|
|
define i16 @combine_fshl_load_i16(ptr %p) nounwind {
|
|
; X86-LABEL: combine_fshl_load_i16:
|
|
; X86: # %bb.0:
|
|
; X86-NEXT: movl {{[0-9]+}}(%esp), %eax
|
|
; X86-NEXT: movzwl 1(%eax), %eax
|
|
; X86-NEXT: retl
|
|
;
|
|
; X64-LABEL: combine_fshl_load_i16:
|
|
; X64: # %bb.0:
|
|
; X64-NEXT: movzwl 1(%rdi), %eax
|
|
; X64-NEXT: retq
|
|
%p1 = getelementptr i16, ptr %p, i32 1
|
|
%ld0 = load i16, ptr%p
|
|
%ld1 = load i16, ptr%p1
|
|
%res = call i16 @llvm.fshl.i16(i16 %ld1, i16 %ld0, i16 8)
|
|
ret i16 %res
|
|
}
|
|
|
|
define i32 @combine_fshl_load_i32(ptr %p) nounwind {
|
|
; X86-LABEL: combine_fshl_load_i32:
|
|
; X86: # %bb.0:
|
|
; X86-NEXT: movl {{[0-9]+}}(%esp), %eax
|
|
; X86-NEXT: movl 11(%eax), %eax
|
|
; X86-NEXT: retl
|
|
;
|
|
; X64-LABEL: combine_fshl_load_i32:
|
|
; X64: # %bb.0:
|
|
; X64-NEXT: movl 11(%rdi), %eax
|
|
; X64-NEXT: retq
|
|
%p0 = getelementptr i32, ptr %p, i32 2
|
|
%p1 = getelementptr i32, ptr %p, i32 3
|
|
%ld0 = load i32, ptr%p0
|
|
%ld1 = load i32, ptr%p1
|
|
%res = call i32 @llvm.fshl.i32(i32 %ld1, i32 %ld0, i32 8)
|
|
ret i32 %res
|
|
}
|
|
|
|
define i64 @combine_fshl_load_i64(ptr %p) nounwind {
|
|
; X86-LABEL: combine_fshl_load_i64:
|
|
; X86: # %bb.0:
|
|
; X86-NEXT: movl {{[0-9]+}}(%esp), %ecx
|
|
; X86-NEXT: movl 13(%ecx), %eax
|
|
; X86-NEXT: movl 17(%ecx), %edx
|
|
; X86-NEXT: retl
|
|
;
|
|
; X64-LABEL: combine_fshl_load_i64:
|
|
; X64: # %bb.0:
|
|
; X64-NEXT: movq 13(%rdi), %rax
|
|
; X64-NEXT: retq
|
|
%p0 = getelementptr i64, ptr %p, i64 1
|
|
%p1 = getelementptr i64, ptr %p, i64 2
|
|
%ld0 = load i64, ptr%p0
|
|
%ld1 = load i64, ptr%p1
|
|
%res = call i64 @llvm.fshl.i64(i64 %ld1, i64 %ld0, i64 24)
|
|
ret i64 %res
|
|
}
|
|
|
|
!llvm.module.flags = !{!0}
|
|
!0 = !{i32 1, !"ProfileSummary", !1}
|
|
!1 = !{!2, !3, !4, !5, !6, !7, !8, !9}
|
|
!2 = !{!"ProfileFormat", !"InstrProf"}
|
|
!3 = !{!"TotalCount", i64 10000}
|
|
!4 = !{!"MaxCount", i64 10}
|
|
!5 = !{!"MaxInternalCount", i64 1}
|
|
!6 = !{!"MaxFunctionCount", i64 1000}
|
|
!7 = !{!"NumCounts", i64 3}
|
|
!8 = !{!"NumFunctions", i64 3}
|
|
!9 = !{!"DetailedSummary", !10}
|
|
!10 = !{!11, !12, !13}
|
|
!11 = !{i32 10000, i64 100, i32 1}
|
|
!12 = !{i32 999000, i64 100, i32 1}
|
|
!13 = !{i32 999999, i64 1, i32 2}
|
|
!14 = !{!"function_entry_count", i64 0}
|