; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 5
; RUN: llc < %s -mtriple=x86_64-- -mcpu=k8 | FileCheck %s --check-prefixes=X64
; RUN: llc < %s -mtriple=x86_64-- -mcpu=opteron | FileCheck %s --check-prefixes=X64
; RUN: llc < %s -mtriple=x86_64-- -mcpu=athlon64 | FileCheck %s --check-prefixes=X64
; RUN: llc < %s -mtriple=x86_64-- -mcpu=athlon-fx | FileCheck %s --check-prefixes=X64
; RUN: llc < %s -mtriple=x86_64-- -mcpu=k8-sse3 | FileCheck %s --check-prefixes=X64
; RUN: llc < %s -mtriple=x86_64-- -mcpu=opteron-sse3 | FileCheck %s --check-prefixes=X64
; RUN: llc < %s -mtriple=x86_64-- -mcpu=athlon64-sse3 | FileCheck %s --check-prefixes=X64
; RUN: llc < %s -mtriple=x86_64-- -mcpu=amdfam10 | FileCheck %s --check-prefixes=X64
; RUN: llc < %s -mtriple=x86_64-- -mcpu=btver1 | FileCheck %s --check-prefixes=X64
; RUN: llc < %s -mtriple=x86_64-- -mcpu=btver2 | FileCheck %s --check-prefixes=BMI
; RUN: llc < %s -mtriple=x86_64-- -mcpu=bdver1 | FileCheck %s --check-prefixes=BMI
; RUN: llc < %s -mtriple=x86_64-- -mcpu=bdver2 | FileCheck %s --check-prefixes=BMI
; RUN: llc < %s -mtriple=x86_64-- -mcpu=bdver3 | FileCheck %s --check-prefixes=BMI
; RUN: llc < %s -mtriple=x86_64-- -mcpu=bdver4 | FileCheck %s --check-prefixes=BMI2-SLOW
; RUN: llc < %s -mtriple=x86_64-- -mcpu=znver1 | FileCheck %s --check-prefixes=BMI2-SLOW
; RUN: llc < %s -mtriple=x86_64-- -mcpu=znver2 | FileCheck %s --check-prefixes=BMI2-SLOW
; RUN: llc < %s -mtriple=x86_64-- -mcpu=znver3 | FileCheck %s --check-prefixes=BMI2-FAST
; RUN: llc < %s -mtriple=x86_64-- -mcpu=znver4 | FileCheck %s --check-prefixes=BMI2-FAST
; RUN: llc < %s -mtriple=x86_64-- -mcpu=znver5 | FileCheck %s --check-prefixes=BMI2-FAST
; RUN: llc < %s -mtriple=x86_64-- -mcpu=znver6 | FileCheck %s --check-prefixes=BMI2-FAST

; Verify that for the X86_64 processors that are known to have poor latency
; double precision shift instructions we do not generate 'shld' or 'shrd'
; instructions.

;uint64_t lshift(uint64_t a, uint64_t b, int c)
;{
;    return (a << c) | (b >> (64-c));
;}

define i64 @lshift(i64 %a, i64 %b, i32 %c) nounwind readnone {
; X64-LABEL: lshift:
; X64:       # %bb.0: # %entry
; X64-NEXT:    movl %edx, %ecx
; X64-NEXT:    movq %rsi, %rax
; X64-NEXT:    shlq %cl, %rdi
; X64-NEXT:    shrq %rax
; X64-NEXT:    notb %cl
; X64-NEXT:    # kill: def $cl killed $cl killed $ecx
; X64-NEXT:    shrq %cl, %rax
; X64-NEXT:    orq %rdi, %rax
; X64-NEXT:    retq
;
; BMI-LABEL: lshift:
; BMI:       # %bb.0: # %entry
; BMI-NEXT:    movq %rsi, %rax
; BMI-NEXT:    movl %edx, %ecx
; BMI-NEXT:    shrq %rax
; BMI-NEXT:    shlq %cl, %rdi
; BMI-NEXT:    notb %cl
; BMI-NEXT:    # kill: def $cl killed $cl killed $ecx
; BMI-NEXT:    shrq %cl, %rax
; BMI-NEXT:    orq %rdi, %rax
; BMI-NEXT:    retq
;
; BMI2-SLOW-LABEL: lshift:
; BMI2-SLOW:       # %bb.0: # %entry
; BMI2-SLOW-NEXT:    # kill: def $edx killed $edx def $rdx
; BMI2-SLOW-NEXT:    shlxq %rdx, %rdi, %rcx
; BMI2-SLOW-NEXT:    notb %dl
; BMI2-SLOW-NEXT:    shrq %rsi
; BMI2-SLOW-NEXT:    shrxq %rdx, %rsi, %rax
; BMI2-SLOW-NEXT:    orq %rcx, %rax
; BMI2-SLOW-NEXT:    retq
;
; BMI2-FAST-LABEL: lshift:
; BMI2-FAST:       # %bb.0: # %entry
; BMI2-FAST-NEXT:    movl %edx, %ecx
; BMI2-FAST-NEXT:    movq %rdi, %rax
; BMI2-FAST-NEXT:    # kill: def $cl killed $cl killed $ecx
; BMI2-FAST-NEXT:    shldq %cl, %rsi, %rax
; BMI2-FAST-NEXT:    retq
entry:
  %sh_prom = zext i32 %c to i64
  %shl = shl i64 %a, %sh_prom
  %sub = sub nsw i32 64, %c
  %sh_prom1 = zext i32 %sub to i64
  %shr = lshr i64 %b, %sh_prom1
  %or = or i64 %shr, %shl
  ret i64 %or
}

;uint64_t rshift(uint64_t a, uint64_t b, int c)
;{
;    return (a >> c) | (b << (64-c));
;}

define i64 @rshift(i64 %a, i64 %b, i32 %c) nounwind readnone {
; X64-LABEL: rshift:
; X64:       # %bb.0: # %entry
; X64-NEXT:    movl %edx, %ecx
; X64-NEXT:    shrq %cl, %rdi
; X64-NEXT:    leaq (%rsi,%rsi), %rax
; X64-NEXT:    notb %cl
; X64-NEXT:    # kill: def $cl killed $cl killed $ecx
; X64-NEXT:    shlq %cl, %rax
; X64-NEXT:    orq %rdi, %rax
; X64-NEXT:    retq
;
; BMI-LABEL: rshift:
; BMI:       # %bb.0: # %entry
; BMI-NEXT:    movl %edx, %ecx
; BMI-NEXT:    leaq (%rsi,%rsi), %rax
; BMI-NEXT:    shrq %cl, %rdi
; BMI-NEXT:    notb %cl
; BMI-NEXT:    # kill: def $cl killed $cl killed $ecx
; BMI-NEXT:    shlq %cl, %rax
; BMI-NEXT:    orq %rdi, %rax
; BMI-NEXT:    retq
;
; BMI2-SLOW-LABEL: rshift:
; BMI2-SLOW:       # %bb.0: # %entry
; BMI2-SLOW-NEXT:    # kill: def $edx killed $edx def $rdx
; BMI2-SLOW-NEXT:    shrxq %rdx, %rdi, %rcx
; BMI2-SLOW-NEXT:    notb %dl
; BMI2-SLOW-NEXT:    addq %rsi, %rsi
; BMI2-SLOW-NEXT:    shlxq %rdx, %rsi, %rax
; BMI2-SLOW-NEXT:    orq %rcx, %rax
; BMI2-SLOW-NEXT:    retq
;
; BMI2-FAST-LABEL: rshift:
; BMI2-FAST:       # %bb.0: # %entry
; BMI2-FAST-NEXT:    movl %edx, %ecx
; BMI2-FAST-NEXT:    movq %rdi, %rax
; BMI2-FAST-NEXT:    # kill: def $cl killed $cl killed $ecx
; BMI2-FAST-NEXT:    shrdq %cl, %rsi, %rax
; BMI2-FAST-NEXT:    retq
entry:
  %sh_prom = zext i32 %c to i64
  %shr = lshr i64 %a, %sh_prom
  %sub = sub nsw i32 64, %c
  %sh_prom1 = zext i32 %sub to i64
  %shl = shl i64 %b, %sh_prom1
  %or = or i64 %shl, %shr
  ret i64 %or
}