[llvm] f16125a - [CodeGenPrepare][X86] Add tests for folding `urem` with loop invariant value; NFC
Noah Goldstein via llvm-commits
llvm-commits at lists.llvm.org
Sun Aug 18 15:58:39 PDT 2024
Author: Noah Goldstein
Date: 2024-08-18T15:58:24-07:00
New Revision: f16125a13ce725b1e936468e08257c0fbb80c0fa
URL: https://github.com/llvm/llvm-project/commit/f16125a13ce725b1e936468e08257c0fbb80c0fa
DIFF: https://github.com/llvm/llvm-project/commit/f16125a13ce725b1e936468e08257c0fbb80c0fa.diff
LOG: [CodeGenPrepare][X86] Add tests for folding `urem` with loop invariant value; NFC
Added:
llvm/test/CodeGen/X86/fold-loop-of-urem.ll
llvm/test/Transforms/CodeGenPrepare/X86/fold-loop-of-urem.ll
Modified:
Removed:
################################################################################
diff --git a/llvm/test/CodeGen/X86/fold-loop-of-urem.ll b/llvm/test/CodeGen/X86/fold-loop-of-urem.ll
new file mode 100644
index 00000000000000..aad2e0dd7bd248
--- /dev/null
+++ b/llvm/test/CodeGen/X86/fold-loop-of-urem.ll
@@ -0,0 +1,1159 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown | FileCheck %s
+
+declare void @use.i32(i32)
+declare void @use.2xi64(<2 x i64>)
+declare void @do_stuff0()
+declare void @do_stuff1()
+declare i1 @get.i1()
+declare i32 @get.i32()
+
+define void @simple_urem_to_sel(i32 %N, i32 %rem_amt) nounwind {
+; CHECK-LABEL: simple_urem_to_sel:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: testl %edi, %edi
+; CHECK-NEXT: je .LBB0_4
+; CHECK-NEXT: # %bb.1: # %for.body.preheader
+; CHECK-NEXT: pushq %rbp
+; CHECK-NEXT: pushq %r14
+; CHECK-NEXT: pushq %rbx
+; CHECK-NEXT: movl %esi, %ebx
+; CHECK-NEXT: movl %edi, %ebp
+; CHECK-NEXT: xorl %r14d, %r14d
+; CHECK-NEXT: .p2align 4, 0x90
+; CHECK-NEXT: .LBB0_2: # %for.body
+; CHECK-NEXT: # =>This Inner Loop Header: Depth=1
+; CHECK-NEXT: movl %r14d, %eax
+; CHECK-NEXT: xorl %edx, %edx
+; CHECK-NEXT: divl %ebx
+; CHECK-NEXT: movl %edx, %edi
+; CHECK-NEXT: callq use.i32 at PLT
+; CHECK-NEXT: incl %r14d
+; CHECK-NEXT: cmpl %r14d, %ebp
+; CHECK-NEXT: jne .LBB0_2
+; CHECK-NEXT: # %bb.3:
+; CHECK-NEXT: popq %rbx
+; CHECK-NEXT: popq %r14
+; CHECK-NEXT: popq %rbp
+; CHECK-NEXT: .LBB0_4: # %for.cond.cleanup
+; CHECK-NEXT: retq
+entry:
+ %cmp3.not = icmp eq i32 %N, 0
+ br i1 %cmp3.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup:
+ ret void
+
+for.body:
+ %i.04 = phi i32 [ %inc, %for.body ], [ 0, %entry ]
+ %rem = urem i32 %i.04, %rem_amt
+ tail call void @use.i32(i32 %rem)
+ %inc = add nuw i32 %i.04, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+define void @simple_urem_to_sel_nested2(i32 %N, i32 %rem_amt) nounwind {
+; CHECK-LABEL: simple_urem_to_sel_nested2:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: testl %edi, %edi
+; CHECK-NEXT: je .LBB1_8
+; CHECK-NEXT: # %bb.1: # %for.body.preheader
+; CHECK-NEXT: pushq %rbp
+; CHECK-NEXT: pushq %r14
+; CHECK-NEXT: pushq %rbx
+; CHECK-NEXT: movl %esi, %ebx
+; CHECK-NEXT: movl %edi, %ebp
+; CHECK-NEXT: xorl %r14d, %r14d
+; CHECK-NEXT: jmp .LBB1_2
+; CHECK-NEXT: .p2align 4, 0x90
+; CHECK-NEXT: .LBB1_5: # %for.body1
+; CHECK-NEXT: # in Loop: Header=BB1_2 Depth=1
+; CHECK-NEXT: movl %r14d, %eax
+; CHECK-NEXT: xorl %edx, %edx
+; CHECK-NEXT: divl %ebx
+; CHECK-NEXT: movl %edx, %edi
+; CHECK-NEXT: callq use.i32 at PLT
+; CHECK-NEXT: .LBB1_6: # %for.body.tail
+; CHECK-NEXT: # in Loop: Header=BB1_2 Depth=1
+; CHECK-NEXT: incl %r14d
+; CHECK-NEXT: cmpl %r14d, %ebp
+; CHECK-NEXT: je .LBB1_7
+; CHECK-NEXT: .LBB1_2: # %for.body
+; CHECK-NEXT: # =>This Inner Loop Header: Depth=1
+; CHECK-NEXT: callq get.i1 at PLT
+; CHECK-NEXT: testb $1, %al
+; CHECK-NEXT: je .LBB1_6
+; CHECK-NEXT: # %bb.3: # %for.body0
+; CHECK-NEXT: # in Loop: Header=BB1_2 Depth=1
+; CHECK-NEXT: callq get.i1 at PLT
+; CHECK-NEXT: testb $1, %al
+; CHECK-NEXT: jne .LBB1_5
+; CHECK-NEXT: # %bb.4: # %for.body2
+; CHECK-NEXT: # in Loop: Header=BB1_2 Depth=1
+; CHECK-NEXT: callq get.i1 at PLT
+; CHECK-NEXT: testb $1, %al
+; CHECK-NEXT: jne .LBB1_5
+; CHECK-NEXT: jmp .LBB1_6
+; CHECK-NEXT: .LBB1_7:
+; CHECK-NEXT: popq %rbx
+; CHECK-NEXT: popq %r14
+; CHECK-NEXT: popq %rbp
+; CHECK-NEXT: .LBB1_8: # %for.cond.cleanup
+; CHECK-NEXT: retq
+entry:
+ %cmp3.not = icmp eq i32 %N, 0
+ br i1 %cmp3.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup:
+ ret void
+
+for.body:
+ %i.04 = phi i32 [ %inc, %for.body.tail ], [ 0, %entry ]
+ %cond0 = call i1 @get.i1()
+ br i1 %cond0, label %for.body0, label %for.body.tail
+for.body0:
+ %cond1 = call i1 @get.i1()
+ br i1 %cond1, label %for.body1, label %for.body2
+for.body2:
+ %cond2 = call i1 @get.i1()
+ br i1 %cond2, label %for.body1, label %for.body.tail
+for.body1:
+ %rem = urem i32 %i.04, %rem_amt
+ tail call void @use.i32(i32 %rem)
+ br label %for.body.tail
+for.body.tail:
+ %inc = add nuw i32 %i.04, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+define void @simple_urem_fail_bad_incr3(i32 %N, i32 %rem_amt) nounwind {
+; CHECK-LABEL: simple_urem_fail_bad_incr3:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: testl %edi, %edi
+; CHECK-NEXT: je .LBB2_9
+; CHECK-NEXT: # %bb.1:
+; CHECK-NEXT: pushq %rbp
+; CHECK-NEXT: pushq %r14
+; CHECK-NEXT: pushq %rbx
+; CHECK-NEXT: movl %esi, %ebx
+; CHECK-NEXT: jmp .LBB2_2
+; CHECK-NEXT: .p2align 4, 0x90
+; CHECK-NEXT: .LBB2_6: # %for.body1
+; CHECK-NEXT: # in Loop: Header=BB2_2 Depth=1
+; CHECK-NEXT: movl %ebp, %eax
+; CHECK-NEXT: xorl %edx, %edx
+; CHECK-NEXT: divl %ebx
+; CHECK-NEXT: movl %edx, %edi
+; CHECK-NEXT: callq use.i32 at PLT
+; CHECK-NEXT: .LBB2_7: # %for.body.tail
+; CHECK-NEXT: # in Loop: Header=BB2_2 Depth=1
+; CHECK-NEXT: callq get.i1 at PLT
+; CHECK-NEXT: testb $1, %al
+; CHECK-NEXT: jne .LBB2_8
+; CHECK-NEXT: .LBB2_2: # %for.body
+; CHECK-NEXT: # =>This Inner Loop Header: Depth=1
+; CHECK-NEXT: callq get.i1 at PLT
+; CHECK-NEXT: testb $1, %al
+; CHECK-NEXT: je .LBB2_5
+; CHECK-NEXT: # %bb.3: # %for.body0
+; CHECK-NEXT: # in Loop: Header=BB2_2 Depth=1
+; CHECK-NEXT: callq get.i1 at PLT
+; CHECK-NEXT: movl %eax, %r14d
+; CHECK-NEXT: callq get.i32 at PLT
+; CHECK-NEXT: testb $1, %r14b
+; CHECK-NEXT: je .LBB2_7
+; CHECK-NEXT: # %bb.4: # in Loop: Header=BB2_2 Depth=1
+; CHECK-NEXT: movl %eax, %ebp
+; CHECK-NEXT: incl %ebp
+; CHECK-NEXT: jmp .LBB2_6
+; CHECK-NEXT: .p2align 4, 0x90
+; CHECK-NEXT: .LBB2_5: # %for.body2
+; CHECK-NEXT: # in Loop: Header=BB2_2 Depth=1
+; CHECK-NEXT: xorl %ebp, %ebp
+; CHECK-NEXT: callq get.i1 at PLT
+; CHECK-NEXT: testb $1, %al
+; CHECK-NEXT: jne .LBB2_6
+; CHECK-NEXT: jmp .LBB2_7
+; CHECK-NEXT: .LBB2_8:
+; CHECK-NEXT: popq %rbx
+; CHECK-NEXT: popq %r14
+; CHECK-NEXT: popq %rbp
+; CHECK-NEXT: .LBB2_9: # %for.cond.cleanup
+; CHECK-NEXT: retq
+entry:
+ %cmp3.not = icmp eq i32 %N, 0
+ br i1 %cmp3.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup:
+ ret void
+
+for.body:
+ %cond0 = call i1 @get.i1()
+ br i1 %cond0, label %for.body0, label %for.body2
+for.body0:
+ %cond1 = call i1 @get.i1()
+ %val = call i32 @get.i32()
+ %inc = add nuw i32 %val, 1
+ br i1 %cond1, label %for.body1, label %for.body.tail
+for.body2:
+ %cond2 = call i1 @get.i1()
+ br i1 %cond2, label %for.body1, label %for.body.tail
+for.body1:
+ %i.04 = phi i32 [ %inc, %for.body0], [ 0, %for.body2 ]
+ %rem = urem i32 %i.04, %rem_amt
+ tail call void @use.i32(i32 %rem)
+ br label %for.body.tail
+for.body.tail:
+ %exitcond.not = call i1 @get.i1()
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+define void @simple_urem_to_sel_vec(<2 x i64> %rem_amt) nounwind {
+; CHECK-LABEL: simple_urem_to_sel_vec:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: pushq %r14
+; CHECK-NEXT: pushq %rbx
+; CHECK-NEXT: subq $24, %rsp
+; CHECK-NEXT: pxor %xmm1, %xmm1
+; CHECK-NEXT: movdqa %xmm1, (%rsp) # 16-byte Spill
+; CHECK-NEXT: movq %xmm0, %rbx
+; CHECK-NEXT: pshufd {{.*#+}} xmm0 = xmm0[2,3,2,3]
+; CHECK-NEXT: movq %xmm0, %r14
+; CHECK-NEXT: .p2align 4, 0x90
+; CHECK-NEXT: .LBB3_1: # %for.body
+; CHECK-NEXT: # =>This Inner Loop Header: Depth=1
+; CHECK-NEXT: movdqa (%rsp), %xmm1 # 16-byte Reload
+; CHECK-NEXT: movdqa %xmm1, (%rsp) # 16-byte Spill
+; CHECK-NEXT: movq %xmm1, %rax
+; CHECK-NEXT: xorl %edx, %edx
+; CHECK-NEXT: divq %rbx
+; CHECK-NEXT: movq %rdx, %xmm0
+; CHECK-NEXT: pshufd {{.*#+}} xmm1 = xmm1[2,3,2,3]
+; CHECK-NEXT: movq %xmm1, %rax
+; CHECK-NEXT: xorl %edx, %edx
+; CHECK-NEXT: divq %r14
+; CHECK-NEXT: movq %rdx, %xmm1
+; CHECK-NEXT: punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
+; CHECK-NEXT: callq use.2xi64 at PLT
+; CHECK-NEXT: movdqa (%rsp), %xmm0 # 16-byte Reload
+; CHECK-NEXT: psubq {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
+; CHECK-NEXT: movdqa %xmm0, (%rsp) # 16-byte Spill
+; CHECK-NEXT: callq get.i1 at PLT
+; CHECK-NEXT: testb $1, %al
+; CHECK-NEXT: je .LBB3_1
+; CHECK-NEXT: # %bb.2: # %for.cond.cleanup
+; CHECK-NEXT: addq $24, %rsp
+; CHECK-NEXT: popq %rbx
+; CHECK-NEXT: popq %r14
+; CHECK-NEXT: retq
+entry:
+ br label %for.body
+
+for.cond.cleanup:
+ ret void
+
+for.body:
+ %i.04 = phi <2 x i64> [ %inc, %for.body ], [ zeroinitializer, %entry ]
+ %rem = urem <2 x i64> %i.04, %rem_amt
+ tail call void @use.2xi64(<2 x i64> %rem)
+ %inc = add nuw <2 x i64> %i.04, <i64 1, i64 1>
+ %exitcond.not = call i1 @get.i1()
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+define void @simple_urem_fail_bad_incr(i32 %N, i32 %rem_amt) nounwind {
+; CHECK-LABEL: simple_urem_fail_bad_incr:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: testl %edi, %edi
+; CHECK-NEXT: je .LBB4_6
+; CHECK-NEXT: # %bb.1: # %for.body.preheader
+; CHECK-NEXT: pushq %rbp
+; CHECK-NEXT: pushq %r14
+; CHECK-NEXT: pushq %rbx
+; CHECK-NEXT: movl %esi, %ebx
+; CHECK-NEXT: movl %edi, %ebp
+; CHECK-NEXT: xorl %r14d, %r14d
+; CHECK-NEXT: jmp .LBB4_2
+; CHECK-NEXT: .p2align 4, 0x90
+; CHECK-NEXT: .LBB4_4: # %for.body.tail
+; CHECK-NEXT: # in Loop: Header=BB4_2 Depth=1
+; CHECK-NEXT: movl %r14d, %eax
+; CHECK-NEXT: xorl %edx, %edx
+; CHECK-NEXT: divl %ebx
+; CHECK-NEXT: movl %edx, %edi
+; CHECK-NEXT: callq use.i32 at PLT
+; CHECK-NEXT: incl %r14d
+; CHECK-NEXT: cmpl %ebp, %r14d
+; CHECK-NEXT: je .LBB4_5
+; CHECK-NEXT: .LBB4_2: # %for.body
+; CHECK-NEXT: # =>This Inner Loop Header: Depth=1
+; CHECK-NEXT: callq get.i1 at PLT
+; CHECK-NEXT: testb $1, %al
+; CHECK-NEXT: je .LBB4_4
+; CHECK-NEXT: # %bb.3: # %for.body0
+; CHECK-NEXT: # in Loop: Header=BB4_2 Depth=1
+; CHECK-NEXT: callq get.i32 at PLT
+; CHECK-NEXT: movl %eax, %r14d
+; CHECK-NEXT: jmp .LBB4_4
+; CHECK-NEXT: .LBB4_5:
+; CHECK-NEXT: popq %rbx
+; CHECK-NEXT: popq %r14
+; CHECK-NEXT: popq %rbp
+; CHECK-NEXT: .LBB4_6: # %for.cond.cleanup
+; CHECK-NEXT: retq
+entry:
+ %cmp3.not = icmp eq i32 %N, 0
+ br i1 %cmp3.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup:
+ ret void
+
+for.body:
+ %i.03 = phi i32 [ %inc, %for.body.tail ], [ 0, %entry ]
+ %cond0 = call i1 @get.i1()
+ br i1 %cond0, label %for.body0, label %for.body.tail
+for.body0:
+ %some_val = call i32 @get.i32()
+ br label %for.body.tail
+
+for.body.tail:
+ %i.04 = phi i32 [ %i.03, %for.body ], [ %some_val, %for.body0 ]
+ %rem = urem i32 %i.04, %rem_amt
+ tail call void @use.i32(i32 %rem)
+ %inc = add nuw i32 %i.04, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+define void @simple_urem_to_sel_second_acc(i32 %N, i32 %rem_amt) nounwind {
+; CHECK-LABEL: simple_urem_to_sel_second_acc:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: cmpl $2, %edi
+; CHECK-NEXT: jb .LBB5_4
+; CHECK-NEXT: # %bb.1: # %for.body.preheader
+; CHECK-NEXT: pushq %rbp
+; CHECK-NEXT: pushq %r15
+; CHECK-NEXT: pushq %r14
+; CHECK-NEXT: pushq %rbx
+; CHECK-NEXT: pushq %rax
+; CHECK-NEXT: movl %esi, %ebx
+; CHECK-NEXT: movl %edi, %ebp
+; CHECK-NEXT: movl $1, %r15d
+; CHECK-NEXT: xorl %r14d, %r14d
+; CHECK-NEXT: .p2align 4, 0x90
+; CHECK-NEXT: .LBB5_2: # %for.body
+; CHECK-NEXT: # =>This Inner Loop Header: Depth=1
+; CHECK-NEXT: movl %r14d, %eax
+; CHECK-NEXT: xorl %edx, %edx
+; CHECK-NEXT: divl %ebx
+; CHECK-NEXT: movl %edx, %edi
+; CHECK-NEXT: callq use.i32 at PLT
+; CHECK-NEXT: incl %r14d
+; CHECK-NEXT: addl $2, %r15d
+; CHECK-NEXT: cmpl %ebp, %r15d
+; CHECK-NEXT: jbe .LBB5_2
+; CHECK-NEXT: # %bb.3:
+; CHECK-NEXT: addq $8, %rsp
+; CHECK-NEXT: popq %rbx
+; CHECK-NEXT: popq %r14
+; CHECK-NEXT: popq %r15
+; CHECK-NEXT: popq %rbp
+; CHECK-NEXT: .LBB5_4: # %for.cond.cleanup
+; CHECK-NEXT: retq
+entry:
+ %cmp3.not = icmp ult i32 %N, 2
+ br i1 %cmp3.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup:
+ ret void
+
+for.body:
+ %i.04 = phi i32 [ %inc, %for.body ], [ 0, %entry ]
+ %i.05 = phi i32 [ %inc2, %for.body ], [ 1, %entry ]
+ %rem = urem i32 %i.04, %rem_amt
+ tail call void @use.i32(i32 %rem)
+ %inc = add nuw i32 %i.04, 1
+ %inc2 = add nuw i32 %i.05, 2
+ %exitcond.not = icmp ugt i32 %inc2, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+define void @simple_urem_fail_srem(i32 %N, i32 %rem_amt) nounwind {
+; CHECK-LABEL: simple_urem_fail_srem:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: testl %edi, %edi
+; CHECK-NEXT: je .LBB6_4
+; CHECK-NEXT: # %bb.1: # %for.body.preheader
+; CHECK-NEXT: pushq %rbp
+; CHECK-NEXT: pushq %r14
+; CHECK-NEXT: pushq %rbx
+; CHECK-NEXT: movl %esi, %ebx
+; CHECK-NEXT: movl %edi, %ebp
+; CHECK-NEXT: xorl %r14d, %r14d
+; CHECK-NEXT: .p2align 4, 0x90
+; CHECK-NEXT: .LBB6_2: # %for.body
+; CHECK-NEXT: # =>This Inner Loop Header: Depth=1
+; CHECK-NEXT: movl %r14d, %eax
+; CHECK-NEXT: cltd
+; CHECK-NEXT: idivl %ebx
+; CHECK-NEXT: movl %edx, %edi
+; CHECK-NEXT: callq use.i32 at PLT
+; CHECK-NEXT: incl %r14d
+; CHECK-NEXT: cmpl %r14d, %ebp
+; CHECK-NEXT: jne .LBB6_2
+; CHECK-NEXT: # %bb.3:
+; CHECK-NEXT: popq %rbx
+; CHECK-NEXT: popq %r14
+; CHECK-NEXT: popq %rbp
+; CHECK-NEXT: .LBB6_4: # %for.cond.cleanup
+; CHECK-NEXT: retq
+entry:
+ %cmp3.not = icmp eq i32 %N, 0
+ br i1 %cmp3.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup:
+ ret void
+
+for.body:
+ %i.04 = phi i32 [ %inc, %for.body ], [ 0, %entry ]
+ %rem = srem i32 %i.04, %rem_amt
+ tail call void @use.i32(i32 %rem)
+ %inc = add nuw i32 %i.04, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+define void @simple_urem_fail_missing_nuw(i32 %N, i32 %rem_amt) nounwind {
+; CHECK-LABEL: simple_urem_fail_missing_nuw:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: testl %edi, %edi
+; CHECK-NEXT: je .LBB7_4
+; CHECK-NEXT: # %bb.1: # %for.body.preheader
+; CHECK-NEXT: pushq %rbp
+; CHECK-NEXT: pushq %r14
+; CHECK-NEXT: pushq %rbx
+; CHECK-NEXT: movl %esi, %ebx
+; CHECK-NEXT: movl %edi, %ebp
+; CHECK-NEXT: xorl %r14d, %r14d
+; CHECK-NEXT: .p2align 4, 0x90
+; CHECK-NEXT: .LBB7_2: # %for.body
+; CHECK-NEXT: # =>This Inner Loop Header: Depth=1
+; CHECK-NEXT: movl %r14d, %eax
+; CHECK-NEXT: xorl %edx, %edx
+; CHECK-NEXT: divl %ebx
+; CHECK-NEXT: movl %edx, %edi
+; CHECK-NEXT: callq use.i32 at PLT
+; CHECK-NEXT: incl %r14d
+; CHECK-NEXT: cmpl %r14d, %ebp
+; CHECK-NEXT: jne .LBB7_2
+; CHECK-NEXT: # %bb.3:
+; CHECK-NEXT: popq %rbx
+; CHECK-NEXT: popq %r14
+; CHECK-NEXT: popq %rbp
+; CHECK-NEXT: .LBB7_4: # %for.cond.cleanup
+; CHECK-NEXT: retq
+entry:
+ %cmp3.not = icmp eq i32 %N, 0
+ br i1 %cmp3.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup:
+ ret void
+
+for.body:
+ %i.04 = phi i32 [ %inc, %for.body ], [ 0, %entry ]
+ %rem = urem i32 %i.04, %rem_amt
+ tail call void @use.i32(i32 %rem)
+ %inc = add nsw i32 %i.04, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+define void @simple_urem_fail_bad_incr2(i32 %N, i32 %rem_amt) nounwind {
+; CHECK-LABEL: simple_urem_fail_bad_incr2:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: testl %edi, %edi
+; CHECK-NEXT: je .LBB8_4
+; CHECK-NEXT: # %bb.1: # %for.body.preheader
+; CHECK-NEXT: pushq %rbp
+; CHECK-NEXT: pushq %r14
+; CHECK-NEXT: pushq %rbx
+; CHECK-NEXT: movl %esi, %ebx
+; CHECK-NEXT: movl %edi, %ebp
+; CHECK-NEXT: xorl %r14d, %r14d
+; CHECK-NEXT: .p2align 4, 0x90
+; CHECK-NEXT: .LBB8_2: # %for.body
+; CHECK-NEXT: # =>This Inner Loop Header: Depth=1
+; CHECK-NEXT: movl %r14d, %eax
+; CHECK-NEXT: xorl %edx, %edx
+; CHECK-NEXT: divl %ebx
+; CHECK-NEXT: movl %edx, %edi
+; CHECK-NEXT: callq use.i32 at PLT
+; CHECK-NEXT: addl $2, %r14d
+; CHECK-NEXT: cmpl %r14d, %ebp
+; CHECK-NEXT: jne .LBB8_2
+; CHECK-NEXT: # %bb.3:
+; CHECK-NEXT: popq %rbx
+; CHECK-NEXT: popq %r14
+; CHECK-NEXT: popq %rbp
+; CHECK-NEXT: .LBB8_4: # %for.cond.cleanup
+; CHECK-NEXT: retq
+entry:
+ %cmp3.not = icmp eq i32 %N, 0
+ br i1 %cmp3.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup:
+ ret void
+
+for.body:
+ %i.04 = phi i32 [ %inc, %for.body ], [ 0, %entry ]
+ %rem = urem i32 %i.04, %rem_amt
+ tail call void @use.i32(i32 %rem)
+ %inc = add nuw i32 %i.04, 2
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+define void @simple_urem_non_zero_entry4(i32 %N, i32 %rem_amt) nounwind {
+; CHECK-LABEL: simple_urem_non_zero_entry4:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: testl %edi, %edi
+; CHECK-NEXT: je .LBB9_4
+; CHECK-NEXT: # %bb.1: # %for.body.preheader
+; CHECK-NEXT: pushq %rbp
+; CHECK-NEXT: pushq %r14
+; CHECK-NEXT: pushq %rbx
+; CHECK-NEXT: movl %esi, %ebx
+; CHECK-NEXT: movl %edi, %ebp
+; CHECK-NEXT: movl $4, %r14d
+; CHECK-NEXT: .p2align 4, 0x90
+; CHECK-NEXT: .LBB9_2: # %for.body
+; CHECK-NEXT: # =>This Inner Loop Header: Depth=1
+; CHECK-NEXT: movl %r14d, %eax
+; CHECK-NEXT: xorl %edx, %edx
+; CHECK-NEXT: divl %ebx
+; CHECK-NEXT: movl %edx, %edi
+; CHECK-NEXT: callq use.i32 at PLT
+; CHECK-NEXT: incl %r14d
+; CHECK-NEXT: cmpl %r14d, %ebp
+; CHECK-NEXT: jne .LBB9_2
+; CHECK-NEXT: # %bb.3:
+; CHECK-NEXT: popq %rbx
+; CHECK-NEXT: popq %r14
+; CHECK-NEXT: popq %rbp
+; CHECK-NEXT: .LBB9_4: # %for.cond.cleanup
+; CHECK-NEXT: retq
+entry:
+ %cmp3.not = icmp eq i32 %N, 0
+ br i1 %cmp3.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup:
+ ret void
+
+for.body:
+ %i.04 = phi i32 [ %inc, %for.body ], [ 4, %entry ]
+ %rem = urem i32 %i.04, %rem_amt
+ tail call void @use.i32(i32 %rem)
+ %inc = add nuw i32 %i.04, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+define void @simple_urem_skip_const_rem_amt(i32 %N) nounwind {
+; CHECK-LABEL: simple_urem_skip_const_rem_amt:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: testl %edi, %edi
+; CHECK-NEXT: je .LBB10_4
+; CHECK-NEXT: # %bb.1: # %for.body.preheader
+; CHECK-NEXT: pushq %rbp
+; CHECK-NEXT: pushq %r14
+; CHECK-NEXT: pushq %rbx
+; CHECK-NEXT: movl %edi, %ebx
+; CHECK-NEXT: addl $-4, %ebx
+; CHECK-NEXT: movl $4, %ebp
+; CHECK-NEXT: movl $2938661835, %r14d # imm = 0xAF286BCB
+; CHECK-NEXT: .p2align 4, 0x90
+; CHECK-NEXT: .LBB10_2: # %for.body
+; CHECK-NEXT: # =>This Inner Loop Header: Depth=1
+; CHECK-NEXT: movl %ebp, %eax
+; CHECK-NEXT: imulq %r14, %rax
+; CHECK-NEXT: shrq $32, %rax
+; CHECK-NEXT: movl %ebp, %ecx
+; CHECK-NEXT: subl %eax, %ecx
+; CHECK-NEXT: shrl %ecx
+; CHECK-NEXT: addl %eax, %ecx
+; CHECK-NEXT: shrl $4, %ecx
+; CHECK-NEXT: leal (%rcx,%rcx,8), %eax
+; CHECK-NEXT: leal (%rcx,%rax,2), %eax
+; CHECK-NEXT: movl %ebp, %edi
+; CHECK-NEXT: subl %eax, %edi
+; CHECK-NEXT: callq use.i32 at PLT
+; CHECK-NEXT: incl %ebp
+; CHECK-NEXT: decl %ebx
+; CHECK-NEXT: jne .LBB10_2
+; CHECK-NEXT: # %bb.3:
+; CHECK-NEXT: popq %rbx
+; CHECK-NEXT: popq %r14
+; CHECK-NEXT: popq %rbp
+; CHECK-NEXT: .LBB10_4: # %for.cond.cleanup
+; CHECK-NEXT: retq
+entry:
+ %cmp3.not = icmp eq i32 %N, 0
+ br i1 %cmp3.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup:
+ ret void
+
+for.body:
+ %i.04 = phi i32 [ %inc, %for.body ], [ 4, %entry ]
+ %rem = urem i32 %i.04, 19
+ tail call void @use.i32(i32 %rem)
+ %inc = add nuw i32 %i.04, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+define void @simple_urem_fail_no_preheader_non_canonical(i32 %N, i32 %rem_amt) nounwind {
+; CHECK-LABEL: simple_urem_fail_no_preheader_non_canonical:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: pushq %rbp
+; CHECK-NEXT: pushq %r14
+; CHECK-NEXT: pushq %rbx
+; CHECK-NEXT: movl %esi, %ebx
+; CHECK-NEXT: movl %edi, %ebp
+; CHECK-NEXT: testl %edi, %edi
+; CHECK-NEXT: je .LBB11_1
+; CHECK-NEXT: # %bb.2: # %for.body1
+; CHECK-NEXT: movl $1, %r14d
+; CHECK-NEXT: jmp .LBB11_3
+; CHECK-NEXT: .LBB11_1:
+; CHECK-NEXT: xorl %r14d, %r14d
+; CHECK-NEXT: .p2align 4, 0x90
+; CHECK-NEXT: .LBB11_3: # %for.body
+; CHECK-NEXT: # =>This Inner Loop Header: Depth=1
+; CHECK-NEXT: movl %r14d, %eax
+; CHECK-NEXT: xorl %edx, %edx
+; CHECK-NEXT: divl %ebx
+; CHECK-NEXT: movl %edx, %edi
+; CHECK-NEXT: callq use.i32 at PLT
+; CHECK-NEXT: incl %r14d
+; CHECK-NEXT: cmpl %r14d, %ebp
+; CHECK-NEXT: jne .LBB11_3
+; CHECK-NEXT: # %bb.4: # %for.cond.cleanup
+; CHECK-NEXT: popq %rbx
+; CHECK-NEXT: popq %r14
+; CHECK-NEXT: popq %rbp
+; CHECK-NEXT: retq
+entry:
+ %cmp3.not = icmp eq i32 %N, 0
+ br i1 %cmp3.not, label %for.body0, label %for.body1
+
+for.cond.cleanup:
+ ret void
+
+for.body0:
+ br label %for.body
+
+for.body1:
+ br label %for.body
+
+for.body:
+ %i.04 = phi i32 [ %inc, %for.body ], [ 0, %for.body0 ], [ 1, %for.body1 ]
+ %rem = urem i32 %i.04, %rem_amt
+ tail call void @use.i32(i32 %rem)
+ %inc = add nuw i32 %i.04, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+define void @simple_urem_multi_latch_non_canonical(i32 %N, i32 %rem_amt) nounwind {
+; CHECK-LABEL: simple_urem_multi_latch_non_canonical:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: testl %edi, %edi
+; CHECK-NEXT: je .LBB12_6
+; CHECK-NEXT: # %bb.1: # %for.body.preheader
+; CHECK-NEXT: pushq %rbp
+; CHECK-NEXT: pushq %r15
+; CHECK-NEXT: pushq %r14
+; CHECK-NEXT: pushq %rbx
+; CHECK-NEXT: pushq %rax
+; CHECK-NEXT: movl %esi, %ebx
+; CHECK-NEXT: movl %edi, %ebp
+; CHECK-NEXT: decl %ebp
+; CHECK-NEXT: xorl %r14d, %r14d
+; CHECK-NEXT: jmp .LBB12_2
+; CHECK-NEXT: .p2align 4, 0x90
+; CHECK-NEXT: .LBB12_3: # %for.body.backedge
+; CHECK-NEXT: # in Loop: Header=BB12_2 Depth=1
+; CHECK-NEXT: incl %r14d
+; CHECK-NEXT: .LBB12_2: # %for.body
+; CHECK-NEXT: # =>This Inner Loop Header: Depth=1
+; CHECK-NEXT: movl %r14d, %eax
+; CHECK-NEXT: xorl %edx, %edx
+; CHECK-NEXT: divl %ebx
+; CHECK-NEXT: movl %edx, %edi
+; CHECK-NEXT: callq use.i32 at PLT
+; CHECK-NEXT: callq get.i1 at PLT
+; CHECK-NEXT: movl %eax, %r15d
+; CHECK-NEXT: callq do_stuff0 at PLT
+; CHECK-NEXT: testb $1, %r15b
+; CHECK-NEXT: je .LBB12_3
+; CHECK-NEXT: # %bb.4: # %for.body0
+; CHECK-NEXT: # in Loop: Header=BB12_2 Depth=1
+; CHECK-NEXT: callq do_stuff1 at PLT
+; CHECK-NEXT: cmpl %r14d, %ebp
+; CHECK-NEXT: jne .LBB12_3
+; CHECK-NEXT: # %bb.5:
+; CHECK-NEXT: addq $8, %rsp
+; CHECK-NEXT: popq %rbx
+; CHECK-NEXT: popq %r14
+; CHECK-NEXT: popq %r15
+; CHECK-NEXT: popq %rbp
+; CHECK-NEXT: .LBB12_6: # %for.cond.cleanup
+; CHECK-NEXT: retq
+entry:
+ %cmp3.not = icmp eq i32 %N, 0
+ br i1 %cmp3.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup:
+ ret void
+
+for.body:
+ %i.04 = phi i32 [ %inc, %for.body ], [ %inc, %for.body0 ], [ 0, %entry ]
+ %rem = urem i32 %i.04, %rem_amt
+ tail call void @use.i32(i32 %rem)
+ %inc = add nuw i32 %i.04, 1
+ %cond = call i1 @get.i1()
+ call void @do_stuff0()
+ br i1 %cond, label %for.body0, label %for.body
+for.body0:
+ call void @do_stuff1()
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+define void @simple_urem_fail_bad_loop(i32 %N, i32 %rem_amt) nounwind {
+; CHECK-LABEL: simple_urem_fail_bad_loop:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: pushq %rbp
+; CHECK-NEXT: pushq %r14
+; CHECK-NEXT: pushq %rbx
+; CHECK-NEXT: movl %esi, %ebx
+; CHECK-NEXT: movl %edi, %ebp
+; CHECK-NEXT: callq get.i32 at PLT
+; CHECK-NEXT: testl %eax, %eax
+; CHECK-NEXT: # implicit-def: $r14d
+; CHECK-NEXT: jne .LBB13_4
+; CHECK-NEXT: # %bb.1:
+; CHECK-NEXT: xorl %r14d, %r14d
+; CHECK-NEXT: .LBB13_2: # %for.cond
+; CHECK-NEXT: cmpl %ebp, %r14d
+; CHECK-NEXT: jae .LBB13_5
+; CHECK-NEXT: # %bb.3: # %for.body
+; CHECK-NEXT: movl %r14d, %edi
+; CHECK-NEXT: xorl $1, %edi
+; CHECK-NEXT: callq use.i32 at PLT
+; CHECK-NEXT: .LBB13_4: # %halfway
+; CHECK-NEXT: movl %r14d, %eax
+; CHECK-NEXT: xorl %edx, %edx
+; CHECK-NEXT: divl %ebx
+; CHECK-NEXT: movl %edx, %edi
+; CHECK-NEXT: callq use.i32 at PLT
+; CHECK-NEXT: incl %r14d
+; CHECK-NEXT: jmp .LBB13_2
+; CHECK-NEXT: .LBB13_5: # %for.end
+; CHECK-NEXT: popq %rbx
+; CHECK-NEXT: popq %r14
+; CHECK-NEXT: popq %rbp
+; CHECK-NEXT: retq
+entry:
+ %call = call i32 @get.i32()
+ %tobool.not = icmp eq i32 %call, 0
+ br i1 %tobool.not, label %for.cond, label %halfway
+
+for.cond:
+ %i.0 = phi i32 [ %inc, %halfway ], [ 0, %entry ]
+ %cmp = icmp ult i32 %i.0, %N
+ br i1 %cmp, label %for.body, label %for.end
+
+for.body:
+ %xor = xor i32 %i.0, 1
+ call void @use.i32(i32 %xor)
+ br label %halfway
+
+halfway:
+ %i.1 = phi i32 [ poison, %entry ], [ %i.0, %for.body ]
+ %rem = urem i32 %i.1, %rem_amt
+ call void @use.i32(i32 %rem)
+ %inc = add nuw i32 %i.1, 1
+ br label %for.cond
+
+for.end:
+ ret void
+}
+
+define void @simple_urem_fail_intermediate_inc(i32 %N, i32 %rem_amt) nounwind {
+; CHECK-LABEL: simple_urem_fail_intermediate_inc:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: testl %edi, %edi
+; CHECK-NEXT: je .LBB14_4
+; CHECK-NEXT: # %bb.1: # %for.body.preheader
+; CHECK-NEXT: pushq %r15
+; CHECK-NEXT: pushq %r14
+; CHECK-NEXT: pushq %rbx
+; CHECK-NEXT: movl %esi, %ebx
+; CHECK-NEXT: movl %edi, %r14d
+; CHECK-NEXT: negl %r14d
+; CHECK-NEXT: movl $1, %r15d
+; CHECK-NEXT: .p2align 4, 0x90
+; CHECK-NEXT: .LBB14_2: # %for.body
+; CHECK-NEXT: # =>This Inner Loop Header: Depth=1
+; CHECK-NEXT: movl %r15d, %eax
+; CHECK-NEXT: xorl %edx, %edx
+; CHECK-NEXT: divl %ebx
+; CHECK-NEXT: movl %edx, %edi
+; CHECK-NEXT: callq use.i32 at PLT
+; CHECK-NEXT: leal 1(%r14,%r15), %eax
+; CHECK-NEXT: movl %r15d, %ecx
+; CHECK-NEXT: incl %ecx
+; CHECK-NEXT: cmpl $1, %eax
+; CHECK-NEXT: movl %ecx, %r15d
+; CHECK-NEXT: jne .LBB14_2
+; CHECK-NEXT: # %bb.3:
+; CHECK-NEXT: popq %rbx
+; CHECK-NEXT: popq %r14
+; CHECK-NEXT: popq %r15
+; CHECK-NEXT: .LBB14_4: # %for.cond.cleanup
+; CHECK-NEXT: retq
+entry:
+ %cmp3.not = icmp eq i32 %N, 0
+ br i1 %cmp3.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup:
+ ret void
+
+for.body:
+ %i.04 = phi i32 [ %inc, %for.body ], [ 0, %entry ]
+ %inc2 = add nuw i32 %i.04, 1
+ %rem = urem i32 %inc2, %rem_amt
+ tail call void @use.i32(i32 %rem)
+ %inc = add nuw i32 %i.04, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+define void @weird_loop(i64 %sub.ptr.div.i56) personality ptr null {
+; CHECK-LABEL: weird_loop:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: .p2align 4, 0x90
+; CHECK-NEXT: .LBB15_1: # %for.body
+; CHECK-NEXT: # =>This Inner Loop Header: Depth=1
+; CHECK-NEXT: jmp .LBB15_1
+entry:
+ br label %for.preheader
+
+for.preheader:
+ %i57.0540.us = phi i64 [ 0, %entry ], [ %add74.us, %for.body ]
+ %add74.us = add nuw i64 %i57.0540.us, 1
+ br label %for.body
+
+for.body:
+ %rem.us = urem i64 %i57.0540.us, %sub.ptr.div.i56
+ br i1 false, label %for.preheader, label %for.body
+}
+
+define void @simple_urem_to_sel_non_zero_start(i32 %N, i32 %rem_amt) nounwind {
+; CHECK-LABEL: simple_urem_to_sel_non_zero_start:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: cmpl $3, %edi
+; CHECK-NEXT: jb .LBB16_4
+; CHECK-NEXT: # %bb.1: # %for.body.preheader
+; CHECK-NEXT: pushq %rbp
+; CHECK-NEXT: pushq %r14
+; CHECK-NEXT: pushq %rbx
+; CHECK-NEXT: movl %esi, %ebx
+; CHECK-NEXT: movl %edi, %ebp
+; CHECK-NEXT: movl $2, %r14d
+; CHECK-NEXT: .p2align 4, 0x90
+; CHECK-NEXT: .LBB16_2: # %for.body
+; CHECK-NEXT: # =>This Inner Loop Header: Depth=1
+; CHECK-NEXT: movl %r14d, %eax
+; CHECK-NEXT: xorl %edx, %edx
+; CHECK-NEXT: divl %ebx
+; CHECK-NEXT: movl %edx, %edi
+; CHECK-NEXT: callq use.i32 at PLT
+; CHECK-NEXT: incl %r14d
+; CHECK-NEXT: cmpl %r14d, %ebp
+; CHECK-NEXT: jne .LBB16_2
+; CHECK-NEXT: # %bb.3:
+; CHECK-NEXT: popq %rbx
+; CHECK-NEXT: popq %r14
+; CHECK-NEXT: popq %rbp
+; CHECK-NEXT: .LBB16_4: # %for.cond.cleanup
+; CHECK-NEXT: retq
+entry:
+ %cmp3.not = icmp ult i32 %N, 3
+ br i1 %cmp3.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup:
+ ret void
+
+for.body:
+ %i.04 = phi i32 [ %inc, %for.body ], [ 2, %entry ]
+ %rem = urem i32 %i.04, %rem_amt
+ tail call void @use.i32(i32 %rem)
+ %inc = add nuw i32 %i.04, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+define void @simple_urem_to_sel_non_zero_start_through_add(i32 %N, i32 %rem_amt_in) nounwind {
+; CHECK-LABEL: simple_urem_to_sel_non_zero_start_through_add:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: cmpl $3, %edi
+; CHECK-NEXT: jb .LBB17_4
+; CHECK-NEXT: # %bb.1: # %for.body.preheader
+; CHECK-NEXT: pushq %r15
+; CHECK-NEXT: pushq %r14
+; CHECK-NEXT: pushq %rbx
+; CHECK-NEXT: movl %esi, %ebx
+; CHECK-NEXT: movl %edi, %r14d
+; CHECK-NEXT: orl $16, %ebx
+; CHECK-NEXT: negl %r14d
+; CHECK-NEXT: movl $7, %r15d
+; CHECK-NEXT: .p2align 4, 0x90
+; CHECK-NEXT: .LBB17_2: # %for.body
+; CHECK-NEXT: # =>This Inner Loop Header: Depth=1
+; CHECK-NEXT: movl %r15d, %eax
+; CHECK-NEXT: xorl %edx, %edx
+; CHECK-NEXT: divl %ebx
+; CHECK-NEXT: movl %edx, %edi
+; CHECK-NEXT: callq use.i32 at PLT
+; CHECK-NEXT: leal 1(%r14,%r15), %eax
+; CHECK-NEXT: movl %r15d, %ecx
+; CHECK-NEXT: incl %ecx
+; CHECK-NEXT: cmpl $5, %eax
+; CHECK-NEXT: movl %ecx, %r15d
+; CHECK-NEXT: jne .LBB17_2
+; CHECK-NEXT: # %bb.3:
+; CHECK-NEXT: popq %rbx
+; CHECK-NEXT: popq %r14
+; CHECK-NEXT: popq %r15
+; CHECK-NEXT: .LBB17_4: # %for.cond.cleanup
+; CHECK-NEXT: retq
+entry:
+ %rem_amt = or i32 %rem_amt_in, 16
+ %cmp3.not = icmp ult i32 %N, 3
+ br i1 %cmp3.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup:
+ ret void
+
+for.body:
+ %i.04 = phi i32 [ %inc, %for.body ], [ 2, %entry ]
+ %i_with_off = add nuw i32 %i.04, 5
+ %rem = urem i32 %i_with_off, %rem_amt
+ tail call void @use.i32(i32 %rem)
+ %inc = add nuw i32 %i.04, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+define void @simple_urem_to_sel_non_zero_start_through_add_fail_missing_nuw(i32 %N, i32 %rem_amt_in) nounwind {
+; CHECK-LABEL: simple_urem_to_sel_non_zero_start_through_add_fail_missing_nuw:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: cmpl $3, %edi
+; CHECK-NEXT: jb .LBB18_4
+; CHECK-NEXT: # %bb.1: # %for.body.preheader
+; CHECK-NEXT: pushq %r15
+; CHECK-NEXT: pushq %r14
+; CHECK-NEXT: pushq %rbx
+; CHECK-NEXT: movl %esi, %ebx
+; CHECK-NEXT: movl %edi, %r14d
+; CHECK-NEXT: orl $16, %ebx
+; CHECK-NEXT: negl %r14d
+; CHECK-NEXT: movl $7, %r15d
+; CHECK-NEXT: .p2align 4, 0x90
+; CHECK-NEXT: .LBB18_2: # %for.body
+; CHECK-NEXT: # =>This Inner Loop Header: Depth=1
+; CHECK-NEXT: movl %r15d, %eax
+; CHECK-NEXT: xorl %edx, %edx
+; CHECK-NEXT: divl %ebx
+; CHECK-NEXT: movl %edx, %edi
+; CHECK-NEXT: callq use.i32 at PLT
+; CHECK-NEXT: leal 1(%r14,%r15), %eax
+; CHECK-NEXT: movl %r15d, %ecx
+; CHECK-NEXT: incl %ecx
+; CHECK-NEXT: cmpl $5, %eax
+; CHECK-NEXT: movl %ecx, %r15d
+; CHECK-NEXT: jne .LBB18_2
+; CHECK-NEXT: # %bb.3:
+; CHECK-NEXT: popq %rbx
+; CHECK-NEXT: popq %r14
+; CHECK-NEXT: popq %r15
+; CHECK-NEXT: .LBB18_4: # %for.cond.cleanup
+; CHECK-NEXT: retq
+entry:
+ %rem_amt = or i32 %rem_amt_in, 16
+ %cmp3.not = icmp ult i32 %N, 3
+ br i1 %cmp3.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup:
+ ret void
+
+for.body:
+ %i.04 = phi i32 [ %inc, %for.body ], [ 2, %entry ]
+ %i_with_off = add i32 %i.04, 5
+ %rem = urem i32 %i_with_off, %rem_amt
+ tail call void @use.i32(i32 %rem)
+ %inc = add nuw i32 %i.04, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+define void @simple_urem_to_sel_non_zero_start_through_add_fail_no_simplify_rem(i32 %N, i32 %rem_amt) nounwind {
+; CHECK-LABEL: simple_urem_to_sel_non_zero_start_through_add_fail_no_simplify_rem:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: cmpl $3, %edi
+; CHECK-NEXT: jb .LBB19_4
+; CHECK-NEXT: # %bb.1: # %for.body.preheader
+; CHECK-NEXT: pushq %r15
+; CHECK-NEXT: pushq %r14
+; CHECK-NEXT: pushq %rbx
+; CHECK-NEXT: movl %esi, %ebx
+; CHECK-NEXT: movl %edi, %r14d
+; CHECK-NEXT: negl %r14d
+; CHECK-NEXT: movl $7, %r15d
+; CHECK-NEXT: .p2align 4, 0x90
+; CHECK-NEXT: .LBB19_2: # %for.body
+; CHECK-NEXT: # =>This Inner Loop Header: Depth=1
+; CHECK-NEXT: movl %r15d, %eax
+; CHECK-NEXT: xorl %edx, %edx
+; CHECK-NEXT: divl %ebx
+; CHECK-NEXT: movl %edx, %edi
+; CHECK-NEXT: callq use.i32 at PLT
+; CHECK-NEXT: leal 1(%r14,%r15), %eax
+; CHECK-NEXT: movl %r15d, %ecx
+; CHECK-NEXT: incl %ecx
+; CHECK-NEXT: cmpl $5, %eax
+; CHECK-NEXT: movl %ecx, %r15d
+; CHECK-NEXT: jne .LBB19_2
+; CHECK-NEXT: # %bb.3:
+; CHECK-NEXT: popq %rbx
+; CHECK-NEXT: popq %r14
+; CHECK-NEXT: popq %r15
+; CHECK-NEXT: .LBB19_4: # %for.cond.cleanup
+; CHECK-NEXT: retq
+entry:
+ %cmp3.not = icmp ult i32 %N, 3
+ br i1 %cmp3.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup:
+ ret void
+
+for.body:
+ %i.04 = phi i32 [ %inc, %for.body ], [ 2, %entry ]
+ %i_with_off = add nuw i32 %i.04, 5
+ %rem = urem i32 %i_with_off, %rem_amt
+ tail call void @use.i32(i32 %rem)
+ %inc = add nuw i32 %i.04, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+define void @simple_urem_to_sel_non_zero_start_through_sub(i32 %N, i32 %rem_amt, i32 %start) nounwind {
+; CHECK-LABEL: simple_urem_to_sel_non_zero_start_through_sub:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: pushq %rbp
+; CHECK-NEXT: pushq %r14
+; CHECK-NEXT: pushq %rbx
+; CHECK-NEXT: movl %edi, %ebp
+; CHECK-NEXT: subl %edx, %ebp
+; CHECK-NEXT: jbe .LBB20_3
+; CHECK-NEXT: # %bb.1: # %for.body.preheader
+; CHECK-NEXT: movl %esi, %ebx
+; CHECK-NEXT: xorl %r14d, %r14d
+; CHECK-NEXT: .p2align 4, 0x90
+; CHECK-NEXT: .LBB20_2: # %for.body
+; CHECK-NEXT: # =>This Inner Loop Header: Depth=1
+; CHECK-NEXT: movl %r14d, %eax
+; CHECK-NEXT: xorl %edx, %edx
+; CHECK-NEXT: divl %ebx
+; CHECK-NEXT: movl %edx, %edi
+; CHECK-NEXT: callq use.i32 at PLT
+; CHECK-NEXT: incl %r14d
+; CHECK-NEXT: cmpl %r14d, %ebp
+; CHECK-NEXT: jne .LBB20_2
+; CHECK-NEXT: .LBB20_3: # %for.cond.cleanup
+; CHECK-NEXT: popq %rbx
+; CHECK-NEXT: popq %r14
+; CHECK-NEXT: popq %rbp
+; CHECK-NEXT: retq
+entry:
+ %cmp3.not = icmp ule i32 %N, %start
+ br i1 %cmp3.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup:
+ ret void
+
+for.body:
+ %i.04 = phi i32 [ %inc, %for.body ], [ %start, %entry ]
+ %i_with_off = sub nuw i32 %i.04, %start
+ %rem = urem i32 %i_with_off, %rem_amt
+ tail call void @use.i32(i32 %rem)
+ %inc = add nuw i32 %i.04, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+define void @simple_urem_to_sel_non_zero_start_through_sub_no_simplfy(i32 %N, i32 %rem_amt, i32 %start) nounwind {
+; CHECK-LABEL: simple_urem_to_sel_non_zero_start_through_sub_no_simplfy:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: cmpl %edx, %edi
+; CHECK-NEXT: jbe .LBB21_4
+; CHECK-NEXT: # %bb.1: # %for.body.preheader
+; CHECK-NEXT: pushq %r15
+; CHECK-NEXT: pushq %r14
+; CHECK-NEXT: pushq %rbx
+; CHECK-NEXT: movl %edx, %r15d
+; CHECK-NEXT: movl %esi, %ebx
+; CHECK-NEXT: movl %edi, %r14d
+; CHECK-NEXT: negl %r14d
+; CHECK-NEXT: addl $-2, %r15d
+; CHECK-NEXT: .p2align 4, 0x90
+; CHECK-NEXT: .LBB21_2: # %for.body
+; CHECK-NEXT: # =>This Inner Loop Header: Depth=1
+; CHECK-NEXT: movl %r15d, %eax
+; CHECK-NEXT: xorl %edx, %edx
+; CHECK-NEXT: divl %ebx
+; CHECK-NEXT: movl %edx, %edi
+; CHECK-NEXT: callq use.i32 at PLT
+; CHECK-NEXT: leal 1(%r14,%r15), %eax
+; CHECK-NEXT: movl %r15d, %ecx
+; CHECK-NEXT: incl %ecx
+; CHECK-NEXT: cmpl $-2, %eax
+; CHECK-NEXT: movl %ecx, %r15d
+; CHECK-NEXT: jne .LBB21_2
+; CHECK-NEXT: # %bb.3:
+; CHECK-NEXT: popq %rbx
+; CHECK-NEXT: popq %r14
+; CHECK-NEXT: popq %r15
+; CHECK-NEXT: .LBB21_4: # %for.cond.cleanup
+; CHECK-NEXT: retq
+entry:
+ %cmp3.not = icmp ule i32 %N, %start
+ br i1 %cmp3.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup:
+ ret void
+
+for.body:
+ %i.04 = phi i32 [ %inc, %for.body ], [ %start, %entry ]
+ %i_with_off = sub nuw i32 %i.04, 2
+ %rem = urem i32 %i_with_off, %rem_amt
+ tail call void @use.i32(i32 %rem)
+ %inc = add nuw i32 %i.04, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
diff --git a/llvm/test/Transforms/CodeGenPrepare/X86/fold-loop-of-urem.ll b/llvm/test/Transforms/CodeGenPrepare/X86/fold-loop-of-urem.ll
new file mode 100644
index 00000000000000..a019679e65905d
--- /dev/null
+++ b/llvm/test/Transforms/CodeGenPrepare/X86/fold-loop-of-urem.ll
@@ -0,0 +1,858 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 5
+; RUN: opt < %s -mtriple=x86_64-unknown-unknown --loop-simplify -codegenprepare -S | FileCheck %s
+
+declare void @use.i32(i32)
+declare void @use.2xi64(<2 x i64>)
+declare void @do_stuff0()
+declare void @do_stuff1()
+declare i1 @get.i1()
+declare i32 @get.i32()
+
+define void @simple_urem_to_sel(i32 %N, i32 %rem_amt) nounwind {
+; CHECK-LABEL: define void @simple_urem_to_sel(
+; CHECK-SAME: i32 [[N:%.*]], i32 [[REM_AMT:%.*]]) #[[ATTR0:[0-9]+]] {
+; CHECK-NEXT: [[ENTRY:.*:]]
+; CHECK-NEXT: [[CMP3_NOT:%.*]] = icmp eq i32 [[N]], 0
+; CHECK-NEXT: br i1 [[CMP3_NOT]], label %[[FOR_COND_CLEANUP:.*]], label %[[FOR_BODY_PREHEADER:.*]]
+; CHECK: [[FOR_BODY_PREHEADER]]:
+; CHECK-NEXT: br label %[[FOR_BODY:.*]]
+; CHECK: [[FOR_COND_CLEANUP]]:
+; CHECK-NEXT: ret void
+; CHECK: [[FOR_BODY]]:
+; CHECK-NEXT: [[I_04:%.*]] = phi i32 [ [[INC:%.*]], %[[FOR_BODY]] ], [ 0, %[[FOR_BODY_PREHEADER]] ]
+; CHECK-NEXT: [[REM:%.*]] = urem i32 [[I_04]], [[REM_AMT]]
+; CHECK-NEXT: tail call void @use.i32(i32 [[REM]])
+; CHECK-NEXT: [[INC]] = add nuw i32 [[I_04]], 1
+; CHECK-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i32 [[INC]], [[N]]
+; CHECK-NEXT: br i1 [[EXITCOND_NOT]], label %[[FOR_COND_CLEANUP]], label %[[FOR_BODY]]
+;
+entry:
+ %cmp3.not = icmp eq i32 %N, 0
+ br i1 %cmp3.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup:
+ ret void
+
+for.body:
+ %i.04 = phi i32 [ %inc, %for.body ], [ 0, %entry ]
+ %rem = urem i32 %i.04, %rem_amt
+ tail call void @use.i32(i32 %rem)
+ %inc = add nuw i32 %i.04, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+define void @simple_urem_to_sel_nested2(i32 %N, i32 %rem_amt) nounwind {
+; CHECK-LABEL: define void @simple_urem_to_sel_nested2(
+; CHECK-SAME: i32 [[N:%.*]], i32 [[REM_AMT:%.*]]) #[[ATTR0]] {
+; CHECK-NEXT: [[ENTRY:.*:]]
+; CHECK-NEXT: [[CMP3_NOT:%.*]] = icmp eq i32 [[N]], 0
+; CHECK-NEXT: br i1 [[CMP3_NOT]], label %[[FOR_COND_CLEANUP:.*]], label %[[FOR_BODY_PREHEADER:.*]]
+; CHECK: [[FOR_BODY_PREHEADER]]:
+; CHECK-NEXT: br label %[[FOR_BODY:.*]]
+; CHECK: [[FOR_COND_CLEANUP]]:
+; CHECK-NEXT: ret void
+; CHECK: [[FOR_BODY]]:
+; CHECK-NEXT: [[I_04:%.*]] = phi i32 [ [[INC:%.*]], %[[FOR_BODY_TAIL:.*]] ], [ 0, %[[FOR_BODY_PREHEADER]] ]
+; CHECK-NEXT: [[COND0:%.*]] = call i1 @get.i1()
+; CHECK-NEXT: br i1 [[COND0]], label %[[FOR_BODY0:.*]], label %[[FOR_BODY_TAIL]]
+; CHECK: [[FOR_BODY0]]:
+; CHECK-NEXT: [[COND1:%.*]] = call i1 @get.i1()
+; CHECK-NEXT: br i1 [[COND1]], label %[[FOR_BODY1:.*]], label %[[FOR_BODY2:.*]]
+; CHECK: [[FOR_BODY2]]:
+; CHECK-NEXT: [[COND2:%.*]] = call i1 @get.i1()
+; CHECK-NEXT: br i1 [[COND2]], label %[[FOR_BODY1]], label %[[FOR_BODY_TAIL]]
+; CHECK: [[FOR_BODY1]]:
+; CHECK-NEXT: [[REM:%.*]] = urem i32 [[I_04]], [[REM_AMT]]
+; CHECK-NEXT: tail call void @use.i32(i32 [[REM]])
+; CHECK-NEXT: br label %[[FOR_BODY_TAIL]]
+; CHECK: [[FOR_BODY_TAIL]]:
+; CHECK-NEXT: [[INC]] = add nuw i32 [[I_04]], 1
+; CHECK-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i32 [[INC]], [[N]]
+; CHECK-NEXT: br i1 [[EXITCOND_NOT]], label %[[FOR_COND_CLEANUP]], label %[[FOR_BODY]]
+;
+entry:
+ %cmp3.not = icmp eq i32 %N, 0
+ br i1 %cmp3.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup:
+ ret void
+
+for.body:
+ %i.04 = phi i32 [ %inc, %for.body.tail ], [ 0, %entry ]
+ %cond0 = call i1 @get.i1()
+ br i1 %cond0, label %for.body0, label %for.body.tail
+for.body0:
+ %cond1 = call i1 @get.i1()
+ br i1 %cond1, label %for.body1, label %for.body2
+for.body2:
+ %cond2 = call i1 @get.i1()
+ br i1 %cond2, label %for.body1, label %for.body.tail
+for.body1:
+ %rem = urem i32 %i.04, %rem_amt
+ tail call void @use.i32(i32 %rem)
+ br label %for.body.tail
+for.body.tail:
+ %inc = add nuw i32 %i.04, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+define void @simple_urem_fail_bad_incr3(i32 %N, i32 %rem_amt) nounwind {
+; CHECK-LABEL: define void @simple_urem_fail_bad_incr3(
+; CHECK-SAME: i32 [[N:%.*]], i32 [[REM_AMT:%.*]]) #[[ATTR0]] {
+; CHECK-NEXT: [[ENTRY:.*:]]
+; CHECK-NEXT: [[CMP3_NOT:%.*]] = icmp eq i32 [[N]], 0
+; CHECK-NEXT: br i1 [[CMP3_NOT]], label %[[FOR_COND_CLEANUP:.*]], label %[[FOR_BODY_PREHEADER:.*]]
+; CHECK: [[FOR_BODY_PREHEADER]]:
+; CHECK-NEXT: br label %[[FOR_BODY:.*]]
+; CHECK: [[FOR_COND_CLEANUP]]:
+; CHECK-NEXT: ret void
+; CHECK: [[FOR_BODY]]:
+; CHECK-NEXT: [[COND0:%.*]] = call i1 @get.i1()
+; CHECK-NEXT: br i1 [[COND0]], label %[[FOR_BODY0:.*]], label %[[FOR_BODY2:.*]]
+; CHECK: [[FOR_BODY0]]:
+; CHECK-NEXT: [[COND1:%.*]] = call i1 @get.i1()
+; CHECK-NEXT: [[VAL:%.*]] = call i32 @get.i32()
+; CHECK-NEXT: [[INC:%.*]] = add nuw i32 [[VAL]], 1
+; CHECK-NEXT: br i1 [[COND1]], label %[[FOR_BODY1:.*]], label %[[FOR_BODY_TAIL:.*]]
+; CHECK: [[FOR_BODY2]]:
+; CHECK-NEXT: [[COND2:%.*]] = call i1 @get.i1()
+; CHECK-NEXT: br i1 [[COND2]], label %[[FOR_BODY1]], label %[[FOR_BODY_TAIL]]
+; CHECK: [[FOR_BODY1]]:
+; CHECK-NEXT: [[I_04:%.*]] = phi i32 [ [[INC]], %[[FOR_BODY0]] ], [ 0, %[[FOR_BODY2]] ]
+; CHECK-NEXT: [[REM:%.*]] = urem i32 [[I_04]], [[REM_AMT]]
+; CHECK-NEXT: tail call void @use.i32(i32 [[REM]])
+; CHECK-NEXT: br label %[[FOR_BODY_TAIL]]
+; CHECK: [[FOR_BODY_TAIL]]:
+; CHECK-NEXT: [[EXITCOND_NOT:%.*]] = call i1 @get.i1()
+; CHECK-NEXT: br i1 [[EXITCOND_NOT]], label %[[FOR_COND_CLEANUP]], label %[[FOR_BODY]]
+;
+entry:
+ %cmp3.not = icmp eq i32 %N, 0
+ br i1 %cmp3.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup:
+ ret void
+
+for.body:
+ %cond0 = call i1 @get.i1()
+ br i1 %cond0, label %for.body0, label %for.body2
+for.body0:
+ %cond1 = call i1 @get.i1()
+ %val = call i32 @get.i32()
+ %inc = add nuw i32 %val, 1
+ br i1 %cond1, label %for.body1, label %for.body.tail
+for.body2:
+ %cond2 = call i1 @get.i1()
+ br i1 %cond2, label %for.body1, label %for.body.tail
+for.body1:
+ %i.04 = phi i32 [ %inc, %for.body0], [ 0, %for.body2 ]
+ %rem = urem i32 %i.04, %rem_amt
+ tail call void @use.i32(i32 %rem)
+ br label %for.body.tail
+for.body.tail:
+ %exitcond.not = call i1 @get.i1()
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+define void @simple_urem_to_sel_vec(<2 x i64> %rem_amt) nounwind {
+; CHECK-LABEL: define void @simple_urem_to_sel_vec(
+; CHECK-SAME: <2 x i64> [[REM_AMT:%.*]]) #[[ATTR0]] {
+; CHECK-NEXT: [[ENTRY:.*]]:
+; CHECK-NEXT: br label %[[FOR_BODY:.*]]
+; CHECK: [[FOR_COND_CLEANUP:.*]]:
+; CHECK-NEXT: ret void
+; CHECK: [[FOR_BODY]]:
+; CHECK-NEXT: [[I_04:%.*]] = phi <2 x i64> [ [[INC:%.*]], %[[FOR_BODY]] ], [ zeroinitializer, %[[ENTRY]] ]
+; CHECK-NEXT: [[REM:%.*]] = urem <2 x i64> [[I_04]], [[REM_AMT]]
+; CHECK-NEXT: tail call void @use.2xi64(<2 x i64> [[REM]])
+; CHECK-NEXT: [[INC]] = add nuw <2 x i64> [[I_04]], <i64 1, i64 1>
+; CHECK-NEXT: [[EXITCOND_NOT:%.*]] = call i1 @get.i1()
+; CHECK-NEXT: br i1 [[EXITCOND_NOT]], label %[[FOR_COND_CLEANUP]], label %[[FOR_BODY]]
+;
+entry:
+ br label %for.body
+
+for.cond.cleanup:
+ ret void
+
+for.body:
+ %i.04 = phi <2 x i64> [ %inc, %for.body ], [ zeroinitializer, %entry ]
+ %rem = urem <2 x i64> %i.04, %rem_amt
+ tail call void @use.2xi64(<2 x i64> %rem)
+ %inc = add nuw <2 x i64> %i.04, <i64 1, i64 1>
+ %exitcond.not = call i1 @get.i1()
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+define void @simple_urem_fail_bad_incr(i32 %N, i32 %rem_amt) nounwind {
+; CHECK-LABEL: define void @simple_urem_fail_bad_incr(
+; CHECK-SAME: i32 [[N:%.*]], i32 [[REM_AMT:%.*]]) #[[ATTR0]] {
+; CHECK-NEXT: [[ENTRY:.*:]]
+; CHECK-NEXT: [[CMP3_NOT:%.*]] = icmp eq i32 [[N]], 0
+; CHECK-NEXT: br i1 [[CMP3_NOT]], label %[[FOR_COND_CLEANUP:.*]], label %[[FOR_BODY_PREHEADER:.*]]
+; CHECK: [[FOR_BODY_PREHEADER]]:
+; CHECK-NEXT: br label %[[FOR_BODY:.*]]
+; CHECK: [[FOR_COND_CLEANUP]]:
+; CHECK-NEXT: ret void
+; CHECK: [[FOR_BODY]]:
+; CHECK-NEXT: [[I_03:%.*]] = phi i32 [ [[INC:%.*]], %[[FOR_BODY_TAIL:.*]] ], [ 0, %[[FOR_BODY_PREHEADER]] ]
+; CHECK-NEXT: [[COND0:%.*]] = call i1 @get.i1()
+; CHECK-NEXT: br i1 [[COND0]], label %[[FOR_BODY0:.*]], label %[[FOR_BODY_TAIL]]
+; CHECK: [[FOR_BODY0]]:
+; CHECK-NEXT: [[SOME_VAL:%.*]] = call i32 @get.i32()
+; CHECK-NEXT: br label %[[FOR_BODY_TAIL]]
+; CHECK: [[FOR_BODY_TAIL]]:
+; CHECK-NEXT: [[I_04:%.*]] = phi i32 [ [[I_03]], %[[FOR_BODY]] ], [ [[SOME_VAL]], %[[FOR_BODY0]] ]
+; CHECK-NEXT: [[REM:%.*]] = urem i32 [[I_04]], [[REM_AMT]]
+; CHECK-NEXT: tail call void @use.i32(i32 [[REM]])
+; CHECK-NEXT: [[INC]] = add nuw i32 [[I_04]], 1
+; CHECK-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i32 [[INC]], [[N]]
+; CHECK-NEXT: br i1 [[EXITCOND_NOT]], label %[[FOR_COND_CLEANUP]], label %[[FOR_BODY]]
+;
+entry:
+ %cmp3.not = icmp eq i32 %N, 0
+ br i1 %cmp3.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup:
+ ret void
+
+for.body:
+ %i.03 = phi i32 [ %inc, %for.body.tail ], [ 0, %entry ]
+ %cond0 = call i1 @get.i1()
+ br i1 %cond0, label %for.body0, label %for.body.tail
+for.body0:
+ %some_val = call i32 @get.i32()
+ br label %for.body.tail
+
+for.body.tail:
+ %i.04 = phi i32 [ %i.03, %for.body ], [ %some_val, %for.body0 ]
+ %rem = urem i32 %i.04, %rem_amt
+ tail call void @use.i32(i32 %rem)
+ %inc = add nuw i32 %i.04, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+define void @simple_urem_to_sel_second_acc(i32 %N, i32 %rem_amt) nounwind {
+; CHECK-LABEL: define void @simple_urem_to_sel_second_acc(
+; CHECK-SAME: i32 [[N:%.*]], i32 [[REM_AMT:%.*]]) #[[ATTR0]] {
+; CHECK-NEXT: [[ENTRY:.*:]]
+; CHECK-NEXT: [[CMP3_NOT:%.*]] = icmp ult i32 [[N]], 2
+; CHECK-NEXT: br i1 [[CMP3_NOT]], label %[[FOR_COND_CLEANUP:.*]], label %[[FOR_BODY_PREHEADER:.*]]
+; CHECK: [[FOR_BODY_PREHEADER]]:
+; CHECK-NEXT: br label %[[FOR_BODY:.*]]
+; CHECK: [[FOR_COND_CLEANUP]]:
+; CHECK-NEXT: ret void
+; CHECK: [[FOR_BODY]]:
+; CHECK-NEXT: [[I_04:%.*]] = phi i32 [ [[INC:%.*]], %[[FOR_BODY]] ], [ 0, %[[FOR_BODY_PREHEADER]] ]
+; CHECK-NEXT: [[I_05:%.*]] = phi i32 [ [[INC2:%.*]], %[[FOR_BODY]] ], [ 1, %[[FOR_BODY_PREHEADER]] ]
+; CHECK-NEXT: [[REM:%.*]] = urem i32 [[I_04]], [[REM_AMT]]
+; CHECK-NEXT: tail call void @use.i32(i32 [[REM]])
+; CHECK-NEXT: [[INC]] = add nuw i32 [[I_04]], 1
+; CHECK-NEXT: [[INC2]] = add nuw i32 [[I_05]], 2
+; CHECK-NEXT: [[EXITCOND_NOT:%.*]] = icmp ugt i32 [[INC2]], [[N]]
+; CHECK-NEXT: br i1 [[EXITCOND_NOT]], label %[[FOR_COND_CLEANUP]], label %[[FOR_BODY]]
+;
+entry:
+ %cmp3.not = icmp ult i32 %N, 2
+ br i1 %cmp3.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup:
+ ret void
+
+for.body:
+ %i.04 = phi i32 [ %inc, %for.body ], [ 0, %entry ]
+ %i.05 = phi i32 [ %inc2, %for.body ], [ 1, %entry ]
+ %rem = urem i32 %i.04, %rem_amt
+ tail call void @use.i32(i32 %rem)
+ %inc = add nuw i32 %i.04, 1
+ %inc2 = add nuw i32 %i.05, 2
+ %exitcond.not = icmp ugt i32 %inc2, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+define void @simple_urem_fail_srem(i32 %N, i32 %rem_amt) nounwind {
+; CHECK-LABEL: define void @simple_urem_fail_srem(
+; CHECK-SAME: i32 [[N:%.*]], i32 [[REM_AMT:%.*]]) #[[ATTR0]] {
+; CHECK-NEXT: [[ENTRY:.*:]]
+; CHECK-NEXT: [[CMP3_NOT:%.*]] = icmp eq i32 [[N]], 0
+; CHECK-NEXT: br i1 [[CMP3_NOT]], label %[[FOR_COND_CLEANUP:.*]], label %[[FOR_BODY_PREHEADER:.*]]
+; CHECK: [[FOR_BODY_PREHEADER]]:
+; CHECK-NEXT: br label %[[FOR_BODY:.*]]
+; CHECK: [[FOR_COND_CLEANUP]]:
+; CHECK-NEXT: ret void
+; CHECK: [[FOR_BODY]]:
+; CHECK-NEXT: [[I_04:%.*]] = phi i32 [ [[INC:%.*]], %[[FOR_BODY]] ], [ 0, %[[FOR_BODY_PREHEADER]] ]
+; CHECK-NEXT: [[REM:%.*]] = srem i32 [[I_04]], [[REM_AMT]]
+; CHECK-NEXT: tail call void @use.i32(i32 [[REM]])
+; CHECK-NEXT: [[INC]] = add nuw i32 [[I_04]], 1
+; CHECK-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i32 [[INC]], [[N]]
+; CHECK-NEXT: br i1 [[EXITCOND_NOT]], label %[[FOR_COND_CLEANUP]], label %[[FOR_BODY]]
+;
+entry:
+ %cmp3.not = icmp eq i32 %N, 0
+ br i1 %cmp3.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup:
+ ret void
+
+for.body:
+ %i.04 = phi i32 [ %inc, %for.body ], [ 0, %entry ]
+ %rem = srem i32 %i.04, %rem_amt
+ tail call void @use.i32(i32 %rem)
+ %inc = add nuw i32 %i.04, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+define void @simple_urem_fail_missing_nuw(i32 %N, i32 %rem_amt) nounwind {
+; CHECK-LABEL: define void @simple_urem_fail_missing_nuw(
+; CHECK-SAME: i32 [[N:%.*]], i32 [[REM_AMT:%.*]]) #[[ATTR0]] {
+; CHECK-NEXT: [[ENTRY:.*:]]
+; CHECK-NEXT: [[CMP3_NOT:%.*]] = icmp eq i32 [[N]], 0
+; CHECK-NEXT: br i1 [[CMP3_NOT]], label %[[FOR_COND_CLEANUP:.*]], label %[[FOR_BODY_PREHEADER:.*]]
+; CHECK: [[FOR_BODY_PREHEADER]]:
+; CHECK-NEXT: br label %[[FOR_BODY:.*]]
+; CHECK: [[FOR_COND_CLEANUP]]:
+; CHECK-NEXT: ret void
+; CHECK: [[FOR_BODY]]:
+; CHECK-NEXT: [[I_04:%.*]] = phi i32 [ [[INC:%.*]], %[[FOR_BODY]] ], [ 0, %[[FOR_BODY_PREHEADER]] ]
+; CHECK-NEXT: [[REM:%.*]] = urem i32 [[I_04]], [[REM_AMT]]
+; CHECK-NEXT: tail call void @use.i32(i32 [[REM]])
+; CHECK-NEXT: [[INC]] = add nsw i32 [[I_04]], 1
+; CHECK-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i32 [[INC]], [[N]]
+; CHECK-NEXT: br i1 [[EXITCOND_NOT]], label %[[FOR_COND_CLEANUP]], label %[[FOR_BODY]]
+;
+entry:
+ %cmp3.not = icmp eq i32 %N, 0
+ br i1 %cmp3.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup:
+ ret void
+
+for.body:
+ %i.04 = phi i32 [ %inc, %for.body ], [ 0, %entry ]
+ %rem = urem i32 %i.04, %rem_amt
+ tail call void @use.i32(i32 %rem)
+ %inc = add nsw i32 %i.04, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+define void @simple_urem_fail_bad_incr2(i32 %N, i32 %rem_amt) nounwind {
+; CHECK-LABEL: define void @simple_urem_fail_bad_incr2(
+; CHECK-SAME: i32 [[N:%.*]], i32 [[REM_AMT:%.*]]) #[[ATTR0]] {
+; CHECK-NEXT: [[ENTRY:.*:]]
+; CHECK-NEXT: [[CMP3_NOT:%.*]] = icmp eq i32 [[N]], 0
+; CHECK-NEXT: br i1 [[CMP3_NOT]], label %[[FOR_COND_CLEANUP:.*]], label %[[FOR_BODY_PREHEADER:.*]]
+; CHECK: [[FOR_BODY_PREHEADER]]:
+; CHECK-NEXT: br label %[[FOR_BODY:.*]]
+; CHECK: [[FOR_COND_CLEANUP]]:
+; CHECK-NEXT: ret void
+; CHECK: [[FOR_BODY]]:
+; CHECK-NEXT: [[I_04:%.*]] = phi i32 [ [[INC:%.*]], %[[FOR_BODY]] ], [ 0, %[[FOR_BODY_PREHEADER]] ]
+; CHECK-NEXT: [[REM:%.*]] = urem i32 [[I_04]], [[REM_AMT]]
+; CHECK-NEXT: tail call void @use.i32(i32 [[REM]])
+; CHECK-NEXT: [[INC]] = add nuw i32 [[I_04]], 2
+; CHECK-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i32 [[INC]], [[N]]
+; CHECK-NEXT: br i1 [[EXITCOND_NOT]], label %[[FOR_COND_CLEANUP]], label %[[FOR_BODY]]
+;
+entry:
+ %cmp3.not = icmp eq i32 %N, 0
+ br i1 %cmp3.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup:
+ ret void
+
+for.body:
+ %i.04 = phi i32 [ %inc, %for.body ], [ 0, %entry ]
+ %rem = urem i32 %i.04, %rem_amt
+ tail call void @use.i32(i32 %rem)
+ %inc = add nuw i32 %i.04, 2
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+define void @simple_urem_non_zero_entry4(i32 %N, i32 %rem_amt) nounwind {
+; CHECK-LABEL: define void @simple_urem_non_zero_entry4(
+; CHECK-SAME: i32 [[N:%.*]], i32 [[REM_AMT:%.*]]) #[[ATTR0]] {
+; CHECK-NEXT: [[ENTRY:.*:]]
+; CHECK-NEXT: [[CMP3_NOT:%.*]] = icmp eq i32 [[N]], 0
+; CHECK-NEXT: br i1 [[CMP3_NOT]], label %[[FOR_COND_CLEANUP:.*]], label %[[FOR_BODY_PREHEADER:.*]]
+; CHECK: [[FOR_BODY_PREHEADER]]:
+; CHECK-NEXT: br label %[[FOR_BODY:.*]]
+; CHECK: [[FOR_COND_CLEANUP]]:
+; CHECK-NEXT: ret void
+; CHECK: [[FOR_BODY]]:
+; CHECK-NEXT: [[I_04:%.*]] = phi i32 [ [[INC:%.*]], %[[FOR_BODY]] ], [ 4, %[[FOR_BODY_PREHEADER]] ]
+; CHECK-NEXT: [[REM:%.*]] = urem i32 [[I_04]], [[REM_AMT]]
+; CHECK-NEXT: tail call void @use.i32(i32 [[REM]])
+; CHECK-NEXT: [[INC]] = add nuw i32 [[I_04]], 1
+; CHECK-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i32 [[INC]], [[N]]
+; CHECK-NEXT: br i1 [[EXITCOND_NOT]], label %[[FOR_COND_CLEANUP]], label %[[FOR_BODY]]
+;
+entry:
+ %cmp3.not = icmp eq i32 %N, 0
+ br i1 %cmp3.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup:
+ ret void
+
+for.body:
+ %i.04 = phi i32 [ %inc, %for.body ], [ 4, %entry ]
+ %rem = urem i32 %i.04, %rem_amt
+ tail call void @use.i32(i32 %rem)
+ %inc = add nuw i32 %i.04, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+define void @simple_urem_skip_const_rem_amt(i32 %N) nounwind {
+; CHECK-LABEL: define void @simple_urem_skip_const_rem_amt(
+; CHECK-SAME: i32 [[N:%.*]]) #[[ATTR0]] {
+; CHECK-NEXT: [[ENTRY:.*:]]
+; CHECK-NEXT: [[CMP3_NOT:%.*]] = icmp eq i32 [[N]], 0
+; CHECK-NEXT: br i1 [[CMP3_NOT]], label %[[FOR_COND_CLEANUP:.*]], label %[[FOR_BODY_PREHEADER:.*]]
+; CHECK: [[FOR_BODY_PREHEADER]]:
+; CHECK-NEXT: br label %[[FOR_BODY:.*]]
+; CHECK: [[FOR_COND_CLEANUP]]:
+; CHECK-NEXT: ret void
+; CHECK: [[FOR_BODY]]:
+; CHECK-NEXT: [[I_04:%.*]] = phi i32 [ [[INC:%.*]], %[[FOR_BODY]] ], [ 4, %[[FOR_BODY_PREHEADER]] ]
+; CHECK-NEXT: [[REM:%.*]] = urem i32 [[I_04]], 19
+; CHECK-NEXT: tail call void @use.i32(i32 [[REM]])
+; CHECK-NEXT: [[INC]] = add nuw i32 [[I_04]], 1
+; CHECK-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i32 [[INC]], [[N]]
+; CHECK-NEXT: br i1 [[EXITCOND_NOT]], label %[[FOR_COND_CLEANUP]], label %[[FOR_BODY]]
+;
+entry:
+ %cmp3.not = icmp eq i32 %N, 0
+ br i1 %cmp3.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup:
+ ret void
+
+for.body:
+ %i.04 = phi i32 [ %inc, %for.body ], [ 4, %entry ]
+ %rem = urem i32 %i.04, 19
+ tail call void @use.i32(i32 %rem)
+ %inc = add nuw i32 %i.04, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+define void @simple_urem_fail_no_preheader_non_canonical(i32 %N, i32 %rem_amt) nounwind {
+; CHECK-LABEL: define void @simple_urem_fail_no_preheader_non_canonical(
+; CHECK-SAME: i32 [[N:%.*]], i32 [[REM_AMT:%.*]]) #[[ATTR0]] {
+; CHECK-NEXT: [[ENTRY:.*]]:
+; CHECK-NEXT: [[CMP3_NOT:%.*]] = icmp eq i32 [[N]], 0
+; CHECK-NEXT: br i1 [[CMP3_NOT]], label %[[FOR_BODY_PREHEADER:.*]], label %[[FOR_BODY1:.*]]
+; CHECK: [[FOR_COND_CLEANUP:.*]]:
+; CHECK-NEXT: ret void
+; CHECK: [[FOR_BODY1]]:
+; CHECK-NEXT: br label %[[FOR_BODY_PREHEADER]]
+; CHECK: [[FOR_BODY_PREHEADER]]:
+; CHECK-NEXT: [[I_04_PH:%.*]] = phi i32 [ 1, %[[FOR_BODY1]] ], [ 0, %[[ENTRY]] ]
+; CHECK-NEXT: br label %[[FOR_BODY:.*]]
+; CHECK: [[FOR_BODY]]:
+; CHECK-NEXT: [[I_04:%.*]] = phi i32 [ [[INC:%.*]], %[[FOR_BODY]] ], [ [[I_04_PH]], %[[FOR_BODY_PREHEADER]] ]
+; CHECK-NEXT: [[REM:%.*]] = urem i32 [[I_04]], [[REM_AMT]]
+; CHECK-NEXT: tail call void @use.i32(i32 [[REM]])
+; CHECK-NEXT: [[INC]] = add nuw i32 [[I_04]], 1
+; CHECK-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i32 [[INC]], [[N]]
+; CHECK-NEXT: br i1 [[EXITCOND_NOT]], label %[[FOR_COND_CLEANUP]], label %[[FOR_BODY]]
+;
+entry:
+ %cmp3.not = icmp eq i32 %N, 0
+ br i1 %cmp3.not, label %for.body0, label %for.body1
+
+for.cond.cleanup:
+ ret void
+
+for.body0:
+ br label %for.body
+
+for.body1:
+ br label %for.body
+
+for.body:
+ %i.04 = phi i32 [ %inc, %for.body ], [ 0, %for.body0 ], [ 1, %for.body1 ]
+ %rem = urem i32 %i.04, %rem_amt
+ tail call void @use.i32(i32 %rem)
+ %inc = add nuw i32 %i.04, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+define void @simple_urem_multi_latch_non_canonical(i32 %N, i32 %rem_amt) nounwind {
+; CHECK-LABEL: define void @simple_urem_multi_latch_non_canonical(
+; CHECK-SAME: i32 [[N:%.*]], i32 [[REM_AMT:%.*]]) #[[ATTR0]] {
+; CHECK-NEXT: [[ENTRY:.*:]]
+; CHECK-NEXT: [[CMP3_NOT:%.*]] = icmp eq i32 [[N]], 0
+; CHECK-NEXT: br i1 [[CMP3_NOT]], label %[[FOR_COND_CLEANUP:.*]], label %[[FOR_BODY_PREHEADER:.*]]
+; CHECK: [[FOR_BODY_PREHEADER]]:
+; CHECK-NEXT: br label %[[FOR_BODY:.*]]
+; CHECK: [[FOR_COND_CLEANUP]]:
+; CHECK-NEXT: ret void
+; CHECK: [[FOR_BODY]]:
+; CHECK-NEXT: [[I_04:%.*]] = phi i32 [ 0, %[[FOR_BODY_PREHEADER]] ], [ [[INC:%.*]], %[[FOR_BODY]] ], [ [[INC]], %[[FOR_BODY0:.*]] ]
+; CHECK-NEXT: [[REM:%.*]] = urem i32 [[I_04]], [[REM_AMT]]
+; CHECK-NEXT: tail call void @use.i32(i32 [[REM]])
+; CHECK-NEXT: [[INC]] = add nuw i32 [[I_04]], 1
+; CHECK-NEXT: [[COND:%.*]] = call i1 @get.i1()
+; CHECK-NEXT: call void @do_stuff0()
+; CHECK-NEXT: br i1 [[COND]], label %[[FOR_BODY0]], label %[[FOR_BODY]]
+; CHECK: [[FOR_BODY0]]:
+; CHECK-NEXT: call void @do_stuff1()
+; CHECK-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i32 [[INC]], [[N]]
+; CHECK-NEXT: br i1 [[EXITCOND_NOT]], label %[[FOR_COND_CLEANUP]], label %[[FOR_BODY]]
+;
+entry:
+ %cmp3.not = icmp eq i32 %N, 0
+ br i1 %cmp3.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup:
+ ret void
+
+for.body:
+ %i.04 = phi i32 [ %inc, %for.body ], [ %inc, %for.body0 ], [ 0, %entry ]
+ %rem = urem i32 %i.04, %rem_amt
+ tail call void @use.i32(i32 %rem)
+ %inc = add nuw i32 %i.04, 1
+ %cond = call i1 @get.i1()
+ call void @do_stuff0()
+ br i1 %cond, label %for.body0, label %for.body
+for.body0:
+ call void @do_stuff1()
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+define void @simple_urem_fail_bad_loop(i32 %N, i32 %rem_amt) nounwind {
+; CHECK-LABEL: define void @simple_urem_fail_bad_loop(
+; CHECK-SAME: i32 [[N:%.*]], i32 [[REM_AMT:%.*]]) #[[ATTR0]] {
+; CHECK-NEXT: [[ENTRY:.*]]:
+; CHECK-NEXT: [[CALL:%.*]] = call i32 @get.i32()
+; CHECK-NEXT: [[TOBOOL_NOT:%.*]] = icmp eq i32 [[CALL]], 0
+; CHECK-NEXT: br i1 [[TOBOOL_NOT]], label %[[FOR_COND:.*]], label %[[HALFWAY:.*]]
+; CHECK: [[FOR_COND]]:
+; CHECK-NEXT: [[I_0:%.*]] = phi i32 [ [[INC:%.*]], %[[HALFWAY]] ], [ 0, %[[ENTRY]] ]
+; CHECK-NEXT: [[CMP:%.*]] = icmp ult i32 [[I_0]], [[N]]
+; CHECK-NEXT: br i1 [[CMP]], label %[[FOR_BODY:.*]], label %[[FOR_END:.*]]
+; CHECK: [[FOR_BODY]]:
+; CHECK-NEXT: [[XOR:%.*]] = xor i32 [[I_0]], 1
+; CHECK-NEXT: call void @use.i32(i32 [[XOR]])
+; CHECK-NEXT: br label %[[HALFWAY]]
+; CHECK: [[HALFWAY]]:
+; CHECK-NEXT: [[I_1:%.*]] = phi i32 [ poison, %[[ENTRY]] ], [ [[I_0]], %[[FOR_BODY]] ]
+; CHECK-NEXT: [[REM:%.*]] = urem i32 [[I_1]], [[REM_AMT]]
+; CHECK-NEXT: call void @use.i32(i32 [[REM]])
+; CHECK-NEXT: [[INC]] = add nuw i32 [[I_1]], 1
+; CHECK-NEXT: br label %[[FOR_COND]]
+; CHECK: [[FOR_END]]:
+; CHECK-NEXT: ret void
+;
+entry:
+ %call = call i32 @get.i32()
+ %tobool.not = icmp eq i32 %call, 0
+ br i1 %tobool.not, label %for.cond, label %halfway
+
+for.cond:
+ %i.0 = phi i32 [ %inc, %halfway ], [ 0, %entry ]
+ %cmp = icmp ult i32 %i.0, %N
+ br i1 %cmp, label %for.body, label %for.end
+
+for.body:
+ %xor = xor i32 %i.0, 1
+ call void @use.i32(i32 %xor)
+ br label %halfway
+
+halfway:
+ %i.1 = phi i32 [ poison, %entry ], [ %i.0, %for.body ]
+ %rem = urem i32 %i.1, %rem_amt
+ call void @use.i32(i32 %rem)
+ %inc = add nuw i32 %i.1, 1
+ br label %for.cond
+
+for.end:
+ ret void
+}
+
+define void @simple_urem_fail_intermediate_inc(i32 %N, i32 %rem_amt) nounwind {
+; CHECK-LABEL: define void @simple_urem_fail_intermediate_inc(
+; CHECK-SAME: i32 [[N:%.*]], i32 [[REM_AMT:%.*]]) #[[ATTR0]] {
+; CHECK-NEXT: [[ENTRY:.*:]]
+; CHECK-NEXT: [[CMP3_NOT:%.*]] = icmp eq i32 [[N]], 0
+; CHECK-NEXT: br i1 [[CMP3_NOT]], label %[[FOR_COND_CLEANUP:.*]], label %[[FOR_BODY_PREHEADER:.*]]
+; CHECK: [[FOR_BODY_PREHEADER]]:
+; CHECK-NEXT: br label %[[FOR_BODY:.*]]
+; CHECK: [[FOR_COND_CLEANUP]]:
+; CHECK-NEXT: ret void
+; CHECK: [[FOR_BODY]]:
+; CHECK-NEXT: [[I_04:%.*]] = phi i32 [ [[INC:%.*]], %[[FOR_BODY]] ], [ 0, %[[FOR_BODY_PREHEADER]] ]
+; CHECK-NEXT: [[INC2:%.*]] = add nuw i32 [[I_04]], 1
+; CHECK-NEXT: [[REM:%.*]] = urem i32 [[INC2]], [[REM_AMT]]
+; CHECK-NEXT: tail call void @use.i32(i32 [[REM]])
+; CHECK-NEXT: [[INC]] = add nuw i32 [[I_04]], 1
+; CHECK-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i32 [[INC]], [[N]]
+; CHECK-NEXT: br i1 [[EXITCOND_NOT]], label %[[FOR_COND_CLEANUP]], label %[[FOR_BODY]]
+;
+entry:
+ %cmp3.not = icmp eq i32 %N, 0
+ br i1 %cmp3.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup:
+ ret void
+
+for.body:
+ %i.04 = phi i32 [ %inc, %for.body ], [ 0, %entry ]
+ %inc2 = add nuw i32 %i.04, 1
+ %rem = urem i32 %inc2, %rem_amt
+ tail call void @use.i32(i32 %rem)
+ %inc = add nuw i32 %i.04, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+define void @weird_loop(i64 %sub.ptr.div.i56) personality ptr null {
+; CHECK-LABEL: define void @weird_loop(
+; CHECK-SAME: i64 [[SUB_PTR_DIV_I56:%.*]]) personality ptr null {
+; CHECK-NEXT: [[ENTRY:.*:]]
+; CHECK-NEXT: [[ADD74_US:%.*]] = add nuw i64 0, 1
+; CHECK-NEXT: br label %[[FOR_BODY:.*]]
+; CHECK: [[FOR_BODY]]:
+; CHECK-NEXT: [[REM_US:%.*]] = urem i64 0, [[SUB_PTR_DIV_I56]]
+; CHECK-NEXT: br label %[[FOR_BODY]]
+;
+entry:
+ br label %for.preheader
+
+for.preheader:
+ %i57.0540.us = phi i64 [ 0, %entry ], [ %add74.us, %for.body ]
+ %add74.us = add nuw i64 %i57.0540.us, 1
+ br label %for.body
+
+for.body:
+ %rem.us = urem i64 %i57.0540.us, %sub.ptr.div.i56
+ br i1 false, label %for.preheader, label %for.body
+}
+
+define void @simple_urem_to_sel_non_zero_start(i32 %N, i32 %rem_amt) nounwind {
+; CHECK-LABEL: define void @simple_urem_to_sel_non_zero_start(
+; CHECK-SAME: i32 [[N:%.*]], i32 [[REM_AMT:%.*]]) #[[ATTR0]] {
+; CHECK-NEXT: [[ENTRY:.*:]]
+; CHECK-NEXT: [[CMP3_NOT:%.*]] = icmp ult i32 [[N]], 3
+; CHECK-NEXT: br i1 [[CMP3_NOT]], label %[[FOR_COND_CLEANUP:.*]], label %[[FOR_BODY_PREHEADER:.*]]
+; CHECK: [[FOR_BODY_PREHEADER]]:
+; CHECK-NEXT: br label %[[FOR_BODY:.*]]
+; CHECK: [[FOR_COND_CLEANUP]]:
+; CHECK-NEXT: ret void
+; CHECK: [[FOR_BODY]]:
+; CHECK-NEXT: [[I_04:%.*]] = phi i32 [ [[INC:%.*]], %[[FOR_BODY]] ], [ 2, %[[FOR_BODY_PREHEADER]] ]
+; CHECK-NEXT: [[REM:%.*]] = urem i32 [[I_04]], [[REM_AMT]]
+; CHECK-NEXT: tail call void @use.i32(i32 [[REM]])
+; CHECK-NEXT: [[INC]] = add nuw i32 [[I_04]], 1
+; CHECK-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i32 [[INC]], [[N]]
+; CHECK-NEXT: br i1 [[EXITCOND_NOT]], label %[[FOR_COND_CLEANUP]], label %[[FOR_BODY]]
+;
+entry:
+ %cmp3.not = icmp ult i32 %N, 3
+ br i1 %cmp3.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup:
+ ret void
+
+for.body:
+ %i.04 = phi i32 [ %inc, %for.body ], [ 2, %entry ]
+ %rem = urem i32 %i.04, %rem_amt
+ tail call void @use.i32(i32 %rem)
+ %inc = add nuw i32 %i.04, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+define void @simple_urem_to_sel_non_zero_start_through_add(i32 %N, i32 %rem_amt_in) nounwind {
+; CHECK-LABEL: define void @simple_urem_to_sel_non_zero_start_through_add(
+; CHECK-SAME: i32 [[N:%.*]], i32 [[REM_AMT_IN:%.*]]) #[[ATTR0]] {
+; CHECK-NEXT: [[ENTRY:.*:]]
+; CHECK-NEXT: [[REM_AMT:%.*]] = or i32 [[REM_AMT_IN]], 16
+; CHECK-NEXT: [[CMP3_NOT:%.*]] = icmp ult i32 [[N]], 3
+; CHECK-NEXT: br i1 [[CMP3_NOT]], label %[[FOR_COND_CLEANUP:.*]], label %[[FOR_BODY_PREHEADER:.*]]
+; CHECK: [[FOR_BODY_PREHEADER]]:
+; CHECK-NEXT: br label %[[FOR_BODY:.*]]
+; CHECK: [[FOR_COND_CLEANUP]]:
+; CHECK-NEXT: ret void
+; CHECK: [[FOR_BODY]]:
+; CHECK-NEXT: [[I_04:%.*]] = phi i32 [ [[INC:%.*]], %[[FOR_BODY]] ], [ 2, %[[FOR_BODY_PREHEADER]] ]
+; CHECK-NEXT: [[I_WITH_OFF:%.*]] = add nuw i32 [[I_04]], 5
+; CHECK-NEXT: [[REM:%.*]] = urem i32 [[I_WITH_OFF]], [[REM_AMT]]
+; CHECK-NEXT: tail call void @use.i32(i32 [[REM]])
+; CHECK-NEXT: [[INC]] = add nuw i32 [[I_04]], 1
+; CHECK-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i32 [[INC]], [[N]]
+; CHECK-NEXT: br i1 [[EXITCOND_NOT]], label %[[FOR_COND_CLEANUP]], label %[[FOR_BODY]]
+;
+entry:
+ %rem_amt = or i32 %rem_amt_in, 16
+ %cmp3.not = icmp ult i32 %N, 3
+ br i1 %cmp3.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup:
+ ret void
+
+for.body:
+ %i.04 = phi i32 [ %inc, %for.body ], [ 2, %entry ]
+ %i_with_off = add nuw i32 %i.04, 5
+ %rem = urem i32 %i_with_off, %rem_amt
+ tail call void @use.i32(i32 %rem)
+ %inc = add nuw i32 %i.04, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+define void @simple_urem_to_sel_non_zero_start_through_add_fail_missing_nuw(i32 %N, i32 %rem_amt_in) nounwind {
+; CHECK-LABEL: define void @simple_urem_to_sel_non_zero_start_through_add_fail_missing_nuw(
+; CHECK-SAME: i32 [[N:%.*]], i32 [[REM_AMT_IN:%.*]]) #[[ATTR0]] {
+; CHECK-NEXT: [[ENTRY:.*:]]
+; CHECK-NEXT: [[REM_AMT:%.*]] = or i32 [[REM_AMT_IN]], 16
+; CHECK-NEXT: [[CMP3_NOT:%.*]] = icmp ult i32 [[N]], 3
+; CHECK-NEXT: br i1 [[CMP3_NOT]], label %[[FOR_COND_CLEANUP:.*]], label %[[FOR_BODY_PREHEADER:.*]]
+; CHECK: [[FOR_BODY_PREHEADER]]:
+; CHECK-NEXT: br label %[[FOR_BODY:.*]]
+; CHECK: [[FOR_COND_CLEANUP]]:
+; CHECK-NEXT: ret void
+; CHECK: [[FOR_BODY]]:
+; CHECK-NEXT: [[I_04:%.*]] = phi i32 [ [[INC:%.*]], %[[FOR_BODY]] ], [ 2, %[[FOR_BODY_PREHEADER]] ]
+; CHECK-NEXT: [[I_WITH_OFF:%.*]] = add i32 [[I_04]], 5
+; CHECK-NEXT: [[REM:%.*]] = urem i32 [[I_WITH_OFF]], [[REM_AMT]]
+; CHECK-NEXT: tail call void @use.i32(i32 [[REM]])
+; CHECK-NEXT: [[INC]] = add nuw i32 [[I_04]], 1
+; CHECK-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i32 [[INC]], [[N]]
+; CHECK-NEXT: br i1 [[EXITCOND_NOT]], label %[[FOR_COND_CLEANUP]], label %[[FOR_BODY]]
+;
+entry:
+ %rem_amt = or i32 %rem_amt_in, 16
+ %cmp3.not = icmp ult i32 %N, 3
+ br i1 %cmp3.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup:
+ ret void
+
+for.body:
+ %i.04 = phi i32 [ %inc, %for.body ], [ 2, %entry ]
+ %i_with_off = add i32 %i.04, 5
+ %rem = urem i32 %i_with_off, %rem_amt
+ tail call void @use.i32(i32 %rem)
+ %inc = add nuw i32 %i.04, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+define void @simple_urem_to_sel_non_zero_start_through_add_fail_no_simplify_rem(i32 %N, i32 %rem_amt) nounwind {
+; CHECK-LABEL: define void @simple_urem_to_sel_non_zero_start_through_add_fail_no_simplify_rem(
+; CHECK-SAME: i32 [[N:%.*]], i32 [[REM_AMT:%.*]]) #[[ATTR0]] {
+; CHECK-NEXT: [[ENTRY:.*:]]
+; CHECK-NEXT: [[CMP3_NOT:%.*]] = icmp ult i32 [[N]], 3
+; CHECK-NEXT: br i1 [[CMP3_NOT]], label %[[FOR_COND_CLEANUP:.*]], label %[[FOR_BODY_PREHEADER:.*]]
+; CHECK: [[FOR_BODY_PREHEADER]]:
+; CHECK-NEXT: br label %[[FOR_BODY:.*]]
+; CHECK: [[FOR_COND_CLEANUP]]:
+; CHECK-NEXT: ret void
+; CHECK: [[FOR_BODY]]:
+; CHECK-NEXT: [[I_04:%.*]] = phi i32 [ [[INC:%.*]], %[[FOR_BODY]] ], [ 2, %[[FOR_BODY_PREHEADER]] ]
+; CHECK-NEXT: [[I_WITH_OFF:%.*]] = add nuw i32 [[I_04]], 5
+; CHECK-NEXT: [[REM:%.*]] = urem i32 [[I_WITH_OFF]], [[REM_AMT]]
+; CHECK-NEXT: tail call void @use.i32(i32 [[REM]])
+; CHECK-NEXT: [[INC]] = add nuw i32 [[I_04]], 1
+; CHECK-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i32 [[INC]], [[N]]
+; CHECK-NEXT: br i1 [[EXITCOND_NOT]], label %[[FOR_COND_CLEANUP]], label %[[FOR_BODY]]
+;
+entry:
+ %cmp3.not = icmp ult i32 %N, 3
+ br i1 %cmp3.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup:
+ ret void
+
+for.body:
+ %i.04 = phi i32 [ %inc, %for.body ], [ 2, %entry ]
+ %i_with_off = add nuw i32 %i.04, 5
+ %rem = urem i32 %i_with_off, %rem_amt
+ tail call void @use.i32(i32 %rem)
+ %inc = add nuw i32 %i.04, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+define void @simple_urem_to_sel_non_zero_start_through_sub(i32 %N, i32 %rem_amt, i32 %start) nounwind {
+; CHECK-LABEL: define void @simple_urem_to_sel_non_zero_start_through_sub(
+; CHECK-SAME: i32 [[N:%.*]], i32 [[REM_AMT:%.*]], i32 [[START:%.*]]) #[[ATTR0]] {
+; CHECK-NEXT: [[ENTRY:.*:]]
+; CHECK-NEXT: [[CMP3_NOT:%.*]] = icmp ule i32 [[N]], [[START]]
+; CHECK-NEXT: br i1 [[CMP3_NOT]], label %[[FOR_COND_CLEANUP:.*]], label %[[FOR_BODY_PREHEADER:.*]]
+; CHECK: [[FOR_BODY_PREHEADER]]:
+; CHECK-NEXT: br label %[[FOR_BODY:.*]]
+; CHECK: [[FOR_COND_CLEANUP]]:
+; CHECK-NEXT: ret void
+; CHECK: [[FOR_BODY]]:
+; CHECK-NEXT: [[I_04:%.*]] = phi i32 [ [[INC:%.*]], %[[FOR_BODY]] ], [ [[START]], %[[FOR_BODY_PREHEADER]] ]
+; CHECK-NEXT: [[I_WITH_OFF:%.*]] = sub nuw i32 [[I_04]], [[START]]
+; CHECK-NEXT: [[REM:%.*]] = urem i32 [[I_WITH_OFF]], [[REM_AMT]]
+; CHECK-NEXT: tail call void @use.i32(i32 [[REM]])
+; CHECK-NEXT: [[INC]] = add nuw i32 [[I_04]], 1
+; CHECK-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i32 [[INC]], [[N]]
+; CHECK-NEXT: br i1 [[EXITCOND_NOT]], label %[[FOR_COND_CLEANUP]], label %[[FOR_BODY]]
+;
+entry:
+ %cmp3.not = icmp ule i32 %N, %start
+ br i1 %cmp3.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup:
+ ret void
+
+for.body:
+ %i.04 = phi i32 [ %inc, %for.body ], [ %start, %entry ]
+ %i_with_off = sub nuw i32 %i.04, %start
+ %rem = urem i32 %i_with_off, %rem_amt
+ tail call void @use.i32(i32 %rem)
+ %inc = add nuw i32 %i.04, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+define void @simple_urem_to_sel_non_zero_start_through_sub_no_simplfy(i32 %N, i32 %rem_amt, i32 %start) nounwind {
+; CHECK-LABEL: define void @simple_urem_to_sel_non_zero_start_through_sub_no_simplfy(
+; CHECK-SAME: i32 [[N:%.*]], i32 [[REM_AMT:%.*]], i32 [[START:%.*]]) #[[ATTR0]] {
+; CHECK-NEXT: [[ENTRY:.*:]]
+; CHECK-NEXT: [[CMP3_NOT:%.*]] = icmp ule i32 [[N]], [[START]]
+; CHECK-NEXT: br i1 [[CMP3_NOT]], label %[[FOR_COND_CLEANUP:.*]], label %[[FOR_BODY_PREHEADER:.*]]
+; CHECK: [[FOR_BODY_PREHEADER]]:
+; CHECK-NEXT: br label %[[FOR_BODY:.*]]
+; CHECK: [[FOR_COND_CLEANUP]]:
+; CHECK-NEXT: ret void
+; CHECK: [[FOR_BODY]]:
+; CHECK-NEXT: [[I_04:%.*]] = phi i32 [ [[INC:%.*]], %[[FOR_BODY]] ], [ [[START]], %[[FOR_BODY_PREHEADER]] ]
+; CHECK-NEXT: [[I_WITH_OFF:%.*]] = sub nuw i32 [[I_04]], 2
+; CHECK-NEXT: [[REM:%.*]] = urem i32 [[I_WITH_OFF]], [[REM_AMT]]
+; CHECK-NEXT: tail call void @use.i32(i32 [[REM]])
+; CHECK-NEXT: [[INC]] = add nuw i32 [[I_04]], 1
+; CHECK-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i32 [[INC]], [[N]]
+; CHECK-NEXT: br i1 [[EXITCOND_NOT]], label %[[FOR_COND_CLEANUP]], label %[[FOR_BODY]]
+;
+entry:
+ %cmp3.not = icmp ule i32 %N, %start
+ br i1 %cmp3.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup:
+ ret void
+
+for.body:
+ %i.04 = phi i32 [ %inc, %for.body ], [ %start, %entry ]
+ %i_with_off = sub nuw i32 %i.04, 2
+ %rem = urem i32 %i_with_off, %rem_amt
+ tail call void @use.i32(i32 %rem)
+ %inc = add nuw i32 %i.04, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
More information about the llvm-commits
mailing list