vendor/llvm/llvm-trunk-r321017

7 files changed, 593 insertions, 1 deletions

diff --git a/test/Transforms/CodeExtractor/PartialInlineDebug.ll b/test/Transforms/CodeExtractor/PartialInlineDebug.ll
new file mode 100644
index 000000000000..f03b38345611
--- /dev/null
+++ b/test/Transforms/CodeExtractor/PartialInlineDebug.ll
@@ -0,0 +1,104 @@
+; RUN: opt < %s -S -partial-inliner -skip-partial-inlining-cost-analysis=true | FileCheck %s
+
+; CHECK-LABEL: @callee
+; CHECK: %mul = mul nsw i32 %v, 10, !dbg ![[DBG1:[0-9]+]]
+define i32 @callee(i32 %v) !dbg !16 {
+entry:
+  %cmp = icmp sgt i32 %v, 2000
+  br i1 %cmp, label %if.then, label %if.end
+
+if.then:                                          ; preds = %entry
+  %mul = mul nsw i32 %v, 10, !dbg !17
+  br label %if.then2
+
+if.then2:
+  %sub = sub i32 %v, 10, !dbg !23
+  br label %if.end
+
+if.end:                                           ; preds = %if.then, %entry
+  %v2 = phi i32 [ %v, %entry ], [ %mul, %if.then2 ]
+  %add = add nsw i32 %v2, 200
+  ret i32 %add
+}
+
+; CHECK-LABEL: @caller
+; CHECK: codeRepl.i:
+; CHECK-NEXT: call void @callee.2_if.then(i32 %v, i32* %mul.loc.i), !dbg ![[DBG2:[0-9]+]]
+define i32 @caller(i32 %v) !dbg !8 {
+entry:
+  %call = call i32 @callee(i32 %v), !dbg !14
+  ret i32 %call
+}
+
+
+; CHECK-LABEL: @callee2
+; CHECK: %sub = sub i32 %v, 10, !dbg ![[DBG3:[0-9]+]]
+define i32 @callee2(i32 %v) !dbg !18 {
+entry:
+  %cmp = icmp sgt i32 %v, 2000
+  br i1 %cmp, label %if.then, label %if.end
+
+if.then:
+  br label %if.then2
+
+if.then2:
+  %sub = sub i32 %v, 10, !dbg !20
+  br label %if.end
+
+if.end:
+  %v2 = phi i32 [ %v, %entry ], [ %sub, %if.then2 ]
+  %add = add nsw i32 %v2, 200
+  ret i32 %add
+}
+
+; CHECK-LABEL: @caller2
+; CHECK: codeRepl.i:
+; CHECK-NEXT: call void @callee2.1_if.then(i32 %v, i32* %sub.loc.i), !dbg ![[DBG4:[0-9]+]]
+define i32 @caller2(i32 %v) !dbg !21 {
+entry:
+  %call = call i32 @callee2(i32 %v), !dbg !22
+  ret i32 %call
+}
+
+; CHECK-LABEL: define internal void @callee2.1_if.then
+; CHECK: br label %if.then, !dbg ![[DBG5:[0-9]+]]
+
+; CHECK-LABEL: define internal void @callee.2_if.then
+; CHECK: br label %if.then, !dbg ![[DBG6:[0-9]+]]
+
+; CHECK: ![[DBG1]] = !DILocation(line: 10, column: 7,
+; CHECK: ![[DBG2]] = !DILocation(line: 10, column: 7,
+; CHECK: ![[DBG3]] = !DILocation(line: 110, column: 17,
+; CHECK: ![[DBG4]] = !DILocation(line: 110, column: 17,
+; CHECK: ![[DBG5]] = !DILocation(line: 110, column: 17,
+; CHECK: ![[DBG6]] = !DILocation(line: 10, column: 7,
+
+
+!llvm.dbg.cu = !{!0}
+!llvm.module.flags = !{!3, !4, !5, !6}
+!llvm.ident = !{!7}
+
+!0 = distinct !DICompileUnit(language: DW_LANG_C99, file: !1, producer: "clang version 6.0.0 (trunk 177881)", isOptimized: true, runtimeVersion: 0, emissionKind: FullDebug, enums: !2)
+!1 = !DIFile(filename: "test.c", directory: "/tmp")
+!2 = !{}
+!3 = !{i32 2, !"Dwarf Version", i32 4}
+!4 = !{i32 2, !"Debug Info Version", i32 3}
+!5 = !{i32 1, !"wchar_size", i32 4}
+!6 = !{i32 1, !"min_enum_size", i32 4}
+!7 = !{!"clang version 6.0.0"}
+!8 = distinct !DISubprogram(name: "caller", scope: !1, file: !1, line: 3, type: !9, isLocal: false, isDefinition: true, scopeLine: 3, flags: DIFlagPrototyped, isOptimized: true, unit: !0, variables: !12)
+!9 = !DISubroutineType(types: !10)
+!10 = !{!11, !11}
+!11 = !DIBasicType(name: "int", size: 19, encoding: DW_ATE_signed)
+!12 = !{!13}
+!13 = !DILocalVariable(name: "v", arg: 1, scope: !8, file: !1, line: 3, type: !11)
+!14 = !DILocation(line: 5, column: 10, scope: !8)
+!15 = distinct !DILexicalBlock(scope: !16, file: !1, line: 9, column: 7)
+!16 = distinct !DISubprogram(name: "callee", scope: !1, file: !1, line: 8, type: !9, isLocal: false, isDefinition: true, scopeLine: 8, flags: DIFlagPrototyped, isOptimized: true, unit: !0, variables: !12)
+!17 = !DILocation(line: 10, column: 7, scope: !15)
+!18 = distinct !DISubprogram(name: "callee2", scope: !1, file: !1, line: 8, type: !9, isLocal: false, isDefinition: true, scopeLine: 8, flags: DIFlagPrototyped, isOptimized: true, unit: !0, variables: !12)
+!19 = distinct !DILexicalBlock(scope: !18, file: !1, line: 100, column: 1)
+!20 = !DILocation(line: 110, column: 17, scope: !19)
+!21 = distinct !DISubprogram(name: "caller2", scope: !1, file: !1, line: 8, type: !9, isLocal: false, isDefinition: true, scopeLine: 8, flags: DIFlagPrototyped, isOptimized: true, unit: !0, variables: !12)
+!22 = !DILocation(line: 110, column: 17, scope: !21)
+!23 = !DILocation(line: 15, column: 7, scope: !15)
diff --git a/test/Transforms/CodeExtractor/PartialInlineNoInline.ll b/test/Transforms/CodeExtractor/PartialInlineNoInline.ll
new file mode 100644
index 000000000000..6c0b83298d23
--- /dev/null
+++ b/test/Transforms/CodeExtractor/PartialInlineNoInline.ll
@@ -0,0 +1,45 @@
+; RUN: opt < %s -partial-inliner -S -stats -pass-remarks=partial-inlining 2>&1 | FileCheck %s
+; RUN: opt < %s -passes=partial-inliner -S -stats -pass-remarks=partial-inlining 2>&1 | FileCheck %s
+
+@stat = external global i32, align 4
+
+define i32 @inline_fail(i32 %count, ...) {
+entry:
+  %vargs = alloca i8*, align 8
+  %vargs1 = bitcast i8** %vargs to i8*
+  call void @llvm.va_start(i8* %vargs1)
+  %stat1 = load i32, i32* @stat, align 4
+  %cmp = icmp slt i32 %stat1, 0
+  br i1 %cmp, label %bb2, label %bb1
+
+bb1:                                              ; preds = %entry
+  %vg1 = add nsw i32 %stat1, 1
+  store i32 %vg1, i32* @stat, align 4
+  %va1 = va_arg i8** %vargs, i32
+  call void @foo(i32 %count, i32 %va1) #2
+  br label %bb2
+
+bb2:                                              ; preds = %bb1, %entry
+  %res = phi i32 [ 1, %bb1 ], [ 0, %entry ]
+  call void @llvm.va_end(i8* %vargs1)
+  ret i32 %res
+}
+
+define i32 @caller(i32 %arg) {
+bb:
+  %res = tail call i32 (i32, ...) @inline_fail(i32 %arg, i32 %arg)
+  ret i32 %res
+}
+
+declare void @foo(i32, i32)
+declare void @llvm.va_start(i8*)
+declare void @llvm.va_end(i8*)
+
+; Check that no remarks have been emitted, inline_fail has not been partial
+; inlined, no code has been extracted and the partial-inlining counter
+; has not been incremented.
+
+; CHECK-NOT: remark
+; CHECK: tail call i32 (i32, ...) @inline_fail(i32 %arg, i32 %arg)
+; CHECK-NOT: inline_fail.1_bb1
+; CHECK-NOT: partial-inlining
diff --git a/test/Transforms/CodeExtractor/PartialInlinePGOMultiRegion.ll b/test/Transforms/CodeExtractor/PartialInlinePGOMultiRegion.ll
new file mode 100644
index 000000000000..a51bdd01df5a
--- /dev/null
+++ b/test/Transforms/CodeExtractor/PartialInlinePGOMultiRegion.ll
@@ -0,0 +1,169 @@
+; RUN: opt -S -partial-inliner -min-block-execution=1 -skip-partial-inlining-cost-analysis < %s | FileCheck %s
+; RUN: opt -S -passes=partial-inliner -min-block-execution=1 -skip-partial-inlining-cost-analysis < %s | FileCheck %s
+; Require a dummy block (if.then.b) as successor to if.then due to PI requirement
+; of region containing more than one BB.
+define signext i32 @bar(i32 signext %value, i32 signext %ub) #0 !prof !30 {
+entry:
+  %value.addr = alloca i32, align 4
+  %ub.addr = alloca i32, align 4
+  %sum = alloca i32, align 4
+  %i = alloca i32, align 4
+  store i32 %value, i32* %value.addr, align 4
+  store i32 %ub, i32* %ub.addr, align 4
+  store i32 0, i32* %sum, align 4
+  store i32 0, i32* %i, align 4
+  br label %for.cond
+
+for.cond:                                         ; preds = %for.inc, %entry
+  %0 = load i32, i32* %i, align 4
+  %1 = load i32, i32* %ub.addr, align 4
+  %cmp = icmp slt i32 %0, %1
+  br i1 %cmp, label %for.body, label %for.cond2, !prof !31
+
+for.body:                                         ; preds = %for.cond
+  %2 = load i32, i32* %value.addr, align 4
+  %rem = srem i32 %2, 20
+  %cmp1 = icmp eq i32 %rem, 0
+  br i1 %cmp1, label %if.then, label %if.else, !prof !32
+
+if.then:                                          ; preds = %for.body
+  %3 = load i32, i32* %value.addr, align 4
+  %4 = load i32, i32* %i, align 4
+  %mul = mul nsw i32 %4, 5
+  %add = add nsw i32 %3, %mul
+  %5 = load i32, i32* %sum, align 4
+  %add2 = add nsw i32 %5, %add
+  store i32 %add2, i32* %sum, align 4
+  br label %if.then.b
+
+if.then.b:                                        ; preds = %if.then
+  br label %if.end
+
+if.else:                                          ; preds = %for.body
+  %6 = load i32, i32* %value.addr, align 4
+  %7 = load i32, i32* %i, align 4
+  %sub = sub nsw i32 %6, %7
+  %8 = load i32, i32* %sum, align 4
+  %add3 = add nsw i32 %8, %sub
+  store i32 %add3, i32* %sum, align 4
+  br label %if.end
+
+if.end:                                           ; preds = %if.else, %if.then
+  br label %for.inc
+
+for.inc:                                          ; preds = %if.end
+  %9 = load i32, i32* %i, align 4
+  %inc = add nsw i32 %9, 1
+  store i32 %inc, i32* %i, align 4
+  br label %for.cond
+
+for.cond2:                                         ; preds = %for.cond
+  %10 = load i32, i32* %i, align 4
+  %11 = load i32, i32* %ub.addr, align 4
+  %cmp2 = icmp slt i32 %10, %11
+  br i1 %cmp2, label %for.body2, label %for.end, !prof !31
+
+for.body2:                                         ; preds = %for.cond2
+  %12 = load i32, i32* %value.addr, align 4
+  %rem2 = srem i32 %12, 20
+  %cmp3 = icmp eq i32 %rem2, 0
+  br i1 %cmp3, label %if.then2, label %if.else2, !prof !32
+
+if.then2:                                          ; preds = %for.body2
+  %13 = load i32, i32* %value.addr, align 4
+  %14 = load i32, i32* %i, align 4
+  %mul2 = mul nsw i32 %14, 5
+  %add4 = add nsw i32 %13, %mul2
+  %15 = load i32, i32* %sum, align 4
+  %add5 = add nsw i32 %15, %add4
+  store i32 %add5, i32* %sum, align 4
+  br label %if.then2.b
+
+if.then2.b:                                        ; preds = %if.then2
+  br label %if.end2
+
+if.else2:                                          ; preds = %for.body2
+  %16 = load i32, i32* %value.addr, align 4
+  %17 = load i32, i32* %i, align 4
+  %sub2 = sub nsw i32 %16, %17
+  %18 = load i32, i32* %sum, align 4
+  %add6 = add nsw i32 %18, %sub2
+  store i32 %add6, i32* %sum, align 4
+  br label %if.end2
+
+if.end2:                                           ; preds = %if.else2, %if.then2
+  br label %for.inc2
+
+for.inc2:                                          ; preds = %if.end2
+  %19 = load i32, i32* %i, align 4
+  %inc2 = add nsw i32 %19, 1
+  store i32 %inc2, i32* %i, align 4
+  br label %for.cond2
+
+for.end:                                          ; preds = %for.cond2
+  %20 = load i32, i32* %sum, align 4
+  ret i32 %20
+}
+
+define signext i32 @foo(i32 signext %value, i32 signext %ub) #0 !prof !30 {
+; CHECK-LABEL: @foo
+; CHECK-NOT: call signext i32 @bar
+; CHECK: codeRepl1.i:
+; CHECK: call void @bar.1_if.then
+; CHECK: codeRepl.i:
+; CHECK: call void @bar.1_if.then2
+entry:
+  %value.addr = alloca i32, align 4
+  %ub.addr = alloca i32, align 4
+  store i32 %value, i32* %value.addr, align 4
+  store i32 %ub, i32* %ub.addr, align 4
+  %0 = load i32, i32* %value.addr, align 4
+  %1 = load i32, i32* %ub.addr, align 4
+  %call = call signext i32 @bar(i32 signext %0, i32 signext %1)
+  ret i32 %call
+}
+
+; CHECK-LABEL: define internal void @bar.1_if.then2
+; CHECK: .exitStub:
+; CHECK: ret void
+
+; CHECK-LABEL: define internal void @bar.1_if.then
+; CHECK: .exitStub:
+; CHECK: ret void
+
+!llvm.module.flags = !{!0, !1, !2}
+!llvm.ident = !{!29}
+
+!0 = !{i32 1, !"wchar_size", i32 4}
+!1 = !{i32 7, !"PIC Level", i32 2}
+!2 = !{i32 1, !"ProfileSummary", !3}
+!3 = !{!4, !5, !6, !7, !8, !9, !10, !11}
+!4 = !{!"ProfileFormat", !"InstrProf"}
+!5 = !{!"TotalCount", i64 103}
+!6 = !{!"MaxCount", i64 100}
+!7 = !{!"MaxInternalCount", i64 1}
+!8 = !{!"MaxFunctionCount", i64 100}
+!9 = !{!"NumCounts", i64 5}
+!10 = !{!"NumFunctions", i64 3}
+!11 = !{!"DetailedSummary", !12}
+!12 = !{!13, !14, !15, !16, !17, !18, !18, !19, !19, !20, !21, !22, !23, !24, !25, !26, !27, !28}
+!13 = !{i32 10000, i64 100, i32 1}
+!14 = !{i32 100000, i64 100, i32 1}
+!15 = !{i32 200000, i64 100, i32 1}
+!16 = !{i32 300000, i64 100, i32 1}
+!17 = !{i32 400000, i64 100, i32 1}
+!18 = !{i32 500000, i64 100, i32 1}
+!19 = !{i32 600000, i64 100, i32 1}
+!20 = !{i32 700000, i64 100, i32 1}
+!21 = !{i32 800000, i64 100, i32 1}
+!22 = !{i32 900000, i64 100, i32 1}
+!23 = !{i32 950000, i64 100, i32 1}
+!24 = !{i32 990000, i64 1, i32 4}
+!25 = !{i32 999000, i64 1, i32 4}
+!26 = !{i32 999900, i64 1, i32 4}
+!27 = !{i32 999990, i64 1, i32 4}
+!28 = !{i32 999999, i64 1, i32 4}
+!29 = !{!"clang version 6.0.0 (123456)"}
+!30 = !{!"function_entry_count", i64 2}
+!31 = !{!"branch_weights", i32 100, i32 1}
+!32 = !{!"branch_weights", i32 0, i32 100}
diff --git a/test/Transforms/CodeExtractor/PartialInlinePGORegion.ll b/test/Transforms/CodeExtractor/PartialInlinePGORegion.ll
new file mode 100644
index 000000000000..27c858f3de60
--- /dev/null
+++ b/test/Transforms/CodeExtractor/PartialInlinePGORegion.ll
@@ -0,0 +1,120 @@
+; RUN: opt -S -partial-inliner -min-block-execution=1 -skip-partial-inlining-cost-analysis < %s | FileCheck %s
+; RUN: opt -S -passes=partial-inliner -min-block-execution=1 -skip-partial-inlining-cost-analysis < %s | FileCheck %s
+; Require a dummy block (if.then.b) as successor to if.then due to PI requirement
+; of region containing more than one BB.
+define signext i32 @bar(i32 signext %value, i32 signext %ub) #0 !prof !30 {
+entry:
+  %value.addr = alloca i32, align 4
+  %ub.addr = alloca i32, align 4
+  %sum = alloca i32, align 4
+  %i = alloca i32, align 4
+  store i32 %value, i32* %value.addr, align 4
+  store i32 %ub, i32* %ub.addr, align 4
+  store i32 0, i32* %sum, align 4
+  store i32 0, i32* %i, align 4
+  br label %for.cond
+
+for.cond:                                         ; preds = %for.inc, %entry
+  %0 = load i32, i32* %i, align 4
+  %1 = load i32, i32* %ub.addr, align 4
+  %cmp = icmp slt i32 %0, %1
+  br i1 %cmp, label %for.body, label %for.end, !prof !31
+
+for.body:                                         ; preds = %for.cond
+  %2 = load i32, i32* %value.addr, align 4
+  %rem = srem i32 %2, 20
+  %cmp1 = icmp eq i32 %rem, 0
+  br i1 %cmp1, label %if.then, label %if.else, !prof !32
+
+if.then:                                          ; preds = %for.body
+  %3 = load i32, i32* %value.addr, align 4
+  %4 = load i32, i32* %i, align 4
+  %mul = mul nsw i32 %4, 5
+  %add = add nsw i32 %3, %mul
+  %5 = load i32, i32* %sum, align 4
+  %add2 = add nsw i32 %5, %add
+  store i32 %add2, i32* %sum, align 4
+  br label %if.then.b
+
+if.then.b:                                        ; preds = %if.then
+  br label %if.end
+
+if.else:                                          ; preds = %for.body
+  %6 = load i32, i32* %value.addr, align 4
+  %7 = load i32, i32* %i, align 4
+  %sub = sub nsw i32 %6, %7
+  %8 = load i32, i32* %sum, align 4
+  %add3 = add nsw i32 %8, %sub
+  store i32 %add3, i32* %sum, align 4
+  br label %if.end
+
+if.end:                                           ; preds = %if.else, %if.then
+  br label %for.inc
+
+for.inc:                                          ; preds = %if.end
+  %9 = load i32, i32* %i, align 4
+  %inc = add nsw i32 %9, 1
+  store i32 %inc, i32* %i, align 4
+  br label %for.cond
+
+for.end:                                          ; preds = %for.cond
+  %10 = load i32, i32* %sum, align 4
+  ret i32 %10
+}
+
+define signext i32 @foo(i32 signext %value, i32 signext %ub) #0 !prof !30 {
+; CHECK-LABEL: @foo
+; CHECK: codeRepl.i:
+; CHECK-NOT: call signext i32 @bar
+; CHECK: call void @bar.1_if.then
+entry:
+  %value.addr = alloca i32, align 4
+  %ub.addr = alloca i32, align 4
+  store i32 %value, i32* %value.addr, align 4
+  store i32 %ub, i32* %ub.addr, align 4
+  %0 = load i32, i32* %value.addr, align 4
+  %1 = load i32, i32* %ub.addr, align 4
+  %call = call signext i32 @bar(i32 signext %0, i32 signext %1)
+  ret i32 %call
+}
+
+; CHECK-LABEL: define internal void @bar.1_if.then
+; CHECK: .exitStub:
+; CHECK: ret void
+
+!llvm.module.flags = !{!0, !1, !2}
+!llvm.ident = !{!29}
+
+!0 = !{i32 1, !"wchar_size", i32 4}
+!1 = !{i32 7, !"PIC Level", i32 2}
+!2 = !{i32 1, !"ProfileSummary", !3}
+!3 = !{!4, !5, !6, !7, !8, !9, !10, !11}
+!4 = !{!"ProfileFormat", !"InstrProf"}
+!5 = !{!"TotalCount", i64 103}
+!6 = !{!"MaxCount", i64 100}
+!7 = !{!"MaxInternalCount", i64 1}
+!8 = !{!"MaxFunctionCount", i64 100}
+!9 = !{!"NumCounts", i64 5}
+!10 = !{!"NumFunctions", i64 3}
+!11 = !{!"DetailedSummary", !12}
+!12 = !{!13, !14, !15, !16, !17, !18, !18, !19, !19, !20, !21, !22, !23, !24, !25, !26, !27, !28}
+!13 = !{i32 10000, i64 100, i32 1}
+!14 = !{i32 100000, i64 100, i32 1}
+!15 = !{i32 200000, i64 100, i32 1}
+!16 = !{i32 300000, i64 100, i32 1}
+!17 = !{i32 400000, i64 100, i32 1}
+!18 = !{i32 500000, i64 100, i32 1}
+!19 = !{i32 600000, i64 100, i32 1}
+!20 = !{i32 700000, i64 100, i32 1}
+!21 = !{i32 800000, i64 100, i32 1}
+!22 = !{i32 900000, i64 100, i32 1}
+!23 = !{i32 950000, i64 100, i32 1}
+!24 = !{i32 990000, i64 1, i32 4}
+!25 = !{i32 999000, i64 1, i32 4}
+!26 = !{i32 999900, i64 1, i32 4}
+!27 = !{i32 999990, i64 1, i32 4}
+!28 = !{i32 999999, i64 1, i32 4}
+!29 = !{!"clang version 6.0.0 (123456)"}
+!30 = !{!"function_entry_count", i64 2}
+!31 = !{!"branch_weights", i32 100, i32 1}
+!32 = !{!"branch_weights", i32 0, i32 100}
diff --git a/test/Transforms/CodeExtractor/PartialInlineVarArg.ll b/test/Transforms/CodeExtractor/PartialInlineVarArg.ll
new file mode 100644
index 000000000000..415eb9ba1754
--- /dev/null
+++ b/test/Transforms/CodeExtractor/PartialInlineVarArg.ll
@@ -0,0 +1,83 @@
+; RUN: opt < %s -partial-inliner -S -skip-partial-inlining-cost-analysis | FileCheck %s
+; RUN: opt < %s -passes=partial-inliner -S -skip-partial-inlining-cost-analysis | FileCheck %s
+
+@stat = external global i32, align 4
+
+define i32 @vararg(i32 %count, ...) {
+entry:
+  %vargs = alloca i8*, align 8
+  %stat1 = load i32, i32* @stat, align 4
+  %cmp = icmp slt i32 %stat1, 0
+  br i1 %cmp, label %bb2, label %bb1
+
+bb1:                                              ; preds = %entry
+  %vg1 = add nsw i32 %stat1, 1
+  store i32 %vg1, i32* @stat, align 4
+  %vargs1 = bitcast i8** %vargs to i8*
+  call void @llvm.va_start(i8* %vargs1)
+  %va1 = va_arg i8** %vargs, i32
+  call void @foo(i32 %count, i32 %va1) #2
+  call void @llvm.va_end(i8* %vargs1)
+  br label %bb2
+
+bb2:                                              ; preds = %bb1, %entry
+  %res = phi i32 [ 1, %bb1 ], [ 0, %entry ]
+  ret i32 %res
+}
+
+declare void @foo(i32, i32)
+declare void @llvm.va_start(i8*)
+declare void @llvm.va_end(i8*)
+
+define i32 @caller1(i32 %arg) {
+bb:
+  %tmp = tail call i32 (i32, ...) @vararg(i32 %arg)
+  ret i32 %tmp
+}
+; CHECK-LABEL: @caller1
+; CHECK: codeRepl.i:
+; CHECK-NEXT:  call void (i32, i8**, i32, ...) @vararg.2_bb1(i32 %stat1.i, i8** %vargs.i, i32 %arg)
+
+define i32 @caller2(i32 %arg, float %arg2) {
+bb:
+  %tmp = tail call i32 (i32, ...) @vararg(i32 %arg, i32 10, float %arg2)
+  ret i32 %tmp
+}
+
+; CHECK-LABEL: @caller2
+; CHECK: codeRepl.i:
+; CHECK-NEXT:  call void (i32, i8**, i32, ...) @vararg.2_bb1(i32 %stat1.i, i8** %vargs.i, i32 %arg, i32 10, float %arg2)
+
+; Test case to check that we do not extract a vararg function, if va_end is in
+; a block that is not outlined.
+define i32 @vararg_not_legal(i32 %count, ...) {
+entry:
+  %vargs = alloca i8*, align 8
+  %vargs0 = bitcast i8** %vargs to i8*
+  %stat1 = load i32, i32* @stat, align 4
+  %cmp = icmp slt i32 %stat1, 0
+  br i1 %cmp, label %bb2, label %bb1
+
+bb1:                                              ; preds = %entry
+  %vg1 = add nsw i32 %stat1, 1
+  store i32 %vg1, i32* @stat, align 4
+  %vargs1 = bitcast i8** %vargs to i8*
+  call void @llvm.va_start(i8* %vargs1)
+  %va1 = va_arg i8** %vargs, i32
+  call void @foo(i32 %count, i32 %va1)
+  br label %bb2
+
+bb2:                                              ; preds = %bb1, %entry
+  %res = phi i32 [ 1, %bb1 ], [ 0, %entry ]
+  %ptr = phi i8* [ %vargs1, %bb1 ], [ %vargs0, %entry]
+  call void @llvm.va_end(i8* %ptr)
+  ret i32 %res
+}
+
+; CHECK-LABEL: @caller3
+; CHECK: tail call i32 (i32, ...) @vararg_not_legal(i32 %arg, i32 %arg)
+define i32 @caller3(i32 %arg) {
+bb:
+  %res = tail call i32 (i32, ...) @vararg_not_legal(i32 %arg, i32 %arg)
+  ret i32 %res
+}
diff --git a/test/Transforms/CodeExtractor/PartialInlineVarArgsDebug.ll b/test/Transforms/CodeExtractor/PartialInlineVarArgsDebug.ll
new file mode 100644
index 000000000000..dc3e77203e0c
--- /dev/null
+++ b/test/Transforms/CodeExtractor/PartialInlineVarArgsDebug.ll
@@ -0,0 +1,67 @@
+; RUN: opt < %s -S -partial-inliner -skip-partial-inlining-cost-analysis=true | FileCheck %s
+
+; CHECK-LABEL: @callee
+; CHECK: %mul = mul nsw i32 %v, 10, !dbg ![[DBG1:[0-9]+]]
+define i32 @callee(i32 %v, ...) !dbg !16 {
+entry:
+  %cmp = icmp sgt i32 %v, 2000, !dbg !17
+  br i1 %cmp, label %if.then, label %if.end, !dbg !19
+
+if.then:                                          ; preds = %entry
+  %mul = mul nsw i32 %v, 10, !dbg !20
+  br label %if.end, !dbg !21
+
+if.end:                                           ; preds = %if.then, %entry
+  %v2 = phi i32 [ %v, %entry ], [ %mul, %if.then ]
+  %add = add nsw i32 %v2, 200, !dbg !22
+  ret i32 %add, !dbg !23
+}
+
+; CHECK-LABEL: @caller
+; CHECK: codeRepl.i:
+; CHECK-NEXT: call void (i32, i32*, ...) @callee.1_if.then(i32 %v, i32* %mul.loc.i, i32 99), !dbg ![[DBG2:[0-9]+]]
+define i32 @caller(i32 %v) !dbg !8 {
+entry:
+  %call = call i32 (i32, ...) @callee(i32 %v, i32 99), !dbg !14
+  ret i32 %call, !dbg !15
+}
+
+; CHECK-LABEL: define internal void @callee.1_if.then
+; CHECK: br label %if.then, !dbg ![[DBG3:[0-9]+]]
+
+; CHECK: ![[DBG1]] = !DILocation(line: 10, column: 7,
+; CHECK: ![[DBG2]] = !DILocation(line: 10, column: 7,
+; CHECK: ![[DBG3]] = !DILocation(line: 10, column: 7,
+
+
+!llvm.dbg.cu = !{!0}
+!llvm.module.flags = !{!3, !4, !5, !6}
+!llvm.ident = !{!7}
+
+!0 = distinct !DICompileUnit(language: DW_LANG_C99, file: !1, producer: "clang version 6.0.0 (trunk 177881)", isOptimized: true, runtimeVersion: 0, emissionKind: FullDebug, enums: !2)
+!1 = !DIFile(filename: "test.c", directory: "/tmp")
+!2 = !{}
+!3 = !{i32 2, !"Dwarf Version", i32 4}
+!4 = !{i32 2, !"Debug Info Version", i32 3}
+!5 = !{i32 1, !"wchar_size", i32 4}
+!6 = !{i32 1, !"min_enum_size", i32 4}
+!7 = !{!"clang version 6.0.0"}
+!8 = distinct !DISubprogram(name: "caller", scope: !1, file: !1, line: 3, type: !9, isLocal: false, isDefinition: true, scopeLine: 3, flags: DIFlagPrototyped, isOptimized: true, unit: !0, variables: !12)
+!9 = !DISubroutineType(types: !10)
+!10 = !{!11, !11}
+!11 = !DIBasicType(name: "int", size: 19, encoding: DW_ATE_signed)
+!12 = !{!13}
+!13 = !DILocalVariable(name: "v", arg: 1, scope: !8, file: !1, line: 3, type: !11)
+!14 = !DILocation(line: 5, column: 10, scope: !8)
+!15 = !DILocation(line: 5, column: 3, scope: !8)
+!16 = distinct !DISubprogram(name: "callee", scope: !1, file: !1, line: 8, type: !9, isLocal: false, isDefinition: true, scopeLine: 8, flags: DIFlagPrototyped, isOptimized: true, unit: !0, variables: !26)
+!26 = !{!27}
+!27 = !DILocalVariable(name: "v", arg: 1, scope: !16, file: !1, line: 8, type: !11)
+!17 = !DILocation(line: 9, column: 9, scope: !18)
+!18 = distinct !DILexicalBlock(scope: !16, file: !1, line: 9, column: 7)
+!19 = !DILocation(line: 9, column: 7, scope: !16)
+!20 = !DILocation(line: 10, column: 7, scope: !18)
+!21 = !DILocation(line: 10, column: 5, scope: !18)
+!22 = !DILocation(line: 11, column: 5, scope: !16)
+!36 = !DILocation(line: 12, column: 10, scope: !16)
+!23 = !DILocation(line: 12, column: 3, scope: !16)
diff --git a/test/Transforms/CodeExtractor/live_shrink_hoist.ll b/test/Transforms/CodeExtractor/live_shrink_hoist.ll
index d1b310f01769..1f57146c9418 100644
--- a/test/Transforms/CodeExtractor/live_shrink_hoist.ll
+++ b/test/Transforms/CodeExtractor/live_shrink_hoist.ll
@@ -1,4 +1,4 @@
-; RUN: opt -S -partial-inliner -max-num-inline-blocks=2 -skip-partial-inlining-cost-analysis  < %s |   FileCheck %s
+; RUN: opt -S -partial-inliner -max-num-inline-blocks=3 -skip-partial-inlining-cost-analysis  < %s |   FileCheck %s
 ; RUN: opt -S -passes=partial-inliner -max-num-inline-blocks=2  -skip-partial-inlining-cost-analysis < %s   | FileCheck %s
 
 %class.A = type { i32 }
@@ -16,6 +16,10 @@ bb:
   br i1 %tmp3, label %bb4, label %bb9
 
 bb4:                                              ; preds = %bb
+  %foo = icmp eq i32 %tmp2, 0
+  br i1 %foo, label %bb5, label %bb9
+
+bb5:                                              ; preds = %bb4
   call void @_ZN1A7memfuncEv(%class.A* nonnull %tmp)
   %tmp5 = getelementptr inbounds %class.A, %class.A* %tmp, i64 0, i32 0
   %tmp6 = load i32, i32* %tmp5, align 4, !tbaa !6