diff --git a/lib/CodeGen/AsmPrinter/DebugHandlerBase.cpp b/lib/CodeGen/AsmPrinter/DebugHandlerBase.cpp
index d30f106a939a..5607242a7f12 100644
--- a/lib/CodeGen/AsmPrinter/DebugHandlerBase.cpp
+++ b/lib/CodeGen/AsmPrinter/DebugHandlerBase.cpp
@@ -201,8 +201,10 @@ void DebugHandlerBase::endInstruction() {
   assert(CurMI != nullptr);
   // Don't create a new label after DBG_VALUE instructions.
   // They don't generate code.
-  if (!CurMI->isDebugValue())
+  if (!CurMI->isDebugValue()) {
     PrevLabel = nullptr;
+    PrevInstBB = CurMI->getParent();
+  }
 
   DenseMap<const MachineInstr *, MCSymbol *>::iterator I =
       LabelsAfterInsn.find(CurMI);
diff --git a/lib/CodeGen/AsmPrinter/DebugHandlerBase.h b/lib/CodeGen/AsmPrinter/DebugHandlerBase.h
index b8bbcec133fd..cd02c4446a78 100644
--- a/lib/CodeGen/AsmPrinter/DebugHandlerBase.h
+++ b/lib/CodeGen/AsmPrinter/DebugHandlerBase.h
@@ -38,10 +38,12 @@ class DebugHandlerBase : public AsmPrinterHandler {
   MachineModuleInfo *MMI;
 
   /// Previous instruction's location information. This is used to
-  /// determine label location to indicate scope boundries in dwarf
-  /// debug info.
+  /// determine label location to indicate scope boundaries in debug info.
+  /// We track the previous instruction's source location (if not line 0),
+  /// whether it was a label, and its parent BB.
   DebugLoc PrevInstLoc;
   MCSymbol *PrevLabel = nullptr;
+  const MachineBasicBlock *PrevInstBB = nullptr;
 
   /// This location indicates end of function prologue and beginning of
   /// function body.
diff --git a/lib/CodeGen/AsmPrinter/DwarfDebug.cpp b/lib/CodeGen/AsmPrinter/DwarfDebug.cpp
index 884e03d3f0a2..949516b17a02 100644
--- a/lib/CodeGen/AsmPrinter/DwarfDebug.cpp
+++ b/lib/CodeGen/AsmPrinter/DwarfDebug.cpp
@@ -1010,31 +1010,70 @@ void DwarfDebug::beginInstruction(const MachineInstr *MI) {
   if (MI->isDebugValue())
     return;
   const DebugLoc &DL = MI->getDebugLoc();
-  if (DL == PrevInstLoc)
+  // When we emit a line-0 record, we don't update PrevInstLoc; so look at
+  // the last line number actually emitted, to see if it was line 0.
+  unsigned LastAsmLine =
+    Asm->OutStreamer->getContext().getCurrentDwarfLoc().getLine();
+
+  if (DL == PrevInstLoc) {
+    // If we have an ongoing unspecified location, nothing to do here.
+    if (!DL)
+      return;
+    // We have an explicit location, same as the previous location.
+    // But we might be coming back to it after a line 0 record.
+    if (LastAsmLine == 0 && DL.getLine() != 0) {
+      // Reinstate the source location but not marked as a statement.
+      const MDNode *Scope = DL.getScope();
+      recordSourceLine(DL.getLine(), DL.getCol(), Scope, /*Flags=*/0);
+    }
     return;
+  }
 
   if (!DL) {
     // We have an unspecified location, which might want to be line 0.
-    if (UnknownLocations) {
-      PrevInstLoc = DL;
-      recordSourceLine(0, 0, nullptr, 0);
+    // If we have already emitted a line-0 record, don't repeat it.
+    if (LastAsmLine == 0)
+      return;
+    // See if we have a reason to emit a line-0 record now.
+    // Reasons to emit a line-0 record include:
+    // - User asked for it (UnknownLocations).
+    // - Instruction has a label, so it's referenced from somewhere else,
+    //   possibly debug information; we want it to have a source location.
+    // - Instruction is at the top of a block; we don't want to inherit the
+    //   location from the physically previous (maybe unrelated) block.
+    if (UnknownLocations || PrevLabel ||
+        (PrevInstBB && PrevInstBB != MI->getParent())) {
+      // Preserve the file number, if we can, to save space in the line table.
+      // Do not update PrevInstLoc, it remembers the last non-0 line.
+      // FIXME: Also preserve the column number, to save more space?
+      const MDNode *Scope = PrevInstLoc ? PrevInstLoc.getScope() : nullptr;
+      recordSourceLine(0, 0, Scope, 0);
     }
     return;
   }
 
-  // We have a new, explicit location.
+  // We have an explicit location, different from the previous location.
+  // Don't repeat a line-0 record, but otherwise emit the new location.
+  // (The new location might be an explicit line 0, which we do emit.)
+  if (DL.getLine() == 0 && LastAsmLine == 0)
+    return;
   unsigned Flags = 0;
-  PrevInstLoc = DL;
   if (DL == PrologEndLoc) {
     Flags |= DWARF2_FLAG_PROLOGUE_END | DWARF2_FLAG_IS_STMT;
     PrologEndLoc = DebugLoc();
   }
-  if (DL.getLine() !=
-      Asm->OutStreamer->getContext().getCurrentDwarfLoc().getLine())
+  // If the line changed, we call that a new statement; unless we went to
+  // line 0 and came back, in which case it is not a new statement.
+  unsigned OldLine = PrevInstLoc ? PrevInstLoc.getLine() : LastAsmLine;
+  if (DL.getLine() && DL.getLine() != OldLine)
     Flags |= DWARF2_FLAG_IS_STMT;
 
   const MDNode *Scope = DL.getScope();
   recordSourceLine(DL.getLine(), DL.getCol(), Scope, Flags);
+
+  // If we're not at line 0, remember this location.
+  if (DL.getLine())
+    PrevInstLoc = DL;
 }
 
 static DebugLoc findPrologueEndLoc(const MachineFunction *MF) {
diff --git a/test/CodeGen/X86/stack-protector.ll b/test/CodeGen/X86/stack-protector.ll
index d3ff61743ed4..5166ed5b02aa 100644
--- a/test/CodeGen/X86/stack-protector.ll
+++ b/test/CodeGen/X86/stack-protector.ll
@@ -3888,23 +3888,27 @@ entry:
 define void @test32() #1 !dbg !7 {
 entry:
 ; LINUX-I386-LABEL: test32:
-; LINUX-I386:       .loc 1 0 0 prologue_end
+; LINUX-I386:       .loc 1 4 2 prologue_end
+; LINUX-I386:       .loc 1 0 0
 ; LINUX-I386-NEXT:  calll __stack_chk_fail
 
 ; LINUX-X64-LABEL: test32:
-; LINUX-X64:       .loc 1 0 0 prologue_end
+; LINUX-X64:       .loc 1 4 2 prologue_end
+; LINUX-X64:       .loc 1 0 0
 ; LINUX-X64-NEXT:  callq __stack_chk_fail
 
 ; LINUX-KERNEL-X64-LABEL: test32:
-; LINUX-KERNEL-X64:       .loc 1 0 0 prologue_end
+; LINUX-KERNEL-X64:       .loc 1 4 2 prologue_end
+; LINUX-KERNEL-X64:       .loc 1 0 0
 ; LINUX-KERNEL-X64-NEXT:  callq __stack_chk_fail
 
 ; OPENBSD-AMD64-LABEL: test32:
-; OPENBSD-AMD64:       .loc 1 0 0 prologue_end
+; OPENBSD-AMD64:       .loc 1 4 2 prologue_end
+; OPENBSD-AMD64:       .loc 1 0 0
 ; OPENBSD-AMD64-NEXT:  movl
 ; OPENBSD-AMD64-NEXT:  callq __stack_smash_handler
   %0 = alloca [5 x i8], align 1
-  ret void
+  ret void, !dbg !9
 }
 
 declare double @testi_aux()
@@ -3940,3 +3944,4 @@ attributes #5 = { ssp "stack-protector-buffer-size"="6" }
 !6 = distinct !DISubprogram(name: "__stack_chk_fail", scope: !1, type: !8, unit: !0)
 !7 = distinct !DISubprogram(name: "test32", scope: !1, type: !8, unit: !0)
 !8 = !DISubroutineType(types: !2)
+!9 = !DILocation(line: 4, column: 2, scope: !7)
diff --git a/test/DebugInfo/AArch64/line-header.ll b/test/DebugInfo/AArch64/line-header.ll
index f8220233faab..38d06f263e25 100644
--- a/test/DebugInfo/AArch64/line-header.ll
+++ b/test/DebugInfo/AArch64/line-header.ll
@@ -3,4 +3,4 @@
 
 ; check line table length is correctly calculated for both big and little endian
 CHECK-LABEL: .debug_line contents:
-CHECK: total_length: 0x0000003c
+CHECK: total_length: 0x0000003e
diff --git a/test/DebugInfo/X86/dwarf-no-source-loc.ll b/test/DebugInfo/X86/dwarf-no-source-loc.ll
new file mode 100644
index 000000000000..d71f95d999e5
--- /dev/null
+++ b/test/DebugInfo/X86/dwarf-no-source-loc.ll
@@ -0,0 +1,74 @@
+; Verify "no source location" directives appear in appropriate places.
+; RUN: llc %s -o - | FileCheck %s
+
+; Generated from this .cpp targeting linux using -g
+; and then removed function attributes as clutter.
+;
+; void bar(int *);
+; void baz(int *);
+; # 5 "no-source-loc.cpp"
+; void foo(int x) {
+;   int z;
+;   if (x)
+; # 20 "include.h"
+;     bar(&z);
+; # 10 "no-source-loc.cpp"
+;   baz(&z);
+; }
+
+target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
+target triple = "x86_64-unknown-linux-gnu"
+
+; Function Attrs: uwtable
+define void @_Z3fooi(i32 %x) !dbg !6 {
+entry:
+  %x.addr = alloca i32, align 4
+  %z = alloca i32, align 4
+  store i32 %x, i32* %x.addr, align 4
+  %0 = load i32, i32* %x.addr, align 4, !dbg !8
+  %tobool = icmp ne i32 %0, 0, !dbg !8
+  br i1 %tobool, label %if.then, label %if.end, !dbg !8
+
+if.then:                                          ; preds = %entry
+  call void @_Z3barPi(i32* %z), !dbg !9
+  br label %if.end, !dbg !9
+
+if.end:                                           ; preds = %if.then, %entry
+  call void @_Z3bazPi(i32* %z), !dbg !12
+  ret void, !dbg !14
+}
+
+; CHECK:      .loc 1 7 7
+; CHECK-NOT:  .loc
+; CHECK:      .loc 1 0 0 is_stmt 0
+; CHECK-NOT:  .loc
+; CHECK:      .loc 2 20 5 is_stmt 1
+; CHECK:      .LBB0_2:
+; CHECK-NEXT: .loc 2 0 0 is_stmt 0
+; CHECK-NOT:  .loc
+; CHECK:      .loc 1 10 3 is_stmt 1
+
+
+declare void @_Z3barPi(i32*)
+
+declare void @_Z3bazPi(i32*)
+
+!llvm.dbg.cu = !{!0}
+!llvm.module.flags = !{!3, !4}
+!llvm.ident = !{!5}
+
+!0 = distinct !DICompileUnit(language: DW_LANG_C_plus_plus, file: !1, producer: "clang version 4.0.0 (trunk 278782)", isOptimized: false, runtimeVersion: 0, emissionKind: LineTablesOnly, enums: !2)
+!1 = !DIFile(filename: "no-source-loc.cpp", directory: "/tests")
+!2 = !{}
+!3 = !{i32 2, !"Dwarf Version", i32 4}
+!4 = !{i32 2, !"Debug Info Version", i32 3}
+!5 = !{!"clang version 4.0.0 (trunk 278782)"}
+!6 = distinct !DISubprogram(name: "foo", scope: !1, file: !1, line: 5, type: !7, isLocal: false, isDefinition: true, scopeLine: 5, flags: DIFlagPrototyped, isOptimized: false, unit: !0, variables: !2)
+!7 = !DISubroutineType(types: !2)
+!8 = !DILocation(line: 7, column: 7, scope: !6)
+!9 = !DILocation(line: 20, column: 5, scope: !10)
+!10 = !DILexicalBlockFile(scope: !6, file: !11, discriminator: 0)
+!11 = !DIFile(filename: "include.h", directory: "/tests")
+!12 = !DILocation(line: 10, column: 3, scope: !13)
+!13 = !DILexicalBlockFile(scope: !6, file: !1, discriminator: 0)
+!14 = !DILocation(line: 11, column: 1, scope: !13)