diff --git a/llvm/lib/AsmParser/LLParser.h b/llvm/lib/AsmParser/LLParser.h
--- a/llvm/lib/AsmParser/LLParser.h
+++ b/llvm/lib/AsmParser/LLParser.h
@@ -308,6 +308,7 @@
     bool ParseTopLevelEntities();
     bool ValidateEndOfModule();
     bool ValidateEndOfIndex();
+    bool ParseTargetDefinitions();
     bool ParseTargetDefinition();
     bool ParseModuleAsm();
     bool ParseSourceFileName();
diff --git a/llvm/lib/AsmParser/LLParser.cpp b/llvm/lib/AsmParser/LLParser.cpp
--- a/llvm/lib/AsmParser/LLParser.cpp
+++ b/llvm/lib/AsmParser/LLParser.cpp
@@ -70,6 +70,11 @@
         Lex.getLoc(),
         "Can't read textual IR with a Context that discards named Values");
 
+  if (M) {
+    if (ParseTargetDefinitions())
+      return true;
+  }
+
   return ParseTopLevelEntities() || ValidateEndOfModule() ||
          ValidateEndOfIndex();
 }
@@ -294,6 +299,20 @@
 // Top-Level Entities
 //===----------------------------------------------------------------------===//
 
+bool LLParser::ParseTargetDefinitions() {
+  while (true) {
+    if (Lex.getKind() == lltok::kw_target) {
+      if (ParseTargetDefinition())
+        return true;
+    } else if (Lex.getKind() == lltok::kw_source_filename) {
+      if (ParseSourceFileName())
+        return true;
+    } else {
+      return false;
+    }
+  }
+}
+
 bool LLParser::ParseTopLevelEntities() {
   // If there is no Module, then parse just the summary index entries.
   if (!M) {
@@ -322,11 +341,6 @@
     case lltok::kw_declare: if (ParseDeclare()) return true; break;
     case lltok::kw_define:  if (ParseDefine()) return true; break;
     case lltok::kw_module:  if (ParseModuleAsm()) return true; break;
-    case lltok::kw_target:  if (ParseTargetDefinition()) return true; break;
-    case lltok::kw_source_filename:
-      if (ParseSourceFileName())
-        return true;
-      break;
     case lltok::kw_deplibs: if (ParseDepLibs()) return true; break;
     case lltok::LocalVarID: if (ParseUnnamedType()) return true; break;
     case lltok::LocalVar:   if (ParseNamedType()) return true; break;
diff --git a/llvm/test/Analysis/DependenceAnalysis/NonCanonicalizedSubscript.ll b/llvm/test/Analysis/DependenceAnalysis/NonCanonicalizedSubscript.ll
--- a/llvm/test/Analysis/DependenceAnalysis/NonCanonicalizedSubscript.ll
+++ b/llvm/test/Analysis/DependenceAnalysis/NonCanonicalizedSubscript.ll
@@ -36,9 +36,6 @@
   ret void
 }
 
-target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
-target triple = "x86_64-unknown-linux-gnu"
-
 ;  unsigned i, j;
 ;  for (i = 1; i < SIZE; i++) {
 ;    for (j = i; j < SIZE; j++) {
diff --git a/llvm/test/Analysis/ScalarEvolution/2012-05-29-MulAddRec.ll b/llvm/test/Analysis/ScalarEvolution/2012-05-29-MulAddRec.ll
--- a/llvm/test/Analysis/ScalarEvolution/2012-05-29-MulAddRec.ll
+++ b/llvm/test/Analysis/ScalarEvolution/2012-05-29-MulAddRec.ll
@@ -8,6 +8,8 @@
 ;
 ; PR12929: cast<Ty>() argument of incompatible type
 
+target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64-S128"
+
 declare void @use(i8 %x)
 
 ; CHECK: @func
@@ -20,7 +22,6 @@
 ; CHECK: %0 = add i8 %inc1, 10
 ; CHECK: br label %for.cond
 
-target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64-S128"
 define void @func() noreturn nounwind uwtable ssp {
 entry:
   br label %for.cond
diff --git a/llvm/test/Transforms/InstCombine/2012-09-17-ZeroSizedAlloca.ll b/llvm/test/Transforms/InstCombine/2012-09-17-ZeroSizedAlloca.ll
--- a/llvm/test/Transforms/InstCombine/2012-09-17-ZeroSizedAlloca.ll
+++ b/llvm/test/Transforms/InstCombine/2012-09-17-ZeroSizedAlloca.ll
@@ -3,12 +3,12 @@
 ; When merging zero sized alloca check that requested alignments of the allocas
 ; are obeyed.
 
-@x = global i8* null, align 8
-@y = global i8* null, align 8
-
 target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64-S128"
 target triple = "x86_64-apple-macosx10.8.0"
 
+@x = global i8* null, align 8
+@y = global i8* null, align 8
+
 ; CHECK-LABEL: @f(
 ; CHECK-NEXT: alloca [0 x i8], align 1024
 ; CHECK-NOT: alloca
diff --git a/llvm/test/Transforms/InstCombine/getelementptr-folding.ll b/llvm/test/Transforms/InstCombine/getelementptr-folding.ll
--- a/llvm/test/Transforms/InstCombine/getelementptr-folding.ll
+++ b/llvm/test/Transforms/InstCombine/getelementptr-folding.ll
@@ -1,13 +1,13 @@
 ; RUN: opt -instcombine -S < %s | FileCheck %s
 
-%struct.matrix_float3x3 = type { [3 x <3 x float>] }
-
 ; We used to fold this by rewriting the indices to 0, 0, 2, 0.  This is
 ; invalid because there is a 4-byte padding after each <3 x float> field.
 
 target datalayout = "e-m:o-i64:64-f80:128-n8:16:32:64-S128"
 target triple = "x86_64-apple-macosx10.11.0"
 
+%struct.matrix_float3x3 = type { [3 x <3 x float>] }
+
 @matrix_identity_float3x3 = external global %struct.matrix_float3x3, align 16
 @bbb = global float* getelementptr inbounds (%struct.matrix_float3x3, %struct.matrix_float3x3* @matrix_identity_float3x3, i64 0, i32 0, i64 1, i64 3)
 ; CHECK: @bbb = global float* getelementptr inbounds (%struct.matrix_float3x3, %struct.matrix_float3x3* @matrix_identity_float3x3, i64 0, i32 0, i64 1, i64 3)
diff --git a/llvm/test/Transforms/InstCombine/overflow-mul.ll b/llvm/test/Transforms/InstCombine/overflow-mul.ll
--- a/llvm/test/Transforms/InstCombine/overflow-mul.ll
+++ b/llvm/test/Transforms/InstCombine/overflow-mul.ll
@@ -1,5 +1,16 @@
 ; RUN: opt -S -instcombine < %s | FileCheck %s
 
+; The last test needs this weird datalayout.
+target datalayout = "i32:8:8"
+; Without it, InstCombine will align the pointed on 4 Bytes
+; The KnownBitsZero that result from the alignment allows to
+; turn:
+;    and i32 %mul, 255
+; to:
+;    and i32 %mul, 252
+; The mask is no longer in the form 2^n-1  and this prevents the transformation.
+
+
 ; return mul(zext x, zext y) > MAX
 define i32 @pr4917_1(i32 %x, i32 %y) nounwind {
 ; CHECK-LABEL: @pr4917_1(
@@ -175,16 +186,6 @@
 }
 
 
-; The last test needs this weird datalayout.
-target datalayout = "i32:8:8"
-; Without it, InstCombine will align the pointed on 4 Bytes
-; The KnownBitsZero that result from the alignment allows to
-; turn:
-;    and i32 %mul, 255
-; to:
-;    and i32 %mul, 252
-; The mask is no longer in the form 2^n-1  and this prevents the transformation.
-
 @pr21445_data = external global i32
 define i1 @pr21445(i8 %a) {
 ; CHECK-LABEL: @pr21445(
diff --git a/llvm/test/Transforms/InstCombine/wcslen-3.ll b/llvm/test/Transforms/InstCombine/wcslen-3.ll
--- a/llvm/test/Transforms/InstCombine/wcslen-3.ll
+++ b/llvm/test/Transforms/InstCombine/wcslen-3.ll
@@ -3,12 +3,12 @@
 ;
 ; RUN: opt < %s -instcombine -S | FileCheck %s
 
+target datalayout = "e-m:o-i64:64-f80:128-n8:16:32:64-S128"
+
 ; Test behavior for wchar_size==2
 !llvm.module.flags = !{!0}
 !0 = !{i32 1, !"wchar_size", i32 2}
 
-target datalayout = "e-m:o-i64:64-f80:128-n8:16:32:64-S128"
-
 declare i64 @wcslen(i16*)
 
 @hello = constant [6 x i16] [i16 104, i16 101, i16 108, i16 108, i16 111, i16 0]
diff --git a/llvm/test/Transforms/LoopIdiom/X86/popcnt.ll b/llvm/test/Transforms/LoopIdiom/X86/popcnt.ll
--- a/llvm/test/Transforms/LoopIdiom/X86/popcnt.ll
+++ b/llvm/test/Transforms/LoopIdiom/X86/popcnt.ll
@@ -1,5 +1,7 @@
 ; RUN: opt -loop-idiom < %s -mtriple=x86_64-apple-darwin -mcpu=corei7 -S | FileCheck %s
 
+target triple = "x86_64-apple-macosx10.8.0"
+
 ;To recognize this pattern:
 ;int popcount(unsigned long long a) {
 ;    int c = 0;
@@ -75,7 +77,6 @@
 }
 
 ; Some variants once cause crash
-target triple = "x86_64-apple-macosx10.8.0"
 
 define i32 @PopCntCrash1(i64 %a) nounwind uwtable readnone ssp {
 entry:
diff --git a/llvm/test/Transforms/LoopVectorize/explicit_outer_nonuniform_inner.ll b/llvm/test/Transforms/LoopVectorize/explicit_outer_nonuniform_inner.ll
--- a/llvm/test/Transforms/LoopVectorize/explicit_outer_nonuniform_inner.ll
+++ b/llvm/test/Transforms/LoopVectorize/explicit_outer_nonuniform_inner.ll
@@ -73,8 +73,6 @@
 ; CHECK: LV: Not vectorizing: Outer loop contains divergent loops.
 ; CHECK: LV: Not vectorizing: Unsupported outer loop.
 
-target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
-
 define void @loop_ub(i32* nocapture %a, i32* nocapture readonly %b, i32 %N, i32 %M) local_unnamed_addr {
 entry:
   %cmp32 = icmp sgt i32 %N, 0
@@ -121,8 +119,6 @@
 ; CHECK: LV: Not vectorizing: Outer loop contains divergent loops.
 ; CHECK: LV: Not vectorizing: Unsupported outer loop.
 
-target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
-
 define void @iv_step(i32* nocapture %a, i32* nocapture readonly %b, i32 %N, i32 %M) local_unnamed_addr {
 entry:
   %cmp33 = icmp sgt i32 %N, 0
diff --git a/llvm/test/Transforms/LoopVectorize/explicit_outer_uniform_diverg_branch.ll b/llvm/test/Transforms/LoopVectorize/explicit_outer_uniform_diverg_branch.ll
--- a/llvm/test/Transforms/LoopVectorize/explicit_outer_uniform_diverg_branch.ll
+++ b/llvm/test/Transforms/LoopVectorize/explicit_outer_uniform_diverg_branch.ll
@@ -77,8 +77,6 @@
 ; CHECK: Unsupported conditional branch.
 ; CHECK: LV: Not vectorizing: Unsupported outer loop.
 
-target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
-
 define void @divergent_branch(i32* nocapture %a, i32* nocapture readonly %b, i32 %N, i32 %M) local_unnamed_addr {
 entry:
   %cmp39 = icmp sgt i32 %N, 0
diff --git a/llvm/test/Transforms/NewGVN/pr33187.ll b/llvm/test/Transforms/NewGVN/pr33187.ll
--- a/llvm/test/Transforms/NewGVN/pr33187.ll
+++ b/llvm/test/Transforms/NewGVN/pr33187.ll
@@ -111,8 +111,6 @@
 attributes #0 = { nounwind uwtable "correctly-rounded-divide-sqrt-fp-math"="false" "disable-tail-calls"="false" "less-precise-fpmad"="false" "frame-pointer"="none" "no-infs-fp-math"="false" "no-jump-tables"="false" "no-nans-fp-math"="false" "no-signed-zeros-fp-math"="false" "no-trapping-math"="false" "stack-protector-buffer-size"="8" "target-cpu"="x86-64" "target-features"="+fxsr,+mmx,+sse,+sse2,+x87" "unsafe-fp-math"="false" "use-soft-float"="false" }
 
 
-source_filename = "pr33187-c.ll"
-
 define void @a() {
 ; CHECK-LABEL: @a(
 ; CHECK-NEXT:  b:
diff --git a/llvm/test/Transforms/NewGVN/refine-stores.ll b/llvm/test/Transforms/NewGVN/refine-stores.ll
--- a/llvm/test/Transforms/NewGVN/refine-stores.ll
+++ b/llvm/test/Transforms/NewGVN/refine-stores.ll
@@ -5,6 +5,10 @@
 ;; We also are testing that various variations that cause stores to move classes
 ;; have the right class movement happen
 ;; All of these tests result in verification failures if it does not.
+
+source_filename = "bugpoint-output-daef094.bc"
+target triple = "x86_64-apple-darwin16.5.0"
+
 %struct.eggs = type {}
 
 define void @spam(i32 *%a) {
@@ -78,10 +82,6 @@
   br i1 undef, label %c, label %e
 }
 
-; ModuleID = 'bugpoint-reduced-simplified.bc'
-source_filename = "bugpoint-output-daef094.bc"
-target triple = "x86_64-apple-darwin16.5.0"
-
 %struct.hoge = type {}
 
 define void @widget(%struct.hoge* %arg) {
@@ -132,8 +132,6 @@
 }
 
 declare void @quux()
-; ModuleID = 'short.ll'
-source_filename = "short.ll"
 
 %struct.a = type {}
 
diff --git a/llvm/test/Transforms/SafeStack/X86/call.ll b/llvm/test/Transforms/SafeStack/X86/call.ll
--- a/llvm/test/Transforms/SafeStack/X86/call.ll
+++ b/llvm/test/Transforms/SafeStack/X86/call.ll
@@ -1,6 +1,9 @@
 ; RUN: opt -safe-stack -S -mtriple=i386-pc-linux-gnu < %s -o - | FileCheck %s
 ; RUN: opt -safe-stack -S -mtriple=x86_64-pc-linux-gnu < %s -o - | FileCheck %s
 
+target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
+target triple = "x86_64-unknown-linux-gnu"
+
 @.str = private unnamed_addr constant [4 x i8] c"%s\0A\00", align 1
 
 ; no arrays / no nested arrays
@@ -20,9 +23,6 @@
 
 declare i32 @printf(i8*, ...)
 
-target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
-target triple = "x86_64-unknown-linux-gnu"
-
 define void @call_memset(i64 %len) safestack {
 entry:
   ; CHECK-LABEL: define void @call_memset