diff --git a/lib/Target/ARM/ARMSubtarget.cpp b/lib/Target/ARM/ARMSubtarget.cpp
index 5be0d2b819a..25e611bee49 100644
--- a/lib/Target/ARM/ARMSubtarget.cpp
+++ b/lib/Target/ARM/ARMSubtarget.cpp
@@ -83,7 +83,7 @@ ARMSubtarget::ARMSubtarget(const std::string &TT, const std::string &CPU,
   // Insert the architecture feature derived from the target triple into the
   // feature string. This is important for setting features that are implied
   // based on the architecture version.
-  std::string ArchFS = ARM_MC::ParseARMTriple(TT);
+  std::string ArchFS = ARM_MC::ParseARMTriple(TT, CPUString);
   if (!FS.empty()) {
     if (!ArchFS.empty())
       ArchFS = ArchFS + "," + FS;
diff --git a/lib/Target/ARM/MCTargetDesc/ARMMCTargetDesc.cpp b/lib/Target/ARM/MCTargetDesc/ARMMCTargetDesc.cpp
index e3512cda3ae..67f6f6f3520 100644
--- a/lib/Target/ARM/MCTargetDesc/ARMMCTargetDesc.cpp
+++ b/lib/Target/ARM/MCTargetDesc/ARMMCTargetDesc.cpp
@@ -35,7 +35,7 @@
 
 using namespace llvm;
 
-std::string ARM_MC::ParseARMTriple(StringRef TT) {
+std::string ARM_MC::ParseARMTriple(StringRef TT, StringRef CPU) {
   // Set the boolean corresponding to the current target triple, or the default
   // if one cannot be determined, to true.
   unsigned Len = TT.size();
@@ -62,9 +62,18 @@ std::string ARM_MC::ParseARMTriple(StringRef TT) {
         // v7em: FeatureNoARM, FeatureDB, FeatureHWDiv, FeatureDSPThumb2,
         //       FeatureT2XtPk, FeatureMClass
         ARMArchFeature = "+v7,+noarm,+db,+hwdiv,+t2dsp,t2xtpk,+mclass";
-      } else
-        // v7a: FeatureNEON, FeatureDB, FeatureDSPThumb2, FeatureT2XtPk
-        ARMArchFeature = "+v7,+neon,+db,+t2dsp,+t2xtpk";
+      } else {
+        // v7 CPUs have lots of different feature sets. If no CPU is specified,
+        // then assume v7a (e.g. cortex-a8) feature set. Otherwise, return
+        // the "minimum" feature set and use CPU string to figure out the exact
+        // features.
+        if (CPU == "generic")
+          // v7a: FeatureNEON, FeatureDB, FeatureDSPThumb2, FeatureT2XtPk
+          ARMArchFeature = "+v7,+neon,+db,+t2dsp,+t2xtpk";
+        else
+          // Use CPU to figure out the exact features.
+          ARMArchFeature = "+v7";
+      }
     } else if (SubVer == '6') {
       if (Len >= Idx+3 && TT[Idx+1] == 't' && TT[Idx+2] == '2')
         ARMArchFeature = "+v6t2";
@@ -94,7 +103,7 @@ std::string ARM_MC::ParseARMTriple(StringRef TT) {
 
 MCSubtargetInfo *ARM_MC::createARMMCSubtargetInfo(StringRef TT, StringRef CPU,
                                                   StringRef FS) {
-  std::string ArchFS = ARM_MC::ParseARMTriple(TT);
+  std::string ArchFS = ARM_MC::ParseARMTriple(TT, CPU);
   if (!FS.empty()) {
     if (!ArchFS.empty())
       ArchFS = ArchFS + "," + FS.str();
diff --git a/lib/Target/ARM/MCTargetDesc/ARMMCTargetDesc.h b/lib/Target/ARM/MCTargetDesc/ARMMCTargetDesc.h
index 88472d7ffc3..57719369c66 100644
--- a/lib/Target/ARM/MCTargetDesc/ARMMCTargetDesc.h
+++ b/lib/Target/ARM/MCTargetDesc/ARMMCTargetDesc.h
@@ -31,7 +31,7 @@ class raw_ostream;
 extern Target TheARMTarget, TheThumbTarget;
 
 namespace ARM_MC {
-  std::string ParseARMTriple(StringRef TT);
+  std::string ParseARMTriple(StringRef TT, StringRef CPU);
 
   /// createARMMCSubtargetInfo - Create a ARM MCSubtargetInfo instance.
   /// This is exposed so Asm parser, etc. do not need to go through
diff --git a/test/CodeGen/Thumb2/thumb2-uxt_rot.ll b/test/CodeGen/Thumb2/thumb2-uxt_rot.ll
index 03189aa7e15..61e849ef4a4 100644
--- a/test/CodeGen/Thumb2/thumb2-uxt_rot.ll
+++ b/test/CodeGen/Thumb2/thumb2-uxt_rot.ll
@@ -1,15 +1,22 @@
-; RUN: llc < %s -march=thumb -mattr=+thumb2,+t2xtpk | FileCheck %s
+; RUN: llc < %s -march=thumb -mcpu=cortex-a8 | FileCheck %s --check-prefix=A8
+; RUN: llc < %s -march=thumb -mcpu=cortex-m3 | FileCheck %s --check-prefix=M3
+; rdar://11318438
 
 define zeroext i8 @test1(i32 %A.u)  {
-; CHECK: test1
-; CHECK: uxtb r0, r0
+; A8: test1
+; A8: uxtb r0, r0
     %B.u = trunc i32 %A.u to i8
     ret i8 %B.u
 }
 
 define zeroext i32 @test2(i32 %A.u, i32 %B.u)  {
-; CHECK: test2
-; CHECK: uxtab  r0, r0, r1
+; A8: test2
+; A8: uxtab  r0, r0, r1
+
+; M3: test2
+; M3: uxtb  r1, r1
+; M3-NOT: uxtab
+; M3: add   r0, r1
     %C.u = trunc i32 %B.u to i8
     %D.u = zext i8 %C.u to i32
     %E.u = add i32 %A.u, %D.u
@@ -17,8 +24,8 @@ define zeroext i32 @test2(i32 %A.u, i32 %B.u)  {
 }
 
 define zeroext i32 @test3(i32 %A.u)  {
-; CHECK: test3
-; CHECK: uxth.w r0, r0, ror #8
+; A8: test3
+; A8: uxth.w r0, r0, ror #8
     %B.u = lshr i32 %A.u, 8
     %C.u = shl i32 %A.u, 24
     %D.u = or i32 %B.u, %C.u