Index: lib/Target/R600/AMDGPUISelLowering.cpp
===================================================================
--- lib/Target/R600/AMDGPUISelLowering.cpp
+++ lib/Target/R600/AMDGPUISelLowering.cpp
@@ -776,6 +776,20 @@
     return DAG.getNode(ISD::getExtForLoadExtType(ExtType), DL, VT, ExtLoad32);
   }
 
+  if (ExtType == ISD::NON_EXTLOAD && VT.getSizeInBits() < 32) {
+    assert(VT == MVT::i1 && "Only i1 non-extloads expected");
+    // FIXME: Copied from PPC
+    // First, load into 32 bits, then truncate to 1 bit.
+
+    SDValue Chain = Load->getChain();
+    SDValue BasePtr = Load->getBasePtr();
+    MachineMemOperand *MMO = Load->getMemOperand();
+
+    SDValue NewLD = DAG.getExtLoad(ISD::EXTLOAD, DL, MVT::i32, Chain,
+                                   BasePtr, MVT::i8, MMO);
+    return DAG.getNode(ISD::TRUNCATE, DL, VT, NewLD);
+  }
+
   // Lower loads constant address space global variable loads
   if (Load->getAddressSpace() == AMDGPUAS::CONSTANT_ADDRESS &&
       isa<GlobalVariable>(GetUnderlyingObject(Load->getPointerInfo().V))) {
Index: lib/Target/R600/SIISelLowering.cpp
===================================================================
--- lib/Target/R600/SIISelLowering.cpp
+++ lib/Target/R600/SIISelLowering.cpp
@@ -126,19 +126,24 @@
 
   setOperationAction(ISD::INTRINSIC_VOID, MVT::Other, Custom);
 
-  setLoadExtAction(ISD::SEXTLOAD, MVT::i32, Expand);
+  setLoadExtAction(ISD::SEXTLOAD, MVT::i1, Promote);
   setLoadExtAction(ISD::SEXTLOAD, MVT::i8, Custom);
   setLoadExtAction(ISD::SEXTLOAD, MVT::i16, Custom);
-  setLoadExtAction(ISD::ZEXTLOAD, MVT::i32, Expand);
-  setLoadExtAction(ISD::ZEXTLOAD, MVT::i8, Custom);
-  setLoadExtAction(ISD::ZEXTLOAD, MVT::i16, Custom);
+  setLoadExtAction(ISD::SEXTLOAD, MVT::i32, Expand);
   setLoadExtAction(ISD::SEXTLOAD, MVT::v8i16, Expand);
   setLoadExtAction(ISD::SEXTLOAD, MVT::v16i16, Expand);
 
+  setLoadExtAction(ISD::ZEXTLOAD, MVT::i1, Promote);
+  setLoadExtAction(ISD::ZEXTLOAD, MVT::i8, Custom);
+  setLoadExtAction(ISD::ZEXTLOAD, MVT::i16, Custom);
+  setLoadExtAction(ISD::ZEXTLOAD, MVT::i32, Expand);
+
+  setLoadExtAction(ISD::EXTLOAD, MVT::i1, Promote);
   setLoadExtAction(ISD::EXTLOAD, MVT::i8, Custom);
   setLoadExtAction(ISD::EXTLOAD, MVT::i16, Custom);
   setLoadExtAction(ISD::EXTLOAD, MVT::i32, Expand);
   setLoadExtAction(ISD::EXTLOAD, MVT::f32, Expand);
+
   setTruncStoreAction(MVT::i32, MVT::i8, Custom);
   setTruncStoreAction(MVT::i32, MVT::i16, Custom);
   setTruncStoreAction(MVT::f64, MVT::f32, Expand);
@@ -147,6 +152,8 @@
   setTruncStoreAction(MVT::v8i32, MVT::v8i16, Expand);
   setTruncStoreAction(MVT::v16i32, MVT::v16i16, Expand);
 
+  setOperationAction(ISD::LOAD, MVT::i1, Custom);
+
   setOperationAction(ISD::GlobalAddress, MVT::i32, Custom);
   setOperationAction(ISD::GlobalAddress, MVT::i64, Custom);
   setOperationAction(ISD::FrameIndex, MVT::i32, Custom);
Index: test/CodeGen/R600/load-i1.ll
===================================================================
--- /dev/null
+++ test/CodeGen/R600/load-i1.ll
@@ -0,0 +1,107 @@
+; RUN: llc -march=r600 -mcpu=SI < %s | FileCheck -check-prefix=SI %s
+
+
+; SI-LABEL: @global_copy_i1_to_i1
+; SI: BUFFER_LOAD_UBYTE
+; SI: V_AND_B32_e32 v{{[0-9]+}}, 1
+; SI: BUFFER_STORE_BYTE
+; SI: S_ENDPGM
+define void @global_copy_i1_to_i1(i1 addrspace(1)* %out, i1 addrspace(1)* %in) nounwind {
+  %load = load i1 addrspace(1)* %in
+  store i1 %load, i1 addrspace(1)* %out, align 1
+  ret void
+}
+
+; SI-LABEL: @global_sextload_i1_to_i32
+; XSI: BUFFER_LOAD_BYTE
+; SI: BUFFER_STORE_DWORD
+; SI: S_ENDPGM
+define void @global_sextload_i1_to_i32(i32 addrspace(1)* %out, i1 addrspace(1)* %in) nounwind {
+  %load = load i1 addrspace(1)* %in
+  %ext = sext i1 %load to i32
+  store i32 %ext, i32 addrspace(1)* %out, align 4
+  ret void
+}
+
+; SI-LABEL: @global_zextload_i1_to_i32
+; SI: BUFFER_LOAD_UBYTE
+; SI: BUFFER_STORE_DWORD
+; SI: S_ENDPGM
+define void @global_zextload_i1_to_i32(i32 addrspace(1)* %out, i1 addrspace(1)* %in) nounwind {
+  %load = load i1 addrspace(1)* %in
+  %ext = zext i1 %load to i32
+  store i32 %ext, i32 addrspace(1)* %out, align 4
+  ret void
+}
+
+; SI-LABEL: @global_sextload_i1_to_i64
+; XSI: BUFFER_LOAD_BYTE
+; SI: BUFFER_STORE_DWORDX2
+; SI: S_ENDPGM
+define void @global_sextload_i1_to_i64(i64 addrspace(1)* %out, i1 addrspace(1)* %in) nounwind {
+  %load = load i1 addrspace(1)* %in
+  %ext = sext i1 %load to i64
+  store i64 %ext, i64 addrspace(1)* %out, align 4
+  ret void
+}
+
+; SI-LABEL: @global_zextload_i1_to_i64
+; SI: BUFFER_LOAD_UBYTE
+; SI: BUFFER_STORE_DWORDX2
+; SI: S_ENDPGM
+define void @global_zextload_i1_to_i64(i64 addrspace(1)* %out, i1 addrspace(1)* %in) nounwind {
+  %load = load i1 addrspace(1)* %in
+  %ext = zext i1 %load to i64
+  store i64 %ext, i64 addrspace(1)* %out, align 4
+  ret void
+}
+
+; SI-LABEL: @i1_arg
+; SI: BUFFER_LOAD_UBYTE
+; SI: V_AND_B32_e32
+; SI: BUFFER_STORE_BYTE
+; SI: S_ENDPGM
+define void @i1_arg(i1 addrspace(1)* %out, i1 %x) nounwind {
+  store i1 %x, i1 addrspace(1)* %out, align 1
+  ret void
+}
+
+; SI-LABEL: @i1_arg_zext_i32
+; SI: BUFFER_LOAD_UBYTE
+; SI: BUFFER_STORE_DWORD
+; SI: S_ENDPGM
+define void @i1_arg_zext_i32(i32 addrspace(1)* %out, i1 %x) nounwind {
+  %ext = zext i1 %x to i32
+  store i32 %ext, i32 addrspace(1)* %out, align 4
+  ret void
+}
+
+; SI-LABEL: @i1_arg_zext_i64
+; SI: BUFFER_LOAD_UBYTE
+; SI: BUFFER_STORE_DWORDX2
+; SI: S_ENDPGM
+define void @i1_arg_zext_i64(i64 addrspace(1)* %out, i1 %x) nounwind {
+  %ext = zext i1 %x to i64
+  store i64 %ext, i64 addrspace(1)* %out, align 8
+  ret void
+}
+
+; SI-LABEL: @i1_arg_sext_i32
+; XSI: BUFFER_LOAD_BYTE
+; SI: BUFFER_STORE_DWORD
+; SI: S_ENDPGM
+define void @i1_arg_sext_i32(i32 addrspace(1)* %out, i1 %x) nounwind {
+  %ext = sext i1 %x to i32
+  store i32 %ext, i32addrspace(1)* %out, align 4
+  ret void
+}
+
+; SI-LABEL: @i1_arg_sext_i64
+; XSI: BUFFER_LOAD_BYTE
+; SI: BUFFER_STORE_DWORDX2
+; SI: S_ENDPGM
+define void @i1_arg_sext_i64(i64 addrspace(1)* %out, i1 %x) nounwind {
+  %ext = sext i1 %x to i64
+  store i64 %ext, i64 addrspace(1)* %out, align 8
+  ret void
+}