3 files changed, 57 insertions, 24 deletions
diff --git a/llvm/test/CodeGen/AMDGPU/amdgcn.private-memory.ll b/llvm/test/CodeGen/AMDGPU/amdgcn.private-memory.ll
new file mode 100644
index 00000000000..d919efe15a1
--- /dev/null
+++ b/llvm/test/CodeGen/AMDGPU/amdgcn.private-memory.ll
@@ -0,0 +1,30 @@
+; RUN: llc -show-mc-encoding -mattr=+promote-alloca -verify-machineinstrs -march=amdgcn < %s | FileCheck %s -check-prefix=SI-PROMOTE -check-prefix=SI -check-prefix=FUNC
+; RUN: llc -show-mc-encoding -mattr=+promote-alloca -verify-machineinstrs -mtriple=amdgcn--amdhsa -mcpu=kaveri < %s | FileCheck %s -check-prefix=SI-PROMOTE -check-prefix=SI -check-prefix=FUNC -check-prefix=HSA-PROMOTE
+; RUN: llc -show-mc-encoding -mattr=-promote-alloca -verify-machineinstrs -march=amdgcn < %s | FileCheck %s -check-prefix=SI-ALLOCA -check-prefix=SI -check-prefix=FUNC
+; RUN: llc -show-mc-encoding -mattr=-promote-alloca -verify-machineinstrs -mtriple=amdgcn-amdhsa -mcpu=kaveri < %s | FileCheck %s -check-prefix=SI-ALLOCA -check-prefix=SI -check-prefix=FUNC -check-prefix=HSA-ALLOCA
+; RUN: llc -show-mc-encoding -mattr=+promote-alloca -verify-machineinstrs -march=amdgcn -mcpu=tonga < %s | FileCheck %s -check-prefix=SI-PROMOTE -check-prefix=SI -check-prefix=FUNC
+; RUN: llc -show-mc-encoding -mattr=-promote-alloca -verify-machineinstrs -march=amdgcn -mcpu=tonga < %s | FileCheck %s -check-prefix=SI-ALLOCA -check-prefix=SI -check-prefix=FUNC
+
+
+declare i32 @llvm.amdgcn.workitem.id.x() nounwind readnone
+
+
+; Make sure we don't overwrite workitem information with private memory
+
+; FUNC-LABEL: {{^}}work_item_info:
+
+; SI-NOT: v_mov_b32_e{{(32|64)}} v0
+define void @work_item_info(i32 addrspace(1)* %out, i32 %in) {
+entry:
+  %0 = alloca [2 x i32]
+  %1 = getelementptr [2 x i32], [2 x i32]* %0, i32 0, i32 0
+  %2 = getelementptr [2 x i32], [2 x i32]* %0, i32 0, i32 1
+  store i32 0, i32* %1
+  store i32 1, i32* %2
+  %3 = getelementptr [2 x i32], [2 x i32]* %0, i32 0, i32 %in
+  %4 = load i32, i32* %3
+  %5 = call i32 @llvm.amdgcn.workitem.id.x()
+  %6 = add i32 %4, %5
+  store i32 %6, i32 addrspace(1)* %out
+  ret void
+}
diff --git a/llvm/test/CodeGen/AMDGPU/private-memory.ll b/llvm/test/CodeGen/AMDGPU/amdgpu.private-memory.ll
index 36dedf94ebf..99c7de40fb4 100644
--- a/llvm/test/CodeGen/AMDGPU/private-memory.ll
+++ b/llvm/test/CodeGen/AMDGPU/amdgpu.private-memory.ll
@@ -1,3 +1,4 @@
+; RUN: llc -march=r600 -mcpu=cypress < %s | FileCheck %s -check-prefix=R600 -check-prefix=FUNC
 ; RUN: llc -show-mc-encoding -mattr=+promote-alloca -verify-machineinstrs -march=amdgcn < %s | FileCheck %s -check-prefix=SI-PROMOTE -check-prefix=SI -check-prefix=FUNC
 ; RUN: llc -show-mc-encoding -mattr=+promote-alloca -verify-machineinstrs -mtriple=amdgcn--amdhsa -mcpu=kaveri < %s | FileCheck %s -check-prefix=SI-PROMOTE -check-prefix=SI -check-prefix=FUNC -check-prefix=HSA-PROMOTE
 ; RUN: llc -show-mc-encoding -mattr=-promote-alloca -verify-machineinstrs -march=amdgcn < %s | FileCheck %s -check-prefix=SI-ALLOCA -check-prefix=SI -check-prefix=FUNC
@@ -8,8 +9,6 @@
 ; RUN: opt -S -mtriple=amdgcn-unknown-amdhsa -mcpu=kaveri -amdgpu-promote-alloca < %s | FileCheck -check-prefix=HSAOPT -check-prefix=OPT %s
 ; RUN: opt -S -mtriple=amdgcn-unknown-unknown -mcpu=kaveri -amdgpu-promote-alloca < %s | FileCheck -check-prefix=NOHSAOPT -check-prefix=OPT %s
 
-declare i32 @llvm.amdgcn.workitem.id.x() nounwind readnone
-
 ; HSAOPT: @mova_same_clause.stack = internal unnamed_addr addrspace(3) global [256 x [5 x i32]] undef, align 4
 ; HSAOPT: @high_alignment.stack = internal unnamed_addr addrspace(3) global [256 x [8 x i32]] undef, align 16
 
@@ -264,28 +263,6 @@ entry:
 
 }
 
-; Make sure we don't overwrite workitem information with private memory
-
-; FUNC-LABEL: {{^}}work_item_info:
-; R600-NOT: MOV T0.X
-; Additional check in case the move ends up in the last slot
-; R600-NOT: MOV * TO.X
-
-; SI-NOT: v_mov_b32_e{{(32|64)}} v0
-define void @work_item_info(i32 addrspace(1)* %out, i32 %in) {
-entry:
-  %0 = alloca [2 x i32]
-  %1 = getelementptr [2 x i32], [2 x i32]* %0, i32 0, i32 0
-  %2 = getelementptr [2 x i32], [2 x i32]* %0, i32 0, i32 1
-  store i32 0, i32* %1
-  store i32 1, i32* %2
-  %3 = getelementptr [2 x i32], [2 x i32]* %0, i32 0, i32 %in
-  %4 = load i32, i32* %3
-  %5 = call i32 @llvm.amdgcn.workitem.id.x()
-  %6 = add i32 %4, %5
-  store i32 %6, i32 addrspace(1)* %out
-  ret void
-}
 
 ; Test that two stack objects are not stored in the same register
 ; The second stack object should be in T3.X
diff --git a/llvm/test/CodeGen/AMDGPU/r600.private-memory.ll b/llvm/test/CodeGen/AMDGPU/r600.private-memory.ll
new file mode 100644
index 00000000000..f406c160cbb
--- /dev/null
+++ b/llvm/test/CodeGen/AMDGPU/r600.private-memory.ll
@@ -0,0 +1,26 @@
+; RUN: llc -march=r600 -mcpu=cypress < %s | FileCheck %s -check-prefix=R600 -check-prefix=FUNC
+
+declare i32 @llvm.r600.read.tidig.x() nounwind readnone
+
+
+; Make sure we don't overwrite workitem information with private memory
+
+; FUNC-LABEL: {{^}}work_item_info:
+; R600-NOT: MOV T0.X
+; Additional check in case the move ends up in the last slot
+; R600-NOT: MOV * TO.X
+
+define void @work_item_info(i32 addrspace(1)* %out, i32 %in) {
+entry:
+  %0 = alloca [2 x i32]
+  %1 = getelementptr [2 x i32], [2 x i32]* %0, i32 0, i32 0
+  %2 = getelementptr [2 x i32], [2 x i32]* %0, i32 0, i32 1
+  store i32 0, i32* %1
+  store i32 1, i32* %2
+  %3 = getelementptr [2 x i32], [2 x i32]* %0, i32 0, i32 %in
+  %4 = load i32, i32* %3
+  %5 = call i32 @llvm.r600.read.tidig.x()
+  %6 = add i32 %4, %5
+  store i32 %6, i32 addrspace(1)* %out
+  ret void
+}