From a396a70c1decc68b2ba9cfba2dd116fa29857168 Mon Sep 17 00:00:00 2001
From: Matt Arsenault <Matthew.Arsenault@amd.com>
Date: Wed, 11 Jun 2014 18:08:45 +0000
Subject: [PATCH] R600/SI: Use LDS atomic inc / dec

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@210677 91177308-0d34-0410-b5e6-96231b3b80d8
---
 lib/Target/R600/SIInstructions.td  | 16 +++++++++++++
 test/CodeGen/R600/local-atomics.ll | 38 ++++++++++++++++++++++++++++++
 2 files changed, 54 insertions(+)

diff --git a/lib/Target/R600/SIInstructions.td b/lib/Target/R600/SIInstructions.td
index a00a8ba9471..f6dc07fbd96 100644
--- a/lib/Target/R600/SIInstructions.td
+++ b/lib/Target/R600/SIInstructions.td
@@ -2222,6 +2222,22 @@ multiclass DSAtomicRetPat<DS inst, ValueType vt, PatFrag frag> {
   >;
 }
 
+// Special case of DSAtomicRetPat for add / sub 1 -> inc / dec
+multiclass DSAtomicIncRetPat<DS inst, ValueType vt, PatFrag frag> {
+  def : Pat <
+    (frag (add i32:$ptr, (i32 IMM16bit:$offset)), (vt 1)),
+    (inst (i1 0), $ptr, (as_i16imm $offset))
+  >;
+
+  def : Pat <
+    (frag i32:$ptr, (vt 1)),
+    (inst 0, $ptr, 0)
+  >;
+}
+
+defm : DSAtomicIncRetPat<DS_INC_RTN_U32, i32, atomic_load_add_local>;
+defm : DSAtomicIncRetPat<DS_DEC_RTN_U32, i32, atomic_load_sub_local>;
+
 defm : DSAtomicRetPat<DS_WRXCHG_RTN_B32, i32, atomic_swap_local>;
 defm : DSAtomicRetPat<DS_ADD_RTN_U32, i32, atomic_load_add_local>;
 defm : DSAtomicRetPat<DS_SUB_RTN_U32, i32, atomic_load_sub_local>;
diff --git a/test/CodeGen/R600/local-atomics.ll b/test/CodeGen/R600/local-atomics.ll
index ef440f994fc..fb70ddf0692 100644
--- a/test/CodeGen/R600/local-atomics.ll
+++ b/test/CodeGen/R600/local-atomics.ll
@@ -47,6 +47,25 @@ define void @lds_atomic_add_ret_i32_offset(i32 addrspace(1)* %out, i32 addrspace
   ret void
 }
 
+; FUNC-LABEL: @lds_atomic_inc_ret_i32:
+; SI: DS_INC_RTN_U32
+; SI: S_ENDPGM
+define void @lds_atomic_inc_ret_i32(i32 addrspace(1)* %out, i32 addrspace(3)* %ptr) nounwind {
+  %result = atomicrmw add i32 addrspace(3)* %ptr, i32 1 seq_cst
+  store i32 %result, i32 addrspace(1)* %out, align 4
+  ret void
+}
+
+; FUNC-LABEL: @lds_atomic_inc_ret_i32_offset:
+; SI: DS_INC_RTN_U32 v{{[0-9]+}}, v{{[0-9]+}}, 0x10
+; SI: S_ENDPGM
+define void @lds_atomic_inc_ret_i32_offset(i32 addrspace(1)* %out, i32 addrspace(3)* %ptr) nounwind {
+  %gep = getelementptr i32 addrspace(3)* %ptr, i32 4
+  %result = atomicrmw add i32 addrspace(3)* %gep, i32 1 seq_cst
+  store i32 %result, i32 addrspace(1)* %out, align 4
+  ret void
+}
+
 ; FUNC-LABEL: @lds_atomic_sub_ret_i32:
 ; SI: DS_SUB_RTN_U32
 ; SI: S_ENDPGM
@@ -66,6 +85,25 @@ define void @lds_atomic_sub_ret_i32_offset(i32 addrspace(1)* %out, i32 addrspace
   ret void
 }
 
+; FUNC-LABEL: @lds_atomic_dec_ret_i32:
+; SI: DS_DEC_RTN_U32
+; SI: S_ENDPGM
+define void @lds_atomic_dec_ret_i32(i32 addrspace(1)* %out, i32 addrspace(3)* %ptr) nounwind {
+  %result = atomicrmw sub i32 addrspace(3)* %ptr, i32 1 seq_cst
+  store i32 %result, i32 addrspace(1)* %out, align 4
+  ret void
+}
+
+; FUNC-LABEL: @lds_atomic_dec_ret_i32_offset:
+; SI: DS_DEC_RTN_U32 v{{[0-9]+}}, v{{[0-9]+}}, 0x10
+; SI: S_ENDPGM
+define void @lds_atomic_dec_ret_i32_offset(i32 addrspace(1)* %out, i32 addrspace(3)* %ptr) nounwind {
+  %gep = getelementptr i32 addrspace(3)* %ptr, i32 4
+  %result = atomicrmw sub i32 addrspace(3)* %gep, i32 1 seq_cst
+  store i32 %result, i32 addrspace(1)* %out, align 4
+  ret void
+}
+
 ; FUNC-LABEL: @lds_atomic_and_ret_i32:
 ; SI: DS_AND_RTN_B32
 ; SI: S_ENDPGM