From 23663b84b909af8bf69c44d6678f151dc0599439 Mon Sep 17 00:00:00 2001
From: Matt Arsenault <Matthew.Arsenault@amd.com>
Date: Sat, 26 Sep 2015 02:25:45 +0000
Subject: [PATCH] AMDGPU: Fix sched model for VOP2b instructions

Trying to use the version with the explicit output operand
would complain because of the missing WriteSALU. I'm not sure
why it doesn't complain about this with the implicit VCC def.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@248646 91177308-0d34-0410-b5e6-96231b3b80d8
---
 lib/Target/AMDGPU/SIInstrInfo.td                   | 13 +++++++------
 test/CodeGen/AMDGPU/si-triv-disjoint-mem-access.ll |  5 +++--
 2 files changed, 10 insertions(+), 8 deletions(-)

diff --git a/lib/Target/AMDGPU/SIInstrInfo.td b/lib/Target/AMDGPU/SIInstrInfo.td
index 0a73419dbbe..dcc44b82627 100644
--- a/lib/Target/AMDGPU/SIInstrInfo.td
+++ b/lib/Target/AMDGPU/SIInstrInfo.td
@@ -1613,14 +1613,15 @@ multiclass VOP2b_Helper <vop2 op, string opName, dag outs32, dag outs64,
                          dag ins32, string asm32, list<dag> pat32,
                          dag ins64, string asm64, list<dag> pat64,
                          string revOp, bit HasMods, bit useSGPRInput> {
+  let SchedRW = [Write32Bit, WriteSALU] in {
+    let Uses = !if(useSGPRInput, [VCC, EXEC], [EXEC]), Defs = [VCC] in {
+      defm _e32 : VOP2_m <op, outs32, ins32, asm32, pat32, opName, revOp>;
+    }
 
-  let Uses = !if(useSGPRInput, [VCC, EXEC], [EXEC]), Defs = [VCC] in {
-    defm _e32 : VOP2_m <op, outs32, ins32, asm32, pat32, opName, revOp>;
+    defm _e64 : VOP3b_2_3_m <op,
+      outs64, ins64, opName#asm64, pat64, opName, revOp, HasMods, useSGPRInput
+    >;
   }
-
-  defm _e64 : VOP3b_2_3_m <op,
-    outs64, ins64, opName#asm64, pat64, opName, revOp, HasMods, useSGPRInput
-  >;
 }
 
 multiclass VOP2bInst <vop2 op, string opName, VOPProfile P,
diff --git a/test/CodeGen/AMDGPU/si-triv-disjoint-mem-access.ll b/test/CodeGen/AMDGPU/si-triv-disjoint-mem-access.ll
index 5a6129aaa3f..35e9ff83463 100644
--- a/test/CodeGen/AMDGPU/si-triv-disjoint-mem-access.ll
+++ b/test/CodeGen/AMDGPU/si-triv-disjoint-mem-access.ll
@@ -155,9 +155,9 @@ define void @reorder_global_load_local_store_global_load(i32 addrspace(1)* %out,
 }
 
 ; FUNC-LABEL: @reorder_local_offsets
-; CI: ds_write_b32 {{v[0-9]+}}, {{v[0-9]+}} offset:12
 ; CI: ds_read_b32 {{v[0-9]+}}, {{v[0-9]+}} offset:400
 ; CI: ds_read_b32 {{v[0-9]+}}, {{v[0-9]+}} offset:404
+; CI: ds_write_b32 {{v[0-9]+}}, {{v[0-9]+}} offset:12
 ; CI: ds_write_b32 {{v[0-9]+}}, {{v[0-9]+}} offset:400
 ; CI: ds_write_b32 {{v[0-9]+}}, {{v[0-9]+}} offset:404
 ; CI: buffer_store_dword
@@ -181,9 +181,10 @@ define void @reorder_local_offsets(i32 addrspace(1)* nocapture %out, i32 addrspa
 }
 
 ; FUNC-LABEL: @reorder_global_offsets
-; CI: buffer_store_dword {{v[0-9]+}}, {{s\[[0-9]+:[0-9]+\]}}, 0 offset:12
 ; CI: buffer_load_dword {{v[0-9]+}}, {{s\[[0-9]+:[0-9]+\]}}, 0 offset:400
 ; CI: buffer_load_dword {{v[0-9]+}}, {{s\[[0-9]+:[0-9]+\]}}, 0 offset:404
+; CI: buffer_store_dword {{v[0-9]+}}, {{s\[[0-9]+:[0-9]+\]}}, 0 offset:12
+; CI: buffer_load_dword {{v[0-9]+}}, {{s\[[0-9]+:[0-9]+\]}}, 0 offset:12
 ; CI: buffer_store_dword {{v[0-9]+}}, {{s\[[0-9]+:[0-9]+\]}}, 0 offset:400
 ; CI: buffer_store_dword {{v[0-9]+}}, {{s\[[0-9]+:[0-9]+\]}}, 0 offset:404
 ; CI: buffer_store_dword
-- 
2.34.1