[llvm] r367881 - AMDGPU/LoadStoreOptimizer: Set the correct offset whem merging MMOs

[Tom Stellard via llvm-commits]

Author: tstellar
Date: Mon Aug  5 09:08:44 2019
New Revision: 367881

URL: http://llvm.org/viewvc/llvm-project?rev=367881&view=rev
Log:
AMDGPU/LoadStoreOptimizer: Set the correct offset whem merging MMOs

Summary:
This is a follow up to r367237.  MachineFunction::getMachineMemOperand()
adds the offset parameter to the existing offset instead of resetting it.
So we need to reset the offset to the correct value after calling this
function.

Reviewers: arsenm

Subscribers: kzhuravl, jvesely, wdng, nhaehnle, yaxunl, dstuttard, tpr, t-tye, hiraditya, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D65557

Modified:
    llvm/trunk/lib/Target/AMDGPU/SILoadStoreOptimizer.cpp
    llvm/trunk/test/CodeGen/AMDGPU/merge-load-store.mir

Modified: llvm/trunk/lib/Target/AMDGPU/SILoadStoreOptimizer.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/AMDGPU/SILoadStoreOptimizer.cpp?rev=367881&r1=367880&r2=367881&view=diff
==============================================================================
--- llvm/trunk/lib/Target/AMDGPU/SILoadStoreOptimizer.cpp (original)
+++ llvm/trunk/lib/Target/AMDGPU/SILoadStoreOptimizer.cpp Mon Aug  5 09:08:44 2019
@@ -313,7 +313,12 @@ static MachineMemOperand *combineKnownAd
                                                    const MachineMemOperand *B) {
   unsigned MinOffset = std::min(A->getOffset(), B->getOffset());
   unsigned Size = A->getSize() + B->getSize();
-  return MF.getMachineMemOperand(A, MinOffset, Size);
+  // This function adds the offset parameter to the existing offset for A,
+  // so we pass 0 here as the offset and then manually set it to the correct
+  // value after the call.
+  MachineMemOperand *MMO = MF.getMachineMemOperand(A, 0, Size);
+  MMO->setOffset(MinOffset);
+  return MMO;
 }
 
 bool SILoadStoreOptimizer::offsetsCanBeCombined(CombineInfo &CI) {

Modified: llvm/trunk/test/CodeGen/AMDGPU/merge-load-store.mir
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/AMDGPU/merge-load-store.mir?rev=367881&r1=367880&r2=367881&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/AMDGPU/merge-load-store.mir (original)
+++ llvm/trunk/test/CodeGen/AMDGPU/merge-load-store.mir Mon Aug  5 09:08:44 2019
@@ -65,7 +65,8 @@
 
   attributes #0 = { convergent nounwind }
 
-  define amdgpu_kernel void @merge_mmos() { ret void }
+  define amdgpu_kernel void @merge_mmos(i32 addrspace(1)* %ptr_addr1) { ret void }
+
 ...
 ---
 name:            mem_dependency
@@ -170,6 +171,8 @@ body:             |
 # CHECK: S_BUFFER_LOAD_DWORDX2_IMM %0, 0, 0, 0 :: (dereferenceable invariant load 8, align 4)
 # CHECK: BUFFER_LOAD_DWORDX2_OFFSET %0, 0, 0, 0, 0, 0, 0, implicit $exec :: (dereferenceable load 8, align 4)
 # CHECK: BUFFER_STORE_DWORDX2_OFFSET_exact killed %{{[0-9]+}}, %0, 0, 0, 0, 0, 0, 0, implicit $exec :: (dereferenceable store 8, align 4)
+# CHECK: BUFFER_LOAD_DWORDX2_OFFSET %0, 0, 64, 0, 0, 0, 0, implicit $exec :: (dereferenceable load 8 from %ir.ptr_addr1 + 64, align 4
+# CHECK: BUFFER_STORE_DWORDX2_OFFSET_exact killed %{{[0-9]+}}, %0, 0, 64, 0, 0, 0, 0, implicit $exec :: (dereferenceable store 8 into %ir.ptr_addr1 + 64, align 4
 name: merge_mmos
 tracksRegLiveness: true
 body: |
@@ -183,6 +186,11 @@ body: |
     %4:vgpr_32 = BUFFER_LOAD_DWORD_OFFSET %0, 0, 4, 0, 0, 0, 0, implicit $exec :: (dereferenceable load 4)
     BUFFER_STORE_DWORD_OFFSET_exact %3, %0, 0, 0, 0, 0, 0, 0, implicit $exec :: (dereferenceable store 4)
     BUFFER_STORE_DWORD_OFFSET_exact %4, %0, 0, 4, 0, 0, 0, 0, implicit $exec :: (dereferenceable store 4)
+    %5:vgpr_32 = BUFFER_LOAD_DWORD_OFFSET %0, 0, 64, 0, 0, 0, 0, implicit $exec :: (dereferenceable load 4 from %ir.ptr_addr1 + 64)
+    %6:vgpr_32 = BUFFER_LOAD_DWORD_OFFSET %0, 0, 68, 0, 0, 0, 0, implicit $exec :: (dereferenceable load 4 from %ir.ptr_addr1 + 68)
+    BUFFER_STORE_DWORD_OFFSET_exact %5, %0, 0, 64, 0, 0, 0, 0, implicit $exec :: (dereferenceable store 4 into %ir.ptr_addr1 + 64)
+    BUFFER_STORE_DWORD_OFFSET_exact %6, %0, 0, 68, 0, 0, 0, 0, implicit $exec :: (dereferenceable store 4 into %ir.ptr_addr1 + 68)
+
     S_ENDPGM 0
 
 ...