[Mlir-commits] [mlir] [MLIR] Pass hostShared flag in gpu.alloc op to runtime wrappers (PR #66401)

[llvmlistbot at llvm.org]

llvmbot wrote:


<!--LLVM PR SUMMARY COMMENT-->

@llvm/pr-subscribers-mlir
            
<details>
<summary>Changes</summary>
This PR is a breakdown of the big PR https://github.com/llvm/llvm-project/pull/65539 which enables intel gpu integration. In this PR we pass hostShared flag to runtime wrappers (required by SyclRuntimeWrappers which will come in subsequent PR) to indicate if the allocation is done on host shared gpu memory or device only memory.
--
Full diff: https://github.com/llvm/llvm-project/pull/66401.diff

5 Files Affected:

- (modified) mlir/lib/Conversion/GPUCommon/GPUToLLVMConversion.cpp (+10-5) 
- (modified) mlir/lib/ExecutionEngine/CudaRuntimeWrappers.cpp (+2-1) 
- (modified) mlir/lib/ExecutionEngine/RocmRuntimeWrappers.cpp (+2-1) 
- (modified) mlir/test/Conversion/GPUCommon/lower-alloc-to-gpu-runtime-calls.mlir (+2-1) 
- (modified) mlir/test/Conversion/GPUCommon/typed-pointers.mlir (+2-1) 


<pre>
diff --git a/mlir/lib/Conversion/GPUCommon/GPUToLLVMConversion.cpp b/mlir/lib/Conversion/GPUCommon/GPUToLLVMConversion.cpp
index eddf3e9a47d0bc8..428d5d1d4b0e944 100644
--- a/mlir/lib/Conversion/GPUCommon/GPUToLLVMConversion.cpp
+++ b/mlir/lib/Conversion/GPUCommon/GPUToLLVMConversion.cpp
@@ -167,7 +167,8 @@ class ConvertOpToGpuRuntimeCallPattern : public ConvertOpToLLVMPattern&lt;OpTy&gt; {
       &quot;mgpuMemAlloc&quot;,
       llvmPointerType /* void * */,
       {llvmIntPtrType /* intptr_t sizeBytes */,
-       llvmPointerType /* void *stream */}};
+       llvmPointerType /* void *stream */,
+       llvmInt64Type /* bool isHostShared */}};
   FunctionCallBuilder deallocCallBuilder = {
       &quot;mgpuMemFree&quot;,
       llvmVoidType,
@@ -786,9 +787,6 @@ LogicalResult ConvertHostUnregisterOpToGpuRuntimeCallPattern::matchAndRewrite(
 LogicalResult ConvertAllocOpToGpuRuntimeCallPattern::matchAndRewrite(
     gpu::AllocOp allocOp, OpAdaptor adaptor,
     ConversionPatternRewriter &amp;rewriter) const {
-  if (adaptor.getHostShared())
-    return rewriter.notifyMatchFailure(
-        allocOp, &quot;host_shared allocation is not supported&quot;);
 
   MemRefType memRefType = allocOp.getType();
 
@@ -799,6 +797,8 @@ LogicalResult ConvertAllocOpToGpuRuntimeCallPattern::matchAndRewrite(
 
   auto loc = allocOp.getLoc();
 
+  bool isShared = allocOp.getHostShared();
+
   // Get shape of the memref as values: static sizes are constant
   // values and dynamic sizes are passed to &#x27;alloc&#x27; as operands.
   SmallVector&lt;Value, 4&gt; shape;
@@ -811,8 +811,13 @@ LogicalResult ConvertAllocOpToGpuRuntimeCallPattern::matchAndRewrite(
   // descriptor.
   Type elementPtrType = this-&gt;getElementPtrType(memRefType);
   auto stream = adaptor.getAsyncDependencies().front();
+
+  auto isHostShared = rewriter.create&lt;mlir::LLVM::ConstantOp&gt;(
+      loc, llvmInt64Type, rewriter.getI64IntegerAttr(isShared));
+
   Value allocatedPtr =
-      allocCallBuilder.create(loc, rewriter, {sizeBytes, stream}).getResult();
+      allocCallBuilder.create(loc, rewriter, {sizeBytes, stream, isHostShared})
+          .getResult();
   if (!getTypeConverter()-&gt;useOpaquePointers())
     allocatedPtr =
         rewriter.create&lt;LLVM::BitcastOp&gt;(loc, elementPtrType, allocatedPtr);
diff --git a/mlir/lib/ExecutionEngine/CudaRuntimeWrappers.cpp b/mlir/lib/ExecutionEngine/CudaRuntimeWrappers.cpp
index 1dba677ebe66365..a0172f85a67a5c0 100644
--- a/mlir/lib/ExecutionEngine/CudaRuntimeWrappers.cpp
+++ b/mlir/lib/ExecutionEngine/CudaRuntimeWrappers.cpp
@@ -210,7 +210,8 @@ extern MLIR_CUDA_WRAPPERS_EXPORT &quot;C&quot; void mgpuEventRecord(CUevent event,
   CUDA_REPORT_IF_ERROR(cuEventRecord(event, stream));
 }
 
-extern &quot;C&quot; void *mgpuMemAlloc(uint64_t sizeBytes, CUstream /*stream*/) {
+extern &quot;C&quot; void *mgpuMemAlloc(uint64_t sizeBytes, CUstream /*stream*/,
+                              bool /*isHostShared*/) {
   ScopedContext scopedContext;
   CUdeviceptr ptr;
   CUDA_REPORT_IF_ERROR(cuMemAlloc(&amp;ptr, sizeBytes));
diff --git a/mlir/lib/ExecutionEngine/RocmRuntimeWrappers.cpp b/mlir/lib/ExecutionEngine/RocmRuntimeWrappers.cpp
index bd3868a8e196f6f..292159536f5522f 100644
--- a/mlir/lib/ExecutionEngine/RocmRuntimeWrappers.cpp
+++ b/mlir/lib/ExecutionEngine/RocmRuntimeWrappers.cpp
@@ -99,7 +99,8 @@ extern &quot;C&quot; void mgpuEventRecord(hipEvent_t event, hipStream_t stream) {
   HIP_REPORT_IF_ERROR(hipEventRecord(event, stream));
 }
 
-extern &quot;C&quot; void *mgpuMemAlloc(uint64_t sizeBytes, hipStream_t /*stream*/) {
+extern &quot;C&quot; void *mgpuMemAlloc(uint64_t sizeBytes, hipStream_t /*stream*/,
+                              bool /*isHostShared*/) {
   void *ptr;
   HIP_REPORT_IF_ERROR(hipMalloc(&amp;ptr, sizeBytes));
   return ptr;
diff --git a/mlir/test/Conversion/GPUCommon/lower-alloc-to-gpu-runtime-calls.mlir b/mlir/test/Conversion/GPUCommon/lower-alloc-to-gpu-runtime-calls.mlir
index 2506c6ceb990ef5..f365dcb02daf4c2 100644
--- a/mlir/test/Conversion/GPUCommon/lower-alloc-to-gpu-runtime-calls.mlir
+++ b/mlir/test/Conversion/GPUCommon/lower-alloc-to-gpu-runtime-calls.mlir
@@ -8,7 +8,8 @@ module attributes {gpu.container_module} {
     %0 = gpu.wait async
     // CHECK: %[[gep:.*]] = llvm.getelementptr {{.*}}[%[[size]]]
     // CHECK: %[[size_bytes:.*]] = llvm.ptrtoint %[[gep]]
-    // CHECK: llvm.call @mgpuMemAlloc(%[[size_bytes]], %[[stream]])
+    // CHECK: %[[isHostShared:.*]] = llvm.mlir.constant 
+    // CHECK: llvm.call @mgpuMemAlloc(%[[size_bytes]], %[[stream]], %[[isHostShared]])
     %1, %2 = gpu.alloc async [%0] (%size) : memref&lt;?xf32&gt;
     // CHECK: %[[float_ptr:.*]] = llvm.extractvalue {{.*}}[0]
     // CHECK: llvm.call @mgpuMemFree(%[[float_ptr]], %[[stream]])
diff --git a/mlir/test/Conversion/GPUCommon/typed-pointers.mlir b/mlir/test/Conversion/GPUCommon/typed-pointers.mlir
index 2fa6c854c567819..e27162c7dbc1902 100644
--- a/mlir/test/Conversion/GPUCommon/typed-pointers.mlir
+++ b/mlir/test/Conversion/GPUCommon/typed-pointers.mlir
@@ -8,7 +8,8 @@ module attributes {gpu.container_module} {
     %0 = gpu.wait async
     // CHECK: %[[gep:.*]] = llvm.getelementptr {{.*}}[%[[size]]]
     // CHECK: %[[size_bytes:.*]] = llvm.ptrtoint %[[gep]]
-    // CHECK: llvm.call @mgpuMemAlloc(%[[size_bytes]], %[[stream]])
+    // CHECK: %[[isHostShared:.*]] = llvm.mlir.constant
+    // CHECK: llvm.call @mgpuMemAlloc(%[[size_bytes]], %[[stream]], %[[isHostShared]])
     %1, %2 = gpu.alloc async [%0] (%size) : memref&lt;?xf32&gt;
     // CHECK: %[[float_ptr:.*]] = llvm.extractvalue {{.*}}[0]
     // CHECK: %[[void_ptr:.*]] = llvm.bitcast %[[float_ptr]]
</pre>
</details>


https://github.com/llvm/llvm-project/pull/66401