[llvm] [Offload] Use the kernel argument size directly in AMDGPU offloading (PR #94667)

[Joseph Huber via llvm-commits]

https://github.com/jhuber6 created https://github.com/llvm/llvm-project/pull/94667

Summary:
The old COV3 implementation of HSA used to omit the implicit arguments
from the kernel argument size. For COV4 and COV5 this is no longer the
case so we can simply use the size reported from the symbol information.


>From 9ea7fc92a057de998f399b992577c0084eb57f76 Mon Sep 17 00:00:00 2001
From: Joseph Huber <huberjn at outlook.com>
Date: Thu, 6 Jun 2024 14:28:36 -0500
Subject: [PATCH] [Offload] Use the kernel argument size directly in AMDGPU
 offloading

Summary:
The old COV3 implementation of HSA used to omit the implicit arguments
from the kernel argument size. For COV4 and COV5 this is no longer the
case so we can simply use the size reported from the symbol information.
---
 offload/plugins-nextgen/amdgpu/src/rtl.cpp | 8 +-------
 1 file changed, 1 insertion(+), 7 deletions(-)

diff --git a/offload/plugins-nextgen/amdgpu/src/rtl.cpp b/offload/plugins-nextgen/amdgpu/src/rtl.cpp
index f088d5d1685df..663cfdc5fdf01 100644
--- a/offload/plugins-nextgen/amdgpu/src/rtl.cpp
+++ b/offload/plugins-nextgen/amdgpu/src/rtl.cpp
@@ -3272,19 +3272,13 @@ Error AMDGPUKernelTy::launchImpl(GenericDeviceTy &GenericDevice,
   if (ArgsSize < KernelArgsSize)
     return Plugin::error("Mismatch of kernel arguments size");
 
-  // The args size reported by HSA may or may not contain the implicit args.
-  // For now, assume that HSA does not consider the implicit arguments when
-  // reporting the arguments of a kernel. In the worst case, we can waste
-  // 56 bytes per allocation.
-  uint32_t AllArgsSize = KernelArgsSize + ImplicitArgsSize;
-
   AMDGPUPluginTy &AMDGPUPlugin =
       static_cast<AMDGPUPluginTy &>(GenericDevice.Plugin);
   AMDHostDeviceTy &HostDevice = AMDGPUPlugin.getHostDevice();
   AMDGPUMemoryManagerTy &ArgsMemoryManager = HostDevice.getArgsMemoryManager();
 
   void *AllArgs = nullptr;
-  if (auto Err = ArgsMemoryManager.allocate(AllArgsSize, &AllArgs))
+  if (auto Err = ArgsMemoryManager.allocate(ArgsSize, &AllArgs))
     return Err;
 
   // Account for user requested dynamic shared memory.