rocr: Quiet some ROCr compile warnings

These are mostly AIE related, but there are a couple of others.

Change-Id: I549e004772160ca282d4c94dc9d94dd2ccae8b1c


[ROCm/ROCR-Runtime commit: 08699069d6]

projects/rocr-runtime/runtime/hsa-runtime/core/inc/amd_aie_agent.h

@@ -95,13 +95,13 @@ public:
 
   // AIE agent methods.
   /// @brief Get the number of columns on this AIE agent.
-  int GetNumCols() const { return num_cols_; }
-  void SetNumCols(int num_cols) { num_cols_ = num_cols; }
+  uint32_t GetNumCols() const { return num_cols_; }
+  void SetNumCols(uint32_t num_cols) { num_cols_ = num_cols; }
   /// @brief Get the number of core tile rows on this AIE agent.
-  int GetNumCoreRows() const { return num_core_rows_; }
-  void SetNumCoreRows(int num_core_rows) { num_core_rows_ = num_core_rows; }
+  uint32_t GetNumCoreRows() const { return num_core_rows_; }
+  void SetNumCoreRows(uint32_t num_core_rows) { num_core_rows_ = num_core_rows; }
   /// @brief Get the number of core tiles on this AIE agent.
-  int GetNumCores() const { return num_cols_ * num_core_rows_; }
+  uint32_t GetNumCores() const { return num_cols_ * num_core_rows_; }
 
 private:
   /// @brief Query the driver to get the region list owned by this agent.
@@ -123,10 +123,10 @@ private:
   const uint32_t max_queues_ = 1;
 
   /// @brief Number of columns in the AIE array.
-  int num_cols_ = 0;
+  uint32_t num_cols_ = 0;
   /// @brief Number of rows of core tiles in the AIE array. Not all rows in a
   /// column are cores. Some can be memory or shim tiles.
-  int num_core_rows_ = 0;
+  uint32_t num_core_rows_ = 0;
 };
 
 } // namespace AMD

projects/rocr-runtime/runtime/hsa-runtime/core/inc/amd_aie_aql_queue.h

@@ -116,7 +116,7 @@ public:
                   hsa_fence_scope_t releaseFence = HSA_FENCE_SCOPE_NONE,
                   hsa_signal_t *signal = NULL) override;
 
-  uint32_t queue_id_ = INVALID_QUEUEID;
+  HSA_QUEUEID queue_id_ = INVALID_QUEUEID;
   /// @brief ID of AIE device on which this queue has been mapped.
   uint32_t node_id_ = std::numeric_limits<uint32_t>::max();
   /// @brief Queue size in bytes.

projects/rocr-runtime/runtime/hsa-runtime/core/inc/amd_aql_queue.h

@@ -346,8 +346,6 @@ class AqlQueue : public core::Queue, private core::LocalSignal, public core::Doo
     static KernelMutex queue_lock_;
     return queue_lock_;
 }
-  // Async scratch single limit - may be modified after init
-  size_t async_scratch_single_limit_;
 
   static __forceinline int& rtti_id() {
     static int rtti_id_ = 0;

projects/rocr-runtime/runtime/hsa-runtime/core/inc/amd_gpu_agent.h

@@ -507,14 +507,14 @@ class GpuAgent : public GpuAgentInt {
   hsa_status_t EnableDmaProfiling(bool enable) override;
 
   hsa_status_t PcSamplingIterateConfig(hsa_ven_amd_pcs_iterate_configuration_callback_t cb,
-                                       void* cb_data);
-  hsa_status_t PcSamplingCreate(pcs::PcsRuntime::PcSamplingSession& session);
+                                       void* cb_data) override;
+  hsa_status_t PcSamplingCreate(pcs::PcsRuntime::PcSamplingSession& session) override;
   hsa_status_t PcSamplingCreateFromId(HsaPcSamplingTraceId pcsId,
-                                      pcs::PcsRuntime::PcSamplingSession& session);
-  hsa_status_t PcSamplingDestroy(pcs::PcsRuntime::PcSamplingSession& session);
-  hsa_status_t PcSamplingStart(pcs::PcsRuntime::PcSamplingSession& session);
-  hsa_status_t PcSamplingStop(pcs::PcsRuntime::PcSamplingSession& session);
-  hsa_status_t PcSamplingFlush(pcs::PcsRuntime::PcSamplingSession& session);
+                            pcs::PcsRuntime::PcSamplingSession& session) override;
+  hsa_status_t PcSamplingDestroy(pcs::PcsRuntime::PcSamplingSession& session) override;
+  hsa_status_t PcSamplingStart(pcs::PcsRuntime::PcSamplingSession& session) override;
+  hsa_status_t PcSamplingStop(pcs::PcsRuntime::PcSamplingSession& session) override;
+  hsa_status_t PcSamplingFlush(pcs::PcsRuntime::PcSamplingSession& session) override;
   hsa_status_t PcSamplingFlushHostTrapDeviceBuffers(pcs::PcsRuntime::PcSamplingSession& session);
 
   static void PcSamplingThreadRun(void* agent);

projects/rocr-runtime/runtime/hsa-runtime/core/inc/hsa_amd_tool_int.hpp

@@ -4,7 +4,8 @@
 #include "inc/hsa_amd_tool.h"
 #include "runtime.h"
 
-namespace rocr::AMD::tool {
+// namespace rocr::AMD::tool {  // C++17
+namespace rocr { namespace AMD { namespace tool {
 
 using scratch_alloc_flag = hsa_amd_event_scratch_alloc_flag_t;
 
@@ -135,7 +136,9 @@ __forceinline void notify_event_scratch_async_reclaim_end(const hsa_queue_t* que
       hsa_amd_tool_event_t{.scratch_async_reclaim_end = &event});
 }
 
-
-}  // namespace rocr::AMD::tool
+// }  // namespace rocr::AMD::tool
+}  // namespace rocr
+}  // namespace AMD
+}  // namespace tool
 
 #endif

projects/rocr-runtime/runtime/hsa-runtime/core/runtime/amd_blit_sdma.cpp

@@ -559,7 +559,7 @@ BlitSdma<RingIndexTy, HwIndexMonotonic, SizeToCountOffset, useGCR>::SubmitCopyRe
   const bool isGFX12Plus = (agent_->isa()->GetMajorVersion() >= 12);
 
   // Common and GFX12 packet must match in size to use same code for vector/append.
-  static_assert(sizeof(SDMA_PKT_COPY_LINEAR_RECT) == sizeof(SDMA_PKT_COPY_LINEAR_RECT_GFX12));
+  static_assert(sizeof(SDMA_PKT_COPY_LINEAR_RECT) == sizeof(SDMA_PKT_COPY_LINEAR_RECT_GFX12), "");
 
   const uint max_pitch = 1 << (isGFX12Plus ? SDMA_PKT_COPY_LINEAR_RECT_GFX12::pitch_bits : SDMA_PKT_COPY_LINEAR_RECT::pitch_bits);