P4 to Git Change 1152825 by bdhanase@bala_workpc_ocl on 2015/05/20 16:19:13

EPR #397491 - According to HSA-Finalizer-ADD, for GPUVM32 private_segment_aperture_base_hi and group_segment_aperture_base_hi should be equal to the 32 bits of the 32 bit private and group segment flat address aperture.

	Reviewed by: German

Affected files ...

... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gpuvirtual.cpp#362 edit


[ROCm/clr commit: 9d37ac1fc8]

projects/clr/rocclr/runtime/device/gpu/gpuvirtual.cpp

@@ -618,10 +618,11 @@ VirtualGPU::allocHsaQueueMem()
     }
     memset(queue, 0, sizeof(amd_queue_t));
     // Provide private and local heap addresses
+    const static uint addressShift = LP64_SWITCH(0, 32);
     queue->private_segment_aperture_base_hi =
-        static_cast<uint32>(dev().gslCtx()->getPrivateApertureBase()>>32);
+        static_cast<uint32>(dev().gslCtx()->getPrivateApertureBase()>>addressShift);
     queue->group_segment_aperture_base_hi =
-        static_cast<uint32>(dev().gslCtx()->getSharedApertureBase()>>32);
+        static_cast<uint32>(dev().gslCtx()->getSharedApertureBase()>>addressShift);
     hsaQueueMem_->unmap(NULL);
     return true;
 }