SWDEV-469825 - Modified the kernel argument readback to use a pointer to volatile

This change modifies the readback mechanism to use a pointer to volatile
instead of a volatile pointer. This ensures that the compiler does not
optimize away the read operation.

Change-Id: I79ff925d615aa8cc4f950e8ff4b7e608fcb179a4

hipamd/src/hip_graph_internal.cpp

@@ -469,11 +469,11 @@ hipError_t GraphExec::CaptureAQLPackets() {
                  kernarg_graph_.back().kernarg_pool_addr_ != 0) {
         address dev_ptr = kernarg_graph_.back().kernarg_pool_addr_ +
                           kernarg_graph_.back().kernarg_pool_size_;
-        auto kSentinel = *reinterpret_cast<volatile address>(dev_ptr - 1);
+        auto kSentinel = *reinterpret_cast<volatile unsigned char*>(dev_ptr - 1);
         _mm_sfence();
         *(dev_ptr - 1) = kSentinel;
         _mm_mfence();
-        kSentinel = *reinterpret_cast<volatile address>(dev_ptr - 1);
+        kSentinel = *reinterpret_cast<volatile unsigned char*>(dev_ptr - 1);
       }
     }
   }

rocclr/device/rocm/rocvirtual.cpp

@@ -3259,7 +3259,8 @@ bool VirtualGPU::submitKernelInternal(const amd::NDRangeContainer& sizes,
           _mm_sfence();
           *(argBuffer + argSize - 1) = *(parameters + argSize - 1);
           _mm_mfence();
-          auto kSentinel = *reinterpret_cast<volatile address>(argBuffer + argSize - 1);
+          auto kSentinel = *reinterpret_cast<volatile unsigned char*>(
+              argBuffer + argSize - 1);
         }
       }
     }