SWDEV-436859 - Enable pitch for COPY_HOST_PTR

Original logic didn't use pitch because, abstraction layer had
a sysmem copy without pitch. Since extra sysmem copy was
disabled, the code has to accept pitch values from the app.

Change-Id: Ia9fba7b33ddff4e9109b4e63d0d6afa52f501c8f


[ROCm/clr commit: fb3dfcf889]

projects/clr/rocclr/device/pal/paldevice.cpp

@@ -1789,9 +1789,11 @@ pal::Memory* Device::createImage(amd::Memory& owner, bool directAccess) const {
         owner.setHostMem(nullptr);
       } else {
         amd::Coord3D origin(0, 0, 0);
-        static const bool Entire = true;
-        if (xferMgr().writeImage(owner.getHostMem(), *gpuImage, origin, image.getRegion(), 0, 0,
-                                 Entire)) {
+        // Copy data with the original pitch values, since runtime doesn't perform
+        // extra sysmem allocation for one device
+        constexpr bool kEntire = true;
+        if (xferMgr().writeImage(owner.getHostMem(), *gpuImage, origin, image.getRegion(),
+            image.getRowPitch(), image.getSlicePitch(), kEntire)) {
           // Clear CHP memory
           owner.setHostMem(nullptr);
         }

projects/clr/rocclr/device/rocm/rocdevice.cpp

@@ -2086,8 +2086,11 @@ device::Memory* Device::createMemory(amd::Memory& owner) const {
 
     imageView->replaceDeviceMemory(this, devImageView);
 
+    // Copy data with the original pitch values, since runtime doesn't perform
+    // extra sysmem allocation for one device
+    const auto image = owner.asImage();
     result = xferMgr().writeImage(owner.getHostMem(), *devImageView, amd::Coord3D(0, 0, 0),
-                                  imageView->getRegion(), 0, 0, true);
+        imageView->getRegion(), image->getRowPitch(), image->getSlicePitch(), true);
 
     // Release host memory, since runtime copied data
     owner.setHostMem(nullptr);