ASoC: Intel: Skylake: Fix CLDMA buffer wrap case

When downloading the firmware/module, if the ring buffer boundary
is reached, we need to wrap to the zeroth position. On next copy
we need to copy till end of buffer and the remaining buffer needs
to be copied from zeroth position.

In this case copy was not handled correctly when wrap condition
is reached which caused invalid data to be copied resulting in
invalid hash failure.

This patch fixes the issue by handling copy at the boundary
condition correctly.
Signed-off-by: NJeeja KP <jeeja.kp@intel.com>
Signed-off-by: NVinod Koul <vinod.koul@intel.com>
Signed-off-by: NMark Brown <broonie@kernel.org>

sound/soc/intel/skylake/skl-sst-cldma.c

@@ -180,6 +180,21 @@ static void skl_cldma_fill_buffer(struct sst_dsp *ctx, unsigned int size,
 			ctx->cl_dev.dma_buffer_offset, trigger);
 	dev_dbg(ctx->dev, "spib position: %d\n", ctx->cl_dev.curr_spib_pos);
 
+	/*
+	 * Check if the size exceeds buffer boundary. If it exceeds
+	 * max_buffer size, then copy till buffer size and then copy
+	 * remaining buffer from the start of ring buffer.
+	 */
+	if (ctx->cl_dev.dma_buffer_offset + size > ctx->cl_dev.bufsize) {
+		unsigned int size_b = ctx->cl_dev.bufsize -
+					ctx->cl_dev.dma_buffer_offset;
+		memcpy(ctx->cl_dev.dmab_data.area + ctx->cl_dev.dma_buffer_offset,
+			curr_pos, size_b);
+		size -= size_b;
+		curr_pos += size_b;
+		ctx->cl_dev.dma_buffer_offset = 0;
+	}
+
 	memcpy(ctx->cl_dev.dmab_data.area + ctx->cl_dev.dma_buffer_offset,
 			curr_pos, size);