r600g/compute: Map only against intermediate buffers

With this we can assure that mapped buffers will never change
its position when relocating the pool.

This patch should finally solve the mapping bug.

v2: Use the new is_item_in_pool util function,
    as suggested by Tom Stellard

Reviewed-by: Tom Stellard <thomas.stellard@amd.com>

diff --git a/src/gallium/drivers/r600/evergreen_compute.c b/src/gallium/drivers/r600/evergreen_compute.c
index 8657071e4c6356893b15f7cef3f2d716d37e69e5..c0dd0f349144f756cecc4b8019b00e2802845e8f 100644 (file)
--- a/src/gallium/drivers/r600/evergreen_compute.c
+++ b/src/gallium/drivers/r600/evergreen_compute.c
@@ -970,14 +970,12 @@ void *r600_compute_global_transfer_map(
        struct pipe_resource *dst;
        unsigned offset = box->x;
 
-       if (buffer->chunk->real_buffer) {
-               dst = (struct pipe_resource*)buffer->chunk->real_buffer;
-       }
-       else {
-               dst = (struct pipe_resource*)buffer->chunk->pool->bo;
-               offset += (buffer->chunk->start_in_dw * 4);
+       if (is_item_in_pool(buffer->chunk)) {
+               compute_memory_demote_item(pool, buffer->chunk, ctx_);
        }
 
+       dst = (struct pipe_resource*)buffer->chunk->real_buffer;
+
        if (usage & PIPE_TRANSFER_READ)
                buffer->chunk->status |= ITEM_MAPPED_FOR_READING;