gallium: split transfer_inline_write into buffer and texture callbacks

to reduce the call indirections with u_resource_vtbl.

The worst call tree you could get was:
  - u_transfer_inline_write_vtbl
    - u_default_transfer_inline_write
      - u_transfer_map_vtbl
        - driver_transfer_map
      - u_transfer_unmap_vtbl
        - driver_transfer_unmap

That's 6 indirect calls. Some drivers only had 5. The goal is to have
1 indirect call for drivers that care. The resource type can be determined
statically at most call sites.

The new interface is:
  pipe_context::buffer_subdata(ctx, resource, usage, offset, size, data)
  pipe_context::texture_subdata(ctx, resource, level, usage, box, data,
                                stride, layer_stride)

v2: fix whitespace, correct ilo's behavior

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Acked-by: Roland Scheidegger <sroland@vmware.com>

3 files changed, 69 insertions, 110 deletions

diff --git a/src/gallium/auxiliary/util/u_inlines.h b/src/gallium/auxiliary/util/u_inlines.h
index 207e2aa..07f7354 100644
--- a/src/gallium/auxiliary/util/u_inlines.h
+++ b/src/gallium/auxiliary/util/u_inlines.h
@@ -339,7 +339,6 @@ pipe_buffer_write(struct pipe_context *pipe,
                   unsigned size,
                   const void *data)
 {
-   struct pipe_box box;
    unsigned access = PIPE_TRANSFER_WRITE;
 
    if (offset == 0 && size == buf->width0) {
@@ -348,16 +347,7 @@ pipe_buffer_write(struct pipe_context *pipe,
       access |= PIPE_TRANSFER_DISCARD_RANGE;
    }
 
-   u_box_1d(offset, size, &box);
-
-   pipe->transfer_inline_write( pipe,
-                                buf,
-                                0,
-                                access,
-                                &box,
-                                data,
-                                size,
-                                0);
+   pipe->buffer_subdata(pipe, buf, access, offset, size, data);
 }
 
 /**
@@ -372,18 +362,10 @@ pipe_buffer_write_nooverlap(struct pipe_context *pipe,
                             unsigned offset, unsigned size,
                             const void *data)
 {
-   struct pipe_box box;
-
-   u_box_1d(offset, size, &box);
-
-   pipe->transfer_inline_write(pipe,
-                               buf,
-                               0,
-                               (PIPE_TRANSFER_WRITE |
-                                PIPE_TRANSFER_UNSYNCHRONIZED),
-                               &box,
-                               data,
-                               0, 0);
+   pipe->buffer_subdata(pipe, buf,
+                        (PIPE_TRANSFER_WRITE |
+                         PIPE_TRANSFER_UNSYNCHRONIZED),
+                        offset, size, data);
 }
 
 
diff --git a/src/gallium/auxiliary/util/u_transfer.c b/src/gallium/auxiliary/util/u_transfer.c
index 0610535..82cf68d 100644
--- a/src/gallium/auxiliary/util/u_transfer.c
+++ b/src/gallium/auxiliary/util/u_transfer.c
@@ -4,34 +4,58 @@
 #include "util/u_transfer.h"
 #include "util/u_memory.h"
 
-/* One-shot transfer operation with data supplied in a user
- * pointer.  XXX: strides??
- */
-void u_default_transfer_inline_write( struct pipe_context *pipe,
-                                      struct pipe_resource *resource,
-                                      unsigned level,
-                                      unsigned usage,
-                                      const struct pipe_box *box,
-                                      const void *data,
-                                      unsigned stride,
-                                      unsigned layer_stride)
+void u_default_buffer_subdata(struct pipe_context *pipe,
+                              struct pipe_resource *resource,
+                              unsigned usage, unsigned offset,
+                              unsigned size, const void *data)
 {
    struct pipe_transfer *transfer = NULL;
+   struct pipe_box box;
    uint8_t *map = NULL;
 
    assert(!(usage & PIPE_TRANSFER_READ));
 
-   /* the write flag is implicit by the nature of transfer_inline_write */
+   /* the write flag is implicit by the nature of buffer_subdata */
    usage |= PIPE_TRANSFER_WRITE;
 
-   /* transfer_inline_write implicitly discards the rewritten buffer range */
-   if (resource->target == PIPE_BUFFER &&
-       box->x == 0 && box->width == resource->width0) {
+   /* buffer_subdata implicitly discards the rewritten buffer range */
+   if (offset == 0 && size == resource->width0) {
       usage |= PIPE_TRANSFER_DISCARD_WHOLE_RESOURCE;
    } else {
       usage |= PIPE_TRANSFER_DISCARD_RANGE;
    }
 
+   u_box_1d(offset, size, &box);
+
+   map = pipe->transfer_map(pipe, resource, 0, usage, &box, &transfer);
+   if (!map)
+      return;
+
+   memcpy(map, data, size);
+   pipe_transfer_unmap(pipe, transfer);
+}
+
+void u_default_texture_subdata(struct pipe_context *pipe,
+                               struct pipe_resource *resource,
+                               unsigned level,
+                               unsigned usage,
+                               const struct pipe_box *box,
+                               const void *data,
+                               unsigned stride,
+                               unsigned layer_stride)
+{
+   struct pipe_transfer *transfer = NULL;
+   const uint8_t *src_data = data;
+   uint8_t *map = NULL;
+
+   assert(!(usage & PIPE_TRANSFER_READ));
+
+   /* the write flag is implicit by the nature of texture_subdata */
+   usage |= PIPE_TRANSFER_WRITE;
+
+   /* texture_subdata implicitly discards the rewritten buffer range */
+   usage |= PIPE_TRANSFER_DISCARD_RANGE;
+
    map = pipe->transfer_map(pipe,
                             resource,
                             level,
@@ -40,28 +64,18 @@ void u_default_transfer_inline_write( struct pipe_context *pipe,
    if (!map)
       return;
 
-   if (resource->target == PIPE_BUFFER) {
-      assert(box->height == 1);
-      assert(box->depth == 1);
-
-      memcpy(map, data, box->width);
-   }
-   else {
-      const uint8_t *src_data = data;
-
-      util_copy_box(map,
-		    resource->format,
-		    transfer->stride, /* bytes */
-		    transfer->layer_stride, /* bytes */
-                    0, 0, 0,
-		    box->width,
-		    box->height,
-		    box->depth,
-		    src_data,
-		    stride,       /* bytes */
-		    layer_stride, /* bytes */
-		    0, 0, 0);
-   }
+   util_copy_box(map,
+                 resource->format,
+                 transfer->stride, /* bytes */
+                 transfer->layer_stride, /* bytes */
+                 0, 0, 0,
+                 box->width,
+                 box->height,
+                 box->depth,
+                 src_data,
+                 stride,       /* bytes */
+                 layer_stride, /* bytes */
+                 0, 0, 0);
 
    pipe_transfer_unmap(pipe, transfer);
 }
@@ -138,27 +152,3 @@ void u_transfer_unmap_vtbl( struct pipe_context *pipe,
    struct u_resource *ur = u_resource(transfer->resource);
    ur->vtbl->transfer_unmap(pipe, transfer);
 }
-
-void u_transfer_inline_write_vtbl( struct pipe_context *pipe,
-                                   struct pipe_resource *resource,
-                                   unsigned level,
-                                   unsigned usage,
-                                   const struct pipe_box *box,
-                                   const void *data,
-                                   unsigned stride,
-                                   unsigned layer_stride)
-{
-   struct u_resource *ur = u_resource(resource);
-   ur->vtbl->transfer_inline_write(pipe,
-                                   resource,
-                                   level,
-                                   usage,
-                                   box,
-                                   data,
-                                   stride,
-                                   layer_stride);
-}
-
-
-
-
diff --git a/src/gallium/auxiliary/util/u_transfer.h b/src/gallium/auxiliary/util/u_transfer.h
index 660dc16..7f680bc 100644
--- a/src/gallium/auxiliary/util/u_transfer.h
+++ b/src/gallium/auxiliary/util/u_transfer.h
@@ -14,14 +14,19 @@ boolean u_default_resource_get_handle(struct pipe_screen *screen,
                                       struct pipe_resource *resource,
                                       struct winsys_handle *handle);
 
-void u_default_transfer_inline_write( struct pipe_context *pipe,
-                                      struct pipe_resource *resource,
-                                      unsigned level,
-                                      unsigned usage,
-                                      const struct pipe_box *box,
-                                      const void *data,
-                                      unsigned stride,
-                                      unsigned layer_stride);
+void u_default_buffer_subdata(struct pipe_context *pipe,
+                              struct pipe_resource *resource,
+                              unsigned usage, unsigned offset,
+                              unsigned size, const void *data);
+
+void u_default_texture_subdata(struct pipe_context *pipe,
+                               struct pipe_resource *resource,
+                               unsigned level,
+                               unsigned usage,
+                               const struct pipe_box *box,
+                               const void *data,
+                               unsigned stride,
+                               unsigned layer_stride);
 
 void u_default_transfer_flush_region( struct pipe_context *pipe,
                                       struct pipe_transfer *transfer,
@@ -58,15 +63,6 @@ struct u_resource_vtbl {
 
    void (*transfer_unmap)( struct pipe_context *,
                            struct pipe_transfer *transfer );
-
-   void (*transfer_inline_write)( struct pipe_context *pipe,
-                                  struct pipe_resource *resource,
-                                  unsigned level,
-                                  unsigned usage,
-                                  const struct pipe_box *box,
-                                  const void *data,
-                                  unsigned stride,
-                                  unsigned layer_stride);
 };
 
 
@@ -98,13 +94,4 @@ void u_transfer_flush_region_vtbl( struct pipe_context *pipe,
 void u_transfer_unmap_vtbl( struct pipe_context *rm_ctx,
                             struct pipe_transfer *transfer );
 
-void u_transfer_inline_write_vtbl( struct pipe_context *rm_ctx,
-                                   struct pipe_resource *resource,
-                                   unsigned level,
-                                   unsigned usage,
-                                   const struct pipe_box *box,
-                                   const void *data,
-                                   unsigned stride,
-                                   unsigned layer_stride);
-
 #endif