While it is preferrable to use a fast manual detiling method for LLC
(does not require synchronisation with a busy GPU and for accessing main
memory both the CPU and GPU have the same bandwidth), if we don't have
such a path then using the GPU to perform the blit is far preferable to
a coherent mmapping.
Signed-off-by: Chris Wilson <ch...@chris-wilson.co.uk>
---
src/mesa/drivers/dri/i965/Makefile.sources | 1 +
src/mesa/drivers/dri/i965/intel_pixel.h | 7 +
src/mesa/drivers/dri/i965/intel_pixel_pbo.c | 252 +++++++++++++++++++++++++++
src/mesa/drivers/dri/i965/intel_pixel_read.c | 38 ++--
4 files changed, 280 insertions(+), 18 deletions(-)
create mode 100644 src/mesa/drivers/dri/i965/intel_pixel_pbo.c
diff --git a/src/mesa/drivers/dri/i965/Makefile.sources
b/src/mesa/drivers/dri/i965/Makefile.sources
index 425c883de8..a5550876cd 100644
--- a/src/mesa/drivers/dri/i965/Makefile.sources
+++ b/src/mesa/drivers/dri/i965/Makefile.sources
@@ -101,6 +101,7 @@ i965_FILES = \
intel_pixel_draw.c \
intel_pixel.h \
intel_pixel_read.c \
+ intel_pixel_pbo.c \
intel_screen.c \
intel_screen.h \
intel_state.c \
diff --git a/src/mesa/drivers/dri/i965/intel_pixel.h
b/src/mesa/drivers/dri/i965/intel_pixel.h
index f5b931f5c1..c3d90d3297 100644
--- a/src/mesa/drivers/dri/i965/intel_pixel.h
+++ b/src/mesa/drivers/dri/i965/intel_pixel.h
@@ -58,4 +58,11 @@ void intelBitmap(struct gl_context * ctx,
const struct gl_pixelstore_attrib *unpack,
const GLubyte * pixels);
+bool intel_readpixels_pbo(struct gl_context * ctx,
+ GLint x, GLint y,
+ GLsizei width, GLsizei height,
+ GLenum format, GLenum type,
+ const struct gl_pixelstore_attrib *pack,
+ GLvoid *pixels);
+
#endif
diff --git a/src/mesa/drivers/dri/i965/intel_pixel_pbo.c
b/src/mesa/drivers/dri/i965/intel_pixel_pbo.c
new file mode 100644
index 0000000000..4e212ee92f
--- /dev/null
+++ b/src/mesa/drivers/dri/i965/intel_pixel_pbo.c
@@ -0,0 +1,252 @@
+/*
+ * Copyright © 2012 Intel Corporation
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ */
+
+#include "main/enums.h"
+#include "main/mtypes.h"
+#include "main/macros.h"
+#include "main/fbobject.h"
+#include "main/image.h"
+#include "main/bufferobj.h"
+#include "main/readpix.h"
+#include "main/state.h"
+#include "main/glformats.h"
+
+#include "brw_blorp.h"
+#include "brw_context.h"
+#include "intel_screen.h"
+#include "intel_batchbuffer.h"
+#include "intel_blit.h"
+#include "intel_buffer_objects.h"
+#include "intel_buffers.h"
+#include "intel_fbo.h"
+#include "intel_mipmap_tree.h"
+#include "intel_pixel.h"
+#include "intel_buffer_objects.h"
+
+#define FILE_DEBUG_FLAG DEBUG_PIXEL
+
+
+static int
+get_texture_swizzle(const struct intel_renderbuffer *irb)
+{
+ if (irb->Base.Base._BaseFormat == GL_LUMINANCE ||
+ irb->Base.Base._BaseFormat == GL_INTENSITY)
+ return MAKE_SWIZZLE4(SWIZZLE_X, SWIZZLE_ZERO, SWIZZLE_ZERO, SWIZZLE_ONE);
+
+ if (irb->Base.Base._BaseFormat == GL_LUMINANCE_ALPHA)
+ return MAKE_SWIZZLE4(SWIZZLE_X, SWIZZLE_ZERO, SWIZZLE_ZERO, SWIZZLE_W);
+
+ if (irb->Base.Base._BaseFormat == GL_ALPHA)
+ return MAKE_SWIZZLE4(SWIZZLE_ZERO, SWIZZLE_ZERO, SWIZZLE_ZERO,
SWIZZLE_W);
+
+ if (irb->Base.Base._BaseFormat == GL_RGB)
+ return MAKE_SWIZZLE4(SWIZZLE_X, SWIZZLE_Y, SWIZZLE_Z, SWIZZLE_ONE);
+
+ return SWIZZLE_XYZW;
+}
+
+static bool
+need_signed_unsigned_int_conversion(mesa_format mesaFormat,
+ GLenum format, GLenum type)
+{
+ const GLenum mesaFormatType = _mesa_get_format_datatype(mesaFormat);
+ const bool is_format_integer = _mesa_is_enum_format_integer(format);
+ return (mesaFormatType == GL_INT &&
+ is_format_integer &&
+ (type == GL_UNSIGNED_INT ||
+ type == GL_UNSIGNED_SHORT ||
+ type == GL_UNSIGNED_BYTE)) ||
+ (mesaFormatType == GL_UNSIGNED_INT &&
+ is_format_integer &&
+ (type == GL_INT ||
+ type == GL_SHORT ||
+ type == GL_BYTE));
+}
+
+bool
+intel_readpixels_pbo(struct gl_context * ctx,
+ GLint x, GLint y,
+ GLsizei width, GLsizei height,
+ GLenum format, GLenum type,
+ const struct gl_pixelstore_attrib *pack,
+ GLvoid *pixels)
+{
+ struct brw_context *brw = brw_context(ctx);
+ bool ok = false;
+
+ struct gl_renderbuffer *rb;
+ switch (format) {
+ case GL_STENCIL_INDEX:
+ rb = ctx->ReadBuffer->Attachment[BUFFER_STENCIL].Renderbuffer;
+ break;
+ case GL_DEPTH_COMPONENT:
+ case GL_DEPTH_STENCIL_EXT:
+ rb = ctx->ReadBuffer->Attachment[BUFFER_DEPTH].Renderbuffer;
+ break;
+ case GL_COLOR_INDEX:
+ rb = NULL;
+ break;
+ default:
+ /* all other formats should be color formats */
+ rb = ctx->ReadBuffer->_ColorReadBuffer;
+ break;
+ }
+ if (!rb)
+ return false;
+
+ if (pack->SwapBytes || pack->LsbFirst) {
+ DBG("%s: bad packing params\n", __func__);
+ return false;
+ }
+
+ if (_mesa_readpixels_needs_slow_path(ctx, format, type, GL_TRUE)) {
+ DBG("%s: non-trivial conversion required\n", __func__);
+ return false;
+ }
+
+ struct intel_renderbuffer *irb = intel_renderbuffer(rb);
+ struct intel_mipmap_tree *src = irb->mt;
+ if (format == GL_STENCIL_INDEX && src->stencil_mt)
+ src = src->stencil_mt;
+
+ if (need_signed_unsigned_int_conversion(src->format, format, type))
+ return false;
+
+ mesa_format dst_format = _mesa_format_from_format_and_type(format, type);
+ if (_mesa_format_is_mesa_array_format(dst_format))
+ dst_format = _mesa_format_from_array_format(dst_format);
+ dst_format = _mesa_get_srgb_format_linear(dst_format);
+
+ if (!brw->mesa_format_supports_render[dst_format]) {
+ dst_format = _mesa_format_fallback_rgbx_to_rgba(dst_format);
+ if (!brw->mesa_format_supports_render[dst_format])
+ return false;
+ }
+
+ bool dst_flip = false;
+ int dst_stride = _mesa_image_row_stride(pack, width, format, type);
+ if (pack->Invert) {
+ dst_stride = -dst_stride;
+ dst_flip = true;
+ }
+
+ int bpp = _mesa_bytes_per_pixel(format, type);
+ pixels += pack->SkipRows * dst_stride + pack->SkipPixels * bpp;
+ int dst_size = dst_stride * (height - 1) + width * bpp;
+
+ struct brw_bo *dst_buffer;
+ int dst_offset;
+ if (_mesa_is_bufferobj(pack->BufferObj)) {
+ dst_offset = (GLintptr)pixels;
+ if ((dst_offset | dst_stride) % bpp)
+ return false;
+
+ dst_buffer = intel_bufferobj_buffer(brw,
+ intel_buffer_object(pack->BufferObj),
+ dst_offset,
+ dst_size,
+ false);
+ } else {
+ dst_offset = (GLintptr)pixels & 4095;
+ if ((dst_offset | dst_stride) % bpp)
+ return false;
+
+ dst_buffer = brw_bo_alloc_userptr(brw->bufmgr, "readpixels",
+ (void *)((GLintptr)pixels & -4096),
+ ALIGN(dst_size + dst_offset, 4096), 0);
+ }
+ if (!dst_buffer)
+ return false;
+
+ struct intel_mipmap_tree *dst =
+ intel_miptree_create_for_bo(brw,
+ dst_buffer,
+ dst_format,
+ dst_offset,
+ width, height, 1,
+ dst_stride,
+ MIPTREE_LAYOUT_TILING_NONE |
+ MIPTREE_LAYOUT_DISABLE_AUX);
+ if (!dst)
+ goto err_dst_buffer;
+
+ if (!rb->Name) {
+ dst_flip = !dst_flip;
+ y = rb->Height - y - height;
+ }
+
+ assert(dst->surf.tiling == ISL_TILING_LINEAR);
+ if (!intel_miptree_blit(brw,
+ src, irb->mt_level, irb->mt_layer, x, y, false,
+ dst, 0, 0, 0, 0, dst_flip,
+ width, height, GL_COPY)) {
+ mesa_format src_format = irb->Base.Base.Format;
+
+ /* blorp restrictions; see try_blorp_blit() */
+ switch (format) {
+ case GL_STENCIL_INDEX:
+ if (src_format != dst_format)
+ goto err_dst_mt;
+
+ /* fallthrough */
+ case GL_DEPTH_COMPONENT:
+ case GL_DEPTH_STENCIL_EXT:
+ if ((src_format == MESA_FORMAT_Z24_UNORM_X8_UINT) !=
+ (dst_format == MESA_FORMAT_Z24_UNORM_X8_UINT))
+ goto err_dst_mt;
+
+ if (_mesa_get_format_base_format(src_format) == GL_DEPTH_STENCIL ||
+ _mesa_get_format_base_format(dst_format) == GL_DEPTH_STENCIL)
+ goto err_dst_mt;
+
+ src_format = MESA_FORMAT_NONE;
+ dst_format = MESA_FORMAT_NONE;
+ break;
+
+ default:
+ break;
+ }
+
+ brw_blorp_blit_miptrees(brw,
+ src, irb->mt_level, irb->mt_layer,
+ src_format, get_texture_swizzle(irb),
+ dst, 0, 0, dst_format,
+ x, y, x + width, y + height,
+ 0, 0, width, height,
+ GL_NEAREST, false, dst_flip,
+ false, false);
+ }
+
+ if (!_mesa_is_bufferobj(pack->BufferObj)) {
+ intel_batchbuffer_flush(brw);
+ brw_bo_wait_rendering(dst_buffer);
+ }
+
+ ok = true;
+err_dst_mt:
+ intel_miptree_release(&dst);
+err_dst_buffer:
+ if (!_mesa_is_bufferobj(pack->BufferObj))
+ brw_bo_unreference(dst_buffer);
+ return ok;
+}
diff --git a/src/mesa/drivers/dri/i965/intel_pixel_read.c
b/src/mesa/drivers/dri/i965/intel_pixel_read.c
index 56ae678573..398a8b686d 100644
--- a/src/mesa/drivers/dri/i965/intel_pixel_read.c
+++ b/src/mesa/drivers/dri/i965/intel_pixel_read.c
@@ -34,10 +34,12 @@
#include "main/glformats.h"
#include "drivers/common/meta.h"
+#include "brw_blorp.h"
#include "brw_context.h"
#include "intel_screen.h"
#include "intel_batchbuffer.h"
#include "intel_blit.h"
+#include "intel_buffer_objects.h"
#include "intel_buffers.h"
#include "intel_fbo.h"
#include "intel_mipmap_tree.h"
@@ -205,16 +207,26 @@ intelReadPixels(struct gl_context * ctx,
GLenum format, GLenum type,
const struct gl_pixelstore_attrib *pack, GLvoid * pixels)
{
- bool ok;
-
struct brw_context *brw = brw_context(ctx);
bool dirty;
DBG("%s\n", __func__);
- if (_mesa_is_bufferobj(pack->BufferObj)) {
- if (_mesa_meta_pbo_GetTexSubImage(ctx, 2, NULL, x, y, 0, width, height,
1,
- format, type, pixels, pack)) {
+ /* Reading pixels wont dirty the front buffer, so reset the dirty
+ * flag after calling intel_prepare_render(). */
+ dirty = brw->front_buffer_dirty;
+ intel_prepare_render(brw);
+ brw->front_buffer_dirty = dirty;
+
+ if (!_mesa_is_bufferobj(pack->BufferObj)) {
+ if (intel_readpixels_tiled_memcpy(ctx, x, y, width, height,
+ format, type, pixels, pack))
+ return;
+ }
+
+ if (intel_readpixels_pbo(ctx, x, y, width, height,
+ format, type, pack, pixels)) {
+ if (_mesa_is_bufferobj(pack->BufferObj)) {
/* _mesa_meta_pbo_GetTexSubImage() implements PBO transfers by
* binding the user-provided BO as a fake framebuffer and rendering
* to it. This breaks the invariant of the GL that nothing is able
@@ -236,22 +248,12 @@ intelReadPixels(struct gl_context * ctx,
* flush here unconditionally.
*/
brw_emit_mi_flush(brw);
- return;
}
-
- perf_debug("%s: fallback to CPU mapping in PBO case\n", __func__);
+ return;
}
- /* Reading pixels wont dirty the front buffer, so reset the dirty
- * flag after calling intel_prepare_render(). */
- dirty = brw->front_buffer_dirty;
- intel_prepare_render(brw);
- brw->front_buffer_dirty = dirty;
-
- ok = intel_readpixels_tiled_memcpy(ctx, x, y, width, height,
- format, type, pixels, pack);
- if(ok)
- return;
+ perf_debug("%s: fallback to CPU mapping%s\n", __func__,
+ _mesa_is_bufferobj(pack->BufferObj) ? " for PBO" : "");
/* Update Mesa state before calling _mesa_readpixels().
* XXX this may not be needed since ReadPixels no longer uses the
--
2.13.3
_______________________________________________
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev
