xserver-multidpi/glamor/glamor_render.c

1532 lines
41 KiB
C
Raw Normal View History

/*
* Copyright © 2009 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.
*
* Authors:
* Eric Anholt <eric@anholt.net>
*
*/
/** @file glamor_render.c
*
* Render acceleration implementation
*/
#include "glamor_priv.h"
#ifdef RENDER
#include "mipict.h"
2010-02-08 11:23:14 +01:00
#include "fbpict.h"
//#include "glu3/glu3.h"
struct shader_key {
enum shader_source source;
enum shader_mask mask;
enum shader_in in;
};
struct blendinfo {
Bool dest_alpha;
Bool source_alpha;
GLenum source_blend;
GLenum dest_blend;
};
static struct blendinfo composite_op_info[] = {
[PictOpClear] = {0, 0, GL_ZERO, GL_ZERO},
[PictOpSrc] = {0, 0, GL_ONE, GL_ZERO},
[PictOpDst] = {0, 0, GL_ZERO, GL_ONE},
[PictOpOver] = {0, 1, GL_ONE, GL_ONE_MINUS_SRC_ALPHA},
[PictOpOverReverse] = {1, 0, GL_ONE_MINUS_DST_ALPHA, GL_ONE},
[PictOpIn] = {1, 0, GL_DST_ALPHA, GL_ZERO},
[PictOpInReverse] = {0, 1, GL_ZERO, GL_SRC_ALPHA},
[PictOpOut] = {1, 0, GL_ONE_MINUS_DST_ALPHA, GL_ZERO},
[PictOpOutReverse] = {0, 1, GL_ZERO, GL_ONE_MINUS_SRC_ALPHA},
[PictOpAtop] = {1, 1, GL_DST_ALPHA, GL_ONE_MINUS_SRC_ALPHA},
[PictOpAtopReverse] = {1, 1, GL_ONE_MINUS_DST_ALPHA, GL_SRC_ALPHA},
[PictOpXor] =
{1, 1, GL_ONE_MINUS_DST_ALPHA, GL_ONE_MINUS_SRC_ALPHA},
[PictOpAdd] = {0, 0, GL_ONE, GL_ONE},
};
static GLuint
glamor_create_composite_fs(glamor_gl_dispatch * dispatch,
struct shader_key *key)
{
const char *source_solid_fetch =
GLAMOR_DEFAULT_PRECISION
"uniform vec4 source;\n"
"vec4 get_source()\n" "{\n" " return source;\n" "}\n";
const char *source_alpha_pixmap_fetch =
GLAMOR_DEFAULT_PRECISION
"varying vec2 source_texture;\n"
"uniform sampler2D source_sampler;\n"
"vec4 get_source()\n"
"{\n" " return texture2D(source_sampler, source_texture);\n"
"}\n";
const char *source_pixmap_fetch =
GLAMOR_DEFAULT_PRECISION "varying vec2 source_texture;\n"
"uniform sampler2D source_sampler;\n" "vec4 get_source()\n"
"{\n"
" return vec4(texture2D(source_sampler, source_texture).rgb, 1);\n"
"}\n";
const char *mask_solid_fetch =
GLAMOR_DEFAULT_PRECISION "uniform vec4 mask;\n"
"vec4 get_mask()\n" "{\n" " return mask;\n" "}\n";
const char *mask_alpha_pixmap_fetch =
GLAMOR_DEFAULT_PRECISION "varying vec2 mask_texture;\n"
"uniform sampler2D mask_sampler;\n" "vec4 get_mask()\n" "{\n"
" return texture2D(mask_sampler, mask_texture);\n" "}\n";
const char *mask_pixmap_fetch =
GLAMOR_DEFAULT_PRECISION "varying vec2 mask_texture;\n"
"uniform sampler2D mask_sampler;\n" "vec4 get_mask()\n" "{\n"
" return vec4(texture2D(mask_sampler, mask_texture).rgb, 1);\n"
"}\n";
const char *in_source_only =
GLAMOR_DEFAULT_PRECISION "void main()\n" "{\n"
" gl_FragColor = get_source();\n" "}\n";
const char *in_normal =
GLAMOR_DEFAULT_PRECISION "void main()\n" "{\n"
" gl_FragColor = get_source() * get_mask().a;\n" "}\n";
const char *in_ca_source =
GLAMOR_DEFAULT_PRECISION "void main()\n" "{\n"
" gl_FragColor = get_source() * get_mask();\n" "}\n";
const char *in_ca_alpha =
GLAMOR_DEFAULT_PRECISION "void main()\n" "{\n"
" gl_FragColor = get_source().a * get_mask();\n" "}\n";
char *source;
const char *source_fetch;
const char *mask_fetch = "";
const char *in;
GLuint prog;
switch (key->source) {
case SHADER_SOURCE_SOLID:
source_fetch = source_solid_fetch;
break;
case SHADER_SOURCE_TEXTURE_ALPHA:
source_fetch = source_alpha_pixmap_fetch;
break;
case SHADER_SOURCE_TEXTURE:
source_fetch = source_pixmap_fetch;
break;
default:
FatalError("Bad composite shader source");
}
switch (key->mask) {
case SHADER_MASK_NONE:
break;
case SHADER_MASK_SOLID:
mask_fetch = mask_solid_fetch;
break;
case SHADER_MASK_TEXTURE_ALPHA:
mask_fetch = mask_alpha_pixmap_fetch;
break;
case SHADER_MASK_TEXTURE:
mask_fetch = mask_pixmap_fetch;
break;
default:
FatalError("Bad composite shader mask");
}
switch (key->in) {
case SHADER_IN_SOURCE_ONLY:
in = in_source_only;
break;
case SHADER_IN_NORMAL:
in = in_normal;
break;
case SHADER_IN_CA_SOURCE:
in = in_ca_source;
break;
case SHADER_IN_CA_ALPHA:
in = in_ca_alpha;
break;
default:
FatalError("Bad composite IN type");
}
XNFasprintf(&source, "%s%s%s", source_fetch, mask_fetch, in);
prog = glamor_compile_glsl_prog(dispatch, GL_FRAGMENT_SHADER,
source);
free(source);
return prog;
}
static GLuint
glamor_create_composite_vs(glamor_gl_dispatch * dispatch,
struct shader_key *key)
{
const char *main_opening =
"attribute vec4 v_position;\n"
"attribute vec4 v_texcoord0;\n"
"attribute vec4 v_texcoord1;\n"
"varying vec2 source_texture;\n"
"varying vec2 mask_texture;\n"
"void main()\n" "{\n" " gl_Position = v_position;\n";
const char *source_coords =
" source_texture = v_texcoord0.xy;\n";
const char *mask_coords = " mask_texture = v_texcoord1.xy;\n";
const char *main_closing = "}\n";
const char *source_coords_setup = "";
const char *mask_coords_setup = "";
char *source;
GLuint prog;
if (key->source != SHADER_SOURCE_SOLID)
source_coords_setup = source_coords;
if (key->mask != SHADER_MASK_NONE
&& key->mask != SHADER_MASK_SOLID)
mask_coords_setup = mask_coords;
XNFasprintf(&source,
"%s%s%s%s",
main_opening,
source_coords_setup, mask_coords_setup, main_closing);
prog =
glamor_compile_glsl_prog(dispatch, GL_VERTEX_SHADER, source);
free(source);
return prog;
}
static void
glamor_create_composite_shader(ScreenPtr screen, struct shader_key *key,
glamor_composite_shader * shader)
{
GLuint vs, fs, prog;
GLint source_sampler_uniform_location,
mask_sampler_uniform_location;
glamor_screen_private *glamor = glamor_get_screen_private(screen);
glamor_gl_dispatch *dispatch = &glamor->dispatch;
vs = glamor_create_composite_vs(dispatch, key);
if (vs == 0)
return;
fs = glamor_create_composite_fs(dispatch, key);
if (fs == 0)
return;
prog = dispatch->glCreateProgram();
dispatch->glAttachShader(prog, vs);
dispatch->glAttachShader(prog, fs);
dispatch->glBindAttribLocation(prog, GLAMOR_VERTEX_POS,
"v_position");
dispatch->glBindAttribLocation(prog, GLAMOR_VERTEX_SOURCE,
"v_texcoord0");
dispatch->glBindAttribLocation(prog, GLAMOR_VERTEX_MASK,
"v_texcoord1");
glamor_link_glsl_prog(dispatch, prog);
shader->prog = prog;
dispatch->glUseProgram(prog);
if (key->source == SHADER_SOURCE_SOLID) {
shader->source_uniform_location =
dispatch->glGetUniformLocation(prog, "source");
} else {
source_sampler_uniform_location =
dispatch->glGetUniformLocation(prog, "source_sampler");
dispatch->glUniform1i(source_sampler_uniform_location, 0);
}
if (key->mask != SHADER_MASK_NONE) {
if (key->mask == SHADER_MASK_SOLID) {
shader->mask_uniform_location =
dispatch->glGetUniformLocation(prog, "mask");
} else {
mask_sampler_uniform_location =
dispatch->glGetUniformLocation(prog,
"mask_sampler");
dispatch->glUniform1i
(mask_sampler_uniform_location, 1);
}
}
}
static glamor_composite_shader *
glamor_lookup_composite_shader(ScreenPtr screen, struct
shader_key
*key)
{
glamor_screen_private *glamor_priv =
glamor_get_screen_private(screen);
glamor_composite_shader *shader;
shader =
&glamor_priv->composite_shader[key->source][key->
mask][key->in];
if (shader->prog == 0)
glamor_create_composite_shader(screen, key, shader);
return shader;
}
#define GLAMOR_COMPOSITE_VBO_SIZE 8192
static void
glamor_reset_composite_vbo(ScreenPtr screen)
{
glamor_screen_private *glamor_priv =
glamor_get_screen_private(screen);
glamor_priv->vbo_offset = 0;
glamor_priv->vbo_size = GLAMOR_COMPOSITE_VBO_SIZE;
glamor_priv->render_nr_verts = 0;
}
void
glamor_init_composite_shaders(ScreenPtr screen)
{
glamor_screen_private *glamor_priv =
glamor_get_screen_private(screen);
glamor_priv->vb = malloc(GLAMOR_COMPOSITE_VBO_SIZE);
assert(glamor_priv->vb != NULL);
glamor_reset_composite_vbo(screen);
}
static Bool
glamor_set_composite_op(ScreenPtr screen,
CARD8 op, PicturePtr dest, PicturePtr mask)
{
GLenum source_blend, dest_blend;
struct blendinfo *op_info;
glamor_screen_private *glamor_priv =
glamor_get_screen_private(screen);
glamor_gl_dispatch *dispatch = &glamor_priv->dispatch;
if (op >= ARRAY_SIZE(composite_op_info)) {
glamor_fallback("unsupported render op %d \n", op);
return GL_FALSE;
}
op_info = &composite_op_info[op];
source_blend = op_info->source_blend;
dest_blend = op_info->dest_blend;
/* If there's no dst alpha channel, adjust the blend op so that we'll treat
* it as always 1.
*/
if (PICT_FORMAT_A(dest->format) == 0 && op_info->dest_alpha) {
if (source_blend == GL_DST_ALPHA)
source_blend = GL_ONE;
else if (source_blend == GL_ONE_MINUS_DST_ALPHA)
source_blend = GL_ZERO;
}
/* Set up the source alpha value for blending in component alpha mode. */
if (mask && mask->componentAlpha
&& PICT_FORMAT_RGB(mask->format) != 0 && op_info->source_alpha)
{
if (source_blend != GL_ZERO) {
glamor_fallback
("Dual-source composite blending not supported\n");
return GL_FALSE;
}
if (dest_blend == GL_SRC_ALPHA)
dest_blend = GL_SRC_COLOR;
else if (dest_blend == GL_ONE_MINUS_SRC_ALPHA)
dest_blend = GL_ONE_MINUS_SRC_COLOR;
}
if (source_blend == GL_ONE && dest_blend == GL_ZERO) {
dispatch->glDisable(GL_BLEND);
} else {
dispatch->glEnable(GL_BLEND);
dispatch->glBlendFunc(source_blend, dest_blend);
}
return TRUE;
}
static void
glamor_set_composite_texture(ScreenPtr screen, int unit,
PicturePtr picture,
glamor_pixmap_private * pixmap_priv)
{
glamor_screen_private *glamor_priv =
glamor_get_screen_private(screen);
glamor_gl_dispatch *dispatch = &glamor_priv->dispatch;
dispatch->glActiveTexture(GL_TEXTURE0 + unit);
dispatch->glBindTexture(GL_TEXTURE_2D, pixmap_priv->tex);
switch (picture->repeatType) {
case RepeatNone:
#ifndef GLAMOR_GLES2
/* XXX GLES2 doesn't support GL_CLAMP_TO_BORDER. */
dispatch->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_S,
GL_CLAMP_TO_BORDER);
dispatch->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T,
GL_CLAMP_TO_BORDER);
#endif
break;
case RepeatNormal:
dispatch->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_S,
GL_REPEAT);
dispatch->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T,
GL_REPEAT);
break;
case RepeatPad:
dispatch->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_S,
GL_CLAMP_TO_EDGE);
dispatch->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T,
GL_CLAMP_TO_EDGE);
break;
case RepeatReflect:
dispatch->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_S,
GL_MIRRORED_REPEAT);
dispatch->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T,
GL_MIRRORED_REPEAT);
break;
}
switch (picture->filter) {
case PictFilterNearest:
dispatch->glTexParameteri(GL_TEXTURE_2D,
GL_TEXTURE_MIN_FILTER,
GL_NEAREST);
dispatch->glTexParameteri(GL_TEXTURE_2D,
GL_TEXTURE_MAG_FILTER,
GL_NEAREST);
break;
case PictFilterBilinear:
default:
dispatch->glTexParameteri(GL_TEXTURE_2D,
GL_TEXTURE_MIN_FILTER,
GL_LINEAR);
dispatch->glTexParameteri(GL_TEXTURE_2D,
GL_TEXTURE_MAG_FILTER,
GL_LINEAR);
break;
}
#ifndef GLAMOR_GLES2
dispatch->glEnable(GL_TEXTURE_2D);
#endif
}
static void
glamor_set_composite_solid(glamor_gl_dispatch * dispatch, float *color,
GLint uniform_location)
{
dispatch->glUniform4fv(uniform_location, 1, color);
}
static int
compatible_formats(CARD8 op, PicturePtr dst, PicturePtr src)
{
if (op == PictOpSrc) {
if (src->format == dst->format)
return 1;
if (src->format == PICT_a8r8g8b8
&& dst->format == PICT_x8r8g8b8)
return 1;
if (src->format == PICT_a8b8g8r8
&& dst->format == PICT_x8b8g8r8)
return 1;
} else if (op == PictOpOver) {
if (src->alphaMap || dst->alphaMap)
return 0;
if (src->format != dst->format)
return 0;
if (src->format == PICT_x8r8g8b8
|| src->format == PICT_x8b8g8r8)
return 1;
}
return 0;
}
static char
glamor_get_picture_location(PicturePtr picture)
{
if (picture == NULL)
return ' ';
if (picture->pDrawable == NULL) {
switch (picture->pSourcePict->type) {
case SourcePictTypeSolidFill:
return 'c';
case SourcePictTypeLinear:
return 'l';
case SourcePictTypeRadial:
return 'r';
default:
return '?';
}
}
return glamor_get_drawable_location(picture->pDrawable);
}
static Bool
glamor_composite_with_copy(CARD8 op,
PicturePtr source,
PicturePtr dest,
INT16 x_source,
INT16 y_source,
INT16 x_dest,
INT16 y_dest, CARD16 width, CARD16 height)
{
RegionRec region;
if (!source->pDrawable)
return FALSE;
if (!compatible_formats(op, dest, source))
return FALSE;
if (source->repeat || source->transform)
return FALSE;
x_dest += dest->pDrawable->x;
y_dest += dest->pDrawable->y;
x_source += source->pDrawable->x;
y_source += source->pDrawable->y;
if (!miComputeCompositeRegion(&region,
source, NULL, dest,
x_source, y_source,
0, 0, x_dest, y_dest, width, height))
return TRUE;
glamor_copy_n_to_n(source->pDrawable,
dest->pDrawable, NULL,
REGION_RECTS(&region),
REGION_NUM_RECTS(&region),
x_source - x_dest, y_source - y_dest,
FALSE, FALSE, 0, NULL);
REGION_UNINIT(dest->pDrawable->pScreen, &region);
return TRUE;
}
static void
glamor_setup_composite_vbo(ScreenPtr screen)
{
glamor_screen_private *glamor_priv =
glamor_get_screen_private(screen);
glamor_gl_dispatch *dispatch = &glamor_priv->dispatch;
glamor_priv->vb_stride = 2 * sizeof(float);
if (glamor_priv->has_source_coords)
glamor_priv->vb_stride += 2 * sizeof(float);
if (glamor_priv->has_mask_coords)
glamor_priv->vb_stride += 2 * sizeof(float);
dispatch->glBindBuffer(GL_ARRAY_BUFFER, glamor_priv->vbo);
dispatch->glVertexAttribPointer(GLAMOR_VERTEX_POS, 2, GL_FLOAT,
GL_FALSE, glamor_priv->vb_stride,
(void *) ((long)
glamor_priv->vbo_offset));
dispatch->glEnableVertexAttribArray(GLAMOR_VERTEX_POS);
if (glamor_priv->has_source_coords) {
dispatch->glVertexAttribPointer(GLAMOR_VERTEX_SOURCE, 2,
GL_FLOAT, GL_FALSE,
glamor_priv->vb_stride,
(void *) ((long)
glamor_priv->vbo_offset
+
2 *
sizeof(float)));
dispatch->glEnableVertexAttribArray(GLAMOR_VERTEX_SOURCE);
}
if (glamor_priv->has_mask_coords) {
dispatch->glVertexAttribPointer(GLAMOR_VERTEX_MASK, 2,
GL_FLOAT, GL_FALSE,
glamor_priv->vb_stride,
(void *) ((long)
glamor_priv->vbo_offset
+
(glamor_priv->has_source_coords
? 4 : 2) *
sizeof(float)));
dispatch->glEnableVertexAttribArray(GLAMOR_VERTEX_MASK);
}
}
static void
glamor_emit_composite_vert(ScreenPtr screen,
const float *src_coords,
const float *mask_coords,
const float *dst_coords, int i)
{
glamor_screen_private *glamor_priv =
glamor_get_screen_private(screen);
float *vb = (float *) (glamor_priv->vb + glamor_priv->vbo_offset);
int j = 0;
vb[j++] = dst_coords[i * 2 + 0];
vb[j++] = dst_coords[i * 2 + 1];
if (glamor_priv->has_source_coords) {
vb[j++] = src_coords[i * 2 + 0];
vb[j++] = src_coords[i * 2 + 1];
}
if (glamor_priv->has_mask_coords) {
vb[j++] = mask_coords[i * 2 + 0];
vb[j++] = mask_coords[i * 2 + 1];
}
glamor_priv->render_nr_verts++;
glamor_priv->vbo_offset += glamor_priv->vb_stride;
}
static void
glamor_flush_composite_rects(ScreenPtr screen)
{
glamor_screen_private *glamor_priv =
glamor_get_screen_private(screen);
glamor_gl_dispatch *dispatch = &glamor_priv->dispatch;
if (!glamor_priv->render_nr_verts)
return;
dispatch->glBufferData(GL_ARRAY_BUFFER, glamor_priv->vbo_offset,
glamor_priv->vb, GL_STREAM_DRAW);
dispatch->glDrawArrays(GL_TRIANGLES, 0,
glamor_priv->render_nr_verts);
glamor_reset_composite_vbo(screen);
}
static void
glamor_emit_composite_rect(ScreenPtr screen,
const float *src_coords,
const float *mask_coords,
const float *dst_coords)
{
glamor_screen_private *glamor_priv =
glamor_get_screen_private(screen);
glamor_gl_dispatch *dispatch = &glamor_priv->dispatch;
if (glamor_priv->vbo_offset + 6 * glamor_priv->vb_stride >
glamor_priv->vbo_size) {
glamor_flush_composite_rects(screen);
}
if (glamor_priv->vbo_offset == 0) {
if (glamor_priv->vbo == 0)
dispatch->glGenBuffers(1, &glamor_priv->vbo);
glamor_setup_composite_vbo(screen);
}
glamor_emit_composite_vert(screen, src_coords, mask_coords,
dst_coords, 0);
glamor_emit_composite_vert(screen, src_coords, mask_coords,
dst_coords, 1);
glamor_emit_composite_vert(screen, src_coords, mask_coords,
dst_coords, 2);
glamor_emit_composite_vert(screen, src_coords, mask_coords,
dst_coords, 0);
glamor_emit_composite_vert(screen, src_coords, mask_coords,
dst_coords, 2);
glamor_emit_composite_vert(screen, src_coords, mask_coords,
dst_coords, 3);
glamor : Add dynamic texture uploading feature. Major refactoring. 1. Rewrite the pixmap texture uploading and downloading functions. Add some new functions for both the prepare/finish access and the new performance feature dynamic texture uploading, which could download and upload the current image to/from a private texture/fbo. In the uploading or downloading phase, we need to handle two things: The first is the yInverted option, If it set, then we don't need to flip y. If not set, if it is from a dynamic texture uploading then we don't need to flip either if the current drawing process will flip it latter. If it is from finish_access, then we must flip the y axis. The second thing is the alpha channel hanlding, if the pixmap's format is something like x8a8r8g8, x1r5g5b5 which means it doesn't has alpha channel, but it do has those extra bits. Then we need to wire those bits to 1. 2. Add almost all the required picture format support. This is not as trivial as it looks like. The previous implementation only support GL_a8,GL_a8r8g8b8,GL_x8r8g8b8. All the other format, we have to fallback to cpu. The reason why we can't simply add those other color format is because the exists of picture. one drawable pixmap may has one or even more container pictures. The drawable pixmap's depth can't map to a specified color format, for example depth 16 can mapped to r5g6b5, x1r5g5b5, a1r5g5b5, or even b5g6r5. So we can't get get the color format just from the depth value. But the pixmap do not has a pict_format element. We have to make a new one in the pixmap private data structure. Reroute the CreatePicture to glamor_create_picture and then store the picture's format to the pixmap's private structure. This is not an ideal solution, as there may be more than one pictures refer to the same pixmap. Then we will have trouble. There is an example in glamor_composite_with_shader. The source and mask often share the same pixmap, but use different picture format. Our current solution is to combine those two different picture formats to one which will not lose any data. Then change the source's format to this new format and then upload the pixmap to texture once. It works. If we fail to find a matched new format then we fallback. There still is a potential problem, if two pictures refer to the same pixmap, and one of them destroy the picture, but the other still remained to be used latter. We don't handle that situation currently. To be fixed. 3. Dynamic texture uploading. This is a performance feature. Although we don't like the client to hold a pixmap data to shared memory and we can't accelerate it. And even worse, we may need to fallback all the required pixmaps to cpu memory and then process them on CPU. This feature is to mitigate this penalty. When the target pixmap has a valid gl fbo attached to it. But the other pixmaps are not. Then it will be more efficient to upload the other pixmaps to GPU and then do the blitting or rendering on GPU than fallback all the pixmaps to CPU. To enable this feature, I experienced a significant performance improvement in the Game "Mines" :). 4. Debug facility. Modify the debug output mechanism. Now add a new macro: glamor_debug_output(_level_, _format_,...) to conditional output some messages according to the environment variable GLAMOR_DEBUG. We have the following levels currently. exports GLAMOR_DEBUG to 3 will enable all the above messages. 5. Changes in pixmap private data structure. Add some for the full color format supports and relate it to the pictures which already described. Also Add the following new elements: gl_fbo - to indicates whether this pixmap is on gpu only. gl_tex - to indicates whether the tex is valid and is containing the pixmap's image originally. As we bring the dynamic pixmap uploading feature, so a cpu memory pixmap may also has a valid fbo or tex attached to it. So we will have to use the above new element to check it true type. After this commit, we can pass the rendercheck testing for all the picture formats. And is much much fater than fallback to cpu when doing rendercheck testing. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2011-06-21 12:31:11 +02:00
}
int pict_format_combine_tab[][3] = {
{PICT_TYPE_ARGB, PICT_TYPE_A, PICT_TYPE_ARGB},
{PICT_TYPE_ABGR, PICT_TYPE_A, PICT_TYPE_ABGR},
};
glamor : Add dynamic texture uploading feature. Major refactoring. 1. Rewrite the pixmap texture uploading and downloading functions. Add some new functions for both the prepare/finish access and the new performance feature dynamic texture uploading, which could download and upload the current image to/from a private texture/fbo. In the uploading or downloading phase, we need to handle two things: The first is the yInverted option, If it set, then we don't need to flip y. If not set, if it is from a dynamic texture uploading then we don't need to flip either if the current drawing process will flip it latter. If it is from finish_access, then we must flip the y axis. The second thing is the alpha channel hanlding, if the pixmap's format is something like x8a8r8g8, x1r5g5b5 which means it doesn't has alpha channel, but it do has those extra bits. Then we need to wire those bits to 1. 2. Add almost all the required picture format support. This is not as trivial as it looks like. The previous implementation only support GL_a8,GL_a8r8g8b8,GL_x8r8g8b8. All the other format, we have to fallback to cpu. The reason why we can't simply add those other color format is because the exists of picture. one drawable pixmap may has one or even more container pictures. The drawable pixmap's depth can't map to a specified color format, for example depth 16 can mapped to r5g6b5, x1r5g5b5, a1r5g5b5, or even b5g6r5. So we can't get get the color format just from the depth value. But the pixmap do not has a pict_format element. We have to make a new one in the pixmap private data structure. Reroute the CreatePicture to glamor_create_picture and then store the picture's format to the pixmap's private structure. This is not an ideal solution, as there may be more than one pictures refer to the same pixmap. Then we will have trouble. There is an example in glamor_composite_with_shader. The source and mask often share the same pixmap, but use different picture format. Our current solution is to combine those two different picture formats to one which will not lose any data. Then change the source's format to this new format and then upload the pixmap to texture once. It works. If we fail to find a matched new format then we fallback. There still is a potential problem, if two pictures refer to the same pixmap, and one of them destroy the picture, but the other still remained to be used latter. We don't handle that situation currently. To be fixed. 3. Dynamic texture uploading. This is a performance feature. Although we don't like the client to hold a pixmap data to shared memory and we can't accelerate it. And even worse, we may need to fallback all the required pixmaps to cpu memory and then process them on CPU. This feature is to mitigate this penalty. When the target pixmap has a valid gl fbo attached to it. But the other pixmaps are not. Then it will be more efficient to upload the other pixmaps to GPU and then do the blitting or rendering on GPU than fallback all the pixmaps to CPU. To enable this feature, I experienced a significant performance improvement in the Game "Mines" :). 4. Debug facility. Modify the debug output mechanism. Now add a new macro: glamor_debug_output(_level_, _format_,...) to conditional output some messages according to the environment variable GLAMOR_DEBUG. We have the following levels currently. exports GLAMOR_DEBUG to 3 will enable all the above messages. 5. Changes in pixmap private data structure. Add some for the full color format supports and relate it to the pictures which already described. Also Add the following new elements: gl_fbo - to indicates whether this pixmap is on gpu only. gl_tex - to indicates whether the tex is valid and is containing the pixmap's image originally. As we bring the dynamic pixmap uploading feature, so a cpu memory pixmap may also has a valid fbo or tex attached to it. So we will have to use the above new element to check it true type. After this commit, we can pass the rendercheck testing for all the picture formats. And is much much fater than fallback to cpu when doing rendercheck testing. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2011-06-21 12:31:11 +02:00
static Bool
combine_pict_format(PictFormatShort * des, const PictFormatShort src,
const PictFormatShort mask, enum shader_in in_ca)
glamor : Add dynamic texture uploading feature. Major refactoring. 1. Rewrite the pixmap texture uploading and downloading functions. Add some new functions for both the prepare/finish access and the new performance feature dynamic texture uploading, which could download and upload the current image to/from a private texture/fbo. In the uploading or downloading phase, we need to handle two things: The first is the yInverted option, If it set, then we don't need to flip y. If not set, if it is from a dynamic texture uploading then we don't need to flip either if the current drawing process will flip it latter. If it is from finish_access, then we must flip the y axis. The second thing is the alpha channel hanlding, if the pixmap's format is something like x8a8r8g8, x1r5g5b5 which means it doesn't has alpha channel, but it do has those extra bits. Then we need to wire those bits to 1. 2. Add almost all the required picture format support. This is not as trivial as it looks like. The previous implementation only support GL_a8,GL_a8r8g8b8,GL_x8r8g8b8. All the other format, we have to fallback to cpu. The reason why we can't simply add those other color format is because the exists of picture. one drawable pixmap may has one or even more container pictures. The drawable pixmap's depth can't map to a specified color format, for example depth 16 can mapped to r5g6b5, x1r5g5b5, a1r5g5b5, or even b5g6r5. So we can't get get the color format just from the depth value. But the pixmap do not has a pict_format element. We have to make a new one in the pixmap private data structure. Reroute the CreatePicture to glamor_create_picture and then store the picture's format to the pixmap's private structure. This is not an ideal solution, as there may be more than one pictures refer to the same pixmap. Then we will have trouble. There is an example in glamor_composite_with_shader. The source and mask often share the same pixmap, but use different picture format. Our current solution is to combine those two different picture formats to one which will not lose any data. Then change the source's format to this new format and then upload the pixmap to texture once. It works. If we fail to find a matched new format then we fallback. There still is a potential problem, if two pictures refer to the same pixmap, and one of them destroy the picture, but the other still remained to be used latter. We don't handle that situation currently. To be fixed. 3. Dynamic texture uploading. This is a performance feature. Although we don't like the client to hold a pixmap data to shared memory and we can't accelerate it. And even worse, we may need to fallback all the required pixmaps to cpu memory and then process them on CPU. This feature is to mitigate this penalty. When the target pixmap has a valid gl fbo attached to it. But the other pixmaps are not. Then it will be more efficient to upload the other pixmaps to GPU and then do the blitting or rendering on GPU than fallback all the pixmaps to CPU. To enable this feature, I experienced a significant performance improvement in the Game "Mines" :). 4. Debug facility. Modify the debug output mechanism. Now add a new macro: glamor_debug_output(_level_, _format_,...) to conditional output some messages according to the environment variable GLAMOR_DEBUG. We have the following levels currently. exports GLAMOR_DEBUG to 3 will enable all the above messages. 5. Changes in pixmap private data structure. Add some for the full color format supports and relate it to the pictures which already described. Also Add the following new elements: gl_fbo - to indicates whether this pixmap is on gpu only. gl_tex - to indicates whether the tex is valid and is containing the pixmap's image originally. As we bring the dynamic pixmap uploading feature, so a cpu memory pixmap may also has a valid fbo or tex attached to it. So we will have to use the above new element to check it true type. After this commit, we can pass the rendercheck testing for all the picture formats. And is much much fater than fallback to cpu when doing rendercheck testing. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2011-06-21 12:31:11 +02:00
{
PictFormatShort new_vis;
int src_type, mask_type, src_bpp, mask_bpp;
int i;
if (src == mask) {
*des = src;
return TRUE;
}
src_bpp = PICT_FORMAT_BPP(src);
mask_bpp = PICT_FORMAT_BPP(mask);
assert(src_bpp == mask_bpp);
new_vis = PICT_FORMAT_VIS(src) | PICT_FORMAT_VIS(mask);
switch (in_ca) {
case SHADER_IN_SOURCE_ONLY:
return TRUE;
case SHADER_IN_NORMAL:
src_type = PICT_FORMAT_TYPE(src);
mask_type = PICT_TYPE_A;
break;
case SHADER_IN_CA_SOURCE:
src_type = PICT_FORMAT_TYPE(src);
mask_type = PICT_FORMAT_TYPE(mask);
break;
case SHADER_IN_CA_ALPHA:
src_type = PICT_TYPE_A;
mask_type = PICT_FORMAT_TYPE(mask);
break;
default:
return FALSE;
}
if (src_type == mask_type) {
*des = PICT_VISFORMAT(src_bpp, src_type, new_vis);
return TRUE;
}
for (i = 0;
i <
sizeof(pict_format_combine_tab) /
sizeof(pict_format_combine_tab[0]); i++) {
if ((src_type == pict_format_combine_tab[i][0]
&& mask_type == pict_format_combine_tab[i][1])
|| (src_type == pict_format_combine_tab[i][1]
&& mask_type == pict_format_combine_tab[i][0])) {
*des = PICT_VISFORMAT(src_bpp,
pict_format_combine_tab[i]
[2], new_vis);
return TRUE;
}
}
return FALSE;
}
static Bool
glamor_composite_with_shader(CARD8 op,
PicturePtr source,
PicturePtr mask,
PicturePtr dest,
int nrect, glamor_composite_rect_t * rects)
{
ScreenPtr screen = dest->pDrawable->pScreen;
glamor_screen_private *glamor_priv =
glamor_get_screen_private(screen);
glamor_gl_dispatch *dispatch = &glamor_priv->dispatch;
PixmapPtr dest_pixmap =
glamor_get_drawable_pixmap(dest->pDrawable);
PixmapPtr source_pixmap = NULL, mask_pixmap = NULL;
glamor_pixmap_private *source_pixmap_priv = NULL;
glamor_pixmap_private *mask_pixmap_priv = NULL;
glamor_pixmap_private *dest_pixmap_priv = NULL;
GLfloat dst_xscale, dst_yscale;
GLfloat mask_xscale = 1, mask_yscale = 1, src_xscale =
1, src_yscale = 1;
struct shader_key key;
glamor_composite_shader *shader;
RegionRec region;
float vertices[8], source_texcoords[8], mask_texcoords[8];
int i;
BoxPtr box;
int dest_x_off, dest_y_off;
int source_x_off, source_y_off;
int mask_x_off, mask_y_off;
enum glamor_pixmap_status source_status = GLAMOR_NONE;
enum glamor_pixmap_status mask_status = GLAMOR_NONE;
PictFormatShort saved_source_format = 0;
float src_matrix[9], mask_matrix[9];
GLfloat source_solid_color[4], mask_solid_color[4];
dest_pixmap_priv = glamor_get_pixmap_private(dest_pixmap);
if (!GLAMOR_PIXMAP_PRIV_HAS_FBO(dest_pixmap_priv)) {
glamor_fallback("dest has no fbo.\n");
goto fail;
}
memset(&key, 0, sizeof(key));
if (!source->pDrawable) {
if (source->pSourcePict->type == SourcePictTypeSolidFill) {
key.source = SHADER_SOURCE_SOLID;
glamor_get_rgba_from_pixel(source->
pSourcePict->solidFill.
color,
&source_solid_color[0],
&source_solid_color[1],
&source_solid_color[2],
&source_solid_color[3],
PICT_a8r8g8b8);
} else {
glamor_fallback("gradient source\n");
goto fail;
}
} else {
key.source = SHADER_SOURCE_TEXTURE_ALPHA;
}
if (mask) {
if (!mask->pDrawable) {
if (mask->pSourcePict->type ==
SourcePictTypeSolidFill) {
key.mask = SHADER_MASK_SOLID;
glamor_get_rgba_from_pixel
(mask->pSourcePict->solidFill.color,
&mask_solid_color[0],
&mask_solid_color[1],
&mask_solid_color[2],
&mask_solid_color[3], PICT_a8r8g8b8);
} else {
glamor_fallback("gradient mask\n");
goto fail;
}
} else {
key.mask = SHADER_MASK_TEXTURE_ALPHA;
}
if (!mask->componentAlpha) {
key.in = SHADER_IN_NORMAL;
} else {
/* We only handle two CA modes. */
if (op == PictOpAdd)
key.in = SHADER_IN_CA_SOURCE;
else if (op == PictOpOutReverse) {
key.in = SHADER_IN_CA_ALPHA;
} else {
glamor_fallback
("Unsupported component alpha op: %d\n",
op);
goto fail;
}
}
} else {
key.mask = SHADER_MASK_NONE;
key.in = SHADER_IN_SOURCE_ONLY;
}
if (source->alphaMap) {
glamor_fallback("source alphaMap\n");
goto fail;
}
if (mask && mask->alphaMap) {
glamor_fallback("mask alphaMap\n");
goto fail;
}
if (key.source == SHADER_SOURCE_TEXTURE ||
key.source == SHADER_SOURCE_TEXTURE_ALPHA) {
source_pixmap =
glamor_get_drawable_pixmap(source->pDrawable);
source_pixmap_priv =
glamor_get_pixmap_private(source_pixmap);
if (source_pixmap == dest_pixmap) {
glamor_fallback("source == dest\n");
goto fail;
}
if (!source_pixmap_priv || source_pixmap_priv->gl_fbo == 0) {
/* XXX in Xephyr, we may have gl_fbo equal to 1 but gl_tex
* equal to zero when the pixmap is screen pixmap. Then we may
* refer the tex zero directly latter in the composition.
* It seems that it works fine, but it may have potential problem*/
glamor : Add dynamic texture uploading feature. Major refactoring. 1. Rewrite the pixmap texture uploading and downloading functions. Add some new functions for both the prepare/finish access and the new performance feature dynamic texture uploading, which could download and upload the current image to/from a private texture/fbo. In the uploading or downloading phase, we need to handle two things: The first is the yInverted option, If it set, then we don't need to flip y. If not set, if it is from a dynamic texture uploading then we don't need to flip either if the current drawing process will flip it latter. If it is from finish_access, then we must flip the y axis. The second thing is the alpha channel hanlding, if the pixmap's format is something like x8a8r8g8, x1r5g5b5 which means it doesn't has alpha channel, but it do has those extra bits. Then we need to wire those bits to 1. 2. Add almost all the required picture format support. This is not as trivial as it looks like. The previous implementation only support GL_a8,GL_a8r8g8b8,GL_x8r8g8b8. All the other format, we have to fallback to cpu. The reason why we can't simply add those other color format is because the exists of picture. one drawable pixmap may has one or even more container pictures. The drawable pixmap's depth can't map to a specified color format, for example depth 16 can mapped to r5g6b5, x1r5g5b5, a1r5g5b5, or even b5g6r5. So we can't get get the color format just from the depth value. But the pixmap do not has a pict_format element. We have to make a new one in the pixmap private data structure. Reroute the CreatePicture to glamor_create_picture and then store the picture's format to the pixmap's private structure. This is not an ideal solution, as there may be more than one pictures refer to the same pixmap. Then we will have trouble. There is an example in glamor_composite_with_shader. The source and mask often share the same pixmap, but use different picture format. Our current solution is to combine those two different picture formats to one which will not lose any data. Then change the source's format to this new format and then upload the pixmap to texture once. It works. If we fail to find a matched new format then we fallback. There still is a potential problem, if two pictures refer to the same pixmap, and one of them destroy the picture, but the other still remained to be used latter. We don't handle that situation currently. To be fixed. 3. Dynamic texture uploading. This is a performance feature. Although we don't like the client to hold a pixmap data to shared memory and we can't accelerate it. And even worse, we may need to fallback all the required pixmaps to cpu memory and then process them on CPU. This feature is to mitigate this penalty. When the target pixmap has a valid gl fbo attached to it. But the other pixmaps are not. Then it will be more efficient to upload the other pixmaps to GPU and then do the blitting or rendering on GPU than fallback all the pixmaps to CPU. To enable this feature, I experienced a significant performance improvement in the Game "Mines" :). 4. Debug facility. Modify the debug output mechanism. Now add a new macro: glamor_debug_output(_level_, _format_,...) to conditional output some messages according to the environment variable GLAMOR_DEBUG. We have the following levels currently. exports GLAMOR_DEBUG to 3 will enable all the above messages. 5. Changes in pixmap private data structure. Add some for the full color format supports and relate it to the pictures which already described. Also Add the following new elements: gl_fbo - to indicates whether this pixmap is on gpu only. gl_tex - to indicates whether the tex is valid and is containing the pixmap's image originally. As we bring the dynamic pixmap uploading feature, so a cpu memory pixmap may also has a valid fbo or tex attached to it. So we will have to use the above new element to check it true type. After this commit, we can pass the rendercheck testing for all the picture formats. And is much much fater than fallback to cpu when doing rendercheck testing. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2011-06-21 12:31:11 +02:00
#ifdef GLAMOR_PIXMAP_DYNAMIC_UPLOAD
source_status = GLAMOR_UPLOAD_PENDING;
glamor : Add dynamic texture uploading feature. Major refactoring. 1. Rewrite the pixmap texture uploading and downloading functions. Add some new functions for both the prepare/finish access and the new performance feature dynamic texture uploading, which could download and upload the current image to/from a private texture/fbo. In the uploading or downloading phase, we need to handle two things: The first is the yInverted option, If it set, then we don't need to flip y. If not set, if it is from a dynamic texture uploading then we don't need to flip either if the current drawing process will flip it latter. If it is from finish_access, then we must flip the y axis. The second thing is the alpha channel hanlding, if the pixmap's format is something like x8a8r8g8, x1r5g5b5 which means it doesn't has alpha channel, but it do has those extra bits. Then we need to wire those bits to 1. 2. Add almost all the required picture format support. This is not as trivial as it looks like. The previous implementation only support GL_a8,GL_a8r8g8b8,GL_x8r8g8b8. All the other format, we have to fallback to cpu. The reason why we can't simply add those other color format is because the exists of picture. one drawable pixmap may has one or even more container pictures. The drawable pixmap's depth can't map to a specified color format, for example depth 16 can mapped to r5g6b5, x1r5g5b5, a1r5g5b5, or even b5g6r5. So we can't get get the color format just from the depth value. But the pixmap do not has a pict_format element. We have to make a new one in the pixmap private data structure. Reroute the CreatePicture to glamor_create_picture and then store the picture's format to the pixmap's private structure. This is not an ideal solution, as there may be more than one pictures refer to the same pixmap. Then we will have trouble. There is an example in glamor_composite_with_shader. The source and mask often share the same pixmap, but use different picture format. Our current solution is to combine those two different picture formats to one which will not lose any data. Then change the source's format to this new format and then upload the pixmap to texture once. It works. If we fail to find a matched new format then we fallback. There still is a potential problem, if two pictures refer to the same pixmap, and one of them destroy the picture, but the other still remained to be used latter. We don't handle that situation currently. To be fixed. 3. Dynamic texture uploading. This is a performance feature. Although we don't like the client to hold a pixmap data to shared memory and we can't accelerate it. And even worse, we may need to fallback all the required pixmaps to cpu memory and then process them on CPU. This feature is to mitigate this penalty. When the target pixmap has a valid gl fbo attached to it. But the other pixmaps are not. Then it will be more efficient to upload the other pixmaps to GPU and then do the blitting or rendering on GPU than fallback all the pixmaps to CPU. To enable this feature, I experienced a significant performance improvement in the Game "Mines" :). 4. Debug facility. Modify the debug output mechanism. Now add a new macro: glamor_debug_output(_level_, _format_,...) to conditional output some messages according to the environment variable GLAMOR_DEBUG. We have the following levels currently. exports GLAMOR_DEBUG to 3 will enable all the above messages. 5. Changes in pixmap private data structure. Add some for the full color format supports and relate it to the pictures which already described. Also Add the following new elements: gl_fbo - to indicates whether this pixmap is on gpu only. gl_tex - to indicates whether the tex is valid and is containing the pixmap's image originally. As we bring the dynamic pixmap uploading feature, so a cpu memory pixmap may also has a valid fbo or tex attached to it. So we will have to use the above new element to check it true type. After this commit, we can pass the rendercheck testing for all the picture formats. And is much much fater than fallback to cpu when doing rendercheck testing. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2011-06-21 12:31:11 +02:00
#else
glamor_fallback("no texture in source\n");
goto fail;
glamor : Add dynamic texture uploading feature. Major refactoring. 1. Rewrite the pixmap texture uploading and downloading functions. Add some new functions for both the prepare/finish access and the new performance feature dynamic texture uploading, which could download and upload the current image to/from a private texture/fbo. In the uploading or downloading phase, we need to handle two things: The first is the yInverted option, If it set, then we don't need to flip y. If not set, if it is from a dynamic texture uploading then we don't need to flip either if the current drawing process will flip it latter. If it is from finish_access, then we must flip the y axis. The second thing is the alpha channel hanlding, if the pixmap's format is something like x8a8r8g8, x1r5g5b5 which means it doesn't has alpha channel, but it do has those extra bits. Then we need to wire those bits to 1. 2. Add almost all the required picture format support. This is not as trivial as it looks like. The previous implementation only support GL_a8,GL_a8r8g8b8,GL_x8r8g8b8. All the other format, we have to fallback to cpu. The reason why we can't simply add those other color format is because the exists of picture. one drawable pixmap may has one or even more container pictures. The drawable pixmap's depth can't map to a specified color format, for example depth 16 can mapped to r5g6b5, x1r5g5b5, a1r5g5b5, or even b5g6r5. So we can't get get the color format just from the depth value. But the pixmap do not has a pict_format element. We have to make a new one in the pixmap private data structure. Reroute the CreatePicture to glamor_create_picture and then store the picture's format to the pixmap's private structure. This is not an ideal solution, as there may be more than one pictures refer to the same pixmap. Then we will have trouble. There is an example in glamor_composite_with_shader. The source and mask often share the same pixmap, but use different picture format. Our current solution is to combine those two different picture formats to one which will not lose any data. Then change the source's format to this new format and then upload the pixmap to texture once. It works. If we fail to find a matched new format then we fallback. There still is a potential problem, if two pictures refer to the same pixmap, and one of them destroy the picture, but the other still remained to be used latter. We don't handle that situation currently. To be fixed. 3. Dynamic texture uploading. This is a performance feature. Although we don't like the client to hold a pixmap data to shared memory and we can't accelerate it. And even worse, we may need to fallback all the required pixmaps to cpu memory and then process them on CPU. This feature is to mitigate this penalty. When the target pixmap has a valid gl fbo attached to it. But the other pixmaps are not. Then it will be more efficient to upload the other pixmaps to GPU and then do the blitting or rendering on GPU than fallback all the pixmaps to CPU. To enable this feature, I experienced a significant performance improvement in the Game "Mines" :). 4. Debug facility. Modify the debug output mechanism. Now add a new macro: glamor_debug_output(_level_, _format_,...) to conditional output some messages according to the environment variable GLAMOR_DEBUG. We have the following levels currently. exports GLAMOR_DEBUG to 3 will enable all the above messages. 5. Changes in pixmap private data structure. Add some for the full color format supports and relate it to the pictures which already described. Also Add the following new elements: gl_fbo - to indicates whether this pixmap is on gpu only. gl_tex - to indicates whether the tex is valid and is containing the pixmap's image originally. As we bring the dynamic pixmap uploading feature, so a cpu memory pixmap may also has a valid fbo or tex attached to it. So we will have to use the above new element to check it true type. After this commit, we can pass the rendercheck testing for all the picture formats. And is much much fater than fallback to cpu when doing rendercheck testing. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2011-06-21 12:31:11 +02:00
#endif
} else if (source_pixmap_priv->pending_op.type ==
GLAMOR_PENDING_FILL) {
key.source = SHADER_SOURCE_SOLID;
memcpy(source_solid_color,
source_pixmap_priv->pending_op.
fill.color4fv, 4 * sizeof(float));
}
}
if (key.mask == SHADER_MASK_TEXTURE ||
key.mask == SHADER_MASK_TEXTURE_ALPHA) {
mask_pixmap = glamor_get_drawable_pixmap(mask->pDrawable);
mask_pixmap_priv = glamor_get_pixmap_private(mask_pixmap);
if (mask_pixmap == dest_pixmap) {
glamor_fallback("mask == dest\n");
goto fail;
}
if (!mask_pixmap_priv || mask_pixmap_priv->gl_fbo == 0) {
glamor : Add dynamic texture uploading feature. Major refactoring. 1. Rewrite the pixmap texture uploading and downloading functions. Add some new functions for both the prepare/finish access and the new performance feature dynamic texture uploading, which could download and upload the current image to/from a private texture/fbo. In the uploading or downloading phase, we need to handle two things: The first is the yInverted option, If it set, then we don't need to flip y. If not set, if it is from a dynamic texture uploading then we don't need to flip either if the current drawing process will flip it latter. If it is from finish_access, then we must flip the y axis. The second thing is the alpha channel hanlding, if the pixmap's format is something like x8a8r8g8, x1r5g5b5 which means it doesn't has alpha channel, but it do has those extra bits. Then we need to wire those bits to 1. 2. Add almost all the required picture format support. This is not as trivial as it looks like. The previous implementation only support GL_a8,GL_a8r8g8b8,GL_x8r8g8b8. All the other format, we have to fallback to cpu. The reason why we can't simply add those other color format is because the exists of picture. one drawable pixmap may has one or even more container pictures. The drawable pixmap's depth can't map to a specified color format, for example depth 16 can mapped to r5g6b5, x1r5g5b5, a1r5g5b5, or even b5g6r5. So we can't get get the color format just from the depth value. But the pixmap do not has a pict_format element. We have to make a new one in the pixmap private data structure. Reroute the CreatePicture to glamor_create_picture and then store the picture's format to the pixmap's private structure. This is not an ideal solution, as there may be more than one pictures refer to the same pixmap. Then we will have trouble. There is an example in glamor_composite_with_shader. The source and mask often share the same pixmap, but use different picture format. Our current solution is to combine those two different picture formats to one which will not lose any data. Then change the source's format to this new format and then upload the pixmap to texture once. It works. If we fail to find a matched new format then we fallback. There still is a potential problem, if two pictures refer to the same pixmap, and one of them destroy the picture, but the other still remained to be used latter. We don't handle that situation currently. To be fixed. 3. Dynamic texture uploading. This is a performance feature. Although we don't like the client to hold a pixmap data to shared memory and we can't accelerate it. And even worse, we may need to fallback all the required pixmaps to cpu memory and then process them on CPU. This feature is to mitigate this penalty. When the target pixmap has a valid gl fbo attached to it. But the other pixmaps are not. Then it will be more efficient to upload the other pixmaps to GPU and then do the blitting or rendering on GPU than fallback all the pixmaps to CPU. To enable this feature, I experienced a significant performance improvement in the Game "Mines" :). 4. Debug facility. Modify the debug output mechanism. Now add a new macro: glamor_debug_output(_level_, _format_,...) to conditional output some messages according to the environment variable GLAMOR_DEBUG. We have the following levels currently. exports GLAMOR_DEBUG to 3 will enable all the above messages. 5. Changes in pixmap private data structure. Add some for the full color format supports and relate it to the pictures which already described. Also Add the following new elements: gl_fbo - to indicates whether this pixmap is on gpu only. gl_tex - to indicates whether the tex is valid and is containing the pixmap's image originally. As we bring the dynamic pixmap uploading feature, so a cpu memory pixmap may also has a valid fbo or tex attached to it. So we will have to use the above new element to check it true type. After this commit, we can pass the rendercheck testing for all the picture formats. And is much much fater than fallback to cpu when doing rendercheck testing. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2011-06-21 12:31:11 +02:00
#ifdef GLAMOR_PIXMAP_DYNAMIC_UPLOAD
mask_status = GLAMOR_UPLOAD_PENDING;
glamor : Add dynamic texture uploading feature. Major refactoring. 1. Rewrite the pixmap texture uploading and downloading functions. Add some new functions for both the prepare/finish access and the new performance feature dynamic texture uploading, which could download and upload the current image to/from a private texture/fbo. In the uploading or downloading phase, we need to handle two things: The first is the yInverted option, If it set, then we don't need to flip y. If not set, if it is from a dynamic texture uploading then we don't need to flip either if the current drawing process will flip it latter. If it is from finish_access, then we must flip the y axis. The second thing is the alpha channel hanlding, if the pixmap's format is something like x8a8r8g8, x1r5g5b5 which means it doesn't has alpha channel, but it do has those extra bits. Then we need to wire those bits to 1. 2. Add almost all the required picture format support. This is not as trivial as it looks like. The previous implementation only support GL_a8,GL_a8r8g8b8,GL_x8r8g8b8. All the other format, we have to fallback to cpu. The reason why we can't simply add those other color format is because the exists of picture. one drawable pixmap may has one or even more container pictures. The drawable pixmap's depth can't map to a specified color format, for example depth 16 can mapped to r5g6b5, x1r5g5b5, a1r5g5b5, or even b5g6r5. So we can't get get the color format just from the depth value. But the pixmap do not has a pict_format element. We have to make a new one in the pixmap private data structure. Reroute the CreatePicture to glamor_create_picture and then store the picture's format to the pixmap's private structure. This is not an ideal solution, as there may be more than one pictures refer to the same pixmap. Then we will have trouble. There is an example in glamor_composite_with_shader. The source and mask often share the same pixmap, but use different picture format. Our current solution is to combine those two different picture formats to one which will not lose any data. Then change the source's format to this new format and then upload the pixmap to texture once. It works. If we fail to find a matched new format then we fallback. There still is a potential problem, if two pictures refer to the same pixmap, and one of them destroy the picture, but the other still remained to be used latter. We don't handle that situation currently. To be fixed. 3. Dynamic texture uploading. This is a performance feature. Although we don't like the client to hold a pixmap data to shared memory and we can't accelerate it. And even worse, we may need to fallback all the required pixmaps to cpu memory and then process them on CPU. This feature is to mitigate this penalty. When the target pixmap has a valid gl fbo attached to it. But the other pixmaps are not. Then it will be more efficient to upload the other pixmaps to GPU and then do the blitting or rendering on GPU than fallback all the pixmaps to CPU. To enable this feature, I experienced a significant performance improvement in the Game "Mines" :). 4. Debug facility. Modify the debug output mechanism. Now add a new macro: glamor_debug_output(_level_, _format_,...) to conditional output some messages according to the environment variable GLAMOR_DEBUG. We have the following levels currently. exports GLAMOR_DEBUG to 3 will enable all the above messages. 5. Changes in pixmap private data structure. Add some for the full color format supports and relate it to the pictures which already described. Also Add the following new elements: gl_fbo - to indicates whether this pixmap is on gpu only. gl_tex - to indicates whether the tex is valid and is containing the pixmap's image originally. As we bring the dynamic pixmap uploading feature, so a cpu memory pixmap may also has a valid fbo or tex attached to it. So we will have to use the above new element to check it true type. After this commit, we can pass the rendercheck testing for all the picture formats. And is much much fater than fallback to cpu when doing rendercheck testing. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2011-06-21 12:31:11 +02:00
#else
glamor_fallback("no texture in mask\n");
goto fail;
glamor : Add dynamic texture uploading feature. Major refactoring. 1. Rewrite the pixmap texture uploading and downloading functions. Add some new functions for both the prepare/finish access and the new performance feature dynamic texture uploading, which could download and upload the current image to/from a private texture/fbo. In the uploading or downloading phase, we need to handle two things: The first is the yInverted option, If it set, then we don't need to flip y. If not set, if it is from a dynamic texture uploading then we don't need to flip either if the current drawing process will flip it latter. If it is from finish_access, then we must flip the y axis. The second thing is the alpha channel hanlding, if the pixmap's format is something like x8a8r8g8, x1r5g5b5 which means it doesn't has alpha channel, but it do has those extra bits. Then we need to wire those bits to 1. 2. Add almost all the required picture format support. This is not as trivial as it looks like. The previous implementation only support GL_a8,GL_a8r8g8b8,GL_x8r8g8b8. All the other format, we have to fallback to cpu. The reason why we can't simply add those other color format is because the exists of picture. one drawable pixmap may has one or even more container pictures. The drawable pixmap's depth can't map to a specified color format, for example depth 16 can mapped to r5g6b5, x1r5g5b5, a1r5g5b5, or even b5g6r5. So we can't get get the color format just from the depth value. But the pixmap do not has a pict_format element. We have to make a new one in the pixmap private data structure. Reroute the CreatePicture to glamor_create_picture and then store the picture's format to the pixmap's private structure. This is not an ideal solution, as there may be more than one pictures refer to the same pixmap. Then we will have trouble. There is an example in glamor_composite_with_shader. The source and mask often share the same pixmap, but use different picture format. Our current solution is to combine those two different picture formats to one which will not lose any data. Then change the source's format to this new format and then upload the pixmap to texture once. It works. If we fail to find a matched new format then we fallback. There still is a potential problem, if two pictures refer to the same pixmap, and one of them destroy the picture, but the other still remained to be used latter. We don't handle that situation currently. To be fixed. 3. Dynamic texture uploading. This is a performance feature. Although we don't like the client to hold a pixmap data to shared memory and we can't accelerate it. And even worse, we may need to fallback all the required pixmaps to cpu memory and then process them on CPU. This feature is to mitigate this penalty. When the target pixmap has a valid gl fbo attached to it. But the other pixmaps are not. Then it will be more efficient to upload the other pixmaps to GPU and then do the blitting or rendering on GPU than fallback all the pixmaps to CPU. To enable this feature, I experienced a significant performance improvement in the Game "Mines" :). 4. Debug facility. Modify the debug output mechanism. Now add a new macro: glamor_debug_output(_level_, _format_,...) to conditional output some messages according to the environment variable GLAMOR_DEBUG. We have the following levels currently. exports GLAMOR_DEBUG to 3 will enable all the above messages. 5. Changes in pixmap private data structure. Add some for the full color format supports and relate it to the pictures which already described. Also Add the following new elements: gl_fbo - to indicates whether this pixmap is on gpu only. gl_tex - to indicates whether the tex is valid and is containing the pixmap's image originally. As we bring the dynamic pixmap uploading feature, so a cpu memory pixmap may also has a valid fbo or tex attached to it. So we will have to use the above new element to check it true type. After this commit, we can pass the rendercheck testing for all the picture formats. And is much much fater than fallback to cpu when doing rendercheck testing. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2011-06-21 12:31:11 +02:00
#endif
} else if (mask_pixmap_priv->pending_op.type ==
GLAMOR_PENDING_FILL) {
key.mask = SHADER_MASK_SOLID;
memcpy(mask_solid_color,
mask_pixmap_priv->pending_op.fill.color4fv,
4 * sizeof(float));
}
}
glamor : Add dynamic texture uploading feature. Major refactoring. 1. Rewrite the pixmap texture uploading and downloading functions. Add some new functions for both the prepare/finish access and the new performance feature dynamic texture uploading, which could download and upload the current image to/from a private texture/fbo. In the uploading or downloading phase, we need to handle two things: The first is the yInverted option, If it set, then we don't need to flip y. If not set, if it is from a dynamic texture uploading then we don't need to flip either if the current drawing process will flip it latter. If it is from finish_access, then we must flip the y axis. The second thing is the alpha channel hanlding, if the pixmap's format is something like x8a8r8g8, x1r5g5b5 which means it doesn't has alpha channel, but it do has those extra bits. Then we need to wire those bits to 1. 2. Add almost all the required picture format support. This is not as trivial as it looks like. The previous implementation only support GL_a8,GL_a8r8g8b8,GL_x8r8g8b8. All the other format, we have to fallback to cpu. The reason why we can't simply add those other color format is because the exists of picture. one drawable pixmap may has one or even more container pictures. The drawable pixmap's depth can't map to a specified color format, for example depth 16 can mapped to r5g6b5, x1r5g5b5, a1r5g5b5, or even b5g6r5. So we can't get get the color format just from the depth value. But the pixmap do not has a pict_format element. We have to make a new one in the pixmap private data structure. Reroute the CreatePicture to glamor_create_picture and then store the picture's format to the pixmap's private structure. This is not an ideal solution, as there may be more than one pictures refer to the same pixmap. Then we will have trouble. There is an example in glamor_composite_with_shader. The source and mask often share the same pixmap, but use different picture format. Our current solution is to combine those two different picture formats to one which will not lose any data. Then change the source's format to this new format and then upload the pixmap to texture once. It works. If we fail to find a matched new format then we fallback. There still is a potential problem, if two pictures refer to the same pixmap, and one of them destroy the picture, but the other still remained to be used latter. We don't handle that situation currently. To be fixed. 3. Dynamic texture uploading. This is a performance feature. Although we don't like the client to hold a pixmap data to shared memory and we can't accelerate it. And even worse, we may need to fallback all the required pixmaps to cpu memory and then process them on CPU. This feature is to mitigate this penalty. When the target pixmap has a valid gl fbo attached to it. But the other pixmaps are not. Then it will be more efficient to upload the other pixmaps to GPU and then do the blitting or rendering on GPU than fallback all the pixmaps to CPU. To enable this feature, I experienced a significant performance improvement in the Game "Mines" :). 4. Debug facility. Modify the debug output mechanism. Now add a new macro: glamor_debug_output(_level_, _format_,...) to conditional output some messages according to the environment variable GLAMOR_DEBUG. We have the following levels currently. exports GLAMOR_DEBUG to 3 will enable all the above messages. 5. Changes in pixmap private data structure. Add some for the full color format supports and relate it to the pictures which already described. Also Add the following new elements: gl_fbo - to indicates whether this pixmap is on gpu only. gl_tex - to indicates whether the tex is valid and is containing the pixmap's image originally. As we bring the dynamic pixmap uploading feature, so a cpu memory pixmap may also has a valid fbo or tex attached to it. So we will have to use the above new element to check it true type. After this commit, we can pass the rendercheck testing for all the picture formats. And is much much fater than fallback to cpu when doing rendercheck testing. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2011-06-21 12:31:11 +02:00
#ifdef GLAMOR_PIXMAP_DYNAMIC_UPLOAD
if (source_status == GLAMOR_UPLOAD_PENDING
&& mask_status == GLAMOR_UPLOAD_PENDING
&& source_pixmap == mask_pixmap) {
if (source->format != mask->format) {
saved_source_format = source->format;
if (!combine_pict_format
(&source->format, source->format,
mask->format, key.in)) {
glamor_fallback
("combine source %x mask %x failed.\n",
source->format, mask->format);
goto fail;
}
if (source->format != saved_source_format) {
glamor_picture_format_fixup(source,
source_pixmap_priv);
}
/* XXX
* By default, glamor_upload_picture_to_texture will wire alpha to 1
* if one picture doesn't have alpha. So we don't do that again in
* rendering function. But here is a special case, as source and
* mask share the same texture but may have different formats. For
* example, source doesn't have alpha, but mask has alpha. Then the
* texture will have the alpha value for the mask. And will not wire
* to 1 for the source. In this case, we have to use different shader
* to wire the source's alpha to 1.
*
* But this may cause a potential problem if the source's repeat mode
* is REPEAT_NONE, and if the source is smaller than the dest, then
* for the region not covered by the source may be painted incorrectly.
* because we wire the alpha to 1.
*
**/
if (!PICT_FORMAT_A(saved_source_format)
&& PICT_FORMAT_A(mask->format))
key.source = SHADER_SOURCE_TEXTURE;
if (!PICT_FORMAT_A(mask->format)
&& PICT_FORMAT_A(saved_source_format))
key.mask = SHADER_MASK_TEXTURE;
mask_status = GLAMOR_NONE;
}
source_status = glamor_upload_picture_to_texture(source);
if (source_status != GLAMOR_UPLOAD_DONE) {
glamor_fallback
("Failed to upload source texture.\n");
goto fail;
}
} else {
if (source_status == GLAMOR_UPLOAD_PENDING) {
source_status =
glamor_upload_picture_to_texture(source);
if (source_status != GLAMOR_UPLOAD_DONE) {
glamor_fallback
("Failed to upload source texture.\n");
goto fail;
}
}
if (mask_status == GLAMOR_UPLOAD_PENDING) {
mask_status =
glamor_upload_picture_to_texture(mask);
if (mask_status != GLAMOR_UPLOAD_DONE) {
glamor_fallback
("Failed to upload mask texture.\n");
goto fail;
}
}
}
glamor : Add dynamic texture uploading feature. Major refactoring. 1. Rewrite the pixmap texture uploading and downloading functions. Add some new functions for both the prepare/finish access and the new performance feature dynamic texture uploading, which could download and upload the current image to/from a private texture/fbo. In the uploading or downloading phase, we need to handle two things: The first is the yInverted option, If it set, then we don't need to flip y. If not set, if it is from a dynamic texture uploading then we don't need to flip either if the current drawing process will flip it latter. If it is from finish_access, then we must flip the y axis. The second thing is the alpha channel hanlding, if the pixmap's format is something like x8a8r8g8, x1r5g5b5 which means it doesn't has alpha channel, but it do has those extra bits. Then we need to wire those bits to 1. 2. Add almost all the required picture format support. This is not as trivial as it looks like. The previous implementation only support GL_a8,GL_a8r8g8b8,GL_x8r8g8b8. All the other format, we have to fallback to cpu. The reason why we can't simply add those other color format is because the exists of picture. one drawable pixmap may has one or even more container pictures. The drawable pixmap's depth can't map to a specified color format, for example depth 16 can mapped to r5g6b5, x1r5g5b5, a1r5g5b5, or even b5g6r5. So we can't get get the color format just from the depth value. But the pixmap do not has a pict_format element. We have to make a new one in the pixmap private data structure. Reroute the CreatePicture to glamor_create_picture and then store the picture's format to the pixmap's private structure. This is not an ideal solution, as there may be more than one pictures refer to the same pixmap. Then we will have trouble. There is an example in glamor_composite_with_shader. The source and mask often share the same pixmap, but use different picture format. Our current solution is to combine those two different picture formats to one which will not lose any data. Then change the source's format to this new format and then upload the pixmap to texture once. It works. If we fail to find a matched new format then we fallback. There still is a potential problem, if two pictures refer to the same pixmap, and one of them destroy the picture, but the other still remained to be used latter. We don't handle that situation currently. To be fixed. 3. Dynamic texture uploading. This is a performance feature. Although we don't like the client to hold a pixmap data to shared memory and we can't accelerate it. And even worse, we may need to fallback all the required pixmaps to cpu memory and then process them on CPU. This feature is to mitigate this penalty. When the target pixmap has a valid gl fbo attached to it. But the other pixmaps are not. Then it will be more efficient to upload the other pixmaps to GPU and then do the blitting or rendering on GPU than fallback all the pixmaps to CPU. To enable this feature, I experienced a significant performance improvement in the Game "Mines" :). 4. Debug facility. Modify the debug output mechanism. Now add a new macro: glamor_debug_output(_level_, _format_,...) to conditional output some messages according to the environment variable GLAMOR_DEBUG. We have the following levels currently. exports GLAMOR_DEBUG to 3 will enable all the above messages. 5. Changes in pixmap private data structure. Add some for the full color format supports and relate it to the pictures which already described. Also Add the following new elements: gl_fbo - to indicates whether this pixmap is on gpu only. gl_tex - to indicates whether the tex is valid and is containing the pixmap's image originally. As we bring the dynamic pixmap uploading feature, so a cpu memory pixmap may also has a valid fbo or tex attached to it. So we will have to use the above new element to check it true type. After this commit, we can pass the rendercheck testing for all the picture formats. And is much much fater than fallback to cpu when doing rendercheck testing. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2011-06-21 12:31:11 +02:00
#endif
glamor_set_destination_pixmap_priv_nc(dest_pixmap_priv);
glamor_validate_pixmap(dest_pixmap);
if (!glamor_set_composite_op(screen, op, dest, mask)) {
goto fail;
}
shader = glamor_lookup_composite_shader(screen, &key);
if (shader->prog == 0) {
glamor_fallback
("no shader program for this render acccel mode\n");
goto fail;
}
dispatch->glUseProgram(shader->prog);
if (key.source == SHADER_SOURCE_SOLID) {
glamor_set_composite_solid(dispatch, source_solid_color,
shader->source_uniform_location);
} else {
glamor_set_composite_texture(screen, 0, source,
source_pixmap_priv);
}
if (key.mask != SHADER_MASK_NONE) {
if (key.mask == SHADER_MASK_SOLID) {
glamor_set_composite_solid(dispatch,
mask_solid_color,
shader->mask_uniform_location);
} else {
glamor_set_composite_texture(screen, 1, mask,
mask_pixmap_priv);
}
}
glamor_priv->has_source_coords = key.source != SHADER_SOURCE_SOLID;
glamor_priv->has_mask_coords = (key.mask != SHADER_MASK_NONE &&
key.mask != SHADER_MASK_SOLID);
glamor_get_drawable_deltas(dest->pDrawable, dest_pixmap,
&dest_x_off, &dest_y_off);
pixmap_priv_get_scale(dest_pixmap_priv, &dst_xscale, &dst_yscale);
if (glamor_priv->has_source_coords) {
glamor_get_drawable_deltas(source->pDrawable,
source_pixmap, &source_x_off,
&source_y_off);
pixmap_priv_get_scale(source_pixmap_priv, &src_xscale,
&src_yscale);
glamor_picture_get_matrixf(source, src_matrix);
}
if (glamor_priv->has_mask_coords) {
glamor_get_drawable_deltas(mask->pDrawable, mask_pixmap,
&mask_x_off, &mask_y_off);
pixmap_priv_get_scale(mask_pixmap_priv, &mask_xscale,
&mask_yscale);
glamor_picture_get_matrixf(mask, mask_matrix);
}
while (nrect--) {
INT16 x_source;
INT16 y_source;
INT16 x_mask;
INT16 y_mask;
INT16 x_dest;
INT16 y_dest;
CARD16 width;
CARD16 height;
x_dest = rects->x_dst;
y_dest = rects->y_dst;
x_source = rects->x_src;
y_source = rects->y_src;
x_mask = rects->x_mask;
y_mask = rects->y_mask;
width = rects->width;
height = rects->height;
x_dest += dest->pDrawable->x;
y_dest += dest->pDrawable->y;
if (source->pDrawable) {
x_source += source->pDrawable->x;
y_source += source->pDrawable->y;
}
if (mask && mask->pDrawable) {
x_mask += mask->pDrawable->x;
y_mask += mask->pDrawable->y;
}
if (!miComputeCompositeRegion(&region,
source, mask, dest,
x_source, y_source,
x_mask, y_mask,
x_dest, y_dest, width,
height))
continue;
x_source += source_x_off;
y_source += source_y_off;
x_mask += mask_x_off;
y_mask += mask_y_off;
box = REGION_RECTS(&region);
for (i = 0; i < REGION_NUM_RECTS(&region); i++) {
int vx1 = box[i].x1 + dest_x_off;
int vx2 = box[i].x2 + dest_x_off;
int vy1 = box[i].y1 + dest_y_off;
int vy2 = box[i].y2 + dest_y_off;
glamor_set_normalize_vcoords(dst_xscale,
dst_yscale, vx1,
vy1, vx2, vy2,
glamor_priv->yInverted,
vertices);
if (key.source != SHADER_SOURCE_SOLID) {
int tx1 = box[i].x1 + x_source - x_dest;
int ty1 = box[i].y1 + y_source - y_dest;
int tx2 = box[i].x2 + x_source - x_dest;
int ty2 = box[i].y2 + y_source - y_dest;
if (source->transform)
glamor_set_transformed_normalize_tcoords
(src_matrix, src_xscale,
src_yscale, tx1, ty1,
tx2, ty2,
glamor_priv->yInverted,
source_texcoords);
else
glamor_set_normalize_tcoords
(src_xscale, src_yscale,
tx1, ty1, tx2, ty2,
glamor_priv->yInverted,
source_texcoords);
}
if (key.mask != SHADER_MASK_NONE
&& key.mask != SHADER_MASK_SOLID) {
float tx1 = box[i].x1 + x_mask - x_dest;
float ty1 = box[i].y1 + y_mask - y_dest;
float tx2 = box[i].x2 + x_mask - x_dest;
float ty2 = box[i].y2 + y_mask - y_dest;
if (mask->transform)
glamor_set_transformed_normalize_tcoords
(mask_matrix,
mask_xscale,
mask_yscale, tx1, ty1,
tx2, ty2,
glamor_priv->yInverted,
mask_texcoords);
else
glamor_set_normalize_tcoords
(mask_xscale,
mask_yscale, tx1, ty1,
tx2, ty2,
glamor_priv->yInverted,
mask_texcoords);
}
glamor_emit_composite_rect(screen,
source_texcoords,
mask_texcoords,
vertices);
}
rects++;
}
glamor_flush_composite_rects(screen);
dispatch->glBindBuffer(GL_ARRAY_BUFFER, 0);
dispatch->glDisableVertexAttribArray(GLAMOR_VERTEX_POS);
dispatch->glDisableVertexAttribArray(GLAMOR_VERTEX_SOURCE);
dispatch->glDisableVertexAttribArray(GLAMOR_VERTEX_MASK);
REGION_UNINIT(dst->pDrawable->pScreen, &region);
dispatch->glDisable(GL_BLEND);
#ifndef GLAMOR_GLES2
dispatch->glActiveTexture(GL_TEXTURE0);
dispatch->glDisable(GL_TEXTURE_2D);
dispatch->glActiveTexture(GL_TEXTURE1);
dispatch->glDisable(GL_TEXTURE_2D);
#endif
dispatch->glUseProgram(0);
if (saved_source_format)
source->format = saved_source_format;
return TRUE;
fail:
if (saved_source_format)
source->format = saved_source_format;
dispatch->glDisable(GL_BLEND);
dispatch->glUseProgram(0);
return FALSE;
}
static PicturePtr
glamor_convert_gradient_picture(ScreenPtr screen,
PicturePtr source,
int x_source,
int y_source, int width, int height)
{
PixmapPtr pixmap;
PicturePtr dst;
int error;
PictFormatShort format;
if (!source->pDrawable)
format = PICT_a8r8g8b8;
else
format = source->format;
pixmap = screen->CreatePixmap(screen,
width,
height,
PIXMAN_FORMAT_DEPTH(format),
GLAMOR_CREATE_PIXMAP_CPU);
if (!pixmap)
return NULL;
dst = CreatePicture(0,
&pixmap->drawable,
PictureMatchFormat(screen,
PIXMAN_FORMAT_DEPTH(format),
format),
0, 0, serverClient, &error);
screen->DestroyPixmap(pixmap);
if (!dst)
return NULL;
ValidatePicture(dst);
fbComposite(PictOpSrc, source, NULL, dst, x_source, y_source,
0, 0, 0, 0, width, height);
return dst;
}
void
glamor_composite(CARD8 op,
PicturePtr source,
PicturePtr mask,
PicturePtr dest,
INT16 x_source,
INT16 y_source,
INT16 x_mask,
INT16 y_mask,
INT16 x_dest, INT16 y_dest, CARD16 width, CARD16 height)
{
ScreenPtr screen = dest->pDrawable->pScreen;
glamor_pixmap_private *dest_pixmap_priv;
glamor_pixmap_private *source_pixmap_priv =
NULL, *mask_pixmap_priv = NULL;
PixmapPtr dest_pixmap =
glamor_get_drawable_pixmap(dest->pDrawable);
PixmapPtr source_pixmap = NULL, mask_pixmap = NULL;
PicturePtr temp_src = source, temp_mask = mask;
int x_temp_src, y_temp_src, x_temp_mask, y_temp_mask;
glamor_composite_rect_t rect;
glamor_screen_private *glamor_priv =
glamor_get_screen_private(screen);
glamor_gl_dispatch *dispatch = &glamor_priv->dispatch;
x_temp_src = x_source;
y_temp_src = y_source;
x_temp_mask = x_mask;
y_temp_mask = y_mask;
dest_pixmap_priv = glamor_get_pixmap_private(dest_pixmap);
/* Currently. Always fallback to cpu if destination is in CPU memory. */
if (!GLAMOR_PIXMAP_PRIV_HAS_FBO(dest_pixmap_priv)) {
goto fail;
}
if (source->pDrawable) {
source_pixmap =
glamor_get_drawable_pixmap(source->pDrawable);
source_pixmap_priv =
glamor_get_pixmap_private(source_pixmap);
}
if (mask && mask->pDrawable) {
mask_pixmap = glamor_get_drawable_pixmap(mask->pDrawable);
mask_pixmap_priv = glamor_get_pixmap_private(mask_pixmap);
}
if ((!source->pDrawable
&& (source->pSourcePict->type != SourcePictTypeSolidFill))
|| (source->pDrawable
&& !GLAMOR_PIXMAP_PRIV_HAS_FBO(source_pixmap_priv)
&&
((width * height * 4 <
(source_pixmap->drawable.width *
source_pixmap->drawable.height))
||
!(glamor_check_fbo_size
(glamor_priv, source_pixmap->drawable.width,
source_pixmap->drawable.height))))) {
temp_src =
glamor_convert_gradient_picture(screen, source,
x_source, y_source,
width, height);
if (!temp_src) {
temp_src = source;
goto fail;
}
x_temp_src = y_temp_src = 0;
}
if (mask
&&
((!mask->pDrawable
&& (mask->pSourcePict->type != SourcePictTypeSolidFill))
|| (mask->pDrawable
&& (!GLAMOR_PIXMAP_PRIV_HAS_FBO(mask_pixmap_priv))
&&
((width * height * 4 <
(mask_pixmap->drawable.width *
mask_pixmap->drawable.height))
||
!(glamor_check_fbo_size
(glamor_priv, mask_pixmap->drawable.width,
mask_pixmap->drawable.height)))))) {
/* XXX if mask->pDrawable is the same as source->pDrawable, we have an opportunity
* to do reduce one convertion. */
temp_mask =
glamor_convert_gradient_picture(screen, mask,
x_mask, y_mask,
width, height);
if (!temp_mask) {
temp_mask = mask;
goto fail;
}
x_temp_mask = y_temp_mask = 0;
}
/* Do two-pass PictOpOver componentAlpha, until we enable
* dual source color blending.
*/
if (mask && mask->componentAlpha) {
if (op == PictOpOver) {
glamor_composite(PictOpOutReverse,
temp_src, temp_mask, dest,
x_temp_src, y_temp_src,
x_temp_mask, y_temp_mask,
x_dest, y_dest, width, height);
glamor_composite(PictOpAdd,
temp_src, temp_mask, dest,
x_temp_src, y_temp_src,
x_temp_mask, y_temp_mask,
x_dest, y_dest, width, height);
goto done;
} else if (op != PictOpAdd && op != PictOpOutReverse) {
glamor_fallback
("glamor_composite(): component alpha\n");
goto fail;
}
}
if (!mask) {
if (glamor_composite_with_copy(op, temp_src, dest,
x_temp_src, y_temp_src,
x_dest, y_dest, width,
height))
goto done;
}
rect.x_src = x_temp_src;
rect.y_src = y_temp_src;
rect.x_mask = x_temp_mask;
rect.y_mask = y_temp_mask;
rect.x_dst = x_dest;
rect.y_dst = y_dest;
rect.width = width;
rect.height = height;
if (glamor_composite_with_shader
(op, temp_src, temp_mask, dest, 1, &rect))
goto done;
fail:
glamor_fallback
("from picts %p:%p %dx%d / %p:%p %d x %d (%c,%c) to pict %p:%p %dx%d (%c)\n",
source, source->pDrawable,
source->pDrawable ? source->pDrawable->width : 0,
source->pDrawable ? source->pDrawable->height : 0, mask,
(!mask) ? NULL : mask->pDrawable, (!mask
|| !mask->pDrawable) ? 0 :
mask->pDrawable->width, (!mask
|| !mask->
pDrawable) ? 0 : mask->pDrawable->
height, glamor_get_picture_location(source),
glamor_get_picture_location(mask), dest, dest->pDrawable,
dest->pDrawable->width, dest->pDrawable->height,
glamor_get_picture_location(dest));
dispatch->glUseProgram(0);
dispatch->glDisable(GL_BLEND);
if (glamor_prepare_access_picture(dest, GLAMOR_ACCESS_RW)) {
if (glamor_prepare_access_picture
(source, GLAMOR_ACCESS_RO)) {
if (!mask
|| glamor_prepare_access_picture(mask,
GLAMOR_ACCESS_RO))
{
fbComposite(op,
source, mask, dest,
x_source, y_source,
x_mask, y_mask, x_dest,
y_dest, width, height);
if (mask)
glamor_finish_access_picture(mask);
}
glamor_finish_access_picture(source);
}
glamor_finish_access_picture(dest);
}
done:
if (temp_src != source)
FreePicture(temp_src, 0);
if (temp_mask != mask)
FreePicture(temp_mask, 0);
}
/**
* Creates an appropriate picture to upload our alpha mask into (which
* we calculated in system memory)
*/
static PicturePtr
glamor_create_mask_picture(ScreenPtr screen,
PicturePtr dst,
PictFormatPtr pict_format,
CARD16 width, CARD16 height)
{
PixmapPtr pixmap;
PicturePtr picture;
int error;
if (!pict_format) {
if (dst->polyEdge == PolyEdgeSharp)
pict_format =
PictureMatchFormat(screen, 1, PICT_a1);
else
pict_format =
PictureMatchFormat(screen, 8, PICT_a8);
if (!pict_format)
return 0;
}
pixmap = screen->CreatePixmap(screen, 0, 0,
pict_format->depth,
GLAMOR_CREATE_PIXMAP_CPU);
if (!pixmap)
return 0;
picture = CreatePicture(0, &pixmap->drawable, pict_format,
0, 0, serverClient, &error);
screen->DestroyPixmap(pixmap);
return picture;
}
/**
* glamor_trapezoids is a copy of miTrapezoids that does all the trapezoid
* accumulation in system memory.
*/
void
glamor_trapezoids(CARD8 op,
PicturePtr src, PicturePtr dst,
PictFormatPtr mask_format, INT16 x_src, INT16 y_src,
int ntrap, xTrapezoid * traps)
{
ScreenPtr screen = dst->pDrawable->pScreen;
BoxRec bounds;
PicturePtr picture;
INT16 x_dst, y_dst;
INT16 x_rel, y_rel;
int width, height, stride;
PixmapPtr pixmap;
pixman_image_t *image;
/* If a mask format wasn't provided, we get to choose, but behavior should
* be as if there was no temporary mask the traps were accumulated into.
*/
if (!mask_format) {
if (dst->polyEdge == PolyEdgeSharp)
mask_format =
PictureMatchFormat(screen, 1, PICT_a1);
else
mask_format =
PictureMatchFormat(screen, 8, PICT_a8);
for (; ntrap; ntrap--, traps++)
glamor_trapezoids(op, src, dst, mask_format, x_src,
y_src, 1, traps);
return;
}
miTrapezoidBounds(ntrap, traps, &bounds);
if (bounds.y1 >= bounds.y2 || bounds.x1 >= bounds.x2)
return;
x_dst = traps[0].left.p1.x >> 16;
y_dst = traps[0].left.p1.y >> 16;
width = bounds.x2 - bounds.x1;
height = bounds.y2 - bounds.y1;
stride = PixmapBytePad(width, mask_format->depth);
picture = glamor_create_mask_picture(screen, dst, mask_format,
width, height);
if (!picture)
return;
image = pixman_image_create_bits(picture->format,
width, height, NULL, stride);
if (!image) {
FreePicture(picture, 0);
return;
}
for (; ntrap; ntrap--, traps++)
pixman_rasterize_trapezoid(image,
(pixman_trapezoid_t *) traps,
-bounds.x1, -bounds.y1);
pixmap = glamor_get_drawable_pixmap(picture->pDrawable);
screen->ModifyPixmapHeader(pixmap, width, height,
mask_format->depth,
BitsPerPixel(mask_format->depth),
PixmapBytePad(width,
mask_format->depth),
pixman_image_get_data(image));
x_rel = bounds.x1 + x_src - x_dst;
y_rel = bounds.y1 + y_src - y_dst;
CompositePicture(op, src, picture, dst,
x_rel, y_rel,
0, 0,
bounds.x1, bounds.y1,
bounds.x2 - bounds.x1, bounds.y2 - bounds.y1);
pixman_image_unref(image);
FreePicture(picture, 0);
}
void
glamor_composite_rects(CARD8 op,
PicturePtr src, PicturePtr mask, PicturePtr dst,
int nrect, glamor_composite_rect_t * rects)
{
int n;
glamor_composite_rect_t *r;
ValidatePicture(src);
ValidatePicture(dst);
if (glamor_composite_with_shader(op, src, mask, dst, nrect, rects))
return;
n = nrect;
r = rects;
while (n--) {
CompositePicture(op,
src,
mask,
dst,
r->x_src, r->y_src,
r->x_mask, r->y_mask,
r->x_dst, r->y_dst, r->width, r->height);
r++;
}
}
#endif /* RENDER */