xserver-multidpi/glamor/glamor_pixmap.c

495 lines
13 KiB
C
Raw Normal View History

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 HAVE_DIX_CONFIG_H
#include <dix-config.h>
#endif
#include <stdlib.h>
#include "glamor_priv.h"
/**
* Sets the offsets to add to coordinates to make them address the same bits in
* the backing drawable. These coordinates are nonzero only for redirected
* windows.
*/
void
glamor_get_drawable_deltas(DrawablePtr drawable, PixmapPtr pixmap,
int *x, int *y)
{
#ifdef COMPOSITE
if (drawable->type == DRAWABLE_WINDOW) {
*x = -pixmap->screen_x;
*y = -pixmap->screen_y;
return;
}
#endif
*x = 0;
*y = 0;
}
void
glamor_set_destination_pixmap_priv_nc(glamor_pixmap_private *pixmap_priv)
{
glBindFramebufferEXT(GL_FRAMEBUFFER_EXT, pixmap_priv->fb);
glViewport(0, 0,
pixmap_priv->container->drawable.width,
pixmap_priv->container->drawable.height);
}
int
glamor_set_destination_pixmap_priv(glamor_pixmap_private *pixmap_priv)
{
if (!GLAMOR_PIXMAP_PRIV_HAS_FBO(pixmap_priv))
return -1;
glamor_set_destination_pixmap_priv_nc(pixmap_priv);
return 0;
}
int
glamor_set_destination_pixmap(PixmapPtr pixmap)
{
glamor_pixmap_private *pixmap_priv = glamor_get_pixmap_private(pixmap);
return glamor_set_destination_pixmap_priv(pixmap_priv);
}
Bool
glamor_set_planemask(PixmapPtr pixmap, unsigned long planemask)
{
if (glamor_pm_is_solid(&pixmap->drawable, planemask)) {
return GL_TRUE;
}
glamor_fallback("unsupported planemask %lx\n", planemask);
return GL_FALSE;
}
void
glamor_set_alu(unsigned char alu)
{
if (alu == GXcopy) {
glDisable(GL_COLOR_LOGIC_OP);
return;
}
glEnable(GL_COLOR_LOGIC_OP);
switch (alu) {
case GXclear:
glLogicOp(GL_CLEAR);
break;
case GXand:
glLogicOp(GL_AND);
break;
case GXandReverse:
glLogicOp(GL_AND_REVERSE);
break;
case GXandInverted:
glLogicOp(GL_AND_INVERTED);
break;
case GXnoop:
glLogicOp(GL_NOOP);
break;
case GXxor:
glLogicOp(GL_XOR);
break;
case GXor:
glLogicOp(GL_OR);
break;
case GXnor:
glLogicOp(GL_NOR);
break;
case GXequiv:
glLogicOp(GL_EQUIV);
break;
case GXinvert:
glLogicOp(GL_INVERT);
break;
case GXorReverse:
glLogicOp(GL_OR_REVERSE);
break;
case GXcopyInverted:
glLogicOp(GL_COPY_INVERTED);
break;
case GXorInverted:
glLogicOp(GL_OR_INVERTED);
break;
case GXnand:
glLogicOp(GL_NAND);
break;
case GXset:
glLogicOp(GL_SET);
break;
default:
FatalError("unknown logic op\n");
}
}
/**
* Upload pixmap to a specified texture.
* This texture may not be the one attached to it.
**/
static void
__glamor_upload_pixmap_to_texture(PixmapPtr pixmap, GLenum format, GLenum type, GLuint tex)
{
glamor_pixmap_private *pixmap_priv = glamor_get_pixmap_private(pixmap);
unsigned int stride, row_length;
void *texels;
GLenum iformat = GL_RGBA;
stride = pixmap->devKind;
row_length = (stride * 8) / pixmap->drawable.bitsPerPixel;
glBindTexture(GL_TEXTURE_2D, tex);
glPixelStorei(GL_UNPACK_ALIGNMENT, 1);
glPixelStorei(GL_UNPACK_ROW_LENGTH, row_length);
if (pixmap_priv->pbo && pixmap_priv->pbo_valid) {
texels = NULL;
glBindBufferARB(GL_PIXEL_UNPACK_BUFFER_EXT, pixmap_priv->pbo);
}
else
texels = pixmap->devPrivate.ptr;
glTexImage2D(GL_TEXTURE_2D,
0,
iformat,
pixmap->drawable.width,
pixmap->drawable.height, 0,
format, type, texels);
}
/*
* Load texture from the pixmap's data pointer and then
* draw the texture to the fbo, and flip the y axis.
* */
static void
_glamor_upload_pixmap_to_texture(PixmapPtr pixmap, GLenum format, GLenum type, int ax, int flip)
{
glamor_pixmap_private *pixmap_priv = glamor_get_pixmap_private(pixmap);
glamor_screen_private *glamor_priv =
glamor_get_screen_private(pixmap->drawable.pScreen);
static float vertices[8] = {-1, -1,
1, -1,
1, 1,
-1, 1};
static float texcoords[8] = {0, 1,
1, 1,
1, 0,
0, 0};
static float texcoords_inv[8] = {0, 0,
1, 0,
1, 1,
0, 1};
float *ptexcoords;
GLuint tex;
int need_flip;
need_flip = (flip && !glamor_priv->yInverted);
/* Try fast path firstly, upload the pixmap to the texture attached
* to the fbo directly. */
if (ax == 0 && !need_flip) {
__glamor_upload_pixmap_to_texture(pixmap, format, type, pixmap_priv->tex);
return;
}
if (need_flip)
ptexcoords = texcoords;
else
ptexcoords = texcoords_inv;
/* Slow path, we need to flip y or wire alpha to 1. */
glVertexPointer(2, GL_FLOAT, sizeof(float) * 2, vertices);
glEnableClientState(GL_VERTEX_ARRAY);
glClientActiveTexture(GL_TEXTURE0);
glTexCoordPointer(2, GL_FLOAT, sizeof(float) * 2, ptexcoords);
glEnableClientState(GL_TEXTURE_COORD_ARRAY);
glBindFramebufferEXT(GL_FRAMEBUFFER_EXT, pixmap_priv->fb);
glViewport(0, 0, pixmap->drawable.width, pixmap->drawable.height);
glGenTextures(1, &tex);
__glamor_upload_pixmap_to_texture(pixmap, format, type, tex);
glActiveTexture(GL_TEXTURE0);
glBindTexture(GL_TEXTURE_2D, tex);
glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_NEAREST);
glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_NEAREST);
glEnable(GL_TEXTURE_2D);
glUseProgramObjectARB(glamor_priv->finish_access_prog[ax]);
glDrawArrays(GL_TRIANGLE_FAN, 0, 4);
glDisable(GL_TEXTURE_2D);
glUseProgramObjectARB(0);
glDisableClientState(GL_VERTEX_ARRAY);
glDisableClientState(GL_TEXTURE_COORD_ARRAY);
glDeleteTextures(1, &tex);
glBindFramebufferEXT(GL_FRAMEBUFFER_EXT, 0);
}
/*
* Prepare to upload a pixmap to texture memory.
* ax 1 means the format needs to wire alpha to 1.
* Two condtion need to setup a fbo for a pixmap
* 1. !yInverted, we need to do flip if we are not yInverted.
* 2. ax != 0, we need to wire the alpha.
* */
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 int
glamor_pixmap_upload_prepare(PixmapPtr pixmap, int ax)
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 need_fbo;
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
glamor_pixmap_private *pixmap_priv = glamor_get_pixmap_private(pixmap);
glamor_screen_private *glamor_priv = glamor_get_screen_private(pixmap->drawable.pScreen);
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
if (!glamor_check_fbo_width_height(pixmap->drawable.width , pixmap->drawable.height)
|| !glamor_check_fbo_depth(pixmap->drawable.depth)) {
glamor_fallback("upload failed reason: bad size or depth %d x %d @depth %d \n",
pixmap->drawable.width, pixmap->drawable.height, pixmap->drawable.depth);
return -1;
}
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
if (GLAMOR_PIXMAP_PRIV_HAS_FBO(pixmap_priv))
return 0;
if (ax != 0 || !glamor_priv->yInverted)
need_fbo = 1;
else
need_fbo = 0;
if (pixmap_priv->tex == 0)
glGenTextures(1, &pixmap_priv->tex);
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
if (need_fbo) {
if (pixmap_priv->fb == 0)
glGenFramebuffersEXT(1, &pixmap_priv->fb);
glBindTexture(GL_TEXTURE_2D, pixmap_priv->tex);
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
glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_NEAREST);
glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_NEAREST);
glTexImage2D(GL_TEXTURE_2D, 0, GL_RGBA, pixmap->drawable.width,
pixmap->drawable.height, 0,
GL_RGBA, GL_UNSIGNED_BYTE, NULL);
glBindFramebufferEXT(GL_FRAMEBUFFER_EXT, pixmap_priv->fb);
glFramebufferTexture2DEXT(GL_FRAMEBUFFER_EXT,
GL_COLOR_ATTACHMENT0_EXT,
GL_TEXTURE_2D,
pixmap_priv->tex,
0);
}
return 0;
}
enum glamor_pixmap_status
glamor_upload_pixmap_to_texture(PixmapPtr pixmap)
{
GLenum format, type;
int ax;
if (glamor_get_tex_format_type_from_pixmap(pixmap,
&format,
&type,
&ax)) {
glamor_fallback("Unknown pixmap depth %d.\n", pixmap->drawable.depth);
return GLAMOR_UPLOAD_FAILED;
}
if (glamor_pixmap_upload_prepare(pixmap, ax))
return GLAMOR_UPLOAD_FAILED;
glamor_debug_output(GLAMOR_DEBUG_TEXTURE_DYNAMIC_UPLOAD,
"Uploading pixmap %p %dx%d depth%d.\n",
pixmap,
pixmap->drawable.width,
pixmap->drawable.height,
pixmap->drawable.depth);
_glamor_upload_pixmap_to_texture(pixmap, format, type, ax, 1);
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
return GLAMOR_UPLOAD_DONE;
}
#if 0
enum glamor_pixmap_status
glamor_upload_pixmap_to_texure_from_data(PixmapPtr pixmap, void *data)
{
enum glamor_pixmap_status upload_status;
glamor_pixmap_private *pixmap_priv = glamor_get_pixmap_private(pixmap);
assert(pixmap_priv->pbo_valid == 0);
assert(pixmap->devPrivate.ptr == NULL);
pixmap->devPrivate.ptr = data;
upload_status = glamor_upload_pixmap_to_texture(pixmap);
pixmap->devPrivate.ptr = NULL;
return upload_status;
}
#endif
void
glamor_restore_pixmap_to_texture(PixmapPtr pixmap)
{
GLenum format, type;
int ax;
if (glamor_get_tex_format_type_from_pixmap(pixmap,
&format,
&type,
&ax)) {
ErrorF("Unknown pixmap depth %d.\n", pixmap->drawable.depth);
assert(0);
}
_glamor_upload_pixmap_to_texture(pixmap, format, type, ax, 1);
}
/**
* Move a pixmap to CPU memory.
* The input data is the pixmap's fbo.
* The output data is at pixmap->devPrivate.ptr. We always use pbo
* to read the fbo and then map it to va. If possible, we will use
* it directly as devPrivate.ptr.
* If successfully download a fbo to cpu then return TRUE.
* Otherwise return FALSE.
**/
Bool
glamor_download_pixmap_to_cpu(PixmapPtr pixmap, glamor_access_t access)
{
glamor_pixmap_private *pixmap_priv = glamor_get_pixmap_private(pixmap);
unsigned int stride, row_length, y;
GLenum format, type, gl_access, gl_usage;
int ax;
uint8_t *data, *read;
glamor_screen_private *glamor_priv =
glamor_get_screen_private(pixmap->drawable.pScreen);
if (!GLAMOR_PIXMAP_PRIV_HAS_FBO(pixmap_priv))
return TRUE;
if (glamor_get_tex_format_type_from_pixmap(pixmap,
&format,
&type,
&ax)) {
ErrorF("Unknown pixmap depth %d.\n", pixmap->drawable.depth);
assert(0); // Should never happen.
return FALSE;
}
pixmap_priv->access_mode = access;
glamor_debug_output(GLAMOR_DEBUG_TEXTURE_DOWNLOAD,
"Downloading pixmap %p %dx%d depth%d\n",
pixmap,
pixmap->drawable.width,
pixmap->drawable.height,
pixmap->drawable.depth);
stride = pixmap->devKind;
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
switch (access) {
case GLAMOR_ACCESS_RO:
gl_access = GL_READ_ONLY_ARB;
gl_usage = GL_STREAM_READ_ARB;
break;
case GLAMOR_ACCESS_WO:
data = malloc(stride * pixmap->drawable.height);
goto done;
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
break;
case GLAMOR_ACCESS_RW:
gl_access = GL_READ_WRITE_ARB;
gl_usage = GL_DYNAMIC_DRAW_ARB;
break;
default:
ErrorF("Glamor: Invalid access code. %d\n", access);
assert(0);
}
row_length = (stride * 8) / pixmap->drawable.bitsPerPixel;
glBindFramebufferEXT(GL_FRAMEBUFFER_EXT, pixmap_priv->fb);
glPixelStorei(GL_PACK_ALIGNMENT, 1);
glPixelStorei(GL_PACK_ROW_LENGTH, row_length);
if (GLEW_MESA_pack_invert || glamor_priv->yInverted) {
if (!glamor_priv->yInverted)
glPixelStorei(GL_PACK_INVERT_MESA, 1);
if (pixmap_priv->pbo == 0)
glGenBuffersARB (1, &pixmap_priv->pbo);
glBindBufferARB (GL_PIXEL_PACK_BUFFER_EXT, pixmap_priv->pbo);
glBufferDataARB (GL_PIXEL_PACK_BUFFER_EXT,
stride * pixmap->drawable.height,
NULL, gl_usage);
glReadPixels (0, 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
row_length, pixmap->drawable.height,
format, type, 0);
data = glMapBufferARB (GL_PIXEL_PACK_BUFFER_EXT, gl_access);
pixmap_priv->pbo_valid = TRUE;
if (!glamor_priv->yInverted)
glPixelStorei(GL_PACK_INVERT_MESA, 0);
glBindBufferARB (GL_PIXEL_PACK_BUFFER_EXT, 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
} else {
data = malloc(stride * pixmap->drawable.height);
assert(data);
if (access != GLAMOR_ACCESS_WO) {
if (pixmap_priv->pbo == 0)
glGenBuffersARB(1, &pixmap_priv->pbo);
glBindBufferARB(GL_PIXEL_PACK_BUFFER_EXT, pixmap_priv->pbo);
glBufferDataARB(GL_PIXEL_PACK_BUFFER_EXT,
stride * pixmap->drawable.height,
NULL, GL_STREAM_READ_ARB);
glReadPixels (0, 0, row_length, pixmap->drawable.height,
format, type, 0);
read = glMapBufferARB(GL_PIXEL_PACK_BUFFER_EXT, GL_READ_ONLY_ARB);
for (y = 0; y < pixmap->drawable.height; y++)
memcpy(data + y * stride,
read + (pixmap->drawable.height - y - 1) * stride, stride);
glUnmapBufferARB(GL_PIXEL_PACK_BUFFER_EXT);
glBindBufferARB(GL_PIXEL_PACK_BUFFER_EXT, 0);
pixmap_priv->pbo_valid = FALSE;
glDeleteBuffersARB(1, &pixmap_priv->pbo);
pixmap_priv->pbo = 0;
}
}
glBindFramebufferEXT(GL_FRAMEBUFFER_EXT, 0);
done:
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
pixmap->devPrivate.ptr = data;
return TRUE;
}
static void
_glamor_destroy_upload_pixmap(PixmapPtr pixmap)
{
glamor_pixmap_private *pixmap_priv = glamor_get_pixmap_private(pixmap);
assert(pixmap_priv->gl_fbo == 0);
if (pixmap_priv->fb)
glDeleteFramebuffersEXT(1, &pixmap_priv->fb);
if (pixmap_priv->tex)
glDeleteTextures(1, &pixmap_priv->tex);
if (pixmap_priv->pbo)
glDeleteBuffersARB(1, &pixmap_priv->pbo);
pixmap_priv->fb = pixmap_priv->tex = pixmap_priv->pbo = 0;
}
void glamor_destroy_upload_pixmap(PixmapPtr pixmap)
{
_glamor_destroy_upload_pixmap(pixmap);
}