andreacavalli/xserver-multidpi
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity
6839996b0b
xserver-multidpi/glamor/glamor_fbo.c

593 lines
15 KiB
C
Raw Normal View History

Added some copyright and author information. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-04-28 08:13:47 +02:00
/*
* Copyright © 2009 Intel Corporation
* Copyright © 1998 Keith Packard
*
* 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:
* Zhigang Gong <zhigang.gong@gmail.com>
*
*/
glamor_fbo: Introduce glamor fbo to manage all the fb/tex. This is the first patch to implement a fbo/tex pool mechanism which is like the sna's BO cache list. We firstly need to decopule the fbo/tex from each pixmap. The new glamor_pixmap_fbo data structure is for that purpose. It's somehow independent to each pixmap and can be reused latter by other pixmaps once it's detached from the current pixmap. And this commit also slightly change the way to create a memory pixmap. We will not create a pixmap private data structure by default, instead we will crete that structure when a memory pixmap is attaching a fbo to it. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-18 12:21:36 +01:00
#include <stdlib.h>
#include "glamor_priv.h"
glamor-fbo: Tweek the cache bucket calculation. And also reduce the expire count to 100 which should be good enough on x11perf and cairo-trace testing. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-21 07:41:44 +01:00
#define GLAMOR_CACHE_EXPIRE_MAX 100
glamor-fbo-pool: Implement fbo cache mechanism. We classify the cache according to the texture's format/width/height. As openGL doesn't allow us to change a texture's format/width/height after the internal texture object is already allocated, we can't just calculate the size and then according ths size to put the fbo to an bucket which is just like SNA does. We can only put the fbo to the corresponding format/width/height bucket. This commit only support the exact size match. The following patch will remove this restriction, just need to handle the repeat/tile case when the size is not exactly match. Should use fls instead of ffs when decide the width/height bucket, thanks for Chris to point this out. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-19 13:47:55 +01:00
#define GLAMOR_CACHE_DEFAULT 0
#define GLAMOR_CACHE_EXACT_SIZE 1
glamor-pixmap-upload: Create a uploading fbo with a texture only. Just an initial implementation and disabled by default. When uploading a pixmap to a texture, we don't really want to attach the texture to any fbo. So add one fbo type which doesn't has a gl FBO attached to it. This commit can increase the cairo-trace's performance by 10-20%. Now the firefox-planet-gnome is 8.3s. SNA is still the best, only take 3.5s. Thanks for Chris to point out the A1 pixmap uploading bug. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-20 13:38:29 +01:00
#define GLAMOR_CACHE_TEXTURE 2
glamor-fbo-pool: Implement fbo cache mechanism. We classify the cache according to the texture's format/width/height. As openGL doesn't allow us to change a texture's format/width/height after the internal texture object is already allocated, we can't just calculate the size and then according ths size to put the fbo to an bucket which is just like SNA does. We can only put the fbo to the corresponding format/width/height bucket. This commit only support the exact size match. The following patch will remove this restriction, just need to handle the repeat/tile case when the size is not exactly match. Should use fls instead of ffs when decide the width/height bucket, thanks for Chris to point this out. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-19 13:47:55 +01:00
glamor_fbo: Added one macro to disable fbo cache. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-03-16 09:42:46 +01:00
//#define NO_FBO_CACHE 1
glamor_fbo: Added a threshold value for the fbo cache pool. Currently set it to 256MB. If cache pool watermark increases to this value, then don't push any fbo to this pool, will purge the fbo directly. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-04-26 12:31:07 +02:00
#define FBO_CACHE_THRESHOLD (256*1024*1024)
glamor_fbo: Added one macro to disable fbo cache. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-03-16 09:42:46 +01:00
glamor-fbo-pool: Implement fbo cache mechanism. We classify the cache according to the texture's format/width/height. As openGL doesn't allow us to change a texture's format/width/height after the internal texture object is already allocated, we can't just calculate the size and then according ths size to put the fbo to an bucket which is just like SNA does. We can only put the fbo to the corresponding format/width/height bucket. This commit only support the exact size match. The following patch will remove this restriction, just need to handle the repeat/tile case when the size is not exactly match. Should use fls instead of ffs when decide the width/height bucket, thanks for Chris to point this out. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-19 13:47:55 +01:00
/* Loop from the tail to the head. */
Fixup For list.h change in xorg Because the file list.h in xorg/include has changed the functions and struct names, adding xorg_ prefix before the original name. So Modify glamor_screen_private struct and the code which use list's functions in glamor_fbo.c. We hack at glamor_priv.h avoid the compile error when using old version xserver header file. Signed-off-by: Junyan He <junyan.he@linux.intel.com>
2012-03-05 01:24:20 +01:00
#define xorg_list_for_each_entry_reverse(pos, head, member) \
glamor-fbo-pool: Implement fbo cache mechanism. We classify the cache according to the texture's format/width/height. As openGL doesn't allow us to change a texture's format/width/height after the internal texture object is already allocated, we can't just calculate the size and then according ths size to put the fbo to an bucket which is just like SNA does. We can only put the fbo to the corresponding format/width/height bucket. This commit only support the exact size match. The following patch will remove this restriction, just need to handle the repeat/tile case when the size is not exactly match. Should use fls instead of ffs when decide the width/height bucket, thanks for Chris to point this out. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-19 13:47:55 +01:00
for (pos = __container_of((head)->prev, pos, member); \
&pos->member != (head); \
pos = __container_of(pos->member.prev, pos, member))
Fixup For list.h change in xorg Because the file list.h in xorg/include has changed the functions and struct names, adding xorg_ prefix before the original name. So Modify glamor_screen_private struct and the code which use list's functions in glamor_fbo.c. We hack at glamor_priv.h avoid the compile error when using old version xserver header file. Signed-off-by: Junyan He <junyan.he@linux.intel.com>
2012-03-05 01:24:20 +01:00
#define xorg_list_for_each_entry_safe_reverse(pos, tmp, head, member) \
glamor-fbo-pool: Implement fbo cache mechanism. We classify the cache according to the texture's format/width/height. As openGL doesn't allow us to change a texture's format/width/height after the internal texture object is already allocated, we can't just calculate the size and then according ths size to put the fbo to an bucket which is just like SNA does. We can only put the fbo to the corresponding format/width/height bucket. This commit only support the exact size match. The following patch will remove this restriction, just need to handle the repeat/tile case when the size is not exactly match. Should use fls instead of ffs when decide the width/height bucket, thanks for Chris to point this out. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-19 13:47:55 +01:00
for (pos = __container_of((head)->prev, pos, member), \
tmp = __container_of(pos->member.prev, pos, member); \
&pos->member != (head); \
pos = tmp, tmp = __container_of(pos->member.prev, tmp, member))
#ifdef __i386__
static inline unsigned long __fls(unsigned long x)
{
asm("bsr %1,%0"
: "=r" (x)
: "rm" (x));
return x;
}
#else
static inline unsigned long __fls(unsigned long x)
{
int n;
if (x == 0) return(0);
n = 0;
if (x <= 0x0000FFFF) {n = n +16; x = x <<16;}
if (x <= 0x00FFFFFF) {n = n + 8; x = x << 8;}
if (x <= 0x0FFFFFFF) {n = n + 4; x = x << 4;}
if (x <= 0x3FFFFFFF) {n = n + 2; x = x << 2;}
if (x <= 0x7FFFFFFF) {n = n + 1;}
return 31 - n;
}
#endif
inline static int cache_wbucket(int size)
{
glamor-fbo: Tweek the cache bucket calculation. And also reduce the expire count to 100 which should be good enough on x11perf and cairo-trace testing. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-21 07:41:44 +01:00
int order = __fls(size / 32);
glamor-fbo-pool: Implement fbo cache mechanism. We classify the cache according to the texture's format/width/height. As openGL doesn't allow us to change a texture's format/width/height after the internal texture object is already allocated, we can't just calculate the size and then according ths size to put the fbo to an bucket which is just like SNA does. We can only put the fbo to the corresponding format/width/height bucket. This commit only support the exact size match. The following patch will remove this restriction, just need to handle the repeat/tile case when the size is not exactly match. Should use fls instead of ffs when decide the width/height bucket, thanks for Chris to point this out. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-19 13:47:55 +01:00
if (order >= CACHE_BUCKET_WCOUNT)
order = CACHE_BUCKET_WCOUNT - 1;
return order;
}
inline static int cache_hbucket(int size)
{
glamor-fbo: Tweek the cache bucket calculation. And also reduce the expire count to 100 which should be good enough on x11perf and cairo-trace testing. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-21 07:41:44 +01:00
int order = __fls(size / 32);
glamor-fbo-pool: Implement fbo cache mechanism. We classify the cache according to the texture's format/width/height. As openGL doesn't allow us to change a texture's format/width/height after the internal texture object is already allocated, we can't just calculate the size and then according ths size to put the fbo to an bucket which is just like SNA does. We can only put the fbo to the corresponding format/width/height bucket. This commit only support the exact size match. The following patch will remove this restriction, just need to handle the repeat/tile case when the size is not exactly match. Should use fls instead of ffs when decide the width/height bucket, thanks for Chris to point this out. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-19 13:47:55 +01:00
if (order >= CACHE_BUCKET_HCOUNT)
order = CACHE_BUCKET_HCOUNT - 1;
return order;
}
glamor_pixmap_fbo *
glamor_pixmap_fbo_cache_get(glamor_screen_private *glamor_priv,
int w, int h, GLenum format, int flag)
{
Fixup For list.h change in xorg Because the file list.h in xorg/include has changed the functions and struct names, adding xorg_ prefix before the original name. So Modify glamor_screen_private struct and the code which use list's functions in glamor_fbo.c. We hack at glamor_priv.h avoid the compile error when using old version xserver header file. Signed-off-by: Junyan He <junyan.he@linux.intel.com>
2012-03-05 01:24:20 +01:00
struct xorg_list *cache;
glamor_fbo: Added a threshold value for the fbo cache pool. Currently set it to 256MB. If cache pool watermark increases to this value, then don't push any fbo to this pool, will purge the fbo directly. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-04-26 12:31:07 +02:00
glamor_pixmap_fbo *fbo_entry, *ret_fbo = NULL;
gles2: Fixed color conversion for the formats except 1555 and 2101010. This patch fixed two major problems when we do the color convesion with GLES2. 1. lack of necessary formats in FBO pool. GLES2 has three different possible texture formats, GL_RGBA, GL_BGRA and GL_ALPHA. Previous implementation only has one bucket for all the three formats which may reuse a incorrect texture format when do the cache lookup. After this fix, we can enable fbo safely when running with GLES2. 2. Refine the format matching method in glamor_get_tex_format_type_from_pictformat. If both revertion and swap_rb are needed, for example use GL_RGBA to represent PICT_b8g8r8a8. Then the downloading and uploading should be handled differently. The picture's format is PICT_b8g8r8a8, Then the expecting color layout is as below (little endian): 0 1 2 3 : address a r g b Now the in GLES2 the supported color format is GL_RGBA, type is GL_UNSIGNED_TYPE, then we need to shuffle the fragment color as : frag_color = sample(texture).argb; before we use glReadPixel to get it back. For the uploading process, the shuffle is a revert shuffle. We still use GL_RGBA, GL_UNSIGNED_BYTE to upload the color to a texture, then let's see 0 1 2 3 : address a r g b : correct colors R G B A : GL_RGBA with GL_UNSIGNED_BYTE Now we need to shuffle again, the mapping rule is r = G, g = B, b = A, a = R. Then the uploading shuffle is as below: frag_color = sample(texture).gbar; After this commit, gles2 version can pass render check with all the formats except those 1555/2101010. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-04-05 15:53:44 +02:00
int n_format;
glamor_fbo: Added one macro to disable fbo cache. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-03-16 09:42:46 +01:00
#ifdef NO_FBO_CACHE
return NULL;
#else
gles2: Fixed color conversion for the formats except 1555 and 2101010. This patch fixed two major problems when we do the color convesion with GLES2. 1. lack of necessary formats in FBO pool. GLES2 has three different possible texture formats, GL_RGBA, GL_BGRA and GL_ALPHA. Previous implementation only has one bucket for all the three formats which may reuse a incorrect texture format when do the cache lookup. After this fix, we can enable fbo safely when running with GLES2. 2. Refine the format matching method in glamor_get_tex_format_type_from_pictformat. If both revertion and swap_rb are needed, for example use GL_RGBA to represent PICT_b8g8r8a8. Then the downloading and uploading should be handled differently. The picture's format is PICT_b8g8r8a8, Then the expecting color layout is as below (little endian): 0 1 2 3 : address a r g b Now the in GLES2 the supported color format is GL_RGBA, type is GL_UNSIGNED_TYPE, then we need to shuffle the fragment color as : frag_color = sample(texture).argb; before we use glReadPixel to get it back. For the uploading process, the shuffle is a revert shuffle. We still use GL_RGBA, GL_UNSIGNED_BYTE to upload the color to a texture, then let's see 0 1 2 3 : address a r g b : correct colors R G B A : GL_RGBA with GL_UNSIGNED_BYTE Now we need to shuffle again, the mapping rule is r = G, g = B, b = A, a = R. Then the uploading shuffle is as below: frag_color = sample(texture).gbar; After this commit, gles2 version can pass render check with all the formats except those 1555/2101010. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-04-05 15:53:44 +02:00
n_format = cache_format(format);
if (n_format == -1)
return NULL;
glamor-pixmap-upload: Create a uploading fbo with a texture only. Just an initial implementation and disabled by default. When uploading a pixmap to a texture, we don't really want to attach the texture to any fbo. So add one fbo type which doesn't has a gl FBO attached to it. This commit can increase the cairo-trace's performance by 10-20%. Now the firefox-planet-gnome is 8.3s. SNA is still the best, only take 3.5s. Thanks for Chris to point out the A1 pixmap uploading bug. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-20 13:38:29 +01:00
if (!(flag & GLAMOR_CACHE_TEXTURE))
gles2: Fixed color conversion for the formats except 1555 and 2101010. This patch fixed two major problems when we do the color convesion with GLES2. 1. lack of necessary formats in FBO pool. GLES2 has three different possible texture formats, GL_RGBA, GL_BGRA and GL_ALPHA. Previous implementation only has one bucket for all the three formats which may reuse a incorrect texture format when do the cache lookup. After this fix, we can enable fbo safely when running with GLES2. 2. Refine the format matching method in glamor_get_tex_format_type_from_pictformat. If both revertion and swap_rb are needed, for example use GL_RGBA to represent PICT_b8g8r8a8. Then the downloading and uploading should be handled differently. The picture's format is PICT_b8g8r8a8, Then the expecting color layout is as below (little endian): 0 1 2 3 : address a r g b Now the in GLES2 the supported color format is GL_RGBA, type is GL_UNSIGNED_TYPE, then we need to shuffle the fragment color as : frag_color = sample(texture).argb; before we use glReadPixel to get it back. For the uploading process, the shuffle is a revert shuffle. We still use GL_RGBA, GL_UNSIGNED_BYTE to upload the color to a texture, then let's see 0 1 2 3 : address a r g b : correct colors R G B A : GL_RGBA with GL_UNSIGNED_BYTE Now we need to shuffle again, the mapping rule is r = G, g = B, b = A, a = R. Then the uploading shuffle is as below: frag_color = sample(texture).gbar; After this commit, gles2 version can pass render check with all the formats except those 1555/2101010. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-04-05 15:53:44 +02:00
cache = &glamor_priv->fbo_cache[n_format]
glamor-pixmap-upload: Create a uploading fbo with a texture only. Just an initial implementation and disabled by default. When uploading a pixmap to a texture, we don't really want to attach the texture to any fbo. So add one fbo type which doesn't has a gl FBO attached to it. This commit can increase the cairo-trace's performance by 10-20%. Now the firefox-planet-gnome is 8.3s. SNA is still the best, only take 3.5s. Thanks for Chris to point out the A1 pixmap uploading bug. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-20 13:38:29 +01:00
[cache_wbucket(w)]
[cache_hbucket(h)];
else
gles2: Fixed color conversion for the formats except 1555 and 2101010. This patch fixed two major problems when we do the color convesion with GLES2. 1. lack of necessary formats in FBO pool. GLES2 has three different possible texture formats, GL_RGBA, GL_BGRA and GL_ALPHA. Previous implementation only has one bucket for all the three formats which may reuse a incorrect texture format when do the cache lookup. After this fix, we can enable fbo safely when running with GLES2. 2. Refine the format matching method in glamor_get_tex_format_type_from_pictformat. If both revertion and swap_rb are needed, for example use GL_RGBA to represent PICT_b8g8r8a8. Then the downloading and uploading should be handled differently. The picture's format is PICT_b8g8r8a8, Then the expecting color layout is as below (little endian): 0 1 2 3 : address a r g b Now the in GLES2 the supported color format is GL_RGBA, type is GL_UNSIGNED_TYPE, then we need to shuffle the fragment color as : frag_color = sample(texture).argb; before we use glReadPixel to get it back. For the uploading process, the shuffle is a revert shuffle. We still use GL_RGBA, GL_UNSIGNED_BYTE to upload the color to a texture, then let's see 0 1 2 3 : address a r g b : correct colors R G B A : GL_RGBA with GL_UNSIGNED_BYTE Now we need to shuffle again, the mapping rule is r = G, g = B, b = A, a = R. Then the uploading shuffle is as below: frag_color = sample(texture).gbar; After this commit, gles2 version can pass render check with all the formats except those 1555/2101010. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-04-05 15:53:44 +02:00
cache = &glamor_priv->tex_cache[n_format]
glamor-pixmap-upload: Create a uploading fbo with a texture only. Just an initial implementation and disabled by default. When uploading a pixmap to a texture, we don't really want to attach the texture to any fbo. So add one fbo type which doesn't has a gl FBO attached to it. This commit can increase the cairo-trace's performance by 10-20%. Now the firefox-planet-gnome is 8.3s. SNA is still the best, only take 3.5s. Thanks for Chris to point out the A1 pixmap uploading bug. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-20 13:38:29 +01:00
[cache_wbucket(w)]
[cache_hbucket(h)];
if (!(flag & GLAMOR_CACHE_EXACT_SIZE)) {
Fixup For list.h change in xorg Because the file list.h in xorg/include has changed the functions and struct names, adding xorg_ prefix before the original name. So Modify glamor_screen_private struct and the code which use list's functions in glamor_fbo.c. We hack at glamor_priv.h avoid the compile error when using old version xserver header file. Signed-off-by: Junyan He <junyan.he@linux.intel.com>
2012-03-05 01:24:20 +01:00
xorg_list_for_each_entry(fbo_entry, cache, list) {
glamor-fbo-pool: Enable to reuse different size fbo/texture. Fixup three special cases, one is in tile and the other is in composite. Both cases are due to repeat texture issue. Maybe we can refine the shader to recalculate texture coords to support partial texture's repeating. The third is when upload a memory pixmap to texture, as now the texture may not have the exact size as the pixmap, we should not use the full rect coords. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-20 02:59:44 +01:00
if (fbo_entry->width >= w && fbo_entry->height >= h) {
glamor-fbo-pool: Implement fbo cache mechanism. We classify the cache according to the texture's format/width/height. As openGL doesn't allow us to change a texture's format/width/height after the internal texture object is already allocated, we can't just calculate the size and then according ths size to put the fbo to an bucket which is just like SNA does. We can only put the fbo to the corresponding format/width/height bucket. This commit only support the exact size match. The following patch will remove this restriction, just need to handle the repeat/tile case when the size is not exactly match. Should use fls instead of ffs when decide the width/height bucket, thanks for Chris to point this out. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-19 13:47:55 +01:00
gles2: Fixed color conversion for the formats except 1555 and 2101010. This patch fixed two major problems when we do the color convesion with GLES2. 1. lack of necessary formats in FBO pool. GLES2 has three different possible texture formats, GL_RGBA, GL_BGRA and GL_ALPHA. Previous implementation only has one bucket for all the three formats which may reuse a incorrect texture format when do the cache lookup. After this fix, we can enable fbo safely when running with GLES2. 2. Refine the format matching method in glamor_get_tex_format_type_from_pictformat. If both revertion and swap_rb are needed, for example use GL_RGBA to represent PICT_b8g8r8a8. Then the downloading and uploading should be handled differently. The picture's format is PICT_b8g8r8a8, Then the expecting color layout is as below (little endian): 0 1 2 3 : address a r g b Now the in GLES2 the supported color format is GL_RGBA, type is GL_UNSIGNED_TYPE, then we need to shuffle the fragment color as : frag_color = sample(texture).argb; before we use glReadPixel to get it back. For the uploading process, the shuffle is a revert shuffle. We still use GL_RGBA, GL_UNSIGNED_BYTE to upload the color to a texture, then let's see 0 1 2 3 : address a r g b : correct colors R G B A : GL_RGBA with GL_UNSIGNED_BYTE Now we need to shuffle again, the mapping rule is r = G, g = B, b = A, a = R. Then the uploading shuffle is as below: frag_color = sample(texture).gbar; After this commit, gles2 version can pass render check with all the formats except those 1555/2101010. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-04-05 15:53:44 +02:00
DEBUGF("Request w %d h %d format %x \n", w, h, format);
DEBUGF("got cache entry %p w %d h %d fbo %d tex %d format %x\n",
glamor-fbo-pool: Implement fbo cache mechanism. We classify the cache according to the texture's format/width/height. As openGL doesn't allow us to change a texture's format/width/height after the internal texture object is already allocated, we can't just calculate the size and then according ths size to put the fbo to an bucket which is just like SNA does. We can only put the fbo to the corresponding format/width/height bucket. This commit only support the exact size match. The following patch will remove this restriction, just need to handle the repeat/tile case when the size is not exactly match. Should use fls instead of ffs when decide the width/height bucket, thanks for Chris to point this out. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-19 13:47:55 +01:00
fbo_entry, fbo_entry->width, fbo_entry->height,
fbo_entry->fb, fbo_entry->tex);
Fixup For list.h change in xorg Because the file list.h in xorg/include has changed the functions and struct names, adding xorg_ prefix before the original name. So Modify glamor_screen_private struct and the code which use list's functions in glamor_fbo.c. We hack at glamor_priv.h avoid the compile error when using old version xserver header file. Signed-off-by: Junyan He <junyan.he@linux.intel.com>
2012-03-05 01:24:20 +01:00
xorg_list_del(&fbo_entry->list);
glamor_fbo: Added a threshold value for the fbo cache pool. Currently set it to 256MB. If cache pool watermark increases to this value, then don't push any fbo to this pool, will purge the fbo directly. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-04-26 12:31:07 +02:00
ret_fbo = fbo_entry;
break;
glamor-fbo-pool: Implement fbo cache mechanism. We classify the cache according to the texture's format/width/height. As openGL doesn't allow us to change a texture's format/width/height after the internal texture object is already allocated, we can't just calculate the size and then according ths size to put the fbo to an bucket which is just like SNA does. We can only put the fbo to the corresponding format/width/height bucket. This commit only support the exact size match. The following patch will remove this restriction, just need to handle the repeat/tile case when the size is not exactly match. Should use fls instead of ffs when decide the width/height bucket, thanks for Chris to point this out. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-19 13:47:55 +01:00
}
}
}
else {
Fixup For list.h change in xorg Because the file list.h in xorg/include has changed the functions and struct names, adding xorg_ prefix before the original name. So Modify glamor_screen_private struct and the code which use list's functions in glamor_fbo.c. We hack at glamor_priv.h avoid the compile error when using old version xserver header file. Signed-off-by: Junyan He <junyan.he@linux.intel.com>
2012-03-05 01:24:20 +01:00
xorg_list_for_each_entry(fbo_entry, cache, list) {
glamor-fbo-pool: Implement fbo cache mechanism. We classify the cache according to the texture's format/width/height. As openGL doesn't allow us to change a texture's format/width/height after the internal texture object is already allocated, we can't just calculate the size and then according ths size to put the fbo to an bucket which is just like SNA does. We can only put the fbo to the corresponding format/width/height bucket. This commit only support the exact size match. The following patch will remove this restriction, just need to handle the repeat/tile case when the size is not exactly match. Should use fls instead of ffs when decide the width/height bucket, thanks for Chris to point this out. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-19 13:47:55 +01:00
if (fbo_entry->width == w && fbo_entry->height == h) {
gles2: Fixed color conversion for the formats except 1555 and 2101010. This patch fixed two major problems when we do the color convesion with GLES2. 1. lack of necessary formats in FBO pool. GLES2 has three different possible texture formats, GL_RGBA, GL_BGRA and GL_ALPHA. Previous implementation only has one bucket for all the three formats which may reuse a incorrect texture format when do the cache lookup. After this fix, we can enable fbo safely when running with GLES2. 2. Refine the format matching method in glamor_get_tex_format_type_from_pictformat. If both revertion and swap_rb are needed, for example use GL_RGBA to represent PICT_b8g8r8a8. Then the downloading and uploading should be handled differently. The picture's format is PICT_b8g8r8a8, Then the expecting color layout is as below (little endian): 0 1 2 3 : address a r g b Now the in GLES2 the supported color format is GL_RGBA, type is GL_UNSIGNED_TYPE, then we need to shuffle the fragment color as : frag_color = sample(texture).argb; before we use glReadPixel to get it back. For the uploading process, the shuffle is a revert shuffle. We still use GL_RGBA, GL_UNSIGNED_BYTE to upload the color to a texture, then let's see 0 1 2 3 : address a r g b : correct colors R G B A : GL_RGBA with GL_UNSIGNED_BYTE Now we need to shuffle again, the mapping rule is r = G, g = B, b = A, a = R. Then the uploading shuffle is as below: frag_color = sample(texture).gbar; After this commit, gles2 version can pass render check with all the formats except those 1555/2101010. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-04-05 15:53:44 +02:00
DEBUGF("Request w %d h %d format %x \n", w, h, format);
DEBUGF("got cache entry %p w %d h %d fbo %d tex %d format %x\n",
glamor-fbo-pool: Implement fbo cache mechanism. We classify the cache according to the texture's format/width/height. As openGL doesn't allow us to change a texture's format/width/height after the internal texture object is already allocated, we can't just calculate the size and then according ths size to put the fbo to an bucket which is just like SNA does. We can only put the fbo to the corresponding format/width/height bucket. This commit only support the exact size match. The following patch will remove this restriction, just need to handle the repeat/tile case when the size is not exactly match. Should use fls instead of ffs when decide the width/height bucket, thanks for Chris to point this out. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-19 13:47:55 +01:00
fbo_entry, fbo_entry->width, fbo_entry->height,
gles2: Fixed color conversion for the formats except 1555 and 2101010. This patch fixed two major problems when we do the color convesion with GLES2. 1. lack of necessary formats in FBO pool. GLES2 has three different possible texture formats, GL_RGBA, GL_BGRA and GL_ALPHA. Previous implementation only has one bucket for all the three formats which may reuse a incorrect texture format when do the cache lookup. After this fix, we can enable fbo safely when running with GLES2. 2. Refine the format matching method in glamor_get_tex_format_type_from_pictformat. If both revertion and swap_rb are needed, for example use GL_RGBA to represent PICT_b8g8r8a8. Then the downloading and uploading should be handled differently. The picture's format is PICT_b8g8r8a8, Then the expecting color layout is as below (little endian): 0 1 2 3 : address a r g b Now the in GLES2 the supported color format is GL_RGBA, type is GL_UNSIGNED_TYPE, then we need to shuffle the fragment color as : frag_color = sample(texture).argb; before we use glReadPixel to get it back. For the uploading process, the shuffle is a revert shuffle. We still use GL_RGBA, GL_UNSIGNED_BYTE to upload the color to a texture, then let's see 0 1 2 3 : address a r g b : correct colors R G B A : GL_RGBA with GL_UNSIGNED_BYTE Now we need to shuffle again, the mapping rule is r = G, g = B, b = A, a = R. Then the uploading shuffle is as below: frag_color = sample(texture).gbar; After this commit, gles2 version can pass render check with all the formats except those 1555/2101010. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-04-05 15:53:44 +02:00
fbo_entry->fb, fbo_entry->tex, fbo_entry->format);
assert(format == fbo_entry->format);
Fixup For list.h change in xorg Because the file list.h in xorg/include has changed the functions and struct names, adding xorg_ prefix before the original name. So Modify glamor_screen_private struct and the code which use list's functions in glamor_fbo.c. We hack at glamor_priv.h avoid the compile error when using old version xserver header file. Signed-off-by: Junyan He <junyan.he@linux.intel.com>
2012-03-05 01:24:20 +01:00
xorg_list_del(&fbo_entry->list);
glamor_fbo: Added a threshold value for the fbo cache pool. Currently set it to 256MB. If cache pool watermark increases to this value, then don't push any fbo to this pool, will purge the fbo directly. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-04-26 12:31:07 +02:00
ret_fbo = fbo_entry;
break;
glamor-fbo-pool: Implement fbo cache mechanism. We classify the cache according to the texture's format/width/height. As openGL doesn't allow us to change a texture's format/width/height after the internal texture object is already allocated, we can't just calculate the size and then according ths size to put the fbo to an bucket which is just like SNA does. We can only put the fbo to the corresponding format/width/height bucket. This commit only support the exact size match. The following patch will remove this restriction, just need to handle the repeat/tile case when the size is not exactly match. Should use fls instead of ffs when decide the width/height bucket, thanks for Chris to point this out. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-19 13:47:55 +01:00
}
}
}
glamor_fbo: Added a threshold value for the fbo cache pool. Currently set it to 256MB. If cache pool watermark increases to this value, then don't push any fbo to this pool, will purge the fbo directly. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-04-26 12:31:07 +02:00
if (ret_fbo)
glamor_priv->fbo_cache_watermark -= ret_fbo->width * ret_fbo->height;
assert(glamor_priv->fbo_cache_watermark >= 0);
return ret_fbo;
glamor_fbo: Added one macro to disable fbo cache. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-03-16 09:42:46 +01:00
#endif
glamor-fbo-pool: Implement fbo cache mechanism. We classify the cache according to the texture's format/width/height. As openGL doesn't allow us to change a texture's format/width/height after the internal texture object is already allocated, we can't just calculate the size and then according ths size to put the fbo to an bucket which is just like SNA does. We can only put the fbo to the corresponding format/width/height bucket. This commit only support the exact size match. The following patch will remove this restriction, just need to handle the repeat/tile case when the size is not exactly match. Should use fls instead of ffs when decide the width/height bucket, thanks for Chris to point this out. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-19 13:47:55 +01:00
}
Refine CloseScreen and FreeScreen processes. This commit move the calling to glamor_close_screen from glamor_egl_free_screen to glamor_egl_close_screen, as this is the right place to do this. We should detach screen fbo and destroy the corresponding KHR image at glamor_egl_close_screen stage. As latter DDX driver will call DestroyPixmap to destroy screen pixmap, if the fbo and image are still there but glamor screen private data pointer has been freed, then it causes segfault. This commit also introduces a new flag GLAMOR_USE_EGL_SCREEN. if DDX driver is using EGL layer then should set this bit when call to glamor_init and then both glamor_close_screen and glamor_egl_close_screen will be registered correctly, DDX layer will not need to call these two functions manually. This way is also the preferred method within Xorg domain. As interfaces changed, bump the version to 0.3.0. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com> Tested-by: Peng Li <peng.li@intel.com>
2012-01-31 11:16:58 +01:00
void
glamor-pixmap-upload: Create a uploading fbo with a texture only. Just an initial implementation and disabled by default. When uploading a pixmap to a texture, we don't really want to attach the texture to any fbo. So add one fbo type which doesn't has a gl FBO attached to it. This commit can increase the cairo-trace's performance by 10-20%. Now the firefox-planet-gnome is 8.3s. SNA is still the best, only take 3.5s. Thanks for Chris to point out the A1 pixmap uploading bug. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-20 13:38:29 +01:00
glamor_purge_fbo(glamor_pixmap_fbo *fbo)
{
Fixup glx support Renaming glamor_priv->dispatch and wrapping the access to the dispatch table with a function that also ensured the context was bound. dispatch = glamor_get_dispatch(glamor_priv); ... glamor_put_dispatch(glamor_priv); So that we catch all places where we attempt to call into GL withouta context. As an optimisation we can then do glamor_get_context(); glamor_put_context() around the rendering entry points to reduce the frequency of having to restore the old context. (Along with allowing the context to be recursively acquired and making the old context part of the glamor_egl state.) Reviewed-by: Zhigang Gong <zhigang.gong@linux.intel.com> Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-02-10 13:54:44 +01:00
glamor_gl_dispatch *dispatch = glamor_get_dispatch(fbo->glamor_priv);
glamor-pixmap-upload: Create a uploading fbo with a texture only. Just an initial implementation and disabled by default. When uploading a pixmap to a texture, we don't really want to attach the texture to any fbo. So add one fbo type which doesn't has a gl FBO attached to it. This commit can increase the cairo-trace's performance by 10-20%. Now the firefox-planet-gnome is 8.3s. SNA is still the best, only take 3.5s. Thanks for Chris to point out the A1 pixmap uploading bug. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-20 13:38:29 +01:00
if (fbo->fb)
dispatch->glDeleteFramebuffers(1, &fbo->fb);
if (fbo->tex)
dispatch->glDeleteTextures(1, &fbo->tex);
if (fbo->pbo)
dispatch->glDeleteBuffers(1, &fbo->pbo);
Fixup glx support Renaming glamor_priv->dispatch and wrapping the access to the dispatch table with a function that also ensured the context was bound. dispatch = glamor_get_dispatch(glamor_priv); ... glamor_put_dispatch(glamor_priv); So that we catch all places where we attempt to call into GL withouta context. As an optimisation we can then do glamor_get_context(); glamor_put_context() around the rendering entry points to reduce the frequency of having to restore the old context. (Along with allowing the context to be recursively acquired and making the old context part of the glamor_egl state.) Reviewed-by: Zhigang Gong <zhigang.gong@linux.intel.com> Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-02-10 13:54:44 +01:00
glamor_put_dispatch(fbo->glamor_priv);
glamor-pixmap-upload: Create a uploading fbo with a texture only. Just an initial implementation and disabled by default. When uploading a pixmap to a texture, we don't really want to attach the texture to any fbo. So add one fbo type which doesn't has a gl FBO attached to it. This commit can increase the cairo-trace's performance by 10-20%. Now the firefox-planet-gnome is 8.3s. SNA is still the best, only take 3.5s. Thanks for Chris to point out the A1 pixmap uploading bug. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-20 13:38:29 +01:00
free(fbo);
}
glamor_fbo: Added one macro to disable fbo cache. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-03-16 09:42:46 +01:00
static void
glamor-fbo-pool: Implement fbo cache mechanism. We classify the cache according to the texture's format/width/height. As openGL doesn't allow us to change a texture's format/width/height after the internal texture object is already allocated, we can't just calculate the size and then according ths size to put the fbo to an bucket which is just like SNA does. We can only put the fbo to the corresponding format/width/height bucket. This commit only support the exact size match. The following patch will remove this restriction, just need to handle the repeat/tile case when the size is not exactly match. Should use fls instead of ffs when decide the width/height bucket, thanks for Chris to point this out. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-19 13:47:55 +01:00
glamor_pixmap_fbo_cache_put(glamor_pixmap_fbo *fbo)
{
Fixup For list.h change in xorg Because the file list.h in xorg/include has changed the functions and struct names, adding xorg_ prefix before the original name. So Modify glamor_screen_private struct and the code which use list's functions in glamor_fbo.c. We hack at glamor_priv.h avoid the compile error when using old version xserver header file. Signed-off-by: Junyan He <junyan.he@linux.intel.com>
2012-03-05 01:24:20 +01:00
struct xorg_list *cache;
gles2: Fixed color conversion for the formats except 1555 and 2101010. This patch fixed two major problems when we do the color convesion with GLES2. 1. lack of necessary formats in FBO pool. GLES2 has three different possible texture formats, GL_RGBA, GL_BGRA and GL_ALPHA. Previous implementation only has one bucket for all the three formats which may reuse a incorrect texture format when do the cache lookup. After this fix, we can enable fbo safely when running with GLES2. 2. Refine the format matching method in glamor_get_tex_format_type_from_pictformat. If both revertion and swap_rb are needed, for example use GL_RGBA to represent PICT_b8g8r8a8. Then the downloading and uploading should be handled differently. The picture's format is PICT_b8g8r8a8, Then the expecting color layout is as below (little endian): 0 1 2 3 : address a r g b Now the in GLES2 the supported color format is GL_RGBA, type is GL_UNSIGNED_TYPE, then we need to shuffle the fragment color as : frag_color = sample(texture).argb; before we use glReadPixel to get it back. For the uploading process, the shuffle is a revert shuffle. We still use GL_RGBA, GL_UNSIGNED_BYTE to upload the color to a texture, then let's see 0 1 2 3 : address a r g b : correct colors R G B A : GL_RGBA with GL_UNSIGNED_BYTE Now we need to shuffle again, the mapping rule is r = G, g = B, b = A, a = R. Then the uploading shuffle is as below: frag_color = sample(texture).gbar; After this commit, gles2 version can pass render check with all the formats except those 1555/2101010. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-04-05 15:53:44 +02:00
int n_format;
glamor_fbo: Added one macro to disable fbo cache. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-03-16 09:42:46 +01:00
#ifdef NO_FBO_CACHE
glamor_purge_fbo(fbo);
return;
#else
gles2: Fixed color conversion for the formats except 1555 and 2101010. This patch fixed two major problems when we do the color convesion with GLES2. 1. lack of necessary formats in FBO pool. GLES2 has three different possible texture formats, GL_RGBA, GL_BGRA and GL_ALPHA. Previous implementation only has one bucket for all the three formats which may reuse a incorrect texture format when do the cache lookup. After this fix, we can enable fbo safely when running with GLES2. 2. Refine the format matching method in glamor_get_tex_format_type_from_pictformat. If both revertion and swap_rb are needed, for example use GL_RGBA to represent PICT_b8g8r8a8. Then the downloading and uploading should be handled differently. The picture's format is PICT_b8g8r8a8, Then the expecting color layout is as below (little endian): 0 1 2 3 : address a r g b Now the in GLES2 the supported color format is GL_RGBA, type is GL_UNSIGNED_TYPE, then we need to shuffle the fragment color as : frag_color = sample(texture).argb; before we use glReadPixel to get it back. For the uploading process, the shuffle is a revert shuffle. We still use GL_RGBA, GL_UNSIGNED_BYTE to upload the color to a texture, then let's see 0 1 2 3 : address a r g b : correct colors R G B A : GL_RGBA with GL_UNSIGNED_BYTE Now we need to shuffle again, the mapping rule is r = G, g = B, b = A, a = R. Then the uploading shuffle is as below: frag_color = sample(texture).gbar; After this commit, gles2 version can pass render check with all the formats except those 1555/2101010. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-04-05 15:53:44 +02:00
n_format = cache_format(fbo->format);
glamor_fbo: Added a threshold value for the fbo cache pool. Currently set it to 256MB. If cache pool watermark increases to this value, then don't push any fbo to this pool, will purge the fbo directly. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-04-26 12:31:07 +02:00
if (fbo->fb == 0 || n_format == -1
|| fbo->glamor_priv->fbo_cache_watermark >= FBO_CACHE_THRESHOLD) {
fbo->glamor_priv->tick ++;
glamor-pixmap-upload: Create a uploading fbo with a texture only. Just an initial implementation and disabled by default. When uploading a pixmap to a texture, we don't really want to attach the texture to any fbo. So add one fbo type which doesn't has a gl FBO attached to it. This commit can increase the cairo-trace's performance by 10-20%. Now the firefox-planet-gnome is 8.3s. SNA is still the best, only take 3.5s. Thanks for Chris to point out the A1 pixmap uploading bug. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-20 13:38:29 +01:00
glamor_purge_fbo(fbo);
return;
}
if (fbo->fb)
gles2: Fixed color conversion for the formats except 1555 and 2101010. This patch fixed two major problems when we do the color convesion with GLES2. 1. lack of necessary formats in FBO pool. GLES2 has three different possible texture formats, GL_RGBA, GL_BGRA and GL_ALPHA. Previous implementation only has one bucket for all the three formats which may reuse a incorrect texture format when do the cache lookup. After this fix, we can enable fbo safely when running with GLES2. 2. Refine the format matching method in glamor_get_tex_format_type_from_pictformat. If both revertion and swap_rb are needed, for example use GL_RGBA to represent PICT_b8g8r8a8. Then the downloading and uploading should be handled differently. The picture's format is PICT_b8g8r8a8, Then the expecting color layout is as below (little endian): 0 1 2 3 : address a r g b Now the in GLES2 the supported color format is GL_RGBA, type is GL_UNSIGNED_TYPE, then we need to shuffle the fragment color as : frag_color = sample(texture).argb; before we use glReadPixel to get it back. For the uploading process, the shuffle is a revert shuffle. We still use GL_RGBA, GL_UNSIGNED_BYTE to upload the color to a texture, then let's see 0 1 2 3 : address a r g b : correct colors R G B A : GL_RGBA with GL_UNSIGNED_BYTE Now we need to shuffle again, the mapping rule is r = G, g = B, b = A, a = R. Then the uploading shuffle is as below: frag_color = sample(texture).gbar; After this commit, gles2 version can pass render check with all the formats except those 1555/2101010. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-04-05 15:53:44 +02:00
cache = &fbo->glamor_priv->fbo_cache[n_format]
glamor-pixmap-upload: Create a uploading fbo with a texture only. Just an initial implementation and disabled by default. When uploading a pixmap to a texture, we don't really want to attach the texture to any fbo. So add one fbo type which doesn't has a gl FBO attached to it. This commit can increase the cairo-trace's performance by 10-20%. Now the firefox-planet-gnome is 8.3s. SNA is still the best, only take 3.5s. Thanks for Chris to point out the A1 pixmap uploading bug. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-20 13:38:29 +01:00
[cache_wbucket(fbo->width)]
[cache_hbucket(fbo->height)];
else
gles2: Fixed color conversion for the formats except 1555 and 2101010. This patch fixed two major problems when we do the color convesion with GLES2. 1. lack of necessary formats in FBO pool. GLES2 has three different possible texture formats, GL_RGBA, GL_BGRA and GL_ALPHA. Previous implementation only has one bucket for all the three formats which may reuse a incorrect texture format when do the cache lookup. After this fix, we can enable fbo safely when running with GLES2. 2. Refine the format matching method in glamor_get_tex_format_type_from_pictformat. If both revertion and swap_rb are needed, for example use GL_RGBA to represent PICT_b8g8r8a8. Then the downloading and uploading should be handled differently. The picture's format is PICT_b8g8r8a8, Then the expecting color layout is as below (little endian): 0 1 2 3 : address a r g b Now the in GLES2 the supported color format is GL_RGBA, type is GL_UNSIGNED_TYPE, then we need to shuffle the fragment color as : frag_color = sample(texture).argb; before we use glReadPixel to get it back. For the uploading process, the shuffle is a revert shuffle. We still use GL_RGBA, GL_UNSIGNED_BYTE to upload the color to a texture, then let's see 0 1 2 3 : address a r g b : correct colors R G B A : GL_RGBA with GL_UNSIGNED_BYTE Now we need to shuffle again, the mapping rule is r = G, g = B, b = A, a = R. Then the uploading shuffle is as below: frag_color = sample(texture).gbar; After this commit, gles2 version can pass render check with all the formats except those 1555/2101010. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-04-05 15:53:44 +02:00
cache = &fbo->glamor_priv->tex_cache[n_format]
glamor-pixmap-upload: Create a uploading fbo with a texture only. Just an initial implementation and disabled by default. When uploading a pixmap to a texture, we don't really want to attach the texture to any fbo. So add one fbo type which doesn't has a gl FBO attached to it. This commit can increase the cairo-trace's performance by 10-20%. Now the firefox-planet-gnome is 8.3s. SNA is still the best, only take 3.5s. Thanks for Chris to point out the A1 pixmap uploading bug. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-20 13:38:29 +01:00
[cache_wbucket(fbo->width)]
[cache_hbucket(fbo->height)];
glamor-fbo-pool: Implement fbo cache mechanism. We classify the cache according to the texture's format/width/height. As openGL doesn't allow us to change a texture's format/width/height after the internal texture object is already allocated, we can't just calculate the size and then according ths size to put the fbo to an bucket which is just like SNA does. We can only put the fbo to the corresponding format/width/height bucket. This commit only support the exact size match. The following patch will remove this restriction, just need to handle the repeat/tile case when the size is not exactly match. Should use fls instead of ffs when decide the width/height bucket, thanks for Chris to point this out. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-19 13:47:55 +01:00
DEBUGF("Put cache entry %p to cache %p w %d h %d format %x fbo %d tex %d \n", fbo, cache,
fbo->width, fbo->height, fbo->format, fbo->fb, fbo->tex);
glamor_fbo: Added a threshold value for the fbo cache pool. Currently set it to 256MB. If cache pool watermark increases to this value, then don't push any fbo to this pool, will purge the fbo directly. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-04-26 12:31:07 +02:00
fbo->glamor_priv->fbo_cache_watermark += fbo->width * fbo->height;
Fixup For list.h change in xorg Because the file list.h in xorg/include has changed the functions and struct names, adding xorg_ prefix before the original name. So Modify glamor_screen_private struct and the code which use list's functions in glamor_fbo.c. We hack at glamor_priv.h avoid the compile error when using old version xserver header file. Signed-off-by: Junyan He <junyan.he@linux.intel.com>
2012-03-05 01:24:20 +01:00
xorg_list_add(&fbo->list, cache);
glamor-fbo-pool: Implement fbo cache mechanism. We classify the cache according to the texture's format/width/height. As openGL doesn't allow us to change a texture's format/width/height after the internal texture object is already allocated, we can't just calculate the size and then according ths size to put the fbo to an bucket which is just like SNA does. We can only put the fbo to the corresponding format/width/height bucket. This commit only support the exact size match. The following patch will remove this restriction, just need to handle the repeat/tile case when the size is not exactly match. Should use fls instead of ffs when decide the width/height bucket, thanks for Chris to point this out. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-19 13:47:55 +01:00
fbo->expire = fbo->glamor_priv->tick + GLAMOR_CACHE_EXPIRE_MAX;
glamor_fbo: Added one macro to disable fbo cache. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-03-16 09:42:46 +01:00
#endif
glamor-fbo-pool: Implement fbo cache mechanism. We classify the cache according to the texture's format/width/height. As openGL doesn't allow us to change a texture's format/width/height after the internal texture object is already allocated, we can't just calculate the size and then according ths size to put the fbo to an bucket which is just like SNA does. We can only put the fbo to the corresponding format/width/height bucket. This commit only support the exact size match. The following patch will remove this restriction, just need to handle the repeat/tile case when the size is not exactly match. Should use fls instead of ffs when decide the width/height bucket, thanks for Chris to point this out. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-19 13:47:55 +01:00
}
fallback_optimize: Prepare for downloading/uploading subregion. Introduced two function glamor_get_sub_pixmap/glamor_put_sub_pixmap, can easily used to get and put sub region of a big textured pixmap. And it can use pbo if possible. To support download a big textured pixmap's sub region to another pixmap's pbo, we introduce a new type of pixmap GLAMOR_MEMORY_MAP. This type of pixmap has a valid devPrivate.ptr pointer, and that pointer points to a pbo mapped address. Now, we are ready to refine those glamor_prepare_access/glamor_finish_access pairs. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-04-09 14:16:07 +02:00
static void
glamor_pixmap_ensure_fb(glamor_pixmap_fbo *fbo)
{
glamor_gl_dispatch *dispatch;
int status;
dispatch = glamor_get_dispatch(fbo->glamor_priv);
if (fbo->fb == 0)
dispatch->glGenFramebuffers(1, &fbo->fb);
assert(fbo->tex != 0);
dispatch->glBindFramebuffer(GL_FRAMEBUFFER, fbo->fb);
dispatch->glFramebufferTexture2D(GL_FRAMEBUFFER,
GL_COLOR_ATTACHMENT0,
GL_TEXTURE_2D, fbo->tex,
0);
status = dispatch->glCheckFramebufferStatus(GL_FRAMEBUFFER);
if (status != GL_FRAMEBUFFER_COMPLETE) {
const char *str;
switch (status) {
case GL_FRAMEBUFFER_INCOMPLETE_ATTACHMENT:
str = "incomplete attachment";
break;
case GL_FRAMEBUFFER_INCOMPLETE_MISSING_ATTACHMENT:
str = "incomplete/missing attachment";
break;
case GL_FRAMEBUFFER_INCOMPLETE_DRAW_BUFFER:
str = "incomplete draw buffer";
break;
case GL_FRAMEBUFFER_INCOMPLETE_READ_BUFFER:
str = "incomplete read buffer";
break;
case GL_FRAMEBUFFER_UNSUPPORTED:
str = "unsupported";
break;
case GL_FRAMEBUFFER_INCOMPLETE_MULTISAMPLE:
str = "incomplete multiple";
break;
default:
str = "unknown error";
break;
}
FatalError("destination is framebuffer incomplete: %s [%#x]\n",
str, status);
}
glamor_put_dispatch(fbo->glamor_priv);
}
glamor_fbo: Introduce glamor fbo to manage all the fb/tex. This is the first patch to implement a fbo/tex pool mechanism which is like the sna's BO cache list. We firstly need to decopule the fbo/tex from each pixmap. The new glamor_pixmap_fbo data structure is for that purpose. It's somehow independent to each pixmap and can be reused latter by other pixmaps once it's detached from the current pixmap. And this commit also slightly change the way to create a memory pixmap. We will not create a pixmap private data structure by default, instead we will crete that structure when a memory pixmap is attaching a fbo to it. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-18 12:21:36 +01:00
glamor_pixmap_fbo *
glamor_create_fbo_from_tex(glamor_screen_private *glamor_priv,
gles2: Fixed color conversion for the formats except 1555 and 2101010. This patch fixed two major problems when we do the color convesion with GLES2. 1. lack of necessary formats in FBO pool. GLES2 has three different possible texture formats, GL_RGBA, GL_BGRA and GL_ALPHA. Previous implementation only has one bucket for all the three formats which may reuse a incorrect texture format when do the cache lookup. After this fix, we can enable fbo safely when running with GLES2. 2. Refine the format matching method in glamor_get_tex_format_type_from_pictformat. If both revertion and swap_rb are needed, for example use GL_RGBA to represent PICT_b8g8r8a8. Then the downloading and uploading should be handled differently. The picture's format is PICT_b8g8r8a8, Then the expecting color layout is as below (little endian): 0 1 2 3 : address a r g b Now the in GLES2 the supported color format is GL_RGBA, type is GL_UNSIGNED_TYPE, then we need to shuffle the fragment color as : frag_color = sample(texture).argb; before we use glReadPixel to get it back. For the uploading process, the shuffle is a revert shuffle. We still use GL_RGBA, GL_UNSIGNED_BYTE to upload the color to a texture, then let's see 0 1 2 3 : address a r g b : correct colors R G B A : GL_RGBA with GL_UNSIGNED_BYTE Now we need to shuffle again, the mapping rule is r = G, g = B, b = A, a = R. Then the uploading shuffle is as below: frag_color = sample(texture).gbar; After this commit, gles2 version can pass render check with all the formats except those 1555/2101010. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-04-05 15:53:44 +02:00
int w, int h, GLenum format, GLint tex, int flag)
glamor_fbo: Introduce glamor fbo to manage all the fb/tex. This is the first patch to implement a fbo/tex pool mechanism which is like the sna's BO cache list. We firstly need to decopule the fbo/tex from each pixmap. The new glamor_pixmap_fbo data structure is for that purpose. It's somehow independent to each pixmap and can be reused latter by other pixmaps once it's detached from the current pixmap. And this commit also slightly change the way to create a memory pixmap. We will not create a pixmap private data structure by default, instead we will crete that structure when a memory pixmap is attaching a fbo to it. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-18 12:21:36 +01:00
{
glamor_pixmap_fbo *fbo;
fbo = calloc(1, sizeof(*fbo));
if (fbo == NULL)
return NULL;
Fixup For list.h change in xorg Because the file list.h in xorg/include has changed the functions and struct names, adding xorg_ prefix before the original name. So Modify glamor_screen_private struct and the code which use list's functions in glamor_fbo.c. We hack at glamor_priv.h avoid the compile error when using old version xserver header file. Signed-off-by: Junyan He <junyan.he@linux.intel.com>
2012-03-05 01:24:20 +01:00
xorg_list_init(&fbo->list);
glamor_fbo: Introduce glamor fbo to manage all the fb/tex. This is the first patch to implement a fbo/tex pool mechanism which is like the sna's BO cache list. We firstly need to decopule the fbo/tex from each pixmap. The new glamor_pixmap_fbo data structure is for that purpose. It's somehow independent to each pixmap and can be reused latter by other pixmaps once it's detached from the current pixmap. And this commit also slightly change the way to create a memory pixmap. We will not create a pixmap private data structure by default, instead we will crete that structure when a memory pixmap is attaching a fbo to it. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-18 12:21:36 +01:00
fbo->tex = tex;
fbo->width = w;
fbo->height = h;
fbo->format = format;
fbo->glamor_priv = glamor_priv;
fallback_optimize: Prepare for downloading/uploading subregion. Introduced two function glamor_get_sub_pixmap/glamor_put_sub_pixmap, can easily used to get and put sub region of a big textured pixmap. And it can use pbo if possible. To support download a big textured pixmap's sub region to another pixmap's pbo, we introduce a new type of pixmap GLAMOR_MEMORY_MAP. This type of pixmap has a valid devPrivate.ptr pointer, and that pointer points to a pbo mapped address. Now, we are ready to refine those glamor_prepare_access/glamor_finish_access pairs. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-04-09 14:16:07 +02:00
if (flag == GLAMOR_CREATE_PIXMAP_MAP) {
glamor_gl_dispatch *dispatch;
dispatch = glamor_get_dispatch(glamor_priv);
dispatch->glGenBuffers(1, &fbo->pbo);
glamor_put_dispatch(glamor_priv);
goto done;
}
glamor-pixmap-upload: Create a uploading fbo with a texture only. Just an initial implementation and disabled by default. When uploading a pixmap to a texture, we don't really want to attach the texture to any fbo. So add one fbo type which doesn't has a gl FBO attached to it. This commit can increase the cairo-trace's performance by 10-20%. Now the firefox-planet-gnome is 8.3s. SNA is still the best, only take 3.5s. Thanks for Chris to point out the A1 pixmap uploading bug. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-20 13:38:29 +01:00
if (flag != GLAMOR_CREATE_FBO_NO_FBO)
glamor_pixmap_ensure_fb(fbo);
glamor_fbo: Introduce glamor fbo to manage all the fb/tex. This is the first patch to implement a fbo/tex pool mechanism which is like the sna's BO cache list. We firstly need to decopule the fbo/tex from each pixmap. The new glamor_pixmap_fbo data structure is for that purpose. It's somehow independent to each pixmap and can be reused latter by other pixmaps once it's detached from the current pixmap. And this commit also slightly change the way to create a memory pixmap. We will not create a pixmap private data structure by default, instead we will crete that structure when a memory pixmap is attaching a fbo to it. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-18 12:21:36 +01:00
fallback_optimize: Prepare for downloading/uploading subregion. Introduced two function glamor_get_sub_pixmap/glamor_put_sub_pixmap, can easily used to get and put sub region of a big textured pixmap. And it can use pbo if possible. To support download a big textured pixmap's sub region to another pixmap's pbo, we introduce a new type of pixmap GLAMOR_MEMORY_MAP. This type of pixmap has a valid devPrivate.ptr pointer, and that pointer points to a pbo mapped address. Now, we are ready to refine those glamor_prepare_access/glamor_finish_access pairs. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-04-09 14:16:07 +02:00
done:
glamor_fbo: Introduce glamor fbo to manage all the fb/tex. This is the first patch to implement a fbo/tex pool mechanism which is like the sna's BO cache list. We firstly need to decopule the fbo/tex from each pixmap. The new glamor_pixmap_fbo data structure is for that purpose. It's somehow independent to each pixmap and can be reused latter by other pixmaps once it's detached from the current pixmap. And this commit also slightly change the way to create a memory pixmap. We will not create a pixmap private data structure by default, instead we will crete that structure when a memory pixmap is attaching a fbo to it. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-18 12:21:36 +01:00
return fbo;
}
glamor-fbo-pool: Implement fbo cache mechanism. We classify the cache according to the texture's format/width/height. As openGL doesn't allow us to change a texture's format/width/height after the internal texture object is already allocated, we can't just calculate the size and then according ths size to put the fbo to an bucket which is just like SNA does. We can only put the fbo to the corresponding format/width/height bucket. This commit only support the exact size match. The following patch will remove this restriction, just need to handle the repeat/tile case when the size is not exactly match. Should use fls instead of ffs when decide the width/height bucket, thanks for Chris to point this out. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-19 13:47:55 +01:00
void
glamor_fbo_expire(glamor_screen_private *glamor_priv)
{
Fixup For list.h change in xorg Because the file list.h in xorg/include has changed the functions and struct names, adding xorg_ prefix before the original name. So Modify glamor_screen_private struct and the code which use list's functions in glamor_fbo.c. We hack at glamor_priv.h avoid the compile error when using old version xserver header file. Signed-off-by: Junyan He <junyan.he@linux.intel.com>
2012-03-05 01:24:20 +01:00
struct xorg_list *cache;
glamor-fbo-pool: Implement fbo cache mechanism. We classify the cache according to the texture's format/width/height. As openGL doesn't allow us to change a texture's format/width/height after the internal texture object is already allocated, we can't just calculate the size and then according ths size to put the fbo to an bucket which is just like SNA does. We can only put the fbo to the corresponding format/width/height bucket. This commit only support the exact size match. The following patch will remove this restriction, just need to handle the repeat/tile case when the size is not exactly match. Should use fls instead of ffs when decide the width/height bucket, thanks for Chris to point this out. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-19 13:47:55 +01:00
glamor_pixmap_fbo *fbo_entry, *tmp;
int i,j,k;
int empty_cache = TRUE;
for(i = 0; i < CACHE_FORMAT_COUNT; i++)
for(j = 0; j < CACHE_BUCKET_WCOUNT; j++)
for(k = 0; k < CACHE_BUCKET_HCOUNT; k++) {
cache = &glamor_priv->fbo_cache[i][j][k];
Fixup For list.h change in xorg Because the file list.h in xorg/include has changed the functions and struct names, adding xorg_ prefix before the original name. So Modify glamor_screen_private struct and the code which use list's functions in glamor_fbo.c. We hack at glamor_priv.h avoid the compile error when using old version xserver header file. Signed-off-by: Junyan He <junyan.he@linux.intel.com>
2012-03-05 01:24:20 +01:00
xorg_list_for_each_entry_safe_reverse(fbo_entry, tmp, cache, list) {
glamor-fbo-pool: Implement fbo cache mechanism. We classify the cache according to the texture's format/width/height. As openGL doesn't allow us to change a texture's format/width/height after the internal texture object is already allocated, we can't just calculate the size and then according ths size to put the fbo to an bucket which is just like SNA does. We can only put the fbo to the corresponding format/width/height bucket. This commit only support the exact size match. The following patch will remove this restriction, just need to handle the repeat/tile case when the size is not exactly match. Should use fls instead of ffs when decide the width/height bucket, thanks for Chris to point this out. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-19 13:47:55 +01:00
if (GLAMOR_TICK_AFTER(fbo_entry->expire, glamor_priv->tick)) {
empty_cache = FALSE;
break;
}
glamor_fbo: Added a threshold value for the fbo cache pool. Currently set it to 256MB. If cache pool watermark increases to this value, then don't push any fbo to this pool, will purge the fbo directly. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-04-26 12:31:07 +02:00
glamor_priv->fbo_cache_watermark -= fbo_entry->width * fbo_entry->height;
Fixup For list.h change in xorg Because the file list.h in xorg/include has changed the functions and struct names, adding xorg_ prefix before the original name. So Modify glamor_screen_private struct and the code which use list's functions in glamor_fbo.c. We hack at glamor_priv.h avoid the compile error when using old version xserver header file. Signed-off-by: Junyan He <junyan.he@linux.intel.com>
2012-03-05 01:24:20 +01:00
xorg_list_del(&fbo_entry->list);
glamor-fbo-pool: Implement fbo cache mechanism. We classify the cache according to the texture's format/width/height. As openGL doesn't allow us to change a texture's format/width/height after the internal texture object is already allocated, we can't just calculate the size and then according ths size to put the fbo to an bucket which is just like SNA does. We can only put the fbo to the corresponding format/width/height bucket. This commit only support the exact size match. The following patch will remove this restriction, just need to handle the repeat/tile case when the size is not exactly match. Should use fls instead of ffs when decide the width/height bucket, thanks for Chris to point this out. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-19 13:47:55 +01:00
DEBUGF("cache %p fbo %p expired %d current %d \n", cache, fbo_entry,
fbo_entry->expire, glamor_priv->tick);
glamor_purge_fbo(fbo_entry);
}
glamor-pixmap-upload: Create a uploading fbo with a texture only. Just an initial implementation and disabled by default. When uploading a pixmap to a texture, we don't really want to attach the texture to any fbo. So add one fbo type which doesn't has a gl FBO attached to it. This commit can increase the cairo-trace's performance by 10-20%. Now the firefox-planet-gnome is 8.3s. SNA is still the best, only take 3.5s. Thanks for Chris to point out the A1 pixmap uploading bug. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-20 13:38:29 +01:00
#if 0
cache = &glamor_priv->tex_cache[i][j][k];
Fixup For list.h change in xorg Because the file list.h in xorg/include has changed the functions and struct names, adding xorg_ prefix before the original name. So Modify glamor_screen_private struct and the code which use list's functions in glamor_fbo.c. We hack at glamor_priv.h avoid the compile error when using old version xserver header file. Signed-off-by: Junyan He <junyan.he@linux.intel.com>
2012-03-05 01:24:20 +01:00
xorg_list_for_each_entry_safe_reverse(fbo_entry, tmp, cache, list) {
glamor-pixmap-upload: Create a uploading fbo with a texture only. Just an initial implementation and disabled by default. When uploading a pixmap to a texture, we don't really want to attach the texture to any fbo. So add one fbo type which doesn't has a gl FBO attached to it. This commit can increase the cairo-trace's performance by 10-20%. Now the firefox-planet-gnome is 8.3s. SNA is still the best, only take 3.5s. Thanks for Chris to point out the A1 pixmap uploading bug. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-20 13:38:29 +01:00
if (GLAMOR_TICK_AFTER(fbo_entry->expire, glamor_priv->tick)) {
empty_cache = FALSE;
break;
}
Fixup For list.h change in xorg Because the file list.h in xorg/include has changed the functions and struct names, adding xorg_ prefix before the original name. So Modify glamor_screen_private struct and the code which use list's functions in glamor_fbo.c. We hack at glamor_priv.h avoid the compile error when using old version xserver header file. Signed-off-by: Junyan He <junyan.he@linux.intel.com>
2012-03-05 01:24:20 +01:00
xorg_list_del(&fbo_entry->list);
glamor-pixmap-upload: Create a uploading fbo with a texture only. Just an initial implementation and disabled by default. When uploading a pixmap to a texture, we don't really want to attach the texture to any fbo. So add one fbo type which doesn't has a gl FBO attached to it. This commit can increase the cairo-trace's performance by 10-20%. Now the firefox-planet-gnome is 8.3s. SNA is still the best, only take 3.5s. Thanks for Chris to point out the A1 pixmap uploading bug. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-20 13:38:29 +01:00
DEBUGF("cache %p fbo %p expired %d current %d \n", cache, fbo_entry,
fbo_entry->expire, glamor_priv->tick);
glamor_purge_fbo(fbo_entry);
}
#endif
glamor-fbo-pool: Implement fbo cache mechanism. We classify the cache according to the texture's format/width/height. As openGL doesn't allow us to change a texture's format/width/height after the internal texture object is already allocated, we can't just calculate the size and then according ths size to put the fbo to an bucket which is just like SNA does. We can only put the fbo to the corresponding format/width/height bucket. This commit only support the exact size match. The following patch will remove this restriction, just need to handle the repeat/tile case when the size is not exactly match. Should use fls instead of ffs when decide the width/height bucket, thanks for Chris to point this out. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-19 13:47:55 +01:00
}
}
void
glamor_init_pixmap_fbo(ScreenPtr screen)
{
glamor_screen_private *glamor_priv;
int i,j,k;
glamor_priv = glamor_get_screen_private(screen);
for(i = 0; i < CACHE_FORMAT_COUNT; i++)
for(j = 0; j < CACHE_BUCKET_WCOUNT; j++)
for(k = 0; k < CACHE_BUCKET_HCOUNT; k++)
{
Fixup For list.h change in xorg Because the file list.h in xorg/include has changed the functions and struct names, adding xorg_ prefix before the original name. So Modify glamor_screen_private struct and the code which use list's functions in glamor_fbo.c. We hack at glamor_priv.h avoid the compile error when using old version xserver header file. Signed-off-by: Junyan He <junyan.he@linux.intel.com>
2012-03-05 01:24:20 +01:00
xorg_list_init(&glamor_priv->fbo_cache[i][j][k]);
xorg_list_init(&glamor_priv->tex_cache[i][j][k]);
glamor-fbo-pool: Implement fbo cache mechanism. We classify the cache according to the texture's format/width/height. As openGL doesn't allow us to change a texture's format/width/height after the internal texture object is already allocated, we can't just calculate the size and then according ths size to put the fbo to an bucket which is just like SNA does. We can only put the fbo to the corresponding format/width/height bucket. This commit only support the exact size match. The following patch will remove this restriction, just need to handle the repeat/tile case when the size is not exactly match. Should use fls instead of ffs when decide the width/height bucket, thanks for Chris to point this out. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-19 13:47:55 +01:00
}
glamor_fbo: Added a threshold value for the fbo cache pool. Currently set it to 256MB. If cache pool watermark increases to this value, then don't push any fbo to this pool, will purge the fbo directly. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-04-26 12:31:07 +02:00
glamor_priv->fbo_cache_watermark = 0;
glamor-fbo-pool: Implement fbo cache mechanism. We classify the cache according to the texture's format/width/height. As openGL doesn't allow us to change a texture's format/width/height after the internal texture object is already allocated, we can't just calculate the size and then according ths size to put the fbo to an bucket which is just like SNA does. We can only put the fbo to the corresponding format/width/height bucket. This commit only support the exact size match. The following patch will remove this restriction, just need to handle the repeat/tile case when the size is not exactly match. Should use fls instead of ffs when decide the width/height bucket, thanks for Chris to point this out. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-19 13:47:55 +01:00
}
void
glamor_fini_pixmap_fbo(ScreenPtr screen)
{
Fixup For list.h change in xorg Because the file list.h in xorg/include has changed the functions and struct names, adding xorg_ prefix before the original name. So Modify glamor_screen_private struct and the code which use list's functions in glamor_fbo.c. We hack at glamor_priv.h avoid the compile error when using old version xserver header file. Signed-off-by: Junyan He <junyan.he@linux.intel.com>
2012-03-05 01:24:20 +01:00
struct xorg_list *cache;
glamor-fbo-pool: Implement fbo cache mechanism. We classify the cache according to the texture's format/width/height. As openGL doesn't allow us to change a texture's format/width/height after the internal texture object is already allocated, we can't just calculate the size and then according ths size to put the fbo to an bucket which is just like SNA does. We can only put the fbo to the corresponding format/width/height bucket. This commit only support the exact size match. The following patch will remove this restriction, just need to handle the repeat/tile case when the size is not exactly match. Should use fls instead of ffs when decide the width/height bucket, thanks for Chris to point this out. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-19 13:47:55 +01:00
glamor_screen_private *glamor_priv;
glamor_pixmap_fbo *fbo_entry, *tmp;
int i,j,k;
glamor_priv = glamor_get_screen_private(screen);
for(i = 0; i < CACHE_FORMAT_COUNT; i++)
for(j = 0; j < CACHE_BUCKET_WCOUNT; j++)
for(k = 0; k < CACHE_BUCKET_HCOUNT; k++)
{
cache = &glamor_priv->fbo_cache[i][j][k];
Fixup For list.h change in xorg Because the file list.h in xorg/include has changed the functions and struct names, adding xorg_ prefix before the original name. So Modify glamor_screen_private struct and the code which use list's functions in glamor_fbo.c. We hack at glamor_priv.h avoid the compile error when using old version xserver header file. Signed-off-by: Junyan He <junyan.he@linux.intel.com>
2012-03-05 01:24:20 +01:00
xorg_list_for_each_entry_safe_reverse(fbo_entry, tmp, cache, list) {
xorg_list_del(&fbo_entry->list);
glamor-fbo-pool: Implement fbo cache mechanism. We classify the cache according to the texture's format/width/height. As openGL doesn't allow us to change a texture's format/width/height after the internal texture object is already allocated, we can't just calculate the size and then according ths size to put the fbo to an bucket which is just like SNA does. We can only put the fbo to the corresponding format/width/height bucket. This commit only support the exact size match. The following patch will remove this restriction, just need to handle the repeat/tile case when the size is not exactly match. Should use fls instead of ffs when decide the width/height bucket, thanks for Chris to point this out. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-19 13:47:55 +01:00
glamor_purge_fbo(fbo_entry);
}
glamor-pixmap-upload: Create a uploading fbo with a texture only. Just an initial implementation and disabled by default. When uploading a pixmap to a texture, we don't really want to attach the texture to any fbo. So add one fbo type which doesn't has a gl FBO attached to it. This commit can increase the cairo-trace's performance by 10-20%. Now the firefox-planet-gnome is 8.3s. SNA is still the best, only take 3.5s. Thanks for Chris to point out the A1 pixmap uploading bug. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-20 13:38:29 +01:00
#if 0
cache = &glamor_priv->tex_cache[i][j][k];
Fixup For list.h change in xorg Because the file list.h in xorg/include has changed the functions and struct names, adding xorg_ prefix before the original name. So Modify glamor_screen_private struct and the code which use list's functions in glamor_fbo.c. We hack at glamor_priv.h avoid the compile error when using old version xserver header file. Signed-off-by: Junyan He <junyan.he@linux.intel.com>
2012-03-05 01:24:20 +01:00
xorg_list_for_each_entry_safe_reverse(fbo_entry, tmp, cache, list) {
xorg_list_del(&fbo_entry->list);
glamor-pixmap-upload: Create a uploading fbo with a texture only. Just an initial implementation and disabled by default. When uploading a pixmap to a texture, we don't really want to attach the texture to any fbo. So add one fbo type which doesn't has a gl FBO attached to it. This commit can increase the cairo-trace's performance by 10-20%. Now the firefox-planet-gnome is 8.3s. SNA is still the best, only take 3.5s. Thanks for Chris to point out the A1 pixmap uploading bug. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-20 13:38:29 +01:00
glamor_purge_fbo(fbo_entry);
}
#endif
glamor-fbo-pool: Implement fbo cache mechanism. We classify the cache according to the texture's format/width/height. As openGL doesn't allow us to change a texture's format/width/height after the internal texture object is already allocated, we can't just calculate the size and then according ths size to put the fbo to an bucket which is just like SNA does. We can only put the fbo to the corresponding format/width/height bucket. This commit only support the exact size match. The following patch will remove this restriction, just need to handle the repeat/tile case when the size is not exactly match. Should use fls instead of ffs when decide the width/height bucket, thanks for Chris to point this out. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-19 13:47:55 +01:00
}
}
void
glamor_destroy_fbo(glamor_pixmap_fbo *fbo)
{
Fixup For list.h change in xorg Because the file list.h in xorg/include has changed the functions and struct names, adding xorg_ prefix before the original name. So Modify glamor_screen_private struct and the code which use list's functions in glamor_fbo.c. We hack at glamor_priv.h avoid the compile error when using old version xserver header file. Signed-off-by: Junyan He <junyan.he@linux.intel.com>
2012-03-05 01:24:20 +01:00
xorg_list_del(&fbo->list);
glamor-fbo-pool: Implement fbo cache mechanism. We classify the cache according to the texture's format/width/height. As openGL doesn't allow us to change a texture's format/width/height after the internal texture object is already allocated, we can't just calculate the size and then according ths size to put the fbo to an bucket which is just like SNA does. We can only put the fbo to the corresponding format/width/height bucket. This commit only support the exact size match. The following patch will remove this restriction, just need to handle the repeat/tile case when the size is not exactly match. Should use fls instead of ffs when decide the width/height bucket, thanks for Chris to point this out. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-19 13:47:55 +01:00
glamor_pixmap_fbo_cache_put(fbo);
}
glamor-pixmap-upload: Create a uploading fbo with a texture only. Just an initial implementation and disabled by default. When uploading a pixmap to a texture, we don't really want to attach the texture to any fbo. So add one fbo type which doesn't has a gl FBO attached to it. This commit can increase the cairo-trace's performance by 10-20%. Now the firefox-planet-gnome is 8.3s. SNA is still the best, only take 3.5s. Thanks for Chris to point out the A1 pixmap uploading bug. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-20 13:38:29 +01:00
glamor_pixmap_fbo *
glamor_create_tex_obj(glamor_screen_private *glamor_priv,
int w, int h, GLenum format, int flag)
{
glamor_gl_dispatch *dispatch;
glamor_pixmap_fbo *fbo;
int cache_flag = GLAMOR_CACHE_TEXTURE;
GLuint tex;
if (flag == GLAMOR_CREATE_TEXTURE_EXACT_SIZE)
cache_flag |= GLAMOR_CACHE_EXACT_SIZE;
fbo = glamor_pixmap_fbo_cache_get(glamor_priv, w, h,
format, cache_flag);
if (fbo)
return fbo;
fbo = calloc(1, sizeof(*fbo));
if (fbo == NULL)
return NULL;
Fixup For list.h change in xorg Because the file list.h in xorg/include has changed the functions and struct names, adding xorg_ prefix before the original name. So Modify glamor_screen_private struct and the code which use list's functions in glamor_fbo.c. We hack at glamor_priv.h avoid the compile error when using old version xserver header file. Signed-off-by: Junyan He <junyan.he@linux.intel.com>
2012-03-05 01:24:20 +01:00
xorg_list_init(&fbo->list);
glamor-pixmap-upload: Create a uploading fbo with a texture only. Just an initial implementation and disabled by default. When uploading a pixmap to a texture, we don't really want to attach the texture to any fbo. So add one fbo type which doesn't has a gl FBO attached to it. This commit can increase the cairo-trace's performance by 10-20%. Now the firefox-planet-gnome is 8.3s. SNA is still the best, only take 3.5s. Thanks for Chris to point out the A1 pixmap uploading bug. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-20 13:38:29 +01:00
Fixup glx support Renaming glamor_priv->dispatch and wrapping the access to the dispatch table with a function that also ensured the context was bound. dispatch = glamor_get_dispatch(glamor_priv); ... glamor_put_dispatch(glamor_priv); So that we catch all places where we attempt to call into GL withouta context. As an optimisation we can then do glamor_get_context(); glamor_put_context() around the rendering entry points to reduce the frequency of having to restore the old context. (Along with allowing the context to be recursively acquired and making the old context part of the glamor_egl state.) Reviewed-by: Zhigang Gong <zhigang.gong@linux.intel.com> Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-02-10 13:54:44 +01:00
dispatch = glamor_get_dispatch(glamor_priv);
glamor-pixmap-upload: Create a uploading fbo with a texture only. Just an initial implementation and disabled by default. When uploading a pixmap to a texture, we don't really want to attach the texture to any fbo. So add one fbo type which doesn't has a gl FBO attached to it. This commit can increase the cairo-trace's performance by 10-20%. Now the firefox-planet-gnome is 8.3s. SNA is still the best, only take 3.5s. Thanks for Chris to point out the A1 pixmap uploading bug. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-20 13:38:29 +01:00
dispatch->glGenTextures(1, &tex);
dispatch->glBindTexture(GL_TEXTURE_2D, tex);
dispatch->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER,
GL_NEAREST);
dispatch->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER,
GL_NEAREST);
dispatch->glTexImage2D(GL_TEXTURE_2D, 0, format, w, h, 0, format,
GL_UNSIGNED_BYTE, NULL);
Fixup glx support Renaming glamor_priv->dispatch and wrapping the access to the dispatch table with a function that also ensured the context was bound. dispatch = glamor_get_dispatch(glamor_priv); ... glamor_put_dispatch(glamor_priv); So that we catch all places where we attempt to call into GL withouta context. As an optimisation we can then do glamor_get_context(); glamor_put_context() around the rendering entry points to reduce the frequency of having to restore the old context. (Along with allowing the context to be recursively acquired and making the old context part of the glamor_egl state.) Reviewed-by: Zhigang Gong <zhigang.gong@linux.intel.com> Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-02-10 13:54:44 +01:00
glamor_put_dispatch(glamor_priv);
glamor-pixmap-upload: Create a uploading fbo with a texture only. Just an initial implementation and disabled by default. When uploading a pixmap to a texture, we don't really want to attach the texture to any fbo. So add one fbo type which doesn't has a gl FBO attached to it. This commit can increase the cairo-trace's performance by 10-20%. Now the firefox-planet-gnome is 8.3s. SNA is still the best, only take 3.5s. Thanks for Chris to point out the A1 pixmap uploading bug. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-20 13:38:29 +01:00
fbo->tex = tex;
fbo->width = w;
fbo->height = h;
fbo->format = format;
fbo->glamor_priv = glamor_priv;
return fbo;
}
void
glamor_destroy_tex_obj(glamor_pixmap_fbo * tex_obj)
{
assert(tex_obj->fb == 0);
Fixup For list.h change in xorg Because the file list.h in xorg/include has changed the functions and struct names, adding xorg_ prefix before the original name. So Modify glamor_screen_private struct and the code which use list's functions in glamor_fbo.c. We hack at glamor_priv.h avoid the compile error when using old version xserver header file. Signed-off-by: Junyan He <junyan.he@linux.intel.com>
2012-03-05 01:24:20 +01:00
xorg_list_del(&tex_obj->list);
glamor-pixmap-upload: Create a uploading fbo with a texture only. Just an initial implementation and disabled by default. When uploading a pixmap to a texture, we don't really want to attach the texture to any fbo. So add one fbo type which doesn't has a gl FBO attached to it. This commit can increase the cairo-trace's performance by 10-20%. Now the firefox-planet-gnome is 8.3s. SNA is still the best, only take 3.5s. Thanks for Chris to point out the A1 pixmap uploading bug. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-20 13:38:29 +01:00
glamor_pixmap_fbo_cache_put(tex_obj);
}
fallback_optimize: Prepare for downloading/uploading subregion. Introduced two function glamor_get_sub_pixmap/glamor_put_sub_pixmap, can easily used to get and put sub region of a big textured pixmap. And it can use pbo if possible. To support download a big textured pixmap's sub region to another pixmap's pbo, we introduce a new type of pixmap GLAMOR_MEMORY_MAP. This type of pixmap has a valid devPrivate.ptr pointer, and that pointer points to a pbo mapped address. Now, we are ready to refine those glamor_prepare_access/glamor_finish_access pairs. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-04-09 14:16:07 +02:00
int
_glamor_create_tex(glamor_screen_private *glamor_priv,
int w, int h, GLenum format)
{
glamor_gl_dispatch *dispatch;
Fixed all unused variables warnings. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-04-28 08:55:34 +02:00
unsigned int tex;
fallback_optimize: Prepare for downloading/uploading subregion. Introduced two function glamor_get_sub_pixmap/glamor_put_sub_pixmap, can easily used to get and put sub region of a big textured pixmap. And it can use pbo if possible. To support download a big textured pixmap's sub region to another pixmap's pbo, we introduce a new type of pixmap GLAMOR_MEMORY_MAP. This type of pixmap has a valid devPrivate.ptr pointer, and that pointer points to a pbo mapped address. Now, we are ready to refine those glamor_prepare_access/glamor_finish_access pairs. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-04-09 14:16:07 +02:00
dispatch = glamor_get_dispatch(glamor_priv);
dispatch->glGenTextures(1, &tex);
dispatch->glBindTexture(GL_TEXTURE_2D, tex);
dispatch->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER,
GL_NEAREST);
dispatch->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER,
GL_NEAREST);
dispatch->glTexImage2D(GL_TEXTURE_2D, 0, format, w, h, 0, format,
GL_UNSIGNED_BYTE, NULL);
glamor_put_dispatch(glamor_priv);
return tex;
}
glamor_fbo: Introduce glamor fbo to manage all the fb/tex. This is the first patch to implement a fbo/tex pool mechanism which is like the sna's BO cache list. We firstly need to decopule the fbo/tex from each pixmap. The new glamor_pixmap_fbo data structure is for that purpose. It's somehow independent to each pixmap and can be reused latter by other pixmaps once it's detached from the current pixmap. And this commit also slightly change the way to create a memory pixmap. We will not create a pixmap private data structure by default, instead we will crete that structure when a memory pixmap is attaching a fbo to it. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-18 12:21:36 +01:00
glamor_pixmap_fbo *
glamor_create_fbo(glamor_screen_private *glamor_priv,
gles2: Fixed color conversion for the formats except 1555 and 2101010. This patch fixed two major problems when we do the color convesion with GLES2. 1. lack of necessary formats in FBO pool. GLES2 has three different possible texture formats, GL_RGBA, GL_BGRA and GL_ALPHA. Previous implementation only has one bucket for all the three formats which may reuse a incorrect texture format when do the cache lookup. After this fix, we can enable fbo safely when running with GLES2. 2. Refine the format matching method in glamor_get_tex_format_type_from_pictformat. If both revertion and swap_rb are needed, for example use GL_RGBA to represent PICT_b8g8r8a8. Then the downloading and uploading should be handled differently. The picture's format is PICT_b8g8r8a8, Then the expecting color layout is as below (little endian): 0 1 2 3 : address a r g b Now the in GLES2 the supported color format is GL_RGBA, type is GL_UNSIGNED_TYPE, then we need to shuffle the fragment color as : frag_color = sample(texture).argb; before we use glReadPixel to get it back. For the uploading process, the shuffle is a revert shuffle. We still use GL_RGBA, GL_UNSIGNED_BYTE to upload the color to a texture, then let's see 0 1 2 3 : address a r g b : correct colors R G B A : GL_RGBA with GL_UNSIGNED_BYTE Now we need to shuffle again, the mapping rule is r = G, g = B, b = A, a = R. Then the uploading shuffle is as below: frag_color = sample(texture).gbar; After this commit, gles2 version can pass render check with all the formats except those 1555/2101010. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-04-05 15:53:44 +02:00
int w, int h,
GLenum format,
int flag)
glamor_fbo: Introduce glamor fbo to manage all the fb/tex. This is the first patch to implement a fbo/tex pool mechanism which is like the sna's BO cache list. We firstly need to decopule the fbo/tex from each pixmap. The new glamor_pixmap_fbo data structure is for that purpose. It's somehow independent to each pixmap and can be reused latter by other pixmaps once it's detached from the current pixmap. And this commit also slightly change the way to create a memory pixmap. We will not create a pixmap private data structure by default, instead we will crete that structure when a memory pixmap is attaching a fbo to it. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-18 12:21:36 +01:00
{
glamor_pixmap_fbo *fbo;
fallback_optimize: Prepare for downloading/uploading subregion. Introduced two function glamor_get_sub_pixmap/glamor_put_sub_pixmap, can easily used to get and put sub region of a big textured pixmap. And it can use pbo if possible. To support download a big textured pixmap's sub region to another pixmap's pbo, we introduce a new type of pixmap GLAMOR_MEMORY_MAP. This type of pixmap has a valid devPrivate.ptr pointer, and that pointer points to a pbo mapped address. Now, we are ready to refine those glamor_prepare_access/glamor_finish_access pairs. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-04-09 14:16:07 +02:00
GLint tex = 0;
glamor-fbo-pool: Implement fbo cache mechanism. We classify the cache according to the texture's format/width/height. As openGL doesn't allow us to change a texture's format/width/height after the internal texture object is already allocated, we can't just calculate the size and then according ths size to put the fbo to an bucket which is just like SNA does. We can only put the fbo to the corresponding format/width/height bucket. This commit only support the exact size match. The following patch will remove this restriction, just need to handle the repeat/tile case when the size is not exactly match. Should use fls instead of ffs when decide the width/height bucket, thanks for Chris to point this out. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-19 13:47:55 +01:00
int cache_flag;
gles2: Fixed color conversion for the formats except 1555 and 2101010. This patch fixed two major problems when we do the color convesion with GLES2. 1. lack of necessary formats in FBO pool. GLES2 has three different possible texture formats, GL_RGBA, GL_BGRA and GL_ALPHA. Previous implementation only has one bucket for all the three formats which may reuse a incorrect texture format when do the cache lookup. After this fix, we can enable fbo safely when running with GLES2. 2. Refine the format matching method in glamor_get_tex_format_type_from_pictformat. If both revertion and swap_rb are needed, for example use GL_RGBA to represent PICT_b8g8r8a8. Then the downloading and uploading should be handled differently. The picture's format is PICT_b8g8r8a8, Then the expecting color layout is as below (little endian): 0 1 2 3 : address a r g b Now the in GLES2 the supported color format is GL_RGBA, type is GL_UNSIGNED_TYPE, then we need to shuffle the fragment color as : frag_color = sample(texture).argb; before we use glReadPixel to get it back. For the uploading process, the shuffle is a revert shuffle. We still use GL_RGBA, GL_UNSIGNED_BYTE to upload the color to a texture, then let's see 0 1 2 3 : address a r g b : correct colors R G B A : GL_RGBA with GL_UNSIGNED_BYTE Now we need to shuffle again, the mapping rule is r = G, g = B, b = A, a = R. Then the uploading shuffle is as below: frag_color = sample(texture).gbar; After this commit, gles2 version can pass render check with all the formats except those 1555/2101010. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-04-05 15:53:44 +02:00
if (!glamor_check_fbo_size(glamor_priv, w, h))
glamor_create_fbo: Concentrate the fbo size/depth checking. Concentrate checking the size/depth when creating fbo. Simply the pixmap creation and the uploading fbo/texture preparing. Also slightly change the uploading fbo's preparation. If don't need fbo, then a fbo only has valid texture should be enough to return. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-21 06:22:14 +01:00
return NULL;
glamor-pixmap-upload: Create a uploading fbo with a texture only. Just an initial implementation and disabled by default. When uploading a pixmap to a texture, we don't really want to attach the texture to any fbo. So add one fbo type which doesn't has a gl FBO attached to it. This commit can increase the cairo-trace's performance by 10-20%. Now the firefox-planet-gnome is 8.3s. SNA is still the best, only take 3.5s. Thanks for Chris to point out the A1 pixmap uploading bug. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-20 13:38:29 +01:00
if (flag == GLAMOR_CREATE_FBO_NO_FBO)
goto new_fbo;
fallback_optimize: Prepare for downloading/uploading subregion. Introduced two function glamor_get_sub_pixmap/glamor_put_sub_pixmap, can easily used to get and put sub region of a big textured pixmap. And it can use pbo if possible. To support download a big textured pixmap's sub region to another pixmap's pbo, we introduce a new type of pixmap GLAMOR_MEMORY_MAP. This type of pixmap has a valid devPrivate.ptr pointer, and that pointer points to a pbo mapped address. Now, we are ready to refine those glamor_prepare_access/glamor_finish_access pairs. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-04-09 14:16:07 +02:00
if (flag == GLAMOR_CREATE_PIXMAP_MAP)
goto no_tex;
glamor-fbo-pool: Implement fbo cache mechanism. We classify the cache according to the texture's format/width/height. As openGL doesn't allow us to change a texture's format/width/height after the internal texture object is already allocated, we can't just calculate the size and then according ths size to put the fbo to an bucket which is just like SNA does. We can only put the fbo to the corresponding format/width/height bucket. This commit only support the exact size match. The following patch will remove this restriction, just need to handle the repeat/tile case when the size is not exactly match. Should use fls instead of ffs when decide the width/height bucket, thanks for Chris to point this out. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-19 13:47:55 +01:00
if (flag == GLAMOR_CREATE_PIXMAP_FIXUP)
cache_flag = GLAMOR_CACHE_EXACT_SIZE;
else
cache_flag = 0;
glamor_fbo: Introduce glamor fbo to manage all the fb/tex. This is the first patch to implement a fbo/tex pool mechanism which is like the sna's BO cache list. We firstly need to decopule the fbo/tex from each pixmap. The new glamor_pixmap_fbo data structure is for that purpose. It's somehow independent to each pixmap and can be reused latter by other pixmaps once it's detached from the current pixmap. And this commit also slightly change the way to create a memory pixmap. We will not create a pixmap private data structure by default, instead we will crete that structure when a memory pixmap is attaching a fbo to it. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-18 12:21:36 +01:00
glamor-fbo-pool: Implement fbo cache mechanism. We classify the cache according to the texture's format/width/height. As openGL doesn't allow us to change a texture's format/width/height after the internal texture object is already allocated, we can't just calculate the size and then according ths size to put the fbo to an bucket which is just like SNA does. We can only put the fbo to the corresponding format/width/height bucket. This commit only support the exact size match. The following patch will remove this restriction, just need to handle the repeat/tile case when the size is not exactly match. Should use fls instead of ffs when decide the width/height bucket, thanks for Chris to point this out. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-19 13:47:55 +01:00
fbo = glamor_pixmap_fbo_cache_get(glamor_priv, w, h,
glamor-fbo-pool: Enable to reuse different size fbo/texture. Fixup three special cases, one is in tile and the other is in composite. Both cases are due to repeat texture issue. Maybe we can refine the shader to recalculate texture coords to support partial texture's repeating. The third is when upload a memory pixmap to texture, as now the texture may not have the exact size as the pixmap, we should not use the full rect coords. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-20 02:59:44 +01:00
format, cache_flag);
glamor-fbo-pool: Implement fbo cache mechanism. We classify the cache according to the texture's format/width/height. As openGL doesn't allow us to change a texture's format/width/height after the internal texture object is already allocated, we can't just calculate the size and then according ths size to put the fbo to an bucket which is just like SNA does. We can only put the fbo to the corresponding format/width/height bucket. This commit only support the exact size match. The following patch will remove this restriction, just need to handle the repeat/tile case when the size is not exactly match. Should use fls instead of ffs when decide the width/height bucket, thanks for Chris to point this out. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-19 13:47:55 +01:00
if (fbo)
return fbo;
glamor-pixmap-upload: Create a uploading fbo with a texture only. Just an initial implementation and disabled by default. When uploading a pixmap to a texture, we don't really want to attach the texture to any fbo. So add one fbo type which doesn't has a gl FBO attached to it. This commit can increase the cairo-trace's performance by 10-20%. Now the firefox-planet-gnome is 8.3s. SNA is still the best, only take 3.5s. Thanks for Chris to point out the A1 pixmap uploading bug. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-20 13:38:29 +01:00
new_fbo:
fallback_optimize: Prepare for downloading/uploading subregion. Introduced two function glamor_get_sub_pixmap/glamor_put_sub_pixmap, can easily used to get and put sub region of a big textured pixmap. And it can use pbo if possible. To support download a big textured pixmap's sub region to another pixmap's pbo, we introduce a new type of pixmap GLAMOR_MEMORY_MAP. This type of pixmap has a valid devPrivate.ptr pointer, and that pointer points to a pbo mapped address. Now, we are ready to refine those glamor_prepare_access/glamor_finish_access pairs. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-04-09 14:16:07 +02:00
tex = _glamor_create_tex(glamor_priv, w, h, format);
no_tex:
gles2: Fixed color conversion for the formats except 1555 and 2101010. This patch fixed two major problems when we do the color convesion with GLES2. 1. lack of necessary formats in FBO pool. GLES2 has three different possible texture formats, GL_RGBA, GL_BGRA and GL_ALPHA. Previous implementation only has one bucket for all the three formats which may reuse a incorrect texture format when do the cache lookup. After this fix, we can enable fbo safely when running with GLES2. 2. Refine the format matching method in glamor_get_tex_format_type_from_pictformat. If both revertion and swap_rb are needed, for example use GL_RGBA to represent PICT_b8g8r8a8. Then the downloading and uploading should be handled differently. The picture's format is PICT_b8g8r8a8, Then the expecting color layout is as below (little endian): 0 1 2 3 : address a r g b Now the in GLES2 the supported color format is GL_RGBA, type is GL_UNSIGNED_TYPE, then we need to shuffle the fragment color as : frag_color = sample(texture).argb; before we use glReadPixel to get it back. For the uploading process, the shuffle is a revert shuffle. We still use GL_RGBA, GL_UNSIGNED_BYTE to upload the color to a texture, then let's see 0 1 2 3 : address a r g b : correct colors R G B A : GL_RGBA with GL_UNSIGNED_BYTE Now we need to shuffle again, the mapping rule is r = G, g = B, b = A, a = R. Then the uploading shuffle is as below: frag_color = sample(texture).gbar; After this commit, gles2 version can pass render check with all the formats except those 1555/2101010. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-04-05 15:53:44 +02:00
fbo = glamor_create_fbo_from_tex(glamor_priv, w, h, format, tex, flag);
glamor_fbo: Introduce glamor fbo to manage all the fb/tex. This is the first patch to implement a fbo/tex pool mechanism which is like the sna's BO cache list. We firstly need to decopule the fbo/tex from each pixmap. The new glamor_pixmap_fbo data structure is for that purpose. It's somehow independent to each pixmap and can be reused latter by other pixmaps once it's detached from the current pixmap. And this commit also slightly change the way to create a memory pixmap. We will not create a pixmap private data structure by default, instead we will crete that structure when a memory pixmap is attaching a fbo to it. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-18 12:21:36 +01:00
return fbo;
}
glamor_pixmap_fbo *
glamor_pixmap_detach_fbo(glamor_pixmap_private *pixmap_priv)
{
glamor_pixmap_fbo *fbo;
if (pixmap_priv == NULL)
return NULL;
fbo = pixmap_priv->fbo;
if (fbo == NULL)
return NULL;
pixmap_priv->fbo = NULL;
return fbo;
}
/* The pixmap must not be attached to another fbo. */
void
glamor_pixmap_attach_fbo(PixmapPtr pixmap, glamor_pixmap_fbo *fbo)
{
glamor_pixmap_private *pixmap_priv;
pixmap_priv = glamor_get_pixmap_private(pixmap);
if (pixmap_priv == NULL) {
glamor_screen_private *glamor_priv;
glamor_priv = glamor_get_screen_private(pixmap->drawable.pScreen);
pixmap_priv = calloc(1, sizeof(*pixmap_priv));
dixSetPrivate(&pixmap->devPrivates,
glamor_pixmap_private_key, pixmap_priv);
pixmap_priv->container = pixmap;
pixmap_priv->glamor_priv = glamor_priv;
pixmap_priv->type = GLAMOR_MEMORY;
}
if (pixmap_priv->fbo)
return;
pixmap_priv->fbo = fbo;
switch (pixmap_priv->type) {
case GLAMOR_TEXTURE_ONLY:
case GLAMOR_TEXTURE_DRM:
pixmap_priv->gl_fbo = 1;
if (fbo->tex != 0)
pixmap_priv->gl_tex = 1;
else {
/* XXX For the Xephyr only, may be broken now.*/
pixmap_priv->gl_tex = 0;
}
fallback_optimize: Prepare for downloading/uploading subregion. Introduced two function glamor_get_sub_pixmap/glamor_put_sub_pixmap, can easily used to get and put sub region of a big textured pixmap. And it can use pbo if possible. To support download a big textured pixmap's sub region to another pixmap's pbo, we introduce a new type of pixmap GLAMOR_MEMORY_MAP. This type of pixmap has a valid devPrivate.ptr pointer, and that pointer points to a pbo mapped address. Now, we are ready to refine those glamor_prepare_access/glamor_finish_access pairs. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-04-09 14:16:07 +02:00
case GLAMOR_MEMORY_MAP:
glamor_fbo: Introduce glamor fbo to manage all the fb/tex. This is the first patch to implement a fbo/tex pool mechanism which is like the sna's BO cache list. We firstly need to decopule the fbo/tex from each pixmap. The new glamor_pixmap_fbo data structure is for that purpose. It's somehow independent to each pixmap and can be reused latter by other pixmaps once it's detached from the current pixmap. And this commit also slightly change the way to create a memory pixmap. We will not create a pixmap private data structure by default, instead we will crete that structure when a memory pixmap is attaching a fbo to it. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-01-18 12:21:36 +01:00
pixmap->devPrivate.ptr = NULL;
break;
default:
break;
}
}
fallback_optimize: Prepare for downloading/uploading subregion. Introduced two function glamor_get_sub_pixmap/glamor_put_sub_pixmap, can easily used to get and put sub region of a big textured pixmap. And it can use pbo if possible. To support download a big textured pixmap's sub region to another pixmap's pbo, we introduce a new type of pixmap GLAMOR_MEMORY_MAP. This type of pixmap has a valid devPrivate.ptr pointer, and that pointer points to a pbo mapped address. Now, we are ready to refine those glamor_prepare_access/glamor_finish_access pairs. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-04-09 14:16:07 +02:00
Bool
glamor_pixmap_ensure_fbo(PixmapPtr pixmap, GLenum format, int flag)
{
glamor_screen_private *glamor_priv;
glamor_pixmap_private *pixmap_priv;
glamor_pixmap_fbo *fbo;
glamor_priv = glamor_get_screen_private(pixmap->drawable.pScreen);
pixmap_priv = glamor_get_pixmap_private(pixmap);
if (pixmap_priv == NULL || pixmap_priv->fbo == NULL) {
fbo = glamor_create_fbo(glamor_priv, pixmap->drawable.width,
pixmap->drawable.height,
format,
flag);
if (fbo == NULL)
return FALSE;
glamor_pixmap_attach_fbo(pixmap, fbo);
} else {
/* We do have a fbo, but it may lack of fb or tex. */
glamor_pixmap_ensure_fbo: Should allocate tex if we don't have one. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-04-13 07:45:42 +02:00
if (!pixmap_priv->fbo->tex)
fallback_optimize: Prepare for downloading/uploading subregion. Introduced two function glamor_get_sub_pixmap/glamor_put_sub_pixmap, can easily used to get and put sub region of a big textured pixmap. And it can use pbo if possible. To support download a big textured pixmap's sub region to another pixmap's pbo, we introduce a new type of pixmap GLAMOR_MEMORY_MAP. This type of pixmap has a valid devPrivate.ptr pointer, and that pointer points to a pbo mapped address. Now, we are ready to refine those glamor_prepare_access/glamor_finish_access pairs. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-04-09 14:16:07 +02:00
pixmap_priv->fbo->tex = _glamor_create_tex(glamor_priv, pixmap->drawable.width,
pixmap->drawable.height, format);
if (flag != GLAMOR_CREATE_FBO_NO_FBO && pixmap_priv->fbo->fb == 0)
glamor_pixmap_ensure_fb(pixmap_priv->fbo);
}
pixmap_priv = glamor_get_pixmap_private(pixmap);
return TRUE;
}
For DRI swap buffers. This commit added two APIs to support the DRI swap buffer. one is glamor_egl_exchange_buffers() which can swap two pixmaps' underlying KHRimages/fbos/texs. The DDX layer should exchange the DRM bos to make them consistent to each other. Another API is glamor_egl_create_textured_screen_ext(), which extent one more parameters to track the DDX layer's back pixmap pointer. This is for the triple buffer support. When using triple buffer, the DDX layer will keep a back pixmap rather then the front pixmap and the pixmap used by the DRI2 client. And during the closing screen stage, we have to dereference all the back pixmap's glamor resources. Thus we have to extent this API to register it when create new screen. Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
2012-02-17 09:56:05 +01:00
_X_EXPORT void
glamor_pixmap_exchange_fbos(PixmapPtr front, PixmapPtr back)
{
glamor_pixmap_private *front_priv, *back_priv;
glamor_pixmap_fbo *temp_fbo;
front_priv = glamor_get_pixmap_private(front);
back_priv = glamor_get_pixmap_private(back);
temp_fbo = front_priv->fbo;
front_priv->fbo = back_priv->fbo;
back_priv->fbo = temp_fbo;
}
Reference in New Issue Copy Permalink