OSDN Git Service
Brian Paul [Thu, 18 Feb 2010 20:02:59 +0000 (13:02 -0700)]
mklib: remove unused -contents_of_archives(), add comments
Brian Paul [Thu, 18 Feb 2010 19:54:28 +0000 (12:54 -0700)]
mesa: silence mklib output for libglapi.c as for other targets
Jon TURNEY [Thu, 18 Feb 2010 19:25:56 +0000 (19:25 +0000)]
Make mklib propogate all errors
Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
Signed-off-by: Brian Paul <brianp@vmware.com>
Brian Paul [Thu, 18 Feb 2010 19:46:48 +0000 (12:46 -0700)]
mklib: Teach mklib to fail build if link fails on cygwin
Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
Signed-off-by: Brian Paul <brianp@vmware.com>
(cherry picked from commit
551c96979e643b409535afe868c42cac0d2285ad)
Brian Paul [Thu, 18 Feb 2010 19:46:12 +0000 (12:46 -0700)]
configure: PTHREADS should be defined on cygwin, as well
Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
Signed-off-by: Brian Paul <brianp@vmware.com>
(cherry picked from commit
c1a4f249f18a766488da0832fbf6700de93af2f8)
Brian Paul [Thu, 18 Feb 2010 16:40:38 +0000 (09:40 -0700)]
swrast: fix left side clipping
Fixes bug 26623. Original patch was submitted by Mathias Frohlich
and modified by Brian.
(cherry picked from commit
7c34c237a2f6732b2c013543523617e375c1f534)
Francisco Jerez [Thu, 18 Feb 2010 12:37:09 +0000 (13:37 +0100)]
Revert "st/mesa: Make the frontbuffer visible on st_flush(PIPE_FLUSH_FRAME)."
We probably don't want to propagate this condition to the pipe driver,
this reverts commit
f455ca6490fcb65781b21f81c7117bd923e250d1 and the
dri_update_buffer flush altogether until an agreement is reached.
Kristian Høgsberg [Thu, 18 Feb 2010 12:49:14 +0000 (07:49 -0500)]
intel: Include main/hash.h using "" instead of <>
Vinson Lee [Thu, 18 Feb 2010 09:24:20 +0000 (01:24 -0800)]
progs/trivial: Update .gitignore.
Vinson Lee [Thu, 18 Feb 2010 09:17:26 +0000 (01:17 -0800)]
progs/redbook: Fix array out-of-bounds access.
Keith Whitwell [Thu, 18 Feb 2010 08:59:19 +0000 (08:59 +0000)]
st/mesa: always call FLUSH_CURRENT before st->pipe->flush()
Undoes part of commit
f455ca6 which would permit pipe->flush() to be
called while the VBO module still had its buffers mapped. Some
drivers care deeply about this. Sorry for not spotting this earlier.
Vinson Lee [Thu, 18 Feb 2010 07:22:51 +0000 (23:22 -0800)]
progs/rbug: Silence printf format warning.
Vinson Lee [Thu, 18 Feb 2010 07:19:24 +0000 (23:19 -0800)]
progs/rbug: Add missing headers.
Vinson Lee [Thu, 18 Feb 2010 07:10:13 +0000 (23:10 -0800)]
util: Fix typo of debug_dump_float_rgba_bmp.
Vinson Lee [Thu, 18 Feb 2010 06:55:16 +0000 (22:55 -0800)]
progs/rbug: Use util_format_name().
Vinson Lee [Thu, 18 Feb 2010 06:10:04 +0000 (22:10 -0800)]
progs/rbug: Add missing header.
Vinson Lee [Thu, 18 Feb 2010 05:50:07 +0000 (21:50 -0800)]
progs/rbug: s/read/rbug_read/
read conflicts with read in /usr/include/unistd.h.
Vinson Lee [Thu, 18 Feb 2010 05:38:17 +0000 (21:38 -0800)]
svga: Silence uninitialized variable warnings.
Marek Olšák [Wed, 17 Feb 2010 23:18:23 +0000 (00:18 +0100)]
r300g: add support for all missing non-FP sampler formats
The idea is to directly parse the format description in
r300_translate_texformat and return ~0 if the format is unsupported.
Chia-I Wu [Thu, 18 Feb 2010 04:38:47 +0000 (12:38 +0800)]
winsys/drm: Do not use --whole-archive in Makefile.egl.
--whole-archive is not portable. Have the drivers reference _eglMain as
a poor man's --whole-archive.
Kristian Høgsberg [Thu, 18 Feb 2010 02:17:55 +0000 (21:17 -0500)]
intel: Implement the DRI2 invalidate function properly
This uses a stamp mechanisms to mark the DRI drawable as invalid.
Instead of immediately updating the buffers we just bump the drawable
stamp and call out to DRI2GetBuffers "later".
"Later" used to be at LOCK_HARDWARE time, and this patch brings back
callouts at the points where we used to call LOCK_HARDWARE. A new function,
intel_prepare_render(), is called where we used to call LOCK_HARDWARE,
and if the buffers are invalid, we call out to DRI2GetBuffers there.
This lets us invalidate buffers only when notified instead of on
every glViewport() call. If the loader calls the DRI invalidate
entrypoint, we disable viewport triggered buffer invalidation.
Additionally, we can clean up the old viewport mechanism a bit,
since we can just invalidate the buffers and not worry about
reentrancy and whatnot.
Francisco Jerez [Wed, 17 Feb 2010 17:56:42 +0000 (18:56 +0100)]
dri/nouveau: Some multithreaded rendering fixes.
Francisco Jerez [Tue, 16 Feb 2010 17:17:11 +0000 (18:17 +0100)]
st/dri2: Flush FRAME instead of RENDER_CACHE on buffer invalidation.
Francisco Jerez [Tue, 16 Feb 2010 16:21:10 +0000 (17:21 +0100)]
st/mesa: Make the frontbuffer visible on st_flush(PIPE_FLUSH_FRAME).
So far the frontbuffer was only being flushed on st_glFlush and
st_glFinish, however, a co-state tracker may need to make sure that
any frontbuffer changes are already on its way to the actual front.
The dri2 state tracker will need this for event-driven GL applications
to resize properly (It could also be done calling "dri_flush_frontbuffer",
but that way we would flush unnecessarily in the double-buffered case).
Additionally this patch avoids flushing the mesa rendering cache if
PIPE_FLUSH_RENDER_CACHE wasn't specified.
Ian Romanick [Wed, 17 Feb 2010 19:40:09 +0000 (11:40 -0800)]
Merge commit '
381d5e209815235911c4aab516037c868c8f695f'
This merges the patches from the series "[PATCH 00/14] More
client-side GLX house cleaning" that were posted to the mesa3d-dev
mailing list. See
http://marc.info/?l=mesa3d-dev&m=
126582985214612&w=2
Patches 01 through 04 eliminate a bunch of annoying warnings that I
get when building Mesa.
Patch 05 fixes an inconsistency between the implementation of
glXSwapIntervalMESA and the spec. I chose to favor the code over the
spec in this case. This also eliminated a warning.
Patches 06 through 12 clean up the way that context creation is
performed on the client. When support for GLX_SGIX_fbconfig and the
related GLX 1.3 functions was added, I refactored a bunch
nuts-and-bolts of context creation to CreateContext. The refactor was
a good idea, I just didn't do it right.
Patches 13 and 14 update glxgears_fbconfig to use GLX 1.3 interfaces.
Brian Paul [Wed, 17 Feb 2010 16:02:33 +0000 (09:02 -0700)]
progs/demos: update GL version test to accept GL 3.x
Brian Paul [Wed, 17 Feb 2010 15:59:02 +0000 (08:59 -0700)]
glsl/apps: print usage info if insufficient args
Brian Paul [Wed, 17 Feb 2010 15:27:53 +0000 (08:27 -0700)]
st/mesa: replace conditional with assertion
Core Mesa will not call ctx->Driver.CheckQuery() if the Ready flag
is already set.
Chia-I Wu [Wed, 17 Feb 2010 14:30:29 +0000 (22:30 +0800)]
progs/openvg: Add .gitignore.
Add all OpenVG demos to .gitignore.
Chia-I Wu [Wed, 17 Feb 2010 14:29:43 +0000 (22:29 +0800)]
progs/es1: Update .gitignore.
Add xegl/bindtex to .gitignore.
Keith Whitwell [Wed, 17 Feb 2010 11:53:31 +0000 (11:53 +0000)]
llvmpipe: use ffs technique for full tiles also
Need to compute two masks here for full and partial 16x16 blocks.
Gives a further good improvement for isosurf particularly:
isosurf 97 -> 108
gears 597 -> 611
Chia-I Wu [Wed, 17 Feb 2010 11:52:29 +0000 (19:52 +0800)]
docs: Update TODOs of egl.html.
Thread-safety is no longer a TODO item. There are surely bugs to be
fixed though.
Chia-I Wu [Wed, 17 Feb 2010 11:03:30 +0000 (19:03 +0800)]
egl: Revisit global data locking.
Lock the global mutex in _eglPreloadDrivers and _eglAllocScreenHandle.
Add comments to why certain pathes do not need locking.
Chia-I Wu [Wed, 17 Feb 2010 09:30:44 +0000 (17:30 +0800)]
egl: Always lock a display before using it.
This gives a simple access control to the display. It is potentially
slow, but a finer grained mutex can always be used in the future. The
benefit of this simple approach is that drivers need not to worry about
thread-safety.
Chia-I Wu [Wed, 17 Feb 2010 10:39:27 +0000 (18:39 +0800)]
egl: Make eglGetDisplay atomic.
Merge _eglNewDisplay and _eglLinkDisplay into _eglFindDisplay. Remove
unused _eglUnlinkDisplay.
Chia-I Wu [Wed, 17 Feb 2010 09:29:52 +0000 (17:29 +0800)]
egl: Add a mutex to _EGLDisplay.
A display may be shared by multiple threads. Add a mutex for access
control.
Chia-I Wu [Wed, 17 Feb 2010 08:05:27 +0000 (16:05 +0800)]
egl: eglGetError should return the status of the last call.
Use macros to record the status of the function call before returning.
This is the only way that eglGetError can return the status of the most
recent function call.
Chia-I Wu [Wed, 17 Feb 2010 08:42:30 +0000 (16:42 +0800)]
egl: Always record error code.
The error code denotes the success or failure of the most recent
function call. A call to _eglError should always update the error code.
Chia-I Wu [Wed, 17 Feb 2010 07:43:47 +0000 (15:43 +0800)]
egl: Use new error checking macros.
Replace all uses of _EGL_DECLARE_* and _eglCheck* by _EGL_CHECK_*.
Chia-I Wu [Wed, 17 Feb 2010 07:22:03 +0000 (15:22 +0800)]
egl: Add new error checking macros.
Add _EGL_CHECK_* which will replace _EGL_DECLARE_* for error checking.
Move _eglCheck* earlier in the file so that the macros and the functions
are grouped together.
Keith Whitwell [Wed, 17 Feb 2010 11:07:45 +0000 (11:07 +0000)]
llvmpipe: rework do_block_16 to use bitmasks and ffs
Some nice speedups:
gears: 547 -> 597
isosurf: 83 -> 98
Others like gloss unchanged. Could do further work in this direction.
Michal Krol [Wed, 17 Feb 2010 10:18:03 +0000 (11:18 +0100)]
glsl/pp: Remove sl_pp_get_extension_state().
Parse extension tokens to track current state of extension enables.
Michal Krol [Wed, 17 Feb 2010 10:07:17 +0000 (11:07 +0100)]
glsl/cl: Parse layout qualifiers only when an appropriate extension enabled.
Michal Krol [Wed, 17 Feb 2010 08:54:59 +0000 (09:54 +0100)]
glsl/cl: Properly emit layout qualifier list.
Vinson Lee [Wed, 17 Feb 2010 10:10:49 +0000 (02:10 -0800)]
glsl: Silence unused value warning.
Vinson Lee [Wed, 17 Feb 2010 09:57:14 +0000 (01:57 -0800)]
progs/redbook: Fix memory leaks on error paths.
Johannes Obermayr [Wed, 17 Feb 2010 06:16:50 +0000 (16:16 +1000)]
nouveau: fix legacy dri driver build
Brian Paul [Tue, 16 Feb 2010 17:42:05 +0000 (10:42 -0700)]
sl/pp: re-do extension testing code
The #extension directive should not effect which extension preprocessor
symbols are defined/undefined; only whether/how the compiler accepts
language features defined by the extension.
Francisco Jerez [Tue, 16 Feb 2010 17:37:43 +0000 (18:37 +0100)]
nv40: Fix incorrect so_new() numbering.
Francisco Jerez [Tue, 16 Feb 2010 17:36:47 +0000 (18:36 +0100)]
nv40: Implement PIPE_CAP_MAX_COMBINED_SAMPLERS.
Kristian Høgsberg [Tue, 16 Feb 2010 15:37:48 +0000 (10:37 -0500)]
glx: Fix a couple of warnings
Kristian Høgsberg [Tue, 16 Feb 2010 15:27:57 +0000 (10:27 -0500)]
glx: Only register wire handlers for the events the server supports
Francisco Jerez [Mon, 8 Feb 2010 18:27:58 +0000 (19:27 +0100)]
st/dri2: Use event-driven buffer validation.
Francisco Jerez [Mon, 8 Feb 2010 18:27:57 +0000 (19:27 +0100)]
dri/nouveau: Use event driven buffer validation.
Francisco Jerez [Mon, 8 Feb 2010 18:27:56 +0000 (19:27 +0100)]
dri2: Event driven buffer validation.
When a buffer invalidation event is received from the X server, the
"invalidate" hook of the DRI2 flush extension is executed: A generic
implementation (dri2InvalidateDrawable) is provided that just bumps
the "pStamp" sequence number in __DRIdrawableRec.
For old servers not supporting buffer invalidation events, the
invalidate hook will be called before flushing the fake front/back
buffer (that's typically once per frame -- not a lot worse than the
situation we were in before).
No effort has been made on preserving backwards compatibility with
version 2 of the flush extension, but I think it's acceptable because
AFAIK no released stack is making use of it.
Signed-off-by: Kristian Høgsberg <krh@bitplanet.net>
Kristian Høgsberg [Mon, 15 Feb 2010 20:15:05 +0000 (15:15 -0500)]
dri2: Allocate cliprect as part of the __DRIdrawableRec
Kristian Høgsberg [Mon, 15 Feb 2010 16:12:28 +0000 (11:12 -0500)]
glx: Move GetGLXDRIDrawable() prototype to glxclient.h
Kristian Høgsberg [Mon, 15 Feb 2010 15:44:05 +0000 (10:44 -0500)]
configure.ac: Lower require dri2proto version to 2.1
Additions after 2.1 are implemented in a conditional way and will be
supported if available.
Brian Paul [Tue, 16 Feb 2010 15:21:38 +0000 (08:21 -0700)]
mesa: Lock mutex around _mesa_HashLookup linked list chase.
Remove const qualifier from _mesa_HashLookup() table parameter to
avoid LOCK/UNLOCK warnings in the function body.
Signed-off-by: Brian Paul <brianp@vmware.com>
(cherry picked from commit
3094adb3caeb90124359db2356df3bf8ee94800a)
Brian Paul [Tue, 16 Feb 2010 15:21:28 +0000 (08:21 -0700)]
mesa: Test for failed malloc in _mesa_HashInsert.
Signed-off-by: Brian Paul <brianp@vmware.com>
(cherry picked from commit
7c7247ddbf6e3f7f93e44c1cb52490044f1a2215)
José Fonseca [Tue, 16 Feb 2010 11:23:18 +0000 (11:23 +0000)]
Revert "progs/redbook: Temporarily disable the demos that extensions not part of windows to fix build."
This reverts commit
1587eff1b6d6ab74c302ca2ccb767d917ce3b1dd.
Michal Krol [Tue, 16 Feb 2010 09:04:58 +0000 (10:04 +0100)]
glsl/pp: Do not try to enable extension `all'.
Ben Skeggs [Tue, 16 Feb 2010 05:43:51 +0000 (15:43 +1000)]
nv50: use hw idx buffers where we can
Brian Paul [Tue, 16 Feb 2010 01:04:00 +0000 (18:04 -0700)]
progs/redbook: use glew
Ben Skeggs [Mon, 15 Feb 2010 06:45:39 +0000 (16:45 +1000)]
nouveau: fix for latest libdrm
José Fonseca [Mon, 15 Feb 2010 19:34:38 +0000 (19:34 +0000)]
wgl: Be lenient when sharing contexts.
My first reading of MS docs was wrong. It says:
All rendering contexts of a shared display list must use an identical
pixel format. Otherwise the results depend on the implementation of
OpenGL used.
That is, it is OK to share contexts with different pixel formats.
Adobe Premiere Pro tries to do that: share lists between a rgbx8 and a
rgba8 pixel format.
José Fonseca [Mon, 15 Feb 2010 21:47:20 +0000 (21:47 +0000)]
progs/redbook: Temporarily disable the demos that extensions not part of windows to fix build.
José Fonseca [Mon, 15 Feb 2010 20:48:24 +0000 (20:48 +0000)]
scons: A less magic way of not building progs by default.
The problem with the old way is that passing 'progs/subdir' as argument
would fail.
Michal Krol [Mon, 15 Feb 2010 20:33:22 +0000 (21:33 +0100)]
glsl/pp: Fix handling of if/elif/else cases.
Once if/elif evalutes to true, all subsequent conditions are always false.
Brian Paul [Mon, 15 Feb 2010 16:55:44 +0000 (09:55 -0700)]
progs/redbook: test image for new redbook tests
Brian Paul [Mon, 15 Feb 2010 16:55:01 +0000 (09:55 -0700)]
progs/redbook: add additional tests for GL 1.4
Brian Paul [Mon, 15 Feb 2010 15:57:34 +0000 (08:57 -0700)]
docs: document glMultiDrawElements() fix
(cherry picked from commit
ab6bcef99a53217db752659ace105ddfd9b25ebb)
Brian Paul [Mon, 15 Feb 2010 15:57:17 +0000 (08:57 -0700)]
vbo: fix broken glMultiDrawElements()
Seems to be a regression from commit
60b08eb1fdf287d28ec66b9282513ab35a61aee0.
(cherry picked from commit
c5a4cfb03ff17955c049f16ac805f5837ea4b633)
Vinson Lee [Mon, 15 Feb 2010 10:14:02 +0000 (02:14 -0800)]
glu/sgi: Add assert to check for null pointer dereference.
Vinson Lee [Mon, 15 Feb 2010 10:01:20 +0000 (02:01 -0800)]
mesa: Remove pointless comparison of unsigned integer with a negative constant.
Vinson Lee [Mon, 15 Feb 2010 08:30:31 +0000 (00:30 -0800)]
glsl/pp: Add asserts to check for null pointer deferences.
Vinson Lee [Mon, 15 Feb 2010 06:58:26 +0000 (22:58 -0800)]
draw: Add assert to check input of memcpy.
Vinson Lee [Mon, 15 Feb 2010 06:26:30 +0000 (22:26 -0800)]
glu/sgi: Move initialization of members to top of Curve constructor.
This is a modification of commit
53d448657bd300ab68b2869b2fba76a627699baf.
The members cpts and spts are possibly used as arguments later in
the constructor.
Vinson Lee [Mon, 15 Feb 2010 06:18:41 +0000 (22:18 -0800)]
glu/sgi: Initialize members of class Curvelist.
Vinson Lee [Mon, 15 Feb 2010 06:04:59 +0000 (22:04 -0800)]
glu/sgi: Initialize members of class Curve.
Vinson Lee [Mon, 15 Feb 2010 05:45:01 +0000 (21:45 -0800)]
glu/sgi: Initialize member of class Pool.
Vinson Lee [Mon, 15 Feb 2010 05:02:18 +0000 (21:02 -0800)]
glx: Silence uninitialized variable warning.
Brian Paul [Mon, 15 Feb 2010 04:54:17 +0000 (21:54 -0700)]
st/mesa: minor assorted clean-ups
Brian Paul [Mon, 15 Feb 2010 04:46:51 +0000 (21:46 -0700)]
st/mesa: change some function return types
Brian Paul [Mon, 15 Feb 2010 03:47:34 +0000 (20:47 -0700)]
st/mesa: move per-fragment shader fields to local vars
Brian Paul [Mon, 15 Feb 2010 01:50:49 +0000 (18:50 -0700)]
st/mesa: use _mesa_clone_fragment_program()
Brian Paul [Mon, 15 Feb 2010 01:46:56 +0000 (18:46 -0700)]
r600: use new program cloning functions
Brian Paul [Mon, 15 Feb 2010 01:46:51 +0000 (18:46 -0700)]
r300: use new program cloning functions
Brian Paul [Mon, 15 Feb 2010 01:46:37 +0000 (18:46 -0700)]
glsl: use new program cloning functions
Brian Paul [Mon, 15 Feb 2010 02:09:04 +0000 (19:09 -0700)]
mesa: added _mesa_clone_vertex/fragment_program()
To reduce casting elsewhere...
Vinson Lee [Mon, 15 Feb 2010 04:49:09 +0000 (20:49 -0800)]
r600: Silence uninitialized variable warnings.
Marek Olšák [Sun, 14 Feb 2010 23:04:32 +0000 (00:04 +0100)]
r300g: fix the size of constant buffers
4 more piglit tests pass, sweet.
José Fonseca [Mon, 15 Feb 2010 00:25:43 +0000 (00:25 +0000)]
python/retrace: Reconstruct blend state from binary blobs.
José Fonseca [Mon, 15 Feb 2010 00:22:59 +0000 (00:22 +0000)]
python: Reconstruct blend state objects from binary blobs.
José Fonseca [Mon, 15 Feb 2010 00:22:29 +0000 (00:22 +0000)]
trace: Dump blend state as a binary blob.
Gallium interfaces are too fluid -- XML or not is just too much work
trying to accommodate for differences in traces.
SWIG generated bindings are also full of quirks, making unmarshalling
the XML from too hard.
Finally, this is the first step for faster tracing.
José Fonseca [Mon, 15 Feb 2010 00:19:22 +0000 (00:19 +0000)]
trace: Fix constant buffer unwrapping.
Pointer unwrapping should happen before dump the pointer, so that
the real pipe driver pointers are shown, instead of the trace driver's.
José Fonseca [Sun, 14 Feb 2010 23:54:22 +0000 (23:54 +0000)]
util: Avoid dumping masked state. Other minor improvements.
José Fonseca [Sun, 14 Feb 2010 23:53:42 +0000 (23:53 +0000)]
os: A stream for debug logging.
Just a wrapper around os_log_message. Although it would probably make
more sense to be the other way around.
Also some comment fixes.
José Fonseca [Sun, 14 Feb 2010 23:25:01 +0000 (23:25 +0000)]
trace: No need to dump reference counts.
José Fonseca [Sun, 14 Feb 2010 23:24:08 +0000 (23:24 +0000)]
python: Pretty-printing of blend state.