José Fonseca [Wed, 24 Feb 2010 15:45:11 +0000 (15:45 +0000)]
llvmpipe: Fix breakeage.
Due to the u_format.csv changes the unrolled code generator was being used
for formats it didn't support.
Bring some bugfixes from u_format_access.py too.
José Fonseca [Wed, 24 Feb 2010 15:41:24 +0000 (15:41 +0000)]
util: Refactor some code.
José Fonseca [Wed, 24 Feb 2010 15:11:28 +0000 (15:11 +0000)]
util: Cope with the fact that formats in u_format.csv are not ordered.
José Fonseca [Wed, 24 Feb 2010 15:10:46 +0000 (15:10 +0000)]
util: Factor out the code to shorten a format name.
José Fonseca [Wed, 24 Feb 2010 14:51:47 +0000 (14:51 +0000)]
llvmpipe: Handle correctly 24bit formats.
José Fonseca [Wed, 24 Feb 2010 14:51:31 +0000 (14:51 +0000)]
util: Handle correctly 24bit formats.
José Fonseca [Wed, 24 Feb 2010 14:36:54 +0000 (14:36 +0000)]
util: Minor tweaks to ambigous/unused format descriptions.
José Fonseca [Wed, 24 Feb 2010 14:33:45 +0000 (14:33 +0000)]
util: Fix swizzles for SRGB formats.
Based on Marek Olšák's fix 8-bits-per-channel formats swizzles, with the
exception that the notation for several formats (the vector formats)
is actually reversed so they should be left alone.
José Fonseca [Wed, 24 Feb 2010 14:15:09 +0000 (14:15 +0000)]
util: Organize formats in groups. Add comments.
José Fonseca [Wed, 24 Feb 2010 13:47:42 +0000 (13:47 +0000)]
util: Allow '#' comments in u_format.csv.
Brian Paul [Wed, 24 Feb 2010 15:13:55 +0000 (08:13 -0700)]
mesa: disable unreachable meta mipmap gen code
More work is needed to support 3D mipmap generation. Disable unreachable
code until then.
See bug 26722.
José Fonseca [Tue, 23 Feb 2010 20:20:19 +0000 (20:20 +0000)]
util: Kill unused UTIL_FORMAT_LAYOUT_SCALAR.
José Fonseca [Tue, 23 Feb 2010 20:19:50 +0000 (20:19 +0000)]
llvmpipe: Kill unused UTIL_FORMAT_LAYOUT_SCALAR.
José Fonseca [Tue, 23 Feb 2010 19:51:42 +0000 (19:51 +0000)]
util: Store more derived data in the the format description.
Dave Airlie [Wed, 24 Feb 2010 09:13:07 +0000 (19:13 +1000)]
radeon/r200: fix the state emission before kernel clear
this moves the emission outside the lock and adds r200 support.
Signed-off-by: Dave Airlie <airlied@redhat.com>
Vinson Lee [Wed, 24 Feb 2010 08:24:51 +0000 (00:24 -0800)]
r600: Assert that array index is not negative.
Dave Airlie [Wed, 24 Feb 2010 07:19:09 +0000 (17:19 +1000)]
radeon: fix bad state emission causes kernel to do bad depth clear
The kernel lets you clear depth without getting a depth offset
from userspace, mesa used to emit state before clear, but that got
lost in the refactoring, which made the kernel bug show up. Fix
mesa driver to emit the state properly now.
Signed-off-by: Dave Airlie <airlied@redhat.com>
Kristian Høgsberg [Wed, 24 Feb 2010 04:15:20 +0000 (23:15 -0500)]
intel: Call intel_prepare_render() in intelMakeCurrent()
This restores old behaviour, where we end up doing a DRI2GetBuffers()
call from intelMakeCurrent(). The idea was that we could do this
lazily, just before we start rendering. However, if we don't do the
DRI2GetBuffers() round-trip we don't get the drawable size and higher
level mesa ends up short-cutting a number of GL calls, such as glClear().
Brian Paul [Wed, 24 Feb 2010 04:16:18 +0000 (21:16 -0700)]
gallivm: added clamp and int_to_float functions
Brian Paul [Mon, 22 Feb 2010 22:19:45 +0000 (15:19 -0700)]
llvmpipe: add lp_tile_image.c to sources
Brian Paul [Mon, 22 Feb 2010 22:18:49 +0000 (15:18 -0700)]
llvmpipe: whole image tile/untile functions
Brian Paul [Wed, 24 Feb 2010 00:13:13 +0000 (17:13 -0700)]
st/mesa: whitespace fixes
Brian Paul [Wed, 24 Feb 2010 01:56:58 +0000 (18:56 -0700)]
llvmpipe: remove Z16 format override hack
Brian Paul [Wed, 24 Feb 2010 01:56:24 +0000 (18:56 -0700)]
llvmpipe: Z16 format is not supported
Brian Paul [Wed, 24 Feb 2010 01:55:28 +0000 (18:55 -0700)]
st/glx: improved depth/stencil format selection code
Actually ask the gallium screen what Z/stencil format is supported.
This will let us remove some hacks in the llvmpipe driver.
Brian Paul [Wed, 24 Feb 2010 01:23:15 +0000 (18:23 -0700)]
st/glx: added PIPE_FORMAT_Z24S8_UNORM code
Brian Paul [Thu, 11 Feb 2010 01:10:16 +0000 (18:10 -0700)]
llvmpipe: added assertions to verify that scene is empty
Brian Paul [Thu, 11 Feb 2010 01:07:38 +0000 (18:07 -0700)]
llvmpipe: added some assertions
George Sapountzis [Mon, 22 Feb 2010 20:27:49 +0000 (22:27 +0200)]
glapi: fix es after mv
George Sapountzis [Mon, 22 Feb 2010 20:27:48 +0000 (22:27 +0200)]
glapi: fix code generation Makefile after mv
George Sapountzis [Mon, 22 Feb 2010 20:27:48 +0000 (22:27 +0200)]
glapi: mv code generation to subdir
This just moves files, code generation Makefile will be fixed in next commit for easier review.
Marek Olšák [Tue, 23 Feb 2010 08:08:28 +0000 (09:08 +0100)]
r300g: fix texture swizzling for the SRGB formats
kleinerm [Fri, 19 Feb 2010 20:28:26 +0000 (12:28 -0800)]
glx: Fix some typos in pointer checking for OML_sync_control entry points.
Signed-off-by: Mario Kleiner <mario.kleiner@tuebingen.mpg.de>
Signed-off-by: Brian Paul <brianp@vmware.com>
Luca Barbieri [Tue, 23 Feb 2010 16:55:24 +0000 (08:55 -0800)]
pipebuffer: slab: if size < alignment, use alignment to find bucket
If the size is lower than the alignment, we must use the alignment to
select the bucket.
Otherwise, the selected bucket won't be able to satisfy our request
and will fail.
Signed-off-by: José Fonseca <jfonseca@vmware.com>
Luca Barbieri [Tue, 23 Feb 2010 16:55:38 +0000 (08:55 -0800)]
pipebuffer: fix inverted signalled checking
A return of 0 means the fence is signalled.
Signed-off-by: José Fonseca <jfonseca@vmware.com>
Luca Barbieri [Tue, 23 Feb 2010 16:50:15 +0000 (08:50 -0800)]
pipebuffer: avoid assert due to increasing a zeroed refcnt
The cache manager stores buffers with a reference count that dropped to 0.
pipe_reference asserts in this case on debug builds,
so use pipe_reference_init instead.
Signed-off-by: José Fonseca <jfonseca@vmware.com>
Eric Anholt [Tue, 23 Feb 2010 18:48:19 +0000 (13:48 -0500)]
i965: Enable GL_ARB_fragment_coord_conventions now that the GLSL is fixed.
Tested with piglit glsl-arb-fragment-coord-conventions.
Chia-I Wu [Mon, 22 Feb 2010 13:59:14 +0000 (21:59 +0800)]
egl: Remove unused _EGL_SKIP_HANDLE_CHECK.
It was added to skip checking EGLDisplay, EGLSurface, and etc. It is
never defined and the spec does not allow the checks to be skipped.
Remove it for good.
Chia-I Wu [Mon, 22 Feb 2010 13:56:58 +0000 (21:56 +0800)]
st/egl: Unify surface creation.
Add a new function egl_g3d_create_surface and use it to create window,
pixmap, buffer, and screen surfaces.
Vinson Lee [Tue, 23 Feb 2010 08:19:30 +0000 (00:19 -0800)]
mesa: Assert that array index is not negative.
Vinson Lee [Tue, 23 Feb 2010 07:49:50 +0000 (23:49 -0800)]
glsl/apps: Assert that ftell does not return an error.
Vinson Lee [Tue, 23 Feb 2010 07:32:43 +0000 (23:32 -0800)]
i965g: Add fallthrough comments in switch statement.
Silences warnings about missing break statements in static analysis.
Vinson Lee [Tue, 23 Feb 2010 07:23:57 +0000 (23:23 -0800)]
draw: Remove dead code.
Marek Olšák [Mon, 22 Feb 2010 12:51:38 +0000 (13:51 +0100)]
r300g: fix draw_elements for "start" != 0
Reported-by: Andre Maasikas <amaasikas@gmail.com>
José Fonseca [Mon, 22 Feb 2010 19:24:18 +0000 (19:24 +0000)]
svga: Upload user buffers only once.
José Fonseca [Mon, 22 Feb 2010 19:20:25 +0000 (19:20 +0000)]
svga: Fix the guest offset of piecewise buffer DMAs.
It was being erroneously set equal to the host offset, but it should be
zero.
José Fonseca [Mon, 22 Feb 2010 18:57:09 +0000 (18:57 +0000)]
svga: Emit a scalar rcp in shadow maps.
Small improvement in Lightsmark 2008.
José Fonseca [Mon, 22 Feb 2010 18:01:32 +0000 (18:01 +0000)]
svga: Don't emit zero writemasks.
This fixes a regression with Lightsmark, where more compact TGSI from Mesa
was causing a zero mask MOV to be emitted for shadow map compare, causing
problems in some backends.
Add a few more assertions to catch cases like this.
José Fonseca [Mon, 22 Feb 2010 21:47:07 +0000 (21:47 +0000)]
tgsi: Report an error when a destination has an empty writemask.
José Fonseca [Thu, 18 Feb 2010 17:21:53 +0000 (17:21 +0000)]
svga: Upload the dirty ranges instead of the whole buffer.
José Fonseca [Thu, 18 Feb 2010 16:40:14 +0000 (16:40 +0000)]
svga: Break uploads of big buffers into smaller pieces.
Fixes crash in Homeworld2 which tries to create a 14MB buffer, because we
now avoid creating GMR buffers larger than 8MB to ensure progress given
we have only a 16MB pool.
José Fonseca [Thu, 18 Feb 2010 14:12:31 +0000 (14:12 +0000)]
svga: Note ranges even when there is no hardware buffer around.
Not having a hardware buffer around doesn't change the fact that the
range is dirty and needs to be uploaded eventually.
José Fonseca [Thu, 18 Feb 2010 14:05:57 +0000 (14:05 +0000)]
svga: Further cleanup/comment svga buffer code.
Just cosmetic changes -- no behavior change.
José Fonseca [Thu, 18 Feb 2010 18:11:50 +0000 (18:11 +0000)]
svga: Temporarily create a sw vertex buf when failed to create a hw buf.
Many apps don't check the return of map buffer so it is better not to
fail.
José Fonseca [Tue, 16 Feb 2010 16:43:53 +0000 (16:43 +0000)]
svga: Remove unused buffer download code.
Unnecessary now that we never destroy buffer storage.
José Fonseca [Tue, 16 Feb 2010 15:51:34 +0000 (15:51 +0000)]
pipebuffer: Don't synchronize when checking for buffer overflows.
To avoid masking synchronization issues in debug builds.
José Fonseca [Thu, 28 Jan 2010 17:11:02 +0000 (17:11 +0000)]
python/retrace: Dump all vertices for verbosity level 3.
José Fonseca [Wed, 27 Jan 2010 15:41:25 +0000 (15:41 +0000)]
svga: Prevent buffer overflow in buffer ranges.
Do this by extending the nearest range to cover the new range.
This fixes an access fault in Call of Duty which was doing many disjoint
glBufferSubData calls.
José Fonseca [Wed, 27 Jan 2010 14:45:56 +0000 (14:45 +0000)]
svga: Fix PIPE_LOGICOP_INVERT.
Actually the current XOR implementation is an INVERT.
This fixes rectangle selection in Maya.
Keith Packard [Sun, 21 Feb 2010 22:30:00 +0000 (14:30 -0800)]
bin/mklib: Clear CDPATH to avoid damaging expand_archive output
The bash 'cd' command tends to emit random stuff to stdout when the
CDPATH variable is set, so clear it to keep extra filenames from being
emitted from the expand_archive function, which would otherwise cause
mklib to fail.
Signed-off-by: Keith Packard <keithp@keithp.com>
Reviewed-by: Dan Nicholson <dbn.lists@gmail.com>
Signed-off-by: Brian Paul <brianp@vmware.com>
Brian Paul [Mon, 22 Feb 2010 15:56:18 +0000 (08:56 -0700)]
glxinfo: Fix User error in glGetProgramivARB
glxinfo needed fixing after commit
4bccd69.
Move fragment program only parameters into their own list
so that they are not queried for a vertex program.
Signed-off-by: Xavier Chantry <chantry.xavier@gmail.com>
Signed-off-by: Brian Paul <brianp@vmware.com>
(cherry picked from commit
b0e5dcb859692707fec8752dd21fc61fc4c3dbcc)
Brian Paul [Mon, 22 Feb 2010 15:14:14 +0000 (08:14 -0700)]
i965g: added missing texture cap queries
Brian Paul [Mon, 22 Feb 2010 15:14:05 +0000 (08:14 -0700)]
i915g: added missing texture cap queries
Brian Paul [Mon, 22 Feb 2010 15:04:53 +0000 (08:04 -0700)]
st/mesa: change viewport Z scale/bias for glBitmap/glDrawPixels
This fixes incorrect Z position of glBitmap, glDraw/CopyPixels for the
svga driver. Now we use 0.5, 0.5 as is typical for ordinary 3D rendering.
(cherry picked from commit
bcd561c66777e58dbb29a573c4d2279772bac6c5)
Brian Paul [Mon, 22 Feb 2010 15:04:39 +0000 (08:04 -0700)]
st/mesa: Add checks for ST_SURFACE_x vs MESA_BUFFER_x
The ST_SURFACE_x values should match the Mesa BUFFER_x values.
Added some assertions to prevent future mix-ups.
(cherry picked from commit
13cbb5fff68ef8831230638e9f0b29a217750e9d)
Brian Paul [Mon, 22 Feb 2010 15:04:16 +0000 (08:04 -0700)]
mesa/st: Make ST_SURFACE_DEPTH index consistent with mesa's BUFFER_DEPTH.
Some st functions assume that they are identical.
(cherry picked from commit
9d17ad2891b58de9e33e943ff918a678c6a3c2bd)
Chris Wilson [Mon, 22 Feb 2010 09:53:36 +0000 (09:53 +0000)]
intel: assert that we do not overflow the batch buffer.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Chris Wilson [Mon, 22 Feb 2010 09:51:33 +0000 (09:51 +0000)]
i915: Fix logic !gen >= 3
The effect of this was that all objects were aligned to 128 bytes
on all generations, rather than just gen2.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Chris Wilson [Mon, 22 Feb 2010 09:50:15 +0000 (09:50 +0000)]
i915: Remove superfluous MI_NOOP from vertex emission
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Chris Wilson [Thu, 21 Jan 2010 17:39:01 +0000 (17:39 +0000)]
intel: Check that we have a bufmgr or bail out when initializing the context.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Dave Airlie [Mon, 22 Feb 2010 07:26:30 +0000 (17:26 +1000)]
Revert "r300g: rebuild winsys/pipe buffer handling and add buffer map"
This reverts commit
fff5be8e7b4557c221f2425dcafc2e7cbbba76ba.
Probably went too soon with this, dileX reported OA not working for him
it works here fine, but the optimisations I wanted aren't working properly
yet so I'll fix that now.
Signed-off-by: Dave Airlie <airlied@redhat.com>
Dave Airlie [Fri, 12 Feb 2010 05:39:51 +0000 (15:39 +1000)]
r300g: rebuild winsys/pipe buffer handling and add buffer map
This creates a cleaner winsys and drop the simple screen stuff.
It makes r300g use pb_bufmgr structs and adds usage of the cached
bufmgr for vertex/index buffers.
It also avoids mapping too often.
I'm not 100% sure this is perfect but it won't find its own bugs.
Signed-off-by: Dave Airlie <airlied@redhat.com>
Marek Olšák [Sun, 21 Feb 2010 19:49:29 +0000 (20:49 +0100)]
r3OOg: support rendering of more than 65535 vertices per draw call (almost)
The path for VAP_ALT_NUM_VERTICES is also in place (and tested)
but not enabled by default due to the missing support of this reg
in the upstream kernel.
Also, a non-zero BUFFER_BASE in the INDX_BUFFER packet3 hangs the machine.
Am I missing something? Because of this, only draw_arrays can render
more than 65535 vertices without the use of VAP_ALT_NUM_VERTICES.
Vinson Lee [Sun, 21 Feb 2010 23:15:52 +0000 (15:15 -0800)]
i965g: Fix memory leak.
Vinson Lee [Sun, 21 Feb 2010 23:10:50 +0000 (15:10 -0800)]
st/mesa: Fix memory leak on out-of-memory error path.
Vinson Lee [Sun, 21 Feb 2010 22:26:53 +0000 (14:26 -0800)]
r300g: Remove unnecessary header.
Ray Kohler [Sun, 21 Feb 2010 03:26:43 +0000 (22:26 -0500)]
Link egl progs to libm as necessary.
Signed-off-by: Ray Kohler <ataraxia937@gmail.com>
Marek Olšák [Sat, 20 Feb 2010 20:12:45 +0000 (21:12 +0100)]
r300g: precompute framebuffer register values
Marek Olšák [Sat, 20 Feb 2010 23:17:04 +0000 (00:17 +0100)]
r300g: add all missing colorbuffer formats
Marek Olšák [Sat, 20 Feb 2010 20:33:53 +0000 (21:33 +0100)]
r300g: make r300_translate_texformat private
Unlikely to increase performance from inlining.
And partially expose it through r300_is_sampler_format_supported.
Marek Olšák [Sat, 20 Feb 2010 17:52:20 +0000 (18:52 +0100)]
r300g: re-enable SRGB formats
Ouch.
Vinson Lee [Sun, 21 Feb 2010 10:13:15 +0000 (02:13 -0800)]
mesa: Add assert to check for null pointer dereference.
Vinson Lee [Sun, 21 Feb 2010 09:57:16 +0000 (01:57 -0800)]
draw: Remove dead code.
Vinson Lee [Sun, 21 Feb 2010 09:30:35 +0000 (01:30 -0800)]
i965g: Remove dead code.
Vinson Lee [Sun, 21 Feb 2010 09:19:47 +0000 (01:19 -0800)]
i965g: Fix assertions that were always true.
The surface_format field is 9 bits wide which has a maximum value
of 511. BRW_SURFACEFORMAT_INVALID has a value of 4095 which exceeds
the width of the surface_format field.
This patch asserts that the format is not BRW_SURFACEFORMAT_INVALID
before assignment to the surface_format field.
Vinson Lee [Sun, 21 Feb 2010 08:36:50 +0000 (00:36 -0800)]
glsl/cl: Prevent possible buffer overflow.
Vinson Lee [Sun, 21 Feb 2010 08:24:05 +0000 (00:24 -0800)]
i965g: Fix use of out-of-scope variable.
Chia-I Wu [Sun, 21 Feb 2010 07:47:45 +0000 (15:47 +0800)]
st/egl: Disable SHM in ximage backend.
With pipe_screen::update_buffer being called in st_validate_state, it is
likely that softpipe will flush tile caches to a detached SHM segment.
Disable SHM for now until a better solution is found.
Vinson Lee [Sun, 21 Feb 2010 07:43:33 +0000 (23:43 -0800)]
radeon: Silence compiler format warning.
Vinson Lee [Sun, 21 Feb 2010 07:27:07 +0000 (23:27 -0800)]
glsl/cl: Prevent possible string overflow.
Chia-I Wu [Sun, 21 Feb 2010 02:11:54 +0000 (10:11 +0800)]
st/egl: Properly convert vrefresh to refresh rate.
Some kernels have vrefresh = (refresh rate * 1000) while others don't.
Use some heuristics here.
Chia-I Wu [Sun, 21 Feb 2010 02:02:13 +0000 (10:02 +0800)]
st/egl: Enable extensions before adding configs.
Configs are validated before added. The validation depends on the
extensions available. If configs are added before enabling extensions,
configs with EGL_SCREEN_SURFACE_MESA bit set will never pass the
validation for example.
Chia-I Wu [Sat, 20 Feb 2010 04:41:30 +0000 (12:41 +0800)]
st/egl: Do not force validate when update_buffer is called.
st/mesa was changed to call pipe_screen::update_buffer in
st_validate_state. In most cases there is no need to validate.
Chia-I Wu [Fri, 19 Feb 2010 14:12:04 +0000 (22:12 +0800)]
st/egl: Clean up DRI2 surface.
Mainly to move the call to DRI2GetBuffers from dri2_surface_validate to
the new dri2_surface_update_buffers.
Chia-I Wu [Fri, 19 Feb 2010 14:36:16 +0000 (22:36 +0800)]
st/egl: Clean up ximage surface.
Mainly to move buffer allocation code from ximage_surface_validate to
the new ximage_surface_update_buffers.
Vinson Lee [Sun, 21 Feb 2010 00:23:44 +0000 (16:23 -0800)]
st/xlib: Add assert to check for null pointer dereference.
Jon TURNEY [Fri, 19 Feb 2010 22:38:00 +0000 (22:38 +0000)]
Cygwin build fix: Fix linkage
Fix the way we make static convenience libraries, such as libmesa.a,
to be the same as linux etc.
Putting archives inside archives doesn't make the objects inside
the archive linkable, so use expand_archives() to get all the objects
inside an archive out again before linking.
Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
Signed-off-by: Brian Paul <brianp@vmware.com>
Vinson Lee [Sat, 20 Feb 2010 10:29:43 +0000 (02:29 -0800)]
gamma: Silence uninitialized variable warnings.
Vinson Lee [Sat, 20 Feb 2010 10:25:29 +0000 (02:25 -0800)]
gamma: Add missing headers.
Vinson Lee [Sat, 20 Feb 2010 10:02:17 +0000 (02:02 -0800)]
gamma: Remove unnecessary header.