mesa.git
15 years agoUse variable library name in pkg-config output.
Tom Fogal [Wed, 29 Apr 2009 16:32:46 +0000 (10:32 -0600)]
Use variable library name in pkg-config output.

Previously the pkg-config output files would contain e.g. `-lGL'
and `-lGLU', even if the user modified their configuration to
build libraries with different names.  This modifies the
pkg-config inputs, and corresponding makery, so that modifying the
output library name will cause the appropriate updated name to
appear in the pkg-config `-l' option.

Signed-off-by: Dan Nicholson <dbn.lists@gmail.com>
15 years agoautoconf: Add switch for optional EGL
Dan Nicholson [Wed, 29 Apr 2009 19:11:43 +0000 (12:11 -0700)]
autoconf: Add switch for optional EGL

EGL doesn't build on all platforms, so allow people to opt out.

15 years agoautoconf: Clean up some m4 usage
Dan Nicholson [Wed, 29 Apr 2009 13:49:27 +0000 (06:49 -0700)]
autoconf: Clean up some m4 usage

m4_fatal is equivalent to m4_errprint + m4_exit.

15 years agotgsi: added tgsi_check_soa_dependencies() and related debug code (disabled)
Brian Paul [Wed, 29 Apr 2009 17:56:57 +0000 (11:56 -0600)]
tgsi: added tgsi_check_soa_dependencies() and related debug code (disabled)

The TGSI interpeter operates in SOA style.  We need to check for data
dependencies in instructions which read from and write to the same register.
For now just adding some debug code to detect that condition.  Actual fixes
to follow.

15 years agomesa: added _mesa_check_soa_dependencies() function
Brian Paul [Wed, 29 Apr 2009 17:52:06 +0000 (11:52 -0600)]
mesa: added _mesa_check_soa_dependencies() function

This function will check an instruction to see if there's data dependencies
between the dst and src registers if executed in an SOA manner.

15 years agosoftpipe: return PIPE_UNREFERENCED in softpipe_is_buffer/texture_referenced()
Brian Paul [Tue, 28 Apr 2009 20:29:27 +0000 (14:29 -0600)]
softpipe: return PIPE_UNREFERENCED in softpipe_is_buffer/texture_referenced()

This allows the engine demo to run again (avoid crash in VBO code).
This stuff still needs to be revisited someday though...

15 years agomesa/st: protect internal flushes with FLUSH_CURRENT
Keith Whitwell [Tue, 28 Apr 2009 16:50:19 +0000 (17:50 +0100)]
mesa/st: protect internal flushes with FLUSH_CURRENT

Already doing this for driver.flush()

15 years agomesa/main: protect driver.finish with FLUSH_CURRENT
Keith Whitwell [Tue, 28 Apr 2009 16:49:50 +0000 (17:49 +0100)]
mesa/main: protect driver.finish with FLUSH_CURRENT

Already doing this for driver.flush()

15 years agomesa/st: remove duplicate offset calculation
Keith Whitwell [Tue, 28 Apr 2009 13:51:11 +0000 (14:51 +0100)]
mesa/st: remove duplicate offset calculation

15 years agogallium/draw: add ability to print out active pipeline stages
Keith Whitwell [Tue, 28 Apr 2009 13:50:05 +0000 (14:50 +0100)]
gallium/draw: add ability to print out active pipeline stages

15 years agomesa/st: translate VERT_ATTRIB_GENERIC8..15 in st_translate_vertex_program
Keith Whitwell [Mon, 27 Apr 2009 14:33:44 +0000 (15:33 +0100)]
mesa/st: translate VERT_ATTRIB_GENERIC8..15 in st_translate_vertex_program

It seems quake4 can hit these attributes sometimes.

15 years agoprogs: add fflushes for cygwin
Keith Whitwell [Tue, 28 Apr 2009 10:49:55 +0000 (11:49 +0100)]
progs: add fflushes for cygwin

15 years agoutil/indices: remove debug prints
Keith Whitwell [Mon, 27 Apr 2009 18:36:42 +0000 (19:36 +0100)]
util/indices: remove debug prints

15 years agomesa/st: workaround for crashes in st_copy_texsubimage
Keith Whitwell [Mon, 27 Apr 2009 17:56:26 +0000 (18:56 +0100)]
mesa/st: workaround for crashes in st_copy_texsubimage

Proper fix for this hasn't been identified, but avoid crashing.

15 years agowgl: Store current HDC/HGLRC in stw_context.
José Fonseca [Mon, 27 Apr 2009 19:24:55 +0000 (20:24 +0100)]
wgl: Store current HDC/HGLRC in stw_context.

Less TLS lookups.

15 years agoR300: add quadpipe overrides
Alex Deucher [Tue, 28 Apr 2009 16:37:29 +0000 (12:37 -0400)]
R300: add quadpipe overrides

RV410 SE chips only have 1 quadpipe.
Also, handle other R300 chip with quadpipe override

15 years agoi965: avoid segfault in intel_update_renderbuffers() if using DRI1
Brian Paul [Tue, 28 Apr 2009 15:58:44 +0000 (09:58 -0600)]
i965: avoid segfault in intel_update_renderbuffers() if using DRI1

15 years agoswrast: add missing break in clamp_rect_coord_linear()
Brian Paul [Tue, 28 Apr 2009 15:05:19 +0000 (09:05 -0600)]
swrast: add missing break in clamp_rect_coord_linear()

See bug 21461.

15 years agodemos: asst. updates, clean-ups
Brian Paul [Mon, 27 Apr 2009 23:01:59 +0000 (17:01 -0600)]
demos: asst. updates, clean-ups

15 years agogallium dri st: Propagate the drawable info when we bind to new drawables.
Thomas Hellstrom [Tue, 28 Apr 2009 12:43:41 +0000 (14:43 +0200)]
gallium dri st: Propagate the drawable info when we bind to new drawables.

Signed-off-by: Thomas Hellstrom <thellstrom-at-vmware-dot-com>
15 years agogallium dri st: Use st_get_current() instead of GET_CURRENT_CONTEXT()
Thomas Hellstrom [Tue, 28 Apr 2009 12:20:45 +0000 (14:20 +0200)]
gallium dri st: Use st_get_current() instead of GET_CURRENT_CONTEXT()

Signed-off-by: Thomas Hellstrom <thellstrom-at-vmware-dot-com>
15 years agost: Add an st_get_current() function.
Thomas Hellstrom [Tue, 28 Apr 2009 12:12:39 +0000 (14:12 +0200)]
st: Add an st_get_current() function.

Signed-off-by: Thomas Hellstrom <thellstrom-at-vmware-dot-com>
15 years agogallium-intel: Create a i965_dri.so symlink
Jakob Bornecrantz [Tue, 28 Apr 2009 11:56:21 +0000 (12:56 +0100)]
gallium-intel: Create a i965_dri.so symlink

This is only used for debuging the gem backend on i965
chipset using the softpipe pipe driver.

Usage: "export INTEL_SOFTPIPE=y" and point LIBGL_DRIVERS_PATH
to "$MESA/lib/gallium" where $MESA is the mesa root.

15 years agogallium-intel: Fix build of dri driver
Jakob Bornecrantz [Mon, 27 Apr 2009 23:00:51 +0000 (01:00 +0200)]
gallium-intel: Fix build of dri driver

15 years agogallium dri st: Fix up some comments and minor bugs.
Thomas Hellstrom [Tue, 28 Apr 2009 11:03:22 +0000 (13:03 +0200)]
gallium dri st: Fix up some comments and minor bugs.

Signed-off-by: Thomas Hellstrom <thellstrom-at-vmware-dot-com>
15 years agoradeon-r300: Fix a bit of breakage.
Corbin Simpson [Tue, 28 Apr 2009 10:28:57 +0000 (03:28 -0700)]
radeon-r300: Fix a bit of breakage.

Not really sure why reordering the ioctls makes them work again.

15 years agoradeon: Use PCI_MATCH_ANY for xorg driver.
Corbin Simpson [Tue, 28 Apr 2009 10:28:37 +0000 (03:28 -0700)]
radeon: Use PCI_MATCH_ANY for xorg driver.

Might as well.

15 years agogallium: Rename the dri state tracker lib to libdridrm.a
Thomas Hellstrom [Tue, 28 Apr 2009 09:58:27 +0000 (11:58 +0200)]
gallium: Rename the dri state tracker lib to libdridrm.a

15 years agogallium: indent and cleanfile the dri state-tracker.
Thomas Hellstrom [Tue, 28 Apr 2009 09:54:25 +0000 (11:54 +0200)]
gallium: indent and cleanfile the dri state-tracker.

15 years agogallium: Update the dri2 state tracker to support dri1.
Thomas Hellstrom [Tue, 28 Apr 2009 09:49:39 +0000 (11:49 +0200)]
gallium: Update the dri2 state tracker to support dri1.

Signed-off-by: Thomas Hellstrom <thellstrom-at-vmware-dot-com>
15 years agogallium: Add a dri1 api
Thomas Hellstrom [Tue, 28 Apr 2009 09:43:06 +0000 (11:43 +0200)]
gallium: Add a dri1 api

that a driver needs to implement on top of the drm api to support dri1.

Signed-off-by: Thomas Hellstrom <thellstrom-at-vmware-dot-com>
15 years agogallium: Move the dri2 state tracker since we're about to extend it to dri1.
Thomas Hellstrom [Tue, 28 Apr 2009 09:32:59 +0000 (11:32 +0200)]
gallium: Move the dri2 state tracker since we're about to extend it to dri1.

Signed-off-by: Thomas Hellstrom <thellstrom-at-vmware-dot-com>
15 years agogallium: Update the drm_api.
Thomas Hellstrom [Tue, 28 Apr 2009 09:23:11 +0000 (11:23 +0200)]
gallium: Update the drm_api.

Make it possible to pass state-tracker-specific data to the
init_screen function, and even open the door for device-specific
state-tracker screen initialization.

Signed-off-by: Thomas Hellstrom <thellstrom-at-vmware-dot-com>
15 years agoAvoid a segfault in shader compilation
Robert Ellison [Mon, 27 Apr 2009 18:08:34 +0000 (12:08 -0600)]
Avoid a segfault in shader compilation

If a shader reaches an out-of-memory condition while adding
a new function (reallocating the function list), a segfault
will occur during cleanup (because the num_functions field
is non-zero, but the functions pointer is NULL).

This fixes that segfault by zeroing out the num_functions
field if reallocation fails.

15 years agowgl: Cope with pre-existing threads.
José Fonseca [Mon, 27 Apr 2009 17:48:11 +0000 (18:48 +0100)]
wgl: Cope with pre-existing threads.

DllMain is called with DLL_THREAD_ATTACH only by threads created after
the DLL is loaded by the process.

15 years agowgl: Implement WGL_EXT_extensions_string extension.
José Fonseca [Mon, 27 Apr 2009 16:30:05 +0000 (17:30 +0100)]
wgl: Implement WGL_EXT_extensions_string extension.

15 years agoutil: Remove unix includes.
José Fonseca [Mon, 27 Apr 2009 16:13:40 +0000 (17:13 +0100)]
util: Remove unix includes.

15 years agoi965: only upload constant buffer data when we actually need the const buffer
Brian Paul [Mon, 27 Apr 2009 15:51:46 +0000 (09:51 -0600)]
i965: only upload constant buffer data when we actually need the const buffer

Make the use_const_buffer field per-program and only call the code which
updates the constant buffer's data if the flag is set.

This should undo the perf regression from 20f3497e4b6756e330f7b3f54e8acaa1d6c92052

15 years agoutil/time: add util_time_sleep() for windows userspace
Keith Whitwell [Mon, 27 Apr 2009 13:43:31 +0000 (14:43 +0100)]
util/time: add util_time_sleep() for windows userspace

Somebody with a clue could probably do a better implemenation...

15 years agomesa/st: fix incorrect face, level in compress_with_blit
Keith Whitwell [Mon, 27 Apr 2009 13:42:23 +0000 (14:42 +0100)]
mesa/st: fix incorrect face, level in compress_with_blit

We were incorrectly applying the destination texture face and level
when requesting a transfer to the temporary texture, which has only
one face and level.  This would obviously cause problems uploading to
compressed cube and mipmap textures.

15 years agodemos: Clean up allocated Textures and Display Lists when demo quit
Shuang He [Mon, 27 Apr 2009 13:13:33 +0000 (07:13 -0600)]
demos: Clean up allocated Textures and Display Lists when demo quit

15 years agoutil: Add debug_printf_once
Micah Dowty [Fri, 24 Apr 2009 21:45:16 +0000 (23:45 +0200)]
util: Add debug_printf_once

15 years agomesa: Call _mesa_snprintf instead of snprintf.
José Fonseca [Mon, 27 Apr 2009 12:09:58 +0000 (13:09 +0100)]
mesa: Call _mesa_snprintf instead of snprintf.

snprintf not directly available on Windows.

15 years agostw: Use a statically initiallized gl proc table.
José Fonseca [Thu, 23 Apr 2009 12:20:06 +0000 (13:20 +0100)]
stw: Use a statically initiallized gl proc table.

It doesn't change anyway.

15 years agor300: always emit output insts after all KIL insts
Maciej Cencora [Tue, 21 Apr 2009 00:38:08 +0000 (02:38 +0200)]
r300: always emit output insts after all KIL insts

15 years agor300-gallium: Correctly flush Draw.
Corbin Simpson [Sun, 26 Apr 2009 17:06:02 +0000 (10:06 -0700)]
r300-gallium: Correctly flush Draw.

Should help with a few non-TCL bugs.

15 years agor300-gallium: Add a draw_flush() to r300_flush().
Mathias Gottschlag [Sun, 26 Apr 2009 10:04:35 +0000 (12:04 +0200)]
r300-gallium: Add a draw_flush() to r300_flush().

This fixes some missing primitives which had been drawn right before the next glClear().

15 years agor300-gallium: Fix vertex shader OVM counting.
Corbin Simpson [Sat, 25 Apr 2009 23:53:38 +0000 (16:53 -0700)]
r300-gallium: Fix vertex shader OVM counting.

Attribs must be packed: position, point size, colors, texcoords.

Thanks to osiris for pointing it out.

15 years agor300-gallium: Clean up FB state emit.
Corbin Simpson [Fri, 24 Apr 2009 23:53:38 +0000 (16:53 -0700)]
r300-gallium: Clean up FB state emit.

15 years agor300-gallium: Set framebuffer pitch on every framebuffer change.
Mathias Gottschlag [Fri, 24 Apr 2009 23:27:23 +0000 (01:27 +0200)]
r300-gallium: Set framebuffer pitch on every framebuffer change.

Signed-off-by: Corbin Simpson <MostAwesomeDude@gmail.com>
15 years agogallium-intel: Fix warning
Jakob Bornecrantz [Sat, 25 Apr 2009 11:56:07 +0000 (12:56 +0100)]
gallium-intel: Fix warning

15 years agogallium-intel: Cache software switch
Jakob Bornecrantz [Sat, 25 Apr 2009 11:55:21 +0000 (12:55 +0100)]
gallium-intel: Cache software switch

15 years agogallium-intel: Link dri2 driver with softpipe
Jakob Bornecrantz [Sat, 25 Apr 2009 11:52:04 +0000 (12:52 +0100)]
gallium-intel: Link dri2 driver with softpipe

15 years agoGNU/Hurd fixes
Samuel Thibault [Thu, 23 Apr 2009 12:43:22 +0000 (05:43 -0700)]
GNU/Hurd fixes

Here is a couple of fixes for GNU/Hurd:
- dri_interface.h: no libdrm support either.
- configure.ac:
 - GNU/Hurd is a GNU OS with _GNU_SOURCE and PTHREADS.
 - GNU needs a couple of flags like other OSes

Signed-off-by: Dan Nicholson <dbn.lists@gmail.com>
15 years agointel: Fix more issues with the combined depth-stencil attachment
Ian Romanick [Fri, 24 Apr 2009 23:39:00 +0000 (16:39 -0700)]
intel: Fix more issues with the combined depth-stencil attachment

15 years agointel: Initialize region ptr to prevent assertion in intel_region_reference
Ian Romanick [Fri, 24 Apr 2009 23:14:51 +0000 (16:14 -0700)]
intel: Initialize region ptr to prevent assertion in intel_region_reference

15 years agotgis: SSE code generator doesn't yet support indirect addressing of temp regs
Brian Paul [Fri, 24 Apr 2009 23:08:59 +0000 (17:08 -0600)]
tgis: SSE code generator doesn't yet support indirect addressing of temp regs

Fall back to interpreter in this case.

15 years agointel / DRI2: When available, use DRI2GetBuffersWithFormat
Ian Romanick [Tue, 21 Apr 2009 03:56:45 +0000 (20:56 -0700)]
intel / DRI2: When available, use DRI2GetBuffersWithFormat

This interface gives the driver two important features.  First, it can
allocate the (fake) front-buffer only when needed.  Second, it can
tell the buffer allocator the format of buffers being allocated.  This
enables support for back-buffer and depth-buffer with different bits
per pixel.

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kristian Høgsberg <krh@redhat.com>
15 years agoDRI2: Implement interface for drivers to access DRI2GetBuffersWithFormat
Ian Romanick [Tue, 21 Apr 2009 03:55:56 +0000 (20:55 -0700)]
DRI2: Implement interface for drivers to access DRI2GetBuffersWithFormat

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kristian Høgsberg <krh@redhat.com>
15 years agoDRI2: Implement protocol for DRI2GetBuffersWithFormat
Ian Romanick [Tue, 21 Apr 2009 03:52:56 +0000 (20:52 -0700)]
DRI2: Implement protocol for DRI2GetBuffersWithFormat

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kristian Høgsberg <krh@redhat.com>
15 years agomesa: fix up error/warning/debug output newlines
Brian Paul [Fri, 24 Apr 2009 16:46:40 +0000 (10:46 -0600)]
mesa: fix up error/warning/debug output newlines

As of commit 23ad86cfb91c294ce85a3116d4b825aaa3988a6e all messages go
through output_if_debug().

Add new parameter to output_if_debug() to indicate whether to emit a newline.

_mesa_warning() and _mesa_error() calls should not end their strings with \n.
_mesa_debug() calls should end their text with \n.

15 years agoi965: use drm_intel_gem_bo_map/unmap_gtt() when possible, otherwise dri_bo_subdata()
Brian Paul [Fri, 24 Apr 2009 15:50:11 +0000 (09:50 -0600)]
i965: use drm_intel_gem_bo_map/unmap_gtt() when possible, otherwise dri_bo_subdata()

This wraps up the unfinished business from commit a9a363f8298e9d534e60e3d2869f8677138a1e7e

15 years agodemos: ensure display lists are destroyed for next generation
Alan Hourihane [Fri, 24 Apr 2009 15:44:58 +0000 (16:44 +0100)]
demos: ensure display lists are destroyed for next generation

15 years agopipebuffer: don't fail when validating mapped buffers
Keith Whitwell [Fri, 24 Apr 2009 11:43:04 +0000 (12:43 +0100)]
pipebuffer: don't fail when validating mapped buffers

This can be almost impossible to avoid - hopefully we won't encounter
a situation where this is a true requirement.  Would probably require
drivers to flush between hardware and software vertex processing.

15 years agoshadowtex: fflush stdout for cygwin
Keith Whitwell [Fri, 24 Apr 2009 11:16:29 +0000 (12:16 +0100)]
shadowtex: fflush stdout for cygwin

15 years agoutil: Add more entry points for dumping to bmp
Jakob Bornecrantz [Thu, 23 Apr 2009 03:42:26 +0000 (05:42 +0200)]
util: Add more entry points for dumping to bmp

15 years agodemos/readpix: add option to draw triangle instead of drawpix
Keith Whitwell [Wed, 22 Apr 2009 14:19:44 +0000 (15:19 +0100)]
demos/readpix:  add option to draw triangle instead of drawpix

15 years agotests/mipmap_view: add a bunch of keystrokes for testing render-to-texture
Keith Whitwell [Tue, 21 Apr 2009 18:49:29 +0000 (19:49 +0100)]
tests/mipmap_view: add a bunch of keystrokes for testing render-to-texture

Move between mipmaps, render a triangle, reload textures with either
the original arch (and GenMipmaps) or via straightforward glTexImage.

15 years agoi965: fix point size issue
Roland Scheidegger [Thu, 23 Apr 2009 21:41:41 +0000 (23:41 +0200)]
i965: fix point size issue

need to clamp point size to user set min/max values, even for constant
point size. Fixes glean pointAtten test.

15 years agomesa: more informative error messages
Brian Paul [Thu, 23 Apr 2009 23:54:34 +0000 (17:54 -0600)]
mesa: more informative error messages

15 years agoi965: revert part of commit 4f4907d69f9020ce17aef21b6431d2dd65e01982
Brian Paul [Thu, 23 Apr 2009 23:41:23 +0000 (17:41 -0600)]
i965: revert part of commit 4f4907d69f9020ce17aef21b6431d2dd65e01982

The drm_intel_gem_bo_map_gtt() call that replaced dri_bo_map() is
producing errors like:

intel_bufmgr_gem.c:689: Error preparing buffer map 39 (vp_const_buffer): Invalid argument .

and returning NULL, causing a segfault in the memcpy().

Just reverting until we can get to the root issue...

15 years agoi915: fix fix for anisotropic filtering
Roland Scheidegger [Thu, 23 Apr 2009 19:55:25 +0000 (21:55 +0200)]
i915: fix fix for anisotropic filtering

forgot to commit the changes to actually support 4x aniso filtering...

15 years agoi965: Support drawing to FBO cube faces other than positive X.
Eric Anholt [Thu, 23 Apr 2009 16:37:55 +0000 (09:37 -0700)]
i965: Support drawing to FBO cube faces other than positive X.

Also fixes drawing to 3D texture depth levels.

15 years agointel: Take advantage of GL_READ_ONLY_ARB to map to GEM bo_map write flag.
Eric Anholt [Wed, 22 Apr 2009 23:24:42 +0000 (16:24 -0700)]
intel: Take advantage of GL_READ_ONLY_ARB to map to GEM bo_map write flag.

This is a CPU win in general, but in particular reduces the pain of
Mesa's calculation of min/max indices in DrawElements (wtf?).

15 years agogallium: Handle non-NULL data pointer in EXA ModifyPixmapHeader hook.
Michel Dänzer [Thu, 23 Apr 2009 13:47:31 +0000 (14:47 +0100)]
gallium: Handle non-NULL data pointer in EXA ModifyPixmapHeader hook.

Need to use the data pointed to for pixmap contents in that case.

Fixes RENDER based text rendering.

15 years agogallium: Make the intel xorg winsys start up with any Intel chipset.
Michel Dänzer [Thu, 23 Apr 2009 12:19:56 +0000 (13:19 +0100)]
gallium: Make the intel xorg winsys start up with any Intel chipset.

For unsupported devices the screen/context creation should fail cleanly later
on.

15 years agogallium/intel/gem: Use softpipe rather than i915simple if INTEL_SOFTPIPE is set.
Michel Dänzer [Thu, 23 Apr 2009 11:57:46 +0000 (12:57 +0100)]
gallium/intel/gem: Use softpipe rather than i915simple if INTEL_SOFTPIPE is set.

15 years agogallium: Always include xorg-server.h before other X server headers.
Michel Dänzer [Thu, 23 Apr 2009 11:20:22 +0000 (12:20 +0100)]
gallium: Always include xorg-server.h before other X server headers.

Various breakage otherwise, e.g. _XSERVER64 not being defined on 64 bit leading
to inconsistent definitions of X server internal structs.

15 years agogallium: Fix up xorg state tracker build.
Michel Dänzer [Thu, 23 Apr 2009 11:17:28 +0000 (12:17 +0100)]
gallium: Fix up xorg state tracker build.

15 years agomesa: minor state-update changes in histogram code
Brian Paul [Wed, 22 Apr 2009 22:41:05 +0000 (16:41 -0600)]
mesa: minor state-update changes in histogram code

Call FLUSH_VERTICES() in _mesa_Histogram().
No need to signal _NEW_PIXEL in ResetHistogram(), ResetMinmax().

15 years agomesa: fix comment typo
Brian Paul [Wed, 22 Apr 2009 22:38:04 +0000 (16:38 -0600)]
mesa: fix comment typo

15 years agogallium: license, copyright
Brian Paul [Wed, 22 Apr 2009 21:12:22 +0000 (15:12 -0600)]
gallium: license, copyright

15 years agost: comments, license, copyright
Brian Paul [Wed, 22 Apr 2009 21:10:36 +0000 (15:10 -0600)]
st: comments, license, copyright

15 years agomesa: fix _mesa_dump_textures(), add null ptr check
Brian Paul [Wed, 22 Apr 2009 21:02:01 +0000 (15:02 -0600)]
mesa: fix _mesa_dump_textures(), add null ptr check

Calling _mesa_dump_textures() deleted the textures... oops!!!

15 years agomesa: Fix buffer overflow when parsing generic vertex attributes.
Brian Paul [Wed, 22 Apr 2009 17:16:03 +0000 (18:16 +0100)]
mesa: Fix buffer overflow when parsing generic vertex attributes.

15 years agointel: fix max anisotropy supported
Roland Scheidegger [Wed, 22 Apr 2009 15:37:18 +0000 (17:37 +0200)]
intel: fix max anisotropy supported

i915 actually supports up to 4 (according to header file - not tested),
i965 up to 16 (code already handled this but slightly broken), so don't use 2
for all chips, even though angular dependency is very high.

15 years agost: also check _NEW_PROGRAM flag for vertex shader constant buffers
Brian Paul [Wed, 22 Apr 2009 13:58:25 +0000 (07:58 -0600)]
st: also check _NEW_PROGRAM flag for vertex shader constant buffers

This is a follow-on to commit c1a3b852807fb160f0cd246c1364b7336b4b947e.
Note that (at this time) wherever _NEW_PROGRAM_CONSTANTS is set we're still
setting _NEW_PROGRAM so this won't really make any difference (for now).

15 years agogallium: Reinstate unconditional flushes.
Thomas Hellstrom [Mon, 20 Apr 2009 18:43:56 +0000 (20:43 +0200)]
gallium: Reinstate unconditional flushes.

Lost in commit e50dd26ca6d0eb0d0f97c2780020ea16e3d4a687.

Signed-off-by: Thomas Hellstrom <thellstrom-at-vmware-dot-com>
15 years agomesa: protect driver.flush() with FLUSH_CURRENT
Keith Whitwell [Tue, 21 Apr 2009 15:55:57 +0000 (16:55 +0100)]
mesa: protect driver.flush() with FLUSH_CURRENT

Need to do this to ensure vbo code unmaps its buffers before calling
the driver, which may be sitting on top of a memory manager which
objects to firing commands from a mapped buffer.

15 years agost: play it safe for now and check _NEW_PROGRAM for shader const buffer atom
Brian Paul [Tue, 21 Apr 2009 23:00:54 +0000 (17:00 -0600)]
st: play it safe for now and check _NEW_PROGRAM for shader const buffer atom

When a new program is bound but no constants are updated we still need
to update the Gallium const buffer.

15 years agoswrast: simplify state update logic for fragment shader const buffers
Brian Paul [Tue, 21 Apr 2009 23:00:01 +0000 (17:00 -0600)]
swrast: simplify state update logic for fragment shader const buffers

15 years agost: use the static atoms[] array directly
Brian Paul [Tue, 21 Apr 2009 22:50:34 +0000 (16:50 -0600)]
st: use the static atoms[] array directly

We can simplify this now that we no longer have any dynamic atoms.

15 years agost: do away with dynamic state atom for const buffers
Brian Paul [Tue, 21 Apr 2009 22:47:30 +0000 (16:47 -0600)]
st: do away with dynamic state atom for const buffers

Just use the new _NEW_PROGRAM_CONSTANTS flag instead.

15 years agomesa: new _NEW_PROGRAM_CONSTANTS flag
Brian Paul [Tue, 21 Apr 2009 21:56:45 +0000 (15:56 -0600)]
mesa: new _NEW_PROGRAM_CONSTANTS flag

This state flag will be used to indicate that vertex/fragment program
constants have changed.  _NEW_PROGRAM will be used to indicate changes
to the vertex/fragment shader itself, or misc related state.

_NEW_PROGRAM_CONSTANTS is also set whenever a program parameter that's
tracking GL state has changed.  For example, if the projection matrix is
in the parameter list, calling glFrustum() will cause _NEW_PROGRAM_CONSTANTS
to be set.  This will let to remove the need for dynamic state atoms in
some drivers.

For now, we still set _NEW_PROGRAM in all the places we used to.  We'll no
longer set _NEW_PROGRAM in glUniform() after drivers/etc have been updated.

15 years agomesa: print internal.current[i] attrib
Brian Paul [Tue, 21 Apr 2009 20:47:57 +0000 (14:47 -0600)]
mesa: print internal.current[i] attrib

15 years agomesa: print parameter list dirty state flag mask
Brian Paul [Tue, 21 Apr 2009 20:44:03 +0000 (14:44 -0600)]
mesa: print parameter list dirty state flag mask

15 years agoi965: const correctness
Brian Paul [Mon, 20 Apr 2009 17:58:09 +0000 (11:58 -0600)]
i965: const correctness

15 years agoUpdate GALLIUM_AUXILIARY_DIRS in configure.ac to match configs/default.
Michel Dänzer [Tue, 21 Apr 2009 17:32:42 +0000 (18:32 +0100)]
Update GALLIUM_AUXILIARY_DIRS in configure.ac to match configs/default.

15 years agor300: r300 hw doesn't support any input modifiers in tex insts
Maciej Cencora [Mon, 20 Apr 2009 23:55:06 +0000 (01:55 +0200)]
r300: r300 hw doesn't support any input modifiers in tex insts

15 years agor300-gallium: Fix CS size mismatch
Mathias Gottschlag [Tue, 21 Apr 2009 15:52:30 +0000 (09:52 -0600)]
r300-gallium: Fix CS size mismatch

This fixes some warnings which appear because the driver assumes a wrong
cs size (13 vs 16 register writes in some cases).