OSDN Git Service
Brian [Wed, 20 Jun 2007 17:01:09 +0000 (11:01 -0600)]
Merge branch 'master' of git+ssh://brianp@git.freedesktop.org/git/mesa/mesa
Brian [Wed, 20 Jun 2007 17:01:04 +0000 (11:01 -0600)]
Effectively disable _TexEnvProgram before calling _swrast_DrawPixels().
It's OK to use _TexEnvProgram regardless of the texture state, but if fog
is also enabled, the fragment program is lacking the actual fog computation
so fogging doesn't appear.
Fixing this might involve a new _MaintainFogProgram field and related code.
For now, just disable the _TexEnvProgram and let swrast handle everything.
Brian [Wed, 20 Jun 2007 16:56:13 +0000 (10:56 -0600)]
fix glDrawPixels + fragment program problem
Brian [Wed, 20 Jun 2007 16:55:31 +0000 (10:55 -0600)]
copy wpos attrib info into zoomed span (fixes fog perspective correction problem)
Brian [Wed, 20 Jun 2007 00:50:26 +0000 (18:50 -0600)]
Merge branch 'master' of git+ssh://brianp@git.freedesktop.org/git/mesa/mesa
Michel Dänzer [Fri, 15 Jun 2007 13:24:35 +0000 (15:24 +0200)]
i915tex: Missing piece of SetTexOffset hook implementation for I830 generation.
Stephane Marchesin [Wed, 13 Jun 2007 22:44:16 +0000 (00:44 +0200)]
nouveau: remove useless viewport xform with id matrix.
Brian [Tue, 12 Jun 2007 23:17:43 +0000 (17:17 -0600)]
fix an error string, refactor _mesa_GetCompressedTexImageARB() to get rid of a goto
Brian [Tue, 12 Jun 2007 23:06:27 +0000 (17:06 -0600)]
fix make clean commands so missing subdirs don't cause infinite loops
Brian [Tue, 12 Jun 2007 21:16:21 +0000 (15:16 -0600)]
added memcpy() to-do item
Brian [Tue, 12 Jun 2007 21:08:43 +0000 (15:08 -0600)]
fix comment
Brian [Tue, 12 Jun 2007 15:47:03 +0000 (09:47 -0600)]
Fix TEXREL issues when using GLX_USE_TLS (see bug 7459).
Brian [Tue, 12 Jun 2007 14:57:12 +0000 (08:57 -0600)]
fix GLX_USE_TLS breakage, fix a warning
Brian [Mon, 11 Jun 2007 23:16:18 +0000 (17:16 -0600)]
Rework _mesa_update_texture_compare_function() to only be called during
state validation/update.
Note that we're still temporarily skipping the test for an active fragment
program. Need to fix shadow2D() ...
Brian [Mon, 11 Jun 2007 22:42:35 +0000 (16:42 -0600)]
rename/clean-up _mesa_validate_texture_wrap_mode
Brian [Mon, 11 Jun 2007 22:36:06 +0000 (16:36 -0600)]
fix typo, added comment
Brian [Mon, 11 Jun 2007 22:32:57 +0000 (16:32 -0600)]
Replace texobj->Complete with texobj->_Complete since it's a derived field.
Brian [Mon, 11 Jun 2007 22:27:33 +0000 (16:27 -0600)]
typo: s/derrived/derived/
Brian [Mon, 11 Jun 2007 22:04:36 +0000 (16:04 -0600)]
Temporarily disable the calls to _mesa_update_texture_compare_function().
This fixes the depth-peel regression reported by Brad King.
Brian [Mon, 11 Jun 2007 16:57:01 +0000 (10:57 -0600)]
In generic_nop() call _mesa_warning() instead of _mesa_problem() since it's an app issue, not a mesa bug.
Brian [Fri, 8 Jun 2007 21:02:05 +0000 (15:02 -0600)]
remove dead code
Dave Airlie [Sun, 10 Jun 2007 02:25:23 +0000 (12:25 +1000)]
r300: i can't see why we fallback for polygon offset line/fill
Brian [Fri, 8 Jun 2007 20:13:13 +0000 (14:13 -0600)]
In supported_buffer_bitmask(), use framebuffer's visual, not context's visual
Brian [Fri, 8 Jun 2007 20:12:27 +0000 (14:12 -0600)]
disable depthBits test in check_compatible(), see bug 11161
Brian [Fri, 8 Jun 2007 19:10:01 +0000 (13:10 -0600)]
fix glReadPixels parameters to handle odd-width windows correctly
Brian [Fri, 8 Jun 2007 19:02:14 +0000 (13:02 -0600)]
disable shader program before glDrawPixels
Brian [Fri, 8 Jun 2007 18:24:35 +0000 (12:24 -0600)]
remove obsolete comment about max DEFAULT_SOFTWARE_DEPTH_BITS
Ian Romanick [Thu, 7 Jun 2007 20:58:50 +0000 (13:58 -0700)]
Add support for GL_ARB_fragment_program_shadow.
Ian Romanick [Thu, 7 Jun 2007 20:38:06 +0000 (13:38 -0700)]
Fix ARB_fp spec conformance bug WRT shadow sampling.
The ARB_fp (and other assembly-level fragment program specs) say that the
depth comparison function is always GL_NONE in fragment program mode.
Ian Romanick [Thu, 7 Jun 2007 19:12:14 +0000 (12:12 -0700)]
Add support for shadow mapping with vertex and fragment programs
Ian Romanick [Thu, 7 Jun 2007 19:09:44 +0000 (12:09 -0700)]
Refactor shadow map modelview and projection code
Explicitly store the modelview matrix and projection matrix used when the
shadow map is created. These matrices only need be updated when
RenderShadowMap is called. Previously, slightly different code was used to
draw and project the shadow map.
Ian Romanick [Thu, 7 Jun 2007 18:59:42 +0000 (11:59 -0700)]
Remove '#if defined' tests for ARB extensions.
Oliver McFadden [Thu, 7 Jun 2007 13:21:18 +0000 (13:21 +0000)]
r300: Added a comment regarding the R300_VAP_CLIP registers.
Oliver McFadden [Thu, 7 Jun 2007 13:20:27 +0000 (13:20 +0000)]
r300: Removed a "don't think this is needed" comment, it actually is needed.
According to the comment you do not need to setup INPUT_CNTL when using vertex
buffers. However, not doing so results in a lockup.
Oliver McFadden [Thu, 7 Jun 2007 13:20:08 +0000 (13:20 +0000)]
r300: Explain the R300_VAP_OUTPUT_VTX_FMT_1 register.
Oliver McFadden [Thu, 7 Jun 2007 13:19:47 +0000 (13:19 +0000)]
r300: Cleaned up long lines in r300Choose8888TexFormat.
Wang Zhenyu [Tue, 5 Jun 2007 18:42:43 +0000 (11:42 -0700)]
Add PCI IDs for the G33, Q33, and Q35 chipsets.
Michel Dänzer [Tue, 5 Jun 2007 17:55:53 +0000 (19:55 +0200)]
i915tex: Better attempt to release miptree when overriding texture image.
The previous approach could lead to crashes in FBO code that dereferences the
miptree struct pointer unconditionally.
Michel Dänzer [Tue, 5 Jun 2007 17:52:10 +0000 (19:52 +0200)]
Remove dubious compile-time test for pre-2.4 Linux kernels.
LINUX_VERSION_CODE shouldn't be used by userspace code, it can be defined empty
these days.
If anybody still cares about 2.2 kernels, they should reinstate this as a
proper runtime test.
Ian Romanick [Tue, 5 Jun 2007 16:24:27 +0000 (09:24 -0700)]
Updates for array texture shadow targets.
Dave Airlie [Mon, 4 Jun 2007 23:38:49 +0000 (09:38 +1000)]
miniglx: implement GetProcAddressARB in miniglx as an alias for GPA
Dave Airlie [Mon, 4 Jun 2007 23:36:28 +0000 (09:36 +1000)]
miniglx: update defines in linux-solo config
Dave Airlie [Sat, 2 Jun 2007 06:21:50 +0000 (16:21 +1000)]
r300: fix non-tcl rs4xx again.
Wang Zhenyu [Wed, 30 May 2007 08:18:26 +0000 (16:18 +0800)]
i915tex: Add support for 945GME
Wang Zhenyu [Wed, 30 May 2007 08:11:12 +0000 (16:11 +0800)]
i915: Add support for 945GME chip
Wang Zhenyu [Wed, 30 May 2007 08:03:50 +0000 (16:03 +0800)]
i965: Add pci info for 965GME/GLE chip.
Michel Dänzer [Wed, 30 May 2007 13:37:42 +0000 (15:37 +0200)]
Fix r300SetTexOffset for big endian platforms.
This was broken by the unification of the texture format table.
Oliver McFadden [Wed, 30 May 2007 03:15:52 +0000 (03:15 +0000)]
r300: Corrected r300PointSize based on dumping the blob.
The OpenGL specification also verifies the default point size should be 1.0.
Oliver McFadden [Wed, 30 May 2007 03:11:49 +0000 (03:11 +0000)]
r300: Corrected r300LineWidth based on dumping the blob.
The OpenGL specification also verifies the default line width should be 1.0.
Oliver McFadden [Tue, 29 May 2007 05:07:33 +0000 (05:07 +0000)]
r300: Removed goto statement in r300SetupRSUnit.
Oliver McFadden [Tue, 29 May 2007 05:02:03 +0000 (05:02 +0000)]
r300: Cleaned up r300SetupRSUnit.
Oliver McFadden [Tue, 29 May 2007 02:58:10 +0000 (02:58 +0000)]
r300: Cleaned up the state atom checking functions.
Oliver McFadden [Tue, 29 May 2007 01:18:33 +0000 (01:18 +0000)]
r300: Cleaned up the state atom debugging code.
Oliver McFadden [Mon, 28 May 2007 01:34:26 +0000 (01:34 +0000)]
r300: Use compile-time endian detection in r300_state.c as well as r300_texstate.c.
Probably best to not mix-and-match compile-time and run-time detection...
Oliver McFadden [Mon, 28 May 2007 01:11:54 +0000 (01:11 +0000)]
r300: Document registers 0x2220 to 0x2230.
These registers are per-pixel and per-vertex X and Y clipping planes.
Oliver McFadden [Mon, 28 May 2007 00:23:47 +0000 (00:23 +0000)]
r300: Only support size 4 ELTs; this is what Mesa provides.
Oliver McFadden [Mon, 28 May 2007 00:16:50 +0000 (00:16 +0000)]
r300: Use the CP_PACKET3 macro for Type 3 packets.
I haven't converted all of the Type 3 packets to the CP_PACKET3 macro yet
because some of the Type 3 packet defines are missing from the R300 register
definition file.
These defines need to be copied from DRM and Mesa into the R300 register
definition file then copied into both DRM and Mesa.
Oliver McFadden [Sun, 27 May 2007 23:35:12 +0000 (23:35 +0000)]
r300: Removed the (completely broken since new VBO branch) OPTIMIZE_ELTS path.
The VTXFMT code was broken by the new VBO branch and the OPTIMIZE_ELTS path
relied on the VTXFMT code...
I'm not even sure if the OPTIMIZE_ELTS path ever worked; the testing that I did
after minimizing the code duplication would have taken the same path as the
non-OPTIMIZE_ELTS code.
Oliver McFadden [Sun, 27 May 2007 23:19:30 +0000 (23:19 +0000)]
r300: Cleaned up the AOS code in r300_render.c.
Oliver McFadden [Sun, 27 May 2007 21:45:19 +0000 (21:45 +0000)]
r300: Added a comment in r300VAPInputCntl0.
Oliver McFadden [Sun, 27 May 2007 20:23:52 +0000 (20:23 +0000)]
r300: Cleaned up r300DestroyTexObj.
Oliver McFadden [Sun, 27 May 2007 15:25:50 +0000 (15:25 +0000)]
r300: R300_SE_VTE_CNTL applies to both non-TCL and TCL hardware.
See r300ResetHwState.
Brian [Tue, 29 May 2007 14:05:10 +0000 (08:05 -0600)]
use glXGetProcAddress to get extension funcs
Brian [Fri, 25 May 2007 22:22:15 +0000 (16:22 -0600)]
check for flat/smooth interp for generic/specular attrib
Oliver McFadden [Sun, 27 May 2007 02:25:31 +0000 (02:25 +0000)]
Revert "r300: Removed the R300_RS_INTERP_[0-9]_UNKNOWN (magic) defines."
This reverts commit
bb3558e6517209086cf8426bbe4743da50351158.
This commit caused a regression reported by Markus Amsler
<markus.amsler@oribi.org>. Apparently these defines are required, although I'm
not sure why.
Oliver McFadden [Sat, 26 May 2007 21:38:52 +0000 (21:38 +0000)]
r300: Rearranged the DWORD construction in r300VAPInputRoute0 for clarity.
Doesn't actually change anything; just makes it easier to read.
Oliver McFadden [Sat, 26 May 2007 20:59:16 +0000 (20:59 +0000)]
r300: Removed duplicate AOS format defines; already defined in r300_reg.h.
Oliver McFadden [Sat, 26 May 2007 19:39:35 +0000 (19:39 +0000)]
r300: Removed unused aos_reg variable.
Oliver McFadden [Sat, 26 May 2007 17:55:30 +0000 (17:55 +0000)]
r300: Use "rmesa" not "r300" in r300_emit.c; some of the macros require "rmesa".
Oliver McFadden [Sat, 26 May 2007 17:36:08 +0000 (17:36 +0000)]
r300: Cleaned up the non-TCL RENDERINPUTS, etc.
Oliver McFadden [Sat, 26 May 2007 17:13:57 +0000 (17:13 +0000)]
r300: Whitespace cleanup in r300_emit.c.
Oliver McFadden [Sat, 26 May 2007 15:49:24 +0000 (15:49 +0000)]
r300: Added the r300VAPInputCntl0 function.
The function just returns the hard-coded value (0x5555) even though we have no
idea what this means...
Oliver McFadden [Sat, 26 May 2007 15:45:04 +0000 (15:45 +0000)]
r300: Added r300VAPOutputCntl0 and r300VAPOutputCntl1 to r300_emit.c.
Oliver McFadden [Sat, 26 May 2007 14:43:04 +0000 (14:43 +0000)]
r300: Removed R300_PFS_NODE_LAST_NODE replaced by R300_PFS_NODE_OUTPUT_COLOR.
Oliver McFadden [Sat, 26 May 2007 12:34:55 +0000 (12:34 +0000)]
r300: Renamed the t_vic function.
Oliver McFadden [Sat, 26 May 2007 00:07:25 +0000 (00:07 +0000)]
r300: Renamed the t_vir0 and t_vir1 functions.
Oliver McFadden [Fri, 25 May 2007 23:50:36 +0000 (23:50 +0000)]
r300: Renamed "dt" to "attribptr" in the t_vir0 function.
Oliver McFadden [Fri, 25 May 2007 23:42:03 +0000 (23:42 +0000)]
r300: Specify the type in the t_vir0 function.
Oliver McFadden [Fri, 25 May 2007 23:02:57 +0000 (23:02 +0000)]
r300: Cleaned up t_vir0 and t_vir1 slightly.
Oliver McFadden [Fri, 25 May 2007 22:01:20 +0000 (22:01 +0000)]
r300: Use C style comments in r300_emit.c.
Oliver McFadden [Fri, 25 May 2007 21:54:45 +0000 (21:54 +0000)]
r300: Removed a (disabled and unneeded) assertion.
The r300EmitVec functions don't touch the buf member so the assertion isn't
needed here.
Oliver McFadden [Fri, 25 May 2007 21:49:53 +0000 (21:49 +0000)]
r300: Improved the r300EmitVec debugging information.
Oliver McFadden [Fri, 25 May 2007 21:43:24 +0000 (21:43 +0000)]
r300: Use #if 0 for disabled code.
Oliver McFadden [Fri, 25 May 2007 21:38:49 +0000 (21:38 +0000)]
r300: Added a TODO note and some tiny cleanups to r300_emit.c.
Dave Airlie [Sat, 26 May 2007 05:58:56 +0000 (15:58 +1000)]
add glxtokens.h to xmesa_xf86.h as discussed on mailing list
Brian [Fri, 25 May 2007 14:57:44 +0000 (08:57 -0600)]
remove #include "GL/glxtokens.h"
Oliver McFadden [Fri, 25 May 2007 13:59:15 +0000 (13:59 +0000)]
r300: Use GL_TRUE rather than 1 for the GLboolean type.
Oliver McFadden [Thu, 24 May 2007 09:15:57 +0000 (09:15 +0000)]
r300: Removed unused vpucount macro from r300_state.c.
Brian [Thu, 24 May 2007 23:07:48 +0000 (17:07 -0600)]
fix logic for calling _swrast_update_deferred_texture()
Brian [Wed, 23 May 2007 22:58:01 +0000 (16:58 -0600)]
doxygen-ize some comments
Brian [Wed, 23 May 2007 22:49:32 +0000 (16:49 -0600)]
restore GL_TEXTURE_LOD_BIAS in _mesa_PopAttrib(), bug 11049
Brian [Wed, 23 May 2007 22:42:29 +0000 (16:42 -0600)]
incr/decr 1D/2DArray texture refcounts like other targets
Oliver McFadden [Wed, 23 May 2007 21:26:35 +0000 (21:26 +0000)]
r300: Call the r300Fogfv function directly within r300_state.c.
This required moving the r300Enable function but there are no actual changes.
Brian [Wed, 23 May 2007 21:33:46 +0000 (15:33 -0600)]
Fix problem w/ two-sided lighting and fragment programs (depth-peel regression)
Oliver McFadden [Wed, 23 May 2007 21:12:11 +0000 (21:12 +0000)]
r300: Use switch statements in r300ResetHwState, etc.
Oliver McFadden [Wed, 23 May 2007 18:48:05 +0000 (18:48 +0000)]
r300: Minor indenting corrections in the texture format table.
Oliver McFadden [Wed, 23 May 2007 18:37:55 +0000 (18:37 +0000)]
r300: Use a single texture format table; reduces duplication.
Oliver McFadden [Wed, 23 May 2007 17:56:47 +0000 (17:56 +0000)]
r300: Corrected the RGB888 texture format entry.
I think this is correct, assuming no endian issues. See commmit
59a08923f51d4ed83effbfcd91473c9ee86465f1.
Oliver McFadden [Mon, 21 May 2007 15:51:38 +0000 (15:51 +0000)]
r300: Removed the R300_RS_INTERP_[0-9]_UNKNOWN (magic) defines.
Supposedly you need to set these values for the interpolaters to work, but they
seem to work fine without these values.
Brian [Wed, 23 May 2007 14:58:08 +0000 (08:58 -0600)]
Fix point attentuation problem (bug 11042)
ctx->Point._Attentuation was computed in wrong place and the VB->Eye coord Z
array wasn't indexed correctly in run_point_stage().