Tag 'mesa_20060325' created by Brian Paul <brian.p...@tungstengraphics.com> at 1970-01-01 00:00 +0000
Changes since mesa_20050715: Aapo Tahkola (79): Fix invalid vsf temp count for rv350. Sync with my local tree. Remove few outdated r200 files. Reduce stderr noise and fix some compiler warnings. Cover my arse. Compiled arrays for vtxfmt_a path. Fix texture mixup when two or more contexts. Silence gcc-4 warnings. Improve tnl program searching performance. Need more than 200 instructions. Fix segmentation fault in _tnl_ProgramCacheDestroy(). -Fix first frame -bug Re-enable fallbacks. Fix broken tnl stage. Remove some ghost code and adjust things a bit. Use ctx->VertexProgram._Current . Fix broken build. Reduce stderr noise. Fix typo causing secondary color not to work properly. Try to detect when native support to given texture width is not available. Couple changes that were missed. Fix AL8 Fix secondary color for VP's Make missing interpolator inputs fatal Add some missing hooks. This should fix various little problems with window movement and 2d windows flashing on top of 3d windows. This problem was previously shadowed by r300ResetHwState as it gets called frequently. Fix (likely) typo. Enable CB_DPATH by default. I think this has been tested well enough already. Warn only when being enabled. No you arent. Drop output checking. Handle fatal case. Fix node setup(from Ben Skeggs) Fix cube maps. Fix pow <small> and a very stypid bug with dummy srcs(0 equals to tmp0.x)</small>. demos/cubemap looks perfect now. Missing patch from Ben Skeggs: Fix invalid tex coord routing and couple other small things. namespace pollution fixes (Tilman Sauerbeck) Fix ARB_position_invariant handling Missing bits of code spotted by Tilman Sauerbeck Fix ARGB1555 component swizzling. Sync from drm. Try to tackle bug #5148 Doesnt apply anymore fix broken get_half & get_zero Fix a problem where offsets from disabled tmus reach drm remove some testing code r300UpdateTextureEnv shouldnt be called. Work around problems with t_vp_build.c R300_PFS_NODE_LAST_NODE -> R300_PFS_NODE_OUTPUT_COLOR drop all that (slow) locking stuff and make u_list per context. radeon driver doesnt have this so i assume we shouldnt either. per vp sw fallbacks enable hw vertex programs by default missing link in the fallback chain remove broken non arb fp path Workaround problems with NWN. (Tilman Sauerbeck) Skip ARL while we dont implement it. Fix broken nv vps but do it with software for now. Fix aliasing bug (Benjamin Herrenschmidt) No need to call *_FIREVERTICES twice. Dont let (possibly) undefined z component reach hw as its unused Free temps when possible Port r200 tex tiling code to run on r300 but keep it disabled. Various fallbacks, checks and tags. Convert to work with current DRM. + tag Enable HW VBOs and other things needed by it. Fix broken DrawElements & DrawRangeElements. Use the force Luke, use the force. Kick VBOs out of GART if nothing else helps. Option to init vp temps(for testing) Fix broken max mipmap leveling that was horribly wrong. Clean build. -fix texrect fallback when using vtxfmt_a (glitz, xgl, etc.) Disable tnl programs when doing software vertex programs. compiz with its texrect and lighting needs was broken because of this. Fix texrect upload conflicts. ARL dst idx was undefined. implement arl and enable hw nv_vp. secondary color fixes. Make ARB_vp backends happy with nv arl Adam Jackson (13): strip out vestigial #ifdef HAVE_CONFIG_H stanzas, they're confusing the XMesaResizeBuffers needs separate paths for client and server builds Add two aliases for some PointParameters entrypoints for libglx compatibility. Convert libGL and DRI drivers to require libdrm. /dri/msg Bug #1155: Define IEEE_ONE on Super-H as well. Add RV350 AQ chip. (popolon at popolon dot org) s/dri.sourceforge.net/dri.freedesktop.org/ Fix the default target to remake readtex.o if needed Start a software-only DRI driver for use in the new accelerated indirect Nuke the old attempt at software DRI, it's broken and awful Search for makedepend in the PATH rather than hardcoding /usr/X11R6/bin Avoid a crash when the user has enabled glVertex{,Attrib}Pointer but hasn't Alan Hourihane (26): pass in the renderbuffer Fix the FFB driver for the renderbuffer changes re-enable the ffb driver Bump the required ddx to 1.4.0 as drmAddress has been removed so that 32/64bit support GLX_DIRECT_COLOR in 1.4.0 typo remove the drmAddress Add a check which compares the passed in size of the DDX driver private DRI Add Egberts fixes for 64bit architectures Commit missing piece from Egbert's 32/64 bit patches Fix Windows build problems. update for mingw Fix the scripts for the cygwin & mingw changes remove drmAddress Fixes from https://bugs.freedesktop.org/attachment.cgi?id=3077 define Bool for solo builds move via_dri.h to server drmUnmap correct size reverse a bogus patch Add Intel 945GM support use screen->front.map instead of pFB from libdri use front.size instead of sPriv->fbSize remove very old i830 code pull in fix from 6.4 branch for XFree86Server definition remove i830 Check DDX for 1.5.0 Brian Paul (615): s/clippping/clipping/ overlay planes test include driverfuncs.h remove fprintf() replace fprintf() calls with _mesa_problem() updates for 6.3 release more updates for 6.3 release added md5 checksums for 6.3 fix some URLs remove ^M chars, disable shading language extensions check for FEATURE_EXT_framebuffer_object to enable EXT_framebuffer_object s/2004/2005/ first 6.3.1 bug fixes make i810PrintSetupFlags() static, first param const bump version to 6.3.1 and include DRI/GLX files in tarballs updated with DRI building instructions 6.3.1 updates bump version to 6.3.1 Make the vertex program source register Index field a signed int since added vp negative addressing bug fix some GLuint idx -> GLint idx changes Break the long list of MAIN_FILES and DEMO_FILES into subgroups. Added -linker option to mklib, used to specify a particular program for Add license/copyright. Fix bug 1858 - array subscript out of range in gamma driver. improve comments about win[0..3] values (bug 2554) SPARC, DRI interface changes for 6.3.1 added directfbgl.h to tarball s/Disaptch/Dispatch/ (bug 3918) free context structure (bug 1252316) and s/free/_mesa_free/ _mesa_free_framebuffer_data() call frees the renderbuffers, no need to call the Delete methods after free display lists and destroy window upon exit delete display lists upon exit actually free the renderbuffer in xmesa_delete_renderbuffer(), fixes a memory leak fix indentation disable the pointer size assertions (bug 4021) fix static assertion problem for gcc (bug 4022) the PIXEL_MAP_I_TO_I table should be GLfloat, not GLint fix X #includes so they're relative to X11/ directory (bug 4047) added FREXPF() macro (bug 4060) test the standard stencil op modes too (Philipp Klaus Krause) include stdint.h if WIN32 and GNUC (Colin Harrison). Patch forwarded to SGI. updates for 6.3.2 Bump version to 6.3.2 Replace mult/div operators with bitwise operators in texel fetch routines updates for 6.3.2 release, plus other assorted clean-ups update MESA_DEBUG with FP info md5 sums for 6.3.2 fix typo added .S DRI files to tarball (Donnie Berkholz), bumped version to 6.5 bump version to 6.5 added GL_X_RENDERABLE to glXChooseFBConfig (bug 4181) update version to 6.5.0 simplify, clean-up texel addressing macros removed lines that caused some files to be included in tarballs multiple times GLw demo doesn't belong here Added multi-window support. Press 'n' to create new window, 'k' to kill also check for texture border in sample_linear_2d() make tiles a little smaller so they're all visible s/GLX_RGBA_BIT/GLX_RGBA_TYPE/ (bug 4237) fix segfault in Fake_glXCreatePBuffer(), bug 4235 removed GL_HP_occlusion_test extension initial 6.5 changes (reflect demo, remove GL_HP_occlusion_test) Update the glut idle function after a new window is created remove stray references to ctx->Depth.OcclusionTest don't compile debug code (bug 4242: rand undefined) fix comment and #if test Demo of how to do (nearly) flicker-free rendering with a single color buffer. added singlebuffer demo Rearrange the code related to GL_ARB_occlusion_object to generalize query check the Ready flag when getting GL_QUERY_RESULT_ARB fix a few proxy color table bugs (bug 4270) added note about IEEE vs VAX floating point (bug 4270) fix APP_LIB_DEPS added some parenthesis to silence warnings silence fprintf and casting warnings fix a cast/assignment warning remove redundant ctx->Driver fallback assignments that were already done in _mesa_init_driver_functions() remove dead code, update comments use 'f' key to toggle front/back buffer rendering Just additional comments. remove the old set_buffer() routine remove old set_buffer routine and misc code remove set_buffer() remove fxDDSetBuffer() added a few more fields to help with DRI drivers use proper datatypes to silence a bunch of warnings use char* instead of GLubyte* in i915_program_error() to silence warnings s/__inline/INLINE/ Finish up some of the gl_renderbuffer work. Finish up some of the gl_renderbuffer work. some missing renderbuffer changes (Roland) fix typo fix page flipping breakage added driFlipRenderbuffers() added an assertion to help debug broken stencil readback set depthHasSurface field for stencil renderbuffer remove unneeded assertion in stencil/depth LOCAL_DEPTH_VARS code change scratch pointer from _int32_t to u_int32_t to silence warning. Is this field actually used for anything? set depthHasSurface for stencil renderbuffer replace -1 with PROGRAM_UNDEFINED when initializing instructions Prototype implementation of new GL_EXT_timer_query extension (not finalized yet). added GL_EXT_timer_query token comments GL_EXT_timer_query fix some strict aliasing problems related to choosing framebuffer configs (Matthias Hopf) silence misc warnings remove SetBuffer(). Misc renderbuffer changes. added SetStencil/DepthPointer functions added flippedData field for page flipping, code to flip stereo color buffers SetBuffer, renderbuffer changes remove SetBuffer remove dead code use new InitStencil/DepthPointer functions SetBuffer, renderbuffer changes renderbuffer updates Remove last remnants of pre-renderbuffer code. fudge set_component_sizes() a bit, see comments replace assertion with a conditional return Added InitPointers helper function Combine GET_DST_PTR and GET_SRC_PTR into just GET_PTR. Use InitPointers helper replace GET_SRC_PTR, GET_DST_PTR with GET_PTR minor clean-ups Fix UpdatePageFlipping() function. Always call driFlipRenderbuffers() Add new void *addr and __DRIdrawablePrivate parameters to Further clean-up and simplification of the span-related functions. Simplify radeonSetCliprects(). Remove the 'mode' parameter. just formatting clean-ups remove unused var remove obsolete CurrentBufferBit make r128SpanRenderStart/Finish static make some functions static don't use DEFARRAY/CHECKARRAY stuff minor clean-up of texture_combine() remove DEFARRAY, CHECKARRAY stuff Use mklib script instead of ar. Make libmesa.a instead of mesa.a added -Wmissing-prototypes to WARN_FLAGS just changes/updates to comments added viaRenderbufferStorage() function minor formatting fix try to fix renderbuffer breakage, primarily in calculate_buffer_parameters() minor tweaks to _mesa_ResizeBuffersMESA() new comment some updates, but code still out of date some clean-up and re-org for renderbuffers fix precision problem in depth clearing (bug 4395) check for either GL_ARB_depth_texture or GL_SGIX_depth_texture in a few places Fix the comment for the 'bound' field: it's a bitmask, not an index. don't generate selection hits for glBitmap, glDraw/CopyPixels. Misc clean-ups remove the redundant textureSize field, use tex.size instead, it's always the same value Add offset and pitch fields to intelRegion. Remove frontOffset, frontPitch, fix GLubyte/GLchan inconsistencies (bug 4331) make some vars static fix a problem w/ color index mode when CHAN_BITS=16 Clean-up of sources/Makefile, bug 4111 put GLUT sources into a separate tarball bunch of updates Silence some warnings. remove rendundant r,g,b,a vars (bug 4331) make lookup_query_object() static use ASM_API variable, bug 4415 restore brace lost in previous check-in added a return upon out-of-memory, remove a needless assignment init StencilOpSeparate to NULL Move the ctx->FirstTimeCurrent initialization Use GLuint/GLint intead of GLstencil for stencil state. Clear either 8 or 16-bit stencil buffers. just some reformatting OpenGL 2.0's two-sided stencil feature wasn't implemented correctly. Replace ctx->Driver.StencilOp/Func/Mask() functions with fix FreeBSD problem (bug 4435) print version with debug info replace semicolon with && update like for MD2 viewer Move _mesa_add_soft_renderbuffers() from framebuffer.c to renderbuffer.c added driUpdateFramebufferSize() Instead of calling _mesa_ResizeBuffersMESA() in the Viewport function, make some functions static updated comment use mklib script to make drivers instead of gcc -shared Instead of calling _mesa_ResizeBuffersMESA() in the Viewport function, make some funcs static Instead of calling _mesa_ResizeBuffersMESA() in the Viewport function, pageflip change from radeon driver remove .a file before creating new one remove rm -f line clean-up some _mesa_make_current() code Remove _tnl_MakeCurrent() and the unused ctx->Driver.MakeCurrent() callback. initial 6.5 release notes remove _tnl_MakeCurrent call remove unused vars Split the s_texture.c file into two new files: Update includes post-splitting of s_texture.c replace s_texture.[ch] with s_texcombine.[ch] and s_texfilter.[ch] New set_fetch_functions() to set a gl_texture_image's FetchTexel functions updated comments Remove a bunch texel fetch functions that can be handled by the new silence warning Code clean-ups, reorg. Replace GLuint with GLbitfield where appropriate. Also replace GLuint more GLuint->GLbitfield changes replace a couple macros with inlined functions Rework the texture filtering functions a bit. alphabetize use mesa import wrappers, bug 4468 remove unused ctx var pass mask=NULL to write all pixels in clear_ci_buffer_with_masking() change some comments lots of little clean-ups, improvements remove unused _swrast_read_depth_span() Fix potential segfault when trying to read pixels outside renderbuffer bounds. compilation fixes, don't know if runtime works s/unsigned long/drm_handle_t/ to get compilation of fb driver to work additional wrapper updates, bug 4468 updated comment s/GLchan/GLubyte/, fix error string, add comment init a=0 change location of ffs() code for windows remove STENCIL_MAX replace STENCIL_BITS with stencilMax value Change zLeft and zval to GLuint. Seems to fix 32-bit Z computation remove unneeded sizeof(GLdepth) test some improvements to _swrast_clear_depth_buffer() compute z[i] as GLuint change mask[] test in put_mono_row_uint() remove 31-bit Z buffer work-around initial check-in s/programss/programs/ replace GLdepth with GLuint replace GLdepth with GLuint and remove GLdepth Remove ACCUM_BITS. add missing scale factor in _swrast_read_depth_span_float() s/GL_DEPTH_BITS/GL_TEXTURE_DEPTH_SIZE_ARB/ in _mesa_GetTexLevelParameteriv() changed a renderbuffer check fix a debug message When testing visual attributes, use the Draw/Read buffer, not the context. Replace gl_renderbuffer.ComponentSizes[] with separate RedBits, GreenBits, improve the test for using software alpha buffer Add Wrapper field to gl_renderbuffer. Used for wrapping one renderbuffer Added a fast xor_line() function. Uses XDrawLine with GXxor mode. added XOR line optimization fix clipped glReadPixels bug Replace gl_framebuffer's _ColorReadBufferMask with _ColorReadBufferIndex, remove printf Added XMesaDrawable to xmesa_renderbuffer. Use it in a few places instead glDraw/CopyPixels of stencil data didn't obey the stencil write mask. Fixed. Lift all the format/type error checking out of the _swrast_Draw/Read/CopyPixels generate GL_INVALID_OPERATION in glAccum if read/draw buffers aren't the same tweak glAccum error checking added _mesa_scale_and_bias_depth() added GL_EXT_packed_depth_stencil change initialization of rb->DepthBits Initial work for GL_EXT_packed_depth_stencil extension. updated assertions (see bug 4590) check for framebuffer completeness, code clean-up add error checks for framebuffer completeness Rewrite much of the pixel zoom code. added _swrast_put_row() implement CopyPixels for GL_DEPTH_STENCIL_EXT call glFinish if drawing to front buffer added _swrast_get_row() minor improvements implemented fallbacks for GL_EXT_packed_depth_stencil use fmod() to prevent overflow of angle variable move around some error checking added _mesa_pack_depth_stencil_span() updated comment fix broken z24_s8 fetcher simplify the read_depth_stencil_pixels() function fix an assertion additional error checking for GL_EXT_packed_depth_stencil GL_EXT_packed_depth_stencil changes added _mesa_texstore_z24_s8() plug in _mesa_texstore_z24_s8() check for GL_DEPTH_STENCIL_EXT finish up some missing code added a few calls to _mesa_update_framebuffer_visual() disable a warning change initialization of renderbuffer DepthBits Use Visual.depthBits rather than Renderbuffer::DepthBits for depth buffer fix behaviour of glIsFrame/Renderbuffer() fix glIsRender/FramebufferEXT() assertions glBindTexture changes, new assertions check if deleting currently bound frame/renderbuffer object In gl_texture_image replace IntFormat with InternalFormat and Format with added GL_INVALID_FRAMEBUFFER_OPERATION_EXT to gluErrorString() s/IntFormat/InternalFormat/ and s/Format/_BaseFormat/ for gl_color_table. Remove _IsPowerOfTwo from gl_texture_object, not really needed. fix byteswapping bug in _mesa_texstore_ycbcr() (Benjamin Herrenschmidt) Fix a number of point size attenuation problems. MATH_DEBUG changes from bug #4468. Fix some issues with state updates and renderbuffers. Querying GL_RED_BITS, Query/print GL_RED/GREEN/BLUE/ALPHA_BITS change test for presence of accum buffer fixed constant fogcoord bug (#4729) check for PIXMAP in clip routine (SF bug 1323215) fix GLAPI/extern problem (bug 4757) change includes (bug 4468) remove unused _mesa_log2() (bug 4468) added some header files to DRI_FILES (Donnie Berkholz) Lots of clean-ups. Replace __inline__ with INLINE macro. Use the move _mesa_exec_malloc/free() up near other malloc/free prototypes s/memcpy/_mesa_memcpy/, s/FREE/_mesa_free/ updated for new mm.c, execmem.c files s/glut_gz2/glut_bz2/ Disable the FASTCALL optimization because it still breaks server-side GLX. fix another renderbuffer wrapping bug added mm.c mm.h execmem.c fix bad loops in TransposeMatrix[fd]() restore FASTCALL stuff fix broken KIL fix broken SWZ instruction Darwin version fix (SF bug 1334274) check span.y >=0 (bug 4842) fix GLX server resize/crash when resizing windows fragment program writing to result.depth.z was broken updates from 6.4 branch silence a bunch of compiler warnings don't use -Wmissing-prototypes for C++ code add newline at end of file TNL module needs to implement ctx->Driver.ProgramStringNotify() function Added assert that result != arg0 in do_RSW. This currently fails and needs fixing. include tnl.h to silence warning Keith's fixes for RSW and XPD (need temporaries) If parsing a program fails, don't change the vertex/fragment program object. fix errors in RoughApproxPower(), do_LIT(), _tnl_program_string() fix a few bugs in SWZ execution fix LIT bug s/lenght/length/ clean-up comments added warning about use of REG_SCRATCH Free old parameter list before assigning new one. FIxes memory leak. reduced swizzle comment Use MAKE_SWIZZLE4() macro in a few more places. added some assertions to catch unexpected swizzle values in cvp_emit_arg(), currently fails sometimes need to use absolute value for RSQ Lots of clean-up in arb program parser. Simplify parse_attrib_binding(). check swrast->_FogEnabled instead of ctx->Fog.Enabled fix problems found with gcc 2.96 (bug 4934) Re-org and clean-up of vertx/fragment program limits (instructions, Rename FRAG_OUTPUT_* tokens to FRAG_RESULT_* to match vertex program convention implement USE_MGL_NAMESPACE for x86-64 dispatch code (Mark Bolstad) added osdemos/ dir to DEMO_FILES fix typo, minor clean-ups added a const, clean-up s/15/VERT_RESULT_MAX/ use WRITEMASK_* tokens Streamline code generation by using a fixed size instruction buffer in fix MinGW problems (bug 4956) remove unneeded HIDDEN definition (bug 4956) MESA_GLX_ALPHA_BITS env var to force visual with alpha channel just pass ctx to do_ndc_cliptest() rather than stuffing it in the arb_vp_machine struct better ffs() function (bug 4956) use VERT_RESULT_HPOS, VERT_BIT_POS tokens minor improvements in _mesa_init_teximage_fields() Replace parameter_type enum in program.[ch], use register_file enums instead. don't need MAKE_SWIZZLE() macro added MESA_GLX_ALPHA_BITS env var remove unneeded #include remove unneeded #includes Unify vertex/fragment program instuctions. s/TexSrcIdx/TexSrcTarget/ s/Negate/NegateBase/ added _mesa_num_inst_src_regs() and _mesa_opcode_string() use _mesa_num_inst_src_regs() instead of nr_args() use _mesa_num_inst_src_regs() and _mesa_opcode_string() Use _mesa_opcode_string() added _mesa_program_file_string() use _mesa_program_file_string() and _mesa_opcode_string() consolidate vertex/fragment program printing into _mesa_print_program() s/_mesa_debug_vp_inst/_mesa_print_program/ s/_mesa_debug_fp_inst/_mesa_print_program/ remove unneeded #includes remove unneeded #includes minor work in _mesa_print_program() remove a bunch of extension stuff that's in glxext.h some initial work on upcoming GL_EXT_framebuffer_blit extension remove _mesa_debug_fp_inst() s/_mesa_debug_fp_inst/_mesa_print_program/ Move COPY_FLOAT() macro down into glide driver. use UNCLAMPED_FLOAT_TO_USHORT/UBYTE in clear_rgba_buffer() (bug #4992) Bug 4996. fix popping of GL_EYE_PLANE state (bug 4996) Move code to enable FP exceptions into debug.c (set MESA_DEBUG=fpexceptions). remove -DMESA_DEBUG fix popping of GL_SPOT_DIRECTION (bug 5005) s/MAX_LIGHTS/ctx->Const.MaxLights/ and spruce up some comments Make _mesa_is_proxy_texture() non-static and use in a few more places. Move stuff common to vertex/fragment_program into the base class, including: update some assertions update an assertion Added OSMesaColorClamp(), bug 4917 Added _mesa_light() helper function so we can avoid transforming then fix comment for TexSrcTarget Improve _mesa_print_program(). remove unneeded RasterPosValid check Fix problem in stencil_and_ztest_span() when using h/w stencil buffer. check for ctx==NULL, bug 4087 fix color interpolation problem reported on VMS add GL_INVALID_FRAMEBUFFER_OPERATION_EXT in _mesa_error() Replace GL_POLYGON+1 with PRIM_OUTSIDE_BEGIN_END change an assertion Added error checking in _tnl_Begin to detect incomplete framebuffer. added _mesa_new_depthstencil_renderbuffer() for combined depth/stencil buffers Support for combined depth/stencil renderbuffers (GL_EXT_packed_depth_stencil). added depthstencil.c Use new gl_framebuffer _Depth/_StencilBuffer fields instead of added depthstencil.[ch] additional comments about GL_DEPTH_STENCIL buffers added -ds option to test GL_EXT_packed_depth_stencil with framebuffer objects improved error checking in error_check_format_type() delete frame/renderbuffers upon exit, more error checking optimize case of reading GL_DEPTH_STENCIL pixels from GL_DEPTH_STENCIL renderbuffer s/MEMCPY/_mesa_memcpy/ and fix typo optimize drawing GL_DEPTH_STENCIL pixels into a GL_DEPTH_STENCIL renderbuffer move _mesa_update_state() calls before error tests that depend on framebuffer completeness In _mesa_RenderbufferStorageEXT(), try to avoid needless reallocation. move a bunch of gl2/shading language structs from mtypes.h to shaderobjects.h update comments, remove ^M chars Remove the _mesa_parse_arb_vertex/fragment_program() functions into removed arbfragparse.[ch] arbvertparse.[ch] remove #include arbfragparse.h, s/PI/M_PI/ remove #include arbfragparse.h No longer derive 'ati_fragment_shader' from 'program' class. Only the update some vars in run_texnorm_stage() check if the texture unit is really enabled before trying to normalize the texcoords call gluDeleteQuadric() to fix mem leak Bunch of little fixes: Saturate is a 1-bit boolean field remove incorrect cast additional clean-ups and improvements Make Saturate a 2-bit field again, renamed to SaturateMode with three s/Saturate/SaturateMode/ move _mesa_init_instruction() to program.c disable freeing of fp->func, see comment (bug 5131) reference bug 5131 in comment in _tnl_free_vertices() New _eglLog() function to replace fprintf/printf calls for debug/info. use _eglLog() change error message fixed wrong return value in radeonShowSurfaceMESA() update some comments Choose the display mode / surface size in an intelligent way. check/print EGL_SCREEN_BIT_MESA, fix eglGetDisplay() call init attribs to EGL_DONT_CARE in _eglParseConfigAttribs() change error test in _eglParseConfigAttribs() report error token in message printed in _eglError() Move all the EGL API function pointers into a new _egl_api struct. use _egl_api struct, make some functions static use new _egl_api struct use ADD_POINTERS macro instead of (uintptr_t) cast include stdint.h for BSD Redo _eglInitSurface() so it can be used with all surface types. Lots of fixes, clean-ups, new comments, etc. rearrange some code to put in more logical order, misc clean-ups check for driver_modes == NULL. Don't need driver_modes for EGL driver apparently need to load colormap after setting the mode to make it work reliably clean up the drmGetMap/drmMap code Put extension flags, string into separate struct. added _EGLExtensions typedef use the files from the drm tree/package remove dri_client/ dir from tarball assorted fixes for server-side direct rendering (bug 5199) fxt1_decode_1() should not be static get depthBits from the current drawbuffer added comment about glTexImage and renderbuffers Added FinishRenderTexture() device driver function to indicate when try removing need for the uintptr_t casts remove uintptr_t cast fix _mesa_ffs() return, test for more platforms (Evgeny Kotsuba) Test sizeof(drm_handle_t) instead of LONG64 when returning handles Fix two failures encountered when running out of memory during XImage Added PCI_CHIP_RV410_5E4B (Adam Kirchhoff) C++ fixes, mostly casts (Stephane Conversy) Updates for GL_EXT_timer_query: remove stray _ from GL[u]int64_EXT do away with 'long long' usage don't need LONGLONG typedefs bump version number 'a' key to toggle animation implement config sorting for eglChooseConfig() lots of updates, mode sorting, etc Added EGL_SCREEN_POSITION_GRANULARITY_MESA query. Some initial per-thread support. rename eglShowSurfaceMESA, add EGL_SCREEN_POSITION_GRANULARITY_MESA updates to match other changes elsewhere Rename eglDRIShowSurfaceMESA. s/eglShowSurfaceMESA/eglShowScreenSurfaceMESA/ silence warning, bug 5318 Removed the GL_SGIX/SGIS_pixel_texture extensions. Same thing can be use GLbitfield instead of GLuint regenerate to pick up glGetQueryObject[u]i64vEXT funcs move va_end() call (Evgeny Kotsuba) minor code movement remove bogus wings3d link special-case Darwin (bug 4509) install GLw headers (bug 5396) check XGetGeometry() return value only call xmesa_set_renderbuffer_funcs() when really necessary s/GLuint/GLbitfield/ added a couple new assertions to help debug XGL problem Free visinfo object (Tilman Sauerbeck) properly free ATI fragment shaders (Tilman Sauerbeck) added wglGetExtensionsStringARB WGL_ARB_extensions_string extension remove the GL_HP_occlusion_test code remove unneeded init of swrast->invalidate_line/triangle New comments, clean-up of fields related to point/line/triangle validation. copypixrate.c - measure the speed of glCopyPixels. Test glTexSubImage and glCopyTexSubImage rate. Don't do swapbuffers if -nodraw specified. Get t0 after setup code. misc updates Test glCopyTexSubImage2D by drawing gears as texture on sides of a box. some initial EGL 1.2 work use fmod() when incrementing CubeRot added more tests (Pedro Maia) silence warning about duplicate M_PI definition updated comment s/malloc/calloc/, fix mem leak fix texcoord unit assertion (bug 5780) add some assertions to check emit_func[] table validity use calloc instead of malloc so try_codegen field is initialized to zero (bug 5791) bring in changes from 6.4 branch fix typo EGL 1.2 tokens/functions/etc. David Reveman's GLX_EXT_texture_from_pixmap extension patch include dix-config.h (Mike Harris) (bug 5835) remove the test for XFree86Server from prev patch (bug 5835) fix AIX issues (bug 5874) replace size_t * w/ unsigned int * (Jurg Billeter) remove stray 'foo' line Fix a few warnings: Set NumInstructions (bug 5864) updated GLUT link silence a variety of compiler warnings add missing prototypes, use _slang_ prefix on non-static functions include proper header file, remove local extern declarations add some #includes, fix some prototypes (bug 5992) update Blender link include inttypes.h if __osf__ defined (Randall Hand) EXT_framebuffer_object (Dave Reveman) Fix _mesa_copy_texture_state() so that we copy texture object bindings, need to check for _NEW_BUFFERS in xmesa_update_state() fix broken _mesa_copy_context() for GL_LIGHTING_BIT display list clean-ups Remove need to initialize the InstSize[] array elements by just storing re-formated with indent. silence a bunch of warnings check for active shader, set FRAGPROG_BIT in raster mask check if active fragment shader for deferred texturing/shading only run shader on active fragments in the span Allow error checking to look for presence of either of two extensions. GL_EXT_framebuffer_blit tokens/functions _swrast_blit_framebuffer() function added s_blit.c remove temporary tokens, more error checks for blit function Added glBlitFramebufferEXT function. updated comment added Flush in glXCopyContext() New comments. Use tuples () instead of lists [] in a few places. added BlitFramebuffer() support No need for _mesa_save_error() to be public. implementation of _swrast_BlitFramebuffer() plug in BlitFramebuffer function check FEATURE_EXT_framebuffer_blit improved error msg test glBlitFramebufferEXT w/ -blit flag added GL_EXT_framebuffer_blit, etc FreeBSD updates (Pedro Giffuni) call _mesa_update_framebuffer_visual() in _mesa_FramebufferRenderbufferEXT() to fix some minor glitches print newline in _mesa_warning() disable double-buffer test in check_compatible(), fixes pbuffer/visual problem just a comment for check_compatible() Instead of adding extra flags to DEFINES with +=, define EXTRA_DEFINES instead. s/GLint/int/, to fit GLX conventions Add prototypes for GLX_MESA_swap_frame_usage, GLX_MESA_swap_control, and replace dprintf() with _glu_dprintf() (bug 6244) Set always_array=true for glGenProgramsNV (Dave Reveman) init Red/Green/EtcBits = 0 in soft_renderbuffer_storage() s/GLuint/GLushort/ in read_depth_stencil_image() use 'p' key to toggle use of GL_EXT_packed_depth_stencil rename enums, fixes compilation breakage (Michal Krol) sync with texmem branch updated some printfs, added comment about sched_yield Need to include glthread.h in glapi.h, not glapi.c so that GET_CURRENT_CONTEXT remove invalid assertion in _glapi_get_dispatch(), fixes additional thread safety problem assorted code clean-ups, comments, etc. finish up some loose ends in _mesa_texstore_z24_s8() Moved _glapi_check_multithread() call into drivers, instead of in added #undef GET_DISPATCH remove another test for IN_DRI_DRIVER Init FogCoordFuncs[0..5] too, remove some unneeded #includes Don't call Driver->Bitmap() if no image to render Queries of GL_READ_BUFFER and GL_DRAW_BUFFER were still returning the minor clean-ups slightly improved error message fix some buffer binding tests in _mesa_make_current() added _mesa_source/dest_buffer_exists() function to easily test if particular renderbuffers are present use new _mesa_sourc/dest_buffer_exists() functions additional error checking to be sure source renderbuffer exists for glCopyTex[Sub]Image() additional assertions and debug code for testing FBO completeness Lots of changes/fixes for rendering to framebuffer objects. GL_EXT_packed_depth_stencil should be off by default Need to check for new state and do readpixels/drawpixels clipping after Added -p option to create a display connection per thread. Added -l option for explicit locking around Xlib and GLX calls in added some missing formats in _mesa_source/dest_buffer_exists() lots of debug options, some code movement, etc Clean up the recent code related to texture_override(). Pass params in a check for incomplete FBO in _mesa_Clear() fix issue in _mesa_free_framebuffer_data() remove carriage returns fix some PFNGL* typedefs merge from texman branch Dave Airlie (39): Add color tiling support to miniglx for radeon fix miniglx setting virtual size of 832 for 800x600 fix up color tiling in miniglx radeon server code fix up libdrm for egl complete fix for miniglx startup issue since driver version checks Check for ATI_fragment_shader enabled fix dri drivers build add 1002:5460 M22 X300 card fix breakage from Brians changes earlier add link for radeon_egl.c Add support for texture compression to R300 driver fixes for r300_tex.c from Aapo Tahkola add the tile bits for offset, and unk bit 31 of size fix up radeon span functions using latest r200 code from Brian, fix missing commit for span changes get rid of some of those unknowns figured out from reg dumper bit 31 of this inserts a nop after the current instruction Fix r300 rectangular texture upload and swtcl coordinate fixing same as radeon update r300 drm minimum to 20 fix from Roland for mipmapped compressed textures fixup driver for some api changes this needs to be calloced update to new API add some r300 information use card_type not isPCI add some card type ioctl support typo rv370 not rv350 add EGL to r300 makefile add support for PCIE card to EGL, doesn't fully work yet but these changes fixup stamp so driver updates drawable info fill out framebuffer egl stuff enable egl on R300 add vblank support to i915 driver commit to fixup MGL namespacing for XGL coverity checker bug: Pointer dereferenced after NULL check coverity check bug fix going over end of array fix coverity 926: use before NULL check fix coverity 1107: use after free make solo build complete Eric Anholt (48): Forced commit to note repocopy from r300.sf.net CVS. Add some missing defines, copied from linux-dri. Connect the r300 bits up to the build. Clean up warnings in r300 code by making some symbols static, adding prototypes Since this isn't a bash script, don't use function and instead inline the usage Dike out the linux-specific EGL bits unless it's a linux build. Attempt to fix the issue reported by Alan Grimes with DRM_RADEON_TEXTURE Fix libGL build on FreeBSD by respecting ASM_FLAGS. Attempt to fix libGL on FreeBSD, where the library was being built without any Fix a missed merge for idr's dispatch changes: failure to do driInitExtensions Bug #4615: Fix the SiS driver for the renderbuffer changes. Previously, all Fix a missed merge for idr's dispatch changes: failure to do driInitExtensions Remove some long-dead code. Make SiS a little more like other drivers: Add SIS_DEBUG (only option being Bump driver dates for the crasher fixes yesterday. Fix FreeBSD build by building libGL in the order desired, and doing a fix to my Add support for GL_EXT_stencil_wrap by guessing that the skip of two values in Fix a typo in a fallback string. Add support for GL_ARB_texture_border_clamp and GL_ARB_texture_mirrored_repeat, Fix texenv by moving a misplaced check for being in fallback (should cover only Fix the GL_BLEND case (and remove the fallback associated with it) by converting - Add a few more hardware internal formats that are supported. Some remaining Since we memcpy texture data synchronously to the card, idle the engine before Turn off the AGP dispatch path yet again, as it seems to be broken (found with Add support for GL_EXT_secondary_color, tested with seccolor test. Fix and enable GL_MESA_ycbcr_texture. Looks fine with yuvrect. I'm slightly In the last commit I added in a fatal error on unknown base formats so I Cut a bunch of code by not trying to precompute the blit commands and instead Initial add of some (disabled) SiS 6326 drawing code integrated from Alan Cox's Move the front.offset setting above a check to shortcut reallocating buffers if Add disabled support for GL_EXT_fog_coord. While it seems correct to me, it's Add a fog mode (normal and fog_coord, leaner/exp/exp2 mode) testing program I Add built programs to .cvsignore. Add support for GL_EXT_fog_coord, tested using tests/fog.c. (Also, matches the Fix the RGB order of the specular color emit, and turn GL_EXT_secondary_color Bump driver date for recent extension adds. Add support for HW stencil spans, mirroring the depth spans support. Used for Bug #4900: Fix the non-fogcoord fog test on r100 (and rv200) at tcl_mode=1 by Bug #1028: Add hardware-accelerated stencil support to r128. Testing with Add support for GL_EXT_stencil_wrap, from hearing that the windows drivers did Correct the ordering of the blend constant color. GC a bit of dead code I noticed while looking at texwrap. Bug #4901: Correct secondary color emission when lighting is turned off, tested First step of Radeon DRI unification: Add support for GL_EXT_secondary_color. It looks like using Regenerate for recent glapi changes (GetQueryObjecti64vEXT). Add the ability to define NO_LIBCWRAPPER to avoid libcwrapper brain damage. Include dix-config.h if HAVE_DIX_CONFIG_H is defined. This will be used by the Ian Romanick (64): Wrap every place that accesses a dispatch table with a macro. A new script- Missed this on the previous commit. All elements of pre-DRI_NEW_INTERFACE_ONLY are removed. This allows Gut a few more dead bits. Replace uses of the CreateNewScreenFunc typedef Fixes the glXGetProcAddress portion of the interface. Most of the functions Quote $(CC) and $(CXX) so that 'CC=ccache gcc' and 'CXX=ccache g++' will No, really, get rid of all the remaining references to DRI_USE_NEW_INTERFACE. Remove the last remnants of GLX_BUILT_IN_XMESA. This allows the removal of Fix SPARC assembly for 64-bit Major rip-up of internal function insertion interface. The old The dri/*/depend files don't exist on a raw check-out, and that makes 'make Remove '#if 0' block that was accidentally left in. glxEnableExtension is a platform-specific function exported by the loader. Use the routines supplied in the __DRIinterfaceMethods structure to create a Add a new config called linux-indirect. This build the GLX libGL just like Remove unused structures and data fields. These had been left in because Fix build problems related to finding Xthreads.h. Convert all instances of XTHREADS to USE_XTHREADS. This fixes one of Fix recent problems with display lists and other parts of the code. In order to create the set of available extensions, we have to know if Fixes bugzillla #812. Fixes bugzilla #1730. Since glcontextmodes.[ch] are no longer used in the DRI drivers, removed the Make the linux-dri-x86 builds work on x86-64 again. mklib now Add rules and target to build src/mesa/x86-64/glapi_x86-64.S. Mesa uses the glVertexAttrib*NV functions to implement fixed-function Remove _glapi_check_multithread from the interface exported by the loader to Make ffb driver build on x86-64. The fix is two fold. First, use the Since Brian's recent commit, "-I/usr/X11R6/include/X11/extensions" is no Don't link with libGL, libEGL, or libEGLdri. On EGL setups libGL isn't Makefile cleanups suggested by George Fufutos (bugzilla #3899). Remove support for the R200_NO_VTXFMT environment variable. Instead use Remove the logic that determines at compile time whether or not HAVE_ALIAS Enable support for EXT_stencil_wrap. The code was already there, but the Put back '-I/usr/X11R6/include'. Make the i810 driver use driGetRendererString like all the other drivers. Add simple regression test for bugzilla #3195. Don't try to update hardware texture fields if there is no hardware texture. Make sure extension entry point offsets are properly initialized. This is Add support for NV_blend_square. This was tested with Enable ARB_texture_compression. The code has been in place for a long time, Fix a realloc problem with indirect vertex arrays. The actual head pointer Fix texture format selection. ChooseTextureFormat is supposed to select the Implement GL_ARB_texture_env_combine, GL_EXT_texture_env_combine, and Enable GL_ARB_texture_mirrored_repeat on Savage4. Refactor the code that Remove the non-existant --get-alias-set option from the help text. Change the 'count' for COMPRESSED_TEXTURE_FORMATS from "?" to "-1". This Add 'get' data for glGetFramebufferAttachmentParameterivEXT enums. ifdef out some code to make glapi.c linkable in libglx.so (on the Add new test "texfilt". This test is a cousin to texwrap. Instead of This patch adds the missing NV_vertex_program2 and NV_vertex_program3 At keithw's request, remove all traces of the "fragment program to C" and Support for "fragment program to C" and TCC has been removed from the Simple test of glInterleavedArrays functionality. Add the ability for interleave to test passing invalid format enums Two subtle things missed on the previous commit. This is a major re-work of the __indirect_glInterleavedArrays routine. The Fix build issues on x86-64 due to missing include path in Test case for bugzilla #3050. Build test for bugzilla #3050 Don't try to bind to proxy texture target. This is defined by the spec to be Display the correct bug number in the information text. Generate server-side GLX protocol decode functions using the 'make server' Later versions of Python handle formats like '% 5u' differently. For Jeremy Kolb (2): Remove old linux-x86-xcb (didn't do anything). Added xcb dri config file. Jerome Glisse (6): Use X11_INCLUDES config macro instead of Fix swizzling table & add struct. Add individual component negation. Indent & wrap around 80 Do individual negetation in swizzling, thus we can fix using a tmp without asking for one Jon Smirl (11): Fix a == vs !- typo for glx_enable_extension Commit Ian's fixes from Bug 3877 Remove OUTREG(RADEON_AGP_BASE, info->ringHandle); from DRI driver Leave the AGP_BASE code in. It is pointless to bump the interface New files for radeon egl driver. More EGL prep. No impact on anything outside of EGL Remove EGL from linux-solo makefile Enable EGL support in 'linux-dri' config Fix up some of the sample to work with the r200 EGL driver Add missing egldri.h header file Remove platform specific bits from radeon_egl.c Jouk Jansen (9): ---------------------------------------------------------------------- Committing in . Committing in . descrip.mms syntax correction Committing in . Update OpenVMS compile support Committing in . Committing in . Committing in . Karl Schultz (14): fix var declaration so that a C compiler can build it. Windows mesa.def file had been moved to the drivers dir tree some time ago. This copy is not needed. remove _swrast_DrawBuffer Add support for memory contexts (Thomas Kaltofen). Cleanup compilation warnings. Fix bugzilla 5253. Add {} around code in RENDER_SPAN macros when code includes a variable declaration. reapply recent changes with Visual Studio instead of direct text editing. handle 64-bit types for Windows. add 64-bit typedefs for portability use 64-bit typedefs for portability add {} to RENDER_SPAN macro to fix compilation problem update for symbol export changes Work In Progress. Move some of the frame buffer, render buffer, etc init code from the create context path to the make current path, where it is lazily initialized. This didn't buy me very much except to make it look closer to the X11 driver. Fix up some of this same code to correct some errors. fix thinko in last revision. add wglGetExtensionsStringARB Keith Whitwell (39): Decrement the context's refcounts on part-used storage structs on Check for prim/vertex store non-null before messing with refcounts. be consistent in populating NegateBase field Treat NegateBase as a bitfield, not a boolean. Keep NumTemporaries uptodate. use an explicit reserved field, rather than magic number in bound Populate arb_fp_instruction negate field correctly. Add _mesa_exec_malloc() and _mesa_exec_free() for allocating A bunch of little tests which exercise each ARB_fp instruction plus Better looking test. Make test visible better looking tests simplify, remove sgi messages new depth tests standalone tests for vp instructions Make colors match the fp tests Make colors match the fp tests (really) Several fp and vp tweaks: Add a NotifyInputChanges() callback so that drivers (particularly tnl add an arl test More useful ARL test Make sure tnl->_DoVertexFog is kept uptodate. Fixes fog in i915 track state flags which might invalidate parameter lists Use correct enums for program output variables. Fixes Use _mesa_exec_free for fp->func. Correct author attribution of mm.h Remove the many aliases for 'struct mem_block' in mm.h Clean up clipping somewhat Remove VB->LastClipped Call ProgramStringNotify after creating tnl programs. Add guard before calling ctx->Driver.BindProgram Split _MaintainTexEnvProgram into two flags _Maintain and _Use. This Fix typo. properly notify driver on scissor change in MakeCurrent recalculate state on drawable changes add simple test for program local parameters Oops - bad file-completion typo. Fix STENCIL_WRITE_MASK defn for bug 5902 Add position_invarient support Kristian Høgsberg (1): Don't use glx_ansic.h functions (__glXMemset). Michal Krol (32): remove slang_*_gc.h files; remove some test code;