[Mesa-dev] glapi dispatch table isn't built correctly with --shared-glapi
This is with git master, but the same issue exists since at least mesa-7.11.2. ./configure --enable-shared-glapi --disable-driglx-direct --with-driver=dri --with-dri-drivers=swrast --with-gallium-drivers=swrast $ gdb --args ./glean -r foo -o -t vertProg1 --quick ... (gdb) b __indirect_glProgramParameters4fvNV Function __indirect_glProgramParameters4fvNV not defined. Make breakpoint pending on future shared library load? (y or [n]) y Breakpoint 1 (__indirect_glProgramParameters4fvNV) pending. (gdb) r Starting program: /opt/jhbuild/git/glean/bin/glean -r foo -o -t vertProg1 --quick Breakpoint 1, __indirect_glProgramParameters4fvNV (target=1, index=135353784, num=20, params=0xb11c) at indirect.c:9306 9306struct glx_context *const gc = __glXGetCurrentContext(); (gdb) bt #0 __indirect_glProgramParameters4fvNV (target=1, index=135353784, num=20, params=0xb11c) at indirect.c:9306 #1 0x080b719b in GLEAN::VertexProgramTest::setup (this=0x8115520) at tvertprog1.cpp:831 #2 0x080b82e3 in GLEAN::VertexProgramTest::runOne (this=0x8115520, r=..., w=...) at tvertprog1.cpp:1112 #3 0x08059885 in GLEAN::BaseTestGLEAN::MultiTestResult::run (this=0x8115520, environment=...) at tbase.h:317 #4 0x08053ec0 in main (argc=7, argv=0xb3d4) at main.cpp:140 (gdb) frame 1 #1 0x080b719b in GLEAN::VertexProgramTest::setup (this=0x8115520) at tvertprog1.cpp:831 831 glGenProgramsARB_func(1, progID); Note that glean is calling glGenPrograms, but we are ending up in glProgramParameters4fvNV. It seems the indirect dispatch table isn't being built correctly. The dispatch table layouts in glapi_map_tmpi.h and glapitable.h are inconsistent when buiding shared glapi, as shared-glapi/glapi_mapi_tmp.h is generated from gl_and_es_API.xml, whereas glapitable.h is generated from gl_API.xml ___ mesa-dev mailing list mesa-dev@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/mesa-dev
Re: [Mesa-dev] glapi dispatch table isn't built correctly with --shared-glapi
On Wed, Feb 1, 2012 at 9:41 AM, Jon TURNEY jon.tur...@dronecode.org.uk wrote: This is with git master, but the same issue exists since at least mesa-7.11.2. ./configure --enable-shared-glapi --disable-driglx-direct --with-driver=dri --with-dri-drivers=swrast --with-gallium-drivers=swrast $ gdb --args ./glean -r foo -o -t vertProg1 --quick ... (gdb) b __indirect_glProgramParameters4fvNV Function __indirect_glProgramParameters4fvNV not defined. Make breakpoint pending on future shared library load? (y or [n]) y Breakpoint 1 (__indirect_glProgramParameters4fvNV) pending. (gdb) r Starting program: /opt/jhbuild/git/glean/bin/glean -r foo -o -t vertProg1 --quick Breakpoint 1, __indirect_glProgramParameters4fvNV (target=1, index=135353784, num=20, params=0xb11c) at indirect.c:9306 9306 struct glx_context *const gc = __glXGetCurrentContext(); (gdb) bt #0 __indirect_glProgramParameters4fvNV (target=1, index=135353784, num=20, params=0xb11c) at indirect.c:9306 #1 0x080b719b in GLEAN::VertexProgramTest::setup (this=0x8115520) at tvertprog1.cpp:831 #2 0x080b82e3 in GLEAN::VertexProgramTest::runOne (this=0x8115520, r=..., w=...) at tvertprog1.cpp:1112 #3 0x08059885 in GLEAN::BaseTestGLEAN::MultiTestResult::run (this=0x8115520, environment=...) at tbase.h:317 #4 0x08053ec0 in main (argc=7, argv=0xb3d4) at main.cpp:140 (gdb) frame 1 #1 0x080b719b in GLEAN::VertexProgramTest::setup (this=0x8115520) at tvertprog1.cpp:831 831 glGenProgramsARB_func(1, progID); Note that glean is calling glGenPrograms, but we are ending up in glProgramParameters4fvNV. It seems the indirect dispatch table isn't being built correctly. The dispatch table layouts in glapi_map_tmpi.h and glapitable.h are inconsistent when buiding shared glapi, as shared-glapi/glapi_mapi_tmp.h is generated from gl_and_es_API.xml, whereas glapitable.h is generated from gl_API.xml Do you think this is related to http://www.mail-archive.com/mesa-dev@lists.freedesktop.org/msg17627.html ? Thanks, Matt ___ mesa-dev mailing list mesa-dev@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/mesa-dev
Re: [Mesa-dev] glapi dispatch table isn't built correctly with --shared-glapi
On 01/02/2012 15:59, Matt Turner wrote: On Wed, Feb 1, 2012 at 9:41 AM, Jon TURNEY wrote: Note that glean is calling glGenPrograms, but we are ending up in glProgramParameters4fvNV. It seems the indirect dispatch table isn't being built correctly. The dispatch table layouts in glapi_map_tmpi.h and glapitable.h are inconsistent when buiding shared glapi, as shared-glapi/glapi_mapi_tmp.h is generated from gl_and_es_API.xml, whereas glapitable.h is generated from gl_API.xml Do you think this is related to http://www.mail-archive.com/mesa-dev@lists.freedesktop.org/msg17627.html ? The part of that thread that says... On 24/01/2012 01:59, Brian Paul wrote: However, any program that uses extension functions acts peculiar and dies: ... I would guess so, yes. Turning --shared-glapi always on just made it more obvious it was broken :S ___ mesa-dev mailing list mesa-dev@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/mesa-dev
