Brian Paul wrote: > Jonathan Bastien-Filiatrault wrote: > >> Brian Paul wrote: >> >> >>> Jonathan Bastien-Filiatrault wrote: >>> >>> >>>> [snip] >>>> at around 45-50 fps with a 960 face model in zpass mode. >>>> >>>> PS: Have you noticed that some versions of the newer mesa does not >>>> seem to like mixing plain old vertex arrays with VBOs ? The mesa in >>>> Debian unstable(7.0.1) does not segfault on me with direct rendering >>>> with mixed code. >>>> >>>> >>> Are you saying 7.0.1 works as expected but Mesa from git (or the swz >>> branch) doesn't work? I don't quite follow what you're saying. >>> >>> >> Sorry for the non-sequitur. The latest git and the swz branch tend to >> segfault when mixing VBOs with plain old vertex arrays with DRI. This >> problem is not present with the mesa in Debian. Disabling hardware >> acceleration prevents the segfault. I am using the latest DDX and DRM >> code. The kernel is 2.6.22.9 vanilla. The machine is a Dell D620, AMD64. >> Here are the few last lines from valgrind: >> >> ==21659== Invalid read of size 8 >> ==21659== at 0x806F3C9: _tnl_draw_prims (t_draw.c:131) >> ==21659== by 0x8067E83: vbo_exec_DrawArrays (vbo_exec_array.c:259) >> ==21659== by 0x805C413: neutral_DrawArrays (vtxfmt_tmp.h:328) >> ==21659== by 0x41E9D3: OglGeometry::drawShadowVolume(RenderPass >> const&) (geometry.cpp:315) >> ^^ vertex array draw >> ==21659== by 0x41EBE8: OglGeometry::render(RenderPass const&) >> (geometry.cpp:357) >> ==21659== by 0x41697F: Drawable::draw(RenderPass const&) (drawable.h:32) >> ==21659== by 0x4169AF: DrawableObject::draw(RenderPass const&) >> (drawableobject.h:14) >> ==21659== by 0x416DC3: Node::draw(RenderPass const&) (node.cpp:24) >> ==21659== by 0x417BBA: Scene::render() (scene.cpp:111) >> ==21659== by 0x4138FE: Window::draw() (window.cpp:242) >> ==21659== by 0x41393D: Window::draw_cb() (window.cpp:36) >> ==21659== by 0x4C4639A: processWindowWorkList (glut_event.c:1306) >> ==21659== Address 0x1C5D7558 is not stack'd, malloc'd or (recently) free'd >> >> I am pretty sure I am not passing invalid pointers to mesa. Shall I >> write a test case ? Do you want to have a look at the code ? >> > > It would be ideal if you could provide a small test program and file a > bug report. I'm working on tracking down a couple other VBO bugs in > between other things... > > -Brian Due to my inexperience with VBO, I was forgetting to unbind the buffer at the draw mentionned above, doing this does not trigger the segfault, since glVertexPointer now knows where to look.
Sorry :-( Jonathan ------------------------------------------------------------------------- This SF.net email is sponsored by: Microsoft Defy all challenges. Microsoft(R) Visual Studio 2005. http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ -- _______________________________________________ Dri-devel mailing list Dri-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/dri-devel
