Roland Scheidegger wrote: > Dennis Schridde wrote: >> Am Sonntag, 27. August 2006 16:56 schrieben Sie: >>> Dennis Schridde wrote: >>>> Hi! >>>> >>>> One of our (Warzone 2100, http://wz.rootzilla.de/) users told us about an >>>> assert that only our game seems to trigger. >>>> >>>> --- >>>> warzone2100: r200_vtxfmt.c:1044: r200VtxFmtFlushVertices: Assertion >>>> `rmesa->dma.flush == 0 || rmesa->dma.flush == flush_prims' failed. >>>> --- >>>> Details and backtrace on >>>> https://gna.org/bugs/?func=detailitem&item_id=6758 >>>> >>>> But as this assert is in the radeon drivers code I have no idea how we >>>> trigger it. Can someone please give me an idea what kinds of events could >>>> trigger that assert? >>> I'm not sure, but it seems somewhat likely it may be related to the >>> _mesa_compile_error shown in the backtracke. This might be an >>> application bug, if it's trying to do something illegal with display >>> lists (but, obviously, it should not cause the driver to abort). From a >>> quick look, that compile error might happen if glLoadIdentity gets >>> called inside a primitive. >> Does glNewList(dlist, GL_COMPILE_AND_EXECUTE); count as "inside a primitive"? >> If yes that this is obviously a bug in our code. > Actually, that "inside a primitive" should have read "inside a begin/end > pair", that it would be inside a display list thus just more or less > coincidential (but obviously mesa wouldn't call _mesa_compile_error if > it wouldn't be inside a display list, but call another error function > instead). But as said, it could be some other problem, debugging symbols > in the back trace really would help. If it is an app bug though then > you should get an opengl error there with any opengl driver, so you > could check for errors there. I can reproduce this here. If the bogus unconditional glEndList is removed (there is a conditional one after all right after that), then the display list code appears to be correct, and mesa (with debug option) will no longer complain about user errors. However, the same assertion will still fail. Since there are other reported cases of this assertion failing, looks like it's quite a commonly hit r200 driver bug (and possibly radeon too, as the vtxfmt code is very similar).
Roland ------------------------------------------------------------------------- Using Tomcat but need to do more? Need to support web services, security? Get stuff done quickly with pre-integrated technology to make your job easier Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642 -- _______________________________________________ Dri-devel mailing list Dri-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/dri-devel
