On 12 Jan 2003, Koenraad Heijlen wrote: >> Aha! I'll test this one. > >I did and it solves my problems, I have no fake cursors anymore. > >The fonts look good antialiased on both heads.
Good to hear! >Something I did notice, I think the refreshrate on my second head >changed (I had to re-calibrate the screen's vertical and horizontal >position). Shouldn't be any real noticable change like that >The new driver also doesn't load the mga_hal.o, very easy to spot if you >diff the now log files. Correct. Our X server is built without Matrox hallib support, which is because it is notoriously difficult to determine if someone reporting a bug is using the matrox hallib or not, and since we don't have the source code to hallib, we are unable to fix problems for people using it. Nonetheless, when the hallib support is enabled, it does cause a lot of wasted time trying to troubleshoot problems. So it has been disabled in our code since around RHL 7.1. You can build your own driver though, merely by adding "#undef RENDER" to the top of the mga_storm.c file and if you enable Matrox Hallib, with: #define UseMatroxHal YES You'll get a hallib enabled driver. >Will this change make the cvs tree? I'm not sure. Only XFree86.org core team members have the ability to commit such changes to the repository. My workaround isn't a real fix, but in lieu of someone fixing the problem properly, I certainly believe that faulty code should be disabled. So I hope that they do commit the change. If nobody else whom is more familiar with the mga render accel code takes a stab at fixing the problem properly, I might do so at some point in the future, but for now, a quick and easy workaround beats nothing at all IMHO, at least for driver stability. Code shouldn't be commited to CVS unless it is stable and receives proper code review and testing, so it's kind of surprising that this one got through the QA processes that various people do for so long. It could also be though, that not many things were using render to actively until recently so nobody noticed. Hard to say. At least we have a workaround for now though. ;o) Take care, TTYL -- Mike A. Harris _______________________________________________ XFree86 mailing list [EMAIL PROTECTED] http://XFree86.Org/mailman/listinfo/xfree86
