> I believe the problem here is that the hardware revision can only be obtained > by root so this is hard-coded by Marks patch to 0x11 which works for VT1622A > but I'm not sure how much the hwrev is used in the code. It looks like the > main is is to test whether it is 0x10 or greater. Not sure of hte best way of > getting round this: I don't like running stuff as root unless it _really_ > needs to!
OK, some confusion here. This is an issue with the DirectFB unichrome gfxdriver code, not with viafb. DirectFB tries to read a byte from the PCI config that is only available to processes running as root. It does this in trying to get the version of the north bridge chip (not the TV encoder). The problems I had were (a) that this version number was being read as zero and (b) that when I used the correct value for my hardware (0x11), some of the subsequent tests on this value were choosing code that didn't work. Having only one hardware platform and no datasheet, it's not possible for me to say which version thresholds should be used to select the different behaviour in the code. All I can say is that some of the code conditional on version >= 0x10 should not apply to version 0x11. Mark _______________________________________________ directfb-users mailing list [email protected] http://mail.directfb.org/cgi-bin/mailman/listinfo/directfb-users
