> After a bit of fiddling I found that the hwrev was not getting set > properly on this board, and it turns out that in driver_init_driver, in > unichrome.c, uc_probe_pci is only called if the hwregs are unable to be > memory mapped, which I believe is incorrect - it should be called > regardless. The following patch resolves this (applied from within > gfxdrivers/unichrome) - it simply moves the call to uc_probe_pci outside > of the "if (!hwregs)" statement.
If you search back a bit on this list (or directfb-users), you should find that patch included in one of my emails. You're absolutely right the code is wrong. However, it still doesn't do the right thing as it will only discover the correct hardware version if the process is running as root: non-superuser processes can't read the necessary byte. A discussion here some time ago concluded that the best way of dealing with this would be to add a utility to read the revision number and then have it as an option in directfbrc. Mark _______________________________________________ directfb-dev mailing list [email protected] http://mail.directfb.org/cgi-bin/mailman/listinfo/directfb-dev
