On 03/12/2013 00:07, Steven Chamberlain wrote: > Had to `kldload radeonkms` myself - that blanks the screen - so have to run > `startx` without looking.
What do you think we should do about this? For radeon, we can disable modesetting in the kernel. Then when userland loads radeonkms, we'd still have something better than the vesa driver without the VT switching problem. If user wants modesetting, it can be enabled via "modeset" flag. Or if we find that we can't make radeonkms behave properly, we could remove it altogether. Then userland would fallback to vesa. Or we could just leave it be, and assume that if user has started X (and X succeeds) they probably don't care that much if their VT became unusable. After all they can always reboot in single mode if they need a console shell. We could also try making it play nice with VTs. Maybe enabling newcons helps. Also, there's some disabled WIP code in radeonkms which is aimed at this purpose (I haven't tested it). As for intel, it's not even clear which module should be preferred by userland. Should xf86-video-intel load i915.ko or i915kms.ko? Unfortunately we need to pick one or the other. On FreeBSD ports they have these NEW_XORG flags which trigger the selection at build time, but we can't have that :-( -- Robert Millan -- To UNSUBSCRIBE, email to debian-bsd-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/52b244b0.5050...@debian.org
