On Wed, 20 Apr 2011 11:43:38 -0700, Keith Packard <[email protected]> wrote: > On Wed, 20 Apr 2011 10:22:22 +0100, Chris Wilson <[email protected]> > wrote: > > > For bonus amusement value, we perform the first load detect before even > > establishing our fbdev. > > It seems like we need to be able to perform load detection in this > case to create a suitable frame buffer at boot time.
Aye, a classic chicken and egg problem. I wanted to ignore it as everything would have been resized upon the first probe by userspace (hotplug polling being disabled for load-detect paths), but only after the probe completes so there is the issue that the fb may be too small and so we don't prevent all potential PGTBL_ER. And VGA is laso likely to be a boot device. I'll spin up a patch for a temporary buffer and see what you think. -Chris -- Chris Wilson, Intel Open Source Technology Centre _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
