On 05/29/2013 01:20 PM, Brendan Conoboy wrote: > On 05/25/2013 10:15 PM, Brendan Conoboy wrote: >> o No graphics - regression since F18 (Kernel 3.6) > > Jon is looking into this for Panda only. He should have an update later > today. So I'm just using a blunt instrument one-liner to request_module the tfp4 panel driver on module init. This is ugly, and needs testing, etc. It'll never go upstream. We can perhaps get away with a synchronous request_module call during module init because that happens in the context of modprobe (kmod) and so we probably won't deadlock stuff, but the right thing in general is probably to DECLARE_WORK a worker that can be scheduled in the kernel at probe time based on what chips are actually present, e.g. after parsing the devicetree for that data. This is just stream-of-consciousness thinking here, I'll think more about it. I'll be ordering some new Panda ES boards for myself and Peter, since the EB3 and other engineering builds seem to generate a backtrace on boot. It's not worth fixing that if production ES boards are working. Jon. _______________________________________________ arm mailing list arm@xxxxxxxxxxxxxxxxxxxxxxx https://admin.fedoraproject.org/mailman/listinfo/arm
