On 31/03/17 00:35, Russell King - ARM Linux wrote: > On Wed, Mar 29, 2017 at 04:15:36PM +0200, Hans Verkuil wrote: >> + cec_notifier_set_phys_addr(hdata->notifier, >> + cec_get_edid_phys_addr(edid)); > > This pattern causes problems - can we have the notifier taking the EDID > please, and stubs in cec-notifier.h to stub that out? > > Maybe something like cec_notifier_set_phys_addr_from_edid(edid) ? Good point. I've added this, and as an extra bonus this allowed me to drop the first cec-edid patch. > > Having converted the tda998x code over to your new notifier, the 0-day > builder reported this tonight: > >>> ERROR: "cec_get_edid_phys_addr" [drivers/gpu/drm/i2c/tda998x.ko] undefined! > > which is caused exactly by this problem. I can add #ifdefs into the > tda998x driver, but as you're already stubbing out > cec_notifier_set_phys_addr() in cec-notifier.h, it would be stupid to > have to resort to #ifdefs in driver code to solve this issue. Will post a new series today. Thanks for pointing this out. A general note: I am considering merging cec-notifier and cec-edid into the CEC module itself. However, I want to get this series in first before I start moving things around. It's been delayed long enough already. Regards, Hans
