On Wed, Aug 17, 2016 at 12:14:53PM +0200, Daniel Vetter wrote:
On Wed, Aug 17, 2016 at 10:26:44AM +0100, Mark Brown wrote:
caused by d25bcfb8c2 (drm/hisilicon: Don't set drm_device->platformdev) which does a a partial conversion to new APIs.
Yeah I screwed up since I naively assumed that compile testing on arm 32bit multiarch is good enough. I'm setting up an arm64 environment now, but would be nice if arm folks would stop loving bazillions of configs so much ;-)
There's only two you really need to care about for this sort of thing, arm allmodconfig and arm64 allmodconfig (the equivalent of i386 and x86-64) - there are some things that don't fit into either but they're all legacy systems, those in active use are pretty much covered by those two.
It's also worth adding || COMPILE_TEST dependencies onto architecture lists to improve coverage if they don't have any actual architecture dependency (which is the norm these days, at least for ARM), the dependencies are useful for normal users so they're not asked about drivers which clearly they can't use.