[quickben]

Anything that breaks transactional states and atomicity isn't 'fit the taste', it's getting fired and we'll get somebody that can do those properly.

If anything, that post highlights the lack of quality among the amd driver team, and doesn't have to do much with 'taste'.

[makomk]

If I'm following the conversation correctly, it doesn't break anything, the kernel developers don't understand it. Also, the followup from Alex Deucher of the AMD team is interesting: https://lists.freedesktop.org/archives/dri-devel/2016-Decemb... Basically, he reckons that the atomic modesetting code is a poorly-thought out and maintained disaster that regularly breaks multiple drivers - and having somewhat followed the changelogs, I can entirely believe this. (Dave Airlie responds by blaming AMD for not sufficiently testing the upstream kernel developers' buggy changes to their drivers.)

[quickben]

You are probably right. But so are the kernel devs.

It seems to be an argument of: we do this for all drivers, vs amd pushing: we want to be the exception for this driver only.