On Fri, 20 Jan 2017 17:45:53 +0100, Pierre-Louis Bossart wrote: > > On 1/20/17 5:15 AM, Takashi Iwai wrote: > > On Fri, 20 Jan 2017 23:22:31 +0100, > > Jerome Anand wrote: > >> > >> + had_ops_v1 = had_ops_v1; /* unused */ > > > > Until now I didn't realize that the whole v1 stuff is never used in > > the current patchset. Will it be ever used in future? If not, can't > > we clean it up? It's a bunch of codes, including the messy union > > definitions. If there is no v3 or whatever, we can even get rid of > > the whole indirect calls. > > > > And if v1 (and the indirect ops calls) should be kept, actually what > > is the difference between v1 and v2, why both implementations do > > exist? Please elaborate in comments. > > v1 refers to Medfield/Clovertrail, v2 to Baytrail/CHT. The differences > are minor and centered on different register definitions or additional > features/bug corrections. We left the v1 code in so far but we could > probably remove it since it's not tested anyway. The question is if we > remove this v1 code and indirect calls now or later, I was planning to > add DP audio support and making more changes would make the > integration more difficult. It's up to you. I prefer the DP implementation sooner so that I can test the driver by myself, which makes me easier working on a later cleanup of the driver code after the merge. My previous question was about the general question, to determine how far we can reduce things. thanks, Takashi _______________________________________________ Intel-gfx mailing list Intel-gfx@xxxxxxxxxxxxxxxxxxxxx https://lists.freedesktop.org/mailman/listinfo/intel-gfx
