Re: [PATCH v8 3/3] drm/fence: add out-fences support

[Sean Paul]

On Fri, Nov 11, 2016 at 12:11 PM, Daniel Vetter <daniel@xxxxxxxx> wrote:
> On Fri, Nov 11, 2016 at 11:48:09AM -0500, Sean Paul wrote:
>> On Fri, Nov 11, 2016 at 9:15 AM, Gustavo Padovan <gustavo@xxxxxxxxxxx> wrote:
>> > +static void complete_crtc_signaling(struct drm_device *dev,
>> > +                                    struct drm_atomic_state *state,
>> > +                                    struct drm_out_fence_state *fence_state,
>> > +                                    unsigned int num_fences, int ret)
>> > +{
>> > +       struct drm_crtc *crtc;
>> > +       struct drm_crtc_state *crtc_state;
>> > +       int i;
>> > +
>> > +       if (!ret) {
>>
>> I don't think there's any reason to smash the fd install and clean-up
>> into one function. I think splitting into 2 functions and calling the
>> right one from atomic_ioctl would be better.
>
> Hm, I suggested this because the control flow in one of Gustavo's earlier
> patches look really funny. I guess it could be split up again, but with
> both callers in the current position. tbh I don't care whether it's this
> or that, both are clear improvement over the older version.

I really don't have a strong opinion either. Perhaps meet in the
middle and pass bool install_fds instead of ret (since that's kind of
an anti-pattern)?

Sean


> -Daniel
> --
> Daniel Vetter
> Software Engineer, Intel Corporation
> http://blog.ffwll.ch