> > > > to be
> > > added.
> > > > A lot of weird things going on here.
> > > >
> > > > 1st one is how is it finding a payload which we do not create
> > > > while we call destroy function
> > > >
> > > > 2nd how is VCPI with id 0 possible from what I see VCPI are 1 at
> > > > least that's what I gather from
> > > > drm_dp_mst_atomic_check_payload_alloc_limits.So what are we
> > > > missing
> > > when we create a payload?
> > > >
> > > > Imre, Jani any idea still new to how payload creation work so am I
> > > > missing something.
> > >
> > > A VCPI ID will be assigned to a payload during an atomic commit only
> > > if the corresponding MST connector is still connected. If the MST
> > > connector gets disconnected by the time of the atomic commit - as in
> > > the above case - no VCPI ID will assigned and the allocation table
> > > in the branch device cannot be updated either for the payload, as
> > > indicated by the above payload creation/removal failed messages.
> > >
> > > I think the fix should be not to clear the VCPI ID if it's 0. Valid
> > > VCPI IDs start from 1.
> >
> > Hmm then in that case should we just return 0 early if vcpi turns out to be > > 0
> here.
> 
> The payload should be still deleted, so only the clearing of VCPI ID from
> payload_mask needs to be avoided if the ID is 0.

Ohkay got it will send the next revision to reflect that

Regards,
Suraj Kandpal

> 
> > Regards,
> > Suraj Kandpal
> >
> > > > Regards
> > > > Suraj Kandpal
> > > >
> > > > > BR,
> > > > > Jani.
> > > > >
> > > > >
> > > > > >
> > > > > >>    }
> > > > > >>
> > > > > >>    return 0;
> > > > > >> --
> > > > > >> 2.34.1
> > > > > >>
> > > > >
> > > > > --
> > > > > Jani Nikula, Intel

Reply via email to