On Tue, Feb 01, 2022 at 12:05:13PM +0200, Ville Syrjälä wrote: > On Tue, Feb 01, 2022 at 10:52:39AM +0200, Lisovskiy, Stanislav wrote: > > On Tue, Jan 18, 2022 at 11:23:50AM +0200, Ville Syrjala wrote: > > > From: Ville Syrjälä <ville.syrjala@xxxxxxxxxxxxxxx> > > > > > > intel_bw_calc_min_cdclk() is entirely pointless. All it manages to do is > > > somehow conflate the per-pipe min cdclk with dbuf min cdclk. There is no > > > (at least documented) dbuf min cdclk limit on pre-skl so let's just get > > > rid of all this confusion. > > > > > > Signed-off-by: Ville Syrjälä <ville.syrjala@xxxxxxxxxxxxxxx> > > > > I think we constantly have a bit contradictional attitude towards such situation. > > >From one perspective you can say, that those kind of "leagcy" callbacks are > > pointless, from the other hand one might say. that we need to have a unified > > approach for all platforms and I think we got, some legacy handlers for old > > platforms for similar purpose as well. > > I'm fine with both approaches, however for example when I was submitting > > that patch, I was asked by reviewer to add this kind of legacy callback, so that we have > > a "uniform" approach. > > We just then need to have some standard agreement on those, which doesn't > > depend on today's cosmic radiation levels :) > > Yes in general I prefer a unified approach across all platforms. > But in this case there is nothing to do for the old platforms as they > don't have any kind of dbuf cdclk limit, or if there is one we don't > know what it is since it's not documented. > > So the only thing the code was really doing was conflating the > per-pipe cdclk limit (which is handled elsewhere for all platforms > in a unified fashion) with something that doesn't even exist. > > Also I don't think it was even correct anyway since it was > using the per-pipe cdclk_state->min_cdclk[] already during > intel_cdclk_atomic_check(), but cdclk_state->min_cdclk[] > isn't even computed until intel_modeset_calc_cdclk() which > is called later. So I guess it was basically just computing > the max of the min_cdclk[] for all the pipes for the _old_ > state, not the new state. No, I think actually the idea was that it was first calculating new_bw_state->min_cdclk, based on plane and dbuf bandwidth requirements in intel_atomic_check_cdclk, however then the final decision which cdclk to choose was is done in intel_cdclk.c, which calculated new_cdclk_state->min_cdclk and then we just choose maximum of those. And intel_compute_min_cdclk is the final arbiter: static int intel_compute_min_cdclk(struct intel_cdclk_state *cdclk_state) { struct intel_atomic_state *state = cdclk_state->base.state; struct drm_i915_private *dev_priv = to_i915(state->base.dev); struct intel_bw_state *bw_state = NULL; struct intel_crtc *crtc; struct intel_crtc_state *crtc_state; int min_cdclk, i; enum pipe pipe; for_each_new_intel_crtc_in_state(state, crtc, crtc_state, i) { int ret; min_cdclk = intel_crtc_compute_min_cdclk(crtc_state); if (min_cdclk < 0) return min_cdclk; bw_state = intel_atomic_get_bw_state(state); if (IS_ERR(bw_state)) return PTR_ERR(bw_state); if (cdclk_state->min_cdclk[crtc->pipe] == min_cdclk) continue; cdclk_state->min_cdclk[crtc->pipe] = min_cdclk; ret = intel_atomic_lock_global_state(&cdclk_state->base); if (ret) return ret; } min_cdclk = cdclk_state->force_min_cdclk; for_each_pipe(dev_priv, pipe) { min_cdclk = max(cdclk_state->min_cdclk[pipe], min_cdclk); if (!bw_state) continue; min_cdclk = max(bw_state->min_cdclk, min_cdclk); } return min_cdclk; } Stan > > -- > Ville Syrjälä > Intel