On Wed, Jan 17, 2024 at 12:12:18PM +0200, Lisovskiy, Stanislav wrote: > On Fri, Jan 12, 2024 at 07:35:24PM +0200, Ville Syrjälä wrote: > > On Tue, Nov 28, 2023 at 10:37:53AM +0200, Stanislav Lisovskiy wrote: > > > We need that in order to force disable SAGV in next patch. > > > Also it is beneficial to separate that code, as in majority cases, > > > when SAGV is enabled, we don't even need those calculations. > > > Also we probably need to determine max PSF GV point as well, however > > > currently we don't do that when we disable SAGV, which might be > > > actually causing some issues in that case. > > > > > > Signed-off-by: Stanislav Lisovskiy <[email protected]> > > > --- > > > drivers/gpu/drm/i915/display/intel_bw.c | 82 ++++++++++++++++++++----- > > > 1 file changed, 65 insertions(+), 17 deletions(-) > > > > > > diff --git a/drivers/gpu/drm/i915/display/intel_bw.c > > > b/drivers/gpu/drm/i915/display/intel_bw.c > > > index 583cd2ebdf89..efd408e96e8a 100644 > > > --- a/drivers/gpu/drm/i915/display/intel_bw.c > > > +++ b/drivers/gpu/drm/i915/display/intel_bw.c > > > @@ -805,6 +805,64 @@ intel_atomic_get_bw_state(struct intel_atomic_state > > > *state) > > > return to_intel_bw_state(bw_state); > > > } > > > > > > +static unsigned int icl_max_bw_qgv_point(struct drm_i915_private *i915, > > > + int num_active_planes) > > > +{ > > > + unsigned int max_bw_point = 0; > > > + unsigned int max_bw = 0; > > > + unsigned int num_qgv_points = i915->display.bw.max[0].num_qgv_points; > > > + int i; > > > + > > > + for (i = 0; i < num_qgv_points; i++) { > > > + unsigned int idx; > > > + unsigned int max_data_rate; > > > + > > > + if (DISPLAY_VER(i915) > 11) > > > + idx = tgl_max_bw_index(i915, num_active_planes, i); > > > + else > > > + idx = icl_max_bw_index(i915, num_active_planes, i); > > > + > > > + if (idx >= ARRAY_SIZE(i915->display.bw.max)) > > > + continue; > > > + > > > + max_data_rate = i915->display.bw.max[idx].deratedbw[i]; > > > > Looks like that that part could be extracted to a helper > > to be used by both codepaths (would be a natural counterpart > > to adl_psf_bw()). > > > > > + > > > + /* > > > + * We need to know which qgv point gives us > > > + * maximum bandwidth in order to disable SAGV > > > + * if we find that we exceed SAGV block time > > > + * with watermarks. By that moment we already > > > + * have those, as it is calculated earlier in > > > + * intel_atomic_check, > > > + */ > > > + if (max_data_rate > max_bw) { > > > + max_bw_point = i; > > > + max_bw = max_data_rate; > > > + } > > > + } > > > + > > > + return max_bw_point; > > > +} > > > + > > > +unsigned int icl_max_bw_psf_gv_point(struct drm_i915_private *i915) > > > +{ > > > + unsigned int num_psf_gv_points = > > > i915->display.bw.max[0].num_psf_gv_points; > > > + unsigned int max_bw = 0; > > > + unsigned int max_bw_point = 0; > > > + int i; > > > + > > > + for (i = 0; i < num_psf_gv_points; i++) { > > > + unsigned int max_data_rate = adl_psf_bw(i915, i); > > > + > > > + if (max_data_rate > max_bw) { > > > + max_bw_point = i; > > > + max_bw = max_data_rate; > > > + } > > > + } > > > + > > > + return max_bw_point; > > > +} > > > + > > > static int mtl_find_qgv_points(struct drm_i915_private *i915, > > > unsigned int data_rate, > > > unsigned int num_active_planes, > > > @@ -882,8 +940,6 @@ static int icl_find_qgv_points(struct > > > drm_i915_private *i915, > > > const struct intel_bw_state *old_bw_state, > > > struct intel_bw_state *new_bw_state) > > > { > > > - unsigned int max_bw_point = 0; > > > - unsigned int max_bw = 0; > > > unsigned int num_psf_gv_points = > > > i915->display.bw.max[0].num_psf_gv_points; > > > unsigned int num_qgv_points = i915->display.bw.max[0].num_qgv_points; > > > u16 psf_points = 0; > > > @@ -909,18 +965,6 @@ static int icl_find_qgv_points(struct > > > drm_i915_private *i915, > > > > > > max_data_rate = i915->display.bw.max[idx].deratedbw[i]; > > > > > > - /* > > > - * We need to know which qgv point gives us > > > - * maximum bandwidth in order to disable SAGV > > > - * if we find that we exceed SAGV block time > > > - * with watermarks. By that moment we already > > > - * have those, as it is calculated earlier in > > > - * intel_atomic_check, > > > - */ > > > - if (max_data_rate > max_bw) { > > > - max_bw_point = i; > > > - max_bw = max_data_rate; > > > - } > > > if (max_data_rate >= data_rate) > > > qgv_points |= BIT(i); > > > > > > @@ -964,9 +1008,13 @@ static int icl_find_qgv_points(struct > > > drm_i915_private *i915, > > > * cause. > > > */ > > > if (!intel_can_enable_sagv(i915, new_bw_state)) { > > > - qgv_points = BIT(max_bw_point); > > > - drm_dbg_kms(&i915->drm, "No SAGV, using single QGV point %d\n", > > > - max_bw_point); > > > + unsigned int max_bw_qgv_point = icl_max_bw_qgv_point(i915, > > > num_active_planes); > > > + unsigned int max_bw_psf_gv_point = > > > icl_max_bw_psf_gv_point(i915); > > > + > > > + qgv_points = BIT(max_bw_qgv_point); > > > + psf_points = BIT(max_bw_psf_gv_point); > > > > We didn't restrict the PSF here previously. > > Yep, but I really suspect we should. BSpec states that we should restrict all > the GV points > except highest one, also that some PSF GV points aren't same or usable, > depending on BW reqs. > So I would restrict that as well, in case if SAGV is off, just to be on safe > side.
Pretty sure it's explicitly noted that PSF doesn't cause issues with latency and hence doesn't need this. In any case, a change like this has no business being in a patch that's just supposed to refactor code. -- Ville Syrjälä Intel
