arm/komeda: Add side_by_side support
Subsystems (4)
v5 cover arm/komeda: Add side_by_side support
Hi: All
Komeda HW (two pipelines) can work on side by side mode, which splits the internal display processing to two halves (LEFT/RIGHT) and handle them by two pipelines separately and simultaneously.
And since one single pipeline only handles the half display frame, so the main engine clock requirement can also be halved.
The data flow of side_by_side as blow:
slave.layer0 ->\ /-> slave.wb_layer -> mem.fb.right_part
... -> slave.compiz ->
slave.layer3 ->/ \-> slave.improcessor->
\ /-> output-link0
master.layer0 ->\ /-> master.improcessor ->\-> output-link1
... -> master.compiz ->master.layer3 ->/ \-> master.wb_layer -> mem.fb.left_part
v3: Rebase
v5: Drop the patch: Expose side_by_side by sysfs/config_id
james qian wang (Arm Technology China) (5):
drm/komeda: Add side by side assembling drm/komeda: Add side by side plane_state split drm/komeda: Build side by side display output pipeline drm/komeda: Add side by side support for writeback drm/komeda: Update writeback signal for side_by_side
.../arm/display/komeda/d71/d71_component.c | 4 + .../gpu/drm/arm/display/komeda/komeda_crtc.c | 54 ++-- .../gpu/drm/arm/display/komeda/komeda_dev.c | 4 + .../gpu/drm/arm/display/komeda/komeda_dev.h | 9 + .../gpu/drm/arm/display/komeda/komeda_kms.h | 8 + .../drm/arm/display/komeda/komeda_pipeline.c | 50 +++- .../drm/arm/display/komeda/komeda_pipeline.h | 39 ++- .../display/komeda/komeda_pipeline_state.c | 277 +++++++++++++++++- .../gpu/drm/arm/display/komeda/komeda_plane.c | 7 +- .../arm/display/komeda/komeda_wb_connector.c | 11 +- 10 files changed, 419 insertions(+), 44 deletions(-)
Signature · 6 lines
2.20.1 _______________________________________________ dri-devel mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/dri-devel
PATCH v5 1/5 drm/komeda: Add side by side assembling
Komeda HW can support side by side, which splits the internal display processing to two single halves (LEFT/RIGHT) and handle them by two pipelines separately.
komeda "side by side" is enabled by DT property: "side_by_side_master", once DT configured side by side, komeda need to verify it with HW's configuration, and assemble it for the further usage.
v3: Correct a typo.
--- .../gpu/drm/arm/display/komeda/komeda_crtc.c | 13 ++++- .../gpu/drm/arm/display/komeda/komeda_dev.c | 4 ++ .../gpu/drm/arm/display/komeda/komeda_dev.h | 9 ++++ .../gpu/drm/arm/display/komeda/komeda_kms.h | 3 ++ .../drm/arm/display/komeda/komeda_pipeline.c | 50 +++++++++++++++++-- .../drm/arm/display/komeda/komeda_pipeline.h | 1 + 6 files changed, 74 insertions(+), 6 deletions(-)
drivers/gpu/drm/arm/display/komeda/komeda_crtc.c +11 -2
@@ -561,21 +561,30 @@ int komeda_kms_setup_crtcs(struct komeda_kms_dev *kms, kms->n_crtcs =0 ;for (i =0 ; i < mdev->n_pipelines; i++) { +/* if sbs, one komeda_dev only can represent one CRTC */ +if (mdev->side_by_side && i != mdev->side_by_side_master) +continue ; + crtc = &kms->crtcs[kms->n_crtcs]; master = mdev->pipelines[i]; crtc->master = master; crtc->slave =komeda_pipeline_get_slave (master); + crtc->side_by_side = mdev->side_by_side;if (crtc->slave)sprintf (str,"pipe-%d" , crtc->slave->id);else sprintf (str,"None" ); -DRM_INFO ("CRTC-%d: master(pipe-%d) slave(%s).\n" , - kms->n_crtcs, master->id, str); +DRM_INFO ("CRTC-%d: master(pipe-%d) slave(%s) sbs(%s).\n" , + kms->n_crtcs, master->id, str, + crtc->side_by_side ?"On" :"Off" ); kms->n_crtcs++; + +if (mdev->side_by_side) +break ; }return 0 ;
drivers/gpu/drm/arm/display/komeda/komeda_dev.c +4 -0
@@ -178,6 +178,10 @@ static int komeda_parse_dt(struct device *dev, struct komeda_dev *mdev) } } +/* this DT node is experimental hence undocumented */ + mdev->side_by_side = !of_property_read_u32 (np,"side_by_side_master" , + &mdev->side_by_side_master); +return ret; }
drivers/gpu/drm/arm/display/komeda/komeda_dev.h +9 -0
@@ -183,6 +183,15 @@ struct komeda_dev {/** @irq: irq number */ int irq; +/** +* @side_by_side: +* +* on sbs the whole display frame will be split to two halves (1:2), +* master pipeline handles the left part, slave for the right part +*/ +bool side_by_side; +/** @side_by_side_master: master pipe id for side by side */ +int side_by_side_master;/** @lock: used to protect dpmode */ struct mutex lock;
drivers/gpu/drm/arm/display/komeda/komeda_kms.h +3 -0
@@ -76,6 +76,9 @@ struct komeda_crtc {*/ struct komeda_pipeline *slave; +/** @side_by_side: if the master and slave works on side by side mode */ +bool side_by_side; +/** @slave_planes: komeda slave planes mask */ u32 slave_planes;
drivers/gpu/drm/arm/display/komeda/komeda_pipeline.c +46 -4
67 lines not on this page · raw message
drivers/gpu/drm/arm/display/komeda/komeda_pipeline.h +1 -0
@@ -237,6 +237,7 @@ struct komeda_layer {* not the source buffer. */ struct komeda_layer *right; +struct komeda_layer *sbs_slave; };struct komeda_layer_state {
Signature · 6 lines
2.20.1 _______________________________________________ dri-devel mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/dri-devel
PATCH v5 2/5 drm/komeda: Add side by side plane_state split
In reply to james qian wang (Arm Technology China)On side by side mode, The full display frame will be split into two parts (Left/Right), and each part will be handled by a single pipeline separately master pipeline for left part, slave for right.
To simplify the usage and implementation, komeda use the following scheme to do the side by side split
1. The planes also have been grouped into two classes:
master-planes and slave-planes.
2. The master plane can display its image on any location of the final/full display frame, komeda will help to split the plane configuration to two parts and fed them into master and slave pipelines.
3. The slave plane only can put its display rect on the right part of the final display frame, and its data is only can be fed into the slave pipeline.
From the perspective of resource usage and assignment:
The master plane can use the resources from the master pipeline and slave pipeline both, but slave plane only can use the slave pipeline resources.
With such scheme, the usage of master planes are same as the none side_by_side mode. user can easily skip the slave planes and no need to consider side_by_side for them.
v4: Address Mihail's review comments.
--- .../drm/arm/display/komeda/komeda_pipeline.h | 33 ++- .../display/komeda/komeda_pipeline_state.c | 188 ++++++++++++++++++ .../gpu/drm/arm/display/komeda/komeda_plane.c | 7 +- 3 files changed, 220 insertions(+), 8 deletions(-)
drivers/gpu/drm/arm/display/komeda/komeda_pipeline.h +26 -7
@@ -521,6 +521,20 @@ komeda_component_pickup_output(struct komeda_component *c, u32 avail_comps)return komeda_pipeline_get_first_component (c->pipeline, avail_inputs); } +static inline const char * +komeda_data_flow_msg (struct komeda_data_flow_cfg *config) +{ +static char str[128 ]; + +snprintf (str,sizeof (str), +"rot: %x src[x/y:%d/%d, w/h:%d/%d] disp[x/y:%d/%d, w/h:%d/%d]" , + config->rot, + config->in_x, config->in_y, config->in_w, config->in_h, + config->out_x, config->out_y, config->out_w, config->out_h); + +return str; +} +struct komeda_plane_state ;struct komeda_crtc_state ;struct komeda_crtc ;
@@ -532,22 +546,27 @@ int komeda_build_layer_data_flow(struct komeda_layer *layer,struct komeda_plane_state *kplane_st, struct komeda_crtc_state *kcrtc_st, struct komeda_data_flow_cfg *dflow); -int komeda_build_wb_data_flow (struct komeda_layer *wb_layer, -struct drm_connector_state *conn_st, -struct komeda_crtc_state *kcrtc_st, -struct komeda_data_flow_cfg *dflow); -int komeda_build_display_data_flow (struct komeda_crtc *kcrtc, -struct komeda_crtc_state *kcrtc_st); -int komeda_build_layer_split_data_flow (struct komeda_layer *left,struct komeda_plane_state *kplane_st,struct komeda_crtc_state *kcrtc_st,struct komeda_data_flow_cfg *dflow); +int komeda_build_layer_sbs_data_flow (struct komeda_layer *layer, +struct komeda_plane_state *kplane_st, +struct komeda_crtc_state *kcrtc_st, +struct komeda_data_flow_cfg *dflow); + +int komeda_build_wb_data_flow (struct komeda_layer *wb_layer, +struct drm_connector_state *conn_st, +struct komeda_crtc_state *kcrtc_st, +struct komeda_data_flow_cfg *dflow);int komeda_build_wb_split_data_flow (struct komeda_layer *wb_layer,struct drm_connector_state *conn_st,struct komeda_crtc_state *kcrtc_st,struct komeda_data_flow_cfg *dflow); +int komeda_build_display_data_flow (struct komeda_crtc *kcrtc, +struct komeda_crtc_state *kcrtc_st); +int komeda_release_unclaimed_resources (struct komeda_pipeline *pipe,struct komeda_crtc_state *kcrtc_st);
drivers/gpu/drm/arm/display/komeda/komeda_pipeline_state.c +188 -0
194 lines not on this page · raw message
drivers/gpu/drm/arm/display/komeda/komeda_plane.c +6 -1
@@ -77,6 +77,7 @@ komeda_plane_atomic_check(struct drm_plane *plane,struct komeda_plane_state *kplane_st =to_kplane_st (state);struct komeda_layer *layer = kplane->layer;struct drm_crtc_state *crtc_st; +struct komeda_crtc *kcrtc;struct komeda_crtc_state *kcrtc_st;struct komeda_data_flow_cfg dflow;int err;
@@ -94,13 +95,17 @@ komeda_plane_atomic_check(struct drm_plane *plane,if (!crtc_st->active)return 0 ; + kcrtc =to_kcrtc (crtc_st->crtc); kcrtc_st =to_kcrtc_st (crtc_st); err =komeda_plane_init_data_flow (state, kcrtc_st, &dflow);if (err)return err; -if (dflow.en_split) +if (kcrtc->side_by_side) + err =komeda_build_layer_sbs_data_flow (layer, + kplane_st, kcrtc_st, &dflow); +else if (dflow.en_split) err =komeda_build_layer_split_data_flow (layer, kplane_st, kcrtc_st, &dflow);else
Signature · 6 lines
2.20.1 _______________________________________________ dri-devel mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/dri-devel
PATCH v5 3/5 drm/komeda: Build side by side display output pipeline
In reply to james qian wang (Arm Technology China)For side by side, the slave pipeline merges to master via image processor
slave-layers -> slave-compiz-> slave-improc-
\
master-layers -> master-compiz -------------> master-improc ->
v3: Rebase.
--- .../arm/display/komeda/d71/d71_component.c | 4 ++ .../gpu/drm/arm/display/komeda/komeda_crtc.c | 18 +++++-- .../drm/arm/display/komeda/komeda_pipeline.h | 1 + .../display/komeda/komeda_pipeline_state.c | 51 ++++++++++++++----- .../arm/display/komeda/komeda_wb_connector.c | 2 +- 5 files changed, 56 insertions(+), 20 deletions(-)
drivers/gpu/drm/arm/display/komeda/d71/d71_component.c +4 -0
@@ -1085,6 +1085,10 @@ static void d71_improc_update(struct komeda_component *c,else if (st->color_format ==DRM_COLOR_FORMAT_YCRCB444 ) ctrl |=IPS_CTRL_YUV ; +/* slave input has been enabled, means side by side */ +if (has_bit (1 , state->active_inputs)) + ctrl |=IPS_CTRL_SBS ; +malidp_write32_mask (reg,BLK_CONTROL , mask, ctrl); }
drivers/gpu/drm/arm/display/komeda/komeda_crtc.c +13 -5
@@ -385,15 +385,23 @@ komeda_crtc_atomic_flush(struct drm_crtc *crtc,komeda_crtc_do_flush (crtc, old); } -/* Returns the minimum frequency of the aclk rate (main engine clock) in Hz */ +/* +* Returns the minimum frequency of the aclk rate (main engine clock) in Hz. +* +* The DPU output can be split into two halves, to stay within the bandwidth +* capabilities of the external link (dual-link mode). +* In these cases, each output link runs at half the pixel clock rate of the +* combined display, and has half the number of pixels. +* Beside split the output, the DPU internal pixel processing also can be split +* into two halves (LEFT/RIGHT) and handles by two pipelines simultaneously. +* So if side by side, the pipeline (main engine clock) also can run at half +* the clock rate of the combined display. +*/ static unsigned long komeda_calc_min_aclk_rate (struct komeda_crtc *kcrtc,unsigned long pxlclk) { -/* Once dual-link one display pipeline drives two display outputs, -* the aclk needs run on the double rate of pxlclk -*/ -if (kcrtc->master->dual_link) +if (kcrtc->master->dual_link && !kcrtc->side_by_side)return pxlclk *2 ;else return pxlclk;
drivers/gpu/drm/arm/display/komeda/komeda_pipeline.h +1 -0
@@ -540,6 +540,7 @@ struct komeda_crtc_state;struct komeda_crtc ;void pipeline_composition_size (struct komeda_crtc_state *kcrtc_st, +bool side_by_side,u16 *hsize,u16 *vsize); int komeda_build_layer_data_flow(struct komeda_layer *layer,
drivers/gpu/drm/arm/display/komeda/komeda_pipeline_state.c +37 -14
136 lines not on this page · raw message
drivers/gpu/drm/arm/display/komeda/komeda_wb_connector.c +1 -1
@@ -21,7 +21,7 @@ komeda_wb_init_data_flow(struct komeda_layer *wb_layer, dflow->out_h = fb->height;/* the write back data comes from the compiz */ -pipeline_composition_size (kcrtc_st, &dflow->in_w, &dflow->in_h); +pipeline_composition_size (kcrtc_st, false, &dflow->in_w, &dflow->in_h); dflow->input.component = &wb_layer->base.pipeline->compiz->base;/* compiz doesn't output alpha */ dflow->pixel_blend_mode =DRM_MODE_BLEND_PIXEL_NONE ;
Signature · 6 lines
2.20.1 _______________________________________________ dri-devel mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/dri-devel
PATCH v5 4/5 drm/komeda: Add side by side support for writeback
In reply to james qian wang (Arm Technology China)In side by side mode, the master pipeline writeback the left frame and the slave writeback the right part, the data flow as below:
slave.compiz -> slave.wb_layer -> fb (right-part) master.compiz -> master.wb_layer -> fb (left-part)
--- .../drm/arm/display/komeda/komeda_pipeline.h | 4 ++ .../display/komeda/komeda_pipeline_state.c | 42 +++++++++++++++++++ .../arm/display/komeda/komeda_wb_connector.c | 6 ++- 3 files changed, 51 insertions(+), 1 deletion(-)
drivers/gpu/drm/arm/display/komeda/komeda_pipeline.h +4 -0
@@ -564,6 +564,10 @@ int komeda_build_wb_split_data_flow(struct komeda_layer *wb_layer,struct drm_connector_state *conn_st, struct komeda_crtc_state *kcrtc_st, struct komeda_data_flow_cfg *dflow); +int komeda_build_wb_sbs_data_flow (struct komeda_crtc *kcrtc, +struct drm_connector_state *conn_st, +struct komeda_crtc_state *kcrtc_st, +struct komeda_data_flow_cfg *wb_dflow);int komeda_build_display_data_flow (struct komeda_crtc *kcrtc,struct komeda_crtc_state *kcrtc_st);
drivers/gpu/drm/arm/display/komeda/komeda_pipeline_state.c +42 -0
48 lines not on this page · raw message
drivers/gpu/drm/arm/display/komeda/komeda_wb_connector.c +5 -1
@@ -37,6 +37,7 @@ komeda_wb_encoder_atomic_check(struct drm_encoder *encoder,struct drm_crtc_state *crtc_st, struct drm_connector_state *conn_st) { + struct komeda_crtc *kcrtc = to_kcrtc(crtc_st->crtc);struct komeda_crtc_state *kcrtc_st =to_kcrtc_st (crtc_st);struct drm_writeback_job *writeback_job = conn_st->writeback_job;struct komeda_layer *wb_layer;
@@ -65,7 +66,10 @@ komeda_wb_encoder_atomic_check(struct drm_encoder *encoder,if (err)return err; -if (dflow.en_split) +if (kcrtc->side_by_side) + err =komeda_build_wb_sbs_data_flow (kcrtc, + conn_st, kcrtc_st, &dflow); +else if (dflow.en_split) err =komeda_build_wb_split_data_flow (wb_layer, conn_st, kcrtc_st, &dflow);else
Signature · 6 lines
2.20.1 _______________________________________________ dri-devel mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/dri-devel
PATCH v5 5/5 drm/komeda: Update writeback signal for side_by_side
In reply to james qian wang (Arm Technology China)In side by side mode, a writeback job is completed by two pipelines: left by master and right by slave, we need to wait both pipeline finished (EOW), then can signal the writeback job completion.
--- .../gpu/drm/arm/display/komeda/komeda_crtc.c | 23 ++++++++++--------- .../gpu/drm/arm/display/komeda/komeda_kms.h | 5 ++++ .../arm/display/komeda/komeda_wb_connector.c | 3 +++ 3 files changed, 20 insertions(+), 11 deletions(-)
drivers/gpu/drm/arm/display/komeda/komeda_crtc.c +12 -11
@@ -193,27 +193,28 @@ komeda_crtc_unprepare(struct komeda_crtc *kcrtc)return err; } -void komeda_crtc_handle_event (struct komeda_crtc *kcrtc, +void komeda_crtc_handle_event (struct komeda_crtc *kcrtc,struct komeda_events *evts) {struct drm_crtc *crtc = &kcrtc->base; +struct komeda_wb_connector *wb_conn = kcrtc->wb_conn;u32 events = evts->pipes[kcrtc->master->id];if (events &KOMEDA_EVENT_VSYNC )drm_crtc_handle_vblank (crtc); -if (events &KOMEDA_EVENT_EOW ) { -struct komeda_wb_connector *wb_conn = kcrtc->wb_conn; +/* handles writeback event */ +if (events &KOMEDA_EVENT_EOW ) + wb_conn->complete_pipes |=BIT (kcrtc->master->id); -if (wb_conn) -drm_writeback_signal_completion (&wb_conn->base,0 ); -else -DRM_WARN ("CRTC[%d]: EOW happen but no wb_connector.\n" , -drm_crtc_index (&kcrtc->base)); +if (kcrtc->side_by_side && + (evts->pipes[kcrtc->slave->id] &KOMEDA_EVENT_EOW )) + wb_conn->complete_pipes |=BIT (kcrtc->slave->id); + +if (wb_conn->expected_pipes == wb_conn->complete_pipes) { + wb_conn->complete_pipes =0 ; +drm_writeback_signal_completion (&wb_conn->base,0 ); } -/* will handle it together with the write back support */ -if (events &KOMEDA_EVENT_EOW ) -DRM_DEBUG ("EOW.\n" );if (events &KOMEDA_EVENT_FLIP ) {unsigned long flags;
drivers/gpu/drm/arm/display/komeda/komeda_kms.h +5 -0
@@ -58,6 +58,11 @@ struct komeda_wb_connector {/** @wb_layer: represents associated writeback pipeline of komeda */ struct komeda_layer *wb_layer; + +/** @expected_pipes: pipelines are used for the writeback job */ +u32 expected_pipes; +/** @complete_pipes: pipelines which have finished writeback */ +u32 complete_pipes; };/**
drivers/gpu/drm/arm/display/komeda/komeda_wb_connector.c +3 -0
@@ -157,6 +157,9 @@ static int komeda_wb_connector_add(struct komeda_kms_dev *kms,return -ENOMEM ; kwb_conn->wb_layer = kcrtc->master->wb_layer; + kwb_conn->expected_pipes =BIT (kcrtc->master->id); +if (kcrtc->side_by_side) + kwb_conn->expected_pipes |=BIT (kcrtc->slave->id); wb_conn = &kwb_conn->base; wb_conn->encoder.possible_crtcs =BIT (drm_crtc_index (&kcrtc->base));
Signature · 6 lines
2.20.1 _______________________________________________ dri-devel mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/dri-devel