Handle link training failure during modeset for DDI
Subsystems (3)
v2 cover Handle link training failure during modeset for DDI
Link training failure is handled by lowering the link rate/lane count as per the Dp Spec 1.2 and sending a hotplug uevent to userspace to notify that it needs to redo a modeset. This is notified by setting connector link status property which should be tracked by userspace to take the action.
This behaviour complies with Dp spec as tested by DP compliance test suite.
Manasi Navare (4):
drm: Add a new connector property for link status drm/i915: Set link status property for DP connector drm/i915: Find fallback link rate/lane count drm/i915: Implement Link Rate fallback on Link training failure
drivers/gpu/drm/drm_atomic_helper.c | 4 + drivers/gpu/drm/drm_connector.c | 17 ++++ drivers/gpu/drm/i915/intel_ddi.c | 23 ++++- drivers/gpu/drm/i915/intel_dp.c | 128 +++++++++++++++++++++++++- drivers/gpu/drm/i915/intel_dp_link_training.c | 12 ++- drivers/gpu/drm/i915/intel_drv.h | 13 ++- include/drm/drm_connector.h | 4 +- include/drm/drm_crtc.h | 5 + include/uapi/drm/drm_mode.h | 4 + 9 files changed, 201 insertions(+), 9 deletions(-)
Signature · 6 lines
1.9.1 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v2 1/4 drm: Add a new connector property for link status
A new default connector property is added for keeping
track of whether the link is good (link training passed) or link is bad (link training failed). If the link status property is not good, then userspace should fire off a new modeset at the current mode even if there have not been any changes in the mode list or connector status.
Also add link status connector member corersponding to the decoded value of link status property.
v3: Add link status member to store property value locally (Ville Syrjala)
v2:
* Make this a default connector property (Daniel Vetter)
--- drivers/gpu/drm/drm_connector.c | 17 +++++++++++++++++ include/drm/drm_connector.h | 4 +++- include/drm/drm_crtc.h | 5 +++++ include/uapi/drm/drm_mode.h | 4 ++++ 4 files changed, 29 insertions(+), 1 deletion(-)
drivers/gpu/drm/drm_connector.c +17 -0
@@ -243,6 +243,10 @@ int drm_connector_init(struct drm_device *dev,drm_object_attach_property (&connector->base, config->dpms_property,0 ); +drm_object_attach_property (&connector->base, + config->link_status_property, +0 ); +if (drm_core_check_feature (dev,DRIVER_ATOMIC )) {drm_object_attach_property (&connector->base, config->prop_crtc_id,0 ); }
@@ -506,6 +510,12 @@ const char *drm_get_subpixel_order_name(enum subpixel_order order) }; DRM_ENUM_NAME_FN(drm_get_dpms_name, drm_dpms_enum_list) +static const struct drm_prop_enum_list drm_link_status_enum_list[] = { + {DRM_MODE_LINK_STATUS_GOOD ,"Good" }, + {DRM_MODE_LINK_STATUS_BAD ,"Bad" }, +}; +DRM_ENUM_NAME_FN (drm_get_link_status_name, drm_link_status_enum_list) +/** * drm_display_info_set_bus_formats - set the supported bus formats * @info: display info to store bus formats in
@@ -622,6 +632,13 @@ int drm_connector_create_standard_properties(struct drm_device *dev)return -ENOMEM ; dev->mode_config.tile_property = prop; + prop =drm_property_create_enum (dev,0 ,"link-status" , + drm_link_status_enum_list, +ARRAY_SIZE (drm_link_status_enum_list)); +if (!prop) +return -ENOMEM ; + dev->mode_config.link_status_property = prop; +return 0 ; }
include/drm/drm_connector.h +3 -1
@@ -682,6 +682,9 @@ struct drm_connector {uint8_t num_h_tile, num_v_tile;uint8_t tile_h_loc, tile_v_loc;uint16_t tile_h_size, tile_v_size; + +/* Connector Link status for link training */ +bool link_status; };#define obj_to_connector (x) container_of(x, struct drm_connector, base)
@@ -754,7 +757,6 @@ int drm_mode_create_tv_properties(struct drm_device *dev,int drm_mode_create_scaling_mode_property (struct drm_device *dev);int drm_mode_create_aspect_ratio_property (struct drm_device *dev);int drm_mode_create_suggested_offset_properties (struct drm_device *dev); -int drm_mode_connector_set_path_property (struct drm_connector *connector,const char *path);int drm_mode_connector_set_tile_property (struct drm_connector *connector);
include/drm/drm_crtc.h +5 -0
@@ -1151,6 +1151,11 @@ struct drm_mode_config {*/ struct drm_property *tile_property;/** +* @link_status_property: Default connector property for link status +* of a connector as a result of link training. +*/ +struct drm_property *link_status_property; +/** * @plane_type_property: Default plane property to differentiate * CURSOR, PRIMARY and OVERLAY legacy uses of planes. */
include/uapi/drm/drm_mode.h +4 -0
@@ -121,6 +121,10 @@#define DRM_MODE_DIRTY_ON 1#define DRM_MODE_DIRTY_ANNOTATE 2 +/* Link Status options */ +#define DRM_MODE_LINK_STATUS_GOOD 0 +#define DRM_MODE_LINK_STATUS_BAD 1 +struct drm_mode_modeinfo {__u32 clock;__u16 hdisplay;
Signature · 6 lines
1.9.1 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Jani Nikula
maintainerRe: [PATCH v2 1/4] drm: Add a new connector property for link status
Quoted 8 lines · Manasi Navare
A new default connector property is added for keeping
track of whether the link is good (link training passed) or link is bad (link training failed). If the link status property is not good, then userspace should fire off a new modeset at the current mode even if there have not been any changes in the mode list or connector status.
Also add link status connector member corersponding to the decoded value of link status property.
I'm still a bit uneasy about the name "link status", because I want it to be evident that this is not just about display port link layer. However, I don't have a good counter proposal, and it's easy to change once we agree on everything else.
Quoted 66 lines
v3: Add link status member to store property value locally (Ville Syrjala)
v2:
* Make this a default connector property (Daniel Vetter)Cc Jani Nikula <[email protected]>Cc Daniel Vetter <[email protected]>Cc Ville Syrjala <[email protected]>Cc Chris Wilson <[email protected]>Signed-off-by Manasi Navare <[email protected]>--- drivers/gpu/drm/drm_connector.c | 17 +++++++++++++++++ include/drm/drm_connector.h | 4 +++- include/drm/drm_crtc.h | 5 +++++ include/uapi/drm/drm_mode.h | 4 ++++ 4 files changed, 29 insertions(+), 1 deletion(-)drivers/gpu/drm/drm_connector.c +17
-0@@ -243,6 +243,10 @@ int drm_connector_init(struct drm_device *dev,drm_object_attach_property (&connector->base, config->dpms_property,0 ); +drm_object_attach_property (&connector->base, + config->link_status_property, +0 ); +if (drm_core_check_feature (dev,DRIVER_ATOMIC )) {drm_object_attach_property (&connector->base, config->prop_crtc_id,0 ); }@@ -506,6 +510,12 @@ const char *drm_get_subpixel_order_name(enum subpixel_order order) }; DRM_ENUM_NAME_FN(drm_get_dpms_name, drm_dpms_enum_list) +static const struct drm_prop_enum_list drm_link_status_enum_list[] = { + {DRM_MODE_LINK_STATUS_GOOD ,"Good" }, + {DRM_MODE_LINK_STATUS_BAD ,"Bad" }, +}; +DRM_ENUM_NAME_FN (drm_get_link_status_name, drm_link_status_enum_list) +/** * drm_display_info_set_bus_formats - set the supported bus formats * @info: display info to store bus formats in @@ -622,6 +632,13 @@ int drm_connector_create_standard_properties(struct drm_device *dev)return -ENOMEM ; dev->mode_config.tile_property = prop; + prop =drm_property_create_enum (dev,0 ,"link-status" , + drm_link_status_enum_list, +ARRAY_SIZE (drm_link_status_enum_list)); +if (!prop) +return -ENOMEM ; + dev->mode_config.link_status_property = prop; +return 0 ; }include/drm/drm_connector.h +0
-0
include/drm/drm_connector.h struct drm_connector { · line 686
Quoted 5 lines
include/drm/drm_connector.h +2
-0@@ -682,6 +682,9 @@ struct drm_connector {uint8_t num_h_tile, num_v_tile;uint8_t tile_h_loc, tile_v_loc;uint16_t tile_h_size, tile_v_size; + +/* Connector Link status for link training */
Drop the "link training" bit, this is not just about DP.
include/drm/drm_connector.h struct drm_connector { · line 687
Quoted 1 line
include/drm/drm_connector.h +1
-0+bool link_status;
What does it mean for link status to be true or false? Which one is good and which one is bad? Please either name this differently, or make this an int and expect it to have the same values as the property.
Quoted 16 lines
include/drm/drm_connector.h +0
-1};#define obj_to_connector (x) container_of(x, struct drm_connector, base)@@ -754,7 +757,6 @@ int drm_mode_create_tv_properties(struct drm_device *dev,int drm_mode_create_scaling_mode_property (struct drm_device *dev);int drm_mode_create_aspect_ratio_property (struct drm_device *dev);int drm_mode_create_suggested_offset_properties (struct drm_device *dev); -int drm_mode_connector_set_path_property (struct drm_connector *connector,const char *path);int drm_mode_connector_set_tile_property (struct drm_connector *connector);include/drm/drm_crtc.h +0
-0
include/drm/drm_crtc.h struct drm_mode_config { · line 1155
Quoted 5 lines
include/drm/drm_crtc.h +2
-0@@ -1151,6 +1151,11 @@ struct drm_mode_config {*/ struct drm_property *tile_property;/** +* @link_status_property: Default connector property for link status +* of a connector as a result of link training.
Same about "link training".
Quoted 21 lines
include/drm/drm_crtc.h +3
-0+*/ +struct drm_property *link_status_property; +/** * @plane_type_property: Default plane property to differentiate * CURSOR, PRIMARY and OVERLAY legacy uses of planes. */ include/uapi/drm/drm_mode.h +4
-0@@ -121,6 +121,10 @@#define DRM_MODE_DIRTY_ON 1#define DRM_MODE_DIRTY_ANNOTATE 2 +/* Link Status options */ +#define DRM_MODE_LINK_STATUS_GOOD 0 +#define DRM_MODE_LINK_STATUS_BAD 1 +struct drm_mode_modeinfo {__u32 clock;__u16 hdisplay;
Signature · 5 lines
Jani Nikula, Intel Open Source Technology Center _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v3 1/4 drm: Add a new connector property for link status
In reply to Manasi NavareA new default connector property is added for keeping
track of whether the link is good (link training passed) or link is bad (link training failed). If the link status property is not good, then userspace should fire off a new modeset at the current mode even if there have not been any changes in the mode list or connector status.
Also add link status connector member corersponding to the decoded value of link status property.
v3:
* Drop "link training" from description since this is not specific to DP (Jani Nikula)
* Add link status member to store property value locally (Ville Syrjala)
v2:
* Make this a default connector property (Daniel Vetter)
--- drivers/gpu/drm/drm_connector.c | 17 +++++++++++++++++ include/drm/drm_connector.h | 7 ++++++- include/drm/drm_crtc.h | 5 +++++ include/uapi/drm/drm_mode.h | 4 ++++ 4 files changed, 32 insertions(+), 1 deletion(-)
drivers/gpu/drm/drm_connector.c +17 -0
@@ -243,6 +243,10 @@ int drm_connector_init(struct drm_device *dev,drm_object_attach_property (&connector->base, config->dpms_property,0 ); +drm_object_attach_property (&connector->base, + config->link_status_property, +0 ); +if (drm_core_check_feature (dev,DRIVER_ATOMIC )) {drm_object_attach_property (&connector->base, config->prop_crtc_id,0 ); }
@@ -506,6 +510,12 @@ const char *drm_get_subpixel_order_name(enum subpixel_order order) }; DRM_ENUM_NAME_FN(drm_get_dpms_name, drm_dpms_enum_list) +static const struct drm_prop_enum_list drm_link_status_enum_list[] = { + {DRM_MODE_LINK_STATUS_GOOD ,"Good" }, + {DRM_MODE_LINK_STATUS_BAD ,"Bad" }, +}; +DRM_ENUM_NAME_FN (drm_get_link_status_name, drm_link_status_enum_list) +/** * drm_display_info_set_bus_formats - set the supported bus formats * @info: display info to store bus formats in
@@ -622,6 +632,13 @@ int drm_connector_create_standard_properties(struct drm_device *dev)return -ENOMEM ; dev->mode_config.tile_property = prop; + prop =drm_property_create_enum (dev,0 ,"link-status" , + drm_link_status_enum_list, +ARRAY_SIZE (drm_link_status_enum_list)); +if (!prop) +return -ENOMEM ; + dev->mode_config.link_status_property = prop; +return 0 ; }
include/drm/drm_connector.h +6 -1
@@ -682,6 +682,12 @@ struct drm_connector {uint8_t num_h_tile, num_v_tile;uint8_t tile_h_loc, tile_v_loc;uint16_t tile_h_size, tile_v_size; + +/* Connector Link status +* 0: If the link is Good +* 1: If the link is Bad +*/ +int link_status; };#define obj_to_connector (x) container_of(x, struct drm_connector, base)
@@ -754,7 +760,6 @@ int drm_mode_create_tv_properties(struct drm_device *dev,int drm_mode_create_scaling_mode_property (struct drm_device *dev);int drm_mode_create_aspect_ratio_property (struct drm_device *dev);int drm_mode_create_suggested_offset_properties (struct drm_device *dev); -int drm_mode_connector_set_path_property (struct drm_connector *connector,const char *path);int drm_mode_connector_set_tile_property (struct drm_connector *connector);
include/drm/drm_crtc.h +5 -0
@@ -1151,6 +1151,11 @@ struct drm_mode_config {*/ struct drm_property *tile_property;/** +* @link_status_property: Default connector property for link status +* of a connector +*/ +struct drm_property *link_status_property; +/** * @plane_type_property: Default plane property to differentiate * CURSOR, PRIMARY and OVERLAY legacy uses of planes. */
include/uapi/drm/drm_mode.h +4 -0
@@ -121,6 +121,10 @@#define DRM_MODE_DIRTY_ON 1#define DRM_MODE_DIRTY_ANNOTATE 2 +/* Link Status options */ +#define DRM_MODE_LINK_STATUS_GOOD 0 +#define DRM_MODE_LINK_STATUS_BAD 1 +struct drm_mode_modeinfo {__u32 clock;__u16 hdisplay;
Signature · 6 lines
1.9.1 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Jani Nikula
maintainerRe: [PATCH v3 1/4] drm: Add a new connector property for link status
Quoted 8 lines · Manasi Navare
A new default connector property is added for keeping
track of whether the link is good (link training passed) or link is bad (link training failed). If the link status property is not good, then userspace should fire off a new modeset at the current mode even if there have not been any changes in the mode list or connector status.
Also add link status connector member corersponding to the decoded value of link status property.
I think it would be good to expand on this, both in the commit message and documentation. This is UABI after all.
When is the property changed, who changes it, what is the userspace expected to do when the status goes from good to bad, how does userspace notice it's gone bad (uevents).
While we want this for handling DP link training failures during mode setting according to the DP spec, in a way that can pass the CTS, it's also needed for async mode sets.
BR,
Jani.
Quoted 120 lines
v3:
* Drop "link training" from description since this is not specific to DP (Jani Nikula)
* Add link status member to store property value locally (Ville Syrjala)
v2:
* Make this a default connector property (Daniel Vetter)Cc Jani Nikula <[email protected]>Cc Daniel Vetter <[email protected]>Cc Ville Syrjala <[email protected]>Cc Chris Wilson <[email protected]>Signed-off-by Manasi Navare <[email protected]>--- drivers/gpu/drm/drm_connector.c | 17 +++++++++++++++++ include/drm/drm_connector.h | 7 ++++++- include/drm/drm_crtc.h | 5 +++++ include/uapi/drm/drm_mode.h | 4 ++++ 4 files changed, 32 insertions(+), 1 deletion(-)drivers/gpu/drm/drm_connector.c +17
-0@@ -243,6 +243,10 @@ int drm_connector_init(struct drm_device *dev,drm_object_attach_property (&connector->base, config->dpms_property,0 ); +drm_object_attach_property (&connector->base, + config->link_status_property, +0 ); +if (drm_core_check_feature (dev,DRIVER_ATOMIC )) {drm_object_attach_property (&connector->base, config->prop_crtc_id,0 ); }@@ -506,6 +510,12 @@ const char *drm_get_subpixel_order_name(enum subpixel_order order) }; DRM_ENUM_NAME_FN(drm_get_dpms_name, drm_dpms_enum_list) +static const struct drm_prop_enum_list drm_link_status_enum_list[] = { + {DRM_MODE_LINK_STATUS_GOOD ,"Good" }, + {DRM_MODE_LINK_STATUS_BAD ,"Bad" }, +}; +DRM_ENUM_NAME_FN (drm_get_link_status_name, drm_link_status_enum_list) +/** * drm_display_info_set_bus_formats - set the supported bus formats * @info: display info to store bus formats in @@ -622,6 +632,13 @@ int drm_connector_create_standard_properties(struct drm_device *dev)return -ENOMEM ; dev->mode_config.tile_property = prop; + prop =drm_property_create_enum (dev,0 ,"link-status" , + drm_link_status_enum_list, +ARRAY_SIZE (drm_link_status_enum_list)); +if (!prop) +return -ENOMEM ; + dev->mode_config.link_status_property = prop; +return 0 ; }include/drm/drm_connector.h +6
-1@@ -682,6 +682,12 @@ struct drm_connector {uint8_t num_h_tile, num_v_tile;uint8_t tile_h_loc, tile_v_loc;uint16_t tile_h_size, tile_v_size; + +/* Connector Link status +* 0: If the link is Good +* 1: If the link is Bad +*/ +int link_status; };#define obj_to_connector (x) container_of(x, struct drm_connector, base)@@ -754,7 +760,6 @@ int drm_mode_create_tv_properties(struct drm_device *dev,int drm_mode_create_scaling_mode_property (struct drm_device *dev);int drm_mode_create_aspect_ratio_property (struct drm_device *dev);int drm_mode_create_suggested_offset_properties (struct drm_device *dev); -int drm_mode_connector_set_path_property (struct drm_connector *connector,const char *path);int drm_mode_connector_set_tile_property (struct drm_connector *connector);include/drm/drm_crtc.h +5
-0@@ -1151,6 +1151,11 @@ struct drm_mode_config {*/ struct drm_property *tile_property;/** +* @link_status_property: Default connector property for link status +* of a connector +*/ +struct drm_property *link_status_property; +/** * @plane_type_property: Default plane property to differentiate * CURSOR, PRIMARY and OVERLAY legacy uses of planes. */ include/uapi/drm/drm_mode.h +4
-0@@ -121,6 +121,10 @@#define DRM_MODE_DIRTY_ON 1#define DRM_MODE_DIRTY_ANNOTATE 2 +/* Link Status options */ +#define DRM_MODE_LINK_STATUS_GOOD 0 +#define DRM_MODE_LINK_STATUS_BAD 1 +struct drm_mode_modeinfo {__u32 clock;__u16 hdisplay;
Signature · 5 lines
Jani Nikula, Intel Open Source Technology Center _______________________________________________ dri-devel mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/dri-devel
PATCH v2 2/4 drm/i915: Set link status property for DP connector
In reply to Manasi NavareThis defines a helper function to set the property value. This will be used to set the link status to Bad in case of link training failures.
v2:
* Simplify the return value (Jani Nikula)
--- drivers/gpu/drm/i915/intel_dp.c | 11 +++++++++++ drivers/gpu/drm/i915/intel_drv.h | 2 ++ 2 files changed, 13 insertions(+)
drivers/gpu/drm/i915/intel_dp.c +11 -0
@@ -4640,6 +4640,17 @@ static int intel_dp_get_modes(struct drm_connector *connector)return 0 ; } +int +intel_dp_set_link_status_property (struct drm_connector *connector, +uint64_t val) +{ +struct drm_device *dev = connector->dev; + +return drm_object_property_set_value (&connector->base, + dev->mode_config.link_status_property, + val); +} +static int intel_dp_connector_register (struct drm_connector *connector) {
drivers/gpu/drm/i915/intel_drv.h +2 -0
@@ -1391,6 +1391,8 @@ u32 skl_plane_stride(const struct drm_framebuffer *fb, int plane,bool intel_dp_init (struct drm_device *dev,i915_reg_t output_reg,enum port port);bool intel_dp_init_connector (struct intel_digital_port *intel_dig_port,struct intel_connector *intel_connector); +int intel_dp_set_link_status_property (struct drm_connector *connector, +uint64_t val);void intel_dp_set_link_params (struct intel_dp *intel_dp,int link_rate,uint8_t lane_count,bool link_mst);
Signature · 6 lines
1.9.1 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH v2 2/4] drm/i915: Set link status property for DP connector
Quoted 7 lines · Manasi Navare
This defines a helper function to set the property value. This will be used to set the link status to Bad in case of link training failures.
v2:
* Simplify the return value (Jani Nikula)
Quoted 6 lines
Cc Jani Nikula <[email protected]>Cc Daniel Vetter <[email protected]>Cc Ville Syrjala <[email protected]>Cc Chris Wilson <[email protected]>Signed-off-by Manasi Navare <[email protected]>
Quoted 40 lines
--- drivers/gpu/drm/i915/intel_dp.c | 11 +++++++++++ drivers/gpu/drm/i915/intel_drv.h | 2 ++ 2 files changed, 13 insertions(+)drivers/gpu/drm/i915/intel_dp.c +11
-0@@ -4640,6 +4640,17 @@ static int intel_dp_get_modes(struct drm_connector *connector)return 0 ; } +int +intel_dp_set_link_status_property (struct drm_connector *connector, +uint64_t val) +{ +struct drm_device *dev = connector->dev; + +return drm_object_property_set_value (&connector->base, + dev->mode_config.link_status_property, + val); +} +static int intel_dp_connector_register (struct drm_connector *connector) {drivers/gpu/drm/i915/intel_drv.h +2
-0@@ -1391,6 +1391,8 @@ u32 skl_plane_stride(const struct drm_framebuffer *fb, int plane,bool intel_dp_init (struct drm_device *dev,i915_reg_t output_reg,enum port port);bool intel_dp_init_connector (struct intel_digital_port *intel_dig_port,struct intel_connector *intel_connector); +int intel_dp_set_link_status_property (struct drm_connector *connector, +uint64_t val);void intel_dp_set_link_params (struct intel_dp *intel_dp,int link_rate,uint8_t lane_count,bool link_mst);
Signature · 5 lines
Jani Nikula, Intel Open Source Technology Center _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v2 3/4 drm/i915: Find fallback link rate/lane count
In reply to Manasi NavareIf link training fails, then we need to fallback to lower link rate first and if link training fails at RBR, then fallback to lower lane count.
This function finds the next lower link rate/lane count value after link training failure.
v2:
Squash the patch that returns the link rate index (Jani Nikula)
--- drivers/gpu/drm/i915/intel_dp.c | 43 ++++++++++++++++++++++++++++++++++++++++ drivers/gpu/drm/i915/intel_drv.h | 6 ++++++ 2 files changed, 49 insertions(+)
drivers/gpu/drm/i915/intel_dp.c +43 -0
@@ -288,6 +288,49 @@ static int intel_dp_common_rates(struct intel_dp *intel_dp, common_rates); } +static int intel_dp_link_rate_index (struct intel_dp *intel_dp, +int *common_rates,int link_rate) +{ +int common_len; +int index; + + common_len =intel_dp_common_rates (intel_dp, common_rates); +for (index =0 ; index < common_len; index++) { +if (link_rate == common_rates[common_len - index -1 ]) +return common_len - index -1 ; + } + +return -1 ; +} + +void intel_dp_get_link_train_fallback_values (struct intel_dp *intel_dp, +int link_rate,uint8_t lane_count) +{ +int common_rates[DP_MAX_SUPPORTED_RATES ] = {}; +int common_len; +int link_rate_index =-1 ; + +if (!intel_dp->link_train_failed) +return ; + + common_len =intel_dp_common_rates (intel_dp, common_rates); + link_rate_index =intel_dp_link_rate_index (intel_dp, + common_rates, + link_rate); +if (link_rate_index >0 ) { + intel_dp->fallback_link_rate_index = link_rate_index -1 ; + intel_dp->fallback_link_rate = common_rates[intel_dp->fallback_link_rate_index]; + intel_dp->fallback_lane_count =intel_dp_max_lane_count (intel_dp); + }else if (lane_count >1 ) { + intel_dp->fallback_link_rate_index = common_len -1 ; + intel_dp->fallback_link_rate = common_rates[intel_dp->fallback_link_rate_index]; + intel_dp->fallback_lane_count = lane_count >>1 ; + }else { +DRM_ERROR ("Link Training Unsuccessful\n" ); + intel_dp->link_train_failed = false; + } +} +static enum drm_mode_status intel_dp_mode_valid (struct drm_connector *connector,struct drm_display_mode *mode)
drivers/gpu/drm/i915/intel_drv.h +6 -0
@@ -897,6 +897,10 @@ struct intel_dp {uint32_t DP ;int link_rate;uint8_t lane_count; +int fallback_link_rate; +uint8_t fallback_lane_count; +int fallback_link_rate_index; +bool link_train_failed;uint8_t sink_count;bool link_mst;bool has_audio;
@@ -1396,6 +1400,8 @@ int intel_dp_set_link_status_property(struct drm_connector *connector,void intel_dp_set_link_params (struct intel_dp *intel_dp,int link_rate,uint8_t lane_count,bool link_mst); +void intel_dp_get_link_train_fallback_values (struct intel_dp *intel_dp, +int link_rate,uint8_t lane_count);void intel_dp_start_link_train (struct intel_dp *intel_dp);void intel_dp_stop_link_train (struct intel_dp *intel_dp);void intel_dp_sink_dpms (struct intel_dp *intel_dp,int mode);
Signature · 6 lines
1.9.1 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v2 4/4 drm/i915: Implement Link Rate fallback on Link training failure
In reply to Manasi NavareIf link training at a link rate optimal for a particular mode fails during modeset's atomic commit phase, then we let the modeset complete and then retry. We save the link rate value at which link training failed, update the link status property to "BAD" and use a lower link rate to prune the modes. It will redo the modeset on the current mode at lower link rate or if the current mode gets pruned due to lower link constraints then, it will send a hotplug uevent for userspace to handle it.
This is also required to pass DP CTS tests 4.3.1.3, 4.3.1.4, 4.3.1.6.
v2:
* Squashed a few patches (Jani Nikula)
--- drivers/gpu/drm/drm_atomic_helper.c | 4 ++ drivers/gpu/drm/i915/intel_ddi.c | 23 ++++++++- drivers/gpu/drm/i915/intel_dp.c | 74 +++++++++++++++++++++++++-- drivers/gpu/drm/i915/intel_dp_link_training.c | 12 +++-- drivers/gpu/drm/i915/intel_drv.h | 5 +- 5 files changed, 110 insertions(+), 8 deletions(-)
drivers/gpu/drm/drm_atomic_helper.c +4 -0
@@ -519,6 +519,10 @@ static int handle_conflicting_encoders(struct drm_atomic_state *state, connector_state);if (ret)return ret; + + crtc_state =drm_atomic_get_existing_crtc_state (state, connector->state->crtc); +if (connector->link_status ==DRM_MODE_LINK_STATUS_BAD ) + crtc_state->connectors_changed = true; }/*
drivers/gpu/drm/i915/intel_ddi.c +22 -1
@@ -1684,6 +1684,8 @@ static void intel_ddi_pre_enable_dp(struct intel_encoder *encoder,struct intel_dp *intel_dp =enc_to_intel_dp (&encoder->base);struct drm_i915_private *dev_priv =to_i915 (encoder->base.dev);enum port port =intel_ddi_get_encoder_port (encoder); +struct intel_connector *intel_connector = intel_dp->attached_connector; +struct drm_connector *connector = &intel_connector->base;intel_dp_set_link_params (intel_dp, link_rate, lane_count, link_mst);
@@ -1694,7 +1696,26 @@ static void intel_ddi_pre_enable_dp(struct intel_encoder *encoder,intel_prepare_dp_ddi_buffers (encoder);intel_ddi_init_dp_buf_reg (encoder);intel_dp_sink_dpms (intel_dp,DRM_MODE_DPMS_ON ); -intel_dp_start_link_train (intel_dp); +if (!intel_dp_start_link_train (intel_dp)) { +DRM_DEBUG_KMS ("Link Training failed at link rate = %d, lane count = %d" , + link_rate, lane_count); + intel_dp->link_train_failed = true; +intel_dp_get_link_train_fallback_values (intel_dp, link_rate, + lane_count); +/* Schedule a Hotplug Uevent to userspace to start modeset */ +schedule_work (&intel_connector->modeset_retry_work); + }else { +DRM_DEBUG_KMS ("Link Training Passed at Link Rate = %d, Lane count = %d" , + link_rate, lane_count); + intel_dp->link_train_failed = false; + intel_dp->fallback_link_rate_index =-1 ; + intel_dp->fallback_link_rate =0 ; + intel_dp->fallback_lane_count =0 ; + connector->link_status =DRM_MODE_LINK_STATUS_GOOD ; +intel_dp_set_link_status_property (connector, +DRM_MODE_LINK_STATUS_GOOD ); + } +if (port !=PORT_A ||INTEL_GEN (dev_priv) >=9 )intel_dp_stop_link_train (intel_dp); }
drivers/gpu/drm/i915/intel_dp.c +71 -3
111 lines not on this page · raw message
drivers/gpu/drm/i915/intel_dp_link_training.c +9 -3
@@ -310,9 +310,15 @@ void intel_dp_stop_link_train(struct intel_dp *intel_dp) DP_TRAINING_PATTERN_DISABLE); } -void +bool intel_dp_start_link_train (struct intel_dp *intel_dp) { -intel_dp_link_training_clock_recovery (intel_dp); -intel_dp_link_training_channel_equalization (intel_dp); +bool ret; + +if (intel_dp_link_training_clock_recovery (intel_dp)) { + ret =intel_dp_link_training_channel_equalization (intel_dp); +if (ret) +return true; + } +return false; }
drivers/gpu/drm/i915/intel_drv.h +4 -1
@@ -312,6 +312,9 @@ struct intel_connector {void *port;/* store this opaque as its illegal to dereference it */ struct intel_dp *mst_port; + +/* Work struct to schedule a uevent on link train failure */ +struct work_struct modeset_retry_work; };struct dpll {
@@ -1402,7 +1405,7 @@ void intel_dp_set_link_params(struct intel_dp *intel_dp, bool link_mst);void intel_dp_get_link_train_fallback_values (struct intel_dp *intel_dp,int link_rate,uint8_t lane_count); -void intel_dp_start_link_train (struct intel_dp *intel_dp); +bool intel_dp_start_link_train (struct intel_dp *intel_dp);void intel_dp_stop_link_train (struct intel_dp *intel_dp);void intel_dp_sink_dpms (struct intel_dp *intel_dp,int mode);void intel_dp_encoder_reset (struct drm_encoder *encoder);
Signature · 6 lines
1.9.1 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Jani Nikula
maintainerRe: [PATCH v2 4/4] drm/i915: Implement Link Rate fallback on Link training failure
Quoted 58 lines · Manasi Navare
If link training at a link rate optimal for a particular mode fails during modeset's atomic commit phase, then we let the modeset complete and then retry. We save the link rate value at which link training failed, update the link status property to "BAD" and use a lower link rate to prune the modes. It will redo the modeset on the current mode at lower link rate or if the current mode gets pruned due to lower link constraints then, it will send a hotplug uevent for userspace to handle it.
This is also required to pass DP CTS tests 4.3.1.3, 4.3.1.4, 4.3.1.6.
v2:
* Squashed a few patches (Jani Nikula)Cc Jani Nikula <[email protected]>Cc Daniel Vetter <[email protected]>Cc Ville Syrjala <[email protected]>Signed-off-by Manasi Navare <[email protected]>--- drivers/gpu/drm/drm_atomic_helper.c | 4 ++ drivers/gpu/drm/i915/intel_ddi.c | 23 ++++++++- drivers/gpu/drm/i915/intel_dp.c | 74 +++++++++++++++++++++++++-- drivers/gpu/drm/i915/intel_dp_link_training.c | 12 +++-- drivers/gpu/drm/i915/intel_drv.h | 5 +- 5 files changed, 110 insertions(+), 8 deletions(-)drivers/gpu/drm/drm_atomic_helper.c +4
-0@@ -519,6 +519,10 @@ static int handle_conflicting_encoders(struct drm_atomic_state *state, connector_state);if (ret)return ret; + + crtc_state =drm_atomic_get_existing_crtc_state (state, connector->state->crtc); +if (connector->link_status ==DRM_MODE_LINK_STATUS_BAD ) + crtc_state->connectors_changed = true; }/* drivers/gpu/drm/i915/intel_ddi.c +2
-0@@ -1684,6 +1684,8 @@ static void intel_ddi_pre_enable_dp(struct intel_encoder *encoder,struct intel_dp *intel_dp =enc_to_intel_dp (&encoder->base);struct drm_i915_private *dev_priv =to_i915 (encoder->base.dev);enum port port =intel_ddi_get_encoder_port (encoder); +struct intel_connector *intel_connector = intel_dp->attached_connector; +struct drm_connector *connector = &intel_connector->base;intel_dp_set_link_params (intel_dp, link_rate, lane_count, link_mst);@@ -1694,7 +1696,26 @@ static void intel_ddi_pre_enable_dp(struct intel_encoder *encoder,intel_prepare_dp_ddi_buffers (encoder);intel_ddi_init_dp_buf_reg (encoder);
drivers/gpu/drm/i915/intel_ddi.c static void intel_ddi_pre_enable_dp(struct intel_encoder *encoder, · line 1706
Quoted 10 lines
drivers/gpu/drm/i915/intel_ddi.c +8
-1intel_dp_sink_dpms (intel_dp,DRM_MODE_DPMS_ON ); -intel_dp_start_link_train (intel_dp); +if (!intel_dp_start_link_train (intel_dp)) { +DRM_DEBUG_KMS ("Link Training failed at link rate = %d, lane count = %d" , + link_rate, lane_count); + intel_dp->link_train_failed = true; +intel_dp_get_link_train_fallback_values (intel_dp, link_rate, + lane_count); +/* Schedule a Hotplug Uevent to userspace to start modeset */ +schedule_work (&intel_connector->modeset_retry_work);
This is not just about DDI. Need to do this for the other cases too.
drivers/gpu/drm/i915/intel_ddi.c static void intel_ddi_pre_enable_dp(struct intel_encoder *encoder, · line 1716
Quoted 10 lines
drivers/gpu/drm/i915/intel_ddi.c +10
-0+ }else { +DRM_DEBUG_KMS ("Link Training Passed at Link Rate = %d, Lane count = %d" , + link_rate, lane_count); + intel_dp->link_train_failed = false; + intel_dp->fallback_link_rate_index =-1 ; + intel_dp->fallback_link_rate =0 ; + intel_dp->fallback_lane_count =0 ; + connector->link_status =DRM_MODE_LINK_STATUS_GOOD ; +intel_dp_set_link_status_property (connector, +DRM_MODE_LINK_STATUS_GOOD );
Looks like you never actually read connector->link_status... Why do you need both connector->link_status and intel_dp->link_train_failed? Do you think you have 4 states? What are they? Can't this all be in sync with the property?
Quoted 27 lines
drivers/gpu/drm/i915/intel_ddi.c +2
-0+ } +if (port !=PORT_A ||INTEL_GEN (dev_priv) >=9 )intel_dp_stop_link_train (intel_dp); }drivers/gpu/drm/i915/intel_dp.c +8
-2@@ -354,8 +354,14 @@ void intel_dp_get_link_train_fallback_values(struct intel_dp *intel_dp, target_clock = fixed_mode->clock; } - max_link_clock =intel_dp_max_link_rate (intel_dp); - max_lanes =intel_dp_max_lane_count (intel_dp); +/* Prune the modes using the fallback link rate/lane count */ +if (intel_dp->link_train_failed) { + max_link_clock = intel_dp->fallback_link_rate; + max_lanes = intel_dp->fallback_lane_count; + }else { + max_link_clock =intel_dp_max_link_rate (intel_dp); + max_lanes =intel_dp_max_lane_count (intel_dp); + } max_rate =intel_dp_max_data_rate (max_link_clock, max_lanes); mode_rate =intel_dp_link_required (target_clock,18 );
drivers/gpu/drm/i915/intel_dp.c static int intel_dp_compute_bpp(struct intel_dp *intel_dp, · line 1654
Quoted 9 lines
drivers/gpu/drm/i915/intel_dp.c +6
-0@@ -1640,6 +1646,12 @@ static int intel_dp_compute_bpp(struct intel_dp *intel_dp,if (adjusted_mode->flags &DRM_MODE_FLAG_DBLCLK )return false; +/* Fall back to lower link rate in case of failure in previous modeset */ +if (intel_dp->link_train_failed) { + min_lane_count = max_lane_count = intel_dp->fallback_lane_count; + min_clock = max_clock = intel_dp->fallback_link_rate_index; + } +
My general feeling is that there's starting to be a bit too much special casing around the fallback values. I'm not decided we need to fix this right away in this series, or whether it can be follow-up work.
One idea is to compute the common rates and lanes once when they're first needed, and all of the helpers would use that info. The fallback code would just trim those, and the conditional fallback stuff could be removed from all over the place.
Quoted 60 lines
drivers/gpu/drm/i915/intel_dp.c +37
-1DRM_DEBUG_KMS("DP link computation with max lane count %i " "max bw %d pixel clock %iKHz\n", max_lane_count, common_rates[max_clock],@@ -4423,6 +4435,13 @@ static bool intel_digital_port_connected(struct drm_i915_private *dev_priv, intel_dp->compliance_test_active =0 ; intel_dp->compliance_test_type =0 ; intel_dp->compliance_test_data =0 ; + intel_dp->link_train_failed = false; + intel_dp->fallback_link_rate_index =-1 ; + intel_dp->fallback_link_rate =0 ; + intel_dp->fallback_lane_count =0 ; + connector->link_status =DRM_MODE_LINK_STATUS_GOOD ; +intel_dp_set_link_status_property (connector, +DRM_MODE_LINK_STATUS_GOOD );if (intel_dp->is_mst) { DRM_DEBUG_KMS("MST device may have disappeared %d vs %d\n",@@ -4514,8 +4533,12 @@ static bool intel_digital_port_connected(struct drm_i915_private *dev_priv,DRM_DEBUG_KMS ("[CONNECTOR:%d:%s]\n" , connector->base.id, connector->name); +/* If this is a retry due to link trianing failure */ +if (status == connector_status_connected && intel_dp->link_train_failed) +return status; +/* If full detect is not performed yet, do a full detect */ -if (!intel_dp->detect_done) +if (!intel_dp->detect_done && !intel_dp->link_train_failed) status =intel_dp_long_pulse (intel_dp->attached_connector); intel_dp->detect_done = false;@@ -5692,6 +5715,47 @@ static bool intel_edp_init_connector(struct intel_dp *intel_dp,return false; } +static void intel_dp_modeset_retry_work_fn (struct work_struct *work) +{ +struct intel_connector *intel_connector; +struct drm_connector *connector; +struct drm_display_mode *mode; +bool verbose_prune = true; +bool reprobe = false; + + intel_connector =container_of (work,typeof (*intel_connector), + modeset_retry_work); + connector = &intel_connector->base; + +/* Grab the locks before changing connector property*/ +mutex_lock (&connector->dev->mode_config.mutex); +DRM_DEBUG_KMS ("[CONNECTOR:%d:%s]\n" , connector->base.id, + connector->name); +list_for_each_entry (mode, &connector->modes, head) { + mode->status =intel_dp_mode_valid (connector, + mode); +if (mode->status !=MODE_OK ) + reprobe = true; + } +drm_mode_prune_invalid (connector->dev, &connector->modes, + verbose_prune); +
drivers/gpu/drm/i915/intel_dp.c static bool intel_edp_init_connector(struct intel_dp *intel_dp, · line 5752
Quoted 10 lines
drivers/gpu/drm/i915/intel_dp.c +10
-0+/* Set connector link status to BAD only if modeset required +* for the current mode, if mode list changed then just send uevent +* so that it can reprobe the connectors and validate modes and do +* a modeset on a different valid mode. +*/ +if (!reprobe) { + connector->link_status =DRM_MODE_LINK_STATUS_BAD ; +intel_dp_set_link_status_property (connector, +DRM_MODE_LINK_STATUS_BAD ); + }
I think the link status property should be set to bad unconditionally here. If the current link is bad, it is bad right now independent of modes fitting into the fallback link parameters.
Which makes me think, unless I'm missing something, that you might be able to prune the invalid modes and set the property right away when link training fails, and only use the work to do drm_kms_helper_hotplug_event.
Quoted 66 lines
drivers/gpu/drm/i915/intel_dp.c +10
-0+mutex_unlock (&connector->dev->mode_config.mutex); + +/* Send Hotplug uevent so userspace can reprobe */ +drm_kms_helper_hotplug_event (connector->dev); +} +bool intel_dp_init_connector (struct intel_digital_port *intel_dig_port,struct intel_connector *intel_connector)@@ -5704,6 +5768,10 @@ static bool intel_edp_init_connector(struct intel_dp *intel_dp,enum port port = intel_dig_port->port;int type; +/* Initialize the work for modeset in case of link train failure */ +INIT_WORK (&intel_connector->modeset_retry_work, + intel_dp_modeset_retry_work_fn); +if (WARN (intel_dig_port->max_lanes <1 ,"Not enough lanes (%d) for DP on port %c\n" , intel_dig_port->max_lanes,port_name (port)))drivers/gpu/drm/i915/intel_dp_link_training.c +9
-3@@ -310,9 +310,15 @@ void intel_dp_stop_link_train(struct intel_dp *intel_dp) DP_TRAINING_PATTERN_DISABLE); } -void +bool intel_dp_start_link_train (struct intel_dp *intel_dp) { -intel_dp_link_training_clock_recovery (intel_dp); -intel_dp_link_training_channel_equalization (intel_dp); +bool ret; + +if (intel_dp_link_training_clock_recovery (intel_dp)) { + ret =intel_dp_link_training_channel_equalization (intel_dp); +if (ret) +return true; + } +return false; }drivers/gpu/drm/i915/intel_drv.h +4
-1@@ -312,6 +312,9 @@ struct intel_connector {void *port;/* store this opaque as its illegal to dereference it */ struct intel_dp *mst_port; + +/* Work struct to schedule a uevent on link train failure */ +struct work_struct modeset_retry_work; };struct dpll {@@ -1402,7 +1405,7 @@ void intel_dp_set_link_params(struct intel_dp *intel_dp, bool link_mst);void intel_dp_get_link_train_fallback_values (struct intel_dp *intel_dp,int link_rate,uint8_t lane_count); -void intel_dp_start_link_train (struct intel_dp *intel_dp); +bool intel_dp_start_link_train (struct intel_dp *intel_dp);void intel_dp_stop_link_train (struct intel_dp *intel_dp);void intel_dp_sink_dpms (struct intel_dp *intel_dp,int mode);void intel_dp_encoder_reset (struct drm_encoder *encoder);
Signature · 5 lines
Jani Nikula, Intel Open Source Technology Center _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH v2 4/4] drm/i915: Implement Link Rate fallback on Link training failure
Quoted 63 lines · Jani Nikula
Quoted 62 lines · Manasi Navare
If link training at a link rate optimal for a particular mode fails during modeset's atomic commit phase, then we let the modeset complete and then retry. We save the link rate value at which link training failed, update the link status property to "BAD" and use a lower link rate to prune the modes. It will redo the modeset on the current mode at lower link rate or if the current mode gets pruned due to lower link constraints then, it will send a hotplug uevent for userspace to handle it.
This is also required to pass DP CTS tests 4.3.1.3, 4.3.1.4, 4.3.1.6.
v2:
* Squashed a few patches (Jani Nikula)Cc Jani Nikula <[email protected]>Cc Daniel Vetter <[email protected]>Cc Ville Syrjala <[email protected]>Signed-off-by Manasi Navare <[email protected]>--- drivers/gpu/drm/drm_atomic_helper.c | 4 ++ drivers/gpu/drm/i915/intel_ddi.c | 23 ++++++++- drivers/gpu/drm/i915/intel_dp.c | 74 +++++++++++++++++++++++++-- drivers/gpu/drm/i915/intel_dp_link_training.c | 12 +++-- drivers/gpu/drm/i915/intel_drv.h | 5 +- 5 files changed, 110 insertions(+), 8 deletions(-)drivers/gpu/drm/drm_atomic_helper.c +4
-0@@ -519,6 +519,10 @@ static int handle_conflicting_encoders(struct drm_atomic_state *state, connector_state);if (ret)return ret; + + crtc_state =drm_atomic_get_existing_crtc_state (state, connector->state->crtc); +if (connector->link_status ==DRM_MODE_LINK_STATUS_BAD ) + crtc_state->connectors_changed = true; }/* drivers/gpu/drm/i915/intel_ddi.c +4
-1@@ -1684,6 +1684,8 @@ static void intel_ddi_pre_enable_dp(struct intel_encoder *encoder,struct intel_dp *intel_dp =enc_to_intel_dp (&encoder->base);struct drm_i915_private *dev_priv =to_i915 (encoder->base.dev);enum port port =intel_ddi_get_encoder_port (encoder); +struct intel_connector *intel_connector = intel_dp->attached_connector; +struct drm_connector *connector = &intel_connector->base;intel_dp_set_link_params (intel_dp, link_rate, lane_count, link_mst);@@ -1694,7 +1696,26 @@ static void intel_ddi_pre_enable_dp(struct intel_encoder *encoder,intel_prepare_dp_ddi_buffers (encoder);intel_ddi_init_dp_buf_reg (encoder);intel_dp_sink_dpms (intel_dp,DRM_MODE_DPMS_ON ); -intel_dp_start_link_train (intel_dp); +if (!intel_dp_start_link_train (intel_dp)) { + DRM_DEBUG_KMS("Link Training failed at link rate = %d, lane count = %d",
Quoted 9 lines
drivers/gpu/drm/i915/intel_ddi.c static void intel_ddi_pre_enable_dp(struct intel_encoder *encoder, · line 1706
Quoted 6 lines
drivers/gpu/drm/i915/intel_ddi.c +6
-0+ link_rate, lane_count); + intel_dp->link_train_failed = true; +intel_dp_get_link_train_fallback_values (intel_dp, link_rate, + lane_count); +/* Schedule a Hotplug Uevent to userspace to start modeset */ +schedule_work (&intel_connector->modeset_retry_work);This is not just about DDI. Need to do this for the other cases too.
Yes, first series will g out for adding this support for DDI, then more patches to expand it to non DDI platforms.
Quoted 7 lines
Quoted 7 lines
drivers/gpu/drm/i915/intel_ddi.c +7
-0+ }else { +DRM_DEBUG_KMS ("Link Training Passed at Link Rate = %d, Lane count = %d" , + link_rate, lane_count); + intel_dp->link_train_failed = false; + intel_dp->fallback_link_rate_index =-1 ; + intel_dp->fallback_link_rate =0 ; + intel_dp->fallback_lane_count =0 ;
Quoted 9 lines
drivers/gpu/drm/i915/intel_ddi.c static void intel_ddi_pre_enable_dp(struct intel_encoder *encoder, · line 1716
Quoted 3 lines
drivers/gpu/drm/i915/intel_ddi.c +3
-0+ connector->link_status =DRM_MODE_LINK_STATUS_GOOD ; +intel_dp_set_link_status_property (connector, +DRM_MODE_LINK_STATUS_GOOD );Looks like you never actually read
connector->link_status... Why do you need bothconnector->link_statusandintel_dp->link_train_failed? Do you think you have 4 states? What are they? Can't this all be in sync with the property?
This connector->link_status member of drm_Connector gets read in drm_atomic_helper_check_modeset() in the driver where it reads this
and sets crtc_state->connector_Changed to true if this link_status
has changed.
This is required so that the driver does a complete modeset.
This connector->link_status was in sync with the property. But reading the drm_object property in drm_atomic_helper_Check_modeset was causing the system to not boot.intel_dp->link_train_failed also just indicates if the link failed, I will have to see if i can just use connector->link_status for this purpose.
Quoted 41 lines
Quoted 27 lines
drivers/gpu/drm/i915/intel_ddi.c +2
-0+ } +if (port !=PORT_A ||INTEL_GEN (dev_priv) >=9 )intel_dp_stop_link_train (intel_dp); }drivers/gpu/drm/i915/intel_dp.c +8
-2@@ -354,8 +354,14 @@ void intel_dp_get_link_train_fallback_values(struct intel_dp *intel_dp, target_clock = fixed_mode->clock; } - max_link_clock =intel_dp_max_link_rate (intel_dp); - max_lanes =intel_dp_max_lane_count (intel_dp); +/* Prune the modes using the fallback link rate/lane count */ +if (intel_dp->link_train_failed) { + max_link_clock = intel_dp->fallback_link_rate; + max_lanes = intel_dp->fallback_lane_count; + }else { + max_link_clock =intel_dp_max_link_rate (intel_dp); + max_lanes =intel_dp_max_lane_count (intel_dp); + } max_rate =intel_dp_max_data_rate (max_link_clock, max_lanes); mode_rate =intel_dp_link_required (target_clock,18 );drivers/gpu/drm/i915/intel_dp.c static int intel_dp_compute_bpp(struct intel_dp *intel_dp, · line 1654
Quoted 9 lines
drivers/gpu/drm/i915/intel_dp.c +6
-0@@ -1640,6 +1646,12 @@ static int intel_dp_compute_bpp(struct intel_dp *intel_dp,if (adjusted_mode->flags &DRM_MODE_FLAG_DBLCLK )return false; +/* Fall back to lower link rate in case of failure in previous modeset */ +if (intel_dp->link_train_failed) { + min_lane_count = max_lane_count = intel_dp->fallback_lane_count; + min_clock = max_clock = intel_dp->fallback_link_rate_index; + } +My general feeling is that there's starting to be a bit too much special casing around the fallback values. I'm not decided we need to fix this right away in this series, or whether it can be follow-up work.
Quoted 5 lines
One idea is to compute the common rates and lanes once when they're first needed, and all of the helpers would use that info. The fallback code would just trim those, and the conditional fallback stuff could be removed from all over the place.
These fallback values get computed in a separate helper function that I have added intel_dp_get_link_train_fallback_values. It is the previous patch.
I store the fallback values in intel_dp one at a time because for that
iteration of modeset we only need to try the fallback_link_rate
and fallback_lane_count so we dont need an array here.
Are you suggesting just changing the common_rates array itself to get trimmed to use the trimmed fallback values after link training fails? Could you please elaborate your thought?
Quoted 76 lines
Quoted 60 lines
drivers/gpu/drm/i915/intel_dp.c +37
-1DRM_DEBUG_KMS("DP link computation with max lane count %i " "max bw %d pixel clock %iKHz\n", max_lane_count, common_rates[max_clock],@@ -4423,6 +4435,13 @@ static bool intel_digital_port_connected(struct drm_i915_private *dev_priv, intel_dp->compliance_test_active =0 ; intel_dp->compliance_test_type =0 ; intel_dp->compliance_test_data =0 ; + intel_dp->link_train_failed = false; + intel_dp->fallback_link_rate_index =-1 ; + intel_dp->fallback_link_rate =0 ; + intel_dp->fallback_lane_count =0 ; + connector->link_status =DRM_MODE_LINK_STATUS_GOOD ; +intel_dp_set_link_status_property (connector, +DRM_MODE_LINK_STATUS_GOOD );if (intel_dp->is_mst) { DRM_DEBUG_KMS("MST device may have disappeared %d vs %d\n",@@ -4514,8 +4533,12 @@ static bool intel_digital_port_connected(struct drm_i915_private *dev_priv,DRM_DEBUG_KMS ("[CONNECTOR:%d:%s]\n" , connector->base.id, connector->name); +/* If this is a retry due to link trianing failure */ +if (status == connector_status_connected && intel_dp->link_train_failed) +return status; +/* If full detect is not performed yet, do a full detect */ -if (!intel_dp->detect_done) +if (!intel_dp->detect_done && !intel_dp->link_train_failed) status =intel_dp_long_pulse (intel_dp->attached_connector); intel_dp->detect_done = false;@@ -5692,6 +5715,47 @@ static bool intel_edp_init_connector(struct intel_dp *intel_dp,return false; } +static void intel_dp_modeset_retry_work_fn (struct work_struct *work) +{ +struct intel_connector *intel_connector; +struct drm_connector *connector; +struct drm_display_mode *mode; +bool verbose_prune = true; +bool reprobe = false; + + intel_connector =container_of (work,typeof (*intel_connector), + modeset_retry_work); + connector = &intel_connector->base; + +/* Grab the locks before changing connector property*/ +mutex_lock (&connector->dev->mode_config.mutex); +DRM_DEBUG_KMS ("[CONNECTOR:%d:%s]\n" , connector->base.id, + connector->name); +list_for_each_entry (mode, &connector->modes, head) { + mode->status =intel_dp_mode_valid (connector, + mode); +if (mode->status !=MODE_OK ) + reprobe = true; + } +drm_mode_prune_invalid (connector->dev, &connector->modes, + verbose_prune); +drivers/gpu/drm/i915/intel_dp.c static bool intel_edp_init_connector(struct intel_dp *intel_dp, · line 5752
Quoted 10 lines
drivers/gpu/drm/i915/intel_dp.c +10
-0+/* Set connector link status to BAD only if modeset required +* for the current mode, if mode list changed then just send uevent +* so that it can reprobe the connectors and validate modes and do +* a modeset on a different valid mode. +*/ +if (!reprobe) { + connector->link_status =DRM_MODE_LINK_STATUS_BAD ; +intel_dp_set_link_status_property (connector, +DRM_MODE_LINK_STATUS_BAD ); + }I think the link status property should be set to bad unconditionally here. If the current link is bad, it is bad right now independent of modes fitting into the fallback link parameters.
Quoted 5 lines
Which makes me think, unless I'm missing something, that you might be able to prune the invalid modes and set the property right away when link training fails, and only use the work to do
drm_kms_helper_hotplug_event.
The problem with setting the link_status property BAD irrespective of the
mode pruning is that, if the modes gets pruned and we set the link_sttaus to BAD Chris Wilson's driver checks that the link status is BAD and he first tries to attempt the modeset at the current mode without calling mode_valid, and that results in a failure in enocder->compute_config since now the mode does not fit and the pipe cannot be configured. This creates a lot of warnings/errors/kernel crash eventually.
So the best way is to set the link status as bad only when we want to force the modeset at the current mode, if the modes get pruned then in any case userspace will do another modeset at the next lower mode.
Manasi
Quoted 56 lines
Quoted 56 lines
drivers/gpu/drm/i915/intel_dp.c +10
-0+mutex_unlock (&connector->dev->mode_config.mutex); + +/* Send Hotplug uevent so userspace can reprobe */ +drm_kms_helper_hotplug_event (connector->dev); +} +bool intel_dp_init_connector (struct intel_digital_port *intel_dig_port,struct intel_connector *intel_connector)@@ -5704,6 +5768,10 @@ static bool intel_edp_init_connector(struct intel_dp *intel_dp,enum port port = intel_dig_port->port;int type; +/* Initialize the work for modeset in case of link train failure */ +INIT_WORK (&intel_connector->modeset_retry_work, + intel_dp_modeset_retry_work_fn); +if (WARN (intel_dig_port->max_lanes <1 ,"Not enough lanes (%d) for DP on port %c\n" , intel_dig_port->max_lanes,port_name (port)))drivers/gpu/drm/i915/intel_dp_link_training.c +9
-3@@ -310,9 +310,15 @@ void intel_dp_stop_link_train(struct intel_dp *intel_dp) DP_TRAINING_PATTERN_DISABLE); } -void +bool intel_dp_start_link_train (struct intel_dp *intel_dp) { -intel_dp_link_training_clock_recovery (intel_dp); -intel_dp_link_training_channel_equalization (intel_dp); +bool ret; + +if (intel_dp_link_training_clock_recovery (intel_dp)) { + ret =intel_dp_link_training_channel_equalization (intel_dp); +if (ret) +return true; + } +return false; }drivers/gpu/drm/i915/intel_drv.h +3
-0@@ -312,6 +312,9 @@ struct intel_connector {void *port;/* store this opaque as its illegal to dereference it */ struct intel_dp *mst_port; + +/* Work struct to schedule a uevent on link train failure */ +struct work_struct modeset_retry_work; };
Quoted 13 lines
Quoted 10 lines
drivers/gpu/drm/i915/intel_drv.h +1
-1struct dpll {@@ -1402,7 +1405,7 @@ void intel_dp_set_link_params(struct intel_dp *intel_dp, bool link_mst);void intel_dp_get_link_train_fallback_values (struct intel_dp *intel_dp,int link_rate,uint8_t lane_count); -void intel_dp_start_link_train (struct intel_dp *intel_dp); +bool intel_dp_start_link_train (struct intel_dp *intel_dp);void intel_dp_stop_link_train (struct intel_dp *intel_dp);void intel_dp_sink_dpms (struct intel_dp *intel_dp,int mode);void intel_dp_encoder_reset (struct drm_encoder *encoder);Signature · 1 line
Jani Nikula, Intel Open Source Technology Center
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Jani Nikula
maintainerRe: [PATCH v2 4/4] drm/i915: Implement Link Rate fallback on Link training failure
Quoted 95 lines · Manasi Navare
Quoted 63 lines · Jani Nikula
Quoted 62 lines · Manasi Navare
If link training at a link rate optimal for a particular mode fails during modeset's atomic commit phase, then we let the modeset complete and then retry. We save the link rate value at which link training failed, update the link status property to "BAD" and use a lower link rate to prune the modes. It will redo the modeset on the current mode at lower link rate or if the current mode gets pruned due to lower link constraints then, it will send a hotplug uevent for userspace to handle it.
This is also required to pass DP CTS tests 4.3.1.3, 4.3.1.4, 4.3.1.6.
v2:
* Squashed a few patches (Jani Nikula)Cc Jani Nikula <[email protected]>Cc Daniel Vetter <[email protected]>Cc Ville Syrjala <[email protected]>Signed-off-by Manasi Navare <[email protected]>--- drivers/gpu/drm/drm_atomic_helper.c | 4 ++ drivers/gpu/drm/i915/intel_ddi.c | 23 ++++++++- drivers/gpu/drm/i915/intel_dp.c | 74 +++++++++++++++++++++++++-- drivers/gpu/drm/i915/intel_dp_link_training.c | 12 +++-- drivers/gpu/drm/i915/intel_drv.h | 5 +- 5 files changed, 110 insertions(+), 8 deletions(-)drivers/gpu/drm/drm_atomic_helper.c +4
-0@@ -519,6 +519,10 @@ static int handle_conflicting_encoders(struct drm_atomic_state *state, connector_state);if (ret)return ret; + + crtc_state =drm_atomic_get_existing_crtc_state (state, connector->state->crtc); +if (connector->link_status ==DRM_MODE_LINK_STATUS_BAD ) + crtc_state->connectors_changed = true; }/* drivers/gpu/drm/i915/intel_ddi.c +4
-1@@ -1684,6 +1684,8 @@ static void intel_ddi_pre_enable_dp(struct intel_encoder *encoder,struct intel_dp *intel_dp =enc_to_intel_dp (&encoder->base);struct drm_i915_private *dev_priv =to_i915 (encoder->base.dev);enum port port =intel_ddi_get_encoder_port (encoder); +struct intel_connector *intel_connector = intel_dp->attached_connector; +struct drm_connector *connector = &intel_connector->base;intel_dp_set_link_params (intel_dp, link_rate, lane_count, link_mst);@@ -1694,7 +1696,26 @@ static void intel_ddi_pre_enable_dp(struct intel_encoder *encoder,intel_prepare_dp_ddi_buffers (encoder);intel_ddi_init_dp_buf_reg (encoder);intel_dp_sink_dpms (intel_dp,DRM_MODE_DPMS_ON ); -intel_dp_start_link_train (intel_dp); +if (!intel_dp_start_link_train (intel_dp)) { + DRM_DEBUG_KMS("Link Training failed at link rate = %d, lane count = %d",Quoted 9 lines
drivers/gpu/drm/i915/intel_ddi.c static void intel_ddi_pre_enable_dp(struct intel_encoder *encoder, · line 1706
Quoted 6 lines
drivers/gpu/drm/i915/intel_ddi.c +6
-0+ link_rate, lane_count); + intel_dp->link_train_failed = true; +intel_dp_get_link_train_fallback_values (intel_dp, link_rate, + lane_count); +/* Schedule a Hotplug Uevent to userspace to start modeset */ +schedule_work (&intel_connector->modeset_retry_work);This is not just about DDI. Need to do this for the other cases too.
Yes, first series will g out for adding this support for DDI, then more patches to expand it to non DDI platforms.
Quoted 7 lines
Quoted 7 lines
drivers/gpu/drm/i915/intel_ddi.c +7
-0+ }else { +DRM_DEBUG_KMS ("Link Training Passed at Link Rate = %d, Lane count = %d" , + link_rate, lane_count); + intel_dp->link_train_failed = false; + intel_dp->fallback_link_rate_index =-1 ; + intel_dp->fallback_link_rate =0 ; + intel_dp->fallback_lane_count =0 ;Quoted 9 lines
drivers/gpu/drm/i915/intel_ddi.c static void intel_ddi_pre_enable_dp(struct intel_encoder *encoder, · line 1716
Quoted 3 lines
drivers/gpu/drm/i915/intel_ddi.c +3
-0+ connector->link_status =DRM_MODE_LINK_STATUS_GOOD ; +intel_dp_set_link_status_property (connector, +DRM_MODE_LINK_STATUS_GOOD );Looks like you never actually read
connector->link_status... Why do you need bothconnector->link_statusandintel_dp->link_train_failed? Do you think you have 4 states? What are they? Can't this all be in sync with the property?
Quoted 10 lines
This
connector->link_statusmember ofdrm_Connectorgets read indrm_atomic_helper_check_modeset()in the driver where it reads this
and setscrtc_state->connector_Changedto true if thislink_status
has changed.
This is required so that the driver does a complete modeset.
Thisconnector->link_statuswas in sync with the property. But reading thedrm_objectproperty indrm_atomic_helper_Check_modesetwas causing the system to not boot.intel_dp->link_train_failedalso just indicates if the link failed, I will have to see if i can just useconnector->link_statusfor this purpose.
Please do. Usually if you add more than one variable for essentially the same thing, you'll end up having combinations of the variables (states) that you should not be in. At least you should have a clear idea what the states are where link_train_failed and link_status disagree.
Quoted 48 lines
Quoted 42 lines
Quoted 28 lines
drivers/gpu/drm/i915/intel_ddi.c +2
-0+ } +if (port !=PORT_A ||INTEL_GEN (dev_priv) >=9 )intel_dp_stop_link_train (intel_dp); }drivers/gpu/drm/i915/intel_dp.c +8
-2@@ -354,8 +354,14 @@ void intel_dp_get_link_train_fallback_values(struct intel_dp *intel_dp, target_clock = fixed_mode->clock; } - max_link_clock =intel_dp_max_link_rate (intel_dp); - max_lanes =intel_dp_max_lane_count (intel_dp); +/* Prune the modes using the fallback link rate/lane count */ +if (intel_dp->link_train_failed) { + max_link_clock = intel_dp->fallback_link_rate; + max_lanes = intel_dp->fallback_lane_count; + }else { + max_link_clock =intel_dp_max_link_rate (intel_dp); + max_lanes =intel_dp_max_lane_count (intel_dp); + } max_rate =intel_dp_max_data_rate (max_link_clock, max_lanes); mode_rate =intel_dp_link_required (target_clock,18 );drivers/gpu/drm/i915/intel_dp.c static int intel_dp_compute_bpp(struct intel_dp *intel_dp, · line 1654
Quoted 9 lines
drivers/gpu/drm/i915/intel_dp.c +6
-0@@ -1640,6 +1646,12 @@ static int intel_dp_compute_bpp(struct intel_dp *intel_dp,if (adjusted_mode->flags &DRM_MODE_FLAG_DBLCLK )return false; +/* Fall back to lower link rate in case of failure in previous modeset */ +if (intel_dp->link_train_failed) { + min_lane_count = max_lane_count = intel_dp->fallback_lane_count; + min_clock = max_clock = intel_dp->fallback_link_rate_index; + } +My general feeling is that there's starting to be a bit too much special casing around the fallback values. I'm not decided we need to fix this right away in this series, or whether it can be follow-up work.
Quoted 5 lines
One idea is to compute the common rates and lanes once when they're first needed, and all of the helpers would use that info. The fallback code would just trim those, and the conditional fallback stuff could be removed from all over the place.
Quoted 6 lines
These fallback values get computed in a separate helper function that I have added
intel_dp_get_link_train_fallback_values. It is the previous patch. I store the fallback values inintel_dpone at a time because for that iteration of modeset we only need to try thefallback_link_rateandfallback_lane_countso we dont need an array here.
Sure. I'm *not* referring to my earlier suggestion of storing failing link rate, lane count pairs (although I think we'll need that eventually).
Quoted 3 lines
Are you suggesting just changing the
common_ratesarray itself to get trimmed to use the trimmed fallback values after link training fails? Could you please elaborate your thought?
I'm not saying you should change this now. But having a lot of code check some fallback stuff is ugly and error prone. I'm just documenting the ideas to make this better.
I think this could be fixed by storing the common rates array and max lanes in intel_dp, instead of having them locally in a few functions, and then making the link rate and lane counting functions aware of the fallback stuff. Either the functions would check for the fallbacks, centralizing the checks in one place, and/or the fallback code would modify the common values stored in intel_dp, so that the link rate and lane counting functions would just work with the fallbacks. Code would be simpler overall.
Quoted 83 lines
Quoted 77 lines
Quoted 61 lines
drivers/gpu/drm/i915/intel_dp.c +37
-1DRM_DEBUG_KMS("DP link computation with max lane count %i " "max bw %d pixel clock %iKHz\n", max_lane_count, common_rates[max_clock],@@ -4423,6 +4435,13 @@ static bool intel_digital_port_connected(struct drm_i915_private *dev_priv, intel_dp->compliance_test_active =0 ; intel_dp->compliance_test_type =0 ; intel_dp->compliance_test_data =0 ; + intel_dp->link_train_failed = false; + intel_dp->fallback_link_rate_index =-1 ; + intel_dp->fallback_link_rate =0 ; + intel_dp->fallback_lane_count =0 ; + connector->link_status =DRM_MODE_LINK_STATUS_GOOD ; +intel_dp_set_link_status_property (connector, +DRM_MODE_LINK_STATUS_GOOD );if (intel_dp->is_mst) { DRM_DEBUG_KMS("MST device may have disappeared %d vs %d\n",@@ -4514,8 +4533,12 @@ static bool intel_digital_port_connected(struct drm_i915_private *dev_priv,DRM_DEBUG_KMS ("[CONNECTOR:%d:%s]\n" , connector->base.id, connector->name); +/* If this is a retry due to link trianing failure */ +if (status == connector_status_connected && intel_dp->link_train_failed) +return status; +/* If full detect is not performed yet, do a full detect */ -if (!intel_dp->detect_done) +if (!intel_dp->detect_done && !intel_dp->link_train_failed) status =intel_dp_long_pulse (intel_dp->attached_connector); intel_dp->detect_done = false;@@ -5692,6 +5715,47 @@ static bool intel_edp_init_connector(struct intel_dp *intel_dp,return false; } +static void intel_dp_modeset_retry_work_fn (struct work_struct *work) +{ +struct intel_connector *intel_connector; +struct drm_connector *connector; +struct drm_display_mode *mode; +bool verbose_prune = true; +bool reprobe = false; + + intel_connector =container_of (work,typeof (*intel_connector), + modeset_retry_work); + connector = &intel_connector->base; + +/* Grab the locks before changing connector property*/ +mutex_lock (&connector->dev->mode_config.mutex); +DRM_DEBUG_KMS ("[CONNECTOR:%d:%s]\n" , connector->base.id, + connector->name); +list_for_each_entry (mode, &connector->modes, head) { + mode->status =intel_dp_mode_valid (connector, + mode); +if (mode->status !=MODE_OK ) + reprobe = true; + } +drm_mode_prune_invalid (connector->dev, &connector->modes, + verbose_prune); +drivers/gpu/drm/i915/intel_dp.c static bool intel_edp_init_connector(struct intel_dp *intel_dp, · line 5752
Quoted 10 lines
drivers/gpu/drm/i915/intel_dp.c +10
-0+/* Set connector link status to BAD only if modeset required +* for the current mode, if mode list changed then just send uevent +* so that it can reprobe the connectors and validate modes and do +* a modeset on a different valid mode. +*/ +if (!reprobe) { + connector->link_status =DRM_MODE_LINK_STATUS_BAD ; +intel_dp_set_link_status_property (connector, +DRM_MODE_LINK_STATUS_BAD ); + }I think the link status property should be set to bad unconditionally here. If the current link is bad, it is bad right now independent of modes fitting into the fallback link parameters.
Quoted 5 lines
Which makes me think, unless I'm missing something, that you might be able to prune the invalid modes and set the property right away when link training fails, and only use the work to do
drm_kms_helper_hotplug_event.
Quoted 11 lines
The problem with setting the
link_statusproperty BAD irrespective of the mode pruning is that, if the modes gets pruned and we set thelink_sttausto BAD Chris Wilson's driver checks that the link status is BAD and he first tries to attempt the modeset at the current mode without callingmode_valid, and that results in a failure inenocder->compute_configsince now the mode does not fit and the pipe cannot be configured. This creates a lot of warnings/errors/kernel crash eventually. So the best way is to set the link status as bad only when we want to force the modeset at the current mode, if the modes get pruned then in any case userspace will do another modeset at the next lower mode.
Why would the userspace driver retry the same mode without refreshing the mode list first if link status is bad?
If the link is bad, the link is *bad*, regardless of whether the current mode might eventually work or not. An interface that reports good on some values of bad is inconsistent (related reading [1]). IMO the userspace should first figure out if the current mode is valid or not, and then decide what to do.
Chris, any comments?
Side note, the kernel must not crash depending on what the userspace does.
BR,
Jani.
[1] http://sweng.the-davies.net/Home/rustys-api-design-manifesto
Quoted 71 lines
Manasi
Quoted 69 lines
Quoted 66 lines
drivers/gpu/drm/i915/intel_dp.c +10
-0+mutex_unlock (&connector->dev->mode_config.mutex); + +/* Send Hotplug uevent so userspace can reprobe */ +drm_kms_helper_hotplug_event (connector->dev); +} +bool intel_dp_init_connector (struct intel_digital_port *intel_dig_port,struct intel_connector *intel_connector)@@ -5704,6 +5768,10 @@ static bool intel_edp_init_connector(struct intel_dp *intel_dp,enum port port = intel_dig_port->port;int type; +/* Initialize the work for modeset in case of link train failure */ +INIT_WORK (&intel_connector->modeset_retry_work, + intel_dp_modeset_retry_work_fn); +if (WARN (intel_dig_port->max_lanes <1 ,"Not enough lanes (%d) for DP on port %c\n" , intel_dig_port->max_lanes,port_name (port)))drivers/gpu/drm/i915/intel_dp_link_training.c +9
-3@@ -310,9 +310,15 @@ void intel_dp_stop_link_train(struct intel_dp *intel_dp) DP_TRAINING_PATTERN_DISABLE); } -void +bool intel_dp_start_link_train (struct intel_dp *intel_dp) { -intel_dp_link_training_clock_recovery (intel_dp); -intel_dp_link_training_channel_equalization (intel_dp); +bool ret; + +if (intel_dp_link_training_clock_recovery (intel_dp)) { + ret =intel_dp_link_training_channel_equalization (intel_dp); +if (ret) +return true; + } +return false; }drivers/gpu/drm/i915/intel_drv.h +4
-1@@ -312,6 +312,9 @@ struct intel_connector {void *port;/* store this opaque as its illegal to dereference it */ struct intel_dp *mst_port; + +/* Work struct to schedule a uevent on link train failure */ +struct work_struct modeset_retry_work; };struct dpll {@@ -1402,7 +1405,7 @@ void intel_dp_set_link_params(struct intel_dp *intel_dp, bool link_mst);void intel_dp_get_link_train_fallback_values (struct intel_dp *intel_dp,int link_rate,uint8_t lane_count); -void intel_dp_start_link_train (struct intel_dp *intel_dp); +bool intel_dp_start_link_train (struct intel_dp *intel_dp);void intel_dp_stop_link_train (struct intel_dp *intel_dp);void intel_dp_sink_dpms (struct intel_dp *intel_dp,int mode);void intel_dp_encoder_reset (struct drm_encoder *encoder);Signature · 1 line
Jani Nikula, Intel Open Source Technology Center
Signature · 5 lines
Jani Nikula, Intel Open Source Technology Center _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH v2 4/4] drm/i915: Implement Link Rate fallback on Link training failure
Quoted 107 lines · Jani Nikula
Quoted 95 lines · Manasi Navare
Quoted 63 lines · Jani Nikula
Quoted 62 lines · Manasi Navare
If link training at a link rate optimal for a particular mode fails during modeset's atomic commit phase, then we let the modeset complete and then retry. We save the link rate value at which link training failed, update the link status property to "BAD" and use a lower link rate to prune the modes. It will redo the modeset on the current mode at lower link rate or if the current mode gets pruned due to lower link constraints then, it will send a hotplug uevent for userspace to handle it.
This is also required to pass DP CTS tests 4.3.1.3, 4.3.1.4, 4.3.1.6.
v2:
* Squashed a few patches (Jani Nikula)Cc Jani Nikula <[email protected]>Cc Daniel Vetter <[email protected]>Cc Ville Syrjala <[email protected]>Signed-off-by Manasi Navare <[email protected]>--- drivers/gpu/drm/drm_atomic_helper.c | 4 ++ drivers/gpu/drm/i915/intel_ddi.c | 23 ++++++++- drivers/gpu/drm/i915/intel_dp.c | 74 +++++++++++++++++++++++++-- drivers/gpu/drm/i915/intel_dp_link_training.c | 12 +++-- drivers/gpu/drm/i915/intel_drv.h | 5 +- 5 files changed, 110 insertions(+), 8 deletions(-)drivers/gpu/drm/drm_atomic_helper.c +4
-0@@ -519,6 +519,10 @@ static int handle_conflicting_encoders(struct drm_atomic_state *state, connector_state);if (ret)return ret; + + crtc_state =drm_atomic_get_existing_crtc_state (state, connector->state->crtc); +if (connector->link_status ==DRM_MODE_LINK_STATUS_BAD ) + crtc_state->connectors_changed = true; }/* drivers/gpu/drm/i915/intel_ddi.c +4
-1@@ -1684,6 +1684,8 @@ static void intel_ddi_pre_enable_dp(struct intel_encoder *encoder,struct intel_dp *intel_dp =enc_to_intel_dp (&encoder->base);struct drm_i915_private *dev_priv =to_i915 (encoder->base.dev);enum port port =intel_ddi_get_encoder_port (encoder); +struct intel_connector *intel_connector = intel_dp->attached_connector; +struct drm_connector *connector = &intel_connector->base;intel_dp_set_link_params (intel_dp, link_rate, lane_count, link_mst);@@ -1694,7 +1696,26 @@ static void intel_ddi_pre_enable_dp(struct intel_encoder *encoder,intel_prepare_dp_ddi_buffers (encoder);intel_ddi_init_dp_buf_reg (encoder);intel_dp_sink_dpms (intel_dp,DRM_MODE_DPMS_ON ); -intel_dp_start_link_train (intel_dp); +if (!intel_dp_start_link_train (intel_dp)) { + DRM_DEBUG_KMS("Link Training failed at link rate = %d, lane count = %d",Quoted 9 lines
drivers/gpu/drm/i915/intel_ddi.c static void intel_ddi_pre_enable_dp(struct intel_encoder *encoder, · line 1706
Quoted 6 lines
drivers/gpu/drm/i915/intel_ddi.c +6
-0+ link_rate, lane_count); + intel_dp->link_train_failed = true; +intel_dp_get_link_train_fallback_values (intel_dp, link_rate, + lane_count); +/* Schedule a Hotplug Uevent to userspace to start modeset */ +schedule_work (&intel_connector->modeset_retry_work);This is not just about DDI. Need to do this for the other cases too.
Yes, first series will g out for adding this support for DDI, then more patches to expand it to non DDI platforms.
Quoted 7 lines
Quoted 7 lines
drivers/gpu/drm/i915/intel_ddi.c +7
-0+ }else { +DRM_DEBUG_KMS ("Link Training Passed at Link Rate = %d, Lane count = %d" , + link_rate, lane_count); + intel_dp->link_train_failed = false; + intel_dp->fallback_link_rate_index =-1 ; + intel_dp->fallback_link_rate =0 ; + intel_dp->fallback_lane_count =0 ;Quoted 9 lines
drivers/gpu/drm/i915/intel_ddi.c static void intel_ddi_pre_enable_dp(struct intel_encoder *encoder, · line 1716
Quoted 3 lines
drivers/gpu/drm/i915/intel_ddi.c +3
-0+ connector->link_status =DRM_MODE_LINK_STATUS_GOOD ; +intel_dp_set_link_status_property (connector, +DRM_MODE_LINK_STATUS_GOOD );Looks like you never actually read
connector->link_status... Why do you need bothconnector->link_statusandintel_dp->link_train_failed? Do you think you have 4 states? What are they? Can't this all be in sync with the property?Quoted 10 lines
This
connector->link_statusmember ofdrm_Connectorgets read indrm_atomic_helper_check_modeset()in the driver where it reads this
and setscrtc_state->connector_Changedto true if thislink_status
has changed.
This is required so that the driver does a complete modeset.
Thisconnector->link_statuswas in sync with the property. But reading thedrm_objectproperty indrm_atomic_helper_Check_modesetwas causing the system to not boot.intel_dp->link_train_failedalso just indicates if the link failed, I will have to see if i can just useconnector->link_statusfor this purpose.
Quoted 5 lines
Please do. Usually if you add more than one variable for essentially the same thing, you'll end up having combinations of the variables (states) that you should not be in. At least you should have a clear idea what the states are where
link_train_failedandlink_statusdisagree.
I will try to converge these.
How do we handle the case where it has tried all possible fallback values, lowest link rate and lowest lane count, now the link is still bad, DPR expects driver to stop link training with ERROR that link training is unsuccessful even after trying fallback. In this case we should not try to retrain again and in that case I was just setting link_train_failed to false again. Do you have any suggestions on how to handle this case, do we send the uevent in this case as well? But then we dont have fallback values since we have exhausted all fallback values.
Quoted 67 lines
Quoted 48 lines
Quoted 42 lines
Quoted 28 lines
drivers/gpu/drm/i915/intel_ddi.c +2
-0+ } +if (port !=PORT_A ||INTEL_GEN (dev_priv) >=9 )intel_dp_stop_link_train (intel_dp); }drivers/gpu/drm/i915/intel_dp.c +8
-2@@ -354,8 +354,14 @@ void intel_dp_get_link_train_fallback_values(struct intel_dp *intel_dp, target_clock = fixed_mode->clock; } - max_link_clock =intel_dp_max_link_rate (intel_dp); - max_lanes =intel_dp_max_lane_count (intel_dp); +/* Prune the modes using the fallback link rate/lane count */ +if (intel_dp->link_train_failed) { + max_link_clock = intel_dp->fallback_link_rate; + max_lanes = intel_dp->fallback_lane_count; + }else { + max_link_clock =intel_dp_max_link_rate (intel_dp); + max_lanes =intel_dp_max_lane_count (intel_dp); + } max_rate =intel_dp_max_data_rate (max_link_clock, max_lanes); mode_rate =intel_dp_link_required (target_clock,18 );drivers/gpu/drm/i915/intel_dp.c static int intel_dp_compute_bpp(struct intel_dp *intel_dp, · line 1654
Quoted 9 lines
drivers/gpu/drm/i915/intel_dp.c +6
-0@@ -1640,6 +1646,12 @@ static int intel_dp_compute_bpp(struct intel_dp *intel_dp,if (adjusted_mode->flags &DRM_MODE_FLAG_DBLCLK )return false; +/* Fall back to lower link rate in case of failure in previous modeset */ +if (intel_dp->link_train_failed) { + min_lane_count = max_lane_count = intel_dp->fallback_lane_count; + min_clock = max_clock = intel_dp->fallback_link_rate_index; + } +My general feeling is that there's starting to be a bit too much special casing around the fallback values. I'm not decided we need to fix this right away in this series, or whether it can be follow-up work.
Quoted 5 lines
One idea is to compute the common rates and lanes once when they're first needed, and all of the helpers would use that info. The fallback code would just trim those, and the conditional fallback stuff could be removed from all over the place.
Quoted 6 lines
These fallback values get computed in a separate helper function that I have added
intel_dp_get_link_train_fallback_values. It is the previous patch. I store the fallback values inintel_dpone at a time because for that iteration of modeset we only need to try thefallback_link_rateandfallback_lane_countso we dont need an array here.Sure. I'm *not* referring to my earlier suggestion of storing failing link rate, lane count pairs (although I think we'll need that eventually).
Quoted 3 lines
Are you suggesting just changing the
common_ratesarray itself to get trimmed to use the trimmed fallback values after link training fails? Could you please elaborate your thought?I'm not saying you should change this now. But having a lot of code check some fallback stuff is ugly and error prone. I'm just documenting the ideas to make this better.
Quoted 9 lines
I think this could be fixed by storing the common rates array and max lanes in
intel_dp, instead of having them locally in a few functions, and then making the link rate and lane counting functions aware of the fallback stuff. Either the functions would check for the fallbacks, centralizing the checks in one place, and/or the fallback code would modify the common values stored inintel_dp, so that the link rate and lane counting functions would just work with the fallbacks. Code would be simpler overall.
Again this might involve a lot more changes since we would have to change all the helper functions that curerntly calculate max lanes and common_rates but this can definitely be done in the next round of clean up after these patches land. But nevertheless I will see if I can update the common_rates array now with fallback values instead of storing fallbck values separately in the intel_dp structure.
Manasi
Quoted 104 lines
Quoted 83 lines
Quoted 77 lines
Quoted 61 lines
drivers/gpu/drm/i915/intel_dp.c +37
-1DRM_DEBUG_KMS("DP link computation with max lane count %i " "max bw %d pixel clock %iKHz\n", max_lane_count, common_rates[max_clock],@@ -4423,6 +4435,13 @@ static bool intel_digital_port_connected(struct drm_i915_private *dev_priv, intel_dp->compliance_test_active =0 ; intel_dp->compliance_test_type =0 ; intel_dp->compliance_test_data =0 ; + intel_dp->link_train_failed = false; + intel_dp->fallback_link_rate_index =-1 ; + intel_dp->fallback_link_rate =0 ; + intel_dp->fallback_lane_count =0 ; + connector->link_status =DRM_MODE_LINK_STATUS_GOOD ; +intel_dp_set_link_status_property (connector, +DRM_MODE_LINK_STATUS_GOOD );if (intel_dp->is_mst) { DRM_DEBUG_KMS("MST device may have disappeared %d vs %d\n",@@ -4514,8 +4533,12 @@ static bool intel_digital_port_connected(struct drm_i915_private *dev_priv,DRM_DEBUG_KMS ("[CONNECTOR:%d:%s]\n" , connector->base.id, connector->name); +/* If this is a retry due to link trianing failure */ +if (status == connector_status_connected && intel_dp->link_train_failed) +return status; +/* If full detect is not performed yet, do a full detect */ -if (!intel_dp->detect_done) +if (!intel_dp->detect_done && !intel_dp->link_train_failed) status =intel_dp_long_pulse (intel_dp->attached_connector); intel_dp->detect_done = false;@@ -5692,6 +5715,47 @@ static bool intel_edp_init_connector(struct intel_dp *intel_dp,return false; } +static void intel_dp_modeset_retry_work_fn (struct work_struct *work) +{ +struct intel_connector *intel_connector; +struct drm_connector *connector; +struct drm_display_mode *mode; +bool verbose_prune = true; +bool reprobe = false; + + intel_connector =container_of (work,typeof (*intel_connector), + modeset_retry_work); + connector = &intel_connector->base; + +/* Grab the locks before changing connector property*/ +mutex_lock (&connector->dev->mode_config.mutex); +DRM_DEBUG_KMS ("[CONNECTOR:%d:%s]\n" , connector->base.id, + connector->name); +list_for_each_entry (mode, &connector->modes, head) { + mode->status =intel_dp_mode_valid (connector, + mode); +if (mode->status !=MODE_OK ) + reprobe = true; + } +drm_mode_prune_invalid (connector->dev, &connector->modes, + verbose_prune); +drivers/gpu/drm/i915/intel_dp.c static bool intel_edp_init_connector(struct intel_dp *intel_dp, · line 5752
Quoted 10 lines
drivers/gpu/drm/i915/intel_dp.c +10
-0+/* Set connector link status to BAD only if modeset required +* for the current mode, if mode list changed then just send uevent +* so that it can reprobe the connectors and validate modes and do +* a modeset on a different valid mode. +*/ +if (!reprobe) { + connector->link_status =DRM_MODE_LINK_STATUS_BAD ; +intel_dp_set_link_status_property (connector, +DRM_MODE_LINK_STATUS_BAD ); + }I think the link status property should be set to bad unconditionally here. If the current link is bad, it is bad right now independent of modes fitting into the fallback link parameters.
Quoted 5 lines
Which makes me think, unless I'm missing something, that you might be able to prune the invalid modes and set the property right away when link training fails, and only use the work to do
drm_kms_helper_hotplug_event.Quoted 11 lines
The problem with setting the
link_statusproperty BAD irrespective of the mode pruning is that, if the modes gets pruned and we set thelink_sttausto BAD Chris Wilson's driver checks that the link status is BAD and he first tries to attempt the modeset at the current mode without callingmode_valid, and that results in a failure inenocder->compute_configsince now the mode does not fit and the pipe cannot be configured. This creates a lot of warnings/errors/kernel crash eventually. So the best way is to set the link status as bad only when we want to force the modeset at the current mode, if the modes get pruned then in any case userspace will do another modeset at the next lower mode.Why would the userspace driver retry the same mode without refreshing the mode list first if link status is bad?
If the link is bad, the link is *bad*, regardless of whether the current mode might eventually work or not. An interface that reports good on some values of bad is inconsistent (related reading [1]). IMO the userspace should first figure out if the current mode is valid or not, and then decide what to do.
Quoted 8 lines
Chris, any comments?
Side note, the kernel must not crash depending on what the userspace does.
BR,
Jani.
Chris, yes I think it is a good idea to always call mode_valid first on recieveing link_status as BAD and then calling setcrtc to set the current mode or whatever valid mode is. Then I would not need to set link status BAD conditionally in the driver. Could you make this change in your link_status patch and resend it to me?
Manasi
Quoted 61 lines
[1] http://sweng.the-davies.net/Home/rustys-api-design-manifesto
Quoted 58 lines
Manasi
Quoted 56 lines
Quoted 56 lines
drivers/gpu/drm/i915/intel_dp.c +10
-0+mutex_unlock (&connector->dev->mode_config.mutex); + +/* Send Hotplug uevent so userspace can reprobe */ +drm_kms_helper_hotplug_event (connector->dev); +} +bool intel_dp_init_connector (struct intel_digital_port *intel_dig_port,struct intel_connector *intel_connector)@@ -5704,6 +5768,10 @@ static bool intel_edp_init_connector(struct intel_dp *intel_dp,enum port port = intel_dig_port->port;int type; +/* Initialize the work for modeset in case of link train failure */ +INIT_WORK (&intel_connector->modeset_retry_work, + intel_dp_modeset_retry_work_fn); +if (WARN (intel_dig_port->max_lanes <1 ,"Not enough lanes (%d) for DP on port %c\n" , intel_dig_port->max_lanes,port_name (port)))drivers/gpu/drm/i915/intel_dp_link_training.c +9
-3@@ -310,9 +310,15 @@ void intel_dp_stop_link_train(struct intel_dp *intel_dp) DP_TRAINING_PATTERN_DISABLE); } -void +bool intel_dp_start_link_train (struct intel_dp *intel_dp) { -intel_dp_link_training_clock_recovery (intel_dp); -intel_dp_link_training_channel_equalization (intel_dp); +bool ret; + +if (intel_dp_link_training_clock_recovery (intel_dp)) { + ret =intel_dp_link_training_channel_equalization (intel_dp); +if (ret) +return true; + } +return false; }drivers/gpu/drm/i915/intel_drv.h +3
-0@@ -312,6 +312,9 @@ struct intel_connector {void *port;/* store this opaque as its illegal to dereference it */ struct intel_dp *mst_port; + +/* Work struct to schedule a uevent on link train failure */ +struct work_struct modeset_retry_work; };
Quoted 16 lines
Quoted 13 lines
Quoted 13 lines
Quoted 10 lines
drivers/gpu/drm/i915/intel_drv.h +1
-1struct dpll {@@ -1402,7 +1405,7 @@ void intel_dp_set_link_params(struct intel_dp *intel_dp, bool link_mst);void intel_dp_get_link_train_fallback_values (struct intel_dp *intel_dp,int link_rate,uint8_t lane_count); -void intel_dp_start_link_train (struct intel_dp *intel_dp); +bool intel_dp_start_link_train (struct intel_dp *intel_dp);void intel_dp_stop_link_train (struct intel_dp *intel_dp);void intel_dp_sink_dpms (struct intel_dp *intel_dp,int mode);void intel_dp_encoder_reset (struct drm_encoder *encoder);Signature · 1 line
Jani Nikula, Intel Open Source Technology CenterSignature · 1 line
Jani Nikula, Intel Open Source Technology Center
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH v2 4/4] drm/i915: Implement Link Rate fallback on Link training failure
Quoted 323 lines · Manasi Navare
Quoted 107 lines · Jani Nikula
Quoted 95 lines · Manasi Navare
Quoted 63 lines · Jani Nikula
Quoted 62 lines · Manasi Navare
If link training at a link rate optimal for a particular mode fails during modeset's atomic commit phase, then we let the modeset complete and then retry. We save the link rate value at which link training failed, update the link status property to "BAD" and use a lower link rate to prune the modes. It will redo the modeset on the current mode at lower link rate or if the current mode gets pruned due to lower link constraints then, it will send a hotplug uevent for userspace to handle it.
This is also required to pass DP CTS tests 4.3.1.3, 4.3.1.4, 4.3.1.6.
v2:
* Squashed a few patches (Jani Nikula)Cc Jani Nikula <[email protected]>Cc Daniel Vetter <[email protected]>Cc Ville Syrjala <[email protected]>Signed-off-by Manasi Navare <[email protected]>--- drivers/gpu/drm/drm_atomic_helper.c | 4 ++ drivers/gpu/drm/i915/intel_ddi.c | 23 ++++++++- drivers/gpu/drm/i915/intel_dp.c | 74 +++++++++++++++++++++++++-- drivers/gpu/drm/i915/intel_dp_link_training.c | 12 +++-- drivers/gpu/drm/i915/intel_drv.h | 5 +- 5 files changed, 110 insertions(+), 8 deletions(-)drivers/gpu/drm/drm_atomic_helper.c +4
-0@@ -519,6 +519,10 @@ static int handle_conflicting_encoders(struct drm_atomic_state *state, connector_state);if (ret)return ret; + + crtc_state =drm_atomic_get_existing_crtc_state (state, connector->state->crtc); +if (connector->link_status ==DRM_MODE_LINK_STATUS_BAD ) + crtc_state->connectors_changed = true; }/* drivers/gpu/drm/i915/intel_ddi.c +4
-1@@ -1684,6 +1684,8 @@ static void intel_ddi_pre_enable_dp(struct intel_encoder *encoder,struct intel_dp *intel_dp =enc_to_intel_dp (&encoder->base);struct drm_i915_private *dev_priv =to_i915 (encoder->base.dev);enum port port =intel_ddi_get_encoder_port (encoder); +struct intel_connector *intel_connector = intel_dp->attached_connector; +struct drm_connector *connector = &intel_connector->base;intel_dp_set_link_params (intel_dp, link_rate, lane_count, link_mst);@@ -1694,7 +1696,26 @@ static void intel_ddi_pre_enable_dp(struct intel_encoder *encoder,intel_prepare_dp_ddi_buffers (encoder);intel_ddi_init_dp_buf_reg (encoder);intel_dp_sink_dpms (intel_dp,DRM_MODE_DPMS_ON ); -intel_dp_start_link_train (intel_dp); +if (!intel_dp_start_link_train (intel_dp)) { + DRM_DEBUG_KMS("Link Training failed at link rate = %d, lane count = %d",Quoted 9 lines
drivers/gpu/drm/i915/intel_ddi.c static void intel_ddi_pre_enable_dp(struct intel_encoder *encoder, · line 1706
Quoted 6 lines
drivers/gpu/drm/i915/intel_ddi.c +6
-0+ link_rate, lane_count); + intel_dp->link_train_failed = true; +intel_dp_get_link_train_fallback_values (intel_dp, link_rate, + lane_count); +/* Schedule a Hotplug Uevent to userspace to start modeset */ +schedule_work (&intel_connector->modeset_retry_work);This is not just about DDI. Need to do this for the other cases too.
Yes, first series will g out for adding this support for DDI, then more patches to expand it to non DDI platforms.
Quoted 7 lines
Quoted 7 lines
drivers/gpu/drm/i915/intel_ddi.c +7
-0+ }else { +DRM_DEBUG_KMS ("Link Training Passed at Link Rate = %d, Lane count = %d" , + link_rate, lane_count); + intel_dp->link_train_failed = false; + intel_dp->fallback_link_rate_index =-1 ; + intel_dp->fallback_link_rate =0 ; + intel_dp->fallback_lane_count =0 ;Quoted 9 lines
drivers/gpu/drm/i915/intel_ddi.c static void intel_ddi_pre_enable_dp(struct intel_encoder *encoder, · line 1716
Quoted 3 lines
drivers/gpu/drm/i915/intel_ddi.c +3
-0+ connector->link_status =DRM_MODE_LINK_STATUS_GOOD ; +intel_dp_set_link_status_property (connector, +DRM_MODE_LINK_STATUS_GOOD );Looks like you never actually read
connector->link_status... Why do you need bothconnector->link_statusandintel_dp->link_train_failed? Do you think you have 4 states? What are they? Can't this all be in sync with the property?Quoted 10 lines
This
connector->link_statusmember ofdrm_Connectorgets read indrm_atomic_helper_check_modeset()in the driver where it reads this
and setscrtc_state->connector_Changedto true if thislink_status
has changed.
This is required so that the driver does a complete modeset.
Thisconnector->link_statuswas in sync with the property. But reading thedrm_objectproperty indrm_atomic_helper_Check_modesetwas causing the system to not boot.intel_dp->link_train_failedalso just indicates if the link failed, I will have to see if i can just useconnector->link_statusfor this purpose.Quoted 5 lines
Please do. Usually if you add more than one variable for essentially the same thing, you'll end up having combinations of the variables (states) that you should not be in. At least you should have a clear idea what the states are where
link_train_failedandlink_statusdisagree.I will try to converge these.
How do we handle the case where it has tried all possible fallback values, lowest link rate and lowest lane count, now the link is still bad, DPR expects driver to stop link training with ERROR that link training is unsuccessful even after trying fallback. In this case we should not try to retrain again and in that case I was just settinglink_train_failedto false again. Do you have any suggestions on how to handle this case, do we send the uevent in this case as well? But then we dont have fallback values since we have exhausted all fallback values.Quoted 67 lines
Quoted 48 lines
Quoted 42 lines
Quoted 28 lines
drivers/gpu/drm/i915/intel_ddi.c +2
-0+ } +if (port !=PORT_A ||INTEL_GEN (dev_priv) >=9 )intel_dp_stop_link_train (intel_dp); }drivers/gpu/drm/i915/intel_dp.c +8
-2@@ -354,8 +354,14 @@ void intel_dp_get_link_train_fallback_values(struct intel_dp *intel_dp, target_clock = fixed_mode->clock; } - max_link_clock =intel_dp_max_link_rate (intel_dp); - max_lanes =intel_dp_max_lane_count (intel_dp); +/* Prune the modes using the fallback link rate/lane count */ +if (intel_dp->link_train_failed) { + max_link_clock = intel_dp->fallback_link_rate; + max_lanes = intel_dp->fallback_lane_count; + }else { + max_link_clock =intel_dp_max_link_rate (intel_dp); + max_lanes =intel_dp_max_lane_count (intel_dp); + } max_rate =intel_dp_max_data_rate (max_link_clock, max_lanes); mode_rate =intel_dp_link_required (target_clock,18 );drivers/gpu/drm/i915/intel_dp.c static int intel_dp_compute_bpp(struct intel_dp *intel_dp, · line 1654
Quoted 9 lines
drivers/gpu/drm/i915/intel_dp.c +6
-0@@ -1640,6 +1646,12 @@ static int intel_dp_compute_bpp(struct intel_dp *intel_dp,if (adjusted_mode->flags &DRM_MODE_FLAG_DBLCLK )return false; +/* Fall back to lower link rate in case of failure in previous modeset */ +if (intel_dp->link_train_failed) { + min_lane_count = max_lane_count = intel_dp->fallback_lane_count; + min_clock = max_clock = intel_dp->fallback_link_rate_index; + } +My general feeling is that there's starting to be a bit too much special casing around the fallback values. I'm not decided we need to fix this right away in this series, or whether it can be follow-up work.
Quoted 5 lines
One idea is to compute the common rates and lanes once when they're first needed, and all of the helpers would use that info. The fallback code would just trim those, and the conditional fallback stuff could be removed from all over the place.
Quoted 6 lines
These fallback values get computed in a separate helper function that I have added
intel_dp_get_link_train_fallback_values. It is the previous patch. I store the fallback values inintel_dpone at a time because for that iteration of modeset we only need to try thefallback_link_rateandfallback_lane_countso we dont need an array here.Sure. I'm *not* referring to my earlier suggestion of storing failing link rate, lane count pairs (although I think we'll need that eventually).
Quoted 3 lines
Are you suggesting just changing the
common_ratesarray itself to get trimmed to use the trimmed fallback values after link training fails? Could you please elaborate your thought?I'm not saying you should change this now. But having a lot of code check some fallback stuff is ugly and error prone. I'm just documenting the ideas to make this better.
Quoted 9 lines
I think this could be fixed by storing the common rates array and max lanes in
intel_dp, instead of having them locally in a few functions, and then making the link rate and lane counting functions aware of the fallback stuff. Either the functions would check for the fallbacks, centralizing the checks in one place, and/or the fallback code would modify the common values stored inintel_dp, so that the link rate and lane counting functions would just work with the fallbacks. Code would be simpler overall.Again this might involve a lot more changes since we would have to change all the helper functions that curerntly calculate max lanes and
common_ratesbut this can definitely be done in the next round of clean up after these patches land. But nevertheless I will see if I can update thecommon_ratesarray now with fallback values instead of storing fallbck values separately in theintel_dpstructure.Manasi
Quoted 104 lines
Quoted 83 lines
Quoted 77 lines
Quoted 61 lines
drivers/gpu/drm/i915/intel_dp.c +37
-1DRM_DEBUG_KMS("DP link computation with max lane count %i " "max bw %d pixel clock %iKHz\n", max_lane_count, common_rates[max_clock],@@ -4423,6 +4435,13 @@ static bool intel_digital_port_connected(struct drm_i915_private *dev_priv, intel_dp->compliance_test_active =0 ; intel_dp->compliance_test_type =0 ; intel_dp->compliance_test_data =0 ; + intel_dp->link_train_failed = false; + intel_dp->fallback_link_rate_index =-1 ; + intel_dp->fallback_link_rate =0 ; + intel_dp->fallback_lane_count =0 ; + connector->link_status =DRM_MODE_LINK_STATUS_GOOD ; +intel_dp_set_link_status_property (connector, +DRM_MODE_LINK_STATUS_GOOD );if (intel_dp->is_mst) { DRM_DEBUG_KMS("MST device may have disappeared %d vs %d\n",@@ -4514,8 +4533,12 @@ static bool intel_digital_port_connected(struct drm_i915_private *dev_priv,DRM_DEBUG_KMS ("[CONNECTOR:%d:%s]\n" , connector->base.id, connector->name); +/* If this is a retry due to link trianing failure */ +if (status == connector_status_connected && intel_dp->link_train_failed) +return status; +/* If full detect is not performed yet, do a full detect */ -if (!intel_dp->detect_done) +if (!intel_dp->detect_done && !intel_dp->link_train_failed) status =intel_dp_long_pulse (intel_dp->attached_connector); intel_dp->detect_done = false;@@ -5692,6 +5715,47 @@ static bool intel_edp_init_connector(struct intel_dp *intel_dp,return false; } +static void intel_dp_modeset_retry_work_fn (struct work_struct *work) +{ +struct intel_connector *intel_connector; +struct drm_connector *connector; +struct drm_display_mode *mode; +bool verbose_prune = true; +bool reprobe = false; + + intel_connector =container_of (work,typeof (*intel_connector), + modeset_retry_work); + connector = &intel_connector->base; + +/* Grab the locks before changing connector property*/ +mutex_lock (&connector->dev->mode_config.mutex); +DRM_DEBUG_KMS ("[CONNECTOR:%d:%s]\n" , connector->base.id, + connector->name); +list_for_each_entry (mode, &connector->modes, head) { + mode->status =intel_dp_mode_valid (connector, + mode); +if (mode->status !=MODE_OK ) + reprobe = true; + } +drm_mode_prune_invalid (connector->dev, &connector->modes, + verbose_prune); +drivers/gpu/drm/i915/intel_dp.c static bool intel_edp_init_connector(struct intel_dp *intel_dp, · line 5752
Quoted 10 lines
drivers/gpu/drm/i915/intel_dp.c +10
-0+/* Set connector link status to BAD only if modeset required +* for the current mode, if mode list changed then just send uevent +* so that it can reprobe the connectors and validate modes and do +* a modeset on a different valid mode. +*/ +if (!reprobe) { + connector->link_status =DRM_MODE_LINK_STATUS_BAD ; +intel_dp_set_link_status_property (connector, +DRM_MODE_LINK_STATUS_BAD ); + }I think the link status property should be set to bad unconditionally here. If the current link is bad, it is bad right now independent of modes fitting into the fallback link parameters.
Quoted 5 lines
Which makes me think, unless I'm missing something, that you might be able to prune the invalid modes and set the property right away when link training fails, and only use the work to do
drm_kms_helper_hotplug_event.Quoted 11 lines
The problem with setting the
link_statusproperty BAD irrespective of the mode pruning is that, if the modes gets pruned and we set thelink_sttausto BAD Chris Wilson's driver checks that the link status is BAD and he first tries to attempt the modeset at the current mode without callingmode_valid, and that results in a failure inenocder->compute_configsince now the mode does not fit and the pipe cannot be configured. This creates a lot of warnings/errors/kernel crash eventually. So the best way is to set the link status as bad only when we want to force the modeset at the current mode, if the modes get pruned then in any case userspace will do another modeset at the next lower mode.Why would the userspace driver retry the same mode without refreshing the mode list first if link status is bad?
If the link is bad, the link is *bad*, regardless of whether the current mode might eventually work or not. An interface that reports good on some values of bad is inconsistent (related reading [1]). IMO the userspace should first figure out if the current mode is valid or not, and then decide what to do.
Quoted 8 lines
Chris, any comments?
Side note, the kernel must not crash depending on what the userspace does.
BR,
Jani.
Quoted 6 lines
Chris, yes I think it is a good idea to always call
mode_validfirst on recieveinglink_statusas BAD and then calling setcrtc to set the current mode or whatever valid mode is. Then I would not need to set link status BAD conditionally in the driver. Could you make this change in yourlink_statuspatch and resend it to me?Manasi
Chris,
I was able to further investigate this and I found the problem in my code, where I had to add a check that only if crtc is present, it will try to configure the pipe else it will return EINVAL to userspace at much earlier stage and that fixed the kernel errors (NULL dereferencing) that I was seeing. So Jani, now I can set the link_status to bad unconditionally and send a hotplug uevent. I will make these changes and submit the new revision of patches.
Chris, your link_status patch is good to go!
Manasi
Quoted 76 lines
Quoted 76 lines
[1] http://sweng.the-davies.net/Home/rustys-api-design-manifesto
Quoted 71 lines
Manasi
Quoted 69 lines
Quoted 66 lines
drivers/gpu/drm/i915/intel_dp.c +10
-0+mutex_unlock (&connector->dev->mode_config.mutex); + +/* Send Hotplug uevent so userspace can reprobe */ +drm_kms_helper_hotplug_event (connector->dev); +} +bool intel_dp_init_connector (struct intel_digital_port *intel_dig_port,struct intel_connector *intel_connector)@@ -5704,6 +5768,10 @@ static bool intel_edp_init_connector(struct intel_dp *intel_dp,enum port port = intel_dig_port->port;int type; +/* Initialize the work for modeset in case of link train failure */ +INIT_WORK (&intel_connector->modeset_retry_work, + intel_dp_modeset_retry_work_fn); +if (WARN (intel_dig_port->max_lanes <1 ,"Not enough lanes (%d) for DP on port %c\n" , intel_dig_port->max_lanes,port_name (port)))drivers/gpu/drm/i915/intel_dp_link_training.c +9
-3@@ -310,9 +310,15 @@ void intel_dp_stop_link_train(struct intel_dp *intel_dp) DP_TRAINING_PATTERN_DISABLE); } -void +bool intel_dp_start_link_train (struct intel_dp *intel_dp) { -intel_dp_link_training_clock_recovery (intel_dp); -intel_dp_link_training_channel_equalization (intel_dp); +bool ret; + +if (intel_dp_link_training_clock_recovery (intel_dp)) { + ret =intel_dp_link_training_channel_equalization (intel_dp); +if (ret) +return true; + } +return false; }drivers/gpu/drm/i915/intel_drv.h +4
-1@@ -312,6 +312,9 @@ struct intel_connector {void *port;/* store this opaque as its illegal to dereference it */ struct intel_dp *mst_port; + +/* Work struct to schedule a uevent on link train failure */ +struct work_struct modeset_retry_work; };struct dpll {@@ -1402,7 +1405,7 @@ void intel_dp_set_link_params(struct intel_dp *intel_dp, bool link_mst);void intel_dp_get_link_train_fallback_values (struct intel_dp *intel_dp,int link_rate,uint8_t lane_count); -void intel_dp_start_link_train (struct intel_dp *intel_dp); +bool intel_dp_start_link_train (struct intel_dp *intel_dp);void intel_dp_stop_link_train (struct intel_dp *intel_dp);void intel_dp_sink_dpms (struct intel_dp *intel_dp,int mode);void intel_dp_encoder_reset (struct drm_encoder *encoder);Signature · 1 line
Jani Nikula, Intel Open Source Technology Center
Quoted 6 lines
Quoted 2 lines
Signature · 1 line
Jani Nikula, Intel Open Source Technology Center_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v3 4/4 drm/i915: Implement Link Rate fallback on Link training failure
In reply to Manasi NavareIf link training at a link rate optimal for a particular mode fails during modeset's atomic commit phase, then we let the modeset complete and then retry. We save the link rate value at which link training failed, update the link status property to "BAD" and use a lower link rate to prune the modes. It will redo the modeset on the current mode at lower link rate or if the current mode gets pruned due to lower link constraints then, it will send a hotplug uevent for userspace to handle it.
This is also required to pass DP CTS tests 4.3.1.3, 4.3.1.4, 4.3.1.6.
v3:
* Set link status property to BAd unconditionally (Jani Nikula)
* Dont use two separate variables link_train_failed and link_status to indicate same thing (Jani Nikula)
v2:
* Squashed a few patches (Jani Nikula)
--- drivers/gpu/drm/drm_atomic_helper.c | 7 +++ drivers/gpu/drm/i915/intel_ddi.c | 22 ++++++++- drivers/gpu/drm/i915/intel_dp.c | 70 ++++++++++++++++++++++++--- drivers/gpu/drm/i915/intel_dp_link_training.c | 12 +++-- drivers/gpu/drm/i915/intel_drv.h | 6 ++- 5 files changed, 105 insertions(+), 12 deletions(-)
drivers/gpu/drm/drm_atomic_helper.c +7 -0
@@ -519,6 +519,13 @@ static int handle_conflicting_encoders(struct drm_atomic_state *state, connector_state);if (ret)return ret; + +if (connector->state->crtc) { + crtc_state =drm_atomic_get_existing_crtc_state (state, + connector->state->crtc); +if (connector->link_status ==DRM_MODE_LINK_STATUS_BAD ) + crtc_state->connectors_changed = true; + } }/*
drivers/gpu/drm/i915/intel_ddi.c +21 -1
@@ -1684,6 +1684,8 @@ static void intel_ddi_pre_enable_dp(struct intel_encoder *encoder,struct intel_dp *intel_dp =enc_to_intel_dp (&encoder->base);struct drm_i915_private *dev_priv =to_i915 (encoder->base.dev);enum port port =intel_ddi_get_encoder_port (encoder); +struct intel_connector *intel_connector = intel_dp->attached_connector; +struct drm_connector *connector = &intel_connector->base;intel_dp_set_link_params (intel_dp, link_rate, lane_count, link_mst);
@@ -1694,7 +1696,25 @@ static void intel_ddi_pre_enable_dp(struct intel_encoder *encoder,intel_prepare_dp_ddi_buffers (encoder);intel_ddi_init_dp_buf_reg (encoder);intel_dp_sink_dpms (intel_dp,DRM_MODE_DPMS_ON ); -intel_dp_start_link_train (intel_dp); +if (!intel_dp_start_link_train (intel_dp)) { +DRM_DEBUG_KMS ("Link Training failed at link rate = %d, lane count = %d" , + link_rate, lane_count); + connector->link_status =DRM_MODE_LINK_STATUS_BAD ; +intel_dp_get_link_train_fallback_values (intel_dp, link_rate, + lane_count); +/* Schedule a Hotplug Uevent to userspace to start modeset */ +schedule_work (&intel_connector->modeset_retry_work); + }else { +DRM_DEBUG_KMS ("Link Training Passed at Link Rate = %d, Lane count = %d" , + link_rate, lane_count); + intel_dp->fallback_link_rate_index =-1 ; + intel_dp->fallback_link_rate =0 ; + intel_dp->fallback_lane_count =0 ; + connector->link_status =DRM_MODE_LINK_STATUS_GOOD ; +intel_dp_set_link_status_property (connector, +DRM_MODE_LINK_STATUS_GOOD ); + } +if (port !=PORT_A ||INTEL_GEN (dev_priv) >=9 )intel_dp_stop_link_train (intel_dp); }
drivers/gpu/drm/i915/intel_dp.c +64 -6
124 lines not on this page · raw message
drivers/gpu/drm/i915/intel_dp_link_training.c +9 -3
@@ -310,9 +310,15 @@ void intel_dp_stop_link_train(struct intel_dp *intel_dp) DP_TRAINING_PATTERN_DISABLE); } -void +bool intel_dp_start_link_train (struct intel_dp *intel_dp) { -intel_dp_link_training_clock_recovery (intel_dp); -intel_dp_link_training_channel_equalization (intel_dp); +bool ret; + +if (intel_dp_link_training_clock_recovery (intel_dp)) { + ret =intel_dp_link_training_channel_equalization (intel_dp); +if (ret) +return true; + } +return false; }
drivers/gpu/drm/i915/intel_drv.h +4 -2
@@ -312,6 +312,9 @@ struct intel_connector {void *port;/* store this opaque as its illegal to dereference it */ struct intel_dp *mst_port; + +/* Work struct to schedule a uevent on link train failure */ +struct work_struct modeset_retry_work; };struct dpll {
@@ -900,7 +903,6 @@ struct intel_dp {int fallback_link_rate;uint8_t fallback_lane_count;int fallback_link_rate_index; -bool link_train_failed;uint8_t sink_count;bool link_mst;bool has_audio;
@@ -1402,7 +1404,7 @@ void intel_dp_set_link_params(struct intel_dp *intel_dp, bool link_mst);void intel_dp_get_link_train_fallback_values (struct intel_dp *intel_dp,int link_rate,uint8_t lane_count); -void intel_dp_start_link_train (struct intel_dp *intel_dp); +bool intel_dp_start_link_train (struct intel_dp *intel_dp);void intel_dp_stop_link_train (struct intel_dp *intel_dp);void intel_dp_sink_dpms (struct intel_dp *intel_dp,int mode);void intel_dp_encoder_reset (struct drm_encoder *encoder);
Signature · 6 lines
1.9.1 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH v3 4/4] drm/i915: Implement Link Rate fallback on Link training failure
Jani, could you please take a look at this patch?
I have addressed your comments on the previous revision. The common_rates array can be moved into Intel dp structure in a follow up patch since that would require changes in a lot of places other than the link rate fallback implementation.
Regards
Manasi
Quoted 283 lines · Manasi Navare
If link training at a link rate optimal for a particular mode fails during modeset's atomic commit phase, then we let the modeset complete and then retry. We save the link rate value at which link training failed, update the link status property to "BAD" and use a lower link rate to prune the modes. It will redo the modeset on the current mode at lower link rate or if the current mode gets pruned due to lower link constraints then, it will send a hotplug uevent for userspace to handle it.
This is also required to pass DP CTS tests 4.3.1.3, 4.3.1.4, 4.3.1.6.
v3:
* Set link status property to BAd unconditionally (Jani Nikula)
* Dont use two separate variableslink_train_failedandlink_statusto indicate same thing (Jani Nikula)
v2:
* Squashed a few patches (Jani Nikula)Cc Jani Nikula <[email protected]>Cc Daniel Vetter <[email protected]>Cc Ville Syrjala <[email protected]>Signed-off-by Manasi Navare <[email protected]>--- drivers/gpu/drm/drm_atomic_helper.c | 7 +++ drivers/gpu/drm/i915/intel_ddi.c | 22 ++++++++- drivers/gpu/drm/i915/intel_dp.c | 70 ++++++++++++++++++++++++--- drivers/gpu/drm/i915/intel_dp_link_training.c | 12 +++-- drivers/gpu/drm/i915/intel_drv.h | 6 ++- 5 files changed, 105 insertions(+), 12 deletions(-)drivers/gpu/drm/drm_atomic_helper.c +7
-0@@ -519,6 +519,13 @@ static int handle_conflicting_encoders(struct drm_atomic_state *state, connector_state);if (ret)return ret; + +if (connector->state->crtc) { + crtc_state =drm_atomic_get_existing_crtc_state (state, + connector->state->crtc); +if (connector->link_status ==DRM_MODE_LINK_STATUS_BAD ) + crtc_state->connectors_changed = true; + } }/* drivers/gpu/drm/i915/intel_ddi.c +21
-1@@ -1684,6 +1684,8 @@ static void intel_ddi_pre_enable_dp(struct intel_encoder *encoder,struct intel_dp *intel_dp =enc_to_intel_dp (&encoder->base);struct drm_i915_private *dev_priv =to_i915 (encoder->base.dev);enum port port =intel_ddi_get_encoder_port (encoder); +struct intel_connector *intel_connector = intel_dp->attached_connector; +struct drm_connector *connector = &intel_connector->base;intel_dp_set_link_params (intel_dp, link_rate, lane_count, link_mst);@@ -1694,7 +1696,25 @@ static void intel_ddi_pre_enable_dp(struct intel_encoder *encoder,intel_prepare_dp_ddi_buffers (encoder);intel_ddi_init_dp_buf_reg (encoder);intel_dp_sink_dpms (intel_dp,DRM_MODE_DPMS_ON ); -intel_dp_start_link_train (intel_dp); +if (!intel_dp_start_link_train (intel_dp)) { +DRM_DEBUG_KMS ("Link Training failed at link rate = %d, lane count = %d" , + link_rate, lane_count); + connector->link_status =DRM_MODE_LINK_STATUS_BAD ; +intel_dp_get_link_train_fallback_values (intel_dp, link_rate, + lane_count); +/* Schedule a Hotplug Uevent to userspace to start modeset */ +schedule_work (&intel_connector->modeset_retry_work); + }else { +DRM_DEBUG_KMS ("Link Training Passed at Link Rate = %d, Lane count = %d" , + link_rate, lane_count); + intel_dp->fallback_link_rate_index =-1 ; + intel_dp->fallback_link_rate =0 ; + intel_dp->fallback_lane_count =0 ; + connector->link_status =DRM_MODE_LINK_STATUS_GOOD ; +intel_dp_set_link_status_property (connector, +DRM_MODE_LINK_STATUS_GOOD ); + } +if (port !=PORT_A ||INTEL_GEN (dev_priv) >=9 )intel_dp_stop_link_train (intel_dp); }drivers/gpu/drm/i915/intel_dp.c +64
-6@@ -310,9 +310,6 @@ void intel_dp_get_link_train_fallback_values(struct intel_dp *intel_dp,int common_len;int link_rate_index =-1 ; -if (!intel_dp->link_train_failed) -return ; - common_len =intel_dp_common_rates (intel_dp, common_rates); link_rate_index = intel_dp_link_rate_index(intel_dp, common_rates,@@ -327,7 +324,6 @@ void intel_dp_get_link_train_fallback_values(struct intel_dp *intel_dp, intel_dp->fallback_lane_count = lane_count >>1 ; }else {DRM_ERROR ("Link Training Unsuccessful\n" ); - intel_dp->link_train_failed = false; } }@@ -354,8 +350,14 @@ void intel_dp_get_link_train_fallback_values(struct intel_dp *intel_dp, target_clock = fixed_mode->clock; } - max_link_clock =intel_dp_max_link_rate (intel_dp); - max_lanes =intel_dp_max_lane_count (intel_dp); +/* Prune the modes using the fallback link rate/lane count */ +if (connector->link_status ==DRM_MODE_LINK_STATUS_BAD ) { + max_link_clock = intel_dp->fallback_link_rate; + max_lanes = intel_dp->fallback_lane_count; + }else { + max_link_clock =intel_dp_max_link_rate (intel_dp); + max_lanes =intel_dp_max_lane_count (intel_dp); + } max_rate =intel_dp_max_data_rate (max_link_clock, max_lanes); mode_rate =intel_dp_link_required (target_clock,18 );@@ -1593,6 +1595,7 @@ static int intel_dp_compute_bpp(struct intel_dp *intel_dp,enum port port =dp_to_dig_port (intel_dp)->port;struct intel_crtc *intel_crtc =to_intel_crtc (pipe_config->base.crtc);struct intel_connector *intel_connector = intel_dp->attached_connector; +struct drm_connector *connector = &intel_connector->base;int lane_count, clock;int min_lane_count =1 ;int max_lane_count =intel_dp_max_lane_count (intel_dp);@@ -1640,6 +1643,12 @@ static int intel_dp_compute_bpp(struct intel_dp *intel_dp,if (adjusted_mode->flags &DRM_MODE_FLAG_DBLCLK )return false; +/* Fall back to lower link rate in case of failure in previous modeset */ +if (connector->link_status ==DRM_MODE_LINK_STATUS_BAD ) { + min_lane_count = max_lane_count = intel_dp->fallback_lane_count; + min_clock = max_clock = intel_dp->fallback_link_rate_index; + } + DRM_DEBUG_KMS("DP link computation with max lane count %i " "max bw %d pixel clock %iKHz\n", max_lane_count, common_rates[max_clock],@@ -4423,6 +4432,12 @@ static bool intel_digital_port_connected(struct drm_i915_private *dev_priv, intel_dp->compliance_test_active =0 ; intel_dp->compliance_test_type =0 ; intel_dp->compliance_test_data =0 ; + intel_dp->fallback_link_rate_index =-1 ; + intel_dp->fallback_link_rate =0 ; + intel_dp->fallback_lane_count =0 ; + connector->link_status =DRM_MODE_LINK_STATUS_GOOD ; +intel_dp_set_link_status_property (connector, +DRM_MODE_LINK_STATUS_GOOD );if (intel_dp->is_mst) { DRM_DEBUG_KMS("MST device may have disappeared %d vs %d\n",@@ -4514,6 +4529,11 @@ static bool intel_digital_port_connected(struct drm_i915_private *dev_priv,DRM_DEBUG_KMS ("[CONNECTOR:%d:%s]\n" , connector->base.id, connector->name); +/* If this is a retry due to link trianing failure */ +if (status == connector_status_connected && + connector->link_status ==DRM_MODE_LINK_STATUS_BAD ) +return status; +/* If full detect is not performed yet, do a full detect */ if (!intel_dp->detect_done) status =intel_dp_long_pulse (intel_dp->attached_connector);@@ -5692,6 +5712,40 @@ static bool intel_edp_init_connector(struct intel_dp *intel_dp,return false; } +static void intel_dp_modeset_retry_work_fn (struct work_struct *work) +{ +struct intel_connector *intel_connector; +struct drm_connector *connector; +struct drm_display_mode *mode; +bool verbose_prune = true; + + intel_connector =container_of (work,typeof (*intel_connector), + modeset_retry_work); + connector = &intel_connector->base; + +/* Grab the locks before changing connector property*/ +mutex_lock (&connector->dev->mode_config.mutex); +DRM_DEBUG_KMS ("[CONNECTOR:%d:%s]\n" , connector->base.id, + connector->name); +list_for_each_entry (mode, &connector->modes, head) { + mode->status =intel_dp_mode_valid (connector, + mode); + } +drm_mode_prune_invalid (connector->dev, &connector->modes, + verbose_prune); + +/* Set connector link status to BAD and send a Uevent to notify +* userspace to do a modeset. +*/ + connector->link_status =DRM_MODE_LINK_STATUS_BAD ; +intel_dp_set_link_status_property (connector, +DRM_MODE_LINK_STATUS_BAD ); +mutex_unlock (&connector->dev->mode_config.mutex); + +/* Send Hotplug uevent so userspace can reprobe */ +drm_kms_helper_hotplug_event (connector->dev); +} +bool intel_dp_init_connector (struct intel_digital_port *intel_dig_port,struct intel_connector *intel_connector)@@ -5704,6 +5758,10 @@ static bool intel_edp_init_connector(struct intel_dp *intel_dp,enum port port = intel_dig_port->port;int type; +/* Initialize the work for modeset in case of link train failure */ +INIT_WORK (&intel_connector->modeset_retry_work, + intel_dp_modeset_retry_work_fn); +if (WARN (intel_dig_port->max_lanes <1 ,"Not enough lanes (%d) for DP on port %c\n" , intel_dig_port->max_lanes,port_name (port)))drivers/gpu/drm/i915/intel_dp_link_training.c +9
-3@@ -310,9 +310,15 @@ void intel_dp_stop_link_train(struct intel_dp *intel_dp) DP_TRAINING_PATTERN_DISABLE); } -void +bool intel_dp_start_link_train (struct intel_dp *intel_dp) { -intel_dp_link_training_clock_recovery (intel_dp); -intel_dp_link_training_channel_equalization (intel_dp); +bool ret; + +if (intel_dp_link_training_clock_recovery (intel_dp)) { + ret =intel_dp_link_training_channel_equalization (intel_dp); +if (ret) +return true; + } +return false; }drivers/gpu/drm/i915/intel_drv.h +4
-2@@ -312,6 +312,9 @@ struct intel_connector {void *port;/* store this opaque as its illegal to dereference it */ struct intel_dp *mst_port; + +/* Work struct to schedule a uevent on link train failure */ +struct work_struct modeset_retry_work; };struct dpll {@@ -900,7 +903,6 @@ struct intel_dp {int fallback_link_rate;uint8_t fallback_lane_count;int fallback_link_rate_index; -bool link_train_failed;uint8_t sink_count;bool link_mst;bool has_audio;@@ -1402,7 +1404,7 @@ void intel_dp_set_link_params(struct intel_dp *intel_dp, bool link_mst);void intel_dp_get_link_train_fallback_values (struct intel_dp *intel_dp,int link_rate,uint8_t lane_count); -void intel_dp_start_link_train (struct intel_dp *intel_dp); +bool intel_dp_start_link_train (struct intel_dp *intel_dp);void intel_dp_stop_link_train (struct intel_dp *intel_dp);void intel_dp_sink_dpms (struct intel_dp *intel_dp,int mode);void intel_dp_encoder_reset (struct drm_encoder *encoder);Signature · 1 line
1.9.1
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Jani Nikula
maintainerRe: [PATCH v3 4/4] drm/i915: Implement Link Rate fallback on Link training failure
Quoted 5 lines · Manasi Navare
Jani, could you please take a look at this patch?
I have addressed your comments on the previous revision. Thecommon_ratesarray can be moved into Intel dp structure in a follow up patch since that would require changes in a lot of places other than the link rate fallback implementation.
I think this is looking good. I think you need to put this together in a series that doesn't add something first, and then remove it, i.e. build a coherent story from the start. It's not that many patches. Then you need to post all patches on both intel-gfx and dri-devel, preferrably in a fresh thread, not replying to previous versions. The thread is too confusing now with so many versions.
Then we need to get r-b and acks for the drm changes. Daniel, can you help with the last part?
BR,
Jani.
Signature · 5 lines
Jani Nikula, Intel Open Source Technology Center _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Jani Nikula
maintainerRe: [PATCH v3 4/4] drm/i915: Implement Link Rate fallback on Link training failure
In reply to Manasi NavareQuoted 37 lines · Manasi Navare
If link training at a link rate optimal for a particular mode fails during modeset's atomic commit phase, then we let the modeset complete and then retry. We save the link rate value at which link training failed, update the link status property to "BAD" and use a lower link rate to prune the modes. It will redo the modeset on the current mode at lower link rate or if the current mode gets pruned due to lower link constraints then, it will send a hotplug uevent for userspace to handle it.
This is also required to pass DP CTS tests 4.3.1.3, 4.3.1.4, 4.3.1.6.
v3:
* Set link status property to BAd unconditionally (Jani Nikula)
* Dont use two separate variableslink_train_failedandlink_statusto indicate same thing (Jani Nikula)
v2:
* Squashed a few patches (Jani Nikula)Cc Jani Nikula <[email protected]>Cc Daniel Vetter <[email protected]>Cc Ville Syrjala <[email protected]>Signed-off-by Manasi Navare <[email protected]>--- drivers/gpu/drm/drm_atomic_helper.c | 7 +++ drivers/gpu/drm/i915/intel_ddi.c | 22 ++++++++- drivers/gpu/drm/i915/intel_dp.c | 70 ++++++++++++++++++++++++--- drivers/gpu/drm/i915/intel_dp_link_training.c | 12 +++-- drivers/gpu/drm/i915/intel_drv.h | 6 ++- 5 files changed, 105 insertions(+), 12 deletions(-)drivers/gpu/drm/drm_atomic_helper.c +0
-0@@ -519,6 +519,13 @@ static int handle_conflicting_encoders(struct drm_atomic_state *state, connector_state);
drivers/gpu/drm/drm_atomic_helper.c static int handle_conflicting_encoders(struct drm_atomic_state *state, · line 529
Quoted 10 lines
drivers/gpu/drm/drm_atomic_helper.c +7
-0if (ret)return ret; + +if (connector->state->crtc) { + crtc_state =drm_atomic_get_existing_crtc_state (state, + connector->state->crtc); +if (connector->link_status ==DRM_MODE_LINK_STATUS_BAD ) + crtc_state->connectors_changed = true; + } }
Could this be a separate patch? Anyway needs to be posted on dri-devel.
Quoted 16 lines
drivers/gpu/drm/drm_atomic_helper.c +0
-0/* drivers/gpu/drm/i915/intel_ddi.c +2
-0@@ -1684,6 +1684,8 @@ static void intel_ddi_pre_enable_dp(struct intel_encoder *encoder,struct intel_dp *intel_dp =enc_to_intel_dp (&encoder->base);struct drm_i915_private *dev_priv =to_i915 (encoder->base.dev);enum port port =intel_ddi_get_encoder_port (encoder); +struct intel_connector *intel_connector = intel_dp->attached_connector; +struct drm_connector *connector = &intel_connector->base;intel_dp_set_link_params (intel_dp, link_rate, lane_count, link_mst);
drivers/gpu/drm/i915/intel_ddi.c static void intel_ddi_pre_enable_dp(struct intel_encoder *encoder, · line 1702
Quoted 8 lines
drivers/gpu/drm/i915/intel_ddi.c +4
-1@@ -1694,7 +1696,25 @@ static void intel_ddi_pre_enable_dp(struct intel_encoder *encoder,intel_prepare_dp_ddi_buffers (encoder);intel_ddi_init_dp_buf_reg (encoder);intel_dp_sink_dpms (intel_dp,DRM_MODE_DPMS_ON ); -intel_dp_start_link_train (intel_dp); +if (!intel_dp_start_link_train (intel_dp)) { +DRM_DEBUG_KMS ("Link Training failed at link rate = %d, lane count = %d" , + link_rate, lane_count); + connector->link_status =DRM_MODE_LINK_STATUS_BAD ;
Why does this not use intel_dp_set_link_status_property()?
Quoted 23 lines
drivers/gpu/drm/i915/intel_ddi.c +15
-0+intel_dp_get_link_train_fallback_values (intel_dp, link_rate, + lane_count); +/* Schedule a Hotplug Uevent to userspace to start modeset */ +schedule_work (&intel_connector->modeset_retry_work); + }else { +DRM_DEBUG_KMS ("Link Training Passed at Link Rate = %d, Lane count = %d" , + link_rate, lane_count); + intel_dp->fallback_link_rate_index =-1 ; + intel_dp->fallback_link_rate =0 ; + intel_dp->fallback_lane_count =0 ; + connector->link_status =DRM_MODE_LINK_STATUS_GOOD ; +intel_dp_set_link_status_property (connector, +DRM_MODE_LINK_STATUS_GOOD ); + } +if (port !=PORT_A ||INTEL_GEN (dev_priv) >=9 )intel_dp_stop_link_train (intel_dp); }drivers/gpu/drm/i915/intel_dp.c +0
-0
drivers/gpu/drm/i915/intel_dp.c void intel_dp_get_link_train_fallback_values(struct intel_dp *intel_dp, · line 315
Quoted 6 lines
drivers/gpu/drm/i915/intel_dp.c +0
-3@@ -310,9 +310,6 @@ void intel_dp_get_link_train_fallback_values(struct intel_dp *intel_dp,int common_len;int link_rate_index =-1 ; -if (!intel_dp->link_train_failed) -return ; -
You add this in an earlier patch and now remove it?
Quoted 50 lines
drivers/gpu/drm/i915/intel_dp.c +15
-3common_len =intel_dp_common_rates (intel_dp, common_rates); link_rate_index = intel_dp_link_rate_index(intel_dp, common_rates,@@ -327,7 +324,6 @@ void intel_dp_get_link_train_fallback_values(struct intel_dp *intel_dp, intel_dp->fallback_lane_count = lane_count >>1 ; }else {DRM_ERROR ("Link Training Unsuccessful\n" ); - intel_dp->link_train_failed = false; } }@@ -354,8 +350,14 @@ void intel_dp_get_link_train_fallback_values(struct intel_dp *intel_dp, target_clock = fixed_mode->clock; } - max_link_clock =intel_dp_max_link_rate (intel_dp); - max_lanes =intel_dp_max_lane_count (intel_dp); +/* Prune the modes using the fallback link rate/lane count */ +if (connector->link_status ==DRM_MODE_LINK_STATUS_BAD ) { + max_link_clock = intel_dp->fallback_link_rate; + max_lanes = intel_dp->fallback_lane_count; + }else { + max_link_clock =intel_dp_max_link_rate (intel_dp); + max_lanes =intel_dp_max_lane_count (intel_dp); + } max_rate =intel_dp_max_data_rate (max_link_clock, max_lanes); mode_rate =intel_dp_link_required (target_clock,18 );@@ -1593,6 +1595,7 @@ static int intel_dp_compute_bpp(struct intel_dp *intel_dp,enum port port =dp_to_dig_port (intel_dp)->port;struct intel_crtc *intel_crtc =to_intel_crtc (pipe_config->base.crtc);struct intel_connector *intel_connector = intel_dp->attached_connector; +struct drm_connector *connector = &intel_connector->base;int lane_count, clock;int min_lane_count =1 ;int max_lane_count =intel_dp_max_lane_count (intel_dp);@@ -1640,6 +1643,12 @@ static int intel_dp_compute_bpp(struct intel_dp *intel_dp,if (adjusted_mode->flags &DRM_MODE_FLAG_DBLCLK )return false; +/* Fall back to lower link rate in case of failure in previous modeset */ +if (connector->link_status ==DRM_MODE_LINK_STATUS_BAD ) { + min_lane_count = max_lane_count = intel_dp->fallback_lane_count; + min_clock = max_clock = intel_dp->fallback_link_rate_index; + } + DRM_DEBUG_KMS("DP link computation with max lane count %i " "max bw %d pixel clock %iKHz\n", max_lane_count, common_rates[max_clock],
drivers/gpu/drm/i915/intel_dp.c static bool intel_digital_port_connected(struct drm_i915_private *dev_priv, · line 4440
Quoted 9 lines
drivers/gpu/drm/i915/intel_dp.c +6
-0@@ -4423,6 +4432,12 @@ static bool intel_digital_port_connected(struct drm_i915_private *dev_priv, intel_dp->compliance_test_active =0 ; intel_dp->compliance_test_type =0 ; intel_dp->compliance_test_data =0 ; + intel_dp->fallback_link_rate_index =-1 ; + intel_dp->fallback_link_rate =0 ; + intel_dp->fallback_lane_count =0 ; + connector->link_status =DRM_MODE_LINK_STATUS_GOOD ; +intel_dp_set_link_status_property (connector, +DRM_MODE_LINK_STATUS_GOOD );
Seems like it would be a good idea to have this intel_dp_set_link_status_property() helper set connector->link_status too.
Quoted 121 lines
drivers/gpu/drm/i915/intel_dp.c +43
-0if (intel_dp->is_mst) { DRM_DEBUG_KMS("MST device may have disappeared %d vs %d\n",@@ -4514,6 +4529,11 @@ static bool intel_digital_port_connected(struct drm_i915_private *dev_priv,DRM_DEBUG_KMS ("[CONNECTOR:%d:%s]\n" , connector->base.id, connector->name); +/* If this is a retry due to link trianing failure */ +if (status == connector_status_connected && + connector->link_status ==DRM_MODE_LINK_STATUS_BAD ) +return status; +/* If full detect is not performed yet, do a full detect */ if (!intel_dp->detect_done) status =intel_dp_long_pulse (intel_dp->attached_connector);@@ -5692,6 +5712,40 @@ static bool intel_edp_init_connector(struct intel_dp *intel_dp,return false; } +static void intel_dp_modeset_retry_work_fn (struct work_struct *work) +{ +struct intel_connector *intel_connector; +struct drm_connector *connector; +struct drm_display_mode *mode; +bool verbose_prune = true; + + intel_connector =container_of (work,typeof (*intel_connector), + modeset_retry_work); + connector = &intel_connector->base; + +/* Grab the locks before changing connector property*/ +mutex_lock (&connector->dev->mode_config.mutex); +DRM_DEBUG_KMS ("[CONNECTOR:%d:%s]\n" , connector->base.id, + connector->name); +list_for_each_entry (mode, &connector->modes, head) { + mode->status =intel_dp_mode_valid (connector, + mode); + } +drm_mode_prune_invalid (connector->dev, &connector->modes, + verbose_prune); + +/* Set connector link status to BAD and send a Uevent to notify +* userspace to do a modeset. +*/ + connector->link_status =DRM_MODE_LINK_STATUS_BAD ; +intel_dp_set_link_status_property (connector, +DRM_MODE_LINK_STATUS_BAD ); +mutex_unlock (&connector->dev->mode_config.mutex); + +/* Send Hotplug uevent so userspace can reprobe */ +drm_kms_helper_hotplug_event (connector->dev); +} +bool intel_dp_init_connector (struct intel_digital_port *intel_dig_port,struct intel_connector *intel_connector)@@ -5704,6 +5758,10 @@ static bool intel_edp_init_connector(struct intel_dp *intel_dp,enum port port = intel_dig_port->port;int type; +/* Initialize the work for modeset in case of link train failure */ +INIT_WORK (&intel_connector->modeset_retry_work, + intel_dp_modeset_retry_work_fn); +if (WARN (intel_dig_port->max_lanes <1 ,"Not enough lanes (%d) for DP on port %c\n" , intel_dig_port->max_lanes,port_name (port)))drivers/gpu/drm/i915/intel_dp_link_training.c +9
-3@@ -310,9 +310,15 @@ void intel_dp_stop_link_train(struct intel_dp *intel_dp) DP_TRAINING_PATTERN_DISABLE); } -void +bool intel_dp_start_link_train (struct intel_dp *intel_dp) { -intel_dp_link_training_clock_recovery (intel_dp); -intel_dp_link_training_channel_equalization (intel_dp); +bool ret; + +if (intel_dp_link_training_clock_recovery (intel_dp)) { + ret =intel_dp_link_training_channel_equalization (intel_dp); +if (ret) +return true; + } +return false; }drivers/gpu/drm/i915/intel_drv.h +4
-2@@ -312,6 +312,9 @@ struct intel_connector {void *port;/* store this opaque as its illegal to dereference it */ struct intel_dp *mst_port; + +/* Work struct to schedule a uevent on link train failure */ +struct work_struct modeset_retry_work; };struct dpll {@@ -900,7 +903,6 @@ struct intel_dp {int fallback_link_rate;uint8_t fallback_lane_count;int fallback_link_rate_index; -bool link_train_failed;uint8_t sink_count;bool link_mst;bool has_audio;@@ -1402,7 +1404,7 @@ void intel_dp_set_link_params(struct intel_dp *intel_dp, bool link_mst);void intel_dp_get_link_train_fallback_values (struct intel_dp *intel_dp,int link_rate,uint8_t lane_count); -void intel_dp_start_link_train (struct intel_dp *intel_dp); +bool intel_dp_start_link_train (struct intel_dp *intel_dp);void intel_dp_stop_link_train (struct intel_dp *intel_dp);void intel_dp_sink_dpms (struct intel_dp *intel_dp,int mode);void intel_dp_encoder_reset (struct drm_encoder *encoder);
Signature · 5 lines
Jani Nikula, Intel Open Source Technology Center _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH v3 4/4] drm/i915: Implement Link Rate fallback on Link training failure
Quoted 42 lines · Jani Nikula
Quoted 41 lines · Manasi Navare
If link training at a link rate optimal for a particular mode fails during modeset's atomic commit phase, then we let the modeset complete and then retry. We save the link rate value at which link training failed, update the link status property to "BAD" and use a lower link rate to prune the modes. It will redo the modeset on the current mode at lower link rate or if the current mode gets pruned due to lower link constraints then, it will send a hotplug uevent for userspace to handle it.
This is also required to pass DP CTS tests 4.3.1.3, 4.3.1.4, 4.3.1.6.
v3:
* Set link status property to BAd unconditionally (Jani Nikula)
* Dont use two separate variableslink_train_failedandlink_statusto indicate same thing (Jani Nikula)
v2:
* Squashed a few patches (Jani Nikula)Cc Jani Nikula <[email protected]>Cc Daniel Vetter <[email protected]>Cc Ville Syrjala <[email protected]>Signed-off-by Manasi Navare <[email protected]>--- drivers/gpu/drm/drm_atomic_helper.c | 7 +++ drivers/gpu/drm/i915/intel_ddi.c | 22 ++++++++- drivers/gpu/drm/i915/intel_dp.c | 70 ++++++++++++++++++++++++--- drivers/gpu/drm/i915/intel_dp_link_training.c | 12 +++-- drivers/gpu/drm/i915/intel_drv.h | 6 ++- 5 files changed, 105 insertions(+), 12 deletions(-)drivers/gpu/drm/drm_atomic_helper.c +2
-0@@ -519,6 +519,13 @@ static int handle_conflicting_encoders(struct drm_atomic_state *state, connector_state);if (ret)return ret; + +if (connector->state->crtc) {
Quoted 9 lines
drivers/gpu/drm/drm_atomic_helper.c static int handle_conflicting_encoders(struct drm_atomic_state *state, · line 529
Quoted 6 lines
drivers/gpu/drm/drm_atomic_helper.c +5
-0+ crtc_state =drm_atomic_get_existing_crtc_state (state, + connector->state->crtc); +if (connector->link_status ==DRM_MODE_LINK_STATUS_BAD ) + crtc_state->connectors_changed = true; + } }Could this be a separate patch? Anyway needs to be posted on dri-devel.
Yes I agree, I will move this into a separate patch right after the patch where I declare the connector link status.
Quoted 13 lines
Quoted 13 lines
drivers/gpu/drm/drm_atomic_helper.c +0
-0/* drivers/gpu/drm/i915/intel_ddi.c +2
-0@@ -1684,6 +1684,8 @@ static void intel_ddi_pre_enable_dp(struct intel_encoder *encoder,struct intel_dp *intel_dp =enc_to_intel_dp (&encoder->base);struct drm_i915_private *dev_priv =to_i915 (encoder->base.dev);enum port port =intel_ddi_get_encoder_port (encoder); +struct intel_connector *intel_connector = intel_dp->attached_connector; +struct drm_connector *connector = &intel_connector->base;
Quoted 14 lines
drivers/gpu/drm/i915/intel_ddi.c static void intel_ddi_pre_enable_dp(struct intel_encoder *encoder, · line 1702
Quoted 11 lines
drivers/gpu/drm/i915/intel_ddi.c +4
-1intel_dp_set_link_params (intel_dp, link_rate, lane_count, link_mst);@@ -1694,7 +1696,25 @@ static void intel_ddi_pre_enable_dp(struct intel_encoder *encoder,intel_prepare_dp_ddi_buffers (encoder);intel_ddi_init_dp_buf_reg (encoder);intel_dp_sink_dpms (intel_dp,DRM_MODE_DPMS_ON ); -intel_dp_start_link_train (intel_dp); +if (!intel_dp_start_link_train (intel_dp)) { +DRM_DEBUG_KMS ("Link Training failed at link rate = %d, lane count = %d" , + link_rate, lane_count); + connector->link_status =DRM_MODE_LINK_STATUS_BAD ;Why does this not use
intel_dp_set_link_status_property()?
I will move this connector->link_status setting inside the set_link_status function.
Quoted 26 lines
Quoted 26 lines
drivers/gpu/drm/i915/intel_ddi.c +15
-0+intel_dp_get_link_train_fallback_values (intel_dp, link_rate, + lane_count); +/* Schedule a Hotplug Uevent to userspace to start modeset */ +schedule_work (&intel_connector->modeset_retry_work); + }else { +DRM_DEBUG_KMS ("Link Training Passed at Link Rate = %d, Lane count = %d" , + link_rate, lane_count); + intel_dp->fallback_link_rate_index =-1 ; + intel_dp->fallback_link_rate =0 ; + intel_dp->fallback_lane_count =0 ; + connector->link_status =DRM_MODE_LINK_STATUS_GOOD ; +intel_dp_set_link_status_property (connector, +DRM_MODE_LINK_STATUS_GOOD ); + } +if (port !=PORT_A ||INTEL_GEN (dev_priv) >=9 )intel_dp_stop_link_train (intel_dp); }drivers/gpu/drm/i915/intel_dp.c +0
-0@@ -310,9 +310,6 @@ void intel_dp_get_link_train_fallback_values(struct intel_dp *intel_dp,int common_len;int link_rate_index =-1 ;
Quoted 6 lines
drivers/gpu/drm/i915/intel_dp.c void intel_dp_get_link_train_fallback_values(struct intel_dp *intel_dp, · line 315
Quoted 3 lines
drivers/gpu/drm/i915/intel_dp.c +0
-3-if (!intel_dp->link_train_failed) -return ; -You add this in an earlier patch and now remove it?
I removed it to address the comment about not using link_train_failed and link_status to indicate the same thing. But this will be cleaned up when I submit a clean patch series so that it does not appear as added in previous patch and removed here.
Quoted 55 lines
Quoted 55 lines
drivers/gpu/drm/i915/intel_dp.c +17
-3common_len =intel_dp_common_rates (intel_dp, common_rates); link_rate_index = intel_dp_link_rate_index(intel_dp, common_rates,@@ -327,7 +324,6 @@ void intel_dp_get_link_train_fallback_values(struct intel_dp *intel_dp, intel_dp->fallback_lane_count = lane_count >>1 ; }else {DRM_ERROR ("Link Training Unsuccessful\n" ); - intel_dp->link_train_failed = false; } }@@ -354,8 +350,14 @@ void intel_dp_get_link_train_fallback_values(struct intel_dp *intel_dp, target_clock = fixed_mode->clock; } - max_link_clock =intel_dp_max_link_rate (intel_dp); - max_lanes =intel_dp_max_lane_count (intel_dp); +/* Prune the modes using the fallback link rate/lane count */ +if (connector->link_status ==DRM_MODE_LINK_STATUS_BAD ) { + max_link_clock = intel_dp->fallback_link_rate; + max_lanes = intel_dp->fallback_lane_count; + }else { + max_link_clock =intel_dp_max_link_rate (intel_dp); + max_lanes =intel_dp_max_lane_count (intel_dp); + } max_rate =intel_dp_max_data_rate (max_link_clock, max_lanes); mode_rate =intel_dp_link_required (target_clock,18 );@@ -1593,6 +1595,7 @@ static int intel_dp_compute_bpp(struct intel_dp *intel_dp,enum port port =dp_to_dig_port (intel_dp)->port;struct intel_crtc *intel_crtc =to_intel_crtc (pipe_config->base.crtc);struct intel_connector *intel_connector = intel_dp->attached_connector; +struct drm_connector *connector = &intel_connector->base;int lane_count, clock;int min_lane_count =1 ;int max_lane_count =intel_dp_max_lane_count (intel_dp);@@ -1640,6 +1643,12 @@ static int intel_dp_compute_bpp(struct intel_dp *intel_dp,if (adjusted_mode->flags &DRM_MODE_FLAG_DBLCLK )return false; +/* Fall back to lower link rate in case of failure in previous modeset */ +if (connector->link_status ==DRM_MODE_LINK_STATUS_BAD ) { + min_lane_count = max_lane_count = intel_dp->fallback_lane_count; + min_clock = max_clock = intel_dp->fallback_link_rate_index; + } + DRM_DEBUG_KMS("DP link computation with max lane count %i " "max bw %d pixel clock %iKHz\n", max_lane_count, common_rates[max_clock],@@ -4423,6 +4432,12 @@ static bool intel_digital_port_connected(struct drm_i915_private *dev_priv, intel_dp->compliance_test_active =0 ; intel_dp->compliance_test_type =0 ; intel_dp->compliance_test_data =0 ; + intel_dp->fallback_link_rate_index =-1 ; + intel_dp->fallback_link_rate =0 ;
Quoted 9 lines
drivers/gpu/drm/i915/intel_dp.c static bool intel_digital_port_connected(struct drm_i915_private *dev_priv, · line 4440
Quoted 4 lines
drivers/gpu/drm/i915/intel_dp.c +4
-0+ intel_dp->fallback_lane_count =0 ; + connector->link_status =DRM_MODE_LINK_STATUS_GOOD ; +intel_dp_set_link_status_property (connector, +DRM_MODE_LINK_STATUS_GOOD );Seems like it would be a good idea to have this
intel_dp_set_link_status_property()helper setconnector->link_statustoo.
Yup, will do this change.
Regards
Manasi
Quoted 113 lines
Quoted 113 lines
drivers/gpu/drm/i915/intel_dp.c +43
-0if (intel_dp->is_mst) { DRM_DEBUG_KMS("MST device may have disappeared %d vs %d\n",@@ -4514,6 +4529,11 @@ static bool intel_digital_port_connected(struct drm_i915_private *dev_priv,DRM_DEBUG_KMS ("[CONNECTOR:%d:%s]\n" , connector->base.id, connector->name); +/* If this is a retry due to link trianing failure */ +if (status == connector_status_connected && + connector->link_status ==DRM_MODE_LINK_STATUS_BAD ) +return status; +/* If full detect is not performed yet, do a full detect */ if (!intel_dp->detect_done) status =intel_dp_long_pulse (intel_dp->attached_connector);@@ -5692,6 +5712,40 @@ static bool intel_edp_init_connector(struct intel_dp *intel_dp,return false; } +static void intel_dp_modeset_retry_work_fn (struct work_struct *work) +{ +struct intel_connector *intel_connector; +struct drm_connector *connector; +struct drm_display_mode *mode; +bool verbose_prune = true; + + intel_connector =container_of (work,typeof (*intel_connector), + modeset_retry_work); + connector = &intel_connector->base; + +/* Grab the locks before changing connector property*/ +mutex_lock (&connector->dev->mode_config.mutex); +DRM_DEBUG_KMS ("[CONNECTOR:%d:%s]\n" , connector->base.id, + connector->name); +list_for_each_entry (mode, &connector->modes, head) { + mode->status =intel_dp_mode_valid (connector, + mode); + } +drm_mode_prune_invalid (connector->dev, &connector->modes, + verbose_prune); + +/* Set connector link status to BAD and send a Uevent to notify +* userspace to do a modeset. +*/ + connector->link_status =DRM_MODE_LINK_STATUS_BAD ; +intel_dp_set_link_status_property (connector, +DRM_MODE_LINK_STATUS_BAD ); +mutex_unlock (&connector->dev->mode_config.mutex); + +/* Send Hotplug uevent so userspace can reprobe */ +drm_kms_helper_hotplug_event (connector->dev); +} +bool intel_dp_init_connector (struct intel_digital_port *intel_dig_port,struct intel_connector *intel_connector)@@ -5704,6 +5758,10 @@ static bool intel_edp_init_connector(struct intel_dp *intel_dp,enum port port = intel_dig_port->port;int type; +/* Initialize the work for modeset in case of link train failure */ +INIT_WORK (&intel_connector->modeset_retry_work, + intel_dp_modeset_retry_work_fn); +if (WARN (intel_dig_port->max_lanes <1 ,"Not enough lanes (%d) for DP on port %c\n" , intel_dig_port->max_lanes,port_name (port)))drivers/gpu/drm/i915/intel_dp_link_training.c +9
-3@@ -310,9 +310,15 @@ void intel_dp_stop_link_train(struct intel_dp *intel_dp) DP_TRAINING_PATTERN_DISABLE); } -void +bool intel_dp_start_link_train (struct intel_dp *intel_dp) { -intel_dp_link_training_clock_recovery (intel_dp); -intel_dp_link_training_channel_equalization (intel_dp); +bool ret; + +if (intel_dp_link_training_clock_recovery (intel_dp)) { + ret =intel_dp_link_training_channel_equalization (intel_dp); +if (ret) +return true; + } +return false; }drivers/gpu/drm/i915/intel_drv.h +3
-1@@ -312,6 +312,9 @@ struct intel_connector {void *port;/* store this opaque as its illegal to dereference it */ struct intel_dp *mst_port; + +/* Work struct to schedule a uevent on link train failure */ +struct work_struct modeset_retry_work; };struct dpll {@@ -900,7 +903,6 @@ struct intel_dp {int fallback_link_rate;uint8_t fallback_lane_count;int fallback_link_rate_index; -bool link_train_failed;uint8_t sink_count;bool link_mst;bool has_audio;@@ -1402,7 +1404,7 @@ void intel_dp_set_link_params(struct intel_dp *intel_dp,
Quoted 11 lines
Quoted 8 lines
drivers/gpu/drm/i915/intel_drv.h +1
-1bool link_mst);void intel_dp_get_link_train_fallback_values (struct intel_dp *intel_dp,int link_rate,uint8_t lane_count); -void intel_dp_start_link_train (struct intel_dp *intel_dp); +bool intel_dp_start_link_train (struct intel_dp *intel_dp);void intel_dp_stop_link_train (struct intel_dp *intel_dp);void intel_dp_sink_dpms (struct intel_dp *intel_dp,int mode);void intel_dp_encoder_reset (struct drm_encoder *encoder);Signature · 1 line
Jani Nikula, Intel Open Source Technology Center
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Patchwork
CI✗ Fi.CI.BAT: failure for Handle link training failure during modeset for DDI
In reply to Manasi Navare== Series Details ==
== Series Details ==
Series: Handle link training failure during modeset for DDI
URL : https://patchwork.freedesktop.org/series/14556/
State : failure
== Summary ==
Series 14556v1 Handle link training failure during modeset for DDI
https://patchwork.freedesktop.org/api/1.0/series/14556/revisions/1/mbox/
Test drv_module_reload_basic:
pass -> INCOMPLETE (fi-skl-6770hq)
dmesg-warn -> PASS (fi-hsw-4770)
Test kms_force_connector_basic:
Subgroup force-edid:
dmesg-warn -> PASS (fi-snb-2520m)
Test kms_pipe_crc_basic:
Subgroup bad-pipe:
dmesg-warn -> PASS (fi-ilk-650)fi-bdw-5557u total:241 pass:226 dwarn:0 dfail:0 fail:0 skip:15 fi-bsw-n3050 total:241 pass:201 dwarn:0 dfail:0 fail:0 skip:40 fi-bxt-t5700 total:241 pass:213 dwarn:0 dfail:0 fail:0 skip:28 fi-byt-j1900 total:241 pass:213 dwarn:0 dfail:0 fail:0 skip:28 fi-byt-n2820 total:241 pass:209 dwarn:0 dfail:0 fail:0 skip:32 fi-hsw-4770 total:241 pass:221 dwarn:0 dfail:0 fail:0 skip:20 fi-hsw-4770r total:241 pass:220 dwarn:0 dfail:0 fail:0 skip:21 fi-ilk-650 total:241 pass:187 dwarn:0 dfail:0 fail:0 skip:54 fi-ivb-3520m total:241 pass:218 dwarn:0 dfail:0 fail:0 skip:23 fi-ivb-3770 total:241 pass:218 dwarn:0 dfail:0 fail:0 skip:23 fi-kbl-7200u total:241 pass:219 dwarn:0 dfail:0 fail:0 skip:22 fi-skl-6260u total:241 pass:227 dwarn:0 dfail:0 fail:0 skip:14 fi-skl-6700hq total:241 pass:220 dwarn:0 dfail:0 fail:0 skip:21 fi-skl-6770hq total:6 pass:5 dwarn:0 dfail:0 fail:0 skip:0 fi-snb-2520m total:241 pass:208 dwarn:0 dfail:0 fail:0 skip:33 fi-snb-2600 total:241 pass:207 dwarn:0 dfail:0 fail:0 skip:34
f3968d51ade8c02e8868c2608ea4adb5a4afe301 drm-intel-nightly: 2016y-10m-28d-19h-57m-33s UTC integration manifest eac571a drm/i915: Implement Link Rate fallback on Link training failure
ff275dd drm/i915: Find fallback link rate/lane count
fd42134 drm/i915: Set link status property for DP connector
7a0359a drm: Add a new connector property for link status
== Logs ==
For more details see: https://intel-gfx-ci.01.org/CI/Patchwork_2854/
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Patchwork
CI✗ Fi.CI.BAT: failure for Handle link training failure during modeset for DDI (rev2)
In reply to Manasi Navare== Series Details ==
== Series Details ==
Series: Handle link training failure during modeset for DDI (rev2)
URL : https://patchwork.freedesktop.org/series/14556/
State : failure
== Summary ==
Series 14556v2 Handle link training failure during modeset for DDI
https://patchwork.freedesktop.org/api/1.0/series/14556/revisions/2/mbox/
Test drv_module_reload_basic:
dmesg-warn -> PASS (fi-ilk-650)
dmesg-warn -> PASS (fi-bdw-5557u)
pass -> INCOMPLETE (fi-skl-6770hq)
dmesg-warn -> INCOMPLETE (fi-skl-6700k)
Test kms_cursor_legacy:
Subgroup basic-busy-flip-before-cursor-legacy:
pass -> DMESG-WARN (fi-ilk-650)
Subgroup basic-flip-before-cursor-varying-size:
pass -> DMESG-WARN (fi-ilk-650)
Test kms_force_connector_basic:
Subgroup force-edid:
dmesg-warn -> PASS (fi-snb-2520m)
Test kms_pipe_crc_basic:
Subgroup bad-nb-words-1:
pass -> DMESG-WARN (fi-ilk-650)
Subgroup bad-nb-words-3:
dmesg-warn -> PASS (fi-ilk-650)
Subgroup nonblocking-crc-pipe-a-frame-sequence:
pass -> DMESG-WARN (fi-ilk-650)
Subgroup suspend-read-crc-pipe-a:
dmesg-warn -> PASS (fi-ilk-650)fi-bdw-5557u total:241 pass:226 dwarn:0 dfail:0 fail:0 skip:15 fi-bsw-n3050 total:241 pass:201 dwarn:0 dfail:0 fail:0 skip:40 fi-bxt-t5700 total:241 pass:213 dwarn:0 dfail:0 fail:0 skip:28 fi-byt-j1900 total:241 pass:213 dwarn:0 dfail:0 fail:0 skip:28 fi-byt-n2820 total:241 pass:209 dwarn:0 dfail:0 fail:0 skip:32 fi-hsw-4770 total:241 pass:221 dwarn:0 dfail:0 fail:0 skip:20 fi-hsw-4770r total:241 pass:220 dwarn:0 dfail:0 fail:0 skip:21 fi-ilk-650 total:241 pass:180 dwarn:7 dfail:0 fail:0 skip:54 fi-ivb-3520m total:241 pass:218 dwarn:0 dfail:0 fail:0 skip:23 fi-ivb-3770 total:241 pass:218 dwarn:0 dfail:0 fail:0 skip:23 fi-kbl-7200u total:241 pass:219 dwarn:0 dfail:0 fail:0 skip:22 fi-skl-6260u total:241 pass:227 dwarn:0 dfail:0 fail:0 skip:14 fi-skl-6700hq total:241 pass:220 dwarn:0 dfail:0 fail:0 skip:21 fi-skl-6700k total:6 pass:5 dwarn:0 dfail:0 fail:0 skip:0 fi-skl-6770hq total:6 pass:5 dwarn:0 dfail:0 fail:0 skip:0 fi-snb-2520m total:241 pass:208 dwarn:0 dfail:0 fail:0 skip:33 fi-snb-2600 total:241 pass:207 dwarn:0 dfail:0 fail:0 skip:34
6a1197bcb5cc18a56ad4ae8e6d706a212bc3db7d drm-intel-nightly: 2016y-10m-31d-14h-58m-16s UTC integration manifest d3c54ba drm/i915: Implement Link Rate fallback on Link training failure
14102d8 drm/i915: Find fallback link rate/lane count
8705bee drm/i915: Set link status property for DP connector
cd53886 drm: Add a new connector property for link status
== Logs ==
For more details see: https://intel-gfx-ci.01.org/CI/Patchwork_2867/
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Patchwork
CI✗ Fi.CI.BAT: failure for Handle link training failure during modeset for DDI (rev3)
In reply to Manasi Navare== Series Details ==
== Series Details ==
Series: Handle link training failure during modeset for DDI (rev3)
URL : https://patchwork.freedesktop.org/series/14556/
State : failure
== Summary ==
Series 14556v3 Handle link training failure during modeset for DDI
https://patchwork.freedesktop.org/api/1.0/series/14556/revisions/3/mbox/
Test drv_module_reload_basic:
pass -> INCOMPLETE (fi-skl-6770hq)
dmesg-warn -> INCOMPLETE (fi-skl-6700k)fi-bdw-5557u total:241 pass:226 dwarn:0 dfail:0 fail:0 skip:15 fi-bsw-n3050 total:241 pass:201 dwarn:0 dfail:0 fail:0 skip:40 fi-byt-j1900 total:241 pass:213 dwarn:0 dfail:0 fail:0 skip:28 fi-byt-n2820 total:241 pass:209 dwarn:0 dfail:0 fail:0 skip:32 fi-hsw-4770 total:241 pass:221 dwarn:0 dfail:0 fail:0 skip:20 fi-hsw-4770r total:241 pass:220 dwarn:0 dfail:0 fail:0 skip:21 fi-ilk-650 total:241 pass:187 dwarn:0 dfail:0 fail:0 skip:54 fi-ivb-3520m total:241 pass:218 dwarn:0 dfail:0 fail:0 skip:23 fi-ivb-3770 total:241 pass:218 dwarn:0 dfail:0 fail:0 skip:23 fi-kbl-7200u total:241 pass:219 dwarn:0 dfail:0 fail:0 skip:22 fi-skl-6260u total:241 pass:227 dwarn:0 dfail:0 fail:0 skip:14 fi-skl-6700hq total:241 pass:220 dwarn:0 dfail:0 fail:0 skip:21 fi-skl-6700k total:6 pass:5 dwarn:0 dfail:0 fail:0 skip:0 fi-skl-6770hq total:6 pass:5 dwarn:0 dfail:0 fail:0 skip:0 fi-snb-2520m total:241 pass:208 dwarn:0 dfail:0 fail:0 skip:33 fi-snb-2600 total:241 pass:207 dwarn:0 dfail:0 fail:0 skip:34
bf6b989af8b0fde56a352d9005c97b2d8e3bbbe3 drm-intel-nightly: 2016y-11m-02d-15h-44m-03s UTC integration manifest 06564f1 drm/i915: Implement Link Rate fallback on Link training failure
3e2e35f drm/i915: Find fallback link rate/lane count
a4fc113 drm/i915: Set link status property for DP connector
0abe534 drm: Add a new connector property for link status
== Logs ==
For more details see: https://intel-gfx-ci.01.org/CI/Patchwork_2896/
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx