DP MST Refactors + debugging tools + suspend/resume reprobing
Subsystems (7)
Lyude Paul
maintainercover DP MST Refactors + debugging tools + suspend/resume reprobing
This is the large series for adding MST suspend/resume reprobing that I've been working on for quite a while now. In addition, I:
- Refactored and cleaned up any code I ended up digging through in the process of understanding how some parts of these helpers worked.
- Added some debugging tools along the way that I ended up needing to figure out some issues in my own code
Note that there's still one important part of this process missing that's not included in this patch series: EDID reprobing, which I believe Stanislav Lisovskiy from Intel is currently working on. The main purpose of this series is to fix the issue of the in-memory topology state (e.g. connectors connected to an MST hub, branch devices, etc.) going out of sync if a topology connected to a connector is swapped out with a different topology while the system is resumed, or while the device housing said connector is in runtime suspend.
As well, the debugging tools that are added in this include:
- A limited debugging utility for dumping the list of topology references on an MST port or branch connector whose topology reference count has reached 0
- Sideband down request dumping, along with some basic selftests for testing our encoding/decoding functions
Lyude Paul (26):
drm/dp_mst: Move link address dumping into a function drm/dp_mst: Destroy mstbs from destroy_connector_work drm/dp_mst: Move test_calc_pbn_mode() into an actual selftest drm/print: Add drm_err_printer()
drm/dp_mst: Add sideband down request tracing + selftests drm/dp_mst: Move PDT teardown for ports into destroy_connector_work drm/dp_mst: Get rid of list clear in drm_dp_finish_destroy_port() drm/dp_mst: Refactor drm_dp_send_enum_path_resources drm/dp_mst: Remove huge conditional in drm_dp_mst_handle_up_req() drm/dp_mst: Constify guid in drm_dp_get_mst_branch_by_guid() drm/dp_mst: Refactor drm_dp_mst_handle_up_req() drm/dp_mst: Refactor drm_dp_mst_handle_down_rep() drm/dp_mst: Destroy topology_mgr mutexes
drm/dp_mst: Cleanup drm_dp_send_link_address() a bit drm/dp_mst: Refactor pdt setup/teardown, add more locking drm/dp_mst: Rename drm_dp_add_port and drm_dp_update_port drm/dp_mst: Remove lies in {up,down}_rep_recv documentation drm/dp_mst: Handle UP requests asynchronously drm/dp_mst: Protect drm_dp_mst_port members with connection_mutex drm/dp_mst: Don't forget to update port->input in
drm_dp_mst_handle_conn_stat()
drm/nouveau: Don't grab runtime PM refs for HPD IRQs drm/amdgpu: Iterate through DRM connectors correctly drm/amdgpu/dm: Resume short HPD IRQs before resuming MST topology drm/dp_mst: Add basic topology reprobing when resuming drm/dp_mst: Also print unhashed pointers for malloc/topology
references
drm/dp_mst: Add topology ref history tracking for debugging
drivers/gpu/drm/Kconfig | 14 + .../gpu/drm/amd/amdgpu/amdgpu_connectors.c | 13 +- drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 20 +- drivers/gpu/drm/amd/amdgpu/amdgpu_display.c | 5 +- drivers/gpu/drm/amd/amdgpu/amdgpu_encoders.c | 40 +- drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c | 5 +- drivers/gpu/drm/amd/amdgpu/dce_v10_0.c | 34 +- drivers/gpu/drm/amd/amdgpu/dce_v11_0.c | 34 +- drivers/gpu/drm/amd/amdgpu/dce_v6_0.c | 40 +- drivers/gpu/drm/amd/amdgpu/dce_v8_0.c | 34 +- .../gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c | 41 +- .../drm/amd/display/amdgpu_dm/amdgpu_dm_irq.c | 10 +- drivers/gpu/drm/drm_dp_mst_topology.c | 1592 +++++++++++++---- .../gpu/drm/drm_dp_mst_topology_internal.h | 24 + drivers/gpu/drm/drm_print.c | 6 + drivers/gpu/drm/i915/display/intel_dp.c | 3 +- drivers/gpu/drm/nouveau/dispnv50/disp.c | 6 +- drivers/gpu/drm/nouveau/nouveau_connector.c | 33 +- drivers/gpu/drm/selftests/Makefile | 2 +- .../gpu/drm/selftests/drm_modeset_selftests.h | 2 + .../drm/selftests/test-drm_dp_mst_helper.c | 248 +++ .../drm/selftests/test-drm_modeset_common.h | 2 + include/drm/drm_dp_mst_helper.h | 127 +- include/drm/drm_print.h | 17 + 24 files changed, 1846 insertions(+), 506 deletions(-) create mode 100644 drivers/gpu/drm/drm_dp_mst_topology_internal.h create mode 100644 drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c
Signature · 6 lines
2.21.0 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Lyude Paul
maintainerPATCH 1/26 drm/dp_mst: Move link address dumping into a function
Since we're about to be calling this from multiple places. Also it makes things easier to read!
--- drivers/gpu/drm/drm_dp_mst_topology.c | 35 ++++++++++++++++++--------- 1 file changed, 23 insertions(+), 12 deletions(-)
drivers/gpu/drm/drm_dp_mst_topology.c +23 -12
@@ -2013,6 +2013,28 @@ static void drm_dp_queue_down_tx(struct drm_dp_mst_topology_mgr *mgr,mutex_unlock (&mgr->qlock); } +static void +drm_dp_dump_link_address (struct drm_dp_link_address_ack_reply *reply) +{ +struct drm_dp_link_addr_reply_port *port_reply; +int i; + +for (i =0 ; i < reply->nports; i++) { + port_reply = &reply->ports[i]; +DRM_DEBUG_KMS ("port %d: input %d, pdt: %d, pn: %d, dpcd_rev: %02x, mcs: %d, ddps: %d, ldps %d, sdp %d/%d\n" , + i, + port_reply->input_port, + port_reply->peer_device_type, + port_reply->port_number, + port_reply->dpcd_revision, + port_reply->mcs, + port_reply->ddps, + port_reply->legacy_device_plug_status, + port_reply->num_sdp_streams, + port_reply->num_sdp_stream_sinks); + } +} +static void drm_dp_send_link_address (struct drm_dp_mst_topology_mgr *mgr,struct drm_dp_mst_branch *mstb) {
@@ -2038,18 +2060,7 @@ static void drm_dp_send_link_address(struct drm_dp_mst_topology_mgr *mgr,DRM_DEBUG_KMS ("link address nak received\n" ); }else {DRM_DEBUG_KMS ("link address reply: %d\n" , txmsg->reply.u.link_addr.nports); -for (i =0 ; i < txmsg->reply.u.link_addr.nports; i++) { -DRM_DEBUG_KMS ("port %d: input %d, pdt: %d, pn: %d, dpcd_rev: %02x, mcs: %d, ddps: %d, ldps %d, sdp %d/%d\n" , i, - txmsg->reply.u.link_addr.ports[i].input_port, - txmsg->reply.u.link_addr.ports[i].peer_device_type, - txmsg->reply.u.link_addr.ports[i].port_number, - txmsg->reply.u.link_addr.ports[i].dpcd_revision, - txmsg->reply.u.link_addr.ports[i].mcs, - txmsg->reply.u.link_addr.ports[i].ddps, - txmsg->reply.u.link_addr.ports[i].legacy_device_plug_status, - txmsg->reply.u.link_addr.ports[i].num_sdp_streams, - txmsg->reply.u.link_addr.ports[i].num_sdp_stream_sinks); - } +drm_dp_dump_link_address (&txmsg->reply.u.link_addr);drm_dp_check_mstb_guid (mstb, txmsg->reply.u.link_addr.guid);
Signature · 1 line
2.21.0
Re: [PATCH 01/26] drm/dp_mst: Move link address dumping into a function
Quoted 8 lines · Lyude Paul
Since we're about to be calling this from multiple places. Also it makes things easier to read!
Cc Juston Li <[email protected]>Cc Imre Deak <[email protected]>Cc Ville Syrjälä <[email protected]>Cc Harry Wentland <[email protected]>Signed-off-by Lyude Paul <[email protected]>
Quoted 60 lines
--- drivers/gpu/drm/drm_dp_mst_topology.c | 35 ++++++++++++++++++--------- 1 file changed, 23 insertions(+), 12 deletions(-)drivers/gpu/drm/drm_dp_mst_topology.c +23
-12@@ -2013,6 +2013,28 @@ static void drm_dp_queue_down_tx(struct drm_dp_mst_topology_mgr *mgr,mutex_unlock (&mgr->qlock); } +static void +drm_dp_dump_link_address (struct drm_dp_link_address_ack_reply *reply) +{ +struct drm_dp_link_addr_reply_port *port_reply; +int i; + +for (i =0 ; i < reply->nports; i++) { + port_reply = &reply->ports[i]; +DRM_DEBUG_KMS ("port %d: input %d, pdt: %d, pn: %d, dpcd_rev: %02x, mcs: %d, ddps: %d, ldps %d, sdp %d/%d\n" , + i, + port_reply->input_port, + port_reply->peer_device_type, + port_reply->port_number, + port_reply->dpcd_revision, + port_reply->mcs, + port_reply->ddps, + port_reply->legacy_device_plug_status, + port_reply->num_sdp_streams, + port_reply->num_sdp_stream_sinks); + } +} +static void drm_dp_send_link_address (struct drm_dp_mst_topology_mgr *mgr,struct drm_dp_mst_branch *mstb) {@@ -2038,18 +2060,7 @@ static void drm_dp_send_link_address(struct drm_dp_mst_topology_mgr *mgr,DRM_DEBUG_KMS ("link address nak received\n" ); }else {DRM_DEBUG_KMS ("link address reply: %d\n" , txmsg->reply.u.link_addr.nports); -for (i =0 ; i < txmsg->reply.u.link_addr.nports; i++) { -DRM_DEBUG_KMS ("port %d: input %d, pdt: %d, pn: %d, dpcd_rev: %02x, mcs: %d, ddps: %d, ldps %d, sdp %d/%d\n" , i, - txmsg->reply.u.link_addr.ports[i].input_port, - txmsg->reply.u.link_addr.ports[i].peer_device_type, - txmsg->reply.u.link_addr.ports[i].port_number, - txmsg->reply.u.link_addr.ports[i].dpcd_revision, - txmsg->reply.u.link_addr.ports[i].mcs, - txmsg->reply.u.link_addr.ports[i].ddps, - txmsg->reply.u.link_addr.ports[i].legacy_device_plug_status, - txmsg->reply.u.link_addr.ports[i].num_sdp_streams, - txmsg->reply.u.link_addr.ports[i].num_sdp_stream_sinks); - } +drm_dp_dump_link_address (&txmsg->reply.u.link_addr);drm_dp_check_mstb_guid (mstb, txmsg->reply.u.link_addr.guid);Signature · 1 line
2.21.0
Signature · 7 lines
Daniel Vetter Software Engineer, Intel Corporation http://blog.ffwll.ch _______________________________________________ dri-devel mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/dri-devel
Lyude Paul
maintainerRe: [PATCH 01/26] drm/dp_mst: Move link address dumping into a function
*sigh* finally have some time to go through these reviews
jfyi: I realized after looking over this patch that it's not actually needed - I had been planning on using drm_dp_dump_link_address() for other things, but ended up deciding to make the final plan to use something that dumps into a format that's identical to the one we're using for dumping DOWN requests. IMHO though, this patch does make things look nicer so I'll probably keep it.
Assuming I can still count your r-b as valid with a change to the commit description?
Quoted 87 lines · Daniel Vetter
Quoted 8 lines · Lyude Paul
Since we're about to be calling this from multiple places. Also it makes things easier to read!
Cc Juston Li <[email protected]>Cc Imre Deak <[email protected]>Cc Ville Syrjälä <[email protected]>Cc Harry Wentland <[email protected]>Signed-off-by Lyude Paul <[email protected]>Reviewed-by Daniel Vetter <[email protected]>Quoted 75 lines
--- drivers/gpu/drm/drm_dp_mst_topology.c | 35 ++++++++++++++++++--------- 1 file changed, 23 insertions(+), 12 deletions(-)drivers/gpu/drm/drm_dp_mst_topology.c +22
-0@@ -2013,6 +2013,28 @@ static void drm_dp_queue_down_tx(struct drm_dp_mst_topology_mgr *mgr,mutex_unlock (&mgr->qlock); } +static void +drm_dp_dump_link_address (struct drm_dp_link_address_ack_reply *reply) +{ +struct drm_dp_link_addr_reply_port *port_reply; +int i; + +for (i =0 ; i < reply->nports; i++) { + port_reply = &reply->ports[i]; + DRM_DEBUG_KMS("port %d: input %d, pdt: %d, pn: %d, dpcd_rev: %02x, mcs: %d, ddps: %d, ldps %d, sdp %d/%d\n", + i, + port_reply->input_port, + port_reply->peer_device_type, + port_reply->port_number, + port_reply->dpcd_revision, + port_reply->mcs, + port_reply->ddps, + port_reply->legacy_device_plug_status, + port_reply->num_sdp_streams, + port_reply->num_sdp_stream_sinks); + } +} + static void drm_dp_send_link_address(struct drm_dp_mst_topology_mgr *mgr,drivers/gpu/drm/drm_dp_mst_topology.c +0
-3struct drm_dp_mst_branch *mstb) {@@ -2038,18 +2060,7 @@ static void drm_dp_send_link_address(struct drm_dp_mst_topology_mgr *mgr,DRM_DEBUG_KMS ("link address nak received\n" ); }else {DRM_DEBUG_KMS ("link address reply: %d\n" , txmsg- >reply.u.link_addr.nports); -for (i =0 ; i < txmsg->reply.u.link_addr.nports; i++) { - DRM_DEBUG_KMS("port %d: input %d, pdt: %d, pn: %d, dpcd_rev: %02x, mcs: %d, ddps: %d, ldps %d, sdp %d/%d\n", i, - txmsg-Quoted 1 line
reply.u.link_addr.ports[i].input_port,
drivers/gpu/drm/drm_dp_mst_topology.c +0
-1- txmsg-Quoted 1 line
reply.u.link_addr.ports[i].peer_device_type,
drivers/gpu/drm/drm_dp_mst_topology.c +0
-1- txmsg-Quoted 1 line
reply.u.link_addr.ports[i].port_number,
drivers/gpu/drm/drm_dp_mst_topology.c +0
-1- txmsg-Quoted 1 line
reply.u.link_addr.ports[i].dpcd_revision,
drivers/gpu/drm/drm_dp_mst_topology.c +0
-3- txmsg->reply.u.link_addr.ports[i].mcs, - txmsg->reply.u.link_addr.ports[i].ddps, - txmsg-Quoted 1 line
reply.u.link_addr.ports[i].legacy_device_plug_status,
drivers/gpu/drm/drm_dp_mst_topology.c +0
-1- txmsg-Quoted 1 line
reply.u.link_addr.ports[i].num_sdp_streams,
drivers/gpu/drm/drm_dp_mst_topology.c +0
-1- txmsg-Quoted 1 line
reply.u.link_addr.ports[i].num_sdp_stream_sinks);
drivers/gpu/drm/drm_dp_mst_topology.c +1
-1- } +drm_dp_dump_link_address (&txmsg->reply.u.link_addr); drm_dp_check_mstb_guid(mstb, txmsg-Quoted 1 line
reply.u.link_addr.guid);
Signature · 1 line
2.21.0
Signature · 2 lines
Cheers, Lyude Paul
Re: [PATCH 01/26] drm/dp_mst: Move link address dumping into a function
Quoted 1 line · Lyude Paul
*sigh* finally have some time to go through these reviews
Hey it took me longer to start even reviewing this, and not even through :-( than it took you to reply here. So no worries!
Quoted 8 lines
jfyi: I realized after looking over this patch that it's not actually needed - I had been planning on using
drm_dp_dump_link_address()for other things, but ended up deciding to make the final plan to use something that dumps into a format that's identical to the one we're using for dumping DOWN requests. IMHO though, this patch does make things look nicer so I'll probably keep it.Assuming I can still count your r-b as valid with a change to the commit description?
Sure.
Cheers, Daniel
Quoted 93 lines · Daniel Vetter
Quoted 87 lines · Lyude Paul
Quoted 8 lines
Since we're about to be calling this from multiple places. Also it makes things easier to read!
Cc Juston Li <[email protected]>Cc Imre Deak <[email protected]>Cc Ville Syrjälä <[email protected]>Cc Harry Wentland <[email protected]>Signed-off-by Lyude Paul <[email protected]>Reviewed-by Daniel Vetter <[email protected]>Quoted 75 lines
--- drivers/gpu/drm/drm_dp_mst_topology.c | 35 ++++++++++++++++++--------- 1 file changed, 23 insertions(+), 12 deletions(-)drivers/gpu/drm/drm_dp_mst_topology.c +22
-0@@ -2013,6 +2013,28 @@ static void drm_dp_queue_down_tx(struct drm_dp_mst_topology_mgr *mgr,mutex_unlock (&mgr->qlock); } +static void +drm_dp_dump_link_address (struct drm_dp_link_address_ack_reply *reply) +{ +struct drm_dp_link_addr_reply_port *port_reply; +int i; + +for (i =0 ; i < reply->nports; i++) { + port_reply = &reply->ports[i]; + DRM_DEBUG_KMS("port %d: input %d, pdt: %d, pn: %d, dpcd_rev: %02x, mcs: %d, ddps: %d, ldps %d, sdp %d/%d\n", + i, + port_reply->input_port, + port_reply->peer_device_type, + port_reply->port_number, + port_reply->dpcd_revision, + port_reply->mcs, + port_reply->ddps, + port_reply->legacy_device_plug_status, + port_reply->num_sdp_streams, + port_reply->num_sdp_stream_sinks); + } +} + static void drm_dp_send_link_address(struct drm_dp_mst_topology_mgr *mgr,drivers/gpu/drm/drm_dp_mst_topology.c +0
-3struct drm_dp_mst_branch *mstb) {@@ -2038,18 +2060,7 @@ static void drm_dp_send_link_address(struct drm_dp_mst_topology_mgr *mgr,DRM_DEBUG_KMS ("link address nak received\n" ); }else {DRM_DEBUG_KMS ("link address reply: %d\n" , txmsg- >reply.u.link_addr.nports); -for (i =0 ; i < txmsg->reply.u.link_addr.nports; i++) { - DRM_DEBUG_KMS("port %d: input %d, pdt: %d, pn: %d, dpcd_rev: %02x, mcs: %d, ddps: %d, ldps %d, sdp %d/%d\n", i, - txmsg-Quoted 1 line
reply.u.link_addr.ports[i].input_port,
drivers/gpu/drm/drm_dp_mst_topology.c +0
-1- txmsg-Quoted 1 line
reply.u.link_addr.ports[i].peer_device_type,
drivers/gpu/drm/drm_dp_mst_topology.c +0
-1- txmsg-Quoted 1 line
reply.u.link_addr.ports[i].port_number,
drivers/gpu/drm/drm_dp_mst_topology.c +0
-1- txmsg-Quoted 1 line
reply.u.link_addr.ports[i].dpcd_revision,
drivers/gpu/drm/drm_dp_mst_topology.c +0
-3- txmsg->reply.u.link_addr.ports[i].mcs, - txmsg->reply.u.link_addr.ports[i].ddps, - txmsg-Quoted 1 line
reply.u.link_addr.ports[i].legacy_device_plug_status,
drivers/gpu/drm/drm_dp_mst_topology.c +0
-1- txmsg-Quoted 1 line
reply.u.link_addr.ports[i].num_sdp_streams,
drivers/gpu/drm/drm_dp_mst_topology.c +0
-1- txmsg-Quoted 1 line
reply.u.link_addr.ports[i].num_sdp_stream_sinks);
drivers/gpu/drm/drm_dp_mst_topology.c +1
-1- } +drm_dp_dump_link_address (&txmsg->reply.u.link_addr); drm_dp_check_mstb_guid(mstb, txmsg-Quoted 1 line
reply.u.link_addr.guid);
Signature · 1 line
2.21.0Signature · 2 lines
Cheers, Lyude Paul
Signature · 3 lines
Daniel Vetter Software Engineer, Intel Corporation http://blog.ffwll.ch
Lyude Paul
maintainerPATCH 2/26 drm/dp_mst: Destroy mstbs from destroy_connector_work
In reply to Lyude PaulCurrently we remove MST branch devices from the in-memory topology immediately when their topology refcount reaches 0. This works just fine at the moment, but since we're about to add suspend/resume reprobing for MST topologies we're going to need to be able to travel through the topology and drop topology refs on branch devices while holding mgr->mutex. Since we currently can't do this due to the circular locking dependencies that would incur, move all of the actual work for drm_dp_destroy_mst_branch_device() into drm_dp_destroy_connector_work() so we can drop topology refs on MSTBs in any locking context.
--- drivers/gpu/drm/drm_dp_mst_topology.c | 121 +++++++++++++++++--------- include/drm/drm_dp_mst_helper.h | 10 +++ 2 files changed, 90 insertions(+), 41 deletions(-)
drivers/gpu/drm/drm_dp_mst_topology.c +80 -41
@@ -1108,34 +1108,17 @@ static void drm_dp_destroy_mst_branch_device(struct kref *kref)struct drm_dp_mst_branch *mstb = container_of(kref, struct drm_dp_mst_branch, topology_kref);struct drm_dp_mst_topology_mgr *mgr = mstb->mgr; -struct drm_dp_mst_port *port, *tmp; -bool wake_tx = false; - -mutex_lock (&mgr->lock); -list_for_each_entry_safe (port, tmp, &mstb->ports, next) { -list_del (&port->next); -drm_dp_mst_topology_put_port (port); - } -mutex_unlock (&mgr->lock); - -/* drop any tx slots msg */ -mutex_lock (&mstb->mgr->qlock); -if (mstb->tx_slots[0 ]) { - mstb->tx_slots[0 ]->state =DRM_DP_SIDEBAND_TX_TIMEOUT ; - mstb->tx_slots[0 ] =NULL ; - wake_tx = true; - } -if (mstb->tx_slots[1 ]) { - mstb->tx_slots[1 ]->state =DRM_DP_SIDEBAND_TX_TIMEOUT ; - mstb->tx_slots[1 ] =NULL ; - wake_tx = true; - } -mutex_unlock (&mstb->mgr->qlock); -if (wake_tx) -wake_up_all (&mstb->mgr->tx_waitq); +INIT_LIST_HEAD (&mstb->destroy_next); -drm_dp_mst_put_mstb_malloc (mstb); +/* +* This can get called under mgr->mutex, so we need to perform the +* actual destruction of the mstb in another worker +*/ +mutex_lock (&mgr->destroy_connector_lock); +list_add (&mstb->destroy_next, &mgr->destroy_branch_device_list); +mutex_unlock (&mgr->destroy_connector_lock); +schedule_work (&mgr->destroy_connector_work); }/**
@@ -3618,10 +3601,56 @@ static void drm_dp_tx_work(struct work_struct *work)mutex_unlock (&mgr->qlock); } +static inline void +drm_dp_finish_destroy_port (struct drm_dp_mst_port *port) +{ +INIT_LIST_HEAD (&port->next); + + port->mgr->cbs->destroy_connector (port->mgr, port->connector); + +drm_dp_port_teardown_pdt (port, port->pdt); + port->pdt =DP_PEER_DEVICE_NONE ; + +drm_dp_mst_put_port_malloc (port); +} + +static inline void +drm_dp_finish_destroy_mst_branch_device (struct drm_dp_mst_branch *mstb) +{ +struct drm_dp_mst_topology_mgr *mgr = mstb->mgr; +struct drm_dp_mst_port *port, *tmp; +bool wake_tx = false; + +mutex_lock (&mgr->lock); +list_for_each_entry_safe (port, tmp, &mstb->ports, next) { +list_del (&port->next); +drm_dp_mst_topology_put_port (port); + } +mutex_unlock (&mgr->lock); + +/* drop any tx slots msg */ +mutex_lock (&mstb->mgr->qlock); +if (mstb->tx_slots[0 ]) { + mstb->tx_slots[0 ]->state =DRM_DP_SIDEBAND_TX_TIMEOUT ; + mstb->tx_slots[0 ] =NULL ; + wake_tx = true; + } +if (mstb->tx_slots[1 ]) { + mstb->tx_slots[1 ]->state =DRM_DP_SIDEBAND_TX_TIMEOUT ; + mstb->tx_slots[1 ] =NULL ; + wake_tx = true; + } +mutex_unlock (&mstb->mgr->qlock); + +if (wake_tx) +wake_up_all (&mstb->mgr->tx_waitq); + +drm_dp_mst_put_mstb_malloc (mstb); +} +static void drm_dp_destroy_connector_work (struct work_struct *work) { struct drm_dp_mst_topology_mgr *mgr = container_of(work, struct drm_dp_mst_topology_mgr, destroy_connector_work); -struct drm_dp_mst_port *port;bool send_hotplug = false;/* * Not a regular list traverse as we have to drop the destroy
@@ -3629,24 +3658,33 @@ static void drm_dp_destroy_connector_work(struct work_struct *work)* ordering between this lock and the config mutex. */ for (;;) { +struct drm_dp_mst_branch *mstb =NULL ; +struct drm_dp_mst_port *port =NULL ; + +/* Destroy any MSTBs first, and then their ports second */ mutex_lock (&mgr->destroy_connector_lock); - port = list_first_entry_or_null(&mgr->destroy_connector_list, struct drm_dp_mst_port, next); -if (!port) { -mutex_unlock (&mgr->destroy_connector_lock); -break ; + mstb =list_first_entry_or_null (&mgr->destroy_branch_device_list, + struct drm_dp_mst_branch, + destroy_next); +if (mstb) { +list_del (&mstb->destroy_next); + }else { + port =list_first_entry_or_null (&mgr->destroy_connector_list, + struct drm_dp_mst_port, + next); +if (port) +list_del (&port->next); } -list_del (&port->next);mutex_unlock (&mgr->destroy_connector_lock); -INIT_LIST_HEAD (&port->next); - - mgr->cbs->destroy_connector (mgr, port->connector); - -drm_dp_port_teardown_pdt (port, port->pdt); - port->pdt =DP_PEER_DEVICE_NONE ; - -drm_dp_mst_put_port_malloc (port); - send_hotplug = true; +if (mstb) { +drm_dp_finish_destroy_mst_branch_device (mstb); + }else if (port) { +drm_dp_finish_destroy_port (port); + send_hotplug = true; + }else { +break ; + } }if (send_hotplug)drm_kms_helper_hotplug_event (mgr->dev);
@@ -3840,6 +3878,7 @@ int drm_dp_mst_topology_mgr_init(struct drm_dp_mst_topology_mgr *mgr,mutex_init (&mgr->destroy_connector_lock);INIT_LIST_HEAD (&mgr->tx_msg_downq);INIT_LIST_HEAD (&mgr->destroy_connector_list); +INIT_LIST_HEAD (&mgr->destroy_branch_device_list);INIT_WORK (&mgr->work, drm_dp_mst_link_probe_work);INIT_WORK (&mgr->tx_work, drm_dp_tx_work);INIT_WORK (&mgr->destroy_connector_work, drm_dp_destroy_connector_work);
include/drm/drm_dp_mst_helper.h +10 -0
@@ -143,6 +143,12 @@ struct drm_dp_mst_branch {*/ struct kref malloc_kref; +/** +* @destroy_next: linked-list entry used by +* drm_dp_destroy_connector_work() +*/ +struct list_head destroy_next; +u8 rad[8 ];u8 lct;int num_ports;
@@ -578,6 +584,10 @@ struct drm_dp_mst_topology_mgr {* @destroy_connector_list: List of to be destroyed connectors. */ struct list_head destroy_connector_list; +/** +* @destroy_branch_device_list: List of to be destroyed branch devices +*/ +struct list_head destroy_branch_device_list;/** * @destroy_connector_lock: Protects @connector_list. */
Signature · 1 line
2.21.0
Re: [PATCH 02/26] drm/dp_mst: Destroy mstbs from destroy_connector_work
Quoted 9 lines · Lyude Paul
Currently we remove MST branch devices from the in-memory topology immediately when their topology refcount reaches 0. This works just fine at the moment, but since we're about to add suspend/resume reprobing for MST topologies we're going to need to be able to travel through the topology and drop topology refs on branch devices while holding
mgr->mutex. Since we currently can't do this due to the circular locking dependencies that would incur, move all of the actual work fordrm_dp_destroy_mst_branch_device()intodrm_dp_destroy_connector_work()so we can drop topology refs on MSTBs in any locking context.
Would be good to point at where exactly the problem is here, maybe also mentioned the exact future patch that causes the problem. I did go look around a bit, but didn't spot it.
Quoted 60 lines
Cc Juston Li <[email protected]>Cc Imre Deak <[email protected]>Cc Ville Syrjälä <[email protected]>Cc Harry Wentland <[email protected]>Signed-off-by Lyude Paul <[email protected]>--- drivers/gpu/drm/drm_dp_mst_topology.c | 121 +++++++++++++++++--------- include/drm/drm_dp_mst_helper.h | 10 +++ 2 files changed, 90 insertions(+), 41 deletions(-)drivers/gpu/drm/drm_dp_mst_topology.c +9
-26@@ -1108,34 +1108,17 @@ static void drm_dp_destroy_mst_branch_device(struct kref *kref)struct drm_dp_mst_branch *mstb = container_of(kref, struct drm_dp_mst_branch, topology_kref);struct drm_dp_mst_topology_mgr *mgr = mstb->mgr; -struct drm_dp_mst_port *port, *tmp; -bool wake_tx = false; - -mutex_lock (&mgr->lock); -list_for_each_entry_safe (port, tmp, &mstb->ports, next) { -list_del (&port->next); -drm_dp_mst_topology_put_port (port); - } -mutex_unlock (&mgr->lock); - -/* drop any tx slots msg */ -mutex_lock (&mstb->mgr->qlock); -if (mstb->tx_slots[0 ]) { - mstb->tx_slots[0 ]->state =DRM_DP_SIDEBAND_TX_TIMEOUT ; - mstb->tx_slots[0 ] =NULL ; - wake_tx = true; - } -if (mstb->tx_slots[1 ]) { - mstb->tx_slots[1 ]->state =DRM_DP_SIDEBAND_TX_TIMEOUT ; - mstb->tx_slots[1 ] =NULL ; - wake_tx = true; - } -mutex_unlock (&mstb->mgr->qlock); -if (wake_tx) -wake_up_all (&mstb->mgr->tx_waitq); +INIT_LIST_HEAD (&mstb->destroy_next); -drm_dp_mst_put_mstb_malloc (mstb); +/* +* This can get called under mgr->mutex, so we need to perform the +* actual destruction of the mstb in another worker +*/ +mutex_lock (&mgr->destroy_connector_lock); +list_add (&mstb->destroy_next, &mgr->destroy_branch_device_list); +mutex_unlock (&mgr->destroy_connector_lock); +schedule_work (&mgr->destroy_connector_work); }/**
drivers/gpu/drm/drm_dp_mst_topology.c static void drm_dp_tx_work(struct work_struct *work) · line 3605
Quoted 5 lines
drivers/gpu/drm/drm_dp_mst_topology.c +2
-0@@ -3618,10 +3601,56 @@ static void drm_dp_tx_work(struct work_struct *work)mutex_unlock (&mgr->qlock); } +static inline void +drm_dp_finish_destroy_port (struct drm_dp_mst_port *port)
Bikeshed: I'd call this _delayed_destroy, I think that makes a bit clearer why there's 2 stages to destroying stuff.
Quoted 67 lines
drivers/gpu/drm/drm_dp_mst_topology.c +51
-5+{ +INIT_LIST_HEAD (&port->next); + + port->mgr->cbs->destroy_connector (port->mgr, port->connector); + +drm_dp_port_teardown_pdt (port, port->pdt); + port->pdt =DP_PEER_DEVICE_NONE ; + +drm_dp_mst_put_port_malloc (port); +} + +static inline void +drm_dp_finish_destroy_mst_branch_device (struct drm_dp_mst_branch *mstb) +{ +struct drm_dp_mst_topology_mgr *mgr = mstb->mgr; +struct drm_dp_mst_port *port, *tmp; +bool wake_tx = false; + +mutex_lock (&mgr->lock); +list_for_each_entry_safe (port, tmp, &mstb->ports, next) { +list_del (&port->next); +drm_dp_mst_topology_put_port (port); + } +mutex_unlock (&mgr->lock); + +/* drop any tx slots msg */ +mutex_lock (&mstb->mgr->qlock); +if (mstb->tx_slots[0 ]) { + mstb->tx_slots[0 ]->state =DRM_DP_SIDEBAND_TX_TIMEOUT ; + mstb->tx_slots[0 ] =NULL ; + wake_tx = true; + } +if (mstb->tx_slots[1 ]) { + mstb->tx_slots[1 ]->state =DRM_DP_SIDEBAND_TX_TIMEOUT ; + mstb->tx_slots[1 ] =NULL ; + wake_tx = true; + } +mutex_unlock (&mstb->mgr->qlock); + +if (wake_tx) +wake_up_all (&mstb->mgr->tx_waitq); + +drm_dp_mst_put_mstb_malloc (mstb); +} +static void drm_dp_destroy_connector_work (struct work_struct *work) { struct drm_dp_mst_topology_mgr *mgr = container_of(work, struct drm_dp_mst_topology_mgr, destroy_connector_work); -struct drm_dp_mst_port *port;bool send_hotplug = false;/* * Not a regular list traverse as we have to drop the destroy @@ -3629,24 +3658,33 @@ static void drm_dp_destroy_connector_work(struct work_struct *work)* ordering between this lock and the config mutex. */ for (;;) { +struct drm_dp_mst_branch *mstb =NULL ; +struct drm_dp_mst_port *port =NULL ; + +/* Destroy any MSTBs first, and then their ports second */ mutex_lock (&mgr->destroy_connector_lock); - port = list_first_entry_or_null(&mgr->destroy_connector_list, struct drm_dp_mst_port, next); -if (!port) { -mutex_unlock (&mgr->destroy_connector_lock); -break ; + mstb = list_first_entry_or_null(&mgr->destroy_branch_device_list, + struct drm_dp_mst_branch,
drivers/gpu/drm/drm_dp_mst_topology.c static void drm_dp_destroy_connector_work(struct work_struct *work) · line 3677
Quoted 10 lines
drivers/gpu/drm/drm_dp_mst_topology.c +9
-0+ destroy_next); +if (mstb) { +list_del (&mstb->destroy_next); + }else { + port =list_first_entry_or_null (&mgr->destroy_connector_list, + struct drm_dp_mst_port, + next); +if (port) +list_del (&port->next); }
Control flow looks rather awkward here. I'd do either two loops, or if you want to have just one, rename it to ->delayed_destroy_list and have a ->delayed_destry_cb next to it?
Cheers, Daniel
Quoted 62 lines
drivers/gpu/drm/drm_dp_mst_topology.c +9
-10-list_del (&port->next);mutex_unlock (&mgr->destroy_connector_lock); -INIT_LIST_HEAD (&port->next); - - mgr->cbs->destroy_connector (mgr, port->connector); - -drm_dp_port_teardown_pdt (port, port->pdt); - port->pdt =DP_PEER_DEVICE_NONE ; - -drm_dp_mst_put_port_malloc (port); - send_hotplug = true; +if (mstb) { +drm_dp_finish_destroy_mst_branch_device (mstb); + }else if (port) { +drm_dp_finish_destroy_port (port); + send_hotplug = true; + }else { +break ; + } }if (send_hotplug)drm_kms_helper_hotplug_event (mgr->dev);@@ -3840,6 +3878,7 @@ int drm_dp_mst_topology_mgr_init(struct drm_dp_mst_topology_mgr *mgr,mutex_init (&mgr->destroy_connector_lock);INIT_LIST_HEAD (&mgr->tx_msg_downq);INIT_LIST_HEAD (&mgr->destroy_connector_list); +INIT_LIST_HEAD (&mgr->destroy_branch_device_list);INIT_WORK (&mgr->work, drm_dp_mst_link_probe_work);INIT_WORK (&mgr->tx_work, drm_dp_tx_work);INIT_WORK (&mgr->destroy_connector_work, drm_dp_destroy_connector_work);include/drm/drm_dp_mst_helper.h +10
-0@@ -143,6 +143,12 @@ struct drm_dp_mst_branch {*/ struct kref malloc_kref; +/** +* @destroy_next: linked-list entry used by +* drm_dp_destroy_connector_work() +*/ +struct list_head destroy_next; +u8 rad[8 ];u8 lct;int num_ports;@@ -578,6 +584,10 @@ struct drm_dp_mst_topology_mgr {* @destroy_connector_list: List of to be destroyed connectors. */ struct list_head destroy_connector_list; +/** +* @destroy_branch_device_list: List of to be destroyed branch devices +*/ +struct list_head destroy_branch_device_list;/** * @destroy_connector_lock: Protects @connector_list. */ Signature · 1 line
2.21.0
Signature · 3 lines
Daniel Vetter Software Engineer, Intel Corporation http://blog.ffwll.ch
Lyude Paul
maintainerRe: [PATCH 02/26] drm/dp_mst: Destroy mstbs from destroy_connector_work
Quoted 11 lines · Daniel Vetter
Quoted 9 lines · Lyude Paul
Currently we remove MST branch devices from the in-memory topology immediately when their topology refcount reaches 0. This works just fine at the moment, but since we're about to add suspend/resume reprobing for MST topologies we're going to need to be able to travel through the topology and drop topology refs on branch devices while holding
mgr->mutex. Since we currently can't do this due to the circular locking dependencies that would incur, move all of the actual work fordrm_dp_destroy_mst_branch_device()intodrm_dp_destroy_connector_work()so we can drop topology refs on MSTBs in any locking context.
Quoted 3 lines
Would be good to point at where exactly the problem is here, maybe also mentioned the exact future patch that causes the problem. I did go look around a bit, but didn't spot it.
Ah, I didn't do a great job of explaining whoops! Basically, the issue is that when reprobing the state during suspend/resume, there's of course going to be a chance that any MST ports in the topology could have had their PDT changed while we were suspended. This in turn means that while we're iterating through each mstb's ports in drm_dp_send_link_address() we want to be able to remove MSTBs from while under &mgr->lock, something we can't do without handling MSTB destruction in another thread. I'll mention this in the next version of this patch
Quoted 228 lines
Quoted 62 lines
Cc Juston Li <[email protected]>Cc Imre Deak <[email protected]>Cc Ville Syrjälä <[email protected]>Cc Harry Wentland <[email protected]>Signed-off-by Lyude Paul <[email protected]>--- drivers/gpu/drm/drm_dp_mst_topology.c | 121 +++++++++++++++++--------- include/drm/drm_dp_mst_helper.h | 10 +++ 2 files changed, 90 insertions(+), 41 deletions(-)drivers/gpu/drm/drm_dp_mst_topology.c +9
-26@@ -1108,34 +1108,17 @@ static void drm_dp_destroy_mst_branch_device(struct kref *kref) struct drm_dp_mst_branch *mstb = container_of(kref, struct drm_dp_mst_branch, topology_kref);struct drm_dp_mst_topology_mgr *mgr = mstb->mgr; -struct drm_dp_mst_port *port, *tmp; -bool wake_tx = false; - -mutex_lock (&mgr->lock); -list_for_each_entry_safe (port, tmp, &mstb->ports, next) { -list_del (&port->next); -drm_dp_mst_topology_put_port (port); - } -mutex_unlock (&mgr->lock); - -/* drop any tx slots msg */ -mutex_lock (&mstb->mgr->qlock); -if (mstb->tx_slots[0 ]) { - mstb->tx_slots[0 ]->state =DRM_DP_SIDEBAND_TX_TIMEOUT ; - mstb->tx_slots[0 ] =NULL ; - wake_tx = true; - } -if (mstb->tx_slots[1 ]) { - mstb->tx_slots[1 ]->state =DRM_DP_SIDEBAND_TX_TIMEOUT ; - mstb->tx_slots[1 ] =NULL ; - wake_tx = true; - } -mutex_unlock (&mstb->mgr->qlock); -if (wake_tx) -wake_up_all (&mstb->mgr->tx_waitq); +INIT_LIST_HEAD (&mstb->destroy_next); -drm_dp_mst_put_mstb_malloc (mstb); +/* +* This can get called under mgr->mutex, so we need to perform the +* actual destruction of the mstb in another worker +*/ +mutex_lock (&mgr->destroy_connector_lock); +list_add (&mstb->destroy_next, &mgr->destroy_branch_device_list); +mutex_unlock (&mgr->destroy_connector_lock); +schedule_work (&mgr->destroy_connector_work); }/**
drivers/gpu/drm/drm_dp_mst_topology.c static void drm_dp_tx_work(struct work_struct · line 3606
Quoted 6 lines
drivers/gpu/drm/drm_dp_mst_topology.c +2
-0@@ -3618,10 +3601,56 @@ static void drm_dp_tx_work(struct work_struct *work)mutex_unlock (&mgr->qlock); } +static inline void +drm_dp_finish_destroy_port (struct drm_dp_mst_port *port)Bikeshed: I'd call this
_delayed_destroy, I think that makes a bit clearer why there's 2 stages to destroying stuff.Quoted 73 lines
drivers/gpu/drm/drm_dp_mst_topology.c +45
-1+{ +INIT_LIST_HEAD (&port->next); + + port->mgr->cbs->destroy_connector (port->mgr, port->connector); + +drm_dp_port_teardown_pdt (port, port->pdt); + port->pdt =DP_PEER_DEVICE_NONE ; + +drm_dp_mst_put_port_malloc (port); +} + +static inline void +drm_dp_finish_destroy_mst_branch_device (struct drm_dp_mst_branch *mstb) +{ +struct drm_dp_mst_topology_mgr *mgr = mstb->mgr; +struct drm_dp_mst_port *port, *tmp; +bool wake_tx = false; + +mutex_lock (&mgr->lock); +list_for_each_entry_safe (port, tmp, &mstb->ports, next) { +list_del (&port->next); +drm_dp_mst_topology_put_port (port); + } +mutex_unlock (&mgr->lock); + +/* drop any tx slots msg */ +mutex_lock (&mstb->mgr->qlock); +if (mstb->tx_slots[0 ]) { + mstb->tx_slots[0 ]->state =DRM_DP_SIDEBAND_TX_TIMEOUT ; + mstb->tx_slots[0 ] =NULL ; + wake_tx = true; + } +if (mstb->tx_slots[1 ]) { + mstb->tx_slots[1 ]->state =DRM_DP_SIDEBAND_TX_TIMEOUT ; + mstb->tx_slots[1 ] =NULL ; + wake_tx = true; + } +mutex_unlock (&mstb->mgr->qlock); + +if (wake_tx) +wake_up_all (&mstb->mgr->tx_waitq); + +drm_dp_mst_put_mstb_malloc (mstb); +} +static void drm_dp_destroy_connector_work (struct work_struct *work) { struct drm_dp_mst_topology_mgr *mgr = container_of(work, struct drm_dp_mst_topology_mgr, destroy_connector_work); -struct drm_dp_mst_port *port;bool send_hotplug = false;drivers/gpu/drm/drm_dp_mst_topology.c +8
-4/* * Not a regular list traverse as we have to drop the destroy @@ -3629,24 +3658,33 @@ static void drm_dp_destroy_connector_work(structwork_struct *work) * ordering between this lock and the config mutex. */ for (;;) { +struct drm_dp_mst_branch *mstb =NULL ; +struct drm_dp_mst_port *port =NULL ; + +/* Destroy any MSTBs first, and then their ports second */ mutex_lock (&mgr->destroy_connector_lock); - port =list_first_entry_or_null (&mgr->destroy_connector_list, struct drm_dp_mst_port, next); -if (!port) { -mutex_unlock (&mgr->destroy_connector_lock); -break ; + mstb =list_first_entry_or_null (&mgr- >destroy_branch_device_list, + struct drm_dp_mst_branch, + destroy_next); +if (mstb) {drivers/gpu/drm/drm_dp_mst_topology.c static void drm_dp_destroy_connector_work(struct · line 3682
Quoted 10 lines
drivers/gpu/drm/drm_dp_mst_topology.c +7
-0+list_del (&mstb->destroy_next); + }else { + port =list_first_entry_or_null (&mgr- >destroy_connector_list, + struct drm_dp_mst_port, + next); +if (port) +list_del (&port->next); }Control flow looks rather awkward here. I'd do either two loops, or if you want to have just one, rename it to ->delayed_destroy_list and have a ->delayed_destry_cb next to it?
Cheers, Daniel
Quoted 65 lines
drivers/gpu/drm/drm_dp_mst_topology.c +9
-10-list_del (&port->next);mutex_unlock (&mgr->destroy_connector_lock); -INIT_LIST_HEAD (&port->next); - - mgr->cbs->destroy_connector (mgr, port->connector); - -drm_dp_port_teardown_pdt (port, port->pdt); - port->pdt =DP_PEER_DEVICE_NONE ; - -drm_dp_mst_put_port_malloc (port); - send_hotplug = true; +if (mstb) { +drm_dp_finish_destroy_mst_branch_device (mstb); + }else if (port) { +drm_dp_finish_destroy_port (port); + send_hotplug = true; + }else { +break ; + } }if (send_hotplug)drm_kms_helper_hotplug_event (mgr->dev);@@ -3840,6 +3878,7 @@ int drm_dp_mst_topology_mgr_init(struct drm_dp_mst_topology_mgr *mgr,mutex_init (&mgr->destroy_connector_lock);INIT_LIST_HEAD (&mgr->tx_msg_downq);INIT_LIST_HEAD (&mgr->destroy_connector_list); +INIT_LIST_HEAD (&mgr->destroy_branch_device_list);INIT_WORK (&mgr->work, drm_dp_mst_link_probe_work);INIT_WORK (&mgr->tx_work, drm_dp_tx_work);INIT_WORK(&mgr->destroy_connector_work,
drm_dp_destroy_connector_work);include/drm/drm_dp_mst_helper.h +10
-0@@ -143,6 +143,12 @@ struct drm_dp_mst_branch {*/ struct kref malloc_kref; +/** +* @destroy_next: linked-list entry used by +* drm_dp_destroy_connector_work() +*/ +struct list_head destroy_next; +u8 rad[8 ];u8 lct;int num_ports;@@ -578,6 +584,10 @@ struct drm_dp_mst_topology_mgr {* @destroy_connector_list: List of to be destroyed connectors. */ struct list_head destroy_connector_list; +/** +* @destroy_branch_device_list: List of to be destroyed branch devices +*/ +struct list_head destroy_branch_device_list;/** * @destroy_connector_lock: Protects @connector_list. */ Signature · 1 line
2.21.0
Signature · 2 lines
Cheers, Lyude Paul
Lyude Paul
maintainerPATCH 3/26 drm/dp_mst: Move test_calc_pbn_mode() into an actual selftest
In reply to Lyude PaulYes, apparently we've been testing this for every single driver load for quite a long time now. At least that means our PBN calculation is solid!
Anyway, introduce self tests for MST and move this into there.
--- drivers/gpu/drm/drm_dp_mst_topology.c | 27 -------------- drivers/gpu/drm/selftests/Makefile | 2 +- .../gpu/drm/selftests/drm_modeset_selftests.h | 1 + .../drm/selftests/test-drm_dp_mst_helper.c | 36 +++++++++++++++++++ .../drm/selftests/test-drm_modeset_common.h | 1 + 5 files changed, 39 insertions(+), 28 deletions(-) create mode 100644 drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c
drivers/gpu/drm/drm_dp_mst_topology.c +0 -27
@@ -45,7 +45,6 @@*/ static bool dump_dp_payload_table (struct drm_dp_mst_topology_mgr *mgr,char *buf); -static int test_calc_pbn_mode (void );static void drm_dp_mst_topology_put_port (struct drm_dp_mst_port *port);
@@ -3439,30 +3438,6 @@ int drm_dp_calc_pbn_mode(int clock, int bpp) }EXPORT_SYMBOL (drm_dp_calc_pbn_mode); -static int test_calc_pbn_mode (void ) -{ -int ret; - ret =drm_dp_calc_pbn_mode (154000 ,30 ); -if (ret !=689 ) { -DRM_ERROR ("PBN calculation test failed - clock %d, bpp %d, expected PBN %d, actual PBN %d.\n" , -154000 ,30 ,689 , ret); -return -EINVAL ; - } - ret =drm_dp_calc_pbn_mode (234000 ,30 ); -if (ret !=1047 ) { -DRM_ERROR ("PBN calculation test failed - clock %d, bpp %d, expected PBN %d, actual PBN %d.\n" , -234000 ,30 ,1047 , ret); -return -EINVAL ; - } - ret =drm_dp_calc_pbn_mode (297000 ,24 ); -if (ret !=1063 ) { -DRM_ERROR ("PBN calculation test failed - clock %d, bpp %d, expected PBN %d, actual PBN %d.\n" , -297000 ,24 ,1063 , ret); -return -EINVAL ; - } -return 0 ; -} -/* we want to kick the TX after we've ack the up/down IRQs. */ static void drm_dp_mst_kick_tx (struct drm_dp_mst_topology_mgr *mgr) {
@@ -3898,8 +3873,6 @@ int drm_dp_mst_topology_mgr_init(struct drm_dp_mst_topology_mgr *mgr,if (!mgr->proposed_vcpis)return -ENOMEM ;set_bit (0 , &mgr->payload_mask); -if (test_calc_pbn_mode () <0 ) -DRM_ERROR ("MST PBN self-test failed\n" ); mst_state =kzalloc (sizeof (*mst_state),GFP_KERNEL );if (mst_state ==NULL )
drivers/gpu/drm/selftests/Makefile +1 -1
@@ -1,6 +1,6 @@ # SPDX-License-Identifier: GPL-2.0-only test-drm_modeset-y := test-drm_modeset_common.o test-drm_plane_helper.o \ test-drm_format.o test-drm_framebuffer.o \ - test-drm_damage_helper.o + test-drm_damage_helper.o test-drm_dp_mst_helper.o obj-$(CONFIG_DRM_DEBUG_SELFTEST) += test-drm_mm.o test-drm_modeset.o test-drm_cmdline_parser.o
drivers/gpu/drm/selftests/drm_modeset_selftests.h +1 -0
@@ -32,3 +32,4 @@ selftest(damage_iter_damage_one_intersect, igt_damage_iter_damage_one_intersect) selftest(damage_iter_damage_one_outside, igt_damage_iter_damage_one_outside) selftest(damage_iter_damage_src_moved, igt_damage_iter_damage_src_moved) selftest(damage_iter_damage_not_visible, igt_damage_iter_damage_not_visible) +selftest (dp_mst_calc_pbn_mode, igt_dp_mst_calc_pbn_mode)
drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c new file +36 -0
@@ -0,0 +1,36 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* +* Test cases for for the DRM DP MST helpers +*/ + +#define pr_fmt (fmt) "drm_dp_mst_helper: " fmt + +#include <drm/drm_dp_mst_helper.h> +#include <drm/drm_print.h> + +#include "test-drm_modeset_common.h" + +int igt_dp_mst_calc_pbn_mode (void *ignored) +{ +int pbn, i; +const struct { +int rate; +int bpp; +int expected; + } test_params[] = { + {154000 ,30 ,689 }, + {234000 ,30 ,1047 }, + {297000 ,24 ,1063 }, + }; + +for (i =0 ; i <ARRAY_SIZE (test_params); i++) { + pbn =drm_dp_calc_pbn_mode (test_params[i].rate, + test_params[i].bpp); +FAIL (pbn != test_params[i].expected, +"Expected PBN %d for clock %d bpp %d, got %d\n" , + test_params[i].expected, test_params[i].rate, + test_params[i].bpp, pbn); + } + +return 0 ; +}
drivers/gpu/drm/selftests/test-drm_modeset_common.h +1 -0
@@ -39,5 +39,6 @@ int igt_damage_iter_damage_one_intersect(void *ignored);int igt_damage_iter_damage_one_outside (void *ignored);int igt_damage_iter_damage_src_moved (void *ignored);int igt_damage_iter_damage_not_visible (void *ignored); +int igt_dp_mst_calc_pbn_mode (void *ignored);#endif
Signature · 1 line
2.21.0
Re: [PATCH 03/26] drm/dp_mst: Move test_calc_pbn_mode() into an actual selftest
Quoted 10 lines · Lyude Paul
Yes, apparently we've been testing this for every single driver load for quite a long time now. At least that means our PBN calculation is solid!
Anyway, introduce self tests for MST and move this into there.
Cc Juston Li <[email protected]>Cc Imre Deak <[email protected]>Cc Ville Syrjälä <[email protected]>Cc Harry Wentland <[email protected]>Signed-off-by Lyude Paul <[email protected]>
More official unit tests, yay!
Quoted 138 lines
--- drivers/gpu/drm/drm_dp_mst_topology.c | 27 -------------- drivers/gpu/drm/selftests/Makefile | 2 +- .../gpu/drm/selftests/drm_modeset_selftests.h | 1 + .../drm/selftests/test-drm_dp_mst_helper.c | 36 +++++++++++++++++++ .../drm/selftests/test-drm_modeset_common.h | 1 + 5 files changed, 39 insertions(+), 28 deletions(-) create mode 100644 drivers/gpu/drm/selftests/test-drm_dp_mst_helper.cdrivers/gpu/drm/drm_dp_mst_topology.c +0
-27@@ -45,7 +45,6 @@*/ static bool dump_dp_payload_table (struct drm_dp_mst_topology_mgr *mgr,char *buf); -static int test_calc_pbn_mode (void );static void drm_dp_mst_topology_put_port (struct drm_dp_mst_port *port);@@ -3439,30 +3438,6 @@ int drm_dp_calc_pbn_mode(int clock, int bpp) }EXPORT_SYMBOL (drm_dp_calc_pbn_mode); -static int test_calc_pbn_mode (void ) -{ -int ret; - ret =drm_dp_calc_pbn_mode (154000 ,30 ); -if (ret !=689 ) { -DRM_ERROR ("PBN calculation test failed - clock %d, bpp %d, expected PBN %d, actual PBN %d.\n" , -154000 ,30 ,689 , ret); -return -EINVAL ; - } - ret =drm_dp_calc_pbn_mode (234000 ,30 ); -if (ret !=1047 ) { -DRM_ERROR ("PBN calculation test failed - clock %d, bpp %d, expected PBN %d, actual PBN %d.\n" , -234000 ,30 ,1047 , ret); -return -EINVAL ; - } - ret =drm_dp_calc_pbn_mode (297000 ,24 ); -if (ret !=1063 ) { -DRM_ERROR ("PBN calculation test failed - clock %d, bpp %d, expected PBN %d, actual PBN %d.\n" , -297000 ,24 ,1063 , ret); -return -EINVAL ; - } -return 0 ; -} -/* we want to kick the TX after we've ack the up/down IRQs. */ static void drm_dp_mst_kick_tx (struct drm_dp_mst_topology_mgr *mgr) {@@ -3898,8 +3873,6 @@ int drm_dp_mst_topology_mgr_init(struct drm_dp_mst_topology_mgr *mgr,if (!mgr->proposed_vcpis)return -ENOMEM ;set_bit (0 , &mgr->payload_mask); -if (test_calc_pbn_mode () <0 ) -DRM_ERROR ("MST PBN self-test failed\n" ); mst_state =kzalloc (sizeof (*mst_state),GFP_KERNEL );if (mst_state ==NULL )drivers/gpu/drm/selftests/Makefile +1
-1@@ -1,6 +1,6 @@ # SPDX-License-Identifier: GPL-2.0-only test-drm_modeset-y := test-drm_modeset_common.o test-drm_plane_helper.o \ test-drm_format.o test-drm_framebuffer.o \ - test-drm_damage_helper.o + test-drm_damage_helper.o test-drm_dp_mst_helper.o obj-$(CONFIG_DRM_DEBUG_SELFTEST) += test-drm_mm.o test-drm_modeset.o test-drm_cmdline_parser.odrivers/gpu/drm/selftests/drm_modeset_selftests.h +1
-0@@ -32,3 +32,4 @@ selftest(damage_iter_damage_one_intersect, igt_damage_iter_damage_one_intersect) selftest(damage_iter_damage_one_outside, igt_damage_iter_damage_one_outside) selftest(damage_iter_damage_src_moved, igt_damage_iter_damage_src_moved) selftest(damage_iter_damage_not_visible, igt_damage_iter_damage_not_visible) +selftest (dp_mst_calc_pbn_mode, igt_dp_mst_calc_pbn_mode)drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c new file +36
-0@@ -0,0 +1,36 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* +* Test cases for for the DRM DP MST helpers +*/ + +#define pr_fmt (fmt) "drm_dp_mst_helper: " fmt + +#include <drm/drm_dp_mst_helper.h> +#include <drm/drm_print.h> + +#include "test-drm_modeset_common.h" + +int igt_dp_mst_calc_pbn_mode (void *ignored) +{ +int pbn, i; +const struct { +int rate; +int bpp; +int expected; + } test_params[] = { + {154000 ,30 ,689 }, + {234000 ,30 ,1047 }, + {297000 ,24 ,1063 }, + }; + +for (i =0 ; i <ARRAY_SIZE (test_params); i++) { + pbn =drm_dp_calc_pbn_mode (test_params[i].rate, + test_params[i].bpp); +FAIL (pbn != test_params[i].expected, +"Expected PBN %d for clock %d bpp %d, got %d\n" , + test_params[i].expected, test_params[i].rate, + test_params[i].bpp, pbn); + } + +return 0 ; +}drivers/gpu/drm/selftests/test-drm_modeset_common.h +1
-0@@ -39,5 +39,6 @@ int igt_damage_iter_damage_one_intersect(void *ignored);int igt_damage_iter_damage_one_outside (void *ignored);int igt_damage_iter_damage_src_moved (void *ignored);int igt_damage_iter_damage_not_visible (void *ignored); +int igt_dp_mst_calc_pbn_mode (void *ignored);#endif Signature · 1 line
2.21.0
Signature · 3 lines
Daniel Vetter Software Engineer, Intel Corporation http://blog.ffwll.ch
Lyude Paul
maintainerPATCH 4/26 drm/print: Add drm_err_printer()
In reply to Lyude PaulA simple convienence function that returns a drm_printer which prints using pr_err()
--- drivers/gpu/drm/drm_print.c | 6 ++++++ include/drm/drm_print.h | 17 +++++++++++++++++ 2 files changed, 23 insertions(+)
drivers/gpu/drm/drm_print.c +6 -0
@@ -147,6 +147,12 @@ void __drm_printfn_debug(struct drm_printer *p, struct va_format *vaf) }EXPORT_SYMBOL (__drm_printfn_debug); +void __drm_printfn_err (struct drm_printer *p,struct va_format *vaf) +{ +pr_err ("%s %pV" , p->prefix, vaf); +} +EXPORT_SYMBOL (__drm_printfn_err); +/** * drm_puts - print a const string to a &drm_printer stream * @p: the &drm printer
include/drm/drm_print.h +17 -0
@@ -83,6 +83,7 @@ void __drm_printfn_seq_file(struct drm_printer *p, struct va_format *vaf);void __drm_puts_seq_file (struct drm_printer *p,const char *str);void __drm_printfn_info (struct drm_printer *p,struct va_format *vaf);void __drm_printfn_debug (struct drm_printer *p,struct va_format *vaf); +void __drm_printfn_err (struct drm_printer *p,struct va_format *vaf); __printf(2, 3)void drm_printf (struct drm_printer *p,const char *f, ...);
@@ -227,6 +228,22 @@ static inline struct drm_printer drm_debug_printer(const char *prefix)return p; } +/** +* drm_err_printer - construct a &drm_printer that outputs to pr_err() +* @prefix: debug output prefix +* +* RETURNS: +* The &drm_printer object +*/ +static inline struct drm_printer drm_err_printer (const char *prefix) +{ +struct drm_printer p = { + .printfn = __drm_printfn_err, + .prefix = prefix + }; +return p; +} +/* * The following categories are defined: *
Signature · 1 line
2.21.0
Re: [PATCH 04/26] drm/print: Add drm_err_printer()
Quoted 18 lines · Lyude Paul
A simple convienence function that returns a
drm_printerwhich prints usingpr_err()Cc Juston Li <[email protected]>Cc Imre Deak <[email protected]>Cc Ville Syrjälä <[email protected]>Cc Harry Wentland <[email protected]>Signed-off-by Lyude Paul <[email protected]>--- drivers/gpu/drm/drm_print.c | 6 ++++++ include/drm/drm_print.h | 17 +++++++++++++++++ 2 files changed, 23 insertions(+)drivers/gpu/drm/drm_print.c +0
-0
drivers/gpu/drm/drm_print.c void __drm_printfn_debug(struct drm_printer *p, struct va_format *vaf) · line 152
Quoted 6 lines
drivers/gpu/drm/drm_print.c +3
-0@@ -147,6 +147,12 @@ void __drm_printfn_debug(struct drm_printer *p, struct va_format *vaf) }EXPORT_SYMBOL (__drm_printfn_debug); +void __drm_printfn_err (struct drm_printer *p,struct va_format *vaf) +{ +pr_err ("%s %pV" , p->prefix, vaf);
DRM printing is a huge bikeshad (or tire fire?). We can't call DRM_ERROR, but for consistency mabye emulate the layout?
Either way: Reviewed-by: Daniel Vetter <[email protected]>
Quoted 44 lines
drivers/gpu/drm/drm_print.c +3
-0+} +EXPORT_SYMBOL (__drm_printfn_err); +/** * drm_puts - print a const string to a &drm_printer stream * @p: the &drm printer include/drm/drm_print.h +17
-0@@ -83,6 +83,7 @@ void __drm_printfn_seq_file(struct drm_printer *p, struct va_format *vaf);void __drm_puts_seq_file (struct drm_printer *p,const char *str);void __drm_printfn_info (struct drm_printer *p,struct va_format *vaf);void __drm_printfn_debug (struct drm_printer *p,struct va_format *vaf); +void __drm_printfn_err (struct drm_printer *p,struct va_format *vaf); __printf(2, 3)void drm_printf (struct drm_printer *p,const char *f, ...);@@ -227,6 +228,22 @@ static inline struct drm_printer drm_debug_printer(const char *prefix)return p; } +/** +* drm_err_printer - construct a &drm_printer that outputs to pr_err() +* @prefix: debug output prefix +* +* RETURNS: +* The &drm_printer object +*/ +static inline struct drm_printer drm_err_printer (const char *prefix) +{ +struct drm_printer p = { + .printfn = __drm_printfn_err, + .prefix = prefix + }; +return p; +} +/* * The following categories are defined: * Signature · 1 line
2.21.0
Signature · 7 lines
Daniel Vetter Software Engineer, Intel Corporation http://blog.ffwll.ch _______________________________________________ dri-devel mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/dri-devel
Lyude Paul
maintainerPATCH 5/26 drm/dp_mst: Add sideband down request tracing + selftests
In reply to Lyude PaulUnfortunately the DP MST helpers do not have much in the way of debugging utilities. So, let's add some!
This adds basic debugging output for down sideband requests that we send from the driver, so that we can actually discern what's happening when sideband requests timeout. Note that with this commit, we'll be dumping out sideband requests under both of the following conditions:
- When the user has enabled DRM_UT_DP output, of course.
- When the user has enabled DRM_UT_KMS or DRM_UT_DP, and a sideband request has failed in some way. This will allow for developers to get a basic idea of what's actually happening with failed modesets on MST, without needing to have DRM_UT_DP explicitly enabled.
Since there wasn't really a good way of testing that any of this worked, I ended up writing simple selftests that lightly test sideband message encoding and decoding as well. Enjoy!
--- drivers/gpu/drm/drm_dp_mst_topology.c | 308 +++++++++++++++++- .../gpu/drm/drm_dp_mst_topology_internal.h | 24 ++ .../gpu/drm/selftests/drm_modeset_selftests.h | 1 + .../drm/selftests/test-drm_dp_mst_helper.c | 212 ++++++++++++ .../drm/selftests/test-drm_modeset_common.h | 1 + include/drm/drm_dp_mst_helper.h | 2 +- 6 files changed, 543 insertions(+), 5 deletions(-) create mode 100644 drivers/gpu/drm/drm_dp_mst_topology_internal.h
drivers/gpu/drm/drm_dp_mst_topology.c +304 -4
@@ -36,6 +36,8 @@#include <drm/drm_print.h> #include <drm/drm_probe_helper.h> +#include "drm_dp_mst_topology_internal.h" +/** * DOC: dp mst helper *
@@ -68,6 +70,8 @@ static int drm_dp_mst_register_i2c_bus(struct drm_dp_aux *aux);static void drm_dp_mst_unregister_i2c_bus (struct drm_dp_aux *aux);static void drm_dp_mst_kick_tx (struct drm_dp_mst_topology_mgr *mgr); +#define DBG_PREFIX "[dp_mst]" +#define DP_STR (x) [DP_ ## x] = #xstatic const char *drm_dp_mst_req_type_str (u8 req_type)
@@ -124,6 +128,43 @@ static const char *drm_dp_mst_nak_reason_str(u8 nak_reason) }#undef DP_STR +#define DP_STR (x) [DRM_DP_SIDEBAND_TX_ ## x] = #x + +static const char *drm_dp_mst_sideband_tx_state_str (int state) +{ +static const char *const sideband_reason_str[] = { +DP_STR (QUEUED ), +DP_STR (START_SEND ), +DP_STR (SENT ), +DP_STR (RX ), +DP_STR (TIMEOUT ), + }; + +if (state >=ARRAY_SIZE (sideband_reason_str) || + !sideband_reason_str[state]) +return "unknown" ; + +return sideband_reason_str[state]; +} + +static int +drm_dp_mst_rad_to_str (const u8 rad[8 ],u8 lct,char *out,size_t len) +{ +int i; +u8 unpacked_rad[16 ]; + +for (i =0 ; i < lct; i++) { +if (i %2 ) + unpacked_rad[i] = rad[i /2 ] >>4 ; +else + unpacked_rad[i] = rad[i /2 ] &BIT_MASK (4 ); + } + +/* TODO: Eventually add something to printk so we can format the rad +* like this: 1.2.3 +*/ +return snprintf (out, len,"%*phC" , lct, unpacked_rad); +}/* sideband msg handling */ static u8 drm_dp_msg_header_crc4 (const uint8_t *data,size_t num_nibbles)
@@ -256,8 +297,9 @@ static bool drm_dp_decode_sideband_msg_hdr(struct drm_dp_sideband_msg_hdr *hdr,return true; } -static void drm_dp_encode_sideband_req (struct drm_dp_sideband_msg_req_body *req, -struct drm_dp_sideband_msg_tx *raw) +void +drm_dp_encode_sideband_req (const struct drm_dp_sideband_msg_req_body *req, +struct drm_dp_sideband_msg_tx *raw) {int idx =0 ;int i;
@@ -362,6 +404,249 @@ static void drm_dp_encode_sideband_req(struct drm_dp_sideband_msg_req_body *req, } raw->cur_len = idx; } +EXPORT_SYMBOL_FOR_TESTS_ONLY (drm_dp_encode_sideband_req); + +/* Decode a sideband request we've encoded, mainly used for debugging */ +int +drm_dp_decode_sideband_req (const struct drm_dp_sideband_msg_tx *raw, +struct drm_dp_sideband_msg_req_body *req) +{ +const u8 *buf = raw->msg; +int i, idx =0 ; + + req->req_type = buf[idx++] &0x7f ; +switch (req->req_type) { +case DP_ENUM_PATH_RESOURCES : +case DP_POWER_DOWN_PHY : +case DP_POWER_UP_PHY : + req->u.port_num.port_number = (buf[idx] >>4 ) &0xf ; +break ; +case DP_ALLOCATE_PAYLOAD : + { +struct drm_dp_allocate_payload *a = + &req->u.allocate_payload; + + a->number_sdp_streams = buf[idx] &0xf ; + a->port_number = (buf[idx] >>4 ) &0xf ; + + a->vcpi = buf[++idx] &0x7f ; + + a->pbn = buf[++idx] <<8 ; + a->pbn |= buf[++idx]; + + idx++; +for (i =0 ; i < a->number_sdp_streams; i++) { + a->sdp_stream_sink[i] = + (buf[idx + (i /2 )] >> ((i %2 ) ?0 :4 )) &0xf ; + } + } +break ; +case DP_QUERY_PAYLOAD : + req->u.query_payload.port_number = (buf[idx] >>4 ) &0xf ; + req->u.query_payload.vcpi = buf[++idx] &0x7f ; +break ; +case DP_REMOTE_DPCD_READ : + { +struct drm_dp_remote_dpcd_read *r = &req->u.dpcd_read; + + r->port_number = (buf[idx] >>4 ) &0xf ; + + r->dpcd_address = (buf[idx] <<16 ) &0xf0000 ; + r->dpcd_address |= (buf[++idx] <<8 ) &0xff00 ; + r->dpcd_address |= buf[++idx] &0xff ; + + r->num_bytes = buf[++idx]; + } +break ; +case DP_REMOTE_DPCD_WRITE : + { +struct drm_dp_remote_dpcd_write *w = + &req->u.dpcd_write; + + w->port_number = (buf[idx] >>4 ) &0xf ; + + w->dpcd_address = (buf[idx] <<16 ) &0xf0000 ; + w->dpcd_address |= (buf[++idx] <<8 ) &0xff00 ; + w->dpcd_address |= buf[++idx] &0xff ; + + w->num_bytes = buf[++idx]; + + w->bytes =kmemdup (&buf[++idx], w->num_bytes, +GFP_KERNEL ); +if (!w->bytes) +return -ENOMEM ; + } +break ; +case DP_REMOTE_I2C_READ : + { +struct drm_dp_remote_i2c_read *r = &req->u.i2c_read; +struct drm_dp_remote_i2c_read_tx *tx; +bool failed = false; + + r->num_transactions = buf[idx] &0x3 ; + r->port_number = (buf[idx] >>4 ) &0xf ; +for (i =0 ; i < r->num_transactions; i++) { + tx = &r->transactions[i]; + + tx->i2c_dev_id = buf[++idx] &0x7f ; + tx->num_bytes = buf[++idx]; + tx->bytes =kmemdup (&buf[++idx], + tx->num_bytes, +GFP_KERNEL ); +if (!tx->bytes) { + failed = true; +break ; + } + idx += tx->num_bytes; + tx->no_stop_bit = (buf[idx] >>5 ) &0x1 ; + tx->i2c_transaction_delay = buf[idx] &0xf ; + } + +if (failed) { +for (i =0 ; i < r->num_transactions; i++) +kfree (tx->bytes); +return -ENOMEM ; + } + + r->read_i2c_device_id = buf[++idx] &0x7f ; + r->num_bytes_read = buf[++idx]; + } +break ; +case DP_REMOTE_I2C_WRITE : + { +struct drm_dp_remote_i2c_write *w = &req->u.i2c_write; + + w->port_number = (buf[idx] >>4 ) &0xf ; + w->write_i2c_device_id = buf[++idx] &0x7f ; + w->num_bytes = buf[++idx]; + w->bytes =kmemdup (&buf[++idx], w->num_bytes, +GFP_KERNEL ); +if (!w->bytes) +return -ENOMEM ; + } +break ; + } + +return 0 ; +} +EXPORT_SYMBOL_FOR_TESTS_ONLY (drm_dp_decode_sideband_req); + +void +drm_dp_dump_sideband_msg_req_body (const struct drm_dp_sideband_msg_req_body *req, +int indent,struct drm_printer *printer) +{ +int i; + +#define P (f, ...) drm_printf_indent(printer, indent, f, ##__VA_ARGS__) +if (req->req_type ==DP_LINK_ADDRESS ) { +/* No contents to print */ +P ("type=%s\n" ,drm_dp_mst_req_type_str (req->req_type)); +return ; + } + +P ("type=%s contents:\n" ,drm_dp_mst_req_type_str (req->req_type)); + indent++; + +switch (req->req_type) { +case DP_ENUM_PATH_RESOURCES : +case DP_POWER_DOWN_PHY : +case DP_POWER_UP_PHY : +P ("port=%d\n" , req->u.port_num.port_number); +break ; +case DP_ALLOCATE_PAYLOAD : +P ("port=%d vcpi=%d pbn=%d sdp_streams=%d %*ph\n" , + req->u.allocate_payload.port_number, + req->u.allocate_payload.vcpi, req->u.allocate_payload.pbn, + req->u.allocate_payload.number_sdp_streams, + req->u.allocate_payload.number_sdp_streams, + req->u.allocate_payload.sdp_stream_sink); +break ; +case DP_QUERY_PAYLOAD : +P ("port=%d vcpi=%d\n" , + req->u.query_payload.port_number, + req->u.query_payload.vcpi); +break ; +case DP_REMOTE_DPCD_READ : +P ("port=%d dpcd_addr=%05x len=%d\n" , + req->u.dpcd_read.port_number, req->u.dpcd_read.dpcd_address, + req->u.dpcd_read.num_bytes); +break ; +case DP_REMOTE_DPCD_WRITE : +P ("port=%d addr=%05x len=%d: %*ph\n" , + req->u.dpcd_write.port_number, + req->u.dpcd_write.dpcd_address, + req->u.dpcd_write.num_bytes, req->u.dpcd_write.num_bytes, + req->u.dpcd_write.bytes); +break ; +case DP_REMOTE_I2C_READ : +P ("port=%d num_tx=%d id=%d size=%d:\n" , + req->u.i2c_read.port_number, + req->u.i2c_read.num_transactions, + req->u.i2c_read.read_i2c_device_id, + req->u.i2c_read.num_bytes_read); + + indent++; +for (i =0 ; i < req->u.i2c_read.num_transactions; i++) { +const struct drm_dp_remote_i2c_read_tx *rtx = + &req->u.i2c_read.transactions[i]; + +P ("%d: id=%03d size=%03d no_stop_bit=%d tx_delay=%03d: %*ph\n" , + i, rtx->i2c_dev_id, rtx->num_bytes, + rtx->no_stop_bit, rtx->i2c_transaction_delay, + rtx->num_bytes, rtx->bytes); + } +break ; +case DP_REMOTE_I2C_WRITE : +P ("port=%d id=%d size=%d: %*ph\n" , + req->u.i2c_write.port_number, + req->u.i2c_write.write_i2c_device_id, + req->u.i2c_write.num_bytes, req->u.i2c_write.num_bytes, + req->u.i2c_write.bytes); +break ; +default : +P ("???\n" ); +break ; + } +#undef P +} +EXPORT_SYMBOL_FOR_TESTS_ONLY (drm_dp_dump_sideband_msg_req_body); + +static inline void +drm_dp_mst_dump_sideband_msg_tx (struct drm_printer *p, +const struct drm_dp_sideband_msg_tx *txmsg) +{ +struct drm_dp_sideband_msg_req_body req; +char buf[64 ]; +int ret; +int i; + +drm_dp_mst_rad_to_str (txmsg->dst->rad, txmsg->dst->lct, buf, +sizeof (buf)); +drm_printf (p,"txmsg cur_offset=%x cur_len=%x seqno=%x state=%s path_msg=%d dst=%s\n" , + txmsg->cur_offset, txmsg->cur_len, txmsg->seqno, +drm_dp_mst_sideband_tx_state_str (txmsg->state), + txmsg->path_msg, buf); + + ret =drm_dp_decode_sideband_req (txmsg, &req); +if (ret) { +drm_printf (p,"<failed to decode sideband req: %d>\n" , ret); +return ; + } +drm_dp_dump_sideband_msg_req_body (&req,1 , p); + +switch (req.req_type) { +case DP_REMOTE_DPCD_WRITE : +kfree (req.u.dpcd_write.bytes); +break ; +case DP_REMOTE_I2C_READ : +for (i =0 ; i < req.u.i2c_read.num_transactions; i++) +kfree (req.u.i2c_read.transactions[i].bytes); +break ; +case DP_REMOTE_I2C_WRITE : +kfree (req.u.i2c_write.bytes); +break ; + } +}static void drm_dp_crc_sideband_chunk_req (u8 *msg,u8 len) {
@@ -893,6 +1178,11 @@ static int drm_dp_mst_wait_tx_reply(struct drm_dp_mst_branch *mstb, } } out: +if (unlikely (ret == -EIO && drm_debug & (DRM_UT_DP |DRM_UT_KMS ))) { +struct drm_printer p =drm_debug_printer (DBG_PREFIX ); + +drm_dp_mst_dump_sideband_msg_tx (&p, txmsg); + }mutex_unlock (&mgr->qlock);return ret;
@@ -1927,8 +2217,11 @@ static int process_single_tx_qlock(struct drm_dp_mst_topology_mgr *mgr, idx += tosend +1 ; ret =drm_dp_send_sideband_msg (mgr, up, chunk, idx); -if (ret) { -DRM_DEBUG_KMS ("sideband msg failed to send\n" ); +if (ret &&unlikely (drm_debug & (DRM_UT_DP |DRM_UT_KMS ))) { +struct drm_printer p =drm_debug_printer (DBG_PREFIX ); + +drm_printf (&p,"sideband msg failed to send\n" ); +drm_dp_mst_dump_sideband_msg_tx (&p, txmsg);return ret; }
@@ -1990,6 +2283,13 @@ static void drm_dp_queue_down_tx(struct drm_dp_mst_topology_mgr *mgr, {mutex_lock (&mgr->qlock);list_add_tail (&txmsg->next, &mgr->tx_msg_downq); + +if (unlikely (drm_debug &DRM_UT_DP )) { +struct drm_printer p =drm_debug_printer (DBG_PREFIX ); + +drm_dp_mst_dump_sideband_msg_tx (&p, txmsg); + } +if (list_is_singular (&mgr->tx_msg_downq))process_single_down_tx_qlock (mgr);mutex_unlock (&mgr->qlock);
drivers/gpu/drm/drm_dp_mst_topology_internal.h new file +24 -0
@@ -0,0 +1,24 @@ +/* SPDX-License-Identifier: GPL-2.0-only +* +* Declarations for DP MST related functions which are only used in selftests +* +* Copyright © 2018 Red Hat +* Authors: +* Lyude Paul <[email protected]> +*/ + +#ifndef _DRM_DP_MST_HELPER_INTERNAL_H_ +#define _DRM_DP_MST_HELPER_INTERNAL_H_ + +#include <drm/drm_dp_mst_helper.h> + +void +drm_dp_encode_sideband_req (const struct drm_dp_sideband_msg_req_body *req, +struct drm_dp_sideband_msg_tx *raw); +int drm_dp_decode_sideband_req (const struct drm_dp_sideband_msg_tx *raw, +struct drm_dp_sideband_msg_req_body *req); +void +drm_dp_dump_sideband_msg_req_body (const struct drm_dp_sideband_msg_req_body *req, +int indent,struct drm_printer *printer); + +#endif /* !_DRM_DP_MST_HELPER_INTERNAL_H_ */
drivers/gpu/drm/selftests/drm_modeset_selftests.h +1 -0
@@ -33,3 +33,4 @@ selftest(damage_iter_damage_one_outside, igt_damage_iter_damage_one_outside) selftest(damage_iter_damage_src_moved, igt_damage_iter_damage_src_moved) selftest(damage_iter_damage_not_visible, igt_damage_iter_damage_not_visible) selftest(dp_mst_calc_pbn_mode, igt_dp_mst_calc_pbn_mode) +selftest (dp_mst_sideband_msg_req_decode, igt_dp_mst_sideband_msg_req_decode)
drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c +212 -0
221 lines not on this page · raw message
drivers/gpu/drm/selftests/test-drm_modeset_common.h +1 -0
@@ -40,5 +40,6 @@ int igt_damage_iter_damage_one_outside(void *ignored);int igt_damage_iter_damage_src_moved (void *ignored);int igt_damage_iter_damage_not_visible (void *ignored);int igt_dp_mst_calc_pbn_mode (void *ignored); +int igt_dp_mst_sideband_msg_req_decode (void *ignored);#endif
include/drm/drm_dp_mst_helper.h +1 -1
@@ -293,7 +293,7 @@ struct drm_dp_remote_dpcd_write {struct drm_dp_remote_i2c_read {u8 num_transactions;u8 port_number; -struct { +struct drm_dp_remote_i2c_read_tx {u8 i2c_dev_id;u8 num_bytes;u8 *bytes;
Signature · 1 line
2.21.0
Re: [PATCH 05/26] drm/dp_mst: Add sideband down request tracing + selftests
Quoted 84 lines · Lyude Paul
Unfortunately the DP MST helpers do not have much in the way of debugging utilities. So, let's add some!
This adds basic debugging output for down sideband requests that we send from the driver, so that we can actually discern what's happening when sideband requests timeout. Note that with this commit, we'll be dumping out sideband requests under both of the following conditions:
- When the user has enabled
DRM_UT_DPoutput, of course.
- When the user has enabledDRM_UT_KMSorDRM_UT_DP, and a sideband request has failed in some way. This will allow for developers to get a basic idea of what's actually happening with failed modesets on MST, without needing to haveDRM_UT_DPexplicitly enabled.Since there wasn't really a good way of testing that any of this worked, I ended up writing simple selftests that lightly test sideband message encoding and decoding as well. Enjoy!
Cc Juston Li <[email protected]>Cc Imre Deak <[email protected]>Cc Ville Syrjälä <[email protected]>Cc Harry Wentland <[email protected]>Signed-off-by Lyude Paul <[email protected]>--- drivers/gpu/drm/drm_dp_mst_topology.c | 308 +++++++++++++++++- .../gpu/drm/drm_dp_mst_topology_internal.h | 24 ++ .../gpu/drm/selftests/drm_modeset_selftests.h | 1 + .../drm/selftests/test-drm_dp_mst_helper.c | 212 ++++++++++++ .../drm/selftests/test-drm_modeset_common.h | 1 + include/drm/drm_dp_mst_helper.h | 2 +- 6 files changed, 543 insertions(+), 5 deletions(-) create mode 100644 drivers/gpu/drm/drm_dp_mst_topology_internal.hdrivers/gpu/drm/drm_dp_mst_topology.c +29
-0@@ -36,6 +36,8 @@#include <drm/drm_print.h> #include <drm/drm_probe_helper.h> +#include "drm_dp_mst_topology_internal.h" +/** * DOC: dp mst helper * @@ -68,6 +70,8 @@ static int drm_dp_mst_register_i2c_bus(struct drm_dp_aux *aux);static void drm_dp_mst_unregister_i2c_bus (struct drm_dp_aux *aux);static void drm_dp_mst_kick_tx (struct drm_dp_mst_topology_mgr *mgr); +#define DBG_PREFIX "[dp_mst]" +#define DP_STR (x) [DP_ ## x] = #xstatic const char *drm_dp_mst_req_type_str (u8 req_type)@@ -124,6 +128,43 @@ static const char *drm_dp_mst_nak_reason_str(u8 nak_reason) }#undef DP_STR +#define DP_STR (x) [DRM_DP_SIDEBAND_TX_ ## x] = #x + +static const char *drm_dp_mst_sideband_tx_state_str (int state) +{ +static const char *const sideband_reason_str[] = { +DP_STR (QUEUED ), +DP_STR (START_SEND ), +DP_STR (SENT ), +DP_STR (RX ), +DP_STR (TIMEOUT ), + }; + +if (state >=ARRAY_SIZE (sideband_reason_str) || + !sideband_reason_str[state]) +return "unknown" ; + +return sideband_reason_str[state]; +} + +static int +drm_dp_mst_rad_to_str (const u8 rad[8 ],u8 lct,char *out,size_t len) +{ +int i; +u8 unpacked_rad[16 ]; +
drivers/gpu/drm/drm_dp_mst_topology.c static const char *drm_dp_mst_nak_reason_str(u8 nak_reason) · line 165
Quoted 10 lines
drivers/gpu/drm/drm_dp_mst_topology.c +10
-0+for (i =0 ; i < lct; i++) { +if (i %2 ) + unpacked_rad[i] = rad[i /2 ] >>4 ; +else + unpacked_rad[i] = rad[i /2 ] &BIT_MASK (4 ); + } + +/* TODO: Eventually add something to printk so we can format the rad +* like this: 1.2.3 +*/
lct *=2;
missing here? And yeah the todo would be sweet, but quite a bit more typing I guess.
Quoted 26 lines
drivers/gpu/drm/drm_dp_mst_topology.c +10
-2+return snprintf (out, len,"%*phC" , lct, unpacked_rad); +}/* sideband msg handling */ static u8 drm_dp_msg_header_crc4 (const uint8_t *data,size_t num_nibbles)@@ -256,8 +297,9 @@ static bool drm_dp_decode_sideband_msg_hdr(struct drm_dp_sideband_msg_hdr *hdr,return true; } -static void drm_dp_encode_sideband_req (struct drm_dp_sideband_msg_req_body *req, -struct drm_dp_sideband_msg_tx *raw) +void +drm_dp_encode_sideband_req (const struct drm_dp_sideband_msg_req_body *req, +struct drm_dp_sideband_msg_tx *raw) {int idx =0 ;int i;@@ -362,6 +404,249 @@ static void drm_dp_encode_sideband_req(struct drm_dp_sideband_msg_req_body *req, } raw->cur_len = idx; } +EXPORT_SYMBOL_FOR_TESTS_ONLY (drm_dp_encode_sideband_req); + +/* Decode a sideband request we've encoded, mainly used for debugging */ +int +drm_dp_decode_sideband_req(const struct drm_dp_sideband_msg_tx *raw,
drivers/gpu/drm/drm_dp_mst_topology.c static void drm_dp_encode_sideband_req(struct drm_dp_sideband_msg_req_body *req, · line 421
Quoted 10 lines
drivers/gpu/drm/drm_dp_mst_topology.c +10
-0+ struct drm_dp_sideband_msg_req_body *req) +{ +const u8 *buf = raw->msg; +int i, idx =0 ; + + req->req_type = buf[idx++] &0x7f ; +switch (req->req_type) { +case DP_ENUM_PATH_RESOURCES : +case DP_POWER_DOWN_PHY : +case DP_POWER_UP_PHY :
OCD warning: the enum isn't ordered the same way here and in drm_dp_encode_sideband_req().
drivers/gpu/drm/drm_dp_mst_topology.c static void drm_dp_encode_sideband_req(struct drm_dp_sideband_msg_req_body *req, · line 432
Quoted 11 lines
drivers/gpu/drm/drm_dp_mst_topology.c +11
-0+ req->u.port_num.port_number = (buf[idx] >>4 ) &0xf ; +break ; +case DP_ALLOCATE_PAYLOAD : + { +struct drm_dp_allocate_payload *a = + &req->u.allocate_payload; + + a->number_sdp_streams = buf[idx] &0xf ; + a->port_number = (buf[idx] >>4 ) &0xf ; + + a->vcpi = buf[++idx] &0x7f ;
Complain here if 0x80 is set?
Quoted 4 lines
drivers/gpu/drm/drm_dp_mst_topology.c +4
-0+ + a->pbn = buf[++idx] <<8 ; + a->pbn |= buf[++idx]; +
drivers/gpu/drm/drm_dp_mst_topology.c static void drm_dp_encode_sideband_req(struct drm_dp_sideband_msg_req_body *req, · line 446
Quoted 10 lines
drivers/gpu/drm/drm_dp_mst_topology.c +10
-0+ idx++; +for (i =0 ; i < a->number_sdp_streams; i++) { + a->sdp_stream_sink[i] = + (buf[idx + (i /2 )] >> ((i %2 ) ?0 :4 )) &0xf ; + } + } +break ; +case DP_QUERY_PAYLOAD : + req->u.query_payload.port_number = (buf[idx] >>4 ) &0xf ; + req->u.query_payload.vcpi = buf[++idx] &0x7f ;
Same here for the highest bit?
Quoted 19 lines
drivers/gpu/drm/drm_dp_mst_topology.c +19
-0+break ; +case DP_REMOTE_DPCD_READ : + { +struct drm_dp_remote_dpcd_read *r = &req->u.dpcd_read; + + r->port_number = (buf[idx] >>4 ) &0xf ; + + r->dpcd_address = (buf[idx] <<16 ) &0xf0000 ; + r->dpcd_address |= (buf[++idx] <<8 ) &0xff00 ; + r->dpcd_address |= buf[++idx] &0xff ; + + r->num_bytes = buf[++idx]; + } +break ; +case DP_REMOTE_DPCD_WRITE : + { +struct drm_dp_remote_dpcd_write *w = + &req->u.dpcd_write; +
drivers/gpu/drm/drm_dp_mst_topology.c static void drm_dp_encode_sideband_req(struct drm_dp_sideband_msg_req_body *req, · line 475
Quoted 10 lines
drivers/gpu/drm/drm_dp_mst_topology.c +10
-0+ w->port_number = (buf[idx] >>4 ) &0xf ; + + w->dpcd_address = (buf[idx] <<16 ) &0xf0000 ; + w->dpcd_address |= (buf[++idx] <<8 ) &0xff00 ; + w->dpcd_address |= buf[++idx] &0xff ; + + w->num_bytes = buf[++idx]; + + w->bytes =kmemdup (&buf[++idx], w->num_bytes, +GFP_KERNEL );
But if we go really strict on validation we'd need to make sure we don't walk past raw->cur_len ... probably not worth it?
Quoted 43 lines
drivers/gpu/drm/drm_dp_mst_topology.c +43
-0+if (!w->bytes) +return -ENOMEM ; + } +break ; +case DP_REMOTE_I2C_READ : + { +struct drm_dp_remote_i2c_read *r = &req->u.i2c_read; +struct drm_dp_remote_i2c_read_tx *tx; +bool failed = false; + + r->num_transactions = buf[idx] &0x3 ; + r->port_number = (buf[idx] >>4 ) &0xf ; +for (i =0 ; i < r->num_transactions; i++) { + tx = &r->transactions[i]; + + tx->i2c_dev_id = buf[++idx] &0x7f ; + tx->num_bytes = buf[++idx]; + tx->bytes =kmemdup (&buf[++idx], + tx->num_bytes, +GFP_KERNEL ); +if (!tx->bytes) { + failed = true; +break ; + } + idx += tx->num_bytes; + tx->no_stop_bit = (buf[idx] >>5 ) &0x1 ; + tx->i2c_transaction_delay = buf[idx] &0xf ; + } + +if (failed) { +for (i =0 ; i < r->num_transactions; i++) +kfree (tx->bytes); +return -ENOMEM ; + } + + r->read_i2c_device_id = buf[++idx] &0x7f ; + r->num_bytes_read = buf[++idx]; + } +break ; +case DP_REMOTE_I2C_WRITE : + { +struct drm_dp_remote_i2c_write *w = &req->u.i2c_write; +
drivers/gpu/drm/drm_dp_mst_topology.c static void drm_dp_encode_sideband_req(struct drm_dp_sideband_msg_req_body *req, · line 528
Quoted 10 lines
drivers/gpu/drm/drm_dp_mst_topology.c +10
-0+ w->port_number = (buf[idx] >>4 ) &0xf ; + w->write_i2c_device_id = buf[++idx] &0x7f ; + w->num_bytes = buf[++idx]; + w->bytes =kmemdup (&buf[++idx], w->num_bytes, +GFP_KERNEL ); +if (!w->bytes) +return -ENOMEM ; + } +break ; + }
For safety: Should we validate cur_len here? Or return it, and let the caller deal with any mismatches.
But I'm kinda leaning towards no safety here, since it's not dealing with any replies we get from the hw/externally, which might be used for attacking us. If we later on have the same thing but for drm_dp_sideband_parse_reply() then a lot more fun.
Quoted 125 lines
drivers/gpu/drm/drm_dp_mst_topology.c +121
-0+ +return 0 ; +} +EXPORT_SYMBOL_FOR_TESTS_ONLY (drm_dp_decode_sideband_req); + +void +drm_dp_dump_sideband_msg_req_body (const struct drm_dp_sideband_msg_req_body *req, +int indent,struct drm_printer *printer) +{ +int i; + +#define P (f, ...) drm_printf_indent(printer, indent, f, ##__VA_ARGS__) +if (req->req_type ==DP_LINK_ADDRESS ) { +/* No contents to print */ +P ("type=%s\n" ,drm_dp_mst_req_type_str (req->req_type)); +return ; + } + +P ("type=%s contents:\n" ,drm_dp_mst_req_type_str (req->req_type)); + indent++; + +switch (req->req_type) { +case DP_ENUM_PATH_RESOURCES : +case DP_POWER_DOWN_PHY : +case DP_POWER_UP_PHY : +P ("port=%d\n" , req->u.port_num.port_number); +break ; +case DP_ALLOCATE_PAYLOAD : +P ("port=%d vcpi=%d pbn=%d sdp_streams=%d %*ph\n" , + req->u.allocate_payload.port_number, + req->u.allocate_payload.vcpi, req->u.allocate_payload.pbn, + req->u.allocate_payload.number_sdp_streams, + req->u.allocate_payload.number_sdp_streams, + req->u.allocate_payload.sdp_stream_sink); +break ; +case DP_QUERY_PAYLOAD : +P ("port=%d vcpi=%d\n" , + req->u.query_payload.port_number, + req->u.query_payload.vcpi); +break ; +case DP_REMOTE_DPCD_READ : +P ("port=%d dpcd_addr=%05x len=%d\n" , + req->u.dpcd_read.port_number, req->u.dpcd_read.dpcd_address, + req->u.dpcd_read.num_bytes); +break ; +case DP_REMOTE_DPCD_WRITE : +P ("port=%d addr=%05x len=%d: %*ph\n" , + req->u.dpcd_write.port_number, + req->u.dpcd_write.dpcd_address, + req->u.dpcd_write.num_bytes, req->u.dpcd_write.num_bytes, + req->u.dpcd_write.bytes); +break ; +case DP_REMOTE_I2C_READ : +P ("port=%d num_tx=%d id=%d size=%d:\n" , + req->u.i2c_read.port_number, + req->u.i2c_read.num_transactions, + req->u.i2c_read.read_i2c_device_id, + req->u.i2c_read.num_bytes_read); + + indent++; +for (i =0 ; i < req->u.i2c_read.num_transactions; i++) { +const struct drm_dp_remote_i2c_read_tx *rtx = + &req->u.i2c_read.transactions[i]; + +P ("%d: id=%03d size=%03d no_stop_bit=%d tx_delay=%03d: %*ph\n" , + i, rtx->i2c_dev_id, rtx->num_bytes, + rtx->no_stop_bit, rtx->i2c_transaction_delay, + rtx->num_bytes, rtx->bytes); + } +break ; +case DP_REMOTE_I2C_WRITE : +P ("port=%d id=%d size=%d: %*ph\n" , + req->u.i2c_write.port_number, + req->u.i2c_write.write_i2c_device_id, + req->u.i2c_write.num_bytes, req->u.i2c_write.num_bytes, + req->u.i2c_write.bytes); +break ; +default : +P ("???\n" ); +break ; + } +#undef P +} +EXPORT_SYMBOL_FOR_TESTS_ONLY (drm_dp_dump_sideband_msg_req_body); + +static inline void +drm_dp_mst_dump_sideband_msg_tx (struct drm_printer *p, +const struct drm_dp_sideband_msg_tx *txmsg) +{ +struct drm_dp_sideband_msg_req_body req; +char buf[64 ]; +int ret; +int i; + +drm_dp_mst_rad_to_str (txmsg->dst->rad, txmsg->dst->lct, buf, +sizeof (buf)); +drm_printf (p,"txmsg cur_offset=%x cur_len=%x seqno=%x state=%s path_msg=%d dst=%s\n" , + txmsg->cur_offset, txmsg->cur_len, txmsg->seqno, +drm_dp_mst_sideband_tx_state_str (txmsg->state), + txmsg->path_msg, buf); + + ret =drm_dp_decode_sideband_req (txmsg, &req); +if (ret) { +drm_printf (p,"<failed to decode sideband req: %d>\n" , ret); +return ; + } +drm_dp_dump_sideband_msg_req_body (&req,1 , p); + +switch (req.req_type) { +case DP_REMOTE_DPCD_WRITE : +kfree (req.u.dpcd_write.bytes); +break ; +case DP_REMOTE_I2C_READ : +for (i =0 ; i < req.u.i2c_read.num_transactions; i++) +kfree (req.u.i2c_read.transactions[i].bytes); +break ; +case DP_REMOTE_I2C_WRITE : +kfree (req.u.i2c_write.bytes); +break ; + } +}static void drm_dp_crc_sideband_chunk_req (u8 *msg,u8 len) {
drivers/gpu/drm/drm_dp_mst_topology.c static int drm_dp_mst_wait_tx_reply(struct drm_dp_mst_branch *mstb, · line 1181
Quoted 4 lines
drivers/gpu/drm/drm_dp_mst_topology.c +1
-0@@ -893,6 +1178,11 @@ static int drm_dp_mst_wait_tx_reply(struct drm_dp_mst_branch *mstb, } } out: +if (unlikely (ret == -EIO && drm_debug & (DRM_UT_DP |DRM_UT_KMS ))) {
Hm I'd only check for DRM_UT_DP here.
Quoted 8 lines
drivers/gpu/drm/drm_dp_mst_topology.c +4
-0+struct drm_printer p =drm_debug_printer (DBG_PREFIX ); + +drm_dp_mst_dump_sideband_msg_tx (&p, txmsg); + }mutex_unlock (&mgr->qlock);return ret;
drivers/gpu/drm/drm_dp_mst_topology.c static int process_single_tx_qlock(struct drm_dp_mst_topology_mgr *mgr, · line 2220
Quoted 6 lines
drivers/gpu/drm/drm_dp_mst_topology.c +1
-2@@ -1927,8 +2217,11 @@ static int process_single_tx_qlock(struct drm_dp_mst_topology_mgr *mgr, idx += tosend +1 ; ret =drm_dp_send_sideband_msg (mgr, up, chunk, idx); -if (ret) { -DRM_DEBUG_KMS ("sideband msg failed to send\n" ); +if (ret &&unlikely (drm_debug & (DRM_UT_DP |DRM_UT_KMS ))) {
Same.
drivers/gpu/drm/drm_dp_mst_topology.c static void drm_dp_queue_down_tx(struct drm_dp_mst_topology_mgr *mgr, · line 2287
Quoted 13 lines
drivers/gpu/drm/drm_dp_mst_topology.c +6
-0+struct drm_printer p =drm_debug_printer (DBG_PREFIX ); + +drm_printf (&p,"sideband msg failed to send\n" ); +drm_dp_mst_dump_sideband_msg_tx (&p, txmsg);return ret; }@@ -1990,6 +2283,13 @@ static void drm_dp_queue_down_tx(struct drm_dp_mst_topology_mgr *mgr, {mutex_lock (&mgr->qlock);list_add_tail (&txmsg->next, &mgr->tx_msg_downq); + +if (unlikely (drm_debug &DRM_UT_DP )) {
Like you do here.
Quoted 72 lines
drivers/gpu/drm/drm_dp_mst_topology.c +5
-0+struct drm_printer p =drm_debug_printer (DBG_PREFIX ); + +drm_dp_mst_dump_sideband_msg_tx (&p, txmsg); + } +if (list_is_singular (&mgr->tx_msg_downq))process_single_down_tx_qlock (mgr);mutex_unlock (&mgr->qlock);drivers/gpu/drm/drm_dp_mst_topology_internal.h new file +24
-0@@ -0,0 +1,24 @@ +/* SPDX-License-Identifier: GPL-2.0-only +* +* Declarations for DP MST related functions which are only used in selftests +* +* Copyright © 2018 Red Hat +* Authors: +* Lyude Paul <[email protected]> +*/ + +#ifndef _DRM_DP_MST_HELPER_INTERNAL_H_ +#define _DRM_DP_MST_HELPER_INTERNAL_H_ + +#include <drm/drm_dp_mst_helper.h> + +void +drm_dp_encode_sideband_req (const struct drm_dp_sideband_msg_req_body *req, +struct drm_dp_sideband_msg_tx *raw); +int drm_dp_decode_sideband_req (const struct drm_dp_sideband_msg_tx *raw, +struct drm_dp_sideband_msg_req_body *req); +void +drm_dp_dump_sideband_msg_req_body (const struct drm_dp_sideband_msg_req_body *req, +int indent,struct drm_printer *printer); + +#endif /* !_DRM_DP_MST_HELPER_INTERNAL_H_ */ drivers/gpu/drm/selftests/drm_modeset_selftests.h +1
-0@@ -33,3 +33,4 @@ selftest(damage_iter_damage_one_outside, igt_damage_iter_damage_one_outside) selftest(damage_iter_damage_src_moved, igt_damage_iter_damage_src_moved) selftest(damage_iter_damage_not_visible, igt_damage_iter_damage_not_visible) selftest(dp_mst_calc_pbn_mode, igt_dp_mst_calc_pbn_mode) +selftest (dp_mst_sideband_msg_req_decode, igt_dp_mst_sideband_msg_req_decode)drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c +10
-0@@ -8,6 +8,7 @@#include <drm/drm_dp_mst_helper.h> #include <drm/drm_print.h> +#include "../drm_dp_mst_topology_internal.h" #include "test-drm_modeset_common.h" int igt_dp_mst_calc_pbn_mode (void *ignored)@@ -34,3 +35,214 @@ int igt_dp_mst_calc_pbn_mode(void *ignored)return 0 ; } + +static bool +sideband_msg_req_equal (const struct drm_dp_sideband_msg_req_body *in, +const struct drm_dp_sideband_msg_req_body *out) +{ +const struct drm_dp_remote_i2c_read_tx *txin, *txout; +int i; + +if (in->req_type != out->req_type)
drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c int igt_dp_mst_calc_pbn_mode(void *ignored) · line 56
Quoted 10 lines
drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c +10
-0+return false; + +switch (in->req_type) { +case DP_ENUM_PATH_RESOURCES : +case DP_POWER_UP_PHY : +case DP_POWER_DOWN_PHY : +case DP_ALLOCATE_PAYLOAD : +case DP_QUERY_PAYLOAD : +case DP_REMOTE_DPCD_READ : +return memcmp (in, out,sizeof (*in)) ==0 ;
Just memcmp the entire thing for everyrone (we assume kzalloc anyway), and then only have the additional checks for the few cases we need it? Would slightly reduce the code.
Quoted 107 lines
drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c +107
-0+ +case DP_REMOTE_I2C_READ : +#define IN in->u.i2c_read +#define OUT out->u.i2c_read +if (IN .num_bytes_read !=OUT .num_bytes_read || +IN .num_transactions !=OUT .num_transactions || +IN .port_number !=OUT .port_number || +IN .read_i2c_device_id !=OUT .read_i2c_device_id) +return false; + +for (i =0 ; i <IN .num_transactions; i++) { + txin = &IN .transactions[i]; + txout = &OUT .transactions[i]; + +if (txin->i2c_dev_id != txout->i2c_dev_id || + txin->no_stop_bit != txout->no_stop_bit || + txin->num_bytes != txout->num_bytes || + txin->i2c_transaction_delay != + txout->i2c_transaction_delay) +return false; + +if (memcmp (txin->bytes, txout->bytes, + txin->num_bytes) !=0 ) +return false; + } +break ; +#undef IN +#undef OUT + +case DP_REMOTE_DPCD_WRITE : +#define IN in->u.dpcd_write +#define OUT out->u.dpcd_write +if (IN .dpcd_address !=OUT .dpcd_address || +IN .num_bytes !=OUT .num_bytes || +IN .port_number !=OUT .port_number) +return false; + +return memcmp (IN .bytes,OUT .bytes,IN .num_bytes) ==0 ; +#undef IN +#undef OUT + +case DP_REMOTE_I2C_WRITE : +#define IN in->u.i2c_write +#define OUT out->u.i2c_write +if (IN .port_number !=OUT .port_number || +IN .write_i2c_device_id !=OUT .write_i2c_device_id || +IN .num_bytes !=OUT .num_bytes) +return false; + +return memcmp (IN .bytes,OUT .bytes,IN .num_bytes) ==0 ; +#undef IN +#undef OUT + } + +return true; +} + +static int +__sideband_msg_req_encode_decode (int line, +struct drm_dp_sideband_msg_req_body *in, +struct drm_dp_sideband_msg_req_body *out) +{ +struct drm_printer p =drm_err_printer (pr_fmt ()); +struct drm_dp_sideband_msg_tx txmsg; +int i, ret; +bool eq; + +drm_dp_encode_sideband_req (in, &txmsg); + ret =drm_dp_decode_sideband_req (&txmsg, out); +if (ret <0 ) { +pr_err (pr_fmt ("Failed to decode sideband request @ line %d: %d\n" ), + line, ret); +return ret; + } + + eq =sideband_msg_req_equal (in, out); +if (!eq) { +pr_err (pr_fmt ("Encode/decode @ line %d failed, expected:\n" ), + line); +drm_dp_dump_sideband_msg_req_body (in,1 , &p); +pr_err (pr_fmt ("Got:\n" )); +drm_dp_dump_sideband_msg_req_body (out,1 , &p); + } + +switch (in->req_type) { +case DP_REMOTE_DPCD_WRITE : +kfree (out->u.dpcd_write.bytes); +break ; +case DP_REMOTE_I2C_READ : +for (i =0 ; i < out->u.i2c_read.num_transactions; i++) +kfree (out->u.i2c_read.transactions[i].bytes); +break ; +case DP_REMOTE_I2C_WRITE : +kfree (out->u.i2c_write.bytes); +break ; + } + +/* Clear everything but the req_type for the input */ +memset (&in->u,0 ,sizeof (in->u)); + +/* Clear the output entirely */ +memset (out,0 ,sizeof (*out)); + +return eq ?0 : -EINVAL ; +} + +int igt_dp_mst_sideband_msg_req_decode (void *unused)
drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c int igt_dp_mst_calc_pbn_mode(void *ignored) · line 173
Quoted 10 lines
drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c +10
-0+{ +struct drm_dp_sideband_msg_req_body in = {0 }, out = {0 }; +u8 data[] = {0xff ,0x0 ,0xdd }; +int i; + +#define DO_TEST () \ + do { \ + if (__sideband_msg_req_encode_decode(__LINE__, &in, &out)) \ + return -EINVAL; \ + } while (0)
I had a tiny wtf moment here until I realized this is a macro ... maybe put the do on the first line and indent everything? Pure bikeshed to help the blind ...
Quoted 65 lines
drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c +65
-0+ + in.req_type =DP_ENUM_PATH_RESOURCES ; + in.u.port_num.port_number =5 ; +DO_TEST (); + + in.req_type =DP_POWER_UP_PHY ; + in.u.port_num.port_number =5 ; +DO_TEST (); + + in.req_type =DP_POWER_DOWN_PHY ; + in.u.port_num.port_number =5 ; +DO_TEST (); + + in.req_type =DP_ALLOCATE_PAYLOAD ; + in.u.allocate_payload.number_sdp_streams =3 ; +for (i =0 ; i < in.u.allocate_payload.number_sdp_streams; i++) + in.u.allocate_payload.sdp_stream_sink[i] = i +1 ; +DO_TEST (); + in.u.allocate_payload.port_number =0xf ; +DO_TEST (); + in.u.allocate_payload.vcpi =0x7f ; +DO_TEST (); + in.u.allocate_payload.pbn =U16_MAX ; +DO_TEST (); + + in.req_type =DP_QUERY_PAYLOAD ; + in.u.query_payload.port_number =0xf ; +DO_TEST (); + in.u.query_payload.vcpi =0x7f ; +DO_TEST (); + + in.req_type =DP_REMOTE_DPCD_READ ; + in.u.dpcd_read.port_number =0xf ; +DO_TEST (); + in.u.dpcd_read.dpcd_address =0xfedcb ; +DO_TEST (); + in.u.dpcd_read.num_bytes =U8_MAX ; +DO_TEST (); + + in.req_type =DP_REMOTE_DPCD_WRITE ; + in.u.dpcd_write.port_number =0xf ; +DO_TEST (); + in.u.dpcd_write.dpcd_address =0xfedcb ; +DO_TEST (); + in.u.dpcd_write.num_bytes =ARRAY_SIZE (data); + in.u.dpcd_write.bytes = data; +DO_TEST (); + + in.req_type =DP_REMOTE_I2C_READ ; + in.u.i2c_read.port_number =0xf ; +DO_TEST (); + in.u.i2c_read.read_i2c_device_id =0x7f ; +DO_TEST (); + in.u.i2c_read.num_transactions =3 ; + in.u.i2c_read.num_bytes_read =ARRAY_SIZE (data) *3 ; +for (i =0 ; i < in.u.i2c_read.num_transactions; i++) { + in.u.i2c_read.transactions[i].bytes = data; + in.u.i2c_read.transactions[i].num_bytes =ARRAY_SIZE (data); + in.u.i2c_read.transactions[i].i2c_dev_id =0x7f & ~i; + in.u.i2c_read.transactions[i].i2c_transaction_delay =0xf & ~i; + } +DO_TEST (); + + in.req_type =DP_REMOTE_I2C_WRITE ; + in.u.i2c_write.port_number =0xf ;
drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c int igt_dp_mst_calc_pbn_mode(void *ignored) · line 248
Quoted 10 lines
drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c +10
-0+DO_TEST (); + in.u.i2c_write.write_i2c_device_id =0x7f ; +DO_TEST (); + in.u.i2c_write.num_bytes =ARRAY_SIZE (data); + in.u.i2c_write.bytes = data; +DO_TEST (); + +#undef DO_TEST +return 0 ; +}
Extremely nice, more unit tests ftw!
With the nits somehow figured out: Reviewed-by: Daniel Vetter <[email protected]>
Quoted 27 lines
drivers/gpu/drm/selftests/test-drm_modeset_common.h +1
-0@@ -40,5 +40,6 @@ int igt_damage_iter_damage_one_outside(void *ignored);int igt_damage_iter_damage_src_moved (void *ignored);int igt_damage_iter_damage_not_visible (void *ignored);int igt_dp_mst_calc_pbn_mode (void *ignored); +int igt_dp_mst_sideband_msg_req_decode (void *ignored);#endif include/drm/drm_dp_mst_helper.h +1
-1@@ -293,7 +293,7 @@ struct drm_dp_remote_dpcd_write {struct drm_dp_remote_i2c_read {u8 num_transactions;u8 port_number; -struct { +struct drm_dp_remote_i2c_read_tx {u8 i2c_dev_id;u8 num_bytes;u8 *bytes;Signature · 1 line
2.21.0
Signature · 7 lines
Daniel Vetter Software Engineer, Intel Corporation http://blog.ffwll.ch _______________________________________________ dri-devel mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/dri-devel
Lyude Paul
maintainerRe: [PATCH 05/26] drm/dp_mst: Add sideband down request tracing + selftests
Quoted 101 lines · Daniel Vetter
Quoted 87 lines · Lyude Paul
Unfortunately the DP MST helpers do not have much in the way of debugging utilities. So, let's add some!
This adds basic debugging output for down sideband requests that we send from the driver, so that we can actually discern what's happening when sideband requests timeout. Note that with this commit, we'll be dumping out sideband requests under both of the following conditions:
- When the user has enabled
DRM_UT_DPoutput, of course.
- When the user has enabledDRM_UT_KMSorDRM_UT_DP, and a sideband request has failed in some way. This will allow for developers to get a basic idea of what's actually happening with failed modesets on MST, without needing to haveDRM_UT_DPexplicitly enabled.Since there wasn't really a good way of testing that any of this worked, I ended up writing simple selftests that lightly test sideband message encoding and decoding as well. Enjoy!
Cc Juston Li <[email protected]>Cc Imre Deak <[email protected]>Cc Ville Syrjälä <[email protected]>Cc Harry Wentland <[email protected]>Signed-off-by Lyude Paul <[email protected]>--- drivers/gpu/drm/drm_dp_mst_topology.c | 308 +++++++++++++++++- .../gpu/drm/drm_dp_mst_topology_internal.h | 24 ++ .../gpu/drm/selftests/drm_modeset_selftests.h | 1 + .../drm/selftests/test-drm_dp_mst_helper.c | 212 ++++++++++++ .../drm/selftests/test-drm_modeset_common.h | 1 + include/drm/drm_dp_mst_helper.h | 2 +- 6 files changed, 543 insertions(+), 5 deletions(-) create mode 100644 drivers/gpu/drm/drm_dp_mst_topology_internal.hdrivers/gpu/drm/drm_dp_mst_topology.c +4
-0@@ -36,6 +36,8 @@#include <drm/drm_print.h> #include <drm/drm_probe_helper.h> +#include "drm_dp_mst_topology_internal.h" +/** * DOC: dp mst helper * @@ -68,6 +70,8 @@ static int drm_dp_mst_register_i2c_bus(struct drm_dp_aux *aux);static void drm_dp_mst_unregister_i2c_bus (struct drm_dp_aux *aux);static void drm_dp_mst_kick_tx (struct drm_dp_mst_topology_mgr *mgr); +#define DBG_PREFIX "[dp_mst]" +#define DP_STR (x) [DP_ ## x] = #xstatic const char *drm_dp_mst_req_type_str(u8
req_type)drivers/gpu/drm/drm_dp_mst_topology.c +25
-0@@ -124,6 +128,43 @@ static const char *drm_dp_mst_nak_reason_str(u8 nak_reason) }#undef DP_STR +#define DP_STR (x) [DRM_DP_SIDEBAND_TX_ ## x] = #x + +static const char *drm_dp_mst_sideband_tx_state_str (int state) +{ +static const char *const sideband_reason_str[] = { +DP_STR (QUEUED ), +DP_STR (START_SEND ), +DP_STR (SENT ), +DP_STR (RX ), +DP_STR (TIMEOUT ), + }; + +if (state >=ARRAY_SIZE (sideband_reason_str) || + !sideband_reason_str[state]) +return "unknown" ; + +return sideband_reason_str[state]; +} + +static int +drm_dp_mst_rad_to_str (const u8 rad[8 ],u8 lct,char *out,size_t len) +{ +int i; +u8 unpacked_rad[16 ]; +drivers/gpu/drm/drm_dp_mst_topology.c static const char *drm_dp_mst_nak_reason_str(u8 · line 166
Quoted 10 lines
drivers/gpu/drm/drm_dp_mst_topology.c +10
-0+for (i =0 ; i < lct; i++) { +if (i %2 ) + unpacked_rad[i] = rad[i /2 ] >>4 ; +else + unpacked_rad[i] = rad[i /2 ] &BIT_MASK (4 ); + } + +/* TODO: Eventually add something to printk so we can format the rad +* like this: 1.2.3 +*/ lct *=2;
Quoted 2 lines
missing here? And yeah the todo would be sweet, but quite a bit more typing I guess.
Nope-although I see why one might assume that. lct == Link Count Total, e.g. how many nodes we have to hop until we reach our destination. Each segment of the RAD is packed as 4 bits, but the LCT is just stored as a plain value.
Quoted 167 lines
Quoted 31 lines
drivers/gpu/drm/drm_dp_mst_topology.c +2
-0+return snprintf (out, len,"%*phC" , lct, unpacked_rad); +}/* sideband msg handling */ static u8
drm_dp_msg_header_crc4(const uint8_t *data, size_t num_nibbles)drivers/gpu/drm/drm_dp_mst_topology.c +3
-2@@ -256,8 +297,9 @@ static bool drm_dp_decode_sideband_msg_hdr(structdrm_dp_sideband_msg_hdr *hdr, return true; } -static void drm_dp_encode_sideband_req (struct drm_dp_sideband_msg_req_body *req, - struct drm_dp_sideband_msg_tx *raw) +void +drm_dp_encode_sideband_req (const struct drm_dp_sideband_msg_req_body *req, +struct drm_dp_sideband_msg_tx *raw){
drivers/gpu/drm/drm_dp_mst_topology.c +5
-0int idx =0 ;int i;@@ -362,6 +404,249 @@ static void drm_dp_encode_sideband_req(struct drm_dp_sideband_msg_req_body *req, } raw->cur_len = idx; } +EXPORT_SYMBOL_FOR_TESTS_ONLY (drm_dp_encode_sideband_req); + +/* Decode a sideband request we've encoded, mainly used for debugging */ +int +drm_dp_decode_sideband_req(const struct drm_dp_sideband_msg_tx *raw,drivers/gpu/drm/drm_dp_mst_topology.c static void drm_dp_encode_sideband_req(struct · line 422
Quoted 10 lines
drivers/gpu/drm/drm_dp_mst_topology.c +10
-0+ struct drm_dp_sideband_msg_req_body *req) +{ +const u8 *buf = raw->msg; +int i, idx =0 ; + + req->req_type = buf[idx++] &0x7f ; +switch (req->req_type) { +case DP_ENUM_PATH_RESOURCES : +case DP_POWER_DOWN_PHY : +case DP_POWER_UP_PHY :OCD warning: the enum isn't ordered the same way here and in
drm_dp_encode_sideband_req().drivers/gpu/drm/drm_dp_mst_topology.c static void drm_dp_encode_sideband_req(struct drm_dp_sideband_msg_req_body *req, · line 432
Quoted 11 lines
drivers/gpu/drm/drm_dp_mst_topology.c +11
-0+ req->u.port_num.port_number = (buf[idx] >>4 ) &0xf ; +break ; +case DP_ALLOCATE_PAYLOAD : + { +struct drm_dp_allocate_payload *a = + &req->u.allocate_payload; + + a->number_sdp_streams = buf[idx] &0xf ; + a->port_number = (buf[idx] >>4 ) &0xf ; + + a->vcpi = buf[++idx] &0x7f ;Complain here if
0x80is set?Quoted 5 lines
drivers/gpu/drm/drm_dp_mst_topology.c +5
-0+ + a->pbn = buf[++idx] <<8 ; + a->pbn |= buf[++idx]; + + idx++;drivers/gpu/drm/drm_dp_mst_topology.c static void drm_dp_encode_sideband_req(struct drm_dp_sideband_msg_req_body *req, · line 446
Quoted 10 lines
drivers/gpu/drm/drm_dp_mst_topology.c +9
-0+for (i =0 ; i < a->number_sdp_streams; i++) { + a->sdp_stream_sink[i] = + (buf[idx + (i /2 )] >> ((i %2 ) ?0 :4 )) &0xf ; + } + } +break ; +case DP_QUERY_PAYLOAD : + req->u.query_payload.port_number = (buf[idx] >>4 ) &0xf ; + req->u.query_payload.vcpi = buf[++idx] &0x7f ;Same here for the highest bit?
Quoted 19 lines
drivers/gpu/drm/drm_dp_mst_topology.c +19
-0+break ; +case DP_REMOTE_DPCD_READ : + { +struct drm_dp_remote_dpcd_read *r = &req->u.dpcd_read; + + r->port_number = (buf[idx] >>4 ) &0xf ; + + r->dpcd_address = (buf[idx] <<16 ) &0xf0000 ; + r->dpcd_address |= (buf[++idx] <<8 ) &0xff00 ; + r->dpcd_address |= buf[++idx] &0xff ; + + r->num_bytes = buf[++idx]; + } +break ; +case DP_REMOTE_DPCD_WRITE : + { +struct drm_dp_remote_dpcd_write *w = + &req->u.dpcd_write; +drivers/gpu/drm/drm_dp_mst_topology.c static void drm_dp_encode_sideband_req(struct drm_dp_sideband_msg_req_body *req, · line 475
Quoted 10 lines
drivers/gpu/drm/drm_dp_mst_topology.c +10
-0+ w->port_number = (buf[idx] >>4 ) &0xf ; + + w->dpcd_address = (buf[idx] <<16 ) &0xf0000 ; + w->dpcd_address |= (buf[++idx] <<8 ) &0xff00 ; + w->dpcd_address |= buf[++idx] &0xff ; + + w->num_bytes = buf[++idx]; + + w->bytes =kmemdup (&buf[++idx], w->num_bytes, +GFP_KERNEL );But if we go really strict on validation we'd need to make sure we don't walk past
raw->cur_len... probably not worth it?Quoted 43 lines
drivers/gpu/drm/drm_dp_mst_topology.c +43
-0+if (!w->bytes) +return -ENOMEM ; + } +break ; +case DP_REMOTE_I2C_READ : + { +struct drm_dp_remote_i2c_read *r = &req->u.i2c_read; +struct drm_dp_remote_i2c_read_tx *tx; +bool failed = false; + + r->num_transactions = buf[idx] &0x3 ; + r->port_number = (buf[idx] >>4 ) &0xf ; +for (i =0 ; i < r->num_transactions; i++) { + tx = &r->transactions[i]; + + tx->i2c_dev_id = buf[++idx] &0x7f ; + tx->num_bytes = buf[++idx]; + tx->bytes =kmemdup (&buf[++idx], + tx->num_bytes, +GFP_KERNEL ); +if (!tx->bytes) { + failed = true; +break ; + } + idx += tx->num_bytes; + tx->no_stop_bit = (buf[idx] >>5 ) &0x1 ; + tx->i2c_transaction_delay = buf[idx] &0xf ; + } + +if (failed) { +for (i =0 ; i < r->num_transactions; i++) +kfree (tx->bytes); +return -ENOMEM ; + } + + r->read_i2c_device_id = buf[++idx] &0x7f ; + r->num_bytes_read = buf[++idx]; + } +break ; +case DP_REMOTE_I2C_WRITE : + { +struct drm_dp_remote_i2c_write *w = &req->u.i2c_write; +drivers/gpu/drm/drm_dp_mst_topology.c static void drm_dp_encode_sideband_req(struct drm_dp_sideband_msg_req_body *req, · line 528
Quoted 10 lines
drivers/gpu/drm/drm_dp_mst_topology.c +10
-0+ w->port_number = (buf[idx] >>4 ) &0xf ; + w->write_i2c_device_id = buf[++idx] &0x7f ; + w->num_bytes = buf[++idx]; + w->bytes =kmemdup (&buf[++idx], w->num_bytes, +GFP_KERNEL ); +if (!w->bytes) +return -ENOMEM ; + } +break ; + }For safety: Should we validate
cur_lenhere? Or return it, and let the caller deal with any mismatches.
Quoted 4 lines
But I'm kinda leaning towards no safety here, since it's not dealing with any replies we get from the hw/externally, which might be used for attacking us. If we later on have the same thing but for
drm_dp_sideband_parse_reply()then a lot more fun.
Yeah-this was my thinking as well. Additionally, I should probably note the next batch of patches I'm going to try getting in afterwards actually adds selftests for our length checking :)
(fun fact: so far, those selftests seem to indicate we're actually doing quite well at verifying length. Mostly.)
Quoted 383 lines
Quoted 129 lines
drivers/gpu/drm/drm_dp_mst_topology.c +65
-0+ +return 0 ; +} +EXPORT_SYMBOL_FOR_TESTS_ONLY (drm_dp_decode_sideband_req); + +void +drm_dp_dump_sideband_msg_req_body (const struct drm_dp_sideband_msg_req_body *req, +int indent,struct drm_printer *printer) +{ +int i; + +#define P (f, ...) drm_printf_indent(printer, indent, f, ##__VA_ARGS__) +if (req->req_type ==DP_LINK_ADDRESS ) { +/* No contents to print */ +P ("type=%s\n" ,drm_dp_mst_req_type_str (req->req_type)); +return ; + } + +P ("type=%s contents:\n" ,drm_dp_mst_req_type_str (req->req_type)); + indent++; + +switch (req->req_type) { +case DP_ENUM_PATH_RESOURCES : +case DP_POWER_DOWN_PHY : +case DP_POWER_UP_PHY : +P ("port=%d\n" , req->u.port_num.port_number); +break ; +case DP_ALLOCATE_PAYLOAD : +P ("port=%d vcpi=%d pbn=%d sdp_streams=%d %*ph\n" , + req->u.allocate_payload.port_number, + req->u.allocate_payload.vcpi, req->u.allocate_payload.pbn, + req->u.allocate_payload.number_sdp_streams, + req->u.allocate_payload.number_sdp_streams, + req->u.allocate_payload.sdp_stream_sink); +break ; +case DP_QUERY_PAYLOAD : +P ("port=%d vcpi=%d\n" , + req->u.query_payload.port_number, + req->u.query_payload.vcpi); +break ; +case DP_REMOTE_DPCD_READ : +P ("port=%d dpcd_addr=%05x len=%d\n" , + req->u.dpcd_read.port_number, req->u.dpcd_read.dpcd_address, + req->u.dpcd_read.num_bytes); +break ; +case DP_REMOTE_DPCD_WRITE : +P ("port=%d addr=%05x len=%d: %*ph\n" , + req->u.dpcd_write.port_number, + req->u.dpcd_write.dpcd_address, + req->u.dpcd_write.num_bytes, req->u.dpcd_write.num_bytes, + req->u.dpcd_write.bytes); +break ; +case DP_REMOTE_I2C_READ : +P ("port=%d num_tx=%d id=%d size=%d:\n" , + req->u.i2c_read.port_number, + req->u.i2c_read.num_transactions, + req->u.i2c_read.read_i2c_device_id, + req->u.i2c_read.num_bytes_read); + + indent++; +for (i =0 ; i < req->u.i2c_read.num_transactions; i++) { +const struct drm_dp_remote_i2c_read_tx *rtx = + &req->u.i2c_read.transactions[i]; + + P("%d: id=%03d size=%03d no_stop_bit=%d tx_delay=%03d:%*ph\n",
drivers/gpu/drm/drm_dp_mst_topology.c +32
-0+ i, rtx->i2c_dev_id, rtx->num_bytes, + rtx->no_stop_bit, rtx->i2c_transaction_delay, + rtx->num_bytes, rtx->bytes); + } +break ; +case DP_REMOTE_I2C_WRITE : +P ("port=%d id=%d size=%d: %*ph\n" , + req->u.i2c_write.port_number, + req->u.i2c_write.write_i2c_device_id, + req->u.i2c_write.num_bytes, req->u.i2c_write.num_bytes, + req->u.i2c_write.bytes); +break ; +default : +P ("???\n" ); +break ; + } +#undef P +} +EXPORT_SYMBOL_FOR_TESTS_ONLY (drm_dp_dump_sideband_msg_req_body); + +static inline void +drm_dp_mst_dump_sideband_msg_tx (struct drm_printer *p, +const struct drm_dp_sideband_msg_tx *txmsg) +{ +struct drm_dp_sideband_msg_req_body req; +char buf[64 ]; +int ret; +int i; + +drm_dp_mst_rad_to_str (txmsg->dst->rad, txmsg->dst->lct, buf, +sizeof (buf)); + drm_printf(p, "txmsg cur_offset=%x cur_len=%x seqno=%x state=%spath_msg=%d dst=%s\n",
drivers/gpu/drm/drm_dp_mst_topology.c +24
-0+ txmsg->cur_offset, txmsg->cur_len, txmsg->seqno, +drm_dp_mst_sideband_tx_state_str (txmsg->state), + txmsg->path_msg, buf); + + ret =drm_dp_decode_sideband_req (txmsg, &req); +if (ret) { +drm_printf (p,"<failed to decode sideband req: %d>\n" , ret); +return ; + } +drm_dp_dump_sideband_msg_req_body (&req,1 , p); + +switch (req.req_type) { +case DP_REMOTE_DPCD_WRITE : +kfree (req.u.dpcd_write.bytes); +break ; +case DP_REMOTE_I2C_READ : +for (i =0 ; i < req.u.i2c_read.num_transactions; i++) +kfree (req.u.i2c_read.transactions[i].bytes); +break ; +case DP_REMOTE_I2C_WRITE : +kfree (req.u.i2c_write.bytes); +break ; + } +}static void drm_dp_crc_sideband_chunk_req (u8 *msg,u8 len) {drivers/gpu/drm/drm_dp_mst_topology.c static int drm_dp_mst_wait_tx_reply(struct · line 1182
Quoted 5 lines
drivers/gpu/drm/drm_dp_mst_topology.c +1
-0@@ -893,6 +1178,11 @@ static int drm_dp_mst_wait_tx_reply(struct drm_dp_mst_branch *mstb, } } out: +if (unlikely (ret == -EIO && drm_debug & (DRM_UT_DP |DRM_UT_KMS ))) {Hm I'd only check for
DRM_UT_DPhere.Quoted 8 lines
drivers/gpu/drm/drm_dp_mst_topology.c +4
-0+struct drm_printer p =drm_debug_printer (DBG_PREFIX ); + +drm_dp_mst_dump_sideband_msg_tx (&p, txmsg); + }mutex_unlock (&mgr->qlock);drivers/gpu/drm/drm_dp_mst_topology.c +0
-0return ret;drivers/gpu/drm/drm_dp_mst_topology.c static int process_single_tx_qlock(struct · line 2221
Quoted 7 lines
drivers/gpu/drm/drm_dp_mst_topology.c +1
-2@@ -1927,8 +2217,11 @@ static int process_single_tx_qlock(struct drm_dp_mst_topology_mgr *mgr, idx += tosend +1 ; ret =drm_dp_send_sideband_msg (mgr, up, chunk, idx); -if (ret) { -DRM_DEBUG_KMS ("sideband msg failed to send\n" ); +if (ret &&unlikely (drm_debug & (DRM_UT_DP |DRM_UT_KMS ))) {Same.
Quoted 8 lines
drivers/gpu/drm/drm_dp_mst_topology.c +4
-0+struct drm_printer p =drm_debug_printer (DBG_PREFIX ); + +drm_printf (&p,"sideband msg failed to send\n" ); +drm_dp_mst_dump_sideband_msg_tx (&p, txmsg);return ret; }drivers/gpu/drm/drm_dp_mst_topology.c static void drm_dp_queue_down_tx(struct · line 2288
Quoted 6 lines
drivers/gpu/drm/drm_dp_mst_topology.c +2
-0@@ -1990,6 +2283,13 @@ static void drm_dp_queue_down_tx(struct drm_dp_mst_topology_mgr *mgr, {mutex_lock (&mgr->qlock);list_add_tail (&txmsg->next, &mgr->tx_msg_downq); + +if (unlikely (drm_debug &DRM_UT_DP )) {Like you do here.
Quoted 81 lines
drivers/gpu/drm/drm_dp_mst_topology.c +5
-0+struct drm_printer p =drm_debug_printer (DBG_PREFIX ); + +drm_dp_mst_dump_sideband_msg_tx (&p, txmsg); + } +if (list_is_singular (&mgr->tx_msg_downq))process_single_down_tx_qlock (mgr);drivers/gpu/drm/drm_dp_mst_topology.c +0
-0mutex_unlock (&mgr->qlock);drivers/gpu/drm/drm_dp_mst_topology_internal.h new file +3
-0@@ -0,0 +1,24 @@ +/* SPDX-License-Identifier: GPL-2.0-only +* +* Declarations for DP MST related functions which are only used in selftests
+ *
+ * Copyright © 2018 Red Hat
+ * Authors:
+ * Lyude Paul <[email protected]>
+ */drivers/gpu/drm/drm_dp_mst_topology_internal.h new file +8
-0+ +#ifndef _DRM_DP_MST_HELPER_INTERNAL_H_ +#define _DRM_DP_MST_HELPER_INTERNAL_H_ + +#include <drm/drm_dp_mst_helper.h> + +void +drm_dp_encode_sideband_req(const struct drm_dp_sideband_msg_req_body*req,
drivers/gpu/drm/drm_dp_mst_topology_internal.h new file +5
-0+ struct drm_dp_sideband_msg_tx *raw); +int drm_dp_decode_sideband_req (const struct drm_dp_sideband_msg_tx *raw, +struct drm_dp_sideband_msg_req_body *req); +void +drm_dp_dump_sideband_msg_req_body(const struct
drm_dp_sideband_msg_req_body*req,drivers/gpu/drm/drm_dp_mst_topology_internal.h new file +3
-0+ int indent, struct drm_printer *printer); + +#endif /* !_DRM_DP_MST_HELPER_INTERNAL_H_ */ drivers/gpu/drm/selftests/drm_modeset_selftests.h +0
-0@@ -33,3 +33,4 @@ selftest(damage_iter_damage_one_outside, igt_damage_iter_damage_one_outside) selftest(damage_iter_damage_src_moved, igt_damage_iter_damage_src_moved) selftest(damage_iter_damage_not_visible,
igt_damage_iter_damage_not_visible)
selftest(dp_mst_calc_pbn_mode, igt_dp_mst_calc_pbn_mode)
+selftest(dp_mst_sideband_msg_req_decode,igt_dp_mst_sideband_msg_req_decode)drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c +10
-0@@ -8,6 +8,7 @@#include <drm/drm_dp_mst_helper.h> #include <drm/drm_print.h> +#include "../drm_dp_mst_topology_internal.h" #include "test-drm_modeset_common.h" int igt_dp_mst_calc_pbn_mode (void *ignored)@@ -34,3 +35,214 @@ int igt_dp_mst_calc_pbn_mode(void *ignored)return 0 ; } + +static bool +sideband_msg_req_equal (const struct drm_dp_sideband_msg_req_body *in, +const struct drm_dp_sideband_msg_req_body *out) +{ +const struct drm_dp_remote_i2c_read_tx *txin, *txout; +int i; + +if (in->req_type != out->req_type)drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c int igt_dp_mst_calc_pbn_mode(void *ignored) · line 56
Quoted 10 lines
drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c +10
-0+return false; + +switch (in->req_type) { +case DP_ENUM_PATH_RESOURCES : +case DP_POWER_UP_PHY : +case DP_POWER_DOWN_PHY : +case DP_ALLOCATE_PAYLOAD : +case DP_QUERY_PAYLOAD : +case DP_REMOTE_DPCD_READ : +return memcmp (in, out,sizeof (*in)) ==0 ;Just memcmp the entire thing for everyrone (we assume kzalloc anyway), and then only have the additional checks for the few cases we need it? Would slightly reduce the code.
Quoted 115 lines
drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c +71
-0+ +case DP_REMOTE_I2C_READ : +#define IN in->u.i2c_read +#define OUT out->u.i2c_read +if (IN .num_bytes_read !=OUT .num_bytes_read || +IN .num_transactions !=OUT .num_transactions || +IN .port_number !=OUT .port_number || +IN .read_i2c_device_id !=OUT .read_i2c_device_id) +return false; + +for (i =0 ; i <IN .num_transactions; i++) { + txin = &IN .transactions[i]; + txout = &OUT .transactions[i]; + +if (txin->i2c_dev_id != txout->i2c_dev_id || + txin->no_stop_bit != txout->no_stop_bit || + txin->num_bytes != txout->num_bytes || + txin->i2c_transaction_delay != + txout->i2c_transaction_delay) +return false; + +if (memcmp (txin->bytes, txout->bytes, + txin->num_bytes) !=0 ) +return false; + } +break ; +#undef IN +#undef OUT + +case DP_REMOTE_DPCD_WRITE : +#define IN in->u.dpcd_write +#define OUT out->u.dpcd_write +if (IN .dpcd_address !=OUT .dpcd_address || +IN .num_bytes !=OUT .num_bytes || +IN .port_number !=OUT .port_number) +return false; + +return memcmp (IN .bytes,OUT .bytes,IN .num_bytes) ==0 ; +#undef IN +#undef OUT + +case DP_REMOTE_I2C_WRITE : +#define IN in->u.i2c_write +#define OUT out->u.i2c_write +if (IN .port_number !=OUT .port_number || +IN .write_i2c_device_id !=OUT .write_i2c_device_id || +IN .num_bytes !=OUT .num_bytes) +return false; + +return memcmp (IN .bytes,OUT .bytes,IN .num_bytes) ==0 ; +#undef IN +#undef OUT + } + +return true; +} + +static int +__sideband_msg_req_encode_decode (int line, +struct drm_dp_sideband_msg_req_body *in, +struct drm_dp_sideband_msg_req_body *out) +{ +struct drm_printer p =drm_err_printer (pr_fmt ()); +struct drm_dp_sideband_msg_tx txmsg; +int i, ret; +bool eq; + +drm_dp_encode_sideband_req (in, &txmsg); + ret =drm_dp_decode_sideband_req (&txmsg, out); +if (ret <0 ) { + pr_err(pr_fmt("Failed to decode sideband request @ line %d:%d\n"),
drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c +42
-0+ line, ret); +return ret; + } + + eq =sideband_msg_req_equal (in, out); +if (!eq) { +pr_err (pr_fmt ("Encode/decode @ line %d failed, expected:\n" ), + line); +drm_dp_dump_sideband_msg_req_body (in,1 , &p); +pr_err (pr_fmt ("Got:\n" )); +drm_dp_dump_sideband_msg_req_body (out,1 , &p); + } + +switch (in->req_type) { +case DP_REMOTE_DPCD_WRITE : +kfree (out->u.dpcd_write.bytes); +break ; +case DP_REMOTE_I2C_READ : +for (i =0 ; i < out->u.i2c_read.num_transactions; i++) +kfree (out->u.i2c_read.transactions[i].bytes); +break ; +case DP_REMOTE_I2C_WRITE : +kfree (out->u.i2c_write.bytes); +break ; + } + +/* Clear everything but the req_type for the input */ +memset (&in->u,0 ,sizeof (in->u)); + +/* Clear the output entirely */ +memset (out,0 ,sizeof (*out)); + +return eq ?0 : -EINVAL ; +} + +int igt_dp_mst_sideband_msg_req_decode (void *unused) +{ +struct drm_dp_sideband_msg_req_body in = {0 }, out = {0 }; +u8 data[] = {0xff ,0x0 ,0xdd }; +int i; + +#define
DO_TEST()\
Quoted 8 lines
drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c int igt_dp_mst_calc_pbn_mode(void *ignored) · line 173
Quoted 4 lines
drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c +4
-0+do { \ +if (__sideband_msg_req_encode_decode (__LINE__, &in, &out)) \ +return -EINVAL ; \ + }while (0 )I had a tiny wtf moment here until I realized this is a macro ... maybe put the do on the first line and indent everything? Pure bikeshed to help the blind ...
Yeah this was kind of silly, in the other round of selftests I've got a cleaner way of doing this - just define DO_TEST() like so:
/* note: I realized, we also don't need to specify an out (just store it on the stack) */
static inline bool sideband_msg_req_encode_decode(in) {
...
}static void actual_test(...) {
...
#define DO_TEST() FAIL_ON(!sideband_msg_req_encode_decode(in))
...
}Will do the same thing for these selftests
Quoted 112 lines
Quoted 67 lines
drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c +60
-0+ + in.req_type =DP_ENUM_PATH_RESOURCES ; + in.u.port_num.port_number =5 ; +DO_TEST (); + + in.req_type =DP_POWER_UP_PHY ; + in.u.port_num.port_number =5 ; +DO_TEST (); + + in.req_type =DP_POWER_DOWN_PHY ; + in.u.port_num.port_number =5 ; +DO_TEST (); + + in.req_type =DP_ALLOCATE_PAYLOAD ; + in.u.allocate_payload.number_sdp_streams =3 ; +for (i =0 ; i < in.u.allocate_payload.number_sdp_streams; i++) + in.u.allocate_payload.sdp_stream_sink[i] = i +1 ; +DO_TEST (); + in.u.allocate_payload.port_number =0xf ; +DO_TEST (); + in.u.allocate_payload.vcpi =0x7f ; +DO_TEST (); + in.u.allocate_payload.pbn =U16_MAX ; +DO_TEST (); + + in.req_type =DP_QUERY_PAYLOAD ; + in.u.query_payload.port_number =0xf ; +DO_TEST (); + in.u.query_payload.vcpi =0x7f ; +DO_TEST (); + + in.req_type =DP_REMOTE_DPCD_READ ; + in.u.dpcd_read.port_number =0xf ; +DO_TEST (); + in.u.dpcd_read.dpcd_address =0xfedcb ; +DO_TEST (); + in.u.dpcd_read.num_bytes =U8_MAX ; +DO_TEST (); + + in.req_type =DP_REMOTE_DPCD_WRITE ; + in.u.dpcd_write.port_number =0xf ; +DO_TEST (); + in.u.dpcd_write.dpcd_address =0xfedcb ; +DO_TEST (); + in.u.dpcd_write.num_bytes =ARRAY_SIZE (data); + in.u.dpcd_write.bytes = data; +DO_TEST (); + + in.req_type =DP_REMOTE_I2C_READ ; + in.u.i2c_read.port_number =0xf ; +DO_TEST (); + in.u.i2c_read.read_i2c_device_id =0x7f ; +DO_TEST (); + in.u.i2c_read.num_transactions =3 ; + in.u.i2c_read.num_bytes_read =ARRAY_SIZE (data) *3 ; +for (i =0 ; i < in.u.i2c_read.num_transactions; i++) { + in.u.i2c_read.transactions[i].bytes = data; + in.u.i2c_read.transactions[i].num_bytes =ARRAY_SIZE (data); + in.u.i2c_read.transactions[i].i2c_dev_id =0x7f & ~i; + in.u.i2c_read.transactions[i].i2c_transaction_delay = 0xf &~i;
drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c +5
-0+ } +DO_TEST (); + + in.req_type =DP_REMOTE_I2C_WRITE ; + in.u.i2c_write.port_number =0xf ;drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c int igt_dp_mst_calc_pbn_mode(void *ignored) · line 248
Quoted 10 lines
drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c +10
-0+DO_TEST (); + in.u.i2c_write.write_i2c_device_id =0x7f ; +DO_TEST (); + in.u.i2c_write.num_bytes =ARRAY_SIZE (data); + in.u.i2c_write.bytes = data; +DO_TEST (); + +#undef DO_TEST +return 0 ; +}Extremely nice, more unit tests ftw!
With the nits somehow figured out: Reviewed-by: Daniel Vetter < [email protected]>
Quoted 29 lines
drivers/gpu/drm/selftests/test-drm_modeset_common.h +1
-0@@ -40,5 +40,6 @@ int igt_damage_iter_damage_one_outside(void *ignored);int igt_damage_iter_damage_src_moved (void *ignored);int igt_damage_iter_damage_not_visible (void *ignored);int igt_dp_mst_calc_pbn_mode (void *ignored); +int igt_dp_mst_sideband_msg_req_decode (void *ignored);#endif include/drm/drm_dp_mst_helper.h +1
-1@@ -293,7 +293,7 @@ struct drm_dp_remote_dpcd_write {struct drm_dp_remote_i2c_read {u8 num_transactions;u8 port_number; -struct { +struct drm_dp_remote_i2c_read_tx {u8 i2c_dev_id;u8 num_bytes;u8 *bytes;Signature · 1 line
2.21.0
Signature · 7 lines
Cheers, Lyude Paul _______________________________________________ dri-devel mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/dri-devel
Lyude Paul
maintainerRe: [PATCH 05/26] drm/dp_mst: Add sideband down request tracing + selftests
In reply to Daniel VetterQuoted 533 lines · Daniel Vetter
Quoted 87 lines · Lyude Paul
Unfortunately the DP MST helpers do not have much in the way of debugging utilities. So, let's add some!
This adds basic debugging output for down sideband requests that we send from the driver, so that we can actually discern what's happening when sideband requests timeout. Note that with this commit, we'll be dumping out sideband requests under both of the following conditions:
- When the user has enabled
DRM_UT_DPoutput, of course.
- When the user has enabledDRM_UT_KMSorDRM_UT_DP, and a sideband request has failed in some way. This will allow for developers to get a basic idea of what's actually happening with failed modesets on MST, without needing to haveDRM_UT_DPexplicitly enabled.Since there wasn't really a good way of testing that any of this worked, I ended up writing simple selftests that lightly test sideband message encoding and decoding as well. Enjoy!
Cc Juston Li <[email protected]>Cc Imre Deak <[email protected]>Cc Ville Syrjälä <[email protected]>Cc Harry Wentland <[email protected]>Signed-off-by Lyude Paul <[email protected]>--- drivers/gpu/drm/drm_dp_mst_topology.c | 308 +++++++++++++++++- .../gpu/drm/drm_dp_mst_topology_internal.h | 24 ++ .../gpu/drm/selftests/drm_modeset_selftests.h | 1 + .../drm/selftests/test-drm_dp_mst_helper.c | 212 ++++++++++++ .../drm/selftests/test-drm_modeset_common.h | 1 + include/drm/drm_dp_mst_helper.h | 2 +- 6 files changed, 543 insertions(+), 5 deletions(-) create mode 100644 drivers/gpu/drm/drm_dp_mst_topology_internal.hdrivers/gpu/drm/drm_dp_mst_topology.c +4
-0@@ -36,6 +36,8 @@#include <drm/drm_print.h> #include <drm/drm_probe_helper.h> +#include "drm_dp_mst_topology_internal.h" +/** * DOC: dp mst helper * @@ -68,6 +70,8 @@ static int drm_dp_mst_register_i2c_bus(struct drm_dp_aux *aux);static void drm_dp_mst_unregister_i2c_bus (struct drm_dp_aux *aux);static void drm_dp_mst_kick_tx (struct drm_dp_mst_topology_mgr *mgr); +#define DBG_PREFIX "[dp_mst]" +#define DP_STR (x) [DP_ ## x] = #xstatic const char *drm_dp_mst_req_type_str(u8
req_type)drivers/gpu/drm/drm_dp_mst_topology.c +25
-0@@ -124,6 +128,43 @@ static const char *drm_dp_mst_nak_reason_str(u8 nak_reason) }#undef DP_STR +#define DP_STR (x) [DRM_DP_SIDEBAND_TX_ ## x] = #x + +static const char *drm_dp_mst_sideband_tx_state_str (int state) +{ +static const char *const sideband_reason_str[] = { +DP_STR (QUEUED ), +DP_STR (START_SEND ), +DP_STR (SENT ), +DP_STR (RX ), +DP_STR (TIMEOUT ), + }; + +if (state >=ARRAY_SIZE (sideband_reason_str) || + !sideband_reason_str[state]) +return "unknown" ; + +return sideband_reason_str[state]; +} + +static int +drm_dp_mst_rad_to_str (const u8 rad[8 ],u8 lct,char *out,size_t len) +{ +int i; +u8 unpacked_rad[16 ]; +drivers/gpu/drm/drm_dp_mst_topology.c static const char *drm_dp_mst_nak_reason_str(u8 · line 166
Quoted 10 lines
drivers/gpu/drm/drm_dp_mst_topology.c +10
-0+for (i =0 ; i < lct; i++) { +if (i %2 ) + unpacked_rad[i] = rad[i /2 ] >>4 ; +else + unpacked_rad[i] = rad[i /2 ] &BIT_MASK (4 ); + } + +/* TODO: Eventually add something to printk so we can format the rad +* like this: 1.2.3 +*/ lct *=2;missing here? And yeah the todo would be sweet, but quite a bit more typing I guess.
Quoted 30 lines
drivers/gpu/drm/drm_dp_mst_topology.c +2
-0+return snprintf (out, len,"%*phC" , lct, unpacked_rad); +}/* sideband msg handling */ static u8
drm_dp_msg_header_crc4(const uint8_t *data, size_t num_nibbles)drivers/gpu/drm/drm_dp_mst_topology.c +3
-2@@ -256,8 +297,9 @@ static bool drm_dp_decode_sideband_msg_hdr(structdrm_dp_sideband_msg_hdr *hdr, return true; } -static void drm_dp_encode_sideband_req (struct drm_dp_sideband_msg_req_body *req, - struct drm_dp_sideband_msg_tx *raw) +void +drm_dp_encode_sideband_req (const struct drm_dp_sideband_msg_req_body *req, +struct drm_dp_sideband_msg_tx *raw){
drivers/gpu/drm/drm_dp_mst_topology.c +5
-0int idx =0 ;int i;@@ -362,6 +404,249 @@ static void drm_dp_encode_sideband_req(struct drm_dp_sideband_msg_req_body *req, } raw->cur_len = idx; } +EXPORT_SYMBOL_FOR_TESTS_ONLY (drm_dp_encode_sideband_req); + +/* Decode a sideband request we've encoded, mainly used for debugging */ +int +drm_dp_decode_sideband_req(const struct drm_dp_sideband_msg_tx *raw,drivers/gpu/drm/drm_dp_mst_topology.c static void drm_dp_encode_sideband_req(struct · line 422
Quoted 10 lines
drivers/gpu/drm/drm_dp_mst_topology.c +10
-0+ struct drm_dp_sideband_msg_req_body *req) +{ +const u8 *buf = raw->msg; +int i, idx =0 ; + + req->req_type = buf[idx++] &0x7f ; +switch (req->req_type) { +case DP_ENUM_PATH_RESOURCES : +case DP_POWER_DOWN_PHY : +case DP_POWER_UP_PHY :OCD warning: the enum isn't ordered the same way here and in
drm_dp_encode_sideband_req().drivers/gpu/drm/drm_dp_mst_topology.c static void drm_dp_encode_sideband_req(struct drm_dp_sideband_msg_req_body *req, · line 432
Quoted 11 lines
drivers/gpu/drm/drm_dp_mst_topology.c +11
-0+ req->u.port_num.port_number = (buf[idx] >>4 ) &0xf ; +break ; +case DP_ALLOCATE_PAYLOAD : + { +struct drm_dp_allocate_payload *a = + &req->u.allocate_payload; + + a->number_sdp_streams = buf[idx] &0xf ; + a->port_number = (buf[idx] >>4 ) &0xf ; + + a->vcpi = buf[++idx] &0x7f ;Complain here if
0x80is set?Quoted 5 lines
drivers/gpu/drm/drm_dp_mst_topology.c +5
-0+ + a->pbn = buf[++idx] <<8 ; + a->pbn |= buf[++idx]; + + idx++;drivers/gpu/drm/drm_dp_mst_topology.c static void drm_dp_encode_sideband_req(struct drm_dp_sideband_msg_req_body *req, · line 446
Quoted 10 lines
drivers/gpu/drm/drm_dp_mst_topology.c +9
-0+for (i =0 ; i < a->number_sdp_streams; i++) { + a->sdp_stream_sink[i] = + (buf[idx + (i /2 )] >> ((i %2 ) ?0 :4 )) &0xf ; + } + } +break ; +case DP_QUERY_PAYLOAD : + req->u.query_payload.port_number = (buf[idx] >>4 ) &0xf ; + req->u.query_payload.vcpi = buf[++idx] &0x7f ;Same here for the highest bit?
Quoted 19 lines
drivers/gpu/drm/drm_dp_mst_topology.c +19
-0+break ; +case DP_REMOTE_DPCD_READ : + { +struct drm_dp_remote_dpcd_read *r = &req->u.dpcd_read; + + r->port_number = (buf[idx] >>4 ) &0xf ; + + r->dpcd_address = (buf[idx] <<16 ) &0xf0000 ; + r->dpcd_address |= (buf[++idx] <<8 ) &0xff00 ; + r->dpcd_address |= buf[++idx] &0xff ; + + r->num_bytes = buf[++idx]; + } +break ; +case DP_REMOTE_DPCD_WRITE : + { +struct drm_dp_remote_dpcd_write *w = + &req->u.dpcd_write; +drivers/gpu/drm/drm_dp_mst_topology.c static void drm_dp_encode_sideband_req(struct drm_dp_sideband_msg_req_body *req, · line 475
Quoted 10 lines
drivers/gpu/drm/drm_dp_mst_topology.c +10
-0+ w->port_number = (buf[idx] >>4 ) &0xf ; + + w->dpcd_address = (buf[idx] <<16 ) &0xf0000 ; + w->dpcd_address |= (buf[++idx] <<8 ) &0xff00 ; + w->dpcd_address |= buf[++idx] &0xff ; + + w->num_bytes = buf[++idx]; + + w->bytes =kmemdup (&buf[++idx], w->num_bytes, +GFP_KERNEL );But if we go really strict on validation we'd need to make sure we don't walk past
raw->cur_len... probably not worth it?Quoted 43 lines
drivers/gpu/drm/drm_dp_mst_topology.c +43
-0+if (!w->bytes) +return -ENOMEM ; + } +break ; +case DP_REMOTE_I2C_READ : + { +struct drm_dp_remote_i2c_read *r = &req->u.i2c_read; +struct drm_dp_remote_i2c_read_tx *tx; +bool failed = false; + + r->num_transactions = buf[idx] &0x3 ; + r->port_number = (buf[idx] >>4 ) &0xf ; +for (i =0 ; i < r->num_transactions; i++) { + tx = &r->transactions[i]; + + tx->i2c_dev_id = buf[++idx] &0x7f ; + tx->num_bytes = buf[++idx]; + tx->bytes =kmemdup (&buf[++idx], + tx->num_bytes, +GFP_KERNEL ); +if (!tx->bytes) { + failed = true; +break ; + } + idx += tx->num_bytes; + tx->no_stop_bit = (buf[idx] >>5 ) &0x1 ; + tx->i2c_transaction_delay = buf[idx] &0xf ; + } + +if (failed) { +for (i =0 ; i < r->num_transactions; i++) +kfree (tx->bytes); +return -ENOMEM ; + } + + r->read_i2c_device_id = buf[++idx] &0x7f ; + r->num_bytes_read = buf[++idx]; + } +break ; +case DP_REMOTE_I2C_WRITE : + { +struct drm_dp_remote_i2c_write *w = &req->u.i2c_write; +drivers/gpu/drm/drm_dp_mst_topology.c static void drm_dp_encode_sideband_req(struct drm_dp_sideband_msg_req_body *req, · line 528
Quoted 10 lines
drivers/gpu/drm/drm_dp_mst_topology.c +10
-0+ w->port_number = (buf[idx] >>4 ) &0xf ; + w->write_i2c_device_id = buf[++idx] &0x7f ; + w->num_bytes = buf[++idx]; + w->bytes =kmemdup (&buf[++idx], w->num_bytes, +GFP_KERNEL ); +if (!w->bytes) +return -ENOMEM ; + } +break ; + }For safety: Should we validate
cur_lenhere? Or return it, and let the caller deal with any mismatches.But I'm kinda leaning towards no safety here, since it's not dealing with any replies we get from the hw/externally, which might be used for attacking us. If we later on have the same thing but for
drm_dp_sideband_parse_reply()then a lot more fun.Quoted 128 lines
drivers/gpu/drm/drm_dp_mst_topology.c +65
-0+ +return 0 ; +} +EXPORT_SYMBOL_FOR_TESTS_ONLY (drm_dp_decode_sideband_req); + +void +drm_dp_dump_sideband_msg_req_body (const struct drm_dp_sideband_msg_req_body *req, +int indent,struct drm_printer *printer) +{ +int i; + +#define P (f, ...) drm_printf_indent(printer, indent, f, ##__VA_ARGS__) +if (req->req_type ==DP_LINK_ADDRESS ) { +/* No contents to print */ +P ("type=%s\n" ,drm_dp_mst_req_type_str (req->req_type)); +return ; + } + +P ("type=%s contents:\n" ,drm_dp_mst_req_type_str (req->req_type)); + indent++; + +switch (req->req_type) { +case DP_ENUM_PATH_RESOURCES : +case DP_POWER_DOWN_PHY : +case DP_POWER_UP_PHY : +P ("port=%d\n" , req->u.port_num.port_number); +break ; +case DP_ALLOCATE_PAYLOAD : +P ("port=%d vcpi=%d pbn=%d sdp_streams=%d %*ph\n" , + req->u.allocate_payload.port_number, + req->u.allocate_payload.vcpi, req->u.allocate_payload.pbn, + req->u.allocate_payload.number_sdp_streams, + req->u.allocate_payload.number_sdp_streams, + req->u.allocate_payload.sdp_stream_sink); +break ; +case DP_QUERY_PAYLOAD : +P ("port=%d vcpi=%d\n" , + req->u.query_payload.port_number, + req->u.query_payload.vcpi); +break ; +case DP_REMOTE_DPCD_READ : +P ("port=%d dpcd_addr=%05x len=%d\n" , + req->u.dpcd_read.port_number, req->u.dpcd_read.dpcd_address, + req->u.dpcd_read.num_bytes); +break ; +case DP_REMOTE_DPCD_WRITE : +P ("port=%d addr=%05x len=%d: %*ph\n" , + req->u.dpcd_write.port_number, + req->u.dpcd_write.dpcd_address, + req->u.dpcd_write.num_bytes, req->u.dpcd_write.num_bytes, + req->u.dpcd_write.bytes); +break ; +case DP_REMOTE_I2C_READ : +P ("port=%d num_tx=%d id=%d size=%d:\n" , + req->u.i2c_read.port_number, + req->u.i2c_read.num_transactions, + req->u.i2c_read.read_i2c_device_id, + req->u.i2c_read.num_bytes_read); + + indent++; +for (i =0 ; i < req->u.i2c_read.num_transactions; i++) { +const struct drm_dp_remote_i2c_read_tx *rtx = + &req->u.i2c_read.transactions[i]; + + P("%d: id=%03d size=%03d no_stop_bit=%d tx_delay=%03d:%*ph\n",
drivers/gpu/drm/drm_dp_mst_topology.c +32
-0+ i, rtx->i2c_dev_id, rtx->num_bytes, + rtx->no_stop_bit, rtx->i2c_transaction_delay, + rtx->num_bytes, rtx->bytes); + } +break ; +case DP_REMOTE_I2C_WRITE : +P ("port=%d id=%d size=%d: %*ph\n" , + req->u.i2c_write.port_number, + req->u.i2c_write.write_i2c_device_id, + req->u.i2c_write.num_bytes, req->u.i2c_write.num_bytes, + req->u.i2c_write.bytes); +break ; +default : +P ("???\n" ); +break ; + } +#undef P +} +EXPORT_SYMBOL_FOR_TESTS_ONLY (drm_dp_dump_sideband_msg_req_body); + +static inline void +drm_dp_mst_dump_sideband_msg_tx (struct drm_printer *p, +const struct drm_dp_sideband_msg_tx *txmsg) +{ +struct drm_dp_sideband_msg_req_body req; +char buf[64 ]; +int ret; +int i; + +drm_dp_mst_rad_to_str (txmsg->dst->rad, txmsg->dst->lct, buf, +sizeof (buf)); + drm_printf(p, "txmsg cur_offset=%x cur_len=%x seqno=%x state=%spath_msg=%d dst=%s\n",
drivers/gpu/drm/drm_dp_mst_topology.c +24
-0+ txmsg->cur_offset, txmsg->cur_len, txmsg->seqno, +drm_dp_mst_sideband_tx_state_str (txmsg->state), + txmsg->path_msg, buf); + + ret =drm_dp_decode_sideband_req (txmsg, &req); +if (ret) { +drm_printf (p,"<failed to decode sideband req: %d>\n" , ret); +return ; + } +drm_dp_dump_sideband_msg_req_body (&req,1 , p); + +switch (req.req_type) { +case DP_REMOTE_DPCD_WRITE : +kfree (req.u.dpcd_write.bytes); +break ; +case DP_REMOTE_I2C_READ : +for (i =0 ; i < req.u.i2c_read.num_transactions; i++) +kfree (req.u.i2c_read.transactions[i].bytes); +break ; +case DP_REMOTE_I2C_WRITE : +kfree (req.u.i2c_write.bytes); +break ; + } +}static void drm_dp_crc_sideband_chunk_req (u8 *msg,u8 len) {drivers/gpu/drm/drm_dp_mst_topology.c static int drm_dp_mst_wait_tx_reply(struct · line 1182
Quoted 5 lines
drivers/gpu/drm/drm_dp_mst_topology.c +1
-0@@ -893,6 +1178,11 @@ static int drm_dp_mst_wait_tx_reply(struct drm_dp_mst_branch *mstb, } } out: +if (unlikely (ret == -EIO && drm_debug & (DRM_UT_DP |DRM_UT_KMS ))) {Hm I'd only check for
DRM_UT_DPhere.Quoted 8 lines
drivers/gpu/drm/drm_dp_mst_topology.c +4
-0+struct drm_printer p =drm_debug_printer (DBG_PREFIX ); + +drm_dp_mst_dump_sideband_msg_tx (&p, txmsg); + }mutex_unlock (&mgr->qlock);drivers/gpu/drm/drm_dp_mst_topology.c +0
-0return ret;drivers/gpu/drm/drm_dp_mst_topology.c static int process_single_tx_qlock(struct · line 2221
Quoted 7 lines
drivers/gpu/drm/drm_dp_mst_topology.c +1
-2@@ -1927,8 +2217,11 @@ static int process_single_tx_qlock(struct drm_dp_mst_topology_mgr *mgr, idx += tosend +1 ; ret =drm_dp_send_sideband_msg (mgr, up, chunk, idx); -if (ret) { -DRM_DEBUG_KMS ("sideband msg failed to send\n" ); +if (ret &&unlikely (drm_debug & (DRM_UT_DP |DRM_UT_KMS ))) {Same.
Quoted 8 lines
drivers/gpu/drm/drm_dp_mst_topology.c +4
-0+struct drm_printer p =drm_debug_printer (DBG_PREFIX ); + +drm_printf (&p,"sideband msg failed to send\n" ); +drm_dp_mst_dump_sideband_msg_tx (&p, txmsg);return ret; }drivers/gpu/drm/drm_dp_mst_topology.c static void drm_dp_queue_down_tx(struct · line 2288
Quoted 6 lines
drivers/gpu/drm/drm_dp_mst_topology.c +2
-0@@ -1990,6 +2283,13 @@ static void drm_dp_queue_down_tx(struct drm_dp_mst_topology_mgr *mgr, {mutex_lock (&mgr->qlock);list_add_tail (&txmsg->next, &mgr->tx_msg_downq); + +if (unlikely (drm_debug &DRM_UT_DP )) {Like you do here.
Quoted 87 lines
drivers/gpu/drm/drm_dp_mst_topology.c +5
-0+struct drm_printer p =drm_debug_printer (DBG_PREFIX ); + +drm_dp_mst_dump_sideband_msg_tx (&p, txmsg); + } +if (list_is_singular (&mgr->tx_msg_downq))process_single_down_tx_qlock (mgr);drivers/gpu/drm/drm_dp_mst_topology.c +0
-0mutex_unlock (&mgr->qlock);drivers/gpu/drm/drm_dp_mst_topology_internal.h new file +3
-0@@ -0,0 +1,24 @@ +/* SPDX-License-Identifier: GPL-2.0-only +* +* Declarations for DP MST related functions which are only used in selftests
+ *
+ * Copyright © 2018 Red Hat
+ * Authors:
+ * Lyude Paul <[email protected]>
+ */drivers/gpu/drm/drm_dp_mst_topology_internal.h new file +8
-0+ +#ifndef _DRM_DP_MST_HELPER_INTERNAL_H_ +#define _DRM_DP_MST_HELPER_INTERNAL_H_ + +#include <drm/drm_dp_mst_helper.h> + +void +drm_dp_encode_sideband_req(const struct drm_dp_sideband_msg_req_body*req,
drivers/gpu/drm/drm_dp_mst_topology_internal.h new file +5
-0+ struct drm_dp_sideband_msg_tx *raw); +int drm_dp_decode_sideband_req (const struct drm_dp_sideband_msg_tx *raw, +struct drm_dp_sideband_msg_req_body *req); +void +drm_dp_dump_sideband_msg_req_body(const struct
drm_dp_sideband_msg_req_body*req,drivers/gpu/drm/drm_dp_mst_topology_internal.h new file +3
-0+ int indent, struct drm_printer *printer); + +#endif /* !_DRM_DP_MST_HELPER_INTERNAL_H_ */ drivers/gpu/drm/selftests/drm_modeset_selftests.h +0
-0@@ -33,3 +33,4 @@ selftest(damage_iter_damage_one_outside, igt_damage_iter_damage_one_outside) selftest(damage_iter_damage_src_moved, igt_damage_iter_damage_src_moved) selftest(damage_iter_damage_not_visible,
igt_damage_iter_damage_not_visible)
selftest(dp_mst_calc_pbn_mode, igt_dp_mst_calc_pbn_mode)
+selftest(dp_mst_sideband_msg_req_decode,igt_dp_mst_sideband_msg_req_decode)drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c +16
-0@@ -8,6 +8,7 @@#include <drm/drm_dp_mst_helper.h> #include <drm/drm_print.h> +#include "../drm_dp_mst_topology_internal.h" #include "test-drm_modeset_common.h" int igt_dp_mst_calc_pbn_mode (void *ignored)@@ -34,3 +35,214 @@ int igt_dp_mst_calc_pbn_mode(void *ignored)return 0 ; } + +static bool +sideband_msg_req_equal (const struct drm_dp_sideband_msg_req_body *in, +const struct drm_dp_sideband_msg_req_body *out) +{ +const struct drm_dp_remote_i2c_read_tx *txin, *txout; +int i; + +if (in->req_type != out->req_type) +return false; + +switch (in->req_type) { +case DP_ENUM_PATH_RESOURCES : +case DP_POWER_UP_PHY : +case DP_POWER_DOWN_PHY :
Quoted 8 lines
drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c int igt_dp_mst_calc_pbn_mode(void *ignored) · line 56
Quoted 4 lines
drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c +4
-0+case DP_ALLOCATE_PAYLOAD : +case DP_QUERY_PAYLOAD : +case DP_REMOTE_DPCD_READ : +return memcmp (in, out,sizeof (*in)) ==0 ;Just memcmp the entire thing for everyrone (we assume kzalloc anyway), and then only have the additional checks for the few cases we need it? Would slightly reduce the code.
Just realized also - this wouldn't work: remember that I2C_READ, DPCD_WRITE, and I2C_WRITE contain pointers which of course will be different even if the contents are the same. Unfortunately I don't think there's really a good solution for this, unless we want to just remove the pointers and store things within the down request struct itself
Quoted 236 lines
Quoted 116 lines
drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c +71
-0+ +case DP_REMOTE_I2C_READ : +#define IN in->u.i2c_read +#define OUT out->u.i2c_read +if (IN .num_bytes_read !=OUT .num_bytes_read || +IN .num_transactions !=OUT .num_transactions || +IN .port_number !=OUT .port_number || +IN .read_i2c_device_id !=OUT .read_i2c_device_id) +return false; + +for (i =0 ; i <IN .num_transactions; i++) { + txin = &IN .transactions[i]; + txout = &OUT .transactions[i]; + +if (txin->i2c_dev_id != txout->i2c_dev_id || + txin->no_stop_bit != txout->no_stop_bit || + txin->num_bytes != txout->num_bytes || + txin->i2c_transaction_delay != + txout->i2c_transaction_delay) +return false; + +if (memcmp (txin->bytes, txout->bytes, + txin->num_bytes) !=0 ) +return false; + } +break ; +#undef IN +#undef OUT + +case DP_REMOTE_DPCD_WRITE : +#define IN in->u.dpcd_write +#define OUT out->u.dpcd_write +if (IN .dpcd_address !=OUT .dpcd_address || +IN .num_bytes !=OUT .num_bytes || +IN .port_number !=OUT .port_number) +return false; + +return memcmp (IN .bytes,OUT .bytes,IN .num_bytes) ==0 ; +#undef IN +#undef OUT + +case DP_REMOTE_I2C_WRITE : +#define IN in->u.i2c_write +#define OUT out->u.i2c_write +if (IN .port_number !=OUT .port_number || +IN .write_i2c_device_id !=OUT .write_i2c_device_id || +IN .num_bytes !=OUT .num_bytes) +return false; + +return memcmp (IN .bytes,OUT .bytes,IN .num_bytes) ==0 ; +#undef IN +#undef OUT + } + +return true; +} + +static int +__sideband_msg_req_encode_decode (int line, +struct drm_dp_sideband_msg_req_body *in, +struct drm_dp_sideband_msg_req_body *out) +{ +struct drm_printer p =drm_err_printer (pr_fmt ()); +struct drm_dp_sideband_msg_tx txmsg; +int i, ret; +bool eq; + +drm_dp_encode_sideband_req (in, &txmsg); + ret =drm_dp_decode_sideband_req (&txmsg, out); +if (ret <0 ) { + pr_err(pr_fmt("Failed to decode sideband request @ line %d:%d\n"),
drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c +42
-0+ line, ret); +return ret; + } + + eq =sideband_msg_req_equal (in, out); +if (!eq) { +pr_err (pr_fmt ("Encode/decode @ line %d failed, expected:\n" ), + line); +drm_dp_dump_sideband_msg_req_body (in,1 , &p); +pr_err (pr_fmt ("Got:\n" )); +drm_dp_dump_sideband_msg_req_body (out,1 , &p); + } + +switch (in->req_type) { +case DP_REMOTE_DPCD_WRITE : +kfree (out->u.dpcd_write.bytes); +break ; +case DP_REMOTE_I2C_READ : +for (i =0 ; i < out->u.i2c_read.num_transactions; i++) +kfree (out->u.i2c_read.transactions[i].bytes); +break ; +case DP_REMOTE_I2C_WRITE : +kfree (out->u.i2c_write.bytes); +break ; + } + +/* Clear everything but the req_type for the input */ +memset (&in->u,0 ,sizeof (in->u)); + +/* Clear the output entirely */ +memset (out,0 ,sizeof (*out)); + +return eq ?0 : -EINVAL ; +} + +int igt_dp_mst_sideband_msg_req_decode (void *unused) +{ +struct drm_dp_sideband_msg_req_body in = {0 }, out = {0 }; +u8 data[] = {0xff ,0x0 ,0xdd }; +int i; + +#define
DO_TEST()\drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c int igt_dp_mst_calc_pbn_mode(void *ignored) · line 173
Quoted 4 lines
drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c +4
-0+do { \ +if (__sideband_msg_req_encode_decode (__LINE__, &in, &out)) \ +return -EINVAL ; \ + }while (0 )I had a tiny wtf moment here until I realized this is a macro ... maybe put the do on the first line and indent everything? Pure bikeshed to help the blind ...
Quoted 66 lines
drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c +60
-0+ + in.req_type =DP_ENUM_PATH_RESOURCES ; + in.u.port_num.port_number =5 ; +DO_TEST (); + + in.req_type =DP_POWER_UP_PHY ; + in.u.port_num.port_number =5 ; +DO_TEST (); + + in.req_type =DP_POWER_DOWN_PHY ; + in.u.port_num.port_number =5 ; +DO_TEST (); + + in.req_type =DP_ALLOCATE_PAYLOAD ; + in.u.allocate_payload.number_sdp_streams =3 ; +for (i =0 ; i < in.u.allocate_payload.number_sdp_streams; i++) + in.u.allocate_payload.sdp_stream_sink[i] = i +1 ; +DO_TEST (); + in.u.allocate_payload.port_number =0xf ; +DO_TEST (); + in.u.allocate_payload.vcpi =0x7f ; +DO_TEST (); + in.u.allocate_payload.pbn =U16_MAX ; +DO_TEST (); + + in.req_type =DP_QUERY_PAYLOAD ; + in.u.query_payload.port_number =0xf ; +DO_TEST (); + in.u.query_payload.vcpi =0x7f ; +DO_TEST (); + + in.req_type =DP_REMOTE_DPCD_READ ; + in.u.dpcd_read.port_number =0xf ; +DO_TEST (); + in.u.dpcd_read.dpcd_address =0xfedcb ; +DO_TEST (); + in.u.dpcd_read.num_bytes =U8_MAX ; +DO_TEST (); + + in.req_type =DP_REMOTE_DPCD_WRITE ; + in.u.dpcd_write.port_number =0xf ; +DO_TEST (); + in.u.dpcd_write.dpcd_address =0xfedcb ; +DO_TEST (); + in.u.dpcd_write.num_bytes =ARRAY_SIZE (data); + in.u.dpcd_write.bytes = data; +DO_TEST (); + + in.req_type =DP_REMOTE_I2C_READ ; + in.u.i2c_read.port_number =0xf ; +DO_TEST (); + in.u.i2c_read.read_i2c_device_id =0x7f ; +DO_TEST (); + in.u.i2c_read.num_transactions =3 ; + in.u.i2c_read.num_bytes_read =ARRAY_SIZE (data) *3 ; +for (i =0 ; i < in.u.i2c_read.num_transactions; i++) { + in.u.i2c_read.transactions[i].bytes = data; + in.u.i2c_read.transactions[i].num_bytes =ARRAY_SIZE (data); + in.u.i2c_read.transactions[i].i2c_dev_id =0x7f & ~i; + in.u.i2c_read.transactions[i].i2c_transaction_delay = 0xf &~i;
drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c +5
-0+ } +DO_TEST (); + + in.req_type =DP_REMOTE_I2C_WRITE ; + in.u.i2c_write.port_number =0xf ;drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c int igt_dp_mst_calc_pbn_mode(void *ignored) · line 248
Quoted 10 lines
drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c +10
-0+DO_TEST (); + in.u.i2c_write.write_i2c_device_id =0x7f ; +DO_TEST (); + in.u.i2c_write.num_bytes =ARRAY_SIZE (data); + in.u.i2c_write.bytes = data; +DO_TEST (); + +#undef DO_TEST +return 0 ; +}Extremely nice, more unit tests ftw!
With the nits somehow figured out: Reviewed-by: Daniel Vetter < [email protected]>
Quoted 29 lines
drivers/gpu/drm/selftests/test-drm_modeset_common.h +1
-0@@ -40,5 +40,6 @@ int igt_damage_iter_damage_one_outside(void *ignored);int igt_damage_iter_damage_src_moved (void *ignored);int igt_damage_iter_damage_not_visible (void *ignored);int igt_dp_mst_calc_pbn_mode (void *ignored); +int igt_dp_mst_sideband_msg_req_decode (void *ignored);#endif include/drm/drm_dp_mst_helper.h +1
-1@@ -293,7 +293,7 @@ struct drm_dp_remote_dpcd_write {struct drm_dp_remote_i2c_read {u8 num_transactions;u8 port_number; -struct { +struct drm_dp_remote_i2c_read_tx {u8 i2c_dev_id;u8 num_bytes;u8 *bytes;Signature · 1 line
2.21.0
Signature · 2 lines
Cheers, Lyude Paul
Re: [PATCH 05/26] drm/dp_mst: Add sideband down request tracing + selftests
Quoted 543 lines · Lyude Paul
Quoted 533 lines · Daniel Vetter
Quoted 87 lines · Lyude Paul
Unfortunately the DP MST helpers do not have much in the way of debugging utilities. So, let's add some!
This adds basic debugging output for down sideband requests that we send from the driver, so that we can actually discern what's happening when sideband requests timeout. Note that with this commit, we'll be dumping out sideband requests under both of the following conditions:
- When the user has enabled
DRM_UT_DPoutput, of course.
- When the user has enabledDRM_UT_KMSorDRM_UT_DP, and a sideband request has failed in some way. This will allow for developers to get a basic idea of what's actually happening with failed modesets on MST, without needing to haveDRM_UT_DPexplicitly enabled.Since there wasn't really a good way of testing that any of this worked, I ended up writing simple selftests that lightly test sideband message encoding and decoding as well. Enjoy!
Cc Juston Li <[email protected]>Cc Imre Deak <[email protected]>Cc Ville Syrjälä <[email protected]>Cc Harry Wentland <[email protected]>Signed-off-by Lyude Paul <[email protected]>--- drivers/gpu/drm/drm_dp_mst_topology.c | 308 +++++++++++++++++- .../gpu/drm/drm_dp_mst_topology_internal.h | 24 ++ .../gpu/drm/selftests/drm_modeset_selftests.h | 1 + .../drm/selftests/test-drm_dp_mst_helper.c | 212 ++++++++++++ .../drm/selftests/test-drm_modeset_common.h | 1 + include/drm/drm_dp_mst_helper.h | 2 +- 6 files changed, 543 insertions(+), 5 deletions(-) create mode 100644 drivers/gpu/drm/drm_dp_mst_topology_internal.hdrivers/gpu/drm/drm_dp_mst_topology.c +4
-0@@ -36,6 +36,8 @@#include <drm/drm_print.h> #include <drm/drm_probe_helper.h> +#include "drm_dp_mst_topology_internal.h" +/** * DOC: dp mst helper * @@ -68,6 +70,8 @@ static int drm_dp_mst_register_i2c_bus(struct drm_dp_aux *aux);static void drm_dp_mst_unregister_i2c_bus (struct drm_dp_aux *aux);static void drm_dp_mst_kick_tx (struct drm_dp_mst_topology_mgr *mgr); +#define DBG_PREFIX "[dp_mst]" +#define DP_STR (x) [DP_ ## x] = #xstatic const char *drm_dp_mst_req_type_str(u8
req_type)drivers/gpu/drm/drm_dp_mst_topology.c +25
-0@@ -124,6 +128,43 @@ static const char *drm_dp_mst_nak_reason_str(u8 nak_reason) }#undef DP_STR +#define DP_STR (x) [DRM_DP_SIDEBAND_TX_ ## x] = #x + +static const char *drm_dp_mst_sideband_tx_state_str (int state) +{ +static const char *const sideband_reason_str[] = { +DP_STR (QUEUED ), +DP_STR (START_SEND ), +DP_STR (SENT ), +DP_STR (RX ), +DP_STR (TIMEOUT ), + }; + +if (state >=ARRAY_SIZE (sideband_reason_str) || + !sideband_reason_str[state]) +return "unknown" ; + +return sideband_reason_str[state]; +} + +static int +drm_dp_mst_rad_to_str (const u8 rad[8 ],u8 lct,char *out,size_t len) +{ +int i; +u8 unpacked_rad[16 ]; +drivers/gpu/drm/drm_dp_mst_topology.c static const char *drm_dp_mst_nak_reason_str(u8 · line 166
Quoted 10 lines
drivers/gpu/drm/drm_dp_mst_topology.c +10
-0+for (i =0 ; i < lct; i++) { +if (i %2 ) + unpacked_rad[i] = rad[i /2 ] >>4 ; +else + unpacked_rad[i] = rad[i /2 ] &BIT_MASK (4 ); + } + +/* TODO: Eventually add something to printk so we can format the rad +* like this: 1.2.3 +*/ lct *=2;missing here? And yeah the todo would be sweet, but quite a bit more typing I guess.
Quoted 30 lines
drivers/gpu/drm/drm_dp_mst_topology.c +2
-0+return snprintf (out, len,"%*phC" , lct, unpacked_rad); +}/* sideband msg handling */ static u8
drm_dp_msg_header_crc4(const uint8_t *data, size_t num_nibbles)drivers/gpu/drm/drm_dp_mst_topology.c +3
-2@@ -256,8 +297,9 @@ static bool drm_dp_decode_sideband_msg_hdr(structdrm_dp_sideband_msg_hdr *hdr, return true; } -static void drm_dp_encode_sideband_req (struct drm_dp_sideband_msg_req_body *req, - struct drm_dp_sideband_msg_tx *raw) +void +drm_dp_encode_sideband_req (const struct drm_dp_sideband_msg_req_body *req, +struct drm_dp_sideband_msg_tx *raw){
drivers/gpu/drm/drm_dp_mst_topology.c +5
-0int idx =0 ;int i;@@ -362,6 +404,249 @@ static void drm_dp_encode_sideband_req(struct drm_dp_sideband_msg_req_body *req, } raw->cur_len = idx; } +EXPORT_SYMBOL_FOR_TESTS_ONLY (drm_dp_encode_sideband_req); + +/* Decode a sideband request we've encoded, mainly used for debugging */ +int +drm_dp_decode_sideband_req(const struct drm_dp_sideband_msg_tx *raw,drivers/gpu/drm/drm_dp_mst_topology.c static void drm_dp_encode_sideband_req(struct · line 422
Quoted 10 lines
drivers/gpu/drm/drm_dp_mst_topology.c +10
-0+ struct drm_dp_sideband_msg_req_body *req) +{ +const u8 *buf = raw->msg; +int i, idx =0 ; + + req->req_type = buf[idx++] &0x7f ; +switch (req->req_type) { +case DP_ENUM_PATH_RESOURCES : +case DP_POWER_DOWN_PHY : +case DP_POWER_UP_PHY :OCD warning: the enum isn't ordered the same way here and in
drm_dp_encode_sideband_req().drivers/gpu/drm/drm_dp_mst_topology.c static void drm_dp_encode_sideband_req(struct drm_dp_sideband_msg_req_body *req, · line 432
Quoted 11 lines
drivers/gpu/drm/drm_dp_mst_topology.c +11
-0+ req->u.port_num.port_number = (buf[idx] >>4 ) &0xf ; +break ; +case DP_ALLOCATE_PAYLOAD : + { +struct drm_dp_allocate_payload *a = + &req->u.allocate_payload; + + a->number_sdp_streams = buf[idx] &0xf ; + a->port_number = (buf[idx] >>4 ) &0xf ; + + a->vcpi = buf[++idx] &0x7f ;Complain here if
0x80is set?Quoted 5 lines
drivers/gpu/drm/drm_dp_mst_topology.c +5
-0+ + a->pbn = buf[++idx] <<8 ; + a->pbn |= buf[++idx]; + + idx++;drivers/gpu/drm/drm_dp_mst_topology.c static void drm_dp_encode_sideband_req(struct drm_dp_sideband_msg_req_body *req, · line 446
Quoted 10 lines
drivers/gpu/drm/drm_dp_mst_topology.c +9
-0+for (i =0 ; i < a->number_sdp_streams; i++) { + a->sdp_stream_sink[i] = + (buf[idx + (i /2 )] >> ((i %2 ) ?0 :4 )) &0xf ; + } + } +break ; +case DP_QUERY_PAYLOAD : + req->u.query_payload.port_number = (buf[idx] >>4 ) &0xf ; + req->u.query_payload.vcpi = buf[++idx] &0x7f ;Same here for the highest bit?
Quoted 19 lines
drivers/gpu/drm/drm_dp_mst_topology.c +19
-0+break ; +case DP_REMOTE_DPCD_READ : + { +struct drm_dp_remote_dpcd_read *r = &req->u.dpcd_read; + + r->port_number = (buf[idx] >>4 ) &0xf ; + + r->dpcd_address = (buf[idx] <<16 ) &0xf0000 ; + r->dpcd_address |= (buf[++idx] <<8 ) &0xff00 ; + r->dpcd_address |= buf[++idx] &0xff ; + + r->num_bytes = buf[++idx]; + } +break ; +case DP_REMOTE_DPCD_WRITE : + { +struct drm_dp_remote_dpcd_write *w = + &req->u.dpcd_write; +drivers/gpu/drm/drm_dp_mst_topology.c static void drm_dp_encode_sideband_req(struct drm_dp_sideband_msg_req_body *req, · line 475
Quoted 10 lines
drivers/gpu/drm/drm_dp_mst_topology.c +10
-0+ w->port_number = (buf[idx] >>4 ) &0xf ; + + w->dpcd_address = (buf[idx] <<16 ) &0xf0000 ; + w->dpcd_address |= (buf[++idx] <<8 ) &0xff00 ; + w->dpcd_address |= buf[++idx] &0xff ; + + w->num_bytes = buf[++idx]; + + w->bytes =kmemdup (&buf[++idx], w->num_bytes, +GFP_KERNEL );But if we go really strict on validation we'd need to make sure we don't walk past
raw->cur_len... probably not worth it?Quoted 43 lines
drivers/gpu/drm/drm_dp_mst_topology.c +43
-0+if (!w->bytes) +return -ENOMEM ; + } +break ; +case DP_REMOTE_I2C_READ : + { +struct drm_dp_remote_i2c_read *r = &req->u.i2c_read; +struct drm_dp_remote_i2c_read_tx *tx; +bool failed = false; + + r->num_transactions = buf[idx] &0x3 ; + r->port_number = (buf[idx] >>4 ) &0xf ; +for (i =0 ; i < r->num_transactions; i++) { + tx = &r->transactions[i]; + + tx->i2c_dev_id = buf[++idx] &0x7f ; + tx->num_bytes = buf[++idx]; + tx->bytes =kmemdup (&buf[++idx], + tx->num_bytes, +GFP_KERNEL ); +if (!tx->bytes) { + failed = true; +break ; + } + idx += tx->num_bytes; + tx->no_stop_bit = (buf[idx] >>5 ) &0x1 ; + tx->i2c_transaction_delay = buf[idx] &0xf ; + } + +if (failed) { +for (i =0 ; i < r->num_transactions; i++) +kfree (tx->bytes); +return -ENOMEM ; + } + + r->read_i2c_device_id = buf[++idx] &0x7f ; + r->num_bytes_read = buf[++idx]; + } +break ; +case DP_REMOTE_I2C_WRITE : + { +struct drm_dp_remote_i2c_write *w = &req->u.i2c_write; +drivers/gpu/drm/drm_dp_mst_topology.c static void drm_dp_encode_sideband_req(struct drm_dp_sideband_msg_req_body *req, · line 528
Quoted 10 lines
drivers/gpu/drm/drm_dp_mst_topology.c +10
-0+ w->port_number = (buf[idx] >>4 ) &0xf ; + w->write_i2c_device_id = buf[++idx] &0x7f ; + w->num_bytes = buf[++idx]; + w->bytes =kmemdup (&buf[++idx], w->num_bytes, +GFP_KERNEL ); +if (!w->bytes) +return -ENOMEM ; + } +break ; + }For safety: Should we validate
cur_lenhere? Or return it, and let the caller deal with any mismatches.But I'm kinda leaning towards no safety here, since it's not dealing with any replies we get from the hw/externally, which might be used for attacking us. If we later on have the same thing but for
drm_dp_sideband_parse_reply()then a lot more fun.Quoted 128 lines
drivers/gpu/drm/drm_dp_mst_topology.c +65
-0+ +return 0 ; +} +EXPORT_SYMBOL_FOR_TESTS_ONLY (drm_dp_decode_sideband_req); + +void +drm_dp_dump_sideband_msg_req_body (const struct drm_dp_sideband_msg_req_body *req, +int indent,struct drm_printer *printer) +{ +int i; + +#define P (f, ...) drm_printf_indent(printer, indent, f, ##__VA_ARGS__) +if (req->req_type ==DP_LINK_ADDRESS ) { +/* No contents to print */ +P ("type=%s\n" ,drm_dp_mst_req_type_str (req->req_type)); +return ; + } + +P ("type=%s contents:\n" ,drm_dp_mst_req_type_str (req->req_type)); + indent++; + +switch (req->req_type) { +case DP_ENUM_PATH_RESOURCES : +case DP_POWER_DOWN_PHY : +case DP_POWER_UP_PHY : +P ("port=%d\n" , req->u.port_num.port_number); +break ; +case DP_ALLOCATE_PAYLOAD : +P ("port=%d vcpi=%d pbn=%d sdp_streams=%d %*ph\n" , + req->u.allocate_payload.port_number, + req->u.allocate_payload.vcpi, req->u.allocate_payload.pbn, + req->u.allocate_payload.number_sdp_streams, + req->u.allocate_payload.number_sdp_streams, + req->u.allocate_payload.sdp_stream_sink); +break ; +case DP_QUERY_PAYLOAD : +P ("port=%d vcpi=%d\n" , + req->u.query_payload.port_number, + req->u.query_payload.vcpi); +break ; +case DP_REMOTE_DPCD_READ : +P ("port=%d dpcd_addr=%05x len=%d\n" , + req->u.dpcd_read.port_number, req->u.dpcd_read.dpcd_address, + req->u.dpcd_read.num_bytes); +break ; +case DP_REMOTE_DPCD_WRITE : +P ("port=%d addr=%05x len=%d: %*ph\n" , + req->u.dpcd_write.port_number, + req->u.dpcd_write.dpcd_address, + req->u.dpcd_write.num_bytes, req->u.dpcd_write.num_bytes, + req->u.dpcd_write.bytes); +break ; +case DP_REMOTE_I2C_READ : +P ("port=%d num_tx=%d id=%d size=%d:\n" , + req->u.i2c_read.port_number, + req->u.i2c_read.num_transactions, + req->u.i2c_read.read_i2c_device_id, + req->u.i2c_read.num_bytes_read); + + indent++; +for (i =0 ; i < req->u.i2c_read.num_transactions; i++) { +const struct drm_dp_remote_i2c_read_tx *rtx = + &req->u.i2c_read.transactions[i]; + + P("%d: id=%03d size=%03d no_stop_bit=%d tx_delay=%03d:%*ph\n",
drivers/gpu/drm/drm_dp_mst_topology.c +32
-0+ i, rtx->i2c_dev_id, rtx->num_bytes, + rtx->no_stop_bit, rtx->i2c_transaction_delay, + rtx->num_bytes, rtx->bytes); + } +break ; +case DP_REMOTE_I2C_WRITE : +P ("port=%d id=%d size=%d: %*ph\n" , + req->u.i2c_write.port_number, + req->u.i2c_write.write_i2c_device_id, + req->u.i2c_write.num_bytes, req->u.i2c_write.num_bytes, + req->u.i2c_write.bytes); +break ; +default : +P ("???\n" ); +break ; + } +#undef P +} +EXPORT_SYMBOL_FOR_TESTS_ONLY (drm_dp_dump_sideband_msg_req_body); + +static inline void +drm_dp_mst_dump_sideband_msg_tx (struct drm_printer *p, +const struct drm_dp_sideband_msg_tx *txmsg) +{ +struct drm_dp_sideband_msg_req_body req; +char buf[64 ]; +int ret; +int i; + +drm_dp_mst_rad_to_str (txmsg->dst->rad, txmsg->dst->lct, buf, +sizeof (buf)); + drm_printf(p, "txmsg cur_offset=%x cur_len=%x seqno=%x state=%spath_msg=%d dst=%s\n",
drivers/gpu/drm/drm_dp_mst_topology.c +24
-0+ txmsg->cur_offset, txmsg->cur_len, txmsg->seqno, +drm_dp_mst_sideband_tx_state_str (txmsg->state), + txmsg->path_msg, buf); + + ret =drm_dp_decode_sideband_req (txmsg, &req); +if (ret) { +drm_printf (p,"<failed to decode sideband req: %d>\n" , ret); +return ; + } +drm_dp_dump_sideband_msg_req_body (&req,1 , p); + +switch (req.req_type) { +case DP_REMOTE_DPCD_WRITE : +kfree (req.u.dpcd_write.bytes); +break ; +case DP_REMOTE_I2C_READ : +for (i =0 ; i < req.u.i2c_read.num_transactions; i++) +kfree (req.u.i2c_read.transactions[i].bytes); +break ; +case DP_REMOTE_I2C_WRITE : +kfree (req.u.i2c_write.bytes); +break ; + } +}static void drm_dp_crc_sideband_chunk_req (u8 *msg,u8 len) {drivers/gpu/drm/drm_dp_mst_topology.c static int drm_dp_mst_wait_tx_reply(struct · line 1182
Quoted 5 lines
drivers/gpu/drm/drm_dp_mst_topology.c +1
-0@@ -893,6 +1178,11 @@ static int drm_dp_mst_wait_tx_reply(struct drm_dp_mst_branch *mstb, } } out: +if (unlikely (ret == -EIO && drm_debug & (DRM_UT_DP |DRM_UT_KMS ))) {Hm I'd only check for
DRM_UT_DPhere.Quoted 8 lines
drivers/gpu/drm/drm_dp_mst_topology.c +4
-0+struct drm_printer p =drm_debug_printer (DBG_PREFIX ); + +drm_dp_mst_dump_sideband_msg_tx (&p, txmsg); + }mutex_unlock (&mgr->qlock);drivers/gpu/drm/drm_dp_mst_topology.c +0
-0return ret;drivers/gpu/drm/drm_dp_mst_topology.c static int process_single_tx_qlock(struct · line 2221
Quoted 7 lines
drivers/gpu/drm/drm_dp_mst_topology.c +1
-2@@ -1927,8 +2217,11 @@ static int process_single_tx_qlock(struct drm_dp_mst_topology_mgr *mgr, idx += tosend +1 ; ret =drm_dp_send_sideband_msg (mgr, up, chunk, idx); -if (ret) { -DRM_DEBUG_KMS ("sideband msg failed to send\n" ); +if (ret &&unlikely (drm_debug & (DRM_UT_DP |DRM_UT_KMS ))) {Same.
Quoted 8 lines
drivers/gpu/drm/drm_dp_mst_topology.c +4
-0+struct drm_printer p =drm_debug_printer (DBG_PREFIX ); + +drm_printf (&p,"sideband msg failed to send\n" ); +drm_dp_mst_dump_sideband_msg_tx (&p, txmsg);return ret; }drivers/gpu/drm/drm_dp_mst_topology.c static void drm_dp_queue_down_tx(struct · line 2288
Quoted 6 lines
drivers/gpu/drm/drm_dp_mst_topology.c +2
-0@@ -1990,6 +2283,13 @@ static void drm_dp_queue_down_tx(struct drm_dp_mst_topology_mgr *mgr, {mutex_lock (&mgr->qlock);list_add_tail (&txmsg->next, &mgr->tx_msg_downq); + +if (unlikely (drm_debug &DRM_UT_DP )) {Like you do here.
Quoted 87 lines
drivers/gpu/drm/drm_dp_mst_topology.c +5
-0+struct drm_printer p =drm_debug_printer (DBG_PREFIX ); + +drm_dp_mst_dump_sideband_msg_tx (&p, txmsg); + } +if (list_is_singular (&mgr->tx_msg_downq))process_single_down_tx_qlock (mgr);drivers/gpu/drm/drm_dp_mst_topology.c +0
-0mutex_unlock (&mgr->qlock);drivers/gpu/drm/drm_dp_mst_topology_internal.h new file +3
-0@@ -0,0 +1,24 @@ +/* SPDX-License-Identifier: GPL-2.0-only +* +* Declarations for DP MST related functions which are only used in selftests
+ *
+ * Copyright © 2018 Red Hat
+ * Authors:
+ * Lyude Paul <[email protected]>
+ */drivers/gpu/drm/drm_dp_mst_topology_internal.h new file +8
-0+ +#ifndef _DRM_DP_MST_HELPER_INTERNAL_H_ +#define _DRM_DP_MST_HELPER_INTERNAL_H_ + +#include <drm/drm_dp_mst_helper.h> + +void +drm_dp_encode_sideband_req(const struct drm_dp_sideband_msg_req_body*req,
drivers/gpu/drm/drm_dp_mst_topology_internal.h new file +5
-0+ struct drm_dp_sideband_msg_tx *raw); +int drm_dp_decode_sideband_req (const struct drm_dp_sideband_msg_tx *raw, +struct drm_dp_sideband_msg_req_body *req); +void +drm_dp_dump_sideband_msg_req_body(const struct
drm_dp_sideband_msg_req_body*req,drivers/gpu/drm/drm_dp_mst_topology_internal.h new file +3
-0+ int indent, struct drm_printer *printer); + +#endif /* !_DRM_DP_MST_HELPER_INTERNAL_H_ */ drivers/gpu/drm/selftests/drm_modeset_selftests.h +0
-0@@ -33,3 +33,4 @@ selftest(damage_iter_damage_one_outside, igt_damage_iter_damage_one_outside) selftest(damage_iter_damage_src_moved, igt_damage_iter_damage_src_moved) selftest(damage_iter_damage_not_visible,
igt_damage_iter_damage_not_visible)
selftest(dp_mst_calc_pbn_mode, igt_dp_mst_calc_pbn_mode)
+selftest(dp_mst_sideband_msg_req_decode,igt_dp_mst_sideband_msg_req_decode)drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c +16
-0@@ -8,6 +8,7 @@#include <drm/drm_dp_mst_helper.h> #include <drm/drm_print.h> +#include "../drm_dp_mst_topology_internal.h" #include "test-drm_modeset_common.h" int igt_dp_mst_calc_pbn_mode (void *ignored)@@ -34,3 +35,214 @@ int igt_dp_mst_calc_pbn_mode(void *ignored)return 0 ; } + +static bool +sideband_msg_req_equal (const struct drm_dp_sideband_msg_req_body *in, +const struct drm_dp_sideband_msg_req_body *out) +{ +const struct drm_dp_remote_i2c_read_tx *txin, *txout; +int i; + +if (in->req_type != out->req_type) +return false; + +switch (in->req_type) { +case DP_ENUM_PATH_RESOURCES : +case DP_POWER_UP_PHY : +case DP_POWER_DOWN_PHY :Quoted 8 lines
drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c int igt_dp_mst_calc_pbn_mode(void *ignored) · line 56
Quoted 4 lines
drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c +4
-0+case DP_ALLOCATE_PAYLOAD : +case DP_QUERY_PAYLOAD : +case DP_REMOTE_DPCD_READ : +return memcmp (in, out,sizeof (*in)) ==0 ;Just memcmp the entire thing for everyrone (we assume kzalloc anyway), and then only have the additional checks for the few cases we need it? Would slightly reduce the code.
Quoted 5 lines
Just realized also - this wouldn't work: remember that
I2C_READ,DPCD_WRITE, andI2C_WRITEcontain pointers which of course will be different even if the contents are the same. Unfortunately I don't think there's really a good solution for this, unless we want to just remove the pointers and store things within the down request struct itself
Uh ... disappoint :-/
Then at least put that into a comment, and maybe make this the default:
handler?
Quoted 241 lines
Quoted 237 lines
Quoted 117 lines
drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c +71
-0+ +case DP_REMOTE_I2C_READ : +#define IN in->u.i2c_read +#define OUT out->u.i2c_read +if (IN .num_bytes_read !=OUT .num_bytes_read || +IN .num_transactions !=OUT .num_transactions || +IN .port_number !=OUT .port_number || +IN .read_i2c_device_id !=OUT .read_i2c_device_id) +return false; + +for (i =0 ; i <IN .num_transactions; i++) { + txin = &IN .transactions[i]; + txout = &OUT .transactions[i]; + +if (txin->i2c_dev_id != txout->i2c_dev_id || + txin->no_stop_bit != txout->no_stop_bit || + txin->num_bytes != txout->num_bytes || + txin->i2c_transaction_delay != + txout->i2c_transaction_delay) +return false; + +if (memcmp (txin->bytes, txout->bytes, + txin->num_bytes) !=0 ) +return false; + } +break ; +#undef IN +#undef OUT + +case DP_REMOTE_DPCD_WRITE : +#define IN in->u.dpcd_write +#define OUT out->u.dpcd_write +if (IN .dpcd_address !=OUT .dpcd_address || +IN .num_bytes !=OUT .num_bytes || +IN .port_number !=OUT .port_number) +return false; + +return memcmp (IN .bytes,OUT .bytes,IN .num_bytes) ==0 ; +#undef IN +#undef OUT + +case DP_REMOTE_I2C_WRITE : +#define IN in->u.i2c_write +#define OUT out->u.i2c_write +if (IN .port_number !=OUT .port_number || +IN .write_i2c_device_id !=OUT .write_i2c_device_id || +IN .num_bytes !=OUT .num_bytes) +return false; + +return memcmp (IN .bytes,OUT .bytes,IN .num_bytes) ==0 ; +#undef IN +#undef OUT + } + +return true; +} + +static int +__sideband_msg_req_encode_decode (int line, +struct drm_dp_sideband_msg_req_body *in, +struct drm_dp_sideband_msg_req_body *out) +{ +struct drm_printer p =drm_err_printer (pr_fmt ()); +struct drm_dp_sideband_msg_tx txmsg; +int i, ret; +bool eq; + +drm_dp_encode_sideband_req (in, &txmsg); + ret =drm_dp_decode_sideband_req (&txmsg, out); +if (ret <0 ) { + pr_err(pr_fmt("Failed to decode sideband request @ line %d:%d\n"),
drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c +42
-0+ line, ret); +return ret; + } + + eq =sideband_msg_req_equal (in, out); +if (!eq) { +pr_err (pr_fmt ("Encode/decode @ line %d failed, expected:\n" ), + line); +drm_dp_dump_sideband_msg_req_body (in,1 , &p); +pr_err (pr_fmt ("Got:\n" )); +drm_dp_dump_sideband_msg_req_body (out,1 , &p); + } + +switch (in->req_type) { +case DP_REMOTE_DPCD_WRITE : +kfree (out->u.dpcd_write.bytes); +break ; +case DP_REMOTE_I2C_READ : +for (i =0 ; i < out->u.i2c_read.num_transactions; i++) +kfree (out->u.i2c_read.transactions[i].bytes); +break ; +case DP_REMOTE_I2C_WRITE : +kfree (out->u.i2c_write.bytes); +break ; + } + +/* Clear everything but the req_type for the input */ +memset (&in->u,0 ,sizeof (in->u)); + +/* Clear the output entirely */ +memset (out,0 ,sizeof (*out)); + +return eq ?0 : -EINVAL ; +} + +int igt_dp_mst_sideband_msg_req_decode (void *unused) +{ +struct drm_dp_sideband_msg_req_body in = {0 }, out = {0 }; +u8 data[] = {0xff ,0x0 ,0xdd }; +int i; + +#define
DO_TEST()\drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c int igt_dp_mst_calc_pbn_mode(void *ignored) · line 173
Quoted 4 lines
drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c +4
-0+do { \ +if (__sideband_msg_req_encode_decode (__LINE__, &in, &out)) \ +return -EINVAL ; \ + }while (0 )I had a tiny wtf moment here until I realized this is a macro ... maybe put the do on the first line and indent everything? Pure bikeshed to help the blind ...
Quoted 66 lines
drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c +60
-0+ + in.req_type =DP_ENUM_PATH_RESOURCES ; + in.u.port_num.port_number =5 ; +DO_TEST (); + + in.req_type =DP_POWER_UP_PHY ; + in.u.port_num.port_number =5 ; +DO_TEST (); + + in.req_type =DP_POWER_DOWN_PHY ; + in.u.port_num.port_number =5 ; +DO_TEST (); + + in.req_type =DP_ALLOCATE_PAYLOAD ; + in.u.allocate_payload.number_sdp_streams =3 ; +for (i =0 ; i < in.u.allocate_payload.number_sdp_streams; i++) + in.u.allocate_payload.sdp_stream_sink[i] = i +1 ; +DO_TEST (); + in.u.allocate_payload.port_number =0xf ; +DO_TEST (); + in.u.allocate_payload.vcpi =0x7f ; +DO_TEST (); + in.u.allocate_payload.pbn =U16_MAX ; +DO_TEST (); + + in.req_type =DP_QUERY_PAYLOAD ; + in.u.query_payload.port_number =0xf ; +DO_TEST (); + in.u.query_payload.vcpi =0x7f ; +DO_TEST (); + + in.req_type =DP_REMOTE_DPCD_READ ; + in.u.dpcd_read.port_number =0xf ; +DO_TEST (); + in.u.dpcd_read.dpcd_address =0xfedcb ; +DO_TEST (); + in.u.dpcd_read.num_bytes =U8_MAX ; +DO_TEST (); + + in.req_type =DP_REMOTE_DPCD_WRITE ; + in.u.dpcd_write.port_number =0xf ; +DO_TEST (); + in.u.dpcd_write.dpcd_address =0xfedcb ; +DO_TEST (); + in.u.dpcd_write.num_bytes =ARRAY_SIZE (data); + in.u.dpcd_write.bytes = data; +DO_TEST (); + + in.req_type =DP_REMOTE_I2C_READ ; + in.u.i2c_read.port_number =0xf ; +DO_TEST (); + in.u.i2c_read.read_i2c_device_id =0x7f ; +DO_TEST (); + in.u.i2c_read.num_transactions =3 ; + in.u.i2c_read.num_bytes_read =ARRAY_SIZE (data) *3 ; +for (i =0 ; i < in.u.i2c_read.num_transactions; i++) { + in.u.i2c_read.transactions[i].bytes = data; + in.u.i2c_read.transactions[i].num_bytes =ARRAY_SIZE (data); + in.u.i2c_read.transactions[i].i2c_dev_id =0x7f & ~i; + in.u.i2c_read.transactions[i].i2c_transaction_delay = 0xf &~i;
drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c +5
-0+ } +DO_TEST (); + + in.req_type =DP_REMOTE_I2C_WRITE ; + in.u.i2c_write.port_number =0xf ;drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c int igt_dp_mst_calc_pbn_mode(void *ignored) · line 248
Quoted 10 lines
drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c +10
-0+DO_TEST (); + in.u.i2c_write.write_i2c_device_id =0x7f ; +DO_TEST (); + in.u.i2c_write.num_bytes =ARRAY_SIZE (data); + in.u.i2c_write.bytes = data; +DO_TEST (); + +#undef DO_TEST +return 0 ; +}Extremely nice, more unit tests ftw!
With the nits somehow figured out: Reviewed-by: Daniel Vetter < [email protected]>
Quoted 29 lines
drivers/gpu/drm/selftests/test-drm_modeset_common.h +1
-0@@ -40,5 +40,6 @@ int igt_damage_iter_damage_one_outside(void *ignored);int igt_damage_iter_damage_src_moved (void *ignored);int igt_damage_iter_damage_not_visible (void *ignored);int igt_dp_mst_calc_pbn_mode (void *ignored); +int igt_dp_mst_sideband_msg_req_decode (void *ignored);#endif include/drm/drm_dp_mst_helper.h +1
-1@@ -293,7 +293,7 @@ struct drm_dp_remote_dpcd_write {struct drm_dp_remote_i2c_read {u8 num_transactions;u8 port_number; -struct { +struct drm_dp_remote_i2c_read_tx {u8 i2c_dev_id;u8 num_bytes;u8 *bytes;Signature · 1 line
2.21.0Signature · 2 lines
Cheers, Lyude Paul
Signature · 7 lines
Daniel Vetter Software Engineer, Intel Corporation http://blog.ffwll.ch _______________________________________________ dri-devel mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/dri-devel
Lyude Paul
maintainerRe: [PATCH 05/26] drm/dp_mst: Add sideband down request tracing + selftests
Two things below
Quoted 436 lines · Daniel Vetter
Quoted 435 lines · Lyude Paul
Quoted 434 lines · Daniel Vetter
Quoted 103 lines · Lyude Paul
Unfortunately the DP MST helpers do not have much in the way of debugging utilities. So, let's add some!
This adds basic debugging output for down sideband requests that we send
from the driver, so that we can actually discern what's happening when sideband requests timeout. Note that with this commit, we'll be dumping
out sideband requests under both of the following conditions:- When the user has enabled
DRM_UT_DPoutput, of course.
- When the user has enabledDRM_UT_KMSorDRM_UT_DP, and a sideband request has failed in some way. This will allow for developers to
get
a basic idea of what's actually happening with failed modesets on
MST,
without needing to haveDRM_UT_DPexplicitly enabled.Since there wasn't really a good way of testing that any of this worked,
I ended up writing simple selftests that lightly test sideband message encoding and decoding as well. Enjoy!Cc Juston Li <[email protected]>Cc Imre Deak <[email protected]>Cc Ville Syrjälä <[email protected]>Cc Harry Wentland <[email protected]>Signed-off-by Lyude Paul <[email protected]>--- drivers/gpu/drm/drm_dp_mst_topology.c | 308 +++++++++++++++++- .../gpu/drm/drm_dp_mst_topology_internal.h | 24 ++ .../gpu/drm/selftests/drm_modeset_selftests.h | 1 + .../drm/selftests/test-drm_dp_mst_helper.c | 212 ++++++++++++ .../drm/selftests/test-drm_modeset_common.h | 1 + include/drm/drm_dp_mst_helper.h | 2 +- 6 files changed, 543 insertions(+), 5 deletions(-) create mode 100644 drivers/gpu/drm/drm_dp_mst_topology_internal.hdrivers/gpu/drm/drm_dp_mst_topology.c +2
-0@@ -36,6 +36,8 @@#include <drm/drm_print.h> #include <drm/drm_probe_helper.h> +#include "drm_dp_mst_topology_internal.h" +/** * DOC: dp mst helper * @@ -68,6 +70,8 @@ static int drm_dp_mst_register_i2c_bus(struct
drm_dp_aux
*aux);
static voiddrm_dp_mst_unregister_i2c_bus(struct drm_dp_aux *aux); static voiddrm_dp_mst_kick_tx(struct drm_dp_mst_topology_mgr *mgr);drivers/gpu/drm/drm_dp_mst_topology.c +35
-0+#define DBG_PREFIX "[dp_mst]" +#define DP_STR (x) [DP_ ## x] = #xstatic const char *drm_dp_mst_req_type_str (u8 req_type)@@ -124,6 +128,43 @@ static const char *drm_dp_mst_nak_reason_str(u8 nak_reason) }#undef DP_STR +#define DP_STR (x) [DRM_DP_SIDEBAND_TX_ ## x] = #x + +static const char *drm_dp_mst_sideband_tx_state_str (int state) +{ +static const char *const sideband_reason_str[] = { +DP_STR (QUEUED ), +DP_STR (START_SEND ), +DP_STR (SENT ), +DP_STR (RX ), +DP_STR (TIMEOUT ), + }; + +if (state >=ARRAY_SIZE (sideband_reason_str) || + !sideband_reason_str[state]) +return "unknown" ; + +return sideband_reason_str[state]; +} + +static int +drm_dp_mst_rad_to_str (const u8 rad[8 ],u8 lct,char *out,size_t len) +{ +int i; +u8 unpacked_rad[16 ]; + +for (i =0 ; i < lct; i++) { +if (i %2 ) + unpacked_rad[i] = rad[i /2 ] >>4 ; +else + unpacked_rad[i] = rad[i /2 ] &BIT_MASK (4 ); + } + +/* TODO: Eventually add something to printk so we can format the rad
drivers/gpu/drm/drm_dp_mst_topology.c static const char *drm_dp_mst_nak_reason_str(u8 nak_reason) · line 165
Quoted 2 lines
drivers/gpu/drm/drm_dp_mst_topology.c +2
-0+* like this: 1.2.3 +*/ lct *=2;missing here? And yeah the todo would be sweet, but quite a bit more typing I guess.
Quoted 33 lines
drivers/gpu/drm/drm_dp_mst_topology.c +2
-0+return snprintf (out, len,"%*phC" , lct, unpacked_rad); +}/* sideband msg handling */ static u8 drm_dp_msg_header_crc4(const uint8_t *data, size_t
num_nibbles)drivers/gpu/drm/drm_dp_mst_topology.c +2
-2@@ -256,8 +297,9 @@ static bool drm_dp_decode_sideband_msg_hdr(structdrm_dp_sideband_msg_hdr *hdr, return true; } -static void drm_dp_encode_sideband_req (struct drm_dp_sideband_msg_req_body *req, -struct drm_dp_sideband_msg_tx *raw) +void +drm_dp_encode_sideband_req(const struct drm_dp_sideband_msg_req_body*req,
drivers/gpu/drm/drm_dp_mst_topology.c +6
-0+ struct drm_dp_sideband_msg_tx *raw) {int idx =0 ;int i;@@ -362,6 +404,249 @@ static void drm_dp_encode_sideband_req(struct drm_dp_sideband_msg_req_body *req, } raw->cur_len = idx; } +EXPORT_SYMBOL_FOR_TESTS_ONLY (drm_dp_encode_sideband_req); + +/* Decode a sideband request we've encoded, mainly used for debugging */ +int +drm_dp_decode_sideband_req(const struct drm_dp_sideband_msg_tx *raw,drivers/gpu/drm/drm_dp_mst_topology.c static void drm_dp_encode_sideband_req(struct · line 423
Quoted 10 lines
drivers/gpu/drm/drm_dp_mst_topology.c +10
-0+ struct drm_dp_sideband_msg_req_body *req) +{ +const u8 *buf = raw->msg; +int i, idx =0 ; + + req->req_type = buf[idx++] &0x7f ; +switch (req->req_type) { +case DP_ENUM_PATH_RESOURCES : +case DP_POWER_DOWN_PHY : +case DP_POWER_UP_PHY :OCD warning: the enum isn't ordered the same way here and in
drm_dp_encode_sideband_req().drivers/gpu/drm/drm_dp_mst_topology.c static void drm_dp_encode_sideband_req(struct drm_dp_sideband_msg_req_body *req, · line 432
Quoted 11 lines
drivers/gpu/drm/drm_dp_mst_topology.c +11
-0+ req->u.port_num.port_number = (buf[idx] >>4 ) &0xf ; +break ; +case DP_ALLOCATE_PAYLOAD : + { +struct drm_dp_allocate_payload *a = + &req->u.allocate_payload; + + a->number_sdp_streams = buf[idx] &0xf ; + a->port_number = (buf[idx] >>4 ) &0xf ; + + a->vcpi = buf[++idx] &0x7f ;Complain here if
0x80is set?Quoted 16 lines
drivers/gpu/drm/drm_dp_mst_topology.c +8
-0+ + a->pbn = buf[++idx] <<8 ; + a->pbn |= buf[++idx]; + + idx++; +for (i =0 ; i < a->number_sdp_streams; i++) { + a->sdp_stream_sink[i] = + (buf[idx + (i / 2)] >> ((i %2) ? 0 :
4)) &0xf;drivers/gpu/drm/drm_dp_mst_topology.c +5
-0+ } + } +break ; +case DP_QUERY_PAYLOAD : + req->u.query_payload.port_number = (buf[idx] >> 4) &
0xf;drivers/gpu/drm/drm_dp_mst_topology.c static void drm_dp_encode_sideband_req(struct drm_dp_sideband_msg_req_body *req, · line 446
Quoted 1 line
drivers/gpu/drm/drm_dp_mst_topology.c +1
-0+ req->u.query_payload.vcpi = buf[++idx] &0x7f ;Same here for the highest bit?
Quoted 20 lines
drivers/gpu/drm/drm_dp_mst_topology.c +4
-0+break ; +case DP_REMOTE_DPCD_READ : + { + struct drm_dp_remote_dpcd_read *r = &req-Quoted 1 line
u.dpcd_read;
drivers/gpu/drm/drm_dp_mst_topology.c +15
-0+ + r->port_number = (buf[idx] >>4 ) &0xf ; + + r->dpcd_address = (buf[idx] <<16 ) &0xf0000 ; + r->dpcd_address |= (buf[++idx] <<8 ) &0xff00 ; + r->dpcd_address |= buf[++idx] &0xff ; + + r->num_bytes = buf[++idx]; + } +break ; +case DP_REMOTE_DPCD_WRITE : + { +struct drm_dp_remote_dpcd_write *w = + &req->u.dpcd_write; +drivers/gpu/drm/drm_dp_mst_topology.c static void drm_dp_encode_sideband_req(struct drm_dp_sideband_msg_req_body *req, · line 475
Quoted 10 lines
drivers/gpu/drm/drm_dp_mst_topology.c +10
-0+ w->port_number = (buf[idx] >>4 ) &0xf ; + + w->dpcd_address = (buf[idx] <<16 ) &0xf0000 ; + w->dpcd_address |= (buf[++idx] <<8 ) &0xff00 ; + w->dpcd_address |= buf[++idx] &0xff ; + + w->num_bytes = buf[++idx]; + + w->bytes =kmemdup (&buf[++idx], w->num_bytes, +GFP_KERNEL );But if we go really strict on validation we'd need to make sure we don't walk past
raw->cur_len... probably not worth it?Quoted 48 lines
drivers/gpu/drm/drm_dp_mst_topology.c +7
-0+if (!w->bytes) +return -ENOMEM ; + } +break ; +case DP_REMOTE_I2C_READ : + { + struct drm_dp_remote_i2c_read *r = &req-Quoted 1 line
u.i2c_read;
drivers/gpu/drm/drm_dp_mst_topology.c +19
-0+struct drm_dp_remote_i2c_read_tx *tx; +bool failed = false; + + r->num_transactions = buf[idx] &0x3 ; + r->port_number = (buf[idx] >>4 ) &0xf ; +for (i =0 ; i < r->num_transactions; i++) { + tx = &r->transactions[i]; + + tx->i2c_dev_id = buf[++idx] &0x7f ; + tx->num_bytes = buf[++idx]; + tx->bytes =kmemdup (&buf[++idx], + tx->num_bytes, +GFP_KERNEL ); +if (!tx->bytes) { + failed = true; +break ; + } + idx += tx->num_bytes; + tx->no_stop_bit = (buf[idx] >> 5) &
0x1;drivers/gpu/drm/drm_dp_mst_topology.c +1
-0+ tx->i2c_transaction_delay = buf[idx] &
0xf;drivers/gpu/drm/drm_dp_mst_topology.c +4
-0+ } + +if (failed) { + for (i = 0; i < r->num_transactions;i++)
drivers/gpu/drm/drm_dp_mst_topology.c +11
-0+kfree (tx->bytes); +return -ENOMEM ; + } + + r->read_i2c_device_id = buf[++idx] &0x7f ; + r->num_bytes_read = buf[++idx]; + } +break ; +case DP_REMOTE_I2C_WRITE : + { + struct drm_dp_remote_i2c_write *w = &req-Quoted 1 line
u.i2c_write;
drivers/gpu/drm/drm_dp_mst_topology.c +1
-0+drivers/gpu/drm/drm_dp_mst_topology.c static void drm_dp_encode_sideband_req(struct drm_dp_sideband_msg_req_body *req, · line 528
Quoted 10 lines
drivers/gpu/drm/drm_dp_mst_topology.c +10
-0+ w->port_number = (buf[idx] >>4 ) &0xf ; + w->write_i2c_device_id = buf[++idx] &0x7f ; + w->num_bytes = buf[++idx]; + w->bytes =kmemdup (&buf[++idx], w->num_bytes, +GFP_KERNEL ); +if (!w->bytes) +return -ENOMEM ; + } +break ; + }For safety: Should we validate
cur_lenhere? Or return it, and let the caller deal with any mismatches.But I'm kinda leaning towards no safety here, since it's not dealing with
any replies we get from the hw/externally, which might be used for attacking us. If we later on have the same thing but fordrm_dp_sideband_parse_reply()then a lot more fun.Quoted 139 lines
drivers/gpu/drm/drm_dp_mst_topology.c +7
-0+ +return 0 ; +} +EXPORT_SYMBOL_FOR_TESTS_ONLY (drm_dp_decode_sideband_req); + +void +drm_dp_dump_sideband_msg_req_body(const struct
drm_dp_sideband_msg_req_body*req,drivers/gpu/drm/drm_dp_mst_topology.c +1
-0+ int indent, struct drm_printer*printer)
drivers/gpu/drm/drm_dp_mst_topology.c +4
-0+{ +int i; + +#define P (f, ...) drm_printf_indent(printer, indent, f,##__VA_ARGS__)
drivers/gpu/drm/drm_dp_mst_topology.c +3
-0+if (req->req_type ==DP_LINK_ADDRESS ) { +/* No contents to print */ + P("type=%s\n", drm_dp_mst_req_type_str(req-Quoted 1 line
req_type));drivers/gpu/drm/drm_dp_mst_topology.c +4
-0+return ; + } + + P("type=%s contents:\n", drm_dp_mst_req_type_str(req-Quoted 1 line
req_type));drivers/gpu/drm/drm_dp_mst_topology.c +12
-0+ indent++; + +switch (req->req_type) { +case DP_ENUM_PATH_RESOURCES : +case DP_POWER_DOWN_PHY : +case DP_POWER_UP_PHY : +P ("port=%d\n" , req->u.port_num.port_number); +break ; +case DP_ALLOCATE_PAYLOAD : + P("port=%d vcpi=%d pbn=%d sdp_streams=%d %*ph\n", + req->u.allocate_payload.port_number, + req->u.allocate_payload.vcpi, req-Quoted 1 line
u.allocate_payload.pbn,
drivers/gpu/drm/drm_dp_mst_topology.c +12
-0+ req->u.allocate_payload.number_sdp_streams, + req->u.allocate_payload.number_sdp_streams, + req->u.allocate_payload.sdp_stream_sink); +break ; +case DP_QUERY_PAYLOAD : +P ("port=%d vcpi=%d\n" , + req->u.query_payload.port_number, + req->u.query_payload.vcpi); +break ; +case DP_REMOTE_DPCD_READ : + P("port=%d dpcd_addr=%05x len=%d\n", + req->u.dpcd_read.port_number, req-Quoted 1 line
u.dpcd_read.dpcd_address,
drivers/gpu/drm/drm_dp_mst_topology.c +7
-0+ req->u.dpcd_read.num_bytes); +break ; +case DP_REMOTE_DPCD_WRITE : + P("port=%d addr=%05x len=%d: %*ph\n", + req->u.dpcd_write.port_number, + req->u.dpcd_write.dpcd_address, + req->u.dpcd_write.num_bytes, req-Quoted 1 line
u.dpcd_write.num_bytes,
drivers/gpu/drm/drm_dp_mst_topology.c +11
-0+ req->u.dpcd_write.bytes); +break ; +case DP_REMOTE_I2C_READ : +P ("port=%d num_tx=%d id=%d size=%d:\n" , + req->u.i2c_read.port_number, + req->u.i2c_read.num_transactions, + req->u.i2c_read.read_i2c_device_id, + req->u.i2c_read.num_bytes_read); + + indent++; +for (i =0 ; i < req->u.i2c_read.num_transactions; i++){
drivers/gpu/drm/drm_dp_mst_topology.c +4
-0+const struct drm_dp_remote_i2c_read_tx *rtx = + &req->u.i2c_read.transactions[i]; + + P("%d: id=%03d size=%03d no_stop_bit=%dtx_delay=%03d:
%*ph\n",drivers/gpu/drm/drm_dp_mst_topology.c +2
-0+ i, rtx->i2c_dev_id, rtx->num_bytes, + rtx->no_stop_bit, rtx-Quoted 1 line
i2c_transaction_delay,drivers/gpu/drm/drm_dp_mst_topology.c +8
-0+ rtx->num_bytes, rtx->bytes); + } +break ; +case DP_REMOTE_I2C_WRITE : + P("port=%d id=%d size=%d: %*ph\n", + req->u.i2c_write.port_number, + req->u.i2c_write.write_i2c_device_id, + req->u.i2c_write.num_bytes, req-Quoted 1 line
u.i2c_write.num_bytes,
drivers/gpu/drm/drm_dp_mst_topology.c +13
-0+ req->u.i2c_write.bytes); +break ; +default : +P ("???\n" ); +break ; + } +#undef P +} +EXPORT_SYMBOL_FOR_TESTS_ONLY (drm_dp_dump_sideband_msg_req_body); + +static inline void +drm_dp_mst_dump_sideband_msg_tx (struct drm_printer *p, + const struct drm_dp_sideband_msg_tx*txmsg)
drivers/gpu/drm/drm_dp_mst_topology.c +9
-0+{ +struct drm_dp_sideband_msg_req_body req; +char buf[64 ]; +int ret; +int i; + +drm_dp_mst_rad_to_str (txmsg->dst->rad, txmsg->dst->lct, buf, +sizeof (buf)); + drm_printf(p, "txmsg cur_offset=%x cur_len=%x seqno=%xstate=%s
path_msg=%d dst=%s\n",drivers/gpu/drm/drm_dp_mst_topology.c +7
-0+ txmsg->cur_offset, txmsg->cur_len, txmsg->seqno, +drm_dp_mst_sideband_tx_state_str (txmsg->state), + txmsg->path_msg, buf); + + ret =drm_dp_decode_sideband_req (txmsg, &req); +if (ret) { + drm_printf(p, "<failed to decode sideband req: %d>\n",ret);
drivers/gpu/drm/drm_dp_mst_topology.c +17
-0+return ; + } +drm_dp_dump_sideband_msg_req_body (&req,1 , p); + +switch (req.req_type) { +case DP_REMOTE_DPCD_WRITE : +kfree (req.u.dpcd_write.bytes); +break ; +case DP_REMOTE_I2C_READ : +for (i =0 ; i < req.u.i2c_read.num_transactions; i++) +kfree (req.u.i2c_read.transactions[i].bytes); +break ; +case DP_REMOTE_I2C_WRITE : +kfree (req.u.i2c_write.bytes); +break ; + } +}
Quoted 11 lines
Quoted 11 lines
Quoted 11 lines
drivers/gpu/drm/drm_dp_mst_topology.c static int drm_dp_mst_wait_tx_reply(struct · line 1183
Quoted 9 lines
static void
drm_dp_crc_sideband_chunk_req(u8 *msg, u8 len){drivers/gpu/drm/drm_dp_mst_topology.c +1
-0@@ -893,6 +1178,11 @@ static int drm_dp_mst_wait_tx_reply(struct drm_dp_mst_branch *mstb, } } out: +if (unlikely (ret == -EIO && drm_debug & (DRM_UT_DP |DRM_UT_KMS ))) {Hm I'd only check for
DRM_UT_DPhere.
Was going to go ahead with this, but realized it might not be a great idea. One of the reasons I went with DRM_UT_DP | DRM_UT_KMS here is because generally, I'd assume that (this is the case for me at least) DRM_UT_DP serves as a way to trace what's going on with DP, and DRM_UT_KMS is for non-atomic specific modesetting debug info. That being said, if we fail sending a message like PAYLOAD_ALLOCATE then the modeset itself is going to not work correctly and thus - the contents of failure would still be relevant to someone working on modesetting issues even if they're not interested in the entirity of what's going on between the sink and the source.
That, and we were already printing when messages timed out under DRM_UT_KMS. Granted though, that printing was added well before I introduced DRM_UT_DP. However, I suppose I could just move that back to using DRM_DEBUG_KMS() and dump the sideband message seperately.
Maybe I'm overthinking this a bit, thoughts?
Quoted 401 lines
Quoted 137 lines
Quoted 128 lines
Quoted 9 lines
drivers/gpu/drm/drm_dp_mst_topology.c +4
-0+struct drm_printer p =drm_debug_printer (DBG_PREFIX ); + +drm_dp_mst_dump_sideband_msg_tx (&p, txmsg); + }mutex_unlock (&mgr->qlock);drivers/gpu/drm/drm_dp_mst_topology.c +0
-0return ret;drivers/gpu/drm/drm_dp_mst_topology.c static int process_single_tx_qlock(struct · line 2221
Quoted 7 lines
drivers/gpu/drm/drm_dp_mst_topology.c +1
-2@@ -1927,8 +2217,11 @@ static int process_single_tx_qlock(struct drm_dp_mst_topology_mgr *mgr, idx += tosend +1 ; ret =drm_dp_send_sideband_msg (mgr, up, chunk, idx); -if (ret) { -DRM_DEBUG_KMS ("sideband msg failed to send\n" ); +if (ret &&unlikely (drm_debug & (DRM_UT_DP |DRM_UT_KMS ))) {Same.
Quoted 8 lines
drivers/gpu/drm/drm_dp_mst_topology.c +4
-0+struct drm_printer p =drm_debug_printer (DBG_PREFIX ); + +drm_printf (&p,"sideband msg failed to send\n" ); +drm_dp_mst_dump_sideband_msg_tx (&p, txmsg);return ret; }drivers/gpu/drm/drm_dp_mst_topology.c static void drm_dp_queue_down_tx(struct · line 2288
Quoted 6 lines
drivers/gpu/drm/drm_dp_mst_topology.c +2
-0@@ -1990,6 +2283,13 @@ static void drm_dp_queue_down_tx(struct drm_dp_mst_topology_mgr *mgr, {mutex_lock (&mgr->qlock);list_add_tail (&txmsg->next, &mgr->tx_msg_downq); + +if (unlikely (drm_debug &DRM_UT_DP )) {Like you do here.
Quoted 92 lines
drivers/gpu/drm/drm_dp_mst_topology.c +5
-0+struct drm_printer p =drm_debug_printer (DBG_PREFIX ); + +drm_dp_mst_dump_sideband_msg_tx (&p, txmsg); + } +if (list_is_singular (&mgr->tx_msg_downq))process_single_down_tx_qlock (mgr);drivers/gpu/drm/drm_dp_mst_topology.c +0
-0mutex_unlock (&mgr->qlock);drivers/gpu/drm/drm_dp_mst_topology_internal.h new file +3
-0@@ -0,0 +1,24 @@ +/* SPDX-License-Identifier: GPL-2.0-only +* +* Declarations for DP MST related functions which are only used in selftests
+ *
+ * Copyright © 2018 Red Hat
+ * Authors:
+ * Lyude Paul <[email protected]>
+ */drivers/gpu/drm/drm_dp_mst_topology_internal.h new file +8
-0+ +#ifndef _DRM_DP_MST_HELPER_INTERNAL_H_ +#define _DRM_DP_MST_HELPER_INTERNAL_H_ + +#include <drm/drm_dp_mst_helper.h> + +void +drm_dp_encode_sideband_req(const struct drm_dp_sideband_msg_req_body*req,
drivers/gpu/drm/drm_dp_mst_topology_internal.h new file +2
-0+ struct drm_dp_sideband_msg_tx *raw); +int drm_dp_decode_sideband_req(const struct drm_dp_sideband_msg_tx*raw,
drivers/gpu/drm/drm_dp_mst_topology_internal.h new file +1
-0+struct drm_dp_sideband_msg_req_body *req);
drivers/gpu/drm/drm_dp_mst_topology_internal.h new file +2
-0+void +drm_dp_dump_sideband_msg_req_body(const struct
drm_dp_sideband_msg_req_body*req,drivers/gpu/drm/drm_dp_mst_topology_internal.h new file +1
-0+ int indent, struct drm_printer*printer);
drivers/gpu/drm/drm_dp_mst_topology_internal.h new file +2
-0+ +#endif /* !_DRM_DP_MST_HELPER_INTERNAL_H_ */ drivers/gpu/drm/selftests/drm_modeset_selftests.h +0
-0@@ -33,3 +33,4 @@ selftest(damage_iter_damage_one_outside, igt_damage_iter_damage_one_outside) selftest(damage_iter_damage_src_moved, igt_damage_iter_damage_src_moved) selftest(damage_iter_damage_not_visible,
igt_damage_iter_damage_not_visible)
selftest(dp_mst_calc_pbn_mode, igt_dp_mst_calc_pbn_mode)
+selftest(dp_mst_sideband_msg_req_decode,igt_dp_mst_sideband_msg_req_decode)drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c +17
-0@@ -8,6 +8,7 @@#include <drm/drm_dp_mst_helper.h> #include <drm/drm_print.h> +#include "../drm_dp_mst_topology_internal.h" #include "test-drm_modeset_common.h" int igt_dp_mst_calc_pbn_mode (void *ignored)@@ -34,3 +35,214 @@ int igt_dp_mst_calc_pbn_mode(void *ignored)return 0 ; } + +static bool +sideband_msg_req_equal (const struct drm_dp_sideband_msg_req_body *in, +const struct drm_dp_sideband_msg_req_body *out) +{ +const struct drm_dp_remote_i2c_read_tx *txin, *txout; +int i; + +if (in->req_type != out->req_type) +return false; + +switch (in->req_type) { +case DP_ENUM_PATH_RESOURCES : +case DP_POWER_UP_PHY : +case DP_POWER_DOWN_PHY : +case DP_ALLOCATE_PAYLOAD :Quoted 8 lines
drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c int igt_dp_mst_calc_pbn_mode(void *ignored) · line 56
Quoted 3 lines
drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c +3
-0+case DP_QUERY_PAYLOAD : +case DP_REMOTE_DPCD_READ : +return memcmp (in, out,sizeof (*in)) ==0 ;Just memcmp the entire thing for everyrone (we assume kzalloc anyway), and
then only have the additional checks for the few cases we need it? Would slightly reduce the code.Quoted 8 lines
Just realized also - this wouldn't work: remember that
I2C_READ,DPCD_WRITE,
andI2C_WRITEcontain pointers which of course will be different even if the
contents are the same. Unfortunately I don't think there's really a good solution for this, unless we want to just remove the pointers and store things
within the down request struct itselfUh ... disappoint :-/
Then at least put that into a comment, and maybe make this the default:
handler?Quoted 251 lines
Quoted 247 lines
Quoted 124 lines
drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c +46
-0+ +case DP_REMOTE_I2C_READ : +#define IN in->u.i2c_read +#define OUT out->u.i2c_read +if (IN .num_bytes_read !=OUT .num_bytes_read || +IN .num_transactions !=OUT .num_transactions || +IN .port_number !=OUT .port_number || +IN .read_i2c_device_id !=OUT .read_i2c_device_id) +return false; + +for (i =0 ; i <IN .num_transactions; i++) { + txin = &IN .transactions[i]; + txout = &OUT .transactions[i]; + +if (txin->i2c_dev_id != txout->i2c_dev_id || + txin->no_stop_bit != txout->no_stop_bit || + txin->num_bytes != txout->num_bytes || + txin->i2c_transaction_delay != + txout->i2c_transaction_delay) +return false; + +if (memcmp (txin->bytes, txout->bytes, + txin->num_bytes) !=0 ) +return false; + } +break ; +#undef IN +#undef OUT + +case DP_REMOTE_DPCD_WRITE : +#define IN in->u.dpcd_write +#define OUT out->u.dpcd_write +if (IN .dpcd_address !=OUT .dpcd_address || +IN .num_bytes !=OUT .num_bytes || +IN .port_number !=OUT .port_number) +return false; + +return memcmp (IN .bytes,OUT .bytes,IN .num_bytes) ==0 ; +#undef IN +#undef OUT + +case DP_REMOTE_I2C_WRITE : +#define IN in->u.i2c_write +#define OUT out->u.i2c_write +if (IN .port_number !=OUT .port_number || + IN.write_i2c_device_id != OUT.write_i2c_device_id||
drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c +14
-0+ IN.num_bytes != OUT.num_bytes) +return false; + +return memcmp (IN .bytes,OUT .bytes,IN .num_bytes) ==0 ; +#undef IN +#undef OUT + } + +return true; +} + +static int +__sideband_msg_req_encode_decode (int line, + struct drm_dp_sideband_msg_req_body*in,
drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c +1
-0+struct drm_dp_sideband_msg_req_body *out)
drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c +10
-0+{ +struct drm_printer p =drm_err_printer (pr_fmt ()); +struct drm_dp_sideband_msg_tx txmsg; +int i, ret; +bool eq; + +drm_dp_encode_sideband_req (in, &txmsg); + ret =drm_dp_decode_sideband_req (&txmsg, out); +if (ret <0 ) { + pr_err(pr_fmt("Failed to decode sideband request @line %d:
%d\n"),drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c +7
-0+ line, ret); +return ret; + } + + eq =sideband_msg_req_equal (in, out); +if (!eq) { + pr_err(pr_fmt("Encode/decode @ line %d failed,expected:\n"),
drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c +35
-0+ line); +drm_dp_dump_sideband_msg_req_body (in,1 , &p); +pr_err (pr_fmt ("Got:\n" )); +drm_dp_dump_sideband_msg_req_body (out,1 , &p); + } + +switch (in->req_type) { +case DP_REMOTE_DPCD_WRITE : +kfree (out->u.dpcd_write.bytes); +break ; +case DP_REMOTE_I2C_READ : +for (i =0 ; i < out->u.i2c_read.num_transactions; i++) +kfree (out->u.i2c_read.transactions[i].bytes); +break ; +case DP_REMOTE_I2C_WRITE : +kfree (out->u.i2c_write.bytes); +break ; + } + +/* Clear everything but the req_type for the input */ +memset (&in->u,0 ,sizeof (in->u)); + +/* Clear the output entirely */ +memset (out,0 ,sizeof (*out)); + +return eq ?0 : -EINVAL ; +} + +int igt_dp_mst_sideband_msg_req_decode (void *unused) +{ +struct drm_dp_sideband_msg_req_body in = {0 }, out = {0 }; +u8 data[] = {0xff ,0x0 ,0xdd }; +int i; + +#define
DO_TEST()\drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c +1
-0+ do{ \
drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c +1
-0+ if (__sideband_msg_req_encode_decode(__LINE__, &in,&out)) \
drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c
Quoted 3 lines
drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c +2
-1+ return -EINVAL ; \ + }while (0 )I had a tiny wtf moment here until I realized this is a macro ... maybe put the do on the first line and indent everything? Pure bikeshed to help
the blind ...Quoted 68 lines
drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c +58
-0+ + in.req_type =DP_ENUM_PATH_RESOURCES ; + in.u.port_num.port_number =5 ; +DO_TEST (); + + in.req_type =DP_POWER_UP_PHY ; + in.u.port_num.port_number =5 ; +DO_TEST (); + + in.req_type =DP_POWER_DOWN_PHY ; + in.u.port_num.port_number =5 ; +DO_TEST (); + + in.req_type =DP_ALLOCATE_PAYLOAD ; + in.u.allocate_payload.number_sdp_streams =3 ; +for (i =0 ; i < in.u.allocate_payload.number_sdp_streams; i++) + in.u.allocate_payload.sdp_stream_sink[i] = i +1 ; +DO_TEST (); + in.u.allocate_payload.port_number =0xf ; +DO_TEST (); + in.u.allocate_payload.vcpi =0x7f ; +DO_TEST (); + in.u.allocate_payload.pbn =U16_MAX ; +DO_TEST (); + + in.req_type =DP_QUERY_PAYLOAD ; + in.u.query_payload.port_number =0xf ; +DO_TEST (); + in.u.query_payload.vcpi =0x7f ; +DO_TEST (); + + in.req_type =DP_REMOTE_DPCD_READ ; + in.u.dpcd_read.port_number =0xf ; +DO_TEST (); + in.u.dpcd_read.dpcd_address =0xfedcb ; +DO_TEST (); + in.u.dpcd_read.num_bytes =U8_MAX ; +DO_TEST (); + + in.req_type =DP_REMOTE_DPCD_WRITE ; + in.u.dpcd_write.port_number =0xf ; +DO_TEST (); + in.u.dpcd_write.dpcd_address =0xfedcb ; +DO_TEST (); + in.u.dpcd_write.num_bytes =ARRAY_SIZE (data); + in.u.dpcd_write.bytes = data; +DO_TEST (); + + in.req_type =DP_REMOTE_I2C_READ ; + in.u.i2c_read.port_number =0xf ; +DO_TEST (); + in.u.i2c_read.read_i2c_device_id =0x7f ; +DO_TEST (); + in.u.i2c_read.num_transactions =3 ; + in.u.i2c_read.num_bytes_read =ARRAY_SIZE (data) *3 ; +for (i =0 ; i < in.u.i2c_read.num_transactions; i++) { + in.u.i2c_read.transactions[i].bytes = data; + in.u.i2c_read.transactions[i].num_bytes =
ARRAY_SIZE(data);drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c +2
-0+ in.u.i2c_read.transactions[i].i2c_dev_id =0x7f & ~i; + in.u.i2c_read.transactions[i].i2c_transaction_delay =
0xf&
~i;drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c +5
-0+ } +DO_TEST (); + + in.req_type =DP_REMOTE_I2C_WRITE ; + in.u.i2c_write.port_number =0xf ;drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c int igt_dp_mst_calc_pbn_mode(void *ignored) · line 248
Quoted 10 lines
drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c +10
-0+DO_TEST (); + in.u.i2c_write.write_i2c_device_id =0x7f ; +DO_TEST (); + in.u.i2c_write.num_bytes =ARRAY_SIZE (data); + in.u.i2c_write.bytes = data; +DO_TEST (); + +#undef DO_TEST +return 0 ; +}Extremely nice, more unit tests ftw!
With the nits somehow figured out: Reviewed-by: Daniel Vetter < [email protected]>
Quoted 30 lines
drivers/gpu/drm/selftests/test-drm_modeset_common.h +1
-0@@ -40,5 +40,6 @@ int igt_damage_iter_damage_one_outside(void *ignored);int igt_damage_iter_damage_src_moved (void *ignored);int igt_damage_iter_damage_not_visible (void *ignored);int igt_dp_mst_calc_pbn_mode (void *ignored); +int igt_dp_mst_sideband_msg_req_decode (void *ignored);#endif
include/drm/drm_dp_mst_helper.h +1
-1@@ -293,7 +293,7 @@ struct drm_dp_remote_dpcd_write {struct drm_dp_remote_i2c_read {u8 num_transactions;u8 port_number; -struct { +struct drm_dp_remote_i2c_read_tx {u8 i2c_dev_id;u8 num_bytes;u8 *bytes;Signature · 1 line
2.21.0Signature · 2 lines
Cheers, Lyude Paul
Signature · 2 lines
Cheers, Lyude Paul
Re: [PATCH 05/26] drm/dp_mst: Add sideband down request tracing + selftests
Quoted 461 lines · Lyude Paul
Two things below
Quoted 436 lines · Daniel Vetter
Quoted 435 lines · Lyude Paul
Quoted 434 lines · Daniel Vetter
Quoted 103 lines · Lyude Paul
Unfortunately the DP MST helpers do not have much in the way of debugging utilities. So, let's add some!
This adds basic debugging output for down sideband requests that we send
from the driver, so that we can actually discern what's happening when sideband requests timeout. Note that with this commit, we'll be dumping
out sideband requests under both of the following conditions:- When the user has enabled
DRM_UT_DPoutput, of course.
- When the user has enabledDRM_UT_KMSorDRM_UT_DP, and a sideband request has failed in some way. This will allow for developers to
get
a basic idea of what's actually happening with failed modesets on
MST,
without needing to haveDRM_UT_DPexplicitly enabled.Since there wasn't really a good way of testing that any of this worked,
I ended up writing simple selftests that lightly test sideband message encoding and decoding as well. Enjoy!Cc Juston Li <[email protected]>Cc Imre Deak <[email protected]>Cc Ville Syrjälä <[email protected]>Cc Harry Wentland <[email protected]>Signed-off-by Lyude Paul <[email protected]>--- drivers/gpu/drm/drm_dp_mst_topology.c | 308 +++++++++++++++++- .../gpu/drm/drm_dp_mst_topology_internal.h | 24 ++ .../gpu/drm/selftests/drm_modeset_selftests.h | 1 + .../drm/selftests/test-drm_dp_mst_helper.c | 212 ++++++++++++ .../drm/selftests/test-drm_modeset_common.h | 1 + include/drm/drm_dp_mst_helper.h | 2 +- 6 files changed, 543 insertions(+), 5 deletions(-) create mode 100644 drivers/gpu/drm/drm_dp_mst_topology_internal.hdrivers/gpu/drm/drm_dp_mst_topology.c +2
-0@@ -36,6 +36,8 @@#include <drm/drm_print.h> #include <drm/drm_probe_helper.h> +#include "drm_dp_mst_topology_internal.h" +/** * DOC: dp mst helper * @@ -68,6 +70,8 @@ static int drm_dp_mst_register_i2c_bus(struct
drm_dp_aux
*aux);
static voiddrm_dp_mst_unregister_i2c_bus(struct drm_dp_aux *aux); static voiddrm_dp_mst_kick_tx(struct drm_dp_mst_topology_mgr *mgr);drivers/gpu/drm/drm_dp_mst_topology.c +35
-0+#define DBG_PREFIX "[dp_mst]" +#define DP_STR (x) [DP_ ## x] = #xstatic const char *drm_dp_mst_req_type_str (u8 req_type)@@ -124,6 +128,43 @@ static const char *drm_dp_mst_nak_reason_str(u8 nak_reason) }#undef DP_STR +#define DP_STR (x) [DRM_DP_SIDEBAND_TX_ ## x] = #x + +static const char *drm_dp_mst_sideband_tx_state_str (int state) +{ +static const char *const sideband_reason_str[] = { +DP_STR (QUEUED ), +DP_STR (START_SEND ), +DP_STR (SENT ), +DP_STR (RX ), +DP_STR (TIMEOUT ), + }; + +if (state >=ARRAY_SIZE (sideband_reason_str) || + !sideband_reason_str[state]) +return "unknown" ; + +return sideband_reason_str[state]; +} + +static int +drm_dp_mst_rad_to_str (const u8 rad[8 ],u8 lct,char *out,size_t len) +{ +int i; +u8 unpacked_rad[16 ]; + +for (i =0 ; i < lct; i++) { +if (i %2 ) + unpacked_rad[i] = rad[i /2 ] >>4 ; +else + unpacked_rad[i] = rad[i /2 ] &BIT_MASK (4 ); + } + +/* TODO: Eventually add something to printk so we can format the rad
drivers/gpu/drm/drm_dp_mst_topology.c static const char *drm_dp_mst_nak_reason_str(u8 nak_reason) · line 165
Quoted 2 lines
drivers/gpu/drm/drm_dp_mst_topology.c +2
-0+* like this: 1.2.3 +*/ lct *=2;missing here? And yeah the todo would be sweet, but quite a bit more typing I guess.
Quoted 33 lines
drivers/gpu/drm/drm_dp_mst_topology.c +2
-0+return snprintf (out, len,"%*phC" , lct, unpacked_rad); +}/* sideband msg handling */ static u8 drm_dp_msg_header_crc4(const uint8_t *data, size_t
num_nibbles)drivers/gpu/drm/drm_dp_mst_topology.c +2
-2@@ -256,8 +297,9 @@ static bool drm_dp_decode_sideband_msg_hdr(structdrm_dp_sideband_msg_hdr *hdr, return true; } -static void drm_dp_encode_sideband_req (struct drm_dp_sideband_msg_req_body *req, -struct drm_dp_sideband_msg_tx *raw) +void +drm_dp_encode_sideband_req(const struct drm_dp_sideband_msg_req_body*req,
drivers/gpu/drm/drm_dp_mst_topology.c +6
-0+ struct drm_dp_sideband_msg_tx *raw) {int idx =0 ;int i;@@ -362,6 +404,249 @@ static void drm_dp_encode_sideband_req(struct drm_dp_sideband_msg_req_body *req, } raw->cur_len = idx; } +EXPORT_SYMBOL_FOR_TESTS_ONLY (drm_dp_encode_sideband_req); + +/* Decode a sideband request we've encoded, mainly used for debugging */ +int +drm_dp_decode_sideband_req(const struct drm_dp_sideband_msg_tx *raw,drivers/gpu/drm/drm_dp_mst_topology.c static void drm_dp_encode_sideband_req(struct · line 423
Quoted 10 lines
drivers/gpu/drm/drm_dp_mst_topology.c +10
-0+ struct drm_dp_sideband_msg_req_body *req) +{ +const u8 *buf = raw->msg; +int i, idx =0 ; + + req->req_type = buf[idx++] &0x7f ; +switch (req->req_type) { +case DP_ENUM_PATH_RESOURCES : +case DP_POWER_DOWN_PHY : +case DP_POWER_UP_PHY :OCD warning: the enum isn't ordered the same way here and in
drm_dp_encode_sideband_req().drivers/gpu/drm/drm_dp_mst_topology.c static void drm_dp_encode_sideband_req(struct drm_dp_sideband_msg_req_body *req, · line 432
Quoted 11 lines
drivers/gpu/drm/drm_dp_mst_topology.c +11
-0+ req->u.port_num.port_number = (buf[idx] >>4 ) &0xf ; +break ; +case DP_ALLOCATE_PAYLOAD : + { +struct drm_dp_allocate_payload *a = + &req->u.allocate_payload; + + a->number_sdp_streams = buf[idx] &0xf ; + a->port_number = (buf[idx] >>4 ) &0xf ; + + a->vcpi = buf[++idx] &0x7f ;Complain here if
0x80is set?Quoted 16 lines
drivers/gpu/drm/drm_dp_mst_topology.c +8
-0+ + a->pbn = buf[++idx] <<8 ; + a->pbn |= buf[++idx]; + + idx++; +for (i =0 ; i < a->number_sdp_streams; i++) { + a->sdp_stream_sink[i] = + (buf[idx + (i / 2)] >> ((i %2) ? 0 :
4)) &0xf;drivers/gpu/drm/drm_dp_mst_topology.c +5
-0+ } + } +break ; +case DP_QUERY_PAYLOAD : + req->u.query_payload.port_number = (buf[idx] >> 4) &
0xf;drivers/gpu/drm/drm_dp_mst_topology.c static void drm_dp_encode_sideband_req(struct drm_dp_sideband_msg_req_body *req, · line 446
Quoted 1 line
drivers/gpu/drm/drm_dp_mst_topology.c +1
-0+ req->u.query_payload.vcpi = buf[++idx] &0x7f ;Same here for the highest bit?
Quoted 20 lines
drivers/gpu/drm/drm_dp_mst_topology.c +4
-0+break ; +case DP_REMOTE_DPCD_READ : + { + struct drm_dp_remote_dpcd_read *r = &req-Quoted 1 line
u.dpcd_read;
drivers/gpu/drm/drm_dp_mst_topology.c +15
-0+ + r->port_number = (buf[idx] >>4 ) &0xf ; + + r->dpcd_address = (buf[idx] <<16 ) &0xf0000 ; + r->dpcd_address |= (buf[++idx] <<8 ) &0xff00 ; + r->dpcd_address |= buf[++idx] &0xff ; + + r->num_bytes = buf[++idx]; + } +break ; +case DP_REMOTE_DPCD_WRITE : + { +struct drm_dp_remote_dpcd_write *w = + &req->u.dpcd_write; +drivers/gpu/drm/drm_dp_mst_topology.c static void drm_dp_encode_sideband_req(struct drm_dp_sideband_msg_req_body *req, · line 475
Quoted 10 lines
drivers/gpu/drm/drm_dp_mst_topology.c +10
-0+ w->port_number = (buf[idx] >>4 ) &0xf ; + + w->dpcd_address = (buf[idx] <<16 ) &0xf0000 ; + w->dpcd_address |= (buf[++idx] <<8 ) &0xff00 ; + w->dpcd_address |= buf[++idx] &0xff ; + + w->num_bytes = buf[++idx]; + + w->bytes =kmemdup (&buf[++idx], w->num_bytes, +GFP_KERNEL );But if we go really strict on validation we'd need to make sure we don't walk past
raw->cur_len... probably not worth it?Quoted 48 lines
drivers/gpu/drm/drm_dp_mst_topology.c +7
-0+if (!w->bytes) +return -ENOMEM ; + } +break ; +case DP_REMOTE_I2C_READ : + { + struct drm_dp_remote_i2c_read *r = &req-Quoted 1 line
u.i2c_read;
drivers/gpu/drm/drm_dp_mst_topology.c +19
-0+struct drm_dp_remote_i2c_read_tx *tx; +bool failed = false; + + r->num_transactions = buf[idx] &0x3 ; + r->port_number = (buf[idx] >>4 ) &0xf ; +for (i =0 ; i < r->num_transactions; i++) { + tx = &r->transactions[i]; + + tx->i2c_dev_id = buf[++idx] &0x7f ; + tx->num_bytes = buf[++idx]; + tx->bytes =kmemdup (&buf[++idx], + tx->num_bytes, +GFP_KERNEL ); +if (!tx->bytes) { + failed = true; +break ; + } + idx += tx->num_bytes; + tx->no_stop_bit = (buf[idx] >> 5) &
0x1;drivers/gpu/drm/drm_dp_mst_topology.c +1
-0+ tx->i2c_transaction_delay = buf[idx] &
0xf;drivers/gpu/drm/drm_dp_mst_topology.c +4
-0+ } + +if (failed) { + for (i = 0; i < r->num_transactions;i++)
drivers/gpu/drm/drm_dp_mst_topology.c +11
-0+kfree (tx->bytes); +return -ENOMEM ; + } + + r->read_i2c_device_id = buf[++idx] &0x7f ; + r->num_bytes_read = buf[++idx]; + } +break ; +case DP_REMOTE_I2C_WRITE : + { + struct drm_dp_remote_i2c_write *w = &req-Quoted 1 line
u.i2c_write;
drivers/gpu/drm/drm_dp_mst_topology.c +1
-0+drivers/gpu/drm/drm_dp_mst_topology.c static void drm_dp_encode_sideband_req(struct drm_dp_sideband_msg_req_body *req, · line 528
Quoted 10 lines
drivers/gpu/drm/drm_dp_mst_topology.c +10
-0+ w->port_number = (buf[idx] >>4 ) &0xf ; + w->write_i2c_device_id = buf[++idx] &0x7f ; + w->num_bytes = buf[++idx]; + w->bytes =kmemdup (&buf[++idx], w->num_bytes, +GFP_KERNEL ); +if (!w->bytes) +return -ENOMEM ; + } +break ; + }For safety: Should we validate
cur_lenhere? Or return it, and let the caller deal with any mismatches.But I'm kinda leaning towards no safety here, since it's not dealing with
any replies we get from the hw/externally, which might be used for attacking us. If we later on have the same thing but fordrm_dp_sideband_parse_reply()then a lot more fun.Quoted 139 lines
drivers/gpu/drm/drm_dp_mst_topology.c +7
-0+ +return 0 ; +} +EXPORT_SYMBOL_FOR_TESTS_ONLY (drm_dp_decode_sideband_req); + +void +drm_dp_dump_sideband_msg_req_body(const struct
drm_dp_sideband_msg_req_body*req,drivers/gpu/drm/drm_dp_mst_topology.c +1
-0+ int indent, struct drm_printer*printer)
drivers/gpu/drm/drm_dp_mst_topology.c +4
-0+{ +int i; + +#define P (f, ...) drm_printf_indent(printer, indent, f,##__VA_ARGS__)
drivers/gpu/drm/drm_dp_mst_topology.c +3
-0+if (req->req_type ==DP_LINK_ADDRESS ) { +/* No contents to print */ + P("type=%s\n", drm_dp_mst_req_type_str(req-Quoted 1 line
req_type));drivers/gpu/drm/drm_dp_mst_topology.c +4
-0+return ; + } + + P("type=%s contents:\n", drm_dp_mst_req_type_str(req-Quoted 1 line
req_type));drivers/gpu/drm/drm_dp_mst_topology.c +12
-0+ indent++; + +switch (req->req_type) { +case DP_ENUM_PATH_RESOURCES : +case DP_POWER_DOWN_PHY : +case DP_POWER_UP_PHY : +P ("port=%d\n" , req->u.port_num.port_number); +break ; +case DP_ALLOCATE_PAYLOAD : + P("port=%d vcpi=%d pbn=%d sdp_streams=%d %*ph\n", + req->u.allocate_payload.port_number, + req->u.allocate_payload.vcpi, req-Quoted 1 line
u.allocate_payload.pbn,
drivers/gpu/drm/drm_dp_mst_topology.c +12
-0+ req->u.allocate_payload.number_sdp_streams, + req->u.allocate_payload.number_sdp_streams, + req->u.allocate_payload.sdp_stream_sink); +break ; +case DP_QUERY_PAYLOAD : +P ("port=%d vcpi=%d\n" , + req->u.query_payload.port_number, + req->u.query_payload.vcpi); +break ; +case DP_REMOTE_DPCD_READ : + P("port=%d dpcd_addr=%05x len=%d\n", + req->u.dpcd_read.port_number, req-Quoted 1 line
u.dpcd_read.dpcd_address,
drivers/gpu/drm/drm_dp_mst_topology.c +7
-0+ req->u.dpcd_read.num_bytes); +break ; +case DP_REMOTE_DPCD_WRITE : + P("port=%d addr=%05x len=%d: %*ph\n", + req->u.dpcd_write.port_number, + req->u.dpcd_write.dpcd_address, + req->u.dpcd_write.num_bytes, req-Quoted 1 line
u.dpcd_write.num_bytes,
drivers/gpu/drm/drm_dp_mst_topology.c +11
-0+ req->u.dpcd_write.bytes); +break ; +case DP_REMOTE_I2C_READ : +P ("port=%d num_tx=%d id=%d size=%d:\n" , + req->u.i2c_read.port_number, + req->u.i2c_read.num_transactions, + req->u.i2c_read.read_i2c_device_id, + req->u.i2c_read.num_bytes_read); + + indent++; +for (i =0 ; i < req->u.i2c_read.num_transactions; i++){
drivers/gpu/drm/drm_dp_mst_topology.c +4
-0+const struct drm_dp_remote_i2c_read_tx *rtx = + &req->u.i2c_read.transactions[i]; + + P("%d: id=%03d size=%03d no_stop_bit=%dtx_delay=%03d:
%*ph\n",drivers/gpu/drm/drm_dp_mst_topology.c +2
-0+ i, rtx->i2c_dev_id, rtx->num_bytes, + rtx->no_stop_bit, rtx-Quoted 1 line
i2c_transaction_delay,drivers/gpu/drm/drm_dp_mst_topology.c +8
-0+ rtx->num_bytes, rtx->bytes); + } +break ; +case DP_REMOTE_I2C_WRITE : + P("port=%d id=%d size=%d: %*ph\n", + req->u.i2c_write.port_number, + req->u.i2c_write.write_i2c_device_id, + req->u.i2c_write.num_bytes, req-Quoted 1 line
u.i2c_write.num_bytes,
drivers/gpu/drm/drm_dp_mst_topology.c +13
-0+ req->u.i2c_write.bytes); +break ; +default : +P ("???\n" ); +break ; + } +#undef P +} +EXPORT_SYMBOL_FOR_TESTS_ONLY (drm_dp_dump_sideband_msg_req_body); + +static inline void +drm_dp_mst_dump_sideband_msg_tx (struct drm_printer *p, + const struct drm_dp_sideband_msg_tx*txmsg)
drivers/gpu/drm/drm_dp_mst_topology.c +9
-0+{ +struct drm_dp_sideband_msg_req_body req; +char buf[64 ]; +int ret; +int i; + +drm_dp_mst_rad_to_str (txmsg->dst->rad, txmsg->dst->lct, buf, +sizeof (buf)); + drm_printf(p, "txmsg cur_offset=%x cur_len=%x seqno=%xstate=%s
path_msg=%d dst=%s\n",drivers/gpu/drm/drm_dp_mst_topology.c +7
-0+ txmsg->cur_offset, txmsg->cur_len, txmsg->seqno, +drm_dp_mst_sideband_tx_state_str (txmsg->state), + txmsg->path_msg, buf); + + ret =drm_dp_decode_sideband_req (txmsg, &req); +if (ret) { + drm_printf(p, "<failed to decode sideband req: %d>\n",ret);
drivers/gpu/drm/drm_dp_mst_topology.c +17
-0+return ; + } +drm_dp_dump_sideband_msg_req_body (&req,1 , p); + +switch (req.req_type) { +case DP_REMOTE_DPCD_WRITE : +kfree (req.u.dpcd_write.bytes); +break ; +case DP_REMOTE_I2C_READ : +for (i =0 ; i < req.u.i2c_read.num_transactions; i++) +kfree (req.u.i2c_read.transactions[i].bytes); +break ; +case DP_REMOTE_I2C_WRITE : +kfree (req.u.i2c_write.bytes); +break ; + } +}Quoted 11 lines
Quoted 11 lines
Quoted 11 lines
drivers/gpu/drm/drm_dp_mst_topology.c static int drm_dp_mst_wait_tx_reply(struct · line 1183
Quoted 9 lines
static void
drm_dp_crc_sideband_chunk_req(u8 *msg, u8 len){drivers/gpu/drm/drm_dp_mst_topology.c +1
-0@@ -893,6 +1178,11 @@ static int drm_dp_mst_wait_tx_reply(struct drm_dp_mst_branch *mstb, } } out: +if (unlikely (ret == -EIO && drm_debug & (DRM_UT_DP |DRM_UT_KMS ))) {Hm I'd only check for
DRM_UT_DPhere.Was going to go ahead with this, but realized it might not be a great idea. One of the reasons I went with
DRM_UT_DP|DRM_UT_KMShere is because generally, I'd assume that (this is the case for me at least)DRM_UT_DPserves as a way to trace what's going on with DP, andDRM_UT_KMSis for non-atomic specific modesetting debug info. That being said, if we fail sending a message likePAYLOAD_ALLOCATEthen the modeset itself is going to not work correctly and thus - the contents of failure would still be relevant to someone working on modesetting issues even if they're not interested in the entirity of what's going on between the sink and the source.
Quoted 6 lines
That, and we were already printing when messages timed out under
DRM_UT_KMS. Granted though, that printing was added well before I introducedDRM_UT_DP. However, I suppose I could just move that back to usingDRM_DEBUG_KMS()and dump the sideband message seperately.Maybe I'm overthinking this a bit, thoughts?
Yes. Or at least, the drm debugging stuff we have is a mess already, let's try to not make it even more fun. I think consistently using DP for this stuff would make a lot more sense (and for general debugging I guess everyone just gets to add the DP bit to their drm.debug option). -Daniel
Quoted 406 lines
Quoted 402 lines
Quoted 138 lines
Quoted 129 lines
Quoted 10 lines
drivers/gpu/drm/drm_dp_mst_topology.c +4
-0+struct drm_printer p =drm_debug_printer (DBG_PREFIX ); + +drm_dp_mst_dump_sideband_msg_tx (&p, txmsg); + }mutex_unlock (&mgr->qlock);drivers/gpu/drm/drm_dp_mst_topology.c +0
-0return ret;drivers/gpu/drm/drm_dp_mst_topology.c static int process_single_tx_qlock(struct · line 2221
Quoted 7 lines
drivers/gpu/drm/drm_dp_mst_topology.c +1
-2@@ -1927,8 +2217,11 @@ static int process_single_tx_qlock(struct drm_dp_mst_topology_mgr *mgr, idx += tosend +1 ; ret =drm_dp_send_sideband_msg (mgr, up, chunk, idx); -if (ret) { -DRM_DEBUG_KMS ("sideband msg failed to send\n" ); +if (ret &&unlikely (drm_debug & (DRM_UT_DP |DRM_UT_KMS ))) {Same.
Quoted 8 lines
drivers/gpu/drm/drm_dp_mst_topology.c +4
-0+struct drm_printer p =drm_debug_printer (DBG_PREFIX ); + +drm_printf (&p,"sideband msg failed to send\n" ); +drm_dp_mst_dump_sideband_msg_tx (&p, txmsg);return ret; }drivers/gpu/drm/drm_dp_mst_topology.c static void drm_dp_queue_down_tx(struct · line 2288
Quoted 6 lines
drivers/gpu/drm/drm_dp_mst_topology.c +2
-0@@ -1990,6 +2283,13 @@ static void drm_dp_queue_down_tx(struct drm_dp_mst_topology_mgr *mgr, {mutex_lock (&mgr->qlock);list_add_tail (&txmsg->next, &mgr->tx_msg_downq); + +if (unlikely (drm_debug &DRM_UT_DP )) {Like you do here.
Quoted 92 lines
drivers/gpu/drm/drm_dp_mst_topology.c +5
-0+struct drm_printer p =drm_debug_printer (DBG_PREFIX ); + +drm_dp_mst_dump_sideband_msg_tx (&p, txmsg); + } +if (list_is_singular (&mgr->tx_msg_downq))process_single_down_tx_qlock (mgr);drivers/gpu/drm/drm_dp_mst_topology.c +0
-0mutex_unlock (&mgr->qlock);drivers/gpu/drm/drm_dp_mst_topology_internal.h new file +3
-0@@ -0,0 +1,24 @@ +/* SPDX-License-Identifier: GPL-2.0-only +* +* Declarations for DP MST related functions which are only used in selftests
+ *
+ * Copyright © 2018 Red Hat
+ * Authors:
+ * Lyude Paul <[email protected]>
+ */drivers/gpu/drm/drm_dp_mst_topology_internal.h new file +8
-0+ +#ifndef _DRM_DP_MST_HELPER_INTERNAL_H_ +#define _DRM_DP_MST_HELPER_INTERNAL_H_ + +#include <drm/drm_dp_mst_helper.h> + +void +drm_dp_encode_sideband_req(const struct drm_dp_sideband_msg_req_body*req,
drivers/gpu/drm/drm_dp_mst_topology_internal.h new file +2
-0+ struct drm_dp_sideband_msg_tx *raw); +int drm_dp_decode_sideband_req(const struct drm_dp_sideband_msg_tx*raw,
drivers/gpu/drm/drm_dp_mst_topology_internal.h new file +1
-0+struct drm_dp_sideband_msg_req_body *req);
drivers/gpu/drm/drm_dp_mst_topology_internal.h new file +2
-0+void +drm_dp_dump_sideband_msg_req_body(const struct
drm_dp_sideband_msg_req_body*req,drivers/gpu/drm/drm_dp_mst_topology_internal.h new file +1
-0+ int indent, struct drm_printer*printer);
drivers/gpu/drm/drm_dp_mst_topology_internal.h new file +2
-0+ +#endif /* !_DRM_DP_MST_HELPER_INTERNAL_H_ */ drivers/gpu/drm/selftests/drm_modeset_selftests.h +0
-0@@ -33,3 +33,4 @@ selftest(damage_iter_damage_one_outside, igt_damage_iter_damage_one_outside) selftest(damage_iter_damage_src_moved, igt_damage_iter_damage_src_moved) selftest(damage_iter_damage_not_visible,
igt_damage_iter_damage_not_visible)
selftest(dp_mst_calc_pbn_mode, igt_dp_mst_calc_pbn_mode)
+selftest(dp_mst_sideband_msg_req_decode,igt_dp_mst_sideband_msg_req_decode)drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c +17
-0@@ -8,6 +8,7 @@#include <drm/drm_dp_mst_helper.h> #include <drm/drm_print.h> +#include "../drm_dp_mst_topology_internal.h" #include "test-drm_modeset_common.h" int igt_dp_mst_calc_pbn_mode (void *ignored)@@ -34,3 +35,214 @@ int igt_dp_mst_calc_pbn_mode(void *ignored)return 0 ; } + +static bool +sideband_msg_req_equal (const struct drm_dp_sideband_msg_req_body *in, +const struct drm_dp_sideband_msg_req_body *out) +{ +const struct drm_dp_remote_i2c_read_tx *txin, *txout; +int i; + +if (in->req_type != out->req_type) +return false; + +switch (in->req_type) { +case DP_ENUM_PATH_RESOURCES : +case DP_POWER_UP_PHY : +case DP_POWER_DOWN_PHY : +case DP_ALLOCATE_PAYLOAD :Quoted 8 lines
drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c int igt_dp_mst_calc_pbn_mode(void *ignored) · line 56
Quoted 3 lines
drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c +3
-0+case DP_QUERY_PAYLOAD : +case DP_REMOTE_DPCD_READ : +return memcmp (in, out,sizeof (*in)) ==0 ;Just memcmp the entire thing for everyrone (we assume kzalloc anyway), and
then only have the additional checks for the few cases we need it? Would slightly reduce the code.Quoted 8 lines
Just realized also - this wouldn't work: remember that
I2C_READ,DPCD_WRITE,
andI2C_WRITEcontain pointers which of course will be different even if the
contents are the same. Unfortunately I don't think there's really a good solution for this, unless we want to just remove the pointers and store things
within the down request struct itselfUh ... disappoint :-/
Then at least put that into a comment, and maybe make this the default:
handler?Quoted 251 lines
Quoted 247 lines
Quoted 124 lines
drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c +46
-0+ +case DP_REMOTE_I2C_READ : +#define IN in->u.i2c_read +#define OUT out->u.i2c_read +if (IN .num_bytes_read !=OUT .num_bytes_read || +IN .num_transactions !=OUT .num_transactions || +IN .port_number !=OUT .port_number || +IN .read_i2c_device_id !=OUT .read_i2c_device_id) +return false; + +for (i =0 ; i <IN .num_transactions; i++) { + txin = &IN .transactions[i]; + txout = &OUT .transactions[i]; + +if (txin->i2c_dev_id != txout->i2c_dev_id || + txin->no_stop_bit != txout->no_stop_bit || + txin->num_bytes != txout->num_bytes || + txin->i2c_transaction_delay != + txout->i2c_transaction_delay) +return false; + +if (memcmp (txin->bytes, txout->bytes, + txin->num_bytes) !=0 ) +return false; + } +break ; +#undef IN +#undef OUT + +case DP_REMOTE_DPCD_WRITE : +#define IN in->u.dpcd_write +#define OUT out->u.dpcd_write +if (IN .dpcd_address !=OUT .dpcd_address || +IN .num_bytes !=OUT .num_bytes || +IN .port_number !=OUT .port_number) +return false; + +return memcmp (IN .bytes,OUT .bytes,IN .num_bytes) ==0 ; +#undef IN +#undef OUT + +case DP_REMOTE_I2C_WRITE : +#define IN in->u.i2c_write +#define OUT out->u.i2c_write +if (IN .port_number !=OUT .port_number || + IN.write_i2c_device_id != OUT.write_i2c_device_id||
drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c +14
-0+ IN.num_bytes != OUT.num_bytes) +return false; + +return memcmp (IN .bytes,OUT .bytes,IN .num_bytes) ==0 ; +#undef IN +#undef OUT + } + +return true; +} + +static int +__sideband_msg_req_encode_decode (int line, + struct drm_dp_sideband_msg_req_body*in,
drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c +1
-0+struct drm_dp_sideband_msg_req_body *out)
drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c +10
-0+{ +struct drm_printer p =drm_err_printer (pr_fmt ()); +struct drm_dp_sideband_msg_tx txmsg; +int i, ret; +bool eq; + +drm_dp_encode_sideband_req (in, &txmsg); + ret =drm_dp_decode_sideband_req (&txmsg, out); +if (ret <0 ) { + pr_err(pr_fmt("Failed to decode sideband request @line %d:
%d\n"),drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c +7
-0+ line, ret); +return ret; + } + + eq =sideband_msg_req_equal (in, out); +if (!eq) { + pr_err(pr_fmt("Encode/decode @ line %d failed,expected:\n"),
drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c +35
-0+ line); +drm_dp_dump_sideband_msg_req_body (in,1 , &p); +pr_err (pr_fmt ("Got:\n" )); +drm_dp_dump_sideband_msg_req_body (out,1 , &p); + } + +switch (in->req_type) { +case DP_REMOTE_DPCD_WRITE : +kfree (out->u.dpcd_write.bytes); +break ; +case DP_REMOTE_I2C_READ : +for (i =0 ; i < out->u.i2c_read.num_transactions; i++) +kfree (out->u.i2c_read.transactions[i].bytes); +break ; +case DP_REMOTE_I2C_WRITE : +kfree (out->u.i2c_write.bytes); +break ; + } + +/* Clear everything but the req_type for the input */ +memset (&in->u,0 ,sizeof (in->u)); + +/* Clear the output entirely */ +memset (out,0 ,sizeof (*out)); + +return eq ?0 : -EINVAL ; +} + +int igt_dp_mst_sideband_msg_req_decode (void *unused) +{ +struct drm_dp_sideband_msg_req_body in = {0 }, out = {0 }; +u8 data[] = {0xff ,0x0 ,0xdd }; +int i; + +#define
DO_TEST()\drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c +1
-0+ do{ \
drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c +1
-0+ if (__sideband_msg_req_encode_decode(__LINE__, &in,&out)) \
drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c
Quoted 3 lines
drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c +2
-1+ return -EINVAL ; \ + }while (0 )I had a tiny wtf moment here until I realized this is a macro ... maybe put the do on the first line and indent everything? Pure bikeshed to help
the blind ...Quoted 68 lines
drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c +58
-0+ + in.req_type =DP_ENUM_PATH_RESOURCES ; + in.u.port_num.port_number =5 ; +DO_TEST (); + + in.req_type =DP_POWER_UP_PHY ; + in.u.port_num.port_number =5 ; +DO_TEST (); + + in.req_type =DP_POWER_DOWN_PHY ; + in.u.port_num.port_number =5 ; +DO_TEST (); + + in.req_type =DP_ALLOCATE_PAYLOAD ; + in.u.allocate_payload.number_sdp_streams =3 ; +for (i =0 ; i < in.u.allocate_payload.number_sdp_streams; i++) + in.u.allocate_payload.sdp_stream_sink[i] = i +1 ; +DO_TEST (); + in.u.allocate_payload.port_number =0xf ; +DO_TEST (); + in.u.allocate_payload.vcpi =0x7f ; +DO_TEST (); + in.u.allocate_payload.pbn =U16_MAX ; +DO_TEST (); + + in.req_type =DP_QUERY_PAYLOAD ; + in.u.query_payload.port_number =0xf ; +DO_TEST (); + in.u.query_payload.vcpi =0x7f ; +DO_TEST (); + + in.req_type =DP_REMOTE_DPCD_READ ; + in.u.dpcd_read.port_number =0xf ; +DO_TEST (); + in.u.dpcd_read.dpcd_address =0xfedcb ; +DO_TEST (); + in.u.dpcd_read.num_bytes =U8_MAX ; +DO_TEST (); + + in.req_type =DP_REMOTE_DPCD_WRITE ; + in.u.dpcd_write.port_number =0xf ; +DO_TEST (); + in.u.dpcd_write.dpcd_address =0xfedcb ; +DO_TEST (); + in.u.dpcd_write.num_bytes =ARRAY_SIZE (data); + in.u.dpcd_write.bytes = data; +DO_TEST (); + + in.req_type =DP_REMOTE_I2C_READ ; + in.u.i2c_read.port_number =0xf ; +DO_TEST (); + in.u.i2c_read.read_i2c_device_id =0x7f ; +DO_TEST (); + in.u.i2c_read.num_transactions =3 ; + in.u.i2c_read.num_bytes_read =ARRAY_SIZE (data) *3 ; +for (i =0 ; i < in.u.i2c_read.num_transactions; i++) { + in.u.i2c_read.transactions[i].bytes = data; + in.u.i2c_read.transactions[i].num_bytes =
ARRAY_SIZE(data);drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c +2
-0+ in.u.i2c_read.transactions[i].i2c_dev_id =0x7f & ~i; + in.u.i2c_read.transactions[i].i2c_transaction_delay =
0xf&
~i;drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c +5
-0+ } +DO_TEST (); + + in.req_type =DP_REMOTE_I2C_WRITE ; + in.u.i2c_write.port_number =0xf ;drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c int igt_dp_mst_calc_pbn_mode(void *ignored) · line 248
Quoted 10 lines
drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c +10
-0+DO_TEST (); + in.u.i2c_write.write_i2c_device_id =0x7f ; +DO_TEST (); + in.u.i2c_write.num_bytes =ARRAY_SIZE (data); + in.u.i2c_write.bytes = data; +DO_TEST (); + +#undef DO_TEST +return 0 ; +}Extremely nice, more unit tests ftw!
With the nits somehow figured out: Reviewed-by: Daniel Vetter < [email protected]>
Quoted 30 lines
drivers/gpu/drm/selftests/test-drm_modeset_common.h +1
-0@@ -40,5 +40,6 @@ int igt_damage_iter_damage_one_outside(void *ignored);int igt_damage_iter_damage_src_moved (void *ignored);int igt_damage_iter_damage_not_visible (void *ignored);int igt_dp_mst_calc_pbn_mode (void *ignored); +int igt_dp_mst_sideband_msg_req_decode (void *ignored);#endif
include/drm/drm_dp_mst_helper.h +1
-1@@ -293,7 +293,7 @@ struct drm_dp_remote_dpcd_write {struct drm_dp_remote_i2c_read {u8 num_transactions;u8 port_number; -struct { +struct drm_dp_remote_i2c_read_tx {u8 i2c_dev_id;u8 num_bytes;u8 *bytes;Signature · 1 line
2.21.0Signature · 2 lines
Cheers, Lyude PaulSignature · 2 lines
Cheers, Lyude Paul
Signature · 3 lines
Daniel Vetter Software Engineer, Intel Corporation http://blog.ffwll.ch
Lyude Paul
maintainerPATCH 6/26 drm/dp_mst: Move PDT teardown for ports into destroy_connector_work
In reply to Lyude PaulThis will allow us to add some locking for port PDTs, which can't be done from drm_dp_destroy_port() since we don't know what locks the caller might be holding. Also, this gets rid of a good bit of unneeded code.
--- drivers/gpu/drm/drm_dp_mst_topology.c | 42 +++++++++++---------------- 1 file changed, 17 insertions(+), 25 deletions(-)
drivers/gpu/drm/drm_dp_mst_topology.c +17 -25
@@ -1509,31 +1509,22 @@ static void drm_dp_destroy_port(struct kref *kref) container_of(kref, struct drm_dp_mst_port, topology_kref);struct drm_dp_mst_topology_mgr *mgr = port->mgr; -if (!port->input) { -kfree (port->cached_edid); - -/* -* The only time we don't have a connector -* on an output port is if the connector init -* fails. -*/ -if (port->connector) { -/* we can't destroy the connector here, as -* we might be holding the mode_config.mutex -* from an EDID retrieval */ - -mutex_lock (&mgr->destroy_connector_lock); -list_add (&port->next, &mgr->destroy_connector_list); -mutex_unlock (&mgr->destroy_connector_lock); -schedule_work (&mgr->destroy_connector_work); -return ; - } -/* no need to clean up vcpi -* as if we have no connector we never setup a vcpi */ -drm_dp_port_teardown_pdt (port, port->pdt); - port->pdt =DP_PEER_DEVICE_NONE ; +/* There's nothing that needs locking to destroy an input port yet */ +if (port->input) { +drm_dp_mst_put_port_malloc (port); +return ; } -drm_dp_mst_put_port_malloc (port); + +kfree (port->cached_edid); + +/* +* we can't destroy the connector here, as we might be holding the +* mode_config.mutex from an EDID retrieval +*/ +mutex_lock (&mgr->destroy_connector_lock); +list_add (&port->next, &mgr->destroy_connector_list); +mutex_unlock (&mgr->destroy_connector_lock); +schedule_work (&mgr->destroy_connector_work); }/**
@@ -3881,7 +3872,8 @@ drm_dp_finish_destroy_port(struct drm_dp_mst_port *port) {INIT_LIST_HEAD (&port->next); - port->mgr->cbs->destroy_connector (port->mgr, port->connector); +if (port->connector) + port->mgr->cbs->destroy_connector (port->mgr, port->connector);drm_dp_port_teardown_pdt (port, port->pdt); port->pdt =DP_PEER_DEVICE_NONE ;
Signature · 1 line
2.21.0
Re: [PATCH 06/26] drm/dp_mst: Move PDT teardown for ports into destroy_connector_work
Quoted 52 lines · Lyude Paul
This will allow us to add some locking for port PDTs, which can't be done from
drm_dp_destroy_port()since we don't know what locks the caller might be holding. Also, this gets rid of a good bit of unneeded code.Cc Juston Li <[email protected]>Cc Imre Deak <[email protected]>Cc Ville Syrjälä <[email protected]>Cc Harry Wentland <[email protected]>Signed-off-by Lyude Paul <[email protected]>--- drivers/gpu/drm/drm_dp_mst_topology.c | 42 +++++++++++---------------- 1 file changed, 17 insertions(+), 25 deletions(-)drivers/gpu/drm/drm_dp_mst_topology.c +5
-24@@ -1509,31 +1509,22 @@ static void drm_dp_destroy_port(struct kref *kref) container_of(kref, struct drm_dp_mst_port, topology_kref);struct drm_dp_mst_topology_mgr *mgr = port->mgr; -if (!port->input) { -kfree (port->cached_edid); - -/* -* The only time we don't have a connector -* on an output port is if the connector init -* fails. -*/ -if (port->connector) { -/* we can't destroy the connector here, as -* we might be holding the mode_config.mutex -* from an EDID retrieval */ - -mutex_lock (&mgr->destroy_connector_lock); -list_add (&port->next, &mgr->destroy_connector_list); -mutex_unlock (&mgr->destroy_connector_lock); -schedule_work (&mgr->destroy_connector_work); -return ; - } -/* no need to clean up vcpi -* as if we have no connector we never setup a vcpi */ -drm_dp_port_teardown_pdt (port, port->pdt); - port->pdt =DP_PEER_DEVICE_NONE ; +/* There's nothing that needs locking to destroy an input port yet */ +if (port->input) { +drm_dp_mst_put_port_malloc (port); +return ; } -drm_dp_mst_put_port_malloc (port); +
drivers/gpu/drm/drm_dp_mst_topology.c static void drm_dp_destroy_port(struct kref *kref) · line 1527
Quoted 10 lines
drivers/gpu/drm/drm_dp_mst_topology.c +10
-0+kfree (port->cached_edid); + +/* +* we can't destroy the connector here, as we might be holding the +* mode_config.mutex from an EDID retrieval +*/ +mutex_lock (&mgr->destroy_connector_lock); +list_add (&port->next, &mgr->destroy_connector_list); +mutex_unlock (&mgr->destroy_connector_lock); +schedule_work (&mgr->destroy_connector_work);
So if I'm not completely blind this just flattens the above code flow (by inverting the if (port->input)).
drivers/gpu/drm/drm_dp_mst_topology.c drm_dp_finish_destroy_port(struct drm_dp_mst_port *port) · line 3875
Quoted 9 lines
drivers/gpu/drm/drm_dp_mst_topology.c +1
-1}/** @@ -3881,7 +3872,8 @@ drm_dp_finish_destroy_port(struct drm_dp_mst_port *port) {INIT_LIST_HEAD (&port->next); - port->mgr->cbs->destroy_connector (port->mgr, port->connector); +if (port->connector)
And this here I can't connect with the commit message. I'm confused, did something go wrong with some rebase here, and this patch should have a different title/summary?
-Daniel
Quoted 7 lines
drivers/gpu/drm/drm_dp_mst_topology.c +1
-0+ port->mgr->cbs->destroy_connector (port->mgr, port->connector);drm_dp_port_teardown_pdt (port, port->pdt); port->pdt =DP_PEER_DEVICE_NONE ;Signature · 1 line
2.21.0
Signature · 7 lines
Daniel Vetter Software Engineer, Intel Corporation http://blog.ffwll.ch _______________________________________________ dri-devel mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/dri-devel
Lyude Paul
maintainerRe: [PATCH 06/26] drm/dp_mst: Move PDT teardown for ports into destroy_connector_work
Quoted 59 lines · Daniel Vetter
Quoted 58 lines · Lyude Paul
This will allow us to add some locking for port PDTs, which can't be done from
drm_dp_destroy_port()since we don't know what locks the caller might be holding. Also, this gets rid of a good bit of unneeded code.Cc Juston Li <[email protected]>Cc Imre Deak <[email protected]>Cc Ville Syrjälä <[email protected]>Cc Harry Wentland <[email protected]>Signed-off-by Lyude Paul <[email protected]>--- drivers/gpu/drm/drm_dp_mst_topology.c | 42 +++++++++++---------------- 1 file changed, 17 insertions(+), 25 deletions(-)drivers/gpu/drm/drm_dp_mst_topology.c +10
-24@@ -1509,31 +1509,22 @@ static void drm_dp_destroy_port(struct kref *kref) container_of(kref, struct drm_dp_mst_port, topology_kref);struct drm_dp_mst_topology_mgr *mgr = port->mgr; -if (!port->input) { -kfree (port->cached_edid); - -/* -* The only time we don't have a connector -* on an output port is if the connector init -* fails. -*/ -if (port->connector) { -/* we can't destroy the connector here, as -* we might be holding the mode_config.mutex -* from an EDID retrieval */ - -mutex_lock (&mgr->destroy_connector_lock); -list_add (&port->next, &mgr->destroy_connector_list); -mutex_unlock (&mgr->destroy_connector_lock); -schedule_work (&mgr->destroy_connector_work); -return ; - } -/* no need to clean up vcpi -* as if we have no connector we never setup a vcpi */ -drm_dp_port_teardown_pdt (port, port->pdt); - port->pdt =DP_PEER_DEVICE_NONE ; +/* There's nothing that needs locking to destroy an input port yet */ +if (port->input) { +drm_dp_mst_put_port_malloc (port); +return ; } -drm_dp_mst_put_port_malloc (port); + +kfree (port->cached_edid); + +/* +* we can't destroy the connector here, as we might be holding the +* mode_config.mutex from an EDID retrieval
Quoted 8 lines
drivers/gpu/drm/drm_dp_mst_topology.c static void drm_dp_destroy_port(struct kref *kref) · line 1527
Quoted 5 lines
drivers/gpu/drm/drm_dp_mst_topology.c +5
-0+*/ +mutex_lock (&mgr->destroy_connector_lock); +list_add (&port->next, &mgr->destroy_connector_list); +mutex_unlock (&mgr->destroy_connector_lock); +schedule_work (&mgr->destroy_connector_work);So if I'm not completely blind this just flattens the above code flow (by inverting the if (
port->input)).
Now I'm really remembering why I refactored this. The control flow on the previous version of this is pretty misleading. To summarize so it's a bit more obvious:
if (port->input) {
drm_dp_mst_put_port_malloc(port);
return;
} else if (port->connector) {
add_connector_to_destroy_list();
return;
/* ^ now, this is where PDT teardown happens */
} else {
drm_dp_port_teardown_pdt(port, port->pdt);
}/* free edid etc etc */
So, I suppose the title of this patch would be more accurate if it was "drm/dp_mst: Remove PDT teardown in destroy_port() and refactor" I'll go ahead and change that
Quoted 9 lines
Quoted 9 lines
drivers/gpu/drm/drm_dp_mst_topology.c +0
-0}/** @@ -3881,7 +3872,8 @@ drm_dp_finish_destroy_port(struct drm_dp_mst_port *port) {INIT_LIST_HEAD (&port->next);
Quoted 7 lines
drivers/gpu/drm/drm_dp_mst_topology.c drm_dp_finish_destroy_port(struct drm_dp_mst_port *port) · line 3875
Quoted 2 lines
drivers/gpu/drm/drm_dp_mst_topology.c +1
-1- port->mgr->cbs->destroy_connector (port->mgr, port->connector); +if (port->connector)And this here I can't connect with the commit message. I'm confused, did something go wrong with some rebase here, and this patch should have a different title/summary?
-Daniel
No, this is correct. In the previous drm_dp_destroy_port() function we only added a port to the delayed destroy work if it had a connector on it. Now that we add ports unconditionally, we have to check port->connector before trying to call port->mgr->cbs->destroy_connector() since port->connector is no longer guaranteed to be != NULL.
Quoted 8 lines
Quoted 8 lines
drivers/gpu/drm/drm_dp_mst_topology.c +1
-0+ port->mgr->cbs->destroy_connector (port->mgr, port->connector);drm_dp_port_teardown_pdt (port, port->pdt);
port->pdt=DP_PEER_DEVICE_NONE;Signature · 1 line
2.21.0
Signature · 2 lines
Cheers, Lyude Paul
Lyude Paul
maintainerPATCH 7/26 drm/dp_mst: Get rid of list clear in drm_dp_finish_destroy_port()
In reply to Lyude PaulThis seems to be some leftover detritus from before the port/mstb kref cleanup and doesn't do anything anymore, so get rid of it.
--- drivers/gpu/drm/drm_dp_mst_topology.c | 2 -- 1 file changed, 2 deletions(-)
drivers/gpu/drm/drm_dp_mst_topology.c +0 -2
@@ -3870,8 +3870,6 @@ static void drm_dp_tx_work(struct work_struct *work)static inline void drm_dp_finish_destroy_port (struct drm_dp_mst_port *port) { -INIT_LIST_HEAD (&port->next); -if (port->connector) port->mgr->cbs->destroy_connector (port->mgr, port->connector);
Signature · 1 line
2.21.0
Re: [PATCH 07/26] drm/dp_mst: Get rid of list clear in drm_dp_finish_destroy_port()
Quoted 8 lines · Lyude Paul
This seems to be some leftover detritus from before the port/mstb kref cleanup and doesn't do anything anymore, so get rid of it.
Cc Juston Li <[email protected]>Cc Imre Deak <[email protected]>Cc Ville Syrjälä <[email protected]>Cc Harry Wentland <[email protected]>Signed-off-by Lyude Paul <[email protected]>
Maybe move this earlier, before patch 2. Either way:
Quoted 20 lines
--- drivers/gpu/drm/drm_dp_mst_topology.c | 2 -- 1 file changed, 2 deletions(-)drivers/gpu/drm/drm_dp_mst_topology.c +0
-2@@ -3870,8 +3870,6 @@ static void drm_dp_tx_work(struct work_struct *work)static inline void drm_dp_finish_destroy_port (struct drm_dp_mst_port *port) { -INIT_LIST_HEAD (&port->next); -if (port->connector) port->mgr->cbs->destroy_connector (port->mgr, port->connector);Signature · 1 line
2.21.0
Signature · 3 lines
Daniel Vetter Software Engineer, Intel Corporation http://blog.ffwll.ch
Lyude Paul
maintainerPATCH 8/26 drm/dp_mst: Refactor drm_dp_send_enum_path_resources
In reply to Lyude PaulUse more pointers so we don't have to write out txmsg->reply.u.path_resources each time. Also, fix line wrapping + rearrange local variables.
--- drivers/gpu/drm/drm_dp_mst_topology.c | 24 ++++++++++++++++-------- 1 file changed, 16 insertions(+), 8 deletions(-)
drivers/gpu/drm/drm_dp_mst_topology.c +16 -8
@@ -2350,12 +2350,14 @@ static void drm_dp_send_link_address(struct drm_dp_mst_topology_mgr *mgr,kfree (txmsg); } -static int drm_dp_send_enum_path_resources (struct drm_dp_mst_topology_mgr *mgr, -struct drm_dp_mst_branch *mstb, -struct drm_dp_mst_port *port) +static int +drm_dp_send_enum_path_resources (struct drm_dp_mst_topology_mgr *mgr, +struct drm_dp_mst_branch *mstb, +struct drm_dp_mst_port *port) { -int len; +struct drm_dp_enum_path_resources_ack_reply *path_res;struct drm_dp_sideband_msg_tx *txmsg; +int len;int ret; txmsg =kzalloc (sizeof (*txmsg),GFP_KERNEL );
@@ -2369,14 +2371,20 @@ static int drm_dp_send_enum_path_resources(struct drm_dp_mst_topology_mgr *mgr, ret =drm_dp_mst_wait_tx_reply (mstb, txmsg);if (ret >0 ) { + path_res = &txmsg->reply.u.path_resources; +if (txmsg->reply.reply_type ==DP_SIDEBAND_REPLY_NAK ) {DRM_DEBUG_KMS ("enum path resources nak received\n" ); }else { -if (port->port_num != txmsg->reply.u.path_resources.port_number) +if (port->port_num != path_res->port_number)DRM_ERROR ("got incorrect port in response\n" ); -DRM_DEBUG_KMS ("enum path resources %d: %d %d\n" , txmsg->reply.u.path_resources.port_number, txmsg->reply.u.path_resources.full_payload_bw_number, - txmsg->reply.u.path_resources.avail_payload_bw_number); - port->available_pbn = txmsg->reply.u.path_resources.avail_payload_bw_number; + +DRM_DEBUG_KMS ("enum path resources %d: %d %d\n" , + path_res->port_number, + path_res->full_payload_bw_number, + path_res->avail_payload_bw_number); + port->available_pbn = + path_res->avail_payload_bw_number; } }
Signature · 6 lines
2.21.0 _______________________________________________ dri-devel mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/dri-devel
Re: [PATCH 08/26] drm/dp_mst: Refactor drm_dp_send_enum_path_resources
Quoted 9 lines · Lyude Paul
Use more pointers so we don't have to write out
txmsg->reply.u.path_resourceseach time. Also, fix line wrapping + rearrange local variables.Cc Juston Li <[email protected]>Cc Imre Deak <[email protected]>Cc Ville Syrjälä <[email protected]>Cc Harry Wentland <[email protected]>Signed-off-by Lyude Paul <[email protected]>
Quoted 55 lines
--- drivers/gpu/drm/drm_dp_mst_topology.c | 24 ++++++++++++++++-------- 1 file changed, 16 insertions(+), 8 deletions(-)drivers/gpu/drm/drm_dp_mst_topology.c +16
-8@@ -2350,12 +2350,14 @@ static void drm_dp_send_link_address(struct drm_dp_mst_topology_mgr *mgr,kfree (txmsg); } -static int drm_dp_send_enum_path_resources (struct drm_dp_mst_topology_mgr *mgr, -struct drm_dp_mst_branch *mstb, -struct drm_dp_mst_port *port) +static int +drm_dp_send_enum_path_resources (struct drm_dp_mst_topology_mgr *mgr, +struct drm_dp_mst_branch *mstb, +struct drm_dp_mst_port *port) { -int len; +struct drm_dp_enum_path_resources_ack_reply *path_res;struct drm_dp_sideband_msg_tx *txmsg; +int len;int ret; txmsg =kzalloc (sizeof (*txmsg),GFP_KERNEL );@@ -2369,14 +2371,20 @@ static int drm_dp_send_enum_path_resources(struct drm_dp_mst_topology_mgr *mgr, ret =drm_dp_mst_wait_tx_reply (mstb, txmsg);if (ret >0 ) { + path_res = &txmsg->reply.u.path_resources; +if (txmsg->reply.reply_type ==DP_SIDEBAND_REPLY_NAK ) {DRM_DEBUG_KMS ("enum path resources nak received\n" ); }else { -if (port->port_num != txmsg->reply.u.path_resources.port_number) +if (port->port_num != path_res->port_number)DRM_ERROR ("got incorrect port in response\n" ); -DRM_DEBUG_KMS ("enum path resources %d: %d %d\n" , txmsg->reply.u.path_resources.port_number, txmsg->reply.u.path_resources.full_payload_bw_number, - txmsg->reply.u.path_resources.avail_payload_bw_number); - port->available_pbn = txmsg->reply.u.path_resources.avail_payload_bw_number; + +DRM_DEBUG_KMS ("enum path resources %d: %d %d\n" , + path_res->port_number, + path_res->full_payload_bw_number, + path_res->avail_payload_bw_number); + port->available_pbn = + path_res->avail_payload_bw_number; } }Signature · 1 line
2.21.0
Signature · 3 lines
Daniel Vetter Software Engineer, Intel Corporation http://blog.ffwll.ch
Lyude Paul
maintainerPATCH 9/26 drm/dp_mst: Remove huge conditional in drm_dp_mst_handle_up_req()
In reply to Lyude PaulWhich reduces indentation and makes this function more legible.
--- drivers/gpu/drm/drm_dp_mst_topology.c | 90 +++++++++++++-------------- 1 file changed, 45 insertions(+), 45 deletions(-)
drivers/gpu/drm/drm_dp_mst_topology.c +45 -45
@@ -3126,7 +3126,9 @@ static int drm_dp_mst_handle_down_rep(struct drm_dp_mst_topology_mgr *mgr)static int drm_dp_mst_handle_up_req (struct drm_dp_mst_topology_mgr *mgr) { -int ret =0 ; +struct drm_dp_sideband_msg_req_body msg; +struct drm_dp_mst_branch *mstb =NULL ; +bool seqno;if (!drm_dp_get_one_sb_msg (mgr, true)) { memset(&mgr->up_req_recv, 0,
@@ -3134,62 +3136,60 @@ static int drm_dp_mst_handle_up_req(struct drm_dp_mst_topology_mgr *mgr)return 0 ; } -if (mgr->up_req_recv.have_eomt) { -struct drm_dp_sideband_msg_req_body msg; -struct drm_dp_mst_branch *mstb =NULL ; -bool seqno; - -if (!mgr->up_req_recv.initial_hdr.broadcast) { - mstb =drm_dp_get_mst_branch_device (mgr, - mgr->up_req_recv.initial_hdr.lct, - mgr->up_req_recv.initial_hdr.rad); -if (!mstb) { -DRM_DEBUG_KMS ("Got MST reply from unknown device %d\n" , mgr->up_req_recv.initial_hdr.lct); -memset (&mgr->up_req_recv,0 ,sizeof (struct drm_dp_sideband_msg_rx )); -return 0 ; - } - } +if (!mgr->up_req_recv.have_eomt) +return 0 ; - seqno = mgr->up_req_recv.initial_hdr.seqno; -drm_dp_sideband_parse_req (&mgr->up_req_recv, &msg); +if (!mgr->up_req_recv.initial_hdr.broadcast) { + mstb =drm_dp_get_mst_branch_device (mgr, + mgr->up_req_recv.initial_hdr.lct, + mgr->up_req_recv.initial_hdr.rad); +if (!mstb) { +DRM_DEBUG_KMS ("Got MST reply from unknown device %d\n" , mgr->up_req_recv.initial_hdr.lct); +memset (&mgr->up_req_recv,0 ,sizeof (struct drm_dp_sideband_msg_rx )); +return 0 ; + } + } -if (msg.req_type ==DP_CONNECTION_STATUS_NOTIFY ) { -drm_dp_send_up_ack_reply (mgr, mgr->mst_primary, msg.req_type, seqno, false); + seqno = mgr->up_req_recv.initial_hdr.seqno; +drm_dp_sideband_parse_req (&mgr->up_req_recv, &msg); -if (!mstb) - mstb =drm_dp_get_mst_branch_device_by_guid (mgr, msg.u.conn_stat.guid); +if (msg.req_type ==DP_CONNECTION_STATUS_NOTIFY ) { +drm_dp_send_up_ack_reply (mgr, mgr->mst_primary, msg.req_type, seqno, false); -if (!mstb) { -DRM_DEBUG_KMS ("Got MST reply from unknown device %d\n" , mgr->up_req_recv.initial_hdr.lct); -memset (&mgr->up_req_recv,0 ,sizeof (struct drm_dp_sideband_msg_rx )); -return 0 ; - } +if (!mstb) + mstb =drm_dp_get_mst_branch_device_by_guid (mgr, msg.u.conn_stat.guid); -drm_dp_update_port (mstb, &msg.u.conn_stat); +if (!mstb) { +DRM_DEBUG_KMS ("Got MST reply from unknown device %d\n" , mgr->up_req_recv.initial_hdr.lct); +memset (&mgr->up_req_recv,0 ,sizeof (struct drm_dp_sideband_msg_rx )); +return 0 ; + } -DRM_DEBUG_KMS ("Got CSN: pn: %d ldps:%d ddps: %d mcs: %d ip: %d pdt: %d\n" , msg.u.conn_stat.port_number, msg.u.conn_stat.legacy_device_plug_status, msg.u.conn_stat.displayport_device_plug_status, msg.u.conn_stat.message_capability_status, msg.u.conn_stat.input_port, msg.u.conn_stat.peer_device_type); -drm_kms_helper_hotplug_event (mgr->dev); +drm_dp_update_port (mstb, &msg.u.conn_stat); - }else if (msg.req_type ==DP_RESOURCE_STATUS_NOTIFY ) { -drm_dp_send_up_ack_reply (mgr, mgr->mst_primary, msg.req_type, seqno, false); -if (!mstb) - mstb =drm_dp_get_mst_branch_device_by_guid (mgr, msg.u.resource_stat.guid); +DRM_DEBUG_KMS ("Got CSN: pn: %d ldps:%d ddps: %d mcs: %d ip: %d pdt: %d\n" , msg.u.conn_stat.port_number, msg.u.conn_stat.legacy_device_plug_status, msg.u.conn_stat.displayport_device_plug_status, msg.u.conn_stat.message_capability_status, msg.u.conn_stat.input_port, msg.u.conn_stat.peer_device_type); +drm_kms_helper_hotplug_event (mgr->dev); -if (!mstb) { -DRM_DEBUG_KMS ("Got MST reply from unknown device %d\n" , mgr->up_req_recv.initial_hdr.lct); -memset (&mgr->up_req_recv,0 ,sizeof (struct drm_dp_sideband_msg_rx )); -return 0 ; - } + }else if (msg.req_type ==DP_RESOURCE_STATUS_NOTIFY ) { +drm_dp_send_up_ack_reply (mgr, mgr->mst_primary, msg.req_type, seqno, false); +if (!mstb) + mstb =drm_dp_get_mst_branch_device_by_guid (mgr, msg.u.resource_stat.guid); -DRM_DEBUG_KMS ("Got RSN: pn: %d avail_pbn %d\n" , msg.u.resource_stat.port_number, msg.u.resource_stat.available_pbn); +if (!mstb) { +DRM_DEBUG_KMS ("Got MST reply from unknown device %d\n" , mgr->up_req_recv.initial_hdr.lct); +memset (&mgr->up_req_recv,0 ,sizeof (struct drm_dp_sideband_msg_rx )); +return 0 ; } -if (mstb) -drm_dp_mst_topology_put_mstb (mstb); - -memset (&mgr->up_req_recv,0 ,sizeof (struct drm_dp_sideband_msg_rx )); +DRM_DEBUG_KMS ("Got RSN: pn: %d avail_pbn %d\n" , msg.u.resource_stat.port_number, msg.u.resource_stat.available_pbn); } -return ret; + +if (mstb) +drm_dp_mst_topology_put_mstb (mstb); + +memset (&mgr->up_req_recv,0 ,sizeof (struct drm_dp_sideband_msg_rx )); + +return 0 ; }/**
Signature · 6 lines
2.21.0 _______________________________________________ dri-devel mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/dri-devel
Re: [PATCH 09/26] drm/dp_mst: Remove huge conditional in drm_dp_mst_handle_up_req()
Quoted 1 line · Lyude Paul
Which reduces indentation and makes this function more legible.
Indeed, nice one.
Quoted 133 lines
Cc Juston Li <[email protected]>Cc Imre Deak <[email protected]>Cc Ville Syrjälä <[email protected]>Cc Harry Wentland <[email protected]>Signed-off-by Lyude Paul <[email protected]>--- drivers/gpu/drm/drm_dp_mst_topology.c | 90 +++++++++++++-------------- 1 file changed, 45 insertions(+), 45 deletions(-)drivers/gpu/drm/drm_dp_mst_topology.c +45
-45@@ -3126,7 +3126,9 @@ static int drm_dp_mst_handle_down_rep(struct drm_dp_mst_topology_mgr *mgr)static int drm_dp_mst_handle_up_req (struct drm_dp_mst_topology_mgr *mgr) { -int ret =0 ; +struct drm_dp_sideband_msg_req_body msg; +struct drm_dp_mst_branch *mstb =NULL ; +bool seqno;if (!drm_dp_get_one_sb_msg (mgr, true)) { memset(&mgr->up_req_recv, 0,@@ -3134,62 +3136,60 @@ static int drm_dp_mst_handle_up_req(struct drm_dp_mst_topology_mgr *mgr)return 0 ; } -if (mgr->up_req_recv.have_eomt) { -struct drm_dp_sideband_msg_req_body msg; -struct drm_dp_mst_branch *mstb =NULL ; -bool seqno; - -if (!mgr->up_req_recv.initial_hdr.broadcast) { - mstb =drm_dp_get_mst_branch_device (mgr, - mgr->up_req_recv.initial_hdr.lct, - mgr->up_req_recv.initial_hdr.rad); -if (!mstb) { -DRM_DEBUG_KMS ("Got MST reply from unknown device %d\n" , mgr->up_req_recv.initial_hdr.lct); -memset (&mgr->up_req_recv,0 ,sizeof (struct drm_dp_sideband_msg_rx )); -return 0 ; - } - } +if (!mgr->up_req_recv.have_eomt) +return 0 ; - seqno = mgr->up_req_recv.initial_hdr.seqno; -drm_dp_sideband_parse_req (&mgr->up_req_recv, &msg); +if (!mgr->up_req_recv.initial_hdr.broadcast) { + mstb =drm_dp_get_mst_branch_device (mgr, + mgr->up_req_recv.initial_hdr.lct, + mgr->up_req_recv.initial_hdr.rad); +if (!mstb) { +DRM_DEBUG_KMS ("Got MST reply from unknown device %d\n" , mgr->up_req_recv.initial_hdr.lct); +memset (&mgr->up_req_recv,0 ,sizeof (struct drm_dp_sideband_msg_rx )); +return 0 ; + } + } -if (msg.req_type ==DP_CONNECTION_STATUS_NOTIFY ) { -drm_dp_send_up_ack_reply (mgr, mgr->mst_primary, msg.req_type, seqno, false); + seqno = mgr->up_req_recv.initial_hdr.seqno; +drm_dp_sideband_parse_req (&mgr->up_req_recv, &msg); -if (!mstb) - mstb =drm_dp_get_mst_branch_device_by_guid (mgr, msg.u.conn_stat.guid); +if (msg.req_type ==DP_CONNECTION_STATUS_NOTIFY ) { +drm_dp_send_up_ack_reply (mgr, mgr->mst_primary, msg.req_type, seqno, false); -if (!mstb) { -DRM_DEBUG_KMS ("Got MST reply from unknown device %d\n" , mgr->up_req_recv.initial_hdr.lct); -memset (&mgr->up_req_recv,0 ,sizeof (struct drm_dp_sideband_msg_rx )); -return 0 ; - } +if (!mstb) + mstb =drm_dp_get_mst_branch_device_by_guid (mgr, msg.u.conn_stat.guid); -drm_dp_update_port (mstb, &msg.u.conn_stat); +if (!mstb) { +DRM_DEBUG_KMS ("Got MST reply from unknown device %d\n" , mgr->up_req_recv.initial_hdr.lct); +memset (&mgr->up_req_recv,0 ,sizeof (struct drm_dp_sideband_msg_rx )); +return 0 ; + } -DRM_DEBUG_KMS ("Got CSN: pn: %d ldps:%d ddps: %d mcs: %d ip: %d pdt: %d\n" , msg.u.conn_stat.port_number, msg.u.conn_stat.legacy_device_plug_status, msg.u.conn_stat.displayport_device_plug_status, msg.u.conn_stat.message_capability_status, msg.u.conn_stat.input_port, msg.u.conn_stat.peer_device_type); -drm_kms_helper_hotplug_event (mgr->dev); +drm_dp_update_port (mstb, &msg.u.conn_stat); - }else if (msg.req_type ==DP_RESOURCE_STATUS_NOTIFY ) { -drm_dp_send_up_ack_reply (mgr, mgr->mst_primary, msg.req_type, seqno, false); -if (!mstb) - mstb =drm_dp_get_mst_branch_device_by_guid (mgr, msg.u.resource_stat.guid); +DRM_DEBUG_KMS ("Got CSN: pn: %d ldps:%d ddps: %d mcs: %d ip: %d pdt: %d\n" , msg.u.conn_stat.port_number, msg.u.conn_stat.legacy_device_plug_status, msg.u.conn_stat.displayport_device_plug_status, msg.u.conn_stat.message_capability_status, msg.u.conn_stat.input_port, msg.u.conn_stat.peer_device_type); +drm_kms_helper_hotplug_event (mgr->dev); -if (!mstb) { -DRM_DEBUG_KMS ("Got MST reply from unknown device %d\n" , mgr->up_req_recv.initial_hdr.lct); -memset (&mgr->up_req_recv,0 ,sizeof (struct drm_dp_sideband_msg_rx )); -return 0 ; - } + }else if (msg.req_type ==DP_RESOURCE_STATUS_NOTIFY ) { +drm_dp_send_up_ack_reply (mgr, mgr->mst_primary, msg.req_type, seqno, false); +if (!mstb) + mstb =drm_dp_get_mst_branch_device_by_guid (mgr, msg.u.resource_stat.guid); -DRM_DEBUG_KMS ("Got RSN: pn: %d avail_pbn %d\n" , msg.u.resource_stat.port_number, msg.u.resource_stat.available_pbn); +if (!mstb) { +DRM_DEBUG_KMS ("Got MST reply from unknown device %d\n" , mgr->up_req_recv.initial_hdr.lct); +memset (&mgr->up_req_recv,0 ,sizeof (struct drm_dp_sideband_msg_rx )); +return 0 ; } -if (mstb) -drm_dp_mst_topology_put_mstb (mstb); - -memset (&mgr->up_req_recv,0 ,sizeof (struct drm_dp_sideband_msg_rx )); +DRM_DEBUG_KMS ("Got RSN: pn: %d avail_pbn %d\n" , msg.u.resource_stat.port_number, msg.u.resource_stat.available_pbn); } -return ret; + +if (mstb) +drm_dp_mst_topology_put_mstb (mstb); + +memset (&mgr->up_req_recv,0 ,sizeof (struct drm_dp_sideband_msg_rx )); + +return 0 ; }/** Signature · 1 line
2.21.0
Signature · 7 lines
Daniel Vetter Software Engineer, Intel Corporation http://blog.ffwll.ch _______________________________________________ dri-devel mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/dri-devel
Lyude Paul
maintainerPATCH 10/26 drm/dp_mst: Constify guid in drm_dp_get_mst_branch_by_guid()
In reply to Lyude PaulAnd it's helper, we'll be using this in just a moment.
--- drivers/gpu/drm/drm_dp_mst_topology.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/gpu/drm/drm_dp_mst_topology.c +2 -2
@@ -1971,7 +1971,7 @@ static struct drm_dp_mst_branch *drm_dp_get_mst_branch_device(struct drm_dp_mst_static struct drm_dp_mst_branch *get_mst_branch_device_by_guid_helper (struct drm_dp_mst_branch *mstb, -uint8_t *guid) +const uint8_t *guid) {struct drm_dp_mst_branch *found_mstb;struct drm_dp_mst_port *port;
@@ -1995,7 +1995,7 @@ static struct drm_dp_mst_branch *get_mst_branch_device_by_guid_helper(static struct drm_dp_mst_branch *drm_dp_get_mst_branch_device_by_guid (struct drm_dp_mst_topology_mgr *mgr, -uint8_t *guid) +const uint8_t *guid) {struct drm_dp_mst_branch *mstb;int ret;
Signature · 1 line
2.21.0
Lyude Paul
maintainerPATCH 11/26 drm/dp_mst: Refactor drm_dp_mst_handle_up_req()
In reply to Lyude PaulThere's a couple of changes here, so to summarize:
* Remove the big ugly mgr->up_req_recv.have_eomt conditional to save on indenting
* Store &mgr->up_req_recv.initial_hdr in a variable so we don't keep going over 80 character long lines
* De-duplicate code for calling drm_dp_send_up_ack_reply() and getting the MSTB via it's GUID
* Remove all of the duplicate calls to memset() and just use a goto instead
* Actually do line wrapping
* Remove the unnecessary if (mstb) check before calling drm_dp_mst_topology_put_mstb() - we are guaranteed to always have mstb != NULL at that point in the function
--- drivers/gpu/drm/drm_dp_mst_topology.c | 75 ++++++++++++++------------- 1 file changed, 38 insertions(+), 37 deletions(-)
drivers/gpu/drm/drm_dp_mst_topology.c +38 -37
@@ -3127,68 +3127,69 @@ static int drm_dp_mst_handle_down_rep(struct drm_dp_mst_topology_mgr *mgr)static int drm_dp_mst_handle_up_req (struct drm_dp_mst_topology_mgr *mgr) {struct drm_dp_sideband_msg_req_body msg; +struct drm_dp_sideband_msg_hdr *hdr = &mgr->up_req_recv.initial_hdr;struct drm_dp_mst_branch *mstb =NULL ; +const u8 *guid;bool seqno; -if (!drm_dp_get_one_sb_msg (mgr, true)) { -memset (&mgr->up_req_recv,0 , -sizeof (struct drm_dp_sideband_msg_rx )); -return 0 ; - } +if (!drm_dp_get_one_sb_msg (mgr, true)) + goto out;if (!mgr->up_req_recv.have_eomt)return 0 ; -if (!mgr->up_req_recv.initial_hdr.broadcast) { - mstb =drm_dp_get_mst_branch_device (mgr, - mgr->up_req_recv.initial_hdr.lct, - mgr->up_req_recv.initial_hdr.rad); +if (!hdr->broadcast) { + mstb =drm_dp_get_mst_branch_device (mgr, hdr->lct, hdr->rad);if (!mstb) { -DRM_DEBUG_KMS ("Got MST reply from unknown device %d\n" , mgr->up_req_recv.initial_hdr.lct); -memset (&mgr->up_req_recv,0 ,sizeof (struct drm_dp_sideband_msg_rx )); -return 0 ; +DRM_DEBUG_KMS ("Got MST reply from unknown device %d\n" , + hdr->lct); + goto out; } } - seqno = mgr->up_req_recv.initial_hdr.seqno; + seqno = hdr->seqno;drm_dp_sideband_parse_req (&mgr->up_req_recv, &msg); -if (msg.req_type ==DP_CONNECTION_STATUS_NOTIFY ) { -drm_dp_send_up_ack_reply (mgr, mgr->mst_primary, msg.req_type, seqno, false); +if (msg.req_type ==DP_CONNECTION_STATUS_NOTIFY ) + guid = msg.u.conn_stat.guid; +else if (msg.req_type ==DP_RESOURCE_STATUS_NOTIFY ) + guid = msg.u.resource_stat.guid; +else + goto out; -if (!mstb) - mstb =drm_dp_get_mst_branch_device_by_guid (mgr, msg.u.conn_stat.guid); +drm_dp_send_up_ack_reply (mgr, mgr->mst_primary, msg.req_type, seqno, + false); +if (!mstb) { + mstb =drm_dp_get_mst_branch_device_by_guid (mgr, guid);if (!mstb) { -DRM_DEBUG_KMS ("Got MST reply from unknown device %d\n" , mgr->up_req_recv.initial_hdr.lct); -memset (&mgr->up_req_recv,0 ,sizeof (struct drm_dp_sideband_msg_rx )); -return 0 ; +DRM_DEBUG_KMS ("Got MST reply from unknown device %d\n" , + hdr->lct); + goto out; } + } +if (msg.req_type ==DP_CONNECTION_STATUS_NOTIFY ) {drm_dp_update_port (mstb, &msg.u.conn_stat); -DRM_DEBUG_KMS ("Got CSN: pn: %d ldps:%d ddps: %d mcs: %d ip: %d pdt: %d\n" , msg.u.conn_stat.port_number, msg.u.conn_stat.legacy_device_plug_status, msg.u.conn_stat.displayport_device_plug_status, msg.u.conn_stat.message_capability_status, msg.u.conn_stat.input_port, msg.u.conn_stat.peer_device_type); -drm_kms_helper_hotplug_event (mgr->dev); +DRM_DEBUG_KMS ("Got CSN: pn: %d ldps:%d ddps: %d mcs: %d ip: %d pdt: %d\n" , + msg.u.conn_stat.port_number, + msg.u.conn_stat.legacy_device_plug_status, + msg.u.conn_stat.displayport_device_plug_status, + msg.u.conn_stat.message_capability_status, + msg.u.conn_stat.input_port, + msg.u.conn_stat.peer_device_type); +drm_kms_helper_hotplug_event (mgr->dev); }else if (msg.req_type ==DP_RESOURCE_STATUS_NOTIFY ) { -drm_dp_send_up_ack_reply (mgr, mgr->mst_primary, msg.req_type, seqno, false); -if (!mstb) - mstb =drm_dp_get_mst_branch_device_by_guid (mgr, msg.u.resource_stat.guid); - -if (!mstb) { -DRM_DEBUG_KMS ("Got MST reply from unknown device %d\n" , mgr->up_req_recv.initial_hdr.lct); -memset (&mgr->up_req_recv,0 ,sizeof (struct drm_dp_sideband_msg_rx )); -return 0 ; - } - -DRM_DEBUG_KMS ("Got RSN: pn: %d avail_pbn %d\n" , msg.u.resource_stat.port_number, msg.u.resource_stat.available_pbn); +DRM_DEBUG_KMS ("Got RSN: pn: %d avail_pbn %d\n" , + msg.u.resource_stat.port_number, + msg.u.resource_stat.available_pbn); } -if (mstb) -drm_dp_mst_topology_put_mstb (mstb); - +drm_dp_mst_topology_put_mstb (mstb); +out:memset (&mgr->up_req_recv,0 ,sizeof (struct drm_dp_sideband_msg_rx )); -return 0 ; }
Signature · 6 lines
2.21.0 _______________________________________________ dri-devel mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/dri-devel
Lyude Paul
maintainerPATCH 12/26 drm/dp_mst: Refactor drm_dp_mst_handle_down_rep()
In reply to Lyude Paul* Remove the big ugly have_eomt conditional
* Store &mgr->down_rep_recv.initial_hdr in a var to make line wrapping easier
* Remove duplicate memset() calls
* Actually wrap lines
--- drivers/gpu/drm/drm_dp_mst_topology.c | 102 +++++++++++++------------- 1 file changed, 50 insertions(+), 52 deletions(-)
drivers/gpu/drm/drm_dp_mst_topology.c +50 -52
@@ -3060,68 +3060,66 @@ static bool drm_dp_get_one_sb_msg(struct drm_dp_mst_topology_mgr *mgr, bool up)static int drm_dp_mst_handle_down_rep (struct drm_dp_mst_topology_mgr *mgr) { -int ret =0 ; +struct drm_dp_sideband_msg_tx *txmsg; +struct drm_dp_mst_branch *mstb; +struct drm_dp_sideband_msg_hdr *hdr = &mgr->down_rep_recv.initial_hdr; +int slot =-1 ; + +if (!drm_dp_get_one_sb_msg (mgr, false)) + goto clear_down_rep_recv; -if (!drm_dp_get_one_sb_msg (mgr, false)) { -memset (&mgr->down_rep_recv,0 , -sizeof (struct drm_dp_sideband_msg_rx )); +if (!mgr->down_rep_recv.have_eomt)return 0 ; + + mstb =drm_dp_get_mst_branch_device (mgr, hdr->lct, hdr->rad); +if (!mstb) { +DRM_DEBUG_KMS ("Got MST reply from unknown device %d\n" , + hdr->lct); + goto clear_down_rep_recv; } -if (mgr->down_rep_recv.have_eomt) { -struct drm_dp_sideband_msg_tx *txmsg; -struct drm_dp_mst_branch *mstb; -int slot =-1 ; - mstb =drm_dp_get_mst_branch_device (mgr, - mgr->down_rep_recv.initial_hdr.lct, - mgr->down_rep_recv.initial_hdr.rad); +/* find the message */ + slot = hdr->seqno; +mutex_lock (&mgr->qlock); + txmsg = mstb->tx_slots[slot]; +/* remove from slots */ +mutex_unlock (&mgr->qlock); -if (!mstb) { -DRM_DEBUG_KMS ("Got MST reply from unknown device %d\n" , mgr->down_rep_recv.initial_hdr.lct); -memset (&mgr->down_rep_recv,0 ,sizeof (struct drm_dp_sideband_msg_rx )); -return 0 ; - } +if (!txmsg) { +DRM_DEBUG_KMS ("Got MST reply with no msg %p %d %d %02x %02x\n" , + mstb, hdr->seqno, hdr->lct, hdr->rad[0 ], + mgr->down_rep_recv.msg[0 ]); + goto no_msg; + } -/* find the message */ - slot = mgr->down_rep_recv.initial_hdr.seqno; -mutex_lock (&mgr->qlock); - txmsg = mstb->tx_slots[slot]; -/* remove from slots */ -mutex_unlock (&mgr->qlock); - -if (!txmsg) { -DRM_DEBUG_KMS ("Got MST reply with no msg %p %d %d %02x %02x\n" , - mstb, - mgr->down_rep_recv.initial_hdr.seqno, - mgr->down_rep_recv.initial_hdr.lct, - mgr->down_rep_recv.initial_hdr.rad[0 ], - mgr->down_rep_recv.msg[0 ]); -drm_dp_mst_topology_put_mstb (mstb); -memset (&mgr->down_rep_recv,0 ,sizeof (struct drm_dp_sideband_msg_rx )); -return 0 ; - } +drm_dp_sideband_parse_reply (&mgr->down_rep_recv, &txmsg->reply); -drm_dp_sideband_parse_reply (&mgr->down_rep_recv, &txmsg->reply); +if (txmsg->reply.reply_type ==DP_SIDEBAND_REPLY_NAK ) +DRM_DEBUG_KMS ("Got NAK reply: req 0x%02x (%s), reason 0x%02x (%s), nak data 0x%02x\n" , + txmsg->reply.req_type, +drm_dp_mst_req_type_str (txmsg->reply.req_type), + txmsg->reply.u.nak.reason, +drm_dp_mst_nak_reason_str (txmsg->reply.u.nak.reason), + txmsg->reply.u.nak.nak_data); -if (txmsg->reply.reply_type ==DP_SIDEBAND_REPLY_NAK ) -DRM_DEBUG_KMS ("Got NAK reply: req 0x%02x (%s), reason 0x%02x (%s), nak data 0x%02x\n" , - txmsg->reply.req_type, -drm_dp_mst_req_type_str (txmsg->reply.req_type), - txmsg->reply.u.nak.reason, -drm_dp_mst_nak_reason_str (txmsg->reply.u.nak.reason), - txmsg->reply.u.nak.nak_data); - -memset (&mgr->down_rep_recv,0 ,sizeof (struct drm_dp_sideband_msg_rx )); -drm_dp_mst_topology_put_mstb (mstb); +memset (&mgr->down_rep_recv,0 ,sizeof (struct drm_dp_sideband_msg_rx )); +drm_dp_mst_topology_put_mstb (mstb); -mutex_lock (&mgr->qlock); - txmsg->state =DRM_DP_SIDEBAND_TX_RX ; - mstb->tx_slots[slot] =NULL ; -mutex_unlock (&mgr->qlock); +mutex_lock (&mgr->qlock); + txmsg->state =DRM_DP_SIDEBAND_TX_RX ; + mstb->tx_slots[slot] =NULL ; +mutex_unlock (&mgr->qlock); -wake_up_all (&mgr->tx_waitq); - } -return ret; +wake_up_all (&mgr->tx_waitq); + +return 0 ; + +no_msg: +drm_dp_mst_topology_put_mstb (mstb); +clear_down_rep_recv: +memset (&mgr->down_rep_recv,0 ,sizeof (struct drm_dp_sideband_msg_rx )); + +return 0 ; }static int drm_dp_mst_handle_up_req (struct drm_dp_mst_topology_mgr *mgr)
Signature · 6 lines
2.21.0 _______________________________________________ dri-devel mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/dri-devel
Lyude Paul
maintainerPATCH 13/26 drm/dp_mst: Destroy topology_mgr mutexes
In reply to Lyude PaulTurns out we've been forgetting for a while now to actually destroy any of the mutexes that we create in drm_dp_mst_topology_mgr. So, let's do that.
--- drivers/gpu/drm/drm_dp_mst_topology.c | 5 +++++ 1 file changed, 5 insertions(+)
drivers/gpu/drm/drm_dp_mst_topology.c +5 -0
@@ -4205,6 +4205,11 @@ void drm_dp_mst_topology_mgr_destroy(struct drm_dp_mst_topology_mgr *mgr) mgr->aux =NULL ;drm_atomic_private_obj_fini (&mgr->base); mgr->funcs =NULL ; + +mutex_destroy (&mgr->destroy_connector_lock); +mutex_destroy (&mgr->payload_lock); +mutex_destroy (&mgr->qlock); +mutex_destroy (&mgr->lock); }EXPORT_SYMBOL (drm_dp_mst_topology_mgr_destroy);
Signature · 6 lines
2.21.0 _______________________________________________ dri-devel mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/dri-devel
Lyude Paul
maintainerPATCH 14/26 drm/dp_mst: Cleanup drm_dp_send_link_address() a bit
In reply to Lyude PaulDeclare local pointer to the drm_dp_link_address_ack_reply struct instead of constantly dereferencing it through the union in txmsg->reply. Then, invert the order of conditionals so we don't have to do the bulk of the work inside them, and can wrap lines even less. Then finally, rearrange variable declarations a bit.
--- drivers/gpu/drm/drm_dp_mst_topology.c | 42 +++++++++++++++------------ 1 file changed, 23 insertions(+), 19 deletions(-)
drivers/gpu/drm/drm_dp_mst_topology.c +23 -19
@@ -2311,9 +2311,9 @@ drm_dp_dump_link_address(struct drm_dp_link_address_ack_reply *reply)static void drm_dp_send_link_address (struct drm_dp_mst_topology_mgr *mgr,struct drm_dp_mst_branch *mstb) { -int len;struct drm_dp_sideband_msg_tx *txmsg; -int ret; +struct drm_dp_link_address_ack_reply *reply; +int i, len, ret; txmsg =kzalloc (sizeof (*txmsg),GFP_KERNEL );if (!txmsg)
@@ -2325,28 +2325,32 @@ static void drm_dp_send_link_address(struct drm_dp_mst_topology_mgr *mgr, mstb->link_address_sent = true;drm_dp_queue_down_tx (mgr, txmsg); +/* FIXME: Actually do some real error handling here */ ret =drm_dp_mst_wait_tx_reply (mstb, txmsg); -if (ret >0 ) { -int i; +if (ret <=0 ) { +DRM_ERROR ("Sending link address failed with %d\n" , ret); + goto out; + } +if (txmsg->reply.reply_type ==DP_SIDEBAND_REPLY_NAK ) { +DRM_ERROR ("link address NAK received\n" ); + ret = -EIO ; + goto out; + } -if (txmsg->reply.reply_type ==DP_SIDEBAND_REPLY_NAK ) { -DRM_DEBUG_KMS ("link address nak received\n" ); - }else { -DRM_DEBUG_KMS ("link address reply: %d\n" , txmsg->reply.u.link_addr.nports); -drm_dp_dump_link_address (&txmsg->reply.u.link_addr); + reply = &txmsg->reply.u.link_addr; +DRM_DEBUG_KMS ("link address reply: %d\n" , reply->nports); +drm_dp_dump_link_address (reply); -drm_dp_check_mstb_guid (mstb, txmsg->reply.u.link_addr.guid); +drm_dp_check_mstb_guid (mstb, reply->guid); -for (i =0 ; i < txmsg->reply.u.link_addr.nports; i++) { -drm_dp_add_port (mstb, mgr->dev, &txmsg->reply.u.link_addr.ports[i]); - } -drm_kms_helper_hotplug_event (mgr->dev); - } - }else { - mstb->link_address_sent = false; -DRM_DEBUG_KMS ("link address failed %d\n" , ret); - } +for (i =0 ; i < reply->nports; i++) +drm_dp_add_port (mstb, mgr->dev, &reply->ports[i]); + +drm_kms_helper_hotplug_event (mgr->dev); +out: +if (ret <=0 ) + mstb->link_address_sent = false;kfree (txmsg); }
Signature · 6 lines
2.21.0 _______________________________________________ dri-devel mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/dri-devel
Lyude Paul
maintainerPATCH 15/26 drm/dp_mst: Refactor pdt setup/teardown, add more locking
In reply to Lyude PaulSince we're going to be implementing suspend/resume reprobing very soon, we need to make sure we are extra careful to ensure that our locking actually protects the topology state where we expect it to. Turns out this isn't the case with drm_dp_port_setup_pdt() and drm_dp_port_teardown_pdt(), both of which change port->mstb without grabbing &mgr->lock.
Additionally, since most callers of these functions are just using it to teardown the port's previous PDT and setup a new one we can simplify things a bit and combine drm_dp_port_setup_pdt() and drm_dp_port_teardown_pdt() into a single function:drm_dp_port_set_pdt(). This function also handles actually ensuring that we grab the correct locks when we need to modify port->mstb.
--- drivers/gpu/drm/drm_dp_mst_topology.c | 181 +++++++++++++++----------- include/drm/drm_dp_mst_helper.h | 6 +- 2 files changed, 110 insertions(+), 77 deletions(-)
drivers/gpu/drm/drm_dp_mst_topology.c +106 -75
@@ -1485,24 +1485,6 @@ drm_dp_mst_topology_put_mstb(struct drm_dp_mst_branch *mstb)kref_put (&mstb->topology_kref, drm_dp_destroy_mst_branch_device); } -static void drm_dp_port_teardown_pdt (struct drm_dp_mst_port *port,int old_pdt) -{ -struct drm_dp_mst_branch *mstb; - -switch (old_pdt) { -case DP_PEER_DEVICE_DP_LEGACY_CONV : -case DP_PEER_DEVICE_SST_SINK : -/* remove i2c over sideband */ -drm_dp_mst_unregister_i2c_bus (&port->aux); -break ; -case DP_PEER_DEVICE_MST_BRANCHING : - mstb = port->mstb; - port->mstb =NULL ; -drm_dp_mst_topology_put_mstb (mstb); -break ; - } -} -static void drm_dp_destroy_port (struct kref *kref) { struct drm_dp_mst_port *port =
@@ -1712,38 +1694,79 @@ static u8 drm_dp_calculate_rad(struct drm_dp_mst_port *port,return parent_lct +1 ; } -/* -* return sends link address for new mstb -*/ -static bool drm_dp_port_setup_pdt (struct drm_dp_mst_port *port) +static int drm_dp_port_set_pdt (struct drm_dp_mst_port *port,u8 new_pdt) { -int ret; -u8 rad[6 ], lct; -bool send_link = false; +struct drm_dp_mst_topology_mgr *mgr = port->mgr; +struct drm_dp_mst_branch *mstb; +u8 rad[8 ], lct; +int ret =0 ; + +if (port->pdt == new_pdt) +return 0 ; + +/* Teardown the old pdt, if there is one */ +switch (port->pdt) { +case DP_PEER_DEVICE_DP_LEGACY_CONV : +case DP_PEER_DEVICE_SST_SINK : +/* +* If the new PDT would also have an i2c bus, don't bother +* with reregistering it +*/ +if (new_pdt ==DP_PEER_DEVICE_DP_LEGACY_CONV || + new_pdt ==DP_PEER_DEVICE_SST_SINK ) { + port->pdt = new_pdt; +return 0 ; + } + +/* remove i2c over sideband */ +drm_dp_mst_unregister_i2c_bus (&port->aux); +break ; +case DP_PEER_DEVICE_MST_BRANCHING : +mutex_lock (&mgr->lock); +drm_dp_mst_topology_put_mstb (port->mstb); + port->mstb =NULL ; +mutex_unlock (&mgr->lock); +break ; + } + + port->pdt = new_pdt;switch (port->pdt) {case DP_PEER_DEVICE_DP_LEGACY_CONV :case DP_PEER_DEVICE_SST_SINK :/* add i2c over sideband */ ret =drm_dp_mst_register_i2c_bus (&port->aux);break ; +case DP_PEER_DEVICE_MST_BRANCHING : lct =drm_dp_calculate_rad (port, rad); + mstb =drm_dp_add_mst_branch_device (lct, rad); +if (!mstb) { + ret = -ENOMEM ; +DRM_ERROR ("Failed to create MSTB for port %p" , port); + goto out; + } - port->mstb =drm_dp_add_mst_branch_device (lct, rad); -if (port->mstb) { - port->mstb->mgr = port->mgr; - port->mstb->port_parent = port; -/* -* Make sure this port's memory allocation stays -* around until its child MSTB releases it -*/ -drm_dp_mst_get_port_malloc (port); +mutex_lock (&mgr->lock); + port->mstb = mstb; + mstb->mgr = port->mgr; + mstb->port_parent = port; - send_link = true; - } +/* +* Make sure this port's memory allocation stays +* around until its child MSTB releases it +*/ +drm_dp_mst_get_port_malloc (port); +mutex_unlock (&mgr->lock); + +/* And make sure we send a link address for this */ + ret =1 ;break ; } -return send_link; + +out: +if (ret <0 ) + port->pdt =DP_PEER_DEVICE_NONE ; +return ret; }static void drm_dp_check_mstb_guid (struct drm_dp_mst_branch *mstb,u8 *guid)
@@ -1793,10 +1816,9 @@ static void drm_dp_add_port(struct drm_dp_mst_branch *mstb,struct drm_device *dev, struct drm_dp_link_addr_reply_port *port_msg) { + struct drm_dp_mst_topology_mgr *mgr = mstb->mgr;struct drm_dp_mst_port *port; -bool ret;bool created = false; -int old_pdt =0 ;int old_ddps =0 ; port =drm_dp_get_port (mstb, port_msg->port_number);
@@ -1808,7 +1830,7 @@ static void drm_dp_add_port(struct drm_dp_mst_branch *mstb,kref_init (&port->malloc_kref); port->parent = mstb; port->port_num = port_msg->port_number; - port->mgr = mstb->mgr; + port->mgr = mgr; port->aux.name ="DPMST" ; port->aux.dev = dev->dev;
@@ -1820,11 +1842,9 @@ static void drm_dp_add_port(struct drm_dp_mst_branch *mstb, created = true; }else { - old_pdt = port->pdt; old_ddps = port->ddps; } - port->pdt = port_msg->peer_device_type; port->input = port_msg->input_port; port->mcs = port_msg->mcs; port->ddps = port_msg->ddps;
@@ -1836,29 +1856,33 @@ static void drm_dp_add_port(struct drm_dp_mst_branch *mstb,/* manage mstb port lists with mgr lock - take a reference for this list */ if (created) { -mutex_lock (&mstb->mgr->lock); +mutex_lock (&mgr->lock);drm_dp_mst_topology_get_port (port);list_add (&port->next, &mstb->ports); -mutex_unlock (&mstb->mgr->lock); +mutex_unlock (&mgr->lock); }if (old_ddps != port->ddps) {if (port->ddps) {if (!port->input) { -drm_dp_send_enum_path_resources (mstb->mgr, - mstb, port); +drm_dp_send_enum_path_resources (mgr, mstb, + port); } }else { port->available_pbn =0 ; } } -if (old_pdt != port->pdt && !port->input) { -drm_dp_port_teardown_pdt (port, old_pdt); - - ret =drm_dp_port_setup_pdt (port); -if (ret == true) -drm_dp_send_link_address (mstb->mgr, port->mstb); +if (!port->input) { +int ret =drm_dp_port_set_pdt (port, + port_msg->peer_device_type); +if (ret ==1 ) { +drm_dp_send_link_address (mgr, port->mstb); + }else if (ret <0 ) { +DRM_ERROR ("Failed to change PDT on port %p: %d\n" , + port, ret); + goto fail; + } }if (created && !port->input) {
@@ -1866,18 +1890,11 @@ static void drm_dp_add_port(struct drm_dp_mst_branch *mstb,build_mst_prop_path (mstb, port->port_num, proppath,sizeof (proppath)); - port->connector = (*mstb->mgr->cbs->add_connector)(mstb->mgr, - port, - proppath); -if (!port->connector) { -/* remove it from the port list */ -mutex_lock (&mstb->mgr->lock); -list_del (&port->next); -mutex_unlock (&mstb->mgr->lock); -/* drop port list reference */ -drm_dp_mst_topology_put_port (port); - goto out; - } + port->connector = (*mgr->cbs->add_connector)(mgr, port, + proppath); +if (!port->connector) + goto fail; +if ((port->pdt ==DP_PEER_DEVICE_DP_LEGACY_CONV || port->pdt ==DP_PEER_DEVICE_SST_SINK ) && port->port_num >=DP_MST_LOGICAL_PORT_0 ) {
@@ -1885,28 +1902,38 @@ static void drm_dp_add_port(struct drm_dp_mst_branch *mstb, &port->aux.ddc);drm_connector_set_tile_property (port->connector); } - (*mstb->mgr->cbs->register_connector)(port->connector); + + (*mgr->cbs->register_connector)(port->connector); } -out:/* put reference to this port */ drm_dp_mst_topology_put_port (port); +return ; + +fail: +/* Remove it from the port list */ +mutex_lock (&mgr->lock); +list_del (&port->next); +mutex_unlock (&mgr->lock); + +/* Drop the port list reference */ +drm_dp_mst_topology_put_port (port); +/* And now drop our reference */ +drm_dp_mst_topology_put_port (port); }static void drm_dp_update_port (struct drm_dp_mst_branch *mstb,struct drm_dp_connection_status_notify *conn_stat) {struct drm_dp_mst_port *port; -int old_pdt;int old_ddps;bool dowork = false; + port =drm_dp_get_port (mstb, conn_stat->port_number);if (!port)return ; old_ddps = port->ddps; - old_pdt = port->pdt; - port->pdt = conn_stat->peer_device_type; port->mcs = conn_stat->message_capability_status; port->ldps = conn_stat->legacy_device_plug_status; port->ddps = conn_stat->displayport_device_plug_status;
@@ -1918,11 +1945,17 @@ static void drm_dp_update_port(struct drm_dp_mst_branch *mstb, port->available_pbn =0 ; } } -if (old_pdt != port->pdt && !port->input) { -drm_dp_port_teardown_pdt (port, old_pdt); -if (drm_dp_port_setup_pdt (port)) +if (!port->input) { +int ret =drm_dp_port_set_pdt (port, + conn_stat->peer_device_type); +if (ret ==1 ) { dowork = true; + }else if (ret <0 ) { +DRM_ERROR ("Failed to change PDT for port %p: %d\n" , + port, ret); + dowork = false; + } }drm_dp_mst_topology_put_port (port);
@@ -3884,9 +3917,7 @@ drm_dp_finish_destroy_port(struct drm_dp_mst_port *port)if (port->connector) port->mgr->cbs->destroy_connector (port->mgr, port->connector); -drm_dp_port_teardown_pdt (port, port->pdt); - port->pdt =DP_PEER_DEVICE_NONE ; - +drm_dp_port_set_pdt (port,DP_PEER_DEVICE_NONE );drm_dp_mst_put_port_malloc (port); }
include/drm/drm_dp_mst_helper.h +4 -2
@@ -55,8 +55,10 @@ struct drm_dp_vcpi {* @num_sdp_stream_sinks: Number of stream sinks * @available_pbn: Available bandwidth for this port. * @next: link to next port on this branch device -* @mstb: branch device attach below this port -* @aux: i2c aux transport to talk to device connected to this port. +* @mstb: branch device on this port, protected by +* &drm_dp_mst_topology_mgr.lock +* @aux: i2c aux transport to talk to device connected to this port, protected +* by &drm_dp_mst_topology_mgr.lock * @parent: branch device parent of this port * @vcpi: Virtual Channel Payload info for this port. * @connector: DRM connector this port is connected to.
Signature · 6 lines
2.21.0 _______________________________________________ dri-devel mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/dri-devel
Lyude Paul
maintainerPATCH 16/26 drm/dp_mst: Rename drm_dp_add_port and drm_dp_update_port
In reply to Lyude PaulThe names for these functions are rather confusing. drm_dp_add_port() sounds like a function that would simply create a port and add it to a topology, and do nothing more. Similarly, drm_dp_update_port() would be assumed to be the function that should be used to update port information after initial creation.
While those assumptions are currently correct in how these functions are used, a quick glance at drm_dp_add_port() reveals that drm_dp_add_port() can also update the information on a port, and seems explicitly designed to do so. This can be explained pretty simply by the fact that there's more situations that would involve updating the port information based on a link address response as opposed to a connection status notification than the driver's initial topology probe. Case in point:
reprobing link addresses after suspend/resume.
Since we're about to start using drm_dp_add_port() differently for suspend/resume reprobing, let's rename both functions to clarify what they actually do.
--- drivers/gpu/drm/drm_dp_mst_topology.c | 17 ++++++++++------- 1 file changed, 10 insertions(+), 7 deletions(-)
drivers/gpu/drm/drm_dp_mst_topology.c +10 -7
@@ -1812,9 +1812,10 @@ static void build_mst_prop_path(const struct drm_dp_mst_branch *mstb,strlcat (proppath, temp, proppath_size); } -static void drm_dp_add_port (struct drm_dp_mst_branch *mstb, -struct drm_device *dev, -struct drm_dp_link_addr_reply_port *port_msg) +static void +drm_dp_mst_handle_link_address_port (struct drm_dp_mst_branch *mstb, +struct drm_device *dev, +struct drm_dp_link_addr_reply_port *port_msg) {struct drm_dp_mst_topology_mgr *mgr = mstb->mgr;struct drm_dp_mst_port *port;
@@ -1922,8 +1923,9 @@ static void drm_dp_add_port(struct drm_dp_mst_branch *mstb,drm_dp_mst_topology_put_port (port); } -static void drm_dp_update_port (struct drm_dp_mst_branch *mstb, -struct drm_dp_connection_status_notify *conn_stat) +static void +drm_dp_mst_handle_conn_stat (struct drm_dp_mst_branch *mstb, +struct drm_dp_connection_status_notify *conn_stat) {struct drm_dp_mst_port *port;int old_ddps;
@@ -2377,7 +2379,8 @@ static void drm_dp_send_link_address(struct drm_dp_mst_topology_mgr *mgr,drm_dp_check_mstb_guid (mstb, reply->guid);for (i =0 ; i < reply->nports; i++) -drm_dp_add_port (mstb, mgr->dev, &reply->ports[i]); +drm_dp_mst_handle_link_address_port (mstb, mgr->dev, + &reply->ports[i]);drm_kms_helper_hotplug_event (mgr->dev);
@@ -3205,7 +3208,7 @@ static int drm_dp_mst_handle_up_req(struct drm_dp_mst_topology_mgr *mgr) }if (msg.req_type ==DP_CONNECTION_STATUS_NOTIFY ) { -drm_dp_update_port (mstb, &msg.u.conn_stat); +drm_dp_mst_handle_conn_stat (mstb, &msg.u.conn_stat); DRM_DEBUG_KMS("Got CSN: pn: %d ldps:%d ddps: %d mcs: %d ip: %d pdt: %d\n", msg.u.conn_stat.port_number,
Signature · 1 line
2.21.0
Lyude Paul
maintainerPATCH 17/26 drm/dp_mst: Remove lies in {up,down}_rep_recv documentation
In reply to Lyude PaulThese are most certainly accessed from far more than the mgr work. In fact, up_req_recv is -only- ever accessed from outside the mgr work.
--- include/drm/drm_dp_mst_helper.h | 8 ++------ 1 file changed, 2 insertions(+), 6 deletions(-)
include/drm/drm_dp_mst_helper.h +2 -6
@@ -489,15 +489,11 @@ struct drm_dp_mst_topology_mgr {int conn_base_id;/** -* @down_rep_recv: Message receiver state for down replies. This and -* @up_req_recv are only ever access from the work item, which is -* serialised. +* @down_rep_recv: Message receiver state for down replies. */ struct drm_dp_sideband_msg_rx down_rep_recv;/** -* @up_req_recv: Message receiver state for up requests. This and -* @down_rep_recv are only ever access from the work item, which is -* serialised. +* @up_req_recv: Message receiver state for up requests. */ struct drm_dp_sideband_msg_rx up_req_recv;
Signature · 1 line
2.21.0
Lyude Paul
maintainerPATCH 18/26 drm/dp_mst: Handle UP requests asynchronously
In reply to Lyude PaulOnce upon a time, hotplugging devices on MST branches actually worked in DRM. Now, it only works in amdgpu (likely because of how it's hotplug handlers are implemented). On both i915 and nouveau, hotplug notifications from MST branches are noticed - but trying to respond to them causes messaging timeouts and causes the whole topology state to go out of sync with reality, usually resulting in the user needing to replug the entire topology in hopes that it actually fixes things.
The reason for this is because the way we currently handle UP requests in MST is completely bogus. drm_dp_mst_handle_up_req() is called from drm_dp_mst_hpd_irq(), which is usually called from the driver's hotplug handler. Because we handle sending the hotplug event from this function, we actually cause the driver's hotplug handler (and in turn, all sideband transactions) to block on drm_device->mode_config.connection_mutex. This makes it impossible to send any sideband messages from the driver's connector probing functions, resulting in the aforementioned sideband message timeout.
There's even more problems with this beyond breaking hotplugging on MST branch devices. It also makes it almost impossible to protect drm_dp_mst_port struct members under a lock because we then have to worry about dealing with all of the lock dependency issues that ensue.
So, let's finally actually fix this issue by handling the processing of up requests asyncronously. This way we can send sideband messages from most contexts without having to deal with getting blocked if we hold connection_mutex. This also fixes MST branch device hotplugging on i915, finally!
--- drivers/gpu/drm/drm_dp_mst_topology.c | 148 +++++++++++++++++++------- include/drm/drm_dp_mst_helper.h | 16 +++ 2 files changed, 123 insertions(+), 41 deletions(-)
drivers/gpu/drm/drm_dp_mst_topology.c +107 -41
@@ -45,6 +45,12 @@* protocol. The helpers contain a topology manager and bandwidth manager. * The helpers encapsulate the sending and received of sideband msgs. */ +struct drm_dp_pending_up_req { +struct drm_dp_sideband_msg_hdr hdr; +struct drm_dp_sideband_msg_req_body msg; +struct list_head next; +}; +static bool dump_dp_payload_table (struct drm_dp_mst_topology_mgr *mgr,char *buf);
@@ -1178,7 +1184,7 @@ static int drm_dp_mst_wait_tx_reply(struct drm_dp_mst_branch *mstb, } } out: -if (unlikely (ret == -EIO && drm_debug & (DRM_UT_DP |DRM_UT_KMS ))) { +if (ret == -EIO &&unlikely (drm_debug & (DRM_UT_DP |DRM_UT_KMS ))) {struct drm_printer p =drm_debug_printer (DBG_PREFIX );drm_dp_mst_dump_sideband_msg_tx (&p, txmsg);
@@ -2990,6 +2996,7 @@ void drm_dp_mst_topology_mgr_suspend(struct drm_dp_mst_topology_mgr *mgr)drm_dp_dpcd_writeb (mgr->aux,DP_MSTM_CTRL ,DP_MST_EN |DP_UPSTREAM_IS_SRC );mutex_unlock (&mgr->lock); +flush_work (&mgr->up_req_work);flush_work (&mgr->work);flush_work (&mgr->destroy_connector_work); }
@@ -3162,12 +3169,70 @@ static int drm_dp_mst_handle_down_rep(struct drm_dp_mst_topology_mgr *mgr)return 0 ; } +static inline void +drm_dp_mst_process_up_req (struct drm_dp_mst_topology_mgr *mgr, +struct drm_dp_pending_up_req *up_req) +{ +struct drm_dp_mst_branch *mstb =NULL ; +struct drm_dp_sideband_msg_req_body *msg = &up_req->msg; +struct drm_dp_sideband_msg_hdr *hdr = &up_req->hdr; + +if (hdr->broadcast) { +const u8 *guid =NULL ; + +if (msg->req_type ==DP_CONNECTION_STATUS_NOTIFY ) + guid = msg->u.conn_stat.guid; +else if (msg->req_type ==DP_RESOURCE_STATUS_NOTIFY ) + guid = msg->u.resource_stat.guid; + + mstb =drm_dp_get_mst_branch_device_by_guid (mgr, guid); + }else { + mstb =drm_dp_get_mst_branch_device (mgr, hdr->lct, hdr->rad); + } + +if (!mstb) { +DRM_DEBUG_KMS ("Got MST reply from unknown device %d\n" , + hdr->lct); +return ; + } + +/* TODO: Add missing handler for DP_RESOURCE_STATUS_NOTIFY events */ +if (msg->req_type ==DP_CONNECTION_STATUS_NOTIFY ) { +drm_dp_mst_handle_conn_stat (mstb, &msg->u.conn_stat); +drm_kms_helper_hotplug_event (mgr->dev); + } + +drm_dp_mst_topology_put_mstb (mstb); +} + +static void drm_dp_mst_up_req_work (struct work_struct *work) +{ +struct drm_dp_mst_topology_mgr *mgr = + container_of(work, struct drm_dp_mst_topology_mgr, + up_req_work); +struct drm_dp_pending_up_req *up_req; + +while (true) { +mutex_lock (&mgr->up_req_lock); + up_req =list_first_entry_or_null (&mgr->up_req_list, + struct drm_dp_pending_up_req, + next); +if (up_req) +list_del (&up_req->next); +mutex_unlock (&mgr->up_req_lock); + +if (!up_req) +break ; + +drm_dp_mst_process_up_req (mgr, up_req); +kfree (up_req); + } +} +static int drm_dp_mst_handle_up_req (struct drm_dp_mst_topology_mgr *mgr) { -struct drm_dp_sideband_msg_req_body msg;struct drm_dp_sideband_msg_hdr *hdr = &mgr->up_req_recv.initial_hdr; -struct drm_dp_mst_branch *mstb =NULL ; -const u8 *guid; +struct drm_dp_pending_up_req *up_req;bool seqno;if (!drm_dp_get_one_sb_msg (mgr, true))
@@ -3176,56 +3241,53 @@ static int drm_dp_mst_handle_up_req(struct drm_dp_mst_topology_mgr *mgr)if (!mgr->up_req_recv.have_eomt)return 0 ; -if (!hdr->broadcast) { - mstb =drm_dp_get_mst_branch_device (mgr, hdr->lct, hdr->rad); -if (!mstb) { -DRM_DEBUG_KMS ("Got MST reply from unknown device %d\n" , - hdr->lct); - goto out; - } + up_req =kzalloc (sizeof (*up_req),GFP_KERNEL ); +if (!up_req) { +DRM_ERROR ("Not enough memory to process MST up req\n" ); +return -ENOMEM ; } +INIT_LIST_HEAD (&up_req->next); seqno = hdr->seqno; -drm_dp_sideband_parse_req (&mgr->up_req_recv, &msg); +drm_dp_sideband_parse_req (&mgr->up_req_recv, &up_req->msg); -if (msg.req_type ==DP_CONNECTION_STATUS_NOTIFY ) - guid = msg.u.conn_stat.guid; -else if (msg.req_type ==DP_RESOURCE_STATUS_NOTIFY ) - guid = msg.u.resource_stat.guid; -else +if (up_req->msg.req_type !=DP_CONNECTION_STATUS_NOTIFY && + up_req->msg.req_type !=DP_RESOURCE_STATUS_NOTIFY ) { +DRM_DEBUG_KMS ("Received unknown up req type, ignoring: %x\n" , + up_req->msg.req_type); +kfree (up_req); goto out; - -drm_dp_send_up_ack_reply (mgr, mgr->mst_primary, msg.req_type, seqno, - false); - -if (!mstb) { - mstb =drm_dp_get_mst_branch_device_by_guid (mgr, guid); -if (!mstb) { -DRM_DEBUG_KMS ("Got MST reply from unknown device %d\n" , - hdr->lct); - goto out; - } } -if (msg.req_type ==DP_CONNECTION_STATUS_NOTIFY ) { -drm_dp_mst_handle_conn_stat (mstb, &msg.u.conn_stat); +drm_dp_send_up_ack_reply (mgr, mgr->mst_primary, up_req->msg.req_type, + seqno, false); + +if (up_req->msg.req_type ==DP_CONNECTION_STATUS_NOTIFY ) { +const struct drm_dp_connection_status_notify *conn_stat = + &up_req->msg.u.conn_stat;DRM_DEBUG_KMS ("Got CSN: pn: %d ldps:%d ddps: %d mcs: %d ip: %d pdt: %d\n" , - msg.u.conn_stat.port_number, - msg.u.conn_stat.legacy_device_plug_status, - msg.u.conn_stat.displayport_device_plug_status, - msg.u.conn_stat.message_capability_status, - msg.u.conn_stat.input_port, - msg.u.conn_stat.peer_device_type); + conn_stat->port_number, + conn_stat->legacy_device_plug_status, + conn_stat->displayport_device_plug_status, + conn_stat->message_capability_status, + conn_stat->input_port, + conn_stat->peer_device_type); + }else if (up_req->msg.req_type ==DP_RESOURCE_STATUS_NOTIFY ) { +const struct drm_dp_resource_status_notify *res_stat = + &up_req->msg.u.resource_stat; -drm_kms_helper_hotplug_event (mgr->dev); - }else if (msg.req_type ==DP_RESOURCE_STATUS_NOTIFY ) {DRM_DEBUG_KMS ("Got RSN: pn: %d avail_pbn %d\n" , - msg.u.resource_stat.port_number, - msg.u.resource_stat.available_pbn); + res_stat->port_number, + res_stat->available_pbn); } -drm_dp_mst_topology_put_mstb (mstb); + up_req->hdr = *hdr; +mutex_lock (&mgr->up_req_lock); +list_add_tail (&up_req->next, &mgr->up_req_list); +mutex_unlock (&mgr->up_req_lock); +queue_work (system_long_wq, &mgr->up_req_work); + out:memset (&mgr->up_req_recv,0 ,sizeof (struct drm_dp_sideband_msg_rx ));return 0 ;
@@ -4186,12 +4248,15 @@ int drm_dp_mst_topology_mgr_init(struct drm_dp_mst_topology_mgr *mgr,mutex_init (&mgr->qlock);mutex_init (&mgr->payload_lock);mutex_init (&mgr->destroy_connector_lock); +mutex_init (&mgr->up_req_lock);INIT_LIST_HEAD (&mgr->tx_msg_downq);INIT_LIST_HEAD (&mgr->destroy_connector_list);INIT_LIST_HEAD (&mgr->destroy_branch_device_list); +INIT_LIST_HEAD (&mgr->up_req_list);INIT_WORK (&mgr->work, drm_dp_mst_link_probe_work);INIT_WORK (&mgr->tx_work, drm_dp_tx_work);INIT_WORK (&mgr->destroy_connector_work, drm_dp_destroy_connector_work); +INIT_WORK (&mgr->up_req_work, drm_dp_mst_up_req_work);init_waitqueue_head (&mgr->tx_waitq); mgr->dev = dev; mgr->aux = aux;
@@ -4248,6 +4313,7 @@ void drm_dp_mst_topology_mgr_destroy(struct drm_dp_mst_topology_mgr *mgr)mutex_destroy (&mgr->payload_lock);mutex_destroy (&mgr->qlock);mutex_destroy (&mgr->lock); +mutex_destroy (&mgr->up_req_lock); }EXPORT_SYMBOL (drm_dp_mst_topology_mgr_destroy);
include/drm/drm_dp_mst_helper.h +16 -0
@@ -595,6 +595,22 @@ struct drm_dp_mst_topology_mgr {* avoid locking inversion. */ struct work_struct destroy_connector_work; + +/** +* @up_req_list: List of pending up requests from the topology that +* need to be processed, in chronological order. +*/ +struct list_head up_req_list; +/** +* @up_req_lock: Protects @up_req_list +*/ +struct mutex up_req_lock; +/** +* @up_req_work: Work item to process up requests received from the +* topology. Needed to avoid blocking hotplug handling and sideband +* transmissions. +*/ +struct work_struct up_req_work; }; int drm_dp_mst_topology_mgr_init(struct drm_dp_mst_topology_mgr *mgr,
Signature · 6 lines
2.21.0 _______________________________________________ dri-devel mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/dri-devel
Lyude Paul
maintainerPATCH 19/26 drm/dp_mst: Protect drm_dp_mst_port members with connection_mutex
In reply to Lyude PaulYes-you read that right. Currently there is literally no locking in place for any of the drm_dp_mst_port struct members that can be modified in response to a link address response, or a connection status response. Which literally means if we're unlucky enough to have any sort of hotplugging event happen before we're finished with reprobing link addresses, we'll race and the contents of said struct members becomes undefined. Fun!
So, finally add some simple locking protections to our MST helpers by protecting any drm_dp_mst_port members which can be changed by link address responses or connection status notifications under drm_device->mode_config.connection_mutex.
--- drivers/gpu/drm/drm_dp_mst_topology.c | 144 +++++++++++++++++++------- include/drm/drm_dp_mst_helper.h | 39 +++++-- 2 files changed, 133 insertions(+), 50 deletions(-)
drivers/gpu/drm/drm_dp_mst_topology.c +105 -39
@@ -1352,6 +1352,7 @@ static void drm_dp_free_mst_port(struct kref *kref) container_of(kref, struct drm_dp_mst_port, malloc_kref);drm_dp_mst_put_mstb_malloc (port->parent); +mutex_destroy (&port->lock);kfree (port); }
@@ -1818,6 +1819,36 @@ static void build_mst_prop_path(const struct drm_dp_mst_branch *mstb,strlcat (proppath, temp, proppath_size); } +static void +drm_dp_mst_port_add_connector (struct drm_dp_mst_branch *mstb, +struct drm_dp_mst_port *port) +{ +struct drm_dp_mst_topology_mgr *mgr = port->mgr; +char proppath[255 ]; +int ret; + +build_mst_prop_path (mstb, port->port_num, proppath,sizeof (proppath)); + port->connector = mgr->cbs->add_connector (mgr, port, proppath); +if (!port->connector) { + ret = -ENOMEM ; + goto error; + } + +if ((port->pdt ==DP_PEER_DEVICE_DP_LEGACY_CONV || + port->pdt ==DP_PEER_DEVICE_SST_SINK ) && + port->port_num >=DP_MST_LOGICAL_PORT_0 ) { + port->cached_edid =drm_get_edid (port->connector, + &port->aux.ddc); +drm_connector_set_tile_property (port->connector); + } + + mgr->cbs->register_connector (port->connector); +return ; + +error: +DRM_ERROR ("Failed to create connector for port %p: %d\n" , port, ret); +} + static void drm_dp_mst_handle_link_address_port(struct drm_dp_mst_branch *mstb, struct drm_device *dev,
@@ -1825,8 +1856,12 @@ drm_dp_mst_handle_link_address_port(struct drm_dp_mst_branch *mstb, {struct drm_dp_mst_topology_mgr *mgr = mstb->mgr;struct drm_dp_mst_port *port; -bool created = false; -int old_ddps =0 ; +struct drm_dp_mst_branch *child_mstb =NULL ; +struct drm_connector *connector_to_destroy =NULL ; +int old_ddps =0 , ret; +u8 new_pdt =DP_PEER_DEVICE_NONE ; +bool created = false, send_link_addr = false, + create_connector = false; port =drm_dp_get_port (mstb, port_msg->port_number);if (!port) {
@@ -1835,6 +1870,7 @@ drm_dp_mst_handle_link_address_port(struct drm_dp_mst_branch *mstb,return ;kref_init (&port->topology_kref);kref_init (&port->malloc_kref); +mutex_init (&port->lock); port->parent = mstb; port->port_num = port_msg->port_number; port->mgr = mgr;
@@ -1848,11 +1884,17 @@ drm_dp_mst_handle_link_address_port(struct drm_dp_mst_branch *mstb,drm_dp_mst_get_mstb_malloc (mstb); created = true; - }else { - old_ddps = port->ddps; } +mutex_lock (&port->lock); +drm_modeset_lock (&dev->mode_config.connection_mutex,NULL ); + +if (!created) + old_ddps = port->ddps; + port->input = port_msg->input_port; +if (!port->input) + new_pdt = port_msg->peer_device_type; port->mcs = port_msg->mcs; port->ddps = port_msg->ddps; port->ldps = port_msg->legacy_device_plug_status;
@@ -1880,44 +1922,58 @@ drm_dp_mst_handle_link_address_port(struct drm_dp_mst_branch *mstb, } } -if (!port->input) { -int ret =drm_dp_port_set_pdt (port, - port_msg->peer_device_type); -if (ret ==1 ) { -drm_dp_send_link_address (mgr, port->mstb); - }else if (ret <0 ) { -DRM_ERROR ("Failed to change PDT on port %p: %d\n" , - port, ret); - goto fail; + ret =drm_dp_port_set_pdt (port, new_pdt); +if (ret ==1 ) { + send_link_addr = true; + }else if (ret <0 ) { +DRM_ERROR ("Failed to change PDT on port %p: %d\n" , + port, ret); + goto fail_unlock; + } + +if (send_link_addr) { +mutex_lock (&mgr->lock); +if (port->mstb) { + child_mstb = port->mstb; +drm_dp_mst_get_mstb_malloc (child_mstb); } +mutex_unlock (&mgr->lock); } -if (created && !port->input) { -char proppath[255 ]; +/* +* We unset port->connector before dropping connection_mutex so that +* there's no chance any of the atomic MST helpers can accidentally +* associate a to-be-destroyed connector with a port. +*/ +if (port->connector && port->input) { + connector_to_destroy = port->connector; + port->connector =NULL ; + }else if (!port->connector && !port->input) { + create_connector = true; + } -build_mst_prop_path (mstb, port->port_num, proppath, -sizeof (proppath)); - port->connector = (*mgr->cbs->add_connector)(mgr, port, - proppath); -if (!port->connector) - goto fail; +drm_modeset_unlock (&dev->mode_config.connection_mutex); -if ((port->pdt ==DP_PEER_DEVICE_DP_LEGACY_CONV || - port->pdt ==DP_PEER_DEVICE_SST_SINK ) && - port->port_num >=DP_MST_LOGICAL_PORT_0 ) { - port->cached_edid =drm_get_edid (port->connector, - &port->aux.ddc); -drm_connector_set_tile_property (port->connector); - } +if (connector_to_destroy) + mgr->cbs->destroy_connector (mgr, connector_to_destroy); +else if (create_connector) +drm_dp_mst_port_add_connector (mstb, port); + +mutex_unlock (&port->lock); - (*mgr->cbs->register_connector)(port->connector); +if (send_link_addr && child_mstb) { +drm_dp_send_link_address (mgr, child_mstb); +drm_dp_mst_put_mstb_malloc (child_mstb); }/* put reference to this port */ drm_dp_mst_topology_put_port (port);return ; -fail: +fail_unlock: +drm_modeset_unlock (&dev->mode_config.connection_mutex); +mutex_unlock (&port->lock); +/* Remove it from the port list */ mutex_lock (&mgr->lock);list_del (&port->next);
@@ -1933,6 +1989,7 @@ static void drm_dp_mst_handle_conn_stat(struct drm_dp_mst_branch *mstb, struct drm_dp_connection_status_notify *conn_stat) { + struct drm_device *dev = mstb->mgr->dev;struct drm_dp_mst_port *port;int old_ddps;bool dowork = false;
@@ -1941,6 +1998,8 @@ drm_dp_mst_handle_conn_stat(struct drm_dp_mst_branch *mstb,if (!port)return ; +drm_modeset_lock (&dev->mode_config.connection_mutex,NULL ); + old_ddps = port->ddps; port->mcs = conn_stat->message_capability_status; port->ldps = conn_stat->legacy_device_plug_status;
@@ -1966,6 +2025,7 @@ drm_dp_mst_handle_conn_stat(struct drm_dp_mst_branch *mstb, } } +drm_modeset_unlock (&dev->mode_config.connection_mutex);drm_dp_mst_topology_put_port (port);if (dowork)queue_work (system_long_wq, &mstb->mgr->work);
@@ -2058,28 +2118,34 @@ drm_dp_get_mst_branch_device_by_guid(struct drm_dp_mst_topology_mgr *mgr,static void drm_dp_check_and_send_link_address (struct drm_dp_mst_topology_mgr *mgr,struct drm_dp_mst_branch *mstb) { +struct drm_device *dev = mgr->dev;struct drm_dp_mst_port *port; -struct drm_dp_mst_branch *mstb_child; +if (!mstb->link_address_sent)drm_dp_send_link_address (mgr, mstb);list_for_each_entry (port, &mstb->ports, next) { -if (port->input) -continue ; +struct drm_dp_mst_branch *mstb_child =NULL ; + +drm_modeset_lock (&dev->mode_config.connection_mutex,NULL ); -if (!port->ddps) +if (port->input || !port->ddps) { +drm_modeset_unlock (&dev->mode_config.connection_mutex);continue ; + }if (!port->available_pbn)drm_dp_send_enum_path_resources (mgr, mstb, port); -if (port->mstb) { +if (port->mstb) mstb_child =drm_dp_mst_topology_get_mstb_validated ( mgr, port->mstb); -if (mstb_child) { -drm_dp_check_and_send_link_address (mgr, mstb_child); -drm_dp_mst_topology_put_mstb (mstb_child); - } + +drm_modeset_unlock (&dev->mode_config.connection_mutex); + +if (mstb_child) { +drm_dp_check_and_send_link_address (mgr, mstb_child); +drm_dp_mst_topology_put_mstb (mstb_child); } } }
include/drm/drm_dp_mst_helper.h +28 -11
@@ -45,23 +45,34 @@ struct drm_dp_vcpi {/** * struct drm_dp_mst_port - MST port * @port_num: port number -* @input: if this port is an input port. -* @mcs: message capability status - DP 1.2 spec. -* @ddps: DisplayPort Device Plug Status - DP 1.2 -* @pdt: Peer Device Type -* @ldps: Legacy Device Plug Status -* @dpcd_rev: DPCD revision of device on this port -* @num_sdp_streams: Number of simultaneous streams -* @num_sdp_stream_sinks: Number of stream sinks -* @available_pbn: Available bandwidth for this port. +* @input: if this port is an input port. Protected by +* &drm_device.mode_config.connection_mutex. +* @mcs: message capability status - DP 1.2 spec. Protected by +* &drm_device.mode_config.connection_mutex. +* @ddps: DisplayPort Device Plug Status - DP 1.2. Protected by +* &drm_device.mode_config.connection_mutex. +* @pdt: Peer Device Type. Protected by +* &drm_device.mode_config.connection_mutex. +* @ldps: Legacy Device Plug Status. Protected by +* &drm_device.mode_config.connection_mutex. +* @dpcd_rev: DPCD revision of device on this port. Protected by +* &drm_device.mode_config.connection_mutex. +* @num_sdp_streams: Number of simultaneous streams. Protected by +* &drm_device.mode_config.connection_mutex. +* @num_sdp_stream_sinks: Number of stream sinks. Protected by +* &drm_device.mode_config.connection_mutex. +* @available_pbn: Available bandwidth for this port. Protected by +* &drm_device.mode_config.connection_mutex. * @next: link to next port on this branch device * @mstb: branch device on this port, protected by * &drm_dp_mst_topology_mgr.lock * @aux: i2c aux transport to talk to device connected to this port, protected -* by &drm_dp_mst_topology_mgr.lock +* by &drm_device.mode_config.connection_mutex. * @parent: branch device parent of this port * @vcpi: Virtual Channel Payload info for this port. -* @connector: DRM connector this port is connected to. +* @connector: DRM connector this port is connected to. Protected by @lock. +* When there is already a connector registered for this port, this is also +* protected by &drm_device.mode_config.connection_mutex. * @mgr: topology manager this port lives under. * * This structure represents an MST port endpoint on a device somewhere
@@ -100,6 +111,12 @@ struct drm_dp_mst_port {struct drm_connector *connector;struct drm_dp_mst_topology_mgr *mgr; +/** +* @lock: Protects @connector. If needed, this lock should be grabbed +* before &drm_device.mode_config.connection_mutex. +*/ +struct mutex lock; +/** * @cached_edid: for DP logical ports - make tiling work by ensuring * that the EDID for all connectors is read immediately.
Signature · 6 lines
2.21.0 _______________________________________________ dri-devel mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/dri-devel
Lyude Paul
maintainerPATCH 20/26 drm/dp_mst: Don't forget to update port->input in drm_dp_mst_handle_conn_stat()
In reply to Lyude PaulThis probably hasn't caused any problems up until now since it's probably nearly impossible to encounter this in the wild, however if we were to receive a connection status notification from the MST hub after resume while we're in the middle of reprobing the link addresses for a topology then there's a much larger chance that a port could have changed from being an output port to input port (or vice versa). If we forget to update this bit of information, we'll potentially ignore a valid PDT change on a downstream port because we think it's an input port.
So, make sure we read the input_port field in connection status notifications in drm_dp_mst_handle_conn_stat() to prevent this from happening once we've implemented suspend/resume reprobing.
--- drivers/gpu/drm/drm_dp_mst_topology.c | 51 +++++++++++++++++++-------- 1 file changed, 37 insertions(+), 14 deletions(-)
drivers/gpu/drm/drm_dp_mst_topology.c +37 -14
@@ -1989,18 +1989,23 @@ static void drm_dp_mst_handle_conn_stat(struct drm_dp_mst_branch *mstb, struct drm_dp_connection_status_notify *conn_stat) { - struct drm_device *dev = mstb->mgr->dev; + struct drm_dp_mst_topology_mgr *mgr = mstb->mgr; +struct drm_device *dev = mgr->dev;struct drm_dp_mst_port *port; -int old_ddps; -bool dowork = false; +struct drm_connector *connector_to_destroy =NULL ; +int old_ddps, ret; +u8 new_pdt; +bool dowork = false, create_connector = false; port =drm_dp_get_port (mstb, conn_stat->port_number);if (!port)return ; +mutex_lock (&port->lock);drm_modeset_lock (&dev->mode_config.connection_mutex,NULL ); old_ddps = port->ddps; + port->input = conn_stat->input_port; port->mcs = conn_stat->message_capability_status; port->ldps = conn_stat->legacy_device_plug_status; port->ddps = conn_stat->displayport_device_plug_status;
@@ -2013,23 +2018,41 @@ drm_dp_mst_handle_conn_stat(struct drm_dp_mst_branch *mstb, } } -if (!port->input) { -int ret =drm_dp_port_set_pdt (port, - conn_stat->peer_device_type); -if (ret ==1 ) { - dowork = true; - }else if (ret <0 ) { -DRM_ERROR ("Failed to change PDT for port %p: %d\n" , - port, ret); - dowork = false; - } + new_pdt = port->input ?DP_PEER_DEVICE_NONE : conn_stat->peer_device_type; + + ret =drm_dp_port_set_pdt (port, new_pdt); +if (ret ==1 ) { + dowork = true; + }else if (ret <0 ) { +DRM_ERROR ("Failed to change PDT for port %p: %d\n" , + port, ret); + dowork = false; + } + +/* +* We unset port->connector before dropping connection_mutex so that +* there's no chance any of the atomic MST helpers can accidentally +* associate a to-be-destroyed connector with a port. +*/ +if (port->connector && port->input) { + connector_to_destroy = port->connector; + port->connector =NULL ; + }else if (!port->connector && !port->input) { + create_connector = true; }drm_modeset_unlock (&dev->mode_config.connection_mutex); + +if (connector_to_destroy) + mgr->cbs->destroy_connector (mgr, connector_to_destroy); +else if (create_connector) +drm_dp_mst_port_add_connector (mstb, port); + +mutex_unlock (&port->lock); +drm_dp_mst_topology_put_port (port);if (dowork)queue_work (system_long_wq, &mstb->mgr->work); - } static struct drm_dp_mst_branch *drm_dp_get_mst_branch_device(struct drm_dp_mst_topology_mgr *mgr,
Signature · 1 line
2.21.0
Lyude Paul
maintainerPATCH 21/26 drm/nouveau: Don't grab runtime PM refs for HPD IRQs
In reply to Lyude PaulIn order for suspend/resume reprobing to work, we need to be able to perform sideband communications during suspend/resume, along with runtime PM suspend/resume. In order to do so, we also need to make sure that nouveau doesn't bother grabbing a runtime PM reference to do so, since otherwise we'll start deadlocking runtime PM again.
Note that we weren't able to do this before, because of the DP MST helpers processing UP requests from topologies in the same context as drm_dp_mst_hpd_irq() which would have caused us to open ourselves up to receiving hotplug events and deadlocking with runtime suspend/resume. Now that those requests are handled asynchronously, this change should be completely safe.
--- drivers/gpu/drm/nouveau/nouveau_connector.c | 33 +++++++++++---------- 1 file changed, 17 insertions(+), 16 deletions(-)
drivers/gpu/drm/nouveau/nouveau_connector.c +17 -16
@@ -1129,6 +1129,16 @@ nouveau_connector_hotplug(struct nvif_notify *notify)const char *name = connector->name;struct nouveau_encoder *nv_encoder;int ret; +bool plugged = (rep->mask !=NVIF_NOTIFY_CONN_V0_UNPLUG ); + +if (rep->mask &NVIF_NOTIFY_CONN_V0_IRQ ) { +NV_DEBUG (drm,"service %s\n" , name); +drm_dp_cec_irq (&nv_connector->aux); +if ((nv_encoder =find_encoder (connector,DCB_OUTPUT_DP ))) +nv50_mstm_service (nv_encoder->dp.mstm); + +return NVIF_NOTIFY_KEEP ; + } ret =pm_runtime_get (drm->dev->dev);if (ret ==0 ) {
@@ -1149,25 +1159,16 @@ nouveau_connector_hotplug(struct nvif_notify *notify)return NVIF_NOTIFY_DROP ; } -if (rep->mask &NVIF_NOTIFY_CONN_V0_IRQ ) { -NV_DEBUG (drm,"service %s\n" , name); -drm_dp_cec_irq (&nv_connector->aux); -if ((nv_encoder =find_encoder (connector,DCB_OUTPUT_DP ))) -nv50_mstm_service (nv_encoder->dp.mstm); - }else { -bool plugged = (rep->mask !=NVIF_NOTIFY_CONN_V0_UNPLUG ); - +if (!plugged) +drm_dp_cec_unset_edid (&nv_connector->aux); +NV_DEBUG (drm,"%splugged %s\n" , plugged ?"" :"un" , name); +if ((nv_encoder =find_encoder (connector,DCB_OUTPUT_DP ))) {if (!plugged) -drm_dp_cec_unset_edid (&nv_connector->aux); -NV_DEBUG (drm,"%splugged %s\n" , plugged ?"" :"un" , name); -if ((nv_encoder =find_encoder (connector,DCB_OUTPUT_DP ))) { -if (!plugged) -nv50_mstm_remove (nv_encoder->dp.mstm); - } - -drm_helper_hpd_irq_event (connector->dev); +nv50_mstm_remove (nv_encoder->dp.mstm); } +drm_helper_hpd_irq_event (connector->dev); +pm_runtime_mark_last_busy (drm->dev->dev);pm_runtime_put_autosuspend (drm->dev->dev);return NVIF_NOTIFY_KEEP ;
Signature · 6 lines
2.21.0 _______________________________________________ dri-devel mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/dri-devel
Lyude Paul
maintainerPATCH 22/26 drm/amdgpu: Iterate through DRM connectors correctly
In reply to Lyude PaulCurrently, every single piece of code in amdgpu that loops through connectors does it incorrectly and doesn't use the proper list iteration helpers, drm_connector_list_iter_begin() and drm_connector_list_iter_end(). Yeesh.
So, do that.
--- .../gpu/drm/amd/amdgpu/amdgpu_connectors.c | 13 +++++- drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 20 +++++++--- drivers/gpu/drm/amd/amdgpu/amdgpu_display.c | 5 ++- drivers/gpu/drm/amd/amdgpu/amdgpu_encoders.c | 40 +++++++++++++------ drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c | 5 ++- drivers/gpu/drm/amd/amdgpu/dce_v10_0.c | 34 ++++++++++++---- drivers/gpu/drm/amd/amdgpu/dce_v11_0.c | 34 ++++++++++++---- drivers/gpu/drm/amd/amdgpu/dce_v6_0.c | 40 ++++++++++++++----- drivers/gpu/drm/amd/amdgpu/dce_v8_0.c | 34 ++++++++++++---- .../gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c | 33 ++++++++------- .../drm/amd/display/amdgpu_dm/amdgpu_dm_irq.c | 10 ++++- 11 files changed, 195 insertions(+), 73 deletions(-)
drivers/gpu/drm/amd/amdgpu/amdgpu_connectors.c +11 -2
@@ -1022,8 +1022,12 @@ amdgpu_connector_dvi_detect(struct drm_connector *connector, bool force)*/ if (amdgpu_connector->shared_ddc && (ret == connector_status_connected)) {struct drm_connector *list_connector; +struct drm_connector_list_iter iter;struct amdgpu_connector *list_amdgpu_connector; -list_for_each_entry (list_connector, &dev->mode_config.connector_list, head) { + +drm_connector_list_iter_begin (dev, &iter); +drm_for_each_connector_iter (list_connector, + &iter) {if (connector == list_connector)continue ; list_amdgpu_connector =to_amdgpu_connector (list_connector);
@@ -1040,6 +1044,7 @@ amdgpu_connector_dvi_detect(struct drm_connector *connector, bool force) } } } +drm_connector_list_iter_end (&iter); } } }
@@ -1501,6 +1506,7 @@ amdgpu_connector_add(struct amdgpu_device *adev, {struct drm_device *dev = adev->ddev;struct drm_connector *connector; +struct drm_connector_list_iter iter;struct amdgpu_connector *amdgpu_connector;struct amdgpu_connector_atom_dig *amdgpu_dig_connector;struct drm_encoder *encoder;
@@ -1514,10 +1520,12 @@ amdgpu_connector_add(struct amdgpu_device *adev,return ;/* see if we already added it */ -list_for_each_entry (connector, &dev->mode_config.connector_list, head) { +drm_connector_list_iter_begin (dev, &iter); +drm_for_each_connector_iter (connector, &iter) { amdgpu_connector =to_amdgpu_connector (connector);if (amdgpu_connector->connector_id == connector_id) { amdgpu_connector->devices |= supported_device; +drm_connector_list_iter_end (&iter);return ; }if (amdgpu_connector->ddc_bus && i2c_bus->valid) {
@@ -1532,6 +1540,7 @@ amdgpu_connector_add(struct amdgpu_device *adev, } } } +drm_connector_list_iter_end (&iter);/* check if it's a dp bridge */ list_for_each_entry (encoder, &dev->mode_config.encoder_list, head) {
drivers/gpu/drm/amd/amdgpu/amdgpu_device.c +14 -6
@@ -2909,6 +2909,7 @@ int amdgpu_device_suspend(struct drm_device *dev, bool suspend, bool fbcon)struct amdgpu_device *adev;struct drm_crtc *crtc;struct drm_connector *connector; +struct drm_connector_list_iter iter;int r;if (dev ==NULL || dev->dev_private ==NULL ) {
@@ -2931,9 +2932,11 @@ int amdgpu_device_suspend(struct drm_device *dev, bool suspend, bool fbcon)if (!amdgpu_device_has_dc_support (adev)) {/* turn off display hw */ drm_modeset_lock_all (dev); -list_for_each_entry (connector, &dev->mode_config.connector_list, head) { -drm_helper_connector_dpms (connector,DRM_MODE_DPMS_OFF ); - } +drm_connector_list_iter_begin (dev, &iter); +drm_for_each_connector_iter (connector, &iter) +drm_helper_connector_dpms (connector, +DRM_MODE_DPMS_OFF ); +drm_connector_list_iter_end (&iter);drm_modeset_unlock_all (dev);/* unpin the front buffers and cursors */ list_for_each_entry (crtc, &dev->mode_config.crtc_list, head) {
@@ -3012,6 +3015,7 @@ int amdgpu_device_suspend(struct drm_device *dev, bool suspend, bool fbcon)int amdgpu_device_resume (struct drm_device *dev,bool resume,bool fbcon) {struct drm_connector *connector; +struct drm_connector_list_iter iter;struct amdgpu_device *adev = dev->dev_private;struct drm_crtc *crtc;int r =0 ;
@@ -3082,9 +3086,13 @@ int amdgpu_device_resume(struct drm_device *dev, bool resume, bool fbcon)/* turn on display hw */ drm_modeset_lock_all (dev); -list_for_each_entry (connector, &dev->mode_config.connector_list, head) { -drm_helper_connector_dpms (connector,DRM_MODE_DPMS_ON ); - } + +drm_connector_list_iter_begin (dev, &iter); +drm_for_each_connector_iter (connector, &iter) +drm_helper_connector_dpms (connector, +DRM_MODE_DPMS_ON ); +drm_connector_list_iter_end (&iter); +drm_modeset_unlock_all (dev); }amdgpu_fbdev_set_suspend (adev,0 );
drivers/gpu/drm/amd/amdgpu/amdgpu_display.c +4 -1
@@ -369,11 +369,13 @@ void amdgpu_display_print_display_setup(struct drm_device *dev)struct amdgpu_connector *amdgpu_connector;struct drm_encoder *encoder;struct amdgpu_encoder *amdgpu_encoder; +struct drm_connector_list_iter iter;uint32_t devices;int i =0 ; +drm_connector_list_iter_begin (dev, &iter);DRM_INFO ("AMDGPU Display Connectors\n" ); -list_for_each_entry (connector, &dev->mode_config.connector_list, head) { +drm_for_each_connector_iter (connector, &iter) { amdgpu_connector =to_amdgpu_connector (connector);DRM_INFO ("Connector %d:\n" , i);DRM_INFO (" %s\n" , connector->name);
@@ -437,6 +439,7 @@ void amdgpu_display_print_display_setup(struct drm_device *dev) } i++; } +drm_connector_list_iter_end (&iter); }/**
drivers/gpu/drm/amd/amdgpu/amdgpu_encoders.c +28 -12
@@ -37,12 +37,14 @@ amdgpu_link_encoder_connector(struct drm_device *dev) {struct amdgpu_device *adev = dev->dev_private;struct drm_connector *connector; +struct drm_connector_list_iter iter;struct amdgpu_connector *amdgpu_connector;struct drm_encoder *encoder;struct amdgpu_encoder *amdgpu_encoder; +drm_connector_list_iter_begin (dev, &iter);/* walk the list and link encoders to connectors */ -list_for_each_entry (connector, &dev->mode_config.connector_list, head) { +drm_for_each_connector_iter (connector, &iter) { amdgpu_connector =to_amdgpu_connector (connector);list_for_each_entry (encoder, &dev->mode_config.encoder_list, head) { amdgpu_encoder =to_amdgpu_encoder (encoder);
@@ -55,6 +57,7 @@ amdgpu_link_encoder_connector(struct drm_device *dev) } } } +drm_connector_list_iter_end (&iter); }void amdgpu_encoder_set_active_device (struct drm_encoder *encoder)
@@ -62,8 +65,10 @@ void amdgpu_encoder_set_active_device(struct drm_encoder *encoder)struct drm_device *dev = encoder->dev;struct amdgpu_encoder *amdgpu_encoder =to_amdgpu_encoder (encoder);struct drm_connector *connector; +struct drm_connector_list_iter iter; -list_for_each_entry (connector, &dev->mode_config.connector_list, head) { +drm_connector_list_iter_begin (dev, &iter); +drm_for_each_connector_iter (connector, &iter) {if (connector->encoder == encoder) {struct amdgpu_connector *amdgpu_connector =to_amdgpu_connector (connector); amdgpu_encoder->active_device = amdgpu_encoder->devices & amdgpu_connector->devices;
@@ -72,6 +77,7 @@ void amdgpu_encoder_set_active_device(struct drm_encoder *encoder) amdgpu_connector->devices, encoder->encoder_type); } } +drm_connector_list_iter_end (&iter); } struct drm_connector *
@@ -79,15 +85,20 @@ amdgpu_get_connector_for_encoder(struct drm_encoder *encoder) {struct drm_device *dev = encoder->dev;struct amdgpu_encoder *amdgpu_encoder =to_amdgpu_encoder (encoder); -struct drm_connector *connector; +struct drm_connector *connector, *found =NULL ; +struct drm_connector_list_iter iter;struct amdgpu_connector *amdgpu_connector; -list_for_each_entry (connector, &dev->mode_config.connector_list, head) { +drm_connector_list_iter_begin (dev, &iter); +drm_for_each_connector_iter (connector, &iter) { amdgpu_connector =to_amdgpu_connector (connector); -if (amdgpu_encoder->active_device & amdgpu_connector->devices) -return connector; +if (amdgpu_encoder->active_device & amdgpu_connector->devices) { + found = connector; +break ; + } } -return NULL ; +drm_connector_list_iter_end (&iter); +return found; } struct drm_connector *
@@ -95,15 +106,20 @@ amdgpu_get_connector_for_encoder_init(struct drm_encoder *encoder) {struct drm_device *dev = encoder->dev;struct amdgpu_encoder *amdgpu_encoder =to_amdgpu_encoder (encoder); -struct drm_connector *connector; +struct drm_connector *connector, *found =NULL ; +struct drm_connector_list_iter iter;struct amdgpu_connector *amdgpu_connector; -list_for_each_entry (connector, &dev->mode_config.connector_list, head) { +drm_connector_list_iter_begin (dev, &iter); +drm_for_each_connector_iter (connector, &iter) { amdgpu_connector =to_amdgpu_connector (connector); -if (amdgpu_encoder->devices & amdgpu_connector->devices) -return connector; +if (amdgpu_encoder->devices & amdgpu_connector->devices) { + found = connector; +break ; + } } -return NULL ; +drm_connector_list_iter_end (&iter); +return found; }struct drm_encoder *amdgpu_get_external_encoder (struct drm_encoder *encoder)
drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c +4 -1
@@ -87,10 +87,13 @@ static void amdgpu_hotplug_work_func(struct work_struct *work)struct drm_device *dev = adev->ddev;struct drm_mode_config *mode_config = &dev->mode_config;struct drm_connector *connector; +struct drm_connector_list_iter iter;mutex_lock (&mode_config->mutex); -list_for_each_entry (connector, &mode_config->connector_list, head) +drm_connector_list_iter_begin (dev, &iter); +drm_for_each_connector_iter (connector, &iter)amdgpu_connector_hotplug (connector); +drm_connector_list_iter_end (&iter);mutex_unlock (&mode_config->mutex);/* Just fire off a uevent and let userspace tell us what to do */ drm_helper_hpd_irq_event (dev);
drivers/gpu/drm/amd/amdgpu/dce_v10_0.c +26 -8
@@ -326,9 +326,11 @@ static void dce_v10_0_hpd_init(struct amdgpu_device *adev) {struct drm_device *dev = adev->ddev;struct drm_connector *connector; +struct drm_connector_list_iter iter;u32 tmp; -list_for_each_entry (connector, &dev->mode_config.connector_list, head) { +drm_connector_list_iter_begin (dev, &iter); +drm_for_each_connector_iter (connector, &iter) {struct amdgpu_connector *amdgpu_connector =to_amdgpu_connector (connector);if (amdgpu_connector->hpd.hpd >= adev->mode_info.num_hpd)
@@ -364,6 +366,7 @@ static void dce_v10_0_hpd_init(struct amdgpu_device *adev)amdgpu_irq_get (adev, &adev->hpd_irq, amdgpu_connector->hpd.hpd); } +drm_connector_list_iter_end (&iter); }/**
@@ -378,9 +381,11 @@ static void dce_v10_0_hpd_fini(struct amdgpu_device *adev) {struct drm_device *dev = adev->ddev;struct drm_connector *connector; +struct drm_connector_list_iter iter;u32 tmp; -list_for_each_entry (connector, &dev->mode_config.connector_list, head) { +drm_connector_list_iter_begin (dev, &iter); +drm_for_each_connector_iter (connector, &iter) {struct amdgpu_connector *amdgpu_connector =to_amdgpu_connector (connector);if (amdgpu_connector->hpd.hpd >= adev->mode_info.num_hpd)
@@ -393,6 +398,7 @@ static void dce_v10_0_hpd_fini(struct amdgpu_device *adev)amdgpu_irq_put (adev, &adev->hpd_irq, amdgpu_connector->hpd.hpd); } +drm_connector_list_iter_end (&iter); }static u32 dce_v10_0_hpd_get_gpio_reg (struct amdgpu_device *adev)
@@ -1215,10 +1221,12 @@ static void dce_v10_0_afmt_audio_select_pin(struct drm_encoder *encoder)static void dce_v10_0_audio_write_latency_fields (struct drm_encoder *encoder,struct drm_display_mode *mode) { -struct amdgpu_device *adev = encoder->dev->dev_private; +struct drm_device *dev = encoder->dev; +struct amdgpu_device *adev = dev->dev_private;struct amdgpu_encoder *amdgpu_encoder =to_amdgpu_encoder (encoder);struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;struct drm_connector *connector; +struct drm_connector_list_iter iter;struct amdgpu_connector *amdgpu_connector =NULL ;u32 tmp;int interlace =0 ;
@@ -1226,12 +1234,14 @@ static void dce_v10_0_audio_write_latency_fields(struct drm_encoder *encoder,if (!dig || !dig->afmt || !dig->afmt->pin)return ; -list_for_each_entry (connector, &encoder->dev->mode_config.connector_list, head) { +drm_connector_list_iter_begin (dev, &iter); +drm_for_each_connector_iter (connector, &iter) {if (connector->encoder == encoder) { amdgpu_connector =to_amdgpu_connector (connector);break ; } } +drm_connector_list_iter_end (&iter);if (!amdgpu_connector) {DRM_ERROR ("Couldn't find encoder's connector\n" );
@@ -1257,10 +1267,12 @@ static void dce_v10_0_audio_write_latency_fields(struct drm_encoder *encoder,static void dce_v10_0_audio_write_speaker_allocation (struct drm_encoder *encoder) { -struct amdgpu_device *adev = encoder->dev->dev_private; +struct drm_device *dev = encoder->dev; +struct amdgpu_device *adev = dev->dev_private;struct amdgpu_encoder *amdgpu_encoder =to_amdgpu_encoder (encoder);struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;struct drm_connector *connector; +struct drm_connector_list_iter iter;struct amdgpu_connector *amdgpu_connector =NULL ;u32 tmp;u8 *sadb =NULL ;
@@ -1269,12 +1281,14 @@ static void dce_v10_0_audio_write_speaker_allocation(struct drm_encoder *encoderif (!dig || !dig->afmt || !dig->afmt->pin)return ; -list_for_each_entry (connector, &encoder->dev->mode_config.connector_list, head) { +drm_connector_list_iter_begin (dev, &iter); +drm_for_each_connector_iter (connector, &iter) {if (connector->encoder == encoder) { amdgpu_connector =to_amdgpu_connector (connector);break ; } } +drm_connector_list_iter_end (&iter);if (!amdgpu_connector) {DRM_ERROR ("Couldn't find encoder's connector\n" );
@@ -1309,10 +1323,12 @@ static void dce_v10_0_audio_write_speaker_allocation(struct drm_encoder *encoderstatic void dce_v10_0_audio_write_sad_regs (struct drm_encoder *encoder) { -struct amdgpu_device *adev = encoder->dev->dev_private; +struct drm_device *dev = encoder->dev; +struct amdgpu_device *adev = dev->dev_private;struct amdgpu_encoder *amdgpu_encoder =to_amdgpu_encoder (encoder);struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;struct drm_connector *connector; +struct drm_connector_list_iter iter;struct amdgpu_connector *amdgpu_connector =NULL ;struct cea_sad *sads;int i, sad_count;
@@ -1335,12 +1351,14 @@ static void dce_v10_0_audio_write_sad_regs(struct drm_encoder *encoder)if (!dig || !dig->afmt || !dig->afmt->pin)return ; -list_for_each_entry (connector, &encoder->dev->mode_config.connector_list, head) { +drm_connector_list_iter_begin (dev, &iter); +drm_for_each_connector_iter (connector, &iter) {if (connector->encoder == encoder) { amdgpu_connector =to_amdgpu_connector (connector);break ; } } +drm_connector_list_iter_end (&iter);if (!amdgpu_connector) {DRM_ERROR ("Couldn't find encoder's connector\n" );
drivers/gpu/drm/amd/amdgpu/dce_v11_0.c +26 -8
@@ -344,9 +344,11 @@ static void dce_v11_0_hpd_init(struct amdgpu_device *adev) {struct drm_device *dev = adev->ddev;struct drm_connector *connector; +struct drm_connector_list_iter iter;u32 tmp; -list_for_each_entry (connector, &dev->mode_config.connector_list, head) { +drm_connector_list_iter_begin (dev, &iter); +drm_for_each_connector_iter (connector, &iter) {struct amdgpu_connector *amdgpu_connector =to_amdgpu_connector (connector);if (amdgpu_connector->hpd.hpd >= adev->mode_info.num_hpd)
@@ -381,6 +383,7 @@ static void dce_v11_0_hpd_init(struct amdgpu_device *adev)dce_v11_0_hpd_set_polarity (adev, amdgpu_connector->hpd.hpd);amdgpu_irq_get (adev, &adev->hpd_irq, amdgpu_connector->hpd.hpd); } +drm_connector_list_iter_end (&iter); }/**
@@ -395,9 +398,11 @@ static void dce_v11_0_hpd_fini(struct amdgpu_device *adev) {struct drm_device *dev = adev->ddev;struct drm_connector *connector; +struct drm_connector_list_iter iter;u32 tmp; -list_for_each_entry (connector, &dev->mode_config.connector_list, head) { +drm_connector_list_iter_begin (dev, &iter); +drm_for_each_connector_iter (connector, &iter) {struct amdgpu_connector *amdgpu_connector =to_amdgpu_connector (connector);if (amdgpu_connector->hpd.hpd >= adev->mode_info.num_hpd)
@@ -409,6 +414,7 @@ static void dce_v11_0_hpd_fini(struct amdgpu_device *adev)amdgpu_irq_put (adev, &adev->hpd_irq, amdgpu_connector->hpd.hpd); } +drm_connector_list_iter_end (&iter); }static u32 dce_v11_0_hpd_get_gpio_reg (struct amdgpu_device *adev)
@@ -1241,10 +1247,12 @@ static void dce_v11_0_afmt_audio_select_pin(struct drm_encoder *encoder)static void dce_v11_0_audio_write_latency_fields (struct drm_encoder *encoder,struct drm_display_mode *mode) { -struct amdgpu_device *adev = encoder->dev->dev_private; +struct drm_device *dev = encoder->dev; +struct amdgpu_device *adev = dev->dev_private;struct amdgpu_encoder *amdgpu_encoder =to_amdgpu_encoder (encoder);struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;struct drm_connector *connector; +struct drm_connector_list_iter iter;struct amdgpu_connector *amdgpu_connector =NULL ;u32 tmp;int interlace =0 ;
@@ -1252,12 +1260,14 @@ static void dce_v11_0_audio_write_latency_fields(struct drm_encoder *encoder,if (!dig || !dig->afmt || !dig->afmt->pin)return ; -list_for_each_entry (connector, &encoder->dev->mode_config.connector_list, head) { +drm_connector_list_iter_begin (dev, &iter); +drm_for_each_connector_iter (connector, &iter) {if (connector->encoder == encoder) { amdgpu_connector =to_amdgpu_connector (connector);break ; } } +drm_connector_list_iter_end (&iter);if (!amdgpu_connector) {DRM_ERROR ("Couldn't find encoder's connector\n" );
@@ -1283,10 +1293,12 @@ static void dce_v11_0_audio_write_latency_fields(struct drm_encoder *encoder,static void dce_v11_0_audio_write_speaker_allocation (struct drm_encoder *encoder) { -struct amdgpu_device *adev = encoder->dev->dev_private; +struct drm_device *dev = encoder->dev; +struct amdgpu_device *adev = dev->dev_private;struct amdgpu_encoder *amdgpu_encoder =to_amdgpu_encoder (encoder);struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;struct drm_connector *connector; +struct drm_connector_list_iter iter;struct amdgpu_connector *amdgpu_connector =NULL ;u32 tmp;u8 *sadb =NULL ;
@@ -1295,12 +1307,14 @@ static void dce_v11_0_audio_write_speaker_allocation(struct drm_encoder *encoderif (!dig || !dig->afmt || !dig->afmt->pin)return ; -list_for_each_entry (connector, &encoder->dev->mode_config.connector_list, head) { +drm_connector_list_iter_begin (dev, &iter); +drm_for_each_connector_iter (connector, &iter) {if (connector->encoder == encoder) { amdgpu_connector =to_amdgpu_connector (connector);break ; } } +drm_connector_list_iter_end (&iter);if (!amdgpu_connector) {DRM_ERROR ("Couldn't find encoder's connector\n" );
@@ -1335,10 +1349,12 @@ static void dce_v11_0_audio_write_speaker_allocation(struct drm_encoder *encoderstatic void dce_v11_0_audio_write_sad_regs (struct drm_encoder *encoder) { -struct amdgpu_device *adev = encoder->dev->dev_private; +struct drm_device *dev = encoder->dev; +struct amdgpu_device *adev = dev->dev_private;struct amdgpu_encoder *amdgpu_encoder =to_amdgpu_encoder (encoder);struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;struct drm_connector *connector; +struct drm_connector_list_iter iter;struct amdgpu_connector *amdgpu_connector =NULL ;struct cea_sad *sads;int i, sad_count;
@@ -1361,12 +1377,14 @@ static void dce_v11_0_audio_write_sad_regs(struct drm_encoder *encoder)if (!dig || !dig->afmt || !dig->afmt->pin)return ; -list_for_each_entry (connector, &encoder->dev->mode_config.connector_list, head) { +drm_connector_list_iter_begin (dev, &iter); +drm_for_each_connector_iter (connector, &iter) {if (connector->encoder == encoder) { amdgpu_connector =to_amdgpu_connector (connector);break ; } } +drm_connector_list_iter_end (&iter);if (!amdgpu_connector) {DRM_ERROR ("Couldn't find encoder's connector\n" );
drivers/gpu/drm/amd/amdgpu/dce_v6_0.c +30 -10
@@ -277,9 +277,11 @@ static void dce_v6_0_hpd_init(struct amdgpu_device *adev) {struct drm_device *dev = adev->ddev;struct drm_connector *connector; +struct drm_connector_list_iter iter;u32 tmp; -list_for_each_entry (connector, &dev->mode_config.connector_list, head) { +drm_connector_list_iter_begin (dev, &iter); +drm_for_each_connector_iter (connector, &iter) {struct amdgpu_connector *amdgpu_connector =to_amdgpu_connector (connector);if (amdgpu_connector->hpd.hpd >= adev->mode_info.num_hpd)
@@ -305,7 +307,7 @@ static void dce_v6_0_hpd_init(struct amdgpu_device *adev)dce_v6_0_hpd_set_polarity (adev, amdgpu_connector->hpd.hpd);amdgpu_irq_get (adev, &adev->hpd_irq, amdgpu_connector->hpd.hpd); } - +drm_connector_list_iter_end (&iter); }/**
@@ -320,9 +322,11 @@ static void dce_v6_0_hpd_fini(struct amdgpu_device *adev) {struct drm_device *dev = adev->ddev;struct drm_connector *connector; +struct drm_connector_list_iter iter;u32 tmp; -list_for_each_entry (connector, &dev->mode_config.connector_list, head) { +drm_connector_list_iter_begin (dev, &iter); +drm_for_each_connector_iter (connector, &iter) {struct amdgpu_connector *amdgpu_connector =to_amdgpu_connector (connector);if (amdgpu_connector->hpd.hpd >= adev->mode_info.num_hpd)
@@ -334,6 +338,7 @@ static void dce_v6_0_hpd_fini(struct amdgpu_device *adev)amdgpu_irq_put (adev, &adev->hpd_irq, amdgpu_connector->hpd.hpd); } +drm_connector_list_iter_end (&iter); }static u32 dce_v6_0_hpd_get_gpio_reg (struct amdgpu_device *adev)
@@ -1120,20 +1125,24 @@ static void dce_v6_0_audio_select_pin(struct drm_encoder *encoder)static void dce_v6_0_audio_write_latency_fields (struct drm_encoder *encoder,struct drm_display_mode *mode) { -struct amdgpu_device *adev = encoder->dev->dev_private; +struct drm_device *dev = encoder->dev; +struct amdgpu_device *adev = dev->dev_private;struct amdgpu_encoder *amdgpu_encoder =to_amdgpu_encoder (encoder);struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;struct drm_connector *connector; +struct drm_connector_list_iter iter;struct amdgpu_connector *amdgpu_connector =NULL ;int interlace =0 ;u32 tmp; -list_for_each_entry (connector, &encoder->dev->mode_config.connector_list, head) { +drm_connector_list_iter_begin (dev, &iter); +drm_for_each_connector_iter (connector, &iter) {if (connector->encoder == encoder) { amdgpu_connector =to_amdgpu_connector (connector);break ; } } +drm_connector_list_iter_end (&iter);if (!amdgpu_connector) {DRM_ERROR ("Couldn't find encoder's connector\n" );
@@ -1160,21 +1169,25 @@ static void dce_v6_0_audio_write_latency_fields(struct drm_encoder *encoder,static void dce_v6_0_audio_write_speaker_allocation (struct drm_encoder *encoder) { -struct amdgpu_device *adev = encoder->dev->dev_private; +struct drm_device *dev = encoder->dev; +struct amdgpu_device *adev = dev->dev_private;struct amdgpu_encoder *amdgpu_encoder =to_amdgpu_encoder (encoder);struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;struct drm_connector *connector; +struct drm_connector_list_iter iter;struct amdgpu_connector *amdgpu_connector =NULL ;u8 *sadb =NULL ;int sad_count;u32 tmp; -list_for_each_entry (connector, &encoder->dev->mode_config.connector_list, head) { +drm_connector_list_iter_begin (dev, &iter); +drm_for_each_connector_iter (connector, &iter) {if (connector->encoder == encoder) { amdgpu_connector =to_amdgpu_connector (connector);break ; } } +drm_connector_list_iter_end (&iter);if (!amdgpu_connector) {DRM_ERROR ("Couldn't find encoder's connector\n" );
@@ -1217,10 +1230,12 @@ static void dce_v6_0_audio_write_speaker_allocation(struct drm_encoder *encoder)static void dce_v6_0_audio_write_sad_regs (struct drm_encoder *encoder) { -struct amdgpu_device *adev = encoder->dev->dev_private; +struct drm_device *dev = encoder->dev; +struct amdgpu_device *adev = dev->dev_private;struct amdgpu_encoder *amdgpu_encoder =to_amdgpu_encoder (encoder);struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;struct drm_connector *connector; +struct drm_connector_list_iter iter;struct amdgpu_connector *amdgpu_connector =NULL ;struct cea_sad *sads;int i, sad_count;
@@ -1240,12 +1255,14 @@ static void dce_v6_0_audio_write_sad_regs(struct drm_encoder *encoder) { ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR13, HDMI_AUDIO_CODING_TYPE_WMA_PRO }, }; -list_for_each_entry (connector, &encoder->dev->mode_config.connector_list, head) { +drm_connector_list_iter_begin (dev, &iter); +drm_for_each_connector_iter (connector, &iter) {if (connector->encoder == encoder) { amdgpu_connector =to_amdgpu_connector (connector);break ; } } +drm_connector_list_iter_end (&iter);if (!amdgpu_connector) {DRM_ERROR ("Couldn't find encoder's connector\n" );
@@ -1628,6 +1645,7 @@ static void dce_v6_0_afmt_setmode(struct drm_encoder *encoder,struct amdgpu_encoder *amdgpu_encoder =to_amdgpu_encoder (encoder);struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;struct drm_connector *connector; +struct drm_connector_list_iter iter;struct amdgpu_connector *amdgpu_connector =NULL ;int em =amdgpu_atombios_encoder_get_encoder_mode (encoder);int bpc =8 ;
@@ -1635,12 +1653,14 @@ static void dce_v6_0_afmt_setmode(struct drm_encoder *encoder,if (!dig || !dig->afmt)return ; -list_for_each_entry (connector, &encoder->dev->mode_config.connector_list, head) { +drm_connector_list_iter_begin (dev, &iter); +drm_for_each_connector_iter (connector, &iter) {if (connector->encoder == encoder) { amdgpu_connector =to_amdgpu_connector (connector);break ; } } +drm_connector_list_iter_end (&iter);if (!amdgpu_connector) {DRM_ERROR ("Couldn't find encoder's connector\n" );
drivers/gpu/drm/amd/amdgpu/dce_v8_0.c +26 -8
@@ -271,9 +271,11 @@ static void dce_v8_0_hpd_init(struct amdgpu_device *adev) {struct drm_device *dev = adev->ddev;struct drm_connector *connector; +struct drm_connector_list_iter iter;u32 tmp; -list_for_each_entry (connector, &dev->mode_config.connector_list, head) { +drm_connector_list_iter_begin (dev, &iter); +drm_for_each_connector_iter (connector, &iter) {struct amdgpu_connector *amdgpu_connector =to_amdgpu_connector (connector);if (amdgpu_connector->hpd.hpd >= adev->mode_info.num_hpd)
@@ -299,6 +301,7 @@ static void dce_v8_0_hpd_init(struct amdgpu_device *adev)dce_v8_0_hpd_set_polarity (adev, amdgpu_connector->hpd.hpd);amdgpu_irq_get (adev, &adev->hpd_irq, amdgpu_connector->hpd.hpd); } +drm_connector_list_iter_end (&iter); }/**
@@ -313,9 +316,11 @@ static void dce_v8_0_hpd_fini(struct amdgpu_device *adev) {struct drm_device *dev = adev->ddev;struct drm_connector *connector; +struct drm_connector_list_iter iter;u32 tmp; -list_for_each_entry (connector, &dev->mode_config.connector_list, head) { +drm_connector_list_iter_begin (dev, &iter); +drm_for_each_connector_iter (connector, &iter) {struct amdgpu_connector *amdgpu_connector =to_amdgpu_connector (connector);if (amdgpu_connector->hpd.hpd >= adev->mode_info.num_hpd)
@@ -327,6 +332,7 @@ static void dce_v8_0_hpd_fini(struct amdgpu_device *adev)amdgpu_irq_put (adev, &adev->hpd_irq, amdgpu_connector->hpd.hpd); } +drm_connector_list_iter_end (&iter); }static u32 dce_v8_0_hpd_get_gpio_reg (struct amdgpu_device *adev)
@@ -1153,10 +1159,12 @@ static void dce_v8_0_afmt_audio_select_pin(struct drm_encoder *encoder)static void dce_v8_0_audio_write_latency_fields (struct drm_encoder *encoder,struct drm_display_mode *mode) { -struct amdgpu_device *adev = encoder->dev->dev_private; +struct drm_device *dev = encoder->dev; +struct amdgpu_device *adev = dev->dev_private;struct amdgpu_encoder *amdgpu_encoder =to_amdgpu_encoder (encoder);struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;struct drm_connector *connector; +struct drm_connector_list_iter iter;struct amdgpu_connector *amdgpu_connector =NULL ;u32 tmp =0 , offset;
@@ -1165,12 +1173,14 @@ static void dce_v8_0_audio_write_latency_fields(struct drm_encoder *encoder, offset = dig->afmt->pin->offset; -list_for_each_entry (connector, &encoder->dev->mode_config.connector_list, head) { +drm_connector_list_iter_begin (dev, &iter); +drm_for_each_connector_iter (connector, &iter) {if (connector->encoder == encoder) { amdgpu_connector =to_amdgpu_connector (connector);break ; } } +drm_connector_list_iter_end (&iter);if (!amdgpu_connector) {DRM_ERROR ("Couldn't find encoder's connector\n" );
@@ -1210,10 +1220,12 @@ static void dce_v8_0_audio_write_latency_fields(struct drm_encoder *encoder,static void dce_v8_0_audio_write_speaker_allocation (struct drm_encoder *encoder) { -struct amdgpu_device *adev = encoder->dev->dev_private; +struct drm_device *dev = encoder->dev; +struct amdgpu_device *adev = dev->dev_private;struct amdgpu_encoder *amdgpu_encoder =to_amdgpu_encoder (encoder);struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;struct drm_connector *connector; +struct drm_connector_list_iter iter;struct amdgpu_connector *amdgpu_connector =NULL ;u32 offset, tmp;u8 *sadb =NULL ;
@@ -1224,12 +1236,14 @@ static void dce_v8_0_audio_write_speaker_allocation(struct drm_encoder *encoder) offset = dig->afmt->pin->offset; -list_for_each_entry (connector, &encoder->dev->mode_config.connector_list, head) { +drm_connector_list_iter_begin (dev, &iter); +drm_for_each_connector_iter (connector, &iter) {if (connector->encoder == encoder) { amdgpu_connector =to_amdgpu_connector (connector);break ; } } +drm_connector_list_iter_end (&iter);if (!amdgpu_connector) {DRM_ERROR ("Couldn't find encoder's connector\n" );
@@ -1259,11 +1273,13 @@ static void dce_v8_0_audio_write_speaker_allocation(struct drm_encoder *encoder)static void dce_v8_0_audio_write_sad_regs (struct drm_encoder *encoder) { -struct amdgpu_device *adev = encoder->dev->dev_private; +struct drm_device *dev = encoder->dev; +struct amdgpu_device *adev = dev->dev_private;struct amdgpu_encoder *amdgpu_encoder =to_amdgpu_encoder (encoder);struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;u32 offset;struct drm_connector *connector; +struct drm_connector_list_iter iter;struct amdgpu_connector *amdgpu_connector =NULL ;struct cea_sad *sads;int i, sad_count;
@@ -1288,12 +1304,14 @@ static void dce_v8_0_audio_write_sad_regs(struct drm_encoder *encoder) offset = dig->afmt->pin->offset; -list_for_each_entry (connector, &encoder->dev->mode_config.connector_list, head) { +drm_connector_list_iter_begin (dev, &iter); +drm_for_each_connector_iter (connector, &iter) {if (connector->encoder == encoder) { amdgpu_connector =to_amdgpu_connector (connector);break ; } } +drm_connector_list_iter_end (&iter);if (!amdgpu_connector) {DRM_ERROR ("Couldn't find encoder's connector\n" );
drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c +18 -15
@@ -752,27 +752,29 @@ static int detect_mst_link_for_all_connectors(struct drm_device *dev) {struct amdgpu_dm_connector *aconnector;struct drm_connector *connector; +struct drm_connector_list_iter iter;int ret =0 ; -drm_modeset_lock (&dev->mode_config.connection_mutex,NULL ); - -list_for_each_entry (connector, &dev->mode_config.connector_list, head) { +drm_connector_list_iter_begin (dev, &iter); +drm_for_each_connector_iter (connector, &iter) { aconnector =to_amdgpu_dm_connector (connector);if (aconnector->dc_link->type == dc_connection_mst_branch && aconnector->mst_mgr.aux) {DRM_DEBUG_DRIVER ("DM_MST: starting TM on aconnector: %p [id: %d]\n" , - aconnector, aconnector->base.base.id); + aconnector, + aconnector->base.base.id); ret =drm_dp_mst_topology_mgr_set_mst (&aconnector->mst_mgr, true);if (ret <0 ) {DRM_ERROR ("DM_MST: Failed to start MST\n" ); - ((struct dc_link *)aconnector->dc_link)->type = dc_connection_single; -return ret; - } + aconnector->dc_link->type = + dc_connection_single; +break ; } + } } +drm_connector_list_iter_end (&iter); -drm_modeset_unlock (&dev->mode_config.connection_mutex);return ret; }
@@ -810,14 +812,13 @@ static void s3_handle_mst(struct drm_device *dev, bool suspend) {struct amdgpu_dm_connector *aconnector;struct drm_connector *connector; +struct drm_connector_list_iter iter;struct drm_dp_mst_topology_mgr *mgr;int ret;bool need_hotplug = false; -drm_modeset_lock (&dev->mode_config.connection_mutex,NULL ); - -list_for_each_entry (connector, &dev->mode_config.connector_list, - head) { +drm_connector_list_iter_begin (dev, &iter); +drm_for_each_connector_iter (connector, &iter) { aconnector =to_amdgpu_dm_connector (connector);if (aconnector->dc_link->type != dc_connection_mst_branch || aconnector->mst_port)
@@ -835,8 +836,7 @@ static void s3_handle_mst(struct drm_device *dev, bool suspend) } } } - -drm_modeset_unlock (&dev->mode_config.connection_mutex); +drm_connector_list_iter_end (&iter);if (need_hotplug)drm_kms_helper_hotplug_event (dev);
@@ -1018,6 +1018,7 @@ static int dm_resume(void *handle)struct amdgpu_display_manager *dm = &adev->dm;struct amdgpu_dm_connector *aconnector;struct drm_connector *connector; +struct drm_connector_list_iter iter;struct drm_crtc *crtc;struct drm_crtc_state *new_crtc_state;struct dm_crtc_state *dm_new_crtc_state;
@@ -1050,7 +1051,8 @@ static int dm_resume(void *handle)amdgpu_dm_irq_resume_early (adev);/* Do detection*/ -list_for_each_entry (connector, &ddev->mode_config.connector_list, head) { +drm_connector_list_iter_begin (ddev, &iter); +drm_for_each_connector_iter (connector, &iter) { aconnector =to_amdgpu_dm_connector (connector);/*
@@ -1078,6 +1080,7 @@ static int dm_resume(void *handle)amdgpu_dm_update_connector_after_detect (aconnector);mutex_unlock (&aconnector->hpd_lock); } +drm_connector_list_iter_end (&iter);/* Force mode set in atomic commit */ for_each_new_crtc_in_state (dm->cached_state, crtc, new_crtc_state, i)
drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_irq.c +8 -2
@@ -732,8 +732,10 @@ void amdgpu_dm_hpd_init(struct amdgpu_device *adev) {struct drm_device *dev = adev->ddev;struct drm_connector *connector; +struct drm_connector_list_iter iter; -list_for_each_entry (connector, &dev->mode_config.connector_list, head) { +drm_connector_list_iter_begin (dev, &iter); +drm_for_each_connector_iter (connector, &iter) {struct amdgpu_dm_connector *amdgpu_dm_connector =to_amdgpu_dm_connector (connector);
@@ -751,6 +753,7 @@ void amdgpu_dm_hpd_init(struct amdgpu_device *adev) true); } } +drm_connector_list_iter_end (&iter); }/**
@@ -765,8 +768,10 @@ void amdgpu_dm_hpd_fini(struct amdgpu_device *adev) {struct drm_device *dev = adev->ddev;struct drm_connector *connector; +struct drm_connector_list_iter iter; -list_for_each_entry (connector, &dev->mode_config.connector_list, head) { +drm_connector_list_iter_begin (dev, &iter); +drm_for_each_connector_iter (connector, &iter) {struct amdgpu_dm_connector *amdgpu_dm_connector =to_amdgpu_dm_connector (connector);const struct dc_link *dc_link = amdgpu_dm_connector->dc_link;
@@ -779,4 +784,5 @@ void amdgpu_dm_hpd_fini(struct amdgpu_device *adev) false); } } +drm_connector_list_iter_end (&iter); }
Signature · 6 lines
2.21.0 _______________________________________________ dri-devel mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/dri-devel
Lyude Paul
maintainerPATCH 23/26 drm/amdgpu/dm: Resume short HPD IRQs before resuming MST topology
In reply to Lyude PaulSince we're going to be reprobing the entire topology state on resume now using sideband transactions, we need to ensure that we actually have short HPD irqs enabled before calling drm_dp_mst_topology_mgr_resume(). So, do that.
--- drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-)
drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c +3 -3
@@ -1041,15 +1041,15 @@ static int dm_resume(void *handle)/* program HPD filter */ dc_resume (dm->dc); -/* On resume we need to rewrite the MSTM control bits to enamble MST*/ -s3_handle_mst (ddev, false); -/* * early enable HPD Rx IRQ, should be done before set mode as short * pulse interrupts are used for MST */ amdgpu_dm_irq_resume_early (adev); +/* On resume we need to rewrite the MSTM control bits to enamble MST*/ +s3_handle_mst (ddev, false); +/* Do detection*/ drm_connector_list_iter_begin (ddev, &iter);drm_for_each_connector_iter (connector, &iter) {
Signature · 1 line
2.21.0
Lyude Paul
maintainerPATCH 24/26 drm/dp_mst: Add basic topology reprobing when resuming
In reply to Lyude PaulFinally! For a very long time, our MST helpers have had one very annoying issue: They don't know how to reprobe the topology state when coming out of suspend. This means that if a user has a machine connected to an MST topology and decides to suspend their machine, we lose all topology changes that happened during that period. That can be a big problem if the machine was connected to a different topology on the same port before resuming, as we won't bother reprobing any of the ports and likely cause the user's monitors not to come back up as expected.
So, we start fixing this by teaching our MST helpers how to reprobe the link addresses of each connected topology when resuming. As it turns out, the behavior that we want here is identical to the behavior we want when initially probing a newly connected MST topology, with a couple of important differences:
- We need to be more careful about handling the potential races between events from the MST hub that could change the topology state as we're performing the link address reprobe
- We need to be more careful about handling unlikely state changes on ports - such as an input port turning into an output port, something that would be far more likely to happen in situations like the MST hub we're connected to being changed while we're suspend
Both of which have been solved by previous commits. That leaves one requirement:
- We need to prune any MST ports in our in-memory topology state that were present when suspending, but have not appeared in the post-resume link address response from their parent branch device
Which we can now handle in this commit by modifying drm_dp_send_link_address(). We then introduce suspend/resume reprobing by introducing drm_dp_mst_topology_mgr_invalidate_mstb(), which we call in drm_dp_mst_topology_mgr_suspend() to traverse the in-memory topology state to indicate that each mstb needs it's link address resent and PBN resources reprobed.
On resume, we start back up &mgr->work and have it reprobe the topology in the same way we would on a hotplug, removing any leftover ports that no longer appear in the topology state.
--- .../gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c | 2 +- drivers/gpu/drm/drm_dp_mst_topology.c | 138 +++++++++++++----- drivers/gpu/drm/i915/display/intel_dp.c | 3 +- drivers/gpu/drm/nouveau/dispnv50/disp.c | 6 +- include/drm/drm_dp_mst_helper.h | 3 +- 5 files changed, 112 insertions(+), 40 deletions(-)
drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c +1 -1
@@ -829,7 +829,7 @@ static void s3_handle_mst(struct drm_device *dev, bool suspend)if (suspend) {drm_dp_mst_topology_mgr_suspend (mgr); }else { - ret =drm_dp_mst_topology_mgr_resume (mgr); + ret =drm_dp_mst_topology_mgr_resume (mgr, true);if (ret <0 ) {drm_dp_mst_topology_mgr_set_mst (mgr, false); need_hotplug = true;
drivers/gpu/drm/drm_dp_mst_topology.c +104 -34
192 lines not on this page · raw message
drivers/gpu/drm/i915/display/intel_dp.c +2 -1
@@ -7366,7 +7366,8 @@ void intel_dp_mst_resume(struct drm_i915_private *dev_priv)if (!intel_dp->can_mst)continue ; - ret =drm_dp_mst_topology_mgr_resume (&intel_dp->mst_mgr); + ret =drm_dp_mst_topology_mgr_resume (&intel_dp->mst_mgr, + true);if (ret) { intel_dp->is_mst = false; drm_dp_mst_topology_mgr_set_mst(&intel_dp->mst_mgr,
drivers/gpu/drm/nouveau/dispnv50/disp.c +3 -3
@@ -1300,14 +1300,14 @@ nv50_mstm_fini(struct nv50_mstm *mstm) }static void -nv50_mstm_init (struct nv50_mstm *mstm) +nv50_mstm_init (struct nv50_mstm *mstm,bool runtime) {int ret;if (!mstm || !mstm->mgr.mst_state)return ; - ret =drm_dp_mst_topology_mgr_resume (&mstm->mgr); + ret =drm_dp_mst_topology_mgr_resume (&mstm->mgr, !runtime);if (ret ==-1 ) {drm_dp_mst_topology_mgr_set_mst (&mstm->mgr, false);drm_kms_helper_hotplug_event (mstm->mgr.dev);
@@ -2257,7 +2257,7 @@ nv50_display_init(struct drm_device *dev, bool resume, bool runtime)if (encoder->encoder_type !=DRM_MODE_ENCODER_DPMST ) {struct nouveau_encoder *nv_encoder =nouveau_encoder (encoder); -nv50_mstm_init (nv_encoder->dp.mstm); +nv50_mstm_init (nv_encoder->dp.mstm, runtime); } }
include/drm/drm_dp_mst_helper.h +2 -1
@@ -683,7 +683,8 @@ void drm_dp_mst_dump_topology(struct seq_file *m,void drm_dp_mst_topology_mgr_suspend (struct drm_dp_mst_topology_mgr *mgr);int __must_check -drm_dp_mst_topology_mgr_resume (struct drm_dp_mst_topology_mgr *mgr); +drm_dp_mst_topology_mgr_resume (struct drm_dp_mst_topology_mgr *mgr, +bool sync);struct drm_dp_mst_topology_state *drm_atomic_get_mst_topology_state (struct drm_atomic_state *state,struct drm_dp_mst_topology_mgr *mgr);int __must_check
Signature · 6 lines
2.21.0 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Lyude Paul
maintainerPATCH 25/26 drm/dp_mst: Also print unhashed pointers for malloc/topology references
In reply to Lyude PaulCurrently we only print mstb/port pointer addresses in our malloc and topology refcount functions using the hashed-by-default %p, but unfortunately if you're trying to debug a use-after-free error caused by a refcounting error then this really isn't terribly useful. On the other hand though, everything in the rest of the DP MST helpers uses hashed pointer values as well and probably isn't useful to convert to unhashed. So, let's just get the best of both worlds and print both the hashed and unhashed pointer in our malloc/topology refcount debugging output. This will hopefully make it a lot easier to figure out which port/mstb is causing KASAN to get upset.
--- drivers/gpu/drm/drm_dp_mst_topology.c | 34 ++++++++++++++++----------- 1 file changed, 20 insertions(+), 14 deletions(-)
drivers/gpu/drm/drm_dp_mst_topology.c +20 -14
@@ -1325,7 +1325,8 @@ static void drm_dp_mst_get_mstb_malloc(struct drm_dp_mst_branch *mstb) {kref_get (&mstb->malloc_kref); -DRM_DEBUG ("mstb %p (%d)\n" , mstb,kref_read (&mstb->malloc_kref)); +DRM_DEBUG ("mstb %p/%px (%d)\n" , + mstb, mstb,kref_read (&mstb->malloc_kref)); }/**
@@ -1342,7 +1343,8 @@ drm_dp_mst_get_mstb_malloc(struct drm_dp_mst_branch *mstb)static void drm_dp_mst_put_mstb_malloc (struct drm_dp_mst_branch *mstb) { -DRM_DEBUG ("mstb %p (%d)\n" , mstb,kref_read (&mstb->malloc_kref) -1 ); +DRM_DEBUG ("mstb %p/%px (%d)\n" , + mstb, mstb,kref_read (&mstb->malloc_kref) -1 );kref_put (&mstb->malloc_kref, drm_dp_free_mst_branch_device); }
@@ -1377,7 +1379,8 @@ void drm_dp_mst_get_port_malloc(struct drm_dp_mst_port *port) {kref_get (&port->malloc_kref); -DRM_DEBUG ("port %p (%d)\n" , port,kref_read (&port->malloc_kref)); +DRM_DEBUG ("port %p/%px (%d)\n" , + port, port,kref_read (&port->malloc_kref)); }EXPORT_SYMBOL (drm_dp_mst_get_port_malloc);
@@ -1394,7 +1397,8 @@ EXPORT_SYMBOL(drm_dp_mst_get_port_malloc);void drm_dp_mst_put_port_malloc (struct drm_dp_mst_port *port) { -DRM_DEBUG ("port %p (%d)\n" , port,kref_read (&port->malloc_kref) -1 ); +DRM_DEBUG ("port %p/%px (%d)\n" , + port, port,kref_read (&port->malloc_kref) -1 );kref_put (&port->malloc_kref, drm_dp_free_mst_port); }EXPORT_SYMBOL (drm_dp_mst_put_port_malloc);
@@ -1445,8 +1449,8 @@ drm_dp_mst_topology_try_get_mstb(struct drm_dp_mst_branch *mstb)int ret =kref_get_unless_zero (&mstb->topology_kref);if (ret) -DRM_DEBUG ("mstb %p (%d)\n" , mstb, -kref_read (&mstb->topology_kref)); +DRM_DEBUG ("mstb %p/%px (%d)\n" , + mstb, mstb,kref_read (&mstb->topology_kref));return ret; }
@@ -1469,7 +1473,8 @@ static void drm_dp_mst_topology_get_mstb(struct drm_dp_mst_branch *mstb) {WARN_ON (kref_read (&mstb->topology_kref) ==0 );kref_get (&mstb->topology_kref); -DRM_DEBUG ("mstb %p (%d)\n" , mstb,kref_read (&mstb->topology_kref)); +DRM_DEBUG ("mstb %p/%px (%d)\n" , + mstb, mstb,kref_read (&mstb->topology_kref)); }/**
@@ -1487,8 +1492,8 @@ static void drm_dp_mst_topology_get_mstb(struct drm_dp_mst_branch *mstb)static void drm_dp_mst_topology_put_mstb (struct drm_dp_mst_branch *mstb) { -DRM_DEBUG ("mstb %p (%d)\n" , - mstb,kref_read (&mstb->topology_kref) -1 ); +DRM_DEBUG ("mstb %p/%px (%d)\n" , + mstb, mstb,kref_read (&mstb->topology_kref) -1 );kref_put (&mstb->topology_kref, drm_dp_destroy_mst_branch_device); }
@@ -1544,8 +1549,8 @@ drm_dp_mst_topology_try_get_port(struct drm_dp_mst_port *port)int ret =kref_get_unless_zero (&port->topology_kref);if (ret) -DRM_DEBUG ("port %p (%d)\n" , port, -kref_read (&port->topology_kref)); +DRM_DEBUG ("port %p/%px (%d)\n" , + port, port,kref_read (&port->topology_kref));return ret; }
@@ -1567,7 +1572,8 @@ static void drm_dp_mst_topology_get_port(struct drm_dp_mst_port *port) {WARN_ON (kref_read (&port->topology_kref) ==0 );kref_get (&port->topology_kref); -DRM_DEBUG ("port %p (%d)\n" , port,kref_read (&port->topology_kref)); +DRM_DEBUG ("port %p/%px (%d)\n" , + port, port,kref_read (&port->topology_kref)); }/**
@@ -1583,8 +1589,8 @@ static void drm_dp_mst_topology_get_port(struct drm_dp_mst_port *port)*/ static void drm_dp_mst_topology_put_port (struct drm_dp_mst_port *port) { -DRM_DEBUG ("port %p (%d)\n" , - port,kref_read (&port->topology_kref) -1 ); +DRM_DEBUG ("port %p/%px (%d)\n" , + port, port,kref_read (&port->topology_kref) -1 );kref_put (&port->topology_kref, drm_dp_destroy_port); }
Signature · 6 lines
2.21.0 _______________________________________________ dri-devel mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/dri-devel
Lyude Paul
maintainerPATCH 26/26 drm/dp_mst: Add topology ref history tracking for debugging
In reply to Lyude PaulFor very subtle mistakes with topology refs, it can be rather difficult to trace them down with the debugging info that we already have. I had one such issue recently while trying to implement suspend/resume reprobing for MST, and ended up coming up with this.
Inspired by Chris Wilson's wakeref tracking for i915, this adds a very similar feature to the DP MST helpers, which allows for partial tracking of topology refs for both ports and branch devices. This is a lot less advanced then wakeref tracking: we merely keep a count of all of the spots where a topology ref has been grabbed or dropped, then dump out that history in chronological order when a port or branch device's topology refcount reaches 0. So far, I've found this incredibly useful for debugging topology refcount errors.
Since this has the potential to be somewhat slow and loud, we add an expert kernel config option to enable or disable this feature, CONFIG_DRM_DEBUG_DP_MST_TOPOLOGY_REFS.
--- drivers/gpu/drm/Kconfig | 14 ++ drivers/gpu/drm/drm_dp_mst_topology.c | 230 +++++++++++++++++++++++++- include/drm/drm_dp_mst_helper.h | 45 +++++ 3 files changed, 285 insertions(+), 4 deletions(-)
drivers/gpu/drm/Kconfig +14 -0
@@ -89,6 +89,20 @@ config DRM_KMS_FB_HELPER help FBDEV helpers for KMS drivers. +config DRM_DEBUG_DP_MST_TOPOLOGY_REFS + bool "Enable refcount backtrace history in the DP MST helpers" + select STACKDEPOT + depends on DRM_KMS_HELPER + depends on DEBUG_KERNEL + depends on EXPERT + help + Enables debug tracing for topology refs in DRM's DP MST helpers. A + history of each topology reference/dereference will be printed to the + kernel log once a port or branch device's topology refcount reaches 0. + + This has the potential to use a lot of memory and print some very + large kernel messages. If in doubt, say "N". + config DRM_FBDEV_EMULATION bool "Enable legacy fbdev support for your modesetting driver" depends on DRM
drivers/gpu/drm/drm_dp_mst_topology.c +226 -4
317 lines not on this page · raw message
include/drm/drm_dp_mst_helper.h +45 -0
69 lines not on this page · raw message
Signature · 6 lines
2.21.0 _______________________________________________ dri-devel mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/dri-devel