X-Git-Url: https://gerrit.automotivelinux.org/gerrit/gitweb?a=blobdiff_plain;f=src%2Fdiagnostic%2Fdiagnostic-manager.cpp;h=ee584bf76b8fbb4923bccd23196bf56642c75b68;hb=dad5753b133c5312f51a4b154b91559bda94aef9;hp=31fbf0b07b2730093da3ceab98987c274cf9d23f;hpb=7175a87b640b9920e666bce53f46002ac19d4c21;p=apps%2Fagl-service-can-low-level.git diff --git a/src/diagnostic/diagnostic-manager.cpp b/src/diagnostic/diagnostic-manager.cpp index 31fbf0b0..ee584bf7 100644 --- a/src/diagnostic/diagnostic-manager.cpp +++ b/src/diagnostic/diagnostic-manager.cpp @@ -20,12 +20,13 @@ #include "diagnostic-manager.hpp" -#include "uds/uds.h" #include "../utils/openxc-utils.hpp" #include "../configuration.hpp" #define MAX_RECURRING_DIAGNOSTIC_FREQUENCY_HZ 10 #define MAX_SIMULTANEOUS_DIAG_REQUESTS 50 +// There are only 8 slots of in flight diagnostic requests +#define MAX_SIMULTANEOUS_IN_FLIGHT_REQUESTS 8 #define TIMERFD_ACCURACY 0 #define MICRO 1000000 @@ -33,9 +34,16 @@ diagnostic_manager_t::diagnostic_manager_t() : initialized_{false} {} +/// @brief Diagnostic manager isn't initialized at launch but after +/// CAN bus devices initialization. For the moment, it is only possible +/// to have 1 diagnostic bus which are the first bus declared in the JSON +/// description file. Configuration instance will return it. +/// +/// this will initialize DiagnosticShims and cancel all active requests +/// if there are any. bool diagnostic_manager_t::initialize() { - // Mandatory to set the bus before intiliaze shims. + // Mandatory to set the bus before intialize shims. bus_ = configuration_t::instance().get_diagnostic_bus(); init_diagnostic_shims(); @@ -46,27 +54,81 @@ bool diagnostic_manager_t::initialize() return initialized_; } -/** - * @brief initialize shims used by UDS lib and set initialized_ to true. - * It is needed before used the diagnostic manager fully because shims are - * required by most member functions. - */ +/// @brief initialize shims used by UDS lib and set initialized_ to true. +/// It is needed before used the diagnostic manager fully because shims are +/// required by most member functions. void diagnostic_manager_t::init_diagnostic_shims() { shims_ = diagnostic_init_shims(shims_logger, shims_send, NULL); DEBUG(binder_interface, "init_diagnostic_shims: Shims initialized"); } +/// @brief Force cleanup all active requests. void diagnostic_manager_t::reset() { - if(initialized_) - { - DEBUG(binder_interface, "Clearing existing diagnostic requests"); - cleanup_active_requests(true); - } + DEBUG(binder_interface, "Clearing existing diagnostic requests"); + cleanup_active_requests(true); +} + +/// @brief send function use by diagnostic library. Only one bus used for now +/// so diagnostic request is sent using the default diagnostic bus not matter of +/// which is specified in the diagnostic message definition. +/// +/// @param[in] arbitration_id - CAN arbitration ID to use when send message. OBD2 broadcast ID +/// is 0x7DF by example. +/// @param[in] data - The data payload for the message. NULL is valid if size is also 0. +/// @param[in] size - The size of the data payload, in bytes. +/// +/// @return true if the CAN message was sent successfully. +bool diagnostic_manager_t::shims_send(const uint32_t arbitration_id, const uint8_t* data, const uint8_t size) +{ + std::shared_ptr can_bus_dev = can_bus_t::get_can_device(configuration_t::instance().get_diagnostic_manager().bus_); + if(can_bus_dev != nullptr) + return can_bus_dev->shims_send(arbitration_id, data, size); + ERROR(binder_interface, "shims_send: Can not retrieve diagnostic bus: %s", configuration_t::instance().get_diagnostic_manager().bus_.c_str()); + return false; +} + +/// @brief The type signature for an optional logging function, if the user +/// wishes to provide one. It should print, store or otherwise display the +/// message. +/// +/// message - A format string to log using the given parameters. +/// ... (vargs) - the parameters for the format string. +/// +void diagnostic_manager_t::shims_logger(const char* format, ...) +{ + va_list args; + va_start(args, format); + + char buffer[256]; + vsnprintf(buffer, 256, format, args); + + DEBUG(binder_interface, "shims_logger: %s", buffer); +} + +/// @brief The type signature for a... OpenXC TODO: not used yet. +void diagnostic_manager_t::shims_timer() +{} + +std::shared_ptr diagnostic_manager_t::get_can_bus_dev() +{ + return can_bus_t::get_can_device(bus_); } +/// @brief Return diagnostic manager shims member. +DiagnosticShims& diagnostic_manager_t::get_shims() +{ + return shims_; +} +/// @brief Search for a specific active diagnostic request in the provided requests list +/// and erase it from the vector. This is useful at unsubscription to clean up the list otherwize +/// all received CAN messages will be passed to DiagnosticRequestHandle of all active diagnostic request +/// contained in the vector but no event if connected to, so we will decode uneeded request. +/// +/// @param[in] entry - a pointer of an active_diagnostic_request instance to clean up +/// @param[in] requests_list - a vector where to make the search and cleaning. void diagnostic_manager_t::find_and_erase(active_diagnostic_request_t* entry, std::vector& requests_list) { auto i = std::find(requests_list.begin(), requests_list.end(), entry); @@ -74,8 +136,7 @@ void diagnostic_manager_t::find_and_erase(active_diagnostic_request_t* entry, st requests_list.erase(i); } -/// Move the entry to the free list and decrement the lock count for any -/// CAN filters it used. +// @brief TODO: implement cancel_request if needed... Don't know. void diagnostic_manager_t::cancel_request(active_diagnostic_request_t* entry) { @@ -97,20 +158,26 @@ void diagnostic_manager_t::cancel_request(active_diagnostic_request_t* entry) }*/ } +/// @brief Cleanup a specific request if it isn't running and get complete. As it is almost +/// impossible to get that state for a recurring request without waiting for that, you can +/// force the cleaning operation. +/// +/// @param[in] entry - the request to clean +/// @param[in] force - Force the cleaning or not ? void diagnostic_manager_t::cleanup_request(active_diagnostic_request_t* entry, bool force) { - if(force || (entry->get_in_flight() && entry->request_completed())) + if((force || (entry != nullptr && entry->get_in_flight() && entry->request_completed()))) { entry->set_in_flight(false); char request_string[128] = {0}; diagnostic_request_to_string(&entry->get_handle()->request, request_string, sizeof(request_string)); - if(entry->get_recurring()) + if(force && entry->get_recurring()) { find_and_erase(entry, recurring_requests_); - if(force) - cancel_request(entry); + cancel_request(entry); + DEBUG(binder_interface, "cleanup_request: Cancelling completed, recurring request: %s", request_string); } else { @@ -121,7 +188,10 @@ void diagnostic_manager_t::cleanup_request(active_diagnostic_request_t* entry, b } } -/// @brief Clean up the request list +/// @brief Clean up all requests lists, recurring and not recurring. +/// +/// @param[in] force - Force the cleaning or not ? If true, that will do +/// the same effect as a call to reset(). void diagnostic_manager_t::cleanup_active_requests(bool force) { for(auto& entry : non_recurring_requests_) @@ -133,8 +203,11 @@ void diagnostic_manager_t::cleanup_active_requests(bool force) cleanup_request(entry, force); } -/// @brief Will return the active_diagnostic_request_t pointer of found request or nullptr if +/// @brief Will return the active_diagnostic_request_t pointer for theDiagnosticRequest or nullptr if /// not found. +/// +/// @param[in] request - Search key, method will go through recurring list to see if it find that request +/// holded by the DiagnosticHandle member. active_diagnostic_request_t* diagnostic_manager_t::find_recurring_request(const DiagnosticRequest* request) { for (auto& entry : recurring_requests_) @@ -151,11 +224,32 @@ active_diagnostic_request_t* diagnostic_manager_t::find_recurring_request(const return nullptr; } -std::shared_ptr diagnostic_manager_t::get_can_bus_dev() -{ - return can_bus_t::get_can_device(bus_); -} - +/// @brief Add and send a new one-time diagnostic request. +/// +/// A one-time (aka non-recurring) request can existing in parallel with a +/// recurring request for the same PID or mode, that's not a problem. +/// +/// For an example, see the docs for addRecurringRequest. This function is very +/// similar but leaves out the frequencyHz parameter. +/// +/// @param[in] request - The parameters for the request. +/// @param[in] name - Human readable name this response, to be used when +/// publishing received responses. TODO: If the name is NULL, the published output +/// will use the raw OBD-II response format. +/// @param[in] wait_for_multiple_responses - If false, When any response is received +/// for this request it will be removed from the active list. If true, the +/// request will remain active until the timeout clock expires, to allow it +/// to receive multiple response. Functional broadcast requests will always +/// waint for the timeout, regardless of this parameter. +/// @param[in] decoder - An optional DiagnosticResponseDecoder to parse the payload of +/// responses to this request. If the decoder is NULL, the output will +/// include the raw payload instead of a parsed value. +/// @param[in] callback - An optional DiagnosticResponseCallback to be notified whenever a +/// response is received for this request. +/// +/// @return true if the request was added successfully. Returns false if there +/// wasn't a free active request entry, if the frequency was too high or if the +/// CAN acceptance filters could not be configured, bool diagnostic_manager_t::add_request(DiagnosticRequest* request, const std::string name, bool wait_for_multiple_responses, const DiagnosticResponseDecoder decoder, const DiagnosticResponseCallback callback) @@ -178,13 +272,13 @@ bool diagnostic_manager_t::add_request(DiagnosticRequest* request, const std::st find_and_erase(entry, non_recurring_requests_); DEBUG(binder_interface, "Added one-time diagnostic request on bus %s: %s", - bus_, request_string); + bus_.c_str(), request_string); non_recurring_requests_.push_back(entry); } else { - WARNING(binder_interface, "There isn't enough request entry. Vector exhausted %d/%d", (int)non_recurring_requests_.size()); + WARNING(binder_interface, "There isn't enough request entry. Vector exhausted %d/%d", (int)non_recurring_requests_.size(), MAX_SIMULTANEOUS_DIAG_REQUESTS); non_recurring_requests_.resize(MAX_SIMULTANEOUS_DIAG_REQUESTS); added = false; } @@ -194,13 +288,66 @@ bool diagnostic_manager_t::add_request(DiagnosticRequest* request, const std::st bool diagnostic_manager_t::validate_optional_request_attributes(float frequencyHz) { if(frequencyHz > MAX_RECURRING_DIAGNOSTIC_FREQUENCY_HZ) { - DEBUG(binder_interface, "Requested recurring diagnostic frequency %d is higher than maximum of %d", + DEBUG(binder_interface, "Requested recurring diagnostic frequency %lf is higher than maximum of %d", frequencyHz, MAX_RECURRING_DIAGNOSTIC_FREQUENCY_HZ); return false; } return true; } +/// @brief Add and send a new recurring diagnostic request. +/// +/// At most one recurring request can be active for the same arbitration ID, mode +/// and (if set) PID on the same bus at one time. If you try and call +/// addRecurringRequest with the same key, it will return an error. +/// +/// TODO: This also adds any neccessary CAN acceptance filters so we can receive the +/// response. If the request is to the functional broadcast ID (0x7df) filters +/// are added for all functional addresses (0x7e8 to 0x7f0). +/// +/// Example: +/// +/// // Creating a functional broadcast, mode 1 request for PID 2. +/// DiagnosticRequest request = { +/// arbitration_id: 0x7df, +/// mode: 1, +/// has_pid: true, +/// pid: 2 +/// }; +/// +/// // Add a recurring request, to be sent at 1Hz, and published with the +/// // name "my_pid_request" +/// addRecurringRequest(&getConfiguration()->diagnosticsManager, +/// canBus, +/// &request, +/// "my_pid_request", +/// false, +/// NULL, +/// NULL, +/// 1); +/// +/// @param[in] request - The parameters for the request. +/// @param[in] name - An optional human readable name this response, to be used when +/// publishing received responses. If the name is NULL, the published output +/// will use the raw OBD-II response format. +/// @param[in] wait_for_multiple_responses - If false, When any response is received +/// for this request it will be removed from the active list. If true, the +/// request will remain active until the timeout clock expires, to allow it +/// to receive multiple response. Functional broadcast requests will always +/// waint for the timeout, regardless of this parameter. +/// @param[in] decoder - An optional DiagnosticResponseDecoder to parse the payload of +/// responses to this request. If the decoder is NULL, the output will +/// include the raw payload instead of a parsed value. +/// @param[in] callback - An optional DiagnosticResponseCallback to be notified whenever a +/// response is received for this request. +/// @param[in] frequencyHz - The frequency (in Hz) to send the request. A frequency above +/// MAX_RECURRING_DIAGNOSTIC_FREQUENCY_HZ is not allowed, and will make this +/// function return false. +/// +/// @return true if the request was added successfully. Returns false if there +/// was too much already running requests, if the frequency was too high TODO:or if the +/// CAN acceptance filters could not be configured, +/// bool diagnostic_manager_t::add_recurring_request(DiagnosticRequest* request, const char* name, bool wait_for_multiple_responses, const DiagnosticResponseDecoder decoder, const DiagnosticResponseCallback callback, float frequencyHz) @@ -226,13 +373,17 @@ bool diagnostic_manager_t::add_recurring_request(DiagnosticRequest* request, con diagnostic_request_to_string(&entry->get_handle()->request, request_string, sizeof(request_string)); - DEBUG(binder_interface, "add_recurring_request: Added recurring diagnostic request (freq: %f) on bus %s: %s", - frequencyHz, bus_.c_str(), request_string); - uint64_t usec; - sd_event_now(afb_daemon_get_event_loop(binder_interface->daemon), CLOCK_MONOTONIC, &usec); + sd_event_now(afb_daemon_get_event_loop(binder_interface->daemon), CLOCK_BOOTTIME, &usec); + + DEBUG(binder_interface, "add_recurring_request: Added recurring diagnostic request (freq: %f) on bus %s: (%s) at %ld", + frequencyHz, + bus_.c_str(), + request_string, + usec); + if(sd_event_add_time(afb_daemon_get_event_loop(binder_interface->daemon), &source, - CLOCK_MONOTONIC, usec, TIMERFD_ACCURACY, send_request, request) < 0) + CLOCK_BOOTTIME, usec, TIMERFD_ACCURACY, send_request, request) < 0) { ERROR(binder_interface, "add_recurring_request: Request fails to be schedule through event loop"); added = false; @@ -254,7 +405,7 @@ bool diagnostic_manager_t::add_recurring_request(DiagnosticRequest* request, con return added; } - +/// @brief Returns true if there are two active requests running for the same arbitration ID. bool diagnostic_manager_t::conflicting(active_diagnostic_request_t* request, active_diagnostic_request_t* candidate) const { return (candidate->get_in_flight() && candidate != request && @@ -263,58 +414,93 @@ bool diagnostic_manager_t::conflicting(active_diagnostic_request_t* request, act } -/// @brief Returns true if there are no other active requests to the same arbitration ID. +/// @brief Returns true if there are no other active requests to the same arbitration ID +/// and if there aren't more than 8 requests in flight at the same time. bool diagnostic_manager_t::clear_to_send(active_diagnostic_request_t* request) const { + int total_in_flight = 0; for ( auto entry : non_recurring_requests_) { if(conflicting(request, entry)) return false; + if(entry->get_in_flight()) + total_in_flight++; } for ( auto entry : recurring_requests_) { if(conflicting(request, entry)) return false; + if(entry->get_in_flight()) + total_in_flight++; } + + if(total_in_flight > MAX_SIMULTANEOUS_IN_FLIGHT_REQUESTS) + return false; return true; } +int diagnostic_manager_t::reschedule_request(sd_event_source *s, uint64_t usec, active_diagnostic_request_t* adr) +{ + usec = usec + (uint64_t)(frequency_clock_t::frequency_to_period(adr->get_frequency_clock().get_frequency())*MICRO); + DEBUG(binder_interface, "send_request: Event loop state: %d. usec: %ld", sd_event_get_state(afb_daemon_get_event_loop(binder_interface->daemon)), usec); + if(sd_event_source_set_time(s, usec) >= 0) + if(sd_event_source_set_enabled(s, SD_EVENT_ON) >= 0) + return 0; + sd_event_source_unref(s); + return -1; +} + +/// @brief Systemd timer event callback use to send CAN messages at regular interval. Depending +/// on the diagnostic message frequency. +/// +/// This should be called from systemd binder event loop and the event is created on add_recurring_request +/// +/// @param[in] s - Systemd event source pointer used to reschedule the new iteration. +/// @param[in] usec - previous call timestamp in microseconds. +/// @param[in] userdata - the DiagnosticRequest struct, use to retrieve the active request from the list. +/// +/// @return positive integer if sent and rescheduled or negative value if something wrong. If an error occurs +/// event will be disabled. int diagnostic_manager_t::send_request(sd_event_source *s, uint64_t usec, void *userdata) { diagnostic_manager_t& dm = configuration_t::instance().get_diagnostic_manager(); DiagnosticRequest* request = (DiagnosticRequest*)userdata; active_diagnostic_request_t* adr = dm.find_recurring_request(request); -// if(adr != nullptr && adr->get_can_bus_dev() == dm.get_can_bus_dev() && adr->should_send() && -// dm.clear_to_send(adr)) - if(adr != nullptr && adr->get_can_bus_dev() == dm.get_can_bus_dev()) + if(adr != nullptr && adr->get_can_bus_dev() == dm.get_can_bus_dev() && adr->should_send() && + dm.clear_to_send(adr)) { adr->get_frequency_clock().tick(); start_diagnostic_request(&dm.shims_, adr->get_handle()); if(adr->get_handle()->completed && !adr->get_handle()->success) { - DEBUG(binder_interface, "send_request: Fatal error sending diagnostic request"); - return 0; + ERROR(binder_interface, "send_request: Fatal error sending diagnostic request"); + sd_event_source_unref(s); + return -1; } + adr->get_timeout_clock().tick(); adr->set_in_flight(true); + } - if(adr->get_recurring()) - { - usec = usec + (uint64_t)(frequency_clock_t::frequency_to_period(adr->get_frequency_clock().get_frequency())*MICRO); - DEBUG(binder_interface, "send_request: Event loop state: %d. usec: %ld", sd_event_get_state(afb_daemon_get_event_loop(binder_interface->daemon)), usec); - if(sd_event_source_set_time(s, usec+1000000) >= 0) - if(sd_event_source_set_enabled(s, SD_EVENT_ON) >= 0) - return 1; - } + if(adr != nullptr && adr->get_recurring()) + { + return dm.reschedule_request(s, usec, adr); } + sd_event_source_unref(s); ERROR(binder_interface, "send_request: Something goes wrong when submitting a new request to the CAN bus"); - return -1; + return -2; } -openxc_VehicleMessage diagnostic_manager_t::relay_diagnostic_response(active_diagnostic_request_t* adr, const DiagnosticResponse& response) const +/// @brief Will decode the diagnostic response and build the final openxc_VehicleMessage to return. +/// +/// @param[in] adr - A pointer to an active diagnostic request holding a valid diagnostic handle +/// @param[in] response - The response to decode from which the Vehicle message will be built and returned +/// +/// @return A filled openxc_VehicleMessage or a zeroed struct if there is an error. +openxc_VehicleMessage diagnostic_manager_t::relay_diagnostic_response(active_diagnostic_request_t* adr, const DiagnosticResponse& response) { openxc_VehicleMessage message = build_VehicleMessage(); float value = (float)diagnostic_payload_to_integer(&response); @@ -338,6 +524,17 @@ openxc_VehicleMessage diagnostic_manager_t::relay_diagnostic_response(active_dia message = build_VehicleMessage(adr, response, value); } + // If not success but completed then the pid isn't supported + if(!response.success) + { + std::vector found_signals; + configuration_t::instance().find_diagnostic_messages( build_DynamicField(adr->get_name()), found_signals ); + found_signals.front()->set_supported(false); + cleanup_request(adr, true); + NOTICE(binder_interface, "relay_diagnostic_response: PID not supported or ill formed. Please unsubscribe from it. Error code : %d", response.negative_response_code); + message = build_VehicleMessage(build_SimpleMessage(adr->get_name(), build_DynamicField("This PID isn't supported by your vehicle."))); + } + if(adr->get_callback() != nullptr) { adr->get_callback()(adr, &response, value); @@ -412,28 +609,3 @@ bool diagnostic_manager_t::is_diagnostic_response(const can_message_t& cm) return true; return false; } - -DiagnosticShims& diagnostic_manager_t::get_shims() -{ - return shims_; -} - -bool diagnostic_manager_t::shims_send(const uint32_t arbitration_id, const uint8_t* data, const uint8_t size) -{ - std::shared_ptr can_bus_dev = can_bus_t::get_can_device(configuration_t::instance().get_diagnostic_manager().bus_); - return can_bus_dev->shims_send(arbitration_id, data, size); -} - -void diagnostic_manager_t::shims_logger(const char* format, ...) -{ - va_list args; - va_start(args, format); - - char buffer[256]; - vsnprintf(buffer, 256, format, args); - - DEBUG(binder_interface, "shims_logger: %s", buffer); -} - -void diagnostic_manager_t::shims_timer() -{}