Jason M. Bills | 1490b14 | 2019-07-01 15:48:43 -0700 | [diff] [blame] | 1 | /* |
| 2 | // Copyright (c) 2019 Intel Corporation |
| 3 | // |
| 4 | // Licensed under the Apache License, Version 2.0 (the "License"); |
| 5 | // you may not use this file except in compliance with the License. |
| 6 | // You may obtain a copy of the License at |
| 7 | // |
| 8 | // http://www.apache.org/licenses/LICENSE-2.0 |
| 9 | // |
| 10 | // Unless required by applicable law or agreed to in writing, software |
| 11 | // distributed under the License is distributed on an "AS IS" BASIS, |
| 12 | // WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. |
| 13 | // See the License for the specific language governing permissions and |
| 14 | // limitations under the License. |
| 15 | */ |
Chen Yugang | e6c0f1c | 2019-08-02 20:36:42 +0800 | [diff] [blame] | 16 | #include <systemd/sd-journal.h> |
| 17 | |
Jason M. Bills | 1490b14 | 2019-07-01 15:48:43 -0700 | [diff] [blame] | 18 | #include <boost/asio/posix/stream_descriptor.hpp> |
| 19 | #include <gpiod.hpp> |
| 20 | #include <iostream> |
| 21 | #include <sdbusplus/asio/object_server.hpp> |
| 22 | |
| 23 | namespace host_error_monitor |
| 24 | { |
| 25 | static boost::asio::io_service io; |
| 26 | static std::shared_ptr<sdbusplus::asio::connection> conn; |
| 27 | |
| 28 | static bool hostOff = true; |
| 29 | |
| 30 | const static constexpr size_t caterrTimeoutMs = 2000; |
| 31 | const static constexpr size_t crashdumpTimeoutS = 300; |
| 32 | |
| 33 | // Timers |
| 34 | // Timer for CATERR asserted |
| 35 | static boost::asio::steady_timer caterrAssertTimer(io); |
| 36 | |
| 37 | // GPIO Lines and Event Descriptors |
| 38 | static gpiod::line caterrLine; |
| 39 | static boost::asio::posix::stream_descriptor caterrEvent(io); |
Chen Yugang | e6c0f1c | 2019-08-02 20:36:42 +0800 | [diff] [blame] | 40 | //---------------------------------- |
| 41 | // PCH_BMC_THERMTRIP function related definition |
| 42 | //---------------------------------- |
| 43 | // GPIO Lines and Event Descriptors |
| 44 | static gpiod::line pchThermtripLine; |
| 45 | static boost::asio::posix::stream_descriptor pchThermtripEvent(io); |
Jason M. Bills | 1490b14 | 2019-07-01 15:48:43 -0700 | [diff] [blame] | 46 | |
Jason M. Bills | a15c252 | 2019-08-16 10:01:44 -0700 | [diff] [blame^] | 47 | static void initializeErrorState(); |
Jason M. Bills | 1490b14 | 2019-07-01 15:48:43 -0700 | [diff] [blame] | 48 | static void initializeHostState() |
| 49 | { |
| 50 | conn->async_method_call( |
| 51 | [](boost::system::error_code ec, |
| 52 | const std::variant<std::string>& property) { |
| 53 | if (ec) |
| 54 | { |
| 55 | return; |
| 56 | } |
| 57 | const std::string* state = std::get_if<std::string>(&property); |
| 58 | if (state == nullptr) |
| 59 | { |
| 60 | std::cerr << "Unable to read host state value\n"; |
| 61 | return; |
| 62 | } |
| 63 | hostOff = *state == "xyz.openbmc_project.State.Host.HostState.Off"; |
Jason M. Bills | a15c252 | 2019-08-16 10:01:44 -0700 | [diff] [blame^] | 64 | // If the system is on, initialize the error state |
| 65 | if (!hostOff) |
| 66 | { |
| 67 | initializeErrorState(); |
| 68 | } |
Jason M. Bills | 1490b14 | 2019-07-01 15:48:43 -0700 | [diff] [blame] | 69 | }, |
| 70 | "xyz.openbmc_project.State.Host", "/xyz/openbmc_project/state/host0", |
| 71 | "org.freedesktop.DBus.Properties", "Get", |
| 72 | "xyz.openbmc_project.State.Host", "CurrentHostState"); |
| 73 | } |
| 74 | |
| 75 | static std::shared_ptr<sdbusplus::bus::match::match> startHostStateMonitor() |
| 76 | { |
| 77 | return std::make_shared<sdbusplus::bus::match::match>( |
| 78 | *conn, |
| 79 | "type='signal',interface='org.freedesktop.DBus.Properties'," |
| 80 | "member='PropertiesChanged',arg0namespace='xyz.openbmc_project.State." |
| 81 | "Host'", |
| 82 | [](sdbusplus::message::message& msg) { |
| 83 | std::string interfaceName; |
| 84 | boost::container::flat_map<std::string, std::variant<std::string>> |
| 85 | propertiesChanged; |
| 86 | std::string state; |
| 87 | try |
| 88 | { |
| 89 | msg.read(interfaceName, propertiesChanged); |
| 90 | state = |
| 91 | std::get<std::string>(propertiesChanged.begin()->second); |
| 92 | } |
| 93 | catch (std::exception& e) |
| 94 | { |
| 95 | std::cerr << "Unable to read host state\n"; |
| 96 | return; |
| 97 | } |
| 98 | hostOff = state == "xyz.openbmc_project.State.Host.HostState.Off"; |
| 99 | |
| 100 | // No host events should fire while off, so cancel any pending |
| 101 | // timers |
| 102 | if (hostOff) |
| 103 | { |
| 104 | caterrAssertTimer.cancel(); |
| 105 | } |
| 106 | }); |
| 107 | } |
| 108 | |
| 109 | static bool requestGPIOEvents( |
| 110 | const std::string& name, const std::function<void()>& handler, |
| 111 | gpiod::line& gpioLine, |
| 112 | boost::asio::posix::stream_descriptor& gpioEventDescriptor) |
| 113 | { |
| 114 | // Find the GPIO line |
| 115 | gpioLine = gpiod::find_line(name); |
| 116 | if (!gpioLine) |
| 117 | { |
| 118 | std::cerr << "Failed to find the " << name << " line\n"; |
| 119 | return false; |
| 120 | } |
| 121 | |
| 122 | try |
| 123 | { |
| 124 | gpioLine.request( |
| 125 | {"host-error-monitor", gpiod::line_request::EVENT_BOTH_EDGES}); |
| 126 | } |
| 127 | catch (std::exception&) |
| 128 | { |
| 129 | std::cerr << "Failed to request events for " << name << "\n"; |
| 130 | return false; |
| 131 | } |
| 132 | |
| 133 | int gpioLineFd = gpioLine.event_get_fd(); |
| 134 | if (gpioLineFd < 0) |
| 135 | { |
| 136 | std::cerr << "Failed to get " << name << " fd\n"; |
| 137 | return false; |
| 138 | } |
| 139 | |
| 140 | gpioEventDescriptor.assign(gpioLineFd); |
| 141 | |
| 142 | gpioEventDescriptor.async_wait( |
| 143 | boost::asio::posix::stream_descriptor::wait_read, |
| 144 | [&name, handler](const boost::system::error_code ec) { |
| 145 | if (ec) |
| 146 | { |
| 147 | std::cerr << name << " fd handler error: " << ec.message() |
| 148 | << "\n"; |
| 149 | return; |
| 150 | } |
| 151 | handler(); |
| 152 | }); |
| 153 | return true; |
| 154 | } |
| 155 | |
| 156 | static void startPowerCycle() |
| 157 | { |
| 158 | conn->async_method_call( |
| 159 | [](boost::system::error_code ec) { |
| 160 | if (ec) |
| 161 | { |
| 162 | std::cerr << "failed to set Chassis State\n"; |
| 163 | } |
| 164 | }, |
| 165 | "xyz.openbmc_project.State.Chassis", |
| 166 | "/xyz/openbmc_project/state/chassis0", |
| 167 | "org.freedesktop.DBus.Properties", "Set", |
| 168 | "xyz.openbmc_project.State.Chassis", "RequestedPowerTransition", |
| 169 | std::variant<std::string>{ |
| 170 | "xyz.openbmc_project.State.Chassis.Transition.PowerCycle"}); |
| 171 | } |
| 172 | |
| 173 | static void startCrashdumpAndRecovery(bool recoverSystem) |
| 174 | { |
| 175 | std::cout << "Starting crashdump\n"; |
| 176 | static std::shared_ptr<sdbusplus::bus::match::match> crashdumpCompleteMatch; |
| 177 | static boost::asio::steady_timer crashdumpTimer(io); |
| 178 | |
| 179 | crashdumpCompleteMatch = std::make_shared<sdbusplus::bus::match::match>( |
| 180 | *conn, |
| 181 | "type='signal',interface='org.freedesktop.DBus.Properties'," |
| 182 | "member='PropertiesChanged',arg0namespace='com.intel.crashdump'", |
| 183 | [recoverSystem](sdbusplus::message::message& msg) { |
| 184 | crashdumpTimer.cancel(); |
| 185 | std::cout << "Crashdump completed\n"; |
| 186 | if (recoverSystem) |
| 187 | { |
| 188 | std::cout << "Recovering the system\n"; |
| 189 | startPowerCycle(); |
| 190 | } |
| 191 | crashdumpCompleteMatch.reset(); |
| 192 | }); |
| 193 | |
| 194 | crashdumpTimer.expires_after(std::chrono::seconds(crashdumpTimeoutS)); |
| 195 | crashdumpTimer.async_wait([](const boost::system::error_code ec) { |
| 196 | if (ec) |
| 197 | { |
| 198 | // operation_aborted is expected if timer is canceled |
| 199 | if (ec != boost::asio::error::operation_aborted) |
| 200 | { |
| 201 | std::cerr << "Crashdump async_wait failed: " << ec.message() |
| 202 | << "\n"; |
| 203 | } |
| 204 | std::cout << "Crashdump timer canceled\n"; |
| 205 | return; |
| 206 | } |
| 207 | std::cerr << "Crashdump failed to complete before timeout\n"; |
| 208 | crashdumpCompleteMatch.reset(); |
| 209 | }); |
| 210 | |
| 211 | conn->async_method_call( |
| 212 | [](boost::system::error_code ec) { |
| 213 | if (ec) |
| 214 | { |
| 215 | std::cerr << "failed to start Crashdump\n"; |
| 216 | crashdumpTimer.cancel(); |
| 217 | crashdumpCompleteMatch.reset(); |
| 218 | } |
| 219 | }, |
| 220 | "com.intel.crashdump", "/com/intel/crashdump", |
| 221 | "com.intel.crashdump.Stored", "GenerateStoredLog"); |
| 222 | } |
| 223 | |
Jason M. Bills | a15c252 | 2019-08-16 10:01:44 -0700 | [diff] [blame^] | 224 | static void caterrAssertHandler() |
| 225 | { |
| 226 | std::cout << "CPU CATERR detected, starting timer\n"; |
| 227 | caterrAssertTimer.expires_after(std::chrono::milliseconds(caterrTimeoutMs)); |
| 228 | caterrAssertTimer.async_wait([](const boost::system::error_code ec) { |
| 229 | if (ec) |
| 230 | { |
| 231 | // operation_aborted is expected if timer is canceled |
| 232 | // before completion. |
| 233 | if (ec != boost::asio::error::operation_aborted) |
| 234 | { |
| 235 | std::cerr << "caterr timeout async_wait failed: " |
| 236 | << ec.message() << "\n"; |
| 237 | } |
| 238 | std::cout << "CATERR assert timer canceled\n"; |
| 239 | return; |
| 240 | } |
| 241 | std::cout << "CATERR asset timer completed\n"; |
| 242 | conn->async_method_call( |
| 243 | [](boost::system::error_code ec, |
| 244 | const std::variant<bool>& property) { |
| 245 | if (ec) |
| 246 | { |
| 247 | return; |
| 248 | } |
| 249 | const bool* reset = std::get_if<bool>(&property); |
| 250 | if (reset == nullptr) |
| 251 | { |
| 252 | std::cerr << "Unable to read reset on CATERR value\n"; |
| 253 | return; |
| 254 | } |
| 255 | startCrashdumpAndRecovery(*reset); |
| 256 | }, |
| 257 | "xyz.openbmc_project.Settings", |
| 258 | "/xyz/openbmc_project/control/processor_error_config", |
| 259 | "org.freedesktop.DBus.Properties", "Get", |
| 260 | "xyz.openbmc_project.Control.Processor.ErrConfig", "ResetOnCATERR"); |
| 261 | }); |
| 262 | } |
| 263 | |
Jason M. Bills | 1490b14 | 2019-07-01 15:48:43 -0700 | [diff] [blame] | 264 | static void caterrHandler() |
| 265 | { |
| 266 | if (!hostOff) |
| 267 | { |
| 268 | gpiod::line_event gpioLineEvent = caterrLine.event_read(); |
| 269 | |
| 270 | bool caterr = |
| 271 | gpioLineEvent.event_type == gpiod::line_event::FALLING_EDGE; |
| 272 | if (caterr) |
| 273 | { |
Jason M. Bills | a15c252 | 2019-08-16 10:01:44 -0700 | [diff] [blame^] | 274 | caterrAssertHandler(); |
Jason M. Bills | 1490b14 | 2019-07-01 15:48:43 -0700 | [diff] [blame] | 275 | } |
| 276 | else |
| 277 | { |
| 278 | caterrAssertTimer.cancel(); |
| 279 | } |
| 280 | } |
| 281 | caterrEvent.async_wait(boost::asio::posix::stream_descriptor::wait_read, |
| 282 | [](const boost::system::error_code ec) { |
| 283 | if (ec) |
| 284 | { |
| 285 | std::cerr << "caterr handler error: " |
| 286 | << ec.message() << "\n"; |
| 287 | return; |
| 288 | } |
| 289 | caterrHandler(); |
| 290 | }); |
| 291 | } |
Chen Yugang | e6c0f1c | 2019-08-02 20:36:42 +0800 | [diff] [blame] | 292 | static void pchThermtripHandler() |
| 293 | { |
| 294 | if (!hostOff) |
| 295 | { |
| 296 | gpiod::line_event gpioLineEvent = pchThermtripLine.event_read(); |
| 297 | |
| 298 | bool pchThermtrip = |
| 299 | gpioLineEvent.event_type == gpiod::line_event::FALLING_EDGE; |
| 300 | if (pchThermtrip) |
| 301 | { |
| 302 | std::cout << "PCH Thermal trip detected \n"; |
| 303 | // log to redfish, call API |
| 304 | sd_journal_send("MESSAGE=SsbThermalTrip: SSB Thermal trip", |
| 305 | "PRIORITY=%i", LOG_INFO, "REDFISH_MESSAGE_ID=%s", |
| 306 | "OpenBMC.0.1.SsbThermalTrip", NULL); |
| 307 | } |
| 308 | } |
| 309 | pchThermtripEvent.async_wait( |
| 310 | boost::asio::posix::stream_descriptor::wait_read, |
| 311 | [](const boost::system::error_code ec) { |
| 312 | if (ec) |
| 313 | { |
| 314 | std::cerr << "PCH Thermal trip handler error: " << ec.message() |
| 315 | << "\n"; |
| 316 | return; |
| 317 | } |
| 318 | pchThermtripHandler(); |
| 319 | }); |
| 320 | } |
| 321 | |
Jason M. Bills | a15c252 | 2019-08-16 10:01:44 -0700 | [diff] [blame^] | 322 | static void initializeErrorState() |
| 323 | { |
| 324 | // Handle CPU_CATERR if it's asserted now |
| 325 | if (caterrLine.get_value() == 0) |
| 326 | { |
| 327 | caterrAssertHandler(); |
| 328 | } |
| 329 | } |
Jason M. Bills | 1490b14 | 2019-07-01 15:48:43 -0700 | [diff] [blame] | 330 | } // namespace host_error_monitor |
| 331 | |
| 332 | int main(int argc, char* argv[]) |
| 333 | { |
| 334 | // setup connection to dbus |
| 335 | host_error_monitor::conn = |
| 336 | std::make_shared<sdbusplus::asio::connection>(host_error_monitor::io); |
| 337 | |
| 338 | // Host Error Monitor Object |
| 339 | host_error_monitor::conn->request_name( |
| 340 | "xyz.openbmc_project.HostErrorMonitor"); |
| 341 | sdbusplus::asio::object_server server = |
| 342 | sdbusplus::asio::object_server(host_error_monitor::conn); |
| 343 | |
| 344 | // Start tracking host state |
| 345 | std::shared_ptr<sdbusplus::bus::match::match> hostStateMonitor = |
| 346 | host_error_monitor::startHostStateMonitor(); |
| 347 | |
| 348 | // Initialize the host state |
| 349 | host_error_monitor::initializeHostState(); |
| 350 | |
| 351 | // Request CPU_CATERR GPIO events |
| 352 | if (!host_error_monitor::requestGPIOEvents( |
| 353 | "CPU_CATERR", host_error_monitor::caterrHandler, |
| 354 | host_error_monitor::caterrLine, host_error_monitor::caterrEvent)) |
| 355 | { |
| 356 | return -1; |
| 357 | } |
| 358 | |
Chen Yugang | e6c0f1c | 2019-08-02 20:36:42 +0800 | [diff] [blame] | 359 | // Request PCH_BMC_THERMTRIP GPIO events |
| 360 | if (!host_error_monitor::requestGPIOEvents( |
| 361 | "PCH_BMC_THERMTRIP", host_error_monitor::pchThermtripHandler, |
| 362 | host_error_monitor::pchThermtripLine, |
| 363 | host_error_monitor::pchThermtripEvent)) |
| 364 | { |
| 365 | return -1; |
| 366 | } |
| 367 | |
Jason M. Bills | 1490b14 | 2019-07-01 15:48:43 -0700 | [diff] [blame] | 368 | host_error_monitor::io.run(); |
| 369 | |
| 370 | return 0; |
| 371 | } |