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 | */ |
| 16 | #include <boost/asio/posix/stream_descriptor.hpp> |
| 17 | #include <gpiod.hpp> |
| 18 | #include <iostream> |
| 19 | #include <sdbusplus/asio/object_server.hpp> |
| 20 | |
| 21 | namespace host_error_monitor |
| 22 | { |
| 23 | static boost::asio::io_service io; |
| 24 | static std::shared_ptr<sdbusplus::asio::connection> conn; |
| 25 | |
| 26 | static bool hostOff = true; |
| 27 | |
| 28 | const static constexpr size_t caterrTimeoutMs = 2000; |
| 29 | const static constexpr size_t crashdumpTimeoutS = 300; |
| 30 | |
| 31 | // Timers |
| 32 | // Timer for CATERR asserted |
| 33 | static boost::asio::steady_timer caterrAssertTimer(io); |
| 34 | |
| 35 | // GPIO Lines and Event Descriptors |
| 36 | static gpiod::line caterrLine; |
| 37 | static boost::asio::posix::stream_descriptor caterrEvent(io); |
| 38 | |
| 39 | static void initializeHostState() |
| 40 | { |
| 41 | conn->async_method_call( |
| 42 | [](boost::system::error_code ec, |
| 43 | const std::variant<std::string>& property) { |
| 44 | if (ec) |
| 45 | { |
| 46 | return; |
| 47 | } |
| 48 | const std::string* state = std::get_if<std::string>(&property); |
| 49 | if (state == nullptr) |
| 50 | { |
| 51 | std::cerr << "Unable to read host state value\n"; |
| 52 | return; |
| 53 | } |
| 54 | hostOff = *state == "xyz.openbmc_project.State.Host.HostState.Off"; |
| 55 | }, |
| 56 | "xyz.openbmc_project.State.Host", "/xyz/openbmc_project/state/host0", |
| 57 | "org.freedesktop.DBus.Properties", "Get", |
| 58 | "xyz.openbmc_project.State.Host", "CurrentHostState"); |
| 59 | } |
| 60 | |
| 61 | static std::shared_ptr<sdbusplus::bus::match::match> startHostStateMonitor() |
| 62 | { |
| 63 | return std::make_shared<sdbusplus::bus::match::match>( |
| 64 | *conn, |
| 65 | "type='signal',interface='org.freedesktop.DBus.Properties'," |
| 66 | "member='PropertiesChanged',arg0namespace='xyz.openbmc_project.State." |
| 67 | "Host'", |
| 68 | [](sdbusplus::message::message& msg) { |
| 69 | std::string interfaceName; |
| 70 | boost::container::flat_map<std::string, std::variant<std::string>> |
| 71 | propertiesChanged; |
| 72 | std::string state; |
| 73 | try |
| 74 | { |
| 75 | msg.read(interfaceName, propertiesChanged); |
| 76 | state = |
| 77 | std::get<std::string>(propertiesChanged.begin()->second); |
| 78 | } |
| 79 | catch (std::exception& e) |
| 80 | { |
| 81 | std::cerr << "Unable to read host state\n"; |
| 82 | return; |
| 83 | } |
| 84 | hostOff = state == "xyz.openbmc_project.State.Host.HostState.Off"; |
| 85 | |
| 86 | // No host events should fire while off, so cancel any pending |
| 87 | // timers |
| 88 | if (hostOff) |
| 89 | { |
| 90 | caterrAssertTimer.cancel(); |
| 91 | } |
| 92 | }); |
| 93 | } |
| 94 | |
| 95 | static bool requestGPIOEvents( |
| 96 | const std::string& name, const std::function<void()>& handler, |
| 97 | gpiod::line& gpioLine, |
| 98 | boost::asio::posix::stream_descriptor& gpioEventDescriptor) |
| 99 | { |
| 100 | // Find the GPIO line |
| 101 | gpioLine = gpiod::find_line(name); |
| 102 | if (!gpioLine) |
| 103 | { |
| 104 | std::cerr << "Failed to find the " << name << " line\n"; |
| 105 | return false; |
| 106 | } |
| 107 | |
| 108 | try |
| 109 | { |
| 110 | gpioLine.request( |
| 111 | {"host-error-monitor", gpiod::line_request::EVENT_BOTH_EDGES}); |
| 112 | } |
| 113 | catch (std::exception&) |
| 114 | { |
| 115 | std::cerr << "Failed to request events for " << name << "\n"; |
| 116 | return false; |
| 117 | } |
| 118 | |
| 119 | int gpioLineFd = gpioLine.event_get_fd(); |
| 120 | if (gpioLineFd < 0) |
| 121 | { |
| 122 | std::cerr << "Failed to get " << name << " fd\n"; |
| 123 | return false; |
| 124 | } |
| 125 | |
| 126 | gpioEventDescriptor.assign(gpioLineFd); |
| 127 | |
| 128 | gpioEventDescriptor.async_wait( |
| 129 | boost::asio::posix::stream_descriptor::wait_read, |
| 130 | [&name, handler](const boost::system::error_code ec) { |
| 131 | if (ec) |
| 132 | { |
| 133 | std::cerr << name << " fd handler error: " << ec.message() |
| 134 | << "\n"; |
| 135 | return; |
| 136 | } |
| 137 | handler(); |
| 138 | }); |
| 139 | return true; |
| 140 | } |
| 141 | |
| 142 | static void startPowerCycle() |
| 143 | { |
| 144 | conn->async_method_call( |
| 145 | [](boost::system::error_code ec) { |
| 146 | if (ec) |
| 147 | { |
| 148 | std::cerr << "failed to set Chassis State\n"; |
| 149 | } |
| 150 | }, |
| 151 | "xyz.openbmc_project.State.Chassis", |
| 152 | "/xyz/openbmc_project/state/chassis0", |
| 153 | "org.freedesktop.DBus.Properties", "Set", |
| 154 | "xyz.openbmc_project.State.Chassis", "RequestedPowerTransition", |
| 155 | std::variant<std::string>{ |
| 156 | "xyz.openbmc_project.State.Chassis.Transition.PowerCycle"}); |
| 157 | } |
| 158 | |
| 159 | static void startCrashdumpAndRecovery(bool recoverSystem) |
| 160 | { |
| 161 | std::cout << "Starting crashdump\n"; |
| 162 | static std::shared_ptr<sdbusplus::bus::match::match> crashdumpCompleteMatch; |
| 163 | static boost::asio::steady_timer crashdumpTimer(io); |
| 164 | |
| 165 | crashdumpCompleteMatch = std::make_shared<sdbusplus::bus::match::match>( |
| 166 | *conn, |
| 167 | "type='signal',interface='org.freedesktop.DBus.Properties'," |
| 168 | "member='PropertiesChanged',arg0namespace='com.intel.crashdump'", |
| 169 | [recoverSystem](sdbusplus::message::message& msg) { |
| 170 | crashdumpTimer.cancel(); |
| 171 | std::cout << "Crashdump completed\n"; |
| 172 | if (recoverSystem) |
| 173 | { |
| 174 | std::cout << "Recovering the system\n"; |
| 175 | startPowerCycle(); |
| 176 | } |
| 177 | crashdumpCompleteMatch.reset(); |
| 178 | }); |
| 179 | |
| 180 | crashdumpTimer.expires_after(std::chrono::seconds(crashdumpTimeoutS)); |
| 181 | crashdumpTimer.async_wait([](const boost::system::error_code ec) { |
| 182 | if (ec) |
| 183 | { |
| 184 | // operation_aborted is expected if timer is canceled |
| 185 | if (ec != boost::asio::error::operation_aborted) |
| 186 | { |
| 187 | std::cerr << "Crashdump async_wait failed: " << ec.message() |
| 188 | << "\n"; |
| 189 | } |
| 190 | std::cout << "Crashdump timer canceled\n"; |
| 191 | return; |
| 192 | } |
| 193 | std::cerr << "Crashdump failed to complete before timeout\n"; |
| 194 | crashdumpCompleteMatch.reset(); |
| 195 | }); |
| 196 | |
| 197 | conn->async_method_call( |
| 198 | [](boost::system::error_code ec) { |
| 199 | if (ec) |
| 200 | { |
| 201 | std::cerr << "failed to start Crashdump\n"; |
| 202 | crashdumpTimer.cancel(); |
| 203 | crashdumpCompleteMatch.reset(); |
| 204 | } |
| 205 | }, |
| 206 | "com.intel.crashdump", "/com/intel/crashdump", |
| 207 | "com.intel.crashdump.Stored", "GenerateStoredLog"); |
| 208 | } |
| 209 | |
| 210 | static void caterrHandler() |
| 211 | { |
| 212 | if (!hostOff) |
| 213 | { |
| 214 | gpiod::line_event gpioLineEvent = caterrLine.event_read(); |
| 215 | |
| 216 | bool caterr = |
| 217 | gpioLineEvent.event_type == gpiod::line_event::FALLING_EDGE; |
| 218 | if (caterr) |
| 219 | { |
| 220 | std::cout << "CPU CATERR detected, starting timer\n"; |
| 221 | caterrAssertTimer.expires_after( |
| 222 | std::chrono::milliseconds(caterrTimeoutMs)); |
| 223 | caterrAssertTimer.async_wait( |
| 224 | [](const boost::system::error_code ec) { |
| 225 | if (ec) |
| 226 | { |
| 227 | // operation_aborted is expected if timer is canceled |
| 228 | // before completion. |
| 229 | if (ec != boost::asio::error::operation_aborted) |
| 230 | { |
| 231 | std::cerr << "caterr timeout async_wait failed: " |
| 232 | << ec.message() << "\n"; |
| 233 | } |
| 234 | std::cout << "CATERR assert timer canceled\n"; |
| 235 | return; |
| 236 | } |
| 237 | std::cout << "CATERR asset timer completed\n"; |
| 238 | conn->async_method_call( |
| 239 | [](boost::system::error_code ec, |
| 240 | const std::variant<bool>& property) { |
| 241 | if (ec) |
| 242 | { |
| 243 | return; |
| 244 | } |
| 245 | const bool* reset = std::get_if<bool>(&property); |
| 246 | if (reset == nullptr) |
| 247 | { |
| 248 | std::cerr |
| 249 | << "Unable to read reset on CATERR value\n"; |
| 250 | return; |
| 251 | } |
| 252 | startCrashdumpAndRecovery(*reset); |
| 253 | }, |
| 254 | "xyz.openbmc_project.Settings", |
| 255 | "/xyz/openbmc_project/control/processor_error_config", |
| 256 | "org.freedesktop.DBus.Properties", "Get", |
| 257 | "xyz.openbmc_project.Control.Processor.ErrConfig", |
| 258 | "ResetOnCATERR"); |
| 259 | }); |
| 260 | } |
| 261 | else |
| 262 | { |
| 263 | caterrAssertTimer.cancel(); |
| 264 | } |
| 265 | } |
| 266 | caterrEvent.async_wait(boost::asio::posix::stream_descriptor::wait_read, |
| 267 | [](const boost::system::error_code ec) { |
| 268 | if (ec) |
| 269 | { |
| 270 | std::cerr << "caterr handler error: " |
| 271 | << ec.message() << "\n"; |
| 272 | return; |
| 273 | } |
| 274 | caterrHandler(); |
| 275 | }); |
| 276 | } |
| 277 | } // namespace host_error_monitor |
| 278 | |
| 279 | int main(int argc, char* argv[]) |
| 280 | { |
| 281 | // setup connection to dbus |
| 282 | host_error_monitor::conn = |
| 283 | std::make_shared<sdbusplus::asio::connection>(host_error_monitor::io); |
| 284 | |
| 285 | // Host Error Monitor Object |
| 286 | host_error_monitor::conn->request_name( |
| 287 | "xyz.openbmc_project.HostErrorMonitor"); |
| 288 | sdbusplus::asio::object_server server = |
| 289 | sdbusplus::asio::object_server(host_error_monitor::conn); |
| 290 | |
| 291 | // Start tracking host state |
| 292 | std::shared_ptr<sdbusplus::bus::match::match> hostStateMonitor = |
| 293 | host_error_monitor::startHostStateMonitor(); |
| 294 | |
| 295 | // Initialize the host state |
| 296 | host_error_monitor::initializeHostState(); |
| 297 | |
| 298 | // Request CPU_CATERR GPIO events |
| 299 | if (!host_error_monitor::requestGPIOEvents( |
| 300 | "CPU_CATERR", host_error_monitor::caterrHandler, |
| 301 | host_error_monitor::caterrLine, host_error_monitor::caterrEvent)) |
| 302 | { |
| 303 | return -1; |
| 304 | } |
| 305 | |
| 306 | host_error_monitor::io.run(); |
| 307 | |
| 308 | return 0; |
| 309 | } |