1 The PCI Express Advanced Error Reporting Driver Guide HOWTO 2 T. Long Nguyen <tom.l.nguyen@intel.com> 3 Yanmin Zhang <yanmin.zhang@intel.com> 4 07/29/2006 5 6 71. Overview 8 91.1 About this guide 10 11This guide describes the basics of the PCI Express Advanced Error 12Reporting (AER) driver and provides information on how to use it, as 13well as how to enable the drivers of endpoint devices to conform with 14PCI Express AER driver. 15 161.2 Copyright (C) Intel Corporation 2006. 17 181.3 What is the PCI Express AER Driver? 19 20PCI Express error signaling can occur on the PCI Express link itself 21or on behalf of transactions initiated on the link. PCI Express 22defines two error reporting paradigms: the baseline capability and 23the Advanced Error Reporting capability. The baseline capability is 24required of all PCI Express components providing a minimum defined 25set of error reporting requirements. Advanced Error Reporting 26capability is implemented with a PCI Express advanced error reporting 27extended capability structure providing more robust error reporting. 28 29The PCI Express AER driver provides the infrastructure to support PCI 30Express Advanced Error Reporting capability. The PCI Express AER 31driver provides three basic functions: 32 33- Gathers the comprehensive error information if errors occurred. 34- Reports error to the users. 35- Performs error recovery actions. 36 37AER driver only attaches root ports which support PCI-Express AER 38capability. 39 40 412. User Guide 42 432.1 Include the PCI Express AER Root Driver into the Linux Kernel 44 45The PCI Express AER Root driver is a Root Port service driver attached 46to the PCI Express Port Bus driver. If a user wants to use it, the driver 47has to be compiled. Option CONFIG_PCIEAER supports this capability. It 48depends on CONFIG_PCIEPORTBUS, so pls. set CONFIG_PCIEPORTBUS=y and 49CONFIG_PCIEAER = y. 50 512.2 Load PCI Express AER Root Driver 52 53Some systems have AER support in firmware. Enabling Linux AER support at 54the same time the firmware handles AER may result in unpredictable 55behavior. Therefore, Linux does not handle AER events unless the firmware 56grants AER control to the OS via the ACPI _OSC method. See the PCI FW 3.0 57Specification for details regarding _OSC usage. 58 592.3 AER error output 60 61When a PCIe AER error is captured, an error message will be output to 62console. If it's a correctable error, it is output as a warning. 63Otherwise, it is printed as an error. So users could choose different 64log level to filter out correctable error messages. 65 66Below shows an example: 670000:50:00.0: PCIe Bus Error: severity=Uncorrected (Fatal), type=Transaction Layer, id=0500(Requester ID) 680000:50:00.0: device [8086:0329] error status/mask=00100000/00000000 690000:50:00.0: [20] Unsupported Request (First) 700000:50:00.0: TLP Header: 04000001 00200a03 05010000 00050100 71 72In the example, 'Requester ID' means the ID of the device who sends 73the error message to root port. Pls. refer to pci express specs for 74other fields. 75 762.4 AER Statistics / Counters 77 78When PCIe AER errors are captured, the counters / statistics are also exposed 79in the form of sysfs attributes which are documented at 80Documentation/ABI/testing/sysfs-bus-pci-devices-aer_stats 81 823. Developer Guide 83 84To enable AER aware support requires a software driver to configure 85the AER capability structure within its device and to provide callbacks. 86 87To support AER better, developers need understand how AER does work 88firstly. 89 90PCI Express errors are classified into two types: correctable errors 91and uncorrectable errors. This classification is based on the impacts 92of those errors, which may result in degraded performance or function 93failure. 94 95Correctable errors pose no impacts on the functionality of the 96interface. The PCI Express protocol can recover without any software 97intervention or any loss of data. These errors are detected and 98corrected by hardware. Unlike correctable errors, uncorrectable 99errors impact functionality of the interface. Uncorrectable errors 100can cause a particular transaction or a particular PCI Express link 101to be unreliable. Depending on those error conditions, uncorrectable 102errors are further classified into non-fatal errors and fatal errors. 103Non-fatal errors cause the particular transaction to be unreliable, 104but the PCI Express link itself is fully functional. Fatal errors, on 105the other hand, cause the link to be unreliable. 106 107When AER is enabled, a PCI Express device will automatically send an 108error message to the PCIe root port above it when the device captures 109an error. The Root Port, upon receiving an error reporting message, 110internally processes and logs the error message in its PCI Express 111capability structure. Error information being logged includes storing 112the error reporting agent's requestor ID into the Error Source 113Identification Registers and setting the error bits of the Root Error 114Status Register accordingly. If AER error reporting is enabled in Root 115Error Command Register, the Root Port generates an interrupt if an 116error is detected. 117 118Note that the errors as described above are related to the PCI Express 119hierarchy and links. These errors do not include any device specific 120errors because device specific errors will still get sent directly to 121the device driver. 122 1233.1 Configure the AER capability structure 124 125AER aware drivers of PCI Express component need change the device 126control registers to enable AER. They also could change AER registers, 127including mask and severity registers. Helper function 128pci_enable_pcie_error_reporting could be used to enable AER. See 129section 3.3. 130 1313.2. Provide callbacks 132 1333.2.1 callback reset_link to reset pci express link 134 135This callback is used to reset the pci express physical link when a 136fatal error happens. The root port aer service driver provides a 137default reset_link function, but different upstream ports might 138have different specifications to reset pci express link, so all 139upstream ports should provide their own reset_link functions. 140 141In struct pcie_port_service_driver, a new pointer, reset_link, is 142added. 143 144pci_ers_result_t (*reset_link) (struct pci_dev *dev); 145 146Section 3.2.2.2 provides more detailed info on when to call 147reset_link. 148 1493.2.2 PCI error-recovery callbacks 150 151The PCI Express AER Root driver uses error callbacks to coordinate 152with downstream device drivers associated with a hierarchy in question 153when performing error recovery actions. 154 155Data struct pci_driver has a pointer, err_handler, to point to 156pci_error_handlers who consists of a couple of callback function 157pointers. AER driver follows the rules defined in 158pci-error-recovery.txt except pci express specific parts (e.g. 159reset_link). Pls. refer to pci-error-recovery.txt for detailed 160definitions of the callbacks. 161 162Below sections specify when to call the error callback functions. 163 1643.2.2.1 Correctable errors 165 166Correctable errors pose no impacts on the functionality of 167the interface. The PCI Express protocol can recover without any 168software intervention or any loss of data. These errors do not 169require any recovery actions. The AER driver clears the device's 170correctable error status register accordingly and logs these errors. 171 1723.2.2.2 Non-correctable (non-fatal and fatal) errors 173 174If an error message indicates a non-fatal error, performing link reset 175at upstream is not required. The AER driver calls error_detected(dev, 176pci_channel_io_normal) to all drivers associated within a hierarchy in 177question. for example, 178EndPoint<==>DownstreamPort B<==>UpstreamPort A<==>RootPort. 179If Upstream port A captures an AER error, the hierarchy consists of 180Downstream port B and EndPoint. 181 182A driver may return PCI_ERS_RESULT_CAN_RECOVER, 183PCI_ERS_RESULT_DISCONNECT, or PCI_ERS_RESULT_NEED_RESET, depending on 184whether it can recover or the AER driver calls mmio_enabled as next. 185 186If an error message indicates a fatal error, kernel will broadcast 187error_detected(dev, pci_channel_io_frozen) to all drivers within 188a hierarchy in question. Then, performing link reset at upstream is 189necessary. As different kinds of devices might use different approaches 190to reset link, AER port service driver is required to provide the 191function to reset link. Firstly, kernel looks for if the upstream 192component has an aer driver. If it has, kernel uses the reset_link 193callback of the aer driver. If the upstream component has no aer driver 194and the port is downstream port, we will perform a hot reset as the 195default by setting the Secondary Bus Reset bit of the Bridge Control 196register associated with the downstream port. As for upstream ports, 197they should provide their own aer service drivers with reset_link 198function. If error_detected returns PCI_ERS_RESULT_CAN_RECOVER and 199reset_link returns PCI_ERS_RESULT_RECOVERED, the error handling goes 200to mmio_enabled. 201 2023.3 helper functions 203 2043.3.1 int pci_enable_pcie_error_reporting(struct pci_dev *dev); 205pci_enable_pcie_error_reporting enables the device to send error 206messages to root port when an error is detected. Note that devices 207don't enable the error reporting by default, so device drivers need 208call this function to enable it. 209 2103.3.2 int pci_disable_pcie_error_reporting(struct pci_dev *dev); 211pci_disable_pcie_error_reporting disables the device to send error 212messages to root port when an error is detected. 213 2143.3.3 int pci_cleanup_aer_uncorrect_error_status(struct pci_dev *dev); 215pci_cleanup_aer_uncorrect_error_status cleanups the uncorrectable 216error status register. 217 2183.4 Frequent Asked Questions 219 220Q: What happens if a PCI Express device driver does not provide an 221error recovery handler (pci_driver->err_handler is equal to NULL)? 222 223A: The devices attached with the driver won't be recovered. If the 224error is fatal, kernel will print out warning messages. Please refer 225to section 3 for more information. 226 227Q: What happens if an upstream port service driver does not provide 228callback reset_link? 229 230A: Fatal error recovery will fail if the errors are reported by the 231upstream ports who are attached by the service driver. 232 233Q: How does this infrastructure deal with driver that is not PCI 234Express aware? 235 236A: This infrastructure calls the error callback functions of the 237driver when an error happens. But if the driver is not aware of 238PCI Express, the device might not report its own errors to root 239port. 240 241Q: What modifications will that driver need to make it compatible 242with the PCI Express AER Root driver? 243 244A: It could call the helper functions to enable AER in devices and 245cleanup uncorrectable status register. Pls. refer to section 3.3. 246 247 2484. Software error injection 249 250Debugging PCIe AER error recovery code is quite difficult because it 251is hard to trigger real hardware errors. Software based error 252injection can be used to fake various kinds of PCIe errors. 253 254First you should enable PCIe AER software error injection in kernel 255configuration, that is, following item should be in your .config. 256 257CONFIG_PCIEAER_INJECT=y or CONFIG_PCIEAER_INJECT=m 258 259After reboot with new kernel or insert the module, a device file named 260/dev/aer_inject should be created. 261 262Then, you need a user space tool named aer-inject, which can be gotten 263from: 264 https://git.kernel.org/cgit/linux/kernel/git/gong.chen/aer-inject.git/ 265 266More information about aer-inject can be found in the document comes 267with its source code. 268
