Based on kernel version 4.16.1. Page generated on 2018-04-09 11:52 EST.
1 APEI Error INJection 2 ~~~~~~~~~~~~~~~~~~~~ 3 4 EINJ provides a hardware error injection mechanism. It is very useful 5 for debugging and testing APEI and RAS features in general. 6 7 You need to check whether your BIOS supports EINJ first. For that, look 8 for early boot messages similar to this one: 9 10 ACPI: EINJ 0x000000007370A000 000150 (v01 INTEL 00000001 INTL 00000001) 11 12 which shows that the BIOS is exposing an EINJ table - it is the 13 mechanism through which the injection is done. 14 15 Alternatively, look in /sys/firmware/acpi/tables for an "EINJ" file, 16 which is a different representation of the same thing. 17 18 It doesn't necessarily mean that EINJ is not supported if those above 19 don't exist: before you give up, go into BIOS setup to see if the BIOS 20 has an option to enable error injection. Look for something called WHEA 21 or similar. Often, you need to enable an ACPI5 support option prior, in 22 order to see the APEI,EINJ,... functionality supported and exposed by 23 the BIOS menu. 24 25 To use EINJ, make sure the following are options enabled in your kernel 26 configuration: 27 28 CONFIG_DEBUG_FS 29 CONFIG_ACPI_APEI 30 CONFIG_ACPI_APEI_EINJ 31 32 The EINJ user interface is in <debugfs mount point>/apei/einj. 33 34 The following files belong to it: 35 36 - available_error_type 37 38 This file shows which error types are supported: 39 40 Error Type Value Error Description 41 ================ ================= 42 0x00000001 Processor Correctable 43 0x00000002 Processor Uncorrectable non-fatal 44 0x00000004 Processor Uncorrectable fatal 45 0x00000008 Memory Correctable 46 0x00000010 Memory Uncorrectable non-fatal 47 0x00000020 Memory Uncorrectable fatal 48 0x00000040 PCI Express Correctable 49 0x00000080 PCI Express Uncorrectable fatal 50 0x00000100 PCI Express Uncorrectable non-fatal 51 0x00000200 Platform Correctable 52 0x00000400 Platform Uncorrectable non-fatal 53 0x00000800 Platform Uncorrectable fatal 54 55 The format of the file contents are as above, except present are only 56 the available error types. 57 58 - error_type 59 60 Set the value of the error type being injected. Possible error types 61 are defined in the file available_error_type above. 62 63 - error_inject 64 65 Write any integer to this file to trigger the error injection. Make 66 sure you have specified all necessary error parameters, i.e. this 67 write should be the last step when injecting errors. 68 69 - flags 70 71 Present for kernel versions 3.13 and above. Used to specify which 72 of param{1..4} are valid and should be used by the firmware during 73 injection. Value is a bitmask as specified in ACPI5.0 spec for the 74 SET_ERROR_TYPE_WITH_ADDRESS data structure: 75 76 Bit 0 - Processor APIC field valid (see param3 below). 77 Bit 1 - Memory address and mask valid (param1 and param2). 78 Bit 2 - PCIe (seg,bus,dev,fn) valid (see param4 below). 79 80 If set to zero, legacy behavior is mimicked where the type of 81 injection specifies just one bit set, and param1 is multiplexed. 82 83 - param1 84 85 This file is used to set the first error parameter value. Its effect 86 depends on the error type specified in error_type. For example, if 87 error type is memory related type, the param1 should be a valid 88 physical memory address. [Unless "flag" is set - see above] 89 90 - param2 91 92 Same use as param1 above. For example, if error type is of memory 93 related type, then param2 should be a physical memory address mask. 94 Linux requires page or narrower granularity, say, 0xfffffffffffff000. 95 96 - param3 97 98 Used when the 0x1 bit is set in "flags" to specify the APIC id 99 100 - param4 101 Used when the 0x4 bit is set in "flags" to specify target PCIe device 102 103 - notrigger 104 105 The error injection mechanism is a two-step process. First inject the 106 error, then perform some actions to trigger it. Setting "notrigger" 107 to 1 skips the trigger phase, which *may* allow the user to cause the 108 error in some other context by a simple access to the CPU, memory 109 location, or device that is the target of the error injection. Whether 110 this actually works depends on what operations the BIOS actually 111 includes in the trigger phase. 112 113 BIOS versions based on the ACPI 4.0 specification have limited options 114 in controlling where the errors are injected. Your BIOS may support an 115 extension (enabled with the param_extension=1 module parameter, or boot 116 command line einj.param_extension=1). This allows the address and mask 117 for memory injections to be specified by the param1 and param2 files in 118 apei/einj. 119 120 BIOS versions based on the ACPI 5.0 specification have more control over 121 the target of the injection. For processor-related errors (type 0x1, 0x2 122 and 0x4), you can set flags to 0x3 (param3 for bit 0, and param1 and 123 param2 for bit 1) so that you have more information added to the error 124 signature being injected. The actual data passed is this: 125 126 memory_address = param1; 127 memory_address_range = param2; 128 apicid = param3; 129 pcie_sbdf = param4; 130 131 For memory errors (type 0x8, 0x10 and 0x20) the address is set using 132 param1 with a mask in param2 (0x0 is equivalent to all ones). For PCI 133 express errors (type 0x40, 0x80 and 0x100) the segment, bus, device and 134 function are specified using param1: 135 136 31 24 23 16 15 11 10 8 7 0 137 +-------------------------------------------------+ 138 | segment | bus | device | function | reserved | 139 +-------------------------------------------------+ 140 141 Anyway, you get the idea, if there's doubt just take a look at the code 142 in drivers/acpi/apei/einj.c. 143 144 An ACPI 5.0 BIOS may also allow vendor-specific errors to be injected. 145 In this case a file named vendor will contain identifying information 146 from the BIOS that hopefully will allow an application wishing to use 147 the vendor-specific extension to tell that they are running on a BIOS 148 that supports it. All vendor extensions have the 0x80000000 bit set in 149 error_type. A file vendor_flags controls the interpretation of param1 150 and param2 (1 = PROCESSOR, 2 = MEMORY, 4 = PCI). See your BIOS vendor 151 documentation for details (and expect changes to this API if vendors 152 creativity in using this feature expands beyond our expectations). 153 154 155 An error injection example: 156 157 # cd /sys/kernel/debug/apei/einj 158 # cat available_error_type # See which errors can be injected 159 0x00000002 Processor Uncorrectable non-fatal 160 0x00000008 Memory Correctable 161 0x00000010 Memory Uncorrectable non-fatal 162 # echo 0x12345000 > param1 # Set memory address for injection 163 # echo $((-1 << 12)) > param2 # Mask 0xfffffffffffff000 - anywhere in this page 164 # echo 0x8 > error_type # Choose correctable memory error 165 # echo 1 > error_inject # Inject now 166 167 You should see something like this in dmesg: 168 169 [22715.830801] EDAC sbridge MC3: HANDLING MCE MEMORY ERROR 170 [22715.834759] EDAC sbridge MC3: CPU 0: Machine Check Event: 0 Bank 7: 8c00004000010090 171 [22715.834759] EDAC sbridge MC3: TSC 0 172 [22715.834759] EDAC sbridge MC3: ADDR 12345000 EDAC sbridge MC3: MISC 144780c86 173 [22715.834759] EDAC sbridge MC3: PROCESSOR 0:306e7 TIME 1422553404 SOCKET 0 APIC 0 174 [22716.616173] EDAC MC3: 1 CE memory read error on CPU_SrcID#0_Channel#0_DIMM#0 (channel:0 slot:0 page:0x12345 offset:0x0 grain:32 syndrome:0x0 - area:DRAM err_code:0001:0090 socket:0 channel_mask:1 rank:0) 175 176 For more information about EINJ, please refer to ACPI specification 177 version 4.0, section 17.5 and ACPI 5.0, section 18.6.