1 /*- 2 * SPDX-License-Identifier: BSD-2-Clause-FreeBSD 3 * 4 * Copyright (c) 2012 Hudson River Trading LLC 5 * Written by: John H. Baldwin <jhb@FreeBSD.org> 6 * All rights reserved. 7 * 8 * Redistribution and use in source and binary forms, with or without 9 * modification, are permitted provided that the following conditions 10 * are met: 11 * 1. Redistributions of source code must retain the above copyright 12 * notice, this list of conditions and the following disclaimer. 13 * 2. Redistributions in binary form must reproduce the above copyright 14 * notice, this list of conditions and the following disclaimer in the 15 * documentation and/or other materials provided with the distribution. 16 * 17 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND 18 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 19 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 20 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE 21 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 22 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 23 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 24 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 25 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 26 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 27 * SUCH DAMAGE. 28 */ 29 30 #ifndef lint 31 static const char rcsid[] = 32 "$FreeBSD$"; 33 #endif /* not lint */ 34 35 #include <sys/param.h> 36 #include <sys/pciio.h> 37 38 #include <err.h> 39 #include <stdio.h> 40 41 #include <dev/pci/pcireg.h> 42 43 #include "pciconf.h" 44 45 struct bit_table { 46 uint32_t mask; 47 const char *desc; 48 }; 49 50 /* Error indicators in the PCI status register (PCIR_STATUS). */ 51 static struct bit_table pci_status[] = { 52 { PCIM_STATUS_MDPERR, "Master Data Parity Error" }, 53 { PCIM_STATUS_STABORT, "Sent Target-Abort" }, 54 { PCIM_STATUS_RTABORT, "Received Target-Abort" }, 55 { PCIM_STATUS_RMABORT, "Received Master-Abort" }, 56 { PCIM_STATUS_SERR, "Signalled System Error" }, 57 { PCIM_STATUS_PERR, "Detected Parity Error" }, 58 { 0, NULL }, 59 }; 60 61 /* Valid error indicator bits in PCIR_STATUS. */ 62 #define PCI_ERRORS (PCIM_STATUS_MDPERR | PCIM_STATUS_STABORT | \ 63 PCIM_STATUS_RTABORT | PCIM_STATUS_RMABORT | \ 64 PCIM_STATUS_SERR | PCIM_STATUS_PERR) 65 66 /* Error indicators in the PCI-Express device status register. */ 67 static struct bit_table pcie_device_status[] = { 68 { PCIEM_STA_CORRECTABLE_ERROR, "Correctable Error Detected" }, 69 { PCIEM_STA_NON_FATAL_ERROR, "Non-Fatal Error Detected" }, 70 { PCIEM_STA_FATAL_ERROR, "Fatal Error Detected" }, 71 { PCIEM_STA_UNSUPPORTED_REQ, "Unsupported Request Detected" }, 72 { 0, NULL }, 73 }; 74 75 /* Valid error indicator bits in the PCI-Express device status register. */ 76 #define PCIE_ERRORS (PCIEM_STA_CORRECTABLE_ERROR | \ 77 PCIEM_STA_NON_FATAL_ERROR | \ 78 PCIEM_STA_FATAL_ERROR | \ 79 PCIEM_STA_UNSUPPORTED_REQ) 80 81 /* AER Uncorrected errors. */ 82 static struct bit_table aer_uc[] = { 83 { PCIM_AER_UC_TRAINING_ERROR, "Link Training Error" }, 84 { PCIM_AER_UC_DL_PROTOCOL_ERROR, "Data Link Protocol Error" }, 85 { PCIM_AER_UC_SURPRISE_LINK_DOWN, "Surprise Link Down Error" }, 86 { PCIM_AER_UC_POISONED_TLP, "Poisoned TLP" }, 87 { PCIM_AER_UC_FC_PROTOCOL_ERROR, "Flow Control Protocol Error" }, 88 { PCIM_AER_UC_COMPLETION_TIMEOUT, "Completion Timeout" }, 89 { PCIM_AER_UC_COMPLETER_ABORT, "Completer Abort" }, 90 { PCIM_AER_UC_UNEXPECTED_COMPLETION, "Unexpected Completion" }, 91 { PCIM_AER_UC_RECEIVER_OVERFLOW, "Receiver Overflow Error" }, 92 { PCIM_AER_UC_MALFORMED_TLP, "Malformed TLP" }, 93 { PCIM_AER_UC_ECRC_ERROR, "ECRC Error" }, 94 { PCIM_AER_UC_UNSUPPORTED_REQUEST, "Unsupported Request" }, 95 { PCIM_AER_UC_ACS_VIOLATION, "ACS Violation" }, 96 { PCIM_AER_UC_INTERNAL_ERROR, "Uncorrectable Internal Error" }, 97 { PCIM_AER_UC_MC_BLOCKED_TLP, "MC Blocked TLP" }, 98 { PCIM_AER_UC_ATOMIC_EGRESS_BLK, "AtomicOp Egress Blocked" }, 99 { PCIM_AER_UC_TLP_PREFIX_BLOCKED, "TLP Prefix Blocked Error" }, 100 { 0, NULL }, 101 }; 102 103 /* AER Corrected errors. */ 104 static struct bit_table aer_cor[] = { 105 { PCIM_AER_COR_RECEIVER_ERROR, "Receiver Error" }, 106 { PCIM_AER_COR_BAD_TLP, "Bad TLP" }, 107 { PCIM_AER_COR_BAD_DLLP, "Bad DLLP" }, 108 { PCIM_AER_COR_REPLAY_ROLLOVER, "REPLAY_NUM Rollover" }, 109 { PCIM_AER_COR_REPLAY_TIMEOUT, "Replay Timer Timeout" }, 110 { PCIM_AER_COR_ADVISORY_NF_ERROR, "Advisory Non-Fatal Error" }, 111 { PCIM_AER_COR_INTERNAL_ERROR, "Corrected Internal Error" }, 112 { PCIM_AER_COR_HEADER_LOG_OVFLOW, "Header Log Overflow" }, 113 { 0, NULL }, 114 }; 115 116 static void 117 print_bits(const char *header, struct bit_table *table, uint32_t mask) 118 { 119 int first; 120 121 first = 1; 122 for (; table->desc != NULL; table++) 123 if (mask & table->mask) { 124 if (first) { 125 printf("%14s = ", header); 126 first = 0; 127 } else 128 printf(" "); 129 printf("%s\n", table->desc); 130 mask &= ~table->mask; 131 } 132 if (mask != 0) { 133 if (first) 134 printf("%14s = ", header); 135 else 136 printf(" "); 137 printf("Unknown: 0x%08x\n", mask); 138 } 139 } 140 141 void 142 list_errors(int fd, struct pci_conf *p) 143 { 144 uint32_t mask, severity; 145 uint16_t sta, aer; 146 uint8_t pcie; 147 148 /* First check for standard PCI errors. */ 149 sta = read_config(fd, &p->pc_sel, PCIR_STATUS, 2); 150 print_bits("PCI errors", pci_status, sta & PCI_ERRORS); 151 152 /* See if this is a PCI-express device. */ 153 pcie = pci_find_cap(fd, p, PCIY_EXPRESS); 154 if (pcie == 0) 155 return; 156 157 /* Check for PCI-e errors. */ 158 sta = read_config(fd, &p->pc_sel, pcie + PCIER_DEVICE_STA, 2); 159 print_bits("PCI-e errors", pcie_device_status, sta & PCIE_ERRORS); 160 161 /* See if this device supports AER. */ 162 aer = pcie_find_cap(fd, p, PCIZ_AER); 163 if (aer == 0) 164 return; 165 166 /* Check for uncorrected errors. */ 167 mask = read_config(fd, &p->pc_sel, aer + PCIR_AER_UC_STATUS, 4); 168 severity = read_config(fd, &p->pc_sel, aer + PCIR_AER_UC_SEVERITY, 4); 169 print_bits("Fatal", aer_uc, mask & severity); 170 print_bits("Non-fatal", aer_uc, mask & ~severity); 171 172 /* Check for corrected errors. */ 173 mask = read_config(fd, &p->pc_sel, aer + PCIR_AER_COR_STATUS, 4); 174 print_bits("Corrected", aer_cor, mask); 175 } 176