1 /* 2 * CDDL HEADER START 3 * 4 * The contents of this file are subject to the terms of the 5 * Common Development and Distribution License (the "License"). 6 * You may not use this file except in compliance with the License. 7 * 8 * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE 9 * or http://www.opensolaris.org/os/licensing. 10 * See the License for the specific language governing permissions 11 * and limitations under the License. 12 * 13 * When distributing Covered Code, include this CDDL HEADER in each 14 * file and include the License file at usr/src/OPENSOLARIS.LICENSE. 15 * If applicable, add the following below this CDDL HEADER, with the 16 * fields enclosed by brackets "[]" replaced with your own identifying 17 * information: Portions Copyright [yyyy] [name of copyright owner] 18 * 19 * CDDL HEADER END 20 */ 21 22 /* 23 * Copyright 2009 Sun Microsystems, Inc. All rights reserved. 24 * Use is subject to license terms. 25 */ 26 27 #ifndef _AO_H 28 #define _AO_H 29 30 #include <sys/types.h> 31 #include <sys/mc.h> 32 #include <sys/mca_amd.h> 33 #include <sys/mc_amd.h> 34 #include <sys/cpu_module_ms_impl.h> 35 #include <sys/nvpair.h> 36 #include <sys/cyclic.h> 37 #include <sys/errorq.h> 38 #include <sys/kobj.h> 39 #include <sys/fm/util.h> 40 41 #ifdef __cplusplus 42 extern "C" { 43 #endif 44 45 #define AO_MAX_CHIPS 8 46 47 #define AO_MCA_MAX_ERRORS 10 48 49 typedef struct ao_ms_data ao_ms_data_t; 50 51 /* 52 * Rather than using torturous conditionals, we match errors using a table of 53 * ao_error_disp_t's. The members in the ao_error_disp_t are matched against 54 * the value of MCi_STATUS, with a successful match indicating that the given 55 * error occurred. 56 * 57 * While aed_stat_code will match most of the status code bits, a few of the 58 * status code fields are either/or, and are treated separately so as to 59 * minimize the number of ao_error_disp_t structures that must be created. 60 * For example, the dc.tag_par error can have r4 values drd or dwr. Rather 61 * than creating two ao_error_disp_t's, we use the separate aed_stat_r4_bits 62 * field to indicate both AO_MCA_R4_BIT_DRD and AO_MCA_R4_BIT_DWD. As the 63 * matching r4 values are drawn from aed_stat_r4_bits, we don't use the r4 64 * bits in aed_stat_code for matching. Similar reasoning lies behind the 65 * creation of the pp and ii fields. 66 */ 67 #define AO_AED_PANIC_NEVER 0x00 68 #define AO_AED_PANIC_IFMCE 0x01 69 #define AO_AED_PANIC_ALWAYS 0x80 70 71 /* 72 * The AO_AED_F_* flags tell us how to interpret aspects of the error 73 * telemetry, such as which bits of the captured address are valid for 74 * this error. 75 */ 76 /* MCi_ADDR ... */ 77 #define AO_AED_F_LINEAR 0x01 /* is a linear address */ 78 #define AO_AED_F_PHYSICAL 0x02 /* is a physical address */ 79 #define AO_AED_F_PAGEALIGNED 0x04 /* aligns to page size */ 80 #define AO_AED_F_L2SETWAY 0x08 /* 3:0 = way, 15/14/13/12:6 = set */ 81 82 #define AO_AED_FLAGS_ADDRTYPE (AO_AED_F_LINEAR | AO_AED_F_PHYSICAL | \ 83 AO_AED_F_PAGEALIGNED | AO_AED_F_L2SETWAY) 84 85 /* 86 * The AO_AED_ET_* flags group individual error dispositions into 87 * error types. This is used to nominate additional telemetry beyond the 88 * architectural bank registers to capture for this error type. 89 */ 90 #define AO_AED_ET_MEMECC 0x0001 /* Main memory ECC error */ 91 92 typedef struct ao_error_disp { 93 const char *aed_class; /* ereport class for use if match */ 94 uint64_t aed_ereport_members; /* ereport contents flags if match */ 95 uint64_t aed_stat_mask; /* status msr bits for match */ 96 uint64_t aed_stat_mask_res; /* status mask result for match */ 97 uint16_t aed_stat_code; /* status code for match */ 98 uint8_t aed_stat_extcode; /* extended status code for match */ 99 uint8_t aed_stat_pp_bits:4; /* AO_MCA_PP_BIT_* for pp matching */ 100 uint8_t aed_stat_ii_bits:4; /* AO_MCA_II_BIT_* for ii matching */ 101 uint16_t aed_stat_r4_bits; /* AO_MCA_R4_BIT_* for r4 matching */ 102 uint8_t aed_addrvalid_hi; /* most significant valid addr bit */ 103 uint8_t aed_addrvalid_lo; /* least significant valid addr bit */ 104 uint8_t aed_panic_when; /* extra conditions for panic */ 105 uint16_t aed_flags; /* AO_AED_F_* */ 106 uint16_t aed_errtype; /* AO_AED_ET_* */ 107 } ao_error_disp_t; 108 109 /* 110 * We store non-architectutal config as inherited from the BIOS to assist 111 * in troubleshooting. 112 */ 113 struct ao_bios_cfg { 114 uint64_t *bcfg_bank_mask; 115 }; 116 117 /* 118 * The master data structure used to hold MCA-related state. 119 */ 120 typedef struct ao_ms_mca { 121 struct ao_bios_cfg ao_mca_bios_cfg; 122 kmutex_t ao_mca_poll_lock; /* keep pollers from colliding */ 123 uint_t ao_mca_flags; /* AO_MCA_F_* */ 124 } ao_ms_mca_t; 125 126 /* 127 * Per-chip shared state 128 */ 129 struct ao_chipshared { 130 uint32_t aos_chiprev; 131 volatile ulong_t aos_cfgonce; /* Config performed once per chip */ 132 hrtime_t aos_nb_poll_timestamp; 133 cmi_hdl_t aos_nb_poll_owner; 134 uint64_t aos_bcfg_nb_misc; /* BIOS value of MC4_MISC MSR */ 135 uint32_t aos_bcfg_nb_cfg; /* BIOS value of NB MCA Config */ 136 uint32_t aos_bcfg_nb_sparectl; /* BIOS value of Online Spare Control */ 137 uint32_t aos_bcfg_dcfg_lo; /* BIOS value of DRAM Config Low */ 138 uint32_t aos_bcfg_dcfg_hi; /* BIOS value of DRAM Config High */ 139 uint32_t aos_bcfg_scrubctl; /* BIOS value of scrub control */ 140 }; 141 142 /* Bit numbers for once-per-chip operations policed by cms_once */ 143 enum ao_cfgonce_bitnum { 144 AO_CFGONCE_NBMCA, 145 AO_CFGONCE_NBCFG, 146 AO_CFGONCE_DRAMCFG 147 }; 148 149 /* 150 * Per-CPU model-specific state 151 */ 152 struct ao_ms_data { 153 cmi_hdl_t ao_ms_hdl; 154 ao_ms_mca_t ao_ms_mca; 155 struct ao_chipshared *ao_ms_shared; 156 uint64_t ao_ms_hwcr_val; 157 }; 158 159 #ifdef _KERNEL 160 161 struct regs; 162 163 /* 164 * Our cms_ops operations and function prototypes for all non-NULL members. 165 */ 166 extern const cms_ops_t _cms_ops; 167 168 extern int ao_ms_init(cmi_hdl_t, void **); 169 extern void ao_ms_post_startup(cmi_hdl_t); 170 extern void ao_ms_post_mpstartup(cmi_hdl_t); 171 extern uint64_t ao_ms_mcgctl_val(cmi_hdl_t, int, uint64_t); 172 extern boolean_t ao_ms_bankctl_skipinit(cmi_hdl_t, int); 173 extern uint64_t ao_ms_bankctl_val(cmi_hdl_t, int, uint64_t); 174 extern void ao_ms_mca_init(cmi_hdl_t, int); 175 extern uint64_t ao_ms_poll_ownermask(cmi_hdl_t, hrtime_t); 176 extern uint32_t ao_ms_error_action(cmi_hdl_t, int, int, uint64_t, 177 uint64_t, uint64_t, void *); 178 extern cms_cookie_t ao_ms_disp_match(cmi_hdl_t, int, uint64_t, uint64_t, 179 uint64_t, void *); 180 extern void ao_ms_ereport_class(cmi_hdl_t, cms_cookie_t, const char **, 181 const char **); 182 extern boolean_t ao_ms_ereport_includestack(cmi_hdl_t, cms_cookie_t); 183 extern void ao_ms_ereport_add_logout(cmi_hdl_t, nvlist_t *, 184 nv_alloc_t *, int, uint64_t, uint64_t, uint64_t, void *, void *); 185 extern cms_errno_t ao_ms_msrinject(cmi_hdl_t, uint_t, uint64_t); 186 187 /* 188 * Local functions 189 */ 190 extern void ao_procnode_scrubber_enable(cmi_hdl_t, ao_ms_data_t *); 191 extern void ao_pcicfg_write(uint_t, uint_t, uint_t, uint32_t); 192 extern uint32_t ao_pcicfg_read(uint_t, uint_t, uint_t); 193 extern void ao_bankstatus_prewrite(cmi_hdl_t, ao_ms_data_t *); 194 extern void ao_bankstatus_postwrite(cmi_hdl_t, ao_ms_data_t *); 195 196 #endif /* _KERNEL */ 197 198 #ifdef __cplusplus 199 } 200 #endif 201 202 #endif /* _AO_H */ 203