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 2008 Sun Microsystems, Inc. All rights reserved. 24 * Use is subject to license terms. 25 */ 26 27 #include <cma.h> 28 29 #include <strings.h> 30 #include <errno.h> 31 #include <time.h> 32 #include <fm/fmd_api.h> 33 #include <sys/fm/protocol.h> 34 #include <sys/systeminfo.h> 35 #include <sys/utsname.h> 36 37 #ifdef sun4v 38 #include <sys/fm/ldom.h> 39 40 static fmd_hdl_t *init_hdl; 41 ldom_hdl_t *cma_lhp; 42 #endif 43 44 extern const char *fmd_fmri_get_platform(); 45 46 cma_t cma; 47 48 cma_stats_t cma_stats = { 49 { "cpu_flts", FMD_TYPE_UINT64, "cpu faults resolved" }, 50 { "cpu_repairs", FMD_TYPE_UINT64, "cpu faults repaired" }, 51 { "cpu_fails", FMD_TYPE_UINT64, "cpu faults unresolveable" }, 52 { "cpu_blfails", FMD_TYPE_UINT64, "failed cpu blacklists" }, 53 { "cpu_supp", FMD_TYPE_UINT64, "cpu offlines suppressed" }, 54 { "cpu_blsupp", FMD_TYPE_UINT64, "cpu blacklists suppressed" }, 55 { "page_flts", FMD_TYPE_UINT64, "page faults resolved" }, 56 { "page_repairs", FMD_TYPE_UINT64, "page faults repaired" }, 57 { "page_fails", FMD_TYPE_UINT64, "page faults unresolveable" }, 58 { "page_supp", FMD_TYPE_UINT64, "page retires suppressed" }, 59 { "page_nonent", FMD_TYPE_UINT64, "retires for non-existent fmris" }, 60 { "bad_flts", FMD_TYPE_UINT64, "invalid fault events received" }, 61 { "nop_flts", FMD_TYPE_UINT64, "inapplicable fault events received" }, 62 { "auto_flts", FMD_TYPE_UINT64, "auto-close faults received" } 63 }; 64 65 typedef struct cma_subscriber { 66 const char *subr_class; 67 const char *subr_sname; 68 uint_t subr_svers; 69 int (*subr_func)(fmd_hdl_t *, nvlist_t *, nvlist_t *, const char *, 70 boolean_t); 71 } cma_subscriber_t; 72 73 static const cma_subscriber_t cma_subrs[] = { 74 #if defined(sun4v) 75 { "fault.memory.page", FM_FMRI_SCHEME_MEM, FM_MEM_SCHEME_VERSION, 76 cma_page_retire }, 77 { "fault.memory.dimm", FM_FMRI_SCHEME_MEM, FM_MEM_SCHEME_VERSION, 78 NULL }, 79 { "fault.memory.dimm_sb", FM_FMRI_SCHEME_MEM, FM_MEM_SCHEME_VERSION, 80 NULL }, 81 { "fault.memory.dimm_ck", FM_FMRI_SCHEME_MEM, FM_MEM_SCHEME_VERSION, 82 NULL }, 83 { "fault.memory.dimm_ue", FM_FMRI_SCHEME_MEM, FM_MEM_SCHEME_VERSION, 84 NULL }, 85 { "fault.memory.bank", FM_FMRI_SCHEME_MEM, FM_MEM_SCHEME_VERSION, 86 NULL }, 87 { "fault.memory.datapath", FM_FMRI_SCHEME_MEM, FM_MEM_SCHEME_VERSION, 88 NULL }, 89 { "fault.memory.link-c", FM_FMRI_SCHEME_MEM, FM_MEM_SCHEME_VERSION, 90 NULL }, 91 { "fault.memory.link-u", FM_FMRI_SCHEME_MEM, FM_MEM_SCHEME_VERSION, 92 NULL }, 93 { "fault.memory.link-f", FM_FMRI_SCHEME_MEM, FM_MEM_SCHEME_VERSION, 94 NULL }, 95 96 /* 97 * The following ultraSPARC-T1/T2 faults do NOT retire a cpu thread, 98 * and therefore must be intercepted before 99 * the default "fault.cpu.*" dispatch to cma_cpu_retire. 100 */ 101 { "fault.cpu.*.l2cachedata", FM_FMRI_SCHEME_CPU, 102 FM_CPU_SCHEME_VERSION, NULL }, 103 { "fault.cpu.*.l2cachetag", FM_FMRI_SCHEME_CPU, 104 FM_CPU_SCHEME_VERSION, NULL }, 105 { "fault.cpu.*.l2cachectl", FM_FMRI_SCHEME_CPU, 106 FM_CPU_SCHEME_VERSION, NULL }, 107 { "fault.cpu.*.l2data-c", FM_FMRI_SCHEME_CPU, 108 FM_CPU_SCHEME_VERSION, NULL }, 109 { "fault.cpu.*.l2data-u", FM_FMRI_SCHEME_CPU, 110 FM_CPU_SCHEME_VERSION, NULL }, 111 { "fault.cpu.*.mau", FM_FMRI_SCHEME_CPU, 112 FM_CPU_SCHEME_VERSION, NULL }, 113 { "fault.cpu.*.lfu-u", FM_FMRI_SCHEME_CPU, 114 FM_CPU_SCHEME_VERSION, NULL }, 115 { "fault.cpu.*.lfu-f", FM_FMRI_SCHEME_CPU, 116 FM_CPU_SCHEME_VERSION, NULL }, 117 { "fault.cpu.*.lfu-p", FM_FMRI_SCHEME_CPU, 118 FM_CPU_SCHEME_VERSION, NULL }, 119 { "fault.cpu.*", FM_FMRI_SCHEME_CPU, FM_CPU_SCHEME_VERSION, 120 cma_cpu_retire }, 121 #elif defined(opl) 122 { "fault.memory.page", FM_FMRI_SCHEME_MEM, FM_MEM_SCHEME_VERSION, 123 cma_page_retire }, 124 { "fault.memory.dimm", FM_FMRI_SCHEME_MEM, FM_MEM_SCHEME_VERSION, 125 NULL }, 126 { "fault.memory.bank", FM_FMRI_SCHEME_MEM, FM_MEM_SCHEME_VERSION, 127 NULL }, 128 { "fault.cpu.SPARC64-VI.*", FM_FMRI_SCHEME_CPU, FM_CPU_SCHEME_VERSION, 129 cma_cpu_retire }, 130 { "fault.cpu.SPARC64-VII.*", FM_FMRI_SCHEME_CPU, FM_CPU_SCHEME_VERSION, 131 cma_cpu_retire }, 132 { "fault.chassis.SPARC-Enterprise.cpu.SPARC64-VI.core.se", 133 FM_FMRI_SCHEME_HC, FM_HC_SCHEME_VERSION, cma_cpu_hc_retire }, 134 { "fault.chassis.SPARC-Enterprise.cpu.SPARC64-VI.core.se-offlinereq", 135 FM_FMRI_SCHEME_HC, FM_HC_SCHEME_VERSION, cma_cpu_hc_retire }, 136 { "fault.chassis.SPARC-Enterprise.cpu.SPARC64-VI.core.ce", 137 FM_FMRI_SCHEME_HC, FM_HC_SCHEME_VERSION, cma_cpu_hc_retire }, 138 { "fault.chassis.SPARC-Enterprise.cpu.SPARC64-VI.core.ce-offlinereq", 139 FM_FMRI_SCHEME_HC, FM_HC_SCHEME_VERSION, cma_cpu_hc_retire }, 140 { "fault.chassis.SPARC-Enterprise.cpu.SPARC64-VII.core.se", 141 FM_FMRI_SCHEME_HC, FM_HC_SCHEME_VERSION, cma_cpu_hc_retire }, 142 { "fault.chassis.SPARC-Enterprise.cpu.SPARC64-VII.core.se-offlinereq", 143 FM_FMRI_SCHEME_HC, FM_HC_SCHEME_VERSION, cma_cpu_hc_retire }, 144 { "fault.chassis.SPARC-Enterprise.cpu.SPARC64-VII.core.ce", 145 FM_FMRI_SCHEME_HC, FM_HC_SCHEME_VERSION, cma_cpu_hc_retire }, 146 { "fault.chassis.SPARC-Enterprise.cpu.SPARC64-VII.core.ce-offlinereq", 147 FM_FMRI_SCHEME_HC, FM_HC_SCHEME_VERSION, cma_cpu_hc_retire }, 148 #else /* Generic */ 149 { "fault.memory.page", FM_FMRI_SCHEME_MEM, FM_MEM_SCHEME_VERSION, 150 cma_page_retire }, 151 { "fault.memory.page_sb", FM_FMRI_SCHEME_MEM, FM_MEM_SCHEME_VERSION, 152 cma_page_retire }, 153 { "fault.memory.page_ck", FM_FMRI_SCHEME_MEM, FM_MEM_SCHEME_VERSION, 154 cma_page_retire }, 155 { "fault.memory.page_ue", FM_FMRI_SCHEME_MEM, FM_MEM_SCHEME_VERSION, 156 cma_page_retire }, 157 { "fault.memory.generic-x86.page_ce", FM_FMRI_SCHEME_MEM, 158 FM_MEM_SCHEME_VERSION, cma_page_retire }, 159 { "fault.memory.generic-x86.page_ue", FM_FMRI_SCHEME_MEM, 160 FM_MEM_SCHEME_VERSION, cma_page_retire }, 161 { "fault.memory.intel.page_ce", FM_FMRI_SCHEME_MEM, 162 FM_MEM_SCHEME_VERSION, cma_page_retire }, 163 { "fault.memory.intel.page_ue", FM_FMRI_SCHEME_MEM, 164 FM_MEM_SCHEME_VERSION, cma_page_retire }, 165 { "fault.memory.dimm", FM_FMRI_SCHEME_MEM, FM_MEM_SCHEME_VERSION, 166 NULL }, 167 { "fault.memory.dimm_sb", FM_FMRI_SCHEME_MEM, FM_MEM_SCHEME_VERSION, 168 NULL }, 169 { "fault.memory.dimm_ck", FM_FMRI_SCHEME_MEM, FM_MEM_SCHEME_VERSION, 170 NULL }, 171 { "fault.memory.dimm_ue", FM_FMRI_SCHEME_MEM, FM_MEM_SCHEME_VERSION, 172 NULL }, 173 { "fault.memory.generic-x86.dimm_ce", FM_FMRI_SCHEME_MEM, 174 FM_MEM_SCHEME_VERSION, NULL }, 175 { "fault.memory.generic-x86.dimm_ue", FM_FMRI_SCHEME_MEM, 176 FM_MEM_SCHEME_VERSION, NULL }, 177 { "fault.memory.intel.dimm_ce", FM_FMRI_SCHEME_MEM, 178 FM_MEM_SCHEME_VERSION, NULL }, 179 { "fault.memory.intel.dimm_ue", FM_FMRI_SCHEME_MEM, 180 FM_MEM_SCHEME_VERSION, NULL }, 181 { "fault.memory.intel.fbd.*", FM_FMRI_SCHEME_HC, 182 FM_HC_SCHEME_VERSION, NULL }, 183 { "fault.memory.dimm_testfail", FM_FMRI_SCHEME_MEM, 184 FM_MEM_SCHEME_VERSION, NULL }, 185 { "fault.memory.bank", FM_FMRI_SCHEME_MEM, FM_MEM_SCHEME_VERSION, 186 NULL }, 187 { "fault.memory.datapath", FM_FMRI_SCHEME_MEM, FM_MEM_SCHEME_VERSION, 188 NULL }, 189 190 /* 191 * The following faults do NOT retire a cpu thread, 192 * and therefore must be intercepted before 193 * the default "fault.cpu.*" dispatch to cma_cpu_retire. 194 */ 195 { "fault.cpu.ultraSPARC-T1.freg", FM_FMRI_SCHEME_CPU, 196 FM_CPU_SCHEME_VERSION, NULL }, 197 { "fault.cpu.ultraSPARC-T1.l2cachedata", FM_FMRI_SCHEME_CPU, 198 FM_CPU_SCHEME_VERSION, NULL }, 199 { "fault.cpu.ultraSPARC-T1.l2cachetag", FM_FMRI_SCHEME_CPU, 200 FM_CPU_SCHEME_VERSION, NULL }, 201 { "fault.cpu.ultraSPARC-T1.l2cachectl", FM_FMRI_SCHEME_CPU, 202 FM_CPU_SCHEME_VERSION, NULL }, 203 { "fault.cpu.ultraSPARC-T1.mau", FM_FMRI_SCHEME_CPU, 204 FM_CPU_SCHEME_VERSION, NULL }, 205 { "fault.cpu.amd.dramchannel", FM_FMRI_SCHEME_HC, FM_HC_SCHEME_VERSION, 206 NULL }, 207 { "fault.cpu.generic-x86.bus_interconnect_memory", FM_FMRI_SCHEME_CPU, 208 FM_CPU_SCHEME_VERSION, NULL }, 209 { "fault.cpu.generic-x86.bus_interconnect_io", FM_FMRI_SCHEME_CPU, 210 FM_CPU_SCHEME_VERSION, NULL }, 211 { "fault.cpu.generic-x86.bus_interconnect", FM_FMRI_SCHEME_CPU, 212 FM_CPU_SCHEME_VERSION, NULL }, 213 { "fault.cpu.intel.bus_interconnect_memory", FM_FMRI_SCHEME_CPU, 214 FM_CPU_SCHEME_VERSION, NULL }, 215 { "fault.cpu.intel.bus_interconnect_io", FM_FMRI_SCHEME_CPU, 216 FM_CPU_SCHEME_VERSION, NULL }, 217 { "fault.cpu.intel.bus_interconnect", FM_FMRI_SCHEME_CPU, 218 FM_CPU_SCHEME_VERSION, NULL }, 219 { "fault.cpu.intel.nb.*", FM_FMRI_SCHEME_HC, 220 FM_HC_SCHEME_VERSION, NULL }, 221 { "fault.cpu.intel.quickpath.*", FM_FMRI_SCHEME_HC, 222 FM_HC_SCHEME_VERSION, NULL }, 223 { "fault.cpu.generic-x86.mc", FM_FMRI_SCHEME_HC, 224 FM_HC_SCHEME_VERSION, NULL }, 225 { "fault.cpu.intel.dma", FM_FMRI_SCHEME_HC, 226 FM_HC_SCHEME_VERSION, NULL }, 227 { "fault.cpu.intel.dma", FM_FMRI_SCHEME_CPU, 228 FM_CPU_SCHEME_VERSION, NULL }, 229 #ifndef sun4v 230 #ifndef opl 231 { "fault.cpu.ultraSPARC-IVplus.l2cachedata-line", 232 FM_FMRI_SCHEME_CPU, FM_CPU_SCHEME_VERSION, 233 cma_cache_way_retire }, 234 { "fault.cpu.ultraSPARC-IVplus.l3cachedata-line", 235 FM_FMRI_SCHEME_CPU, FM_CPU_SCHEME_VERSION, 236 cma_cache_way_retire }, 237 { "fault.cpu.ultraSPARC-IVplus.l2cachetag-line", 238 FM_FMRI_SCHEME_CPU, FM_CPU_SCHEME_VERSION, 239 cma_cache_way_retire }, 240 { "fault.cpu.ultraSPARC-IVplus.l3cachetag-line", 241 FM_FMRI_SCHEME_CPU, FM_CPU_SCHEME_VERSION, 242 cma_cache_way_retire }, 243 #endif 244 #endif 245 /* 246 * Default "fault.cpu.*" for "mem" scheme ASRU dispatch. 247 */ 248 { "fault.cpu.*", FM_FMRI_SCHEME_CPU, FM_CPU_SCHEME_VERSION, 249 cma_cpu_retire }, 250 #endif 251 { NULL, NULL, 0, NULL } 252 }; 253 254 static const cma_subscriber_t * 255 nvl2subr(fmd_hdl_t *hdl, nvlist_t *nvl, nvlist_t **asrup) 256 { 257 const cma_subscriber_t *sp; 258 nvlist_t *asru; 259 char *scheme; 260 uint8_t version; 261 char *fltclass = "(unknown)"; 262 boolean_t retire; 263 264 if (nvlist_lookup_boolean_value(nvl, FM_SUSPECT_RETIRE, &retire) == 0 && 265 retire == 0) { 266 fmd_hdl_debug(hdl, "cma_recv: retire suppressed"); 267 return (NULL); 268 } 269 270 if (nvlist_lookup_nvlist(nvl, FM_FAULT_ASRU, &asru) != 0 || 271 nvlist_lookup_string(asru, FM_FMRI_SCHEME, &scheme) != 0 || 272 nvlist_lookup_uint8(asru, FM_VERSION, &version) != 0) { 273 cma_stats.bad_flts.fmds_value.ui64++; 274 return (NULL); 275 } 276 277 for (sp = cma_subrs; sp->subr_class != NULL; sp++) { 278 if (fmd_nvl_class_match(hdl, nvl, sp->subr_class) && 279 strcmp(scheme, sp->subr_sname) == 0 && 280 version <= sp->subr_svers) { 281 *asrup = asru; 282 return (sp); 283 } 284 } 285 286 (void) nvlist_lookup_string(nvl, FM_CLASS, &fltclass); 287 fmd_hdl_error(hdl, "No handling disposition for %s with asru in " 288 "scheme \"%s\"\n", fltclass, scheme); 289 cma_stats.nop_flts.fmds_value.ui64++; 290 return (NULL); 291 } 292 293 static void 294 cma_recv_list(fmd_hdl_t *hdl, nvlist_t *nvl, const char *class) 295 { 296 char *uuid = NULL; 297 nvlist_t **nva; 298 uint_t nvc = 0; 299 uint_t keepopen; 300 int err = 0; 301 nvlist_t *asru; 302 uint32_t index; 303 304 err |= nvlist_lookup_string(nvl, FM_SUSPECT_UUID, &uuid); 305 err |= nvlist_lookup_nvlist_array(nvl, FM_SUSPECT_FAULT_LIST, 306 &nva, &nvc); 307 if (err != 0) { 308 cma_stats.bad_flts.fmds_value.ui64++; 309 return; 310 } 311 312 keepopen = nvc; 313 while (nvc-- != 0 && (strcmp(class, FM_LIST_SUSPECT_CLASS) != 0 || 314 !fmd_case_uuclosed(hdl, uuid))) { 315 nvlist_t *nvl = *nva++; 316 const cma_subscriber_t *subr; 317 int has_fault; 318 319 if ((subr = nvl2subr(hdl, nvl, &asru)) == NULL) 320 continue; 321 322 /* 323 * A handler returns CMA_RA_SUCCESS to indicate that 324 * from this suspects point-of-view the case may be 325 * closed, CMA_RA_FAILURE otherwise. 326 * A handler must not close the case itself. 327 */ 328 if (subr->subr_func != NULL) { 329 has_fault = fmd_nvl_fmri_has_fault(hdl, asru, 330 FMD_HAS_FAULT_ASRU, NULL); 331 if (strcmp(class, FM_LIST_SUSPECT_CLASS) == 0) { 332 if (has_fault == 1) 333 err = subr->subr_func(hdl, nvl, asru, 334 uuid, 0); 335 } else { 336 if (has_fault == 0) 337 err = subr->subr_func(hdl, nvl, asru, 338 uuid, 1); 339 } 340 if (err == CMA_RA_SUCCESS) 341 keepopen--; 342 } 343 } 344 /* 345 * Do not close the case if we are handling cache faults. 346 */ 347 if (nvlist_lookup_uint32(asru, FM_FMRI_CPU_CACHE_INDEX, &index) != 0) { 348 if (!keepopen && strcmp(class, FM_LIST_SUSPECT_CLASS) == 0) { 349 fmd_case_uuclose(hdl, uuid); 350 } 351 } 352 if (!keepopen && strcmp(class, FM_LIST_REPAIRED_CLASS) == 0) 353 fmd_case_uuresolved(hdl, uuid); 354 } 355 356 static void 357 cma_recv_one(fmd_hdl_t *hdl, nvlist_t *nvl) 358 { 359 const cma_subscriber_t *subr; 360 nvlist_t *asru; 361 362 if ((subr = nvl2subr(hdl, nvl, &asru)) == NULL) 363 return; 364 365 if (subr->subr_func != NULL) { 366 if (fmd_nvl_fmri_has_fault(hdl, asru, 367 FMD_HAS_FAULT_ASRU, NULL) == 1) 368 (void) subr->subr_func(hdl, nvl, asru, NULL, 0); 369 } 370 } 371 372 /*ARGSUSED*/ 373 static void 374 cma_recv(fmd_hdl_t *hdl, fmd_event_t *ep, nvlist_t *nvl, const char *class) 375 { 376 fmd_hdl_debug(hdl, "received %s\n", class); 377 378 if (strcmp(class, FM_LIST_SUSPECT_CLASS) == 0 || 379 strcmp(class, FM_LIST_REPAIRED_CLASS) == 0 || 380 strcmp(class, FM_LIST_UPDATED_CLASS) == 0) 381 cma_recv_list(hdl, nvl, class); 382 else 383 cma_recv_one(hdl, nvl); 384 } 385 386 /*ARGSUSED*/ 387 static void 388 cma_timeout(fmd_hdl_t *hdl, id_t id, void *arg) 389 { 390 if (id == cma.cma_page_timerid) 391 cma_page_retry(hdl); 392 #ifdef sun4v 393 /* 394 * cpu offline/online needs to be retried on sun4v because 395 * ldom request can be asynchronous. 396 */ 397 else if (id == cma.cma_cpu_timerid) 398 cma_cpu_retry(hdl); 399 #endif 400 } 401 402 #ifdef sun4v 403 static void * 404 cma_init_alloc(size_t size) 405 { 406 return (fmd_hdl_alloc(init_hdl, size, FMD_SLEEP)); 407 } 408 409 static void 410 cma_init_free(void *addr, size_t size) 411 { 412 fmd_hdl_free(init_hdl, addr, size); 413 } 414 #endif 415 416 static const fmd_hdl_ops_t fmd_ops = { 417 cma_recv, /* fmdo_recv */ 418 cma_timeout, /* fmdo_timeout */ 419 NULL, /* fmdo_close */ 420 NULL, /* fmdo_stats */ 421 NULL, /* fmdo_gc */ 422 }; 423 424 static const fmd_prop_t fmd_props[] = { 425 { "cpu_tries", FMD_TYPE_UINT32, "10" }, 426 { "cpu_delay", FMD_TYPE_TIME, "1sec" }, 427 #ifdef sun4v 428 { "cpu_ret_mindelay", FMD_TYPE_TIME, "5sec" }, 429 { "cpu_ret_maxdelay", FMD_TYPE_TIME, "5min" }, 430 #endif /* sun4v */ 431 { "cpu_offline_enable", FMD_TYPE_BOOL, "true" }, 432 { "cpu_online_enable", FMD_TYPE_BOOL, "true" }, 433 { "cpu_forced_offline", FMD_TYPE_BOOL, "true" }, 434 #ifdef opl 435 { "cpu_blacklist_enable", FMD_TYPE_BOOL, "false" }, 436 { "cpu_unblacklist_enable", FMD_TYPE_BOOL, "false" }, 437 #else 438 { "cpu_blacklist_enable", FMD_TYPE_BOOL, "true" }, 439 { "cpu_unblacklist_enable", FMD_TYPE_BOOL, "true" }, 440 #endif /* opl */ 441 { "page_ret_mindelay", FMD_TYPE_TIME, "1sec" }, 442 { "page_ret_maxdelay", FMD_TYPE_TIME, "5min" }, 443 { "page_retire_enable", FMD_TYPE_BOOL, "true" }, 444 { "page_unretire_enable", FMD_TYPE_BOOL, "true" }, 445 { NULL, 0, NULL } 446 }; 447 448 static const fmd_hdl_info_t fmd_info = { 449 "CPU/Memory Retire Agent", CMA_VERSION, &fmd_ops, fmd_props 450 }; 451 452 void 453 _fmd_init(fmd_hdl_t *hdl) 454 { 455 hrtime_t nsec; 456 #ifdef i386 457 /* 458 * Abort the cpumem-retire module if Solaris is running under the Xen 459 * hypervisor. 460 */ 461 if (strcmp(fmd_fmri_get_platform(), "i86xpv") == 0) 462 return; 463 #endif 464 465 if (fmd_hdl_register(hdl, FMD_API_VERSION, &fmd_info) != 0) 466 return; /* invalid data in configuration file */ 467 468 fmd_hdl_subscribe(hdl, "fault.cpu.*"); 469 fmd_hdl_subscribe(hdl, "fault.memory.*"); 470 #ifdef opl 471 fmd_hdl_subscribe(hdl, "fault.chassis.SPARC-Enterprise.cpu.*"); 472 #endif 473 474 (void) fmd_stat_create(hdl, FMD_STAT_NOALLOC, sizeof (cma_stats) / 475 sizeof (fmd_stat_t), (fmd_stat_t *)&cma_stats); 476 477 cma.cma_cpu_tries = fmd_prop_get_int32(hdl, "cpu_tries"); 478 479 nsec = fmd_prop_get_int64(hdl, "cpu_delay"); 480 cma.cma_cpu_delay.tv_sec = nsec / NANOSEC; 481 cma.cma_cpu_delay.tv_nsec = nsec % NANOSEC; 482 483 cma.cma_page_mindelay = fmd_prop_get_int64(hdl, "page_ret_mindelay"); 484 cma.cma_page_maxdelay = fmd_prop_get_int64(hdl, "page_ret_maxdelay"); 485 486 #ifdef sun4v 487 cma.cma_cpu_mindelay = fmd_prop_get_int64(hdl, "cpu_ret_mindelay"); 488 cma.cma_cpu_maxdelay = fmd_prop_get_int64(hdl, "cpu_ret_maxdelay"); 489 #endif 490 491 cma.cma_cpu_dooffline = fmd_prop_get_int32(hdl, "cpu_offline_enable"); 492 cma.cma_cpu_forcedoffline = fmd_prop_get_int32(hdl, 493 "cpu_forced_offline"); 494 cma.cma_cpu_doonline = fmd_prop_get_int32(hdl, "cpu_online_enable"); 495 cma.cma_cpu_doblacklist = fmd_prop_get_int32(hdl, 496 "cpu_blacklist_enable"); 497 cma.cma_cpu_dounblacklist = fmd_prop_get_int32(hdl, 498 "cpu_unblacklist_enable"); 499 cma.cma_page_doretire = fmd_prop_get_int32(hdl, "page_retire_enable"); 500 cma.cma_page_dounretire = fmd_prop_get_int32(hdl, 501 "page_unretire_enable"); 502 503 if (cma.cma_page_maxdelay < cma.cma_page_mindelay) 504 fmd_hdl_abort(hdl, "page retirement delays conflict\n"); 505 506 #ifdef sun4v 507 init_hdl = hdl; 508 cma_lhp = ldom_init(cma_init_alloc, cma_init_free); 509 #endif 510 } 511 512 void 513 _fmd_fini(fmd_hdl_t *hdl) 514 { 515 #ifdef sun4v 516 ldom_fini(cma_lhp); 517 cma_cpu_fini(hdl); 518 #endif 519 cma_page_fini(hdl); 520 } 521