1 /* $NetBSD: kern_cpu.c,v 1.55 2012/01/29 22:55:40 rmind Exp $ */ 2 3 /*- 4 * Copyright (c) 2007, 2008, 2009, 2010, 2012 The NetBSD Foundation, Inc. 5 * All rights reserved. 6 * 7 * This code is derived from software contributed to The NetBSD Foundation 8 * by Andrew Doran. 9 * 10 * Redistribution and use in source and binary forms, with or without 11 * modification, are permitted provided that the following conditions 12 * are met: 13 * 1. Redistributions of source code must retain the above copyright 14 * notice, this list of conditions and the following disclaimer. 15 * 2. Redistributions in binary form must reproduce the above copyright 16 * notice, this list of conditions and the following disclaimer in the 17 * documentation and/or other materials provided with the distribution. 18 * 19 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS 20 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED 21 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR 22 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS 23 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR 24 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF 25 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 26 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN 27 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) 28 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 29 * POSSIBILITY OF SUCH DAMAGE. 30 */ 31 32 /*- 33 * Copyright (c)2007 YAMAMOTO Takashi, 34 * All rights reserved. 35 * 36 * Redistribution and use in source and binary forms, with or without 37 * modification, are permitted provided that the following conditions 38 * are met: 39 * 1. Redistributions of source code must retain the above copyright 40 * notice, this list of conditions and the following disclaimer. 41 * 2. Redistributions in binary form must reproduce the above copyright 42 * notice, this list of conditions and the following disclaimer in the 43 * documentation and/or other materials provided with the distribution. 44 * 45 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND 46 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 47 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 48 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE 49 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 50 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 51 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 52 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 53 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 54 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 55 * SUCH DAMAGE. 56 */ 57 58 #include <sys/cdefs.h> 59 __KERNEL_RCSID(0, "$NetBSD: kern_cpu.c,v 1.55 2012/01/29 22:55:40 rmind Exp $"); 60 61 #include "opt_cpu_ucode.h" 62 63 #include <sys/param.h> 64 #include <sys/systm.h> 65 #include <sys/idle.h> 66 #include <sys/sched.h> 67 #include <sys/intr.h> 68 #include <sys/conf.h> 69 #include <sys/cpu.h> 70 #include <sys/cpuio.h> 71 #include <sys/proc.h> 72 #include <sys/percpu.h> 73 #include <sys/kernel.h> 74 #include <sys/kauth.h> 75 #include <sys/xcall.h> 76 #include <sys/pool.h> 77 #include <sys/kmem.h> 78 #include <sys/select.h> 79 #include <sys/namei.h> 80 #include <sys/callout.h> 81 82 #include <uvm/uvm_extern.h> 83 84 /* 85 * If the port has stated that cpu_data is the first thing in cpu_info, 86 * verify that the claim is true. This will prevent them from getting out 87 * of sync. 88 */ 89 #ifdef __HAVE_CPU_DATA_FIRST 90 CTASSERT(offsetof(struct cpu_info, ci_data) == 0); 91 #else 92 CTASSERT(offsetof(struct cpu_info, ci_data) != 0); 93 #endif 94 95 void cpuctlattach(int); 96 97 static void cpu_xc_online(struct cpu_info *); 98 static void cpu_xc_offline(struct cpu_info *); 99 100 dev_type_ioctl(cpuctl_ioctl); 101 102 const struct cdevsw cpuctl_cdevsw = { 103 nullopen, nullclose, nullread, nullwrite, cpuctl_ioctl, 104 nullstop, notty, nopoll, nommap, nokqfilter, 105 D_OTHER | D_MPSAFE 106 }; 107 108 kmutex_t cpu_lock __cacheline_aligned; 109 int ncpu __read_mostly; 110 int ncpuonline __read_mostly; 111 bool mp_online __read_mostly; 112 113 /* Note: set on mi_cpu_attach() and idle_loop(). */ 114 kcpuset_t * kcpuset_attached __read_mostly = NULL; 115 kcpuset_t * kcpuset_running __read_mostly = NULL; 116 117 struct cpuqueue cpu_queue __cacheline_aligned 118 = CIRCLEQ_HEAD_INITIALIZER(cpu_queue); 119 120 static struct cpu_info **cpu_infos __read_mostly; 121 122 /* 123 * mi_cpu_init: early initialisation of MI CPU related structures. 124 * 125 * Note: may not block and memory allocator is not yet available. 126 */ 127 void 128 mi_cpu_init(void) 129 { 130 131 mutex_init(&cpu_lock, MUTEX_DEFAULT, IPL_NONE); 132 133 kcpuset_create(&kcpuset_attached, true); 134 kcpuset_create(&kcpuset_running, true); 135 kcpuset_set(kcpuset_running, 0); 136 } 137 138 int 139 mi_cpu_attach(struct cpu_info *ci) 140 { 141 int error; 142 143 KASSERT(maxcpus > 0); 144 145 ci->ci_index = ncpu; 146 kcpuset_set(kcpuset_attached, cpu_index(ci)); 147 148 CIRCLEQ_INSERT_TAIL(&cpu_queue, ci, ci_data.cpu_qchain); 149 TAILQ_INIT(&ci->ci_data.cpu_ld_locks); 150 __cpu_simple_lock_init(&ci->ci_data.cpu_ld_lock); 151 152 /* This is useful for eg, per-cpu evcnt */ 153 snprintf(ci->ci_data.cpu_name, sizeof(ci->ci_data.cpu_name), "cpu%d", 154 cpu_index(ci)); 155 156 if (__predict_false(cpu_infos == NULL)) { 157 cpu_infos = 158 kmem_zalloc(sizeof(cpu_infos[0]) * maxcpus, KM_SLEEP); 159 } 160 cpu_infos[cpu_index(ci)] = ci; 161 162 sched_cpuattach(ci); 163 164 error = create_idle_lwp(ci); 165 if (error != 0) { 166 /* XXX revert sched_cpuattach */ 167 return error; 168 } 169 170 if (ci == curcpu()) 171 ci->ci_data.cpu_onproc = curlwp; 172 else 173 ci->ci_data.cpu_onproc = ci->ci_data.cpu_idlelwp; 174 175 percpu_init_cpu(ci); 176 softint_init(ci); 177 callout_init_cpu(ci); 178 xc_init_cpu(ci); 179 pool_cache_cpu_init(ci); 180 selsysinit(ci); 181 cache_cpu_init(ci); 182 TAILQ_INIT(&ci->ci_data.cpu_biodone); 183 ncpu++; 184 ncpuonline++; 185 186 return 0; 187 } 188 189 void 190 cpuctlattach(int dummy) 191 { 192 193 KASSERT(cpu_infos != NULL); 194 } 195 196 int 197 cpuctl_ioctl(dev_t dev, u_long cmd, void *data, int flag, lwp_t *l) 198 { 199 CPU_INFO_ITERATOR cii; 200 cpustate_t *cs; 201 struct cpu_info *ci; 202 int error, i; 203 u_int id; 204 205 error = 0; 206 207 mutex_enter(&cpu_lock); 208 switch (cmd) { 209 case IOC_CPU_SETSTATE: 210 if (error == 0) 211 cs = data; 212 error = kauth_authorize_system(l->l_cred, 213 KAUTH_SYSTEM_CPU, KAUTH_REQ_SYSTEM_CPU_SETSTATE, cs, NULL, 214 NULL); 215 if (error != 0) 216 break; 217 if (cs->cs_id >= maxcpus || 218 (ci = cpu_lookup(cs->cs_id)) == NULL) { 219 error = ESRCH; 220 break; 221 } 222 error = cpu_setintr(ci, cs->cs_intr); 223 error = cpu_setstate(ci, cs->cs_online); 224 break; 225 226 case IOC_CPU_GETSTATE: 227 if (error == 0) 228 cs = data; 229 id = cs->cs_id; 230 memset(cs, 0, sizeof(*cs)); 231 cs->cs_id = id; 232 if (cs->cs_id >= maxcpus || 233 (ci = cpu_lookup(id)) == NULL) { 234 error = ESRCH; 235 break; 236 } 237 if ((ci->ci_schedstate.spc_flags & SPCF_OFFLINE) != 0) 238 cs->cs_online = false; 239 else 240 cs->cs_online = true; 241 if ((ci->ci_schedstate.spc_flags & SPCF_NOINTR) != 0) 242 cs->cs_intr = false; 243 else 244 cs->cs_intr = true; 245 cs->cs_lastmod = (int32_t)ci->ci_schedstate.spc_lastmod; 246 cs->cs_lastmodhi = (int32_t) 247 (ci->ci_schedstate.spc_lastmod >> 32); 248 cs->cs_intrcnt = cpu_intr_count(ci) + 1; 249 cs->cs_hwid = ci->ci_cpuid; 250 break; 251 252 case IOC_CPU_MAPID: 253 i = 0; 254 for (CPU_INFO_FOREACH(cii, ci)) { 255 if (i++ == *(int *)data) 256 break; 257 } 258 if (ci == NULL) 259 error = ESRCH; 260 else 261 *(int *)data = cpu_index(ci); 262 break; 263 264 case IOC_CPU_GETCOUNT: 265 *(int *)data = ncpu; 266 break; 267 268 #ifdef CPU_UCODE 269 case IOC_CPU_UCODE_GET_VERSION: 270 error = cpu_ucode_get_version(data); 271 break; 272 273 case IOC_CPU_UCODE_APPLY: 274 error = kauth_authorize_machdep(l->l_cred, 275 KAUTH_MACHDEP_CPU_UCODE_APPLY, 276 NULL, NULL, NULL, NULL); 277 if (error != 0) 278 break; 279 error = cpu_ucode_apply(data); 280 break; 281 #endif 282 283 default: 284 error = ENOTTY; 285 break; 286 } 287 mutex_exit(&cpu_lock); 288 289 return error; 290 } 291 292 struct cpu_info * 293 cpu_lookup(u_int idx) 294 { 295 struct cpu_info *ci; 296 297 KASSERT(idx < maxcpus); 298 299 if (__predict_false(cpu_infos == NULL)) { 300 KASSERT(idx == 0); 301 return curcpu(); 302 } 303 304 ci = cpu_infos[idx]; 305 KASSERT(ci == NULL || cpu_index(ci) == idx); 306 307 return ci; 308 } 309 310 static void 311 cpu_xc_offline(struct cpu_info *ci) 312 { 313 struct schedstate_percpu *spc, *mspc = NULL; 314 struct cpu_info *target_ci; 315 struct lwp *l; 316 CPU_INFO_ITERATOR cii; 317 int s; 318 319 /* 320 * Thread that made the cross call (separate context) holds 321 * cpu_lock on our behalf. 322 */ 323 spc = &ci->ci_schedstate; 324 s = splsched(); 325 spc->spc_flags |= SPCF_OFFLINE; 326 splx(s); 327 328 /* Take the first available CPU for the migration. */ 329 for (CPU_INFO_FOREACH(cii, target_ci)) { 330 mspc = &target_ci->ci_schedstate; 331 if ((mspc->spc_flags & SPCF_OFFLINE) == 0) 332 break; 333 } 334 KASSERT(target_ci != NULL); 335 336 /* 337 * Migrate all non-bound threads to the other CPU. Note that this 338 * runs from the xcall thread, thus handling of LSONPROC is not needed. 339 */ 340 mutex_enter(proc_lock); 341 LIST_FOREACH(l, &alllwp, l_list) { 342 struct cpu_info *mci; 343 344 lwp_lock(l); 345 if (l->l_cpu != ci || (l->l_pflag & (LP_BOUND | LP_INTR))) { 346 lwp_unlock(l); 347 continue; 348 } 349 /* Regular case - no affinity. */ 350 if (l->l_affinity == NULL) { 351 lwp_migrate(l, target_ci); 352 continue; 353 } 354 /* Affinity is set, find an online CPU in the set. */ 355 for (CPU_INFO_FOREACH(cii, mci)) { 356 mspc = &mci->ci_schedstate; 357 if ((mspc->spc_flags & SPCF_OFFLINE) == 0 && 358 kcpuset_isset(l->l_affinity, cpu_index(mci))) 359 break; 360 } 361 if (mci == NULL) { 362 lwp_unlock(l); 363 mutex_exit(proc_lock); 364 goto fail; 365 } 366 lwp_migrate(l, mci); 367 } 368 mutex_exit(proc_lock); 369 370 #ifdef __HAVE_MD_CPU_OFFLINE 371 cpu_offline_md(); 372 #endif 373 return; 374 fail: 375 /* Just unset the SPCF_OFFLINE flag, caller will check */ 376 s = splsched(); 377 spc->spc_flags &= ~SPCF_OFFLINE; 378 splx(s); 379 } 380 381 static void 382 cpu_xc_online(struct cpu_info *ci) 383 { 384 struct schedstate_percpu *spc; 385 int s; 386 387 spc = &ci->ci_schedstate; 388 s = splsched(); 389 spc->spc_flags &= ~SPCF_OFFLINE; 390 splx(s); 391 } 392 393 int 394 cpu_setstate(struct cpu_info *ci, bool online) 395 { 396 struct schedstate_percpu *spc; 397 CPU_INFO_ITERATOR cii; 398 struct cpu_info *ci2; 399 uint64_t where; 400 xcfunc_t func; 401 int nonline; 402 403 spc = &ci->ci_schedstate; 404 405 KASSERT(mutex_owned(&cpu_lock)); 406 407 if (online) { 408 if ((spc->spc_flags & SPCF_OFFLINE) == 0) 409 return 0; 410 func = (xcfunc_t)cpu_xc_online; 411 ncpuonline++; 412 } else { 413 if ((spc->spc_flags & SPCF_OFFLINE) != 0) 414 return 0; 415 nonline = 0; 416 /* 417 * Ensure that at least one CPU within the processor set 418 * stays online. Revisit this later. 419 */ 420 for (CPU_INFO_FOREACH(cii, ci2)) { 421 if ((ci2->ci_schedstate.spc_flags & SPCF_OFFLINE) != 0) 422 continue; 423 if (ci2->ci_schedstate.spc_psid != spc->spc_psid) 424 continue; 425 nonline++; 426 } 427 if (nonline == 1) 428 return EBUSY; 429 func = (xcfunc_t)cpu_xc_offline; 430 ncpuonline--; 431 } 432 433 where = xc_unicast(0, func, ci, NULL, ci); 434 xc_wait(where); 435 if (online) { 436 KASSERT((spc->spc_flags & SPCF_OFFLINE) == 0); 437 } else if ((spc->spc_flags & SPCF_OFFLINE) == 0) { 438 /* If was not set offline, then it is busy */ 439 return EBUSY; 440 } 441 442 spc->spc_lastmod = time_second; 443 return 0; 444 } 445 446 #ifdef __HAVE_INTR_CONTROL 447 static void 448 cpu_xc_intr(struct cpu_info *ci) 449 { 450 struct schedstate_percpu *spc; 451 int s; 452 453 spc = &ci->ci_schedstate; 454 s = splsched(); 455 spc->spc_flags &= ~SPCF_NOINTR; 456 splx(s); 457 } 458 459 static void 460 cpu_xc_nointr(struct cpu_info *ci) 461 { 462 struct schedstate_percpu *spc; 463 int s; 464 465 spc = &ci->ci_schedstate; 466 s = splsched(); 467 spc->spc_flags |= SPCF_NOINTR; 468 splx(s); 469 } 470 471 int 472 cpu_setintr(struct cpu_info *ci, bool intr) 473 { 474 struct schedstate_percpu *spc; 475 CPU_INFO_ITERATOR cii; 476 struct cpu_info *ci2; 477 uint64_t where; 478 xcfunc_t func; 479 int nintr; 480 481 spc = &ci->ci_schedstate; 482 483 KASSERT(mutex_owned(&cpu_lock)); 484 485 if (intr) { 486 if ((spc->spc_flags & SPCF_NOINTR) == 0) 487 return 0; 488 func = (xcfunc_t)cpu_xc_intr; 489 } else { 490 if ((spc->spc_flags & SPCF_NOINTR) != 0) 491 return 0; 492 /* 493 * Ensure that at least one CPU within the system 494 * is handing device interrupts. 495 */ 496 nintr = 0; 497 for (CPU_INFO_FOREACH(cii, ci2)) { 498 if ((ci2->ci_schedstate.spc_flags & SPCF_NOINTR) != 0) 499 continue; 500 if (ci2 == ci) 501 continue; 502 nintr++; 503 } 504 if (nintr == 0) 505 return EBUSY; 506 func = (xcfunc_t)cpu_xc_nointr; 507 } 508 509 where = xc_unicast(0, func, ci, NULL, ci); 510 xc_wait(where); 511 if (intr) { 512 KASSERT((spc->spc_flags & SPCF_NOINTR) == 0); 513 } else if ((spc->spc_flags & SPCF_NOINTR) == 0) { 514 /* If was not set offline, then it is busy */ 515 return EBUSY; 516 } 517 518 /* Direct interrupts away from the CPU and record the change. */ 519 cpu_intr_redistribute(); 520 spc->spc_lastmod = time_second; 521 return 0; 522 } 523 #else /* __HAVE_INTR_CONTROL */ 524 int 525 cpu_setintr(struct cpu_info *ci, bool intr) 526 { 527 528 return EOPNOTSUPP; 529 } 530 531 u_int 532 cpu_intr_count(struct cpu_info *ci) 533 { 534 535 return 0; /* 0 == "don't know" */ 536 } 537 #endif /* __HAVE_INTR_CONTROL */ 538 539 bool 540 cpu_softintr_p(void) 541 { 542 543 return (curlwp->l_pflag & LP_INTR) != 0; 544 } 545 546 #ifdef CPU_UCODE 547 int 548 cpu_ucode_load(struct cpu_ucode_softc *sc, const char *fwname) 549 { 550 firmware_handle_t fwh; 551 int error; 552 553 if (sc->sc_blob != NULL) { 554 firmware_free(sc->sc_blob, 0); 555 sc->sc_blob = NULL; 556 sc->sc_blobsize = 0; 557 } 558 559 error = cpu_ucode_md_open(&fwh, fwname); 560 if (error != 0) { 561 aprint_error("ucode: firmware_open failed: %i\n", error); 562 goto err0; 563 } 564 565 sc->sc_blobsize = firmware_get_size(fwh); 566 sc->sc_blob = firmware_malloc(sc->sc_blobsize); 567 if (sc->sc_blob == NULL) { 568 error = ENOMEM; 569 firmware_close(fwh); 570 goto err0; 571 } 572 573 error = firmware_read(fwh, 0, sc->sc_blob, sc->sc_blobsize); 574 firmware_close(fwh); 575 if (error != 0) 576 goto err1; 577 578 return 0; 579 580 err1: 581 firmware_free(sc->sc_blob, 0); 582 sc->sc_blob = NULL; 583 sc->sc_blobsize = 0; 584 err0: 585 return error; 586 } 587 #endif 588