| 1 | /* $NetBSD: uvm_glue.c,v 1.163 2016/05/22 09:10:37 maxv Exp $ */ |
| 2 | |
| 3 | /* |
| 4 | * Copyright (c) 1997 Charles D. Cranor and Washington University. |
| 5 | * Copyright (c) 1991, 1993, The Regents of the University of California. |
| 6 | * |
| 7 | * All rights reserved. |
| 8 | * |
| 9 | * This code is derived from software contributed to Berkeley by |
| 10 | * The Mach Operating System project at Carnegie-Mellon University. |
| 11 | * |
| 12 | * Redistribution and use in source and binary forms, with or without |
| 13 | * modification, are permitted provided that the following conditions |
| 14 | * are met: |
| 15 | * 1. Redistributions of source code must retain the above copyright |
| 16 | * notice, this list of conditions and the following disclaimer. |
| 17 | * 2. Redistributions in binary form must reproduce the above copyright |
| 18 | * notice, this list of conditions and the following disclaimer in the |
| 19 | * documentation and/or other materials provided with the distribution. |
| 20 | * 3. Neither the name of the University nor the names of its contributors |
| 21 | * may be used to endorse or promote products derived from this software |
| 22 | * without specific prior written permission. |
| 23 | * |
| 24 | * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND |
| 25 | * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE |
| 26 | * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE |
| 27 | * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE |
| 28 | * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL |
| 29 | * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS |
| 30 | * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) |
| 31 | * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT |
| 32 | * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY |
| 33 | * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF |
| 34 | * SUCH DAMAGE. |
| 35 | * |
| 36 | * @(#)vm_glue.c 8.6 (Berkeley) 1/5/94 |
| 37 | * from: Id: uvm_glue.c,v 1.1.2.8 1998/02/07 01:16:54 chs Exp |
| 38 | * |
| 39 | * |
| 40 | * Copyright (c) 1987, 1990 Carnegie-Mellon University. |
| 41 | * All rights reserved. |
| 42 | * |
| 43 | * Permission to use, copy, modify and distribute this software and |
| 44 | * its documentation is hereby granted, provided that both the copyright |
| 45 | * notice and this permission notice appear in all copies of the |
| 46 | * software, derivative works or modified versions, and any portions |
| 47 | * thereof, and that both notices appear in supporting documentation. |
| 48 | * |
| 49 | * CARNEGIE MELLON ALLOWS FREE USE OF THIS SOFTWARE IN ITS "AS IS" |
| 50 | * CONDITION. CARNEGIE MELLON DISCLAIMS ANY LIABILITY OF ANY KIND |
| 51 | * FOR ANY DAMAGES WHATSOEVER RESULTING FROM THE USE OF THIS SOFTWARE. |
| 52 | * |
| 53 | * Carnegie Mellon requests users of this software to return to |
| 54 | * |
| 55 | * Software Distribution Coordinator or Software.Distribution@CS.CMU.EDU |
| 56 | * School of Computer Science |
| 57 | * Carnegie Mellon University |
| 58 | * Pittsburgh PA 15213-3890 |
| 59 | * |
| 60 | * any improvements or extensions that they make and grant Carnegie the |
| 61 | * rights to redistribute these changes. |
| 62 | */ |
| 63 | |
| 64 | #include <sys/cdefs.h> |
| 65 | __KERNEL_RCSID(0, "$NetBSD: uvm_glue.c,v 1.163 2016/05/22 09:10:37 maxv Exp $" ); |
| 66 | |
| 67 | #include "opt_kgdb.h" |
| 68 | #include "opt_kstack.h" |
| 69 | #include "opt_uvmhist.h" |
| 70 | |
| 71 | /* |
| 72 | * uvm_glue.c: glue functions |
| 73 | */ |
| 74 | |
| 75 | #include <sys/param.h> |
| 76 | #include <sys/kernel.h> |
| 77 | |
| 78 | #include <sys/systm.h> |
| 79 | #include <sys/proc.h> |
| 80 | #include <sys/resourcevar.h> |
| 81 | #include <sys/buf.h> |
| 82 | #include <sys/syncobj.h> |
| 83 | #include <sys/cpu.h> |
| 84 | #include <sys/atomic.h> |
| 85 | #include <sys/lwp.h> |
| 86 | |
| 87 | #include <uvm/uvm.h> |
| 88 | |
| 89 | /* |
| 90 | * uvm_kernacc: test if kernel can access a memory region. |
| 91 | * |
| 92 | * => Currently used only by /dev/kmem driver (dev/mm.c). |
| 93 | */ |
| 94 | bool |
| 95 | uvm_kernacc(void *addr, size_t len, vm_prot_t prot) |
| 96 | { |
| 97 | vaddr_t saddr = trunc_page((vaddr_t)addr); |
| 98 | vaddr_t eaddr = round_page(saddr + len); |
| 99 | bool rv; |
| 100 | |
| 101 | vm_map_lock_read(kernel_map); |
| 102 | rv = uvm_map_checkprot(kernel_map, saddr, eaddr, prot); |
| 103 | vm_map_unlock_read(kernel_map); |
| 104 | |
| 105 | return rv; |
| 106 | } |
| 107 | |
| 108 | #ifdef KGDB |
| 109 | /* |
| 110 | * Change protections on kernel pages from addr to addr+len |
| 111 | * (presumably so debugger can plant a breakpoint). |
| 112 | * |
| 113 | * We force the protection change at the pmap level. If we were |
| 114 | * to use vm_map_protect a change to allow writing would be lazily- |
| 115 | * applied meaning we would still take a protection fault, something |
| 116 | * we really don't want to do. It would also fragment the kernel |
| 117 | * map unnecessarily. We cannot use pmap_protect since it also won't |
| 118 | * enforce a write-enable request. Using pmap_enter is the only way |
| 119 | * we can ensure the change takes place properly. |
| 120 | */ |
| 121 | void |
| 122 | uvm_chgkprot(void *addr, size_t len, int rw) |
| 123 | { |
| 124 | vm_prot_t prot; |
| 125 | paddr_t pa; |
| 126 | vaddr_t sva, eva; |
| 127 | |
| 128 | prot = rw == B_READ ? VM_PROT_READ : VM_PROT_READ|VM_PROT_WRITE; |
| 129 | eva = round_page((vaddr_t)addr + len); |
| 130 | for (sva = trunc_page((vaddr_t)addr); sva < eva; sva += PAGE_SIZE) { |
| 131 | /* |
| 132 | * Extract physical address for the page. |
| 133 | */ |
| 134 | if (pmap_extract(pmap_kernel(), sva, &pa) == false) |
| 135 | panic("%s: invalid page" , __func__); |
| 136 | pmap_enter(pmap_kernel(), sva, pa, prot, PMAP_WIRED); |
| 137 | } |
| 138 | pmap_update(pmap_kernel()); |
| 139 | } |
| 140 | #endif |
| 141 | |
| 142 | /* |
| 143 | * uvm_vslock: wire user memory for I/O |
| 144 | * |
| 145 | * - called from physio and sys___sysctl |
| 146 | * - XXXCDC: consider nuking this (or making it a macro?) |
| 147 | */ |
| 148 | |
| 149 | int |
| 150 | uvm_vslock(struct vmspace *vs, void *addr, size_t len, vm_prot_t access_type) |
| 151 | { |
| 152 | struct vm_map *map; |
| 153 | vaddr_t start, end; |
| 154 | int error; |
| 155 | |
| 156 | map = &vs->vm_map; |
| 157 | start = trunc_page((vaddr_t)addr); |
| 158 | end = round_page((vaddr_t)addr + len); |
| 159 | error = uvm_fault_wire(map, start, end, access_type, 0); |
| 160 | return error; |
| 161 | } |
| 162 | |
| 163 | /* |
| 164 | * uvm_vsunlock: unwire user memory wired by uvm_vslock() |
| 165 | * |
| 166 | * - called from physio and sys___sysctl |
| 167 | * - XXXCDC: consider nuking this (or making it a macro?) |
| 168 | */ |
| 169 | |
| 170 | void |
| 171 | uvm_vsunlock(struct vmspace *vs, void *addr, size_t len) |
| 172 | { |
| 173 | uvm_fault_unwire(&vs->vm_map, trunc_page((vaddr_t)addr), |
| 174 | round_page((vaddr_t)addr + len)); |
| 175 | } |
| 176 | |
| 177 | /* |
| 178 | * uvm_proc_fork: fork a virtual address space |
| 179 | * |
| 180 | * - the address space is copied as per parent map's inherit values |
| 181 | */ |
| 182 | void |
| 183 | uvm_proc_fork(struct proc *p1, struct proc *p2, bool shared) |
| 184 | { |
| 185 | |
| 186 | if (shared == true) { |
| 187 | p2->p_vmspace = NULL; |
| 188 | uvmspace_share(p1, p2); |
| 189 | } else { |
| 190 | p2->p_vmspace = uvmspace_fork(p1->p_vmspace); |
| 191 | } |
| 192 | |
| 193 | cpu_proc_fork(p1, p2); |
| 194 | } |
| 195 | |
| 196 | /* |
| 197 | * uvm_lwp_fork: fork a thread |
| 198 | * |
| 199 | * - a new PCB structure is allocated for the child process, |
| 200 | * and filled in by MD layer |
| 201 | * - if specified, the child gets a new user stack described by |
| 202 | * stack and stacksize |
| 203 | * - NOTE: the kernel stack may be at a different location in the child |
| 204 | * process, and thus addresses of automatic variables may be invalid |
| 205 | * after cpu_lwp_fork returns in the child process. We do nothing here |
| 206 | * after cpu_lwp_fork returns. |
| 207 | */ |
| 208 | void |
| 209 | uvm_lwp_fork(struct lwp *l1, struct lwp *l2, void *stack, size_t stacksize, |
| 210 | void (*func)(void *), void *arg) |
| 211 | { |
| 212 | |
| 213 | /* Fill stack with magic number. */ |
| 214 | kstack_setup_magic(l2); |
| 215 | |
| 216 | /* |
| 217 | * cpu_lwp_fork() copy and update the pcb, and make the child ready |
| 218 | * to run. If this is a normal user fork, the child will exit |
| 219 | * directly to user mode via child_return() on its first time |
| 220 | * slice and will not return here. If this is a kernel thread, |
| 221 | * the specified entry point will be executed. |
| 222 | */ |
| 223 | cpu_lwp_fork(l1, l2, stack, stacksize, func, arg); |
| 224 | |
| 225 | /* Inactive emap for new LWP. */ |
| 226 | l2->l_emap_gen = UVM_EMAP_INACTIVE; |
| 227 | } |
| 228 | |
| 229 | #ifndef USPACE_ALIGN |
| 230 | #define USPACE_ALIGN 0 |
| 231 | #endif |
| 232 | |
| 233 | static pool_cache_t uvm_uarea_cache; |
| 234 | #if defined(__HAVE_CPU_UAREA_ROUTINES) |
| 235 | static pool_cache_t uvm_uarea_system_cache; |
| 236 | #else |
| 237 | #define uvm_uarea_system_cache uvm_uarea_cache |
| 238 | #endif |
| 239 | |
| 240 | static void * |
| 241 | uarea_poolpage_alloc(struct pool *pp, int flags) |
| 242 | { |
| 243 | #if defined(PMAP_MAP_POOLPAGE) |
| 244 | if (USPACE == PAGE_SIZE && USPACE_ALIGN == 0) { |
| 245 | struct vm_page *pg; |
| 246 | vaddr_t va; |
| 247 | |
| 248 | #if defined(PMAP_ALLOC_POOLPAGE) |
| 249 | pg = PMAP_ALLOC_POOLPAGE( |
| 250 | ((flags & PR_WAITOK) == 0 ? UVM_KMF_NOWAIT : 0)); |
| 251 | #else |
| 252 | pg = uvm_pagealloc(NULL, 0, NULL, |
| 253 | ((flags & PR_WAITOK) == 0 ? UVM_KMF_NOWAIT : 0)); |
| 254 | #endif |
| 255 | if (pg == NULL) |
| 256 | return NULL; |
| 257 | va = PMAP_MAP_POOLPAGE(VM_PAGE_TO_PHYS(pg)); |
| 258 | if (va == 0) |
| 259 | uvm_pagefree(pg); |
| 260 | return (void *)va; |
| 261 | } |
| 262 | #endif |
| 263 | #if defined(__HAVE_CPU_UAREA_ROUTINES) |
| 264 | void *va = cpu_uarea_alloc(false); |
| 265 | if (va) |
| 266 | return (void *)va; |
| 267 | #endif |
| 268 | return (void *)uvm_km_alloc(kernel_map, pp->pr_alloc->pa_pagesz, |
| 269 | USPACE_ALIGN, UVM_KMF_WIRED | |
| 270 | ((flags & PR_WAITOK) ? UVM_KMF_WAITVA : |
| 271 | (UVM_KMF_NOWAIT | UVM_KMF_TRYLOCK))); |
| 272 | } |
| 273 | |
| 274 | static void |
| 275 | uarea_poolpage_free(struct pool *pp, void *addr) |
| 276 | { |
| 277 | #if defined(PMAP_MAP_POOLPAGE) |
| 278 | if (USPACE == PAGE_SIZE && USPACE_ALIGN == 0) { |
| 279 | paddr_t pa; |
| 280 | |
| 281 | pa = PMAP_UNMAP_POOLPAGE((vaddr_t) addr); |
| 282 | KASSERT(pa != 0); |
| 283 | uvm_pagefree(PHYS_TO_VM_PAGE(pa)); |
| 284 | return; |
| 285 | } |
| 286 | #endif |
| 287 | #if defined(__HAVE_CPU_UAREA_ROUTINES) |
| 288 | if (cpu_uarea_free(addr)) |
| 289 | return; |
| 290 | #endif |
| 291 | uvm_km_free(kernel_map, (vaddr_t)addr, pp->pr_alloc->pa_pagesz, |
| 292 | UVM_KMF_WIRED); |
| 293 | } |
| 294 | |
| 295 | static struct pool_allocator uvm_uarea_allocator = { |
| 296 | .pa_alloc = uarea_poolpage_alloc, |
| 297 | .pa_free = uarea_poolpage_free, |
| 298 | .pa_pagesz = USPACE, |
| 299 | }; |
| 300 | |
| 301 | #if defined(__HAVE_CPU_UAREA_ROUTINES) |
| 302 | static void * |
| 303 | uarea_system_poolpage_alloc(struct pool *pp, int flags) |
| 304 | { |
| 305 | void * const va = cpu_uarea_alloc(true); |
| 306 | if (va != NULL) |
| 307 | return va; |
| 308 | |
| 309 | return (void *)uvm_km_alloc(kernel_map, pp->pr_alloc->pa_pagesz, |
| 310 | USPACE_ALIGN, UVM_KMF_WIRED | |
| 311 | ((flags & PR_WAITOK) ? UVM_KMF_WAITVA : |
| 312 | (UVM_KMF_NOWAIT | UVM_KMF_TRYLOCK))); |
| 313 | } |
| 314 | |
| 315 | static void |
| 316 | uarea_system_poolpage_free(struct pool *pp, void *addr) |
| 317 | { |
| 318 | if (cpu_uarea_free(addr)) |
| 319 | return; |
| 320 | |
| 321 | uvm_km_free(kernel_map, (vaddr_t)addr, pp->pr_alloc->pa_pagesz, |
| 322 | UVM_KMF_WIRED); |
| 323 | } |
| 324 | |
| 325 | static struct pool_allocator uvm_uarea_system_allocator = { |
| 326 | .pa_alloc = uarea_system_poolpage_alloc, |
| 327 | .pa_free = uarea_system_poolpage_free, |
| 328 | .pa_pagesz = USPACE, |
| 329 | }; |
| 330 | #endif /* __HAVE_CPU_UAREA_ROUTINES */ |
| 331 | |
| 332 | void |
| 333 | uvm_uarea_init(void) |
| 334 | { |
| 335 | int flags = PR_NOTOUCH; |
| 336 | |
| 337 | /* |
| 338 | * specify PR_NOALIGN unless the alignment provided by |
| 339 | * the backend (USPACE_ALIGN) is sufficient to provide |
| 340 | * pool page size (UPSACE) alignment. |
| 341 | */ |
| 342 | |
| 343 | if ((USPACE_ALIGN == 0 && USPACE != PAGE_SIZE) || |
| 344 | (USPACE_ALIGN % USPACE) != 0) { |
| 345 | flags |= PR_NOALIGN; |
| 346 | } |
| 347 | |
| 348 | uvm_uarea_cache = pool_cache_init(USPACE, USPACE_ALIGN, 0, flags, |
| 349 | "uarea" , &uvm_uarea_allocator, IPL_NONE, NULL, NULL, NULL); |
| 350 | #if defined(__HAVE_CPU_UAREA_ROUTINES) |
| 351 | uvm_uarea_system_cache = pool_cache_init(USPACE, USPACE_ALIGN, |
| 352 | 0, flags, "uareasys" , &uvm_uarea_system_allocator, |
| 353 | IPL_NONE, NULL, NULL, NULL); |
| 354 | #endif |
| 355 | } |
| 356 | |
| 357 | /* |
| 358 | * uvm_uarea_alloc: allocate a u-area |
| 359 | */ |
| 360 | |
| 361 | vaddr_t |
| 362 | uvm_uarea_alloc(void) |
| 363 | { |
| 364 | |
| 365 | return (vaddr_t)pool_cache_get(uvm_uarea_cache, PR_WAITOK); |
| 366 | } |
| 367 | |
| 368 | vaddr_t |
| 369 | uvm_uarea_system_alloc(struct cpu_info *ci) |
| 370 | { |
| 371 | #ifdef __HAVE_CPU_UAREA_ALLOC_IDLELWP |
| 372 | if (__predict_false(ci != NULL)) |
| 373 | return cpu_uarea_alloc_idlelwp(ci); |
| 374 | #endif |
| 375 | |
| 376 | return (vaddr_t)pool_cache_get(uvm_uarea_system_cache, PR_WAITOK); |
| 377 | } |
| 378 | |
| 379 | /* |
| 380 | * uvm_uarea_free: free a u-area |
| 381 | */ |
| 382 | |
| 383 | void |
| 384 | uvm_uarea_free(vaddr_t uaddr) |
| 385 | { |
| 386 | |
| 387 | pool_cache_put(uvm_uarea_cache, (void *)uaddr); |
| 388 | } |
| 389 | |
| 390 | void |
| 391 | uvm_uarea_system_free(vaddr_t uaddr) |
| 392 | { |
| 393 | |
| 394 | pool_cache_put(uvm_uarea_system_cache, (void *)uaddr); |
| 395 | } |
| 396 | |
| 397 | vaddr_t |
| 398 | uvm_lwp_getuarea(lwp_t *l) |
| 399 | { |
| 400 | |
| 401 | return (vaddr_t)l->l_addr - UAREA_PCB_OFFSET; |
| 402 | } |
| 403 | |
| 404 | void |
| 405 | uvm_lwp_setuarea(lwp_t *l, vaddr_t addr) |
| 406 | { |
| 407 | |
| 408 | l->l_addr = (void *)(addr + UAREA_PCB_OFFSET); |
| 409 | } |
| 410 | |
| 411 | /* |
| 412 | * uvm_proc_exit: exit a virtual address space |
| 413 | * |
| 414 | * - borrow proc0's address space because freeing the vmspace |
| 415 | * of the dead process may block. |
| 416 | */ |
| 417 | |
| 418 | void |
| 419 | uvm_proc_exit(struct proc *p) |
| 420 | { |
| 421 | struct lwp *l = curlwp; /* XXX */ |
| 422 | struct vmspace *ovm; |
| 423 | |
| 424 | KASSERT(p == l->l_proc); |
| 425 | ovm = p->p_vmspace; |
| 426 | KASSERT(ovm != NULL); |
| 427 | |
| 428 | if (__predict_false(ovm == proc0.p_vmspace)) |
| 429 | return; |
| 430 | |
| 431 | /* |
| 432 | * borrow proc0's address space. |
| 433 | */ |
| 434 | kpreempt_disable(); |
| 435 | pmap_deactivate(l); |
| 436 | p->p_vmspace = proc0.p_vmspace; |
| 437 | pmap_activate(l); |
| 438 | kpreempt_enable(); |
| 439 | |
| 440 | uvmspace_free(ovm); |
| 441 | } |
| 442 | |
| 443 | void |
| 444 | uvm_lwp_exit(struct lwp *l) |
| 445 | { |
| 446 | vaddr_t va = uvm_lwp_getuarea(l); |
| 447 | bool system = (l->l_flag & LW_SYSTEM) != 0; |
| 448 | |
| 449 | if (system) |
| 450 | uvm_uarea_system_free(va); |
| 451 | else |
| 452 | uvm_uarea_free(va); |
| 453 | #ifdef DIAGNOSTIC |
| 454 | uvm_lwp_setuarea(l, (vaddr_t)NULL); |
| 455 | #endif |
| 456 | } |
| 457 | |
| 458 | /* |
| 459 | * uvm_init_limit: init per-process VM limits |
| 460 | * |
| 461 | * - called for process 0 and then inherited by all others. |
| 462 | */ |
| 463 | |
| 464 | void |
| 465 | uvm_init_limits(struct proc *p) |
| 466 | { |
| 467 | |
| 468 | /* |
| 469 | * Set up the initial limits on process VM. Set the maximum |
| 470 | * resident set size to be all of (reasonably) available memory. |
| 471 | * This causes any single, large process to start random page |
| 472 | * replacement once it fills memory. |
| 473 | */ |
| 474 | |
| 475 | p->p_rlimit[RLIMIT_STACK].rlim_cur = DFLSSIZ; |
| 476 | p->p_rlimit[RLIMIT_STACK].rlim_max = maxsmap; |
| 477 | p->p_rlimit[RLIMIT_DATA].rlim_cur = DFLDSIZ; |
| 478 | p->p_rlimit[RLIMIT_DATA].rlim_max = maxdmap; |
| 479 | p->p_rlimit[RLIMIT_AS].rlim_cur = RLIM_INFINITY; |
| 480 | p->p_rlimit[RLIMIT_AS].rlim_max = RLIM_INFINITY; |
| 481 | p->p_rlimit[RLIMIT_RSS].rlim_cur = MIN( |
| 482 | VM_MAXUSER_ADDRESS, ctob((rlim_t)uvmexp.free)); |
| 483 | } |
| 484 | |
| 485 | /* |
| 486 | * uvm_scheduler: process zero main loop. |
| 487 | */ |
| 488 | |
| 489 | extern struct loadavg averunnable; |
| 490 | |
| 491 | void |
| 492 | uvm_scheduler(void) |
| 493 | { |
| 494 | lwp_t *l = curlwp; |
| 495 | |
| 496 | lwp_lock(l); |
| 497 | l->l_priority = PRI_VM; |
| 498 | l->l_class = SCHED_FIFO; |
| 499 | lwp_unlock(l); |
| 500 | |
| 501 | for (;;) { |
| 502 | sched_pstats(); |
| 503 | (void)kpause("uvm" , false, hz, NULL); |
| 504 | } |
| 505 | } |
| 506 | |