bsd/vm/dp_backing_file.c

   1 /*
   2  * Copyright (c) 2000-2007 Apple Inc. All rights reserved.
   3  *
   4  * @APPLE_OSREFERENCE_LICENSE_HEADER_START@
   5  *
   6  * This file contains Original Code and/or Modifications of Original Code
   7  * as defined in and that are subject to the Apple Public Source License
   8  * Version 2.0 (the 'License'). You may not use this file except in
   9  * compliance with the License. The rights granted to you under the License
  10  * may not be used to create, or enable the creation or redistribution of,
  11  * unlawful or unlicensed copies of an Apple operating system, or to
  12  * circumvent, violate, or enable the circumvention or violation of, any
  13  * terms of an Apple operating system software license agreement.
  14  *
  15  * Please obtain a copy of the License at
  16  * http://www.opensource.apple.com/apsl/ and read it before using this file.
  17  *
  18  * The Original Code and all software distributed under the License are
  19  * distributed on an 'AS IS' basis, WITHOUT WARRANTY OF ANY KIND, EITHER
  20  * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
  21  * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,
  22  * FITNESS FOR A PARTICULAR PURPOSE, QUIET ENJOYMENT OR NON-INFRINGEMENT.
  23  * Please see the License for the specific language governing rights and
  24  * limitations under the License.
  25  *
  26  * @APPLE_OSREFERENCE_LICENSE_HEADER_END@
  27  */
  28 /*
  29  * NOTICE: This file was modified by SPARTA, Inc. in 2005 to introduce
  30  * support for mandatory and extensible security protections.  This notice
  31  * is included in support of clause 2.2 (b) of the Apple Public License,
  32  * Version 2.0.
  33  */
  34
  35 #include <sys/param.h>
  36 #include <sys/systm.h>
  37 #include <sys/lock.h>
  38 #include <sys/proc_internal.h>
  39 #include <sys/kauth.h>
  40 #include <sys/buf.h>
  41 #include <sys/uio.h>
  42 #include <sys/vnode_internal.h>
  43 #include <sys/namei.h>
  44 #include <sys/ubc_internal.h>
  45 #include <sys/mount_internal.h>
  46 #include <sys/malloc.h>
  47
  48 #include <default_pager/default_pager_types.h>
  49 #include <default_pager/default_pager_object.h>
  50
  51 #include <security/audit/audit.h>
  52 #include <bsm/audit_kevents.h>
  53
  54 #include <mach/mach_types.h>
  55 #include <mach/host_priv.h>
  56 #include <mach/mach_traps.h>
  57 #include <mach/boolean.h>
  58
  59 #include <kern/kern_types.h>
  60 #include <kern/host.h>
  61 #include <kern/task.h>
  62 #include <kern/zalloc.h>
  63 #include <kern/kalloc.h>
  64 #include <kern/assert.h>
  65
  66 #include <libkern/libkern.h>
  67
  68 #include <vm/vm_pageout.h>
  69 #include <vm/vm_map.h>
  70 #include <vm/vm_kern.h>
  71 #include <vm/vnode_pager.h>
  72 #include <vm/vm_protos.h>
  73 #if CONFIG_MACF
  74 #include <security/mac_framework.h>
  75 #endif
  76
  77 /*
  78  * temporary support for delayed instantiation
  79  * of default_pager
  80  */
  81 int default_pager_init_flag = 0;
  82
  83 struct bs_map           bs_port_table[MAX_BACKING_STORE] = {
  84         {0,0},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},
  85         {0,0},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},
  86         {0,0},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},
  87         {0,0},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},
  88         {0,0},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},
  89         {0,0},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},
  90         {0,0},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},
  91         {0,0},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},
  92         {0,0},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},
  93         {0,0},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0}};
  94
  95 /* ###################################################### */
  96
  97
  98 /*
  99  *      Routine:        macx_backing_store_recovery
 100  *      Function:
 101  *              Syscall interface to set a tasks privilege
 102  *              level so that it is not subject to
 103  *              macx_backing_store_suspend
 104  */
 105 int
 106 macx_backing_store_recovery(
 107         struct macx_backing_store_recovery_args *args)
 108 {
 109         int             pid = args->pid;
 110         int             error;
 111         struct proc     *p =  current_proc();
 112         boolean_t       funnel_state;
 113
 114         funnel_state = thread_funnel_set(kernel_flock, TRUE);
 115         if ((error = suser(kauth_cred_get(), 0)))
 116                 goto backing_store_recovery_return;
 117
 118         /* for now restrict backing_store_recovery */
 119         /* usage to only present task */
 120         if(pid != proc_selfpid()) {
 121                 error = EINVAL;
 122                 goto backing_store_recovery_return;
 123         }
 124
 125         task_backing_store_privileged(p->task);
 126
 127 backing_store_recovery_return:
 128         (void) thread_funnel_set(kernel_flock, FALSE);
 129         return(error);
 130 }
 131
 132 /*
 133  *      Routine:        macx_backing_store_suspend
 134  *      Function:
 135  *              Syscall interface to stop new demand for
 136  *              backing store when backing store is low
 137  */
 138
 139 int
 140 macx_backing_store_suspend(
 141         struct macx_backing_store_suspend_args *args)
 142 {
 143         boolean_t       suspend = args->suspend;
 144         int             error;
 145         boolean_t       funnel_state;
 146
 147         funnel_state = thread_funnel_set(kernel_flock, TRUE);
 148         if ((error = suser(kauth_cred_get(), 0)))
 149                 goto backing_store_suspend_return;
 150
 151         vm_backing_store_disable(suspend);
 152
 153 backing_store_suspend_return:
 154         (void) thread_funnel_set(kernel_flock, FALSE);
 155         return(error);
 156 }
 157
 158 extern boolean_t backing_store_stop_compaction;
 159
 160 /*
 161  *      Routine:        macx_backing_store_compaction
 162  *      Function:
 163  *              Turn compaction of swap space on or off.  This is
 164  *              used during shutdown/restart so that the kernel
 165  *              doesn't waste time compacting swap files that are
 166  *              about to be deleted anyway.  Compaction is always
 167  *              on by default when the system comes up and is turned
 168  *              off when a shutdown/restart is requested.  It is
 169  *              re-enabled if the shutdown/restart is aborted for any reason.
 170  */
 171
 172 int
 173 macx_backing_store_compaction(int flags)
 174 {
 175         int error;
 176
 177         if ((error = suser(kauth_cred_get(), 0)))
 178                 return error;
 179
 180         if (flags & SWAP_COMPACT_DISABLE) {
 181                 backing_store_stop_compaction = TRUE;
 182
 183         } else if (flags & SWAP_COMPACT_ENABLE) {
 184                 backing_store_stop_compaction = FALSE;
 185         }
 186
 187         return 0;
 188 }
 189
 190 /*
 191  *      Routine:        macx_triggers
 192  *      Function:
 193  *              Syscall interface to set the call backs for low and
 194  *              high water marks.
 195  */
 196 int
 197 macx_triggers(
 198         struct macx_triggers_args *args)
 199 {
 200         int     error;
 201
 202         error = suser(kauth_cred_get(), 0);
 203         if (error)
 204                 return error;
 205
 206         return mach_macx_triggers(args);
 207 }
 208
 209
 210 extern boolean_t dp_isssd;
 211
 212 /*
 213  *      Routine:        macx_swapon
 214  *      Function:
 215  *              Syscall interface to add a file to backing store
 216  */
 217 int
 218 macx_swapon(
 219         struct macx_swapon_args *args)
 220 {
 221         int                     size = args->size;
 222         vnode_t                 vp = (vnode_t)NULL;
 223         struct nameidata        nd, *ndp;
 224         register int            error;
 225         kern_return_t           kr;
 226         mach_port_t             backing_store;
 227         memory_object_default_t default_pager;
 228         int                     i;
 229         boolean_t               funnel_state;
 230         off_t                   file_size;
 231         vfs_context_t           ctx = vfs_context_current();
 232         struct proc             *p =  current_proc();
 233         int                     dp_cluster_size;
 234
 235
 236         AUDIT_MACH_SYSCALL_ENTER(AUE_SWAPON);
 237         AUDIT_ARG(value32, args->priority);
 238
 239         funnel_state = thread_funnel_set(kernel_flock, TRUE);
 240         ndp = &nd;
 241
 242         if ((error = suser(kauth_cred_get(), 0)))
 243                 goto swapon_bailout;
 244
 245         /*
 246          * Get a vnode for the paging area.
 247          */
 248         NDINIT(ndp, LOOKUP, FOLLOW | LOCKLEAF | AUDITVNPATH1,
 249                ((IS_64BIT_PROCESS(p)) ? UIO_USERSPACE64 : UIO_USERSPACE32),
 250                (user_addr_t) args->filename, ctx);
 251
 252         if ((error = namei(ndp)))
 253                 goto swapon_bailout;
 254         nameidone(ndp);
 255         vp = ndp->ni_vp;
 256
 257         if (vp->v_type != VREG) {
 258                 error = EINVAL;
 259                 goto swapon_bailout;
 260         }
 261
 262         /* get file size */
 263         if ((error = vnode_size(vp, &file_size, ctx)) != 0)
 264                 goto swapon_bailout;
 265 #if CONFIG_MACF
 266         vnode_lock(vp);
 267         error = mac_system_check_swapon(vfs_context_ucred(ctx), vp);
 268         vnode_unlock(vp);
 269         if (error)
 270                 goto swapon_bailout;
 271 #endif
 272
 273         /* resize to desired size if it's too small */
 274         if ((file_size < (off_t)size) && ((error = vnode_setsize(vp, (off_t)size, 0, ctx)) != 0))
 275                 goto swapon_bailout;
 276
 277         if (default_pager_init_flag == 0) {
 278                 start_def_pager(NULL);
 279                 default_pager_init_flag = 1;
 280         }
 281
 282         /* add new backing store to list */
 283         i = 0;
 284         while(bs_port_table[i].vp != 0) {
 285                 if(i == MAX_BACKING_STORE)
 286                         break;
 287                 i++;
 288         }
 289         if(i == MAX_BACKING_STORE) {
 290                 error = ENOMEM;
 291                 goto swapon_bailout;
 292         }
 293
 294         /* remember the vnode. This vnode has namei() reference */
 295         bs_port_table[i].vp = vp;
 296
 297         /*
 298          * Look to see if we are already paging to this file.
 299          */
 300         /* make certain the copy send of kernel call will work */
 301         default_pager = MEMORY_OBJECT_DEFAULT_NULL;
 302         kr = host_default_memory_manager(host_priv_self(), &default_pager, 0);
 303         if(kr != KERN_SUCCESS) {
 304            error = EAGAIN;
 305            bs_port_table[i].vp = 0;
 306            goto swapon_bailout;
 307         }
 308
 309         if (vp->v_mount->mnt_kern_flag & MNTK_SSD) {
 310                 /*
 311                  * keep the cluster size small since the
 312                  * seek cost is effectively 0 which means
 313                  * we don't care much about fragmentation
 314                  */
 315                 dp_isssd = TRUE;
 316                 dp_cluster_size = 2 * PAGE_SIZE;
 317         } else {
 318                 /*
 319                  * use the default cluster size
 320                  */
 321                 dp_isssd = FALSE;
 322                 dp_cluster_size = 0;
 323         }
 324         kr = default_pager_backing_store_create(default_pager,
 325                                         -1, /* default priority */
 326                                         dp_cluster_size,
 327                                         &backing_store);
 328         memory_object_default_deallocate(default_pager);
 329
 330         if(kr != KERN_SUCCESS) {
 331            error = ENOMEM;
 332            bs_port_table[i].vp = 0;
 333            goto swapon_bailout;
 334         }
 335
 336         /* Mark this vnode as being used for swapfile */
 337         vnode_lock_spin(vp);
 338         SET(vp->v_flag, VSWAP);
 339         vnode_unlock(vp);
 340
 341         /*
 342          * NOTE: we are able to supply PAGE_SIZE here instead of
 343          *      an actual record size or block number because:
 344          *      a: we do not support offsets from the beginning of the
 345          *              file (allowing for non page size/record modulo offsets.
 346          *      b: because allow paging will be done modulo page size
 347          */
 348
 349         kr = default_pager_add_file(backing_store, (vnode_ptr_t) vp,
 350                                 PAGE_SIZE, (int)(file_size/PAGE_SIZE));
 351         if(kr != KERN_SUCCESS) {
 352            bs_port_table[i].vp = 0;
 353            if(kr == KERN_INVALID_ARGUMENT)
 354                 error = EINVAL;
 355            else
 356                 error = ENOMEM;
 357
 358            /* This vnode is not to be used for swapfile */
 359            vnode_lock_spin(vp);
 360            CLR(vp->v_flag, VSWAP);
 361            vnode_unlock(vp);
 362
 363            goto swapon_bailout;
 364         }
 365         bs_port_table[i].bs = (void *)backing_store;
 366         error = 0;
 367
 368         ubc_setthreadcred(vp, p, current_thread());
 369
 370         /*
 371          * take a long term reference on the vnode to keep
 372          * vnreclaim() away from this vnode.
 373          */
 374         vnode_ref(vp);
 375
 376 swapon_bailout:
 377         if (vp) {
 378                 vnode_put(vp);
 379         }
 380         (void) thread_funnel_set(kernel_flock, FALSE);
 381         AUDIT_MACH_SYSCALL_EXIT(error);
 382         return(error);
 383 }
 384
 385 /*
 386  *      Routine:        macx_swapoff
 387  *      Function:
 388  *              Syscall interface to remove a file from backing store
 389  */
 390 int
 391 macx_swapoff(
 392         struct macx_swapoff_args *args)
 393 {
 394         __unused int    flags = args->flags;
 395         kern_return_t   kr;
 396         mach_port_t     backing_store;
 397
 398         struct vnode            *vp = 0;
 399         struct nameidata        nd, *ndp;
 400         struct proc             *p =  current_proc();
 401         int                     i;
 402         int                     error;
 403         boolean_t               funnel_state;
 404         vfs_context_t ctx = vfs_context_current();
 405
 406         AUDIT_MACH_SYSCALL_ENTER(AUE_SWAPOFF);
 407
 408         funnel_state = thread_funnel_set(kernel_flock, TRUE);
 409         backing_store = NULL;
 410         ndp = &nd;
 411
 412         if ((error = suser(kauth_cred_get(), 0)))
 413                 goto swapoff_bailout;
 414
 415         /*
 416          * Get the vnode for the paging area.
 417          */
 418         NDINIT(ndp, LOOKUP, FOLLOW | LOCKLEAF | AUDITVNPATH1,
 419                ((IS_64BIT_PROCESS(p)) ? UIO_USERSPACE64 : UIO_USERSPACE32),
 420                (user_addr_t) args->filename, ctx);
 421
 422         if ((error = namei(ndp)))
 423                 goto swapoff_bailout;
 424         nameidone(ndp);
 425         vp = ndp->ni_vp;
 426
 427         if (vp->v_type != VREG) {
 428                 error = EINVAL;
 429                 goto swapoff_bailout;
 430         }
 431 #if CONFIG_MACF
 432         vnode_lock(vp);
 433         error = mac_system_check_swapoff(vfs_context_ucred(ctx), vp);
 434         vnode_unlock(vp);
 435         if (error)
 436                 goto swapoff_bailout;
 437 #endif
 438
 439         for(i = 0; i < MAX_BACKING_STORE; i++) {
 440                 if(bs_port_table[i].vp == vp) {
 441                         break;
 442                 }
 443         }
 444         if (i == MAX_BACKING_STORE) {
 445                 error = EINVAL;
 446                 goto swapoff_bailout;
 447         }
 448         backing_store = (mach_port_t)bs_port_table[i].bs;
 449
 450         kr = default_pager_backing_store_delete(backing_store);
 451         switch (kr) {
 452                 case KERN_SUCCESS:
 453                         error = 0;
 454                         bs_port_table[i].vp = 0;
 455                         /* This vnode is no longer used for swapfile */
 456                         vnode_lock_spin(vp);
 457                         CLR(vp->v_flag, VSWAP);
 458                         vnode_unlock(vp);
 459
 460                         /* get rid of macx_swapon() "long term" reference */
 461                         vnode_rele(vp);
 462
 463                         break;
 464                 case KERN_FAILURE:
 465                         error = EAGAIN;
 466                         break;
 467                 default:
 468                         error = EAGAIN;
 469                         break;
 470         }
 471
 472 swapoff_bailout:
 473         /* get rid of macx_swapoff() namei() reference */
 474         if (vp)
 475                 vnode_put(vp);
 476
 477         (void) thread_funnel_set(kernel_flock, FALSE);
 478         AUDIT_MACH_SYSCALL_EXIT(error);
 479         return(error);
 480 }
 481
 482 /*
 483  *      Routine:        macx_swapinfo
 484  *      Function:
 485  *              Syscall interface to get general swap statistics
 486  */
 487 int
 488 macx_swapinfo(
 489         memory_object_size_t    *total_p,
 490         memory_object_size_t    *avail_p,
 491         vm_size_t               *pagesize_p,
 492         boolean_t               *encrypted_p)
 493 {
 494         int                     error;
 495         memory_object_default_t default_pager;
 496         default_pager_info_64_t dpi64;
 497         kern_return_t           kr;
 498
 499         error = 0;
 500
 501         /*
 502          * Get a handle on the default pager.
 503          */
 504         default_pager = MEMORY_OBJECT_DEFAULT_NULL;
 505         kr = host_default_memory_manager(host_priv_self(), &default_pager, 0);
 506         if (kr != KERN_SUCCESS) {
 507                 error = EAGAIN; /* XXX why EAGAIN ? */
 508                 goto done;
 509         }
 510         if (default_pager == MEMORY_OBJECT_DEFAULT_NULL) {
 511                 /*
 512                  * The default pager has not initialized yet,
 513                  * so it can't be using any swap space at all.
 514                  */
 515                 *total_p = 0;
 516                 *avail_p = 0;
 517                 *pagesize_p = 0;
 518                 *encrypted_p = FALSE;
 519                 goto done;
 520         }
 521
 522         /*
 523          * Get swap usage data from default pager.
 524          */
 525         kr = default_pager_info_64(default_pager, &dpi64);
 526         if (kr != KERN_SUCCESS) {
 527                 error = ENOTSUP;
 528                 goto done;
 529         }
 530
 531         /*
 532          * Provide default pager info to caller.
 533          */
 534         *total_p = dpi64.dpi_total_space;
 535         *avail_p = dpi64.dpi_free_space;
 536         *pagesize_p = dpi64.dpi_page_size;
 537         if (dpi64.dpi_flags & DPI_ENCRYPTED) {
 538                 *encrypted_p = TRUE;
 539         } else {
 540                 *encrypted_p = FALSE;
 541         }
 542
 543 done:
 544         if (default_pager != MEMORY_OBJECT_DEFAULT_NULL) {
 545                 /* release our handle on default pager */
 546                 memory_object_default_deallocate(default_pager);
 547         }
 548         return error;
 549 }