gcore.tproj/sparse.c

   1 /*
   2  * Copyright (c) 2016 Apple Inc.  All rights reserved.
   3  */
   4
   5 #include "options.h"
   6 #include "vm.h"
   7 #include "region.h"
   8 #include "utils.h"
   9 #include "dyld.h"
  10 #include "threads.h"
  11 #include "sparse.h"
  12 #include "vanilla.h"
  13 #include "corefile.h"
  14
  15 #include <sys/types.h>
  16 #include <sys/sysctl.h>
  17 #include <sys/stat.h>
  18 #include <sys/mman.h>
  19 #include <libproc.h>
  20
  21 #include <stdio.h>
  22 #include <string.h>
  23 #include <strings.h>
  24 #include <stdlib.h>
  25 #include <stdarg.h>
  26 #include <signal.h>
  27 #include <unistd.h>
  28 #include <errno.h>
  29 #include <ctype.h>
  30 #include <fcntl.h>
  31 #include <assert.h>
  32
  33 #include <mach/mach.h>
  34
  35 static struct subregion *
  36 new_subregion(
  37     const mach_vm_offset_t vmaddr,
  38     const mach_vm_offset_t vmsize,
  39     const native_segment_command_t *sc,
  40     const struct libent *le)
  41 {
  42     struct subregion *s = malloc(sizeof (*s));
  43
  44     assert(vmaddr != 0 && vmsize != 0);
  45     assert(vmaddr < vmaddr + vmsize);
  46     s->s_segcmd = *sc;
  47
  48     S_SETADDR(s, vmaddr);
  49     S_SETSIZE(s, vmsize);
  50
  51     s->s_libent = le;
  52     s->s_isuuidref = false;
  53     return s;
  54 }
  55
  56 static void
  57 del_subregion(struct subregion *s)
  58 {
  59     poison(s, 0xfacefac1, sizeof (*s));
  60     free(s);
  61 }
  62
  63 static walk_return_t
  64 clean_subregions(struct region *r)
  65 {
  66         if (r->r_nsubregions) {
  67                 assert(r->r_subregions);
  68                 for (unsigned i = 0; i < r->r_nsubregions; i++)
  69                         del_subregion(r->r_subregions[i]);
  70                 poison(r->r_subregions, 0xfac1fac1, sizeof (r->r_subregions[0]) * r->r_nsubregions);
  71                 free(r->r_subregions);
  72                 r->r_nsubregions = 0;
  73                 r->r_subregions = NULL;
  74         } else {
  75                 assert(NULL == r->r_subregions);
  76         }
  77     return WALK_CONTINUE;
  78 }
  79
  80 void
  81 del_sparse_region(struct region *r)
  82 {
  83     clean_subregions(r);
  84     poison(r, 0xcafecaff, sizeof (*r));
  85     free(r);
  86 }
  87
  88 #define NULLsc  ((native_segment_command_t *)0)
  89
  90 static bool
  91 issamesubregiontype(const struct subregion *s0, const struct subregion *s1) {
  92     return 0 == strncmp(S_MACHO_TYPE(s0), S_MACHO_TYPE(s1), sizeof (NULLsc->segname));
  93 }
  94
  95 bool
  96 issubregiontype(const struct subregion *s, const char *sctype) {
  97     return 0 == strncmp(S_MACHO_TYPE(s), sctype, sizeof (NULLsc->segname));
  98 }
  99
 100 static void
 101 elide_subregion(struct region *r, unsigned ind)
 102 {
 103     del_subregion(r->r_subregions[ind]);
 104     for (unsigned j = ind; j < r->r_nsubregions - 1; j++)
 105         r->r_subregions[j] = r->r_subregions[j+1];
 106     assert(r->r_nsubregions != 0);
 107     r->r_subregions[--r->r_nsubregions] = NULL;
 108 }
 109
 110 struct subregionlist {
 111     STAILQ_ENTRY(subregionlist) srl_linkage;
 112     struct subregion *srl_s;
 113 };
 114 typedef STAILQ_HEAD(, subregionlist) subregionlisthead_t;
 115
 116 static walk_return_t
 117 add_subregions_for_libent(
 118     subregionlisthead_t *srlh,
 119     const struct region *r,
 120     const native_mach_header_t *mh,
 121     const mach_vm_offset_t __unused mh_taddr,   // address in target
 122     const struct libent *le)
 123 {
 124     const struct load_command *lc = (const void *)(mh + 1);
 125         mach_vm_offset_t objoff = le->le_objoff;
 126     for (unsigned n = 0; n < mh->ncmds; n++) {
 127
 128         const native_segment_command_t *sc;
 129
 130         switch (lc->cmd) {
 131             case NATIVE_LC_SEGMENT:
 132                 sc = (const void *)lc;
 133
 134                 if (0 == sc->vmaddr && strcmp(sc->segname, SEG_PAGEZERO) == 0)
 135                     break;
 136                                 mach_vm_offset_t lo = sc->vmaddr + objoff;
 137                                 mach_vm_offset_t hi = lo + sc->vmsize;
 138
 139                 /* Eliminate non-overlapping sections first */
 140
 141                 if (R_ENDADDR(r) - 1 < lo)
 142                     break;
 143                 if (hi - 1 < R_ADDR(r))
 144                     break;
 145
 146                 /*
 147                  * Some part of this segment is in the region.
 148                  * Trim the edges in the case where we span regions.
 149                  */
 150                 if (lo < R_ADDR(r))
 151                     lo = R_ADDR(r);
 152                 if (hi > R_ENDADDR(r))
 153                     hi = R_ENDADDR(r);
 154
 155                 struct subregionlist *srl = calloc(1, sizeof (*srl));
 156                 struct subregion *s = new_subregion(lo, hi - lo, sc, le);
 157                 assert(sc->fileoff >= 0);
 158                 srl->srl_s = s;
 159                 STAILQ_INSERT_HEAD(srlh, srl, srl_linkage);
 160
 161                 if (OPTIONS_DEBUG(opt, 2)) {
 162                     hsize_str_t hstr;
 163                     printr(r, "subregion %llx-%llx %7s %12s\t%s [%s off %zd for %zd nsects %u flags %x]\n",
 164                            S_ADDR(s), S_ENDADDR(s),
 165                            str_hsize(hstr, S_SIZE(s)),
 166                            sc->segname,
 167                            S_FILENAME(s),
 168                            str_prot(sc->initprot),
 169                            sc->fileoff, sc->filesize,
 170                            sc->nsects, sc->flags);
 171                 }
 172                 break;
 173             default:
 174                 break;
 175         }
 176         if (lc->cmdsize)
 177             lc = (const void *)((caddr_t)lc + lc->cmdsize);
 178         else
 179             break;
 180     }
 181     return WALK_CONTINUE;
 182 }
 183
 184 /*
 185  * Because we aggregate information from multiple sources, there may
 186  * be duplicate subregions.  Eliminate them here.
 187  *
 188  * Note that the each library in the shared cache points
 189  * separately at a single, unified (large!) __LINKEDIT section; these
 190  * get removed here too.
 191  *
 192  * Assumes the subregion array is sorted by address!
 193  */
 194 static void
 195 eliminate_duplicate_subregions(struct region *r)
 196 {
 197     unsigned i = 1;
 198     while (i < r->r_nsubregions) {
 199         struct subregion *s0 = r->r_subregions[i-1];
 200         struct subregion *s1 = r->r_subregions[i];
 201
 202         if (S_ADDR(s0) != S_ADDR(s1) || S_SIZE(s0) != S_SIZE(s1)) {
 203             i++;
 204             continue;
 205         }
 206         if (memcmp(&s0->s_segcmd, &s1->s_segcmd, sizeof (s0->s_segcmd)) != 0) {
 207             i++;
 208             continue;
 209         }
 210         if (OPTIONS_DEBUG(opt, 3))
 211             printr(r, "eliding duplicate %s subregion (%llx-%llx) file %s\n",
 212                    S_MACHO_TYPE(s1), S_ADDR(s1), S_ENDADDR(s1), S_FILENAME(s1));
 213         /* If the duplicate subregions aren't mapping the same file (?), forget the name */
 214         if (s0->s_libent != s1->s_libent)
 215             s0->s_libent = s1->s_libent = NULL;
 216         elide_subregion(r, i);
 217     }
 218 }
 219
 220 /*
 221  * See if any of the dyld information we have can better describe this
 222  * region of the target address space.
 223  */
 224 walk_return_t
 225 decorate_memory_region(struct region *r, void *arg)
 226 {
 227         if (r->r_inzfodregion || r->r_incommregion)
 228                 return WALK_CONTINUE;
 229
 230     const dyld_process_info dpi = arg;
 231
 232     __block walk_return_t retval = WALK_CONTINUE;
 233     __block subregionlisthead_t srlhead = STAILQ_HEAD_INITIALIZER(srlhead);
 234
 235     _dyld_process_info_for_each_image(dpi, ^(uint64_t __unused mhaddr, const uuid_t uuid, __unused const char *path) {
 236         if (WALK_CONTINUE == retval) {
 237             const struct libent *le = libent_lookup_byuuid(uuid);
 238             assert(le->le_mhaddr == mhaddr);
 239                         bool shouldskip = false;
 240                         if (V_SIZE(&le->le_vr))
 241                                 shouldskip = (R_ENDADDR(r) < V_ADDR(&le->le_vr) ||
 242                                                           R_ADDR(r) > V_ENDADDR(&le->le_vr));
 243                         if (!shouldskip)
 244                 retval = add_subregions_for_libent(&srlhead, r, le->le_mh, le->le_mhaddr, le);
 245         }
 246     });
 247     if (WALK_CONTINUE != retval)
 248         goto done;
 249
 250     /*
 251      * Take the unsorted list of subregions, if any,
 252      * and hang a sorted array of ranges on the region structure.
 253      */
 254     if (!STAILQ_EMPTY(&srlhead)) {
 255         struct subregionlist *srl;
 256         STAILQ_FOREACH(srl, &srlhead, srl_linkage) {
 257             r->r_nsubregions++;
 258         }
 259         assert(r->r_nsubregions);
 260
 261         r->r_subregions = calloc(r->r_nsubregions, sizeof (void *));
 262         unsigned i = 0;
 263         STAILQ_FOREACH(srl, &srlhead, srl_linkage) {
 264             r->r_subregions[i++] = srl->srl_s;
 265         }
 266         qsort_b(r->r_subregions, r->r_nsubregions, sizeof (void *),
 267                 ^(const void *a, const void *b) {
 268                     const struct subregion *lhs = *(struct subregion **)a;
 269                     const struct subregion *rhs = *(struct subregion **)b;
 270                     if (S_ADDR(lhs) > S_ADDR(rhs))
 271                         return 1;
 272                     if (S_ADDR(lhs) < S_ADDR(rhs))
 273                         return -1;
 274                     return 0;
 275                 });
 276
 277         eliminate_duplicate_subregions(r);
 278
 279                 if (r->r_info.external_pager) {
 280                         /*
 281                          * Only very specific segment types get to be filerefs
 282                          */
 283                         for (i = 0; i < r->r_nsubregions; i++) {
 284                                 struct subregion *s = r->r_subregions[i];
 285                                 /*
 286                                  * Anything marked writable is trivially disqualified; we're
 287                                  * going to copy it anyway.
 288                                  */
 289                                 if (s->s_segcmd.initprot & VM_PROT_WRITE)
 290                                         continue;
 291
 292                                 /* __TEXT and __LINKEDIT are our real targets */
 293                                 if (!issubregiontype(s, SEG_TEXT) && !issubregiontype(s, SEG_LINKEDIT) && !issubregiontype(s, "__UNICODE")) {
 294                                         if (OPTIONS_DEBUG(opt, 3)) {
 295                                                 hsize_str_t hstr;
 296                                                 printvr(S_RANGE(s), "skipping read-only %s segment %s\n", S_MACHO_TYPE(s), str_hsize(hstr, S_SIZE(s)));
 297                                         }
 298                                         continue;
 299                                 }
 300                                 if (r->r_insharedregion) {
 301                                         /*
 302                                          * Part of the shared region: things get more complicated.
 303                                          */
 304                                         if (r->r_fileref) {
 305                                                 /*
 306                                                  * There's a file reference here for the whole region.
 307                                                  * For __TEXT subregions, we could, in principle (though
 308                                                  * see below) generate references to the individual
 309                                                  * dylibs that dyld reports in the region. If the
 310                                                  * debugger could then use the __LINKEDIT info in the
 311                                                  * file, then we'd be done.  But as long as the dump
 312                                                  * includes __LINKEDIT sections, we're going to
 313                                                  * end up generating a file reference to the combined
 314                                                  * __LINKEDIT section in the shared cache anyway, so
 315                                                  * we might as well do that for the __TEXT regions as
 316                                                  * well.
 317                                                  */
 318                                                 s->s_libent = r->r_fileref->fr_libent;
 319                                                 s->s_isuuidref = true;
 320                                         } else {
 321                                                 /*
 322                                                  * If we get here, it's likely that the shared cache
 323                                                  * name can't be found e.g. update_dyld_shared_cache(1).
 324                                                  * For __TEXT subregions, we could generate refs to
 325                                                  * the individual dylibs, but note that the mach header
 326                                                  * and segment commands in memory are still pointing
 327                                                  * into the shared cache so any act of reconstruction
 328                                                  * is fiendishly complex.  So copy it.
 329                                                  */
 330                                                 assert(!s->s_isuuidref);
 331                                         }
 332                                 } else {
 333                                         /* Just a regular dylib? */
 334                                         if (s->s_libent)
 335                                                 s->s_isuuidref = true;
 336                                 }
 337                         }
 338                 }
 339         }
 340     assert(WALK_CONTINUE == retval);
 341
 342 done:
 343     if (!STAILQ_EMPTY(&srlhead)) {
 344         struct subregionlist *srl, *trl;
 345         STAILQ_FOREACH_SAFE(srl, &srlhead, srl_linkage, trl) {
 346             free(srl);
 347         }
 348     }
 349     return retval;
 350 }
 351
 352 /*
 353  * Strip region of all decoration
 354  *
 355  * Invoked (on every region!) after an error during the initial
 356  * 'decoration' phase to discard potentially incomplete information.
 357  */
 358 walk_return_t
 359 undecorate_memory_region(struct region *r, __unused void *arg)
 360 {
 361     assert(&sparse_ops != r->r_op);
 362     return r->r_nsubregions ? clean_subregions(r) : WALK_CONTINUE;
 363 }
 364
 365 /*
 366  * This optimization occurs -after- the vanilla_region_optimizations(),
 367  * and -after- we've tagged zfod and first-pass fileref's.
 368  */
 369 walk_return_t
 370 sparse_region_optimization(struct region *r, __unused void *arg)
 371 {
 372     assert(&sparse_ops != r->r_op);
 373
 374     if (r->r_inzfodregion) {
 375         /*
 376          * Pure zfod region: almost certainly a more compact
 377          * representation - keep it that way.
 378          */
 379                 if (OPTIONS_DEBUG(opt, 3))
 380                         printr(r, "retaining zfod region\n");
 381         assert(&zfod_ops == r->r_op);
 382         return clean_subregions(r);
 383     }
 384
 385         if (r->r_insharedregion && 0 == r->r_nsubregions) {
 386                 /*
 387                  * A segment in the shared region needs to be
 388                  * identified with an LC_SEGMENT that dyld claims,
 389                  * otherwise (we assert) it's not useful to the dump.
 390                  */
 391                 if (OPTIONS_DEBUG(opt, 2)) {
 392                         hsize_str_t hstr;
 393                         printr(r, "not referenced in dyld info => "
 394                                    "eliding %s range in shared region\n",
 395                                    str_hsize(hstr, R_SIZE(r)));
 396                 }
 397                 if (0 == r->r_info.pages_dirtied && 0 == r->r_info.pages_swapped_out)
 398                         return WALK_DELETE_REGION;
 399                 if (OPTIONS_DEBUG(opt, 2)) {
 400                         hsize_str_t hstr;
 401                         printr(r, "dirty pages, but not referenced in dyld info => "
 402                                    "NOT eliding %s range in shared region\n",
 403                                    str_hsize(hstr, R_SIZE(r)));
 404                 }
 405         }
 406
 407         if (r->r_fileref) {
 408                 /*
 409                  * Already have a fileref for the whole region: already
 410                  * a more compact representation - keep it that way.
 411                  */
 412                 if (OPTIONS_DEBUG(opt, 3))
 413                         printr(r, "retaining fileref region\n");
 414                 assert(&fileref_ops == r->r_op);
 415                 return clean_subregions(r);
 416         }
 417
 418     if (r->r_nsubregions > 1) {
 419         /*
 420          * Merge adjacent or identical subregions that have no file reference
 421          * (Reducing the number of subregions reduces header overhead and
 422          * improves compressability)
 423          */
 424         unsigned i = 1;
 425         while (i < r->r_nsubregions) {
 426             struct subregion *s0 = r->r_subregions[i-1];
 427             struct subregion *s1 = r->r_subregions[i];
 428
 429             if (s0->s_isuuidref) {
 430                 i++;
 431                 continue; /* => destined to be a fileref */
 432             }
 433             if (!issamesubregiontype(s0, s1)) {
 434                 i++;
 435                 continue; /* merge-able subregions must have same "type" */
 436             }
 437
 438             if (S_ENDADDR(s0) == S_ADDR(s1)) {
 439                 /* directly adjacent subregions */
 440                 if (OPTIONS_DEBUG(opt, 2))
 441                     printr(r, "merging subregions (%llx-%llx + %llx-%llx) -- adjacent\n",
 442                            S_ADDR(s0), S_ENDADDR(s0), S_ADDR(s1), S_ENDADDR(s1));
 443                 S_SETSIZE(s0, S_ENDADDR(s1) - S_ADDR(s0));
 444                 elide_subregion(r, i);
 445                 continue;
 446             }
 447
 448             const mach_vm_size_t pfn[2] = {
 449                 S_ADDR(s0) >> pageshift_host,
 450                 S_ADDR(s1) >> pageshift_host
 451             };
 452             const mach_vm_size_t endpfn[2] = {
 453                 (S_ENDADDR(s0) - 1) >> pageshift_host,
 454                 (S_ENDADDR(s1) - 1) >> pageshift_host
 455             };
 456
 457             if (pfn[0] == pfn[1] && pfn[0] == endpfn[0] && pfn[0] == endpfn[1]) {
 458                 /* two small subregions share a host page */
 459                 if (OPTIONS_DEBUG(opt, 2))
 460                     printr(r, "merging subregions (%llx-%llx + %llx-%llx) -- same page\n",
 461                            S_ADDR(s0), S_ENDADDR(s0), S_ADDR(s1), S_ENDADDR(s1));
 462                 S_SETSIZE(s0, S_ENDADDR(s1) - S_ADDR(s0));
 463                 elide_subregion(r, i);
 464                 continue;
 465             }
 466
 467             if (pfn[1] == 1 + endpfn[0]) {
 468                 /* subregions are pagewise-adjacent: bigger chunks to compress */
 469                 if (OPTIONS_DEBUG(opt, 2))
 470                     printr(r, "merging subregions (%llx-%llx + %llx-%llx) -- adjacent pages\n",
 471                            S_ADDR(s0), S_ENDADDR(s0), S_ADDR(s1), S_ENDADDR(s1));
 472                 S_SETSIZE(s0, S_ENDADDR(s1) - S_ADDR(s0));
 473                 elide_subregion(r, i);
 474                 continue;
 475             }
 476
 477             i++;    /* this isn't the subregion we're looking for */
 478         }
 479     }
 480
 481         if (1 == r->r_nsubregions) {
 482                 struct subregion *s = r->r_subregions[0];
 483                 if (!s->s_isuuidref &&
 484                         R_ADDR(r) == S_ADDR(s) && R_ENDADDR(r) == S_ENDADDR(s)) {
 485                         if (OPTIONS_DEBUG(opt, 3))
 486                                 printr(r, "subregion (%llx-%llx) reverts to region\n",
 487                                            S_ADDR(s), S_ENDADDR(s));
 488                         return clean_subregions(r);
 489                 }
 490         }
 491
 492     if (r->r_nsubregions)
 493         r->r_op = &sparse_ops;
 494
 495     return WALK_CONTINUE;
 496 }