bsd-user/main.c

   1 /*
   2  *  qemu bsd user main
   3  *
   4  *  Copyright (c) 2003-2008 Fabrice Bellard
   5  *  Copyright (c) 2013-14 Stacey Son
   6  *
   7  *  This program is free software; you can redistribute it and/or modify
   8  *  it under the terms of the GNU General Public License as published by
   9  *  the Free Software Foundation; either version 2 of the License, or
  10  *  (at your option) any later version.
  11  *
  12  *  This program is distributed in the hope that it will be useful,
  13  *  but WITHOUT ANY WARRANTY; without even the implied warranty of
  14  *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  15  *  GNU General Public License for more details.
  16  *
  17  *  You should have received a copy of the GNU General Public License
  18  *  along with this program; if not, see <http://www.gnu.org/licenses/>.
  19  */
  20
  21 #include "qemu/osdep.h"
  22 #include <sys/resource.h>
  23 #include <sys/sysctl.h>
  24
  25 #include "qemu/help-texts.h"
  26 #include "qemu/units.h"
  27 #include "qemu/accel.h"
  28 #include "qemu-version.h"
  29 #include <machine/trap.h>
  30
  31 #include "qapi/error.h"
  32 #include "qemu.h"
  33 #include "qemu/config-file.h"
  34 #include "qemu/error-report.h"
  35 #include "qemu/path.h"
  36 #include "qemu/help_option.h"
  37 #include "qemu/module.h"
  38 #include "exec/exec-all.h"
  39 #include "tcg/startup.h"
  40 #include "qemu/timer.h"
  41 #include "qemu/envlist.h"
  42 #include "qemu/cutils.h"
  43 #include "exec/log.h"
  44 #include "trace/control.h"
  45 #include "crypto/init.h"
  46 #include "qemu/guest-random.h"
  47 #include "gdbstub/user.h"
  48
  49 #include "host-os.h"
  50 #include "target_arch_cpu.h"
  51
  52 static bool opt_one_insn_per_tb;
  53 uintptr_t guest_base;
  54 bool have_guest_base;
  55 /*
  56  * When running 32-on-64 we should make sure we can fit all of the possible
  57  * guest address space into a contiguous chunk of virtual host memory.
  58  *
  59  * This way we will never overlap with our own libraries or binaries or stack
  60  * or anything else that QEMU maps.
  61  *
  62  * Many cpus reserve the high bit (or more than one for some 64-bit cpus)
  63  * of the address for the kernel.  Some cpus rely on this and user space
  64  * uses the high bit(s) for pointer tagging and the like.  For them, we
  65  * must preserve the expected address space.
  66  */
  67 #ifndef MAX_RESERVED_VA
  68 # if HOST_LONG_BITS > TARGET_VIRT_ADDR_SPACE_BITS
  69 #  if TARGET_VIRT_ADDR_SPACE_BITS == 32 && \
  70       (TARGET_LONG_BITS == 32 || defined(TARGET_ABI32))
  71 #   define MAX_RESERVED_VA  0xfffffffful
  72 #  else
  73 #   define MAX_RESERVED_VA  ((1ul << TARGET_VIRT_ADDR_SPACE_BITS) - 1)
  74 #  endif
  75 # else
  76 #  define MAX_RESERVED_VA  0
  77 # endif
  78 #endif
  79
  80 /*
  81  * That said, reserving *too* much vm space via mmap can run into problems
  82  * with rlimits, oom due to page table creation, etc.  We will still try it,
  83  * if directed by the command-line option, but not by default.
  84  */
  85 #if HOST_LONG_BITS == 64 && TARGET_VIRT_ADDR_SPACE_BITS <= 32
  86 unsigned long reserved_va = MAX_RESERVED_VA;
  87 #else
  88 unsigned long reserved_va;
  89 #endif
  90
  91 const char *interp_prefix = CONFIG_QEMU_INTERP_PREFIX;
  92 const char *qemu_uname_release;
  93 char qemu_proc_pathname[PATH_MAX];  /* full path to exeutable */
  94
  95 unsigned long target_maxtsiz = TARGET_MAXTSIZ;   /* max text size */
  96 unsigned long target_dfldsiz = TARGET_DFLDSIZ;   /* initial data size limit */
  97 unsigned long target_maxdsiz = TARGET_MAXDSIZ;   /* max data size */
  98 unsigned long target_dflssiz = TARGET_DFLSSIZ;   /* initial data size limit */
  99 unsigned long target_maxssiz = TARGET_MAXSSIZ;   /* max stack size */
 100 unsigned long target_sgrowsiz = TARGET_SGROWSIZ; /* amount to grow stack */
 101
 102 /* Helper routines for implementing atomic operations. */
 103
 104 void fork_start(void)
 105 {
 106     start_exclusive();
 107     cpu_list_lock();
 108     mmap_fork_start();
 109 }
 110
 111 void fork_end(int child)
 112 {
 113     if (child) {
 114         CPUState *cpu, *next_cpu;
 115         /*
 116          * Child processes created by fork() only have a single thread.  Discard
 117          * information about the parent threads.
 118          */
 119         CPU_FOREACH_SAFE(cpu, next_cpu) {
 120             if (cpu != thread_cpu) {
 121                 QTAILQ_REMOVE_RCU(&cpus_queue, cpu, node);
 122             }
 123         }
 124         mmap_fork_end(child);
 125         /*
 126          * qemu_init_cpu_list() takes care of reinitializing the exclusive
 127          * state, so we don't need to end_exclusive() here.
 128          */
 129         qemu_init_cpu_list();
 130         gdbserver_fork(thread_cpu);
 131     } else {
 132         mmap_fork_end(child);
 133         cpu_list_unlock();
 134         end_exclusive();
 135     }
 136 }
 137
 138 void cpu_loop(CPUArchState *env)
 139 {
 140     target_cpu_loop(env);
 141 }
 142
 143 static void usage(void)
 144 {
 145     printf("qemu-" TARGET_NAME " version " QEMU_FULL_VERSION
 146            "\n" QEMU_COPYRIGHT "\n"
 147            "usage: qemu-" TARGET_NAME " [options] program [arguments...]\n"
 148            "BSD CPU emulator (compiled for %s emulation)\n"
 149            "\n"
 150            "Standard options:\n"
 151            "-h                print this help\n"
 152            "-g port           wait gdb connection to port\n"
 153            "-L path           set the elf interpreter prefix (default=%s)\n"
 154            "-s size           set the stack size in bytes (default=%ld)\n"
 155            "-cpu model        select CPU (-cpu help for list)\n"
 156            "-drop-ld-preload  drop LD_PRELOAD for target process\n"
 157            "-E var=value      sets/modifies targets environment variable(s)\n"
 158            "-U var            unsets targets environment variable(s)\n"
 159            "-B address        set guest_base address to address\n"
 160            "\n"
 161            "Debug options:\n"
 162            "-d item1[,...]    enable logging of specified items\n"
 163            "                  (use '-d help' for a list of log items)\n"
 164            "-D logfile        write logs to 'logfile' (default stderr)\n"
 165            "-one-insn-per-tb  run with one guest instruction per emulated TB\n"
 166            "-strace           log system calls\n"
 167            "-trace            [[enable=]<pattern>][,events=<file>][,file=<file>]\n"
 168            "                  specify tracing options\n"
 169            "\n"
 170            "Environment variables:\n"
 171            "QEMU_STRACE       Print system calls and arguments similar to the\n"
 172            "                  'strace' program.  Enable by setting to any value.\n"
 173            "You can use -E and -U options to set/unset environment variables\n"
 174            "for target process.  It is possible to provide several variables\n"
 175            "by repeating the option.  For example:\n"
 176            "    -E var1=val2 -E var2=val2 -U LD_PRELOAD -U LD_DEBUG\n"
 177            "Note that if you provide several changes to single variable\n"
 178            "last change will stay in effect.\n"
 179            "\n"
 180            QEMU_HELP_BOTTOM "\n"
 181            ,
 182            TARGET_NAME,
 183            interp_prefix,
 184            target_dflssiz);
 185     exit(1);
 186 }
 187
 188 __thread CPUState *thread_cpu;
 189
 190 void stop_all_tasks(void)
 191 {
 192     /*
 193      * We trust when using NPTL (pthreads) start_exclusive() handles thread
 194      * stopping correctly.
 195      */
 196     start_exclusive();
 197 }
 198
 199 bool qemu_cpu_is_self(CPUState *cpu)
 200 {
 201     return thread_cpu == cpu;
 202 }
 203
 204 void qemu_cpu_kick(CPUState *cpu)
 205 {
 206     cpu_exit(cpu);
 207 }
 208
 209 /* Assumes contents are already zeroed.  */
 210 static void init_task_state(TaskState *ts)
 211 {
 212     ts->sigaltstack_used = (struct target_sigaltstack) {
 213         .ss_sp = 0,
 214         .ss_size = 0,
 215         .ss_flags = TARGET_SS_DISABLE,
 216     };
 217 }
 218
 219 void gemu_log(const char *fmt, ...)
 220 {
 221     va_list ap;
 222
 223     va_start(ap, fmt);
 224     vfprintf(stderr, fmt, ap);
 225     va_end(ap);
 226 }
 227
 228 static void
 229 adjust_ssize(void)
 230 {
 231     struct rlimit rl;
 232
 233     if (getrlimit(RLIMIT_STACK, &rl) != 0) {
 234         return;
 235     }
 236
 237     target_maxssiz = MIN(target_maxssiz, rl.rlim_max);
 238     target_dflssiz = MIN(MAX(target_dflssiz, rl.rlim_cur), target_maxssiz);
 239
 240     rl.rlim_max = target_maxssiz;
 241     rl.rlim_cur = target_dflssiz;
 242     setrlimit(RLIMIT_STACK, &rl);
 243 }
 244
 245 static void save_proc_pathname(char *argv0)
 246 {
 247     int mib[4];
 248     size_t len;
 249
 250     mib[0] = CTL_KERN;
 251     mib[1] = KERN_PROC;
 252     mib[2] = KERN_PROC_PATHNAME;
 253     mib[3] = -1;
 254
 255     len = sizeof(qemu_proc_pathname);
 256     if (sysctl(mib, 4, qemu_proc_pathname, &len, NULL, 0)) {
 257         perror("sysctl");
 258     }
 259 }
 260
 261 int main(int argc, char **argv)
 262 {
 263     const char *filename;
 264     const char *cpu_model;
 265     const char *cpu_type;
 266     const char *log_file = NULL;
 267     const char *log_mask = NULL;
 268     const char *seed_optarg = NULL;
 269     struct target_pt_regs regs1, *regs = &regs1;
 270     struct image_info info1, *info = &info1;
 271     struct bsd_binprm bprm;
 272     TaskState *ts;
 273     CPUArchState *env;
 274     CPUState *cpu;
 275     int optind, rv;
 276     const char *r;
 277     const char *gdbstub = NULL;
 278     char **target_environ, **wrk;
 279     envlist_t *envlist = NULL;
 280     char *argv0 = NULL;
 281
 282     adjust_ssize();
 283
 284     if (argc <= 1) {
 285         usage();
 286     }
 287
 288     save_proc_pathname(argv[0]);
 289
 290     error_init(argv[0]);
 291     module_call_init(MODULE_INIT_TRACE);
 292     qemu_init_cpu_list();
 293     module_call_init(MODULE_INIT_QOM);
 294
 295     envlist = envlist_create();
 296
 297     /*
 298      * add current environment into the list
 299      * envlist_setenv adds to the front of the list; to preserve environ
 300      * order add from back to front
 301      */
 302     for (wrk = environ; *wrk != NULL; wrk++) {
 303         continue;
 304     }
 305     while (wrk != environ) {
 306         wrk--;
 307         (void) envlist_setenv(envlist, *wrk);
 308     }
 309
 310     cpu_model = NULL;
 311
 312     qemu_add_opts(&qemu_trace_opts);
 313
 314     optind = 1;
 315     for (;;) {
 316         if (optind >= argc) {
 317             break;
 318         }
 319         r = argv[optind];
 320         if (r[0] != '-') {
 321             break;
 322         }
 323         optind++;
 324         r++;
 325         if (!strcmp(r, "-")) {
 326             break;
 327         } else if (!strcmp(r, "d")) {
 328             if (optind >= argc) {
 329                 break;
 330             }
 331             log_mask = argv[optind++];
 332         } else if (!strcmp(r, "D")) {
 333             if (optind >= argc) {
 334                 break;
 335             }
 336             log_file = argv[optind++];
 337         } else if (!strcmp(r, "E")) {
 338             r = argv[optind++];
 339             if (envlist_setenv(envlist, r) != 0) {
 340                 usage();
 341             }
 342         } else if (!strcmp(r, "ignore-environment")) {
 343             envlist_free(envlist);
 344             envlist = envlist_create();
 345         } else if (!strcmp(r, "U")) {
 346             r = argv[optind++];
 347             if (envlist_unsetenv(envlist, r) != 0) {
 348                 usage();
 349             }
 350         } else if (!strcmp(r, "s")) {
 351             r = argv[optind++];
 352             rv = qemu_strtoul(r, &r, 0, &target_dflssiz);
 353             if (rv < 0 || target_dflssiz <= 0) {
 354                 usage();
 355             }
 356             if (*r == 'M') {
 357                 target_dflssiz *= 1024 * 1024;
 358             } else if (*r == 'k' || *r == 'K') {
 359                 target_dflssiz *= 1024;
 360             }
 361             if (target_dflssiz > target_maxssiz) {
 362                 usage();
 363             }
 364         } else if (!strcmp(r, "L")) {
 365             interp_prefix = argv[optind++];
 366         } else if (!strcmp(r, "p")) {
 367             qemu_host_page_size = atoi(argv[optind++]);
 368             if (qemu_host_page_size == 0 ||
 369                 (qemu_host_page_size & (qemu_host_page_size - 1)) != 0) {
 370                 fprintf(stderr, "page size must be a power of two\n");
 371                 exit(1);
 372             }
 373         } else if (!strcmp(r, "g")) {
 374             gdbstub = g_strdup(argv[optind++]);
 375         } else if (!strcmp(r, "r")) {
 376             qemu_uname_release = argv[optind++];
 377         } else if (!strcmp(r, "cpu")) {
 378             cpu_model = argv[optind++];
 379             if (is_help_option(cpu_model)) {
 380                 list_cpus();
 381                 exit(1);
 382             }
 383         } else if (!strcmp(r, "B")) {
 384             rv = qemu_strtoul(argv[optind++], NULL, 0, &guest_base);
 385             if (rv < 0) {
 386                 usage();
 387             }
 388             have_guest_base = true;
 389         } else if (!strcmp(r, "drop-ld-preload")) {
 390             (void) envlist_unsetenv(envlist, "LD_PRELOAD");
 391         } else if (!strcmp(r, "seed")) {
 392             seed_optarg = optarg;
 393         } else if (!strcmp(r, "one-insn-per-tb")) {
 394             opt_one_insn_per_tb = true;
 395         } else if (!strcmp(r, "strace")) {
 396             do_strace = 1;
 397         } else if (!strcmp(r, "trace")) {
 398             trace_opt_parse(optarg);
 399         } else if (!strcmp(r, "0")) {
 400             argv0 = argv[optind++];
 401         } else {
 402             usage();
 403         }
 404     }
 405
 406     /* init debug */
 407     {
 408         int mask = 0;
 409         if (log_mask) {
 410             mask = qemu_str_to_log_mask(log_mask);
 411             if (!mask) {
 412                 qemu_print_log_usage(stdout);
 413                 exit(1);
 414             }
 415         }
 416         qemu_set_log_filename_flags(log_file, mask, &error_fatal);
 417     }
 418
 419     if (optind >= argc) {
 420         usage();
 421     }
 422     filename = argv[optind];
 423     if (argv0) {
 424         argv[optind] = argv0;
 425     }
 426
 427     if (!trace_init_backends()) {
 428         exit(1);
 429     }
 430     trace_init_file();
 431
 432     /* Zero out regs */
 433     memset(regs, 0, sizeof(struct target_pt_regs));
 434
 435     /* Zero bsd params */
 436     memset(&bprm, 0, sizeof(bprm));
 437
 438     /* Zero out image_info */
 439     memset(info, 0, sizeof(struct image_info));
 440
 441     /* Scan interp_prefix dir for replacement files. */
 442     init_paths(interp_prefix);
 443
 444     if (cpu_model == NULL) {
 445         cpu_model = TARGET_DEFAULT_CPU_MODEL;
 446     }
 447
 448     cpu_type = parse_cpu_option(cpu_model);
 449
 450     /* init tcg before creating CPUs and to get qemu_host_page_size */
 451     {
 452         AccelState *accel = current_accel();
 453         AccelClass *ac = ACCEL_GET_CLASS(accel);
 454
 455         accel_init_interfaces(ac);
 456         object_property_set_bool(OBJECT(accel), "one-insn-per-tb",
 457                                  opt_one_insn_per_tb, &error_abort);
 458         ac->init_machine(NULL);
 459     }
 460     cpu = cpu_create(cpu_type);
 461     env = cpu_env(cpu);
 462     cpu_reset(cpu);
 463     thread_cpu = cpu;
 464
 465     if (getenv("QEMU_STRACE")) {
 466         do_strace = 1;
 467     }
 468
 469     target_environ = envlist_to_environ(envlist, NULL);
 470     envlist_free(envlist);
 471
 472     {
 473         Error *err = NULL;
 474         if (seed_optarg != NULL) {
 475             qemu_guest_random_seed_main(seed_optarg, &err);
 476         } else {
 477             qcrypto_init(&err);
 478         }
 479         if (err) {
 480             error_reportf_err(err, "cannot initialize crypto: ");
 481             exit(1);
 482         }
 483     }
 484
 485     /*
 486      * Now that page sizes are configured we can do
 487      * proper page alignment for guest_base.
 488      */
 489     if (have_guest_base) {
 490         if (guest_base & ~qemu_host_page_mask) {
 491             error_report("Selected guest base not host page aligned");
 492             exit(1);
 493         }
 494     }
 495
 496     /*
 497      * If reserving host virtual address space, do so now.
 498      * Combined with '-B', ensure that the chosen range is free.
 499      */
 500     if (reserved_va) {
 501         void *p;
 502
 503         if (have_guest_base) {
 504             p = mmap((void *)guest_base, reserved_va + 1, PROT_NONE,
 505                      MAP_ANON | MAP_PRIVATE | MAP_FIXED | MAP_EXCL, -1, 0);
 506         } else {
 507             p = mmap(NULL, reserved_va + 1, PROT_NONE,
 508                      MAP_ANON | MAP_PRIVATE, -1, 0);
 509         }
 510         if (p == MAP_FAILED) {
 511             const char *err = strerror(errno);
 512             char *sz = size_to_str(reserved_va + 1);
 513
 514             if (have_guest_base) {
 515                 error_report("Cannot allocate %s bytes at -B %p for guest "
 516                              "address space: %s", sz, (void *)guest_base, err);
 517             } else {
 518                 error_report("Cannot allocate %s bytes for guest "
 519                              "address space: %s", sz, err);
 520             }
 521             exit(1);
 522         }
 523         guest_base = (uintptr_t)p;
 524         have_guest_base = true;
 525
 526         /* Ensure that mmap_next_start is within range. */
 527         if (reserved_va <= mmap_next_start) {
 528             mmap_next_start = (reserved_va / 4 * 3)
 529                               & TARGET_PAGE_MASK & qemu_host_page_mask;
 530         }
 531     }
 532
 533     if (loader_exec(filename, argv + optind, target_environ, regs, info,
 534                     &bprm) != 0) {
 535         printf("Error loading %s\n", filename);
 536         _exit(1);
 537     }
 538
 539     for (wrk = target_environ; *wrk; wrk++) {
 540         g_free(*wrk);
 541     }
 542
 543     g_free(target_environ);
 544
 545     if (qemu_loglevel_mask(CPU_LOG_PAGE)) {
 546         FILE *f = qemu_log_trylock();
 547         if (f) {
 548             fprintf(f, "guest_base  %p\n", (void *)guest_base);
 549             fprintf(f, "page layout changed following binary load\n");
 550             page_dump(f);
 551
 552             fprintf(f, "end_code    0x" TARGET_ABI_FMT_lx "\n",
 553                     info->end_code);
 554             fprintf(f, "start_code  0x" TARGET_ABI_FMT_lx "\n",
 555                     info->start_code);
 556             fprintf(f, "start_data  0x" TARGET_ABI_FMT_lx "\n",
 557                     info->start_data);
 558             fprintf(f, "end_data    0x" TARGET_ABI_FMT_lx "\n",
 559                     info->end_data);
 560             fprintf(f, "start_stack 0x" TARGET_ABI_FMT_lx "\n",
 561                     info->start_stack);
 562             fprintf(f, "brk         0x" TARGET_ABI_FMT_lx "\n", info->brk);
 563             fprintf(f, "entry       0x" TARGET_ABI_FMT_lx "\n", info->entry);
 564
 565             qemu_log_unlock(f);
 566         }
 567     }
 568
 569     /* build Task State */
 570     ts = g_new0(TaskState, 1);
 571     init_task_state(ts);
 572     ts->info = info;
 573     ts->bprm = &bprm;
 574     cpu->opaque = ts;
 575
 576     target_set_brk(info->brk);
 577     syscall_init();
 578     signal_init();
 579
 580     /*
 581      * Now that we've loaded the binary, GUEST_BASE is fixed.  Delay
 582      * generating the prologue until now so that the prologue can take
 583      * the real value of GUEST_BASE into account.
 584      */
 585     tcg_prologue_init();
 586
 587     target_cpu_init(env, regs);
 588
 589     if (gdbstub) {
 590         gdbserver_start(gdbstub);
 591         gdb_handlesig(cpu, 0);
 592     }
 593     cpu_loop(env);
 594     /* never exits */
 595     return 0;
 596 }