kvm: dirty-ring: Fix race with vcpu creation
[qemu/ar7.git] / bsd-user / main.c
blobbabc3b009b624ab058d30377c038bbf3fa363e70
1 /*
2 * qemu bsd user main
4 * Copyright (c) 2003-2008 Fabrice Bellard
5 * Copyright (c) 2013-14 Stacey Son
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.
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.
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/>.
21 #include "qemu/osdep.h"
22 #include <sys/resource.h>
23 #include <sys/sysctl.h>
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>
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/tcg.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"
49 #include "host-os.h"
50 #include "target_arch_cpu.h"
52 int singlestep;
53 uintptr_t guest_base;
54 bool have_guest_base;
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.
59 * This way we will never overlap with our own libraries or binaries or stack
60 * or anything else that QEMU maps.
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.
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
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.
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
91 static 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 */
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 */
102 /* Helper routines for implementing atomic operations. */
104 void fork_start(void)
106 start_exclusive();
107 cpu_list_lock();
108 mmap_fork_start();
111 void fork_end(int child)
113 if (child) {
114 CPUState *cpu, *next_cpu;
116 * Child processes created by fork() only have a single thread. Discard
117 * information about the parent threads.
119 CPU_FOREACH_SAFE(cpu, next_cpu) {
120 if (cpu != thread_cpu) {
121 QTAILQ_REMOVE_RCU(&cpus, cpu, node);
124 mmap_fork_end(child);
126 * qemu_init_cpu_list() takes care of reinitializing the exclusive
127 * state, so we don't need to end_exclusive() here.
129 qemu_init_cpu_list();
130 gdbserver_fork(thread_cpu);
131 } else {
132 mmap_fork_end(child);
133 cpu_list_unlock();
134 end_exclusive();
138 void cpu_loop(CPUArchState *env)
140 target_cpu_loop(env);
143 static void usage(void)
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 "-singlestep always run in singlestep mode\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"
182 TARGET_NAME,
183 interp_prefix,
184 target_dflssiz);
185 exit(1);
188 __thread CPUState *thread_cpu;
190 void stop_all_tasks(void)
193 * We trust when using NPTL (pthreads) start_exclusive() handles thread
194 * stopping correctly.
196 start_exclusive();
199 bool qemu_cpu_is_self(CPUState *cpu)
201 return thread_cpu == cpu;
204 void qemu_cpu_kick(CPUState *cpu)
206 cpu_exit(cpu);
209 /* Assumes contents are already zeroed. */
210 static void init_task_state(TaskState *ts)
212 ts->sigaltstack_used = (struct target_sigaltstack) {
213 .ss_sp = 0,
214 .ss_size = 0,
215 .ss_flags = TARGET_SS_DISABLE,
219 void gemu_log(const char *fmt, ...)
221 va_list ap;
223 va_start(ap, fmt);
224 vfprintf(stderr, fmt, ap);
225 va_end(ap);
228 static void
229 adjust_ssize(void)
231 struct rlimit rl;
233 if (getrlimit(RLIMIT_STACK, &rl) != 0) {
234 return;
237 target_maxssiz = MIN(target_maxssiz, rl.rlim_max);
238 target_dflssiz = MIN(MAX(target_dflssiz, rl.rlim_cur), target_maxssiz);
240 rl.rlim_max = target_maxssiz;
241 rl.rlim_cur = target_dflssiz;
242 setrlimit(RLIMIT_STACK, &rl);
245 static void save_proc_pathname(char *argv0)
247 int mib[4];
248 size_t len;
250 mib[0] = CTL_KERN;
251 mib[1] = KERN_PROC;
252 mib[2] = KERN_PROC_PATHNAME;
253 mib[3] = -1;
255 len = sizeof(qemu_proc_pathname);
256 if (sysctl(mib, 4, qemu_proc_pathname, &len, NULL, 0)) {
257 perror("sysctl");
261 int main(int argc, char **argv)
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;
282 adjust_ssize();
284 if (argc <= 1) {
285 usage();
288 save_proc_pathname(argv[0]);
290 error_init(argv[0]);
291 module_call_init(MODULE_INIT_TRACE);
292 qemu_init_cpu_list();
293 module_call_init(MODULE_INIT_QOM);
295 envlist = envlist_create();
297 /* add current environment into the list */
298 for (wrk = environ; *wrk != NULL; wrk++) {
299 (void) envlist_setenv(envlist, *wrk);
302 cpu_model = NULL;
304 qemu_add_opts(&qemu_trace_opts);
306 optind = 1;
307 for (;;) {
308 if (optind >= argc) {
309 break;
311 r = argv[optind];
312 if (r[0] != '-') {
313 break;
315 optind++;
316 r++;
317 if (!strcmp(r, "-")) {
318 break;
319 } else if (!strcmp(r, "d")) {
320 if (optind >= argc) {
321 break;
323 log_mask = argv[optind++];
324 } else if (!strcmp(r, "D")) {
325 if (optind >= argc) {
326 break;
328 log_file = argv[optind++];
329 } else if (!strcmp(r, "E")) {
330 r = argv[optind++];
331 if (envlist_setenv(envlist, r) != 0) {
332 usage();
334 } else if (!strcmp(r, "ignore-environment")) {
335 envlist_free(envlist);
336 envlist = envlist_create();
337 } else if (!strcmp(r, "U")) {
338 r = argv[optind++];
339 if (envlist_unsetenv(envlist, r) != 0) {
340 usage();
342 } else if (!strcmp(r, "s")) {
343 r = argv[optind++];
344 rv = qemu_strtoul(r, &r, 0, &target_dflssiz);
345 if (rv < 0 || target_dflssiz <= 0) {
346 usage();
348 if (*r == 'M') {
349 target_dflssiz *= 1024 * 1024;
350 } else if (*r == 'k' || *r == 'K') {
351 target_dflssiz *= 1024;
353 if (target_dflssiz > target_maxssiz) {
354 usage();
356 } else if (!strcmp(r, "L")) {
357 interp_prefix = argv[optind++];
358 } else if (!strcmp(r, "p")) {
359 qemu_host_page_size = atoi(argv[optind++]);
360 if (qemu_host_page_size == 0 ||
361 (qemu_host_page_size & (qemu_host_page_size - 1)) != 0) {
362 fprintf(stderr, "page size must be a power of two\n");
363 exit(1);
365 } else if (!strcmp(r, "g")) {
366 gdbstub = g_strdup(argv[optind++]);
367 } else if (!strcmp(r, "r")) {
368 qemu_uname_release = argv[optind++];
369 } else if (!strcmp(r, "cpu")) {
370 cpu_model = argv[optind++];
371 if (is_help_option(cpu_model)) {
372 /* XXX: implement xxx_cpu_list for targets that still miss it */
373 #if defined(cpu_list)
374 cpu_list();
375 #endif
376 exit(1);
378 } else if (!strcmp(r, "B")) {
379 rv = qemu_strtoul(argv[optind++], NULL, 0, &guest_base);
380 if (rv < 0) {
381 usage();
383 have_guest_base = true;
384 } else if (!strcmp(r, "drop-ld-preload")) {
385 (void) envlist_unsetenv(envlist, "LD_PRELOAD");
386 } else if (!strcmp(r, "seed")) {
387 seed_optarg = optarg;
388 } else if (!strcmp(r, "singlestep")) {
389 singlestep = 1;
390 } else if (!strcmp(r, "strace")) {
391 do_strace = 1;
392 } else if (!strcmp(r, "trace")) {
393 trace_opt_parse(optarg);
394 } else if (!strcmp(r, "0")) {
395 argv0 = argv[optind++];
396 } else {
397 usage();
401 /* init debug */
403 int mask = 0;
404 if (log_mask) {
405 mask = qemu_str_to_log_mask(log_mask);
406 if (!mask) {
407 qemu_print_log_usage(stdout);
408 exit(1);
411 qemu_set_log_filename_flags(log_file, mask, &error_fatal);
414 if (optind >= argc) {
415 usage();
417 filename = argv[optind];
418 if (argv0) {
419 argv[optind] = argv0;
422 if (!trace_init_backends()) {
423 exit(1);
425 trace_init_file();
427 /* Zero out regs */
428 memset(regs, 0, sizeof(struct target_pt_regs));
430 /* Zero bsd params */
431 memset(&bprm, 0, sizeof(bprm));
433 /* Zero out image_info */
434 memset(info, 0, sizeof(struct image_info));
436 /* Scan interp_prefix dir for replacement files. */
437 init_paths(interp_prefix);
439 if (cpu_model == NULL) {
440 cpu_model = TARGET_DEFAULT_CPU_MODEL;
443 cpu_type = parse_cpu_option(cpu_model);
445 /* init tcg before creating CPUs and to get qemu_host_page_size */
447 AccelClass *ac = ACCEL_GET_CLASS(current_accel());
449 accel_init_interfaces(ac);
450 ac->init_machine(NULL);
452 cpu = cpu_create(cpu_type);
453 env = cpu->env_ptr;
454 cpu_reset(cpu);
455 thread_cpu = cpu;
457 if (getenv("QEMU_STRACE")) {
458 do_strace = 1;
461 target_environ = envlist_to_environ(envlist, NULL);
462 envlist_free(envlist);
464 if (reserved_va) {
465 mmap_next_start = reserved_va + 1;
469 Error *err = NULL;
470 if (seed_optarg != NULL) {
471 qemu_guest_random_seed_main(seed_optarg, &err);
472 } else {
473 qcrypto_init(&err);
475 if (err) {
476 error_reportf_err(err, "cannot initialize crypto: ");
477 exit(1);
482 * Now that page sizes are configured we can do
483 * proper page alignment for guest_base.
485 guest_base = HOST_PAGE_ALIGN(guest_base);
487 if (loader_exec(filename, argv + optind, target_environ, regs, info,
488 &bprm) != 0) {
489 printf("Error loading %s\n", filename);
490 _exit(1);
493 for (wrk = target_environ; *wrk; wrk++) {
494 g_free(*wrk);
497 g_free(target_environ);
499 if (qemu_loglevel_mask(CPU_LOG_PAGE)) {
500 FILE *f = qemu_log_trylock();
501 if (f) {
502 fprintf(f, "guest_base %p\n", (void *)guest_base);
503 fprintf(f, "page layout changed following binary load\n");
504 page_dump(f);
506 fprintf(f, "start_brk 0x" TARGET_ABI_FMT_lx "\n",
507 info->start_brk);
508 fprintf(f, "end_code 0x" TARGET_ABI_FMT_lx "\n",
509 info->end_code);
510 fprintf(f, "start_code 0x" TARGET_ABI_FMT_lx "\n",
511 info->start_code);
512 fprintf(f, "start_data 0x" TARGET_ABI_FMT_lx "\n",
513 info->start_data);
514 fprintf(f, "end_data 0x" TARGET_ABI_FMT_lx "\n",
515 info->end_data);
516 fprintf(f, "start_stack 0x" TARGET_ABI_FMT_lx "\n",
517 info->start_stack);
518 fprintf(f, "brk 0x" TARGET_ABI_FMT_lx "\n", info->brk);
519 fprintf(f, "entry 0x" TARGET_ABI_FMT_lx "\n", info->entry);
521 qemu_log_unlock(f);
525 /* build Task State */
526 ts = g_new0(TaskState, 1);
527 init_task_state(ts);
528 ts->info = info;
529 ts->bprm = &bprm;
530 cpu->opaque = ts;
532 target_set_brk(info->brk);
533 syscall_init();
534 signal_init();
537 * Now that we've loaded the binary, GUEST_BASE is fixed. Delay
538 * generating the prologue until now so that the prologue can take
539 * the real value of GUEST_BASE into account.
541 tcg_prologue_init(tcg_ctx);
543 target_cpu_init(env, regs);
545 if (gdbstub) {
546 gdbserver_start(gdbstub);
547 gdb_handlesig(cpu, 0);
549 cpu_loop(env);
550 /* never exits */
551 return 0;