virtio-net: vhost control virtqueue support
[qemu/kevin.git] / bsd-user / main.c
blobcb5ea402361fbc695b7fc7bc4296fbc5aabd2e47
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 <sys/types.h>
22 #include <sys/time.h>
23 #include <sys/resource.h>
24 #include <sys/sysctl.h>
26 #include "qemu/osdep.h"
27 #include "qemu-common.h"
28 #include "qemu/units.h"
29 #include "qemu/accel.h"
30 #include "sysemu/tcg.h"
31 #include "qemu-version.h"
32 #include <machine/trap.h>
34 #include "qapi/error.h"
35 #include "qemu.h"
36 #include "qemu/config-file.h"
37 #include "qemu/error-report.h"
38 #include "qemu/path.h"
39 #include "qemu/help_option.h"
40 #include "qemu/module.h"
41 #include "exec/exec-all.h"
42 #include "tcg/tcg.h"
43 #include "qemu/timer.h"
44 #include "qemu/envlist.h"
45 #include "qemu/cutils.h"
46 #include "exec/log.h"
47 #include "trace/control.h"
48 #include "crypto/init.h"
49 #include "qemu/guest-random.h"
51 #include "host-os.h"
52 #include "target_arch_cpu.h"
54 int singlestep;
55 uintptr_t guest_base;
56 bool have_guest_base;
58 * When running 32-on-64 we should make sure we can fit all of the possible
59 * guest address space into a contiguous chunk of virtual host memory.
61 * This way we will never overlap with our own libraries or binaries or stack
62 * or anything else that QEMU maps.
64 * Many cpus reserve the high bit (or more than one for some 64-bit cpus)
65 * of the address for the kernel. Some cpus rely on this and user space
66 * uses the high bit(s) for pointer tagging and the like. For them, we
67 * must preserve the expected address space.
69 #ifndef MAX_RESERVED_VA
70 # if HOST_LONG_BITS > TARGET_VIRT_ADDR_SPACE_BITS
71 # if TARGET_VIRT_ADDR_SPACE_BITS == 32 && \
72 (TARGET_LONG_BITS == 32 || defined(TARGET_ABI32))
74 * There are a number of places where we assign reserved_va to a variable
75 * of type abi_ulong and expect it to fit. Avoid the last page.
77 # define MAX_RESERVED_VA (0xfffffffful & TARGET_PAGE_MASK)
78 # else
79 # define MAX_RESERVED_VA (1ul << TARGET_VIRT_ADDR_SPACE_BITS)
80 # endif
81 # else
82 # define MAX_RESERVED_VA 0
83 # endif
84 #endif
87 * That said, reserving *too* much vm space via mmap can run into problems
88 * with rlimits, oom due to page table creation, etc. We will still try it,
89 * if directed by the command-line option, but not by default.
91 #if HOST_LONG_BITS == 64 && TARGET_VIRT_ADDR_SPACE_BITS <= 32
92 unsigned long reserved_va = MAX_RESERVED_VA;
93 #else
94 unsigned long reserved_va;
95 #endif
97 static const char *interp_prefix = CONFIG_QEMU_INTERP_PREFIX;
98 const char *qemu_uname_release;
99 enum BSDType bsd_type;
100 char qemu_proc_pathname[PATH_MAX]; /* full path to exeutable */
102 unsigned long target_maxtsiz = TARGET_MAXTSIZ; /* max text size */
103 unsigned long target_dfldsiz = TARGET_DFLDSIZ; /* initial data size limit */
104 unsigned long target_maxdsiz = TARGET_MAXDSIZ; /* max data size */
105 unsigned long target_dflssiz = TARGET_DFLSSIZ; /* initial data size limit */
106 unsigned long target_maxssiz = TARGET_MAXSSIZ; /* max stack size */
107 unsigned long target_sgrowsiz = TARGET_SGROWSIZ; /* amount to grow stack */
109 /* Helper routines for implementing atomic operations. */
111 void fork_start(void)
113 start_exclusive();
114 cpu_list_lock();
115 mmap_fork_start();
118 void fork_end(int child)
120 if (child) {
121 CPUState *cpu, *next_cpu;
123 * Child processes created by fork() only have a single thread. Discard
124 * information about the parent threads.
126 CPU_FOREACH_SAFE(cpu, next_cpu) {
127 if (cpu != thread_cpu) {
128 QTAILQ_REMOVE_RCU(&cpus, cpu, node);
131 mmap_fork_end(child);
133 * qemu_init_cpu_list() takes care of reinitializing the exclusive
134 * state, so we don't need to end_exclusive() here.
136 qemu_init_cpu_list();
137 gdbserver_fork(thread_cpu);
138 } else {
139 mmap_fork_end(child);
140 cpu_list_unlock();
141 end_exclusive();
145 void cpu_loop(CPUArchState *env)
147 target_cpu_loop(env);
150 static void usage(void)
152 printf("qemu-" TARGET_NAME " version " QEMU_FULL_VERSION
153 "\n" QEMU_COPYRIGHT "\n"
154 "usage: qemu-" TARGET_NAME " [options] program [arguments...]\n"
155 "BSD CPU emulator (compiled for %s emulation)\n"
156 "\n"
157 "Standard options:\n"
158 "-h print this help\n"
159 "-g port wait gdb connection to port\n"
160 "-L path set the elf interpreter prefix (default=%s)\n"
161 "-s size set the stack size in bytes (default=%ld)\n"
162 "-cpu model select CPU (-cpu help for list)\n"
163 "-drop-ld-preload drop LD_PRELOAD for target process\n"
164 "-E var=value sets/modifies targets environment variable(s)\n"
165 "-U var unsets targets environment variable(s)\n"
166 "-B address set guest_base address to address\n"
167 "-bsd type select emulated BSD type FreeBSD/NetBSD/OpenBSD (default)\n"
168 "\n"
169 "Debug options:\n"
170 "-d item1[,...] enable logging of specified items\n"
171 " (use '-d help' for a list of log items)\n"
172 "-D logfile write logs to 'logfile' (default stderr)\n"
173 "-singlestep always run in singlestep mode\n"
174 "-strace log system calls\n"
175 "-trace [[enable=]<pattern>][,events=<file>][,file=<file>]\n"
176 " specify tracing options\n"
177 "\n"
178 "Environment variables:\n"
179 "QEMU_STRACE Print system calls and arguments similar to the\n"
180 " 'strace' program. Enable by setting to any value.\n"
181 "You can use -E and -U options to set/unset environment variables\n"
182 "for target process. It is possible to provide several variables\n"
183 "by repeating the option. For example:\n"
184 " -E var1=val2 -E var2=val2 -U LD_PRELOAD -U LD_DEBUG\n"
185 "Note that if you provide several changes to single variable\n"
186 "last change will stay in effect.\n"
187 "\n"
188 QEMU_HELP_BOTTOM "\n"
190 TARGET_NAME,
191 interp_prefix,
192 target_dflssiz);
193 exit(1);
196 __thread CPUState *thread_cpu;
198 void stop_all_tasks(void)
201 * We trust when using NPTL (pthreads) start_exclusive() handles thread
202 * stopping correctly.
204 start_exclusive();
207 bool qemu_cpu_is_self(CPUState *cpu)
209 return thread_cpu == cpu;
212 void qemu_cpu_kick(CPUState *cpu)
214 cpu_exit(cpu);
217 /* Assumes contents are already zeroed. */
218 void init_task_state(TaskState *ts)
220 int i;
222 ts->first_free = ts->sigqueue_table;
223 for (i = 0; i < MAX_SIGQUEUE_SIZE - 1; i++) {
224 ts->sigqueue_table[i].next = &ts->sigqueue_table[i + 1];
226 ts->sigqueue_table[i].next = NULL;
229 void gemu_log(const char *fmt, ...)
231 va_list ap;
233 va_start(ap, fmt);
234 vfprintf(stderr, fmt, ap);
235 va_end(ap);
238 static void
239 adjust_ssize(void)
241 struct rlimit rl;
243 if (getrlimit(RLIMIT_STACK, &rl) != 0) {
244 return;
247 target_maxssiz = MIN(target_maxssiz, rl.rlim_max);
248 target_dflssiz = MIN(MAX(target_dflssiz, rl.rlim_cur), target_maxssiz);
250 rl.rlim_max = target_maxssiz;
251 rl.rlim_cur = target_dflssiz;
252 setrlimit(RLIMIT_STACK, &rl);
255 static void save_proc_pathname(char *argv0)
257 int mib[4];
258 size_t len;
260 mib[0] = CTL_KERN;
261 mib[1] = KERN_PROC;
262 mib[2] = KERN_PROC_PATHNAME;
263 mib[3] = -1;
265 len = sizeof(qemu_proc_pathname);
266 if (sysctl(mib, 4, qemu_proc_pathname, &len, NULL, 0)) {
267 perror("sysctl");
271 int main(int argc, char **argv)
273 const char *filename;
274 const char *cpu_model;
275 const char *cpu_type;
276 const char *log_file = NULL;
277 const char *log_mask = NULL;
278 const char *seed_optarg = NULL;
279 struct target_pt_regs regs1, *regs = &regs1;
280 struct image_info info1, *info = &info1;
281 struct bsd_binprm bprm;
282 TaskState *ts;
283 CPUArchState *env;
284 CPUState *cpu;
285 int optind, rv;
286 const char *r;
287 const char *gdbstub = NULL;
288 char **target_environ, **wrk;
289 envlist_t *envlist = NULL;
290 bsd_type = HOST_DEFAULT_BSD_TYPE;
291 char *argv0 = NULL;
293 adjust_ssize();
295 if (argc <= 1) {
296 usage();
299 save_proc_pathname(argv[0]);
301 error_init(argv[0]);
302 module_call_init(MODULE_INIT_TRACE);
303 qemu_init_cpu_list();
304 module_call_init(MODULE_INIT_QOM);
306 envlist = envlist_create();
308 /* add current environment into the list */
309 for (wrk = environ; *wrk != NULL; wrk++) {
310 (void) envlist_setenv(envlist, *wrk);
313 cpu_model = NULL;
315 qemu_add_opts(&qemu_trace_opts);
317 optind = 1;
318 for (;;) {
319 if (optind >= argc) {
320 break;
322 r = argv[optind];
323 if (r[0] != '-') {
324 break;
326 optind++;
327 r++;
328 if (!strcmp(r, "-")) {
329 break;
330 } else if (!strcmp(r, "d")) {
331 if (optind >= argc) {
332 break;
334 log_mask = argv[optind++];
335 } else if (!strcmp(r, "D")) {
336 if (optind >= argc) {
337 break;
339 log_file = argv[optind++];
340 } else if (!strcmp(r, "E")) {
341 r = argv[optind++];
342 if (envlist_setenv(envlist, r) != 0) {
343 usage();
345 } else if (!strcmp(r, "ignore-environment")) {
346 envlist_free(envlist);
347 envlist = envlist_create();
348 } else if (!strcmp(r, "U")) {
349 r = argv[optind++];
350 if (envlist_unsetenv(envlist, r) != 0) {
351 usage();
353 } else if (!strcmp(r, "s")) {
354 r = argv[optind++];
355 rv = qemu_strtoul(r, &r, 0, &target_dflssiz);
356 if (rv < 0 || target_dflssiz <= 0) {
357 usage();
359 if (*r == 'M') {
360 target_dflssiz *= 1024 * 1024;
361 } else if (*r == 'k' || *r == 'K') {
362 target_dflssiz *= 1024;
364 if (target_dflssiz > target_maxssiz) {
365 usage();
367 } else if (!strcmp(r, "L")) {
368 interp_prefix = argv[optind++];
369 } else if (!strcmp(r, "p")) {
370 qemu_host_page_size = atoi(argv[optind++]);
371 if (qemu_host_page_size == 0 ||
372 (qemu_host_page_size & (qemu_host_page_size - 1)) != 0) {
373 fprintf(stderr, "page size must be a power of two\n");
374 exit(1);
376 } else if (!strcmp(r, "g")) {
377 gdbstub = g_strdup(argv[optind++]);
378 } else if (!strcmp(r, "r")) {
379 qemu_uname_release = argv[optind++];
380 } else if (!strcmp(r, "cpu")) {
381 cpu_model = argv[optind++];
382 if (is_help_option(cpu_model)) {
383 /* XXX: implement xxx_cpu_list for targets that still miss it */
384 #if defined(cpu_list)
385 cpu_list();
386 #endif
387 exit(1);
389 } else if (!strcmp(r, "B")) {
390 rv = qemu_strtoul(argv[optind++], NULL, 0, &guest_base);
391 if (rv < 0) {
392 usage();
394 have_guest_base = true;
395 } else if (!strcmp(r, "drop-ld-preload")) {
396 (void) envlist_unsetenv(envlist, "LD_PRELOAD");
397 } else if (!strcmp(r, "bsd")) {
398 if (!strcasecmp(argv[optind], "freebsd")) {
399 bsd_type = target_freebsd;
400 } else if (!strcasecmp(argv[optind], "netbsd")) {
401 bsd_type = target_netbsd;
402 } else if (!strcasecmp(argv[optind], "openbsd")) {
403 bsd_type = target_openbsd;
404 } else {
405 usage();
407 optind++;
408 } else if (!strcmp(r, "seed")) {
409 seed_optarg = optarg;
410 } else if (!strcmp(r, "singlestep")) {
411 singlestep = 1;
412 } else if (!strcmp(r, "strace")) {
413 do_strace = 1;
414 } else if (!strcmp(r, "trace")) {
415 trace_opt_parse(optarg);
416 } else if (!strcmp(r, "0")) {
417 argv0 = argv[optind++];
418 } else {
419 usage();
423 /* init debug */
424 qemu_log_needs_buffers();
425 qemu_set_log_filename(log_file, &error_fatal);
426 if (log_mask) {
427 int mask;
429 mask = qemu_str_to_log_mask(log_mask);
430 if (!mask) {
431 qemu_print_log_usage(stdout);
432 exit(1);
434 qemu_set_log(mask);
437 if (optind >= argc) {
438 usage();
440 filename = argv[optind];
441 if (argv0) {
442 argv[optind] = argv0;
445 if (!trace_init_backends()) {
446 exit(1);
448 trace_init_file();
450 /* Zero out regs */
451 memset(regs, 0, sizeof(struct target_pt_regs));
453 /* Zero bsd params */
454 memset(&bprm, 0, sizeof(bprm));
456 /* Zero out image_info */
457 memset(info, 0, sizeof(struct image_info));
459 /* Scan interp_prefix dir for replacement files. */
460 init_paths(interp_prefix);
462 if (cpu_model == NULL) {
463 cpu_model = TARGET_DEFAULT_CPU_MODEL;
466 cpu_type = parse_cpu_option(cpu_model);
468 /* init tcg before creating CPUs and to get qemu_host_page_size */
470 AccelClass *ac = ACCEL_GET_CLASS(current_accel());
472 accel_init_interfaces(ac);
473 ac->init_machine(NULL);
475 cpu = cpu_create(cpu_type);
476 env = cpu->env_ptr;
477 cpu_reset(cpu);
478 thread_cpu = cpu;
480 if (getenv("QEMU_STRACE")) {
481 do_strace = 1;
484 target_environ = envlist_to_environ(envlist, NULL);
485 envlist_free(envlist);
487 if (reserved_va) {
488 mmap_next_start = reserved_va;
492 Error *err = NULL;
493 if (seed_optarg != NULL) {
494 qemu_guest_random_seed_main(seed_optarg, &err);
495 } else {
496 qcrypto_init(&err);
498 if (err) {
499 error_reportf_err(err, "cannot initialize crypto: ");
500 exit(1);
505 * Now that page sizes are configured we can do
506 * proper page alignment for guest_base.
508 guest_base = HOST_PAGE_ALIGN(guest_base);
510 if (loader_exec(filename, argv + optind, target_environ, regs, info,
511 &bprm) != 0) {
512 printf("Error loading %s\n", filename);
513 _exit(1);
516 for (wrk = target_environ; *wrk; wrk++) {
517 g_free(*wrk);
520 g_free(target_environ);
522 if (qemu_loglevel_mask(CPU_LOG_PAGE)) {
523 qemu_log("guest_base %p\n", (void *)guest_base);
524 log_page_dump("binary load");
526 qemu_log("start_brk 0x" TARGET_ABI_FMT_lx "\n", info->start_brk);
527 qemu_log("end_code 0x" TARGET_ABI_FMT_lx "\n", info->end_code);
528 qemu_log("start_code 0x" TARGET_ABI_FMT_lx "\n",
529 info->start_code);
530 qemu_log("start_data 0x" TARGET_ABI_FMT_lx "\n",
531 info->start_data);
532 qemu_log("end_data 0x" TARGET_ABI_FMT_lx "\n", info->end_data);
533 qemu_log("start_stack 0x" TARGET_ABI_FMT_lx "\n",
534 info->start_stack);
535 qemu_log("brk 0x" TARGET_ABI_FMT_lx "\n", info->brk);
536 qemu_log("entry 0x" TARGET_ABI_FMT_lx "\n", info->entry);
539 /* build Task State */
540 ts = g_new0(TaskState, 1);
541 init_task_state(ts);
542 ts->info = info;
543 ts->bprm = &bprm;
544 cpu->opaque = ts;
546 target_set_brk(info->brk);
547 syscall_init();
548 signal_init();
551 * Now that we've loaded the binary, GUEST_BASE is fixed. Delay
552 * generating the prologue until now so that the prologue can take
553 * the real value of GUEST_BASE into account.
555 tcg_prologue_init(tcg_ctx);
557 target_cpu_init(env, regs);
559 if (gdbstub) {
560 gdbserver_start(gdbstub);
561 gdb_handlesig(cpu, 0);
563 cpu_loop(env);
564 /* never exits */
565 return 0;