1 /* SPDX-License-Identifier: LGPL-2.1+ */
3 #ifndef __LXC_PROCESS_UTILS_H
4 #define __LXC_PROCESS_UTILS_H
8 #include <linux/sched.h>
15 #include <sys/syscall.h>
19 #include "syscall_numbers.h"
22 #define CSIGNAL 0x000000ff /* signal mask to be sent at exit */
26 #define CLONE_VM 0x00000100 /* set if VM shared between processes */
30 #define CLONE_FS 0x00000200 /* set if fs info shared between processes */
34 #define CLONE_FILES 0x00000400 /* set if open files shared between processes */
38 #define CLONE_SIGHAND 0x00000800 /* set if signal handlers and blocked signals shared */
42 #define CLONE_PIDFD 0x00001000 /* set if a pidfd should be placed in parent */
46 #define CLONE_PTRACE 0x00002000 /* set if we want to let tracing continue on the child too */
50 #define CLONE_VFORK 0x00004000 /* set if the parent wants the child to wake it up on mm_release */
54 #define CLONE_PARENT 0x00008000 /* set if we want to have the same parent as the cloner */
58 #define CLONE_THREAD 0x00010000 /* Same thread group? */
62 #define CLONE_NEWNS 0x00020000 /* New mount namespace group */
66 #define CLONE_SYSVSEM 0x00040000 /* share system V SEM_UNDO semantics */
70 #define CLONE_SETTLS 0x00080000 /* create a new TLS for the child */
73 #ifndef CLONE_PARENT_SETTID
74 #define CLONE_PARENT_SETTID 0x00100000 /* set the TID in the parent */
77 #ifndef CLONE_CHILD_CLEARTID
78 #define CLONE_CHILD_CLEARTID 0x00200000 /* clear the TID in the child */
81 #ifndef CLONE_DETACHED
82 #define CLONE_DETACHED 0x00400000 /* Unused, ignored */
85 #ifndef CLONE_UNTRACED
86 #define CLONE_UNTRACED 0x00800000 /* set if the tracing process can't force CLONE_PTRACE on this clone */
89 #ifndef CLONE_CHILD_SETTID
90 #define CLONE_CHILD_SETTID 0x01000000 /* set the TID in the child */
93 #ifndef CLONE_NEWCGROUP
94 #define CLONE_NEWCGROUP 0x02000000 /* New cgroup namespace */
98 #define CLONE_NEWUTS 0x04000000 /* New utsname namespace */
102 #define CLONE_NEWIPC 0x08000000 /* New ipc namespace */
105 #ifndef CLONE_NEWUSER
106 #define CLONE_NEWUSER 0x10000000 /* New user namespace */
110 #define CLONE_NEWPID 0x20000000 /* New pid namespace */
114 #define CLONE_NEWNET 0x40000000 /* New network namespace */
118 #define CLONE_IO 0x80000000 /* Clone io context */
121 /* Flags for the clone3() syscall. */
122 #ifndef CLONE_CLEAR_SIGHAND
123 #define CLONE_CLEAR_SIGHAND 0x100000000ULL /* Clear any signal handler and reset to SIG_DFL. */
126 #ifndef CLONE_INTO_CGROUP
127 #define CLONE_INTO_CGROUP 0x200000000ULL /* Clone into a specific cgroup given the right permissions. */
131 * cloning flags intersect with CSIGNAL so can be used with unshare and clone3
134 #ifndef CLONE_NEWTIME
135 #define CLONE_NEWTIME 0x00000080 /* New time namespace */
143 #ifndef CLONE_ARGS_SIZE_VER0
144 #define CLONE_ARGS_SIZE_VER0 64 /* sizeof first published struct */
147 #ifndef CLONE_ARGS_SIZE_VER1
148 #define CLONE_ARGS_SIZE_VER1 80 /* sizeof second published struct */
151 #ifndef CLONE_ARGS_SIZE_VER2
152 #define CLONE_ARGS_SIZE_VER2 88 /* sizeof third published struct */
156 #define ptr_to_u64(ptr) ((__u64)((uintptr_t)(ptr)))
159 #define u64_to_ptr(x) ((void *)(uintptr_t)x)
162 struct lxc_clone_args
{
165 __aligned_u64 child_tid
;
166 __aligned_u64 parent_tid
;
167 __aligned_u64 exit_signal
;
169 __aligned_u64 stack_size
;
171 __aligned_u64 set_tid
;
172 __aligned_u64 set_tid_size
;
173 __aligned_u64 cgroup
;
176 __returns_twice
static inline pid_t
lxc_clone3(struct lxc_clone_args
*args
, size_t size
)
178 return syscall(__NR_clone3
, args
, size
);
181 #if defined(__ia64__)
182 int __clone2(int (*__fn
)(void *__arg
), void *__child_stack_base
,
183 size_t __child_stack_size
, int __flags
, void *__arg
, ...);
185 int clone(int (*fn
)(void *), void *child_stack
, int flags
, void *arg
, ...
186 /* pid_t *ptid, struct user_desc *tls, pid_t *ctid */);
190 * lxc_clone() - create a new process
193 * This function allocates a new stack the size of page and passes it to the
196 * - support all CLONE_*flags:
197 * This function supports all CLONE_* flags. If in doubt or not sufficiently
198 * familiar with process creation in the kernel and interactions with libcs
199 * this function should be used.
201 * - pthread_atfork() handlers depending on libc:
202 * Whether this function runs pthread_atfork() handlers depends on the
203 * corresponding libc wrapper. glibc currently does not run pthread_atfork()
204 * handlers but does not guarantee that they are not. Other libcs might or
205 * might not run pthread_atfork() handlers. If you require guarantees please
206 * refer to the lxc_raw_clone*() functions in process_utils.{c,h}.
208 * - should call lxc_raw_getpid():
209 * The child should use lxc_raw_getpid() to retrieve its pid.
211 __hidden
extern pid_t
lxc_clone(int (*fn
)(void *), void *arg
, int flags
, int *pidfd
);
215 * lxc_raw_clone() - create a new process
218 * This function returns 0 in the child and > 0 in the parent.
221 * This function does not allocate a new stack and relies on copy-on-write
224 * - supports subset of ClONE_* flags:
225 * lxc_raw_clone() intentionally only supports a subset of the flags available
226 * to the actual system call. Please refer to the implementation what flags
227 * cannot be used. Also, please don't assume that just because a flag isn't
228 * explicitly checked for as being unsupported that it is supported. If in
229 * doubt or not sufficiently familiar with process creation in the kernel and
230 * interactions with libcs this function should be used.
232 * - no pthread_atfork() handlers:
233 * This function circumvents - as much as this this is possible - any libc
234 * wrappers and thus does not run any pthread_atfork() handlers. Make sure
235 * that this is safe to do in the context you are trying to call this
238 * - must call lxc_raw_getpid():
239 * The child must use lxc_raw_getpid() to retrieve its pid.
241 __hidden
extern pid_t
lxc_raw_clone(unsigned long flags
, int *pidfd
);
242 __hidden
extern pid_t
lxc_raw_legacy_clone(unsigned long flags
, int *pidfd
);
245 * lxc_raw_clone_cb() - create a new process
247 * - non-fork() behavior:
248 * Function does return pid of the child or -1 on error. Pass in a callback
249 * function via the "fn" argument that gets executed in the child process.
250 * The "args" argument is passed to "fn".
252 * All other comments that apply to lxc_raw_clone() apply to lxc_raw_clone_cb()
255 __hidden
extern pid_t
lxc_raw_clone_cb(int (*fn
)(void *), void *args
, unsigned long flags
,
259 static inline int execveat(int dirfd
, const char *pathname
, char *const argv
[],
260 char *const envp
[], int flags
)
262 return syscall(__NR_execveat
, dirfd
, pathname
, argv
, envp
, flags
);
265 extern int execveat(int dirfd
, const char *pathname
, char *const argv
[],
266 char *const envp
[], int flags
);
270 * Because of older glibc's pid cache (up to 2.25) whenever clone() is called
271 * the child must must retrieve it's own pid via lxc_raw_getpid().
273 static inline pid_t
lxc_raw_getpid(void)
275 return (pid_t
)syscall(SYS_getpid
);
278 static inline pid_t
lxc_raw_gettid(void)
281 return syscall(__NR_gettid
);
283 return lxc_raw_getpid();
287 __hidden
extern int lxc_raw_pidfd_send_signal(int pidfd
, int sig
, siginfo_t
*info
,
290 static inline const char *signal_name(int sig
)
295 s
= sigdescr_np(sig
);
299 return s
?: "INVALID_SIGNAL_NUMBER";
302 #endif /* __LXC_PROCESS_UTILS_H */