зеркало из https://github.com/github/ruby.git
9560 строки
259 KiB
C
9560 строки
259 KiB
C
/**********************************************************************
|
|
|
|
process.c -
|
|
|
|
$Author$
|
|
created at: Tue Aug 10 14:30:50 JST 1993
|
|
|
|
Copyright (C) 1993-2007 Yukihiro Matsumoto
|
|
Copyright (C) 2000 Network Applied Communication Laboratory, Inc.
|
|
Copyright (C) 2000 Information-technology Promotion Agency, Japan
|
|
|
|
**********************************************************************/
|
|
|
|
#include "ruby/internal/config.h"
|
|
|
|
#include "ruby/fiber/scheduler.h"
|
|
|
|
#include <ctype.h>
|
|
#include <errno.h>
|
|
#include <signal.h>
|
|
#include <stdarg.h>
|
|
#include <stdio.h>
|
|
#include <time.h>
|
|
|
|
#ifdef HAVE_STDLIB_H
|
|
# include <stdlib.h>
|
|
#endif
|
|
|
|
#ifdef HAVE_UNISTD_H
|
|
# include <unistd.h>
|
|
#endif
|
|
|
|
#ifdef HAVE_FCNTL_H
|
|
# include <fcntl.h>
|
|
#endif
|
|
|
|
#ifdef HAVE_PROCESS_H
|
|
# include <process.h>
|
|
#endif
|
|
|
|
#ifndef EXIT_SUCCESS
|
|
# define EXIT_SUCCESS 0
|
|
#endif
|
|
|
|
#ifndef EXIT_FAILURE
|
|
# define EXIT_FAILURE 1
|
|
#endif
|
|
|
|
#ifdef HAVE_SYS_WAIT_H
|
|
# include <sys/wait.h>
|
|
#endif
|
|
|
|
#ifdef HAVE_SYS_RESOURCE_H
|
|
# include <sys/resource.h>
|
|
#endif
|
|
|
|
#ifdef HAVE_VFORK_H
|
|
# include <vfork.h>
|
|
#endif
|
|
|
|
#ifdef HAVE_SYS_PARAM_H
|
|
# include <sys/param.h>
|
|
#endif
|
|
|
|
#ifndef MAXPATHLEN
|
|
# define MAXPATHLEN 1024
|
|
#endif
|
|
|
|
#include <sys/stat.h>
|
|
|
|
#ifdef HAVE_SYS_TIME_H
|
|
# include <sys/time.h>
|
|
#endif
|
|
|
|
#ifdef HAVE_SYS_TIMES_H
|
|
# include <sys/times.h>
|
|
#endif
|
|
|
|
#ifdef HAVE_PWD_H
|
|
# include <pwd.h>
|
|
#endif
|
|
|
|
#ifdef HAVE_GRP_H
|
|
# include <grp.h>
|
|
# ifdef __CYGWIN__
|
|
int initgroups(const char *, rb_gid_t);
|
|
# endif
|
|
#endif
|
|
|
|
#ifdef HAVE_SYS_ID_H
|
|
# include <sys/id.h>
|
|
#endif
|
|
|
|
#ifdef __APPLE__
|
|
# include <mach/mach_time.h>
|
|
#endif
|
|
|
|
#include "dln.h"
|
|
#include "hrtime.h"
|
|
#include "internal.h"
|
|
#include "internal/bits.h"
|
|
#include "internal/dir.h"
|
|
#include "internal/error.h"
|
|
#include "internal/eval.h"
|
|
#include "internal/hash.h"
|
|
#include "internal/io.h"
|
|
#include "internal/numeric.h"
|
|
#include "internal/object.h"
|
|
#include "internal/process.h"
|
|
#include "internal/thread.h"
|
|
#include "internal/variable.h"
|
|
#include "internal/warnings.h"
|
|
#include "rjit.h"
|
|
#include "ruby/io.h"
|
|
#include "ruby/st.h"
|
|
#include "ruby/thread.h"
|
|
#include "ruby/util.h"
|
|
#include "vm_core.h"
|
|
#include "vm_sync.h"
|
|
#include "ruby/ractor.h"
|
|
|
|
/* define system APIs */
|
|
#ifdef _WIN32
|
|
#undef open
|
|
#define open rb_w32_uopen
|
|
#endif
|
|
|
|
#if defined(HAVE_TIMES) || defined(_WIN32)
|
|
/*********************************************************************
|
|
*
|
|
* Document-class: Process::Tms
|
|
*
|
|
* Placeholder for rusage
|
|
*/
|
|
static VALUE rb_cProcessTms;
|
|
#endif
|
|
|
|
#ifndef WIFEXITED
|
|
#define WIFEXITED(w) (((w) & 0xff) == 0)
|
|
#endif
|
|
#ifndef WIFSIGNALED
|
|
#define WIFSIGNALED(w) (((w) & 0x7f) > 0 && (((w) & 0x7f) < 0x7f))
|
|
#endif
|
|
#ifndef WIFSTOPPED
|
|
#define WIFSTOPPED(w) (((w) & 0xff) == 0x7f)
|
|
#endif
|
|
#ifndef WEXITSTATUS
|
|
#define WEXITSTATUS(w) (((w) >> 8) & 0xff)
|
|
#endif
|
|
#ifndef WTERMSIG
|
|
#define WTERMSIG(w) ((w) & 0x7f)
|
|
#endif
|
|
#ifndef WSTOPSIG
|
|
#define WSTOPSIG WEXITSTATUS
|
|
#endif
|
|
|
|
#if defined(__FreeBSD__) || defined(__NetBSD__) || defined(__OpenBSD__) || defined(__bsdi__)
|
|
#define HAVE_44BSD_SETUID 1
|
|
#define HAVE_44BSD_SETGID 1
|
|
#endif
|
|
|
|
#ifdef __NetBSD__
|
|
#undef HAVE_SETRUID
|
|
#undef HAVE_SETRGID
|
|
#endif
|
|
|
|
#ifdef BROKEN_SETREUID
|
|
#define setreuid ruby_setreuid
|
|
int setreuid(rb_uid_t ruid, rb_uid_t euid);
|
|
#endif
|
|
#ifdef BROKEN_SETREGID
|
|
#define setregid ruby_setregid
|
|
int setregid(rb_gid_t rgid, rb_gid_t egid);
|
|
#endif
|
|
|
|
#if defined(HAVE_44BSD_SETUID) || defined(__APPLE__)
|
|
#if !defined(USE_SETREUID) && !defined(BROKEN_SETREUID)
|
|
#define OBSOLETE_SETREUID 1
|
|
#endif
|
|
#if !defined(USE_SETREGID) && !defined(BROKEN_SETREGID)
|
|
#define OBSOLETE_SETREGID 1
|
|
#endif
|
|
#endif
|
|
|
|
static void check_uid_switch(void);
|
|
static void check_gid_switch(void);
|
|
static int exec_async_signal_safe(const struct rb_execarg *, char *, size_t);
|
|
|
|
VALUE rb_envtbl(void);
|
|
VALUE rb_env_to_hash(void);
|
|
|
|
#if 1
|
|
#define p_uid_from_name p_uid_from_name
|
|
#define p_gid_from_name p_gid_from_name
|
|
#endif
|
|
|
|
#if defined(HAVE_UNISTD_H)
|
|
# if defined(HAVE_GETLOGIN_R)
|
|
# define USE_GETLOGIN_R 1
|
|
# define GETLOGIN_R_SIZE_DEFAULT 0x100
|
|
# define GETLOGIN_R_SIZE_LIMIT 0x1000
|
|
# if defined(_SC_LOGIN_NAME_MAX)
|
|
# define GETLOGIN_R_SIZE_INIT sysconf(_SC_LOGIN_NAME_MAX)
|
|
# else
|
|
# define GETLOGIN_R_SIZE_INIT GETLOGIN_R_SIZE_DEFAULT
|
|
# endif
|
|
# elif defined(HAVE_GETLOGIN)
|
|
# define USE_GETLOGIN 1
|
|
# endif
|
|
#endif
|
|
|
|
#if defined(HAVE_PWD_H)
|
|
# if defined(HAVE_GETPWUID_R)
|
|
# define USE_GETPWUID_R 1
|
|
# elif defined(HAVE_GETPWUID)
|
|
# define USE_GETPWUID 1
|
|
# endif
|
|
# if defined(HAVE_GETPWNAM_R)
|
|
# define USE_GETPWNAM_R 1
|
|
# elif defined(HAVE_GETPWNAM)
|
|
# define USE_GETPWNAM 1
|
|
# endif
|
|
# if defined(HAVE_GETPWNAM_R) || defined(HAVE_GETPWUID_R)
|
|
# define GETPW_R_SIZE_DEFAULT 0x1000
|
|
# define GETPW_R_SIZE_LIMIT 0x10000
|
|
# if defined(_SC_GETPW_R_SIZE_MAX)
|
|
# define GETPW_R_SIZE_INIT sysconf(_SC_GETPW_R_SIZE_MAX)
|
|
# else
|
|
# define GETPW_R_SIZE_INIT GETPW_R_SIZE_DEFAULT
|
|
# endif
|
|
# endif
|
|
# ifdef USE_GETPWNAM_R
|
|
# define PREPARE_GETPWNAM \
|
|
VALUE getpw_buf = 0
|
|
# define FINISH_GETPWNAM \
|
|
(getpw_buf ? (void)rb_str_resize(getpw_buf, 0) : (void)0)
|
|
# define OBJ2UID1(id) obj2uid((id), &getpw_buf)
|
|
# define OBJ2UID(id) obj2uid0(id)
|
|
static rb_uid_t obj2uid(VALUE id, VALUE *getpw_buf);
|
|
static inline rb_uid_t
|
|
obj2uid0(VALUE id)
|
|
{
|
|
rb_uid_t uid;
|
|
PREPARE_GETPWNAM;
|
|
uid = OBJ2UID1(id);
|
|
FINISH_GETPWNAM;
|
|
return uid;
|
|
}
|
|
# else
|
|
# define PREPARE_GETPWNAM /* do nothing */
|
|
# define FINISH_GETPWNAM /* do nothing */
|
|
# define OBJ2UID1(id) obj2uid((id))
|
|
# define OBJ2UID(id) obj2uid((id))
|
|
static rb_uid_t obj2uid(VALUE id);
|
|
# endif
|
|
#else
|
|
# define PREPARE_GETPWNAM /* do nothing */
|
|
# define FINISH_GETPWNAM /* do nothing */
|
|
# define OBJ2UID1(id) NUM2UIDT(id)
|
|
# define OBJ2UID(id) NUM2UIDT(id)
|
|
# ifdef p_uid_from_name
|
|
# undef p_uid_from_name
|
|
# define p_uid_from_name rb_f_notimplement
|
|
# endif
|
|
#endif
|
|
|
|
#if defined(HAVE_GRP_H)
|
|
# if defined(HAVE_GETGRNAM_R) && defined(_SC_GETGR_R_SIZE_MAX)
|
|
# define USE_GETGRNAM_R
|
|
# define GETGR_R_SIZE_INIT sysconf(_SC_GETGR_R_SIZE_MAX)
|
|
# define GETGR_R_SIZE_DEFAULT 0x1000
|
|
# define GETGR_R_SIZE_LIMIT 0x10000
|
|
# endif
|
|
# ifdef USE_GETGRNAM_R
|
|
# define PREPARE_GETGRNAM \
|
|
VALUE getgr_buf = 0
|
|
# define FINISH_GETGRNAM \
|
|
(getgr_buf ? (void)rb_str_resize(getgr_buf, 0) : (void)0)
|
|
# define OBJ2GID1(id) obj2gid((id), &getgr_buf)
|
|
# define OBJ2GID(id) obj2gid0(id)
|
|
static rb_gid_t obj2gid(VALUE id, VALUE *getgr_buf);
|
|
static inline rb_gid_t
|
|
obj2gid0(VALUE id)
|
|
{
|
|
rb_gid_t gid;
|
|
PREPARE_GETGRNAM;
|
|
gid = OBJ2GID1(id);
|
|
FINISH_GETGRNAM;
|
|
return gid;
|
|
}
|
|
static rb_gid_t obj2gid(VALUE id, VALUE *getgr_buf);
|
|
# else
|
|
# define PREPARE_GETGRNAM /* do nothing */
|
|
# define FINISH_GETGRNAM /* do nothing */
|
|
# define OBJ2GID1(id) obj2gid((id))
|
|
# define OBJ2GID(id) obj2gid((id))
|
|
static rb_gid_t obj2gid(VALUE id);
|
|
# endif
|
|
#else
|
|
# define PREPARE_GETGRNAM /* do nothing */
|
|
# define FINISH_GETGRNAM /* do nothing */
|
|
# define OBJ2GID1(id) NUM2GIDT(id)
|
|
# define OBJ2GID(id) NUM2GIDT(id)
|
|
# ifdef p_gid_from_name
|
|
# undef p_gid_from_name
|
|
# define p_gid_from_name rb_f_notimplement
|
|
# endif
|
|
#endif
|
|
|
|
#if SIZEOF_CLOCK_T == SIZEOF_INT
|
|
typedef unsigned int unsigned_clock_t;
|
|
#elif SIZEOF_CLOCK_T == SIZEOF_LONG
|
|
typedef unsigned long unsigned_clock_t;
|
|
#elif defined(HAVE_LONG_LONG) && SIZEOF_CLOCK_T == SIZEOF_LONG_LONG
|
|
typedef unsigned LONG_LONG unsigned_clock_t;
|
|
#endif
|
|
#ifndef HAVE_SIG_T
|
|
typedef void (*sig_t) (int);
|
|
#endif
|
|
|
|
#define id_exception idException
|
|
static ID id_in, id_out, id_err, id_pid, id_uid, id_gid;
|
|
static ID id_close, id_child;
|
|
#ifdef HAVE_SETPGID
|
|
static ID id_pgroup;
|
|
#endif
|
|
#ifdef _WIN32
|
|
static ID id_new_pgroup;
|
|
#endif
|
|
static ID id_unsetenv_others, id_chdir, id_umask, id_close_others;
|
|
static ID id_nanosecond, id_microsecond, id_millisecond, id_second;
|
|
static ID id_float_microsecond, id_float_millisecond, id_float_second;
|
|
static ID id_GETTIMEOFDAY_BASED_CLOCK_REALTIME, id_TIME_BASED_CLOCK_REALTIME;
|
|
#ifdef CLOCK_REALTIME
|
|
static ID id_CLOCK_REALTIME;
|
|
# define RUBY_CLOCK_REALTIME ID2SYM(id_CLOCK_REALTIME)
|
|
#endif
|
|
#ifdef CLOCK_MONOTONIC
|
|
static ID id_CLOCK_MONOTONIC;
|
|
# define RUBY_CLOCK_MONOTONIC ID2SYM(id_CLOCK_MONOTONIC)
|
|
#endif
|
|
#ifdef CLOCK_PROCESS_CPUTIME_ID
|
|
static ID id_CLOCK_PROCESS_CPUTIME_ID;
|
|
# define RUBY_CLOCK_PROCESS_CPUTIME_ID ID2SYM(id_CLOCK_PROCESS_CPUTIME_ID)
|
|
#endif
|
|
#ifdef CLOCK_THREAD_CPUTIME_ID
|
|
static ID id_CLOCK_THREAD_CPUTIME_ID;
|
|
# define RUBY_CLOCK_THREAD_CPUTIME_ID ID2SYM(id_CLOCK_THREAD_CPUTIME_ID)
|
|
#endif
|
|
#ifdef HAVE_TIMES
|
|
static ID id_TIMES_BASED_CLOCK_MONOTONIC;
|
|
static ID id_TIMES_BASED_CLOCK_PROCESS_CPUTIME_ID;
|
|
#endif
|
|
#ifdef RUSAGE_SELF
|
|
static ID id_GETRUSAGE_BASED_CLOCK_PROCESS_CPUTIME_ID;
|
|
#endif
|
|
static ID id_CLOCK_BASED_CLOCK_PROCESS_CPUTIME_ID;
|
|
#ifdef __APPLE__
|
|
static ID id_MACH_ABSOLUTE_TIME_BASED_CLOCK_MONOTONIC;
|
|
# define RUBY_MACH_ABSOLUTE_TIME_BASED_CLOCK_MONOTONIC ID2SYM(id_MACH_ABSOLUTE_TIME_BASED_CLOCK_MONOTONIC)
|
|
#endif
|
|
static ID id_hertz;
|
|
|
|
static rb_pid_t cached_pid;
|
|
|
|
/* execv and execl are async-signal-safe since SUSv4 (POSIX.1-2008, XPG7) */
|
|
#if defined(__sun) && !defined(_XPG7) /* Solaris 10, 9, ... */
|
|
#define execv(path, argv) (rb_async_bug_errno("unreachable: async-signal-unsafe execv() is called", 0))
|
|
#define execl(path, arg0, arg1, arg2, term) do { extern char **environ; execle((path), (arg0), (arg1), (arg2), (term), (environ)); } while (0)
|
|
#define ALWAYS_NEED_ENVP 1
|
|
#else
|
|
#define ALWAYS_NEED_ENVP 0
|
|
#endif
|
|
|
|
static void
|
|
assert_close_on_exec(int fd)
|
|
{
|
|
#if VM_CHECK_MODE > 0
|
|
#if defined(HAVE_FCNTL) && defined(F_GETFD) && defined(FD_CLOEXEC)
|
|
int flags = fcntl(fd, F_GETFD);
|
|
if (flags == -1) {
|
|
static const char m[] = "reserved FD closed unexpectedly?\n";
|
|
(void)!write(2, m, sizeof(m) - 1);
|
|
return;
|
|
}
|
|
if (flags & FD_CLOEXEC) return;
|
|
rb_bug("reserved FD did not have close-on-exec set");
|
|
#else
|
|
rb_bug("reserved FD without close-on-exec support");
|
|
#endif /* FD_CLOEXEC */
|
|
#endif /* VM_CHECK_MODE */
|
|
}
|
|
|
|
static inline int
|
|
close_unless_reserved(int fd)
|
|
{
|
|
if (rb_reserved_fd_p(fd)) { /* async-signal-safe */
|
|
assert_close_on_exec(fd);
|
|
return 0;
|
|
}
|
|
return close(fd); /* async-signal-safe */
|
|
}
|
|
|
|
/*#define DEBUG_REDIRECT*/
|
|
#if defined(DEBUG_REDIRECT)
|
|
|
|
static void
|
|
ttyprintf(const char *fmt, ...)
|
|
{
|
|
va_list ap;
|
|
FILE *tty;
|
|
int save = errno;
|
|
#ifdef _WIN32
|
|
tty = fopen("con", "w");
|
|
#else
|
|
tty = fopen("/dev/tty", "w");
|
|
#endif
|
|
if (!tty)
|
|
return;
|
|
|
|
va_start(ap, fmt);
|
|
vfprintf(tty, fmt, ap);
|
|
va_end(ap);
|
|
fclose(tty);
|
|
errno = save;
|
|
}
|
|
|
|
static int
|
|
redirect_dup(int oldfd)
|
|
{
|
|
int ret;
|
|
ret = dup(oldfd);
|
|
ttyprintf("dup(%d) => %d\n", oldfd, ret);
|
|
return ret;
|
|
}
|
|
|
|
static int
|
|
redirect_dup2(int oldfd, int newfd)
|
|
{
|
|
int ret;
|
|
ret = dup2(oldfd, newfd);
|
|
ttyprintf("dup2(%d, %d) => %d\n", oldfd, newfd, ret);
|
|
return ret;
|
|
}
|
|
|
|
static int
|
|
redirect_cloexec_dup(int oldfd)
|
|
{
|
|
int ret;
|
|
ret = rb_cloexec_dup(oldfd);
|
|
ttyprintf("cloexec_dup(%d) => %d\n", oldfd, ret);
|
|
return ret;
|
|
}
|
|
|
|
static int
|
|
redirect_cloexec_dup2(int oldfd, int newfd)
|
|
{
|
|
int ret;
|
|
ret = rb_cloexec_dup2(oldfd, newfd);
|
|
ttyprintf("cloexec_dup2(%d, %d) => %d\n", oldfd, newfd, ret);
|
|
return ret;
|
|
}
|
|
|
|
static int
|
|
redirect_close(int fd)
|
|
{
|
|
int ret;
|
|
ret = close_unless_reserved(fd);
|
|
ttyprintf("close(%d) => %d\n", fd, ret);
|
|
return ret;
|
|
}
|
|
|
|
static int
|
|
parent_redirect_open(const char *pathname, int flags, mode_t perm)
|
|
{
|
|
int ret;
|
|
ret = rb_cloexec_open(pathname, flags, perm);
|
|
ttyprintf("parent_open(\"%s\", 0x%x, 0%o) => %d\n", pathname, flags, perm, ret);
|
|
return ret;
|
|
}
|
|
|
|
static int
|
|
parent_redirect_close(int fd)
|
|
{
|
|
int ret;
|
|
ret = close_unless_reserved(fd);
|
|
ttyprintf("parent_close(%d) => %d\n", fd, ret);
|
|
return ret;
|
|
}
|
|
|
|
#else
|
|
#define redirect_dup(oldfd) dup(oldfd)
|
|
#define redirect_dup2(oldfd, newfd) dup2((oldfd), (newfd))
|
|
#define redirect_cloexec_dup(oldfd) rb_cloexec_dup(oldfd)
|
|
#define redirect_cloexec_dup2(oldfd, newfd) rb_cloexec_dup2((oldfd), (newfd))
|
|
#define redirect_close(fd) close_unless_reserved(fd)
|
|
#define parent_redirect_open(pathname, flags, perm) rb_cloexec_open((pathname), (flags), (perm))
|
|
#define parent_redirect_close(fd) close_unless_reserved(fd)
|
|
#endif
|
|
|
|
static VALUE
|
|
get_pid(void)
|
|
{
|
|
if (UNLIKELY(!cached_pid)) { /* 0 is not a valid pid */
|
|
cached_pid = getpid();
|
|
}
|
|
/* pid should be likely POSFIXABLE() */
|
|
return PIDT2NUM(cached_pid);
|
|
}
|
|
|
|
#if defined HAVE_WORKING_FORK || defined HAVE_DAEMON
|
|
static void
|
|
clear_pid_cache(void)
|
|
{
|
|
cached_pid = 0;
|
|
}
|
|
#endif
|
|
|
|
/*
|
|
* call-seq:
|
|
* Process.pid -> integer
|
|
*
|
|
* Returns the process ID of the current process:
|
|
*
|
|
* Process.pid # => 15668
|
|
*
|
|
*/
|
|
|
|
static VALUE
|
|
proc_get_pid(VALUE _)
|
|
{
|
|
return get_pid();
|
|
}
|
|
|
|
static VALUE
|
|
get_ppid(void)
|
|
{
|
|
return PIDT2NUM(getppid());
|
|
}
|
|
|
|
/*
|
|
* call-seq:
|
|
* Process.ppid -> integer
|
|
*
|
|
* Returns the process ID of the parent of the current process:
|
|
*
|
|
* puts "Pid is #{Process.pid}."
|
|
* fork { puts "Parent pid is #{Process.ppid}." }
|
|
*
|
|
* Output:
|
|
*
|
|
* Pid is 271290.
|
|
* Parent pid is 271290.
|
|
*
|
|
* May not return a trustworthy value on certain platforms.
|
|
*/
|
|
|
|
static VALUE
|
|
proc_get_ppid(VALUE _)
|
|
{
|
|
return get_ppid();
|
|
}
|
|
|
|
|
|
/*********************************************************************
|
|
*
|
|
* Document-class: Process::Status
|
|
*
|
|
* A Process::Status contains information about a system process.
|
|
*
|
|
* Thread-local variable <tt>$?</tt> is initially +nil+.
|
|
* Some methods assign to it a Process::Status object
|
|
* that represents a system process (either running or terminated):
|
|
*
|
|
* `ruby -e "exit 99"`
|
|
* stat = $? # => #<Process::Status: pid 1262862 exit 99>
|
|
* stat.class # => Process::Status
|
|
* stat.to_i # => 25344
|
|
* stat >> 8 # => 99
|
|
* stat.stopped? # => false
|
|
* stat.exited? # => true
|
|
* stat.exitstatus # => 99
|
|
*
|
|
*/
|
|
|
|
static VALUE rb_cProcessStatus;
|
|
|
|
struct rb_process_status {
|
|
rb_pid_t pid;
|
|
int status;
|
|
int error;
|
|
};
|
|
|
|
static const rb_data_type_t rb_process_status_type = {
|
|
.wrap_struct_name = "Process::Status",
|
|
.function = {
|
|
.dfree = RUBY_DEFAULT_FREE,
|
|
},
|
|
.data = NULL,
|
|
.flags = RUBY_TYPED_FREE_IMMEDIATELY,
|
|
};
|
|
|
|
static VALUE
|
|
rb_process_status_allocate(VALUE klass)
|
|
{
|
|
struct rb_process_status *data = NULL;
|
|
|
|
return TypedData_Make_Struct(klass, struct rb_process_status, &rb_process_status_type, data);
|
|
}
|
|
|
|
VALUE
|
|
rb_last_status_get(void)
|
|
{
|
|
return GET_THREAD()->last_status;
|
|
}
|
|
|
|
/*
|
|
* call-seq:
|
|
* Process.last_status -> Process::Status or nil
|
|
*
|
|
* Returns a Process::Status object representing the most recently exited
|
|
* child process in the current thread, or +nil+ if none:
|
|
*
|
|
* Process.spawn('ruby', '-e', 'exit 13')
|
|
* Process.wait
|
|
* Process.last_status # => #<Process::Status: pid 14396 exit 13>
|
|
*
|
|
* Process.spawn('ruby', '-e', 'exit 14')
|
|
* Process.wait
|
|
* Process.last_status # => #<Process::Status: pid 4692 exit 14>
|
|
*
|
|
* Process.spawn('ruby', '-e', 'exit 15')
|
|
* # 'exit 15' has not been reaped by #wait.
|
|
* Process.last_status # => #<Process::Status: pid 4692 exit 14>
|
|
* Process.wait
|
|
* Process.last_status # => #<Process::Status: pid 1380 exit 15>
|
|
*
|
|
*/
|
|
static VALUE
|
|
proc_s_last_status(VALUE mod)
|
|
{
|
|
return rb_last_status_get();
|
|
}
|
|
|
|
VALUE
|
|
rb_process_status_new(rb_pid_t pid, int status, int error)
|
|
{
|
|
VALUE last_status = rb_process_status_allocate(rb_cProcessStatus);
|
|
|
|
struct rb_process_status *data = RTYPEDDATA_DATA(last_status);
|
|
data->pid = pid;
|
|
data->status = status;
|
|
data->error = error;
|
|
|
|
rb_obj_freeze(last_status);
|
|
return last_status;
|
|
}
|
|
|
|
static VALUE
|
|
process_status_dump(VALUE status)
|
|
{
|
|
VALUE dump = rb_class_new_instance(0, 0, rb_cObject);
|
|
struct rb_process_status *data = RTYPEDDATA_DATA(status);
|
|
if (data->pid) {
|
|
rb_ivar_set(dump, id_status, INT2NUM(data->status));
|
|
rb_ivar_set(dump, id_pid, PIDT2NUM(data->pid));
|
|
}
|
|
return dump;
|
|
}
|
|
|
|
static VALUE
|
|
process_status_load(VALUE real_obj, VALUE load_obj)
|
|
{
|
|
struct rb_process_status *data = rb_check_typeddata(real_obj, &rb_process_status_type);
|
|
VALUE status = rb_attr_get(load_obj, id_status);
|
|
VALUE pid = rb_attr_get(load_obj, id_pid);
|
|
data->pid = NIL_P(pid) ? 0 : NUM2PIDT(pid);
|
|
data->status = NIL_P(status) ? 0 : NUM2INT(status);
|
|
return real_obj;
|
|
}
|
|
|
|
void
|
|
rb_last_status_set(int status, rb_pid_t pid)
|
|
{
|
|
GET_THREAD()->last_status = rb_process_status_new(pid, status, 0);
|
|
}
|
|
|
|
void
|
|
rb_last_status_clear(void)
|
|
{
|
|
GET_THREAD()->last_status = Qnil;
|
|
}
|
|
|
|
static rb_pid_t
|
|
pst_pid(VALUE pst)
|
|
{
|
|
struct rb_process_status *data = RTYPEDDATA_DATA(pst);
|
|
return data->pid;
|
|
}
|
|
|
|
static int
|
|
pst_status(VALUE pst)
|
|
{
|
|
struct rb_process_status *data = RTYPEDDATA_DATA(pst);
|
|
return data->status;
|
|
}
|
|
|
|
/*
|
|
* call-seq:
|
|
* to_i -> integer
|
|
*
|
|
* Returns the system-dependent integer status of +self+:
|
|
*
|
|
* `cat /nop`
|
|
* $?.to_i # => 256
|
|
*/
|
|
|
|
static VALUE
|
|
pst_to_i(VALUE self)
|
|
{
|
|
int status = pst_status(self);
|
|
return RB_INT2NUM(status);
|
|
}
|
|
|
|
#define PST2INT(st) pst_status(st)
|
|
|
|
/*
|
|
* call-seq:
|
|
* pid -> integer
|
|
*
|
|
* Returns the process ID of the process:
|
|
*
|
|
* system("false")
|
|
* $?.pid # => 1247002
|
|
*
|
|
*/
|
|
|
|
static VALUE
|
|
pst_pid_m(VALUE self)
|
|
{
|
|
rb_pid_t pid = pst_pid(self);
|
|
return PIDT2NUM(pid);
|
|
}
|
|
|
|
static VALUE pst_message_status(VALUE str, int status);
|
|
|
|
static void
|
|
pst_message(VALUE str, rb_pid_t pid, int status)
|
|
{
|
|
rb_str_catf(str, "pid %ld", (long)pid);
|
|
pst_message_status(str, status);
|
|
}
|
|
|
|
static VALUE
|
|
pst_message_status(VALUE str, int status)
|
|
{
|
|
if (WIFSTOPPED(status)) {
|
|
int stopsig = WSTOPSIG(status);
|
|
const char *signame = ruby_signal_name(stopsig);
|
|
if (signame) {
|
|
rb_str_catf(str, " stopped SIG%s (signal %d)", signame, stopsig);
|
|
}
|
|
else {
|
|
rb_str_catf(str, " stopped signal %d", stopsig);
|
|
}
|
|
}
|
|
if (WIFSIGNALED(status)) {
|
|
int termsig = WTERMSIG(status);
|
|
const char *signame = ruby_signal_name(termsig);
|
|
if (signame) {
|
|
rb_str_catf(str, " SIG%s (signal %d)", signame, termsig);
|
|
}
|
|
else {
|
|
rb_str_catf(str, " signal %d", termsig);
|
|
}
|
|
}
|
|
if (WIFEXITED(status)) {
|
|
rb_str_catf(str, " exit %d", WEXITSTATUS(status));
|
|
}
|
|
#ifdef WCOREDUMP
|
|
if (WCOREDUMP(status)) {
|
|
rb_str_cat2(str, " (core dumped)");
|
|
}
|
|
#endif
|
|
return str;
|
|
}
|
|
|
|
|
|
/*
|
|
* call-seq:
|
|
* to_s -> string
|
|
*
|
|
* Returns a string representation of +self+:
|
|
*
|
|
* `cat /nop`
|
|
* $?.to_s # => "pid 1262141 exit 1"
|
|
*
|
|
*
|
|
*/
|
|
|
|
static VALUE
|
|
pst_to_s(VALUE st)
|
|
{
|
|
rb_pid_t pid;
|
|
int status;
|
|
VALUE str;
|
|
|
|
pid = pst_pid(st);
|
|
status = PST2INT(st);
|
|
|
|
str = rb_str_buf_new(0);
|
|
pst_message(str, pid, status);
|
|
return str;
|
|
}
|
|
|
|
|
|
/*
|
|
* call-seq:
|
|
* inspect -> string
|
|
*
|
|
* Returns a string representation of +self+:
|
|
*
|
|
* system("false")
|
|
* $?.inspect # => "#<Process::Status: pid 1303494 exit 1>"
|
|
*
|
|
*/
|
|
|
|
static VALUE
|
|
pst_inspect(VALUE st)
|
|
{
|
|
rb_pid_t pid;
|
|
int status;
|
|
VALUE str;
|
|
|
|
pid = pst_pid(st);
|
|
if (!pid) {
|
|
return rb_sprintf("#<%s: uninitialized>", rb_class2name(CLASS_OF(st)));
|
|
}
|
|
status = PST2INT(st);
|
|
|
|
str = rb_sprintf("#<%s: ", rb_class2name(CLASS_OF(st)));
|
|
pst_message(str, pid, status);
|
|
rb_str_cat2(str, ">");
|
|
return str;
|
|
}
|
|
|
|
|
|
/*
|
|
* call-seq:
|
|
* stat == other -> true or false
|
|
*
|
|
* Returns whether the value of #to_i == +other+:
|
|
*
|
|
* `cat /nop`
|
|
* stat = $? # => #<Process::Status: pid 1170366 exit 1>
|
|
* sprintf('%x', stat.to_i) # => "100"
|
|
* stat == 0x100 # => true
|
|
*
|
|
*/
|
|
|
|
static VALUE
|
|
pst_equal(VALUE st1, VALUE st2)
|
|
{
|
|
if (st1 == st2) return Qtrue;
|
|
return rb_equal(pst_to_i(st1), st2);
|
|
}
|
|
|
|
|
|
/*
|
|
* call-seq:
|
|
* stat & mask -> integer
|
|
*
|
|
* This method is deprecated; use other attribute methods.
|
|
*
|
|
* Returns the logical AND of the value of #to_i with +mask+:
|
|
*
|
|
* `cat /nop`
|
|
* stat = $? # => #<Process::Status: pid 1155508 exit 1>
|
|
* sprintf('%x', stat.to_i) # => "100"
|
|
* stat & 0x00 # => 0
|
|
*
|
|
* ArgumentError is raised if +mask+ is negative.
|
|
*/
|
|
|
|
static VALUE
|
|
pst_bitand(VALUE st1, VALUE st2)
|
|
{
|
|
int status = PST2INT(st1);
|
|
int mask = NUM2INT(st2);
|
|
|
|
if (mask < 0) {
|
|
rb_raise(rb_eArgError, "negative mask value: %d", mask);
|
|
}
|
|
#define WARN_SUGGEST(suggest) \
|
|
rb_warn_deprecated_to_remove_at(3.4, "Process::Status#&", suggest)
|
|
|
|
switch (mask) {
|
|
case 0x80:
|
|
WARN_SUGGEST("Process::Status#coredump?");
|
|
break;
|
|
case 0x7f:
|
|
WARN_SUGGEST("Process::Status#signaled? or Process::Status#termsig");
|
|
break;
|
|
case 0xff:
|
|
WARN_SUGGEST("Process::Status#exited?, Process::Status#stopped? or Process::Status#coredump?");
|
|
break;
|
|
case 0xff00:
|
|
WARN_SUGGEST("Process::Status#exitstatus or Process::Status#stopsig");
|
|
break;
|
|
default:
|
|
WARN_SUGGEST("other Process::Status predicates");
|
|
break;
|
|
}
|
|
#undef WARN_SUGGEST
|
|
status &= mask;
|
|
|
|
return INT2NUM(status);
|
|
}
|
|
|
|
|
|
/*
|
|
* call-seq:
|
|
* stat >> places -> integer
|
|
*
|
|
* This method is deprecated; use other predicate methods.
|
|
*
|
|
* Returns the value of #to_i, shifted +places+ to the right:
|
|
*
|
|
* `cat /nop`
|
|
* stat = $? # => #<Process::Status: pid 1155508 exit 1>
|
|
* stat.to_i # => 256
|
|
* stat >> 1 # => 128
|
|
* stat >> 2 # => 64
|
|
*
|
|
* ArgumentError is raised if +places+ is negative.
|
|
*/
|
|
|
|
static VALUE
|
|
pst_rshift(VALUE st1, VALUE st2)
|
|
{
|
|
int status = PST2INT(st1);
|
|
int places = NUM2INT(st2);
|
|
|
|
if (places < 0) {
|
|
rb_raise(rb_eArgError, "negative shift value: %d", places);
|
|
}
|
|
#define WARN_SUGGEST(suggest) \
|
|
rb_warn_deprecated_to_remove_at(3.4, "Process::Status#>>", suggest)
|
|
|
|
switch (places) {
|
|
case 7:
|
|
WARN_SUGGEST("Process::Status#coredump?");
|
|
break;
|
|
case 8:
|
|
WARN_SUGGEST("Process::Status#exitstatus or Process::Status#stopsig");
|
|
break;
|
|
default:
|
|
WARN_SUGGEST("other Process::Status attributes");
|
|
break;
|
|
}
|
|
#undef WARN_SUGGEST
|
|
status >>= places;
|
|
|
|
return INT2NUM(status);
|
|
}
|
|
|
|
|
|
/*
|
|
* call-seq:
|
|
* stopped? -> true or false
|
|
*
|
|
* Returns +true+ if this process is stopped,
|
|
* and if the corresponding #wait call had the Process::WUNTRACED flag set,
|
|
* +false+ otherwise.
|
|
*/
|
|
|
|
static VALUE
|
|
pst_wifstopped(VALUE st)
|
|
{
|
|
int status = PST2INT(st);
|
|
|
|
return RBOOL(WIFSTOPPED(status));
|
|
}
|
|
|
|
|
|
/*
|
|
* call-seq:
|
|
* stopsig -> integer or nil
|
|
*
|
|
* Returns the number of the signal that caused the process to stop,
|
|
* or +nil+ if the process is not stopped.
|
|
*/
|
|
|
|
static VALUE
|
|
pst_wstopsig(VALUE st)
|
|
{
|
|
int status = PST2INT(st);
|
|
|
|
if (WIFSTOPPED(status))
|
|
return INT2NUM(WSTOPSIG(status));
|
|
return Qnil;
|
|
}
|
|
|
|
|
|
/*
|
|
* call-seq:
|
|
* signaled? -> true or false
|
|
*
|
|
* Returns +true+ if the process terminated because of an uncaught signal,
|
|
* +false+ otherwise.
|
|
*/
|
|
|
|
static VALUE
|
|
pst_wifsignaled(VALUE st)
|
|
{
|
|
int status = PST2INT(st);
|
|
|
|
return RBOOL(WIFSIGNALED(status));
|
|
}
|
|
|
|
|
|
/*
|
|
* call-seq:
|
|
* termsig -> integer or nil
|
|
*
|
|
* Returns the number of the signal that caused the process to terminate
|
|
* or +nil+ if the process was not terminated by an uncaught signal.
|
|
*/
|
|
|
|
static VALUE
|
|
pst_wtermsig(VALUE st)
|
|
{
|
|
int status = PST2INT(st);
|
|
|
|
if (WIFSIGNALED(status))
|
|
return INT2NUM(WTERMSIG(status));
|
|
return Qnil;
|
|
}
|
|
|
|
|
|
/*
|
|
* call-seq:
|
|
* exited? -> true or false
|
|
*
|
|
* Returns +true+ if the process exited normally
|
|
* (for example using an <code>exit()</code> call or finishing the
|
|
* program), +false+ if not.
|
|
*/
|
|
|
|
static VALUE
|
|
pst_wifexited(VALUE st)
|
|
{
|
|
int status = PST2INT(st);
|
|
|
|
return RBOOL(WIFEXITED(status));
|
|
}
|
|
|
|
|
|
/*
|
|
* call-seq:
|
|
* exitstatus -> integer or nil
|
|
*
|
|
* Returns the least significant eight bits of the return code
|
|
* of the process if it has exited;
|
|
* +nil+ otherwise:
|
|
*
|
|
* `exit 99`
|
|
* $?.exitstatus # => 99
|
|
*
|
|
*/
|
|
|
|
static VALUE
|
|
pst_wexitstatus(VALUE st)
|
|
{
|
|
int status = PST2INT(st);
|
|
|
|
if (WIFEXITED(status))
|
|
return INT2NUM(WEXITSTATUS(status));
|
|
return Qnil;
|
|
}
|
|
|
|
|
|
/*
|
|
* call-seq:
|
|
* success? -> true, false, or nil
|
|
*
|
|
* Returns:
|
|
*
|
|
* - +true+ if the process has completed successfully and exited.
|
|
* - +false+ if the process has completed unsuccessfully and exited.
|
|
* - +nil+ if the process has not exited.
|
|
*
|
|
*/
|
|
|
|
static VALUE
|
|
pst_success_p(VALUE st)
|
|
{
|
|
int status = PST2INT(st);
|
|
|
|
if (!WIFEXITED(status))
|
|
return Qnil;
|
|
return RBOOL(WEXITSTATUS(status) == EXIT_SUCCESS);
|
|
}
|
|
|
|
|
|
/*
|
|
* call-seq:
|
|
* coredump? -> true or false
|
|
*
|
|
* Returns +true+ if the process generated a coredump
|
|
* when it terminated, +false+ if not.
|
|
*
|
|
* Not available on all platforms.
|
|
*/
|
|
|
|
static VALUE
|
|
pst_wcoredump(VALUE st)
|
|
{
|
|
#ifdef WCOREDUMP
|
|
int status = PST2INT(st);
|
|
|
|
return RBOOL(WCOREDUMP(status));
|
|
#else
|
|
return Qfalse;
|
|
#endif
|
|
}
|
|
|
|
static rb_pid_t
|
|
do_waitpid(rb_pid_t pid, int *st, int flags)
|
|
{
|
|
#if defined HAVE_WAITPID
|
|
return waitpid(pid, st, flags);
|
|
#elif defined HAVE_WAIT4
|
|
return wait4(pid, st, flags, NULL);
|
|
#else
|
|
# error waitpid or wait4 is required.
|
|
#endif
|
|
}
|
|
|
|
struct waitpid_state {
|
|
struct ccan_list_node wnode;
|
|
rb_execution_context_t *ec;
|
|
rb_nativethread_cond_t *cond;
|
|
rb_pid_t ret;
|
|
rb_pid_t pid;
|
|
int status;
|
|
int options;
|
|
int errnum;
|
|
};
|
|
|
|
static void
|
|
waitpid_state_init(struct waitpid_state *w, rb_pid_t pid, int options)
|
|
{
|
|
w->ret = 0;
|
|
w->pid = pid;
|
|
w->options = options;
|
|
w->errnum = 0;
|
|
w->status = 0;
|
|
}
|
|
|
|
static void *
|
|
waitpid_blocking_no_SIGCHLD(void *x)
|
|
{
|
|
struct waitpid_state *w = x;
|
|
|
|
w->ret = do_waitpid(w->pid, &w->status, w->options);
|
|
|
|
return 0;
|
|
}
|
|
|
|
static void
|
|
waitpid_no_SIGCHLD(struct waitpid_state *w)
|
|
{
|
|
if (w->options & WNOHANG) {
|
|
w->ret = do_waitpid(w->pid, &w->status, w->options);
|
|
}
|
|
else {
|
|
do {
|
|
rb_thread_call_without_gvl(waitpid_blocking_no_SIGCHLD, w, RUBY_UBF_PROCESS, 0);
|
|
} while (w->ret < 0 && errno == EINTR && (RUBY_VM_CHECK_INTS(w->ec),1));
|
|
}
|
|
if (w->ret == -1)
|
|
w->errnum = errno;
|
|
}
|
|
|
|
VALUE
|
|
rb_process_status_wait(rb_pid_t pid, int flags)
|
|
{
|
|
// We only enter the scheduler if we are "blocking":
|
|
if (!(flags & WNOHANG)) {
|
|
VALUE scheduler = rb_fiber_scheduler_current();
|
|
VALUE result = rb_fiber_scheduler_process_wait(scheduler, pid, flags);
|
|
if (!UNDEF_P(result)) return result;
|
|
}
|
|
|
|
struct waitpid_state waitpid_state;
|
|
|
|
waitpid_state_init(&waitpid_state, pid, flags);
|
|
waitpid_state.ec = GET_EC();
|
|
|
|
waitpid_no_SIGCHLD(&waitpid_state);
|
|
|
|
if (waitpid_state.ret == 0) return Qnil;
|
|
|
|
return rb_process_status_new(waitpid_state.ret, waitpid_state.status, waitpid_state.errnum);
|
|
}
|
|
|
|
/*
|
|
* call-seq:
|
|
* Process::Status.wait(pid = -1, flags = 0) -> Process::Status
|
|
*
|
|
* Like Process.wait, but returns a Process::Status object
|
|
* (instead of an integer pid or nil);
|
|
* see Process.wait for the values of +pid+ and +flags+.
|
|
*
|
|
* If there are child processes,
|
|
* waits for a child process to exit and returns a Process::Status object
|
|
* containing information on that process;
|
|
* sets thread-local variable <tt>$?</tt>:
|
|
*
|
|
* Process.spawn('cat /nop') # => 1155880
|
|
* Process::Status.wait # => #<Process::Status: pid 1155880 exit 1>
|
|
* $? # => #<Process::Status: pid 1155508 exit 1>
|
|
*
|
|
* If there is no child process,
|
|
* returns an "empty" Process::Status object
|
|
* that does not represent an actual process;
|
|
* does not set thread-local variable <tt>$?</tt>:
|
|
*
|
|
* Process::Status.wait # => #<Process::Status: pid -1 exit 0>
|
|
* $? # => #<Process::Status: pid 1155508 exit 1> # Unchanged.
|
|
*
|
|
* May invoke the scheduler hook Fiber::Scheduler#process_wait.
|
|
*
|
|
* Not available on all platforms.
|
|
*/
|
|
|
|
static VALUE
|
|
rb_process_status_waitv(int argc, VALUE *argv, VALUE _)
|
|
{
|
|
rb_check_arity(argc, 0, 2);
|
|
|
|
rb_pid_t pid = -1;
|
|
int flags = 0;
|
|
|
|
if (argc >= 1) {
|
|
pid = NUM2PIDT(argv[0]);
|
|
}
|
|
|
|
if (argc >= 2) {
|
|
flags = RB_NUM2INT(argv[1]);
|
|
}
|
|
|
|
return rb_process_status_wait(pid, flags);
|
|
}
|
|
|
|
rb_pid_t
|
|
rb_waitpid(rb_pid_t pid, int *st, int flags)
|
|
{
|
|
VALUE status = rb_process_status_wait(pid, flags);
|
|
if (NIL_P(status)) return 0;
|
|
|
|
struct rb_process_status *data = rb_check_typeddata(status, &rb_process_status_type);
|
|
pid = data->pid;
|
|
|
|
if (st) *st = data->status;
|
|
|
|
if (pid == -1) {
|
|
errno = data->error;
|
|
}
|
|
else {
|
|
GET_THREAD()->last_status = status;
|
|
}
|
|
|
|
return pid;
|
|
}
|
|
|
|
static VALUE
|
|
proc_wait(int argc, VALUE *argv)
|
|
{
|
|
rb_pid_t pid;
|
|
int flags, status;
|
|
|
|
flags = 0;
|
|
if (rb_check_arity(argc, 0, 2) == 0) {
|
|
pid = -1;
|
|
}
|
|
else {
|
|
VALUE vflags;
|
|
pid = NUM2PIDT(argv[0]);
|
|
if (argc == 2 && !NIL_P(vflags = argv[1])) {
|
|
flags = NUM2UINT(vflags);
|
|
}
|
|
}
|
|
|
|
if ((pid = rb_waitpid(pid, &status, flags)) < 0)
|
|
rb_sys_fail(0);
|
|
|
|
if (pid == 0) {
|
|
rb_last_status_clear();
|
|
return Qnil;
|
|
}
|
|
|
|
return PIDT2NUM(pid);
|
|
}
|
|
|
|
/* [MG]:FIXME: I wasn't sure how this should be done, since ::wait()
|
|
has historically been documented as if it didn't take any arguments
|
|
despite the fact that it's just an alias for ::waitpid(). The way I
|
|
have it below is more truthful, but a little confusing.
|
|
|
|
I also took the liberty of putting in the pid values, as they're
|
|
pretty useful, and it looked as if the original 'ri' output was
|
|
supposed to contain them after "[...]depending on the value of
|
|
aPid:".
|
|
|
|
The 'ansi' and 'bs' formats of the ri output don't display the
|
|
definition list for some reason, but the plain text one does.
|
|
*/
|
|
|
|
/*
|
|
* call-seq:
|
|
* Process.wait(pid = -1, flags = 0) -> integer
|
|
*
|
|
* Waits for a suitable child process to exit, returns its process ID,
|
|
* and sets <tt>$?</tt> to a Process::Status object
|
|
* containing information on that process.
|
|
* Which child it waits for depends on the value of the given +pid+:
|
|
*
|
|
* - Positive integer: Waits for the child process whose process ID is +pid+:
|
|
*
|
|
* pid0 = Process.spawn('ruby', '-e', 'exit 13') # => 230866
|
|
* pid1 = Process.spawn('ruby', '-e', 'exit 14') # => 230891
|
|
* Process.wait(pid0) # => 230866
|
|
* $? # => #<Process::Status: pid 230866 exit 13>
|
|
* Process.wait(pid1) # => 230891
|
|
* $? # => #<Process::Status: pid 230891 exit 14>
|
|
* Process.wait(pid0) # Raises Errno::ECHILD
|
|
*
|
|
* - <tt>0</tt>: Waits for any child process whose group ID
|
|
* is the same as that of the current process:
|
|
*
|
|
* parent_pgpid = Process.getpgid(Process.pid)
|
|
* puts "Parent process group ID is #{parent_pgpid}."
|
|
* child0_pid = fork do
|
|
* puts "Child 0 pid is #{Process.pid}"
|
|
* child0_pgid = Process.getpgid(Process.pid)
|
|
* puts "Child 0 process group ID is #{child0_pgid} (same as parent's)."
|
|
* end
|
|
* child1_pid = fork do
|
|
* puts "Child 1 pid is #{Process.pid}"
|
|
* Process.setpgid(0, Process.pid)
|
|
* child1_pgid = Process.getpgid(Process.pid)
|
|
* puts "Child 1 process group ID is #{child1_pgid} (different from parent's)."
|
|
* end
|
|
* retrieved_pid = Process.wait(0)
|
|
* puts "Process.wait(0) returned pid #{retrieved_pid}, which is child 0 pid."
|
|
* begin
|
|
* Process.wait(0)
|
|
* rescue Errno::ECHILD => x
|
|
* puts "Raised #{x.class}, because child 1 process group ID differs from parent process group ID."
|
|
* end
|
|
*
|
|
* Output:
|
|
*
|
|
* Parent process group ID is 225764.
|
|
* Child 0 pid is 225788
|
|
* Child 0 process group ID is 225764 (same as parent's).
|
|
* Child 1 pid is 225789
|
|
* Child 1 process group ID is 225789 (different from parent's).
|
|
* Process.wait(0) returned pid 225788, which is child 0 pid.
|
|
* Raised Errno::ECHILD, because child 1 process group ID differs from parent process group ID.
|
|
*
|
|
* - <tt>-1</tt> (default): Waits for any child process:
|
|
*
|
|
* parent_pgpid = Process.getpgid(Process.pid)
|
|
* puts "Parent process group ID is #{parent_pgpid}."
|
|
* child0_pid = fork do
|
|
* puts "Child 0 pid is #{Process.pid}"
|
|
* child0_pgid = Process.getpgid(Process.pid)
|
|
* puts "Child 0 process group ID is #{child0_pgid} (same as parent's)."
|
|
* end
|
|
* child1_pid = fork do
|
|
* puts "Child 1 pid is #{Process.pid}"
|
|
* Process.setpgid(0, Process.pid)
|
|
* child1_pgid = Process.getpgid(Process.pid)
|
|
* puts "Child 1 process group ID is #{child1_pgid} (different from parent's)."
|
|
* sleep 3 # To force child 1 to exit later than child 0 exit.
|
|
* end
|
|
* child_pids = [child0_pid, child1_pid]
|
|
* retrieved_pid = Process.wait(-1)
|
|
* puts child_pids.include?(retrieved_pid)
|
|
* retrieved_pid = Process.wait(-1)
|
|
* puts child_pids.include?(retrieved_pid)
|
|
*
|
|
* Output:
|
|
*
|
|
* Parent process group ID is 228736.
|
|
* Child 0 pid is 228758
|
|
* Child 0 process group ID is 228736 (same as parent's).
|
|
* Child 1 pid is 228759
|
|
* Child 1 process group ID is 228759 (different from parent's).
|
|
* true
|
|
* true
|
|
*
|
|
* - Less than <tt>-1</tt>: Waits for any child whose process group ID is <tt>-pid</tt>:
|
|
*
|
|
* parent_pgpid = Process.getpgid(Process.pid)
|
|
* puts "Parent process group ID is #{parent_pgpid}."
|
|
* child0_pid = fork do
|
|
* puts "Child 0 pid is #{Process.pid}"
|
|
* child0_pgid = Process.getpgid(Process.pid)
|
|
* puts "Child 0 process group ID is #{child0_pgid} (same as parent's)."
|
|
* end
|
|
* child1_pid = fork do
|
|
* puts "Child 1 pid is #{Process.pid}"
|
|
* Process.setpgid(0, Process.pid)
|
|
* child1_pgid = Process.getpgid(Process.pid)
|
|
* puts "Child 1 process group ID is #{child1_pgid} (different from parent's)."
|
|
* end
|
|
* sleep 1
|
|
* retrieved_pid = Process.wait(-child1_pid)
|
|
* puts "Process.wait(-child1_pid) returned pid #{retrieved_pid}, which is child 1 pid."
|
|
* begin
|
|
* Process.wait(-child1_pid)
|
|
* rescue Errno::ECHILD => x
|
|
* puts "Raised #{x.class}, because there's no longer a child with process group id #{child1_pid}."
|
|
* end
|
|
*
|
|
* Output:
|
|
*
|
|
* Parent process group ID is 230083.
|
|
* Child 0 pid is 230108
|
|
* Child 0 process group ID is 230083 (same as parent's).
|
|
* Child 1 pid is 230109
|
|
* Child 1 process group ID is 230109 (different from parent's).
|
|
* Process.wait(-child1_pid) returned pid 230109, which is child 1 pid.
|
|
* Raised Errno::ECHILD, because there's no longer a child with process group id 230109.
|
|
*
|
|
* Argument +flags+ should be given as one of the following constants,
|
|
* or as the logical OR of both:
|
|
*
|
|
* - Process::WNOHANG: Does not block if no child process is available.
|
|
* - Process:WUNTRACED: May return a stopped child process, even if not yet reported.
|
|
*
|
|
* Not all flags are available on all platforms.
|
|
*
|
|
* Raises Errno::ECHILD if there is no suitable child process.
|
|
*
|
|
* Not available on all platforms.
|
|
*
|
|
* Process.waitpid is an alias for Process.wait.
|
|
*/
|
|
static VALUE
|
|
proc_m_wait(int c, VALUE *v, VALUE _)
|
|
{
|
|
return proc_wait(c, v);
|
|
}
|
|
|
|
/*
|
|
* call-seq:
|
|
* Process.wait2(pid = -1, flags = 0) -> [pid, status]
|
|
*
|
|
* Like Process.waitpid, but returns an array
|
|
* containing the child process +pid+ and Process::Status +status+:
|
|
*
|
|
* pid = Process.spawn('ruby', '-e', 'exit 13') # => 309581
|
|
* Process.wait2(pid)
|
|
* # => [309581, #<Process::Status: pid 309581 exit 13>]
|
|
*
|
|
* Process.waitpid2 is an alias for Process.waitpid.
|
|
*/
|
|
|
|
static VALUE
|
|
proc_wait2(int argc, VALUE *argv, VALUE _)
|
|
{
|
|
VALUE pid = proc_wait(argc, argv);
|
|
if (NIL_P(pid)) return Qnil;
|
|
return rb_assoc_new(pid, rb_last_status_get());
|
|
}
|
|
|
|
|
|
/*
|
|
* call-seq:
|
|
* Process.waitall -> array
|
|
*
|
|
* Waits for all children, returns an array of 2-element arrays;
|
|
* each subarray contains the integer pid and Process::Status status
|
|
* for one of the reaped child processes:
|
|
*
|
|
* pid0 = Process.spawn('ruby', '-e', 'exit 13') # => 325470
|
|
* pid1 = Process.spawn('ruby', '-e', 'exit 14') # => 325495
|
|
* Process.waitall
|
|
* # => [[325470, #<Process::Status: pid 325470 exit 13>], [325495, #<Process::Status: pid 325495 exit 14>]]
|
|
*
|
|
*/
|
|
|
|
static VALUE
|
|
proc_waitall(VALUE _)
|
|
{
|
|
VALUE result;
|
|
rb_pid_t pid;
|
|
int status;
|
|
|
|
result = rb_ary_new();
|
|
rb_last_status_clear();
|
|
|
|
for (pid = -1;;) {
|
|
pid = rb_waitpid(-1, &status, 0);
|
|
if (pid == -1) {
|
|
int e = errno;
|
|
if (e == ECHILD)
|
|
break;
|
|
rb_syserr_fail(e, 0);
|
|
}
|
|
rb_ary_push(result, rb_assoc_new(PIDT2NUM(pid), rb_last_status_get()));
|
|
}
|
|
return result;
|
|
}
|
|
|
|
static VALUE rb_cWaiter;
|
|
|
|
static VALUE
|
|
detach_process_pid(VALUE thread)
|
|
{
|
|
return rb_thread_local_aref(thread, id_pid);
|
|
}
|
|
|
|
static VALUE
|
|
detach_process_watcher(void *arg)
|
|
{
|
|
rb_pid_t cpid, pid = (rb_pid_t)(VALUE)arg;
|
|
int status;
|
|
|
|
while ((cpid = rb_waitpid(pid, &status, 0)) == 0) {
|
|
/* wait while alive */
|
|
}
|
|
return rb_last_status_get();
|
|
}
|
|
|
|
VALUE
|
|
rb_detach_process(rb_pid_t pid)
|
|
{
|
|
VALUE watcher = rb_thread_create(detach_process_watcher, (void*)(VALUE)pid);
|
|
rb_thread_local_aset(watcher, id_pid, PIDT2NUM(pid));
|
|
RBASIC_SET_CLASS(watcher, rb_cWaiter);
|
|
return watcher;
|
|
}
|
|
|
|
|
|
/*
|
|
* call-seq:
|
|
* Process.detach(pid) -> thread
|
|
*
|
|
* Avoids the potential for a child process to become a
|
|
* {zombie process}[https://en.wikipedia.org/wiki/Zombie_process].
|
|
* Process.detach prevents this by setting up a separate Ruby thread
|
|
* whose sole job is to reap the status of the process _pid_ when it terminates.
|
|
*
|
|
* This method is needed only when the parent process will never wait
|
|
* for the child process.
|
|
*
|
|
* This example does not reap the second child process;
|
|
* that process appears as a zombie in the process status (+ps+) output:
|
|
*
|
|
* pid = Process.spawn('ruby', '-e', 'exit 13') # => 312691
|
|
* sleep(1)
|
|
* # Find zombies.
|
|
* system("ps -ho pid,state -p #{pid}")
|
|
*
|
|
* Output:
|
|
*
|
|
* 312716 Z
|
|
*
|
|
* This example also does not reap the second child process,
|
|
* but it does detach the process so that it does not become a zombie:
|
|
*
|
|
* pid = Process.spawn('ruby', '-e', 'exit 13') # => 313213
|
|
* thread = Process.detach(pid)
|
|
* sleep(1)
|
|
* # => #<Process::Waiter:0x00007f038f48b838 run>
|
|
* system("ps -ho pid,state -p #{pid}") # Finds no zombies.
|
|
*
|
|
* The waiting thread can return the pid of the detached child process:
|
|
*
|
|
* thread.join.pid # => 313262
|
|
*
|
|
*/
|
|
|
|
static VALUE
|
|
proc_detach(VALUE obj, VALUE pid)
|
|
{
|
|
return rb_detach_process(NUM2PIDT(pid));
|
|
}
|
|
|
|
/* This function should be async-signal-safe. Actually it is. */
|
|
static void
|
|
before_exec_async_signal_safe(void)
|
|
{
|
|
}
|
|
|
|
static void
|
|
before_exec_non_async_signal_safe(void)
|
|
{
|
|
/*
|
|
* On Mac OS X 10.5.x (Leopard) or earlier, exec() may return ENOTSUP
|
|
* if the process have multiple threads. Therefore we have to kill
|
|
* internal threads temporary. [ruby-core:10583]
|
|
* This is also true on Haiku. It returns Errno::EPERM against exec()
|
|
* in multiple threads.
|
|
*
|
|
* Nowadays, we always stop the timer thread completely to allow redirects.
|
|
*/
|
|
rb_thread_stop_timer_thread();
|
|
}
|
|
|
|
#define WRITE_CONST(fd, str) (void)(write((fd),(str),sizeof(str)-1)<0)
|
|
#ifdef _WIN32
|
|
int rb_w32_set_nonblock2(int fd, int nonblock);
|
|
#endif
|
|
|
|
static int
|
|
set_blocking(int fd)
|
|
{
|
|
#ifdef _WIN32
|
|
return rb_w32_set_nonblock2(fd, 0);
|
|
#elif defined(F_GETFL) && defined(F_SETFL)
|
|
int fl = fcntl(fd, F_GETFL); /* async-signal-safe */
|
|
|
|
/* EBADF ought to be possible */
|
|
if (fl == -1) return fl;
|
|
if (fl & O_NONBLOCK) {
|
|
fl &= ~O_NONBLOCK;
|
|
return fcntl(fd, F_SETFL, fl);
|
|
}
|
|
return 0;
|
|
#endif
|
|
}
|
|
|
|
static void
|
|
stdfd_clear_nonblock(void)
|
|
{
|
|
/* many programs cannot deal with non-blocking stdin/stdout/stderr */
|
|
int fd;
|
|
for (fd = 0; fd < 3; fd++) {
|
|
(void)set_blocking(fd); /* can't do much about errors anyhow */
|
|
}
|
|
}
|
|
|
|
static void
|
|
before_exec(void)
|
|
{
|
|
before_exec_non_async_signal_safe();
|
|
before_exec_async_signal_safe();
|
|
}
|
|
|
|
/* This function should be async-signal-safe. Actually it is. */
|
|
static void
|
|
after_exec_async_signal_safe(void)
|
|
{
|
|
}
|
|
|
|
static void
|
|
after_exec_non_async_signal_safe(void)
|
|
{
|
|
rb_thread_reset_timer_thread();
|
|
rb_thread_start_timer_thread();
|
|
}
|
|
|
|
static void
|
|
after_exec(void)
|
|
{
|
|
after_exec_async_signal_safe();
|
|
after_exec_non_async_signal_safe();
|
|
}
|
|
|
|
#if defined HAVE_WORKING_FORK || defined HAVE_DAEMON
|
|
static void
|
|
before_fork_ruby(void)
|
|
{
|
|
before_exec();
|
|
}
|
|
|
|
static void
|
|
after_fork_ruby(rb_pid_t pid)
|
|
{
|
|
rb_threadptr_pending_interrupt_clear(GET_THREAD());
|
|
if (pid == 0) {
|
|
clear_pid_cache();
|
|
rb_thread_atfork();
|
|
}
|
|
after_exec();
|
|
}
|
|
#endif
|
|
|
|
#if defined(HAVE_WORKING_FORK)
|
|
|
|
COMPILER_WARNING_PUSH
|
|
#if __has_warning("-Wdeprecated-declarations") || RBIMPL_COMPILER_IS(GCC)
|
|
COMPILER_WARNING_IGNORED(-Wdeprecated-declarations)
|
|
#endif
|
|
static inline rb_pid_t
|
|
rb_fork(void)
|
|
{
|
|
return fork();
|
|
}
|
|
COMPILER_WARNING_POP
|
|
|
|
/* try_with_sh and exec_with_sh should be async-signal-safe. Actually it is.*/
|
|
#define try_with_sh(err, prog, argv, envp) ((err == ENOEXEC) ? exec_with_sh((prog), (argv), (envp)) : (void)0)
|
|
static void
|
|
exec_with_sh(const char *prog, char **argv, char **envp)
|
|
{
|
|
*argv = (char *)prog;
|
|
*--argv = (char *)"sh";
|
|
if (envp)
|
|
execve("/bin/sh", argv, envp); /* async-signal-safe */
|
|
else
|
|
execv("/bin/sh", argv); /* async-signal-safe (since SUSv4) */
|
|
}
|
|
|
|
#else
|
|
#define try_with_sh(err, prog, argv, envp) (void)0
|
|
#endif
|
|
|
|
/* This function should be async-signal-safe. Actually it is. */
|
|
static int
|
|
proc_exec_cmd(const char *prog, VALUE argv_str, VALUE envp_str)
|
|
{
|
|
char **argv;
|
|
#ifndef _WIN32
|
|
char **envp;
|
|
int err;
|
|
#endif
|
|
|
|
argv = ARGVSTR2ARGV(argv_str);
|
|
|
|
if (!prog) {
|
|
return ENOENT;
|
|
}
|
|
|
|
#ifdef _WIN32
|
|
rb_w32_uaspawn(P_OVERLAY, prog, argv);
|
|
return errno;
|
|
#else
|
|
envp = envp_str ? RB_IMEMO_TMPBUF_PTR(envp_str) : NULL;
|
|
if (envp_str)
|
|
execve(prog, argv, envp); /* async-signal-safe */
|
|
else
|
|
execv(prog, argv); /* async-signal-safe (since SUSv4) */
|
|
err = errno;
|
|
try_with_sh(err, prog, argv, envp); /* try_with_sh() is async-signal-safe. */
|
|
return err;
|
|
#endif
|
|
}
|
|
|
|
/* This function should be async-signal-safe. Actually it is. */
|
|
static int
|
|
proc_exec_sh(const char *str, VALUE envp_str)
|
|
{
|
|
const char *s;
|
|
|
|
s = str;
|
|
while (*s == ' ' || *s == '\t' || *s == '\n')
|
|
s++;
|
|
|
|
if (!*s) {
|
|
return ENOENT;
|
|
}
|
|
|
|
#ifdef _WIN32
|
|
rb_w32_uspawn(P_OVERLAY, (char *)str, 0);
|
|
#elif defined(__CYGWIN32__)
|
|
{
|
|
char fbuf[MAXPATHLEN];
|
|
char *shell = dln_find_exe_r("sh", 0, fbuf, sizeof(fbuf));
|
|
int status = -1;
|
|
if (shell)
|
|
execl(shell, "sh", "-c", str, (char *) NULL);
|
|
else
|
|
status = system(str);
|
|
if (status != -1)
|
|
exit(status);
|
|
}
|
|
#else
|
|
if (envp_str)
|
|
execle("/bin/sh", "sh", "-c", str, (char *)NULL, RB_IMEMO_TMPBUF_PTR(envp_str)); /* async-signal-safe */
|
|
else
|
|
execl("/bin/sh", "sh", "-c", str, (char *)NULL); /* async-signal-safe (since SUSv4) */
|
|
#endif /* _WIN32 */
|
|
return errno;
|
|
}
|
|
|
|
int
|
|
rb_proc_exec(const char *str)
|
|
{
|
|
int ret;
|
|
before_exec();
|
|
ret = proc_exec_sh(str, Qfalse);
|
|
after_exec();
|
|
errno = ret;
|
|
return -1;
|
|
}
|
|
|
|
static void
|
|
mark_exec_arg(void *ptr)
|
|
{
|
|
struct rb_execarg *eargp = ptr;
|
|
if (eargp->use_shell)
|
|
rb_gc_mark(eargp->invoke.sh.shell_script);
|
|
else {
|
|
rb_gc_mark(eargp->invoke.cmd.command_name);
|
|
rb_gc_mark(eargp->invoke.cmd.command_abspath);
|
|
rb_gc_mark(eargp->invoke.cmd.argv_str);
|
|
rb_gc_mark(eargp->invoke.cmd.argv_buf);
|
|
}
|
|
rb_gc_mark(eargp->redirect_fds);
|
|
rb_gc_mark(eargp->envp_str);
|
|
rb_gc_mark(eargp->envp_buf);
|
|
rb_gc_mark(eargp->dup2_tmpbuf);
|
|
rb_gc_mark(eargp->rlimit_limits);
|
|
rb_gc_mark(eargp->fd_dup2);
|
|
rb_gc_mark(eargp->fd_close);
|
|
rb_gc_mark(eargp->fd_open);
|
|
rb_gc_mark(eargp->fd_dup2_child);
|
|
rb_gc_mark(eargp->env_modification);
|
|
rb_gc_mark(eargp->path_env);
|
|
rb_gc_mark(eargp->chdir_dir);
|
|
}
|
|
|
|
static size_t
|
|
memsize_exec_arg(const void *ptr)
|
|
{
|
|
return sizeof(struct rb_execarg);
|
|
}
|
|
|
|
static const rb_data_type_t exec_arg_data_type = {
|
|
"exec_arg",
|
|
{mark_exec_arg, RUBY_TYPED_DEFAULT_FREE, memsize_exec_arg},
|
|
0, 0, RUBY_TYPED_FREE_IMMEDIATELY
|
|
};
|
|
|
|
#ifdef _WIN32
|
|
# define DEFAULT_PROCESS_ENCODING rb_utf8_encoding()
|
|
#endif
|
|
#ifdef DEFAULT_PROCESS_ENCODING
|
|
# define EXPORT_STR(str) rb_str_export_to_enc((str), DEFAULT_PROCESS_ENCODING)
|
|
# define EXPORT_DUP(str) export_dup(str)
|
|
static VALUE
|
|
export_dup(VALUE str)
|
|
{
|
|
VALUE newstr = EXPORT_STR(str);
|
|
if (newstr == str) newstr = rb_str_dup(str);
|
|
return newstr;
|
|
}
|
|
#else
|
|
# define EXPORT_STR(str) (str)
|
|
# define EXPORT_DUP(str) rb_str_dup(str)
|
|
#endif
|
|
|
|
#if !defined(HAVE_WORKING_FORK) && defined(HAVE_SPAWNV)
|
|
# define USE_SPAWNV 1
|
|
#else
|
|
# define USE_SPAWNV 0
|
|
#endif
|
|
#ifndef P_NOWAIT
|
|
# define P_NOWAIT _P_NOWAIT
|
|
#endif
|
|
|
|
#if USE_SPAWNV
|
|
#if defined(_WIN32)
|
|
#define proc_spawn_cmd_internal(argv, prog) rb_w32_uaspawn(P_NOWAIT, (prog), (argv))
|
|
#else
|
|
static rb_pid_t
|
|
proc_spawn_cmd_internal(char **argv, char *prog)
|
|
{
|
|
char fbuf[MAXPATHLEN];
|
|
rb_pid_t status;
|
|
|
|
if (!prog)
|
|
prog = argv[0];
|
|
prog = dln_find_exe_r(prog, 0, fbuf, sizeof(fbuf));
|
|
if (!prog)
|
|
return -1;
|
|
|
|
before_exec();
|
|
status = spawnv(P_NOWAIT, prog, (const char **)argv);
|
|
if (status == -1 && errno == ENOEXEC) {
|
|
*argv = (char *)prog;
|
|
*--argv = (char *)"sh";
|
|
status = spawnv(P_NOWAIT, "/bin/sh", (const char **)argv);
|
|
after_exec();
|
|
if (status == -1) errno = ENOEXEC;
|
|
}
|
|
return status;
|
|
}
|
|
#endif
|
|
|
|
static rb_pid_t
|
|
proc_spawn_cmd(char **argv, VALUE prog, struct rb_execarg *eargp)
|
|
{
|
|
rb_pid_t pid = -1;
|
|
|
|
if (argv[0]) {
|
|
#if defined(_WIN32)
|
|
DWORD flags = 0;
|
|
if (eargp->new_pgroup_given && eargp->new_pgroup_flag) {
|
|
flags = CREATE_NEW_PROCESS_GROUP;
|
|
}
|
|
pid = rb_w32_uaspawn_flags(P_NOWAIT, prog ? RSTRING_PTR(prog) : 0, argv, flags);
|
|
#else
|
|
pid = proc_spawn_cmd_internal(argv, prog ? RSTRING_PTR(prog) : 0);
|
|
#endif
|
|
}
|
|
return pid;
|
|
}
|
|
|
|
#if defined(_WIN32)
|
|
#define proc_spawn_sh(str) rb_w32_uspawn(P_NOWAIT, (str), 0)
|
|
#else
|
|
static rb_pid_t
|
|
proc_spawn_sh(char *str)
|
|
{
|
|
char fbuf[MAXPATHLEN];
|
|
rb_pid_t status;
|
|
|
|
char *shell = dln_find_exe_r("sh", 0, fbuf, sizeof(fbuf));
|
|
before_exec();
|
|
status = spawnl(P_NOWAIT, (shell ? shell : "/bin/sh"), "sh", "-c", str, (char*)NULL);
|
|
after_exec();
|
|
return status;
|
|
}
|
|
#endif
|
|
#endif
|
|
|
|
static VALUE
|
|
hide_obj(VALUE obj)
|
|
{
|
|
RBASIC_CLEAR_CLASS(obj);
|
|
return obj;
|
|
}
|
|
|
|
static VALUE
|
|
check_exec_redirect_fd(VALUE v, int iskey)
|
|
{
|
|
VALUE tmp;
|
|
int fd;
|
|
if (FIXNUM_P(v)) {
|
|
fd = FIX2INT(v);
|
|
}
|
|
else if (SYMBOL_P(v)) {
|
|
ID id = rb_check_id(&v);
|
|
if (id == id_in)
|
|
fd = 0;
|
|
else if (id == id_out)
|
|
fd = 1;
|
|
else if (id == id_err)
|
|
fd = 2;
|
|
else
|
|
goto wrong;
|
|
}
|
|
else if (!NIL_P(tmp = rb_io_check_io(v))) {
|
|
rb_io_t *fptr;
|
|
GetOpenFile(tmp, fptr);
|
|
if (fptr->tied_io_for_writing)
|
|
rb_raise(rb_eArgError, "duplex IO redirection");
|
|
fd = fptr->fd;
|
|
}
|
|
else {
|
|
goto wrong;
|
|
}
|
|
if (fd < 0) {
|
|
rb_raise(rb_eArgError, "negative file descriptor");
|
|
}
|
|
#ifdef _WIN32
|
|
else if (fd >= 3 && iskey) {
|
|
rb_raise(rb_eArgError, "wrong file descriptor (%d)", fd);
|
|
}
|
|
#endif
|
|
return INT2FIX(fd);
|
|
|
|
wrong:
|
|
rb_raise(rb_eArgError, "wrong exec redirect");
|
|
UNREACHABLE_RETURN(Qundef);
|
|
}
|
|
|
|
static VALUE
|
|
check_exec_redirect1(VALUE ary, VALUE key, VALUE param)
|
|
{
|
|
if (ary == Qfalse) {
|
|
ary = hide_obj(rb_ary_new());
|
|
}
|
|
if (!RB_TYPE_P(key, T_ARRAY)) {
|
|
VALUE fd = check_exec_redirect_fd(key, !NIL_P(param));
|
|
rb_ary_push(ary, hide_obj(rb_assoc_new(fd, param)));
|
|
}
|
|
else {
|
|
int i;
|
|
for (i = 0 ; i < RARRAY_LEN(key); i++) {
|
|
VALUE v = RARRAY_AREF(key, i);
|
|
VALUE fd = check_exec_redirect_fd(v, !NIL_P(param));
|
|
rb_ary_push(ary, hide_obj(rb_assoc_new(fd, param)));
|
|
}
|
|
}
|
|
return ary;
|
|
}
|
|
|
|
static void
|
|
check_exec_redirect(VALUE key, VALUE val, struct rb_execarg *eargp)
|
|
{
|
|
VALUE param;
|
|
VALUE path, flags, perm;
|
|
VALUE tmp;
|
|
ID id;
|
|
|
|
switch (TYPE(val)) {
|
|
case T_SYMBOL:
|
|
id = rb_check_id(&val);
|
|
if (id == id_close) {
|
|
param = Qnil;
|
|
eargp->fd_close = check_exec_redirect1(eargp->fd_close, key, param);
|
|
}
|
|
else if (id == id_in) {
|
|
param = INT2FIX(0);
|
|
eargp->fd_dup2 = check_exec_redirect1(eargp->fd_dup2, key, param);
|
|
}
|
|
else if (id == id_out) {
|
|
param = INT2FIX(1);
|
|
eargp->fd_dup2 = check_exec_redirect1(eargp->fd_dup2, key, param);
|
|
}
|
|
else if (id == id_err) {
|
|
param = INT2FIX(2);
|
|
eargp->fd_dup2 = check_exec_redirect1(eargp->fd_dup2, key, param);
|
|
}
|
|
else {
|
|
rb_raise(rb_eArgError, "wrong exec redirect symbol: %"PRIsVALUE,
|
|
val);
|
|
}
|
|
break;
|
|
|
|
case T_FILE:
|
|
io:
|
|
val = check_exec_redirect_fd(val, 0);
|
|
/* fall through */
|
|
case T_FIXNUM:
|
|
param = val;
|
|
eargp->fd_dup2 = check_exec_redirect1(eargp->fd_dup2, key, param);
|
|
break;
|
|
|
|
case T_ARRAY:
|
|
path = rb_ary_entry(val, 0);
|
|
if (RARRAY_LEN(val) == 2 && SYMBOL_P(path) &&
|
|
path == ID2SYM(id_child)) {
|
|
param = check_exec_redirect_fd(rb_ary_entry(val, 1), 0);
|
|
eargp->fd_dup2_child = check_exec_redirect1(eargp->fd_dup2_child, key, param);
|
|
}
|
|
else {
|
|
FilePathValue(path);
|
|
flags = rb_ary_entry(val, 1);
|
|
if (NIL_P(flags))
|
|
flags = INT2NUM(O_RDONLY);
|
|
else if (RB_TYPE_P(flags, T_STRING))
|
|
flags = INT2NUM(rb_io_modestr_oflags(StringValueCStr(flags)));
|
|
else
|
|
flags = rb_to_int(flags);
|
|
perm = rb_ary_entry(val, 2);
|
|
perm = NIL_P(perm) ? INT2FIX(0644) : rb_to_int(perm);
|
|
param = hide_obj(rb_ary_new3(4, hide_obj(EXPORT_DUP(path)),
|
|
flags, perm, Qnil));
|
|
eargp->fd_open = check_exec_redirect1(eargp->fd_open, key, param);
|
|
}
|
|
break;
|
|
|
|
case T_STRING:
|
|
path = val;
|
|
FilePathValue(path);
|
|
if (RB_TYPE_P(key, T_FILE))
|
|
key = check_exec_redirect_fd(key, 1);
|
|
if (FIXNUM_P(key) && (FIX2INT(key) == 1 || FIX2INT(key) == 2))
|
|
flags = INT2NUM(O_WRONLY|O_CREAT|O_TRUNC);
|
|
else if (RB_TYPE_P(key, T_ARRAY)) {
|
|
int i;
|
|
for (i = 0; i < RARRAY_LEN(key); i++) {
|
|
VALUE v = RARRAY_AREF(key, i);
|
|
VALUE fd = check_exec_redirect_fd(v, 1);
|
|
if (FIX2INT(fd) != 1 && FIX2INT(fd) != 2) break;
|
|
}
|
|
if (i == RARRAY_LEN(key))
|
|
flags = INT2NUM(O_WRONLY|O_CREAT|O_TRUNC);
|
|
else
|
|
flags = INT2NUM(O_RDONLY);
|
|
}
|
|
else
|
|
flags = INT2NUM(O_RDONLY);
|
|
perm = INT2FIX(0644);
|
|
param = hide_obj(rb_ary_new3(4, hide_obj(EXPORT_DUP(path)),
|
|
flags, perm, Qnil));
|
|
eargp->fd_open = check_exec_redirect1(eargp->fd_open, key, param);
|
|
break;
|
|
|
|
default:
|
|
tmp = val;
|
|
val = rb_io_check_io(tmp);
|
|
if (!NIL_P(val)) goto io;
|
|
rb_raise(rb_eArgError, "wrong exec redirect action");
|
|
}
|
|
|
|
}
|
|
|
|
#if defined(HAVE_SETRLIMIT) && defined(NUM2RLIM)
|
|
static int rlimit_type_by_sym(VALUE key);
|
|
|
|
static void
|
|
rb_execarg_addopt_rlimit(struct rb_execarg *eargp, int rtype, VALUE val)
|
|
{
|
|
VALUE ary = eargp->rlimit_limits;
|
|
VALUE tmp, softlim, hardlim;
|
|
if (eargp->rlimit_limits == Qfalse)
|
|
ary = eargp->rlimit_limits = hide_obj(rb_ary_new());
|
|
else
|
|
ary = eargp->rlimit_limits;
|
|
tmp = rb_check_array_type(val);
|
|
if (!NIL_P(tmp)) {
|
|
if (RARRAY_LEN(tmp) == 1)
|
|
softlim = hardlim = rb_to_int(rb_ary_entry(tmp, 0));
|
|
else if (RARRAY_LEN(tmp) == 2) {
|
|
softlim = rb_to_int(rb_ary_entry(tmp, 0));
|
|
hardlim = rb_to_int(rb_ary_entry(tmp, 1));
|
|
}
|
|
else {
|
|
rb_raise(rb_eArgError, "wrong exec rlimit option");
|
|
}
|
|
}
|
|
else {
|
|
softlim = hardlim = rb_to_int(val);
|
|
}
|
|
tmp = hide_obj(rb_ary_new3(3, INT2NUM(rtype), softlim, hardlim));
|
|
rb_ary_push(ary, tmp);
|
|
}
|
|
#endif
|
|
|
|
#define TO_BOOL(val, name) (NIL_P(val) ? 0 : rb_bool_expected((val), name, TRUE))
|
|
int
|
|
rb_execarg_addopt(VALUE execarg_obj, VALUE key, VALUE val)
|
|
{
|
|
struct rb_execarg *eargp = rb_execarg_get(execarg_obj);
|
|
|
|
ID id;
|
|
|
|
switch (TYPE(key)) {
|
|
case T_SYMBOL:
|
|
#if defined(HAVE_SETRLIMIT) && defined(NUM2RLIM)
|
|
{
|
|
int rtype = rlimit_type_by_sym(key);
|
|
if (rtype != -1) {
|
|
rb_execarg_addopt_rlimit(eargp, rtype, val);
|
|
RB_GC_GUARD(execarg_obj);
|
|
return ST_CONTINUE;
|
|
}
|
|
}
|
|
#endif
|
|
if (!(id = rb_check_id(&key))) return ST_STOP;
|
|
#ifdef HAVE_SETPGID
|
|
if (id == id_pgroup) {
|
|
rb_pid_t pgroup;
|
|
if (eargp->pgroup_given) {
|
|
rb_raise(rb_eArgError, "pgroup option specified twice");
|
|
}
|
|
if (!RTEST(val))
|
|
pgroup = -1; /* asis(-1) means "don't call setpgid()". */
|
|
else if (val == Qtrue)
|
|
pgroup = 0; /* new process group. */
|
|
else {
|
|
pgroup = NUM2PIDT(val);
|
|
if (pgroup < 0) {
|
|
rb_raise(rb_eArgError, "negative process group ID : %ld", (long)pgroup);
|
|
}
|
|
}
|
|
eargp->pgroup_given = 1;
|
|
eargp->pgroup_pgid = pgroup;
|
|
}
|
|
else
|
|
#endif
|
|
#ifdef _WIN32
|
|
if (id == id_new_pgroup) {
|
|
if (eargp->new_pgroup_given) {
|
|
rb_raise(rb_eArgError, "new_pgroup option specified twice");
|
|
}
|
|
eargp->new_pgroup_given = 1;
|
|
eargp->new_pgroup_flag = TO_BOOL(val, "new_pgroup");
|
|
}
|
|
else
|
|
#endif
|
|
if (id == id_unsetenv_others) {
|
|
if (eargp->unsetenv_others_given) {
|
|
rb_raise(rb_eArgError, "unsetenv_others option specified twice");
|
|
}
|
|
eargp->unsetenv_others_given = 1;
|
|
eargp->unsetenv_others_do = TO_BOOL(val, "unsetenv_others");
|
|
}
|
|
else if (id == id_chdir) {
|
|
if (eargp->chdir_given) {
|
|
rb_raise(rb_eArgError, "chdir option specified twice");
|
|
}
|
|
FilePathValue(val);
|
|
val = rb_str_encode_ospath(val);
|
|
eargp->chdir_given = 1;
|
|
eargp->chdir_dir = hide_obj(EXPORT_DUP(val));
|
|
}
|
|
else if (id == id_umask) {
|
|
mode_t cmask = NUM2MODET(val);
|
|
if (eargp->umask_given) {
|
|
rb_raise(rb_eArgError, "umask option specified twice");
|
|
}
|
|
eargp->umask_given = 1;
|
|
eargp->umask_mask = cmask;
|
|
}
|
|
else if (id == id_close_others) {
|
|
if (eargp->close_others_given) {
|
|
rb_raise(rb_eArgError, "close_others option specified twice");
|
|
}
|
|
eargp->close_others_given = 1;
|
|
eargp->close_others_do = TO_BOOL(val, "close_others");
|
|
}
|
|
else if (id == id_in) {
|
|
key = INT2FIX(0);
|
|
goto redirect;
|
|
}
|
|
else if (id == id_out) {
|
|
key = INT2FIX(1);
|
|
goto redirect;
|
|
}
|
|
else if (id == id_err) {
|
|
key = INT2FIX(2);
|
|
goto redirect;
|
|
}
|
|
else if (id == id_uid) {
|
|
#ifdef HAVE_SETUID
|
|
if (eargp->uid_given) {
|
|
rb_raise(rb_eArgError, "uid option specified twice");
|
|
}
|
|
check_uid_switch();
|
|
{
|
|
eargp->uid = OBJ2UID(val);
|
|
eargp->uid_given = 1;
|
|
}
|
|
#else
|
|
rb_raise(rb_eNotImpError,
|
|
"uid option is unimplemented on this machine");
|
|
#endif
|
|
}
|
|
else if (id == id_gid) {
|
|
#ifdef HAVE_SETGID
|
|
if (eargp->gid_given) {
|
|
rb_raise(rb_eArgError, "gid option specified twice");
|
|
}
|
|
check_gid_switch();
|
|
{
|
|
eargp->gid = OBJ2GID(val);
|
|
eargp->gid_given = 1;
|
|
}
|
|
#else
|
|
rb_raise(rb_eNotImpError,
|
|
"gid option is unimplemented on this machine");
|
|
#endif
|
|
}
|
|
else if (id == id_exception) {
|
|
if (eargp->exception_given) {
|
|
rb_raise(rb_eArgError, "exception option specified twice");
|
|
}
|
|
eargp->exception_given = 1;
|
|
eargp->exception = TO_BOOL(val, "exception");
|
|
}
|
|
else {
|
|
return ST_STOP;
|
|
}
|
|
break;
|
|
|
|
case T_FIXNUM:
|
|
case T_FILE:
|
|
case T_ARRAY:
|
|
redirect:
|
|
check_exec_redirect(key, val, eargp);
|
|
break;
|
|
|
|
default:
|
|
return ST_STOP;
|
|
}
|
|
|
|
RB_GC_GUARD(execarg_obj);
|
|
return ST_CONTINUE;
|
|
}
|
|
|
|
static int
|
|
check_exec_options_i(st_data_t st_key, st_data_t st_val, st_data_t arg)
|
|
{
|
|
VALUE key = (VALUE)st_key;
|
|
VALUE val = (VALUE)st_val;
|
|
VALUE execarg_obj = (VALUE)arg;
|
|
if (rb_execarg_addopt(execarg_obj, key, val) != ST_CONTINUE) {
|
|
if (SYMBOL_P(key))
|
|
rb_raise(rb_eArgError, "wrong exec option symbol: % "PRIsVALUE,
|
|
key);
|
|
rb_raise(rb_eArgError, "wrong exec option");
|
|
}
|
|
return ST_CONTINUE;
|
|
}
|
|
|
|
static int
|
|
check_exec_options_i_extract(st_data_t st_key, st_data_t st_val, st_data_t arg)
|
|
{
|
|
VALUE key = (VALUE)st_key;
|
|
VALUE val = (VALUE)st_val;
|
|
VALUE *args = (VALUE *)arg;
|
|
VALUE execarg_obj = args[0];
|
|
if (rb_execarg_addopt(execarg_obj, key, val) != ST_CONTINUE) {
|
|
VALUE nonopts = args[1];
|
|
if (NIL_P(nonopts)) args[1] = nonopts = rb_hash_new();
|
|
rb_hash_aset(nonopts, key, val);
|
|
}
|
|
return ST_CONTINUE;
|
|
}
|
|
|
|
static int
|
|
check_exec_fds_1(struct rb_execarg *eargp, VALUE h, int maxhint, VALUE ary)
|
|
{
|
|
long i;
|
|
|
|
if (ary != Qfalse) {
|
|
for (i = 0; i < RARRAY_LEN(ary); i++) {
|
|
VALUE elt = RARRAY_AREF(ary, i);
|
|
int fd = FIX2INT(RARRAY_AREF(elt, 0));
|
|
if (RTEST(rb_hash_lookup(h, INT2FIX(fd)))) {
|
|
rb_raise(rb_eArgError, "fd %d specified twice", fd);
|
|
}
|
|
if (ary == eargp->fd_dup2)
|
|
rb_hash_aset(h, INT2FIX(fd), Qtrue);
|
|
else if (ary == eargp->fd_dup2_child)
|
|
rb_hash_aset(h, INT2FIX(fd), RARRAY_AREF(elt, 1));
|
|
else /* ary == eargp->fd_close */
|
|
rb_hash_aset(h, INT2FIX(fd), INT2FIX(-1));
|
|
if (maxhint < fd)
|
|
maxhint = fd;
|
|
if (ary == eargp->fd_dup2 || ary == eargp->fd_dup2_child) {
|
|
fd = FIX2INT(RARRAY_AREF(elt, 1));
|
|
if (maxhint < fd)
|
|
maxhint = fd;
|
|
}
|
|
}
|
|
}
|
|
return maxhint;
|
|
}
|
|
|
|
static VALUE
|
|
check_exec_fds(struct rb_execarg *eargp)
|
|
{
|
|
VALUE h = rb_hash_new();
|
|
VALUE ary;
|
|
int maxhint = -1;
|
|
long i;
|
|
|
|
maxhint = check_exec_fds_1(eargp, h, maxhint, eargp->fd_dup2);
|
|
maxhint = check_exec_fds_1(eargp, h, maxhint, eargp->fd_close);
|
|
maxhint = check_exec_fds_1(eargp, h, maxhint, eargp->fd_dup2_child);
|
|
|
|
if (eargp->fd_dup2_child) {
|
|
ary = eargp->fd_dup2_child;
|
|
for (i = 0; i < RARRAY_LEN(ary); i++) {
|
|
VALUE elt = RARRAY_AREF(ary, i);
|
|
int newfd = FIX2INT(RARRAY_AREF(elt, 0));
|
|
int oldfd = FIX2INT(RARRAY_AREF(elt, 1));
|
|
int lastfd = oldfd;
|
|
VALUE val = rb_hash_lookup(h, INT2FIX(lastfd));
|
|
long depth = 0;
|
|
while (FIXNUM_P(val) && 0 <= FIX2INT(val)) {
|
|
lastfd = FIX2INT(val);
|
|
val = rb_hash_lookup(h, val);
|
|
if (RARRAY_LEN(ary) < depth)
|
|
rb_raise(rb_eArgError, "cyclic child fd redirection from %d", oldfd);
|
|
depth++;
|
|
}
|
|
if (val != Qtrue)
|
|
rb_raise(rb_eArgError, "child fd %d is not redirected", oldfd);
|
|
if (oldfd != lastfd) {
|
|
VALUE val2;
|
|
rb_ary_store(elt, 1, INT2FIX(lastfd));
|
|
rb_hash_aset(h, INT2FIX(newfd), INT2FIX(lastfd));
|
|
val = INT2FIX(oldfd);
|
|
while (FIXNUM_P(val2 = rb_hash_lookup(h, val))) {
|
|
rb_hash_aset(h, val, INT2FIX(lastfd));
|
|
val = val2;
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
eargp->close_others_maxhint = maxhint;
|
|
return h;
|
|
}
|
|
|
|
static void
|
|
rb_check_exec_options(VALUE opthash, VALUE execarg_obj)
|
|
{
|
|
if (RHASH_EMPTY_P(opthash))
|
|
return;
|
|
rb_hash_stlike_foreach(opthash, check_exec_options_i, (st_data_t)execarg_obj);
|
|
}
|
|
|
|
VALUE
|
|
rb_execarg_extract_options(VALUE execarg_obj, VALUE opthash)
|
|
{
|
|
VALUE args[2];
|
|
if (RHASH_EMPTY_P(opthash))
|
|
return Qnil;
|
|
args[0] = execarg_obj;
|
|
args[1] = Qnil;
|
|
rb_hash_stlike_foreach(opthash, check_exec_options_i_extract, (st_data_t)args);
|
|
return args[1];
|
|
}
|
|
|
|
#ifdef ENV_IGNORECASE
|
|
#define ENVMATCH(s1, s2) (STRCASECMP((s1), (s2)) == 0)
|
|
#else
|
|
#define ENVMATCH(n1, n2) (strcmp((n1), (n2)) == 0)
|
|
#endif
|
|
|
|
static int
|
|
check_exec_env_i(st_data_t st_key, st_data_t st_val, st_data_t arg)
|
|
{
|
|
VALUE key = (VALUE)st_key;
|
|
VALUE val = (VALUE)st_val;
|
|
VALUE env = ((VALUE *)arg)[0];
|
|
VALUE *path = &((VALUE *)arg)[1];
|
|
char *k;
|
|
|
|
k = StringValueCStr(key);
|
|
if (strchr(k, '='))
|
|
rb_raise(rb_eArgError, "environment name contains a equal : %"PRIsVALUE, key);
|
|
|
|
if (!NIL_P(val))
|
|
StringValueCStr(val);
|
|
|
|
key = EXPORT_STR(key);
|
|
if (!NIL_P(val)) val = EXPORT_STR(val);
|
|
|
|
if (ENVMATCH(k, PATH_ENV)) {
|
|
*path = val;
|
|
}
|
|
rb_ary_push(env, hide_obj(rb_assoc_new(key, val)));
|
|
|
|
return ST_CONTINUE;
|
|
}
|
|
|
|
static VALUE
|
|
rb_check_exec_env(VALUE hash, VALUE *path)
|
|
{
|
|
VALUE env[2];
|
|
|
|
env[0] = hide_obj(rb_ary_new());
|
|
env[1] = Qfalse;
|
|
rb_hash_stlike_foreach(hash, check_exec_env_i, (st_data_t)env);
|
|
*path = env[1];
|
|
|
|
return env[0];
|
|
}
|
|
|
|
static VALUE
|
|
rb_check_argv(int argc, VALUE *argv)
|
|
{
|
|
VALUE tmp, prog;
|
|
int i;
|
|
|
|
rb_check_arity(argc, 1, UNLIMITED_ARGUMENTS);
|
|
|
|
prog = 0;
|
|
tmp = rb_check_array_type(argv[0]);
|
|
if (!NIL_P(tmp)) {
|
|
if (RARRAY_LEN(tmp) != 2) {
|
|
rb_raise(rb_eArgError, "wrong first argument");
|
|
}
|
|
prog = RARRAY_AREF(tmp, 0);
|
|
argv[0] = RARRAY_AREF(tmp, 1);
|
|
SafeStringValue(prog);
|
|
StringValueCStr(prog);
|
|
prog = rb_str_new_frozen(prog);
|
|
}
|
|
for (i = 0; i < argc; i++) {
|
|
SafeStringValue(argv[i]);
|
|
argv[i] = rb_str_new_frozen(argv[i]);
|
|
StringValueCStr(argv[i]);
|
|
}
|
|
return prog;
|
|
}
|
|
|
|
static VALUE
|
|
check_hash(VALUE obj)
|
|
{
|
|
if (RB_SPECIAL_CONST_P(obj)) return Qnil;
|
|
switch (RB_BUILTIN_TYPE(obj)) {
|
|
case T_STRING:
|
|
case T_ARRAY:
|
|
return Qnil;
|
|
default:
|
|
break;
|
|
}
|
|
return rb_check_hash_type(obj);
|
|
}
|
|
|
|
static VALUE
|
|
rb_exec_getargs(int *argc_p, VALUE **argv_p, int accept_shell, VALUE *env_ret, VALUE *opthash_ret)
|
|
{
|
|
VALUE hash, prog;
|
|
|
|
if (0 < *argc_p) {
|
|
hash = check_hash((*argv_p)[*argc_p-1]);
|
|
if (!NIL_P(hash)) {
|
|
*opthash_ret = hash;
|
|
(*argc_p)--;
|
|
}
|
|
}
|
|
|
|
if (0 < *argc_p) {
|
|
hash = check_hash((*argv_p)[0]);
|
|
if (!NIL_P(hash)) {
|
|
*env_ret = hash;
|
|
(*argc_p)--;
|
|
(*argv_p)++;
|
|
}
|
|
}
|
|
prog = rb_check_argv(*argc_p, *argv_p);
|
|
if (!prog) {
|
|
prog = (*argv_p)[0];
|
|
if (accept_shell && *argc_p == 1) {
|
|
*argc_p = 0;
|
|
*argv_p = 0;
|
|
}
|
|
}
|
|
return prog;
|
|
}
|
|
|
|
#ifndef _WIN32
|
|
struct string_part {
|
|
const char *ptr;
|
|
size_t len;
|
|
};
|
|
|
|
static int
|
|
compare_posix_sh(const void *key, const void *el)
|
|
{
|
|
const struct string_part *word = key;
|
|
int ret = strncmp(word->ptr, el, word->len);
|
|
if (!ret && ((const char *)el)[word->len]) ret = -1;
|
|
return ret;
|
|
}
|
|
#endif
|
|
|
|
static void
|
|
rb_exec_fillarg(VALUE prog, int argc, VALUE *argv, VALUE env, VALUE opthash, VALUE execarg_obj)
|
|
{
|
|
struct rb_execarg *eargp = rb_execarg_get(execarg_obj);
|
|
char fbuf[MAXPATHLEN];
|
|
|
|
MEMZERO(eargp, struct rb_execarg, 1);
|
|
|
|
if (!NIL_P(opthash)) {
|
|
rb_check_exec_options(opthash, execarg_obj);
|
|
}
|
|
if (!NIL_P(env)) {
|
|
env = rb_check_exec_env(env, &eargp->path_env);
|
|
eargp->env_modification = env;
|
|
}
|
|
|
|
prog = EXPORT_STR(prog);
|
|
eargp->use_shell = argc == 0;
|
|
if (eargp->use_shell)
|
|
eargp->invoke.sh.shell_script = prog;
|
|
else
|
|
eargp->invoke.cmd.command_name = prog;
|
|
|
|
#ifndef _WIN32
|
|
if (eargp->use_shell) {
|
|
static const char posix_sh_cmds[][9] = {
|
|
"!", /* reserved */
|
|
".", /* special built-in */
|
|
":", /* special built-in */
|
|
"break", /* special built-in */
|
|
"case", /* reserved */
|
|
"continue", /* special built-in */
|
|
"do", /* reserved */
|
|
"done", /* reserved */
|
|
"elif", /* reserved */
|
|
"else", /* reserved */
|
|
"esac", /* reserved */
|
|
"eval", /* special built-in */
|
|
"exec", /* special built-in */
|
|
"exit", /* special built-in */
|
|
"export", /* special built-in */
|
|
"fi", /* reserved */
|
|
"for", /* reserved */
|
|
"if", /* reserved */
|
|
"in", /* reserved */
|
|
"readonly", /* special built-in */
|
|
"return", /* special built-in */
|
|
"set", /* special built-in */
|
|
"shift", /* special built-in */
|
|
"then", /* reserved */
|
|
"times", /* special built-in */
|
|
"trap", /* special built-in */
|
|
"unset", /* special built-in */
|
|
"until", /* reserved */
|
|
"while", /* reserved */
|
|
};
|
|
const char *p;
|
|
struct string_part first = {0, 0};
|
|
int has_meta = 0;
|
|
/*
|
|
* meta characters:
|
|
*
|
|
* * Pathname Expansion
|
|
* ? Pathname Expansion
|
|
* {} Grouping Commands
|
|
* [] Pathname Expansion
|
|
* <> Redirection
|
|
* () Grouping Commands
|
|
* ~ Tilde Expansion
|
|
* & AND Lists, Asynchronous Lists
|
|
* | OR Lists, Pipelines
|
|
* \ Escape Character
|
|
* $ Parameter Expansion
|
|
* ; Sequential Lists
|
|
* ' Single-Quotes
|
|
* ` Command Substitution
|
|
* " Double-Quotes
|
|
* \n Lists
|
|
*
|
|
* # Comment
|
|
* = Assignment preceding command name
|
|
* % (used in Parameter Expansion)
|
|
*/
|
|
for (p = RSTRING_PTR(prog); *p; p++) {
|
|
if (*p == ' ' || *p == '\t') {
|
|
if (first.ptr && !first.len) first.len = p - first.ptr;
|
|
}
|
|
else {
|
|
if (!first.ptr) first.ptr = p;
|
|
}
|
|
if (!has_meta && strchr("*?{}[]<>()~&|\\$;'`\"\n#", *p))
|
|
has_meta = 1;
|
|
if (!first.len) {
|
|
if (*p == '=') {
|
|
has_meta = 1;
|
|
}
|
|
else if (*p == '/') {
|
|
first.len = 0x100; /* longer than any posix_sh_cmds */
|
|
}
|
|
}
|
|
if (has_meta)
|
|
break;
|
|
}
|
|
if (!has_meta && first.ptr) {
|
|
if (!first.len) first.len = p - first.ptr;
|
|
if (first.len > 0 && first.len <= sizeof(posix_sh_cmds[0]) &&
|
|
bsearch(&first, posix_sh_cmds, numberof(posix_sh_cmds), sizeof(posix_sh_cmds[0]), compare_posix_sh))
|
|
has_meta = 1;
|
|
}
|
|
if (!has_meta) {
|
|
/* avoid shell since no shell meta character found. */
|
|
eargp->use_shell = 0;
|
|
}
|
|
if (!eargp->use_shell) {
|
|
VALUE argv_buf;
|
|
argv_buf = hide_obj(rb_str_buf_new(0));
|
|
p = RSTRING_PTR(prog);
|
|
while (*p) {
|
|
while (*p == ' ' || *p == '\t')
|
|
p++;
|
|
if (*p) {
|
|
const char *w = p;
|
|
while (*p && *p != ' ' && *p != '\t')
|
|
p++;
|
|
rb_str_buf_cat(argv_buf, w, p-w);
|
|
rb_str_buf_cat(argv_buf, "", 1); /* append '\0' */
|
|
}
|
|
}
|
|
eargp->invoke.cmd.argv_buf = argv_buf;
|
|
eargp->invoke.cmd.command_name =
|
|
hide_obj(rb_str_subseq(argv_buf, 0, strlen(RSTRING_PTR(argv_buf))));
|
|
rb_enc_copy(eargp->invoke.cmd.command_name, prog);
|
|
}
|
|
}
|
|
#endif
|
|
|
|
if (!eargp->use_shell) {
|
|
const char *abspath;
|
|
const char *path_env = 0;
|
|
if (RTEST(eargp->path_env)) path_env = RSTRING_PTR(eargp->path_env);
|
|
abspath = dln_find_exe_r(RSTRING_PTR(eargp->invoke.cmd.command_name),
|
|
path_env, fbuf, sizeof(fbuf));
|
|
if (abspath)
|
|
eargp->invoke.cmd.command_abspath = rb_str_new_cstr(abspath);
|
|
else
|
|
eargp->invoke.cmd.command_abspath = Qnil;
|
|
}
|
|
|
|
if (!eargp->use_shell && !eargp->invoke.cmd.argv_buf) {
|
|
int i;
|
|
VALUE argv_buf;
|
|
argv_buf = rb_str_buf_new(0);
|
|
hide_obj(argv_buf);
|
|
for (i = 0; i < argc; i++) {
|
|
VALUE arg = argv[i];
|
|
const char *s = StringValueCStr(arg);
|
|
#ifdef DEFAULT_PROCESS_ENCODING
|
|
arg = EXPORT_STR(arg);
|
|
s = RSTRING_PTR(arg);
|
|
#endif
|
|
rb_str_buf_cat(argv_buf, s, RSTRING_LEN(arg) + 1); /* include '\0' */
|
|
}
|
|
eargp->invoke.cmd.argv_buf = argv_buf;
|
|
}
|
|
|
|
if (!eargp->use_shell) {
|
|
const char *p, *ep, *null=NULL;
|
|
VALUE argv_str;
|
|
argv_str = hide_obj(rb_str_buf_new(sizeof(char*) * (argc + 2)));
|
|
rb_str_buf_cat(argv_str, (char *)&null, sizeof(null)); /* place holder for /bin/sh of try_with_sh. */
|
|
p = RSTRING_PTR(eargp->invoke.cmd.argv_buf);
|
|
ep = p + RSTRING_LEN(eargp->invoke.cmd.argv_buf);
|
|
while (p < ep) {
|
|
rb_str_buf_cat(argv_str, (char *)&p, sizeof(p));
|
|
p += strlen(p) + 1;
|
|
}
|
|
rb_str_buf_cat(argv_str, (char *)&null, sizeof(null)); /* terminator for execve. */
|
|
eargp->invoke.cmd.argv_str =
|
|
rb_imemo_tmpbuf_auto_free_pointer_new_from_an_RString(argv_str);
|
|
}
|
|
RB_GC_GUARD(execarg_obj);
|
|
}
|
|
|
|
struct rb_execarg *
|
|
rb_execarg_get(VALUE execarg_obj)
|
|
{
|
|
struct rb_execarg *eargp;
|
|
TypedData_Get_Struct(execarg_obj, struct rb_execarg, &exec_arg_data_type, eargp);
|
|
return eargp;
|
|
}
|
|
|
|
static VALUE
|
|
rb_execarg_init(int argc, const VALUE *orig_argv, int accept_shell, VALUE execarg_obj)
|
|
{
|
|
struct rb_execarg *eargp = rb_execarg_get(execarg_obj);
|
|
VALUE prog, ret;
|
|
VALUE env = Qnil, opthash = Qnil;
|
|
VALUE argv_buf;
|
|
VALUE *argv = ALLOCV_N(VALUE, argv_buf, argc);
|
|
MEMCPY(argv, orig_argv, VALUE, argc);
|
|
prog = rb_exec_getargs(&argc, &argv, accept_shell, &env, &opthash);
|
|
rb_exec_fillarg(prog, argc, argv, env, opthash, execarg_obj);
|
|
ALLOCV_END(argv_buf);
|
|
ret = eargp->use_shell ? eargp->invoke.sh.shell_script : eargp->invoke.cmd.command_name;
|
|
RB_GC_GUARD(execarg_obj);
|
|
return ret;
|
|
}
|
|
|
|
VALUE
|
|
rb_execarg_new(int argc, const VALUE *argv, int accept_shell, int allow_exc_opt)
|
|
{
|
|
VALUE execarg_obj;
|
|
struct rb_execarg *eargp;
|
|
execarg_obj = TypedData_Make_Struct(0, struct rb_execarg, &exec_arg_data_type, eargp);
|
|
rb_execarg_init(argc, argv, accept_shell, execarg_obj);
|
|
if (!allow_exc_opt && eargp->exception_given) {
|
|
rb_raise(rb_eArgError, "exception option is not allowed");
|
|
}
|
|
return execarg_obj;
|
|
}
|
|
|
|
void
|
|
rb_execarg_setenv(VALUE execarg_obj, VALUE env)
|
|
{
|
|
struct rb_execarg *eargp = rb_execarg_get(execarg_obj);
|
|
env = !NIL_P(env) ? rb_check_exec_env(env, &eargp->path_env) : Qfalse;
|
|
eargp->env_modification = env;
|
|
}
|
|
|
|
static int
|
|
fill_envp_buf_i(st_data_t st_key, st_data_t st_val, st_data_t arg)
|
|
{
|
|
VALUE key = (VALUE)st_key;
|
|
VALUE val = (VALUE)st_val;
|
|
VALUE envp_buf = (VALUE)arg;
|
|
|
|
rb_str_buf_cat2(envp_buf, StringValueCStr(key));
|
|
rb_str_buf_cat2(envp_buf, "=");
|
|
rb_str_buf_cat2(envp_buf, StringValueCStr(val));
|
|
rb_str_buf_cat(envp_buf, "", 1); /* append '\0' */
|
|
|
|
return ST_CONTINUE;
|
|
}
|
|
|
|
|
|
static long run_exec_dup2_tmpbuf_size(long n);
|
|
|
|
struct open_struct {
|
|
VALUE fname;
|
|
int oflags;
|
|
mode_t perm;
|
|
int ret;
|
|
int err;
|
|
};
|
|
|
|
static void *
|
|
open_func(void *ptr)
|
|
{
|
|
struct open_struct *data = ptr;
|
|
const char *fname = RSTRING_PTR(data->fname);
|
|
data->ret = parent_redirect_open(fname, data->oflags, data->perm);
|
|
data->err = errno;
|
|
return NULL;
|
|
}
|
|
|
|
static void
|
|
rb_execarg_allocate_dup2_tmpbuf(struct rb_execarg *eargp, long len)
|
|
{
|
|
VALUE tmpbuf = rb_imemo_tmpbuf_auto_free_pointer();
|
|
rb_imemo_tmpbuf_set_ptr(tmpbuf, ruby_xmalloc(run_exec_dup2_tmpbuf_size(len)));
|
|
eargp->dup2_tmpbuf = tmpbuf;
|
|
}
|
|
|
|
static VALUE
|
|
rb_execarg_parent_start1(VALUE execarg_obj)
|
|
{
|
|
struct rb_execarg *eargp = rb_execarg_get(execarg_obj);
|
|
int unsetenv_others;
|
|
VALUE envopts;
|
|
VALUE ary;
|
|
|
|
ary = eargp->fd_open;
|
|
if (ary != Qfalse) {
|
|
long i;
|
|
for (i = 0; i < RARRAY_LEN(ary); i++) {
|
|
VALUE elt = RARRAY_AREF(ary, i);
|
|
int fd = FIX2INT(RARRAY_AREF(elt, 0));
|
|
VALUE param = RARRAY_AREF(elt, 1);
|
|
VALUE vpath = RARRAY_AREF(param, 0);
|
|
int flags = NUM2INT(RARRAY_AREF(param, 1));
|
|
mode_t perm = NUM2MODET(RARRAY_AREF(param, 2));
|
|
VALUE fd2v = RARRAY_AREF(param, 3);
|
|
int fd2;
|
|
if (NIL_P(fd2v)) {
|
|
struct open_struct open_data;
|
|
again:
|
|
open_data.fname = vpath;
|
|
open_data.oflags = flags;
|
|
open_data.perm = perm;
|
|
open_data.ret = -1;
|
|
open_data.err = EINTR;
|
|
rb_thread_call_without_gvl2(open_func, (void *)&open_data, RUBY_UBF_IO, 0);
|
|
if (open_data.ret == -1) {
|
|
if (open_data.err == EINTR) {
|
|
rb_thread_check_ints();
|
|
goto again;
|
|
}
|
|
rb_syserr_fail_str(open_data.err, vpath);
|
|
}
|
|
fd2 = open_data.ret;
|
|
rb_update_max_fd(fd2);
|
|
RARRAY_ASET(param, 3, INT2FIX(fd2));
|
|
rb_thread_check_ints();
|
|
}
|
|
else {
|
|
fd2 = NUM2INT(fd2v);
|
|
}
|
|
rb_execarg_addopt(execarg_obj, INT2FIX(fd), INT2FIX(fd2));
|
|
}
|
|
}
|
|
|
|
eargp->redirect_fds = check_exec_fds(eargp);
|
|
|
|
ary = eargp->fd_dup2;
|
|
if (ary != Qfalse) {
|
|
rb_execarg_allocate_dup2_tmpbuf(eargp, RARRAY_LEN(ary));
|
|
}
|
|
|
|
unsetenv_others = eargp->unsetenv_others_given && eargp->unsetenv_others_do;
|
|
envopts = eargp->env_modification;
|
|
if (ALWAYS_NEED_ENVP || unsetenv_others || envopts != Qfalse) {
|
|
VALUE envtbl, envp_str, envp_buf;
|
|
char *p, *ep;
|
|
if (unsetenv_others) {
|
|
envtbl = rb_hash_new();
|
|
}
|
|
else {
|
|
envtbl = rb_env_to_hash();
|
|
}
|
|
hide_obj(envtbl);
|
|
if (envopts != Qfalse) {
|
|
st_table *stenv = RHASH_TBL_RAW(envtbl);
|
|
long i;
|
|
for (i = 0; i < RARRAY_LEN(envopts); i++) {
|
|
VALUE pair = RARRAY_AREF(envopts, i);
|
|
VALUE key = RARRAY_AREF(pair, 0);
|
|
VALUE val = RARRAY_AREF(pair, 1);
|
|
if (NIL_P(val)) {
|
|
st_data_t stkey = (st_data_t)key;
|
|
st_delete(stenv, &stkey, NULL);
|
|
}
|
|
else {
|
|
st_insert(stenv, (st_data_t)key, (st_data_t)val);
|
|
RB_OBJ_WRITTEN(envtbl, Qundef, key);
|
|
RB_OBJ_WRITTEN(envtbl, Qundef, val);
|
|
}
|
|
}
|
|
}
|
|
envp_buf = rb_str_buf_new(0);
|
|
hide_obj(envp_buf);
|
|
rb_hash_stlike_foreach(envtbl, fill_envp_buf_i, (st_data_t)envp_buf);
|
|
envp_str = rb_str_buf_new(sizeof(char*) * (RHASH_SIZE(envtbl) + 1));
|
|
hide_obj(envp_str);
|
|
p = RSTRING_PTR(envp_buf);
|
|
ep = p + RSTRING_LEN(envp_buf);
|
|
while (p < ep) {
|
|
rb_str_buf_cat(envp_str, (char *)&p, sizeof(p));
|
|
p += strlen(p) + 1;
|
|
}
|
|
p = NULL;
|
|
rb_str_buf_cat(envp_str, (char *)&p, sizeof(p));
|
|
eargp->envp_str =
|
|
rb_imemo_tmpbuf_auto_free_pointer_new_from_an_RString(envp_str);
|
|
eargp->envp_buf = envp_buf;
|
|
|
|
/*
|
|
char **tmp_envp = (char **)RSTRING_PTR(envp_str);
|
|
while (*tmp_envp) {
|
|
printf("%s\n", *tmp_envp);
|
|
tmp_envp++;
|
|
}
|
|
*/
|
|
}
|
|
|
|
RB_GC_GUARD(execarg_obj);
|
|
return Qnil;
|
|
}
|
|
|
|
void
|
|
rb_execarg_parent_start(VALUE execarg_obj)
|
|
{
|
|
int state;
|
|
rb_protect(rb_execarg_parent_start1, execarg_obj, &state);
|
|
if (state) {
|
|
rb_execarg_parent_end(execarg_obj);
|
|
rb_jump_tag(state);
|
|
}
|
|
}
|
|
|
|
static VALUE
|
|
execarg_parent_end(VALUE execarg_obj)
|
|
{
|
|
struct rb_execarg *eargp = rb_execarg_get(execarg_obj);
|
|
int err = errno;
|
|
VALUE ary;
|
|
|
|
ary = eargp->fd_open;
|
|
if (ary != Qfalse) {
|
|
long i;
|
|
for (i = 0; i < RARRAY_LEN(ary); i++) {
|
|
VALUE elt = RARRAY_AREF(ary, i);
|
|
VALUE param = RARRAY_AREF(elt, 1);
|
|
VALUE fd2v;
|
|
int fd2;
|
|
fd2v = RARRAY_AREF(param, 3);
|
|
if (!NIL_P(fd2v)) {
|
|
fd2 = FIX2INT(fd2v);
|
|
parent_redirect_close(fd2);
|
|
RARRAY_ASET(param, 3, Qnil);
|
|
}
|
|
}
|
|
}
|
|
|
|
errno = err;
|
|
return execarg_obj;
|
|
}
|
|
|
|
void
|
|
rb_execarg_parent_end(VALUE execarg_obj)
|
|
{
|
|
execarg_parent_end(execarg_obj);
|
|
RB_GC_GUARD(execarg_obj);
|
|
}
|
|
|
|
static void
|
|
rb_exec_fail(struct rb_execarg *eargp, int err, const char *errmsg)
|
|
{
|
|
if (!errmsg || !*errmsg) return;
|
|
if (strcmp(errmsg, "chdir") == 0) {
|
|
rb_sys_fail_str(eargp->chdir_dir);
|
|
}
|
|
rb_sys_fail(errmsg);
|
|
}
|
|
|
|
#if 0
|
|
void
|
|
rb_execarg_fail(VALUE execarg_obj, int err, const char *errmsg)
|
|
{
|
|
if (!errmsg || !*errmsg) return;
|
|
rb_exec_fail(rb_execarg_get(execarg_obj), err, errmsg);
|
|
RB_GC_GUARD(execarg_obj);
|
|
}
|
|
#endif
|
|
|
|
VALUE
|
|
rb_f_exec(int argc, const VALUE *argv)
|
|
{
|
|
VALUE execarg_obj, fail_str;
|
|
struct rb_execarg *eargp;
|
|
#define CHILD_ERRMSG_BUFLEN 80
|
|
char errmsg[CHILD_ERRMSG_BUFLEN] = { '\0' };
|
|
int err, state;
|
|
|
|
execarg_obj = rb_execarg_new(argc, argv, TRUE, FALSE);
|
|
eargp = rb_execarg_get(execarg_obj);
|
|
before_exec(); /* stop timer thread before redirects */
|
|
|
|
rb_protect(rb_execarg_parent_start1, execarg_obj, &state);
|
|
if (state) {
|
|
execarg_parent_end(execarg_obj);
|
|
after_exec(); /* restart timer thread */
|
|
rb_jump_tag(state);
|
|
}
|
|
|
|
fail_str = eargp->use_shell ? eargp->invoke.sh.shell_script : eargp->invoke.cmd.command_name;
|
|
|
|
err = exec_async_signal_safe(eargp, errmsg, sizeof(errmsg));
|
|
after_exec(); /* restart timer thread */
|
|
|
|
rb_exec_fail(eargp, err, errmsg);
|
|
RB_GC_GUARD(execarg_obj);
|
|
rb_syserr_fail_str(err, fail_str);
|
|
UNREACHABLE_RETURN(Qnil);
|
|
}
|
|
|
|
NORETURN(static VALUE f_exec(int c, const VALUE *a, VALUE _));
|
|
|
|
/*
|
|
* call-seq:
|
|
* exec([env, ] command_line, options = {})
|
|
* exec([env, ] exe_path, *args, options = {})
|
|
*
|
|
* Replaces the current process by doing one of the following:
|
|
*
|
|
* - Passing string +command_line+ to the shell.
|
|
* - Invoking the executable at +exe_path+.
|
|
*
|
|
* This method has potential security vulnerabilities if called with untrusted input;
|
|
* see {Command Injection}[rdoc-ref:command_injection.rdoc].
|
|
*
|
|
* The new process is created using the
|
|
* {exec system call}[https://pubs.opengroup.org/onlinepubs/9699919799.2018edition/functions/execve.html];
|
|
* it may inherit some of its environment from the calling program
|
|
* (possibly including open file descriptors).
|
|
*
|
|
* Argument +env+, if given, is a hash that affects +ENV+ for the new process;
|
|
* see {Execution Environment}[rdoc-ref:Process@Execution+Environment].
|
|
*
|
|
* Argument +options+ is a hash of options for the new process;
|
|
* see {Execution Options}[rdoc-ref:Process@Execution+Options].
|
|
*
|
|
* The first required argument is one of the following:
|
|
*
|
|
* - +command_line+ if it is a string,
|
|
* and if it begins with a shell reserved word or special built-in,
|
|
* or if it contains one or more meta characters.
|
|
* - +exe_path+ otherwise.
|
|
*
|
|
* <b>Argument +command_line+</b>
|
|
*
|
|
* \String argument +command_line+ is a command line to be passed to a shell;
|
|
* it must begin with a shell reserved word, begin with a special built-in,
|
|
* or contain meta characters:
|
|
*
|
|
* exec('if true; then echo "Foo"; fi') # Shell reserved word.
|
|
* exec('echo') # Built-in.
|
|
* exec('date > date.tmp') # Contains meta character.
|
|
*
|
|
* The command line may also contain arguments and options for the command:
|
|
*
|
|
* exec('echo "Foo"')
|
|
*
|
|
* Output:
|
|
*
|
|
* Foo
|
|
*
|
|
* See {Execution Shell}[rdoc-ref:Process@Execution+Shell] for details about the shell.
|
|
*
|
|
* Raises an exception if the new process could not execute.
|
|
*
|
|
* <b>Argument +exe_path+</b>
|
|
*
|
|
* Argument +exe_path+ is one of the following:
|
|
*
|
|
* - The string path to an executable to be called.
|
|
* - A 2-element array containing the path to an executable
|
|
* and the string to be used as the name of the executing process.
|
|
*
|
|
* Example:
|
|
*
|
|
* exec('/usr/bin/date')
|
|
*
|
|
* Output:
|
|
*
|
|
* Sat Aug 26 09:38:00 AM CDT 2023
|
|
*
|
|
* Ruby invokes the executable directly, with no shell and no shell expansion:
|
|
*
|
|
* exec('doesnt_exist') # Raises Errno::ENOENT
|
|
*
|
|
* If one or more +args+ is given, each is an argument or option
|
|
* to be passed to the executable:
|
|
*
|
|
* exec('echo', 'C*')
|
|
* exec('echo', 'hello', 'world')
|
|
*
|
|
* Output:
|
|
*
|
|
* C*
|
|
* hello world
|
|
*
|
|
* Raises an exception if the new process could not execute.
|
|
*/
|
|
|
|
static VALUE
|
|
f_exec(int c, const VALUE *a, VALUE _)
|
|
{
|
|
rb_f_exec(c, a);
|
|
UNREACHABLE_RETURN(Qnil);
|
|
}
|
|
|
|
#define ERRMSG(str) do { if (errmsg && 0 < errmsg_buflen) strlcpy(errmsg, (str), errmsg_buflen); } while (0)
|
|
#define ERRMSG1(str, a) do { if (errmsg && 0 < errmsg_buflen) snprintf(errmsg, errmsg_buflen, (str), (a)); } while (0)
|
|
#define ERRMSG2(str, a, b) do { if (errmsg && 0 < errmsg_buflen) snprintf(errmsg, errmsg_buflen, (str), (a), (b)); } while (0)
|
|
|
|
static int fd_get_cloexec(int fd, char *errmsg, size_t errmsg_buflen);
|
|
static int fd_set_cloexec(int fd, char *errmsg, size_t errmsg_buflen);
|
|
static int fd_clear_cloexec(int fd, char *errmsg, size_t errmsg_buflen);
|
|
|
|
static int
|
|
save_redirect_fd(int fd, struct rb_execarg *sargp, char *errmsg, size_t errmsg_buflen)
|
|
{
|
|
if (sargp) {
|
|
VALUE newary, redirection;
|
|
int save_fd = redirect_cloexec_dup(fd), cloexec;
|
|
if (save_fd == -1) {
|
|
if (errno == EBADF)
|
|
return 0;
|
|
ERRMSG("dup");
|
|
return -1;
|
|
}
|
|
rb_update_max_fd(save_fd);
|
|
newary = sargp->fd_dup2;
|
|
if (newary == Qfalse) {
|
|
newary = hide_obj(rb_ary_new());
|
|
sargp->fd_dup2 = newary;
|
|
}
|
|
cloexec = fd_get_cloexec(fd, errmsg, errmsg_buflen);
|
|
redirection = hide_obj(rb_assoc_new(INT2FIX(fd), INT2FIX(save_fd)));
|
|
if (cloexec) rb_ary_push(redirection, Qtrue);
|
|
rb_ary_push(newary, redirection);
|
|
|
|
newary = sargp->fd_close;
|
|
if (newary == Qfalse) {
|
|
newary = hide_obj(rb_ary_new());
|
|
sargp->fd_close = newary;
|
|
}
|
|
rb_ary_push(newary, hide_obj(rb_assoc_new(INT2FIX(save_fd), Qnil)));
|
|
}
|
|
|
|
return 0;
|
|
}
|
|
|
|
static int
|
|
intcmp(const void *a, const void *b)
|
|
{
|
|
return *(int*)a - *(int*)b;
|
|
}
|
|
|
|
static int
|
|
intrcmp(const void *a, const void *b)
|
|
{
|
|
return *(int*)b - *(int*)a;
|
|
}
|
|
|
|
struct run_exec_dup2_fd_pair {
|
|
int oldfd;
|
|
int newfd;
|
|
long older_index;
|
|
long num_newer;
|
|
int cloexec;
|
|
};
|
|
|
|
static long
|
|
run_exec_dup2_tmpbuf_size(long n)
|
|
{
|
|
return sizeof(struct run_exec_dup2_fd_pair) * n;
|
|
}
|
|
|
|
/* This function should be async-signal-safe. Actually it is. */
|
|
static int
|
|
fd_get_cloexec(int fd, char *errmsg, size_t errmsg_buflen)
|
|
{
|
|
#ifdef F_GETFD
|
|
int ret = 0;
|
|
ret = fcntl(fd, F_GETFD); /* async-signal-safe */
|
|
if (ret == -1) {
|
|
ERRMSG("fcntl(F_GETFD)");
|
|
return -1;
|
|
}
|
|
if (ret & FD_CLOEXEC) return 1;
|
|
#endif
|
|
return 0;
|
|
}
|
|
|
|
/* This function should be async-signal-safe. Actually it is. */
|
|
static int
|
|
fd_set_cloexec(int fd, char *errmsg, size_t errmsg_buflen)
|
|
{
|
|
#ifdef F_GETFD
|
|
int ret = 0;
|
|
ret = fcntl(fd, F_GETFD); /* async-signal-safe */
|
|
if (ret == -1) {
|
|
ERRMSG("fcntl(F_GETFD)");
|
|
return -1;
|
|
}
|
|
if (!(ret & FD_CLOEXEC)) {
|
|
ret |= FD_CLOEXEC;
|
|
ret = fcntl(fd, F_SETFD, ret); /* async-signal-safe */
|
|
if (ret == -1) {
|
|
ERRMSG("fcntl(F_SETFD)");
|
|
return -1;
|
|
}
|
|
}
|
|
#endif
|
|
return 0;
|
|
}
|
|
|
|
/* This function should be async-signal-safe. Actually it is. */
|
|
static int
|
|
fd_clear_cloexec(int fd, char *errmsg, size_t errmsg_buflen)
|
|
{
|
|
#ifdef F_GETFD
|
|
int ret;
|
|
ret = fcntl(fd, F_GETFD); /* async-signal-safe */
|
|
if (ret == -1) {
|
|
ERRMSG("fcntl(F_GETFD)");
|
|
return -1;
|
|
}
|
|
if (ret & FD_CLOEXEC) {
|
|
ret &= ~FD_CLOEXEC;
|
|
ret = fcntl(fd, F_SETFD, ret); /* async-signal-safe */
|
|
if (ret == -1) {
|
|
ERRMSG("fcntl(F_SETFD)");
|
|
return -1;
|
|
}
|
|
}
|
|
#endif
|
|
return 0;
|
|
}
|
|
|
|
/* This function should be async-signal-safe when sargp is NULL. Hopefully it is. */
|
|
static int
|
|
run_exec_dup2(VALUE ary, VALUE tmpbuf, struct rb_execarg *sargp, char *errmsg, size_t errmsg_buflen)
|
|
{
|
|
long n, i;
|
|
int ret;
|
|
int extra_fd = -1;
|
|
struct rb_imemo_tmpbuf_struct *buf = (void *)tmpbuf;
|
|
struct run_exec_dup2_fd_pair *pairs = (void *)buf->ptr;
|
|
|
|
n = RARRAY_LEN(ary);
|
|
|
|
/* initialize oldfd and newfd: O(n) */
|
|
for (i = 0; i < n; i++) {
|
|
VALUE elt = RARRAY_AREF(ary, i);
|
|
pairs[i].oldfd = FIX2INT(RARRAY_AREF(elt, 1));
|
|
pairs[i].newfd = FIX2INT(RARRAY_AREF(elt, 0)); /* unique */
|
|
pairs[i].cloexec = RARRAY_LEN(elt) > 2 && RTEST(RARRAY_AREF(elt, 2));
|
|
pairs[i].older_index = -1;
|
|
}
|
|
|
|
/* sort the table by oldfd: O(n log n) */
|
|
if (!sargp)
|
|
qsort(pairs, n, sizeof(struct run_exec_dup2_fd_pair), intcmp); /* hopefully async-signal-safe */
|
|
else
|
|
qsort(pairs, n, sizeof(struct run_exec_dup2_fd_pair), intrcmp);
|
|
|
|
/* initialize older_index and num_newer: O(n log n) */
|
|
for (i = 0; i < n; i++) {
|
|
int newfd = pairs[i].newfd;
|
|
struct run_exec_dup2_fd_pair key, *found;
|
|
key.oldfd = newfd;
|
|
found = bsearch(&key, pairs, n, sizeof(struct run_exec_dup2_fd_pair), intcmp); /* hopefully async-signal-safe */
|
|
pairs[i].num_newer = 0;
|
|
if (found) {
|
|
while (pairs < found && (found-1)->oldfd == newfd)
|
|
found--;
|
|
while (found < pairs+n && found->oldfd == newfd) {
|
|
pairs[i].num_newer++;
|
|
found->older_index = i;
|
|
found++;
|
|
}
|
|
}
|
|
}
|
|
|
|
/* non-cyclic redirection: O(n) */
|
|
for (i = 0; i < n; i++) {
|
|
long j = i;
|
|
while (j != -1 && pairs[j].oldfd != -1 && pairs[j].num_newer == 0) {
|
|
if (save_redirect_fd(pairs[j].newfd, sargp, errmsg, errmsg_buflen) < 0) /* async-signal-safe */
|
|
goto fail;
|
|
ret = redirect_dup2(pairs[j].oldfd, pairs[j].newfd); /* async-signal-safe */
|
|
if (ret == -1) {
|
|
ERRMSG("dup2");
|
|
goto fail;
|
|
}
|
|
if (pairs[j].cloexec &&
|
|
fd_set_cloexec(pairs[j].newfd, errmsg, errmsg_buflen)) {
|
|
goto fail;
|
|
}
|
|
rb_update_max_fd(pairs[j].newfd); /* async-signal-safe but don't need to call it in a child process. */
|
|
pairs[j].oldfd = -1;
|
|
j = pairs[j].older_index;
|
|
if (j != -1)
|
|
pairs[j].num_newer--;
|
|
}
|
|
}
|
|
|
|
/* cyclic redirection: O(n) */
|
|
for (i = 0; i < n; i++) {
|
|
long j;
|
|
if (pairs[i].oldfd == -1)
|
|
continue;
|
|
if (pairs[i].oldfd == pairs[i].newfd) { /* self cycle */
|
|
if (fd_clear_cloexec(pairs[i].oldfd, errmsg, errmsg_buflen) == -1) /* async-signal-safe */
|
|
goto fail;
|
|
pairs[i].oldfd = -1;
|
|
continue;
|
|
}
|
|
if (extra_fd == -1) {
|
|
extra_fd = redirect_dup(pairs[i].oldfd); /* async-signal-safe */
|
|
if (extra_fd == -1) {
|
|
ERRMSG("dup");
|
|
goto fail;
|
|
}
|
|
rb_update_max_fd(extra_fd);
|
|
}
|
|
else {
|
|
ret = redirect_dup2(pairs[i].oldfd, extra_fd); /* async-signal-safe */
|
|
if (ret == -1) {
|
|
ERRMSG("dup2");
|
|
goto fail;
|
|
}
|
|
rb_update_max_fd(extra_fd);
|
|
}
|
|
pairs[i].oldfd = extra_fd;
|
|
j = pairs[i].older_index;
|
|
pairs[i].older_index = -1;
|
|
while (j != -1) {
|
|
ret = redirect_dup2(pairs[j].oldfd, pairs[j].newfd); /* async-signal-safe */
|
|
if (ret == -1) {
|
|
ERRMSG("dup2");
|
|
goto fail;
|
|
}
|
|
rb_update_max_fd(ret);
|
|
pairs[j].oldfd = -1;
|
|
j = pairs[j].older_index;
|
|
}
|
|
}
|
|
if (extra_fd != -1) {
|
|
ret = redirect_close(extra_fd); /* async-signal-safe */
|
|
if (ret == -1) {
|
|
ERRMSG("close");
|
|
goto fail;
|
|
}
|
|
}
|
|
|
|
return 0;
|
|
|
|
fail:
|
|
return -1;
|
|
}
|
|
|
|
/* This function should be async-signal-safe. Actually it is. */
|
|
static int
|
|
run_exec_close(VALUE ary, char *errmsg, size_t errmsg_buflen)
|
|
{
|
|
long i;
|
|
int ret;
|
|
|
|
for (i = 0; i < RARRAY_LEN(ary); i++) {
|
|
VALUE elt = RARRAY_AREF(ary, i);
|
|
int fd = FIX2INT(RARRAY_AREF(elt, 0));
|
|
ret = redirect_close(fd); /* async-signal-safe */
|
|
if (ret == -1) {
|
|
ERRMSG("close");
|
|
return -1;
|
|
}
|
|
}
|
|
return 0;
|
|
}
|
|
|
|
/* This function should be async-signal-safe when sargp is NULL. Actually it is. */
|
|
static int
|
|
run_exec_dup2_child(VALUE ary, struct rb_execarg *sargp, char *errmsg, size_t errmsg_buflen)
|
|
{
|
|
long i;
|
|
int ret;
|
|
|
|
for (i = 0; i < RARRAY_LEN(ary); i++) {
|
|
VALUE elt = RARRAY_AREF(ary, i);
|
|
int newfd = FIX2INT(RARRAY_AREF(elt, 0));
|
|
int oldfd = FIX2INT(RARRAY_AREF(elt, 1));
|
|
|
|
if (save_redirect_fd(newfd, sargp, errmsg, errmsg_buflen) < 0) /* async-signal-safe */
|
|
return -1;
|
|
ret = redirect_dup2(oldfd, newfd); /* async-signal-safe */
|
|
if (ret == -1) {
|
|
ERRMSG("dup2");
|
|
return -1;
|
|
}
|
|
rb_update_max_fd(newfd);
|
|
}
|
|
return 0;
|
|
}
|
|
|
|
#ifdef HAVE_SETPGID
|
|
/* This function should be async-signal-safe when sargp is NULL. Actually it is. */
|
|
static int
|
|
run_exec_pgroup(const struct rb_execarg *eargp, struct rb_execarg *sargp, char *errmsg, size_t errmsg_buflen)
|
|
{
|
|
/*
|
|
* If FD_CLOEXEC is available, rb_fork_async_signal_safe waits the child's execve.
|
|
* So setpgid is done in the child when rb_fork_async_signal_safe is returned in
|
|
* the parent.
|
|
* No race condition, even without setpgid from the parent.
|
|
* (Is there an environment which has setpgid but no FD_CLOEXEC?)
|
|
*/
|
|
int ret;
|
|
rb_pid_t pgroup;
|
|
|
|
pgroup = eargp->pgroup_pgid;
|
|
if (pgroup == -1)
|
|
return 0;
|
|
|
|
if (sargp) {
|
|
/* maybe meaningless with no fork environment... */
|
|
sargp->pgroup_given = 1;
|
|
sargp->pgroup_pgid = getpgrp();
|
|
}
|
|
|
|
if (pgroup == 0) {
|
|
pgroup = getpid(); /* async-signal-safe */
|
|
}
|
|
ret = setpgid(getpid(), pgroup); /* async-signal-safe */
|
|
if (ret == -1) ERRMSG("setpgid");
|
|
return ret;
|
|
}
|
|
#endif
|
|
|
|
#if defined(HAVE_SETRLIMIT) && defined(RLIM2NUM)
|
|
/* This function should be async-signal-safe when sargp is NULL. Hopefully it is. */
|
|
static int
|
|
run_exec_rlimit(VALUE ary, struct rb_execarg *sargp, char *errmsg, size_t errmsg_buflen)
|
|
{
|
|
long i;
|
|
for (i = 0; i < RARRAY_LEN(ary); i++) {
|
|
VALUE elt = RARRAY_AREF(ary, i);
|
|
int rtype = NUM2INT(RARRAY_AREF(elt, 0));
|
|
struct rlimit rlim;
|
|
if (sargp) {
|
|
VALUE tmp, newary;
|
|
if (getrlimit(rtype, &rlim) == -1) {
|
|
ERRMSG("getrlimit");
|
|
return -1;
|
|
}
|
|
tmp = hide_obj(rb_ary_new3(3, RARRAY_AREF(elt, 0),
|
|
RLIM2NUM(rlim.rlim_cur),
|
|
RLIM2NUM(rlim.rlim_max)));
|
|
if (sargp->rlimit_limits == Qfalse)
|
|
newary = sargp->rlimit_limits = hide_obj(rb_ary_new());
|
|
else
|
|
newary = sargp->rlimit_limits;
|
|
rb_ary_push(newary, tmp);
|
|
}
|
|
rlim.rlim_cur = NUM2RLIM(RARRAY_AREF(elt, 1));
|
|
rlim.rlim_max = NUM2RLIM(RARRAY_AREF(elt, 2));
|
|
if (setrlimit(rtype, &rlim) == -1) { /* hopefully async-signal-safe */
|
|
ERRMSG("setrlimit");
|
|
return -1;
|
|
}
|
|
}
|
|
return 0;
|
|
}
|
|
#endif
|
|
|
|
#if !defined(HAVE_WORKING_FORK)
|
|
static VALUE
|
|
save_env_i(RB_BLOCK_CALL_FUNC_ARGLIST(i, ary))
|
|
{
|
|
rb_ary_push(ary, hide_obj(rb_ary_dup(argv[0])));
|
|
return Qnil;
|
|
}
|
|
|
|
static void
|
|
save_env(struct rb_execarg *sargp)
|
|
{
|
|
if (!sargp)
|
|
return;
|
|
if (sargp->env_modification == Qfalse) {
|
|
VALUE env = rb_envtbl();
|
|
if (RTEST(env)) {
|
|
VALUE ary = hide_obj(rb_ary_new());
|
|
rb_block_call(env, idEach, 0, 0, save_env_i,
|
|
(VALUE)ary);
|
|
sargp->env_modification = ary;
|
|
}
|
|
sargp->unsetenv_others_given = 1;
|
|
sargp->unsetenv_others_do = 1;
|
|
}
|
|
}
|
|
#endif
|
|
|
|
#ifdef _WIN32
|
|
#undef chdir
|
|
#define chdir(p) rb_w32_uchdir(p)
|
|
#endif
|
|
|
|
/* This function should be async-signal-safe when sargp is NULL. Hopefully it is. */
|
|
int
|
|
rb_execarg_run_options(const struct rb_execarg *eargp, struct rb_execarg *sargp, char *errmsg, size_t errmsg_buflen)
|
|
{
|
|
VALUE obj;
|
|
|
|
if (sargp) {
|
|
/* assume that sargp is always NULL on fork-able environments */
|
|
MEMZERO(sargp, struct rb_execarg, 1);
|
|
sargp->redirect_fds = Qnil;
|
|
}
|
|
|
|
#ifdef HAVE_SETPGID
|
|
if (eargp->pgroup_given) {
|
|
if (run_exec_pgroup(eargp, sargp, errmsg, errmsg_buflen) == -1) /* async-signal-safe */
|
|
return -1;
|
|
}
|
|
#endif
|
|
|
|
#if defined(HAVE_SETRLIMIT) && defined(RLIM2NUM)
|
|
obj = eargp->rlimit_limits;
|
|
if (obj != Qfalse) {
|
|
if (run_exec_rlimit(obj, sargp, errmsg, errmsg_buflen) == -1) /* hopefully async-signal-safe */
|
|
return -1;
|
|
}
|
|
#endif
|
|
|
|
#if !defined(HAVE_WORKING_FORK)
|
|
if (eargp->unsetenv_others_given && eargp->unsetenv_others_do) {
|
|
save_env(sargp);
|
|
rb_env_clear();
|
|
}
|
|
|
|
obj = eargp->env_modification;
|
|
if (obj != Qfalse) {
|
|
long i;
|
|
save_env(sargp);
|
|
for (i = 0; i < RARRAY_LEN(obj); i++) {
|
|
VALUE pair = RARRAY_AREF(obj, i);
|
|
VALUE key = RARRAY_AREF(pair, 0);
|
|
VALUE val = RARRAY_AREF(pair, 1);
|
|
if (NIL_P(val))
|
|
ruby_setenv(StringValueCStr(key), 0);
|
|
else
|
|
ruby_setenv(StringValueCStr(key), StringValueCStr(val));
|
|
}
|
|
}
|
|
#endif
|
|
|
|
if (eargp->umask_given) {
|
|
mode_t mask = eargp->umask_mask;
|
|
mode_t oldmask = umask(mask); /* never fail */ /* async-signal-safe */
|
|
if (sargp) {
|
|
sargp->umask_given = 1;
|
|
sargp->umask_mask = oldmask;
|
|
}
|
|
}
|
|
|
|
obj = eargp->fd_dup2;
|
|
if (obj != Qfalse) {
|
|
if (run_exec_dup2(obj, eargp->dup2_tmpbuf, sargp, errmsg, errmsg_buflen) == -1) /* hopefully async-signal-safe */
|
|
return -1;
|
|
}
|
|
|
|
obj = eargp->fd_close;
|
|
if (obj != Qfalse) {
|
|
if (sargp)
|
|
rb_warn("cannot close fd before spawn");
|
|
else {
|
|
if (run_exec_close(obj, errmsg, errmsg_buflen) == -1) /* async-signal-safe */
|
|
return -1;
|
|
}
|
|
}
|
|
|
|
#ifdef HAVE_WORKING_FORK
|
|
if (eargp->close_others_do) {
|
|
rb_close_before_exec(3, eargp->close_others_maxhint, eargp->redirect_fds); /* async-signal-safe */
|
|
}
|
|
#endif
|
|
|
|
obj = eargp->fd_dup2_child;
|
|
if (obj != Qfalse) {
|
|
if (run_exec_dup2_child(obj, sargp, errmsg, errmsg_buflen) == -1) /* async-signal-safe */
|
|
return -1;
|
|
}
|
|
|
|
if (eargp->chdir_given) {
|
|
if (sargp) {
|
|
sargp->chdir_given = 1;
|
|
sargp->chdir_dir = hide_obj(rb_dir_getwd_ospath());
|
|
}
|
|
if (chdir(RSTRING_PTR(eargp->chdir_dir)) == -1) { /* async-signal-safe */
|
|
ERRMSG("chdir");
|
|
return -1;
|
|
}
|
|
}
|
|
|
|
#ifdef HAVE_SETGID
|
|
if (eargp->gid_given) {
|
|
if (setgid(eargp->gid) < 0) {
|
|
ERRMSG("setgid");
|
|
return -1;
|
|
}
|
|
}
|
|
#endif
|
|
#ifdef HAVE_SETUID
|
|
if (eargp->uid_given) {
|
|
if (setuid(eargp->uid) < 0) {
|
|
ERRMSG("setuid");
|
|
return -1;
|
|
}
|
|
}
|
|
#endif
|
|
|
|
if (sargp) {
|
|
VALUE ary = sargp->fd_dup2;
|
|
if (ary != Qfalse) {
|
|
rb_execarg_allocate_dup2_tmpbuf(sargp, RARRAY_LEN(ary));
|
|
}
|
|
}
|
|
{
|
|
int preserve = errno;
|
|
stdfd_clear_nonblock();
|
|
errno = preserve;
|
|
}
|
|
|
|
return 0;
|
|
}
|
|
|
|
/* This function should be async-signal-safe. Hopefully it is. */
|
|
int
|
|
rb_exec_async_signal_safe(const struct rb_execarg *eargp, char *errmsg, size_t errmsg_buflen)
|
|
{
|
|
errno = exec_async_signal_safe(eargp, errmsg, errmsg_buflen);
|
|
return -1;
|
|
}
|
|
|
|
static int
|
|
exec_async_signal_safe(const struct rb_execarg *eargp, char *errmsg, size_t errmsg_buflen)
|
|
{
|
|
#if !defined(HAVE_WORKING_FORK)
|
|
struct rb_execarg sarg, *const sargp = &sarg;
|
|
#else
|
|
struct rb_execarg *const sargp = NULL;
|
|
#endif
|
|
int err;
|
|
|
|
if (rb_execarg_run_options(eargp, sargp, errmsg, errmsg_buflen) < 0) { /* hopefully async-signal-safe */
|
|
return errno;
|
|
}
|
|
|
|
if (eargp->use_shell) {
|
|
err = proc_exec_sh(RSTRING_PTR(eargp->invoke.sh.shell_script), eargp->envp_str); /* async-signal-safe */
|
|
}
|
|
else {
|
|
char *abspath = NULL;
|
|
if (!NIL_P(eargp->invoke.cmd.command_abspath))
|
|
abspath = RSTRING_PTR(eargp->invoke.cmd.command_abspath);
|
|
err = proc_exec_cmd(abspath, eargp->invoke.cmd.argv_str, eargp->envp_str); /* async-signal-safe */
|
|
}
|
|
#if !defined(HAVE_WORKING_FORK)
|
|
rb_execarg_run_options(sargp, NULL, errmsg, errmsg_buflen);
|
|
#endif
|
|
|
|
return err;
|
|
}
|
|
|
|
#ifdef HAVE_WORKING_FORK
|
|
/* This function should be async-signal-safe. Hopefully it is. */
|
|
static int
|
|
rb_exec_atfork(void* arg, char *errmsg, size_t errmsg_buflen)
|
|
{
|
|
return rb_exec_async_signal_safe(arg, errmsg, errmsg_buflen); /* hopefully async-signal-safe */
|
|
}
|
|
|
|
static VALUE
|
|
proc_syswait(VALUE pid)
|
|
{
|
|
rb_syswait((rb_pid_t)pid);
|
|
return Qnil;
|
|
}
|
|
|
|
static int
|
|
move_fds_to_avoid_crash(int *fdp, int n, VALUE fds)
|
|
{
|
|
int min = 0;
|
|
int i;
|
|
for (i = 0; i < n; i++) {
|
|
int ret;
|
|
while (RTEST(rb_hash_lookup(fds, INT2FIX(fdp[i])))) {
|
|
if (min <= fdp[i])
|
|
min = fdp[i]+1;
|
|
while (RTEST(rb_hash_lookup(fds, INT2FIX(min))))
|
|
min++;
|
|
ret = rb_cloexec_fcntl_dupfd(fdp[i], min);
|
|
if (ret == -1)
|
|
return -1;
|
|
rb_update_max_fd(ret);
|
|
close(fdp[i]);
|
|
fdp[i] = ret;
|
|
}
|
|
}
|
|
return 0;
|
|
}
|
|
|
|
static int
|
|
pipe_nocrash(int filedes[2], VALUE fds)
|
|
{
|
|
int ret;
|
|
ret = rb_pipe(filedes);
|
|
if (ret == -1)
|
|
return -1;
|
|
if (RTEST(fds)) {
|
|
int save = errno;
|
|
if (move_fds_to_avoid_crash(filedes, 2, fds) == -1) {
|
|
close(filedes[0]);
|
|
close(filedes[1]);
|
|
return -1;
|
|
}
|
|
errno = save;
|
|
}
|
|
return ret;
|
|
}
|
|
|
|
#ifndef O_BINARY
|
|
#define O_BINARY 0
|
|
#endif
|
|
|
|
static VALUE
|
|
rb_thread_sleep_that_takes_VALUE_as_sole_argument(VALUE n)
|
|
{
|
|
rb_thread_sleep(NUM2INT(n));
|
|
return Qundef;
|
|
}
|
|
|
|
static int
|
|
handle_fork_error(int err, struct rb_process_status *status, int *ep, volatile int *try_gc_p)
|
|
{
|
|
int state = 0;
|
|
|
|
switch (err) {
|
|
case ENOMEM:
|
|
if ((*try_gc_p)-- > 0 && !rb_during_gc()) {
|
|
rb_gc();
|
|
return 0;
|
|
}
|
|
break;
|
|
case EAGAIN:
|
|
#if defined(EWOULDBLOCK) && EWOULDBLOCK != EAGAIN
|
|
case EWOULDBLOCK:
|
|
#endif
|
|
if (!status && !ep) {
|
|
rb_thread_sleep(1);
|
|
return 0;
|
|
}
|
|
else {
|
|
rb_protect(rb_thread_sleep_that_takes_VALUE_as_sole_argument, INT2FIX(1), &state);
|
|
if (status) status->status = state;
|
|
if (!state) return 0;
|
|
}
|
|
break;
|
|
}
|
|
if (ep) {
|
|
close(ep[0]);
|
|
close(ep[1]);
|
|
errno = err;
|
|
}
|
|
if (state && !status) rb_jump_tag(state);
|
|
return -1;
|
|
}
|
|
|
|
#define prefork() ( \
|
|
rb_io_flush(rb_stdout), \
|
|
rb_io_flush(rb_stderr) \
|
|
)
|
|
|
|
/*
|
|
* Forks child process, and returns the process ID in the parent
|
|
* process.
|
|
*
|
|
* If +status+ is given, protects from any exceptions and sets the
|
|
* jump status to it, and returns -1. If failed to fork new process
|
|
* but no exceptions occurred, sets 0 to it. Otherwise, if forked
|
|
* successfully, the value of +status+ is undetermined.
|
|
*
|
|
* In the child process, just returns 0 if +chfunc+ is +NULL+.
|
|
* Otherwise +chfunc+ will be called with +charg+, and then the child
|
|
* process exits with +EXIT_SUCCESS+ when it returned zero.
|
|
*
|
|
* In the case of the function is called and returns non-zero value,
|
|
* the child process exits with non-+EXIT_SUCCESS+ value (normally
|
|
* 127). And, on the platforms where +FD_CLOEXEC+ is available,
|
|
* +errno+ is propagated to the parent process, and this function
|
|
* returns -1 in the parent process. On the other platforms, just
|
|
* returns pid.
|
|
*
|
|
* If fds is not Qnil, internal pipe for the errno propagation is
|
|
* arranged to avoid conflicts of the hash keys in +fds+.
|
|
*
|
|
* +chfunc+ must not raise any exceptions.
|
|
*/
|
|
|
|
static ssize_t
|
|
write_retry(int fd, const void *buf, size_t len)
|
|
{
|
|
ssize_t w;
|
|
|
|
do {
|
|
w = write(fd, buf, len);
|
|
} while (w < 0 && errno == EINTR);
|
|
|
|
return w;
|
|
}
|
|
|
|
static ssize_t
|
|
read_retry(int fd, void *buf, size_t len)
|
|
{
|
|
ssize_t r;
|
|
|
|
if (set_blocking(fd) != 0) {
|
|
#ifndef _WIN32
|
|
rb_async_bug_errno("set_blocking failed reading child error", errno);
|
|
#endif
|
|
}
|
|
|
|
do {
|
|
r = read(fd, buf, len);
|
|
} while (r < 0 && errno == EINTR);
|
|
|
|
return r;
|
|
}
|
|
|
|
static void
|
|
send_child_error(int fd, char *errmsg, size_t errmsg_buflen)
|
|
{
|
|
int err;
|
|
|
|
err = errno;
|
|
if (write_retry(fd, &err, sizeof(err)) < 0) err = errno;
|
|
if (errmsg && 0 < errmsg_buflen) {
|
|
errmsg[errmsg_buflen-1] = '\0';
|
|
errmsg_buflen = strlen(errmsg);
|
|
if (errmsg_buflen > 0 && write_retry(fd, errmsg, errmsg_buflen) < 0)
|
|
err = errno;
|
|
}
|
|
}
|
|
|
|
static int
|
|
recv_child_error(int fd, int *errp, char *errmsg, size_t errmsg_buflen)
|
|
{
|
|
int err;
|
|
ssize_t size;
|
|
if ((size = read_retry(fd, &err, sizeof(err))) < 0) {
|
|
err = errno;
|
|
}
|
|
*errp = err;
|
|
if (size == sizeof(err) &&
|
|
errmsg && 0 < errmsg_buflen) {
|
|
ssize_t ret = read_retry(fd, errmsg, errmsg_buflen-1);
|
|
if (0 <= ret) {
|
|
errmsg[ret] = '\0';
|
|
}
|
|
}
|
|
close(fd);
|
|
return size != 0;
|
|
}
|
|
|
|
#ifdef HAVE_WORKING_VFORK
|
|
#if !defined(HAVE_GETRESUID) && defined(HAVE_GETUIDX)
|
|
/* AIX 7.1 */
|
|
static int
|
|
getresuid(rb_uid_t *ruid, rb_uid_t *euid, rb_uid_t *suid)
|
|
{
|
|
rb_uid_t ret;
|
|
|
|
*ruid = getuid();
|
|
*euid = geteuid();
|
|
ret = getuidx(ID_SAVED);
|
|
if (ret == (rb_uid_t)-1)
|
|
return -1;
|
|
*suid = ret;
|
|
return 0;
|
|
}
|
|
#define HAVE_GETRESUID
|
|
#endif
|
|
|
|
#if !defined(HAVE_GETRESGID) && defined(HAVE_GETGIDX)
|
|
/* AIX 7.1 */
|
|
static int
|
|
getresgid(rb_gid_t *rgid, rb_gid_t *egid, rb_gid_t *sgid)
|
|
{
|
|
rb_gid_t ret;
|
|
|
|
*rgid = getgid();
|
|
*egid = getegid();
|
|
ret = getgidx(ID_SAVED);
|
|
if (ret == (rb_gid_t)-1)
|
|
return -1;
|
|
*sgid = ret;
|
|
return 0;
|
|
}
|
|
#define HAVE_GETRESGID
|
|
#endif
|
|
|
|
static int
|
|
has_privilege(void)
|
|
{
|
|
/*
|
|
* has_privilege() is used to choose vfork() or fork().
|
|
*
|
|
* If the process has privilege, the parent process or
|
|
* the child process can change UID/GID.
|
|
* If vfork() is used to create the child process and
|
|
* the parent or child process change effective UID/GID,
|
|
* different privileged processes shares memory.
|
|
* It is a bad situation.
|
|
* So, fork() should be used.
|
|
*/
|
|
|
|
rb_uid_t ruid, euid;
|
|
rb_gid_t rgid, egid;
|
|
|
|
#if defined HAVE_ISSETUGID
|
|
if (issetugid())
|
|
return 1;
|
|
#endif
|
|
|
|
#ifdef HAVE_GETRESUID
|
|
{
|
|
int ret;
|
|
rb_uid_t suid;
|
|
ret = getresuid(&ruid, &euid, &suid);
|
|
if (ret == -1)
|
|
rb_sys_fail("getresuid(2)");
|
|
if (euid != suid)
|
|
return 1;
|
|
}
|
|
#else
|
|
ruid = getuid();
|
|
euid = geteuid();
|
|
#endif
|
|
|
|
if (euid == 0 || euid != ruid)
|
|
return 1;
|
|
|
|
#ifdef HAVE_GETRESGID
|
|
{
|
|
int ret;
|
|
rb_gid_t sgid;
|
|
ret = getresgid(&rgid, &egid, &sgid);
|
|
if (ret == -1)
|
|
rb_sys_fail("getresgid(2)");
|
|
if (egid != sgid)
|
|
return 1;
|
|
}
|
|
#else
|
|
rgid = getgid();
|
|
egid = getegid();
|
|
#endif
|
|
|
|
if (egid != rgid)
|
|
return 1;
|
|
|
|
return 0;
|
|
}
|
|
#endif
|
|
|
|
struct child_handler_disabler_state
|
|
{
|
|
sigset_t sigmask;
|
|
};
|
|
|
|
static void
|
|
disable_child_handler_before_fork(struct child_handler_disabler_state *old)
|
|
{
|
|
#ifdef HAVE_PTHREAD_SIGMASK
|
|
int ret;
|
|
sigset_t all;
|
|
|
|
ret = sigfillset(&all);
|
|
if (ret == -1)
|
|
rb_sys_fail("sigfillset");
|
|
|
|
ret = pthread_sigmask(SIG_SETMASK, &all, &old->sigmask); /* not async-signal-safe */
|
|
if (ret != 0) {
|
|
rb_syserr_fail(ret, "pthread_sigmask");
|
|
}
|
|
#else
|
|
# pragma GCC warning "pthread_sigmask on fork is not available. potentially dangerous"
|
|
#endif
|
|
}
|
|
|
|
static void
|
|
disable_child_handler_fork_parent(struct child_handler_disabler_state *old)
|
|
{
|
|
#ifdef HAVE_PTHREAD_SIGMASK
|
|
int ret;
|
|
|
|
ret = pthread_sigmask(SIG_SETMASK, &old->sigmask, NULL); /* not async-signal-safe */
|
|
if (ret != 0) {
|
|
rb_syserr_fail(ret, "pthread_sigmask");
|
|
}
|
|
#else
|
|
# pragma GCC warning "pthread_sigmask on fork is not available. potentially dangerous"
|
|
#endif
|
|
}
|
|
|
|
/* This function should be async-signal-safe. Actually it is. */
|
|
static int
|
|
disable_child_handler_fork_child(struct child_handler_disabler_state *old, char *errmsg, size_t errmsg_buflen)
|
|
{
|
|
int sig;
|
|
int ret;
|
|
|
|
for (sig = 1; sig < NSIG; sig++) {
|
|
sig_t handler = signal(sig, SIG_DFL);
|
|
|
|
if (handler == SIG_ERR && errno == EINVAL) {
|
|
continue; /* Ignore invalid signal number */
|
|
}
|
|
if (handler == SIG_ERR) {
|
|
ERRMSG("signal to obtain old action");
|
|
return -1;
|
|
}
|
|
#ifdef SIGPIPE
|
|
if (sig == SIGPIPE) {
|
|
continue;
|
|
}
|
|
#endif
|
|
/* it will be reset to SIG_DFL at execve time, instead */
|
|
if (handler == SIG_IGN) {
|
|
signal(sig, SIG_IGN);
|
|
}
|
|
}
|
|
|
|
/* non-Ruby child process, ensure cmake can see SIGCHLD */
|
|
sigemptyset(&old->sigmask);
|
|
ret = sigprocmask(SIG_SETMASK, &old->sigmask, NULL); /* async-signal-safe */
|
|
if (ret != 0) {
|
|
ERRMSG("sigprocmask");
|
|
return -1;
|
|
}
|
|
return 0;
|
|
}
|
|
|
|
static rb_pid_t
|
|
retry_fork_async_signal_safe(struct rb_process_status *status, int *ep,
|
|
int (*chfunc)(void*, char *, size_t), void *charg,
|
|
char *errmsg, size_t errmsg_buflen,
|
|
struct waitpid_state *w)
|
|
{
|
|
rb_pid_t pid;
|
|
volatile int try_gc = 1;
|
|
struct child_handler_disabler_state old;
|
|
int err;
|
|
|
|
while (1) {
|
|
prefork();
|
|
disable_child_handler_before_fork(&old);
|
|
#ifdef HAVE_WORKING_VFORK
|
|
if (!has_privilege())
|
|
pid = vfork();
|
|
else
|
|
pid = rb_fork();
|
|
#else
|
|
pid = rb_fork();
|
|
#endif
|
|
if (pid == 0) {/* fork succeed, child process */
|
|
int ret;
|
|
close(ep[0]);
|
|
ret = disable_child_handler_fork_child(&old, errmsg, errmsg_buflen); /* async-signal-safe */
|
|
if (ret == 0) {
|
|
ret = chfunc(charg, errmsg, errmsg_buflen);
|
|
if (!ret) _exit(EXIT_SUCCESS);
|
|
}
|
|
send_child_error(ep[1], errmsg, errmsg_buflen);
|
|
#if EXIT_SUCCESS == 127
|
|
_exit(EXIT_FAILURE);
|
|
#else
|
|
_exit(127);
|
|
#endif
|
|
}
|
|
err = errno;
|
|
disable_child_handler_fork_parent(&old);
|
|
if (0 < pid) /* fork succeed, parent process */
|
|
return pid;
|
|
/* fork failed */
|
|
if (handle_fork_error(err, status, ep, &try_gc))
|
|
return -1;
|
|
}
|
|
}
|
|
|
|
static rb_pid_t
|
|
fork_check_err(struct rb_process_status *status, int (*chfunc)(void*, char *, size_t), void *charg,
|
|
VALUE fds, char *errmsg, size_t errmsg_buflen,
|
|
struct rb_execarg *eargp)
|
|
{
|
|
rb_pid_t pid;
|
|
int err;
|
|
int ep[2];
|
|
int error_occurred;
|
|
|
|
struct waitpid_state *w = eargp && eargp->waitpid_state ? eargp->waitpid_state : 0;
|
|
|
|
if (status) status->status = 0;
|
|
|
|
if (pipe_nocrash(ep, fds)) return -1;
|
|
|
|
pid = retry_fork_async_signal_safe(status, ep, chfunc, charg, errmsg, errmsg_buflen, w);
|
|
|
|
if (status) status->pid = pid;
|
|
|
|
if (pid < 0) {
|
|
if (status) status->error = errno;
|
|
|
|
return pid;
|
|
}
|
|
|
|
close(ep[1]);
|
|
|
|
error_occurred = recv_child_error(ep[0], &err, errmsg, errmsg_buflen);
|
|
|
|
if (error_occurred) {
|
|
if (status) {
|
|
int state = 0;
|
|
status->error = err;
|
|
|
|
VM_ASSERT((w == 0) && "only used by extensions");
|
|
rb_protect(proc_syswait, (VALUE)pid, &state);
|
|
|
|
status->status = state;
|
|
}
|
|
else if (!w) {
|
|
rb_syswait(pid);
|
|
}
|
|
|
|
errno = err;
|
|
return -1;
|
|
}
|
|
|
|
return pid;
|
|
}
|
|
|
|
/*
|
|
* The "async_signal_safe" name is a lie, but it is used by pty.c and
|
|
* maybe other exts. fork() is not async-signal-safe due to pthread_atfork
|
|
* and future POSIX revisions will remove it from a list of signal-safe
|
|
* functions. rb_waitpid is not async-signal-safe since RJIT, either.
|
|
* For our purposes, we do not need async-signal-safety, here
|
|
*/
|
|
rb_pid_t
|
|
rb_fork_async_signal_safe(int *status,
|
|
int (*chfunc)(void*, char *, size_t), void *charg,
|
|
VALUE fds, char *errmsg, size_t errmsg_buflen)
|
|
{
|
|
struct rb_process_status process_status;
|
|
|
|
rb_pid_t result = fork_check_err(&process_status, chfunc, charg, fds, errmsg, errmsg_buflen, 0);
|
|
|
|
if (status) {
|
|
*status = process_status.status;
|
|
}
|
|
|
|
return result;
|
|
}
|
|
|
|
static rb_pid_t
|
|
rb_fork_ruby2(struct rb_process_status *status)
|
|
{
|
|
rb_pid_t pid;
|
|
int try_gc = 1, err;
|
|
struct child_handler_disabler_state old;
|
|
|
|
if (status) status->status = 0;
|
|
|
|
while (1) {
|
|
prefork();
|
|
disable_child_handler_before_fork(&old);
|
|
before_fork_ruby();
|
|
pid = rb_fork();
|
|
err = errno;
|
|
if (status) {
|
|
status->pid = pid;
|
|
status->error = err;
|
|
}
|
|
after_fork_ruby(pid);
|
|
disable_child_handler_fork_parent(&old); /* yes, bad name */
|
|
|
|
if (pid >= 0) { /* fork succeed */
|
|
return pid;
|
|
}
|
|
|
|
/* fork failed */
|
|
if (handle_fork_error(err, status, NULL, &try_gc)) {
|
|
return -1;
|
|
}
|
|
}
|
|
}
|
|
|
|
rb_pid_t
|
|
rb_fork_ruby(int *status)
|
|
{
|
|
struct rb_process_status process_status = {0};
|
|
|
|
rb_pid_t pid = rb_fork_ruby2(&process_status);
|
|
|
|
if (status) *status = process_status.status;
|
|
|
|
return pid;
|
|
}
|
|
|
|
static rb_pid_t
|
|
proc_fork_pid(void)
|
|
{
|
|
rb_pid_t pid = rb_fork_ruby(NULL);
|
|
|
|
if (pid == -1) {
|
|
rb_sys_fail("fork(2)");
|
|
}
|
|
|
|
return pid;
|
|
}
|
|
|
|
rb_pid_t
|
|
rb_call_proc__fork(void)
|
|
{
|
|
ID id__fork;
|
|
CONST_ID(id__fork, "_fork");
|
|
if (rb_method_basic_definition_p(CLASS_OF(rb_mProcess), id__fork)) {
|
|
return proc_fork_pid();
|
|
}
|
|
else {
|
|
VALUE pid = rb_funcall(rb_mProcess, id__fork, 0);
|
|
return NUM2PIDT(pid);
|
|
}
|
|
}
|
|
#endif
|
|
|
|
#if defined(HAVE_WORKING_FORK) && !defined(CANNOT_FORK_WITH_PTHREAD)
|
|
/*
|
|
* call-seq:
|
|
* Process._fork -> integer
|
|
*
|
|
* An internal API for fork. Do not call this method directly.
|
|
* Currently, this is called via Kernel#fork, Process.fork, and
|
|
* IO.popen with <tt>"-"</tt>.
|
|
*
|
|
* This method is not for casual code but for application monitoring
|
|
* libraries. You can add custom code before and after fork events
|
|
* by overriding this method.
|
|
*
|
|
* Note: Process.daemon may be implemented using fork(2) BUT does not go
|
|
* through this method.
|
|
* Thus, depending on your reason to hook into this method, you
|
|
* may also want to hook into that one.
|
|
* See {this issue}[https://bugs.ruby-lang.org/issues/18911] for a
|
|
* more detailed discussion of this.
|
|
*/
|
|
VALUE
|
|
rb_proc__fork(VALUE _obj)
|
|
{
|
|
rb_pid_t pid = proc_fork_pid();
|
|
return PIDT2NUM(pid);
|
|
}
|
|
|
|
/*
|
|
* call-seq:
|
|
* Process.fork { ... } -> integer or nil
|
|
* Process.fork -> integer or nil
|
|
*
|
|
* Creates a child process.
|
|
*
|
|
* With a block given, runs the block in the child process;
|
|
* on block exit, the child terminates with a status of zero:
|
|
*
|
|
* puts "Before the fork: #{Process.pid}"
|
|
* fork do
|
|
* puts "In the child process: #{Process.pid}"
|
|
* end # => 382141
|
|
* puts "After the fork: #{Process.pid}"
|
|
*
|
|
* Output:
|
|
*
|
|
* Before the fork: 420496
|
|
* After the fork: 420496
|
|
* In the child process: 420520
|
|
*
|
|
* With no block given, the +fork+ call returns twice:
|
|
*
|
|
* - Once in the parent process, returning the pid of the child process.
|
|
* - Once in the child process, returning +nil+.
|
|
*
|
|
* Example:
|
|
*
|
|
* puts "This is the first line before the fork (pid #{Process.pid})"
|
|
* puts fork
|
|
* puts "This is the second line after the fork (pid #{Process.pid})"
|
|
*
|
|
* Output:
|
|
*
|
|
* This is the first line before the fork (pid 420199)
|
|
* 420223
|
|
* This is the second line after the fork (pid 420199)
|
|
*
|
|
* This is the second line after the fork (pid 420223)
|
|
*
|
|
* In either case, the child process may exit using
|
|
* Kernel.exit! to avoid the call to Kernel#at_exit.
|
|
*
|
|
* To avoid zombie processes, the parent process should call either:
|
|
*
|
|
* - Process.wait, to collect the termination statuses of its children.
|
|
* - Process.detach, to register disinterest in their status.
|
|
*
|
|
* The thread calling +fork+ is the only thread in the created child process;
|
|
* +fork+ doesn't copy other threads.
|
|
*
|
|
* Note that method +fork+ is available on some platforms,
|
|
* but not on others:
|
|
*
|
|
* Process.respond_to?(:fork) # => true # Would be false on some.
|
|
*
|
|
* If not, you may use ::spawn instead of +fork+.
|
|
*/
|
|
|
|
static VALUE
|
|
rb_f_fork(VALUE obj)
|
|
{
|
|
rb_pid_t pid;
|
|
|
|
pid = rb_call_proc__fork();
|
|
|
|
if (pid == 0) {
|
|
if (rb_block_given_p()) {
|
|
int status;
|
|
rb_protect(rb_yield, Qundef, &status);
|
|
ruby_stop(status);
|
|
}
|
|
return Qnil;
|
|
}
|
|
|
|
return PIDT2NUM(pid);
|
|
}
|
|
#else
|
|
#define rb_proc__fork rb_f_notimplement
|
|
#define rb_f_fork rb_f_notimplement
|
|
#endif
|
|
|
|
static int
|
|
exit_status_code(VALUE status)
|
|
{
|
|
int istatus;
|
|
|
|
switch (status) {
|
|
case Qtrue:
|
|
istatus = EXIT_SUCCESS;
|
|
break;
|
|
case Qfalse:
|
|
istatus = EXIT_FAILURE;
|
|
break;
|
|
default:
|
|
istatus = NUM2INT(status);
|
|
#if EXIT_SUCCESS != 0
|
|
if (istatus == 0)
|
|
istatus = EXIT_SUCCESS;
|
|
#endif
|
|
break;
|
|
}
|
|
return istatus;
|
|
}
|
|
|
|
NORETURN(static VALUE rb_f_exit_bang(int argc, VALUE *argv, VALUE obj));
|
|
/*
|
|
* call-seq:
|
|
* exit!(status = false)
|
|
* Process.exit!(status = false)
|
|
*
|
|
* Exits the process immediately; no exit handlers are called.
|
|
* Returns exit status +status+ to the underlying operating system.
|
|
*
|
|
* Process.exit!(true)
|
|
*
|
|
* Values +true+ and +false+ for argument +status+
|
|
* indicate, respectively, success and failure;
|
|
* The meanings of integer values are system-dependent.
|
|
*
|
|
*/
|
|
|
|
static VALUE
|
|
rb_f_exit_bang(int argc, VALUE *argv, VALUE obj)
|
|
{
|
|
int istatus;
|
|
|
|
if (rb_check_arity(argc, 0, 1) == 1) {
|
|
istatus = exit_status_code(argv[0]);
|
|
}
|
|
else {
|
|
istatus = EXIT_FAILURE;
|
|
}
|
|
_exit(istatus);
|
|
|
|
UNREACHABLE_RETURN(Qnil);
|
|
}
|
|
|
|
void
|
|
rb_exit(int status)
|
|
{
|
|
if (GET_EC()->tag) {
|
|
VALUE args[2];
|
|
|
|
args[0] = INT2NUM(status);
|
|
args[1] = rb_str_new2("exit");
|
|
rb_exc_raise(rb_class_new_instance(2, args, rb_eSystemExit));
|
|
}
|
|
ruby_stop(status);
|
|
}
|
|
|
|
VALUE
|
|
rb_f_exit(int argc, const VALUE *argv)
|
|
{
|
|
int istatus;
|
|
|
|
if (rb_check_arity(argc, 0, 1) == 1) {
|
|
istatus = exit_status_code(argv[0]);
|
|
}
|
|
else {
|
|
istatus = EXIT_SUCCESS;
|
|
}
|
|
rb_exit(istatus);
|
|
|
|
UNREACHABLE_RETURN(Qnil);
|
|
}
|
|
|
|
NORETURN(static VALUE f_exit(int c, const VALUE *a, VALUE _));
|
|
/*
|
|
* call-seq:
|
|
* exit(status = true)
|
|
* Process.exit(status = true)
|
|
*
|
|
* Initiates termination of the Ruby script by raising SystemExit;
|
|
* the exception may be caught.
|
|
* Returns exit status +status+ to the underlying operating system.
|
|
*
|
|
* Values +true+ and +false+ for argument +status+
|
|
* indicate, respectively, success and failure;
|
|
* The meanings of integer values are system-dependent.
|
|
*
|
|
* Example:
|
|
*
|
|
* begin
|
|
* exit
|
|
* puts 'Never get here.'
|
|
* rescue SystemExit
|
|
* puts 'Rescued a SystemExit exception.'
|
|
* end
|
|
* puts 'After begin block.'
|
|
*
|
|
* Output:
|
|
*
|
|
* Rescued a SystemExit exception.
|
|
* After begin block.
|
|
*
|
|
* Just prior to final termination,
|
|
* Ruby executes any at-exit procedures (see Kernel::at_exit)
|
|
* and any object finalizers (see ObjectSpace::define_finalizer).
|
|
*
|
|
* Example:
|
|
*
|
|
* at_exit { puts 'In at_exit function.' }
|
|
* ObjectSpace.define_finalizer('string', proc { puts 'In finalizer.' })
|
|
* exit
|
|
*
|
|
* Output:
|
|
*
|
|
* In at_exit function.
|
|
* In finalizer.
|
|
*
|
|
*/
|
|
|
|
static VALUE
|
|
f_exit(int c, const VALUE *a, VALUE _)
|
|
{
|
|
rb_f_exit(c, a);
|
|
UNREACHABLE_RETURN(Qnil);
|
|
}
|
|
|
|
VALUE
|
|
rb_f_abort(int argc, const VALUE *argv)
|
|
{
|
|
rb_check_arity(argc, 0, 1);
|
|
if (argc == 0) {
|
|
rb_execution_context_t *ec = GET_EC();
|
|
VALUE errinfo = rb_ec_get_errinfo(ec);
|
|
if (!NIL_P(errinfo)) {
|
|
rb_ec_error_print(ec, errinfo);
|
|
}
|
|
rb_exit(EXIT_FAILURE);
|
|
}
|
|
else {
|
|
VALUE args[2];
|
|
|
|
args[1] = args[0] = argv[0];
|
|
StringValue(args[0]);
|
|
rb_io_puts(1, args, rb_ractor_stderr());
|
|
args[0] = INT2NUM(EXIT_FAILURE);
|
|
rb_exc_raise(rb_class_new_instance(2, args, rb_eSystemExit));
|
|
}
|
|
|
|
UNREACHABLE_RETURN(Qnil);
|
|
}
|
|
|
|
NORETURN(static VALUE f_abort(int c, const VALUE *a, VALUE _));
|
|
|
|
/*
|
|
* call-seq:
|
|
* abort
|
|
* Process.abort(msg = nil)
|
|
*
|
|
* Terminates execution immediately, effectively by calling
|
|
* <tt>Kernel.exit(false)</tt>.
|
|
*
|
|
* If string argument +msg+ is given,
|
|
* it is written to STDERR prior to termination;
|
|
* otherwise, if an exception was raised,
|
|
* prints its message and backtrace.
|
|
*/
|
|
|
|
static VALUE
|
|
f_abort(int c, const VALUE *a, VALUE _)
|
|
{
|
|
rb_f_abort(c, a);
|
|
UNREACHABLE_RETURN(Qnil);
|
|
}
|
|
|
|
void
|
|
rb_syswait(rb_pid_t pid)
|
|
{
|
|
int status;
|
|
|
|
rb_waitpid(pid, &status, 0);
|
|
}
|
|
|
|
#if !defined HAVE_WORKING_FORK && !defined HAVE_SPAWNV && !defined __EMSCRIPTEN__
|
|
char *
|
|
rb_execarg_commandline(const struct rb_execarg *eargp, VALUE *prog)
|
|
{
|
|
VALUE cmd = *prog;
|
|
if (eargp && !eargp->use_shell) {
|
|
VALUE str = eargp->invoke.cmd.argv_str;
|
|
VALUE buf = eargp->invoke.cmd.argv_buf;
|
|
char *p, **argv = ARGVSTR2ARGV(str);
|
|
long i, argc = ARGVSTR2ARGC(str);
|
|
const char *start = RSTRING_PTR(buf);
|
|
cmd = rb_str_new(start, RSTRING_LEN(buf));
|
|
p = RSTRING_PTR(cmd);
|
|
for (i = 1; i < argc; ++i) {
|
|
p[argv[i] - start - 1] = ' ';
|
|
}
|
|
*prog = cmd;
|
|
return p;
|
|
}
|
|
return StringValueCStr(*prog);
|
|
}
|
|
#endif
|
|
|
|
static rb_pid_t
|
|
rb_spawn_process(struct rb_execarg *eargp, char *errmsg, size_t errmsg_buflen)
|
|
{
|
|
rb_pid_t pid;
|
|
#if !defined HAVE_WORKING_FORK || USE_SPAWNV
|
|
VALUE prog;
|
|
struct rb_execarg sarg;
|
|
# if !defined HAVE_SPAWNV
|
|
int status;
|
|
# endif
|
|
#endif
|
|
|
|
#if defined HAVE_WORKING_FORK && !USE_SPAWNV
|
|
pid = fork_check_err(eargp->status, rb_exec_atfork, eargp, eargp->redirect_fds, errmsg, errmsg_buflen, eargp);
|
|
#else
|
|
prog = eargp->use_shell ? eargp->invoke.sh.shell_script : eargp->invoke.cmd.command_name;
|
|
|
|
if (rb_execarg_run_options(eargp, &sarg, errmsg, errmsg_buflen) < 0) {
|
|
return -1;
|
|
}
|
|
|
|
if (prog && !eargp->use_shell) {
|
|
char **argv = ARGVSTR2ARGV(eargp->invoke.cmd.argv_str);
|
|
argv[0] = RSTRING_PTR(prog);
|
|
}
|
|
# if defined HAVE_SPAWNV
|
|
if (eargp->use_shell) {
|
|
pid = proc_spawn_sh(RSTRING_PTR(prog));
|
|
}
|
|
else {
|
|
char **argv = ARGVSTR2ARGV(eargp->invoke.cmd.argv_str);
|
|
pid = proc_spawn_cmd(argv, prog, eargp);
|
|
}
|
|
|
|
if (pid == -1) {
|
|
rb_last_status_set(0x7f << 8, pid);
|
|
}
|
|
# else
|
|
status = system(rb_execarg_commandline(eargp, &prog));
|
|
pid = 1; /* dummy */
|
|
rb_last_status_set((status & 0xff) << 8, pid);
|
|
# endif
|
|
|
|
if (eargp->waitpid_state) {
|
|
eargp->waitpid_state->pid = pid;
|
|
}
|
|
|
|
rb_execarg_run_options(&sarg, NULL, errmsg, errmsg_buflen);
|
|
#endif
|
|
|
|
return pid;
|
|
}
|
|
|
|
struct spawn_args {
|
|
VALUE execarg;
|
|
struct {
|
|
char *ptr;
|
|
size_t buflen;
|
|
} errmsg;
|
|
};
|
|
|
|
static VALUE
|
|
do_spawn_process(VALUE arg)
|
|
{
|
|
struct spawn_args *argp = (struct spawn_args *)arg;
|
|
rb_execarg_parent_start1(argp->execarg);
|
|
return (VALUE)rb_spawn_process(DATA_PTR(argp->execarg),
|
|
argp->errmsg.ptr, argp->errmsg.buflen);
|
|
}
|
|
|
|
static rb_pid_t
|
|
rb_execarg_spawn(VALUE execarg_obj, char *errmsg, size_t errmsg_buflen)
|
|
{
|
|
struct spawn_args args;
|
|
|
|
args.execarg = execarg_obj;
|
|
args.errmsg.ptr = errmsg;
|
|
args.errmsg.buflen = errmsg_buflen;
|
|
return (rb_pid_t)rb_ensure(do_spawn_process, (VALUE)&args,
|
|
execarg_parent_end, execarg_obj);
|
|
}
|
|
|
|
static rb_pid_t
|
|
rb_spawn_internal(int argc, const VALUE *argv, char *errmsg, size_t errmsg_buflen)
|
|
{
|
|
VALUE execarg_obj;
|
|
|
|
execarg_obj = rb_execarg_new(argc, argv, TRUE, FALSE);
|
|
return rb_execarg_spawn(execarg_obj, errmsg, errmsg_buflen);
|
|
}
|
|
|
|
rb_pid_t
|
|
rb_spawn_err(int argc, const VALUE *argv, char *errmsg, size_t errmsg_buflen)
|
|
{
|
|
return rb_spawn_internal(argc, argv, errmsg, errmsg_buflen);
|
|
}
|
|
|
|
rb_pid_t
|
|
rb_spawn(int argc, const VALUE *argv)
|
|
{
|
|
return rb_spawn_internal(argc, argv, NULL, 0);
|
|
}
|
|
|
|
/*
|
|
* call-seq:
|
|
* system([env, ] command_line, options = {}, exception: false) -> true, false, or nil
|
|
* system([env, ] exe_path, *args, options = {}, exception: false) -> true, false, or nil
|
|
*
|
|
* Creates a new child process by doing one of the following
|
|
* in that process:
|
|
*
|
|
* - Passing string +command_line+ to the shell.
|
|
* - Invoking the executable at +exe_path+.
|
|
*
|
|
* This method has potential security vulnerabilities if called with untrusted input;
|
|
* see {Command Injection}[rdoc-ref:command_injection.rdoc].
|
|
*
|
|
* Returns:
|
|
*
|
|
* - +true+ if the command exits with status zero.
|
|
* - +false+ if the exit status is a non-zero integer.
|
|
* - +nil+ if the command could not execute.
|
|
*
|
|
* Raises an exception (instead of returning +false+ or +nil+)
|
|
* if keyword argument +exception+ is set to +true+.
|
|
*
|
|
* Assigns the command's error status to <tt>$?</tt>.
|
|
*
|
|
* The new process is created using the
|
|
* {system system call}[https://pubs.opengroup.org/onlinepubs/9699919799.2018edition/functions/system.html];
|
|
* it may inherit some of its environment from the calling program
|
|
* (possibly including open file descriptors).
|
|
*
|
|
* Argument +env+, if given, is a hash that affects +ENV+ for the new process;
|
|
* see {Execution Environment}[rdoc-ref:Process@Execution+Environment].
|
|
*
|
|
* Argument +options+ is a hash of options for the new process;
|
|
* see {Execution Options}[rdoc-ref:Process@Execution+Options].
|
|
*
|
|
* The first required argument is one of the following:
|
|
*
|
|
* - +command_line+ if it is a string,
|
|
* and if it begins with a shell reserved word or special built-in,
|
|
* or if it contains one or more meta characters.
|
|
* - +exe_path+ otherwise.
|
|
*
|
|
* <b>Argument +command_line+</b>
|
|
*
|
|
* \String argument +command_line+ is a command line to be passed to a shell;
|
|
* it must begin with a shell reserved word, begin with a special built-in,
|
|
* or contain meta characters:
|
|
*
|
|
* system('if true; then echo "Foo"; fi') # => true # Shell reserved word.
|
|
* system('echo') # => true # Built-in.
|
|
* system('date > /tmp/date.tmp') # => true # Contains meta character.
|
|
* system('date > /nop/date.tmp') # => false
|
|
* system('date > /nop/date.tmp', exception: true) # Raises RuntimeError.
|
|
*
|
|
* Assigns the command's error status to <tt>$?</tt>:
|
|
*
|
|
* system('echo') # => true # Built-in.
|
|
* $? # => #<Process::Status: pid 640610 exit 0>
|
|
* system('date > /nop/date.tmp') # => false
|
|
* $? # => #<Process::Status: pid 640742 exit 2>
|
|
*
|
|
* The command line may also contain arguments and options for the command:
|
|
*
|
|
* system('echo "Foo"') # => true
|
|
*
|
|
* Output:
|
|
*
|
|
* Foo
|
|
*
|
|
* See {Execution Shell}[rdoc-ref:Process@Execution+Shell] for details about the shell.
|
|
*
|
|
* Raises an exception if the new process could not execute.
|
|
*
|
|
* <b>Argument +exe_path+</b>
|
|
*
|
|
* Argument +exe_path+ is one of the following:
|
|
*
|
|
* - The string path to an executable to be called.
|
|
* - A 2-element array containing the path to an executable
|
|
* and the string to be used as the name of the executing process.
|
|
*
|
|
* Example:
|
|
*
|
|
* system('/usr/bin/date') # => true # Path to date on Unix-style system.
|
|
* system('foo') # => nil # Command failed.
|
|
*
|
|
* Output:
|
|
*
|
|
* Mon Aug 28 11:43:10 AM CDT 2023
|
|
*
|
|
* Assigns the command's error status to <tt>$?</tt>:
|
|
*
|
|
* system('/usr/bin/date') # => true
|
|
* $? # => #<Process::Status: pid 645605 exit 0>
|
|
* system('foo') # => nil
|
|
* $? # => #<Process::Status: pid 645608 exit 127>
|
|
*
|
|
* Ruby invokes the executable directly, with no shell and no shell expansion:
|
|
*
|
|
* system('doesnt_exist') # => nil
|
|
*
|
|
* If one or more +args+ is given, each is an argument or option
|
|
* to be passed to the executable:
|
|
*
|
|
* system('echo', 'C*') # => true
|
|
* system('echo', 'hello', 'world') # => true
|
|
*
|
|
* Output:
|
|
*
|
|
* C*
|
|
* hello world
|
|
*
|
|
* Raises an exception if the new process could not execute.
|
|
*/
|
|
|
|
static VALUE
|
|
rb_f_system(int argc, VALUE *argv, VALUE _)
|
|
{
|
|
VALUE execarg_obj = rb_execarg_new(argc, argv, TRUE, TRUE);
|
|
struct rb_execarg *eargp = rb_execarg_get(execarg_obj);
|
|
|
|
struct rb_process_status status = {0};
|
|
eargp->status = &status;
|
|
|
|
rb_last_status_clear();
|
|
|
|
// This function can set the thread's last status.
|
|
// May be different from waitpid_state.pid on exec failure.
|
|
rb_pid_t pid = rb_execarg_spawn(execarg_obj, 0, 0);
|
|
|
|
if (pid > 0) {
|
|
VALUE status = rb_process_status_wait(pid, 0);
|
|
|
|
struct rb_process_status *data = rb_check_typeddata(status, &rb_process_status_type);
|
|
|
|
// Set the last status:
|
|
rb_obj_freeze(status);
|
|
GET_THREAD()->last_status = status;
|
|
|
|
if (data->status == EXIT_SUCCESS) {
|
|
return Qtrue;
|
|
}
|
|
|
|
if (data->error != 0) {
|
|
if (eargp->exception) {
|
|
VALUE command = eargp->invoke.sh.shell_script;
|
|
RB_GC_GUARD(execarg_obj);
|
|
rb_syserr_fail_str(data->error, command);
|
|
}
|
|
else {
|
|
return Qnil;
|
|
}
|
|
}
|
|
else if (eargp->exception) {
|
|
VALUE command = eargp->invoke.sh.shell_script;
|
|
VALUE str = rb_str_new_cstr("Command failed with");
|
|
rb_str_cat_cstr(pst_message_status(str, data->status), ": ");
|
|
rb_str_append(str, command);
|
|
RB_GC_GUARD(execarg_obj);
|
|
rb_exc_raise(rb_exc_new_str(rb_eRuntimeError, str));
|
|
}
|
|
else {
|
|
return Qfalse;
|
|
}
|
|
|
|
RB_GC_GUARD(status);
|
|
}
|
|
|
|
if (eargp->exception) {
|
|
VALUE command = eargp->invoke.sh.shell_script;
|
|
RB_GC_GUARD(execarg_obj);
|
|
rb_syserr_fail_str(errno, command);
|
|
}
|
|
else {
|
|
return Qnil;
|
|
}
|
|
}
|
|
|
|
/*
|
|
* call-seq:
|
|
* spawn([env, ] command_line, options = {}) -> pid
|
|
* spawn([env, ] exe_path, *args, options = {}) -> pid
|
|
*
|
|
* Creates a new child process by doing one of the following
|
|
* in that process:
|
|
*
|
|
* - Passing string +command_line+ to the shell.
|
|
* - Invoking the executable at +exe_path+.
|
|
*
|
|
* This method has potential security vulnerabilities if called with untrusted input;
|
|
* see {Command Injection}[rdoc-ref:command_injection.rdoc].
|
|
*
|
|
* Returns the process ID (pid) of the new process,
|
|
* without waiting for it to complete.
|
|
*
|
|
* To avoid zombie processes, the parent process should call either:
|
|
*
|
|
* - Process.wait, to collect the termination statuses of its children.
|
|
* - Process.detach, to register disinterest in their status.
|
|
*
|
|
* The new process is created using the
|
|
* {exec system call}[https://pubs.opengroup.org/onlinepubs/9699919799.2018edition/functions/execve.html];
|
|
* it may inherit some of its environment from the calling program
|
|
* (possibly including open file descriptors).
|
|
*
|
|
* Argument +env+, if given, is a hash that affects +ENV+ for the new process;
|
|
* see {Execution Environment}[rdoc-ref:Process@Execution+Environment].
|
|
*
|
|
* Argument +options+ is a hash of options for the new process;
|
|
* see {Execution Options}[rdoc-ref:Process@Execution+Options].
|
|
*
|
|
* The first required argument is one of the following:
|
|
*
|
|
* - +command_line+ if it is a string,
|
|
* and if it begins with a shell reserved word or special built-in,
|
|
* or if it contains one or more meta characters.
|
|
* - +exe_path+ otherwise.
|
|
*
|
|
* <b>Argument +command_line+</b>
|
|
*
|
|
* \String argument +command_line+ is a command line to be passed to a shell;
|
|
* it must begin with a shell reserved word, begin with a special built-in,
|
|
* or contain meta characters:
|
|
*
|
|
* spawn('if true; then echo "Foo"; fi') # => 798847 # Shell reserved word.
|
|
* Process.wait # => 798847
|
|
* spawn('echo') # => 798848 # Built-in.
|
|
* Process.wait # => 798848
|
|
* spawn('date > /tmp/date.tmp') # => 798879 # Contains meta character.
|
|
* Process.wait # => 798849
|
|
* spawn('date > /nop/date.tmp') # => 798882 # Issues error message.
|
|
* Process.wait # => 798882
|
|
*
|
|
* The command line may also contain arguments and options for the command:
|
|
*
|
|
* spawn('echo "Foo"') # => 799031
|
|
* Process.wait # => 799031
|
|
*
|
|
* Output:
|
|
*
|
|
* Foo
|
|
*
|
|
* See {Execution Shell}[rdoc-ref:Process@Execution+Shell] for details about the shell.
|
|
*
|
|
* Raises an exception if the new process could not execute.
|
|
*
|
|
* <b>Argument +exe_path+</b>
|
|
*
|
|
* Argument +exe_path+ is one of the following:
|
|
*
|
|
* - The string path to an executable to be called.
|
|
* - A 2-element array containing the path to an executable
|
|
* and the string to be used as the name of the executing process.
|
|
*
|
|
* Example:
|
|
*
|
|
* spawn('/usr/bin/date') # => 799198 # Path to date on Unix-style system.
|
|
* Process.wait # => 799198
|
|
*
|
|
* Output:
|
|
*
|
|
* Thu Aug 31 10:06:48 AM CDT 2023
|
|
*
|
|
* Ruby invokes the executable directly, with no shell and no shell expansion.
|
|
*
|
|
* If one or more +args+ is given, each is an argument or option
|
|
* to be passed to the executable:
|
|
*
|
|
* spawn('echo', 'C*') # => 799392
|
|
* Process.wait # => 799392
|
|
* spawn('echo', 'hello', 'world') # => 799393
|
|
* Process.wait # => 799393
|
|
*
|
|
* Output:
|
|
*
|
|
* C*
|
|
* hello world
|
|
*
|
|
* Raises an exception if the new process could not execute.
|
|
*/
|
|
|
|
static VALUE
|
|
rb_f_spawn(int argc, VALUE *argv, VALUE _)
|
|
{
|
|
rb_pid_t pid;
|
|
char errmsg[CHILD_ERRMSG_BUFLEN] = { '\0' };
|
|
VALUE execarg_obj, fail_str;
|
|
struct rb_execarg *eargp;
|
|
|
|
execarg_obj = rb_execarg_new(argc, argv, TRUE, FALSE);
|
|
eargp = rb_execarg_get(execarg_obj);
|
|
fail_str = eargp->use_shell ? eargp->invoke.sh.shell_script : eargp->invoke.cmd.command_name;
|
|
|
|
pid = rb_execarg_spawn(execarg_obj, errmsg, sizeof(errmsg));
|
|
|
|
if (pid == -1) {
|
|
int err = errno;
|
|
rb_exec_fail(eargp, err, errmsg);
|
|
RB_GC_GUARD(execarg_obj);
|
|
rb_syserr_fail_str(err, fail_str);
|
|
}
|
|
#if defined(HAVE_WORKING_FORK) || defined(HAVE_SPAWNV)
|
|
return PIDT2NUM(pid);
|
|
#else
|
|
return Qnil;
|
|
#endif
|
|
}
|
|
|
|
/*
|
|
* call-seq:
|
|
* sleep(secs = nil) -> slept_secs
|
|
*
|
|
* Suspends execution of the current thread for the number of seconds
|
|
* specified by numeric argument +secs+, or forever if +secs+ is +nil+;
|
|
* returns the integer number of seconds suspended (rounded).
|
|
*
|
|
* Time.new # => 2008-03-08 19:56:19 +0900
|
|
* sleep 1.2 # => 1
|
|
* Time.new # => 2008-03-08 19:56:20 +0900
|
|
* sleep 1.9 # => 2
|
|
* Time.new # => 2008-03-08 19:56:22 +0900
|
|
*
|
|
*/
|
|
|
|
static VALUE
|
|
rb_f_sleep(int argc, VALUE *argv, VALUE _)
|
|
{
|
|
time_t beg = time(0);
|
|
VALUE scheduler = rb_fiber_scheduler_current();
|
|
|
|
if (scheduler != Qnil) {
|
|
rb_fiber_scheduler_kernel_sleepv(scheduler, argc, argv);
|
|
}
|
|
else {
|
|
if (argc == 0 || (argc == 1 && NIL_P(argv[0]))) {
|
|
rb_thread_sleep_forever();
|
|
}
|
|
else {
|
|
rb_check_arity(argc, 0, 1);
|
|
rb_thread_wait_for(rb_time_interval(argv[0]));
|
|
}
|
|
}
|
|
|
|
time_t end = time(0) - beg;
|
|
|
|
return TIMET2NUM(end);
|
|
}
|
|
|
|
|
|
#if (defined(HAVE_GETPGRP) && defined(GETPGRP_VOID)) || defined(HAVE_GETPGID)
|
|
/*
|
|
* call-seq:
|
|
* Process.getpgrp -> integer
|
|
*
|
|
* Returns the process group ID for the current process:
|
|
*
|
|
* Process.getpgid(0) # => 25527
|
|
* Process.getpgrp # => 25527
|
|
*
|
|
*/
|
|
|
|
static VALUE
|
|
proc_getpgrp(VALUE _)
|
|
{
|
|
rb_pid_t pgrp;
|
|
|
|
#if defined(HAVE_GETPGRP) && defined(GETPGRP_VOID)
|
|
pgrp = getpgrp();
|
|
if (pgrp < 0) rb_sys_fail(0);
|
|
return PIDT2NUM(pgrp);
|
|
#else /* defined(HAVE_GETPGID) */
|
|
pgrp = getpgid(0);
|
|
if (pgrp < 0) rb_sys_fail(0);
|
|
return PIDT2NUM(pgrp);
|
|
#endif
|
|
}
|
|
#else
|
|
#define proc_getpgrp rb_f_notimplement
|
|
#endif
|
|
|
|
|
|
#if defined(HAVE_SETPGID) || (defined(HAVE_SETPGRP) && defined(SETPGRP_VOID))
|
|
/*
|
|
* call-seq:
|
|
* Process.setpgrp -> 0
|
|
*
|
|
* Equivalent to <tt>setpgid(0, 0)</tt>.
|
|
*
|
|
* Not available on all platforms.
|
|
*/
|
|
|
|
static VALUE
|
|
proc_setpgrp(VALUE _)
|
|
{
|
|
/* check for posix setpgid() first; this matches the posix */
|
|
/* getpgrp() above. It appears that configure will set SETPGRP_VOID */
|
|
/* even though setpgrp(0,0) would be preferred. The posix call avoids */
|
|
/* this confusion. */
|
|
#ifdef HAVE_SETPGID
|
|
if (setpgid(0,0) < 0) rb_sys_fail(0);
|
|
#elif defined(HAVE_SETPGRP) && defined(SETPGRP_VOID)
|
|
if (setpgrp() < 0) rb_sys_fail(0);
|
|
#endif
|
|
return INT2FIX(0);
|
|
}
|
|
#else
|
|
#define proc_setpgrp rb_f_notimplement
|
|
#endif
|
|
|
|
|
|
#if defined(HAVE_GETPGID)
|
|
/*
|
|
* call-seq:
|
|
* Process.getpgid(pid) -> integer
|
|
*
|
|
* Returns the process group ID for the given process ID +pid+:
|
|
*
|
|
* Process.getpgid(Process.ppid) # => 25527
|
|
*
|
|
* Not available on all platforms.
|
|
*/
|
|
|
|
static VALUE
|
|
proc_getpgid(VALUE obj, VALUE pid)
|
|
{
|
|
rb_pid_t i;
|
|
|
|
i = getpgid(NUM2PIDT(pid));
|
|
if (i < 0) rb_sys_fail(0);
|
|
return PIDT2NUM(i);
|
|
}
|
|
#else
|
|
#define proc_getpgid rb_f_notimplement
|
|
#endif
|
|
|
|
|
|
#ifdef HAVE_SETPGID
|
|
/*
|
|
* call-seq:
|
|
* Process.setpgid(pid, pgid) -> 0
|
|
*
|
|
* Sets the process group ID for the process given by process ID +pid+
|
|
* to +pgid+.
|
|
*
|
|
* Not available on all platforms.
|
|
*/
|
|
|
|
static VALUE
|
|
proc_setpgid(VALUE obj, VALUE pid, VALUE pgrp)
|
|
{
|
|
rb_pid_t ipid, ipgrp;
|
|
|
|
ipid = NUM2PIDT(pid);
|
|
ipgrp = NUM2PIDT(pgrp);
|
|
|
|
if (setpgid(ipid, ipgrp) < 0) rb_sys_fail(0);
|
|
return INT2FIX(0);
|
|
}
|
|
#else
|
|
#define proc_setpgid rb_f_notimplement
|
|
#endif
|
|
|
|
|
|
#ifdef HAVE_GETSID
|
|
/*
|
|
* call-seq:
|
|
* Process.getsid(pid = nil) -> integer
|
|
*
|
|
* Returns the session ID of the given process ID +pid+,
|
|
* or of the current process if not given:
|
|
*
|
|
* Process.getsid # => 27422
|
|
* Process.getsid(0) # => 27422
|
|
* Process.getsid(Process.pid()) # => 27422
|
|
*
|
|
* Not available on all platforms.
|
|
*/
|
|
static VALUE
|
|
proc_getsid(int argc, VALUE *argv, VALUE _)
|
|
{
|
|
rb_pid_t sid;
|
|
rb_pid_t pid = 0;
|
|
|
|
if (rb_check_arity(argc, 0, 1) == 1 && !NIL_P(argv[0]))
|
|
pid = NUM2PIDT(argv[0]);
|
|
|
|
sid = getsid(pid);
|
|
if (sid < 0) rb_sys_fail(0);
|
|
return PIDT2NUM(sid);
|
|
}
|
|
#else
|
|
#define proc_getsid rb_f_notimplement
|
|
#endif
|
|
|
|
|
|
#if defined(HAVE_SETSID) || (defined(HAVE_SETPGRP) && defined(TIOCNOTTY))
|
|
#if !defined(HAVE_SETSID)
|
|
static rb_pid_t ruby_setsid(void);
|
|
#define setsid() ruby_setsid()
|
|
#endif
|
|
/*
|
|
* call-seq:
|
|
* Process.setsid -> integer
|
|
*
|
|
* Establishes the current process as a new session and process group leader,
|
|
* with no controlling tty;
|
|
* returns the session ID:
|
|
*
|
|
* Process.setsid # => 27422
|
|
*
|
|
* Not available on all platforms.
|
|
*/
|
|
|
|
static VALUE
|
|
proc_setsid(VALUE _)
|
|
{
|
|
rb_pid_t pid;
|
|
|
|
pid = setsid();
|
|
if (pid < 0) rb_sys_fail(0);
|
|
return PIDT2NUM(pid);
|
|
}
|
|
|
|
#if !defined(HAVE_SETSID)
|
|
#define HAVE_SETSID 1
|
|
static rb_pid_t
|
|
ruby_setsid(void)
|
|
{
|
|
rb_pid_t pid;
|
|
int ret;
|
|
|
|
pid = getpid();
|
|
#if defined(SETPGRP_VOID)
|
|
ret = setpgrp();
|
|
/* If `pid_t setpgrp(void)' is equivalent to setsid(),
|
|
`ret' will be the same value as `pid', and following open() will fail.
|
|
In Linux, `int setpgrp(void)' is equivalent to setpgid(0, 0). */
|
|
#else
|
|
ret = setpgrp(0, pid);
|
|
#endif
|
|
if (ret == -1) return -1;
|
|
|
|
if ((fd = rb_cloexec_open("/dev/tty", O_RDWR, 0)) >= 0) {
|
|
rb_update_max_fd(fd);
|
|
ioctl(fd, TIOCNOTTY, NULL);
|
|
close(fd);
|
|
}
|
|
return pid;
|
|
}
|
|
#endif
|
|
#else
|
|
#define proc_setsid rb_f_notimplement
|
|
#endif
|
|
|
|
|
|
#ifdef HAVE_GETPRIORITY
|
|
/*
|
|
* call-seq:
|
|
* Process.getpriority(kind, id) -> integer
|
|
*
|
|
* Returns the scheduling priority for specified process, process group,
|
|
* or user.
|
|
*
|
|
* Argument +kind+ is one of:
|
|
*
|
|
* - Process::PRIO_PROCESS: return priority for process.
|
|
* - Process::PRIO_PGRP: return priority for process group.
|
|
* - Process::PRIO_USER: return priority for user.
|
|
*
|
|
* Argument +id+ is the ID for the process, process group, or user;
|
|
* zero specified the current ID for +kind+.
|
|
*
|
|
* Examples:
|
|
*
|
|
* Process.getpriority(Process::PRIO_USER, 0) # => 19
|
|
* Process.getpriority(Process::PRIO_PROCESS, 0) # => 19
|
|
*
|
|
* Not available on all platforms.
|
|
*/
|
|
|
|
static VALUE
|
|
proc_getpriority(VALUE obj, VALUE which, VALUE who)
|
|
{
|
|
int prio, iwhich, iwho;
|
|
|
|
iwhich = NUM2INT(which);
|
|
iwho = NUM2INT(who);
|
|
|
|
errno = 0;
|
|
prio = getpriority(iwhich, iwho);
|
|
if (errno) rb_sys_fail(0);
|
|
return INT2FIX(prio);
|
|
}
|
|
#else
|
|
#define proc_getpriority rb_f_notimplement
|
|
#endif
|
|
|
|
|
|
#ifdef HAVE_GETPRIORITY
|
|
/*
|
|
* call-seq:
|
|
* Process.setpriority(kind, integer, priority) -> 0
|
|
*
|
|
* See Process.getpriority.
|
|
*
|
|
* Examples:
|
|
*
|
|
* Process.setpriority(Process::PRIO_USER, 0, 19) # => 0
|
|
* Process.setpriority(Process::PRIO_PROCESS, 0, 19) # => 0
|
|
* Process.getpriority(Process::PRIO_USER, 0) # => 19
|
|
* Process.getpriority(Process::PRIO_PROCESS, 0) # => 19
|
|
*
|
|
* Not available on all platforms.
|
|
*/
|
|
|
|
static VALUE
|
|
proc_setpriority(VALUE obj, VALUE which, VALUE who, VALUE prio)
|
|
{
|
|
int iwhich, iwho, iprio;
|
|
|
|
iwhich = NUM2INT(which);
|
|
iwho = NUM2INT(who);
|
|
iprio = NUM2INT(prio);
|
|
|
|
if (setpriority(iwhich, iwho, iprio) < 0)
|
|
rb_sys_fail(0);
|
|
return INT2FIX(0);
|
|
}
|
|
#else
|
|
#define proc_setpriority rb_f_notimplement
|
|
#endif
|
|
|
|
#if defined(HAVE_SETRLIMIT) && defined(NUM2RLIM)
|
|
static int
|
|
rlimit_resource_name2int(const char *name, long len, int casetype)
|
|
{
|
|
int resource;
|
|
const char *p;
|
|
#define RESCHECK(r) \
|
|
do { \
|
|
if (len == rb_strlen_lit(#r) && STRCASECMP(name, #r) == 0) { \
|
|
resource = RLIMIT_##r; \
|
|
goto found; \
|
|
} \
|
|
} while (0)
|
|
|
|
switch (TOUPPER(*name)) {
|
|
case 'A':
|
|
#ifdef RLIMIT_AS
|
|
RESCHECK(AS);
|
|
#endif
|
|
break;
|
|
|
|
case 'C':
|
|
#ifdef RLIMIT_CORE
|
|
RESCHECK(CORE);
|
|
#endif
|
|
#ifdef RLIMIT_CPU
|
|
RESCHECK(CPU);
|
|
#endif
|
|
break;
|
|
|
|
case 'D':
|
|
#ifdef RLIMIT_DATA
|
|
RESCHECK(DATA);
|
|
#endif
|
|
break;
|
|
|
|
case 'F':
|
|
#ifdef RLIMIT_FSIZE
|
|
RESCHECK(FSIZE);
|
|
#endif
|
|
break;
|
|
|
|
case 'M':
|
|
#ifdef RLIMIT_MEMLOCK
|
|
RESCHECK(MEMLOCK);
|
|
#endif
|
|
#ifdef RLIMIT_MSGQUEUE
|
|
RESCHECK(MSGQUEUE);
|
|
#endif
|
|
break;
|
|
|
|
case 'N':
|
|
#ifdef RLIMIT_NOFILE
|
|
RESCHECK(NOFILE);
|
|
#endif
|
|
#ifdef RLIMIT_NPROC
|
|
RESCHECK(NPROC);
|
|
#endif
|
|
#ifdef RLIMIT_NPTS
|
|
RESCHECK(NPTS);
|
|
#endif
|
|
#ifdef RLIMIT_NICE
|
|
RESCHECK(NICE);
|
|
#endif
|
|
break;
|
|
|
|
case 'R':
|
|
#ifdef RLIMIT_RSS
|
|
RESCHECK(RSS);
|
|
#endif
|
|
#ifdef RLIMIT_RTPRIO
|
|
RESCHECK(RTPRIO);
|
|
#endif
|
|
#ifdef RLIMIT_RTTIME
|
|
RESCHECK(RTTIME);
|
|
#endif
|
|
break;
|
|
|
|
case 'S':
|
|
#ifdef RLIMIT_STACK
|
|
RESCHECK(STACK);
|
|
#endif
|
|
#ifdef RLIMIT_SBSIZE
|
|
RESCHECK(SBSIZE);
|
|
#endif
|
|
#ifdef RLIMIT_SIGPENDING
|
|
RESCHECK(SIGPENDING);
|
|
#endif
|
|
break;
|
|
}
|
|
return -1;
|
|
|
|
found:
|
|
switch (casetype) {
|
|
case 0:
|
|
for (p = name; *p; p++)
|
|
if (!ISUPPER(*p))
|
|
return -1;
|
|
break;
|
|
|
|
case 1:
|
|
for (p = name; *p; p++)
|
|
if (!ISLOWER(*p))
|
|
return -1;
|
|
break;
|
|
|
|
default:
|
|
rb_bug("unexpected casetype");
|
|
}
|
|
return resource;
|
|
#undef RESCHECK
|
|
}
|
|
|
|
static int
|
|
rlimit_type_by_hname(const char *name, long len)
|
|
{
|
|
return rlimit_resource_name2int(name, len, 0);
|
|
}
|
|
|
|
static int
|
|
rlimit_type_by_lname(const char *name, long len)
|
|
{
|
|
return rlimit_resource_name2int(name, len, 1);
|
|
}
|
|
|
|
static int
|
|
rlimit_type_by_sym(VALUE key)
|
|
{
|
|
VALUE name = rb_sym2str(key);
|
|
const char *rname = RSTRING_PTR(name);
|
|
long len = RSTRING_LEN(name);
|
|
int rtype = -1;
|
|
static const char prefix[] = "rlimit_";
|
|
enum {prefix_len = sizeof(prefix)-1};
|
|
|
|
if (len > prefix_len && strncmp(prefix, rname, prefix_len) == 0) {
|
|
rtype = rlimit_type_by_lname(rname + prefix_len, len - prefix_len);
|
|
}
|
|
|
|
RB_GC_GUARD(key);
|
|
return rtype;
|
|
}
|
|
|
|
static int
|
|
rlimit_resource_type(VALUE rtype)
|
|
{
|
|
const char *name;
|
|
long len;
|
|
VALUE v;
|
|
int r;
|
|
|
|
switch (TYPE(rtype)) {
|
|
case T_SYMBOL:
|
|
v = rb_sym2str(rtype);
|
|
name = RSTRING_PTR(v);
|
|
len = RSTRING_LEN(v);
|
|
break;
|
|
|
|
default:
|
|
v = rb_check_string_type(rtype);
|
|
if (!NIL_P(v)) {
|
|
rtype = v;
|
|
case T_STRING:
|
|
name = StringValueCStr(rtype);
|
|
len = RSTRING_LEN(rtype);
|
|
break;
|
|
}
|
|
/* fall through */
|
|
|
|
case T_FIXNUM:
|
|
case T_BIGNUM:
|
|
return NUM2INT(rtype);
|
|
}
|
|
|
|
r = rlimit_type_by_hname(name, len);
|
|
if (r != -1)
|
|
return r;
|
|
|
|
rb_raise(rb_eArgError, "invalid resource name: % "PRIsVALUE, rtype);
|
|
|
|
UNREACHABLE_RETURN(-1);
|
|
}
|
|
|
|
static rlim_t
|
|
rlimit_resource_value(VALUE rval)
|
|
{
|
|
const char *name;
|
|
VALUE v;
|
|
|
|
switch (TYPE(rval)) {
|
|
case T_SYMBOL:
|
|
v = rb_sym2str(rval);
|
|
name = RSTRING_PTR(v);
|
|
break;
|
|
|
|
default:
|
|
v = rb_check_string_type(rval);
|
|
if (!NIL_P(v)) {
|
|
rval = v;
|
|
case T_STRING:
|
|
name = StringValueCStr(rval);
|
|
break;
|
|
}
|
|
/* fall through */
|
|
|
|
case T_FIXNUM:
|
|
case T_BIGNUM:
|
|
return NUM2RLIM(rval);
|
|
}
|
|
|
|
#ifdef RLIM_INFINITY
|
|
if (strcmp(name, "INFINITY") == 0) return RLIM_INFINITY;
|
|
#endif
|
|
#ifdef RLIM_SAVED_MAX
|
|
if (strcmp(name, "SAVED_MAX") == 0) return RLIM_SAVED_MAX;
|
|
#endif
|
|
#ifdef RLIM_SAVED_CUR
|
|
if (strcmp(name, "SAVED_CUR") == 0) return RLIM_SAVED_CUR;
|
|
#endif
|
|
rb_raise(rb_eArgError, "invalid resource value: %"PRIsVALUE, rval);
|
|
|
|
UNREACHABLE_RETURN((rlim_t)-1);
|
|
}
|
|
#endif
|
|
|
|
#if defined(HAVE_GETRLIMIT) && defined(RLIM2NUM)
|
|
/*
|
|
* call-seq:
|
|
* Process.getrlimit(resource) -> [cur_limit, max_limit]
|
|
*
|
|
* Returns a 2-element array of the current (soft) limit
|
|
* and maximum (hard) limit for the given +resource+.
|
|
*
|
|
* Argument +resource+ specifies the resource whose limits are to be returned;
|
|
* see Process.setrlimit.
|
|
*
|
|
* Each of the returned values +cur_limit+ and +max_limit+ is an integer;
|
|
* see Process.setrlimit.
|
|
*
|
|
* Example:
|
|
*
|
|
* Process.getrlimit(:CORE) # => [0, 18446744073709551615]
|
|
*
|
|
* See Process.setrlimit.
|
|
*
|
|
* Not available on all platforms.
|
|
*/
|
|
|
|
static VALUE
|
|
proc_getrlimit(VALUE obj, VALUE resource)
|
|
{
|
|
struct rlimit rlim;
|
|
|
|
if (getrlimit(rlimit_resource_type(resource), &rlim) < 0) {
|
|
rb_sys_fail("getrlimit");
|
|
}
|
|
return rb_assoc_new(RLIM2NUM(rlim.rlim_cur), RLIM2NUM(rlim.rlim_max));
|
|
}
|
|
#else
|
|
#define proc_getrlimit rb_f_notimplement
|
|
#endif
|
|
|
|
#if defined(HAVE_SETRLIMIT) && defined(NUM2RLIM)
|
|
/*
|
|
* call-seq:
|
|
* Process.setrlimit(resource, cur_limit, max_limit = cur_limit) -> nil
|
|
*
|
|
* Sets limits for the current process for the given +resource+
|
|
* to +cur_limit+ (soft limit) and +max_limit+ (hard limit);
|
|
* returns +nil+.
|
|
*
|
|
* Argument +resource+ specifies the resource whose limits are to be set;
|
|
* the argument may be given as a symbol, as a string, or as a constant
|
|
* beginning with <tt>Process::RLIMIT_</tt>
|
|
* (e.g., +:CORE+, <tt>'CORE'</tt>, or <tt>Process::RLIMIT_CORE</tt>.
|
|
*
|
|
* The resources available and supported are system-dependent,
|
|
* and may include (here expressed as symbols):
|
|
*
|
|
* - +:AS+: Total available memory (bytes) (SUSv3, NetBSD, FreeBSD, OpenBSD except 4.4BSD-Lite).
|
|
* - +:CORE+: Core size (bytes) (SUSv3).
|
|
* - +:CPU+: CPU time (seconds) (SUSv3).
|
|
* - +:DATA+: Data segment (bytes) (SUSv3).
|
|
* - +:FSIZE+: File size (bytes) (SUSv3).
|
|
* - +:MEMLOCK+: Total size for mlock(2) (bytes) (4.4BSD, GNU/Linux).
|
|
* - +:MSGQUEUE+: Allocation for POSIX message queues (bytes) (GNU/Linux).
|
|
* - +:NICE+: Ceiling on process's nice(2) value (number) (GNU/Linux).
|
|
* - +:NOFILE+: File descriptors (number) (SUSv3).
|
|
* - +:NPROC+: Number of processes for the user (number) (4.4BSD, GNU/Linux).
|
|
* - +:NPTS+: Number of pseudo terminals (number) (FreeBSD).
|
|
* - +:RSS+: Resident memory size (bytes) (4.2BSD, GNU/Linux).
|
|
* - +:RTPRIO+: Ceiling on the process's real-time priority (number) (GNU/Linux).
|
|
* - +:RTTIME+: CPU time for real-time process (us) (GNU/Linux).
|
|
* - +:SBSIZE+: All socket buffers (bytes) (NetBSD, FreeBSD).
|
|
* - +:SIGPENDING+: Number of queued signals allowed (signals) (GNU/Linux).
|
|
* - +:STACK+: Stack size (bytes) (SUSv3).
|
|
*
|
|
* Arguments +cur_limit+ and +max_limit+ may be:
|
|
*
|
|
* - Integers (+max_limit+ should not be smaller than +cur_limit+).
|
|
* - Symbol +:SAVED_MAX+, string <tt>'SAVED_MAX'</tt>,
|
|
* or constant <tt>Process::RLIM_SAVED_MAX</tt>: saved maximum limit.
|
|
* - Symbol +:SAVED_CUR+, string <tt>'SAVED_CUR'</tt>,
|
|
* or constant <tt>Process::RLIM_SAVED_CUR</tt>: saved current limit.
|
|
* - Symbol +:INFINITY+, string <tt>'INFINITY'</tt>,
|
|
* or constant <tt>Process::RLIM_INFINITY</tt>: no limit on resource.
|
|
*
|
|
* This example raises the soft limit of core size to
|
|
* the hard limit to try to make core dump possible:
|
|
*
|
|
* Process.setrlimit(:CORE, Process.getrlimit(:CORE)[1])
|
|
*
|
|
* Not available on all platforms.
|
|
*/
|
|
|
|
static VALUE
|
|
proc_setrlimit(int argc, VALUE *argv, VALUE obj)
|
|
{
|
|
VALUE resource, rlim_cur, rlim_max;
|
|
struct rlimit rlim;
|
|
|
|
rb_check_arity(argc, 2, 3);
|
|
resource = argv[0];
|
|
rlim_cur = argv[1];
|
|
if (argc < 3 || NIL_P(rlim_max = argv[2]))
|
|
rlim_max = rlim_cur;
|
|
|
|
rlim.rlim_cur = rlimit_resource_value(rlim_cur);
|
|
rlim.rlim_max = rlimit_resource_value(rlim_max);
|
|
|
|
if (setrlimit(rlimit_resource_type(resource), &rlim) < 0) {
|
|
rb_sys_fail("setrlimit");
|
|
}
|
|
return Qnil;
|
|
}
|
|
#else
|
|
#define proc_setrlimit rb_f_notimplement
|
|
#endif
|
|
|
|
static int under_uid_switch = 0;
|
|
static void
|
|
check_uid_switch(void)
|
|
{
|
|
if (under_uid_switch) {
|
|
rb_raise(rb_eRuntimeError, "can't handle UID while evaluating block given to Process::UID.switch method");
|
|
}
|
|
}
|
|
|
|
static int under_gid_switch = 0;
|
|
static void
|
|
check_gid_switch(void)
|
|
{
|
|
if (under_gid_switch) {
|
|
rb_raise(rb_eRuntimeError, "can't handle GID while evaluating block given to Process::UID.switch method");
|
|
}
|
|
}
|
|
|
|
|
|
#if defined(HAVE_PWD_H)
|
|
/**
|
|
* Best-effort attempt to obtain the name of the login user, if any,
|
|
* associated with the process. Processes not descended from login(1) (or
|
|
* similar) may not have a logged-in user; returns Qnil in that case.
|
|
*/
|
|
VALUE
|
|
rb_getlogin(void)
|
|
{
|
|
#if ( !defined(USE_GETLOGIN_R) && !defined(USE_GETLOGIN) )
|
|
return Qnil;
|
|
#else
|
|
char MAYBE_UNUSED(*login) = NULL;
|
|
|
|
# ifdef USE_GETLOGIN_R
|
|
|
|
#if defined(__FreeBSD__)
|
|
typedef int getlogin_r_size_t;
|
|
#else
|
|
typedef size_t getlogin_r_size_t;
|
|
#endif
|
|
|
|
long loginsize = GETLOGIN_R_SIZE_INIT; /* maybe -1 */
|
|
|
|
if (loginsize < 0)
|
|
loginsize = GETLOGIN_R_SIZE_DEFAULT;
|
|
|
|
VALUE maybe_result = rb_str_buf_new(loginsize);
|
|
|
|
login = RSTRING_PTR(maybe_result);
|
|
loginsize = rb_str_capacity(maybe_result);
|
|
rb_str_set_len(maybe_result, loginsize);
|
|
|
|
int gle;
|
|
errno = 0;
|
|
while ((gle = getlogin_r(login, (getlogin_r_size_t)loginsize)) != 0) {
|
|
|
|
if (gle == ENOTTY || gle == ENXIO || gle == ENOENT) {
|
|
rb_str_resize(maybe_result, 0);
|
|
return Qnil;
|
|
}
|
|
|
|
if (gle != ERANGE || loginsize >= GETLOGIN_R_SIZE_LIMIT) {
|
|
rb_str_resize(maybe_result, 0);
|
|
rb_syserr_fail(gle, "getlogin_r");
|
|
}
|
|
|
|
rb_str_modify_expand(maybe_result, loginsize);
|
|
login = RSTRING_PTR(maybe_result);
|
|
loginsize = rb_str_capacity(maybe_result);
|
|
}
|
|
|
|
if (login == NULL) {
|
|
rb_str_resize(maybe_result, 0);
|
|
return Qnil;
|
|
}
|
|
|
|
return maybe_result;
|
|
|
|
# elif USE_GETLOGIN
|
|
|
|
errno = 0;
|
|
login = getlogin();
|
|
if (errno) {
|
|
if (errno == ENOTTY || errno == ENXIO || errno == ENOENT) {
|
|
return Qnil;
|
|
}
|
|
rb_syserr_fail(errno, "getlogin");
|
|
}
|
|
|
|
return login ? rb_str_new_cstr(login) : Qnil;
|
|
# endif
|
|
|
|
#endif
|
|
}
|
|
|
|
VALUE
|
|
rb_getpwdirnam_for_login(VALUE login_name)
|
|
{
|
|
#if ( !defined(USE_GETPWNAM_R) && !defined(USE_GETPWNAM) )
|
|
return Qnil;
|
|
#else
|
|
|
|
if (NIL_P(login_name)) {
|
|
/* nothing to do; no name with which to query the password database */
|
|
return Qnil;
|
|
}
|
|
|
|
char *login = RSTRING_PTR(login_name);
|
|
|
|
struct passwd *pwptr;
|
|
|
|
# ifdef USE_GETPWNAM_R
|
|
|
|
struct passwd pwdnm;
|
|
char *bufnm;
|
|
long bufsizenm = GETPW_R_SIZE_INIT; /* maybe -1 */
|
|
|
|
if (bufsizenm < 0)
|
|
bufsizenm = GETPW_R_SIZE_DEFAULT;
|
|
|
|
VALUE getpwnm_tmp = rb_str_tmp_new(bufsizenm);
|
|
|
|
bufnm = RSTRING_PTR(getpwnm_tmp);
|
|
bufsizenm = rb_str_capacity(getpwnm_tmp);
|
|
rb_str_set_len(getpwnm_tmp, bufsizenm);
|
|
|
|
int enm;
|
|
errno = 0;
|
|
while ((enm = getpwnam_r(login, &pwdnm, bufnm, bufsizenm, &pwptr)) != 0) {
|
|
|
|
if (enm == ENOENT || enm== ESRCH || enm == EBADF || enm == EPERM) {
|
|
/* not found; non-errors */
|
|
rb_str_resize(getpwnm_tmp, 0);
|
|
return Qnil;
|
|
}
|
|
|
|
if (enm != ERANGE || bufsizenm >= GETPW_R_SIZE_LIMIT) {
|
|
rb_str_resize(getpwnm_tmp, 0);
|
|
rb_syserr_fail(enm, "getpwnam_r");
|
|
}
|
|
|
|
rb_str_modify_expand(getpwnm_tmp, bufsizenm);
|
|
bufnm = RSTRING_PTR(getpwnm_tmp);
|
|
bufsizenm = rb_str_capacity(getpwnm_tmp);
|
|
}
|
|
|
|
if (pwptr == NULL) {
|
|
/* no record in the password database for the login name */
|
|
rb_str_resize(getpwnm_tmp, 0);
|
|
return Qnil;
|
|
}
|
|
|
|
/* found it */
|
|
VALUE result = rb_str_new_cstr(pwptr->pw_dir);
|
|
rb_str_resize(getpwnm_tmp, 0);
|
|
return result;
|
|
|
|
# elif USE_GETPWNAM
|
|
|
|
errno = 0;
|
|
pwptr = getpwnam(login);
|
|
if (pwptr) {
|
|
/* found it */
|
|
return rb_str_new_cstr(pwptr->pw_dir);
|
|
}
|
|
if (errno
|
|
/* avoid treating as errors errno values that indicate "not found" */
|
|
&& ( errno != ENOENT && errno != ESRCH && errno != EBADF && errno != EPERM)) {
|
|
rb_syserr_fail(errno, "getpwnam");
|
|
}
|
|
|
|
return Qnil; /* not found */
|
|
# endif
|
|
|
|
#endif
|
|
}
|
|
|
|
/**
|
|
* Look up the user's dflt home dir in the password db, by uid.
|
|
*/
|
|
VALUE
|
|
rb_getpwdiruid(void)
|
|
{
|
|
# if !defined(USE_GETPWUID_R) && !defined(USE_GETPWUID)
|
|
/* Should never happen... </famous-last-words> */
|
|
return Qnil;
|
|
# else
|
|
uid_t ruid = getuid();
|
|
|
|
struct passwd *pwptr;
|
|
|
|
# ifdef USE_GETPWUID_R
|
|
|
|
struct passwd pwdid;
|
|
char *bufid;
|
|
long bufsizeid = GETPW_R_SIZE_INIT; /* maybe -1 */
|
|
|
|
if (bufsizeid < 0)
|
|
bufsizeid = GETPW_R_SIZE_DEFAULT;
|
|
|
|
VALUE getpwid_tmp = rb_str_tmp_new(bufsizeid);
|
|
|
|
bufid = RSTRING_PTR(getpwid_tmp);
|
|
bufsizeid = rb_str_capacity(getpwid_tmp);
|
|
rb_str_set_len(getpwid_tmp, bufsizeid);
|
|
|
|
int eid;
|
|
errno = 0;
|
|
while ((eid = getpwuid_r(ruid, &pwdid, bufid, bufsizeid, &pwptr)) != 0) {
|
|
|
|
if (eid == ENOENT || eid== ESRCH || eid == EBADF || eid == EPERM) {
|
|
/* not found; non-errors */
|
|
rb_str_resize(getpwid_tmp, 0);
|
|
return Qnil;
|
|
}
|
|
|
|
if (eid != ERANGE || bufsizeid >= GETPW_R_SIZE_LIMIT) {
|
|
rb_str_resize(getpwid_tmp, 0);
|
|
rb_syserr_fail(eid, "getpwuid_r");
|
|
}
|
|
|
|
rb_str_modify_expand(getpwid_tmp, bufsizeid);
|
|
bufid = RSTRING_PTR(getpwid_tmp);
|
|
bufsizeid = rb_str_capacity(getpwid_tmp);
|
|
}
|
|
|
|
if (pwptr == NULL) {
|
|
/* no record in the password database for the uid */
|
|
rb_str_resize(getpwid_tmp, 0);
|
|
return Qnil;
|
|
}
|
|
|
|
/* found it */
|
|
VALUE result = rb_str_new_cstr(pwptr->pw_dir);
|
|
rb_str_resize(getpwid_tmp, 0);
|
|
return result;
|
|
|
|
# elif defined(USE_GETPWUID)
|
|
|
|
errno = 0;
|
|
pwptr = getpwuid(ruid);
|
|
if (pwptr) {
|
|
/* found it */
|
|
return rb_str_new_cstr(pwptr->pw_dir);
|
|
}
|
|
if (errno
|
|
/* avoid treating as errors errno values that indicate "not found" */
|
|
&& ( errno == ENOENT || errno == ESRCH || errno == EBADF || errno == EPERM)) {
|
|
rb_syserr_fail(errno, "getpwuid");
|
|
}
|
|
|
|
return Qnil; /* not found */
|
|
# endif
|
|
|
|
#endif /* !defined(USE_GETPWUID_R) && !defined(USE_GETPWUID) */
|
|
}
|
|
#endif /* HAVE_PWD_H */
|
|
|
|
|
|
/*********************************************************************
|
|
* Document-class: Process::Sys
|
|
*
|
|
* The Process::Sys module contains UID and GID
|
|
* functions which provide direct bindings to the system calls of the
|
|
* same names instead of the more-portable versions of the same
|
|
* functionality found in the Process,
|
|
* Process::UID, and Process::GID modules.
|
|
*/
|
|
|
|
#if defined(HAVE_PWD_H)
|
|
static rb_uid_t
|
|
obj2uid(VALUE id
|
|
# ifdef USE_GETPWNAM_R
|
|
, VALUE *getpw_tmp
|
|
# endif
|
|
)
|
|
{
|
|
rb_uid_t uid;
|
|
VALUE tmp;
|
|
|
|
if (FIXNUM_P(id) || NIL_P(tmp = rb_check_string_type(id))) {
|
|
uid = NUM2UIDT(id);
|
|
}
|
|
else {
|
|
const char *usrname = StringValueCStr(id);
|
|
struct passwd *pwptr;
|
|
#ifdef USE_GETPWNAM_R
|
|
struct passwd pwbuf;
|
|
char *getpw_buf;
|
|
long getpw_buf_len;
|
|
int e;
|
|
if (!*getpw_tmp) {
|
|
getpw_buf_len = GETPW_R_SIZE_INIT;
|
|
if (getpw_buf_len < 0) getpw_buf_len = GETPW_R_SIZE_DEFAULT;
|
|
*getpw_tmp = rb_str_tmp_new(getpw_buf_len);
|
|
}
|
|
getpw_buf = RSTRING_PTR(*getpw_tmp);
|
|
getpw_buf_len = rb_str_capacity(*getpw_tmp);
|
|
rb_str_set_len(*getpw_tmp, getpw_buf_len);
|
|
errno = 0;
|
|
while ((e = getpwnam_r(usrname, &pwbuf, getpw_buf, getpw_buf_len, &pwptr)) != 0) {
|
|
if (e != ERANGE || getpw_buf_len >= GETPW_R_SIZE_LIMIT) {
|
|
rb_str_resize(*getpw_tmp, 0);
|
|
rb_syserr_fail(e, "getpwnam_r");
|
|
}
|
|
rb_str_modify_expand(*getpw_tmp, getpw_buf_len);
|
|
getpw_buf = RSTRING_PTR(*getpw_tmp);
|
|
getpw_buf_len = rb_str_capacity(*getpw_tmp);
|
|
}
|
|
#else
|
|
pwptr = getpwnam(usrname);
|
|
#endif
|
|
if (!pwptr) {
|
|
#ifndef USE_GETPWNAM_R
|
|
endpwent();
|
|
#endif
|
|
rb_raise(rb_eArgError, "can't find user for %"PRIsVALUE, id);
|
|
}
|
|
uid = pwptr->pw_uid;
|
|
#ifndef USE_GETPWNAM_R
|
|
endpwent();
|
|
#endif
|
|
}
|
|
return uid;
|
|
}
|
|
|
|
# ifdef p_uid_from_name
|
|
/*
|
|
* call-seq:
|
|
* Process::UID.from_name(name) -> uid
|
|
*
|
|
* Get the user ID by the _name_.
|
|
* If the user is not found, +ArgumentError+ will be raised.
|
|
*
|
|
* Process::UID.from_name("root") #=> 0
|
|
* Process::UID.from_name("nosuchuser") #=> can't find user for nosuchuser (ArgumentError)
|
|
*/
|
|
|
|
static VALUE
|
|
p_uid_from_name(VALUE self, VALUE id)
|
|
{
|
|
return UIDT2NUM(OBJ2UID(id));
|
|
}
|
|
# endif
|
|
#endif
|
|
|
|
#if defined(HAVE_GRP_H)
|
|
static rb_gid_t
|
|
obj2gid(VALUE id
|
|
# ifdef USE_GETGRNAM_R
|
|
, VALUE *getgr_tmp
|
|
# endif
|
|
)
|
|
{
|
|
rb_gid_t gid;
|
|
VALUE tmp;
|
|
|
|
if (FIXNUM_P(id) || NIL_P(tmp = rb_check_string_type(id))) {
|
|
gid = NUM2GIDT(id);
|
|
}
|
|
else {
|
|
const char *grpname = StringValueCStr(id);
|
|
struct group *grptr;
|
|
#ifdef USE_GETGRNAM_R
|
|
struct group grbuf;
|
|
char *getgr_buf;
|
|
long getgr_buf_len;
|
|
int e;
|
|
if (!*getgr_tmp) {
|
|
getgr_buf_len = GETGR_R_SIZE_INIT;
|
|
if (getgr_buf_len < 0) getgr_buf_len = GETGR_R_SIZE_DEFAULT;
|
|
*getgr_tmp = rb_str_tmp_new(getgr_buf_len);
|
|
}
|
|
getgr_buf = RSTRING_PTR(*getgr_tmp);
|
|
getgr_buf_len = rb_str_capacity(*getgr_tmp);
|
|
rb_str_set_len(*getgr_tmp, getgr_buf_len);
|
|
errno = 0;
|
|
while ((e = getgrnam_r(grpname, &grbuf, getgr_buf, getgr_buf_len, &grptr)) != 0) {
|
|
if (e != ERANGE || getgr_buf_len >= GETGR_R_SIZE_LIMIT) {
|
|
rb_str_resize(*getgr_tmp, 0);
|
|
rb_syserr_fail(e, "getgrnam_r");
|
|
}
|
|
rb_str_modify_expand(*getgr_tmp, getgr_buf_len);
|
|
getgr_buf = RSTRING_PTR(*getgr_tmp);
|
|
getgr_buf_len = rb_str_capacity(*getgr_tmp);
|
|
}
|
|
#elif defined(HAVE_GETGRNAM)
|
|
grptr = getgrnam(grpname);
|
|
#else
|
|
grptr = NULL;
|
|
#endif
|
|
if (!grptr) {
|
|
#if !defined(USE_GETGRNAM_R) && defined(HAVE_ENDGRENT)
|
|
endgrent();
|
|
#endif
|
|
rb_raise(rb_eArgError, "can't find group for %"PRIsVALUE, id);
|
|
}
|
|
gid = grptr->gr_gid;
|
|
#if !defined(USE_GETGRNAM_R) && defined(HAVE_ENDGRENT)
|
|
endgrent();
|
|
#endif
|
|
}
|
|
return gid;
|
|
}
|
|
|
|
# ifdef p_gid_from_name
|
|
/*
|
|
* call-seq:
|
|
* Process::GID.from_name(name) -> gid
|
|
*
|
|
* Get the group ID by the _name_.
|
|
* If the group is not found, +ArgumentError+ will be raised.
|
|
*
|
|
* Process::GID.from_name("wheel") #=> 0
|
|
* Process::GID.from_name("nosuchgroup") #=> can't find group for nosuchgroup (ArgumentError)
|
|
*/
|
|
|
|
static VALUE
|
|
p_gid_from_name(VALUE self, VALUE id)
|
|
{
|
|
return GIDT2NUM(OBJ2GID(id));
|
|
}
|
|
# endif
|
|
#endif
|
|
|
|
#if defined HAVE_SETUID
|
|
/*
|
|
* call-seq:
|
|
* Process::Sys.setuid(user) -> nil
|
|
*
|
|
* Set the user ID of the current process to _user_. Not
|
|
* available on all platforms.
|
|
*
|
|
*/
|
|
|
|
static VALUE
|
|
p_sys_setuid(VALUE obj, VALUE id)
|
|
{
|
|
check_uid_switch();
|
|
if (setuid(OBJ2UID(id)) != 0) rb_sys_fail(0);
|
|
return Qnil;
|
|
}
|
|
#else
|
|
#define p_sys_setuid rb_f_notimplement
|
|
#endif
|
|
|
|
|
|
#if defined HAVE_SETRUID
|
|
/*
|
|
* call-seq:
|
|
* Process::Sys.setruid(user) -> nil
|
|
*
|
|
* Set the real user ID of the calling process to _user_.
|
|
* Not available on all platforms.
|
|
*
|
|
*/
|
|
|
|
static VALUE
|
|
p_sys_setruid(VALUE obj, VALUE id)
|
|
{
|
|
check_uid_switch();
|
|
if (setruid(OBJ2UID(id)) != 0) rb_sys_fail(0);
|
|
return Qnil;
|
|
}
|
|
#else
|
|
#define p_sys_setruid rb_f_notimplement
|
|
#endif
|
|
|
|
|
|
#if defined HAVE_SETEUID
|
|
/*
|
|
* call-seq:
|
|
* Process::Sys.seteuid(user) -> nil
|
|
*
|
|
* Set the effective user ID of the calling process to
|
|
* _user_. Not available on all platforms.
|
|
*
|
|
*/
|
|
|
|
static VALUE
|
|
p_sys_seteuid(VALUE obj, VALUE id)
|
|
{
|
|
check_uid_switch();
|
|
if (seteuid(OBJ2UID(id)) != 0) rb_sys_fail(0);
|
|
return Qnil;
|
|
}
|
|
#else
|
|
#define p_sys_seteuid rb_f_notimplement
|
|
#endif
|
|
|
|
|
|
#if defined HAVE_SETREUID
|
|
/*
|
|
* call-seq:
|
|
* Process::Sys.setreuid(rid, eid) -> nil
|
|
*
|
|
* Sets the (user) real and/or effective user IDs of the current
|
|
* process to _rid_ and _eid_, respectively. A value of
|
|
* <code>-1</code> for either means to leave that ID unchanged. Not
|
|
* available on all platforms.
|
|
*
|
|
*/
|
|
|
|
static VALUE
|
|
p_sys_setreuid(VALUE obj, VALUE rid, VALUE eid)
|
|
{
|
|
rb_uid_t ruid, euid;
|
|
PREPARE_GETPWNAM;
|
|
check_uid_switch();
|
|
ruid = OBJ2UID1(rid);
|
|
euid = OBJ2UID1(eid);
|
|
FINISH_GETPWNAM;
|
|
if (setreuid(ruid, euid) != 0) rb_sys_fail(0);
|
|
return Qnil;
|
|
}
|
|
#else
|
|
#define p_sys_setreuid rb_f_notimplement
|
|
#endif
|
|
|
|
|
|
#if defined HAVE_SETRESUID
|
|
/*
|
|
* call-seq:
|
|
* Process::Sys.setresuid(rid, eid, sid) -> nil
|
|
*
|
|
* Sets the (user) real, effective, and saved user IDs of the
|
|
* current process to _rid_, _eid_, and _sid_ respectively. A
|
|
* value of <code>-1</code> for any value means to
|
|
* leave that ID unchanged. Not available on all platforms.
|
|
*
|
|
*/
|
|
|
|
static VALUE
|
|
p_sys_setresuid(VALUE obj, VALUE rid, VALUE eid, VALUE sid)
|
|
{
|
|
rb_uid_t ruid, euid, suid;
|
|
PREPARE_GETPWNAM;
|
|
check_uid_switch();
|
|
ruid = OBJ2UID1(rid);
|
|
euid = OBJ2UID1(eid);
|
|
suid = OBJ2UID1(sid);
|
|
FINISH_GETPWNAM;
|
|
if (setresuid(ruid, euid, suid) != 0) rb_sys_fail(0);
|
|
return Qnil;
|
|
}
|
|
#else
|
|
#define p_sys_setresuid rb_f_notimplement
|
|
#endif
|
|
|
|
|
|
/*
|
|
* call-seq:
|
|
* Process.uid -> integer
|
|
* Process::UID.rid -> integer
|
|
* Process::Sys.getuid -> integer
|
|
*
|
|
* Returns the (real) user ID of the current process.
|
|
*
|
|
* Process.uid # => 1000
|
|
*
|
|
*/
|
|
|
|
static VALUE
|
|
proc_getuid(VALUE obj)
|
|
{
|
|
rb_uid_t uid = getuid();
|
|
return UIDT2NUM(uid);
|
|
}
|
|
|
|
|
|
#if defined(HAVE_SETRESUID) || defined(HAVE_SETREUID) || defined(HAVE_SETRUID) || defined(HAVE_SETUID)
|
|
/*
|
|
* call-seq:
|
|
* Process.uid = new_uid -> new_uid
|
|
*
|
|
* Sets the (user) user ID for the current process to +new_uid+:
|
|
*
|
|
* Process.uid = 1000 # => 1000
|
|
*
|
|
* Not available on all platforms.
|
|
*/
|
|
|
|
static VALUE
|
|
proc_setuid(VALUE obj, VALUE id)
|
|
{
|
|
rb_uid_t uid;
|
|
|
|
check_uid_switch();
|
|
|
|
uid = OBJ2UID(id);
|
|
#if defined(HAVE_SETRESUID)
|
|
if (setresuid(uid, -1, -1) < 0) rb_sys_fail(0);
|
|
#elif defined HAVE_SETREUID
|
|
if (setreuid(uid, -1) < 0) rb_sys_fail(0);
|
|
#elif defined HAVE_SETRUID
|
|
if (setruid(uid) < 0) rb_sys_fail(0);
|
|
#elif defined HAVE_SETUID
|
|
{
|
|
if (geteuid() == uid) {
|
|
if (setuid(uid) < 0) rb_sys_fail(0);
|
|
}
|
|
else {
|
|
rb_notimplement();
|
|
}
|
|
}
|
|
#endif
|
|
return id;
|
|
}
|
|
#else
|
|
#define proc_setuid rb_f_notimplement
|
|
#endif
|
|
|
|
|
|
/********************************************************************
|
|
*
|
|
* Document-class: Process::UID
|
|
*
|
|
* The Process::UID module contains a collection of
|
|
* module functions which can be used to portably get, set, and
|
|
* switch the current process's real, effective, and saved user IDs.
|
|
*
|
|
*/
|
|
|
|
static rb_uid_t SAVED_USER_ID = -1;
|
|
|
|
#ifdef BROKEN_SETREUID
|
|
int
|
|
setreuid(rb_uid_t ruid, rb_uid_t euid)
|
|
{
|
|
if (ruid != (rb_uid_t)-1 && ruid != getuid()) {
|
|
if (euid == (rb_uid_t)-1) euid = geteuid();
|
|
if (setuid(ruid) < 0) return -1;
|
|
}
|
|
if (euid != (rb_uid_t)-1 && euid != geteuid()) {
|
|
if (seteuid(euid) < 0) return -1;
|
|
}
|
|
return 0;
|
|
}
|
|
#endif
|
|
|
|
/*
|
|
* call-seq:
|
|
* Process::UID.change_privilege(user) -> integer
|
|
*
|
|
* Change the current process's real and effective user ID to that
|
|
* specified by _user_. Returns the new user ID. Not
|
|
* available on all platforms.
|
|
*
|
|
* [Process.uid, Process.euid] #=> [0, 0]
|
|
* Process::UID.change_privilege(31) #=> 31
|
|
* [Process.uid, Process.euid] #=> [31, 31]
|
|
*/
|
|
|
|
static VALUE
|
|
p_uid_change_privilege(VALUE obj, VALUE id)
|
|
{
|
|
rb_uid_t uid;
|
|
|
|
check_uid_switch();
|
|
|
|
uid = OBJ2UID(id);
|
|
|
|
if (geteuid() == 0) { /* root-user */
|
|
#if defined(HAVE_SETRESUID)
|
|
if (setresuid(uid, uid, uid) < 0) rb_sys_fail(0);
|
|
SAVED_USER_ID = uid;
|
|
#elif defined(HAVE_SETUID)
|
|
if (setuid(uid) < 0) rb_sys_fail(0);
|
|
SAVED_USER_ID = uid;
|
|
#elif defined(HAVE_SETREUID) && !defined(OBSOLETE_SETREUID)
|
|
if (getuid() == uid) {
|
|
if (SAVED_USER_ID == uid) {
|
|
if (setreuid(-1, uid) < 0) rb_sys_fail(0);
|
|
}
|
|
else {
|
|
if (uid == 0) { /* (r,e,s) == (root, root, x) */
|
|
if (setreuid(-1, SAVED_USER_ID) < 0) rb_sys_fail(0);
|
|
if (setreuid(SAVED_USER_ID, 0) < 0) rb_sys_fail(0);
|
|
SAVED_USER_ID = 0; /* (r,e,s) == (x, root, root) */
|
|
if (setreuid(uid, uid) < 0) rb_sys_fail(0);
|
|
SAVED_USER_ID = uid;
|
|
}
|
|
else {
|
|
if (setreuid(0, -1) < 0) rb_sys_fail(0);
|
|
SAVED_USER_ID = 0;
|
|
if (setreuid(uid, uid) < 0) rb_sys_fail(0);
|
|
SAVED_USER_ID = uid;
|
|
}
|
|
}
|
|
}
|
|
else {
|
|
if (setreuid(uid, uid) < 0) rb_sys_fail(0);
|
|
SAVED_USER_ID = uid;
|
|
}
|
|
#elif defined(HAVE_SETRUID) && defined(HAVE_SETEUID)
|
|
if (getuid() == uid) {
|
|
if (SAVED_USER_ID == uid) {
|
|
if (seteuid(uid) < 0) rb_sys_fail(0);
|
|
}
|
|
else {
|
|
if (uid == 0) {
|
|
if (setruid(SAVED_USER_ID) < 0) rb_sys_fail(0);
|
|
SAVED_USER_ID = 0;
|
|
if (setruid(0) < 0) rb_sys_fail(0);
|
|
}
|
|
else {
|
|
if (setruid(0) < 0) rb_sys_fail(0);
|
|
SAVED_USER_ID = 0;
|
|
if (seteuid(uid) < 0) rb_sys_fail(0);
|
|
if (setruid(uid) < 0) rb_sys_fail(0);
|
|
SAVED_USER_ID = uid;
|
|
}
|
|
}
|
|
}
|
|
else {
|
|
if (seteuid(uid) < 0) rb_sys_fail(0);
|
|
if (setruid(uid) < 0) rb_sys_fail(0);
|
|
SAVED_USER_ID = uid;
|
|
}
|
|
#else
|
|
(void)uid;
|
|
rb_notimplement();
|
|
#endif
|
|
}
|
|
else { /* unprivileged user */
|
|
#if defined(HAVE_SETRESUID)
|
|
if (setresuid((getuid() == uid)? (rb_uid_t)-1: uid,
|
|
(geteuid() == uid)? (rb_uid_t)-1: uid,
|
|
(SAVED_USER_ID == uid)? (rb_uid_t)-1: uid) < 0) rb_sys_fail(0);
|
|
SAVED_USER_ID = uid;
|
|
#elif defined(HAVE_SETREUID) && !defined(OBSOLETE_SETREUID)
|
|
if (SAVED_USER_ID == uid) {
|
|
if (setreuid((getuid() == uid)? (rb_uid_t)-1: uid,
|
|
(geteuid() == uid)? (rb_uid_t)-1: uid) < 0)
|
|
rb_sys_fail(0);
|
|
}
|
|
else if (getuid() != uid) {
|
|
if (setreuid(uid, (geteuid() == uid)? (rb_uid_t)-1: uid) < 0)
|
|
rb_sys_fail(0);
|
|
SAVED_USER_ID = uid;
|
|
}
|
|
else if (/* getuid() == uid && */ geteuid() != uid) {
|
|
if (setreuid(geteuid(), uid) < 0) rb_sys_fail(0);
|
|
SAVED_USER_ID = uid;
|
|
if (setreuid(uid, -1) < 0) rb_sys_fail(0);
|
|
}
|
|
else { /* getuid() == uid && geteuid() == uid */
|
|
if (setreuid(-1, SAVED_USER_ID) < 0) rb_sys_fail(0);
|
|
if (setreuid(SAVED_USER_ID, uid) < 0) rb_sys_fail(0);
|
|
SAVED_USER_ID = uid;
|
|
if (setreuid(uid, -1) < 0) rb_sys_fail(0);
|
|
}
|
|
#elif defined(HAVE_SETRUID) && defined(HAVE_SETEUID)
|
|
if (SAVED_USER_ID == uid) {
|
|
if (geteuid() != uid && seteuid(uid) < 0) rb_sys_fail(0);
|
|
if (getuid() != uid && setruid(uid) < 0) rb_sys_fail(0);
|
|
}
|
|
else if (/* SAVED_USER_ID != uid && */ geteuid() == uid) {
|
|
if (getuid() != uid) {
|
|
if (setruid(uid) < 0) rb_sys_fail(0);
|
|
SAVED_USER_ID = uid;
|
|
}
|
|
else {
|
|
if (setruid(SAVED_USER_ID) < 0) rb_sys_fail(0);
|
|
SAVED_USER_ID = uid;
|
|
if (setruid(uid) < 0) rb_sys_fail(0);
|
|
}
|
|
}
|
|
else if (/* geteuid() != uid && */ getuid() == uid) {
|
|
if (seteuid(uid) < 0) rb_sys_fail(0);
|
|
if (setruid(SAVED_USER_ID) < 0) rb_sys_fail(0);
|
|
SAVED_USER_ID = uid;
|
|
if (setruid(uid) < 0) rb_sys_fail(0);
|
|
}
|
|
else {
|
|
rb_syserr_fail(EPERM, 0);
|
|
}
|
|
#elif defined HAVE_44BSD_SETUID
|
|
if (getuid() == uid) {
|
|
/* (r,e,s)==(uid,?,?) ==> (uid,uid,uid) */
|
|
if (setuid(uid) < 0) rb_sys_fail(0);
|
|
SAVED_USER_ID = uid;
|
|
}
|
|
else {
|
|
rb_syserr_fail(EPERM, 0);
|
|
}
|
|
#elif defined HAVE_SETEUID
|
|
if (getuid() == uid && SAVED_USER_ID == uid) {
|
|
if (seteuid(uid) < 0) rb_sys_fail(0);
|
|
}
|
|
else {
|
|
rb_syserr_fail(EPERM, 0);
|
|
}
|
|
#elif defined HAVE_SETUID
|
|
if (getuid() == uid && SAVED_USER_ID == uid) {
|
|
if (setuid(uid) < 0) rb_sys_fail(0);
|
|
}
|
|
else {
|
|
rb_syserr_fail(EPERM, 0);
|
|
}
|
|
#else
|
|
rb_notimplement();
|
|
#endif
|
|
}
|
|
return id;
|
|
}
|
|
|
|
|
|
|
|
#if defined HAVE_SETGID
|
|
/*
|
|
* call-seq:
|
|
* Process::Sys.setgid(group) -> nil
|
|
*
|
|
* Set the group ID of the current process to _group_. Not
|
|
* available on all platforms.
|
|
*
|
|
*/
|
|
|
|
static VALUE
|
|
p_sys_setgid(VALUE obj, VALUE id)
|
|
{
|
|
check_gid_switch();
|
|
if (setgid(OBJ2GID(id)) != 0) rb_sys_fail(0);
|
|
return Qnil;
|
|
}
|
|
#else
|
|
#define p_sys_setgid rb_f_notimplement
|
|
#endif
|
|
|
|
|
|
#if defined HAVE_SETRGID
|
|
/*
|
|
* call-seq:
|
|
* Process::Sys.setrgid(group) -> nil
|
|
*
|
|
* Set the real group ID of the calling process to _group_.
|
|
* Not available on all platforms.
|
|
*
|
|
*/
|
|
|
|
static VALUE
|
|
p_sys_setrgid(VALUE obj, VALUE id)
|
|
{
|
|
check_gid_switch();
|
|
if (setrgid(OBJ2GID(id)) != 0) rb_sys_fail(0);
|
|
return Qnil;
|
|
}
|
|
#else
|
|
#define p_sys_setrgid rb_f_notimplement
|
|
#endif
|
|
|
|
|
|
#if defined HAVE_SETEGID
|
|
/*
|
|
* call-seq:
|
|
* Process::Sys.setegid(group) -> nil
|
|
*
|
|
* Set the effective group ID of the calling process to
|
|
* _group_. Not available on all platforms.
|
|
*
|
|
*/
|
|
|
|
static VALUE
|
|
p_sys_setegid(VALUE obj, VALUE id)
|
|
{
|
|
check_gid_switch();
|
|
if (setegid(OBJ2GID(id)) != 0) rb_sys_fail(0);
|
|
return Qnil;
|
|
}
|
|
#else
|
|
#define p_sys_setegid rb_f_notimplement
|
|
#endif
|
|
|
|
|
|
#if defined HAVE_SETREGID
|
|
/*
|
|
* call-seq:
|
|
* Process::Sys.setregid(rid, eid) -> nil
|
|
*
|
|
* Sets the (group) real and/or effective group IDs of the current
|
|
* process to <em>rid</em> and <em>eid</em>, respectively. A value of
|
|
* <code>-1</code> for either means to leave that ID unchanged. Not
|
|
* available on all platforms.
|
|
*
|
|
*/
|
|
|
|
static VALUE
|
|
p_sys_setregid(VALUE obj, VALUE rid, VALUE eid)
|
|
{
|
|
rb_gid_t rgid, egid;
|
|
check_gid_switch();
|
|
rgid = OBJ2GID(rid);
|
|
egid = OBJ2GID(eid);
|
|
if (setregid(rgid, egid) != 0) rb_sys_fail(0);
|
|
return Qnil;
|
|
}
|
|
#else
|
|
#define p_sys_setregid rb_f_notimplement
|
|
#endif
|
|
|
|
#if defined HAVE_SETRESGID
|
|
/*
|
|
* call-seq:
|
|
* Process::Sys.setresgid(rid, eid, sid) -> nil
|
|
*
|
|
* Sets the (group) real, effective, and saved user IDs of the
|
|
* current process to <em>rid</em>, <em>eid</em>, and <em>sid</em>
|
|
* respectively. A value of <code>-1</code> for any value means to
|
|
* leave that ID unchanged. Not available on all platforms.
|
|
*
|
|
*/
|
|
|
|
static VALUE
|
|
p_sys_setresgid(VALUE obj, VALUE rid, VALUE eid, VALUE sid)
|
|
{
|
|
rb_gid_t rgid, egid, sgid;
|
|
check_gid_switch();
|
|
rgid = OBJ2GID(rid);
|
|
egid = OBJ2GID(eid);
|
|
sgid = OBJ2GID(sid);
|
|
if (setresgid(rgid, egid, sgid) != 0) rb_sys_fail(0);
|
|
return Qnil;
|
|
}
|
|
#else
|
|
#define p_sys_setresgid rb_f_notimplement
|
|
#endif
|
|
|
|
|
|
#if defined HAVE_ISSETUGID
|
|
/*
|
|
* call-seq:
|
|
* Process::Sys.issetugid -> true or false
|
|
*
|
|
* Returns +true+ if the process was created as a result
|
|
* of an execve(2) system call which had either of the setuid or
|
|
* setgid bits set (and extra privileges were given as a result) or
|
|
* if it has changed any of its real, effective or saved user or
|
|
* group IDs since it began execution.
|
|
*
|
|
*/
|
|
|
|
static VALUE
|
|
p_sys_issetugid(VALUE obj)
|
|
{
|
|
return RBOOL(issetugid());
|
|
}
|
|
#else
|
|
#define p_sys_issetugid rb_f_notimplement
|
|
#endif
|
|
|
|
|
|
/*
|
|
* call-seq:
|
|
* Process.gid -> integer
|
|
* Process::GID.rid -> integer
|
|
* Process::Sys.getgid -> integer
|
|
*
|
|
* Returns the (real) group ID for the current process:
|
|
*
|
|
* Process.gid # => 1000
|
|
*
|
|
*/
|
|
|
|
static VALUE
|
|
proc_getgid(VALUE obj)
|
|
{
|
|
rb_gid_t gid = getgid();
|
|
return GIDT2NUM(gid);
|
|
}
|
|
|
|
|
|
#if defined(HAVE_SETRESGID) || defined(HAVE_SETREGID) || defined(HAVE_SETRGID) || defined(HAVE_SETGID)
|
|
/*
|
|
* call-seq:
|
|
* Process.gid = new_gid -> new_gid
|
|
*
|
|
* Sets the group ID for the current process to +new_gid+:
|
|
*
|
|
* Process.gid = 1000 # => 1000
|
|
*
|
|
*/
|
|
|
|
static VALUE
|
|
proc_setgid(VALUE obj, VALUE id)
|
|
{
|
|
rb_gid_t gid;
|
|
|
|
check_gid_switch();
|
|
|
|
gid = OBJ2GID(id);
|
|
#if defined(HAVE_SETRESGID)
|
|
if (setresgid(gid, -1, -1) < 0) rb_sys_fail(0);
|
|
#elif defined HAVE_SETREGID
|
|
if (setregid(gid, -1) < 0) rb_sys_fail(0);
|
|
#elif defined HAVE_SETRGID
|
|
if (setrgid(gid) < 0) rb_sys_fail(0);
|
|
#elif defined HAVE_SETGID
|
|
{
|
|
if (getegid() == gid) {
|
|
if (setgid(gid) < 0) rb_sys_fail(0);
|
|
}
|
|
else {
|
|
rb_notimplement();
|
|
}
|
|
}
|
|
#endif
|
|
return GIDT2NUM(gid);
|
|
}
|
|
#else
|
|
#define proc_setgid rb_f_notimplement
|
|
#endif
|
|
|
|
|
|
#if defined(_SC_NGROUPS_MAX) || defined(NGROUPS_MAX)
|
|
/*
|
|
* Maximum supplementary groups are platform dependent.
|
|
* FWIW, 65536 is enough big for our supported OSs.
|
|
*
|
|
* OS Name max groups
|
|
* -----------------------------------------------
|
|
* Linux Kernel >= 2.6.3 65536
|
|
* Linux Kernel < 2.6.3 32
|
|
* IBM AIX 5.2 64
|
|
* IBM AIX 5.3 ... 6.1 128
|
|
* IBM AIX 7.1 128 (can be configured to be up to 2048)
|
|
* OpenBSD, NetBSD 16
|
|
* FreeBSD < 8.0 16
|
|
* FreeBSD >=8.0 1023
|
|
* Darwin (Mac OS X) 16
|
|
* Sun Solaris 7,8,9,10 16
|
|
* Sun Solaris 11 / OpenSolaris 1024
|
|
* Windows 1015
|
|
*/
|
|
static int _maxgroups = -1;
|
|
static int
|
|
get_sc_ngroups_max(void)
|
|
{
|
|
#ifdef _SC_NGROUPS_MAX
|
|
return (int)sysconf(_SC_NGROUPS_MAX);
|
|
#elif defined(NGROUPS_MAX)
|
|
return (int)NGROUPS_MAX;
|
|
#else
|
|
return -1;
|
|
#endif
|
|
}
|
|
static int
|
|
maxgroups(void)
|
|
{
|
|
if (_maxgroups < 0) {
|
|
_maxgroups = get_sc_ngroups_max();
|
|
if (_maxgroups < 0)
|
|
_maxgroups = RB_MAX_GROUPS;
|
|
}
|
|
|
|
return _maxgroups;
|
|
}
|
|
#endif
|
|
|
|
|
|
|
|
#ifdef HAVE_GETGROUPS
|
|
/*
|
|
* call-seq:
|
|
* Process.groups -> array
|
|
*
|
|
* Returns an array of the group IDs
|
|
* in the supplemental group access list for the current process:
|
|
*
|
|
* Process.groups # => [4, 24, 27, 30, 46, 122, 135, 136, 1000]
|
|
*
|
|
* These properties of the returned array are system-dependent:
|
|
*
|
|
* - Whether (and how) the array is sorted.
|
|
* - Whether the array includes effective group IDs.
|
|
* - Whether the array includes duplicate group IDs.
|
|
* - Whether the array size exceeds the value of Process.maxgroups.
|
|
*
|
|
* Use this call to get a sorted and unique array:
|
|
*
|
|
* Process.groups.uniq.sort
|
|
*
|
|
*/
|
|
|
|
static VALUE
|
|
proc_getgroups(VALUE obj)
|
|
{
|
|
VALUE ary, tmp;
|
|
int i, ngroups;
|
|
rb_gid_t *groups;
|
|
|
|
ngroups = getgroups(0, NULL);
|
|
if (ngroups == -1)
|
|
rb_sys_fail(0);
|
|
|
|
groups = ALLOCV_N(rb_gid_t, tmp, ngroups);
|
|
|
|
ngroups = getgroups(ngroups, groups);
|
|
if (ngroups == -1)
|
|
rb_sys_fail(0);
|
|
|
|
ary = rb_ary_new();
|
|
for (i = 0; i < ngroups; i++)
|
|
rb_ary_push(ary, GIDT2NUM(groups[i]));
|
|
|
|
ALLOCV_END(tmp);
|
|
|
|
return ary;
|
|
}
|
|
#else
|
|
#define proc_getgroups rb_f_notimplement
|
|
#endif
|
|
|
|
|
|
#ifdef HAVE_SETGROUPS
|
|
/*
|
|
* call-seq:
|
|
* Process.groups = new_groups -> new_groups
|
|
*
|
|
* Sets the supplemental group access list to the given
|
|
* array of group IDs.
|
|
*
|
|
* Process.groups # => [0, 1, 2, 3, 4, 6, 10, 11, 20, 26, 27]
|
|
* Process.groups = [27, 6, 10, 11] # => [27, 6, 10, 11]
|
|
* Process.groups # => [27, 6, 10, 11]
|
|
*
|
|
*/
|
|
|
|
static VALUE
|
|
proc_setgroups(VALUE obj, VALUE ary)
|
|
{
|
|
int ngroups, i;
|
|
rb_gid_t *groups;
|
|
VALUE tmp;
|
|
PREPARE_GETGRNAM;
|
|
|
|
Check_Type(ary, T_ARRAY);
|
|
|
|
ngroups = RARRAY_LENINT(ary);
|
|
if (ngroups > maxgroups())
|
|
rb_raise(rb_eArgError, "too many groups, %d max", maxgroups());
|
|
|
|
groups = ALLOCV_N(rb_gid_t, tmp, ngroups);
|
|
|
|
for (i = 0; i < ngroups; i++) {
|
|
VALUE g = RARRAY_AREF(ary, i);
|
|
|
|
groups[i] = OBJ2GID1(g);
|
|
}
|
|
FINISH_GETGRNAM;
|
|
|
|
if (setgroups(ngroups, groups) == -1) /* ngroups <= maxgroups */
|
|
rb_sys_fail(0);
|
|
|
|
ALLOCV_END(tmp);
|
|
|
|
return proc_getgroups(obj);
|
|
}
|
|
#else
|
|
#define proc_setgroups rb_f_notimplement
|
|
#endif
|
|
|
|
|
|
#ifdef HAVE_INITGROUPS
|
|
/*
|
|
* call-seq:
|
|
* Process.initgroups(username, gid) -> array
|
|
*
|
|
* Sets the supplemental group access list;
|
|
* the new list includes:
|
|
*
|
|
* - The group IDs of those groups to which the user given by +username+ belongs.
|
|
* - The group ID +gid+.
|
|
*
|
|
* Example:
|
|
*
|
|
* Process.groups # => [0, 1, 2, 3, 4, 6, 10, 11, 20, 26, 27]
|
|
* Process.initgroups('me', 30) # => [30, 6, 10, 11]
|
|
* Process.groups # => [30, 6, 10, 11]
|
|
*
|
|
* Not available on all platforms.
|
|
*/
|
|
|
|
static VALUE
|
|
proc_initgroups(VALUE obj, VALUE uname, VALUE base_grp)
|
|
{
|
|
if (initgroups(StringValueCStr(uname), OBJ2GID(base_grp)) != 0) {
|
|
rb_sys_fail(0);
|
|
}
|
|
return proc_getgroups(obj);
|
|
}
|
|
#else
|
|
#define proc_initgroups rb_f_notimplement
|
|
#endif
|
|
|
|
#if defined(_SC_NGROUPS_MAX) || defined(NGROUPS_MAX)
|
|
/*
|
|
* call-seq:
|
|
* Process.maxgroups -> integer
|
|
*
|
|
* Returns the maximum number of group IDs allowed
|
|
* in the supplemental group access list:
|
|
*
|
|
* Process.maxgroups # => 32
|
|
*
|
|
*/
|
|
|
|
static VALUE
|
|
proc_getmaxgroups(VALUE obj)
|
|
{
|
|
return INT2FIX(maxgroups());
|
|
}
|
|
#else
|
|
#define proc_getmaxgroups rb_f_notimplement
|
|
#endif
|
|
|
|
#ifdef HAVE_SETGROUPS
|
|
/*
|
|
* call-seq:
|
|
* Process.maxgroups = new_max -> new_max
|
|
*
|
|
* Sets the maximum number of group IDs allowed
|
|
* in the supplemental group access list.
|
|
*/
|
|
|
|
static VALUE
|
|
proc_setmaxgroups(VALUE obj, VALUE val)
|
|
{
|
|
int ngroups = FIX2INT(val);
|
|
int ngroups_max = get_sc_ngroups_max();
|
|
|
|
if (ngroups <= 0)
|
|
rb_raise(rb_eArgError, "maxgroups %d should be positive", ngroups);
|
|
|
|
if (ngroups > RB_MAX_GROUPS)
|
|
ngroups = RB_MAX_GROUPS;
|
|
|
|
if (ngroups_max > 0 && ngroups > ngroups_max)
|
|
ngroups = ngroups_max;
|
|
|
|
_maxgroups = ngroups;
|
|
|
|
return INT2FIX(_maxgroups);
|
|
}
|
|
#else
|
|
#define proc_setmaxgroups rb_f_notimplement
|
|
#endif
|
|
|
|
#if defined(HAVE_DAEMON) || (defined(HAVE_WORKING_FORK) && defined(HAVE_SETSID))
|
|
static int rb_daemon(int nochdir, int noclose);
|
|
|
|
/*
|
|
* call-seq:
|
|
* Process.daemon(nochdir = nil, noclose = nil) -> 0
|
|
*
|
|
* Detaches the current process from its controlling terminal
|
|
* and runs it in the background as system daemon;
|
|
* returns zero.
|
|
*
|
|
* By default:
|
|
*
|
|
* - Changes the current working directory to the root directory.
|
|
* - Redirects $stdin, $stdout, and $stderr to the null device.
|
|
*
|
|
* If optional argument +nochdir+ is +true+,
|
|
* does not change the current working directory.
|
|
*
|
|
* If optional argument +noclose+ is +true+,
|
|
* does not redirect $stdin, $stdout, or $stderr.
|
|
*/
|
|
|
|
static VALUE
|
|
proc_daemon(int argc, VALUE *argv, VALUE _)
|
|
{
|
|
int n, nochdir = FALSE, noclose = FALSE;
|
|
|
|
switch (rb_check_arity(argc, 0, 2)) {
|
|
case 2: noclose = TO_BOOL(argv[1], "noclose");
|
|
case 1: nochdir = TO_BOOL(argv[0], "nochdir");
|
|
}
|
|
|
|
prefork();
|
|
n = rb_daemon(nochdir, noclose);
|
|
if (n < 0) rb_sys_fail("daemon");
|
|
return INT2FIX(n);
|
|
}
|
|
|
|
extern const char ruby_null_device[];
|
|
|
|
static int
|
|
rb_daemon(int nochdir, int noclose)
|
|
{
|
|
int err = 0;
|
|
#ifdef HAVE_DAEMON
|
|
before_fork_ruby();
|
|
err = daemon(nochdir, noclose);
|
|
after_fork_ruby(0);
|
|
#else
|
|
int n;
|
|
|
|
switch (rb_fork_ruby(NULL)) {
|
|
case -1: return -1;
|
|
case 0: break;
|
|
default: _exit(EXIT_SUCCESS);
|
|
}
|
|
|
|
/* ignore EPERM which means already being process-leader */
|
|
if (setsid() < 0) (void)0;
|
|
|
|
if (!nochdir)
|
|
err = chdir("/");
|
|
|
|
if (!noclose && (n = rb_cloexec_open(ruby_null_device, O_RDWR, 0)) != -1) {
|
|
rb_update_max_fd(n);
|
|
(void)dup2(n, 0);
|
|
(void)dup2(n, 1);
|
|
(void)dup2(n, 2);
|
|
if (n > 2)
|
|
(void)close (n);
|
|
}
|
|
#endif
|
|
return err;
|
|
}
|
|
#else
|
|
#define proc_daemon rb_f_notimplement
|
|
#endif
|
|
|
|
/********************************************************************
|
|
*
|
|
* Document-class: Process::GID
|
|
*
|
|
* The Process::GID module contains a collection of
|
|
* module functions which can be used to portably get, set, and
|
|
* switch the current process's real, effective, and saved group IDs.
|
|
*
|
|
*/
|
|
|
|
static rb_gid_t SAVED_GROUP_ID = -1;
|
|
|
|
#ifdef BROKEN_SETREGID
|
|
int
|
|
setregid(rb_gid_t rgid, rb_gid_t egid)
|
|
{
|
|
if (rgid != (rb_gid_t)-1 && rgid != getgid()) {
|
|
if (egid == (rb_gid_t)-1) egid = getegid();
|
|
if (setgid(rgid) < 0) return -1;
|
|
}
|
|
if (egid != (rb_gid_t)-1 && egid != getegid()) {
|
|
if (setegid(egid) < 0) return -1;
|
|
}
|
|
return 0;
|
|
}
|
|
#endif
|
|
|
|
/*
|
|
* call-seq:
|
|
* Process::GID.change_privilege(group) -> integer
|
|
*
|
|
* Change the current process's real and effective group ID to that
|
|
* specified by _group_. Returns the new group ID. Not
|
|
* available on all platforms.
|
|
*
|
|
* [Process.gid, Process.egid] #=> [0, 0]
|
|
* Process::GID.change_privilege(33) #=> 33
|
|
* [Process.gid, Process.egid] #=> [33, 33]
|
|
*/
|
|
|
|
static VALUE
|
|
p_gid_change_privilege(VALUE obj, VALUE id)
|
|
{
|
|
rb_gid_t gid;
|
|
|
|
check_gid_switch();
|
|
|
|
gid = OBJ2GID(id);
|
|
|
|
if (geteuid() == 0) { /* root-user */
|
|
#if defined(HAVE_SETRESGID)
|
|
if (setresgid(gid, gid, gid) < 0) rb_sys_fail(0);
|
|
SAVED_GROUP_ID = gid;
|
|
#elif defined HAVE_SETGID
|
|
if (setgid(gid) < 0) rb_sys_fail(0);
|
|
SAVED_GROUP_ID = gid;
|
|
#elif defined(HAVE_SETREGID) && !defined(OBSOLETE_SETREGID)
|
|
if (getgid() == gid) {
|
|
if (SAVED_GROUP_ID == gid) {
|
|
if (setregid(-1, gid) < 0) rb_sys_fail(0);
|
|
}
|
|
else {
|
|
if (gid == 0) { /* (r,e,s) == (root, y, x) */
|
|
if (setregid(-1, SAVED_GROUP_ID) < 0) rb_sys_fail(0);
|
|
if (setregid(SAVED_GROUP_ID, 0) < 0) rb_sys_fail(0);
|
|
SAVED_GROUP_ID = 0; /* (r,e,s) == (x, root, root) */
|
|
if (setregid(gid, gid) < 0) rb_sys_fail(0);
|
|
SAVED_GROUP_ID = gid;
|
|
}
|
|
else { /* (r,e,s) == (z, y, x) */
|
|
if (setregid(0, 0) < 0) rb_sys_fail(0);
|
|
SAVED_GROUP_ID = 0;
|
|
if (setregid(gid, gid) < 0) rb_sys_fail(0);
|
|
SAVED_GROUP_ID = gid;
|
|
}
|
|
}
|
|
}
|
|
else {
|
|
if (setregid(gid, gid) < 0) rb_sys_fail(0);
|
|
SAVED_GROUP_ID = gid;
|
|
}
|
|
#elif defined(HAVE_SETRGID) && defined (HAVE_SETEGID)
|
|
if (getgid() == gid) {
|
|
if (SAVED_GROUP_ID == gid) {
|
|
if (setegid(gid) < 0) rb_sys_fail(0);
|
|
}
|
|
else {
|
|
if (gid == 0) {
|
|
if (setegid(gid) < 0) rb_sys_fail(0);
|
|
if (setrgid(SAVED_GROUP_ID) < 0) rb_sys_fail(0);
|
|
SAVED_GROUP_ID = 0;
|
|
if (setrgid(0) < 0) rb_sys_fail(0);
|
|
}
|
|
else {
|
|
if (setrgid(0) < 0) rb_sys_fail(0);
|
|
SAVED_GROUP_ID = 0;
|
|
if (setegid(gid) < 0) rb_sys_fail(0);
|
|
if (setrgid(gid) < 0) rb_sys_fail(0);
|
|
SAVED_GROUP_ID = gid;
|
|
}
|
|
}
|
|
}
|
|
else {
|
|
if (setegid(gid) < 0) rb_sys_fail(0);
|
|
if (setrgid(gid) < 0) rb_sys_fail(0);
|
|
SAVED_GROUP_ID = gid;
|
|
}
|
|
#else
|
|
rb_notimplement();
|
|
#endif
|
|
}
|
|
else { /* unprivileged user */
|
|
#if defined(HAVE_SETRESGID)
|
|
if (setresgid((getgid() == gid)? (rb_gid_t)-1: gid,
|
|
(getegid() == gid)? (rb_gid_t)-1: gid,
|
|
(SAVED_GROUP_ID == gid)? (rb_gid_t)-1: gid) < 0) rb_sys_fail(0);
|
|
SAVED_GROUP_ID = gid;
|
|
#elif defined(HAVE_SETREGID) && !defined(OBSOLETE_SETREGID)
|
|
if (SAVED_GROUP_ID == gid) {
|
|
if (setregid((getgid() == gid)? (rb_uid_t)-1: gid,
|
|
(getegid() == gid)? (rb_uid_t)-1: gid) < 0)
|
|
rb_sys_fail(0);
|
|
}
|
|
else if (getgid() != gid) {
|
|
if (setregid(gid, (getegid() == gid)? (rb_uid_t)-1: gid) < 0)
|
|
rb_sys_fail(0);
|
|
SAVED_GROUP_ID = gid;
|
|
}
|
|
else if (/* getgid() == gid && */ getegid() != gid) {
|
|
if (setregid(getegid(), gid) < 0) rb_sys_fail(0);
|
|
SAVED_GROUP_ID = gid;
|
|
if (setregid(gid, -1) < 0) rb_sys_fail(0);
|
|
}
|
|
else { /* getgid() == gid && getegid() == gid */
|
|
if (setregid(-1, SAVED_GROUP_ID) < 0) rb_sys_fail(0);
|
|
if (setregid(SAVED_GROUP_ID, gid) < 0) rb_sys_fail(0);
|
|
SAVED_GROUP_ID = gid;
|
|
if (setregid(gid, -1) < 0) rb_sys_fail(0);
|
|
}
|
|
#elif defined(HAVE_SETRGID) && defined(HAVE_SETEGID)
|
|
if (SAVED_GROUP_ID == gid) {
|
|
if (getegid() != gid && setegid(gid) < 0) rb_sys_fail(0);
|
|
if (getgid() != gid && setrgid(gid) < 0) rb_sys_fail(0);
|
|
}
|
|
else if (/* SAVED_GROUP_ID != gid && */ getegid() == gid) {
|
|
if (getgid() != gid) {
|
|
if (setrgid(gid) < 0) rb_sys_fail(0);
|
|
SAVED_GROUP_ID = gid;
|
|
}
|
|
else {
|
|
if (setrgid(SAVED_GROUP_ID) < 0) rb_sys_fail(0);
|
|
SAVED_GROUP_ID = gid;
|
|
if (setrgid(gid) < 0) rb_sys_fail(0);
|
|
}
|
|
}
|
|
else if (/* getegid() != gid && */ getgid() == gid) {
|
|
if (setegid(gid) < 0) rb_sys_fail(0);
|
|
if (setrgid(SAVED_GROUP_ID) < 0) rb_sys_fail(0);
|
|
SAVED_GROUP_ID = gid;
|
|
if (setrgid(gid) < 0) rb_sys_fail(0);
|
|
}
|
|
else {
|
|
rb_syserr_fail(EPERM, 0);
|
|
}
|
|
#elif defined HAVE_44BSD_SETGID
|
|
if (getgid() == gid) {
|
|
/* (r,e,s)==(gid,?,?) ==> (gid,gid,gid) */
|
|
if (setgid(gid) < 0) rb_sys_fail(0);
|
|
SAVED_GROUP_ID = gid;
|
|
}
|
|
else {
|
|
rb_syserr_fail(EPERM, 0);
|
|
}
|
|
#elif defined HAVE_SETEGID
|
|
if (getgid() == gid && SAVED_GROUP_ID == gid) {
|
|
if (setegid(gid) < 0) rb_sys_fail(0);
|
|
}
|
|
else {
|
|
rb_syserr_fail(EPERM, 0);
|
|
}
|
|
#elif defined HAVE_SETGID
|
|
if (getgid() == gid && SAVED_GROUP_ID == gid) {
|
|
if (setgid(gid) < 0) rb_sys_fail(0);
|
|
}
|
|
else {
|
|
rb_syserr_fail(EPERM, 0);
|
|
}
|
|
#else
|
|
(void)gid;
|
|
rb_notimplement();
|
|
#endif
|
|
}
|
|
return id;
|
|
}
|
|
|
|
|
|
/*
|
|
* call-seq:
|
|
* Process.euid -> integer
|
|
* Process::UID.eid -> integer
|
|
* Process::Sys.geteuid -> integer
|
|
*
|
|
* Returns the effective user ID for the current process.
|
|
*
|
|
* Process.euid # => 501
|
|
*
|
|
*/
|
|
|
|
static VALUE
|
|
proc_geteuid(VALUE obj)
|
|
{
|
|
rb_uid_t euid = geteuid();
|
|
return UIDT2NUM(euid);
|
|
}
|
|
|
|
#if defined(HAVE_SETRESUID) || defined(HAVE_SETREUID) || defined(HAVE_SETEUID) || defined(HAVE_SETUID) || defined(_POSIX_SAVED_IDS)
|
|
static void
|
|
proc_seteuid(rb_uid_t uid)
|
|
{
|
|
#if defined(HAVE_SETRESUID)
|
|
if (setresuid(-1, uid, -1) < 0) rb_sys_fail(0);
|
|
#elif defined HAVE_SETREUID
|
|
if (setreuid(-1, uid) < 0) rb_sys_fail(0);
|
|
#elif defined HAVE_SETEUID
|
|
if (seteuid(uid) < 0) rb_sys_fail(0);
|
|
#elif defined HAVE_SETUID
|
|
if (uid == getuid()) {
|
|
if (setuid(uid) < 0) rb_sys_fail(0);
|
|
}
|
|
else {
|
|
rb_notimplement();
|
|
}
|
|
#else
|
|
rb_notimplement();
|
|
#endif
|
|
}
|
|
#endif
|
|
|
|
#if defined(HAVE_SETRESUID) || defined(HAVE_SETREUID) || defined(HAVE_SETEUID) || defined(HAVE_SETUID)
|
|
/*
|
|
* call-seq:
|
|
* Process.euid = new_euid -> new_euid
|
|
*
|
|
* Sets the effective user ID for the current process.
|
|
*
|
|
* Not available on all platforms.
|
|
*/
|
|
|
|
static VALUE
|
|
proc_seteuid_m(VALUE mod, VALUE euid)
|
|
{
|
|
check_uid_switch();
|
|
proc_seteuid(OBJ2UID(euid));
|
|
return euid;
|
|
}
|
|
#else
|
|
#define proc_seteuid_m rb_f_notimplement
|
|
#endif
|
|
|
|
static rb_uid_t
|
|
rb_seteuid_core(rb_uid_t euid)
|
|
{
|
|
#if defined(HAVE_SETRESUID) || (defined(HAVE_SETREUID) && !defined(OBSOLETE_SETREUID))
|
|
rb_uid_t uid;
|
|
#endif
|
|
|
|
check_uid_switch();
|
|
|
|
#if defined(HAVE_SETRESUID) || (defined(HAVE_SETREUID) && !defined(OBSOLETE_SETREUID))
|
|
uid = getuid();
|
|
#endif
|
|
|
|
#if defined(HAVE_SETRESUID)
|
|
if (uid != euid) {
|
|
if (setresuid(-1,euid,euid) < 0) rb_sys_fail(0);
|
|
SAVED_USER_ID = euid;
|
|
}
|
|
else {
|
|
if (setresuid(-1,euid,-1) < 0) rb_sys_fail(0);
|
|
}
|
|
#elif defined(HAVE_SETREUID) && !defined(OBSOLETE_SETREUID)
|
|
if (setreuid(-1, euid) < 0) rb_sys_fail(0);
|
|
if (uid != euid) {
|
|
if (setreuid(euid,uid) < 0) rb_sys_fail(0);
|
|
if (setreuid(uid,euid) < 0) rb_sys_fail(0);
|
|
SAVED_USER_ID = euid;
|
|
}
|
|
#elif defined HAVE_SETEUID
|
|
if (seteuid(euid) < 0) rb_sys_fail(0);
|
|
#elif defined HAVE_SETUID
|
|
if (geteuid() == 0) rb_sys_fail(0);
|
|
if (setuid(euid) < 0) rb_sys_fail(0);
|
|
#else
|
|
rb_notimplement();
|
|
#endif
|
|
return euid;
|
|
}
|
|
|
|
|
|
/*
|
|
* call-seq:
|
|
* Process::UID.grant_privilege(user) -> integer
|
|
* Process::UID.eid= user -> integer
|
|
*
|
|
* Set the effective user ID, and if possible, the saved user ID of
|
|
* the process to the given _user_. Returns the new
|
|
* effective user ID. Not available on all platforms.
|
|
*
|
|
* [Process.uid, Process.euid] #=> [0, 0]
|
|
* Process::UID.grant_privilege(31) #=> 31
|
|
* [Process.uid, Process.euid] #=> [0, 31]
|
|
*/
|
|
|
|
static VALUE
|
|
p_uid_grant_privilege(VALUE obj, VALUE id)
|
|
{
|
|
rb_seteuid_core(OBJ2UID(id));
|
|
return id;
|
|
}
|
|
|
|
|
|
/*
|
|
* call-seq:
|
|
* Process.egid -> integer
|
|
* Process::GID.eid -> integer
|
|
* Process::Sys.geteid -> integer
|
|
*
|
|
* Returns the effective group ID for the current process:
|
|
*
|
|
* Process.egid # => 500
|
|
*
|
|
* Not available on all platforms.
|
|
*/
|
|
|
|
static VALUE
|
|
proc_getegid(VALUE obj)
|
|
{
|
|
rb_gid_t egid = getegid();
|
|
|
|
return GIDT2NUM(egid);
|
|
}
|
|
|
|
#if defined(HAVE_SETRESGID) || defined(HAVE_SETREGID) || defined(HAVE_SETEGID) || defined(HAVE_SETGID) || defined(_POSIX_SAVED_IDS)
|
|
/*
|
|
* call-seq:
|
|
* Process.egid = new_egid -> new_egid
|
|
*
|
|
* Sets the effective group ID for the current process.
|
|
*
|
|
* Not available on all platforms.
|
|
*/
|
|
|
|
static VALUE
|
|
proc_setegid(VALUE obj, VALUE egid)
|
|
{
|
|
#if defined(HAVE_SETRESGID) || defined(HAVE_SETREGID) || defined(HAVE_SETEGID) || defined(HAVE_SETGID)
|
|
rb_gid_t gid;
|
|
#endif
|
|
|
|
check_gid_switch();
|
|
|
|
#if defined(HAVE_SETRESGID) || defined(HAVE_SETREGID) || defined(HAVE_SETEGID) || defined(HAVE_SETGID)
|
|
gid = OBJ2GID(egid);
|
|
#endif
|
|
|
|
#if defined(HAVE_SETRESGID)
|
|
if (setresgid(-1, gid, -1) < 0) rb_sys_fail(0);
|
|
#elif defined HAVE_SETREGID
|
|
if (setregid(-1, gid) < 0) rb_sys_fail(0);
|
|
#elif defined HAVE_SETEGID
|
|
if (setegid(gid) < 0) rb_sys_fail(0);
|
|
#elif defined HAVE_SETGID
|
|
if (gid == getgid()) {
|
|
if (setgid(gid) < 0) rb_sys_fail(0);
|
|
}
|
|
else {
|
|
rb_notimplement();
|
|
}
|
|
#else
|
|
rb_notimplement();
|
|
#endif
|
|
return egid;
|
|
}
|
|
#endif
|
|
|
|
#if defined(HAVE_SETRESGID) || defined(HAVE_SETREGID) || defined(HAVE_SETEGID) || defined(HAVE_SETGID)
|
|
#define proc_setegid_m proc_setegid
|
|
#else
|
|
#define proc_setegid_m rb_f_notimplement
|
|
#endif
|
|
|
|
static rb_gid_t
|
|
rb_setegid_core(rb_gid_t egid)
|
|
{
|
|
#if defined(HAVE_SETRESGID) || (defined(HAVE_SETREGID) && !defined(OBSOLETE_SETREGID))
|
|
rb_gid_t gid;
|
|
#endif
|
|
|
|
check_gid_switch();
|
|
|
|
#if defined(HAVE_SETRESGID) || (defined(HAVE_SETREGID) && !defined(OBSOLETE_SETREGID))
|
|
gid = getgid();
|
|
#endif
|
|
|
|
#if defined(HAVE_SETRESGID)
|
|
if (gid != egid) {
|
|
if (setresgid(-1,egid,egid) < 0) rb_sys_fail(0);
|
|
SAVED_GROUP_ID = egid;
|
|
}
|
|
else {
|
|
if (setresgid(-1,egid,-1) < 0) rb_sys_fail(0);
|
|
}
|
|
#elif defined(HAVE_SETREGID) && !defined(OBSOLETE_SETREGID)
|
|
if (setregid(-1, egid) < 0) rb_sys_fail(0);
|
|
if (gid != egid) {
|
|
if (setregid(egid,gid) < 0) rb_sys_fail(0);
|
|
if (setregid(gid,egid) < 0) rb_sys_fail(0);
|
|
SAVED_GROUP_ID = egid;
|
|
}
|
|
#elif defined HAVE_SETEGID
|
|
if (setegid(egid) < 0) rb_sys_fail(0);
|
|
#elif defined HAVE_SETGID
|
|
if (geteuid() == 0 /* root user */) rb_sys_fail(0);
|
|
if (setgid(egid) < 0) rb_sys_fail(0);
|
|
#else
|
|
rb_notimplement();
|
|
#endif
|
|
return egid;
|
|
}
|
|
|
|
|
|
/*
|
|
* call-seq:
|
|
* Process::GID.grant_privilege(group) -> integer
|
|
* Process::GID.eid = group -> integer
|
|
*
|
|
* Set the effective group ID, and if possible, the saved group ID of
|
|
* the process to the given _group_. Returns the new
|
|
* effective group ID. Not available on all platforms.
|
|
*
|
|
* [Process.gid, Process.egid] #=> [0, 0]
|
|
* Process::GID.grant_privilege(31) #=> 33
|
|
* [Process.gid, Process.egid] #=> [0, 33]
|
|
*/
|
|
|
|
static VALUE
|
|
p_gid_grant_privilege(VALUE obj, VALUE id)
|
|
{
|
|
rb_setegid_core(OBJ2GID(id));
|
|
return id;
|
|
}
|
|
|
|
|
|
/*
|
|
* call-seq:
|
|
* Process::UID.re_exchangeable? -> true or false
|
|
*
|
|
* Returns +true+ if the real and effective user IDs of a
|
|
* process may be exchanged on the current platform.
|
|
*
|
|
*/
|
|
|
|
static VALUE
|
|
p_uid_exchangeable(VALUE _)
|
|
{
|
|
#if defined(HAVE_SETRESUID)
|
|
return Qtrue;
|
|
#elif defined(HAVE_SETREUID) && !defined(OBSOLETE_SETREUID)
|
|
return Qtrue;
|
|
#else
|
|
return Qfalse;
|
|
#endif
|
|
}
|
|
|
|
|
|
/*
|
|
* call-seq:
|
|
* Process::UID.re_exchange -> integer
|
|
*
|
|
* Exchange real and effective user IDs and return the new effective
|
|
* user ID. Not available on all platforms.
|
|
*
|
|
* [Process.uid, Process.euid] #=> [0, 31]
|
|
* Process::UID.re_exchange #=> 0
|
|
* [Process.uid, Process.euid] #=> [31, 0]
|
|
*/
|
|
|
|
static VALUE
|
|
p_uid_exchange(VALUE obj)
|
|
{
|
|
rb_uid_t uid;
|
|
#if defined(HAVE_SETRESUID) || (defined(HAVE_SETREUID) && !defined(OBSOLETE_SETREUID))
|
|
rb_uid_t euid;
|
|
#endif
|
|
|
|
check_uid_switch();
|
|
|
|
uid = getuid();
|
|
#if defined(HAVE_SETRESUID) || (defined(HAVE_SETREUID) && !defined(OBSOLETE_SETREUID))
|
|
euid = geteuid();
|
|
#endif
|
|
|
|
#if defined(HAVE_SETRESUID)
|
|
if (setresuid(euid, uid, uid) < 0) rb_sys_fail(0);
|
|
SAVED_USER_ID = uid;
|
|
#elif defined(HAVE_SETREUID) && !defined(OBSOLETE_SETREUID)
|
|
if (setreuid(euid,uid) < 0) rb_sys_fail(0);
|
|
SAVED_USER_ID = uid;
|
|
#else
|
|
rb_notimplement();
|
|
#endif
|
|
return UIDT2NUM(uid);
|
|
}
|
|
|
|
|
|
/*
|
|
* call-seq:
|
|
* Process::GID.re_exchangeable? -> true or false
|
|
*
|
|
* Returns +true+ if the real and effective group IDs of a
|
|
* process may be exchanged on the current platform.
|
|
*
|
|
*/
|
|
|
|
static VALUE
|
|
p_gid_exchangeable(VALUE _)
|
|
{
|
|
#if defined(HAVE_SETRESGID)
|
|
return Qtrue;
|
|
#elif defined(HAVE_SETREGID) && !defined(OBSOLETE_SETREGID)
|
|
return Qtrue;
|
|
#else
|
|
return Qfalse;
|
|
#endif
|
|
}
|
|
|
|
|
|
/*
|
|
* call-seq:
|
|
* Process::GID.re_exchange -> integer
|
|
*
|
|
* Exchange real and effective group IDs and return the new effective
|
|
* group ID. Not available on all platforms.
|
|
*
|
|
* [Process.gid, Process.egid] #=> [0, 33]
|
|
* Process::GID.re_exchange #=> 0
|
|
* [Process.gid, Process.egid] #=> [33, 0]
|
|
*/
|
|
|
|
static VALUE
|
|
p_gid_exchange(VALUE obj)
|
|
{
|
|
rb_gid_t gid;
|
|
#if defined(HAVE_SETRESGID) || (defined(HAVE_SETREGID) && !defined(OBSOLETE_SETREGID))
|
|
rb_gid_t egid;
|
|
#endif
|
|
|
|
check_gid_switch();
|
|
|
|
gid = getgid();
|
|
#if defined(HAVE_SETRESGID) || (defined(HAVE_SETREGID) && !defined(OBSOLETE_SETREGID))
|
|
egid = getegid();
|
|
#endif
|
|
|
|
#if defined(HAVE_SETRESGID)
|
|
if (setresgid(egid, gid, gid) < 0) rb_sys_fail(0);
|
|
SAVED_GROUP_ID = gid;
|
|
#elif defined(HAVE_SETREGID) && !defined(OBSOLETE_SETREGID)
|
|
if (setregid(egid,gid) < 0) rb_sys_fail(0);
|
|
SAVED_GROUP_ID = gid;
|
|
#else
|
|
rb_notimplement();
|
|
#endif
|
|
return GIDT2NUM(gid);
|
|
}
|
|
|
|
/* [MG] :FIXME: Is this correct? I'm not sure how to phrase this. */
|
|
|
|
/*
|
|
* call-seq:
|
|
* Process::UID.sid_available? -> true or false
|
|
*
|
|
* Returns +true+ if the current platform has saved user
|
|
* ID functionality.
|
|
*
|
|
*/
|
|
|
|
static VALUE
|
|
p_uid_have_saved_id(VALUE _)
|
|
{
|
|
#if defined(HAVE_SETRESUID) || defined(HAVE_SETEUID) || defined(_POSIX_SAVED_IDS)
|
|
return Qtrue;
|
|
#else
|
|
return Qfalse;
|
|
#endif
|
|
}
|
|
|
|
|
|
#if defined(HAVE_SETRESUID) || defined(HAVE_SETEUID) || defined(_POSIX_SAVED_IDS)
|
|
static VALUE
|
|
p_uid_sw_ensure(VALUE i)
|
|
{
|
|
rb_uid_t id = (rb_uid_t/* narrowing */)i;
|
|
under_uid_switch = 0;
|
|
id = rb_seteuid_core(id);
|
|
return UIDT2NUM(id);
|
|
}
|
|
|
|
|
|
/*
|
|
* call-seq:
|
|
* Process::UID.switch -> integer
|
|
* Process::UID.switch {|| block} -> object
|
|
*
|
|
* Switch the effective and real user IDs of the current process. If
|
|
* a <em>block</em> is given, the user IDs will be switched back
|
|
* after the block is executed. Returns the new effective user ID if
|
|
* called without a block, and the return value of the block if one
|
|
* is given.
|
|
*
|
|
*/
|
|
|
|
static VALUE
|
|
p_uid_switch(VALUE obj)
|
|
{
|
|
rb_uid_t uid, euid;
|
|
|
|
check_uid_switch();
|
|
|
|
uid = getuid();
|
|
euid = geteuid();
|
|
|
|
if (uid != euid) {
|
|
proc_seteuid(uid);
|
|
if (rb_block_given_p()) {
|
|
under_uid_switch = 1;
|
|
return rb_ensure(rb_yield, Qnil, p_uid_sw_ensure, SAVED_USER_ID);
|
|
}
|
|
else {
|
|
return UIDT2NUM(euid);
|
|
}
|
|
}
|
|
else if (euid != SAVED_USER_ID) {
|
|
proc_seteuid(SAVED_USER_ID);
|
|
if (rb_block_given_p()) {
|
|
under_uid_switch = 1;
|
|
return rb_ensure(rb_yield, Qnil, p_uid_sw_ensure, euid);
|
|
}
|
|
else {
|
|
return UIDT2NUM(uid);
|
|
}
|
|
}
|
|
else {
|
|
rb_syserr_fail(EPERM, 0);
|
|
}
|
|
|
|
UNREACHABLE_RETURN(Qnil);
|
|
}
|
|
#else
|
|
static VALUE
|
|
p_uid_sw_ensure(VALUE obj)
|
|
{
|
|
under_uid_switch = 0;
|
|
return p_uid_exchange(obj);
|
|
}
|
|
|
|
static VALUE
|
|
p_uid_switch(VALUE obj)
|
|
{
|
|
rb_uid_t uid, euid;
|
|
|
|
check_uid_switch();
|
|
|
|
uid = getuid();
|
|
euid = geteuid();
|
|
|
|
if (uid == euid) {
|
|
rb_syserr_fail(EPERM, 0);
|
|
}
|
|
p_uid_exchange(obj);
|
|
if (rb_block_given_p()) {
|
|
under_uid_switch = 1;
|
|
return rb_ensure(rb_yield, Qnil, p_uid_sw_ensure, obj);
|
|
}
|
|
else {
|
|
return UIDT2NUM(euid);
|
|
}
|
|
}
|
|
#endif
|
|
|
|
|
|
/* [MG] :FIXME: Is this correct? I'm not sure how to phrase this. */
|
|
|
|
/*
|
|
* call-seq:
|
|
* Process::GID.sid_available? -> true or false
|
|
*
|
|
* Returns +true+ if the current platform has saved group
|
|
* ID functionality.
|
|
*
|
|
*/
|
|
|
|
static VALUE
|
|
p_gid_have_saved_id(VALUE _)
|
|
{
|
|
#if defined(HAVE_SETRESGID) || defined(HAVE_SETEGID) || defined(_POSIX_SAVED_IDS)
|
|
return Qtrue;
|
|
#else
|
|
return Qfalse;
|
|
#endif
|
|
}
|
|
|
|
#if defined(HAVE_SETRESGID) || defined(HAVE_SETEGID) || defined(_POSIX_SAVED_IDS)
|
|
static VALUE
|
|
p_gid_sw_ensure(VALUE i)
|
|
{
|
|
rb_gid_t id = (rb_gid_t/* narrowing */)i;
|
|
under_gid_switch = 0;
|
|
id = rb_setegid_core(id);
|
|
return GIDT2NUM(id);
|
|
}
|
|
|
|
|
|
/*
|
|
* call-seq:
|
|
* Process::GID.switch -> integer
|
|
* Process::GID.switch {|| block} -> object
|
|
*
|
|
* Switch the effective and real group IDs of the current process. If
|
|
* a <em>block</em> is given, the group IDs will be switched back
|
|
* after the block is executed. Returns the new effective group ID if
|
|
* called without a block, and the return value of the block if one
|
|
* is given.
|
|
*
|
|
*/
|
|
|
|
static VALUE
|
|
p_gid_switch(VALUE obj)
|
|
{
|
|
rb_gid_t gid, egid;
|
|
|
|
check_gid_switch();
|
|
|
|
gid = getgid();
|
|
egid = getegid();
|
|
|
|
if (gid != egid) {
|
|
proc_setegid(obj, GIDT2NUM(gid));
|
|
if (rb_block_given_p()) {
|
|
under_gid_switch = 1;
|
|
return rb_ensure(rb_yield, Qnil, p_gid_sw_ensure, SAVED_GROUP_ID);
|
|
}
|
|
else {
|
|
return GIDT2NUM(egid);
|
|
}
|
|
}
|
|
else if (egid != SAVED_GROUP_ID) {
|
|
proc_setegid(obj, GIDT2NUM(SAVED_GROUP_ID));
|
|
if (rb_block_given_p()) {
|
|
under_gid_switch = 1;
|
|
return rb_ensure(rb_yield, Qnil, p_gid_sw_ensure, egid);
|
|
}
|
|
else {
|
|
return GIDT2NUM(gid);
|
|
}
|
|
}
|
|
else {
|
|
rb_syserr_fail(EPERM, 0);
|
|
}
|
|
|
|
UNREACHABLE_RETURN(Qnil);
|
|
}
|
|
#else
|
|
static VALUE
|
|
p_gid_sw_ensure(VALUE obj)
|
|
{
|
|
under_gid_switch = 0;
|
|
return p_gid_exchange(obj);
|
|
}
|
|
|
|
static VALUE
|
|
p_gid_switch(VALUE obj)
|
|
{
|
|
rb_gid_t gid, egid;
|
|
|
|
check_gid_switch();
|
|
|
|
gid = getgid();
|
|
egid = getegid();
|
|
|
|
if (gid == egid) {
|
|
rb_syserr_fail(EPERM, 0);
|
|
}
|
|
p_gid_exchange(obj);
|
|
if (rb_block_given_p()) {
|
|
under_gid_switch = 1;
|
|
return rb_ensure(rb_yield, Qnil, p_gid_sw_ensure, obj);
|
|
}
|
|
else {
|
|
return GIDT2NUM(egid);
|
|
}
|
|
}
|
|
#endif
|
|
|
|
|
|
#if defined(HAVE_TIMES)
|
|
static long
|
|
get_clk_tck(void)
|
|
{
|
|
#ifdef HAVE__SC_CLK_TCK
|
|
return sysconf(_SC_CLK_TCK);
|
|
#elif defined CLK_TCK
|
|
return CLK_TCK;
|
|
#elif defined HZ
|
|
return HZ;
|
|
#else
|
|
return 60;
|
|
#endif
|
|
}
|
|
|
|
/*
|
|
* call-seq:
|
|
* Process.times -> process_tms
|
|
*
|
|
* Returns a Process::Tms structure that contains user and system CPU times
|
|
* for the current process, and for its children processes:
|
|
*
|
|
* Process.times
|
|
* # => #<struct Process::Tms utime=55.122118, stime=35.533068, cutime=0.0, cstime=0.002846>
|
|
*
|
|
*/
|
|
|
|
VALUE
|
|
rb_proc_times(VALUE obj)
|
|
{
|
|
VALUE utime, stime, cutime, cstime, ret;
|
|
#if defined(RUSAGE_SELF) && defined(RUSAGE_CHILDREN)
|
|
struct rusage usage_s, usage_c;
|
|
|
|
if (getrusage(RUSAGE_SELF, &usage_s) != 0 || getrusage(RUSAGE_CHILDREN, &usage_c) != 0)
|
|
rb_sys_fail("getrusage");
|
|
utime = DBL2NUM((double)usage_s.ru_utime.tv_sec + (double)usage_s.ru_utime.tv_usec/1e6);
|
|
stime = DBL2NUM((double)usage_s.ru_stime.tv_sec + (double)usage_s.ru_stime.tv_usec/1e6);
|
|
cutime = DBL2NUM((double)usage_c.ru_utime.tv_sec + (double)usage_c.ru_utime.tv_usec/1e6);
|
|
cstime = DBL2NUM((double)usage_c.ru_stime.tv_sec + (double)usage_c.ru_stime.tv_usec/1e6);
|
|
#else
|
|
const double hertz = (double)get_clk_tck();
|
|
struct tms buf;
|
|
|
|
times(&buf);
|
|
utime = DBL2NUM(buf.tms_utime / hertz);
|
|
stime = DBL2NUM(buf.tms_stime / hertz);
|
|
cutime = DBL2NUM(buf.tms_cutime / hertz);
|
|
cstime = DBL2NUM(buf.tms_cstime / hertz);
|
|
#endif
|
|
ret = rb_struct_new(rb_cProcessTms, utime, stime, cutime, cstime);
|
|
RB_GC_GUARD(utime);
|
|
RB_GC_GUARD(stime);
|
|
RB_GC_GUARD(cutime);
|
|
RB_GC_GUARD(cstime);
|
|
return ret;
|
|
}
|
|
#else
|
|
#define rb_proc_times rb_f_notimplement
|
|
#endif
|
|
|
|
#ifdef HAVE_LONG_LONG
|
|
typedef LONG_LONG timetick_int_t;
|
|
#define TIMETICK_INT_MIN LLONG_MIN
|
|
#define TIMETICK_INT_MAX LLONG_MAX
|
|
#define TIMETICK_INT2NUM(v) LL2NUM(v)
|
|
#define MUL_OVERFLOW_TIMETICK_P(a, b) MUL_OVERFLOW_LONG_LONG_P(a, b)
|
|
#else
|
|
typedef long timetick_int_t;
|
|
#define TIMETICK_INT_MIN LONG_MIN
|
|
#define TIMETICK_INT_MAX LONG_MAX
|
|
#define TIMETICK_INT2NUM(v) LONG2NUM(v)
|
|
#define MUL_OVERFLOW_TIMETICK_P(a, b) MUL_OVERFLOW_LONG_P(a, b)
|
|
#endif
|
|
|
|
CONSTFUNC(static timetick_int_t gcd_timetick_int(timetick_int_t, timetick_int_t));
|
|
static timetick_int_t
|
|
gcd_timetick_int(timetick_int_t a, timetick_int_t b)
|
|
{
|
|
timetick_int_t t;
|
|
|
|
if (a < b) {
|
|
t = a;
|
|
a = b;
|
|
b = t;
|
|
}
|
|
|
|
while (1) {
|
|
t = a % b;
|
|
if (t == 0)
|
|
return b;
|
|
a = b;
|
|
b = t;
|
|
}
|
|
}
|
|
|
|
static void
|
|
reduce_fraction(timetick_int_t *np, timetick_int_t *dp)
|
|
{
|
|
timetick_int_t gcd = gcd_timetick_int(*np, *dp);
|
|
if (gcd != 1) {
|
|
*np /= gcd;
|
|
*dp /= gcd;
|
|
}
|
|
}
|
|
|
|
static void
|
|
reduce_factors(timetick_int_t *numerators, int num_numerators,
|
|
timetick_int_t *denominators, int num_denominators)
|
|
{
|
|
int i, j;
|
|
for (i = 0; i < num_numerators; i++) {
|
|
if (numerators[i] == 1)
|
|
continue;
|
|
for (j = 0; j < num_denominators; j++) {
|
|
if (denominators[j] == 1)
|
|
continue;
|
|
reduce_fraction(&numerators[i], &denominators[j]);
|
|
}
|
|
}
|
|
}
|
|
|
|
struct timetick {
|
|
timetick_int_t giga_count;
|
|
int32_t count; /* 0 .. 999999999 */
|
|
};
|
|
|
|
static VALUE
|
|
timetick2dblnum(struct timetick *ttp,
|
|
timetick_int_t *numerators, int num_numerators,
|
|
timetick_int_t *denominators, int num_denominators)
|
|
{
|
|
double d;
|
|
int i;
|
|
|
|
reduce_factors(numerators, num_numerators,
|
|
denominators, num_denominators);
|
|
|
|
d = ttp->giga_count * 1e9 + ttp->count;
|
|
|
|
for (i = 0; i < num_numerators; i++)
|
|
d *= numerators[i];
|
|
for (i = 0; i < num_denominators; i++)
|
|
d /= denominators[i];
|
|
|
|
return DBL2NUM(d);
|
|
}
|
|
|
|
static VALUE
|
|
timetick2dblnum_reciprocal(struct timetick *ttp,
|
|
timetick_int_t *numerators, int num_numerators,
|
|
timetick_int_t *denominators, int num_denominators)
|
|
{
|
|
double d;
|
|
int i;
|
|
|
|
reduce_factors(numerators, num_numerators,
|
|
denominators, num_denominators);
|
|
|
|
d = 1.0;
|
|
for (i = 0; i < num_denominators; i++)
|
|
d *= denominators[i];
|
|
for (i = 0; i < num_numerators; i++)
|
|
d /= numerators[i];
|
|
d /= ttp->giga_count * 1e9 + ttp->count;
|
|
|
|
return DBL2NUM(d);
|
|
}
|
|
|
|
#define NDIV(x,y) (-(-((x)+1)/(y))-1)
|
|
#define DIV(n,d) ((n)<0 ? NDIV((n),(d)) : (n)/(d))
|
|
|
|
static VALUE
|
|
timetick2integer(struct timetick *ttp,
|
|
timetick_int_t *numerators, int num_numerators,
|
|
timetick_int_t *denominators, int num_denominators)
|
|
{
|
|
VALUE v;
|
|
int i;
|
|
|
|
reduce_factors(numerators, num_numerators,
|
|
denominators, num_denominators);
|
|
|
|
if (!MUL_OVERFLOW_SIGNED_INTEGER_P(1000000000, ttp->giga_count,
|
|
TIMETICK_INT_MIN, TIMETICK_INT_MAX-ttp->count)) {
|
|
timetick_int_t t = ttp->giga_count * 1000000000 + ttp->count;
|
|
for (i = 0; i < num_numerators; i++) {
|
|
timetick_int_t factor = numerators[i];
|
|
if (MUL_OVERFLOW_TIMETICK_P(factor, t))
|
|
goto generic;
|
|
t *= factor;
|
|
}
|
|
for (i = 0; i < num_denominators; i++) {
|
|
t = DIV(t, denominators[i]);
|
|
}
|
|
return TIMETICK_INT2NUM(t);
|
|
}
|
|
|
|
generic:
|
|
v = TIMETICK_INT2NUM(ttp->giga_count);
|
|
v = rb_funcall(v, '*', 1, LONG2FIX(1000000000));
|
|
v = rb_funcall(v, '+', 1, LONG2FIX(ttp->count));
|
|
for (i = 0; i < num_numerators; i++) {
|
|
timetick_int_t factor = numerators[i];
|
|
if (factor == 1)
|
|
continue;
|
|
v = rb_funcall(v, '*', 1, TIMETICK_INT2NUM(factor));
|
|
}
|
|
for (i = 0; i < num_denominators; i++) {
|
|
v = rb_funcall(v, '/', 1, TIMETICK_INT2NUM(denominators[i])); /* Ruby's '/' is div. */
|
|
}
|
|
return v;
|
|
}
|
|
|
|
static VALUE
|
|
make_clock_result(struct timetick *ttp,
|
|
timetick_int_t *numerators, int num_numerators,
|
|
timetick_int_t *denominators, int num_denominators,
|
|
VALUE unit)
|
|
{
|
|
if (unit == ID2SYM(id_nanosecond)) {
|
|
numerators[num_numerators++] = 1000000000;
|
|
return timetick2integer(ttp, numerators, num_numerators, denominators, num_denominators);
|
|
}
|
|
else if (unit == ID2SYM(id_microsecond)) {
|
|
numerators[num_numerators++] = 1000000;
|
|
return timetick2integer(ttp, numerators, num_numerators, denominators, num_denominators);
|
|
}
|
|
else if (unit == ID2SYM(id_millisecond)) {
|
|
numerators[num_numerators++] = 1000;
|
|
return timetick2integer(ttp, numerators, num_numerators, denominators, num_denominators);
|
|
}
|
|
else if (unit == ID2SYM(id_second)) {
|
|
return timetick2integer(ttp, numerators, num_numerators, denominators, num_denominators);
|
|
}
|
|
else if (unit == ID2SYM(id_float_microsecond)) {
|
|
numerators[num_numerators++] = 1000000;
|
|
return timetick2dblnum(ttp, numerators, num_numerators, denominators, num_denominators);
|
|
}
|
|
else if (unit == ID2SYM(id_float_millisecond)) {
|
|
numerators[num_numerators++] = 1000;
|
|
return timetick2dblnum(ttp, numerators, num_numerators, denominators, num_denominators);
|
|
}
|
|
else if (NIL_P(unit) || unit == ID2SYM(id_float_second)) {
|
|
return timetick2dblnum(ttp, numerators, num_numerators, denominators, num_denominators);
|
|
}
|
|
else
|
|
rb_raise(rb_eArgError, "unexpected unit: %"PRIsVALUE, unit);
|
|
}
|
|
|
|
#ifdef __APPLE__
|
|
static const mach_timebase_info_data_t *
|
|
get_mach_timebase_info(void)
|
|
{
|
|
static mach_timebase_info_data_t sTimebaseInfo;
|
|
|
|
if ( sTimebaseInfo.denom == 0 ) {
|
|
(void) mach_timebase_info(&sTimebaseInfo);
|
|
}
|
|
|
|
return &sTimebaseInfo;
|
|
}
|
|
|
|
double
|
|
ruby_real_ms_time(void)
|
|
{
|
|
const mach_timebase_info_data_t *info = get_mach_timebase_info();
|
|
uint64_t t = mach_absolute_time();
|
|
return (double)t * info->numer / info->denom / 1e6;
|
|
}
|
|
#endif
|
|
|
|
#if defined(NUM2CLOCKID)
|
|
# define NUMERIC_CLOCKID 1
|
|
#else
|
|
# define NUMERIC_CLOCKID 0
|
|
# define NUM2CLOCKID(x) 0
|
|
#endif
|
|
|
|
#define clock_failed(name, err, arg) do { \
|
|
int clock_error = (err); \
|
|
rb_syserr_fail_str(clock_error, rb_sprintf("clock_" name "(%+"PRIsVALUE")", (arg))); \
|
|
} while (0)
|
|
|
|
/*
|
|
* call-seq:
|
|
* Process.clock_gettime(clock_id, unit = :float_second) -> number
|
|
*
|
|
* Returns a clock time as determined by POSIX function
|
|
* {clock_gettime()}[https://man7.org/linux/man-pages/man3/clock_gettime.3.html]:
|
|
*
|
|
* Process.clock_gettime(:CLOCK_PROCESS_CPUTIME_ID) # => 198.650379677
|
|
*
|
|
* Argument +clock_id+ should be a symbol or a constant that specifies
|
|
* the clock whose time is to be returned;
|
|
* see below.
|
|
*
|
|
* Optional argument +unit+ should be a symbol that specifies
|
|
* the unit to be used in the returned clock time;
|
|
* see below.
|
|
*
|
|
* <b>Argument +clock_id+</b>
|
|
*
|
|
* Argument +clock_id+ specifies the clock whose time is to be returned;
|
|
* it may be a constant such as <tt>Process::CLOCK_REALTIME</tt>,
|
|
* or a symbol shorthand such as +:CLOCK_REALTIME+.
|
|
*
|
|
* The supported clocks depend on the underlying operating system;
|
|
* this method supports the following clocks on the indicated platforms
|
|
* (raises Errno::EINVAL if called with an unsupported clock):
|
|
*
|
|
* - +:CLOCK_BOOTTIME+: Linux 2.6.39.
|
|
* - +:CLOCK_BOOTTIME_ALARM+: Linux 3.0.
|
|
* - +:CLOCK_MONOTONIC+: SUSv3 to 4, Linux 2.5.63, FreeBSD 3.0, NetBSD 2.0, OpenBSD 3.4, macOS 10.12, Windows-2000.
|
|
* - +:CLOCK_MONOTONIC_COARSE+: Linux 2.6.32.
|
|
* - +:CLOCK_MONOTONIC_FAST+: FreeBSD 8.1.
|
|
* - +:CLOCK_MONOTONIC_PRECISE+: FreeBSD 8.1.
|
|
* - +:CLOCK_MONOTONIC_RAW+: Linux 2.6.28, macOS 10.12.
|
|
* - +:CLOCK_MONOTONIC_RAW_APPROX+: macOS 10.12.
|
|
* - +:CLOCK_PROCESS_CPUTIME_ID+: SUSv3 to 4, Linux 2.5.63, FreeBSD 9.3, OpenBSD 5.4, macOS 10.12.
|
|
* - +:CLOCK_PROF+: FreeBSD 3.0, OpenBSD 2.1.
|
|
* - +:CLOCK_REALTIME+: SUSv2 to 4, Linux 2.5.63, FreeBSD 3.0, NetBSD 2.0, OpenBSD 2.1, macOS 10.12, Windows-8/Server-2012.
|
|
* Time.now is recommended over +:CLOCK_REALTIME:.
|
|
* - +:CLOCK_REALTIME_ALARM+: Linux 3.0.
|
|
* - +:CLOCK_REALTIME_COARSE+: Linux 2.6.32.
|
|
* - +:CLOCK_REALTIME_FAST+: FreeBSD 8.1.
|
|
* - +:CLOCK_REALTIME_PRECISE+: FreeBSD 8.1.
|
|
* - +:CLOCK_SECOND+: FreeBSD 8.1.
|
|
* - +:CLOCK_TAI+: Linux 3.10.
|
|
* - +:CLOCK_THREAD_CPUTIME_ID+: SUSv3 to 4, Linux 2.5.63, FreeBSD 7.1, OpenBSD 5.4, macOS 10.12.
|
|
* - +:CLOCK_UPTIME+: FreeBSD 7.0, OpenBSD 5.5.
|
|
* - +:CLOCK_UPTIME_FAST+: FreeBSD 8.1.
|
|
* - +:CLOCK_UPTIME_PRECISE+: FreeBSD 8.1.
|
|
* - +:CLOCK_UPTIME_RAW+: macOS 10.12.
|
|
* - +:CLOCK_UPTIME_RAW_APPROX+: macOS 10.12.
|
|
* - +:CLOCK_VIRTUAL+: FreeBSD 3.0, OpenBSD 2.1.
|
|
*
|
|
* Note that SUS stands for Single Unix Specification.
|
|
* SUS contains POSIX and clock_gettime is defined in the POSIX part.
|
|
* SUS defines +:CLOCK_REALTIME+ as mandatory but
|
|
* +:CLOCK_MONOTONIC+, +:CLOCK_PROCESS_CPUTIME_ID+,
|
|
* and +:CLOCK_THREAD_CPUTIME_ID+ are optional.
|
|
*
|
|
* Certain emulations are used when the given +clock_id+
|
|
* is not supported directly:
|
|
*
|
|
* - Emulations for +:CLOCK_REALTIME+:
|
|
*
|
|
* - +:GETTIMEOFDAY_BASED_CLOCK_REALTIME+:
|
|
* Use gettimeofday() defined by SUS (deprecated in SUSv4).
|
|
* The resolution is 1 microsecond.
|
|
* - +:TIME_BASED_CLOCK_REALTIME+:
|
|
* Use time() defined by ISO C.
|
|
* The resolution is 1 second.
|
|
*
|
|
* - Emulations for +:CLOCK_MONOTONIC+:
|
|
*
|
|
* - +:MACH_ABSOLUTE_TIME_BASED_CLOCK_MONOTONIC+:
|
|
* Use mach_absolute_time(), available on Darwin.
|
|
* The resolution is CPU dependent.
|
|
* - +:TIMES_BASED_CLOCK_MONOTONIC+:
|
|
* Use the result value of times() defined by POSIX, thus:
|
|
* >>>
|
|
* Upon successful completion, times() shall return the elapsed real time,
|
|
* in clock ticks, since an arbitrary point in the past
|
|
* (for example, system start-up time).
|
|
*
|
|
* For example, GNU/Linux returns a value based on jiffies and it is monotonic.
|
|
* However, 4.4BSD uses gettimeofday() and it is not monotonic.
|
|
* (FreeBSD uses +:CLOCK_MONOTONIC+ instead, though.)
|
|
*
|
|
* The resolution is the clock tick.
|
|
* "getconf CLK_TCK" command shows the clock ticks per second.
|
|
* (The clock ticks-per-second is defined by HZ macro in older systems.)
|
|
* If it is 100 and clock_t is 32 bits integer type,
|
|
* the resolution is 10 millisecond and cannot represent over 497 days.
|
|
*
|
|
* - Emulations for +:CLOCK_PROCESS_CPUTIME_ID+:
|
|
*
|
|
* - +:GETRUSAGE_BASED_CLOCK_PROCESS_CPUTIME_ID+:
|
|
* Use getrusage() defined by SUS.
|
|
* getrusage() is used with RUSAGE_SELF to obtain the time only for
|
|
* the calling process (excluding the time for child processes).
|
|
* The result is addition of user time (ru_utime) and system time (ru_stime).
|
|
* The resolution is 1 microsecond.
|
|
* - +:TIMES_BASED_CLOCK_PROCESS_CPUTIME_ID+:
|
|
* Use times() defined by POSIX.
|
|
* The result is addition of user time (tms_utime) and system time (tms_stime).
|
|
* tms_cutime and tms_cstime are ignored to exclude the time for child processes.
|
|
* The resolution is the clock tick.
|
|
* "getconf CLK_TCK" command shows the clock ticks per second.
|
|
* (The clock ticks per second is defined by HZ macro in older systems.)
|
|
* If it is 100, the resolution is 10 millisecond.
|
|
* - +:CLOCK_BASED_CLOCK_PROCESS_CPUTIME_ID+:
|
|
* Use clock() defined by ISO C.
|
|
* The resolution is <tt>1/CLOCKS_PER_SEC</tt>.
|
|
* +CLOCKS_PER_SEC+ is the C-level macro defined by time.h.
|
|
* SUS defines +CLOCKS_PER_SEC+ as 1000000;
|
|
* other systems may define it differently.
|
|
* If +CLOCKS_PER_SEC+ is 1000000 (as in SUS),
|
|
* the resolution is 1 microsecond.
|
|
* If +CLOCKS_PER_SEC+ is 1000000 and clock_t is a 32-bit integer type,
|
|
* it cannot represent over 72 minutes.
|
|
*
|
|
* <b>Argument +unit+</b>
|
|
*
|
|
* Optional argument +unit+ (default +:float_second+)
|
|
* specifies the unit for the returned value.
|
|
*
|
|
* - +:float_microsecond+: Number of microseconds as a float.
|
|
* - +:float_millisecond+: Number of milliseconds as a float.
|
|
* - +:float_second+: Number of seconds as a float.
|
|
* - +:microsecond+: Number of microseconds as an integer.
|
|
* - +:millisecond+: Number of milliseconds as an integer.
|
|
* - +:nanosecond+: Number of nanoseconds as an integer.
|
|
* - +::second+: Number of seconds as an integer.
|
|
*
|
|
* Examples:
|
|
*
|
|
* Process.clock_gettime(:CLOCK_PROCESS_CPUTIME_ID, :float_microsecond)
|
|
* # => 203605054.825
|
|
* Process.clock_gettime(:CLOCK_PROCESS_CPUTIME_ID, :float_millisecond)
|
|
* # => 203643.696848
|
|
* Process.clock_gettime(:CLOCK_PROCESS_CPUTIME_ID, :float_second)
|
|
* # => 203.762181929
|
|
* Process.clock_gettime(:CLOCK_PROCESS_CPUTIME_ID, :microsecond)
|
|
* # => 204123212
|
|
* Process.clock_gettime(:CLOCK_PROCESS_CPUTIME_ID, :millisecond)
|
|
* # => 204298
|
|
* Process.clock_gettime(:CLOCK_PROCESS_CPUTIME_ID, :nanosecond)
|
|
* # => 204602286036
|
|
* Process.clock_gettime(:CLOCK_PROCESS_CPUTIME_ID, :second)
|
|
* # => 204
|
|
*
|
|
* The underlying function, clock_gettime(), returns a number of nanoseconds.
|
|
* Float object (IEEE 754 double) is not enough to represent
|
|
* the return value for +:CLOCK_REALTIME+.
|
|
* If the exact nanoseconds value is required, use +:nanosecond+ as the +unit+.
|
|
*
|
|
* The origin (time zero) of the returned value is system-dependent,
|
|
* and may be, for example, system start up time,
|
|
* process start up time, the Epoch, etc.
|
|
*
|
|
* The origin in +:CLOCK_REALTIME+ is defined as the Epoch:
|
|
* <tt>1970-01-01 00:00:00 UTC</tt>;
|
|
* some systems count leap seconds and others don't,
|
|
* so the result may vary across systems.
|
|
*/
|
|
static VALUE
|
|
rb_clock_gettime(int argc, VALUE *argv, VALUE _)
|
|
{
|
|
int ret;
|
|
|
|
struct timetick tt;
|
|
timetick_int_t numerators[2];
|
|
timetick_int_t denominators[2];
|
|
int num_numerators = 0;
|
|
int num_denominators = 0;
|
|
|
|
VALUE unit = (rb_check_arity(argc, 1, 2) == 2) ? argv[1] : Qnil;
|
|
VALUE clk_id = argv[0];
|
|
#ifdef HAVE_CLOCK_GETTIME
|
|
clockid_t c;
|
|
#endif
|
|
|
|
if (SYMBOL_P(clk_id)) {
|
|
#ifdef CLOCK_REALTIME
|
|
if (clk_id == RUBY_CLOCK_REALTIME) {
|
|
c = CLOCK_REALTIME;
|
|
goto gettime;
|
|
}
|
|
#endif
|
|
|
|
#ifdef CLOCK_MONOTONIC
|
|
if (clk_id == RUBY_CLOCK_MONOTONIC) {
|
|
c = CLOCK_MONOTONIC;
|
|
goto gettime;
|
|
}
|
|
#endif
|
|
|
|
#ifdef CLOCK_PROCESS_CPUTIME_ID
|
|
if (clk_id == RUBY_CLOCK_PROCESS_CPUTIME_ID) {
|
|
c = CLOCK_PROCESS_CPUTIME_ID;
|
|
goto gettime;
|
|
}
|
|
#endif
|
|
|
|
#ifdef CLOCK_THREAD_CPUTIME_ID
|
|
if (clk_id == RUBY_CLOCK_THREAD_CPUTIME_ID) {
|
|
c = CLOCK_THREAD_CPUTIME_ID;
|
|
goto gettime;
|
|
}
|
|
#endif
|
|
|
|
/*
|
|
* Non-clock_gettime clocks are provided by symbol clk_id.
|
|
*/
|
|
#ifdef HAVE_GETTIMEOFDAY
|
|
/*
|
|
* GETTIMEOFDAY_BASED_CLOCK_REALTIME is used for
|
|
* CLOCK_REALTIME if clock_gettime is not available.
|
|
*/
|
|
#define RUBY_GETTIMEOFDAY_BASED_CLOCK_REALTIME ID2SYM(id_GETTIMEOFDAY_BASED_CLOCK_REALTIME)
|
|
if (clk_id == RUBY_GETTIMEOFDAY_BASED_CLOCK_REALTIME) {
|
|
struct timeval tv;
|
|
ret = gettimeofday(&tv, 0);
|
|
if (ret != 0)
|
|
rb_sys_fail("gettimeofday");
|
|
tt.giga_count = tv.tv_sec;
|
|
tt.count = (int32_t)tv.tv_usec * 1000;
|
|
denominators[num_denominators++] = 1000000000;
|
|
goto success;
|
|
}
|
|
#endif
|
|
|
|
#define RUBY_TIME_BASED_CLOCK_REALTIME ID2SYM(id_TIME_BASED_CLOCK_REALTIME)
|
|
if (clk_id == RUBY_TIME_BASED_CLOCK_REALTIME) {
|
|
time_t t;
|
|
t = time(NULL);
|
|
if (t == (time_t)-1)
|
|
rb_sys_fail("time");
|
|
tt.giga_count = t;
|
|
tt.count = 0;
|
|
denominators[num_denominators++] = 1000000000;
|
|
goto success;
|
|
}
|
|
|
|
#ifdef HAVE_TIMES
|
|
#define RUBY_TIMES_BASED_CLOCK_MONOTONIC \
|
|
ID2SYM(id_TIMES_BASED_CLOCK_MONOTONIC)
|
|
if (clk_id == RUBY_TIMES_BASED_CLOCK_MONOTONIC) {
|
|
struct tms buf;
|
|
clock_t c;
|
|
unsigned_clock_t uc;
|
|
c = times(&buf);
|
|
if (c == (clock_t)-1)
|
|
rb_sys_fail("times");
|
|
uc = (unsigned_clock_t)c;
|
|
tt.count = (int32_t)(uc % 1000000000);
|
|
tt.giga_count = (uc / 1000000000);
|
|
denominators[num_denominators++] = get_clk_tck();
|
|
goto success;
|
|
}
|
|
#endif
|
|
|
|
#ifdef RUSAGE_SELF
|
|
#define RUBY_GETRUSAGE_BASED_CLOCK_PROCESS_CPUTIME_ID \
|
|
ID2SYM(id_GETRUSAGE_BASED_CLOCK_PROCESS_CPUTIME_ID)
|
|
if (clk_id == RUBY_GETRUSAGE_BASED_CLOCK_PROCESS_CPUTIME_ID) {
|
|
struct rusage usage;
|
|
int32_t usec;
|
|
ret = getrusage(RUSAGE_SELF, &usage);
|
|
if (ret != 0)
|
|
rb_sys_fail("getrusage");
|
|
tt.giga_count = usage.ru_utime.tv_sec + usage.ru_stime.tv_sec;
|
|
usec = (int32_t)(usage.ru_utime.tv_usec + usage.ru_stime.tv_usec);
|
|
if (1000000 <= usec) {
|
|
tt.giga_count++;
|
|
usec -= 1000000;
|
|
}
|
|
tt.count = usec * 1000;
|
|
denominators[num_denominators++] = 1000000000;
|
|
goto success;
|
|
}
|
|
#endif
|
|
|
|
#ifdef HAVE_TIMES
|
|
#define RUBY_TIMES_BASED_CLOCK_PROCESS_CPUTIME_ID \
|
|
ID2SYM(id_TIMES_BASED_CLOCK_PROCESS_CPUTIME_ID)
|
|
if (clk_id == RUBY_TIMES_BASED_CLOCK_PROCESS_CPUTIME_ID) {
|
|
struct tms buf;
|
|
unsigned_clock_t utime, stime;
|
|
if (times(&buf) == (clock_t)-1)
|
|
rb_sys_fail("times");
|
|
utime = (unsigned_clock_t)buf.tms_utime;
|
|
stime = (unsigned_clock_t)buf.tms_stime;
|
|
tt.count = (int32_t)((utime % 1000000000) + (stime % 1000000000));
|
|
tt.giga_count = (utime / 1000000000) + (stime / 1000000000);
|
|
if (1000000000 <= tt.count) {
|
|
tt.count -= 1000000000;
|
|
tt.giga_count++;
|
|
}
|
|
denominators[num_denominators++] = get_clk_tck();
|
|
goto success;
|
|
}
|
|
#endif
|
|
|
|
#define RUBY_CLOCK_BASED_CLOCK_PROCESS_CPUTIME_ID \
|
|
ID2SYM(id_CLOCK_BASED_CLOCK_PROCESS_CPUTIME_ID)
|
|
if (clk_id == RUBY_CLOCK_BASED_CLOCK_PROCESS_CPUTIME_ID) {
|
|
clock_t c;
|
|
unsigned_clock_t uc;
|
|
errno = 0;
|
|
c = clock();
|
|
if (c == (clock_t)-1)
|
|
rb_sys_fail("clock");
|
|
uc = (unsigned_clock_t)c;
|
|
tt.count = (int32_t)(uc % 1000000000);
|
|
tt.giga_count = uc / 1000000000;
|
|
denominators[num_denominators++] = CLOCKS_PER_SEC;
|
|
goto success;
|
|
}
|
|
|
|
#ifdef __APPLE__
|
|
if (clk_id == RUBY_MACH_ABSOLUTE_TIME_BASED_CLOCK_MONOTONIC) {
|
|
const mach_timebase_info_data_t *info = get_mach_timebase_info();
|
|
uint64_t t = mach_absolute_time();
|
|
tt.count = (int32_t)(t % 1000000000);
|
|
tt.giga_count = t / 1000000000;
|
|
numerators[num_numerators++] = info->numer;
|
|
denominators[num_denominators++] = info->denom;
|
|
denominators[num_denominators++] = 1000000000;
|
|
goto success;
|
|
}
|
|
#endif
|
|
}
|
|
else if (NUMERIC_CLOCKID) {
|
|
#if defined(HAVE_CLOCK_GETTIME)
|
|
struct timespec ts;
|
|
c = NUM2CLOCKID(clk_id);
|
|
gettime:
|
|
ret = clock_gettime(c, &ts);
|
|
if (ret == -1)
|
|
clock_failed("gettime", errno, clk_id);
|
|
tt.count = (int32_t)ts.tv_nsec;
|
|
tt.giga_count = ts.tv_sec;
|
|
denominators[num_denominators++] = 1000000000;
|
|
goto success;
|
|
#endif
|
|
}
|
|
else {
|
|
rb_unexpected_type(clk_id, T_SYMBOL);
|
|
}
|
|
clock_failed("gettime", EINVAL, clk_id);
|
|
|
|
success:
|
|
return make_clock_result(&tt, numerators, num_numerators, denominators, num_denominators, unit);
|
|
}
|
|
|
|
/*
|
|
* call-seq:
|
|
* Process.clock_getres(clock_id, unit = :float_second) -> number
|
|
*
|
|
* Returns a clock resolution as determined by POSIX function
|
|
* {clock_getres()}[https://man7.org/linux/man-pages/man3/clock_getres.3.html]:
|
|
*
|
|
* Process.clock_getres(:CLOCK_REALTIME) # => 1.0e-09
|
|
*
|
|
* See Process.clock_gettime for the values of +clock_id+ and +unit+.
|
|
*
|
|
* Examples:
|
|
*
|
|
* Process.clock_getres(:CLOCK_PROCESS_CPUTIME_ID, :float_microsecond) # => 0.001
|
|
* Process.clock_getres(:CLOCK_PROCESS_CPUTIME_ID, :float_millisecond) # => 1.0e-06
|
|
* Process.clock_getres(:CLOCK_PROCESS_CPUTIME_ID, :float_second) # => 1.0e-09
|
|
* Process.clock_getres(:CLOCK_PROCESS_CPUTIME_ID, :microsecond) # => 0
|
|
* Process.clock_getres(:CLOCK_PROCESS_CPUTIME_ID, :millisecond) # => 0
|
|
* Process.clock_getres(:CLOCK_PROCESS_CPUTIME_ID, :nanosecond) # => 1
|
|
* Process.clock_getres(:CLOCK_PROCESS_CPUTIME_ID, :second) # => 0
|
|
*
|
|
* In addition to the values for +unit+ supported in Process.clock_gettime,
|
|
* this method supports +:hertz+, the integer number of clock ticks per second
|
|
* (which is the reciprocal of +:float_second+):
|
|
*
|
|
* Process.clock_getres(:TIMES_BASED_CLOCK_PROCESS_CPUTIME_ID, :hertz) # => 100.0
|
|
* Process.clock_getres(:TIMES_BASED_CLOCK_PROCESS_CPUTIME_ID, :float_second) # => 0.01
|
|
*
|
|
* <b>Accuracy</b>:
|
|
* Note that the returned resolution may be inaccurate on some platforms
|
|
* due to underlying bugs.
|
|
* Inaccurate resolutions have been reported for various clocks including
|
|
* +:CLOCK_MONOTONIC+ and +:CLOCK_MONOTONIC_RAW+
|
|
* on Linux, macOS, BSD or AIX platforms, when using ARM processors,
|
|
* or when using virtualization.
|
|
*/
|
|
static VALUE
|
|
rb_clock_getres(int argc, VALUE *argv, VALUE _)
|
|
{
|
|
int ret;
|
|
|
|
struct timetick tt;
|
|
timetick_int_t numerators[2];
|
|
timetick_int_t denominators[2];
|
|
int num_numerators = 0;
|
|
int num_denominators = 0;
|
|
#ifdef HAVE_CLOCK_GETRES
|
|
clockid_t c;
|
|
#endif
|
|
|
|
VALUE unit = (rb_check_arity(argc, 1, 2) == 2) ? argv[1] : Qnil;
|
|
VALUE clk_id = argv[0];
|
|
|
|
if (SYMBOL_P(clk_id)) {
|
|
#ifdef CLOCK_REALTIME
|
|
if (clk_id == RUBY_CLOCK_REALTIME) {
|
|
c = CLOCK_REALTIME;
|
|
goto getres;
|
|
}
|
|
#endif
|
|
|
|
#ifdef CLOCK_MONOTONIC
|
|
if (clk_id == RUBY_CLOCK_MONOTONIC) {
|
|
c = CLOCK_MONOTONIC;
|
|
goto getres;
|
|
}
|
|
#endif
|
|
|
|
#ifdef CLOCK_PROCESS_CPUTIME_ID
|
|
if (clk_id == RUBY_CLOCK_PROCESS_CPUTIME_ID) {
|
|
c = CLOCK_PROCESS_CPUTIME_ID;
|
|
goto getres;
|
|
}
|
|
#endif
|
|
|
|
#ifdef CLOCK_THREAD_CPUTIME_ID
|
|
if (clk_id == RUBY_CLOCK_THREAD_CPUTIME_ID) {
|
|
c = CLOCK_THREAD_CPUTIME_ID;
|
|
goto getres;
|
|
}
|
|
#endif
|
|
|
|
#ifdef RUBY_GETTIMEOFDAY_BASED_CLOCK_REALTIME
|
|
if (clk_id == RUBY_GETTIMEOFDAY_BASED_CLOCK_REALTIME) {
|
|
tt.giga_count = 0;
|
|
tt.count = 1000;
|
|
denominators[num_denominators++] = 1000000000;
|
|
goto success;
|
|
}
|
|
#endif
|
|
|
|
#ifdef RUBY_TIME_BASED_CLOCK_REALTIME
|
|
if (clk_id == RUBY_TIME_BASED_CLOCK_REALTIME) {
|
|
tt.giga_count = 1;
|
|
tt.count = 0;
|
|
denominators[num_denominators++] = 1000000000;
|
|
goto success;
|
|
}
|
|
#endif
|
|
|
|
#ifdef RUBY_TIMES_BASED_CLOCK_MONOTONIC
|
|
if (clk_id == RUBY_TIMES_BASED_CLOCK_MONOTONIC) {
|
|
tt.count = 1;
|
|
tt.giga_count = 0;
|
|
denominators[num_denominators++] = get_clk_tck();
|
|
goto success;
|
|
}
|
|
#endif
|
|
|
|
#ifdef RUBY_GETRUSAGE_BASED_CLOCK_PROCESS_CPUTIME_ID
|
|
if (clk_id == RUBY_GETRUSAGE_BASED_CLOCK_PROCESS_CPUTIME_ID) {
|
|
tt.giga_count = 0;
|
|
tt.count = 1000;
|
|
denominators[num_denominators++] = 1000000000;
|
|
goto success;
|
|
}
|
|
#endif
|
|
|
|
#ifdef RUBY_TIMES_BASED_CLOCK_PROCESS_CPUTIME_ID
|
|
if (clk_id == RUBY_TIMES_BASED_CLOCK_PROCESS_CPUTIME_ID) {
|
|
tt.count = 1;
|
|
tt.giga_count = 0;
|
|
denominators[num_denominators++] = get_clk_tck();
|
|
goto success;
|
|
}
|
|
#endif
|
|
|
|
#ifdef RUBY_CLOCK_BASED_CLOCK_PROCESS_CPUTIME_ID
|
|
if (clk_id == RUBY_CLOCK_BASED_CLOCK_PROCESS_CPUTIME_ID) {
|
|
tt.count = 1;
|
|
tt.giga_count = 0;
|
|
denominators[num_denominators++] = CLOCKS_PER_SEC;
|
|
goto success;
|
|
}
|
|
#endif
|
|
|
|
#ifdef RUBY_MACH_ABSOLUTE_TIME_BASED_CLOCK_MONOTONIC
|
|
if (clk_id == RUBY_MACH_ABSOLUTE_TIME_BASED_CLOCK_MONOTONIC) {
|
|
const mach_timebase_info_data_t *info = get_mach_timebase_info();
|
|
tt.count = 1;
|
|
tt.giga_count = 0;
|
|
numerators[num_numerators++] = info->numer;
|
|
denominators[num_denominators++] = info->denom;
|
|
denominators[num_denominators++] = 1000000000;
|
|
goto success;
|
|
}
|
|
#endif
|
|
}
|
|
else if (NUMERIC_CLOCKID) {
|
|
#if defined(HAVE_CLOCK_GETRES)
|
|
struct timespec ts;
|
|
c = NUM2CLOCKID(clk_id);
|
|
getres:
|
|
ret = clock_getres(c, &ts);
|
|
if (ret == -1)
|
|
clock_failed("getres", errno, clk_id);
|
|
tt.count = (int32_t)ts.tv_nsec;
|
|
tt.giga_count = ts.tv_sec;
|
|
denominators[num_denominators++] = 1000000000;
|
|
goto success;
|
|
#endif
|
|
}
|
|
else {
|
|
rb_unexpected_type(clk_id, T_SYMBOL);
|
|
}
|
|
clock_failed("getres", EINVAL, clk_id);
|
|
|
|
success:
|
|
if (unit == ID2SYM(id_hertz)) {
|
|
return timetick2dblnum_reciprocal(&tt, numerators, num_numerators, denominators, num_denominators);
|
|
}
|
|
else {
|
|
return make_clock_result(&tt, numerators, num_numerators, denominators, num_denominators, unit);
|
|
}
|
|
}
|
|
|
|
static VALUE
|
|
get_CHILD_STATUS(ID _x, VALUE *_y)
|
|
{
|
|
return rb_last_status_get();
|
|
}
|
|
|
|
static VALUE
|
|
get_PROCESS_ID(ID _x, VALUE *_y)
|
|
{
|
|
return get_pid();
|
|
}
|
|
|
|
/*
|
|
* call-seq:
|
|
* Process.kill(signal, *ids) -> count
|
|
*
|
|
* Sends a signal to each process specified by +ids+
|
|
* (which must specify at least one ID);
|
|
* returns the count of signals sent.
|
|
*
|
|
* For each given +id+, if +id+ is:
|
|
*
|
|
* - Positive, sends the signal to the process whose process ID is +id+.
|
|
* - Zero, send the signal to all processes in the current process group.
|
|
* - Negative, sends the signal to a system-dependent collection of processes.
|
|
*
|
|
* Argument +signal+ specifies the signal to be sent;
|
|
* the argument may be:
|
|
*
|
|
* - An integer signal number: e.g., +-29+, +0+, +29+.
|
|
* - A signal name (string), with or without leading <tt>'SIG'</tt>,
|
|
* and with or without a further prefixed minus sign (<tt>'-'</tt>):
|
|
* e.g.:
|
|
*
|
|
* - <tt>'SIGPOLL'</tt>.
|
|
* - <tt>'POLL'</tt>,
|
|
* - <tt>'-SIGPOLL'</tt>.
|
|
* - <tt>'-POLL'</tt>.
|
|
*
|
|
* - A signal symbol, with or without leading <tt>'SIG'</tt>,
|
|
* and with or without a further prefixed minus sign (<tt>'-'</tt>):
|
|
* e.g.:
|
|
*
|
|
* - +:SIGPOLL+.
|
|
* - +:POLL+.
|
|
* - <tt>:'-SIGPOLL'</tt>.
|
|
* - <tt>:'-POLL'</tt>.
|
|
*
|
|
* If +signal+ is:
|
|
*
|
|
* - A non-negative integer, or a signal name or symbol
|
|
* without prefixed <tt>'-'</tt>,
|
|
* each process with process ID +id+ is signalled.
|
|
* - A negative integer, or a signal name or symbol
|
|
* with prefixed <tt>'-'</tt>,
|
|
* each process group with group ID +id+ is signalled.
|
|
*
|
|
* Use method Signal.list to see which signals are supported
|
|
* by Ruby on the underlying platform;
|
|
* the method returns a hash of the string names
|
|
* and non-negative integer values of the supported signals.
|
|
* The size and content of the returned hash varies widely
|
|
* among platforms.
|
|
*
|
|
* Additionally, signal +0+ is useful to determine if the process exists.
|
|
*
|
|
* Example:
|
|
*
|
|
* pid = fork do
|
|
* Signal.trap('HUP') { puts 'Ouch!'; exit }
|
|
* # ... do some work ...
|
|
* end
|
|
* # ...
|
|
* Process.kill('HUP', pid)
|
|
* Process.wait
|
|
*
|
|
* Output:
|
|
*
|
|
* Ouch!
|
|
*
|
|
* Exceptions:
|
|
*
|
|
* - Raises Errno::EINVAL or RangeError if +signal+ is an integer
|
|
* but invalid.
|
|
* - Raises ArgumentError if +signal+ is a string or symbol
|
|
* but invalid.
|
|
* - Raises Errno::ESRCH or RangeError if one of +ids+ is invalid.
|
|
* - Raises Errno::EPERM if needed permissions are not in force.
|
|
*
|
|
* In the last two cases, signals may have been sent to some processes.
|
|
*/
|
|
|
|
static VALUE
|
|
proc_rb_f_kill(int c, const VALUE *v, VALUE _)
|
|
{
|
|
return rb_f_kill(c, v);
|
|
}
|
|
|
|
VALUE rb_mProcess;
|
|
static VALUE rb_mProcUID;
|
|
static VALUE rb_mProcGID;
|
|
static VALUE rb_mProcID_Syscall;
|
|
|
|
/*
|
|
* call-seq:
|
|
* Process.warmup -> true
|
|
*
|
|
* Notify the Ruby virtual machine that the boot sequence is finished,
|
|
* and that now is a good time to optimize the application. This is useful
|
|
* for long running applications.
|
|
*
|
|
* This method is expected to be called at the end of the application boot.
|
|
* If the application is deployed using a pre-forking model, +Process.warmup+
|
|
* should be called in the original process before the first fork.
|
|
*
|
|
* The actual optimizations performed are entirely implementation specific
|
|
* and may change in the future without notice.
|
|
*
|
|
* On CRuby, +Process.warmup+:
|
|
*
|
|
* * Performs a major GC.
|
|
* * Compacts the heap.
|
|
* * Promotes all surviving objects to the old generation.
|
|
* * Precomputes the coderange of all strings.
|
|
* * Frees all empty heap pages and increments the allocatable pages counter
|
|
* by the number of pages freed.
|
|
* * Invoke +malloc_trim+ if available to free empty malloc pages.
|
|
*/
|
|
|
|
static VALUE
|
|
proc_warmup(VALUE _)
|
|
{
|
|
RB_VM_LOCK_ENTER();
|
|
rb_gc_prepare_heap();
|
|
RB_VM_LOCK_LEAVE();
|
|
return Qtrue;
|
|
}
|
|
|
|
/*
|
|
* Document-module: Process
|
|
*
|
|
* \Module +Process+ represents a process in the underlying operating system.
|
|
* Its methods support management of the current process and its child processes.
|
|
*
|
|
* == \Process Creation
|
|
*
|
|
* Each of these methods creates a process:
|
|
*
|
|
* - Process.exec: Replaces the current process by running a given external command.
|
|
* - Process.spawn, Kernel#spawn: Executes the given command and returns its pid without waiting for completion.
|
|
* - Kernel#system: Executes the given command in a subshell.
|
|
*
|
|
* Each of these methods accepts:
|
|
*
|
|
* - An optional hash of environment variable names and values.
|
|
* - An optional hash of execution options.
|
|
*
|
|
* === Execution Environment
|
|
*
|
|
* Optional leading argument +env+ is a hash of name/value pairs,
|
|
* where each name is a string and each value is a string or +nil+;
|
|
* each name/value pair is added to ENV in the new process.
|
|
*
|
|
* Process.spawn( 'ruby -e "p ENV[\"Foo\"]"')
|
|
* Process.spawn({'Foo' => '0'}, 'ruby -e "p ENV[\"Foo\"]"')
|
|
*
|
|
* Output:
|
|
*
|
|
* nil
|
|
* "0"
|
|
*
|
|
* The effect is usually similar to that of calling ENV#update with argument +env+,
|
|
* where each named environment variable is created or updated
|
|
* (if the value is non-+nil+),
|
|
* or deleted (if the value is +nil+).
|
|
*
|
|
* However, some modifications to the calling process may remain
|
|
* if the new process fails.
|
|
* For example, hard resource limits are not restored.
|
|
*
|
|
* === Execution Options
|
|
*
|
|
* Optional trailing argument +options+ is a hash of execution options.
|
|
*
|
|
* ==== Working Directory (+:chdir+)
|
|
*
|
|
* By default, the working directory for the new process is the same as
|
|
* that of the current process:
|
|
*
|
|
* Dir.chdir('/var')
|
|
* Process.spawn('ruby -e "puts Dir.pwd"')
|
|
*
|
|
* Output:
|
|
*
|
|
* /var
|
|
*
|
|
* Use option +:chdir+ to set the working directory for the new process:
|
|
*
|
|
* Process.spawn('ruby -e "puts Dir.pwd"', {chdir: '/tmp'})
|
|
*
|
|
* Output:
|
|
*
|
|
* /tmp
|
|
*
|
|
* The working directory of the current process is not changed:
|
|
*
|
|
* Dir.pwd # => "/var"
|
|
*
|
|
* ==== \File Redirection (\File Descriptor)
|
|
*
|
|
* Use execution options for file redirection in the new process.
|
|
*
|
|
* The key for such an option may be an integer file descriptor (fd),
|
|
* specifying a source,
|
|
* or an array of fds, specifying multiple sources.
|
|
|
|
* An integer source fd may be specified as:
|
|
*
|
|
* - _n_: Specifies file descriptor _n_.
|
|
*
|
|
* There are these shorthand symbols for fds:
|
|
*
|
|
* - +:in+: Specifies file descriptor 0 (STDIN).
|
|
* - +:out+: Specifies file descriptor 1 (STDOUT).
|
|
* - +:err+: Specifies file descriptor 2 (STDERR).
|
|
*
|
|
* The value given with a source is one of:
|
|
*
|
|
* - _n_:
|
|
* Redirects to fd _n_ in the parent process.
|
|
* - +filepath+:
|
|
* Redirects from or to the file at +filepath+ via <tt>open(filepath, mode, 0644)</tt>,
|
|
* where +mode+ is <tt>'r'</tt> for source +:in+,
|
|
* or <tt>'w'</tt> for source +:out+ or +:err+.
|
|
* - <tt>[filepath]</tt>:
|
|
* Redirects from the file at +filepath+ via <tt>open(filepath, 'r', 0644)</tt>.
|
|
* - <tt>[filepath, mode]</tt>:
|
|
* Redirects from or to the file at +filepath+ via <tt>open(filepath, mode, 0644)</tt>.
|
|
* - <tt>[filepath, mode, perm]</tt>:
|
|
* Redirects from or to the file at +filepath+ via <tt>open(filepath, mode, perm)</tt>.
|
|
* - <tt>[:child, fd]</tt>:
|
|
* Redirects to the redirected +fd+.
|
|
* - +:close+: Closes the file descriptor in child process.
|
|
*
|
|
* See {Access Modes}[rdoc-ref:File@Access+Modes]
|
|
* and {File Permissions}[rdoc-ref:File@File+Permissions].
|
|
*
|
|
* ==== Environment Variables (+:unsetenv_others+)
|
|
*
|
|
* By default, the new process inherits environment variables
|
|
* from the parent process;
|
|
* use execution option key +:unsetenv_others+ with value +true+
|
|
* to clear environment variables in the new process.
|
|
*
|
|
* Any changes specified by execution option +env+ are made after the new process
|
|
* inherits or clears its environment variables;
|
|
* see {Execution Environment}[rdoc-ref:Process@Execution+Environment].
|
|
*
|
|
* ==== \File-Creation Access (+:umask+)
|
|
*
|
|
* Use execution option +:umask+ to set the file-creation access
|
|
* for the new process;
|
|
* see {Access Modes}[rdoc-ref:File@Access+Modes]:
|
|
*
|
|
* command = 'ruby -e "puts sprintf(\"0%o\", File.umask)"'
|
|
* options = {:umask => 0644}
|
|
* Process.spawn(command, options)
|
|
*
|
|
* Output:
|
|
*
|
|
* 0644
|
|
*
|
|
* ==== \Process Groups (+:pgroup+ and +:new_pgroup+)
|
|
*
|
|
* By default, the new process belongs to the same
|
|
* {process group}[https://en.wikipedia.org/wiki/Process_group]
|
|
* as the parent process.
|
|
*
|
|
* To specify a different process group.
|
|
* use execution option +:pgroup+ with one of the following values:
|
|
*
|
|
* - +true+: Create a new process group for the new process.
|
|
* - _pgid_: Create the new process in the process group
|
|
* whose id is _pgid_.
|
|
*
|
|
* On Windows only, use execution option +:new_pgroup+ with value +true+
|
|
* to create a new process group for the new process.
|
|
*
|
|
* ==== Resource Limits
|
|
*
|
|
* Use execution options to set resource limits.
|
|
*
|
|
* The keys for these options are symbols of the form
|
|
* <tt>:rlimit_<i>resource_name</i></tt>,
|
|
* where _resource_name_ is the downcased form of one of the string
|
|
* resource names described at method Process.setrlimit.
|
|
* For example, key +:rlimit_cpu+ corresponds to resource limit <tt>'CPU'</tt>.
|
|
*
|
|
* The value for such as key is one of:
|
|
*
|
|
* - An integer, specifying both the current and maximum limits.
|
|
* - A 2-element array of integers, specifying the current and maximum limits.
|
|
*
|
|
* ==== \File Descriptor Inheritance
|
|
*
|
|
* By default, the new process inherits file descriptors from the parent process.
|
|
*
|
|
* Use execution option <tt>:close_others => true</tt> to modify that inheritance
|
|
* by closing non-standard fds (3 and greater) that are not otherwise redirected.
|
|
*
|
|
* === Execution Shell
|
|
*
|
|
* On a Unix-like system, the shell invoked is <tt>/bin/sh</tt>;
|
|
* otherwise the shell invoked is determined by environment variable
|
|
* <tt>ENV['RUBYSHELL']</tt>, if defined, or <tt>ENV['COMSPEC']</tt> otherwise.
|
|
*
|
|
* Except for the +COMSPEC+ case,
|
|
* the entire string +command_line+ is passed as an argument
|
|
* to {shell option -c}[https://pubs.opengroup.org/onlinepubs/9699919799.2018edition/utilities/sh.html].
|
|
*
|
|
* The shell performs normal shell expansion on the command line:
|
|
*
|
|
* spawn('echo C*') # => 799139
|
|
* Process.wait # => 799139
|
|
*
|
|
* Output:
|
|
*
|
|
* CONTRIBUTING.md COPYING COPYING.ja
|
|
*
|
|
* == What's Here
|
|
*
|
|
* === Current-Process Getters
|
|
*
|
|
* - ::argv0: Returns the process name as a frozen string.
|
|
* - ::egid: Returns the effective group ID.
|
|
* - ::euid: Returns the effective user ID.
|
|
* - ::getpgrp: Return the process group ID.
|
|
* - ::getrlimit: Returns the resource limit.
|
|
* - ::gid: Returns the (real) group ID.
|
|
* - ::pid: Returns the process ID.
|
|
* - ::ppid: Returns the process ID of the parent process.
|
|
* - ::uid: Returns the (real) user ID.
|
|
*
|
|
* === Current-Process Setters
|
|
*
|
|
* - ::egid=: Sets the effective group ID.
|
|
* - ::euid=: Sets the effective user ID.
|
|
* - ::gid=: Sets the (real) group ID.
|
|
* - ::setproctitle: Sets the process title.
|
|
* - ::setpgrp: Sets the process group ID of the process to zero.
|
|
* - ::setrlimit: Sets a resource limit.
|
|
* - ::setsid: Establishes the process as a new session and process group leader,
|
|
* with no controlling tty.
|
|
* - ::uid=: Sets the user ID.
|
|
*
|
|
* === Current-Process Execution
|
|
*
|
|
* - ::abort: Immediately terminates the process.
|
|
* - ::daemon: Detaches the process from its controlling terminal
|
|
* and continues running it in the background as system daemon.
|
|
* - ::exec: Replaces the process by running a given external command.
|
|
* - ::exit: Initiates process termination by raising exception SystemExit
|
|
* (which may be caught).
|
|
* - ::exit!: Immediately exits the process.
|
|
* - ::warmup: Notifies the Ruby virtual machine that the boot sequence
|
|
* for the application is completed,
|
|
* and that the VM may begin optimizing the application.
|
|
*
|
|
* === Child Processes
|
|
*
|
|
* - ::detach: Guards against a child process becoming a zombie.
|
|
* - ::fork: Creates a child process.
|
|
* - ::kill: Sends a given signal to processes.
|
|
* - ::spawn: Creates a child process.
|
|
* - ::wait, ::waitpid: Waits for a child process to exit; returns its process ID.
|
|
* - ::wait2, ::waitpid2: Waits for a child process to exit; returns its process ID and status.
|
|
* - ::waitall: Waits for all child processes to exit;
|
|
* returns their process IDs and statuses.
|
|
*
|
|
* === \Process Groups
|
|
*
|
|
* - ::getpgid: Returns the process group ID for a process.
|
|
* - ::getpriority: Returns the scheduling priority
|
|
* for a process, process group, or user.
|
|
* - ::getsid: Returns the session ID for a process.
|
|
* - ::groups: Returns an array of the group IDs
|
|
* in the supplemental group access list for this process.
|
|
* - ::groups=: Sets the supplemental group access list
|
|
* to the given array of group IDs.
|
|
* - ::initgroups: Initializes the supplemental group access list.
|
|
* - ::last_status: Returns the status of the last executed child process
|
|
* in the current thread.
|
|
* - ::maxgroups: Returns the maximum number of group IDs allowed
|
|
* in the supplemental group access list.
|
|
* - ::maxgroups=: Sets the maximum number of group IDs allowed
|
|
* in the supplemental group access list.
|
|
* - ::setpgid: Sets the process group ID of a process.
|
|
* - ::setpriority: Sets the scheduling priority
|
|
* for a process, process group, or user.
|
|
*
|
|
* === Timing
|
|
*
|
|
* - ::clock_getres: Returns the resolution of a system clock.
|
|
* - ::clock_gettime: Returns the time from a system clock.
|
|
* - ::times: Returns a Process::Tms object containing times
|
|
* for the current process and its child processes.
|
|
*
|
|
*/
|
|
|
|
void
|
|
InitVM_process(void)
|
|
{
|
|
rb_define_virtual_variable("$?", get_CHILD_STATUS, 0);
|
|
rb_define_virtual_variable("$$", get_PROCESS_ID, 0);
|
|
|
|
rb_gvar_ractor_local("$$");
|
|
rb_gvar_ractor_local("$?");
|
|
|
|
rb_define_global_function("exec", f_exec, -1);
|
|
rb_define_global_function("fork", rb_f_fork, 0);
|
|
rb_define_global_function("exit!", rb_f_exit_bang, -1);
|
|
rb_define_global_function("system", rb_f_system, -1);
|
|
rb_define_global_function("spawn", rb_f_spawn, -1);
|
|
rb_define_global_function("sleep", rb_f_sleep, -1);
|
|
rb_define_global_function("exit", f_exit, -1);
|
|
rb_define_global_function("abort", f_abort, -1);
|
|
|
|
rb_mProcess = rb_define_module("Process");
|
|
|
|
#ifdef WNOHANG
|
|
/* see Process.wait */
|
|
rb_define_const(rb_mProcess, "WNOHANG", INT2FIX(WNOHANG));
|
|
#else
|
|
/* see Process.wait */
|
|
rb_define_const(rb_mProcess, "WNOHANG", INT2FIX(0));
|
|
#endif
|
|
#ifdef WUNTRACED
|
|
/* see Process.wait */
|
|
rb_define_const(rb_mProcess, "WUNTRACED", INT2FIX(WUNTRACED));
|
|
#else
|
|
/* see Process.wait */
|
|
rb_define_const(rb_mProcess, "WUNTRACED", INT2FIX(0));
|
|
#endif
|
|
|
|
rb_define_singleton_method(rb_mProcess, "exec", f_exec, -1);
|
|
rb_define_singleton_method(rb_mProcess, "fork", rb_f_fork, 0);
|
|
rb_define_singleton_method(rb_mProcess, "spawn", rb_f_spawn, -1);
|
|
rb_define_singleton_method(rb_mProcess, "exit!", rb_f_exit_bang, -1);
|
|
rb_define_singleton_method(rb_mProcess, "exit", f_exit, -1);
|
|
rb_define_singleton_method(rb_mProcess, "abort", f_abort, -1);
|
|
rb_define_singleton_method(rb_mProcess, "last_status", proc_s_last_status, 0);
|
|
rb_define_singleton_method(rb_mProcess, "_fork", rb_proc__fork, 0);
|
|
|
|
rb_define_module_function(rb_mProcess, "kill", proc_rb_f_kill, -1);
|
|
rb_define_module_function(rb_mProcess, "wait", proc_m_wait, -1);
|
|
rb_define_module_function(rb_mProcess, "wait2", proc_wait2, -1);
|
|
rb_define_module_function(rb_mProcess, "waitpid", proc_m_wait, -1);
|
|
rb_define_module_function(rb_mProcess, "waitpid2", proc_wait2, -1);
|
|
rb_define_module_function(rb_mProcess, "waitall", proc_waitall, 0);
|
|
rb_define_module_function(rb_mProcess, "detach", proc_detach, 1);
|
|
|
|
/* :nodoc: */
|
|
rb_cWaiter = rb_define_class_under(rb_mProcess, "Waiter", rb_cThread);
|
|
rb_undef_alloc_func(rb_cWaiter);
|
|
rb_undef_method(CLASS_OF(rb_cWaiter), "new");
|
|
rb_define_method(rb_cWaiter, "pid", detach_process_pid, 0);
|
|
|
|
rb_cProcessStatus = rb_define_class_under(rb_mProcess, "Status", rb_cObject);
|
|
rb_define_alloc_func(rb_cProcessStatus, rb_process_status_allocate);
|
|
rb_undef_method(CLASS_OF(rb_cProcessStatus), "new");
|
|
rb_marshal_define_compat(rb_cProcessStatus, rb_cObject,
|
|
process_status_dump, process_status_load);
|
|
|
|
rb_define_singleton_method(rb_cProcessStatus, "wait", rb_process_status_waitv, -1);
|
|
|
|
rb_define_method(rb_cProcessStatus, "==", pst_equal, 1);
|
|
rb_define_method(rb_cProcessStatus, "&", pst_bitand, 1);
|
|
rb_define_method(rb_cProcessStatus, ">>", pst_rshift, 1);
|
|
rb_define_method(rb_cProcessStatus, "to_i", pst_to_i, 0);
|
|
rb_define_method(rb_cProcessStatus, "to_s", pst_to_s, 0);
|
|
rb_define_method(rb_cProcessStatus, "inspect", pst_inspect, 0);
|
|
|
|
rb_define_method(rb_cProcessStatus, "pid", pst_pid_m, 0);
|
|
|
|
rb_define_method(rb_cProcessStatus, "stopped?", pst_wifstopped, 0);
|
|
rb_define_method(rb_cProcessStatus, "stopsig", pst_wstopsig, 0);
|
|
rb_define_method(rb_cProcessStatus, "signaled?", pst_wifsignaled, 0);
|
|
rb_define_method(rb_cProcessStatus, "termsig", pst_wtermsig, 0);
|
|
rb_define_method(rb_cProcessStatus, "exited?", pst_wifexited, 0);
|
|
rb_define_method(rb_cProcessStatus, "exitstatus", pst_wexitstatus, 0);
|
|
rb_define_method(rb_cProcessStatus, "success?", pst_success_p, 0);
|
|
rb_define_method(rb_cProcessStatus, "coredump?", pst_wcoredump, 0);
|
|
|
|
rb_define_module_function(rb_mProcess, "pid", proc_get_pid, 0);
|
|
rb_define_module_function(rb_mProcess, "ppid", proc_get_ppid, 0);
|
|
|
|
rb_define_module_function(rb_mProcess, "getpgrp", proc_getpgrp, 0);
|
|
rb_define_module_function(rb_mProcess, "setpgrp", proc_setpgrp, 0);
|
|
rb_define_module_function(rb_mProcess, "getpgid", proc_getpgid, 1);
|
|
rb_define_module_function(rb_mProcess, "setpgid", proc_setpgid, 2);
|
|
|
|
rb_define_module_function(rb_mProcess, "getsid", proc_getsid, -1);
|
|
rb_define_module_function(rb_mProcess, "setsid", proc_setsid, 0);
|
|
|
|
rb_define_module_function(rb_mProcess, "getpriority", proc_getpriority, 2);
|
|
rb_define_module_function(rb_mProcess, "setpriority", proc_setpriority, 3);
|
|
|
|
rb_define_module_function(rb_mProcess, "warmup", proc_warmup, 0);
|
|
|
|
#ifdef HAVE_GETPRIORITY
|
|
/* see Process.setpriority */
|
|
rb_define_const(rb_mProcess, "PRIO_PROCESS", INT2FIX(PRIO_PROCESS));
|
|
/* see Process.setpriority */
|
|
rb_define_const(rb_mProcess, "PRIO_PGRP", INT2FIX(PRIO_PGRP));
|
|
/* see Process.setpriority */
|
|
rb_define_const(rb_mProcess, "PRIO_USER", INT2FIX(PRIO_USER));
|
|
#endif
|
|
|
|
rb_define_module_function(rb_mProcess, "getrlimit", proc_getrlimit, 1);
|
|
rb_define_module_function(rb_mProcess, "setrlimit", proc_setrlimit, -1);
|
|
#if defined(RLIM2NUM) && defined(RLIM_INFINITY)
|
|
{
|
|
VALUE inf = RLIM2NUM(RLIM_INFINITY);
|
|
#ifdef RLIM_SAVED_MAX
|
|
{
|
|
VALUE v = RLIM_INFINITY == RLIM_SAVED_MAX ? inf : RLIM2NUM(RLIM_SAVED_MAX);
|
|
/* see Process.setrlimit */
|
|
rb_define_const(rb_mProcess, "RLIM_SAVED_MAX", v);
|
|
}
|
|
#endif
|
|
/* see Process.setrlimit */
|
|
rb_define_const(rb_mProcess, "RLIM_INFINITY", inf);
|
|
#ifdef RLIM_SAVED_CUR
|
|
{
|
|
VALUE v = RLIM_INFINITY == RLIM_SAVED_CUR ? inf : RLIM2NUM(RLIM_SAVED_CUR);
|
|
/* see Process.setrlimit */
|
|
rb_define_const(rb_mProcess, "RLIM_SAVED_CUR", v);
|
|
}
|
|
#endif
|
|
}
|
|
#ifdef RLIMIT_AS
|
|
/* Maximum size of the process's virtual memory (address space) in bytes.
|
|
*
|
|
* see the system getrlimit(2) manual for details.
|
|
*/
|
|
rb_define_const(rb_mProcess, "RLIMIT_AS", INT2FIX(RLIMIT_AS));
|
|
#endif
|
|
#ifdef RLIMIT_CORE
|
|
/* Maximum size of the core file.
|
|
*
|
|
* see the system getrlimit(2) manual for details.
|
|
*/
|
|
rb_define_const(rb_mProcess, "RLIMIT_CORE", INT2FIX(RLIMIT_CORE));
|
|
#endif
|
|
#ifdef RLIMIT_CPU
|
|
/* CPU time limit in seconds.
|
|
*
|
|
* see the system getrlimit(2) manual for details.
|
|
*/
|
|
rb_define_const(rb_mProcess, "RLIMIT_CPU", INT2FIX(RLIMIT_CPU));
|
|
#endif
|
|
#ifdef RLIMIT_DATA
|
|
/* Maximum size of the process's data segment.
|
|
*
|
|
* see the system getrlimit(2) manual for details.
|
|
*/
|
|
rb_define_const(rb_mProcess, "RLIMIT_DATA", INT2FIX(RLIMIT_DATA));
|
|
#endif
|
|
#ifdef RLIMIT_FSIZE
|
|
/* Maximum size of files that the process may create.
|
|
*
|
|
* see the system getrlimit(2) manual for details.
|
|
*/
|
|
rb_define_const(rb_mProcess, "RLIMIT_FSIZE", INT2FIX(RLIMIT_FSIZE));
|
|
#endif
|
|
#ifdef RLIMIT_MEMLOCK
|
|
/* Maximum number of bytes of memory that may be locked into RAM.
|
|
*
|
|
* see the system getrlimit(2) manual for details.
|
|
*/
|
|
rb_define_const(rb_mProcess, "RLIMIT_MEMLOCK", INT2FIX(RLIMIT_MEMLOCK));
|
|
#endif
|
|
#ifdef RLIMIT_MSGQUEUE
|
|
/* Specifies the limit on the number of bytes that can be allocated
|
|
* for POSIX message queues for the real user ID of the calling process.
|
|
*
|
|
* see the system getrlimit(2) manual for details.
|
|
*/
|
|
rb_define_const(rb_mProcess, "RLIMIT_MSGQUEUE", INT2FIX(RLIMIT_MSGQUEUE));
|
|
#endif
|
|
#ifdef RLIMIT_NICE
|
|
/* Specifies a ceiling to which the process's nice value can be raised.
|
|
*
|
|
* see the system getrlimit(2) manual for details.
|
|
*/
|
|
rb_define_const(rb_mProcess, "RLIMIT_NICE", INT2FIX(RLIMIT_NICE));
|
|
#endif
|
|
#ifdef RLIMIT_NOFILE
|
|
/* Specifies a value one greater than the maximum file descriptor
|
|
* number that can be opened by this process.
|
|
*
|
|
* see the system getrlimit(2) manual for details.
|
|
*/
|
|
rb_define_const(rb_mProcess, "RLIMIT_NOFILE", INT2FIX(RLIMIT_NOFILE));
|
|
#endif
|
|
#ifdef RLIMIT_NPROC
|
|
/* The maximum number of processes that can be created for the
|
|
* real user ID of the calling process.
|
|
*
|
|
* see the system getrlimit(2) manual for details.
|
|
*/
|
|
rb_define_const(rb_mProcess, "RLIMIT_NPROC", INT2FIX(RLIMIT_NPROC));
|
|
#endif
|
|
#ifdef RLIMIT_NPTS
|
|
/* The maximum number of pseudo-terminals that can be created for the
|
|
* real user ID of the calling process.
|
|
*
|
|
* see the system getrlimit(2) manual for details.
|
|
*/
|
|
rb_define_const(rb_mProcess, "RLIMIT_NPTS", INT2FIX(RLIMIT_NPTS));
|
|
#endif
|
|
#ifdef RLIMIT_RSS
|
|
/* Specifies the limit (in pages) of the process's resident set.
|
|
*
|
|
* see the system getrlimit(2) manual for details.
|
|
*/
|
|
rb_define_const(rb_mProcess, "RLIMIT_RSS", INT2FIX(RLIMIT_RSS));
|
|
#endif
|
|
#ifdef RLIMIT_RTPRIO
|
|
/* Specifies a ceiling on the real-time priority that may be set for this process.
|
|
*
|
|
* see the system getrlimit(2) manual for details.
|
|
*/
|
|
rb_define_const(rb_mProcess, "RLIMIT_RTPRIO", INT2FIX(RLIMIT_RTPRIO));
|
|
#endif
|
|
#ifdef RLIMIT_RTTIME
|
|
/* Specifies limit on CPU time this process scheduled under a real-time
|
|
* scheduling policy can consume.
|
|
*
|
|
* see the system getrlimit(2) manual for details.
|
|
*/
|
|
rb_define_const(rb_mProcess, "RLIMIT_RTTIME", INT2FIX(RLIMIT_RTTIME));
|
|
#endif
|
|
#ifdef RLIMIT_SBSIZE
|
|
/* Maximum size of the socket buffer.
|
|
*/
|
|
rb_define_const(rb_mProcess, "RLIMIT_SBSIZE", INT2FIX(RLIMIT_SBSIZE));
|
|
#endif
|
|
#ifdef RLIMIT_SIGPENDING
|
|
/* Specifies a limit on the number of signals that may be queued for
|
|
* the real user ID of the calling process.
|
|
*
|
|
* see the system getrlimit(2) manual for details.
|
|
*/
|
|
rb_define_const(rb_mProcess, "RLIMIT_SIGPENDING", INT2FIX(RLIMIT_SIGPENDING));
|
|
#endif
|
|
#ifdef RLIMIT_STACK
|
|
/* Maximum size of the stack, in bytes.
|
|
*
|
|
* see the system getrlimit(2) manual for details.
|
|
*/
|
|
rb_define_const(rb_mProcess, "RLIMIT_STACK", INT2FIX(RLIMIT_STACK));
|
|
#endif
|
|
#endif
|
|
|
|
rb_define_module_function(rb_mProcess, "uid", proc_getuid, 0);
|
|
rb_define_module_function(rb_mProcess, "uid=", proc_setuid, 1);
|
|
rb_define_module_function(rb_mProcess, "gid", proc_getgid, 0);
|
|
rb_define_module_function(rb_mProcess, "gid=", proc_setgid, 1);
|
|
rb_define_module_function(rb_mProcess, "euid", proc_geteuid, 0);
|
|
rb_define_module_function(rb_mProcess, "euid=", proc_seteuid_m, 1);
|
|
rb_define_module_function(rb_mProcess, "egid", proc_getegid, 0);
|
|
rb_define_module_function(rb_mProcess, "egid=", proc_setegid_m, 1);
|
|
rb_define_module_function(rb_mProcess, "initgroups", proc_initgroups, 2);
|
|
rb_define_module_function(rb_mProcess, "groups", proc_getgroups, 0);
|
|
rb_define_module_function(rb_mProcess, "groups=", proc_setgroups, 1);
|
|
rb_define_module_function(rb_mProcess, "maxgroups", proc_getmaxgroups, 0);
|
|
rb_define_module_function(rb_mProcess, "maxgroups=", proc_setmaxgroups, 1);
|
|
|
|
rb_define_module_function(rb_mProcess, "daemon", proc_daemon, -1);
|
|
|
|
rb_define_module_function(rb_mProcess, "times", rb_proc_times, 0);
|
|
|
|
#if defined(RUBY_CLOCK_REALTIME)
|
|
#elif defined(RUBY_GETTIMEOFDAY_BASED_CLOCK_REALTIME)
|
|
# define RUBY_CLOCK_REALTIME RUBY_GETTIMEOFDAY_BASED_CLOCK_REALTIME
|
|
#elif defined(RUBY_TIME_BASED_CLOCK_REALTIME)
|
|
# define RUBY_CLOCK_REALTIME RUBY_TIME_BASED_CLOCK_REALTIME
|
|
#endif
|
|
#if defined(CLOCK_REALTIME) && defined(CLOCKID2NUM)
|
|
/* see Process.clock_gettime */
|
|
rb_define_const(rb_mProcess, "CLOCK_REALTIME", CLOCKID2NUM(CLOCK_REALTIME));
|
|
#elif defined(RUBY_CLOCK_REALTIME)
|
|
rb_define_const(rb_mProcess, "CLOCK_REALTIME", RUBY_CLOCK_REALTIME);
|
|
#endif
|
|
|
|
#if defined(RUBY_CLOCK_MONOTONIC)
|
|
#elif defined(RUBY_MACH_ABSOLUTE_TIME_BASED_CLOCK_MONOTONIC)
|
|
# define RUBY_CLOCK_MONOTONIC RUBY_MACH_ABSOLUTE_TIME_BASED_CLOCK_MONOTONIC
|
|
#endif
|
|
#if defined(CLOCK_MONOTONIC) && defined(CLOCKID2NUM)
|
|
/* see Process.clock_gettime */
|
|
rb_define_const(rb_mProcess, "CLOCK_MONOTONIC", CLOCKID2NUM(CLOCK_MONOTONIC));
|
|
#elif defined(RUBY_CLOCK_MONOTONIC)
|
|
rb_define_const(rb_mProcess, "CLOCK_MONOTONIC", RUBY_CLOCK_MONOTONIC);
|
|
#endif
|
|
|
|
#if defined(RUBY_CLOCK_PROCESS_CPUTIME_ID)
|
|
#elif defined(RUBY_GETRUSAGE_BASED_CLOCK_PROCESS_CPUTIME_ID)
|
|
# define RUBY_CLOCK_PROCESS_CPUTIME_ID RUBY_GETRUSAGE_BASED_CLOCK_PROCESS_CPUTIME_ID
|
|
#endif
|
|
#if defined(CLOCK_PROCESS_CPUTIME_ID) && defined(CLOCKID2NUM)
|
|
/* see Process.clock_gettime */
|
|
rb_define_const(rb_mProcess, "CLOCK_PROCESS_CPUTIME_ID", CLOCKID2NUM(CLOCK_PROCESS_CPUTIME_ID));
|
|
#elif defined(RUBY_CLOCK_PROCESS_CPUTIME_ID)
|
|
rb_define_const(rb_mProcess, "CLOCK_PROCESS_CPUTIME_ID", RUBY_CLOCK_PROCESS_CPUTIME_ID);
|
|
#endif
|
|
|
|
#if defined(CLOCK_THREAD_CPUTIME_ID) && defined(CLOCKID2NUM)
|
|
/* see Process.clock_gettime */
|
|
rb_define_const(rb_mProcess, "CLOCK_THREAD_CPUTIME_ID", CLOCKID2NUM(CLOCK_THREAD_CPUTIME_ID));
|
|
#elif defined(RUBY_CLOCK_THREAD_CPUTIME_ID)
|
|
rb_define_const(rb_mProcess, "CLOCK_THREAD_CPUTIME_ID", RUBY_CLOCK_THREAD_CPUTIME_ID);
|
|
#endif
|
|
|
|
#ifdef CLOCKID2NUM
|
|
#ifdef CLOCK_VIRTUAL
|
|
/* see Process.clock_gettime */
|
|
rb_define_const(rb_mProcess, "CLOCK_VIRTUAL", CLOCKID2NUM(CLOCK_VIRTUAL));
|
|
#endif
|
|
#ifdef CLOCK_PROF
|
|
/* see Process.clock_gettime */
|
|
rb_define_const(rb_mProcess, "CLOCK_PROF", CLOCKID2NUM(CLOCK_PROF));
|
|
#endif
|
|
#ifdef CLOCK_REALTIME_FAST
|
|
/* see Process.clock_gettime */
|
|
rb_define_const(rb_mProcess, "CLOCK_REALTIME_FAST", CLOCKID2NUM(CLOCK_REALTIME_FAST));
|
|
#endif
|
|
#ifdef CLOCK_REALTIME_PRECISE
|
|
/* see Process.clock_gettime */
|
|
rb_define_const(rb_mProcess, "CLOCK_REALTIME_PRECISE", CLOCKID2NUM(CLOCK_REALTIME_PRECISE));
|
|
#endif
|
|
#ifdef CLOCK_REALTIME_COARSE
|
|
/* see Process.clock_gettime */
|
|
rb_define_const(rb_mProcess, "CLOCK_REALTIME_COARSE", CLOCKID2NUM(CLOCK_REALTIME_COARSE));
|
|
#endif
|
|
#ifdef CLOCK_REALTIME_ALARM
|
|
/* see Process.clock_gettime */
|
|
rb_define_const(rb_mProcess, "CLOCK_REALTIME_ALARM", CLOCKID2NUM(CLOCK_REALTIME_ALARM));
|
|
#endif
|
|
#ifdef CLOCK_MONOTONIC_FAST
|
|
/* see Process.clock_gettime */
|
|
rb_define_const(rb_mProcess, "CLOCK_MONOTONIC_FAST", CLOCKID2NUM(CLOCK_MONOTONIC_FAST));
|
|
#endif
|
|
#ifdef CLOCK_MONOTONIC_PRECISE
|
|
/* see Process.clock_gettime */
|
|
rb_define_const(rb_mProcess, "CLOCK_MONOTONIC_PRECISE", CLOCKID2NUM(CLOCK_MONOTONIC_PRECISE));
|
|
#endif
|
|
#ifdef CLOCK_MONOTONIC_RAW
|
|
/* see Process.clock_gettime */
|
|
rb_define_const(rb_mProcess, "CLOCK_MONOTONIC_RAW", CLOCKID2NUM(CLOCK_MONOTONIC_RAW));
|
|
#endif
|
|
#ifdef CLOCK_MONOTONIC_RAW_APPROX
|
|
/* see Process.clock_gettime */
|
|
rb_define_const(rb_mProcess, "CLOCK_MONOTONIC_RAW_APPROX", CLOCKID2NUM(CLOCK_MONOTONIC_RAW_APPROX));
|
|
#endif
|
|
#ifdef CLOCK_MONOTONIC_COARSE
|
|
/* see Process.clock_gettime */
|
|
rb_define_const(rb_mProcess, "CLOCK_MONOTONIC_COARSE", CLOCKID2NUM(CLOCK_MONOTONIC_COARSE));
|
|
#endif
|
|
#ifdef CLOCK_BOOTTIME
|
|
/* see Process.clock_gettime */
|
|
rb_define_const(rb_mProcess, "CLOCK_BOOTTIME", CLOCKID2NUM(CLOCK_BOOTTIME));
|
|
#endif
|
|
#ifdef CLOCK_BOOTTIME_ALARM
|
|
/* see Process.clock_gettime */
|
|
rb_define_const(rb_mProcess, "CLOCK_BOOTTIME_ALARM", CLOCKID2NUM(CLOCK_BOOTTIME_ALARM));
|
|
#endif
|
|
#ifdef CLOCK_UPTIME
|
|
/* see Process.clock_gettime */
|
|
rb_define_const(rb_mProcess, "CLOCK_UPTIME", CLOCKID2NUM(CLOCK_UPTIME));
|
|
#endif
|
|
#ifdef CLOCK_UPTIME_FAST
|
|
/* see Process.clock_gettime */
|
|
rb_define_const(rb_mProcess, "CLOCK_UPTIME_FAST", CLOCKID2NUM(CLOCK_UPTIME_FAST));
|
|
#endif
|
|
#ifdef CLOCK_UPTIME_PRECISE
|
|
/* see Process.clock_gettime */
|
|
rb_define_const(rb_mProcess, "CLOCK_UPTIME_PRECISE", CLOCKID2NUM(CLOCK_UPTIME_PRECISE));
|
|
#endif
|
|
#ifdef CLOCK_UPTIME_RAW
|
|
/* see Process.clock_gettime */
|
|
rb_define_const(rb_mProcess, "CLOCK_UPTIME_RAW", CLOCKID2NUM(CLOCK_UPTIME_RAW));
|
|
#endif
|
|
#ifdef CLOCK_UPTIME_RAW_APPROX
|
|
/* see Process.clock_gettime */
|
|
rb_define_const(rb_mProcess, "CLOCK_UPTIME_RAW_APPROX", CLOCKID2NUM(CLOCK_UPTIME_RAW_APPROX));
|
|
#endif
|
|
#ifdef CLOCK_SECOND
|
|
/* see Process.clock_gettime */
|
|
rb_define_const(rb_mProcess, "CLOCK_SECOND", CLOCKID2NUM(CLOCK_SECOND));
|
|
#endif
|
|
#ifdef CLOCK_TAI
|
|
/* see Process.clock_gettime */
|
|
rb_define_const(rb_mProcess, "CLOCK_TAI", CLOCKID2NUM(CLOCK_TAI));
|
|
#endif
|
|
#endif
|
|
rb_define_module_function(rb_mProcess, "clock_gettime", rb_clock_gettime, -1);
|
|
rb_define_module_function(rb_mProcess, "clock_getres", rb_clock_getres, -1);
|
|
|
|
#if defined(HAVE_TIMES) || defined(_WIN32)
|
|
rb_cProcessTms = rb_struct_define_under(rb_mProcess, "Tms", "utime", "stime", "cutime", "cstime", NULL);
|
|
#if 0 /* for RDoc */
|
|
/* user time used in this process */
|
|
rb_define_attr(rb_cProcessTms, "utime", TRUE, TRUE);
|
|
/* system time used in this process */
|
|
rb_define_attr(rb_cProcessTms, "stime", TRUE, TRUE);
|
|
/* user time used in the child processes */
|
|
rb_define_attr(rb_cProcessTms, "cutime", TRUE, TRUE);
|
|
/* system time used in the child processes */
|
|
rb_define_attr(rb_cProcessTms, "cstime", TRUE, TRUE);
|
|
#endif
|
|
#endif
|
|
|
|
SAVED_USER_ID = geteuid();
|
|
SAVED_GROUP_ID = getegid();
|
|
|
|
rb_mProcUID = rb_define_module_under(rb_mProcess, "UID");
|
|
rb_mProcGID = rb_define_module_under(rb_mProcess, "GID");
|
|
|
|
rb_define_module_function(rb_mProcUID, "rid", proc_getuid, 0);
|
|
rb_define_module_function(rb_mProcGID, "rid", proc_getgid, 0);
|
|
rb_define_module_function(rb_mProcUID, "eid", proc_geteuid, 0);
|
|
rb_define_module_function(rb_mProcGID, "eid", proc_getegid, 0);
|
|
rb_define_module_function(rb_mProcUID, "change_privilege", p_uid_change_privilege, 1);
|
|
rb_define_module_function(rb_mProcGID, "change_privilege", p_gid_change_privilege, 1);
|
|
rb_define_module_function(rb_mProcUID, "grant_privilege", p_uid_grant_privilege, 1);
|
|
rb_define_module_function(rb_mProcGID, "grant_privilege", p_gid_grant_privilege, 1);
|
|
rb_define_alias(rb_singleton_class(rb_mProcUID), "eid=", "grant_privilege");
|
|
rb_define_alias(rb_singleton_class(rb_mProcGID), "eid=", "grant_privilege");
|
|
rb_define_module_function(rb_mProcUID, "re_exchange", p_uid_exchange, 0);
|
|
rb_define_module_function(rb_mProcGID, "re_exchange", p_gid_exchange, 0);
|
|
rb_define_module_function(rb_mProcUID, "re_exchangeable?", p_uid_exchangeable, 0);
|
|
rb_define_module_function(rb_mProcGID, "re_exchangeable?", p_gid_exchangeable, 0);
|
|
rb_define_module_function(rb_mProcUID, "sid_available?", p_uid_have_saved_id, 0);
|
|
rb_define_module_function(rb_mProcGID, "sid_available?", p_gid_have_saved_id, 0);
|
|
rb_define_module_function(rb_mProcUID, "switch", p_uid_switch, 0);
|
|
rb_define_module_function(rb_mProcGID, "switch", p_gid_switch, 0);
|
|
#ifdef p_uid_from_name
|
|
rb_define_module_function(rb_mProcUID, "from_name", p_uid_from_name, 1);
|
|
#endif
|
|
#ifdef p_gid_from_name
|
|
rb_define_module_function(rb_mProcGID, "from_name", p_gid_from_name, 1);
|
|
#endif
|
|
|
|
rb_mProcID_Syscall = rb_define_module_under(rb_mProcess, "Sys");
|
|
|
|
rb_define_module_function(rb_mProcID_Syscall, "getuid", proc_getuid, 0);
|
|
rb_define_module_function(rb_mProcID_Syscall, "geteuid", proc_geteuid, 0);
|
|
rb_define_module_function(rb_mProcID_Syscall, "getgid", proc_getgid, 0);
|
|
rb_define_module_function(rb_mProcID_Syscall, "getegid", proc_getegid, 0);
|
|
|
|
rb_define_module_function(rb_mProcID_Syscall, "setuid", p_sys_setuid, 1);
|
|
rb_define_module_function(rb_mProcID_Syscall, "setgid", p_sys_setgid, 1);
|
|
|
|
rb_define_module_function(rb_mProcID_Syscall, "setruid", p_sys_setruid, 1);
|
|
rb_define_module_function(rb_mProcID_Syscall, "setrgid", p_sys_setrgid, 1);
|
|
|
|
rb_define_module_function(rb_mProcID_Syscall, "seteuid", p_sys_seteuid, 1);
|
|
rb_define_module_function(rb_mProcID_Syscall, "setegid", p_sys_setegid, 1);
|
|
|
|
rb_define_module_function(rb_mProcID_Syscall, "setreuid", p_sys_setreuid, 2);
|
|
rb_define_module_function(rb_mProcID_Syscall, "setregid", p_sys_setregid, 2);
|
|
|
|
rb_define_module_function(rb_mProcID_Syscall, "setresuid", p_sys_setresuid, 3);
|
|
rb_define_module_function(rb_mProcID_Syscall, "setresgid", p_sys_setresgid, 3);
|
|
rb_define_module_function(rb_mProcID_Syscall, "issetugid", p_sys_issetugid, 0);
|
|
}
|
|
|
|
void
|
|
Init_process(void)
|
|
{
|
|
#define define_id(name) id_##name = rb_intern_const(#name)
|
|
define_id(in);
|
|
define_id(out);
|
|
define_id(err);
|
|
define_id(pid);
|
|
define_id(uid);
|
|
define_id(gid);
|
|
define_id(close);
|
|
define_id(child);
|
|
#ifdef HAVE_SETPGID
|
|
define_id(pgroup);
|
|
#endif
|
|
#ifdef _WIN32
|
|
define_id(new_pgroup);
|
|
#endif
|
|
define_id(unsetenv_others);
|
|
define_id(chdir);
|
|
define_id(umask);
|
|
define_id(close_others);
|
|
define_id(nanosecond);
|
|
define_id(microsecond);
|
|
define_id(millisecond);
|
|
define_id(second);
|
|
define_id(float_microsecond);
|
|
define_id(float_millisecond);
|
|
define_id(float_second);
|
|
define_id(GETTIMEOFDAY_BASED_CLOCK_REALTIME);
|
|
define_id(TIME_BASED_CLOCK_REALTIME);
|
|
#ifdef CLOCK_REALTIME
|
|
define_id(CLOCK_REALTIME);
|
|
#endif
|
|
#ifdef CLOCK_MONOTONIC
|
|
define_id(CLOCK_MONOTONIC);
|
|
#endif
|
|
#ifdef CLOCK_PROCESS_CPUTIME_ID
|
|
define_id(CLOCK_PROCESS_CPUTIME_ID);
|
|
#endif
|
|
#ifdef CLOCK_THREAD_CPUTIME_ID
|
|
define_id(CLOCK_THREAD_CPUTIME_ID);
|
|
#endif
|
|
#ifdef HAVE_TIMES
|
|
define_id(TIMES_BASED_CLOCK_MONOTONIC);
|
|
define_id(TIMES_BASED_CLOCK_PROCESS_CPUTIME_ID);
|
|
#endif
|
|
#ifdef RUSAGE_SELF
|
|
define_id(GETRUSAGE_BASED_CLOCK_PROCESS_CPUTIME_ID);
|
|
#endif
|
|
define_id(CLOCK_BASED_CLOCK_PROCESS_CPUTIME_ID);
|
|
#ifdef __APPLE__
|
|
define_id(MACH_ABSOLUTE_TIME_BASED_CLOCK_MONOTONIC);
|
|
#endif
|
|
define_id(hertz);
|
|
|
|
InitVM(process);
|
|
}
|