ruby/ext/pty/pty.c
Jeremy Evans a3f5a043fa Handle getlogin failure in PTY.spawn
getlogin is only called if USER environment variable is not set,
but if getlogin returns NULL in that case, then do not call
getpwnam, and assume /bin/sh as shell.

Mentioned in comment to bug 20586.
2024-08-22 11:20:47 +09:00

823 lines
22 KiB
C

#include "ruby/config.h"
#ifdef RUBY_EXTCONF_H
# include RUBY_EXTCONF_H
#endif
#include <ctype.h>
#include <errno.h>
#include <signal.h>
#include <stdio.h>
#include <stdlib.h>
#include <sys/types.h>
#include <sys/stat.h>
#include <sys/file.h>
#include <fcntl.h>
#ifdef HAVE_PWD_H
# include <pwd.h>
#endif
#ifdef HAVE_SYS_IOCTL_H
# include <sys/ioctl.h>
#endif
#ifdef HAVE_LIBUTIL_H
# include <libutil.h>
#endif
#ifdef HAVE_UTIL_H
# include <util.h>
#endif
#ifdef HAVE_PTY_H
# include <pty.h>
#endif
#if defined(HAVE_SYS_PARAM_H)
/* for __FreeBSD_version */
# include <sys/param.h>
#endif
#ifdef HAVE_SYS_WAIT_H
# include <sys/wait.h>
#else
# define WIFSTOPPED(status) (((status) & 0xff) == 0x7f)
#endif
#ifdef HAVE_SYS_STROPTS_H
#include <sys/stropts.h>
#endif
#ifdef HAVE_UNISTD_H
#include <unistd.h>
#endif
#include "internal.h"
#include "internal/process.h"
#include "internal/signal.h"
#include "ruby/io.h"
#include "ruby/util.h"
#define DEVICELEN 16
#ifndef HAVE_SETEUID
# ifdef HAVE_SETREUID
# define seteuid(e) setreuid(-1, (e))
# else /* NOT HAVE_SETREUID */
# ifdef HAVE_SETRESUID
# define seteuid(e) setresuid(-1, (e), -1)
# else /* NOT HAVE_SETRESUID */
/* I can't set euid. (;_;) */
# endif /* HAVE_SETRESUID */
# endif /* HAVE_SETREUID */
#endif /* NO_SETEUID */
static VALUE eChildExited;
/* Returns the exit status of the child for which PTY#check
* raised this exception
*/
static VALUE
echild_status(VALUE self)
{
return rb_ivar_get(self, rb_intern("status"));
}
struct pty_info {
int fd;
rb_pid_t child_pid;
};
static void getDevice(int*, int*, char [DEVICELEN], int);
static int start_new_session(char *errbuf, size_t errbuf_len);
static int obtain_ctty(int master, int slave, const char *slavename, char *errbuf, size_t errbuf_len);
static int drop_privilige(char *errbuf, size_t errbuf_len);
struct child_info {
int master, slave;
const char *slavename;
VALUE execarg_obj;
struct rb_execarg *eargp;
};
static int
chfunc(void *data, char *errbuf, size_t errbuf_len)
{
const struct child_info *carg = data;
int master = carg->master;
int slave = carg->slave;
const char *slavename = carg->slavename;
if (start_new_session(errbuf, errbuf_len))
return -1;
if (obtain_ctty(master, slave, slavename, errbuf, errbuf_len))
return -1;
if (drop_privilige(errbuf, errbuf_len))
return -1;
return rb_exec_async_signal_safe(carg->eargp, errbuf, errbuf_len);
}
#define ERROR_EXIT(str) do { \
strlcpy(errbuf, (str), errbuf_len); \
return -1; \
} while (0)
/*
* Set free from process group and controlling terminal
*/
static int
start_new_session(char *errbuf, size_t errbuf_len)
{
#ifdef HAVE_SETSID
(void) setsid();
#else /* HAS_SETSID */
# ifdef HAVE_SETPGRP
# ifdef SETPGRP_VOID
if (setpgrp() == -1)
ERROR_EXIT("setpgrp()");
# else /* SETPGRP_VOID */
if (setpgrp(0, getpid()) == -1)
ERROR_EXIT("setpgrp()");
{
int i = rb_cloexec_open("/dev/tty", O_RDONLY, 0);
if (i < 0) ERROR_EXIT("/dev/tty");
rb_update_max_fd(i);
if (ioctl(i, TIOCNOTTY, (char *)0))
ERROR_EXIT("ioctl(TIOCNOTTY)");
close(i);
}
# endif /* SETPGRP_VOID */
# endif /* HAVE_SETPGRP */
#endif /* HAS_SETSID */
return 0;
}
/*
* obtain new controlling terminal
*/
static int
obtain_ctty(int master, int slave, const char *slavename, char *errbuf, size_t errbuf_len)
{
#if defined(TIOCSCTTY)
close(master);
(void) ioctl(slave, TIOCSCTTY, (char *)0);
/* errors ignored for sun */
#else
close(slave);
slave = rb_cloexec_open(slavename, O_RDWR, 0);
if (slave < 0) {
ERROR_EXIT("open: pty slave");
}
rb_update_max_fd(slave);
close(master);
#endif
dup2(slave,0);
dup2(slave,1);
dup2(slave,2);
if (slave < 0 || slave > 2) (void)!close(slave);
return 0;
}
static int
drop_privilige(char *errbuf, size_t errbuf_len)
{
#if defined(HAVE_SETEUID) || defined(HAVE_SETREUID) || defined(HAVE_SETRESUID)
if (seteuid(getuid())) ERROR_EXIT("seteuid()");
#endif
return 0;
}
#undef ERROR_EXIT
static void
establishShell(int argc, VALUE *argv, struct pty_info *info,
char SlaveName[DEVICELEN])
{
int master, slave, status = 0;
rb_pid_t pid;
char *p;
VALUE v;
struct child_info carg;
char errbuf[32];
if (argc == 0) {
const char *shellname = "/bin/sh";
if ((p = getenv("SHELL")) != NULL) {
shellname = p;
}
else {
#if defined HAVE_PWD_H
const char *username = getenv("USER");
if (username == NULL)
username = getlogin();
if (username != NULL) {
struct passwd *pwent = getpwnam(username);
if (pwent && pwent->pw_shell)
shellname = pwent->pw_shell;
}
#endif
}
v = rb_str_new2(shellname);
argc = 1;
argv = &v;
}
carg.execarg_obj = rb_execarg_new(argc, argv, 1, 0);
carg.eargp = rb_execarg_get(carg.execarg_obj);
rb_execarg_parent_start(carg.execarg_obj);
getDevice(&master, &slave, SlaveName, 0);
carg.master = master;
carg.slave = slave;
carg.slavename = SlaveName;
errbuf[0] = '\0';
pid = rb_fork_async_signal_safe(&status, chfunc, &carg, Qnil, errbuf, sizeof(errbuf));
if (pid < 0) {
int e = errno;
close(master);
close(slave);
rb_execarg_parent_end(carg.execarg_obj);
errno = e;
if (status) rb_jump_tag(status);
rb_sys_fail(errbuf[0] ? errbuf : "fork failed");
}
close(slave);
rb_execarg_parent_end(carg.execarg_obj);
info->child_pid = pid;
info->fd = master;
RB_GC_GUARD(carg.execarg_obj);
}
#if (defined(HAVE_POSIX_OPENPT) || defined(HAVE_PTSNAME)) && !defined(HAVE_PTSNAME_R)
/* glibc only, not obsolete interface on Tru64 or HP-UX */
static int
ptsname_r(int fd, char *buf, size_t buflen)
{
extern char *ptsname(int);
char *name = ptsname(fd);
if (!name) return -1;
strlcpy(buf, name, buflen);
return 0;
}
# define HAVE_PTSNAME_R 1
#endif
#if defined(HAVE_POSIX_OPENPT) || defined(HAVE_OPENPTY) || defined(HAVE_PTSNAME_R)
static int
no_mesg(char *slavedevice, int nomesg)
{
if (nomesg)
return chmod(slavedevice, 0600);
else
return 0;
}
#endif
#if defined(I_PUSH) && !defined(__linux__) && !defined(_AIX)
static inline int
ioctl_I_PUSH(int fd, const char *const name)
{
int ret = 0;
# if defined(I_FIND)
ret = ioctl(fd, I_FIND, name);
# endif
if (ret == 0) {
ret = ioctl(fd, I_PUSH, name);
}
return ret;
}
#endif
static int
get_device_once(int *master, int *slave, char SlaveName[DEVICELEN], int nomesg, int fail)
{
#if defined(HAVE_POSIX_OPENPT)
/* Unix98 PTY */
int masterfd = -1, slavefd = -1;
char *slavedevice;
struct sigaction dfl, old;
dfl.sa_handler = SIG_DFL;
dfl.sa_flags = 0;
sigemptyset(&dfl.sa_mask);
#if defined(__sun) || defined(__OpenBSD__) || (defined(__FreeBSD__) && __FreeBSD_version < 902000)
/* workaround for Solaris 10: grantpt() doesn't work if FD_CLOEXEC is set. [ruby-dev:44688] */
/* FreeBSD 9.2 or later supports O_CLOEXEC
* http://www.freebsd.org/cgi/query-pr.cgi?pr=162374 */
if ((masterfd = posix_openpt(O_RDWR|O_NOCTTY)) == -1) goto error;
if (sigaction(SIGCHLD, &dfl, &old) == -1) goto error;
if (grantpt(masterfd) == -1) goto grantpt_error;
rb_fd_fix_cloexec(masterfd);
#else
{
int flags = O_RDWR|O_NOCTTY;
# if defined(O_CLOEXEC)
/* glibc posix_openpt() in GNU/Linux calls open("/dev/ptmx", flags) internally.
* So version dependency on GNU/Linux is the same as O_CLOEXEC with open().
* O_CLOEXEC is available since Linux 2.6.23. Linux 2.6.18 silently ignore it. */
flags |= O_CLOEXEC;
# endif
if ((masterfd = posix_openpt(flags)) == -1) goto error;
}
rb_fd_fix_cloexec(masterfd);
if (sigaction(SIGCHLD, &dfl, &old) == -1) goto error;
if (grantpt(masterfd) == -1) goto grantpt_error;
#endif
if (sigaction(SIGCHLD, &old, NULL) == -1) goto error;
if (unlockpt(masterfd) == -1) goto error;
if (ptsname_r(masterfd, SlaveName, DEVICELEN) != 0) goto error;
slavedevice = SlaveName;
if (no_mesg(slavedevice, nomesg) == -1) goto error;
if ((slavefd = rb_cloexec_open(slavedevice, O_RDWR|O_NOCTTY, 0)) == -1) goto error;
rb_update_max_fd(slavefd);
#if defined(I_PUSH) && !defined(__linux__) && !defined(_AIX)
if (ioctl_I_PUSH(slavefd, "ptem") == -1) goto error;
if (ioctl_I_PUSH(slavefd, "ldterm") == -1) goto error;
if (ioctl_I_PUSH(slavefd, "ttcompat") == -1) goto error;
#endif
*master = masterfd;
*slave = slavefd;
return 0;
grantpt_error:
sigaction(SIGCHLD, &old, NULL);
error:
if (slavefd != -1) close(slavefd);
if (masterfd != -1) close(masterfd);
if (fail) {
rb_raise(rb_eRuntimeError, "can't get Master/Slave device");
}
return -1;
#elif defined HAVE_OPENPTY
/*
* Use openpty(3) of 4.3BSD Reno and later,
* or the same interface function.
*/
if (openpty(master, slave, SlaveName,
(struct termios *)0, (struct winsize *)0) == -1) {
if (!fail) return -1;
rb_raise(rb_eRuntimeError, "openpty() failed");
}
rb_fd_fix_cloexec(*master);
rb_fd_fix_cloexec(*slave);
if (no_mesg(SlaveName, nomesg) == -1) {
if (!fail) return -1;
rb_raise(rb_eRuntimeError, "can't chmod slave pty");
}
return 0;
#elif defined HAVE__GETPTY
/* SGI IRIX */
char *name;
mode_t mode = nomesg ? 0600 : 0622;
if (!(name = _getpty(master, O_RDWR, mode, 0))) {
if (!fail) return -1;
rb_raise(rb_eRuntimeError, "_getpty() failed");
}
rb_fd_fix_cloexec(*master);
*slave = rb_cloexec_open(name, O_RDWR, 0);
/* error check? */
rb_update_max_fd(*slave);
strlcpy(SlaveName, name, DEVICELEN);
return 0;
#elif defined(HAVE_PTSNAME)
/* System V */
int masterfd = -1, slavefd = -1;
char *slavedevice;
void (*s)();
extern int unlockpt(int);
extern int grantpt(int);
#if defined(__sun)
/* workaround for Solaris 10: grantpt() doesn't work if FD_CLOEXEC is set. [ruby-dev:44688] */
if((masterfd = open("/dev/ptmx", O_RDWR, 0)) == -1) goto error;
s = signal(SIGCHLD, SIG_DFL);
if(grantpt(masterfd) == -1) goto error;
rb_fd_fix_cloexec(masterfd);
#else
if((masterfd = rb_cloexec_open("/dev/ptmx", O_RDWR, 0)) == -1) goto error;
rb_update_max_fd(masterfd);
s = signal(SIGCHLD, SIG_DFL);
if(grantpt(masterfd) == -1) goto error;
#endif
signal(SIGCHLD, s);
if(unlockpt(masterfd) == -1) goto error;
if (ptsname_r(masterfd, SlaveName, DEVICELEN) != 0) goto error;
slavedevice = SlaveName;
if (no_mesg(slavedevice, nomesg) == -1) goto error;
if((slavefd = rb_cloexec_open(slavedevice, O_RDWR, 0)) == -1) goto error;
rb_update_max_fd(slavefd);
#if defined(I_PUSH) && !defined(__linux__) && !defined(_AIX)
if(ioctl_I_PUSH(slavefd, "ptem") == -1) goto error;
if(ioctl_I_PUSH(slavefd, "ldterm") == -1) goto error;
ioctl_I_PUSH(slavefd, "ttcompat");
#endif
*master = masterfd;
*slave = slavefd;
return 0;
error:
if (slavefd != -1) close(slavefd);
if (masterfd != -1) close(masterfd);
if (fail) rb_raise(rb_eRuntimeError, "can't get Master/Slave device");
return -1;
#else
/* BSD */
int masterfd = -1, slavefd = -1;
int i;
char MasterName[DEVICELEN];
#define HEX1(c) \
c"0",c"1",c"2",c"3",c"4",c"5",c"6",c"7", \
c"8",c"9",c"a",c"b",c"c",c"d",c"e",c"f"
#if defined(_IBMESA) /* AIX/ESA */
static const char MasterDevice[] = "/dev/ptyp%s";
static const char SlaveDevice[] = "/dev/ttyp%s";
static const char deviceNo[][3] = {
HEX1("0"), HEX1("1"), HEX1("2"), HEX1("3"),
HEX1("4"), HEX1("5"), HEX1("6"), HEX1("7"),
HEX1("8"), HEX1("9"), HEX1("a"), HEX1("b"),
HEX1("c"), HEX1("d"), HEX1("e"), HEX1("f"),
};
#else /* 4.2BSD */
static const char MasterDevice[] = "/dev/pty%s";
static const char SlaveDevice[] = "/dev/tty%s";
static const char deviceNo[][3] = {
HEX1("p"), HEX1("q"), HEX1("r"), HEX1("s"),
};
#endif
#undef HEX1
for (i = 0; i < numberof(deviceNo); i++) {
const char *const devno = deviceNo[i];
snprintf(MasterName, sizeof MasterName, MasterDevice, devno);
if ((masterfd = rb_cloexec_open(MasterName,O_RDWR,0)) >= 0) {
rb_update_max_fd(masterfd);
*master = masterfd;
snprintf(SlaveName, DEVICELEN, SlaveDevice, devno);
if ((slavefd = rb_cloexec_open(SlaveName,O_RDWR,0)) >= 0) {
rb_update_max_fd(slavefd);
*slave = slavefd;
if (chown(SlaveName, getuid(), getgid()) != 0) goto error;
if (chmod(SlaveName, nomesg ? 0600 : 0622) != 0) goto error;
return 0;
}
close(masterfd);
}
}
error:
if (slavefd != -1) close(slavefd);
if (masterfd != -1) close(masterfd);
if (fail) rb_raise(rb_eRuntimeError, "can't get %s", SlaveName);
return -1;
#endif
}
static void
getDevice(int *master, int *slave, char SlaveName[DEVICELEN], int nomesg)
{
if (get_device_once(master, slave, SlaveName, nomesg, 0)) {
rb_gc();
get_device_once(master, slave, SlaveName, nomesg, 1);
}
}
static VALUE
pty_close_pty(VALUE assoc)
{
VALUE io;
int i;
for (i = 0; i < 2; i++) {
io = rb_ary_entry(assoc, i);
if (RB_TYPE_P(io, T_FILE)) {
/* it's OK to call rb_io_close again even if it's already closed */
rb_io_close(io);
}
}
return Qnil;
}
/*
* call-seq:
* PTY.open => [master_io, slave_file]
* PTY.open {|(master_io, slave_file)| ... } => block value
*
* Allocates a pty (pseudo-terminal).
*
* In the block form, yields an array of two elements (<tt>master_io, slave_file</tt>)
* and the value of the block is returned from +open+.
*
* The IO and File are both closed after the block completes if they haven't
* been already closed.
*
* PTY.open {|master, slave|
* p master #=> #<IO:masterpty:/dev/pts/1>
* p slave #=> #<File:/dev/pts/1>
* p slave.path #=> "/dev/pts/1"
* }
*
* In the non-block form, returns a two element array, <tt>[master_io,
* slave_file]</tt>.
*
* master, slave = PTY.open
* # do something with master for IO, or the slave file
*
* The arguments in both forms are:
*
* +master_io+:: the master of the pty, as an IO.
* +slave_file+:: the slave of the pty, as a File. The path to the
* terminal device is available via +slave_file.path+
*
* IO#raw! is usable to disable newline conversions:
*
* require 'io/console'
* PTY.open {|m, s|
* s.raw!
* # ...
* }
*
*/
static VALUE
pty_open(VALUE klass)
{
int master_fd, slave_fd;
char slavename[DEVICELEN];
getDevice(&master_fd, &slave_fd, slavename, 1);
VALUE master_path = rb_obj_freeze(rb_sprintf("masterpty:%s", slavename));
VALUE master_io = rb_io_open_descriptor(rb_cIO, master_fd, FMODE_READWRITE | FMODE_SYNC | FMODE_DUPLEX, master_path, RUBY_IO_TIMEOUT_DEFAULT, NULL);
VALUE slave_path = rb_obj_freeze(rb_str_new_cstr(slavename));
VALUE slave_file = rb_io_open_descriptor(rb_cFile, slave_fd, FMODE_READWRITE | FMODE_SYNC | FMODE_DUPLEX | FMODE_TTY, slave_path, RUBY_IO_TIMEOUT_DEFAULT, NULL);
VALUE assoc = rb_assoc_new(master_io, slave_file);
if (rb_block_given_p()) {
return rb_ensure(rb_yield, assoc, pty_close_pty, assoc);
}
return assoc;
}
static VALUE
pty_detach_process(VALUE v)
{
struct pty_info *info = (void *)v;
#ifdef WNOHANG
int st;
if (rb_waitpid(info->child_pid, &st, WNOHANG) <= 0)
return Qnil;
#endif
rb_detach_process(info->child_pid);
return Qnil;
}
/*
* call-seq:
* PTY.spawn([env,] command_line) { |r, w, pid| ... }
* PTY.spawn([env,] command_line) => [r, w, pid]
* PTY.spawn([env,] command, arguments, ...) { |r, w, pid| ... }
* PTY.spawn([env,] command, arguments, ...) => [r, w, pid]
*
* Spawns the specified command on a newly allocated pty. You can also use the
* alias ::getpty.
*
* The command's controlling tty is set to the slave device of the pty
* and its standard input/output/error is redirected to the slave device.
*
* +env+ is an optional hash that provides additional environment variables to the spawned pty.
*
* # sets FOO to "bar"
* PTY.spawn({"FOO"=>"bar"}, "printenv", "FOO") { |r,w,pid| p r.read } #=> "bar\r\n"
* # unsets FOO
* PTY.spawn({"FOO"=>nil}, "printenv", "FOO") { |r,w,pid| p r.read } #=> ""
*
* +command+ and +command_line+ are the full commands to run, given a String.
* Any additional +arguments+ will be passed to the command.
*
* === Return values
*
* In the non-block form this returns an array of size three,
* <tt>[r, w, pid]</tt>.
*
* In the block form these same values will be yielded to the block:
*
* +r+:: A readable IO that contains the command's
* standard output and standard error
* +w+:: A writable IO that is the command's standard input
* +pid+:: The process identifier for the command.
*/
static VALUE
pty_getpty(int argc, VALUE *argv, VALUE self)
{
VALUE res;
struct pty_info info;
char SlaveName[DEVICELEN];
establishShell(argc, argv, &info, SlaveName);
VALUE pty_path = rb_obj_freeze(rb_str_new_cstr(SlaveName));
VALUE rport = rb_io_open_descriptor(
rb_cFile, info.fd, FMODE_READABLE, pty_path, RUBY_IO_TIMEOUT_DEFAULT, NULL
);
int wpty_fd = rb_cloexec_dup(info.fd);
if (wpty_fd == -1) {
rb_sys_fail("dup()");
}
VALUE wport = rb_io_open_descriptor(
rb_cFile, wpty_fd, FMODE_WRITABLE | FMODE_TRUNC | FMODE_CREATE | FMODE_SYNC,
pty_path, RUBY_IO_TIMEOUT_DEFAULT, NULL
);
res = rb_ary_new2(3);
rb_ary_store(res, 0, rport);
rb_ary_store(res, 1, wport);
rb_ary_store(res,2,PIDT2NUM(info.child_pid));
if (rb_block_given_p()) {
rb_ensure(rb_yield, res, pty_detach_process, (VALUE)&info);
return Qnil;
}
return res;
}
NORETURN(static void raise_from_check(rb_pid_t pid, int status));
static void
raise_from_check(rb_pid_t pid, int status)
{
const char *state;
VALUE msg;
VALUE exc;
#if defined(WIFSTOPPED)
#elif defined(IF_STOPPED)
#define WIFSTOPPED(status) IF_STOPPED(status)
#else
---->> Either IF_STOPPED or WIFSTOPPED is needed <<----
#endif /* WIFSTOPPED | IF_STOPPED */
if (WIFSTOPPED(status)) { /* suspend */
state = "stopped";
}
else if (kill(pid, 0) == 0) {
state = "changed";
}
else {
state = "exited";
}
msg = rb_sprintf("pty - %s: %ld", state, (long)pid);
exc = rb_exc_new_str(eChildExited, msg);
rb_iv_set(exc, "status", rb_last_status_get());
rb_exc_raise(exc);
}
/*
* call-seq:
* PTY.check(pid, raise = false) => Process::Status or nil
* PTY.check(pid, true) => nil or raises PTY::ChildExited
*
* Checks the status of the child process specified by +pid+.
* Returns +nil+ if the process is still alive.
*
* If the process is not alive, and +raise+ was true, a PTY::ChildExited
* exception will be raised. Otherwise it will return a Process::Status
* instance.
*
* +pid+:: The process id of the process to check
* +raise+:: If +true+ and the process identified by +pid+ is no longer
* alive a PTY::ChildExited is raised.
*
*/
static VALUE
pty_check(int argc, VALUE *argv, VALUE self)
{
VALUE pid, exc;
rb_pid_t cpid;
int status;
const int flag =
#ifdef WNOHANG
WNOHANG|
#endif
#ifdef WUNTRACED
WUNTRACED|
#endif
0;
rb_scan_args(argc, argv, "11", &pid, &exc);
cpid = rb_waitpid(NUM2PIDT(pid), &status, flag);
if (cpid == -1 || cpid == 0) return Qnil;
if (!RTEST(exc)) return rb_last_status_get();
raise_from_check(cpid, status);
UNREACHABLE_RETURN(Qnil);
}
static VALUE cPTY;
/*
* Document-class: PTY::ChildExited
*
* Thrown when PTY::check is called for a pid that represents a process that
* has exited.
*/
/*
* Document-class: PTY
*
* Creates and manages pseudo terminals (PTYs). See also
* https://en.wikipedia.org/wiki/Pseudo_terminal
*
* PTY allows you to allocate new terminals using ::open or ::spawn a new
* terminal with a specific command.
*
* == Example
*
* In this example we will change the buffering type in the +factor+ command,
* assuming that factor uses stdio for stdout buffering.
*
* If IO.pipe is used instead of PTY.open, this code deadlocks because factor's
* stdout is fully buffered.
*
* # start by requiring the standard library PTY
* require 'pty'
*
* master, slave = PTY.open
* read, write = IO.pipe
* pid = spawn("factor", :in=>read, :out=>slave)
* read.close # we dont need the read
* slave.close # or the slave
*
* # pipe "42" to the factor command
* write.puts "42"
* # output the response from factor
* p master.gets #=> "42: 2 3 7\n"
*
* # pipe "144" to factor and print out the response
* write.puts "144"
* p master.gets #=> "144: 2 2 2 2 3 3\n"
* write.close # close the pipe
*
* # The result of read operation when pty slave is closed is platform
* # dependent.
* ret = begin
* master.gets # FreeBSD returns nil.
* rescue Errno::EIO # GNU/Linux raises EIO.
* nil
* end
* p ret #=> nil
*
* == License
*
* (c) Copyright 1998 by Akinori Ito.
*
* This software may be redistributed freely for this purpose, in full
* or in part, provided that this entire copyright notice is included
* on any copies of this software and applications and derivations thereof.
*
* This software is provided on an "as is" basis, without warranty of any
* kind, either expressed or implied, as to any matter including, but not
* limited to warranty of fitness of purpose, or merchantability, or
* results obtained from use of this software.
*/
void
Init_pty(void)
{
cPTY = rb_define_module("PTY");
#if 1
rb_define_module_function(cPTY,"get""pty",pty_getpty,-1);
#else /* for RDoc */
/* show getpty as an alias of spawn */
VALUE sPTY = rb_singleton_class(cPTY);
rb_define_alias(sPTY, "getpty", "spawn");
#endif
rb_define_module_function(cPTY,"spawn",pty_getpty,-1);
rb_define_singleton_method(cPTY,"check",pty_check,-1);
rb_define_singleton_method(cPTY,"open",pty_open,0);
eChildExited = rb_define_class_under(cPTY,"ChildExited",rb_eRuntimeError);
rb_define_method(eChildExited,"status",echild_status,0);
}