/*

 * Copyright (c) 1995, 2008, Oracle and/or its affiliates. All rights reserved.

 * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.

 *

 * This code is free software; you can redistribute it and/or modify it

 * under the terms of the GNU General Public License version 2 only, as

 * published by the Free Software Foundation.  Oracle designates this

 * particular file as subject to the "Classpath" exception as provided

 * by Oracle in the LICENSE file that accompanied this code.

 *

 * This code is distributed in the hope that it will be useful, but WITHOUT

 * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or

 * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License

 * version 2 for more details (a copy is included in the LICENSE file that

 * accompanied this code).

 *

 * You should have received a copy of the GNU General Public License version

 * 2 along with this work; if not, write to the Free Software Foundation,

 * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.

 *

 * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA

 * or visit www.oracle.com if you need additional information or have any

 * questions.

 */

#undef  _LARGEFILE64_SOURCE

#define _LARGEFILE64_SOURCE 1

#include "jni.h"

#include "jvm.h"

#include "jvm_md.h"

#include "jni_util.h"

#include "io_util.h"

/*

 * Platform-specific support for java.lang.Process

 */

#include <assert.h>

#include <stddef.h>

#include <stdlib.h>

#include <sys/types.h>

#include <ctype.h>

#include <signal.h>

#include <string.h>

#include <errno.h>

#include <dirent.h>

#include <unistd.h>

#include <fcntl.h>

#include <limits.h>

/*

 * There are 3 possible strategies we might use to "fork":

 *

 * - fork(2).  Very portable and reliable but subject to

 *   failure due to overcommit (see the documentation on

 *   /proc/sys/vm/overcommit_memory in Linux proc(5)).

 *   This is the ancient problem of spurious failure whenever a large

 *   process starts a small subprocess.

 *

 * - vfork().  Using this is scary because all relevant man pages

 *   contain dire warnings, e.g. Linux vfork(2).  But at least it's

 *   documented in the glibc docs and is standardized by XPG4.

 *   http://www.opengroup.org/onlinepubs/000095399/functions/vfork.html

 *   On Linux, one might think that vfork() would be implemented using

 *   the clone system call with flag CLONE_VFORK, but in fact vfork is

 *   a separate system call (which is a good sign, suggesting that

 *   vfork will continue to be supported at least on Linux).

 *   Another good sign is that glibc implements posix_spawn using

 *   vfork whenever possible.  Note that we cannot use posix_spawn

 *   ourselves because there's no reliable way to close all inherited

 *   file descriptors.

 *

 * - clone() with flags CLONE_VM but not CLONE_THREAD.  clone() is

 *   Linux-specific, but this ought to work - at least the glibc

 *   sources contain code to handle different combinations of CLONE_VM

 *   and CLONE_THREAD.  However, when this was implemented, it

 *   appeared to fail on 32-bit i386 (but not 64-bit x86_64) Linux with

 *   the simple program

 *     Runtime.getRuntime().exec("/bin/true").waitFor();

 *   with:

 *     #  Internal Error (os_linux_x86.cpp:683), pid=19940, tid=2934639536

 *     #  Error: pthread_getattr_np failed with errno = 3 (ESRCH)

 *   We believe this is a glibc bug, reported here:

 *     http://sources.redhat.com/bugzilla/show_bug.cgi?id=10311

 *   but the glibc maintainers closed it as WONTFIX.

 *

 * Based on the above analysis, we are currently using vfork() on

 * Linux and fork() on other Unix systems, but the code to use clone()

 * remains.

 */

#define START_CHILD_USE_CLONE 0  /* clone() currently disabled; see above. */

#ifndef START_CHILD_USE_CLONE

  #ifdef __linux__

    #define START_CHILD_USE_CLONE 1

  #else

    #define START_CHILD_USE_CLONE 0

  #endif

#endif

/* By default, use vfork() on Linux. */

#ifndef START_CHILD_USE_VFORK

  #ifdef __linux__

    #define START_CHILD_USE_VFORK 1

  #else

    #define START_CHILD_USE_VFORK 0

  #endif

#endif

#if START_CHILD_USE_CLONE

#include <sched.h>

#define START_CHILD_SYSTEM_CALL "clone"

#elif START_CHILD_USE_VFORK

#define START_CHILD_SYSTEM_CALL "vfork"

#else

#define START_CHILD_SYSTEM_CALL "fork"

#endif

#ifndef STDIN_FILENO

#define STDIN_FILENO 0

#endif

#ifndef STDOUT_FILENO

#define STDOUT_FILENO 1

#endif

#ifndef STDERR_FILENO

#define STDERR_FILENO 2

#endif

#ifndef SA_NOCLDSTOP

#define SA_NOCLDSTOP 0

#endif

#ifndef SA_RESTART

#define SA_RESTART 0

#endif

#define FAIL_FILENO (STDERR_FILENO + 1)

/* TODO: Refactor. */

#define RESTARTABLE(_cmd, _result) do { \

  do { \

    _result = _cmd; \

  } while((_result == -1) && (errno == EINTR)); \

} while(0)

/* This is one of the rare times it's more portable to declare an

 * external symbol explicitly, rather than via a system header.

 * The declaration is standardized as part of UNIX98, but there is

 * no standard (not even de-facto) header file where the

 * declaration is to be found.  See:

 * http://www.opengroup.org/onlinepubs/009695399/functions/environ.html

 * http://www.opengroup.org/onlinepubs/009695399/functions/xsh_chap02_02.html

 *

 * "All identifiers in this volume of IEEE Std 1003.1-2001, except

 * environ, are defined in at least one of the headers" (!)

 */

extern char **environ;

static void

setSIGCHLDHandler(JNIEnv *env)

{

    /* There is a subtle difference between having the signal handler

     * for SIGCHLD be SIG_DFL and SIG_IGN.  We cannot obtain process

     * termination information for child processes if the signal

     * handler is SIG_IGN.  It must be SIG_DFL.

     *

     * We used to set the SIGCHLD handler only on Linux, but it's

     * safest to set it unconditionally.

     *

     * Consider what happens if java's parent process sets the SIGCHLD

     * handler to SIG_IGN.  Normally signal handlers are inherited by

     * children, but SIGCHLD is a controversial case.  Solaris appears

     * to always reset it to SIG_DFL, but this behavior may be

     * non-standard-compliant, and we shouldn't rely on it.

     *

     * References:

     * http://www.opengroup.org/onlinepubs/7908799/xsh/exec.html

     * http://www.pasc.org/interps/unofficial/db/p1003.1/pasc-1003.1-132.html

     */

    struct sigaction sa;

    sa.sa_handler = SIG_DFL;

    sigemptyset(&sa.sa_mask);

    sa.sa_flags = SA_NOCLDSTOP | SA_RESTART;

    if (sigaction(SIGCHLD, &sa, NULL) < 0)

        JNU_ThrowInternalError(env, "Can't set SIGCHLD handler");

}

static void*

xmalloc(JNIEnv *env, size_t size)

{

    void *p = malloc(size);

    if (p == NULL)

        JNU_ThrowOutOfMemoryError(env, NULL);

    return p;

}

#define NEW(type, n) ((type *) xmalloc(env, (n) * sizeof(type)))

/**

 * If PATH is not defined, the OS provides some default value.

 * Unfortunately, there's no portable way to get this value.

 * Fortunately, it's only needed if the child has PATH while we do not.

 */

static const char*

defaultPath(void)

{

#ifdef __solaris__

    /* These really are the Solaris defaults! */

    return (geteuid() == 0 || getuid() == 0) ?

        "/usr/xpg4/bin:/usr/ccs/bin:/usr/bin:/opt/SUNWspro/bin:/usr/sbin" :

        "/usr/xpg4/bin:/usr/ccs/bin:/usr/bin:/opt/SUNWspro/bin:";

#else

    return ":/bin:/usr/bin";    /* glibc */

#endif

}

static const char*

effectivePath(void)

{

    const char *s = getenv("PATH");

    return (s != NULL) ? s : defaultPath();

}

static int

countOccurrences(const char *s, char c)

{

    int count;

    for (count = 0; *s != '\0'; s++)

        count += (*s == c);

    return count;

}

static const char * const *

splitPath(JNIEnv *env, const char *path)

{

    const char *p, *q;

    char **pathv;

    int i;

    int count = countOccurrences(path, ':') + 1;

    pathv = NEW(char*, count+1);

    pathv[count] = NULL;

    for (p = path, i = 0; i < count; i++, p = q + 1) {

        for (q = p; (*q != ':') && (*q != '\0'); q++)

            ;

        if (q == p)             /* empty PATH component => "." */

            pathv[i] = "./";

        else {

            int addSlash = ((*(q - 1)) != '/');

            pathv[i] = NEW(char, q - p + addSlash + 1);

            memcpy(pathv[i], p, q - p);

            if (addSlash)

                pathv[i][q - p] = '/';

            pathv[i][q - p + addSlash] = '\0';

        }

    }

    return (const char * const *) pathv;

}

/**

 * Cached value of JVM's effective PATH.

 * (We don't support putenv("PATH=...") in native code)

 */

static const char *parentPath;

/**

 * Split, canonicalized version of parentPath

 */

static const char * const *parentPathv;

static jfieldID field_exitcode;

JNIEXPORT void JNICALL

Java_java_lang_UNIXProcess_initIDs(JNIEnv *env, jclass clazz)

{

    field_exitcode = (*env)->GetFieldID(env, clazz, "exitcode", "I");

    parentPath  = effectivePath();

    parentPathv = splitPath(env, parentPath);

    setSIGCHLDHandler(env);

}

#ifndef WIFEXITED

#define WIFEXITED(status) (((status)&0xFF) == 0)

#endif

#ifndef WEXITSTATUS

#define WEXITSTATUS(status) (((status)>>8)&0xFF)

#endif

#ifndef WIFSIGNALED

#define WIFSIGNALED(status) (((status)&0xFF) > 0 && ((status)&0xFF00) == 0)

#endif

#ifndef WTERMSIG

#define WTERMSIG(status) ((status)&0x7F)

#endif

/* Block until a child process exits and return its exit code.

   Note, can only be called once for any given pid. */

JNIEXPORT jint JNICALL

Java_java_lang_UNIXProcess_waitForProcessExit(JNIEnv* env,

                                              jobject junk,

                                              jint pid)

{

    /* We used to use waitid() on Solaris, waitpid() on Linux, but

     * waitpid() is more standard, so use it on all POSIX platforms. */

    int status;

    /* Wait for the child process to exit.  This returns immediately if

       the child has already exited. */

    while (waitpid(pid, &status, 0) < 0) {

        switch (errno) {

        case ECHILD: return 0;

        case EINTR: break;

        default: return -1;

        }

    }

    if (WIFEXITED(status)) {

        /*

         * The child exited normally; get its exit code.

         */

        return WEXITSTATUS(status);

    } else if (WIFSIGNALED(status)) {

        /* The child exited because of a signal.

         * The best value to return is 0x80 + signal number,

         * because that is what all Unix shells do, and because

         * it allows callers to distinguish between process exit and

         * process death by signal.

         * Unfortunately, the historical behavior on Solaris is to return

         * the signal number, and we preserve this for compatibility. */

#ifdef __solaris__

        return WTERMSIG(status);

#else

        return 0x80 + WTERMSIG(status);

#endif

    } else {

        /*

         * Unknown exit code; pass it through.

         */

        return status;

    }

}

static ssize_t

restartableWrite(int fd, const void *buf, size_t count)

{

    ssize_t result;

    RESTARTABLE(write(fd, buf, count), result);

    return result;

}

static int

restartableDup2(int fd_from, int fd_to)

{

    int err;

    RESTARTABLE(dup2(fd_from, fd_to), err);

    return err;

}

static int

restartableClose(int fd)

{

    int err;

    RESTARTABLE(close(fd), err);

    return err;

}

static int

closeSafely(int fd)

{

    return (fd == -1) ? 0 : restartableClose(fd);

}

static int

isAsciiDigit(char c)

{

  return c >= '0' && c <= '9';

}

static int

closeDescriptors(void)

{

    DIR *dp;

    struct dirent64 *dirp;

    int from_fd = FAIL_FILENO + 1;

    /* We're trying to close all file descriptors, but opendir() might

     * itself be implemented using a file descriptor, and we certainly

     * don't want to close that while it's in use.  We assume that if

     * opendir() is implemented using a file descriptor, then it uses

     * the lowest numbered file descriptor, just like open().  So we

     * close a couple explicitly.  */

    restartableClose(from_fd);          /* for possible use by opendir() */

    restartableClose(from_fd + 1);      /* another one for good luck */

    if ((dp = opendir("/proc/self/fd")) == NULL)

        return 0;

    /* We use readdir64 instead of readdir to work around Solaris bug

     * 6395699: /proc/self/fd fails to report file descriptors >= 1024 on Solaris 9

     */

    while ((dirp = readdir64(dp)) != NULL) {

        int fd;

        if (isAsciiDigit(dirp->d_name[0]) &&

            (fd = strtol(dirp->d_name, NULL, 10)) >= from_fd + 2)

            restartableClose(fd);

    }

    closedir(dp);

    return 1;

}

static int

moveDescriptor(int fd_from, int fd_to)

{

    if (fd_from != fd_to) {

        if ((restartableDup2(fd_from, fd_to) == -1) ||

            (restartableClose(fd_from) == -1))

            return -1;

    }

    return 0;

}

static const char *

getBytes(JNIEnv *env, jbyteArray arr)

{

    return arr == NULL ? NULL :

        (const char*) (*env)->GetByteArrayElements(env, arr, NULL);

}

static void

releaseBytes(JNIEnv *env, jbyteArray arr, const char* parr)

{

    if (parr != NULL)

        (*env)->ReleaseByteArrayElements(env, arr, (jbyte*) parr, JNI_ABORT);

}

static void

initVectorFromBlock(const char**vector, const char* block, int count)

{

    int i;

    const char *p;

    for (i = 0, p = block; i < count; i++) {

        /* Invariant: p always points to the start of a C string. */

        vector[i] = p;

        while (*(p++));

    }

    vector[count] = NULL;

}

static void

throwIOException(JNIEnv *env, int errnum, const char *defaultDetail)

{

    static const char * const format = "error=%d, %s";

    const char *detail = defaultDetail;

    char *errmsg;

    jstring s;

    if (errnum != 0) {

        const char *s = strerror(errnum);

        if (strcmp(s, "Unknown error") != 0)

            detail = s;

    }

    /* ASCII Decimal representation uses 2.4 times as many bits as binary. */

    errmsg = NEW(char, strlen(format) + strlen(detail) + 3 * sizeof(errnum));

    sprintf(errmsg, format, errnum, detail);

    s = JNU_NewStringPlatform(env, errmsg);

    if (s != NULL) {

        jobject x = JNU_NewObjectByName(env, "java/io/IOException",

                                        "(Ljava/lang/String;)V", s);

        if (x != NULL)

            (*env)->Throw(env, x);

    }

    free(errmsg);

}

#ifdef DEBUG_PROCESS

/* Debugging process code is difficult; where to write debug output? */

static void

debugPrint(char *format, ...)

{

    FILE *tty = fopen("/dev/tty", "w");

    va_list ap;

    va_start(ap, format);

    vfprintf(tty, format, ap);

    va_end(ap);

    fclose(tty);

}

#endif /* DEBUG_PROCESS */

/**

 * Exec FILE as a traditional Bourne shell script (i.e. one without #!).

 * If we could do it over again, we would probably not support such an ancient

 * misfeature, but compatibility wins over sanity.  The original support for

 * this was imported accidentally from execvp().

 */

static void

execve_as_traditional_shell_script(const char *file,

                                   const char *argv[],

                                   const char *const envp[])

{

    /* Use the extra word of space provided for us in argv by caller. */

    const char *argv0 = argv[0];

    const char *const *end = argv;

    while (*end != NULL)