Fixes issue #8052: The posix subprocess module's close_fds behavior was

suboptimal by closing all possible file descriptors rather than just
the open ones in the child process before exec().

It now closes only the open fds when it is possible to safely determine what
those are.

Lib/test/test_subprocess.py

@@ -1295,6 +1295,11 @@ class POSIXProcessTestCase(BaseTestCase):
         self.addCleanup(os.close, fds[1])
 
         open_fds = set(fds)
+        # add a bunch more fds
+        for _ in range(9):
+            fd = os.open("/dev/null", os.O_RDONLY)
+            self.addCleanup(os.close, fd)
+            open_fds.add(fd)
 
         p = subprocess.Popen([sys.executable, fd_status],
                              stdout=subprocess.PIPE, close_fds=False)
@@ -1313,6 +1318,19 @@ class POSIXProcessTestCase(BaseTestCase):
                          "Some fds were left open")
         self.assertIn(1, remaining_fds, "Subprocess failed")
 
+        # Keep some of the fd's we opened open in the subprocess.
+        # This tests _posixsubprocess.c's proper handling of fds_to_keep.
+        fds_to_keep = set(open_fds.pop() for _ in range(8))
+        p = subprocess.Popen([sys.executable, fd_status],
+                             stdout=subprocess.PIPE, close_fds=True,
+                             pass_fds=())
+        output, ignored = p.communicate()
+        remaining_fds = set(map(int, output.split(b',')))
+
+        self.assertFalse(remaining_fds & fds_to_keep & open_fds,
+                         "Some fds not in pass_fds were left open")
+        self.assertIn(1, remaining_fds, "Subprocess failed")
+
     # Mac OS X Tiger (10.4) has a kernel bug: sometimes, the file
     # descriptor of a pipe closed in the parent process is valid in the
     # child process according to fstat(), but the mode of the file

Modules/_posixsubprocess.c

@@ -5,7 +5,26 @@
 #endif
 #include <unistd.h>
 #include <fcntl.h>
+#ifdef HAVE_SYS_TYPES_H
+#include <sys/types.h>
+#endif
+#ifdef HAVE_SYS_SYSCALL_H
+#include <sys/syscall.h>
+#endif
+#ifdef HAVE_DIRENT_H
+#include <dirent.h>
+#endif
 
+#if defined(sun) && !defined(HAVE_DIRFD)
+/* Some versions of Solaris lack dirfd(). */
+# define DIRFD(dirp) ((dirp)->dd_fd)
+# define HAVE_DIRFD
+#else
+# define DIRFD(dirp) (dirfd(dirp))
+#endif
+
+#define LINUX_SOLARIS_FD_DIR "/proc/self/fd"
+#define BSD_OSX_FD_DIR "/dev/fd"
 
 #define POSIX_CALL(call)   if ((call) == -1) goto error
 
@@ -26,6 +45,233 @@ static int _enable_gc(PyObject *gc_module)
 }
 
 
+/* Convert ASCII to a positive int, no libc call. no overflow. -1 on error. */
+static int _pos_int_from_ascii(char *name)
+{
+    int num = 0;
+    while (*name >= '0' && *name <= '9') {
+        num = num * 10 + (*name - '0');
+        ++name;
+    }
+    if (*name)
+        return -1;  /* Non digit found, not a number. */
+    return num;
+}
+
+
+/* Returns 1 if there is a problem with fd_sequence, 0 otherwise. */
+static int _sanity_check_python_fd_sequence(PyObject *fd_sequence)
+{
+    Py_ssize_t seq_idx, seq_len = PySequence_Length(fd_sequence);
+    long prev_fd = -1;
+    for (seq_idx = 0; seq_idx < seq_len; ++seq_idx) {
+        PyObject* py_fd = PySequence_Fast_GET_ITEM(fd_sequence, seq_idx);
+        long iter_fd = PyLong_AsLong(py_fd);
+        if (iter_fd < 0 || iter_fd < prev_fd || iter_fd > INT_MAX) {
+            /* Negative, overflow, not a Long, unsorted, too big for a fd. */
+            return 1;
+        }
+    }
+    return 0;
+}
+
+
+/* Is fd found in the sorted Python Sequence? */
+static int _is_fd_in_sorted_fd_sequence(int fd, PyObject *fd_sequence)
+{
+    /* Binary search. */
+    Py_ssize_t search_min = 0;
+    Py_ssize_t search_max = PySequence_Length(fd_sequence) - 1;
+    if (search_max < 0)
+        return 0;
+    do {
+        long middle = (search_min + search_max) / 2;
+        long middle_fd = PyLong_AsLong(
+                PySequence_Fast_GET_ITEM(fd_sequence, middle));
+        if (fd == middle_fd)
+            return 1;
+        if (fd > middle_fd)
+            search_min = middle + 1;
+        else
+            search_max = middle - 1;
+    } while (search_min <= search_max);
+    return 0;
+}
+
+
+/* Close all file descriptors in the range start_fd inclusive to
+ * end_fd exclusive except for those in py_fds_to_keep.  If the
+ * range defined by [start_fd, end_fd) is large this will take a
+ * long time as it calls close() on EVERY possible fd.
+ */
+static void _close_fds_by_brute_force(int start_fd, int end_fd,
+                                      PyObject *py_fds_to_keep)
+{
+    Py_ssize_t num_fds_to_keep = PySequence_Length(py_fds_to_keep);
+    Py_ssize_t keep_seq_idx;
+    int fd_num;
+    /* As py_fds_to_keep is sorted we can loop through the list closing
+     * fds inbetween any in the keep list falling within our range. */
+    for (keep_seq_idx = 0; keep_seq_idx < num_fds_to_keep; ++keep_seq_idx) {
+        PyObject* py_keep_fd = PySequence_Fast_GET_ITEM(py_fds_to_keep,
+                                                        keep_seq_idx);
+        int keep_fd = PyLong_AsLong(py_keep_fd);
+        if (keep_fd < start_fd)
+            continue;
+        for (fd_num = start_fd; fd_num < keep_fd; ++fd_num) {
+            while (close(fd_num) < 0 && errno == EINTR);
+        }
+        start_fd = keep_fd + 1;
+    }
+    if (start_fd <= end_fd) {
+        for (fd_num = start_fd; fd_num < end_fd; ++fd_num) {
+            while (close(fd_num) < 0 && errno == EINTR);
+        }
+    }
+}
+
+
+#if defined(__linux__) && defined(HAVE_SYS_SYSCALL_H)
+/* It doesn't matter if d_name has room for NAME_MAX chars; we're using this
+ * only to read a directory of short file descriptor number names.  The kernel
+ * will return an error if we didn't give it enough space.  Highly Unlikely.
+ * This structure is very old and stable: It will not change unless the kernel
+ * chooses to break compatibility with all existing binaries.  Highly Unlikely.
+ */
+struct linux_dirent {
+   unsigned long  d_ino;        /* Inode number */
+   unsigned long  d_off;        /* Offset to next linux_dirent */
+   unsigned short d_reclen;     /* Length of this linux_dirent */
+   char           d_name[256];  /* Filename (null-terminated) */
+};
+
+/* Close all open file descriptors in the range start_fd inclusive to end_fd
+ * exclusive. Do not close any in the sorted py_fds_to_keep list.
+ *
+ * This version is async signal safe as it does not make any unsafe C library
+ * calls, malloc calls or handle any locks.  It is _unfortunate_ to be forced
+ * to resort to making a kernel system call directly but this is the ONLY api
+ * available that does no harm.  opendir/readdir/closedir perform memory
+ * allocation and locking so while they usually work they are not guaranteed
+ * to (especially if you have replaced your malloc implementation).  A version
+ * of this function that uses those can be found in the _maybe_unsafe variant.
+ *
+ * This is Linux specific because that is all I am ready to test it on.  It
+ * should be easy to add OS specific dirent or dirent64 structures and modify
+ * it with some cpp #define magic to work on other OSes as well if you want.
+ */
+static void _close_open_fd_range_safe(int start_fd, int end_fd,
+                                      PyObject* py_fds_to_keep)
+{
+    int fd_dir_fd;
+    if (start_fd >= end_fd)
+        return;
+    fd_dir_fd = open(LINUX_SOLARIS_FD_DIR, O_RDONLY | O_CLOEXEC, 0);
+    /* Not trying to open the BSD_OSX path as this is currently Linux only. */
+    if (fd_dir_fd == -1) {
+        /* No way to get a list of open fds. */
+        _close_fds_by_brute_force(start_fd, end_fd, py_fds_to_keep);
+        return;
+    } else {
+        char buffer[sizeof(struct linux_dirent)];
+        int bytes;
+        while ((bytes = syscall(SYS_getdents, fd_dir_fd,
+                                (struct linux_dirent *)buffer,
+                                sizeof(buffer))) > 0) {
+            struct linux_dirent *entry;
+            int offset;
+            for (offset = 0; offset < bytes; offset += entry->d_reclen) {
+                int fd;
+                entry = (struct linux_dirent *)(buffer + offset);
+                if ((fd = _pos_int_from_ascii(entry->d_name)) < 0)
+                    continue;  /* Not a number. */
+                if (fd != fd_dir_fd && fd >= start_fd && fd < end_fd &&
+                    !_is_fd_in_sorted_fd_sequence(fd, py_fds_to_keep)) {
+                    while (close(fd) < 0 && errno == EINTR);
+                }
+            }
+        }
+        close(fd_dir_fd);
+    }
+}
+
+#define _close_open_fd_range _close_open_fd_range_safe
+
+#else  /* NOT (defined(__linux__) && defined(HAVE_SYS_SYSCALL_H)) */
+
+
+/* Close all open file descriptors in the range start_fd inclusive to end_fd
+ * exclusive. Do not close any in the sorted py_fds_to_keep list.
+ *
+ * This function violates the strict use of async signal safe functions. :(
+ * It calls opendir(), readdir64() and closedir().  Of these, the one most
+ * likely to ever cause a problem is opendir() as it performs an internal
+ * malloc().  Practically this should not be a problem.  The Java VM makes the
+ * same calls between fork and exec in its own UNIXProcess_md.c implementation.
+ *
+ * readdir_r() is not used because it provides no benefit.  It is typically
+ * implemented as readdir() followed by memcpy().  See also:
+ *   http://womble.decadent.org.uk/readdir_r-advisory.html
+ */
+static void _close_open_fd_range_maybe_unsafe(int start_fd, int end_fd,
+                                              PyObject* py_fds_to_keep)
+{
+    DIR *proc_fd_dir;
+#ifndef HAVE_DIRFD
+    while (_is_fd_in_sorted_fd_sequence(start_fd, py_fds_to_keep) &&
+           (start_fd < end_fd)) {
+        ++start_fd;
+    }
+    if (start_fd >= end_fd)
+        return;
+    /* Close our lowest fd before we call opendir so that it is likely to
+     * reuse that fd otherwise we might close opendir's file descriptor in
+     * our loop.  This trick assumes that fd's are allocated on a lowest
+     * available basis. */
+    while (close(start_fd) < 0 && errno == EINTR);
+    ++start_fd;
+#endif
+    if (start_fd >= end_fd)
+        return;
+
+    proc_fd_dir = opendir(BSD_OSX_FD_DIR);
+    if (!proc_fd_dir)
+        proc_fd_dir = opendir(LINUX_SOLARIS_FD_DIR);
+    if (!proc_fd_dir) {
+        /* No way to get a list of open fds. */
+        _close_fds_by_brute_force(start_fd, end_fd, py_fds_to_keep);
+    } else {
+        struct dirent64 *dir_entry;
+#ifdef HAVE_DIRFD
+        int fd_used_by_opendir = DIRFD(proc_fd_dir);
+#else
+        int fd_used_by_opendir = start_fd - 1;
+#endif
+        errno = 0;
+        /* readdir64 is used to work around Solaris 9 bug 6395699. */
+        while ((dir_entry = readdir64(proc_fd_dir))) {
+            int fd;
+            if ((fd = _pos_int_from_ascii(dir_entry->d_name)) < 0)
+                continue;  /* Not a number. */
+            if (fd != fd_used_by_opendir && fd >= start_fd && fd < end_fd &&
+                !_is_fd_in_sorted_fd_sequence(fd, py_fds_to_keep)) {
+                while (close(fd) < 0 && errno == EINTR);
+            }
+            errno = 0;
+        }
+        if (errno) {
+            /* readdir error, revert behavior. Highly Unlikely. */
+            _close_fds_by_brute_force(start_fd, end_fd, py_fds_to_keep);
+        }
+        closedir(proc_fd_dir);
+    }
+}
+
+#define _close_open_fd_range _close_open_fd_range_maybe_unsafe
+
+#endif  /* else NOT (defined(__linux__) && defined(HAVE_SYS_SYSCALL_H)) */
+
+
 /*
  * This function is code executed in the child process immediately after fork
  * to set things up and call exec().
@@ -46,12 +292,12 @@ static void child_exec(char *const exec_array[],
                        int errread, int errwrite,
                        int errpipe_read, int errpipe_write,
                        int close_fds, int restore_signals,
-                       int call_setsid, Py_ssize_t num_fds_to_keep,
+                       int call_setsid,
                        PyObject *py_fds_to_keep,
                        PyObject *preexec_fn,
                        PyObject *preexec_fn_args_tuple)
 {
-    int i, saved_errno, fd_num, unused;
+    int i, saved_errno, unused;
     PyObject *result;
     const char* err_msg = "";
     /* Buffer large enough to hold a hex integer.  We can't malloc. */
@@ -113,33 +359,8 @@ static void child_exec(char *const exec_array[],
         POSIX_CALL(close(errwrite));
     }
 
-    /* close() is intentionally not checked for errors here as we are closing */
+    if (close_fds)
-    /* a large range of fds, some of which may be invalid. */
+        _close_open_fd_range(3, max_fd, py_fds_to_keep);
-    if (close_fds) {
-        Py_ssize_t keep_seq_idx;
-        int start_fd = 3;
-        for (keep_seq_idx = 0; keep_seq_idx < num_fds_to_keep; ++keep_seq_idx) {
-            PyObject* py_keep_fd = PySequence_Fast_GET_ITEM(py_fds_to_keep,
-                                                            keep_seq_idx);
-            int keep_fd = PyLong_AsLong(py_keep_fd);
-            if (keep_fd < 0) {  /* Negative number, overflow or not a Long. */
-                err_msg = "bad value in fds_to_keep.";
-                errno = 0;  /* We don't want to report an OSError. */
-                goto error;
-            }
-            if (keep_fd < start_fd)
-                continue;
-            for (fd_num = start_fd; fd_num < keep_fd; ++fd_num) {
-                close(fd_num);
-            }
-            start_fd = keep_fd + 1;
-        }
-        if (start_fd <= max_fd) {
-            for (fd_num = start_fd; fd_num < max_fd; ++fd_num) {
-                close(fd_num);
-            }
-        }
-    }
 
     if (cwd)
         POSIX_CALL(chdir(cwd));
@@ -227,7 +448,7 @@ subprocess_fork_exec(PyObject* self, PyObject *args)
     pid_t pid;
     int need_to_reenable_gc = 0;
     char *const *exec_array, *const *argv = NULL, *const *envp = NULL;
-    Py_ssize_t arg_num, num_fds_to_keep;
+    Py_ssize_t arg_num;
 
     if (!PyArg_ParseTuple(
             args, "OOOOOOiiiiiiiiiiO:fork_exec",
@@ -243,9 +464,12 @@ subprocess_fork_exec(PyObject* self, PyObject *args)
         PyErr_SetString(PyExc_ValueError, "errpipe_write must be >= 3");
         return NULL;
     }
-    num_fds_to_keep = PySequence_Length(py_fds_to_keep);
+    if (PySequence_Length(py_fds_to_keep) < 0) {
-    if (num_fds_to_keep < 0) {
+        PyErr_SetString(PyExc_ValueError, "cannot get length of fds_to_keep");
-        PyErr_SetString(PyExc_ValueError, "bad fds_to_keep");
+        return NULL;
+    }
+    if (_sanity_check_python_fd_sequence(py_fds_to_keep)) {
+        PyErr_SetString(PyExc_ValueError, "bad value(s) in fds_to_keep");
         return NULL;
     }
 
@@ -348,8 +572,7 @@ subprocess_fork_exec(PyObject* self, PyObject *args)
                    p2cread, p2cwrite, c2pread, c2pwrite,
                    errread, errwrite, errpipe_read, errpipe_write,
                    close_fds, restore_signals, call_setsid,
-                   num_fds_to_keep, py_fds_to_keep,
+                   py_fds_to_keep, preexec_fn, preexec_fn_args_tuple);
-                   preexec_fn, preexec_fn_args_tuple);
         _exit(255);
         return NULL;  /* Dead code to avoid a potential compiler warning. */
     }

configure

@@ -6165,7 +6165,7 @@ unistd.h utime.h \
 sys/audioio.h sys/bsdtty.h sys/epoll.h sys/event.h sys/file.h sys/loadavg.h \
 sys/lock.h sys/mkdev.h sys/modem.h \
 sys/param.h sys/poll.h sys/select.h sys/socket.h sys/statvfs.h sys/stat.h \
-sys/termio.h sys/time.h \
+sys/syscall.h sys/termio.h sys/time.h \
 sys/times.h sys/types.h sys/un.h sys/utsname.h sys/wait.h pty.h libutil.h \
 sys/resource.h netpacket/packet.h sysexits.h bluetooth.h \
 bluetooth/bluetooth.h linux/tipc.h spawn.h util.h

configure.in

@@ -1341,7 +1341,7 @@ unistd.h utime.h \
 sys/audioio.h sys/bsdtty.h sys/epoll.h sys/event.h sys/file.h sys/loadavg.h \
 sys/lock.h sys/mkdev.h sys/modem.h \
 sys/param.h sys/poll.h sys/select.h sys/socket.h sys/statvfs.h sys/stat.h \
-sys/termio.h sys/time.h \
+sys/syscall.h sys/termio.h sys/time.h \
 sys/times.h sys/types.h sys/un.h sys/utsname.h sys/wait.h pty.h libutil.h \
 sys/resource.h netpacket/packet.h sysexits.h bluetooth.h \
 bluetooth/bluetooth.h linux/tipc.h spawn.h util.h)

pyconfig.h.in

@@ -789,6 +789,9 @@
 /* Define to 1 if you have the <sys/stat.h> header file. */
 #undef HAVE_SYS_STAT_H
 
+/* Define to 1 if you have the <sys/syscall.h> header file. */
+#undef HAVE_SYS_SYSCALL_H
+
 /* Define to 1 if you have the <sys/termio.h> header file. */
 #undef HAVE_SYS_TERMIO_H