1
0
mirror of git://git.sv.gnu.org/coreutils.git synced 2026-08-18 17:39:22 +02:00

Add a test for cp -p and existing file permissions.

* tests/cp/Makefile.am (TESTS): Add existing-perm-race.
* tests/cp/existing-perm-race: New test.  It isn't much of a
test yet, since it's hard to catch the race, but it has a FIXME
that will let us do a better test later.
This commit is contained in:
Paul Eggert
2007-12-01 10:10:52 +01:00
committed by Jim Meyering
parent b64119bc54
commit 20ef6c7b9c
3 changed files with 103 additions and 0 deletions
+8
View File
@@ -1,3 +1,11 @@
2007-11-29 Paul Eggert <eggert@cs.ucla.edu>
Add a test for cp -p and existing file permissions.
* tests/cp/Makefile.am (TESTS): Add existing-perm-race.
* tests/cp/existing-perm-race: New test. It isn't much of a
test yet, since it's hard to catch the race, but it has a FIXME
that will let us do a better test later.
2007-11-28 Paul Eggert <eggert@cs.ucla.edu>
Fix a security race with "cp -p A B" when B already exists.
+1
View File
@@ -21,6 +21,7 @@ TESTS = \
thru-dangling \
cp-a-selinux \
file-perm-race parent-perm-race \
existing-perm-race \
backup-dir \
src-base-dot \
sparse \
+94
View File
@@ -0,0 +1,94 @@
#!/bin/sh
# Make sure cp -p isn't too generous with existing file permissions.
# Copyright (C) 2006-2007 Free Software Foundation, Inc.
# This program is free software: you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
# the Free Software Foundation, either version 3 of the License, or
# (at your option) any later version.
# This program 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 for more details.
# You should have received a copy of the GNU General Public License
# along with this program. If not, see <http://www.gnu.org/licenses/>.
if test "$VERBOSE" = yes; then
set -x
cp --version
fi
. $srcdir/../group-names
. $srcdir/../test-lib.sh
set _ $groups; shift
g1=$1
g2=$2
fail=0
umask 077
mkfifo fifo || {
echo "$0: fifos not supported; skipping this test." 1>&2
(exit 77); exit 77
}
touch fifo-copy &&
chgrp $g1 fifo &&
chgrp $g2 fifo-copy &&
chmod g+r fifo-copy || framework-failure
# Copy a fifo's contents. That way, we can examine the
# destination permissions before they're finalized.
cp -p --copy-contents fifo fifo-copy &
cp_pid=$!
(
# Now 'cp' is reading the fifo. Wait for the destination file to
# be written to, encouraging things along by echoing to the fifo.
while test ! -s fifo-copy; do
echo foo
done
# Check the permissions of the destination.
ls -l -n fifo-copy >ls.out &&
# Close the fifo so that "cp" can continue. But output first,
# before exiting, otherwise some shells would optimize away the file
# descriptor that holds the fifo open.
echo foo
) >fifo || fail=1
# Check that the destination mode is safe while the file is being copied.
read mode links owner group etc <ls.out || fail=1
case $mode in
-rw-------*) ;;
# FIXME: Remove the following case; the file mode should always be
# 600 while the data are being copied. This will require changing
# cp so that it also does not put $g1's data in a file that is
# accessible to $g2. This fix will not close a security hole, since
# a $g2 process can maintain an open file descriptor to the
# destination, but it's safer anyway.
-rw-r-----*)
# If the file has group $g1 and is group-readable, that is definitely
# bogus, as neither the source nor the destination was readable to group $g1.
test "$group" = "$g1" && fail=1;;
*) fail=1;;
esac
wait $cp_pid || fail=1
# Check that the final mode and group are right.
ls -l -n fifo-copy >ls.out &&
read mode links owner group etc <ls.out || fail=1
case $mode in
-rw-------*) test "$group" = "$g1" || fail=1;;
*) fail=1;;
esac
(exit $fail); exit $fail