Hi, On Fri, Sep 21, 2007 at 12:42:36AM -0700, mpb wrote: > I believe the following script demonstrates that readlink cannot > always canonicalize paths. It may be possible to create a simpler > demonstration of the bug, but I could not trivially do so. [...] > path=/tmp/c0/b3/b4/a5/a6 [...] > readlink -e $path
$ readlink -ev /tmp/c0/b3/b4/a5/a6
readlink: /tmp/c0/b3/b4/a5/a6: Too many levels of symbolic links
$ readlink -ev /tmp/c0/b3/b4/a5
readlink: /tmp/c0/b3/b4/a5: Too many levels of symbolic links
$ strace -e lstat64 stat -c '' /tmp/c0/b3/b4/a5/a6
lstat64("/tmp/c0/b3/b4/a5/a6", {st_mode=S_IFREG|0644, st_size=9, ...}) = 0
$ strace -e lstat64,readlink readlink -e /tmp/c0/b3/b4/a5/a6
lstat64("/tmp", {st_mode=S_IFDIR|S_ISVTX|0777, st_size=4096, ...}) = 0
lstat64("/tmp/c0", {st_mode=S_IFDIR|0755, st_size=4096, ...}) = 0
lstat64("/tmp/c0/b3", {st_mode=S_IFLNK|0777, st_size=13, ...}) = 0
readlink("/tmp/c0/b3", "/tmp/a0/a2/b3", 14) = 13
lstat64("/tmp", {st_mode=S_IFDIR|S_ISVTX|0777, st_size=4096, ...}) = 0
lstat64("/tmp/a0", {st_mode=S_IFDIR|0755, st_size=4096, ...}) = 0
lstat64("/tmp/a0/a2", {st_mode=S_IFLNK|0777, st_size=5, ...}) = 0
readlink("/tmp/a0/a2", "a1/a2", 6) = 5
lstat64("/tmp/a0/a1", {st_mode=S_IFDIR|0755, st_size=4096, ...}) = 0
lstat64("/tmp/a0/a1/a2", {st_mode=S_IFDIR|0755, st_size=4096, ...}) = 0
lstat64("/tmp/a0/a1/a2/b3", {st_mode=S_IFDIR|0755, st_size=4096, ...}) = 0
lstat64("/tmp/a0/a1/a2/b3/b4", {st_mode=S_IFDIR|0755, st_size=4096, ...}) = 0
lstat64("/tmp/a0/a1/a2/b3/b4/a5", {st_mode=S_IFLNK|0777, st_size=19, ...}) = 0
readlink("/tmp/a0/a1/a2/b3/b4/a5", "/tmp/a0/a2/a3/a4/a5", 20) = 19
lstat64("/tmp", {st_mode=S_IFDIR|S_ISVTX|0777, st_size=4096, ...}) = 0
lstat64("/tmp/a0", {st_mode=S_IFDIR|0755, st_size=4096, ...}) = 0
lstat64("/tmp/a0/a2", {st_mode=S_IFLNK|0777, st_size=5, ...}) = 0
That is, /tmp/a0/a2 is checked twice, and cycle_check() aborts the
traversal.
--
ldv
pgpIjTi4ntcB6.pgp
Description: PGP signature
_______________________________________________ Bug-coreutils mailing list Bug-coreutils@gnu.org http://lists.gnu.org/mailman/listinfo/bug-coreutils
