On Thu, Sep 24, 2009 at 09:31:45PM -0600, Eric Blake wrote:
>-----BEGIN PGP SIGNED MESSAGE-----
>Hash: SHA1
>
>According to Christopher Faylor on 9/23/2009 10:41 AM:
>>> Also less risky would be to make changes locally in mkdir, link, and
>>> rename for now.
>
>Done - this patch narrows the scope of the changes to just the interfaces
>in question. I've also tested that it made it through the coreutils
>testsuite without any regressions.
>
>>
>> I'm not clear if this is a regression or not. If it isn't a regression,
>> I'd opt for leaving it until 1.7.2.
>
>Now that I'm not touching path.cc, these are all much more self-contained,
>and make cygwin more consistent with Linux. For example:
>
>touch a
>ln -s c b
>link a b/
>
>should fail because b/ is not an existing directory, but without this
>patch, it succeeds and creates the regular file c as a link to a.
>
>2009-09-24 Eric Blake <e...@byu.net>
>
> * syscalls.cc (link): Delete obsolete comment. Reject directories
> and missing source up front.
> (rename): Use correct errno for trailing '.'. Allow trailing
> slash to newpath iff oldpath is directory.
> * dir.cc (mkdir): Reject dangling symlink with trailing slash.
> * fhandler_disk_file.cc (fhandler_disk_file::link): Reject
> trailing slash.
> * fhandler.cc (fhandler_base::link): Match Linux errno.
>
>- --
>Don't work too hard, make some time for fun as well!
>
>Eric Blake e...@byu.net
>-----BEGIN PGP SIGNATURE-----
>Version: GnuPG v1.4.9 (Cygwin)
>Comment: Public key at home.comcast.net/~ericblake/eblake.gpg
>Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/
>
>iEYEARECAAYFAkq8OaEACgkQ84KuGfSFAYAiLACghYLCFaIGmFR4AuzKAmBuQcg/
>kFoAoJcX+ufE6YUq7S1AeVRvHtyZ30wc
>=4otJ
>-----END PGP SIGNATURE-----
>diff --git a/winsup/cygwin/dir.cc b/winsup/cygwin/dir.cc
>index 2b9125f..b7c31e4 100644
>--- a/winsup/cygwin/dir.cc
>+++ b/winsup/cygwin/dir.cc
>@@ -1,6 +1,7 @@
> /* dir.cc: Posix directory-related routines
>
>- Copyright 1996, 1997, 1998, 1999, 2000, 2001, 2002, 2006, 2007 Red Hat,
>Inc.
>+ Copyright 1996, 1997, 1998, 1999, 2000, 2001, 2002, 2006, 2007,
>+ 2009 Red Hat, Inc.
>
> This file is part of Cygwin.
>
>@@ -21,6 +22,7 @@ details. */
> #include "dtable.h"
> #include "cygheap.h"
> #include "cygtls.h"
>+#include "tls_pbuf.h"
>
> extern "C" int
> dirfd (DIR *dir)
>@@ -273,11 +275,26 @@ mkdir (const char *dir, mode_t mode)
> {
> int res = -1;
> fhandler_base *fh = NULL;
>+ size_t dlen;
>+ char *newbuf;
>+ tmp_pathbuf tp;
>
> myfault efault;
> if (efault.faulted (EFAULT))
> return -1;
>
>+ /* POSIX says mkdir("symlink-to-missing/") should create the
>+ directory "missing", but Linux rejects it with EEXIST. Copy
>+ Linux behavior for now. */
>+
>+ dlen = strlen (dir);
>+ if (isdirsep (dir[dlen - 1]))
Couldn't this index negatively if dir is zero length?
>+ {
>+ stpcpy (newbuf = tp.c_get (), dir);
Since stpcpy returns a pointer to the end of the buffer couldn't we use that
and do pointer arithmetic rather than index arithmetic?
>+ while (dlen > 0 && isdirsep (dir[dlen - 1]))
>+ newbuf[--dlen] = '\0';
>+ dir = newbuf;
>+ }
> if (!(fh = build_fh_name (dir, NULL, PC_SYM_NOFOLLOW)))
> goto done; /* errno already set */;
>
>diff --git a/winsup/cygwin/fhandler.cc b/winsup/cygwin/fhandler.cc
>index b52d7c2..44311ca 100644
>--- a/winsup/cygwin/fhandler.cc
>+++ b/winsup/cygwin/fhandler.cc
>@@ -1541,7 +1541,7 @@ fhandler_base::ftruncate (_off64_t length, bool
>allow_truncate)
> int
> fhandler_base::link (const char *newpath)
> {
>- set_errno (EINVAL);
>+ set_errno (EPERM);
> return -1;
> }
>
>diff --git a/winsup/cygwin/fhandler_disk_file.cc
>b/winsup/cygwin/fhandler_disk_file.cc
>index 214be47..4c4b559 100644
>--- a/winsup/cygwin/fhandler_disk_file.cc
>+++ b/winsup/cygwin/fhandler_disk_file.cc
>@@ -1186,6 +1186,7 @@ fhandler_disk_file::ftruncate (_off64_t length, bool
>allow_truncate)
> int
> fhandler_disk_file::link (const char *newpath)
> {
>+ size_t nlen = strlen (newpath);
> path_conv newpc (newpath, PC_SYM_NOFOLLOW | PC_POSIX, stat_suffixes);
> if (newpc.error)
> {
>@@ -1200,7 +1201,13 @@ fhandler_disk_file::link (const char *newpath)
> return -1;
> }
>
>- char new_buf[strlen (newpath) + 5];
>+ if (isdirsep (newpath[nlen - 1]) || has_dot_last_component (newpath, false))
Same observation: Couldn't newpath be zero length?
cgf
