Date: Mon, 4 May 2009 19:30:03 GMT From: Jaakko Heinonen <jh@saunalahti.fi> To: freebsd-bugs@FreeBSD.org Subject: Re: bin/133907: cp(1) wrongly reports errors in vacuous copy Message-ID: <200905041930.n44JU3nG077786@freefall.freebsd.org>
next in thread | raw e-mail | index | archive | help
The following reply was made to PR bin/133907; it has been noted by GNATS.
From: Jaakko Heinonen <jh@saunalahti.fi>
To: james@jrv.org
Cc: bug-followup@FreeBSD.org
Subject: Re: bin/133907: cp(1) wrongly reports errors in vacuous copy
Date: Mon, 4 May 2009 22:21:38 +0300
Hi,
On 2009-04-22, james@jrv.org wrote:
> $ mkdir x y
> $ cp -Rp x/ y
> cp: utimes: y/x: No such file or directory
> cp: chown: y/x: No such file or directory
> cp: chmod: y/x: No such file or directory
> cp: chflags: y/x: No such file or directory
> $
>
> Appears to affect only the vacuous case of arg x being an empty
> directory, and only if -p is used.
The problem is that in empty directory case fts(3) returns the path name
without trailing slash in post-order phase. This results an incorrect
value for "base".
It's somewhat unexpected that fts(3) returns with different path name in
post-order because the fts(3) manual page states:
FTS_DP A directory being visited in post-order. The
contents of the FTSENT structure will be
unchanged from when it was returned in pre-
order, i.e., with the fts_info field set to
FTS_D.
This happens only with FTS_NOCHDIR option used by cp(1).
However there is actually no need to set the base value in post-order
phase. This patch should fix the problem.
--- patch begins here ---
Index: bin/cp/cp.c
===================================================================
--- bin/cp/cp.c (revision 191680)
+++ bin/cp/cp.c (working copy)
@@ -316,7 +316,8 @@ copy(char *argv[], enum op type, int fts
* Since the first level MUST be FTS_ROOTLEVEL, base
* is always initialized.
*/
- if (curr->fts_level == FTS_ROOTLEVEL) {
+ if (curr->fts_level == FTS_ROOTLEVEL &&
+ curr->fts_info != FTS_DP) {
if (type != DIR_TO_DNE) {
p = strrchr(curr->fts_path, '/');
base = (p == NULL) ? 0 :
--- patch ends here ---
The fts(3) problem may be worth of fixing but I think that the cp(1)
change is justified because the fts(3) (mis)behavior seems to exist on
other OSes too.
--
Jaakko
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200905041930.n44JU3nG077786>