On Tue, Oct 09, 2012 at 08:18:21AM -0600, Eric Blake wrote:
> On 10/09/2012 03:05 AM, Richard W.M. Jones wrote:
> > On Tue, Oct 09, 2012 at 09:54:45AM +0100, Richard W.M. Jones wrote:
> >> The F_DUPFD_CLOEXEC fix that Al Viro posted fixes 3/4 of the
> >> bugs, but I'm still investigating this one:
> >>
> >>> test-dup3.c:108: assertion failed
> >>> # ASSERT (dup3 (fd, fd, o_flags) == -1);
> >
> > Thanks to Jim for pointing out this is a kernel bug, because the
> > proposed standard for dup3 specifies that fd1 != fd2 else error:
> >
> > http://austingroupbugs.net/view.php?id=411
>
> I agree that it is a regression in behavior, and wrote that Austin Group
> wording based on the previous Linux behavior. However, rather than
> claiming it is a kernel bug, we must also consider whether it should be
> fixed in glibc; after all, if the kernel is easier to implement dup2 and
> dup3 alike (with no special casing of dup-to-self), it's not too hard
> for glibc to special-case a dup-to-self check without even calling into
> the kernel.
The fix is pretty simple. It looks as if some code got dropped
when it was moved from one file to another.
Rich.
--
Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones
libguestfs lets you edit virtual machines. Supports shell scripting,
bindings from many languages. http://libguestfs.org
From 0944e30e12dec6544b3602626b60ff412375c78f Mon Sep 17 00:00:00 2001
From: "Richard W.M. Jones" <rjo...@redhat.com>
Date: Tue, 9 Oct 2012 14:42:45 +0100
Subject: [PATCH] dup3: Return an error when oldfd == newfd.
The following commit:
commit fe17f22d7fd0e344ef6447238f799bb49f670c6f
Author: Al Viro <v...@zeniv.linux.org.uk>
Date: Tue Aug 21 11:48:11 2012 -0400
take purely descriptor-related stuff from fcntl.c to file.c
was supposed to be just code motion, but it dropped the following two
lines:
if (unlikely(oldfd == newfd))
return -EINVAL;
from the dup3 system call. dup3 is not specified by POSIX, so Linux
can do what it likes. However the POSIX proposal for dup3 [1] states
that it should return an error if oldfd == newfd.
[1] http://austingroupbugs.net/view.php?id=411
Signed-off-by: Richard W.M. Jones <rjo...@redhat.com>
Tested-by: Richard W.M. Jones <rjo...@redhat.com>
---
fs/file.c | 3 +++
1 file changed, 3 insertions(+)
diff --git a/fs/file.c b/fs/file.c
index 0f1bda4..d3b5fa8 100644
--- a/fs/file.c
+++ b/fs/file.c
@@ -922,6 +922,9 @@ SYSCALL_DEFINE3(dup3, unsigned int, oldfd, unsigned int,
newfd, int, flags)
if ((flags & ~O_CLOEXEC) != 0)
return -EINVAL;
+ if (unlikely(oldfd == newfd))
+ return -EINVAL;
+
if (newfd >= rlimit(RLIMIT_NOFILE))
return -EMFILE;
--
1.7.10.4
