Hi,
couple of things stopped working on my box
where I have /dev/vc/XX as symlinks to /dev/ttyXX, and some
things use /dev/vc/XX and some /dev/ttyXX. After last update
hour ago things which use /dev/vc/XX stopped working for
non-root - they now fail with EACCES error if they attempt
to redirect its input or output through '>' or '<>' bash
redirection operators:
$ touch /tmp/xx
$ ln -s /tmp/xx yy
$ cat > yy
-bash: yy: Permission denied
Strace says:
[pid 3268] open("yy", O_WRONLY|O_CREAT|O_TRUNC|O_LARGEFILE, 0666) = -1 EACCES (Permission denied)
This command succeeds on kernel from thursday. Simillar problem is
that
$ rm /tmp/xx
$ ln -s /tmp/xx yy
$ cat > yy
-bash: yy: No such file or directory
while it creates /tmp/xx file on older kernels.
Currently I suspect Trond's LOOKUP_CONTINUE change in
'[PATCH] Add open intent information to the 'struct nameidata', but I
did not tried reverting it yet to find whether it is culprit or no. But
other changes than these four from Trond looks completely innocent.
Thanks,
Petr Vandrovec
[email protected]
On Mon, 2003-07-07 at 17:46, Petr Vandrovec wrote:
> Hi,
> couple of things stopped working on my box
> where I have /dev/vc/XX as symlinks to /dev/ttyXX, and some
> things use /dev/vc/XX and some /dev/ttyXX. After last update
> hour ago things which use /dev/vc/XX stopped working for
> non-root - they now fail with EACCES error if they attempt
> to redirect its input or output through '>' or '<>' bash
> redirection operators:
>
> $ touch /tmp/xx
> $ ln -s /tmp/xx yy
> $ cat > yy
> -bash: yy: Permission denied
>
> Strace says:
>
> [pid 3268] open("yy", O_WRONLY|O_CREAT|O_TRUNC|O_LARGEFILE, 0666) = -1 EACCES (Permission denied)
>
> This command succeeds on kernel from thursday. Simillar problem is
> that
>
> $ rm /tmp/xx
> $ ln -s /tmp/xx yy
> $ cat > yy
> -bash: yy: No such file or directory
>
> while it creates /tmp/xx file on older kernels.
>
> Currently I suspect Trond's LOOKUP_CONTINUE change in
> '[PATCH] Add open intent information to the 'struct nameidata', but I
> did not tried reverting it yet to find whether it is culprit or no. But
> other changes than these four from Trond looks completely innocent.
JFYI, on Red Hat 9, I'm having problems when using "vi" to edit a file
through a symbolic link. For example, I can't save changes with ":wq!"
in vi when editing "/etc/grub.conf" on my RHL9 box (which is a symbolic
link to "/boot/grub/grub.conf").
On Mon, Jul 07, 2003 at 08:49:29PM +0200, Felipe Alfaro Solana wrote:
> On Mon, 2003-07-07 at 17:46, Petr Vandrovec wrote:
> > Hi,
> > couple of things stopped working on my box
> > where I have /dev/vc/XX as symlinks to /dev/ttyXX, and some
> > things use /dev/vc/XX and some /dev/ttyXX. After last update
> > hour ago things which use /dev/vc/XX stopped working for
> > non-root - they now fail with EACCES error if they attempt
> > to redirect its input or output through '>' or '<>' bash
> > redirection operators:
> >
> > $ touch /tmp/xx
> > $ ln -s /tmp/xx yy
> > $ cat > yy
> > -bash: yy: Permission denied
> >
> > Strace says:
> >
> > [pid 3268] open("yy", O_WRONLY|O_CREAT|O_TRUNC|O_LARGEFILE, 0666) = -1 EACCES (Permission denied)
> > Currently I suspect Trond's LOOKUP_CONTINUE change in
> > '[PATCH] Add open intent information to the 'struct nameidata', but I
> > did not tried reverting it yet to find whether it is culprit or no. But
> > other changes than these four from Trond looks completely innocent.
Reverting "[email protected]|ChangeSet|20030704190606|14242" indeed
fixes problem.
> JFYI, on Red Hat 9, I'm having problems when using "vi" to edit a file
> through a symbolic link. For example, I can't save changes with ":wq!"
> in vi when editing "/etc/grub.conf" on my RHL9 box (which is a symbolic
> link to "/boot/grub/grub.conf").
I think that it is unrelated.
Petr Vandrovec
[email protected]
>>>>> " " == Petr Vandrovec <[email protected]> writes:
> Hi,
> couple of things stopped working on my box
> where I have /dev/vc/XX as symlinks to /dev/ttyXX, and some
> things use /dev/vc/XX and some /dev/ttyXX. After last update
> hour ago things which use /dev/vc/XX stopped working for
> non-root - they now fail with EACCES error if they attempt to
> redirect its input or output through '>' or '<>' bash
> redirection operators:
Whoops. Looks like I missed an assumption in open_namei().
Does the following patch fix the problem?
Cheers,
Trond
--- linux-2.5.74-up/fs/namei.c.orig 2003-06-30 08:49:25.000000000 +0200
+++ linux-2.5.74-up/fs/namei.c 2003-07-07 21:25:00.000000000 +0200
@@ -1344,6 +1344,7 @@
* stored in nd->last.name and we will have to putname() it when we
* are done. Procfs-like symlinks just set LAST_BIND.
*/
+ nd->flags |= LOOKUP_PARENT;
error = security_inode_follow_link(dentry, nd);
if (error)
goto exit_dput;
@@ -1352,6 +1353,7 @@
dput(dentry);
if (error)
return error;
+ nd->flags &= ~LOOKUP_PARENT;
if (nd->last_type == LAST_BIND) {
dentry = nd->dentry;
goto ok;
On Mon, Jul 07, 2003 at 09:30:34PM +0200, Trond Myklebust wrote:
> >>>>> " " == Petr Vandrovec <[email protected]> writes:
>
> > Hi,
> > couple of things stopped working on my box
> > where I have /dev/vc/XX as symlinks to /dev/ttyXX, and some
> > things use /dev/vc/XX and some /dev/ttyXX. After last update
> > hour ago things which use /dev/vc/XX stopped working for
> > non-root - they now fail with EACCES error if they attempt to
> > redirect its input or output through '>' or '<>' bash
> > redirection operators:
>
> Whoops. Looks like I missed an assumption in open_namei().
> Does the following patch fix the problem?
Yes, it fixes problem. Thanks,
Petr Vandrovec
> Cheers,
> Trond
>
> --- linux-2.5.74-up/fs/namei.c.orig 2003-06-30 08:49:25.000000000 +0200
> +++ linux-2.5.74-up/fs/namei.c 2003-07-07 21:25:00.000000000 +0200
> @@ -1344,6 +1344,7 @@
> * stored in nd->last.name and we will have to putname() it when we
> * are done. Procfs-like symlinks just set LAST_BIND.
> */
> + nd->flags |= LOOKUP_PARENT;
> error = security_inode_follow_link(dentry, nd);
> if (error)
> goto exit_dput;
> @@ -1352,6 +1353,7 @@
> dput(dentry);
> if (error)
> return error;
> + nd->flags &= ~LOOKUP_PARENT;
> if (nd->last_type == LAST_BIND) {
> dentry = nd->dentry;
> goto ok;
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to [email protected]
> More majordomo info at http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at http://www.tux.org/lkml/
>
Trond Myklebust <[email protected]> wrote:
>
> Whoops. Looks like I missed an assumption in open_namei().
> Does the following patch fix the problem?
Yes.