Yes.  I am completely at a loss.  The Linux kernel version I updated to
is 2.6.15.3.  After chmoding 666 on /dev/tty, I changed it back to 777
because it is definitely a directory.  Evidence below:

[EMAIL PROTECTED]:/dev/tty# ls -l
total 0
crw-------  1 root root 3, 10 2007-03-21 00:58 s
crw-------  1 root root 3,  0 2007-03-21 00:58 s0
crw-------  1 root root 3,  1 2007-03-21 00:58 s1
crw-------  1 root root 3,  2 2007-03-21 00:58 s2
crw-------  1 root root 3,  3 2007-03-21 00:58 s3
crw-------  1 root root 3,  4 2007-03-21 00:58 s4
crw-------  1 root root 3,  5 2007-03-21 00:58 s5
crw-------  1 root root 3,  6 2007-03-21 00:58 s6
crw-------  1 root root 3,  7 2007-03-21 00:58 s7
crw-------  1 root root 3,  8 2007-03-21 00:58 s8
crw-------  1 root root 3,  9 2007-03-21 00:58 s9

On Wed, 2006-04-19 at 08:11 -0400, Greg Wooledge wrote:
> On Tue, Apr 18, 2006 at 09:58:24AM -0500, Christ, Bryan wrote:
> > Most of the suggestions I have read say to chmod 666 /dev/tty, but
> > my /dev/tty is a directory.
> 
> That's bad.  That's very, very bad.  I'd suggest you get in touch with
> one of the support forums (mailing lists, IRC channels, etc.) for your
> operating system.
> 
> > ssh [EMAIL PROTECTED]
> > Permission denied, please try again.
> > Permission denied, please try again.
> > Permission denied (publickey,gssapi-with-mic,password).
> 
> If you did indeed issue "chmod 666" on a directory, that might explain
> part of the problem -- a directory which lacks the "execute" bit would
> be untraversable.
> 
> > debug1: read_passphrase: can't open /dev/tty: Is a directory
> > debug3: packet_send2: adding 8 (len 51 padlen 5 extra_pad 64)
> > debug2: we sent a password packet, wait for reply
> > debug1: Authentications that can continue:
> > publickey,gssapi-with-mic,password
> > Permission denied, please try again.
> 
> *nod*  Whatever your Linux distribution has done, fixing it is probably
> outside the scope of this mailing list.  /dev/tty is supposed to be a
> character device node.  Shell scripts and other Unix programs have *always*
> been able to count on "read foo < /dev/tty" working.  If /dev/tty is a
> directory, that will break a *lot* of stuff.
> 
> I'm hesitant to suggest even something as simple as "man MAKEDEV", for
> fear that any attempt to fix this snafu (without understanding the
> primary cause) will just make it worse.

Reply via email to