[226] in Athena Bugs
Re: 6.0a VAX /bin/login password problems
daemon@ATHENA.MIT.EDU (Mark W. Eichin)
Tue Apr 19 03:39:35 1988
Date: Tue, 19 Apr 88 03:38:29 EDT
From: Mark W. Eichin <eichin@ATHENA.MIT.EDU>
To: Jerome H Saltzer <Saltzer@ATHENA.MIT.EDU>
Cc: henry@GARP.MIT.EDU, vs-testers@ATHENA.MIT.EDU, bugs@ATHENA.MIT.EDU,
In-Reply-To: Jerome H. Saltzer's message of Mon, 18 Apr 88 13:13:51 EST <8804181813.AA13739@HERACLES.MIT.EDU>
>Date: Mon, 18 Apr 88 13:13:51 EST
>From: Jerome H. Saltzer <Saltzer@ATHENA.MIT.EDU>
>> Well, Mark, I am unable to duplicate this on your workstation. I
>I wonder if the problem is that some /bin/login bug is triggered by
>the passage of time rather than the exact number of password
>attempts. It may be that some signal goes off after N (= 60?)
>seconds that leaves /bin/login uninterested in attaching the home
>directory. Under that hypothesis, if you get your second
>password-typing attempt in before the deadline it still works right.
> Jerry
I type very fast; N must be under 15 seconds for that to be true.
Some further investigation reveals that /bin/athena/attach and
/etc/mount are definitely being run, it seems that the mount is just
failing for some reason... and @flame{of course, there is no
reasonable error return from mount, nor is it checked by attach, so}
we are unable to find any error returns...
Further note: I tried logging into EPIMETHEUS (6.0a vs2000) HEIMDALL
(VS2/RC) and THOR (VS2) and had no problem (got my homedir, the bug
did *not* show up.). Note that the last two had users, so any
load/timing based comparisons are off.
Maybe we should ask the whole staff to try mistyping their passwords
once, and then getting it right, just to see if any other combinations
of user/machine have the problem. Then again, I haven't yet tested it
on ttyv0 of fyunch.
To my knowledge, the only recorded and repeatable observation of this
is *myself* telnet'ing into *FYUNCH*. It is however consistently
repeatable. If we can find other examples, it might deserve some
serious attention; if we can make this example go away, as it seems to
be just one weird combination, that might suffice.
_Mark_
