Date: Thu, 8 Sep 2005 02:51:32 +0300 From: Giorgos Keramidas <keramida@freebsd.org> To: Craig Edwards <brain@winbot.co.uk> Cc: freebsd-security@freebsd.org, talonz <talonz@gmail.com> Subject: Re: ee using 99% cpu after user ssh session terminates abnormaly Message-ID: <20050907235132.GB13522@orion.daedalusnetworks.priv> In-Reply-To: <431F7183.7080405@winbot.co.uk> References: <431F6941.20006@gmail.com> <20050907223748.GB563@odin.ac.hmc.edu> <431F7183.7080405@winbot.co.uk>
next in thread | previous in thread | raw e-mail | index | archive | help
On 2005-09-08 00:02, Craig Edwards <brain@winbot.co.uk> wrote: > At least this is what i suspect happens. Lazy programming somewhere... > Brooks Davis wrote: > >On Thu, Sep 08, 2005 at 08:27:13AM +1000, talonz wrote: > >>Recently i have been using a dialup 56k account to access the net > >>and have noticed that when my ssh session times out and I am editing > >>a file in ` ee ' the system goes to 99% cpu usage and stays like > >>this till the pid is killed. This is a standard user account (not > >>root/su) > >> > >>Would a user be able to create a denial of service condition on the > >>remote system using this bug? > > > >No more then they could with the ablity to run any other program that > >loops. > > I can duplicate this with nano on freebsd 5.4 and 5.2.1 > > It seems that the process ignores the HUP signal maybe or ignores the > EOF condition on stdin, and the select loop, or whatever it uses, just > loops infinitely with nothing to read, constantly returning an error > condition. FWIW, pico seems to handle HUP just fine. So whatever causes nano to enter a loop is something that is done differently in nano.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20050907235132.GB13522>