Re: permissions of /proc/tty/driver

Re: permissions of /proc/tty/driver

[Christoph Hellwig]

On Fri, Jan 14, 2005 at 06:45:25PM +0100, Thomas Viehmann wrote:
> Hi.
> 
> This may not be stritly on topic, but I couln't figure out a better 
> place to ask:
> 
> During the packaging of an application, I have the following problem:
> I would like to run a daemon as non-root. The daemon likes to 
> (continually) check /proc/tty/driver/usbserial to see whether or not 
> interesting USB devices  are connected. The permissions of this actual 
> file is (on a kernel compiled from Debian's kernel-source-2.6.10) 0444, 
> so this isn't a problem. However, the parent directory /proc/tty/driver 
> is 0500. I'm not sure whether this is related to Debian DSAs 358 or 423 
> (where /proc/tty/driver/serial is mentioned as leaking sensitive 
> information), to me the contents of usbserial look innocent enough.
> Do you have any hints on what might be a good solution?

The permissions on the directory look indeed too strict to me.  It might
be better to just use strict permissions on /proc/tty/driver/serial
indeed.

Counter-question:  What information is available in
/proc/tty/driver/usbserial but not in sysfs?

Re: permissions of /proc/tty/driver

[Thomas Viehmann]

Hi.

Christoph Hellwig wrote:
> Counter-question:  What information is available in
> /proc/tty/driver/usbserial but not in sysfs?

Thanks for this hint, is there a way of finding vendor and product ids 
of all ttyUSB devices better than
looking for /sys/bus/usb/devices/*-*/*-*:*/ttyUSB* and then ckecking the 
obvious files in the grandparent directory?

Kind regards

Thomas
-- 
Thomas Viehmann, <http://thomas.viehmann.net/>

Re: permissions of /proc/tty/driver

[Sergey Vlasov]

On Sun, 16 Jan 2005 14:13:46 +0100, Thomas Viehmann wrote:

> Christoph Hellwig wrote:
>> Counter-question:  What information is available in
>> /proc/tty/driver/usbserial but not in sysfs?
> 
> Thanks for this hint, is there a way of finding vendor and product ids 
> of all ttyUSB devices better than
> looking for /sys/bus/usb/devices/*-*/*-*:*/ttyUSB* and then ckecking the 
> obvious files in the grandparent directory?

/sys/bus/usb-serial/devices/* looks like what you need...

Re: permissions of /proc/tty/driver

[Christoph Hellwig]

On Sun, Jan 16, 2005 at 02:13:46PM +0100, Thomas Viehmann wrote:
> Hi.
> 
> Christoph Hellwig wrote:
> >Counter-question:  What information is available in
> >/proc/tty/driver/usbserial but not in sysfs?
> 
> Thanks for this hint, is there a way of finding vendor and product ids 
> of all ttyUSB devices better than
> looking for /sys/bus/usb/devices/*-*/*-*:*/ttyUSB* and then ckecking the 
> obvious files in the grandparent directory?

I think that's the obvious way.  Using libsysfs will make your life much
easier when doing that, though.

Re: permissions of /proc/tty/driver

[Alan Cox]

On Sul, 2005-01-16 at 12:04, Christoph Hellwig wrote:
> > (where /proc/tty/driver/serial is mentioned as leaking sensitive 
> > information), to me the contents of usbserial look innocent enough.
> > Do you have any hints on what might be a good solution?
> 
> The permissions on the directory look indeed too strict to me.  It might
> be better to just use strict permissions on /proc/tty/driver/serial
> indeed.

The file containts transmit and receive byte counts, which means you can
both measure intercharacter delay and character count. Thats a big help
to password guessers

Re: permissions of /proc/tty/driver

[Christoph Hellwig]

On Sun, Jan 16, 2005 at 09:11:03PM +0000, Alan Cox wrote:
> On Sul, 2005-01-16 at 12:04, Christoph Hellwig wrote:
> > > (where /proc/tty/driver/serial is mentioned as leaking sensitive 
> > > information), to me the contents of usbserial look innocent enough.
> > > Do you have any hints on what might be a good solution?
> > 
> > The permissions on the directory look indeed too strict to me.  It might
> > be better to just use strict permissions on /proc/tty/driver/serial
> > indeed.
> 
> The file containts transmit and receive byte counts, which means you can
> both measure intercharacter delay and character count. Thats a big help
> to password guessers

I know.  But that doesn't explain why we don't keep strict permissions
only on that file but on the directory.

Re: permissions of /proc/tty/driver

[Arjan van de Ven]

On Sun, 2005-01-16 at 23:26 +0100, Christoph Hellwig wrote:
> On Sun, Jan 16, 2005 at 09:11:03PM +0000, Alan Cox wrote:
> > On Sul, 2005-01-16 at 12:04, Christoph Hellwig wrote:
> > > > (where /proc/tty/driver/serial is mentioned as leaking sensitive 
> > > > information), to me the contents of usbserial look innocent enough.
> > > > Do you have any hints on what might be a good solution?
> > > 
> > > The permissions on the directory look indeed too strict to me.  It might
> > > be better to just use strict permissions on /proc/tty/driver/serial
> > > indeed.
> > 
> > The file containts transmit and receive byte counts, which means you can
> > both measure intercharacter delay and character count. Thats a big help
> > to password guessers
> 
> I know.  But that doesn't explain why we don't keep strict permissions
> only on that file but on the directory.

ls -la on the file gives you the size maybe ?

Re: permissions of /proc/tty/driver

[Alan Cox]

On Llu, 2005-01-17 at 08:14, Arjan van de Ven wrote:
> On Sun, 2005-01-16 at 23:26 +0100, Christoph Hellwig wrote:
> > I know.  But that doesn't explain why we don't keep strict permissions
> > only on that file but on the directory.
> 
> ls -la on the file gives you the size maybe ?

I went for a dig in the archives

The original fix removed the tx/rx data from the file if you weren't
priviledged. Linus did the directory hack because he didn't want to
worry about drivers that got missed out/not fixed.

So there's a janitor project there - to go through all the tty/serial
drivers and make sure they don't give out excessively useful information
to non CAP_SYS_RAWIO users, then loosen permissions.