netdev
[Top] [All Lists]

Re: IFF_PROMISC bug?

To: jgarzik@xxxxxxxxxxxxxxxx
Subject: Re: IFF_PROMISC bug?
From: "David S. Miller" <davem@xxxxxxxxxx>
Date: Wed, 13 Feb 2002 03:43:24 -0800 (PST)
Cc: ak@xxxxxxx, linux-net@xxxxxxxxxxxxxxx, netdev@xxxxxxxxxxx, kuznet@xxxxxxxxxxxxx
In-reply-to: <3C6A2789.5DCC1C8F@mandrakesoft.com>
References: <3C6A1EF5.1BF97B99@mandrakesoft.com> <20020213.001736.03366378.davem@redhat.com> <3C6A2789.5DCC1C8F@mandrakesoft.com>
Sender: owner-netdev@xxxxxxxxxxx
   From: Jeff Garzik <jgarzik@xxxxxxxxxxxxxxxx>
   Date: Wed, 13 Feb 2002 03:44:57 -0500

   [sigh, I meant to have said that we can support IFF_PROMISC for
   SIOCGIFFLAGS -only-, until the end of time.  I agree with you that
   SIOCSIFFLAGS+IFF_PROMISC clearly is unsupportable.]
   
Then you will break more than what you claim is broken right now.

At least the current code RESPECTS setting IFF_PROMISC if an app uses
SIOCSIFFLAGS to do it, your version would not.

I think you'll be breaking more apps that way, in fact right off the
bat your version will break all current NMAP binaries out there.
That's just the first thing I checked, there are surely several other
examples.

The more and more I look at this situation, the more and more I
respect Alexey's decision to implement things how they are now.
It is the the only reasonable choice, really...

<Prev in Thread] Current Thread [Next in Thread>