netdev - Re: should __LINK_STATE_* enums really be restricted to generic queueing layer?

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <20180824134917.GB1999@lunn.ch>
Date:   Fri, 24 Aug 2018 15:49:17 +0200
From:   Andrew Lunn <andrew@...n.ch>
To:     rpjday@...shcourse.ca
Cc:     netdev@...r.kernel.org
Subject: Re: should __LINK_STATE_* enums really be restricted to generic
 queueing layer?

On Fri, Aug 24, 2018 at 07:17:38AM -0400, rpjday@...shcourse.ca wrote:
>   more curious pedantry ... in netdevice.h, one reads:
> 
>   /* These flag bits are private to the generic network queueing
>    * layer; they may not be explicitly referenced by any other
>    * code.
>    */
> 
>   enum netdev_state_t {
>         __LINK_STATE_START,
>         __LINK_STATE_PRESENT,
>         __LINK_STATE_NOCARRIER,
>         __LINK_STATE_LINKWATCH_PENDING,
>         __LINK_STATE_DORMANT,
>   };
> 
> ok, but there are definitely a few examples of what look like
> non-generic network queueing code referencing those enums.
> 
>   one example, drivers/net/ethernet/amd/sun3lance.c, which openly
> (and amusingly) admits that it shouldn't be doing this:
> 
>         dev->netdev_ops = &lance_netdev_ops;
>   //      KLUDGE -- REMOVE ME
>         set_bit(__LINK_STATE_PRESENT, &dev->state);
>         return 1;
>   }
> 
> there are a few more places (admittedly, not many) -- is this
> acceptable behaviour?

Not in a new driver.

The sun3lance is a very old driver. I have misty memories of a Sun
3/60. There is nothing in this code about a driving a PHY. Maybe using
netif_carrier_on() would work, but without having access to the
hardware to test, i would just leave it alone.

       Andrew