netdev - Re: uapi: MAX_ADDR_LEN vs. numeric 32

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 for Android: free password hash cracker in your pocket

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

Date:   Sat, 5 Aug 2017 01:25:19 +0300
From:   "Dmitry V. Levin" <ldv@...linux.org>
To:     Mikko Rapeli <mikko.rapeli@....fi>
Cc:     netdev@...r.kernel.org, Pavel Emelyanov <xemul@...tuozzo.com>
Subject: Re: uapi: MAX_ADDR_LEN vs. numeric 32

Hi,

On Sat, Aug 05, 2017 at 12:33:25AM +0300, Mikko Rapeli wrote:
> First, thanks Dmitry for fixing several uapi compilation problems in
> user space. I got a bit demotivated

That's quite understandable.

> about the slow review progress, e.g.
> no feedback what so ever, on some of the patches, but lets try again...
> 
> I rebased my tree now and saw
> 
> commit 745cb7f8a5de0805cade3de3991b7a95317c7c73
> Author: Dmitry V. Levin <ldv@...linux.org>
> Date:   Tue Mar 7 23:50:50 2017 +0300
> 
>     uapi: fix linux/packet_diag.h userspace compilation error
> 
> which does:
> 
> --- a/include/uapi/linux/packet_diag.h
> +++ b/include/uapi/linux/packet_diag.h
> @@ -64,7 +64,7 @@ struct packet_diag_mclist {
>         __u32   pdmc_count;
>         __u16   pdmc_type;
>         __u16   pdmc_alen;
> -       __u8    pdmc_addr[MAX_ADDR_LEN];
> +       __u8    pdmc_addr[32]; /* MAX_ADDR_LEN */
>  };
>  
>  struct packet_diag_ring {
> 
> In my tree I had fixed that case with:
> 
> --- a/include/uapi/linux/packet_diag.h
> +++ b/include/uapi/linux/packet_diag.h
> @@ -2,6 +2,7 @@
>  #define __PACKET_DIAG_H__
>  
>  #include <linux/types.h>
> +#include <linux/netdevice.h>
>  
>  struct packet_diag_req {
>         __u8    sdiag_family;
> 
> since netdevice.h has the definition also in user space
> 
> #define MAX_ADDR_LEN    32              /* Largest hardware address length */
> 
> I find using MAX_ADDR_LEN better than numeric 32, though I doubt this will
> change any time soon. Would you mind if I change packet_diag.h and
> if_link.h to use that instead and fix the userspace compilation
> problems by including netdevice.h?

The alternative fix, that is, to include <linux/netdevice.h>
which pulls in other headers and a lot of definitions with them,
has been mentioned in the discussion, too.
We decided that the fix that was applied would be the least of all evils.


-- 
ldv

Download attachment "signature.asc" of type "application/pgp-signature" (802 bytes)