netdev.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* Re: SIOCETHTOOL history ?
  2003-05-06 12:37 arun
@ 2003-05-06 11:49 ` David S. Miller
  0 siblings, 0 replies; 3+ messages in thread
From: David S. Miller @ 2003-05-06 11:49 UTC (permalink / raw)
  To: akumar; +Cc: netdev

   From: "arun" <akumar@omnesysindia.com>
   Date: Tue, 6 May 2003 18:07:48 +0530
   
   I have used SIOCGMIIPHY before and it seemed not to require 
   any su rights or capabilities.

Yes it does:

                /*
                 *      These ioctl calls:
                 *      - require superuser power.
                 *      - require strict serialization.
                 *      - return a value
                 */
                case SIOCETHTOOL:
                case SIOCGMIIPHY:
                case SIOCGMIIREG:
                        if (!capable(CAP_NET_ADMIN))
                                return -EPERM;

^ permalink raw reply	[flat|nested] 3+ messages in thread

* SIOCETHTOOL history ?
@ 2003-05-06 12:37 arun
  2003-05-06 11:49 ` David S. Miller
  0 siblings, 1 reply; 3+ messages in thread
From: arun @ 2003-05-06 12:37 UTC (permalink / raw)
  To: netdev

 Hi,
 
 Can anyone of the veterans please help me out with a 
small code curiosity

Why does the SIOCETHTOOL ioctl need 
CAP_NET_ADMIN even for harmless commands  like 
GSET ?

Some sample drivers that I managed to download over the net
that supports SIOCETHTOOL  has comments like 

"no need to check for CAP_NET_ADMIN since that is already
present in net/core/dev.c".  (And sure enough it was there)

Why would such a blanket checking be present when this ioctl
is only an entry point for many sub commands ? 

I have used SIOCGMIIPHY before and it seemed not to require 
any su rights or capabilities. What's then the difference between these
2 ioctls that require more capabilities for SIOCETHTOOL ?

arun

^ permalink raw reply	[flat|nested] 3+ messages in thread

* Re: SIOCETHTOOL history ?
@ 2003-05-06 13:24 arun
  0 siblings, 0 replies; 3+ messages in thread
From: arun @ 2003-05-06 13:24 UTC (permalink / raw)
  To: netdev


----- Original Message ----- 
From: "arun" <akumar@omnesysindia.com>
To: "David S. Miller" <davem@redhat.com>
Sent: Tuesday, May 06, 2003 6:53 PM
Subject: Re: SIOCETHTOOL history ?


> ----- Original Message ----- 
> From: "David S. Miller" <davem@redhat.com>
> To: <akumar@omnesysindia.com>
> Cc: <netdev@oss.sgi.com>
> Sent: Tuesday, May 06, 2003 5:19 PM
> Subject: Re: SIOCETHTOOL history ?
>  
 > Yes it does:
 
 Iam really stumped in that case since, mii-diag tool based
 on Donald Beckers code, ftp://ftp.scyld.com/pub/diag/mii-diag.c
 seems to work ok without CAP_NET_ADMIN on RH 7.1 
 and RH8.0 distros. 
 
 (I dont set any capabilities just tried to read speed / card id
 and it never returned EPERM for me)
 
 I chanced on the ethtool since mii failed to report anything on
 newer Gigabit cards (specifically a Broadcom Corporation 
 NetXtreme BCM5701 card)
 
 However ethtool required running with su root even for reading
 the capabilities (I recompiled ethtool code with only a GSET 
 command to confirm this) and that is why i have this doubt on the
 reason of the behavioural differnces between the 2 ioctl's
 
 arun
 
 

^ permalink raw reply	[flat|nested] 3+ messages in thread

end of thread, other threads:[~2003-05-06 13:24 UTC | newest]

Thread overview: 3+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2003-05-06 13:24 SIOCETHTOOL history ? arun
  -- strict thread matches above, loose matches on Subject: below --
2003-05-06 12:37 arun
2003-05-06 11:49 ` David S. Miller

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).