lkml.org 
[lkml]   [2012]   [Jan]   [30]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
SubjectRE: [07/22] Cyclades PC300 driver: ioctl() fixes for mixed 64/32-bit systems
From
Date
On Mon, 2012-01-30 at 11:44 +0000, David Laight wrote:
> > +struct pc300_net_stats {
> > + u64 rx_packets;
> ...
> > + u64 rx_compressed;
> > + u64 tx_compressed;
> > +};
> > +
> > typedef struct pc300stats {
> > int hw_type;
> > u32 line_on;
> > u32 line_off;
> > - struct net_device_stats gen_stats;
> > + /* Use this instead of net_device_stats, since passing
> > + * net_device_stats breaks 32-bit user processes on
> > 64-bit kernels,
> > + * and rtnetlink is unreasonably complicated just to get
> > + * some statistics.
> > + */
> > + struct pc300_net_stats net_stats;
> > falc_t te_stats;
> > } pc300stats_t;
>
> Since 'struct net_device_stats' looks like a common structure,
> maybe the 64bit version should be added in the same place??
[...]

It's called struct rtnl_link_stats64 and is already defined in
<linux/rtnetlink.h> (where it must remain).

This change doesn't seem to be a proper fix. If the ioctl interface to
this driver uses struct net_device_stats (which I didn't realise,
otherwise I would have left that structure exported from
<linux/netdevice.h>) then it should continue doing so to maintain
compatibility. The driver will have to fix up the outgoing structure
with the aid of is_compat_task().

If backward-compatibility is really not a concern for the driver then
this ioctl should be completely removed in favour of ethtool extended
stats.

Ben.

--
Ben Hutchings, Staff Engineer, Solarflare
Not speaking for my employer; that's the marketing department's job.
They asked us to note that Solarflare product names are trademarked.



\
 
 \ /
  Last update: 2012-01-30 15:41    [W:0.035 / U:11.712 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site