Re: CVS commit: src/sys/net

[yamt%mwd.biglobe.ne.jp@localhost (YAMAMOTO Takashi)]

> On Sun, Aug 16, 2009 at 05:01:12AM +0000, YAMAMOTO Takashi wrote:
>> hi,
>> 
>> > On Fri, Aug 14, 2009 at 06:48:10AM +0000, YAMAMOTO Takashi wrote:
>> >> hi,
>> >> 
>> >> > Module Name:    src
>> >> > Committed By:   dyoung
>> >> > Date:           Thu Aug 13 00:23:32 UTC 2009
>> >> > 
>> >> > Modified Files:
>> >> >         src/sys/net: if.c if.h
>> >> > 
>> >> > Log Message:
>> >> > Use sysctl(9) to expose to userland each interface transmission
>> >> > queue's maximum length, current length, and number of drops.  E.g.,
>> >> > 
>> >> > % sysctl net.interfaces.bnx0
>> >> > net.interfaces.bnx0.sndq.len = 0
>> >> > net.interfaces.bnx0.sndq.maxlen = 509
>> >> > net.interfaces.bnx0.sndq.drops = 0
>> >> 
>> >> does it work for xvif interfaces?  cf. PR/35074
>> > 
>> > Probably not.  Thanks for bringing this case to my attention.
>> > 
>> > I could change illegal characters in if_xname to dashes or to
>> > underscores in the sysctl node name.  Or I could name the node after the
>> > if_index instead of after the if_xname, if the if_xname contains illegal
>> > characters.  I guess that I like the latter idea better.  What do you
>> > think?
>> 
>> why do you want to use sysctl rather than eg. ioctl?
> 
> I like sysctl interfaces because they are self-describing, granular,
> hierarchical, and direct.

sorry, honestly speaking, none of them sounds like a good excuse to
start using sysctl for purposes which already have an existing api.
i guess that it's too late to throw ioctl away.

> It is easy for userland to name sysctl
> interfaces to permit/restrict their visibility and use by a user or a
> process, and that is something I want to do in the future.

in that case, what to do for existing ioctl-based operations?

YAMAMOTO Takashi

> 
> Dave
> 
> -- 
> David Young             OJC Technologies
> dyoung%ojctech.com@localhost      Urbana, IL * (217) 278-3933