cvs commit: src/sys/net rtsock.c

[Max Laier]

On Saturday 14 August 2004 00:37, Robert Watson wrote:
> On Sat, 14 Aug 2004, Max Laier wrote:
> > On Saturday 14 August 2004 00:23, Robert Watson wrote:
> > > rwatson     2004-08-13 22:23:21 UTC
> > >
> > >   FreeBSD src repository
> > >
> > >   Modified files:
> > >     sys/net              rtsock.c
> > >   Log:
> > >   Use IFQ_SET_MAXLEN() to set the maximum queue depth of the routing
> > >   socket netisr queue.
> >
> > Well, I suggest to use IFQ_* macros only on queues that are
> > IFQ_SET_READY'ed as well. i.e. as long as you don't use IFQ_{EN,DE}QUEUE
> > with the queue there is not much point in using IFQ_SET_MAXLEN either.
> > It'd just add noise to searches for IFQ_*
> >
> > I don't think that there are plans to use ALTQ disciplines on the netisr
> > queues, are there?
>
> No, that wouldn't really make sense, as we want a reliable record stream,
> really, and the protocol in question wouldn't benefit from RED, etc.  Do
> you suggest, then, just backing this out?  Some of the other ifq
> abstractions are used, such as IFF_HANDOFF()...

It does not hurt (much) to use IFQ_* macros on normal FIFO queues as long as 
IFQ_SET_READY is not used to activate ALTQ processing. Nevertheless, there is 
some overhead and I don't see us removing the IF_* macros.

So long story short: For normal FIFO queues use IF_* macros, for ALTQ-enabled 
queues use IFQ_* or IFQ_DRV_* (see altq(9) for details on the difference and 
conversion).

-- 
/"\  Best regards,			| mlaier at freebsd.org
\ /  Max Laier				| ICQ #67774661
 X   http://pf4freebsd.love2party.net/	| mlaier at EFnet
/ \  ASCII Ribbon Campaign		| Against HTML Mail and News
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 187 bytes
Desc: signature
Url : http://lists.freebsd.org/pipermail/cvs-src/attachments/20040814/7781b279/attachment.bin