conf/104884: Add support EtherChannel configuration to rc.conf

[Florent Thoumie]

Norikatsu Shigemura wrote:
> The following reply was made to PR conf/104884; it has been noted by GNATS.
> 
> From: Norikatsu Shigemura <nork at FreeBSD.org>
> To: Florent Thoumie <flz at FreeBSD.org>
> Cc: Norikatsu Shigemura <nork at FreeBSD.org>, freebsd-bugs at FreeBSD.org,
>         FreeBSD-gnats-submit at FreeBSD.org, freebsd-rc at FreeBSD.org
> Subject: Re: conf/104884: Add support EtherChannel configuration to rc.conf
> Date: Tue, 30 Jan 2007 01:53:17 +0900
> 
>  On Sun, 28 Jan 2007 19:37:03 +0000
>  Florent Thoumie <flz at freebsd.org> wrote:
>  > >> http://www.freebsd.org/cgi/query-pr.cgi?pr=104884
>  > >>> Category:       conf
>  > >>> Responsible:    freebsd-bugs
>  > >>> Synopsis:       Add support EtherChannel configuration to rc.conf
>  > >>> Arrival-Date:   Sat Oct 28 16:10:18 GMT 2006
>  > > 	I chased HEAD.  Please see following patch.
>  > > 	Anyone, please handle this PR?
>  > > 	And I'll make a patch for 6-stable.
>  > I'm sorry, I meant to answer but forgot about it. I don't know much
>  
>  	Thanks for your handling.
>  
>  > about technical details, so I'll only focus on style. Here are my
>  > comments on the patch:
>  
>  	I think that the answer of all questions is 'BECAUSE netgraph's
>  	specification.'. :-)

... which I don't really know, so better not take everything I write as
face value :-)

>  > > Index: network.subr
>  > > ===================================================================
>  > > RCS file: /home/ncvs/src/etc/network.subr,v
>  > > retrieving revision 1.176
>  > > diff -u -r1.176 network.subr
>  > > --- network.subr	29 Oct 2006 13:29:49 -0000	1.176
>  > > +++ network.subr	28 Jan 2007 14:52:36 -0000
>  > > @@ -907,3 +907,78 @@
>  > >  		esac
>  > >  	done
>  > >  }
>  > > +
>  > > +ng_mkpeer() {
>  > > +	ngctl -f - 2> /dev/null <<EOF
>  > > +mkpeer $*
>  > > +msg dummy nodeinfo
>  > > +EOF
>  > > +}
>  > > +
>  > > +ng_create_one() {
>  > > +	ng_mkpeer $* | while read line; do
>  > > +		t=`expr "${line}" : '.* name="\([a-z]*[0-9]*\)" .*'`
>  > > +		if [ -n "${t}" ]; then
>  > > +			echo ${t}
>  > > +			return
>  > > +		fi
>  > > +	done
>  > > +}
>  
>  	I implemented ng_mkpeer and ng_create_one as generic functions.
>  	If anyone want to add other netgraph function, they can use
>  	these functions.

I wasn't commenting this one, seemed alright to me.


>  > > +ng_fec_create() {
>  > > +	local req_iface iface bogus
>  > > +	req_iface="$1"
>  > > +	if [ -z "${req_iface}" ]; then
>  > Why are you testing this? It's only called in fec_up() and can't be
>  > called with a empty argument. Or do you want to "export" the function to
>  > other scripts?
>  
>  	Ah! This code's meaning was changed from original code.  Yes,
>  	I think that this code should be removed.
>  
>  > > +	ngctl shutdown ${req_iface}: > /dev/null 2>&1
>  > > +	bogus=""
>  > > +	while true; do
>  > > +		iface=`ng_create_one fec dummy fec`
>  > > +		if [ -z "${iface}" ]; then
>  > > +			exit 2
>  > > +		fi
>  > > +		if [ "${iface}" = "${req_iface}" ]; then
>  > > +			echo ${iface}
>  > > +			break
>  > > +		fi
>  > > +		bogus="${bogus} ${iface}"
>  > > +	done
>  > > +	for iface in ${bogus}; do
>  > > +		ngctl shutdown ${iface}:
>  > > +	done
>  > 
>  > These loops are a bit confusing. If I understand correctly, you're
>  > creating interfaces until they reach the right number and then you
>  > delete all the ones which have been created unnecessarily? Could it be
>  
>  	Your understanding is right:-).  But we cannot control unit number
>  	in 'ngctl mkpeer', because 'Find the first free unit number for a
>  	new interface' strategy in sys/netgraph/ng_fec.c#ng_fec_get_unit.
>  
>  > that iface is higher than req_iface (which would loop undefinitely)?
>  
>  	Previously, I removed req_iface by ngctl shutdown.  So not reache
>  	infinity:-).

Fair enough.

>  > > +# fec_up ifn
>  > > +# Configure Fast EtherChannel for interface $ifn. Returns 0 if FEC
>  > > +# arguments were found and configured; returns 1 otherwise.
>  > > +fec_up() {
>  > > +	case ${fec_interfaces} in
>  > > +	[Nn][Oo] | '')
>  > > +		;;
>  > What's the point of this? The 'case' seems useless to me. Just got with
>  > the 'for' loop. If it's an empty list, then it just won't do anything.
>  > Default has to be '' and not 'NO' (but it seems more sensible anyway).
>  
>  	I obtained gif_up code.  I don't know why/where are problem in it.

No problem, just seemed bad style to me :-)

-- 
Florent Thoumie
flz at FreeBSD.org
FreeBSD Committer

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 250 bytes
Desc: OpenPGP digital signature
Url : http://lists.freebsd.org/pipermail/freebsd-rc/attachments/20070129/b230cf71/signature.pgp