Date: Sun, 20 May 2007 14:24:28 +0930 From: Greg 'groggy' Lehey <grog@FreeBSD.org> To: Doug Barton <dougb@FreeBSD.org> Cc: cvs-src@FreeBSD.org, src-committers@FreeBSD.org, cvs-all@FreeBSD.org, edwin@freebsd.org Subject: Re: cvs commit: src/etc services Message-ID: <20070520045428.GG2288@wantadilla.lemis.com> In-Reply-To: <464FD1F7.2010907@FreeBSD.org> References: <200705200331.l4K3Vq3V079479@repoman.freebsd.org> <464FD1F7.2010907@FreeBSD.org>
next in thread | previous in thread | raw e-mail | index | archive | help
--KIzF6Cje4W/osXrF
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
On Saturday, 19 May 2007 at 21:43:35 -0700, Doug Barton wrote:
> Greg Lehey wrote:
>> grog 2007-05-20 03:31:52 UTC
>>
>> FreeBSD src repository
>>
>> Modified files:
>> etc services
>> Log:
>> Bring the well known ports of /etc/services into sync with the IANA
>> list.
>>
>> This is only for the well known known ports (port 1-1023) for tcp and
>> udp only.
>>
>> Changes:
>> - Removed "problems" comments around port 57, 77 and 87
>> - Removed audionews (port 114)
>> - Added imap3 (port 220)
>> - Removed yak-chat (port 258)
>> - Removed concert (port 786)
>> - Added a lot of new allocations
>
> In the past we've avoided adding entries to /etc/services that were
> not actually being used by something we care about since every extra
> entry increases response time for getservby*() calls.
Understood. It's a factor I thought about.
> Do we feel this is no longer a factor worth worrying about?
It's certainly worth worrying about, and I thought about it before
committing. My thoughts:
- Clearly a lot of this stuff was just corrections.
- We don't know in advance which IP ports are used by ports from the
Ports Collection.
- Maybe the performance issue isn't that serious after all.
To quote from services(5):
BUGS
A name server should be used instead of a static file.
I was just thinking that it might be relatively simple to build a
"compiled" file in tabular format and then mmap it. That should be
relatively simple, and should greatly improve performance modulo mmap
overhead. How does the mmap overhead compare with querying a name
server?
Greg
--
See complete headers for address and phone numbers.
--KIzF6Cje4W/osXrF
Content-Type: application/pgp-signature
Content-Disposition: inline
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.3 (FreeBSD)
iD8DBQFGT9SEIubykFB6QiMRAhoNAKCLzhGeVT2ohrjhwGH6AeEBeIQ3jwCgiPEv
j5M33ghp1UK86ubWmiuXSi8=
=nkad
-----END PGP SIGNATURE-----
--KIzF6Cje4W/osXrF--
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20070520045428.GG2288>