Mailing-List: contact cygwin-developers-help AT sourceware DOT cygnus DOT com; run by ezmlm
List-Subscribe: <mailto:cygwin-developers-subscribe AT sources DOT redhat DOT com>
List-Archive: <http://sources.redhat.com/ml/cygwin-developers/>
List-Post: <mailto:cygwin-developers AT sources DOT redhat DOT com>
List-Help: <mailto:cygwin-developers-help AT sources DOT redhat DOT com>, <http://sources.redhat.com/ml/#faqs>
Sender: cygwin-developers-owner AT sources DOT redhat DOT com
Delivered-To: mailing list cygwin-developers AT sources DOT redhat DOT com
To: cygdev <cygwin-developers AT cygwin DOT com>
Subject: Re: Proposal: check and read /etc/{passwd,group} less frequently.
References: <EA18B9FA0FE4194AA2B4CDB91F73C0EF08F177 AT itdomain002 DOT itdomain DOT net DOT au>
	<s1su1y6qlam DOT fsf AT jaist DOT ac DOT jp> <s1sd74uufha DOT fsf AT jaist DOT ac DOT jp>
	<20010914150107 DOT E25339 AT cygbert DOT vinschen DOT de>
Mime-Version: 1.0 (generated by tm-edit 7.106)
Content-Type: text/plain; charset=US-ASCII
From: Kazuhiro Fujieda <fujieda AT jaist DOT ac DOT jp>
Date: 15 Sep 2001 06:11:27 +0900
In-Reply-To: Corinna Vinschen's message of Fri, 14 Sep 2001 15:01:07 +0200
Message-ID: <s1sbskdv5og.fsf@jaist.ac.jp>
Lines: 29
X-Mailer: Gnus v5.3/Emacs 19.34

>>> On Fri, 14 Sep 2001 15:01:07 +0200
>>> Corinna Vinschen <vinschen AT redhat DOT com> said:

> What about a long running server process e.g. started from cygrunsrv?
> It runs uinfo_init() just once at the beginning.  It would never get
> the changes to /etc/passwd and /etc/group since it needs that info
> _before_ changing the user context (which would call internal_getlogin()
> directly from seteuid(), btw. -- no uinfo_init() at all).

There is no problem in regard to inetd because it spawns real
servers which refer /etc/{passwd,group}.

Sshd can't have the chance to get the changes.  But it less help
sshd to check and read /etc/{passwd,group} in internal_getlogin(),
because it calls getpwnam() before seteuid().

I don't hate to restart sshd so much. I feel it much more
comfortable and reasonable than the previous way, that is,
I must terminate all Cygwin processes to get changes to
/etc/{passwd,group}.

Anyway, I implemented my idea and got benchmarks on Win98 SE and
WinNT 4.0 SP6a via compiling bash. The code became much simpler,
but I couldn't observe any performance improvement on both
platforms. So I withdraw my proposal.
____
  | AIST      Kazuhiro Fujieda <fujieda AT jaist DOT ac DOT jp>
  | HOKURIKU  Center for Information Science
o_/ 1990      Japan Advanced Institute of Science and Technology