X-Recipient: archive-cygwin AT delorie DOT com
X-Spam-Check-By: sourceware.org
From: Tatsuro MATSUOKA <matsuoka AT mol DOT nagoya-u DOT ac DOT jp>
To: Eric Blake <ebb9 AT byu DOT net>
Cc: cygwin AT cygwin DOT com, matsuoka AT nuce DOT nagoya-u DOT ac DOT jp
Date: Wed, 12 Sep 2007 11:29:57 +0900
Subject: Re: gcc-dw2? or fast sjlj-exceptions EH
Reply-To: matsuoka AT nuce DOT nagoya-u DOT ac DOT jp
In-Reply-To: <46E732EC.2080400@byu.net>
References: <20070912054905 DOT 177F7D98 DOT matsuoka AT mol DOT nagoya-u DOT ac DOT jp> <46E732EC DOT 2080400 AT byu DOT net>
Message-Id: <20070912112957.BB0EAB88.matsuoka@mol.nagoya-u.ac.jp>
MIME-Version: 1.0
Content-Type: text/plain; charset="US-ASCII"
Content-Transfer-Encoding: 7bit
X-Mailer: nPOP Ver 1.0.9
X-IsSubscribed: yes
Mailing-List: contact cygwin-help AT cygwin DOT com; run by ezmlm
Precedence: bulk
List-Id: <cygwin.cygwin.com>
List-Unsubscribe: <mailto:cygwin-unsubscribe-archive-cygwin=delorie DOT com AT cygwin DOT com>
List-Subscribe: <mailto:cygwin-subscribe AT cygwin DOT com>
List-Archive: <http://sourceware.org/ml/cygwin/>
List-Post: <mailto:cygwin AT cygwin DOT com>
List-Help: <mailto:cygwin-help AT cygwin DOT com>, <http://sourceware.org/ml/#faqs>
Sender: cygwin-owner AT cygwin DOT com
Mail-Followup-To: cygwin AT cygwin DOT com
Delivered-To: mailing list cygwin AT cygwin DOT com

Eric Blake <ebb9 AT byu DOT net> wrote:
(2007/09/12 09:29)

>I think you are misunderstanding a fundamental concept.  sjlj exception
>handling is inherently slower than stack-tracing exception handling,
>because it must assume the worst and store the entire stack state (the
>'sj' or setjump) prior to any call site where an exception might occur,
>whether or not an exception actually happens (the 'lj' or longjump part).
> Dwarf error handling, on the other hand, is a form of stack-tracing,
>where there is NO penalty UNLESS there was an exception.  Exception
>handling is slower, because it must crawl backwards through the stack to
>find all catch points with handlers to run, but more efficient when there
>is no exception because you don't waste time saving state that never gets
>jumped to.
>
>In other words, the ONLY way to speed up octave is to avoid sjlj
>exceptions.  And there are two ways to do that - either switch to
>stack-based exceptions (such as dwarf), or avoid exceptions to begin with.
> The octave maintainer has already expressed a reluctance to disable
>exceptions, and the current gcc for cygwin, 3.4.4, doesn't have a way to
>use anything other than sjlj, so you will just have to wait.

Dear Eric Blake 
I really appreciate for your comprehensible reply.
I'll wait for a while.

Sincerely yours

Tasturo MATSUOKA



--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Problem reports:       http://cygwin.com/problems.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/