From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 27230 invoked by alias); 3 Sep 2009 18:40:51 -0000 Received: (qmail 27221 invoked by uid 22791); 3 Sep 2009 18:40:51 -0000 X-SWARE-Spam-Status: No, hits=-2.1 required=5.0 tests=AWL,BAYES_00,J_CHICKENPOX_32 X-Spam-Check-By: sourceware.org Received: from rock.gnat.com (HELO rock.gnat.com) (205.232.38.15) by sourceware.org (qpsmtpd/0.43rc1) with ESMTP; Thu, 03 Sep 2009 18:40:47 +0000 Received: from localhost (localhost.localdomain [127.0.0.1]) by filtered-rock.gnat.com (Postfix) with ESMTP id 553AE2BAB50 for ; Thu, 3 Sep 2009 14:40:45 -0400 (EDT) Received: from rock.gnat.com ([127.0.0.1]) by localhost (rock.gnat.com [127.0.0.1]) (amavisd-new, port 10024) with LMTP id w1SPYEldh2Ga for ; Thu, 3 Sep 2009 14:40:45 -0400 (EDT) Received: from joel.gnat.com (localhost.localdomain [127.0.0.1]) by rock.gnat.com (Postfix) with ESMTP id 1A5332BAB41 for ; Thu, 3 Sep 2009 14:40:45 -0400 (EDT) Received: by joel.gnat.com (Postfix, from userid 1000) id 6D82EF589A; Thu, 3 Sep 2009 11:40:35 -0700 (PDT) Date: Thu, 03 Sep 2009 18:40:00 -0000 From: Joel Brobecker To: gdb@sourceware.org Subject: Re: More info on PR/9711 (quadratic slowdown for deep stack traces) Message-ID: <20090903184035.GB4379@adacore.com> References: <20090901204815.GK4379@adacore.com> <20090902114303.GA30223@caradoc.them.org> <20090902162402.GQ4379@adacore.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20090902162402.GQ4379@adacore.com> User-Agent: Mutt/1.5.18 (2008-05-17) Mailing-List: contact gdb-help@sourceware.org; run by ezmlm Precedence: bulk List-Id: List-Subscribe: List-Archive: List-Post: List-Help: , Sender: gdb-owner@sourceware.org X-SW-Source: 2009-09/txt/msg00047.txt.bz2 > I suppose we could immediately flush the extra frames when the setting > changes to "off", but on the other hand, I think that the right thing > to do in frame_find_by_id is to call get_prev_frame_1 anyway. I will > test that next. Just for the record, this turned out to be a bad idea. In the case when the previous frames had not been computed, we might end up trying to unwind past main. This causes a few regressions in one of gdb.mi testcases. I didn't investigate precisely what was happening since the suggestion became pretty much irrelevant when I implemented the cache. See: http://www.sourceware.org/ml/gdb-patches/2009-09/msg00085.html -- Joel