From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 21119 invoked by alias); 11 Sep 2008 18:38:16 -0000 Received: (qmail 21107 invoked by uid 22791); 11 Sep 2008 18:38:15 -0000 X-Spam-Check-By: sourceware.org Received: from NaN.false.org (HELO nan.false.org) (208.75.86.248) by sourceware.org (qpsmtpd/0.31) with ESMTP; Thu, 11 Sep 2008 18:37:34 +0000 Received: from nan.false.org (localhost [127.0.0.1]) by nan.false.org (Postfix) with ESMTP id 7638E1019F; Thu, 11 Sep 2008 18:37:31 +0000 (GMT) Received: from caradoc.them.org (22.svnf5.xdsl.nauticom.net [209.195.183.55]) by nan.false.org (Postfix) with ESMTP id 573B81019E; Thu, 11 Sep 2008 18:37:31 +0000 (GMT) Received: from drow by caradoc.them.org with local (Exim 4.69) (envelope-from ) id 1Kdr2o-00028E-8C; Thu, 11 Sep 2008 14:37:30 -0400 Date: Thu, 11 Sep 2008 18:38:00 -0000 From: Daniel Jacobowitz To: Joel Brobecker Cc: gdb-patches@sourceware.org Subject: Re: [RFA/commit/dwarf] Create partial symbols for nested subprograms Message-ID: <20080911183730.GA17809@caradoc.them.org> Mail-Followup-To: Joel Brobecker , gdb-patches@sourceware.org References: <20080910201959.GC10133@adacore.com> <20080910203437.GA26162@caradoc.them.org> <20080911175422.GS12222@adacore.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20080911175422.GS12222@adacore.com> User-Agent: Mutt/1.5.17 (2008-05-11) X-IsSubscribed: yes Mailing-List: contact gdb-patches-help@sourceware.org; run by ezmlm Precedence: bulk List-Id: List-Subscribe: List-Archive: List-Post: List-Help: , Sender: gdb-patches-owner@sourceware.org X-SW-Source: 2008-09/txt/msg00239.txt.bz2 On Thu, Sep 11, 2008 at 10:54:22AM -0700, Joel Brobecker wrote: > > Do you have any performance data for GDB startup with this change? > > I am not sure I am doing the right kind of performance test, so if you > have some better suggestions, please let me know. For now, here is what > I did: > > . I used one of the large Ada applications we have: > It is 290MB in size - reduced to 160MB when I strip the debugging > info => the debugging info size is roughly 130MB. > The number of real lines of code is 700K spread in ~3500 files, > and it has a total of 39,000 subprograms declared in the debugging > info. > > . To measure the startup time, I used the --statistics command-line > switch. I usually use 'gdb -batch foo.exe' to time reading partial symbols, and 'gdb -readnow -batch foo.exe' to time full symbols (only partial symbols are at issue here). > It looks like roughly a 4% increase in startup time. Not sure whether > that's considered a large increase or not - I just think that it's not > noticeable. None of our users have reported issues with startup time. Your users must be more patient than mine or Tom's :-) I consider startup time to be pretty important, and I've been working on bringing it down... Tom's been working on an even more drastic version. > But, to make sure that only languages that have this feature get hit, > I adjusted the patch to only search nested procedure when the language > allows them. Does it work / help to do the check here too: > && (load_all > || last_die->tag == DW_TAG_namespace > || last_die->tag == DW_TAG_enumeration_type > + || last_die->tag == DW_TAG_subprogram > + || last_die->tag == DW_TAG_lexical_block > || (cu->language != language_c > && (last_die->tag == DW_TAG_class_type > || last_die->tag == DW_TAG_interface_type That'll bring the memory usage back down for non-Ada. -- Daniel Jacobowitz CodeSourcery