From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 28301 invoked by alias); 21 Jan 2007 22:35:49 -0000 Received: (qmail 28281 invoked by uid 22791); 21 Jan 2007 22:35:44 -0000 X-Spam-Check-By: sourceware.org Received: from 195.22.55.53.adsl.nextra.cz (HELO host0.dyn.jankratochvil.net) (195.22.55.53) by sourceware.org (qpsmtpd/0.31) with ESMTP; Sun, 21 Jan 2007 22:35:39 +0000 Received: from host0.dyn.jankratochvil.net (localhost [127.0.0.1]) by host0.dyn.jankratochvil.net (8.13.8/8.13.8) with ESMTP id l0LMZXJx010875; Sun, 21 Jan 2007 23:35:33 +0100 Received: (from jkratoch@localhost) by host0.dyn.jankratochvil.net (8.13.8/8.13.8/Submit) id l0LMZXRr010874; Sun, 21 Jan 2007 23:35:33 +0100 Date: Sun, 21 Jan 2007 22:35:00 -0000 From: Jan Kratochvil To: gdb-patches@sourceware.org Cc: Daniel Jacobowitz Subject: Re: [PATCH] Follow specific symbol's DW_AT_decl_file Message-ID: <20070121223533.GA1890@host0.dyn.jankratochvil.net> References: <20070111223940.GA8958@host0.dyn.jankratochvil.net> <20070121165451.GA12463@nevyn.them.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20070121165451.GA12463@nevyn.them.org> User-Agent: Mutt/1.4.2.2i 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: 2007-01/txt/msg00448.txt.bz2 On Sun, 21 Jan 2007 17:54:51 +0100, Daniel Jacobowitz wrote: ... > /* This structure is space critical. See space comments at the top. */ > > So I wrote an implementation that used an on-the-side hash table. But > in the end, it was a bit slower, and most symbols needed one. Even > with -readnow the memory increase was only 3%, so it will be less than > that in practice. So I went back to doing it in the symbol after all > :-) Due to that comment I originally valgrind(1)ed it and on my testcase the memory requirement increase was visible but negligible. > The other changes I made were to move the index handling into > dwarf2read.c, since the concept of a file index is dwarf specific, > and to move the test into gdb.base, since in theory other debug > formats could get it right. Thanks, it unexpectedly cleaned up the patch a lot. Regards, Jan > I tested this on x86_64-pc-linux-gnu and committed it. (fine on local RH i686-pc-linux-gnu)