From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 5216 invoked by alias); 11 Jul 2002 01:55:59 -0000 Mailing-List: contact gdb-patches-help@sources.redhat.com; run by ezmlm Precedence: bulk List-Subscribe: List-Archive: List-Post: List-Help: , Sender: gdb-patches-owner@sources.redhat.com Received: (qmail 5209 invoked from network); 11 Jul 2002 01:55:59 -0000 Received: from unknown (HELO potter.sfbay.redhat.com) (205.180.83.107) by sources.redhat.com with SMTP; 11 Jul 2002 01:55:59 -0000 Received: from localhost.redhat.com (IDENT:MrLHYUA0J0fe6SVPMmGy9gOaIdZYMniT@romulus.sfbay.redhat.com [172.16.27.251]) by potter.sfbay.redhat.com (8.11.6/8.11.6) with ESMTP id g6B1umQ25366 for ; Wed, 10 Jul 2002 18:56:49 -0700 Received: by localhost.redhat.com (Postfix, from userid 469) id 713AF107D5; Wed, 10 Jul 2002 21:55:05 -0400 (EDT) From: Elena Zannoni MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Message-ID: <15660.58745.329143.773104@localhost.redhat.com> Date: Wed, 10 Jul 2002 19:08:00 -0000 To: Daniel Jacobowitz Cc: gdb-patches@sources.redhat.com Subject: Re: [rfa] Symbol hashing (for the last time?) In-Reply-To: <20020711005652.GA17895@nevyn.them.org> References: <20011030112756.A1546@nevyn.them.org> <20020711005652.GA17895@nevyn.them.org> X-SW-Source: 2002-07/txt/msg00220.txt.bz2 Daniel Jacobowitz writes: > Here's a patch from last October, dusted off and merged to the current > sources. The only substantial changes were some fixes for ada-lang.c, > merged after I wrote the original patch. I've verified no regressions > on i386-linux for GCC (2.95,3.0.4,3.1)/(stabs,dwarf2). > On my plate as well. I think I reviewed a few precursor patches to this. I have to reread the old threads. Of course if anybody else has comments, please feel free. Elena > This converts the normal symbol table lookups into hash tables. A few > sorts of symbol tables aren't hashed: those produced by mdebugread.c > and dstread.c, because they build symbol tables in lots of ad-hoc code, > and symbol tables which are actually the arguments to a function > (because order matters, or at least comments suggest so). A next step > will be to convert mdebugread.c, delete dstread.c (it's marked for an > upcoming obsoletion, isn't it?), and then delete all the complicated > binary search code since the only remaining unhashed symtabs will be > argument lists, which are small. > > This should help performance a bit on large programs. Ok to commit? > Anyone see any problems with it? >