From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 9521 invoked by alias); 20 Jul 2010 21:59:16 -0000 Received: (qmail 9511 invoked by uid 22791); 20 Jul 2010 21:59:15 -0000 X-SWARE-Spam-Status: No, hits=-2.1 required=5.0 tests=AWL,BAYES_00,TW_TM X-Spam-Check-By: sourceware.org Received: from dair.pair.com (HELO dair.pair.com) (209.68.1.49) by sourceware.org (qpsmtpd/0.43rc1) with SMTP; Tue, 20 Jul 2010 21:59:08 +0000 Received: (qmail 61467 invoked by uid 20157); 20 Jul 2010 21:59:06 -0000 Received: from localhost (sendmail-bs@127.0.0.1) by localhost with SMTP; 20 Jul 2010 21:59:06 -0000 Date: Tue, 20 Jul 2010 21:59:00 -0000 From: Hans-Peter Nilsson To: Jakub Jelinek cc: gcc-patches@gcc.gnu.org Subject: cris-elf still broken with "[PATCH] Unbreak ia64 bootstrap (PR debug/45006)" In-Reply-To: <20100720163614.GF19172@tyan-ft48-01.lab.bos.redhat.com> Message-ID: References: <20100720163614.GF19172@tyan-ft48-01.lab.bos.redhat.com> User-Agent: Alpine 2.00 (BSF 1167 2008-08-23) MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII X-IsSubscribed: yes Mailing-List: contact gcc-patches-help@gcc.gnu.org; run by ezmlm Precedence: bulk List-Id: List-Archive: List-Post: List-Help: Sender: gcc-patches-owner@gcc.gnu.org X-SW-Source: 2010-07/txt/msg01608.txt.bz2 On Tue, 20 Jul 2010, Jakub Jelinek wrote: > Hi! > > My PR45003 fix apparently broke IA-64 bootstrap, because there > FUNCTION_DECLs have DECL_MODE DImode, while TYPE_MODE of FUNCTION_TYPE > is TImode. Guess that's related to fn descriptors. Probably not; I saw the same for my cris-elf autotester. No fn descriptors there. > The following patch is an attempt to fix that. But doesn't fix all introduced problems (read: r162348 doesn't work), see PR45009. > For FUNCTION_DECLs, > I don't see how SIGN_EXTEND/ZERO_EXTEND on the MEM would be ever useful, > FUNCTION_DECLs ought to appear in DEBUG stmts only inside of ADDR_EXPR > where the MEM is not used anyway and we just use the address. It might not matter in the context of a FUNCTION_DECL, but anyway FWIW, sign-extend and zero-extend on mem are valid insns (by themselves or as operands for add or sub) for cris/crisv32. > PR debug/45006 > * cfgexpand.c (expand_debug_expr): Don't go through adjust_mode > path for FUNCTION_DECLs. Only look at TYPE_UNSIGNED of > operand's type if exp is tcc_unary class tree. brgds, H-P