From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 13779 invoked by alias); 8 Apr 2011 19:22:32 -0000 Received: (qmail 13765 invoked by uid 22791); 8 Apr 2011 19:22:31 -0000 X-SWARE-Spam-Status: No, hits=-1.8 required=5.0 tests=AWL,BAYES_00,T_RP_MATCHES_RCVD X-Spam-Check-By: sourceware.org Received: from mail.codesourcery.com (HELO mail.codesourcery.com) (38.113.113.100) by sourceware.org (qpsmtpd/0.43rc1) with ESMTP; Fri, 08 Apr 2011 19:22:27 +0000 Received: (qmail 23205 invoked from network); 8 Apr 2011 19:22:26 -0000 Received: from unknown (HELO digraph.polyomino.org.uk) (joseph@127.0.0.2) by mail.codesourcery.com with ESMTPA; 8 Apr 2011 19:22:26 -0000 Received: from jsm28 (helo=localhost) by digraph.polyomino.org.uk with local-esmtp (Exim 4.72) (envelope-from ) id 1Q8HGC-0001AL-Or; Fri, 08 Apr 2011 19:22:24 +0000 Date: Fri, 08 Apr 2011 19:22:00 -0000 From: "Joseph S. Myers" To: Paul Brook cc: binutils@sourceware.org Subject: Re: [PATCH] C6X unwinding table generation In-Reply-To: <201104071355.13054.paul@codesourcery.com> Message-ID: References: <201104071355.13054.paul@codesourcery.com> MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Mailing-List: contact binutils-help@sourceware.org; run by ezmlm Precedence: bulk List-Id: List-Subscribe: List-Archive: List-Post: List-Help: , Sender: binutils-owner@sourceware.org X-SW-Source: 2011-04/txt/msg00129.txt.bz2 Regarding the whole patch series, there still seem to be some bits missing - are they going to come in subsequent patches? * Testcase for linker handling of R_C6000_PREL31. * Will the assembler properly generate CANTUNWIND markers for functions without unwind information, if a function with unwind information is followed by one without in the .s file? Will the linker generate such markers for input objects without unwind information, and when doing a final link for the address after the end of the input objects? This (similar to what you implemented for ARM a while back) is expressly required by the ABI (section 10.8.1 in version 0.9). General comments on this individual patch (more later once I've gone through the details of it): * Please name functions and variables in tc-tic6x.c to contain "tic6x" where possible. * Assembler diagnostics should start with a lowercase letter. * Each diagnostic in tc-tic6x.c should be tested by a testcase (if you think it's unreachable, abort instead). * It looks to me like the state you put in the "unwind" variable should actually go in the tic6x_segment_info_type structure (maybe the c6x_unwind structure as an element of the tic6x_segment_info_type structure); the code generally tries to ensure that switching between sections works properly by keeping state there rather than in global variables. -- Joseph S. Myers joseph@codesourcery.com