From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 108405 invoked by alias); 8 Jun 2015 11:27:59 -0000 Mailing-List: contact libc-alpha-help@sourceware.org; run by ezmlm Precedence: bulk List-Id: List-Subscribe: List-Archive: List-Post: List-Help: , Sender: libc-alpha-owner@sourceware.org Received: (qmail 108393 invoked by uid 89); 8 Jun 2015 11:27:58 -0000 Authentication-Results: sourceware.org; auth=none X-Virus-Found: No X-Spam-SWARE-Status: No, score=-1.4 required=5.0 tests=AWL,BAYES_00,KAM_LAZY_DOMAIN_SECURITY,SPF_HELO_PASS,T_RP_MATCHES_RCVD autolearn=no version=3.3.2 X-HELO: mx1.redhat.com Date: Mon, 08 Jun 2015 11:48:00 -0000 From: Jonathan Wakely To: Torvald Riegel Cc: Martin Sebor , Rich Felker , Szabolcs Nagy , GNU C Library Subject: Re: [PATCH] pthread_once hangs when init routine throws an exception [BZ #18435] Message-ID: <20150608112755.GV12728@redhat.com> References: <556B7F10.40209@redhat.com> <556C31DE.4020803@arm.com> <556CA772.2060207@redhat.com> <1433329427.21461.101.camel@triegel.csb> <20150603191444.GM17573@brightrain.aerifal.cx> <556F6012.6090502@redhat.com> <1433404311.21461.231.camel@triegel.csb> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii; format=flowed Content-Disposition: inline In-Reply-To: <1433404311.21461.231.camel@triegel.csb> User-Agent: Mutt/1.5.23 (2014-03-12) X-SW-Source: 2015-06/txt/msg00265.txt.bz2 On 04/06/15 09:51 +0200, Torvald Riegel wrote: >We wanted to change to void* with GCC-5, but missed that boat >unfortunately. We can document that we don't give ABI/API guarantees >for it, but this would be much clearer to users if this would return >void* I believe. I think it's still OK to change it to void* now, and document that it is currently the address of the POSIX primitive, but that the function could be modified or removed in future. A disadvantage of returning void* is that if we do change what it returns then code using it still compiles and links. If we change it from returning pthread_xxx_t* to something else, or just remove it entirely, then code using it fails earlier, at compile-time. So we need to document that there are no guarantees for it, whatever it returns. Simply changing to void* isn't enough.