From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 22873 invoked by alias); 1 May 2013 12:53:07 -0000 Mailing-List: contact libc-ports-help@sourceware.org; run by ezmlm Precedence: bulk List-Id: List-Subscribe: List-Post: List-Help: , Sender: libc-ports-owner@sourceware.org Received: (qmail 22861 invoked by uid 89); 1 May 2013 12:53:06 -0000 X-Spam-SWARE-Status: No, score=-3.8 required=5.0 tests=BAYES_00,FREEMAIL_ENVFROM_END_DIGIT,FREEMAIL_FROM,KHOP_THREADED,RCVD_IN_DNSWL_LOW,RCVD_IN_HOSTKARMA_YE,SPF_PASS,TW_CP autolearn=ham version=3.3.1 Received: from mail-bk0-f48.google.com (HELO mail-bk0-f48.google.com) (209.85.214.48) by sourceware.org (qpsmtpd/0.84/v0.84-167-ge50287c) with ESMTP; Wed, 01 May 2013 12:53:05 +0000 Received: by mail-bk0-f48.google.com with SMTP id it19so628946bkc.35 for ; Wed, 01 May 2013 05:53:03 -0700 (PDT) X-Received: by 10.204.231.198 with SMTP id jr6mr817106bkb.104.1367412783047; Wed, 01 May 2013 05:53:03 -0700 (PDT) Received: from pebble.twiddle.net (host217-43-57-145.range217-43.btcentralplus.com. [217.43.57.145]) by mx.google.com with ESMTPSA id rm3sm435093bkb.20.2013.05.01.05.53.01 for (version=TLSv1 cipher=RC4-SHA bits=128/128); Wed, 01 May 2013 05:53:02 -0700 (PDT) Message-ID: <5181102B.8060006@twiddle.net> Date: Wed, 01 May 2013 12:53:00 -0000 From: Richard Henderson User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:17.0) Gecko/20130402 Thunderbird/17.0.5 MIME-Version: 1.0 To: Will Newton CC: libc-ports@sourceware.org, patches@linaro.org, Roland McGrath Subject: Re: [PATCH, v5] ARM: Add Cortex-A15 optimized NEON and VFP memcpy routines, with IFUNC. References: <517FF73E.5020509@linaro.org> In-Reply-To: <517FF73E.5020509@linaro.org> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-SW-Source: 2013-05/txt/msg00010.txt.bz2 On 2013-04-30 17:54, Will Newton wrote: > +. */ > + > +#define NO_THUMB Please add a comment as to why NO_THUMB is being set. C.f. the other uses of that macro within sysdeps/arm/. (It's the negative offsets for memories as I recall, but that won't be obvious from this file.) > + /* Jump directly into the sequence below at the correct offset. */ > + add pc, pc, tmp1, lsl #1 This is going to fail for NaCl, defining ARM_ALWAYS_BX. I'm not quite sure all of what needs to be done to fix that. I'm sure Roland will have some advice... Perhaps the answer is "nothing" and NaCl should be vectored back to the non-ifunc version of the code. r~