From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mail-pa0-f48.google.com (mail-pa0-f48.google.com [209.85.220.48]) (using TLSv1 with cipher ECDHE-RSA-RC4-SHA (128/128 bits)) (Client CN "smtp.gmail.com", Issuer "Google Internet Authority" (not verified)) by ozlabs.org (Postfix) with ESMTPS id E91112C007A for ; Tue, 19 Feb 2013 04:06:14 +1100 (EST) Received: by mail-pa0-f48.google.com with SMTP id hz10so2928968pad.7 for ; Mon, 18 Feb 2013 09:06:12 -0800 (PST) Sender: Richard Henderson Message-ID: <51225F80.3030706@twiddle.net> Date: Mon, 18 Feb 2013 09:06:08 -0800 From: Richard Henderson MIME-Version: 1.0 To: linuxppc-dev@lists.ozlabs.org Subject: Accumulating PPC_FEATURE_ARCH bits Content-Type: text/plain; charset=ISO-8859-1; format=flowed List-Id: Linux on PowerPC Developers Mail List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , If one wants to test the PPC_FEATURE_ARCH_ISA_2_0[56] bits, one has to be careful that if 2_06 is set, 2_05 won't be. This seems illogical to me: given that 2.06 includes all of the 2.05 instructions, it would make most sense if both bits were set in the hwcap. Now, obviously I have to code around existing practice, but going forward this isn't tennable. When a Power 2.07 ISA is released, I'd rather not have existing programs degrade because suddenly the 2_06 bit is no longer set. Thanks, r~