From mboxrd@z Thu Jan 1 00:00:00 1970 From: Paul Mackerras Subject: Re: [PATCH 4/8] Added ONE_REG interface for debug instruction Date: Mon, 11 Feb 2013 11:22:35 +1100 Message-ID: <20130211002235.GA32734@drongo> References: <1358324685-30225-1-git-send-email-bharat.bhushan@freescale.com> <1358324685-30225-3-git-send-email-bharat.bhushan@freescale.com> <20130204004104.GA17537@drongo> <40A423D4-6230-4ED5-A1EB-6589E7016DE4@suse.de> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: Bharat Bhushan , kvm-ppc@vger.kernel.org, kvm@vger.kernel.org, Bharat Bhushan To: Alexander Graf Return-path: Content-Disposition: inline In-Reply-To: <40A423D4-6230-4ED5-A1EB-6589E7016DE4@suse.de> Sender: kvm-ppc-owner@vger.kernel.org List-Id: kvm.vger.kernel.org On Thu, Feb 07, 2013 at 03:29:50PM +0100, Alexander Graf wrote: > > On 04.02.2013, at 01:41, Paul Mackerras wrote: > > > On Wed, Jan 16, 2013 at 01:54:41PM +0530, Bharat Bhushan wrote: > >> This patch adds the one_reg interface to get the special instruction > >> to be used for setting software breakpoint from userspace. > > > > Since this presumably is constant for any given platform, wouldn't > > a capability be more appropriate? > > How so? A capability only tells you "I can do debug". Or "I can do debug on e500". I don't want to reteach QEMU how to do debug for every core we implement. Capabilities aren't just binary - the get-capability (check_extension) ioctl returns a 32-bit value, and we already have some that return values other than 0 or 1. So I was thinking that we could add a capability which when queried returns the special instruction. It's up to you whether you want to do it that way or not. Paul.