From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([140.186.70.92]:54755) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1RaW7A-0006HJ-SJ for qemu-devel@nongnu.org; Tue, 13 Dec 2011 12:26:08 -0500 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1RaW75-0006aT-4D for qemu-devel@nongnu.org; Tue, 13 Dec 2011 12:26:04 -0500 Received: from relay1.mentorg.com ([192.94.38.131]:55325) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1RaW74-0006aI-Pl for qemu-devel@nongnu.org; Tue, 13 Dec 2011 12:25:58 -0500 From: Paul Brook Date: Tue, 13 Dec 2011 17:25:51 +0000 References: <1323717136-21661-1-git-send-email-stefanb@linux.vnet.ibm.com> <201112130451.44284.paul@codesourcery.com> <4EE74A45.8000306@linux.vnet.ibm.com> In-Reply-To: <4EE74A45.8000306@linux.vnet.ibm.com> MIME-Version: 1.0 Content-Type: Text/Plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Message-Id: <201112131725.52323.paul@codesourcery.com> Subject: Re: [Qemu-devel] [PATCH V13 6/7] Introduce --enable-tpm-passthrough configure option List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Stefan Berger Cc: andreas.niederl@iaik.tugraz.at, qemu-devel@nongnu.org, mst@redhat.com > On 12/12/2011 11:51 PM, Paul Brook wrote: > >>>> +tpm_passthrough="no" > >>> > >>> Same as before, please probe for existence. > >> > >> We would be probing for /dev/tpm0. Is that really what we want that this > >> driver only gets compiled if /dev/tpm0 is (currently) available? > > > > If what you say is true then this code should always be enabled. > > Michael Tsirkin previously requested that there be an option for the TPM > passthrough driver to be selectively enabled since at least using > /dev/tpm0 may not be what everybody wants. The passthrough driver at > some point will also be able to use sockets to communicate with a TPM > when a file descriptor is passed to Qemu, so maybe that changes then? Surely that's a runtime decision made by the qemu user, not a compile time decision made by the distribution vendor. Testing functionality of the build machine is fundamentally wrong. At best it completely breaks cross compiling, at worst it subtly breaks building on a dev machine and running on a production server. Configure time probing only makes sense for code that will fail to build if external compile time dependencies are not present. If you really think building qemu without this functionality is useful, then maybe add a configure option to disable it. With the possible exception of developer debugging aids, code that is not enabled by default is clearly not worth having and should be removed altogether. Paul