From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([208.118.235.92]:52838) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1SvqZw-0001q8-Ax for qemu-devel@nongnu.org; Mon, 30 Jul 2012 10:04:17 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1SvqZq-0000Kh-Fd for qemu-devel@nongnu.org; Mon, 30 Jul 2012 10:04:12 -0400 Received: from mx1.redhat.com ([209.132.183.28]:45759) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1SvqZq-0000Kc-83 for qemu-devel@nongnu.org; Mon, 30 Jul 2012 10:04:06 -0400 Message-ID: <5016944E.4010300@redhat.com> Date: Mon, 30 Jul 2012 16:03:58 +0200 From: Paolo Bonzini MIME-Version: 1.0 References: <1343127865-16608-1-git-send-email-pbonzini@redhat.com> <1343127865-16608-44-git-send-email-pbonzini@redhat.com> <5013EC8A.3000409@redhat.com> <50168ED0.7070502@redhat.com> <501691DA.50808@redhat.com> In-Reply-To: <501691DA.50808@redhat.com> Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 7bit Subject: Re: [Qemu-devel] [PATCH 43/47] mirror: allow customizing the granularity List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Eric Blake Cc: kwolf@redhat.com, jcody@redhat.com, qemu-devel@nongnu.org, stefanha@linux.vnet.ibm.com Il 30/07/2012 15:53, Eric Blake ha scritto: > On 07/30/2012 07:40 AM, Paolo Bonzini wrote: >> Il 28/07/2012 15:43, Eric Blake ha scritto: >>>>> + if (granularity < 512 || granularity > 1048576 * 64) { >>>>> + error_set(errp, QERR_INVALID_PARAMETER, device); >>>>> + return; >>>>> + } >>>>> + if (granularity & (granularity - 1)) { >>>>> + error_set(errp, QERR_INVALID_PARAMETER, device); >>>>> + return; >>> In the XBZLRE migration series, we decided to round the users input down >>> to a power of two instead of reject it. Should we do that here? >> >> I can certainly do that, but do you have a pointer to the discussion so >> that I can understand the rationale? > > https://lists.gnu.org/archive/html/qemu-devel/2012-05/msg02421.html Hmm, a buffer size (the cache size in the case of XBZRLE) is different however. There is no reason in principle why it needs to be a power of two (except if you know how the hash table is implemented, or something like that). For example, the buf_size argument in this series does indeed support a non-power-of-two size. Requesting a granularity to be a power-of-two shouldn't be surprising to anybody who has an idea of what a bit shift is and how it is used... Paolo