From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:52695) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1UrBmp-0008Ou-Gs for qemu-devel@nongnu.org; Mon, 24 Jun 2013 14:46:52 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1UrBml-0002eE-AH for qemu-devel@nongnu.org; Mon, 24 Jun 2013 14:46:47 -0400 Received: from mx.ipv6.kamp.de ([2a02:248:0:51::16]:34613 helo=mx01.kamp.de) by eggs.gnu.org with smtp (Exim 4.71) (envelope-from ) id 1UrBml-0002dc-02 for qemu-devel@nongnu.org; Mon, 24 Jun 2013 14:46:43 -0400 Message-ID: <51C8940E.9010508@kamp.de> Date: Mon, 24 Jun 2013 20:46:38 +0200 From: Peter Lieven MIME-Version: 1.0 References: <1371934712-11714-1-git-send-email-pl@kamp.de> <1371934712-11714-7-git-send-email-pl@kamp.de> <51C858B3.7010200@redhat.com> <51C87120.9010105@kamp.de> <51C872FA.9060404@redhat.com> In-Reply-To: <51C872FA.9060404@redhat.com> Content-Type: text/plain; charset=ISO-8859-15 Content-Transfer-Encoding: 7bit Subject: Re: [Qemu-devel] [PATCH 6/8] qemu-img: use bdrv_write_zeroes to write zeroes List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Paolo Bonzini Cc: Kevin Wolf , ronniesahlberg@gmail.com, qemu-devel@nongnu.org, Stefan Hajnoczi Am 24.06.2013 18:25, schrieb Paolo Bonzini: > Il 24/06/2013 18:17, Peter Lieven ha scritto: >> Am 24.06.2013 16:33, schrieb Paolo Bonzini: >>> Il 22/06/2013 22:58, Peter Lieven ha scritto: >>>> Signed-off-by: Peter Lieven >>>> --- >>>> qemu-img.c | 10 +++++++--- >>>> 1 file changed, 7 insertions(+), 3 deletions(-) >>>> >>>> diff --git a/qemu-img.c b/qemu-img.c >>>> index 809b4f1..5aa53ab 100644 >>>> --- a/qemu-img.c >>>> +++ b/qemu-img.c >>>> @@ -1513,9 +1513,13 @@ static int img_convert(int argc, char **argv) >>>> If the output is to a host device, we also write out >>>> sectors that are entirely 0, since whatever data was >>>> already there is garbage, not 0s. */ >>>> - if (!has_zero_init || out_baseimg || >>>> - is_allocated_sectors_min(buf1, n, &n1, min_sparse)) { >>>> - ret = bdrv_write(out_bs, sector_num, buf1, n1); >>>> + int allocated = is_allocated_sectors_min(buf1, n, &n1, min_sparse); >>>> + if (!has_zero_init || out_baseimg || allocated) { >>>> + if (allocated || out_baseimg) { >>>> + ret = bdrv_write(out_bs, sector_num, buf1, n1); >>>> + } else { >>>> + ret = bdrv_write_zeroes(out_bs, sector_num, n1); >>> I think it should still do the write only if !has_zero_init. >> With this I am still in trouble with iSCSI, but Kevin pointed out another >> possible solution for the allocating everything problem. >> >> a) let iscsi_create discard the whole device if lbpz && lbprz >> b) return 1 for has_zero_init if lbprz. >> >> What do you think? > Fine by me (but the discard may take a looong time! :)). fyi: i have created an optimized unmap function which transmits the maximum allowed unmap_list length (read from block_limits page). for 1TB it takes approx. 1 minute on my storage. i will include this in v2 of the series and also add the optimization to the other unmap calls (which currently do not check max_unmap size at all). Peter