From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from merlin.infradead.org ([205.233.59.134]:38587 "EHLO merlin.infradead.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751650Ab3H0RDL (ORCPT ); Tue, 27 Aug 2013 13:03:11 -0400 Message-ID: <521CDBC2.8050705@kernel.dk> Date: Tue, 27 Aug 2013 11:02:58 -0600 From: Jens Axboe MIME-Version: 1.0 Subject: Re: [PATCH 1/2] Adds check for numberio during verify phase. References: <1377552162-31741-1-git-send-email-jcasse@chromium.org> In-Reply-To: <1377552162-31741-1-git-send-email-jcasse@chromium.org> Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Sender: fio-owner@vger.kernel.org List-Id: fio@vger.kernel.org To: Juan Casse Cc: Grant Grundler , fio@vger.kernel.org On 08/26/2013 03:22 PM, Juan Casse wrote: > We check numberio to detect stale blocks. > > Signed-off-by: Juan Casse > Reviewed-by: Grant Grundler > --- > HOWTO | 7 +++++++ > init.c | 33 +++++++++++++++++++++++++++++++++ > ioengine.h | 1 + > iolog.c | 1 + > iolog.h | 1 + > libfio.c | 3 ++- > options.c | 9 +++++++++ > thread_options.h | 2 ++ > verify.c | 10 ++++++++-- > 9 files changed, 64 insertions(+), 3 deletions(-) > > diff --git a/HOWTO b/HOWTO > index 005dac2..14ff8b5 100644 > --- a/HOWTO > +++ b/HOWTO > @@ -657,6 +657,13 @@ ioengine=str Defines how the job issues io to the file. The following > filename, eg ioengine=external:/tmp/foo.o > to load ioengine foo.o in /tmp. > > +data_integrity_check > + If this option is given, fio will check the i/o number of > + each block read back during the verification phase. Fio > + checks numberio to detect stale blocks. Currently, this > + option requires synchronous i/o, and equal-sized read and > + write blocks. This option requires workloads that write data. I think this use case is just too narrow. Why does it require sync IO and equal read/write sizes? Can't you just replay and re-generate and compare? -- Jens Axboe