From mboxrd@z Thu Jan 1 00:00:00 1970 From: Xiao Ni Subject: Re: One patch just review the code Date: Thu, 7 Aug 2014 06:06:19 -0400 (EDT) Message-ID: <835401588.19232017.1407405979159.JavaMail.zimbra@redhat.com> References: <705044700.17914512.1407222497904.JavaMail.zimbra@redhat.com> <1122994629.17915693.1407222727992.JavaMail.zimbra@redhat.com> <20140806164420.09e4df13@notabene.brown> Mime-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <20140806164420.09e4df13@notabene.brown> Sender: linux-raid-owner@vger.kernel.org To: NeilBrown Cc: linux-raid@vger.kernel.org, Jes Sorensen List-Id: linux-raid.ids ----- Original Message ----- > From: "NeilBrown" > To: "Xiao Ni" > Cc: linux-raid@vger.kernel.org, "Jes Sorensen" > Sent: Wednesday, August 6, 2014 2:44:20 PM > Subject: Re: One patch just review the code > > On Tue, 5 Aug 2014 03:12:07 -0400 (EDT) Xiao Ni wrote: > > > Hi all > > > > I'm reading the code of md. I find there is a problem. I know now there > > are arrays mark[SYNC_MARKS] mark_cnt[SYNC_MARKS] > > store the information about how many sectors finish recovery and the > > moment. > > > > > > 7825 currspeed = ((unsigned > > long)(io_sectors-mddev->resync_mark_cnt))/2 > > 7826 /((jiffies-mddev->resync_mark)/HZ +1) +1; > > > > But when calculate the speed of recovery, the sectors used to calculate > > contains > > the sectors which are not finished recovery. > > > > When assign value to mark_cnt[next], it subtract the sectors which don't > > finish recovery. > > So I think when calculate the recovery speed we should subtract the sectors > > not finishing > > recovery too. > > > > 7638 mark_cnt[next] = io_sectors - > > atomic_read(&mddev->recovery_active); > > > > So I try to modify and the patch is: > > > > --- linux-stable/drivers/md/md.c 2014-07-30 14:36:37.327535805 +0800 > > +++ fix/md.c 2014-07-31 16:40:57.151493177 +0800 > > @@ -7652,7 +7652,7 @@ > > */ > > cond_resched(); > > > > - currspeed = ((unsigned long)(io_sectors-mddev->resync_mark_cnt))/2 > > + currspeed = ((unsigned > > long)(io_sectors-atomic_read(&mddev->recovery_active)-mddev->resync_mark_cnt))/2 > > /((jiffies-mddev->resync_mark)/HZ +1) +1; > > > > if (currspeed > speed_min(mddev)) { > > > > Am I right? > > Yes, that looks right. > If you create a properly formatted patch, and wrap that long line nicely I'll > apply it. > > Thanks, > NeilBrown > I definite a new variable, do you allow me to do by this way? Signed-off-by: Xiao Ni diff -urN linux-stable/drivers/md/md.c fix/md.c --- linux-stable/drivers/md/md.c 2014-07-30 14:36:37.327535805 +0800 +++ fix/md.c 2014-08-07 16:07:12.559503942 +0800 @@ -7376,7 +7376,7 @@ struct mddev *mddev2; unsigned int currspeed = 0, window; - sector_t max_sectors,j, io_sectors; + sector_t max_sectors,j, io_sectors, recovery_done; unsigned long mark[SYNC_MARKS]; unsigned long update_time; sector_t mark_cnt[SYNC_MARKS]; @@ -7652,7 +7652,8 @@ */ cond_resched(); - currspeed = ((unsigned long)(io_sectors-mddev->resync_mark_cnt))/2 + recovery_done = io_sectors - atomic_read(&mddev->recovery_active); + currspeed = ((unsigned long)recovery_done - mddev->resync_mark_cnt)/2 /((jiffies-mddev->resync_mark)/HZ +1) +1; if (currspeed > speed_min(mddev)) {