Hi, I've made sync_completed sysfs entry writeable to let md know that configuration prepared by mdmon is ready. Other thing that this is used for, is that after write by mdmon 0 to sync_completed, mdadm has to be able read this information to proceed with reshape limits. To achieve this, flag MD_SYNC_COMPLETED_SHOW_0 is used, so after 0 is put to sysfs it is exposed to user space until it is read by mdadm. Algorithm is as follow: 1. mdadm puts "reshape" to action => clear MD_SYNC_COMPLETED_SHOW_0 (0 is not exposed) 2. kernel starts reshape and waits in wait_reshape() for reconfiguration by mdmon 3. mdmon adds devices and "0" is put to sync_completed => MD_SYNC_COMPLETED_SHOW_0 is set 4. md goes with reshape (it probably stops due to max_sync is not set by mdadm yet) 5. mdadm read from sync completed gives "0" 6. mdadm stores sync_max (so md knows it can go forward) => MD_SYNC_COMPLETED_SHOW_0 is cleared during sync_max store Considering above comment can be "add possibility to use "0" value in sync complete during reshape start handshake/synchronization" If we decide to remove wait_reshape() this synchronization will be not necessary. Adam -----Original Message----- From: Neil Brown [mailto:neilb@xxxxxxx] Sent: Thursday, March 04, 2010 6:59 AM To: Kwolek, Adam Cc: linux-raid@xxxxxxxxxxxxxxx; Williams, Dan J; Ciechanowski, Ed Subject: Re: [md PATCH 1/8] Enable OLCE for external IMSM metadata On Wed, 24 Feb 2010 07:42:48 +0000 "Kwolek, Adam" <adam.kwolek@xxxxxxxxx> wrote: > >From 464c542abb3f107771a4d97611ef5035e40755e9 Mon Sep 17 00:00:00 2001 > From: Adam Kwolek <adam.kwolek@xxxxxxxxx> > Date: Thu, 18 Feb 2010 11:24:19 +0100 > Subject: [PATCH] OLCE: add sync_completed handler > > Changes to be committed: > modified: md.c > modified: md.h > > Add possibility to send notification to md via sync_completed I'm sorry, but this patch doesn't make much sense to me. The comment above doesn't seem to match the code below. Can I have a more complete explanation please. NeilBrown > > Signed-off-by: Adam Kwolek <adam.kwolek@xxxxxxxxx> > --- > drivers/md/md.c | 53 ++++++++++++++++++++++++++++++++++++++++++++++++++++- > drivers/md/md.h | 2 ++ > 2 files changed, 54 insertions(+), 1 deletions(-) > > diff --git a/drivers/md/md.c b/drivers/md/md.c > index f836926..d7e1053 100644 > --- a/drivers/md/md.c > +++ b/drivers/md/md.c > @@ -3461,6 +3461,10 @@ action_store(mddev_t *mddev, const char *page, size_t len) > int err; > if (mddev->pers->start_reshape == NULL) > return -EINVAL; > + > + /* OLCE: correct value saved, cannot show zero */ > + clear_bit(MD_SYNC_COMPLETED_SHOW_0, &mddev->flags); > + > err = mddev->pers->start_reshape(mddev); > if (err) > return err; > @@ -3601,6 +3605,14 @@ sync_completed_show(mddev_t *mddev, char *page) > { > unsigned long max_sectors, resync; > > + /* bit is cleared by saving to sysfs: > + * action: reshape > + * setting valid sync_max > + * seting other value than 0 to sync_compl > + */ > + if (test_bit(MD_SYNC_COMPLETED_SHOW_0, &mddev->flags)) > + return sprintf(page, "0\n"); > + > if (!test_bit(MD_RECOVERY_RUNNING, &mddev->recovery)) > return sprintf(page, "none\n"); > > @@ -3613,7 +3625,40 @@ sync_completed_show(mddev_t *mddev, char *page) > return sprintf(page, "%lu / %lu\n", resync, max_sectors); > } > > -static struct md_sysfs_entry md_sync_completed = __ATTR_RO(sync_completed); > +static ssize_t > +sync_completed_store(mddev_t *mddev, const char *buf, size_t len) > +{ > + /* accept set to 0 only as signal */ > + unsigned long long passedValue; > + > + /* store for external meta only */ > + if (!mddev->external) > + return -EINVAL; > + > + /* check if numeric */ > + if (strict_strtoull(buf, 10, &passedValue)) > + return -EINVAL; > + > + /* use 0 only */ > + if (passedValue) > + return -EINVAL; > + > + if ((mddev->reshape_position == 0) && (*buf == '0') && (len == 1)) { > + /* not started yet - display passed '0' to signal user space*/ > + set_bit(MD_SYNC_COMPLETED_SHOW_0, &mddev->flags); > + } else { > + if ((*buf != '0') || (len != 1)) > + clear_bit(MD_SYNC_COMPLETED_SHOW_0, &mddev->flags); > + } > + > + return len; > +} > + > +static struct md_sysfs_entry md_sync_completed = > +__ATTR(sync_completed, S_IRUGO|S_IWUSR, > + sync_completed_show, > + sync_completed_store); > + > > static ssize_t > min_sync_show(mddev_t *mddev, char *page) > @@ -3679,6 +3724,12 @@ max_sync_store(mddev_t *mddev, const char *buf, size_t len) > } > mddev->resync_max = max; > } > + /* OLCE: correct value saved, cannot show zero */ > + if (test_bit(MD_SYNC_COMPLETED_SHOW_0, &mddev->flags)) { > + /* on begin some check points can be skipped */ > + clear_bit(MD_SYNC_COMPLETED_SHOW_0, &mddev->flags); > + } > + > wake_up(&mddev->recovery_wait); > return len; > } > diff --git a/drivers/md/md.h b/drivers/md/md.h > index f184b69..070e7c6 100644 > --- a/drivers/md/md.h > +++ b/drivers/md/md.h > @@ -119,6 +119,8 @@ struct mddev_s > #define MD_CHANGE_CLEAN 1 /* transition to or from 'clean' */ > #define MD_CHANGE_PENDING 2 /* superblock update in progress */ > > +#define MD_SYNC_COMPLETED_SHOW_0 3 /* force to show in sync complete 0 */ > + /* when it is set from user space */ > int suspended; > atomic_t active_io; > int ro; -- To unsubscribe from this list: send the line "unsubscribe linux-raid" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
