From: Chandra Seetharaman <sekharan@us.ibm.com>
To: device-mapper development <dm-devel@redhat.com>
Cc: Alasdair G Kergon <agk@redhat.com>
Subject: Re: [PATCH] dm mpath: delay retry activate_path on SCSI_DH_RETRY
Date: Wed, 18 Feb 2009 17:55:46 -0800 [thread overview]
Message-ID: <1235008546.15552.60.camel@chandra-ubuntu> (raw)
In-Reply-To: <200902171917.38252.knikanth@suse.de>
Hi Nikanth,
Thanks for the patch.
Please see my comment below.
chandra
On Tue, 2009-02-17 at 19:17 +0530, Nikanth Karthikesan wrote:
> Delay retry to activate_path if it returns SCSI_DH_RETRY.
>
> Signed-off-by: Nikanth Karthikesan <knikanth@suse.de>
>
> ---
>
> diff --git a/drivers/md/dm-mpath.c b/drivers/md/dm-mpath.c
> index 095f77b..af54632 100644
> --- a/drivers/md/dm-mpath.c
> +++ b/drivers/md/dm-mpath.c
> @@ -65,12 +65,14 @@ struct multipath {
> spinlock_t lock;
>
> const char *hw_handler_name;
> - struct work_struct activate_path;
> + struct delayed_work activate_path;
> struct pgpath *pgpath_to_activate;
> unsigned nr_priority_groups;
> struct list_head priority_groups;
> unsigned pg_init_required; /* pg_init needs calling? */
> unsigned pg_init_in_progress; /* Only one pg_init allowed at once */
> + unsigned long pg_init_jiffy; /* To delay retry if SCSI_DH_RETRY */
> +#define SCSI_DH_RETRY_DELAY ((HZ * 2))
>
> unsigned nr_valid_paths; /* Total number of usable paths */
> struct pgpath *current_pgpath;
> @@ -203,7 +205,7 @@ static struct multipath *alloc_multipath(struct dm_target *ti)
> m->queue_io = 1;
> INIT_WORK(&m->process_queued_ios, process_queued_ios);
> INIT_WORK(&m->trigger_event, trigger_event);
> - INIT_WORK(&m->activate_path, activate_path);
> + INIT_DELAYED_WORK(&m->activate_path, activate_path);
> m->mpio_pool = mempool_create_slab_pool(MIN_IOS, _mpio_cache);
> if (!m->mpio_pool) {
> kfree(m);
> @@ -431,6 +433,8 @@ static void process_queued_ios(struct work_struct *work)
> struct pgpath *pgpath = NULL;
> unsigned init_required = 0, must_queue = 1;
> unsigned long flags;
> + unsigned long delay = 0;
> + unsigned long now;
>
> spin_lock_irqsave(&m->lock, flags);
>
> @@ -452,13 +456,20 @@ static void process_queued_ios(struct work_struct *work)
> m->pg_init_required = 0;
> m->pg_init_in_progress = 1;
> init_required = 1;
> + /* Delay retry due to SCSI_DH_RETRY */
> + if (m->pg_init_jiffy) {
> + now = jiffies;
> + if (time_after(now, m->pg_init_jiffy))
> + delay = now - m->pg_init_jiffy;
I think the logic is reversed. Acc to linux/jiffies.h, "time_after(a,b)
returns true if time a is after time b",
We want it other way around, don't we ?
IMO, we need _not_ be so critical of the time. We could just set a flag
in pg_init_done and use 2 seconds in queue_delayed_work().
> + m->pg_init_jiffy = 0;
> + }
> }
>
> out:
> spin_unlock_irqrestore(&m->lock, flags);
>
> if (init_required)
> - queue_work(kmpath_handlerd, &m->activate_path);
> + queue_delayed_work(kmpath_handlerd, &m->activate_path, delay);
>
> if (!must_queue)
> dispatch_queued_ios(m);
> @@ -1060,6 +1071,7 @@ static void pg_init_done(struct dm_path *path, int errors)
> struct priority_group *pg = pgpath->pg;
> struct multipath *m = pg->m;
> unsigned long flags;
> + bool delay = false;
>
> /* device or driver problems */
> switch (errors) {
> @@ -1084,8 +1096,11 @@ static void pg_init_done(struct dm_path *path, int errors)
> */
> bypass_pg(m, pg, 1);
> break;
> - /* TODO: For SCSI_DH_RETRY we should wait a couple seconds */
> + /*
> + * For SCSI_DH_RETRY we wait for a couple seconds.
> + */
> case SCSI_DH_RETRY:
> + delay = true;
> case SCSI_DH_IMM_RETRY:
> case SCSI_DH_RES_TEMP_UNAVAIL:
> if (pg_init_limit_reached(m, pgpath))
> @@ -1112,6 +1127,10 @@ static void pg_init_done(struct dm_path *path, int errors)
> }
>
> m->pg_init_in_progress = 0;
> + if (delay)
> + m->pg_init_jiffy = jiffies + SCSI_DH_RETRY_DELAY;
> + else
> + m->pg_init_jiffy = 0;
> queue_work(kmultipathd, &m->process_queued_ios);
> spin_unlock_irqrestore(&m->lock, flags);
> }
> @@ -1120,7 +1139,7 @@ static void activate_path(struct work_struct *work)
> {
> int ret;
> struct multipath *m =
> - container_of(work, struct multipath, activate_path);
> + container_of(work, struct multipath, activate_path.work);
> struct dm_path *path;
> unsigned long flags;
>
>
> --
> dm-devel mailing list
> dm-devel@redhat.com
> https://www.redhat.com/mailman/listinfo/dm-devel
next prev parent reply other threads:[~2009-02-19 1:55 UTC|newest]
Thread overview: 16+ messages / expand[flat|nested] mbox.gz Atom feed top
2009-02-17 13:47 [PATCH] dm mpath: delay retry activate_path on SCSI_DH_RETRY Nikanth Karthikesan
2009-02-19 1:55 ` Chandra Seetharaman [this message]
2009-02-19 2:11 ` Alasdair G Kergon
2009-02-19 7:10 ` Nikanth Karthikesan
2009-02-20 0:45 ` Chandra Seetharaman
2009-02-20 5:03 ` Nikanth Karthikesan
2009-02-20 15:08 ` Konrad Rzeszutek
2009-02-20 21:11 ` Chandra Seetharaman
2009-03-02 10:48 ` Nikanth Karthikesan
-- strict thread matches above, loose matches on Subject: below --
2009-04-28 6:15 Nikanth Karthikesan
2009-04-28 19:35 ` Chandra Seetharaman
2009-04-28 22:34 ` Alasdair G Kergon
2009-05-05 3:18 Chandra Seetharaman
2009-05-15 3:10 ` Chandra Seetharaman
2009-05-15 13:38 ` Mike Christie
2009-06-09 20:54 ` Chandra Seetharaman
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=1235008546.15552.60.camel@chandra-ubuntu \
--to=sekharan@us.ibm.com \
--cc=agk@redhat.com \
--cc=dm-devel@redhat.com \
--cc=sekharan@linux.vnet.ibm.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).