Re: [PATCH 2/2] imsm: Update metadata for second array

On Mon, 31 Jan 2011 09:28:54 +0000 "Kwolek, Adam" [off-list ref]
wrote:

quoted

-----Original Message-----
From: linux-raid-owner@vger.kernel.org [mailto:linux-raid-
owner@vger.kernel.org] On Behalf Of NeilBrown
Sent: Monday, January 31, 2011 10:19 AM
To: Kwolek, Adam
Cc: linux-raid@vger.kernel.org; Williams, Dan J; Ciechanowski, Ed;
Neubauer, Wojciech
Subject: Re: [PATCH 2/2] imsm: Update metadata for second array

On Mon, 31 Jan 2011 09:59:30 +0100 Adam Kwolek [off-list ref]
wrote:

quoted

When second array reshape is about to start external metadata should

be updated

quoted

by mdmon in imsm_set_array_state().
For this purposes imsm_progress_container_reshape() is reused.

Signed-off-by: Adam Kwolek <redacted>
---

 super-intel.c |    6 +++++-
 1 files changed, 5 insertions(+), 1 deletions(-)

diff --git a/super-intel.c b/super-intel.c
index 8484df6..0ab4355 100644
--- a/super-intel.c
+++ b/super-intel.c

@@ -5249,13 +5249,17 @@ static int imsm_set_array_state(struct

active_array *a, int consistent)

quoted

 		super->updates_pending++;
 	}

-	/* finalize online capacity expansion/reshape */
+	/* manage online capacity expansion/reshape */
 	if ((a->curr_action != reshape) &&
 	    (a->prev_action == reshape)) {
 		struct mdinfo *mdi;

+		/* finalize online capacity expansion/reshape */
 		for (mdi = a->info.devs; mdi; mdi = mdi->next)
 			imsm_set_disk(a, mdi->disk.raid_disk, mdi-
curr_state);
+
+		/* check next volume reshape */
+		imsm_progress_container_reshape(super);
 	}

 	return consistent;

You still haven't explained why you need this extra call to
imsm_progress_container_reshape.
Does the other call never get reached?  or does it do the wrong thing?
or is
it called too early? or too late or .....

NeilBrown

Call to imsm_progress_container_reshape() placed above (super-intel.c:5186), cannot be used as it is called for currently migrated volume only.
If we finish migration it will be never be called (guard for migration in progress: super-intel.c:5133)

Answer is: It is never called for no migration in progress (not too early and not too late: never).

This a reason I've add second call (after finalizing migration) for next array reshape initiation .

Thanks.
I've applied this for now, but I think something is badly messed up in the
handling of migration state - it is terribly convoluted.
However I don't have time to untangle it just now so I'll just take you patch
and hope it doesn't make anything worse :-)

Thanks,
NeilBrown