Re: [PATCH] imsm: FIX: Limit migration record operation by disk slot not by index

[NeilBrown <neilb@suse.de>]

[-- Attachment #1: Type: text/plain, Size: 1103 bytes --]

On Thu, 08 Dec 2011 09:46:56 +0100 Adam Kwolek <adam.kwolek@intel.com> wrote:

> imsm should store migration record in to 2 first disks in array.
> This should be evaluated based on disk slots, not on disks index.
> It is not guaranteed that indexes are equal to slots.
> 
> Signed-off-by: Adam Kwolek <adam.kwolek@intel.com>
> ---
...

> @@ -2200,11 +2245,37 @@ static int write_imsm_migr_rec(struct supertype *st)
>  	struct dl *sd;
>  	int len;
>  	struct imsm_update_general_migration_checkpoint *u;
> +	struct imsm_dev *dev;
> +	struct imsm_map *map = NULL;
> +
> +	/* find map under migration */
> +	dev = imsm_get_device_during_migration(super);
> +	/* if no migration, write buffer anyway to clear migr_record
> +	 * on disk based on first available device
> +	*/
> +	if (dev == NULL) {
> +		int vol;
> +
> +		if (super->current_vol >= 0)
> +			vol = super->current_vol;
> +		else
> +			vol = 0;
> +		dev = get_imsm_dev(super, super->current_vol);
> +	}

'vol' is unused here.  Should it be:
                dev = get_imsm_dev(super, vol);
??

NeilBrown



[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 828 bytes --]