On Wed, Feb 06, 2013 at 01:31:40PM +0100, Benoît Canet wrote:
> +/* Deduplicate all the cluster that can be deduplicated.
> + *
> + * Next it compute the number of non deduplicable sectors to come while
> storing
s/compute/computes/
> + * the hashes of these sectors in a linked list for later usage.
> + * Then it compute the first duplicated cluster hash that come after non
s/compute/computes/
s/come/comes/
> + * deduplicable cluster, this hash will be used at next call of the function
> + *
> + * @ds: a structure containing the state of the deduplication
> + * for this write request
> + * @sector_num: The logical sector
> + * @data: the buffer containing the data to deduplicate
> + * @data_nr: the size of the buffer in sectors
> + *
> + */
> +int qcow2_dedup(BlockDriverState *bs,
> + QCowDedupState *ds,
> + uint64_t sector_num,
> + uint8_t *data,
> + int data_nr)
> +{
> + BDRVQcowState *s = bs->opaque;
> + int ret = 0;
> + int deduped_clusters_nr = 0;
> + int left_to_process;
> + int begining_index;
> +
> + begining_index = sector_num & (s->cluster_sectors - 1);
s/begining/beginning/ but start_index would be shorter.
> +
> + left_to_process = (data_nr / s->cluster_sectors) -
> + ds->nb_clusters_processed;
data_nr must be a multiple of cluster_sectors?
