[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Qemu-devel] [RFC V6 24/33] qcow2: Integrate deduplication in qcow2_
From: |
Stefan Hajnoczi |
Subject: |
Re: [Qemu-devel] [RFC V6 24/33] qcow2: Integrate deduplication in qcow2_co_writev loop. |
Date: |
Thu, 28 Feb 2013 10:38:02 +0100 |
User-agent: |
Mutt/1.5.21 (2010-09-15) |
On Wed, Feb 27, 2013 at 03:31:48PM +0100, Benoît Canet wrote:
> > I find this expression hard to understand.
> >
> > We only get here if ds.nb_undedupable_sectors > 0. In other words, we
> > tried to dedup but failed, so we must write data into the image file.
> >
> > Can we ensure that ds.nb_undedupable_sectors is limited to at most
> > remaining_sectors? Then the expression becomes clearer:
>
> ds.nb_undedupable_sectors can be bigger than remaining_sectors if
> dedup_cluster_data was completed at its begining and/or at its ending in case
> of
> an unaligned write.
>
> Maybe I could grow/replace qiov at the begining of writev to ingest the
> completed data and increase remaining_sectors so its will be alway >=
> ds.nb_undedupable_sectors ?
>
> What do you think about this solution ?
Sure, if the semantics of the rest of this function can cope with
extending qiov.
Stefan
- Re: [Qemu-devel] [RFC V6 13/33] qcow2: make the deduplication forget a cluster hash when a cluster is to dedupe, (continued)
- [Qemu-devel] [RFC V6 14/33] qcow2: Create qcow2_is_cluster_to_dedup., Benoît Canet, 2013/02/06
- [Qemu-devel] [RFC V6 28/33] qcow2: Add check_dedup_l2 in order to check l2 of dedup table., Benoît Canet, 2013/02/06
- [Qemu-devel] [RFC V6 09/33] qcow2: Implement qcow2_compute_cluster_hash., Benoît Canet, 2013/02/06
- [Qemu-devel] [RFC V6 10/33] qcow2: Extract qcow2_dedup_grow_table, Benoît Canet, 2013/02/06
- Re: [Qemu-devel] [RFC V6 00/33] QCOW2 deduplication core functionality, Stefan Hajnoczi, 2013/02/08