Re: [PATCH 2 of 2] DM Snapshot: dont insert before existing chunk

[Mike Snitzer <snitzer@redhat.com>]

On Thu, Sep 24 2009 at  8:48am -0400,
Alasdair G Kergon <agk@redhat.com> wrote:

> On Wed, Sep 23, 2009 at 10:25:08AM -0500, Jon Brassow wrote:
> > Don't insert into hash before an existing chunk.
> > Most inserts are after an existing extent anyway so this code is used
> > very rarely.
>  
> Again, I'm not particularly keen on taking this one.  Some systems are very
> short of memory and everything we can save matters and we have no statistics
> about how "rare" this actually is.
> 
> Is it really that difficult to split into two when removing?

Jon, Mikulas and I examined/discussed this in detail today.  We had a
nice break-through: snapshot-merge no longer needs to avoid inserting
before an existing chunk.  As such snapshot-merge no longer imposes less
efficient packing of the exception cache.

We needed a small change to the snapshot-merge clear_exception() code so
that it increments e->old_chunk and e->new_chunk when it makes its call
to dm_consecutive_chunk_count_dec(), e.g.:

        if (dm_consecutive_chunk_count(e)) {
                if ((de->old_chunk == e->old_chunk) &&
                    (de->new_chunk == dm_chunk_number(e->new_chunk))) {
                        e->old_chunk++;
                        e->new_chunk++;
                }
                dm_consecutive_chunk_count_dec(e);
        }

This is the inverse of what is done in dm_insert_exception():

        /* Insert before an existing chunk? */
        if (new_e->old_chunk == (e->old_chunk - 1) &&
            new_e->new_chunk == (dm_chunk_number(e->new_chunk) - 1)) {
                dm_consecutive_chunk_count_inc(e);
                e->old_chunk--;
                e->new_chunk--;
                dm_free_exception(eh, new_e);
                return;
        }

Updated DM snapshot-merge patches are available here:
http://people.redhat.com/msnitzer/patches/snapshot-merge/kernel_unified/2.6.31/