ovsdb: Add lazy-copy support for ovsdb_datum objects.

Currently ovsdb-server is using shallow copies of some JSON objects
by keeping a reference counter.  JSON string objects are also used
directly as ovsdb atoms in database rows to avoid extra copies.

Taking this approach one step further ovsdb_datum objects can also
be mostly deduplicated by postponing the copy until it actually
needed.  datum object itself contains a type and 2 pointers to
data arrays.  Adding a one more pointer to a reference counter
we may create a shallow copy of the datum by simply copying type
and pointers and increasing the reference counter.

Before modifying the datum, special function needs to be called
to perform an actual copy of the object, a.k.a. unshare it.
Most of the datum modifications are performed inside the special
functions in ovsdb-data.c, so that is not very hard to track.
A few places like ovsdb-server.c and column mutations are accessing
and changing the data directly, so a few extra unshare() calls
has to be added there.

This change doesn't affect the maximum memory consumption too much,
because most of the copies are short-living.  However, not actually
performing these copies saves up to 40% of CPU time on operations
with large sets.

Reported-at: https://bugzilla.redhat.com/2069089
Acked-by: Dumitru Ceara <dceara@redhat.com>
Signed-off-by: Ilya Maximets <i.maximets@ovn.org>

1 files changed, 7 insertions, 6 deletions

diff --git a/lib/ovsdb-data.h b/lib/ovsdb-data.h
index ba5d179a6..dcb620513 100644
--- a/lib/ovsdb-data.h
+++ b/lib/ovsdb-data.h
@@ -146,6 +146,8 @@ struct ovsdb_datum {
     unsigned int n;             /* Number of 'keys' and 'values'. */
     union ovsdb_atom *keys;     /* Each of the ovsdb_type's 'key_type'. */
     union ovsdb_atom *values;   /* Each of the ovsdb_type's 'value_type'. */
+    unsigned int *refcnt;       /* Number of copies with the same
+                                 * 'keys' and 'values'. */
 };
 #define OVSDB_DATUM_INITIALIZER { 0, NULL, NULL }
 
@@ -155,22 +157,21 @@ void ovsdb_datum_init_default(struct ovsdb_datum *, const struct ovsdb_type *);
 bool ovsdb_datum_is_default(const struct ovsdb_datum *,
                             const struct ovsdb_type *);
 const struct ovsdb_datum *ovsdb_datum_default(const struct ovsdb_type *);
-void ovsdb_datum_clone(struct ovsdb_datum *, const struct ovsdb_datum *,
-                       const struct ovsdb_type *);
+void ovsdb_datum_clone(struct ovsdb_datum *, const struct ovsdb_datum *);
 void ovsdb_datum_destroy(struct ovsdb_datum *, const struct ovsdb_type *);
+void ovsdb_datum_unshare(struct ovsdb_datum *, const struct ovsdb_type *);
 void ovsdb_datum_swap(struct ovsdb_datum *, struct ovsdb_datum *);
 
 /* Checking and maintaining invariants. */
 struct ovsdb_error *ovsdb_datum_sort(struct ovsdb_datum *,
-                                     enum ovsdb_atomic_type key_type)
+                                     const struct ovsdb_type *type)
     OVS_WARN_UNUSED_RESULT;
 
 void ovsdb_datum_sort_assert(struct ovsdb_datum *,
-                             enum ovsdb_atomic_type key_type);
+                             const struct ovsdb_type *type);
 
 size_t ovsdb_datum_sort_unique(struct ovsdb_datum *,
-                               enum ovsdb_atomic_type key_type,
-                               enum ovsdb_atomic_type value_type);
+                               const struct ovsdb_type *type);
 
 struct ovsdb_error *ovsdb_datum_check_constraints(
     const struct ovsdb_datum *, const struct ovsdb_type *)