[PATCH v9 5/6] box: introduce offset_slot cache in key_part

[Vladimir Davydov]

On Sun, Feb 03, 2019 at 01:20:25PM +0300, Kirill Shcherbatov wrote:
> tuple_field_by_part looks up the tuple_field corresponding to the
> given key part in tuple_format in order to quickly retrieve the offset
> of indexed data from the tuple field map. For regular indexes this
> operation is blazing fast, however of JSON indexes it is not as we
> have to parse the path to data and then do multiple lookups in a JSON
> tree. Since tuple_field_by_part is used by comparators, we should
> strive to make this routine as fast as possible for all kinds of
> indexes.
> 
> This patch introduces an optimization that is supposed to make
> tuple_field_by_part for JSON indexes as fast as it is for regular
> indexes in most cases. We do that by caching the offset slot right in
> key_part. There's a catch here however - we create a new format
> whenever an index is dropped or created and we don't reindex old
> tuples. As a result, there may be several generations of tuples in the
> same space, all using different formats while there's the only key_def
> used for comparison.
> 
> To overcome this problem, we introduce the notion of tuple_format
> epoch. This is a counter incremented each time a new format is
> created. We store it in tuple_format and key_def, and we only use
> the offset slot cached in a key_def if it's epoch coincides with the
> epoch of the tuple format. If they don't, we look up a tuple_field as
> before, and then update the cached value provided the epoch of the
> tuple format.
> 
> Part of #1012
> ---
>  src/box/key_def.c      | 15 ++++++++++-----
>  src/box/key_def.h      | 14 ++++++++++++++
>  src/box/tuple.h        |  2 +-
>  src/box/tuple_format.c |  6 +++++-
>  src/box/tuple_format.h | 41 +++++++++++++++++++++++++++++++++++------
>  5 files changed, 65 insertions(+), 13 deletions(-)

Pushed to 2.1 with the following minor changes:

diff --git a/src/box/tuple_format.c b/src/box/tuple_format.c
index fc152cbb..2d9b71ee 100644
--- a/src/box/tuple_format.c
+++ b/src/box/tuple_format.c
@@ -674,7 +674,6 @@ tuple_format_reuse(struct tuple_format **p_format)
 		tuple_format_destroy(format);
 		free(format);
 		*p_format = *entry;
-		(*p_format)->epoch = ++formats_epoch;
 		return true;
 	}
 	return false;
@@ -738,12 +737,12 @@ tuple_format_new(struct tuple_format_vtab *vtab, void *engine,
 	format->is_temporary = is_temporary;
 	format->is_ephemeral = is_ephemeral;
 	format->exact_field_count = exact_field_count;
+	format->epoch = ++formats_epoch;
 	if (tuple_format_create(format, keys, key_count, space_fields,
 				space_field_count) < 0)
 		goto err;
 	if (tuple_format_reuse(&format))
 		return format;
-	format->epoch = ++formats_epoch;
 	if (tuple_format_register(format) < 0)
 		goto err;
 	if (tuple_format_add_to_hash(format) < 0) {
diff --git a/src/box/tuple_format.h b/src/box/tuple_format.h
index aedd3e91..01ed97ae 100644
--- a/src/box/tuple_format.h
+++ b/src/box/tuple_format.h
@@ -137,12 +137,6 @@ tuple_field_is_nullable(struct tuple_field *tuple_field)
  * Tuple format describes how tuple is stored and information about its fields
  */
 struct tuple_format {
-	/**
-	 * Counter that grows incrementally on space rebuild
-	 * used for caching offset slot in key_part, for more
-	 * details see key_part::offset_slot_cache.
-	 */
-	uint64_t epoch;
 	/** Virtual function table */
 	struct tuple_format_vtab vtab;
 	/** Pointer to engine-specific data. */
@@ -155,6 +149,12 @@ struct tuple_format {
 	 * ephemeral spaces.
 	 */
 	uint32_t hash;
+	/**
+	 * Counter that grows incrementally on space rebuild
+	 * used for caching offset slot in key_part, for more
+	 * details see key_part::offset_slot_cache.
+	 */
+	uint64_t epoch;
 	/** Reference counter */
 	int refs;
 	/**