git/list[1] front-page[2] threads[3] people[4] search[5] about
 

[PATCH 27/34] commit-graph: load modified path Bloom filters for merge commits

From
SZEDER Gábor <szeder.dev@gmail.com>
Date
May 29, 2020, 08:50 UTC
Message-ID
<20200529085038.26008-28-szeder.dev@gmail.com>
In-Reply-To
<20200529085038.26008-1-szeder.dev@gmail.com>
---
 commit-graph.c | 53 +++++++++++++++++++++++++++++++++++++++++++++-----
 commit-graph.h |  2 ++
 2 files changed, 50 insertions(+), 5 deletions(-)
diff --git a/commit-graph.c b/commit-graph.c
index 178bbf0113..32738ba4b8 100644
--- a/commit-graph.c
+++ b/commit-graph.c
@@ -26,6 +26,7 @@
 #define GRAPH_CHUNKID_EXTRAEDGES 0x45444745 /* "EDGE" */
 #define GRAPH_CHUNKID_BASE 0x42415345 /* "BASE" */
 #define GRAPH_CHUNKID_MODIFIED_PATH_BLOOM_FILTER_INDEX 0x4d504249 /* "MPBI" */
+#define GRAPH_CHUNKID_MODIFIED_PATH_BLOOM_FILTER_MERGE_INDEX 0x4d50424d /* "MPBM" */
 #define GRAPH_CHUNKID_MODIFIED_PATH_BLOOM_FILTERS 0x4d504246 /* "MPBF" */
 #define GRAPH_CHUNKID_MODIFIED_PATH_BLOOM_FILTER_EXCLUDES 0x4d504258 /* "MPBX" */
 
@@ -328,6 +329,15 @@ struct commit_graph *parse_commit_graph(void *graph_map, int fd,
 			}
 			break;
 
+		case GRAPH_CHUNKID_MODIFIED_PATH_BLOOM_FILTER_MERGE_INDEX:
+			if (graph->chunk_mpbf_merge_index)
+				chunk_repeated = 1;
+			else {
+				graph->chunk_mpbf_merge_index = data + chunk_offset;
+				graph->chunk_mpbf_merge_index_size = next_chunk_offset - chunk_offset;
+			}
+			break;
+
 		case GRAPH_CHUNKID_MODIFIED_PATH_BLOOM_FILTERS:
 			if (graph->chunk_mpbf_filters)
 				chunk_repeated = 1;
@@ -869,11 +879,44 @@ static int load_modified_path_bloom_filter_from_graph(
 		bf->bits = (uint8_t*) bloom_index;
 		return 1;
 	} else if (bloom_index[0] & (1 << 6)) {
-		/*
-		 * Modified path Bloom filters for second..nth parents of
-		 * merge commits are not implemented yet.
-		 */
-		return 0;
+		struct commit_list *p;
+		uint32_t pos;
+		int found = 0;
+
+		pos = get_be32(bloom_index + sizeof(uint32_t));
+
+		if (!graph->chunk_mpbf_merge_index)
+			BUG("commit %s refers to position %u in the Modified Path Bloom Filter Merge Index chunk, but that chunk is missing",
+			    oid_to_hex(&commit->object.oid), pos);
+
+		for (p = commit->parents; p; p = p->next, pos++)
+			if (p->item == parent) {
+				found = 1;
+				break;
+			}
+		if (!found)
+			BUG("commit %s has no parent %s\n",
+			    oid_to_hex(&commit->object.oid),
+			    oid_to_hex(&parent->object.oid));
+
+		if (pos * sizeof(uint64_t) > graph->chunk_mpbf_merge_index_size)
+			BUG("commit %s and parent %s refer to position %u in the Modified Path Bloom Filter Merge Index chunk, but that's too large for a chunk of size %lu bytes",
+			    oid_to_hex(&commit->object.oid),
+			    oid_to_hex(&parent->object.oid), pos,
+			    graph->chunk_mpbf_merge_index_size);
+		bloom_index = graph->chunk_mpbf_merge_index + sizeof(uint64_t) * pos;
+		if (bloom_index[0] & (1 << 7)) {
+			uint64_t v;
+			memcpy(&v, bloom_index, sizeof(v));
+			if (v == GRAPH_MODIFIED_PATH_BLOOM_FILTER_NONE)
+				return 0;
+
+			/* embedded modified path Bloom filter */
+			bf->nr_bits = GRAPH_MODIFIED_PATH_BLOOM_FILTER_EMBEDDED_NR_BITS;
+			bf->bits = (uint8_t*) bloom_index;
+			return 1;
+		} else
+			offset = get_be64(bloom_index);
 	} else {
 		if (!first_parent)
 			return 0;
diff --git a/commit-graph.h b/commit-graph.h
index cde0d7fa30..b2605a0e3b 100644
--- a/commit-graph.h
+++ b/commit-graph.h
@@ -69,6 +69,8 @@ struct commit_graph {
 	const unsigned char *chunk_base_graphs;
 	const unsigned char *chunk_mpbf_index;
 	uint64_t chunk_mpbf_index_size;
+	const unsigned char *chunk_mpbf_merge_index;
+	uint64_t chunk_mpbf_merge_index_size;
 	const unsigned char *chunk_mpbf_filters;
 	uint64_t chunk_mpbf_filters_size;
 	const unsigned char *chunk_mpbf_excludes;
-- 
2.27.0.rc1.431.g5c813f95dc
Previous: SZEDER GáborNext: SZEDER Gábor
Message 27 of 43 in “An alternative modified path Bloom filters implementation”
  1. 00/34 An alternative modified path Bloom filters implementationSZEDER Gábor, May 29, 2020
  2. 01/34 tree-walk.c: don't match submodule entries for 'submod/anything'SZEDER Gábor, May 29, 2020
  3. 02/34 commit-graph: fix parsing the Chunk Lookup tableSZEDER Gábor, May 29, 2020
  4. 03/34 commit-graph-format.txt: all multi-byte numbers are in network byte orderSZEDER Gábor, May 29, 2020
  5. 04/34 commit-slab: add a function to deep free entries on the slabSZEDER Gábor, May 29, 2020
  6. Derrick StoleeJun 4, 2020
  7. SZEDER GáborJun 27, 2020
  8. Derrick StoleeJun 29, 2020
  9. 05/34 diff.h: drop diff_tree_oid() & friends' return valueSZEDER Gábor, May 29, 2020
  10. 07/34 commit-graph: simplify parse_commit_graph() #1SZEDER Gábor, May 29, 2020
  11. 06/34 commit-graph: clean up #includesSZEDER Gábor, May 29, 2020
  12. 09/34 commit-graph: simplify write_commit_graph_file() #1SZEDER Gábor, May 29, 2020
  13. 10/34 commit-graph: simplify write_commit_graph_file() #2SZEDER Gábor, May 29, 2020
  14. 08/34 commit-graph: simplify parse_commit_graph() #2SZEDER Gábor, May 29, 2020
  15. 11/34 commit-graph: allocate the 'struct chunk_info' array dinamicallySZEDER Gábor, May 29, 2020
  16. 12/34 commit-graph: unify the signatures of all write_graph_chunk_*() functionsSZEDER Gábor, May 29, 2020
  17. 13/34 commit-graph: simplify write_commit_graph_file() #3SZEDER Gábor, May 29, 2020
  18. 14/34 commit-graph: check chunk sizes after writingSZEDER Gábor, May 29, 2020
  19. 16/34 Add a generic and minimal Bloom filter implementationSZEDER Gábor, May 29, 2020
  20. 17/34 Import a streaming-capable Murmur3 hash function implementationSZEDER Gábor, May 29, 2020
  21. 15/34 commit-graph-format.txt: document the modified path Bloom filter chunksSZEDER Gábor, May 29, 2020
  22. SZEDER GáborJun 2, 2020
  23. 18/34 commit-graph: write "empty" Modified Path Bloom Filter Index chunkSZEDER Gábor, May 29, 2020
  24. 19/34 commit-graph: add commit slab for modified path Bloom filtersSZEDER Gábor, May 29, 2020
  25. 20/34 commit-graph: fill the Modified Path Bloom Filter Index chunkSZEDER Gábor, May 29, 2020
  26. 21/34 commit-graph: load and use the Modified Path Bloom Filter Index chunkSZEDER Gábor, May 29, 2020
  27. 27/34 commit-graph: load modified path Bloom filters for merge commitsSZEDER Gábor, May 29, 2020
  28. 28/34 commit-graph: write Modified Path Bloom Filter Merge Index chunkSZEDER Gábor, May 29, 2020
  29. 30/34 commit-graph: move write_commit_graph_reachable below write_commit_graphSZEDER Gábor, May 29, 2020
  30. 29/34 commit-graph: extract init and free write_commit_graph_contextSZEDER Gábor, May 29, 2020
  31. 31/34 t7007-show: make the first test compatible with the next patchSZEDER Gábor, May 29, 2020
  32. 26/34 commit-graph: deduplicate modified path Bloom filtersSZEDER Gábor, May 29, 2020
  33. 25/34 commit-graph: check embedded modified path Bloom filters with a maskSZEDER Gábor, May 29, 2020
  34. 24/34 commit-graph: check all leading directories in modified path Bloom filtersSZEDER Gábor, May 29, 2020
  35. 23/34 commit-graph: load and use the Modified Path Bloom Filters chunkSZEDER Gábor, May 29, 2020
  36. 22/34 commit-graph: write the Modified Path Bloom Filters chunkSZEDER Gábor, May 29, 2020
  37. 34/34 commit-graph: use modified path Bloom filters with wildcards, if possibleSZEDER Gábor, May 29, 2020
  38. 33/34 commit-graph: write modified path Bloom filters in "history order"SZEDER Gábor, May 29, 2020
  39. 32/34 PoC commit-graph: use revision walk machinery for '--reachable'SZEDER Gábor, May 29, 2020
  40. Derrick StoleeMay 29, 2020
  41. Johannes SchindelinMay 28, 2020
  42. Taylor BlauJun 1, 2020
  43. Junio C HamanoJun 2, 2020

Read the whole thread, see it on lore, or plain text.

$ cat FOOTERMessages come from the public archive at lore.kernel.org/git, fetched every hour. The front page is chosen and written each morning by an AI editor and can be wrong; the threads themselves are the record. About and API. For agents: an MCP server at https://gitlist.dev/mcp, and any thread, story or person page as Markdown by adding .md to its URL (or sending Accept: text/markdown). Details in /llms.txt.