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

[PATCH 10/15] midx: use chunk-format API in write_midx_internal()

From
Derrick Stolee via GitGitGadget <gitgitgadget@gmail.com>
Date
Dec 3, 2020, 16:16 UTC
Message-ID
<f2f78ee1054f64fe767335c4b0c77ba2fbec5fae.1607012215.git.gitgitgadget@gmail.com>
In-Reply-To
<pull.804.git.1607012215.gitgitgadget@gmail.com>
From: Derrick Stolee <dstolee@microsoft.com>

The chunk-format API allows automatically writing the table of contents for a chunk-based file format when using an array of "struct chunk_info"s. Update write_midx_internal() to use this strategy, which also simplifies the chunk writing loop. This loop will be replaced with a chunk-format API call in an upcoming change.

Signed-off-by: Derrick Stolee <dstolee@microsoft.com>
---
 midx.c | 96 +++++++++++++---------------------------------------------
 1 file changed, 21 insertions(+), 75 deletions(-)
diff --git a/midx.c b/midx.c
index ce6d4339bd..0548266bea 100644
--- a/midx.c
+++ b/midx.c
@@ -11,6 +11,7 @@
 #include "trace2.h"
 #include "run-command.h"
 #include "repository.h"
+#include "chunk-format.h"
 
 #define MIDX_SIGNATURE 0x4d494458 /* "MIDX" */
 #define MIDX_VERSION 1
@@ -799,15 +800,14 @@ static int write_midx_large_offsets(struct hashfile *f,
 static int write_midx_internal(const char *object_dir, struct multi_pack_index *m,
 			       struct string_list *packs_to_drop, unsigned flags)
 {
-	unsigned char cur_chunk, num_chunks = 0;
+	unsigned char num_chunks = 0;
 	char *midx_name;
 	uint32_t i;
 	struct hashfile *f = NULL;
 	struct lock_file lk;
 	struct write_midx_context ctx = { 0 };
 	uint64_t header_size = 0;
-	uint32_t chunk_ids[MIDX_MAX_CHUNKS + 1];
-	uint64_t chunk_offsets[MIDX_MAX_CHUNKS + 1];
+	struct chunk_info chunks[MIDX_MAX_CHUNKS];
 	int pack_name_concat_len = 0;
 	int dropped_packs = 0;
 	int result = 0;
@@ -923,7 +923,6 @@ static int write_midx_internal(const char *object_dir, struct multi_pack_index *
 	if (ctx.m)
 		close_midx(ctx.m);
 
-	cur_chunk = 0;
 	num_chunks = ctx.large_offsets_needed ? 5 : 4;
 
 	if (ctx.nr - dropped_packs == 0) {
@@ -934,85 +933,32 @@ static int write_midx_internal(const char *object_dir, struct multi_pack_index *
 
 	header_size = write_midx_header(f, num_chunks, ctx.nr - dropped_packs);
 
-	chunk_ids[cur_chunk] = MIDX_CHUNKID_PACKNAMES;
-	chunk_offsets[cur_chunk] = header_size + (num_chunks + 1) * MIDX_CHUNKLOOKUP_WIDTH;
+	chunks[0].id = MIDX_CHUNKID_PACKNAMES;
+	chunks[0].size = pack_name_concat_len;
+	chunks[0].write_fn = write_midx_pack_names;
 
-	cur_chunk++;
-	chunk_ids[cur_chunk] = MIDX_CHUNKID_OIDFANOUT;
-	chunk_offsets[cur_chunk] = chunk_offsets[cur_chunk - 1] + pack_name_concat_len;
+	chunks[1].id = MIDX_CHUNKID_OIDFANOUT;
+	chunks[1].size = MIDX_CHUNK_FANOUT_SIZE;
+	chunks[1].write_fn = write_midx_oid_fanout;
 
-	cur_chunk++;
-	chunk_ids[cur_chunk] = MIDX_CHUNKID_OIDLOOKUP;
-	chunk_offsets[cur_chunk] = chunk_offsets[cur_chunk - 1] + MIDX_CHUNK_FANOUT_SIZE;
+	chunks[2].id = MIDX_CHUNKID_OIDLOOKUP;
+	chunks[2].size = ctx.entries_nr * the_hash_algo->rawsz;
+	chunks[2].write_fn = write_midx_oid_lookup;
 
-	cur_chunk++;
-	chunk_ids[cur_chunk] = MIDX_CHUNKID_OBJECTOFFSETS;
-	chunk_offsets[cur_chunk] = chunk_offsets[cur_chunk - 1] + ctx.entries_nr * the_hash_algo->rawsz;
+	chunks[3].id = MIDX_CHUNKID_OBJECTOFFSETS;
+	chunks[3].size = ctx.entries_nr * MIDX_CHUNK_OFFSET_WIDTH;
+	chunks[3].write_fn = write_midx_object_offsets;
 
-	cur_chunk++;
-	chunk_offsets[cur_chunk] = chunk_offsets[cur_chunk - 1] + ctx.entries_nr * MIDX_CHUNK_OFFSET_WIDTH;
 	if (ctx.large_offsets_needed) {
-		chunk_ids[cur_chunk] = MIDX_CHUNKID_LARGEOFFSETS;
-
-		cur_chunk++;
-		chunk_offsets[cur_chunk] = chunk_offsets[cur_chunk - 1] +
-					   ctx.num_large_offsets * MIDX_CHUNK_LARGE_OFFSET_WIDTH;
+		chunks[4].id = MIDX_CHUNKID_LARGEOFFSETS;
+		chunks[4].size = ctx.num_large_offsets * MIDX_CHUNK_LARGE_OFFSET_WIDTH;
+		chunks[4].write_fn = write_midx_large_offsets;
 	}
 
-	chunk_ids[cur_chunk] = 0;
-
-	for (i = 0; i <= num_chunks; i++) {
-		if (i && chunk_offsets[i] < chunk_offsets[i - 1])
-			BUG("incorrect chunk offsets: %"PRIu64" before %"PRIu64,
-			    chunk_offsets[i - 1],
-			    chunk_offsets[i]);
-
-		if (chunk_offsets[i] % MIDX_CHUNK_ALIGNMENT)
-			BUG("chunk offset %"PRIu64" is not properly aligned",
-			    chunk_offsets[i]);
-
-		hashwrite_be32(f, chunk_ids[i]);
-		hashwrite_be64(f, chunk_offsets[i]);
-	}
-
-	for (i = 0; i < num_chunks; i++) {
-		if (f->total + f->offset != chunk_offsets[i])
-			BUG("incorrect chunk offset (%"PRIu64" != %"PRIu64") for chunk id %"PRIx32,
-			    chunk_offsets[i],
-			    f->total + f->offset,
-			    chunk_ids[i]);
-
-		switch (chunk_ids[i]) {
-			case MIDX_CHUNKID_PACKNAMES:
-				write_midx_pack_names(f, &ctx);
-				break;
-
-			case MIDX_CHUNKID_OIDFANOUT:
-				write_midx_oid_fanout(f, &ctx);
-				break;
-
-			case MIDX_CHUNKID_OIDLOOKUP:
-				write_midx_oid_lookup(f, &ctx);
-				break;
-
-			case MIDX_CHUNKID_OBJECTOFFSETS:
-				write_midx_object_offsets(f, &ctx);
-				break;
-
-			case MIDX_CHUNKID_LARGEOFFSETS:
-				write_midx_large_offsets(f, &ctx);
-				break;
-
-			default:
-				BUG("trying to write unknown chunk id %"PRIx32,
-				    chunk_ids[i]);
-		}
-	}
+	write_table_of_contents(f, header_size, chunks, num_chunks);
 
-	if (f->total + f->offset != chunk_offsets[num_chunks])
-		BUG("incorrect final offset %"PRIu64" != %"PRIu64,
-		    f->total + f->offset,
-		    chunk_offsets[num_chunks]);
+	for (i = 0; i < num_chunks; i++)
+		chunks[i].write_fn(f, &ctx);
 
 	finalize_hashfile(f, NULL, CSUM_FSYNC | CSUM_HASH_IN_STREAM);
 	commit_lock_file(&lk);
-- 
gitgitgadget
Previous: Derrick StoleeNext: Taylor Blau
Message 31 of 40 in “Refactor chunk-format into an API”
  1. 00/15 Refactor chunk-format into an APIDerrick Stolee via GitGitGadget, Dec 3, 2020
  2. 02/15 chunk-format: add API for writing table of contentsDerrick Stolee via GitGitGadget, Dec 3, 2020
  3. Taylor BlauDec 8, 2020
  4. 01/15 commit-graph: anonymize data in chunk_write_fnDerrick Stolee via GitGitGadget, Dec 3, 2020
  5. 04/15 midx: use context in write_midx_pack_names()Derrick Stolee via GitGitGadget, Dec 3, 2020
  6. 03/15 midx: rename pack_info to write_midx_contextDerrick Stolee via GitGitGadget, Dec 3, 2020
  7. 05/15 midx: add entries to write_midx_contextDerrick Stolee via GitGitGadget, Dec 3, 2020
  8. Junio C HamanoDec 3, 2020
  9. Derrick StoleeDec 4, 2020
  10. Taylor BlauDec 8, 2020
  11. 07/15 midx: add num_large_offsets to write_midx_contextDerrick Stolee via GitGitGadget, Dec 3, 2020
  12. 06/15 midx: add pack_perm to write_midx_contextDerrick Stolee via GitGitGadget, Dec 3, 2020
  13. 09/15 midx: drop chunk progress during writeDerrick Stolee via GitGitGadget, Dec 3, 2020
  14. 08/15 midx: convert chunk write methods to return intDerrick Stolee via GitGitGadget, Dec 3, 2020
  15. Junio C HamanoDec 3, 2020
  16. Derrick StoleeDec 4, 2020
  17. 11/15 midx: use 64-bit multiplication for chunk sizesDerrick Stolee via GitGitGadget, Dec 3, 2020
  18. Junio C HamanoDec 3, 2020
  19. Taylor BlauDec 8, 2020
  20. 12/15 chunk-format: create write_chunks()Derrick Stolee via GitGitGadget, Dec 3, 2020
  21. Taylor BlauDec 8, 2020
  22. 13/15 chunk-format: create chunk reading APIDerrick Stolee via GitGitGadget, Dec 3, 2020
  23. Junio C HamanoDec 3, 2020
  24. Derrick StoleeDec 4, 2020
  25. Junio C HamanoDec 4, 2020
  26. Junio C HamanoDec 3, 2020
  27. Derrick StoleeDec 4, 2020
  28. 15/15 chunk-format: add technical docsDerrick Stolee via GitGitGadget, Dec 3, 2020
  29. 14/15 commit-graph: restore duplicate chunk checksDerrick Stolee via GitGitGadget, Dec 3, 2020
  30. Derrick StoleeDec 7, 2020
  31. 10/15 midx: use chunk-format API in write_midx_internal()Derrick Stolee via GitGitGadget, Dec 3, 2020
  32. Taylor BlauDec 8, 2020
  33. Derrick StoleeDec 10, 2020
  34. René ScharfeDec 4, 2020
  35. Derrick StoleeDec 4, 2020
  36. Junio C HamanoDec 4, 2020
  37. Taylor BlauDec 8, 2020
  38. René ScharfeDec 9, 2020
  39. Taylor BlauDec 10, 2020
  40. Derrick StoleeDec 10, 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.