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

[PATCH v2 24/25] repack: pack objects mentioned by the index

From
Jeff King <peff@peff.net>
Date
Oct 15, 2014, 22:46 UTC
Message-ID
<20141015224632.GX25630@peff.net>
In-Reply-To
<20141015223244.GA25368@peff.net>

When we pack all objects, we use only the objects reachable from references and reflogs. This misses any objects which are reachable from the index, but not yet referenced.

By itself this isn't a big deal; the objects can remain loose until they are actually used in a commit. However, it does create a problem when we drop packed but unreachable objects. We try to optimize out the writing of objects that we will immediately prune, which means we must follow the same rules as prune in determining what is reachable. And prune uses the index for this purpose.

This is rather uncommon in practice, as objects in the index would not usually have been packed in the first place. But it could happen in a sequence like:

  1. You make a commit on a branch that references blob X.
  2. You repack, moving X into the pack.
  3. You delete the branch (and its reflog), so that X is
     unreferenced.
  4. You "git add" blob X so that it is now referenced only
     by the index.
  5. You repack again with git-gc. The pack-objects we
     invoke will see that X is neither referenced nor
     recent and not bother loosening it.
Signed-off-by: Jeff King <peff@peff.net>
---
In addition to fixing the safety, this is actually _packing_ those index
objects. I don't see anything wrong with that, but if would prefer not
to, it should be possible to do a separate traversal over the index
objects (and mark them as "to keep", but not "to pack").
 builtin/pack-objects.c               |  8 ++++++++
 builtin/repack.c                     |  1 +
 t/t7701-repack-unpack-unreachable.sh | 13 +++++++++++++
 3 files changed, 22 insertions(+)
diff --git a/builtin/pack-objects.c b/builtin/pack-objects.c
index b26276b..3dc9108 100644
--- a/builtin/pack-objects.c
+++ b/builtin/pack-objects.c
@@ -2617,6 +2617,7 @@ int cmd_pack_objects(int argc, const char **argv, const char *prefix)
 	int all_progress_implied = 0;
 	struct argv_array rp = ARGV_ARRAY_INIT;
 	int rev_list_unpacked = 0, rev_list_all = 0, rev_list_reflog = 0;
+	int rev_list_index = 0;
 	struct option pack_objects_options[] = {
 		OPT_SET_INT('q', "quiet", &progress,
 			    N_("do not show progress meter"), 0),
@@ -2663,6 +2664,9 @@ int cmd_pack_objects(int argc, const char **argv, const char *prefix)
 		{ OPTION_SET_INT, 0, "reflog", &rev_list_reflog, NULL,
 		  N_("include objects referred by reflog entries"),
 		  PARSE_OPT_NOARG | PARSE_OPT_NONEG, NULL, 1 },
+		{ OPTION_SET_INT, 0, "index-objects", &rev_list_index, NULL,
+		  N_("include objects referred to by the index"),
+		  PARSE_OPT_NOARG | PARSE_OPT_NONEG, NULL, 1 },
 		OPT_BOOL(0, "stdout", &pack_to_stdout,
 			 N_("output pack to stdout")),
 		OPT_BOOL(0, "include-tag", &include_tag,
@@ -2720,6 +2724,10 @@ int cmd_pack_objects(int argc, const char **argv, const char *prefix)
 		use_internal_rev_list = 1;
 		argv_array_push(&rp, "--reflog");
 	}
+	if (rev_list_index) {
+		use_internal_rev_list = 1;
+		argv_array_push(&rp, "--index-objects");
+	}
 	if (rev_list_unpacked) {
 		use_internal_rev_list = 1;
 		argv_array_push(&rp, "--unpacked");
diff --git a/builtin/repack.c b/builtin/repack.c
index 2aae05d..80b6021 100644
--- a/builtin/repack.c
+++ b/builtin/repack.c
@@ -209,6 +209,7 @@ int cmd_repack(int argc, const char **argv, const char *prefix)
 	argv_array_push(&cmd_args, "--non-empty");
 	argv_array_push(&cmd_args, "--all");
 	argv_array_push(&cmd_args, "--reflog");
+	argv_array_push(&cmd_args, "--index-objects");
 	if (window)
 		argv_array_pushf(&cmd_args, "--window=%s", window);
 	if (window_memory)
diff --git a/t/t7701-repack-unpack-unreachable.sh b/t/t7701-repack-unpack-unreachable.sh
index b8d4cde..aad8a9c 100755
--- a/t/t7701-repack-unpack-unreachable.sh
+++ b/t/t7701-repack-unpack-unreachable.sh
@@ -109,4 +109,17 @@ test_expect_success 'do not bother loosening old objects' '
 	test_must_fail git cat-file -p $obj2
 '
 
+test_expect_success 'keep packed objects found only in index' '
+	echo my-unique-content >file &&
+	git add file &&
+	git commit -m "make it reachable" &&
+	git gc &&
+	git reset HEAD^ &&
+	git reflog expire --expire=now --all &&
+	git add file &&
+	test-chmtime =-86400 .git/objects/pack/* &&
+	git gc --prune=1.hour.ago &&
+	git cat-file blob :file
+'
+
 test_done

-- 
2.1.2.596.g7379948
Previous: Jeff KingNext: Jeff King
Message 39 of 48 in “prune-safety”
  1. 0/25 prune-safetyJeff King, Oct 15, 2014
  2. 01/25 foreach_alt_odb: propagate return value from callbackJeff King, Oct 15, 2014
  3. 02/25 isxdigit: cast input to unsigned charJeff King, Oct 15, 2014
  4. Junio C HamanoOct 16, 2014
  5. 03/25 object_array: factor out slopbuf-freeing logicJeff King, Oct 15, 2014
  6. Junio C HamanoOct 16, 2014
  7. git-bundle rev handling and de-dupingJeff King, Oct 17, 2014
  8. Philip OakleyOct 17, 2014
  9. Junio C HamanoOct 17, 2014
  10. 04/25 object_array: add a "clear" functionJeff King, Oct 15, 2014
  11. 05/25 clean up name allocation in prepare_revision_walkJeff King, Oct 15, 2014
  12. 06/25 reachable: use traverse_commit_list instead of custom walkJeff King, Oct 15, 2014
  13. Junio C HamanoOct 16, 2014
  14. 07/25 reachable: reuse revision.c "add all reflogs" codeJeff King, Oct 15, 2014
  15. 08/25 prune: factor out loose-object directory traversalJeff King, Oct 15, 2014
  16. 09/25 reachable: mark index blobs as SEENJeff King, Oct 15, 2014
  17. 10/25 prune-packed: use for_each_loose_file_in_objdirJeff King, Oct 15, 2014
  18. 11/25 count-objects: do not use xsize_t when counting object sizeJeff King, Oct 15, 2014
  19. 12/25 count-objects: use for_each_loose_file_in_objdirJeff King, Oct 15, 2014
  20. 13/25 sha1_file: add for_each iterators for loose and packed objectsJeff King, Oct 15, 2014
  21. 14/25 prune: keep objects reachable from recent objectsJeff King, Oct 15, 2014
  22. 15/25 pack-objects: refactor unpack-unreachable expiration checkJeff King, Oct 15, 2014
  23. 16/25 pack-objects: match prune logic for discarding objectsJeff King, Oct 15, 2014
  24. 17/25 write_sha1_file: freshen existing objectsJeff King, Oct 15, 2014
  25. 18/25 make add_object_array_with_context interface more saneJeff King, Oct 15, 2014
  26. 19/25 traverse_commit_list: support pending blobs/trees with pathsJeff King, Oct 15, 2014
  27. 20/25 rev-list: document --reflog optionJeff King, Oct 15, 2014
  28. 21/25 rev-list: add --index-objects optionJeff King, Oct 15, 2014
  29. Junio C HamanoOct 16, 2014
  30. Jeff KingOct 17, 2014
  31. Jeff KingOct 17, 2014
  32. 22/26 rev-list: add --indexed-objects optionJeff King, Oct 17, 2014
  33. 23/26 reachable: use revision machinery's --indexed-objects codeJeff King, Oct 17, 2014
  34. 24/26 pack-objects: use argv_arrayJeff King, Oct 17, 2014
  35. 25/26 repack: pack objects mentioned by the indexJeff King, Oct 17, 2014
  36. 26/26 pack-objects: double-check options before discarding objectsJeff King, Oct 17, 2014
  37. 22/25 reachable: use revision machinery's --index-objects codeJeff King, Oct 15, 2014
  38. 23/25 pack-objects: use argv_arrayJeff King, Oct 15, 2014
  39. 24/25 repack: pack objects mentioned by the indexJeff King, Oct 15, 2014
  40. 25/25 pack-objects: double-check options before discarding objectsJeff King, Oct 15, 2014
  41. Junio C HamanoOct 16, 2014
  42. Junio C HamanoOct 16, 2014
  43. Jeff KingOct 16, 2014
  44. Jeff KingOct 16, 2014
  45. Junio C HamanoOct 16, 2014
  46. Jeff KingOct 17, 2014
  47. Jeff KingOct 18, 2014
  48. Jeff KingOct 17, 2014

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.