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

[PATCH] commit: detect commits that exist in commit-graph but not in the ODB

From
Patrick Steinhardt <ps@pks.im>
Date
Oct 13, 2023, 12:37 UTC
Message-ID
<b0bf576c51a706367a758b8e30eca37edb9c2734.1697200576.git.ps@pks.im>
In-Reply-To
<ZSkCGS3JPEQ71dOF@tanuki>

Commit graphs can become stale and contain references to commits that do not exist in the object database anymore. Theoretically, this can lead to a scenario where we are able to successfully look up any such commit via the commit graph even though such a lookup would fail if done via the object database directly.

As the commit graph is mostly intended as a sort of cache to speed up parsing of commits we do not want to have diverging behaviour in a repository with and a repository without commit graphs, no matter whether they are stale or not. As commits are otherwise immutable, the only thing that we really need to care about is thus the presence or absence of a commit.

To address potentially stale commit data that may exist in the graph, our `lookup_commit_in_graph()` function will check for the commit's existence in both the commit graph, but also in the object database. So even if we were able to look up the commit's data in the graph, we would still pretend as if the commit didn't exist if it is missing in the object database.

We don't have the same safety net in `parse_commit_in_graph_one()` though. This function is mostly used internally in "commit-graph.c" itself to validate the commit graph, and this usage is fine. We do expose its functionality via `parse_commit_in_graph()` though, which gets called by `repo_parse_commit_internal()`, and that function is in turn used in many places in our codebase.

For all I can see this function is never used to directly turn an object ID into a commit object without additional safety checks before or after this lookup. What it is being used for though is to walk history via the parent chain of commits. So when commits in the parent chain of a graph walk are missing it is possible that we wouldn't notice if that missing commit was part of the commit graph. Thus, a query like `git rev-parse HEAD~2` can succeed even if the intermittent commit is missing.

It's unclear whether there are additional ways in which such stale commit graphs can lead to problems. In any case, it feels like this is a bigger bug waiting to happen when we gain additional direct or indirect callers of `repo_parse_commit_internal()`. So let's fix the inconsistent behaviour by checking for object existence via the object database, as well.

Signed-off-by: Patrick Steinhardt <ps@pks.im>
---
 commit.c                |  7 ++++++-
 t/t5318-commit-graph.sh | 23 +++++++++++++++++++++++
 2 files changed, 29 insertions(+), 1 deletion(-)
diff --git a/commit.c b/commit.c
index b3223478bc..109e9217e3 100644
--- a/commit.c
+++ b/commit.c
@@ -572,8 +572,13 @@ int repo_parse_commit_internal(struct repository *r,
 		return -1;
 	if (item->object.parsed)
 		return 0;
-	if (use_commit_graph && parse_commit_in_graph(r, item))
+	if (use_commit_graph && parse_commit_in_graph(r, item)) {
+		if (!has_object(r, &item->object.oid, 0))
+			return quiet_on_missing ? -1 :
+				error(_("commit %s exists in commit-graph but not in the object database"),
+				      oid_to_hex(&item->object.oid));
 		return 0;
+	}
 
 	if (oid_object_info_extended(r, &item->object.oid, &oi, flags) < 0)
 		return quiet_on_missing ? -1 :
diff --git a/t/t5318-commit-graph.sh b/t/t5318-commit-graph.sh
index ba65f17dd9..25f8e9e2d3 100755
--- a/t/t5318-commit-graph.sh
+++ b/t/t5318-commit-graph.sh
@@ -821,4 +821,27 @@ test_expect_success 'overflow during generation version upgrade' '
 	)
 '
 
+test_expect_success 'commit exists in commit-graph but not in object database' '
+	test_when_finished "rm -rf repo" &&
+	git init repo &&
+	(
+		cd repo &&
+
+		test_commit A &&
+		test_commit B &&
+		test_commit C &&
+		git commit-graph write --reachable &&
+
+		# Corrupt the repository by deleting the intermittent commit
+		# object. Commands should notice that this object is absent and
+		# thus that the repository is corrupt even if the commit graph
+		# exists.
+		oid=$(git rev-parse B) &&
+		rm .git/objects/"$(test_oid_to_path "$oid")" &&
+
+		test_must_fail git rev-parse HEAD~2 2>error &&
+		grep "error: commit $oid exists in commit-graph but not in the object database" error
+	)
+'
+
 test_done
-- 
2.42.0
Previous: Patrick SteinhardtNext: Junio C Hamano
Message 16 of 61 in “rev-list: add support for commits in `--missing`”
  1. 0/3 rev-list: add support for commits in `--missing`Karthik Nayak, Oct 9, 2023
  2. 1/3 revision: rename bit to `do_not_die_on_missing_objects`Karthik Nayak, Oct 9, 2023
  3. 3/3 rev-list: add commit object support in `--missing` optionKarthik Nayak, Oct 9, 2023
  4. 2/3 rev-list: move `show_commit()` to the bottomKarthik Nayak, Oct 9, 2023
  5. Junio C HamanoOct 9, 2023
  6. Patrick SteinhardtOct 10, 2023
  7. Junio C HamanoOct 10, 2023
  8. Karthik NayakOct 11, 2023
  9. Junio C HamanoOct 11, 2023
  10. Karthik NayakOct 12, 2023
  11. Patrick SteinhardtOct 12, 2023
  12. Karthik NayakOct 12, 2023
  13. Junio C HamanoOct 12, 2023
  14. Patrick SteinhardtOct 13, 2023
  15. Patrick SteinhardtOct 13, 2023
  16. commit: detect commits that exist in commit-graph but not in the ODBPatrick Steinhardt, Oct 13, 2023
  17. Junio C HamanoOct 13, 2023
  18. Patrick SteinhardtOct 17, 2023
  19. Junio C HamanoOct 17, 2023
  20. Patrick SteinhardtOct 19, 2023
  21. Patrick SteinhardtOct 19, 2023
  22. Junio C HamanoOct 19, 2023
  23. Jeff KingOct 20, 2023
  24. Junio C HamanoOct 20, 2023
  25. Patrick SteinhardtOct 23, 2023
  26. Junio C HamanoOct 13, 2023
  27. Junio C HamanoOct 12, 2023
  28. 0/3 rev-list: add support for commits in `--missing`Karthik Nayak, Oct 16, 2023
  29. 1/3 revision: rename bit to `do_not_die_on_missing_objects`Karthik Nayak, Oct 16, 2023
  30. 2/3 rev-list: move `show_commit()` to the bottomKarthik Nayak, Oct 16, 2023
  31. 3/3 rev-list: add commit object support in `--missing` optionKarthik Nayak, Oct 16, 2023
  32. Junio C HamanoOct 16, 2023
  33. Karthik NayakOct 16, 2023
  34. Junio C HamanoOct 16, 2023
  35. 0/3 rev-list: add support for commits in `--missing`Karthik Nayak, Oct 19, 2023
  36. 1/3 revision: rename bit to `do_not_die_on_missing_objects`Karthik Nayak, Oct 19, 2023
  37. 2/3 rev-list: move `show_commit()` to the bottomKarthik Nayak, Oct 19, 2023
  38. 3/3 rev-list: add commit object support in `--missing` optionKarthik Nayak, Oct 19, 2023
  39. Junio C HamanoOct 19, 2023
  40. Junio C HamanoOct 19, 2023
  41. Karthik NayakOct 20, 2023
  42. Karthik NayakOct 20, 2023
  43. Junio C HamanoOct 20, 2023
  44. Junio C HamanoOct 20, 2023
  45. Karthik NayakOct 24, 2023
  46. 0/3 rev-list: add support for commits in `--missing`Karthik Nayak, Oct 24, 2023
  47. 1/3 revision: rename bit to `do_not_die_on_missing_objects`Karthik Nayak, Oct 24, 2023
  48. 2/3 rev-list: move `show_commit()` to the bottomKarthik Nayak, Oct 24, 2023
  49. 3/3 rev-list: add commit object support in `--missing` optionKarthik Nayak, Oct 24, 2023
  50. Junio C HamanoOct 24, 2023
  51. Junio C HamanoOct 25, 2023
  52. Karthik NayakOct 25, 2023
  53. Patrick SteinhardtOct 25, 2023
  54. Junio C HamanoOct 26, 2023
  55. 0/3 rev-list: add support for commits in `--missing`Karthik Nayak, Oct 26, 2023
  56. 1/3 revision: rename bit to `do_not_die_on_missing_objects`Karthik Nayak, Oct 26, 2023
  57. 2/3 rev-list: move `show_commit()` to the bottomKarthik Nayak, Oct 26, 2023
  58. 3/3 rev-list: add commit object support in `--missing` optionKarthik Nayak, Oct 26, 2023
  59. Patrick SteinhardtOct 27, 2023
  60. Karthik NayakOct 27, 2023
  61. 3/3 rev-list: add commit object support in `--missing` optionKarthik Nayak, Oct 27, 2023

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.