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

[PATCH 02/23] parse_commit_buffer(): treat lookup_tree() failure as parse error

From
Jeff King <peff@peff.net>
Date
Oct 18, 2019, 04:43 UTC
Message-ID
<20191018044328.GB17879@sigill.intra.peff.net>
In-Reply-To
<20191018044103.GA17625@sigill.intra.peff.net>

If parsing a commit yields a valid tree oid, but we've seen that same oid as a non-tree in the same process, the resulting commit struct will end up with a NULL tree pointer, but not otherwise report a parsing failure.

That's perhaps convenient for callers which want to operate on even partially corrupt commits (e.g., by still looking at the parents). But it leaves a potential trap for most callers, who now have to manually check for a NULL tree. Most do not, and it's likely that there are possible segfaults lurking. I say "possible" because there are many candidates, and I don't think it's worth following through on reproducing them when we can just fix them all in one spot. And certainly we _have_ seen real-world cases, such as the one fixed by 806278dead (commit-graph.c: handle corrupt/missing trees, 2019-09-05).

Note that we can't quite drop the check in the caller added by that commit yet, as there's some subtlety with repeated parsings (which will be addressed in a future commit).

Signed-off-by: Jeff King <peff@peff.net>
---
 commit.c | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)
diff --git a/commit.c b/commit.c
index 6467c9e175..810419a168 100644
--- a/commit.c
+++ b/commit.c
@@ -401,6 +401,7 @@ int parse_commit_buffer(struct repository *r, struct commit *item, const void *b
 	struct commit_graft *graft;
 	const int tree_entry_len = the_hash_algo->hexsz + 5;
 	const int parent_entry_len = the_hash_algo->hexsz + 7;
+	struct tree *tree;
 
 	if (item->object.parsed)
 		return 0;
@@ -412,7 +413,12 @@ int parse_commit_buffer(struct repository *r, struct commit *item, const void *b
 	if (get_oid_hex(bufptr + 5, &parent) < 0)
 		return error("bad tree pointer in commit %s",
 			     oid_to_hex(&item->object.oid));
-	set_commit_tree(item, lookup_tree(r, &parent));
+	tree = lookup_tree(r, &parent);
+	if (!tree)
+		return error("bad tree pointer %s in commit %s",
+			     oid_to_hex(&parent),
+			     oid_to_hex(&item->object.oid));
+	set_commit_tree(item, tree);
 	bufptr += tree_entry_len + 1; /* "tree " + "hex sha1" + "\n" */
 	pptr = &item->parents;
 
-- 
2.23.0.1228.gee29b05929
Previous: Jeff KingNext: Jonathan Tan
Message 34 of 39 in “parsing and fsck cleanups”
  1. 0/23 parsing and fsck cleanupsJeff King, Oct 18, 2019
  2. 08/23 fsck: require an actual buffer for non-blobsJeff King, Oct 18, 2019
  3. 04/23 remember commit/tag parse failuresJeff King, Oct 18, 2019
  4. Junio C HamanoOct 24, 2019
  5. Jonathan TanOct 24, 2019
  6. Jeff KingOct 24, 2019
  7. 05/23 fsck: stop checking commit->tree valueJeff King, Oct 18, 2019
  8. Junio C HamanoOct 24, 2019
  9. 06/23 fsck: stop checking commit->parent countsJeff King, Oct 18, 2019
  10. 09/23 fsck: unify object-name codeJeff King, Oct 18, 2019
  11. Junio C HamanoOct 24, 2019
  12. Jeff KingOct 24, 2019
  13. Junio C HamanoOct 25, 2019
  14. Jeff KingOct 25, 2019
  15. 10/23 fsck_describe_object(): build on our get_object_name() primitiveJeff King, Oct 18, 2019
  16. Junio C HamanoOct 24, 2019
  17. 11/23 fsck: use oids rather than objects for object_name APIJeff King, Oct 18, 2019
  18. 07/23 fsck: stop checking tag->taggedJeff King, Oct 18, 2019
  19. 12/23 fsck: don't require object structs for display functionsJeff King, Oct 18, 2019
  20. 13/23 fsck: only provide oid/type in fsck_error callbackJeff King, Oct 18, 2019
  21. 14/23 fsck: only require an oid for skiplist functionsJeff King, Oct 18, 2019
  22. 15/23 fsck: don't require an object struct for report()Jeff King, Oct 18, 2019
  23. 16/23 fsck: accept an oid instead of a "struct blob" for fsck_blob()Jeff King, Oct 18, 2019
  24. 17/23 fsck: drop blob struct from fsck_finish()Jeff King, Oct 18, 2019
  25. 18/23 fsck: don't require an object struct for fsck_ident()Jeff King, Oct 18, 2019
  26. 19/23 fsck: don't require an object struct in verify_headers()Jeff King, Oct 18, 2019
  27. 20/23 fsck: rename vague "oid" local variablesJeff King, Oct 18, 2019
  28. 21/23 fsck: accept an oid instead of a "struct tag" for fsck_tag()Jeff King, Oct 18, 2019
  29. 22/23 fsck: accept an oid instead of a "struct commit" for fsck_commit()Jeff King, Oct 18, 2019
  30. 23/23 fsck: accept an oid instead of a "struct tree" for fsck_tree()Jeff King, Oct 18, 2019
  31. 01/23 parse_commit_buffer(): treat lookup_commit() failure as parse errorJeff King, Oct 18, 2019
  32. Junio C HamanoOct 24, 2019
  33. Jeff KingOct 24, 2019
  34. 02/23 parse_commit_buffer(): treat lookup_tree() failure as parse errorJeff King, Oct 18, 2019
  35. Jonathan TanOct 24, 2019
  36. Jeff KingOct 24, 2019
  37. 03/23 parse_tag_buffer(): treat NULL tag pointer as parse errorJeff King, Oct 18, 2019
  38. Jonathan TanOct 24, 2019
  39. Junio C HamanoOct 25, 2019

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.