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

[PATCH v2 07/11] ls-refs: ignore very long ref-prefix counts

From
Jeff King <peff@peff.net>
Date
Sep 14, 2021, 23:51 UTC
Message-ID
<YUE1hz/KKM0XebCP@coredump.intra.peff.net>
In-Reply-To
<YUE1alo58cGyTw6/@coredump.intra.peff.net>

Because each "ref-prefix" capability from the client comes in its own pkt-line, there's no limit to the number of them that a misbehaving client may send. We read them all into a strvec, which means the client can waste arbitrary amounts of our memory by just sending us "ref-prefix foo" over and over.

One possible solution is to just drop the connection when the limit is reached. If we set it high enough, then only misbehaving or malicious clients would hit it. But "high enough" is vague, and it's unfriendly if we guess wrong and a legitimate client hits this.

But we can do better. Since supporting the ref-prefix capability is optional anyway, the client has to further cull the response based on their own patterns. So we can simply ignore the patterns once we cross a certain threshold. Note that we have to ignore _all_ patterns, not just the ones past our limit (since otherwise we'd send too little data).

The limit here is fairly arbitrary, and probably much higher than anyone would need in practice. It might be worth limiting it further, if only because we check it linearly (so with "m" local refs and "n" patterns, we do "m * n" string comparisons). But if we care about optimizing this, an even better solution may be a more advanced data structure anyway.

I didn't bother making the limit configurable, since it's so high and since Git should behave correctly in either case. It wouldn't be too hard to do, but it makes both the code and documentation more complex.

Signed-off-by: Jeff King <peff@peff.net>
---
 ls-refs.c            | 20 ++++++++++++++++++--
 t/t5701-git-serve.sh | 31 +++++++++++++++++++++++++++++++
 2 files changed, 49 insertions(+), 2 deletions(-)
diff --git a/ls-refs.c b/ls-refs.c
index a1a0250607..18c4f41e87 100644
--- a/ls-refs.c
+++ b/ls-refs.c
@@ -40,6 +40,12 @@ static void ensure_config_read(void)
 	config_read = 1;
 }
 
+/*
+ * If we see this many or more "ref-prefix" lines from the client, we consider
+ * it "too many" and will avoid using the prefix feature entirely.
+ */
+#define TOO_MANY_PREFIXES 65536
+
 /*
  * Check if one of the prefixes is a prefix of the ref.
  * If no prefixes were provided, all refs match.
@@ -156,15 +162,25 @@ int ls_refs(struct repository *r, struct packet_reader *request)
 			data.peel = 1;
 		else if (!strcmp("symrefs", arg))
 			data.symrefs = 1;
-		else if (skip_prefix(arg, "ref-prefix ", &out))
-			strvec_push(&data.prefixes, out);
+		else if (skip_prefix(arg, "ref-prefix ", &out)) {
+			if (data.prefixes.nr < TOO_MANY_PREFIXES)
+				strvec_push(&data.prefixes, out);
+		}
 		else if (!strcmp("unborn", arg))
 			data.unborn = allow_unborn;
 	}
 
 	if (request->status != PACKET_READ_FLUSH)
 		die(_("expected flush after ls-refs arguments"));
 
+	/*
+	 * If we saw too many prefixes, we must avoid using them at all; as
+	 * soon as we have any prefix, they are meant to form a comprehensive
+	 * list.
+	 */
+	if (data.prefixes.nr >= TOO_MANY_PREFIXES)
+		strvec_clear(&data.prefixes);
+
 	send_possibly_unborn_head(&data);
 	if (!data.prefixes.nr)
 		strvec_push(&data.prefixes, "");
diff --git a/t/t5701-git-serve.sh b/t/t5701-git-serve.sh
index 930721f053..3bc96ebcde 100755
--- a/t/t5701-git-serve.sh
+++ b/t/t5701-git-serve.sh
@@ -158,6 +158,37 @@ test_expect_success 'refs/heads prefix' '
 	test_cmp expect actual
 '
 
+test_expect_success 'ignore very large set of prefixes' '
+	# generate a large number of ref-prefixes that we expect
+	# to match nothing; the value here exceeds TOO_MANY_PREFIXES
+	# from ls-refs.c.
+	{
+		echo command=ls-refs &&
+		echo object-format=$(test_oid algo)
+		echo 0001 &&
+		perl -le "print \"ref-prefix refs/heads/\$_\" for (1..65536)" &&
+		echo 0000
+	} |
+	test-tool pkt-line pack >in &&
+
+	# and then confirm that we see unmatched prefixes anyway (i.e.,
+	# that the prefix was not applied).
+	cat >expect <<-EOF &&
+	$(git rev-parse HEAD) HEAD
+	$(git rev-parse refs/heads/dev) refs/heads/dev
+	$(git rev-parse refs/heads/main) refs/heads/main
+	$(git rev-parse refs/heads/release) refs/heads/release
+	$(git rev-parse refs/tags/annotated-tag) refs/tags/annotated-tag
+	$(git rev-parse refs/tags/one) refs/tags/one
+	$(git rev-parse refs/tags/two) refs/tags/two
+	0000
+	EOF
+
+	test-tool serve-v2 --stateless-rpc <in >out &&
+	test-tool pkt-line unpack <out >actual &&
+	test_cmp expect actual
+'
+
 test_expect_success 'peel parameter' '
 	test-tool pkt-line pack >in <<-EOF &&
 	command=ls-refs
-- 
2.33.0.917.gae6ecbedc7
Previous: Junio C HamanoNext: Taylor Blau
Message 43 of 77 in “reducing memory allocations for v2 servers”
  1. 0/9 reducing memory allocations for v2 serversJeff King, Sep 14, 2021
  2. 1/9 serve: rename is_command() to parse_command()Jeff King, Sep 14, 2021
  3. 2/9 serve: return capability "value" from get_capability()Jeff King, Sep 14, 2021
  4. 3/9 serve: add "receive" method for v2 capabilities tableJeff King, Sep 14, 2021
  5. 4/9 serve: provide "receive" function for object-format capabilityJeff King, Sep 14, 2021
  6. Martin ÅgrenSep 14, 2021
  7. 5/9 serve: provide "receive" function for session-id capabilityJeff King, Sep 14, 2021
  8. Taylor BlauSep 14, 2021
  9. Jeff KingSep 14, 2021
  10. Taylor BlauSep 14, 2021
  11. Martin ÅgrenSep 14, 2021
  12. Jeff KingSep 14, 2021
  13. 6/9 serve: drop "keys" strvecJeff King, Sep 14, 2021
  14. Taylor BlauSep 14, 2021
  15. Jeff KingSep 14, 2021
  16. 7/9 ls-refs: ignore very long ref-prefix countsJeff King, Sep 14, 2021
  17. Taylor BlauSep 14, 2021
  18. Jeff KingSep 14, 2021
  19. Jeff KingSep 14, 2021
  20. Martin ÅgrenSep 14, 2021
  21. Jeff KingSep 14, 2021
  22. Jeff KingSep 14, 2021
  23. Taylor BlauSep 14, 2021
  24. Jeff KingSep 14, 2021
  25. 8/9 serve: reject bogus v2 "command=ls-refs=foo"Jeff King, Sep 14, 2021
  26. Taylor BlauSep 14, 2021
  27. 9/9 serve: reject commands used as capabilitiesJeff King, Sep 14, 2021
  28. Taylor BlauSep 14, 2021
  29. Junio C HamanoSep 14, 2021
  30. Jeff KingSep 14, 2021
  31. 0/11 limit memory allocations for v2 serversJeff King, Sep 14, 2021
  32. 01/11 serve: rename is_command() to parse_command()Jeff King, Sep 14, 2021
  33. 02/11 serve: return capability "value" from get_capability()Jeff King, Sep 14, 2021
  34. 03/11 serve: add "receive" method for v2 capabilities tableJeff King, Sep 14, 2021
  35. Ævar Arnfjörð BjarmasonSep 15, 2021
  36. Jeff KingSep 15, 2021
  37. Junio C HamanoSep 15, 2021
  38. Jeff KingSep 15, 2021
  39. 04/11 serve: provide "receive" function for object-format capabilityJeff King, Sep 14, 2021
  40. Junio C HamanoSep 15, 2021
  41. 05/11 serve: provide "receive" function for session-id capabilityJeff King, Sep 14, 2021
  42. Junio C HamanoSep 15, 2021
  43. 07/11 ls-refs: ignore very long ref-prefix countsJeff King, Sep 14, 2021
  44. Taylor BlauSep 15, 2021
  45. Jeff KingSep 15, 2021
  46. Eric SunshineSep 15, 2021
  47. Jeff KingSep 15, 2021
  48. 06/11 serve: drop "keys" strvecJeff King, Sep 14, 2021
  49. Junio C HamanoSep 15, 2021
  50. 08/11 docs/protocol-v2: clarify some ls-refs ref-prefix detailsJeff King, Sep 14, 2021
  51. 09/11 serve: reject bogus v2 "command=ls-refs=foo"Jeff King, Sep 14, 2021
  52. Ævar Arnfjörð BjarmasonSep 15, 2021
  53. Jeff KingSep 15, 2021
  54. Eric SunshineSep 15, 2021
  55. Jeff KingSep 15, 2021
  56. Junio C HamanoSep 15, 2021
  57. Jeff KingSep 15, 2021
  58. 10/11 serve: reject commands used as capabilitiesJeff King, Sep 14, 2021
  59. 11/11 ls-refs: reject unknown argumentsJeff King, Sep 14, 2021
  60. Ævar Arnfjörð BjarmasonSep 15, 2021
  61. Jeff KingSep 15, 2021
  62. Taylor BlauSep 15, 2021
  63. Jeff KingSep 15, 2021
  64. 0/11 limit memory allocations for v2 serversJeff King, Sep 15, 2021
  65. 01/11 serve: rename is_command() to parse_command()Jeff King, Sep 15, 2021
  66. 02/11 serve: return capability "value" from get_capability()Jeff King, Sep 15, 2021
  67. 03/11 serve: add "receive" method for v2 capabilities tableJeff King, Sep 15, 2021
  68. 04/11 serve: provide "receive" function for object-format capabilityJeff King, Sep 15, 2021
  69. 05/11 serve: provide "receive" function for session-id capabilityJeff King, Sep 15, 2021
  70. 06/11 serve: drop "keys" strvecJeff King, Sep 15, 2021
  71. 07/11 ls-refs: ignore very long ref-prefix countsJeff King, Sep 15, 2021
  72. 08/11 docs/protocol-v2: clarify some ls-refs ref-prefix detailsJeff King, Sep 15, 2021
  73. 09/11 serve: reject bogus v2 "command=ls-refs=foo"Jeff King, Sep 15, 2021
  74. 10/11 serve: reject commands used as capabilitiesJeff King, Sep 15, 2021
  75. 11/11 ls-refs: reject unknown argumentsJeff King, Sep 15, 2021
  76. Ævar Arnfjörð BjarmasonSep 15, 2021
  77. Jeff KingSep 15, 2021

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.