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

[PATCH 1/2] http: advertise capabilities when cloning empty repos

From
brian m. carlson <sandals@crustytoothpaste.net>
Date
Apr 26, 2023, 20:53 UTC
Message-ID
<20230426205324.326501-2-sandals@crustytoothpaste.net>
In-Reply-To
<20230426205324.326501-1-sandals@crustytoothpaste.net>
From: "brian m. carlson" <bk2204@github.com>

When cloning an empty repository, the HTTP protocol version 0 currently offers nothing but the header and flush packets for the /info/refs endpoint. This means that no capabilities are provided, so the client side doesn't know what capabilities are present.

However, this does pose a problem when working with SHA-256 repositories, since we use the capabilities to know the remote side's object format (hash algorithm). It used to be possible to set the correct algorithm with `GIT_DEFAULT_HASH` (which is what the Git LFS testsuite did), but this no longer works as of 8b214c2e9d ("clone: propagate object-format when cloning from void", 2023-04-05), since there we always read the hash algorithm from the remote. If there is no hash algorithm provided, we default to SHA-1 for backwards compatibility.

Fortunately, the push version of the protocol already indicates a clue for how to solve this. When the /info/refs endpoint is accessed for a push and the remote is empty, we include a dummy "capabilities^{}" ref pointing to the all-zeros object ID. The protocol documentation already indicates this should _always_ be sent, even for fetches and clones, so let's just do that, which means we'll properly announce the hash algorithm as part of the capabilities. This just works with the existing code because we share the same ref code for fetches and clones, and libgit2 does as well.

Signed-off-by: brian m. carlson <bk2204@github.com>
---
 t/t5551-http-fetch-smart.sh | 27 +++++++++++++++++++++++++++
 t/t5700-protocol-v1.sh      | 21 +++++++++++++++++++--
 upload-pack.c               |  4 ++++
 3 files changed, 50 insertions(+), 2 deletions(-)
diff --git a/t/t5551-http-fetch-smart.sh b/t/t5551-http-fetch-smart.sh
index 0908534f25..21b7767cbd 100755
--- a/t/t5551-http-fetch-smart.sh
+++ b/t/t5551-http-fetch-smart.sh
@@ -611,6 +611,33 @@ test_expect_success 'client falls back from v2 to v0 to match server' '
 	grep symref=HEAD:refs/heads/ trace
 '
 
+test_expect_success 'create empty http-accessible SHA-256 repository' '
+	mkdir "$HTTPD_DOCUMENT_ROOT_PATH/sha256.git" &&
+	(cd "$HTTPD_DOCUMENT_ROOT_PATH/sha256.git" &&
+	 git --bare init --object-format=sha256
+	)
+'
+
+test_expect_success 'clone empty SHA-256 repository with protocol v2' '
+	rm -fr sha256 &&
+	echo sha256 >expected &&
+	git -c protocol.version=2 clone "$HTTPD_URL/smart/sha256.git" &&
+	git -C sha256 rev-parse --show-object-format >actual &&
+	test_cmp actual expected &&
+	git ls-remote "$HTTPD_URL/smart/sha256.git" >actual &&
+	test_must_be_empty actual
+'
+
+test_expect_success 'clone empty SHA-256 repository with protocol v0' '
+	rm -fr sha256 &&
+	echo sha256 >expected &&
+	GIT_TRACE=1 GIT_TRACE_PACKET=1 git -c protocol.version=0 clone "$HTTPD_URL/smart/sha256.git" &&
+	git -C sha256 rev-parse --show-object-format >actual &&
+	test_cmp actual expected &&
+	git ls-remote "$HTTPD_URL/smart/sha256.git" >actual &&
+	test_must_be_empty actual
+'
+
 test_expect_success 'passing hostname resolution information works' '
 	BOGUS_HOST=gitbogusexamplehost.invalid &&
 	BOGUS_HTTPD_URL=$HTTPD_PROTO://$BOGUS_HOST:$LIB_HTTPD_PORT &&
diff --git a/t/t5700-protocol-v1.sh b/t/t5700-protocol-v1.sh
index 6c8d4c6cf1..3cd9db9012 100755
--- a/t/t5700-protocol-v1.sh
+++ b/t/t5700-protocol-v1.sh
@@ -249,10 +249,12 @@ test_expect_success 'push with ssh:// using protocol v1' '
 . "$TEST_DIRECTORY"/lib-httpd.sh
 start_httpd
 
-test_expect_success 'create repo to be served by http:// transport' '
+test_expect_success 'create repos to be served by http:// transport' '
 	git init "$HTTPD_DOCUMENT_ROOT_PATH/http_parent" &&
 	git -C "$HTTPD_DOCUMENT_ROOT_PATH/http_parent" config http.receivepack true &&
-	test_commit -C "$HTTPD_DOCUMENT_ROOT_PATH/http_parent" one
+	test_commit -C "$HTTPD_DOCUMENT_ROOT_PATH/http_parent" one &&
+	git init --object-format=sha256 "$HTTPD_DOCUMENT_ROOT_PATH/sha256" &&
+	git -C "$HTTPD_DOCUMENT_ROOT_PATH/sha256" config http.receivepack true
 '
 
 test_expect_success 'clone with http:// using protocol v1' '
@@ -269,6 +271,21 @@ test_expect_success 'clone with http:// using protocol v1' '
 	grep "git< version 1" log
 '
 
+test_expect_success 'clone with http:// using protocol v1 with empty SHA-256 repo' '
+	GIT_TRACE_PACKET=1 GIT_TRACE_CURL=1 git -c protocol.version=1 \
+		clone "$HTTPD_URL/smart/sha256" sha256 2>log &&
+
+	cat log &&
+	echo sha256 >expect &&
+	git -C sha256 rev-parse --show-object-format >actual &&
+	test_cmp expect actual &&
+
+	# Client requested to use protocol v1
+	grep "Git-Protocol: version=1" log &&
+	# Server responded using protocol v1
+	grep "git< version 1" log
+'
+
 test_expect_success 'fetch with http:// using protocol v1' '
 	test_commit -C "$HTTPD_DOCUMENT_ROOT_PATH/http_parent" two &&
 
diff --git a/upload-pack.c b/upload-pack.c
index 08633dc121..5ef9b162b6 100644
--- a/upload-pack.c
+++ b/upload-pack.c
@@ -120,6 +120,7 @@ struct upload_pack_data {
 	unsigned allow_ref_in_want : 1;				/* v2 only */
 	unsigned allow_sideband_all : 1;			/* v2 only */
 	unsigned advertise_sid : 1;
+	unsigned sent_capabilities : 1;
 };
 
 static void upload_pack_data_init(struct upload_pack_data *data)
@@ -1240,6 +1241,7 @@ static int send_ref(const char *refname, const struct object_id *oid,
 			     git_user_agent_sanitized());
 		strbuf_release(&symref_info);
 		strbuf_release(&session_id);
+		data->sent_capabilities = 1;
 	} else {
 		packet_fwrite_fmt(stdout, "%s %s\n", oid_to_hex(oid), refname_nons);
 	}
@@ -1379,6 +1381,8 @@ void upload_pack(const int advertise_refs, const int stateless_rpc,
 			data.no_done = 1;
 		head_ref_namespaced(send_ref, &data);
 		for_each_namespaced_ref(send_ref, &data);
+		if (!data.sent_capabilities && advertise_refs)
+			send_ref("capabilities^{}", null_oid(), 0, &data);
 		/*
 		 * fflush stdout before calling advertise_shallow_grafts because send_ref
 		 * uses stdio.
Previous: brian m. carlsonNext: Junio C Hamano
Message 23 of 58 in “git clone of empty repositories doesn't preserve hash”
  1. Adam MajerApr 5, 2023
  2. Junio C HamanoApr 5, 2023
  3. Adam MajerApr 5, 2023
  4. Jeff KingApr 5, 2023
  5. Junio C HamanoApr 5, 2023
  6. Junio C HamanoApr 5, 2023
  7. Jeff KingApr 5, 2023
  8. brian m. carlsonApr 5, 2023
  9. Adam MajerApr 6, 2023
  10. brian m. carlsonApr 25, 2023
  11. Junio C HamanoApr 25, 2023
  12. Junio C HamanoApr 25, 2023
  13. brian m. carlsonApr 26, 2023
  14. Jeff KingApr 26, 2023
  15. Junio C HamanoApr 26, 2023
  16. doc: GIT_DEFAULT_HASH is and will be ignored during "clone"Junio C Hamano, Apr 26, 2023
  17. brian m. carlsonApr 26, 2023
  18. Jeff KingApr 27, 2023
  19. Jeff KingApr 26, 2023
  20. Junio C HamanoApr 26, 2023
  21. brian m. carlsonApr 26, 2023
  22. 0/2 Fix empty SHA-256 clones with v0 and v1brian m. carlson, Apr 26, 2023
  23. 1/2 http: advertise capabilities when cloning empty reposbrian m. carlson, Apr 26, 2023
  24. Junio C HamanoApr 26, 2023
  25. brian m. carlsonApr 26, 2023
  26. Jeff KingApr 27, 2023
  27. Jeff KingApr 27, 2023
  28. Junio C HamanoApr 27, 2023
  29. 2/2 Honor GIT_DEFAULT_HASH for empty clones without remote algobrian m. carlson, Apr 26, 2023
  30. Junio C HamanoApr 26, 2023
  31. Junio C HamanoApr 26, 2023
  32. Jeff KingApr 27, 2023
  33. Is GIT_DEFAULT_HASH flawed?Felipe Contreras, May 2, 2023
  34. Adam MajerMay 3, 2023
  35. Felipe ContrerasMay 3, 2023
  36. Adam MajerMay 3, 2023
  37. Felipe ContrerasMay 8, 2023
  38. demerphqMay 3, 2023
  39. Felipe ContrerasMay 3, 2023
  40. brian m. carlsonMay 3, 2023
  41. Felipe ContrerasMay 8, 2023
  42. brian m. carlsonMay 8, 2023
  43. Oswald BuddenhagenMay 9, 2023
  44. Junio C HamanoMay 9, 2023
  45. Junio C HamanoApr 26, 2023
  46. Jeff KingApr 27, 2023
  47. 0/1 Fix empty SHA-256 clones with v0 and v1brian m. carlson, May 1, 2023
  48. 1/1 upload-pack: advertise capabilities when cloning empty reposbrian m. carlson, May 1, 2023
  49. Jeff KingMay 1, 2023
  50. Junio C HamanoMay 1, 2023
  51. Junio C HamanoMay 1, 2023
  52. 0/1 Fix empty SHA-256 clones with v0 and v1brian m. carlson, May 17, 2023
  53. 1/1 upload-pack: advertise capabilities when cloning empty reposbrian m. carlson, May 17, 2023
  54. Junio C HamanoMay 17, 2023
  55. brian m. carlsonMay 17, 2023
  56. Jeff KingMay 18, 2023
  57. brian m. carlsonMay 19, 2023
  58. Jeff KingApr 5, 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.