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

[PATCH v5 6/8] send-pack: new return code "ERROR_SEND_PACK_BAD_REF_STATUS"

From
Patrick Steinhardt <ps@pks.im>
Date
Feb 3, 2025, 06:29 UTC
Message-ID
<20250203-pks-push-atomic-respect-exit-code-v5-6-d66481e36622@pks.im>
In-Reply-To
<20250203-pks-push-atomic-respect-exit-code-v5-0-d66481e36622@pks.im>
From: Jiang Xin <zhiyou.jx@alibaba-inc.com>

The "push_refs" function in the transport_vtable is the handler for git-push operation. All the "push_refs" functions for different transports (protocols) should have the same behavior, but the behavior of "git_transport_push()" function for builtin_smart_vtable in "transport.c" (which calls "send_pack()" in "send-pack.c") differs from the handler of the HTTP protocol.

The "push_refs()" function for the HTTP protocol which calls the "push_refs_with_push()" function in "transport-helper.c" will return 0 even when a bad REF_STATUS (such as REF_STATUS_REJECT_NONFASTFORWARD) was found. But "send_pack()" for Git smart protocol will return -1 for a bad REF_STATUS.

We cannot ignore bad REF_STATUS directly in the "send_pack()" function, because the function is also used in "builtin/send-pack.c". So we add a new non-zero error code "SEND_PACK_ERROR_REF_STATUS" for "send_pack()".

Ignore the specific error code in the "git_transport_push()" function to have the same behavior as "push_refs()" for HTTP protocol. Note that even though we ignore the error here, we'll ultimately still end up detecting that a subset of refs was not pushed in `transport_push()` because we eventually call `push_had_errors()` on the remote refs.

Signed-off-by: Jiang Xin <zhiyou.jx@alibaba-inc.com>
Signed-off-by: Patrick Steinhardt <ps@pks.im>
---
 send-pack.c |  9 ++-------
 send-pack.h | 13 +++++++++++++
 transport.c |  7 +++++++
 3 files changed, 22 insertions(+), 7 deletions(-)
diff --git a/send-pack.c b/send-pack.c
index 772c7683a0..4448c081cc 100644
--- a/send-pack.c
+++ b/send-pack.c
@@ -632,7 +632,7 @@ int send_pack(struct repository *r,
 				reject_atomic_push(remote_refs, args->send_mirror);
 				error("atomic push failed for ref %s. status: %d",
 				      ref->name, ref->status);
-				ret = args->porcelain ? 0 : -1;
+				ret = ERROR_SEND_PACK_BAD_REF_STATUS;
 				goto out;
 			}
 			/* else fallthrough */
@@ -763,11 +763,6 @@ int send_pack(struct repository *r,
 	if (ret < 0)
 		goto out;
 
-	if (args->porcelain) {
-		ret = 0;
-		goto out;
-	}
-
 	for (ref = remote_refs; ref; ref = ref->next) {
 		switch (ref->status) {
 		case REF_STATUS_NONE:
@@ -775,7 +770,7 @@ int send_pack(struct repository *r,
 		case REF_STATUS_OK:
 			break;
 		default:
-			ret = -1;
+			ret = ERROR_SEND_PACK_BAD_REF_STATUS;
 			goto out;
 		}
 	}
diff --git a/send-pack.h b/send-pack.h
index d256715681..c5ded2d200 100644
--- a/send-pack.h
+++ b/send-pack.h
@@ -13,6 +13,9 @@ struct repository;
 #define SEND_PACK_PUSH_CERT_IF_ASKED 1
 #define SEND_PACK_PUSH_CERT_ALWAYS 2
 
+/* At least one reference has been rejected by the remote side. */
+#define ERROR_SEND_PACK_BAD_REF_STATUS 1
+
 struct send_pack_args {
 	const char *url;
 	unsigned verbose:1,
@@ -36,6 +39,16 @@ struct option;
 int option_parse_push_signed(const struct option *opt,
 			     const char *arg, int unset);
 
+/*
+ * Compute a packfile and write it to a file descriptor. The `fd` array needs
+ * to contain two file descriptors: `fd[0]` is the file descriptor used as
+ * input for the packet reader, whereas `fd[1]` is the file descriptor the
+ * packfile will be written to.
+ *
+ * Returns 0 on success, non-zero otherwise. Negative return values indicate a
+ * generic error, whereas positive return values indicate specific error
+ * conditions as documented with the `ERROR_SEND_PACK_*` constants.
+ */
 int send_pack(struct repository *r, struct send_pack_args *args,
 	      int fd[], struct child_process *conn,
 	      struct ref *remote_refs, struct oid_array *extra_have);
diff --git a/transport.c b/transport.c
index 81ae8243b9..d064aff33e 100644
--- a/transport.c
+++ b/transport.c
@@ -934,6 +934,13 @@ static int git_transport_push(struct transport *transport, struct ref *remote_re
 	case protocol_v0:
 		ret = send_pack(the_repository, &args, data->fd, data->conn, remote_refs,
 				&data->extra_have);
+		/*
+		 * Ignore the specific error code to maintain consistent behavior
+		 * with the "push_refs()" function across different transports,
+		 * such as "push_refs_with_push()" for HTTP protocol.
+		 */
+		if (ret == ERROR_SEND_PACK_BAD_REF_STATUS)
+			ret = 0;
 		break;
 	case protocol_unknown_version:
 		BUG("unknown protocol version");
-- 
2.48.1.502.g6dc24dfdaf.dirty
Previous: Patrick SteinhardtNext: Patrick Steinhardt
Message 51 of 54 in “transport: don't ignore git-receive-pack(1) exit code on atomic push”
  1. 0/2 transport: don't ignore git-receive-pack(1) exit code on atomic pushPatrick Steinhardt, Nov 13, 2024
  2. 1/2 t5504: modernize test by moving heredocs into test bodiesPatrick Steinhardt, Nov 13, 2024
  3. 2/2 transport: don't ignore git-receive-pack(1) exit code on atomic pushPatrick Steinhardt, Nov 13, 2024
  4. Jiang XinNov 14, 2024
  5. 0/6 fix behaviors of git-push --porcelainJiang Xin, Nov 14, 2024
  6. 1/6 t5548: new test cases for push --porcelain and --dry-runJiang Xin, Nov 14, 2024
  7. Patrick SteinhardtNov 25, 2024
  8. Jiang XinDec 3, 2024
  9. 2/6 push: fix the behavior of the Done message for porcelainJiang Xin, Nov 14, 2024
  10. Patrick SteinhardtNov 25, 2024
  11. 3/6 t5504: modernize test by moving heredocs into test bodiesJiang Xin, Nov 14, 2024
  12. 4/6 t5543: atomic push reports exit code failureJiang Xin, Nov 14, 2024
  13. 5/6 push: only ignore finish_connect() for dry-run modeJiang Xin, Nov 14, 2024
  14. Patrick SteinhardtNov 25, 2024
  15. 6/6 push: not send push-options to server with --dry-runJiang Xin, Nov 14, 2024
  16. Patrick SteinhardtNov 25, 2024
  17. 0/8 fix behaviors of git-push --porcelainJiang Xin, Dec 10, 2024
  18. 1/8 t5504: modernize test by moving heredocs into test bodiesJiang Xin, Dec 10, 2024
  19. 2/8 t5548: refactor to reuse setup_upstream() functionJiang Xin, Dec 10, 2024
  20. 3/8 t5548: refactor test cases by resetting upstreamJiang Xin, Dec 10, 2024
  21. 4/8 t5548: add new porcelain test casesJiang Xin, Dec 10, 2024
  22. 5/8 t5548: add porcelain push test cases for dry-run modeJiang Xin, Dec 10, 2024
  23. Jiang XinDec 10, 2024
  24. 6/8 send-pack: new return code "ERROR_SEND_PACK_BAD_REF_STATUS"Jiang Xin, Dec 10, 2024
  25. Patrick SteinhardtDec 16, 2024
  26. 7/8 t5543: atomic push reports exit code failureJiang Xin, Dec 10, 2024
  27. 8/8 send-pack: gracefully close the connection for atomic pushJiang Xin, Dec 10, 2024
  28. Patrick SteinhardtDec 16, 2024
  29. Junio C HamanoNov 14, 2024
  30. Patrick SteinhardtNov 25, 2024
  31. 0/8 transport: don't ignore git-receive-pack(1) exit code on atomic pushPatrick Steinhardt, Jan 31, 2025
  32. 1/8 t5504: modernize test by moving heredocs into test bodiesPatrick Steinhardt, Jan 31, 2025
  33. Eric SunshineJan 31, 2025
  34. Junio C HamanoJan 31, 2025
  35. Patrick SteinhardtFeb 3, 2025
  36. 2/8 t5548: refactor to reuse setup_upstream() functionPatrick Steinhardt, Jan 31, 2025
  37. Eric SunshineJan 31, 2025
  38. 4/8 t5548: add new porcelain test casesPatrick Steinhardt, Jan 31, 2025
  39. Eric SunshineJan 31, 2025
  40. 5/8 t5548: add porcelain push test cases for dry-run modePatrick Steinhardt, Jan 31, 2025
  41. 3/8 t5548: refactor test cases by resetting upstreamPatrick Steinhardt, Jan 31, 2025
  42. 7/8 t5543: atomic push reports exit code failurePatrick Steinhardt, Jan 31, 2025
  43. 6/8 send-pack: new return code "ERROR_SEND_PACK_BAD_REF_STATUS"Patrick Steinhardt, Jan 31, 2025
  44. 8/8 send-pack: gracefully close the connection for atomic pushPatrick Steinhardt, Jan 31, 2025
  45. 0/8 transport: don't ignore git-receive-pack(1) exit code on atomic pushPatrick Steinhardt, Feb 3, 2025
  46. 1/8 t5504: modernize test by moving heredocs into test bodiesPatrick Steinhardt, Feb 3, 2025
  47. 2/8 t5548: refactor to reuse setup_upstream() functionPatrick Steinhardt, Feb 3, 2025
  48. 3/8 t5548: refactor test cases by resetting upstreamPatrick Steinhardt, Feb 3, 2025
  49. 4/8 t5548: add new porcelain test casesPatrick Steinhardt, Feb 3, 2025
  50. 5/8 t5548: add porcelain push test cases for dry-run modePatrick Steinhardt, Feb 3, 2025
  51. 6/8 send-pack: new return code "ERROR_SEND_PACK_BAD_REF_STATUS"Patrick Steinhardt, Feb 3, 2025
  52. 7/8 t5543: atomic push reports exit code failurePatrick Steinhardt, Feb 3, 2025
  53. 8/8 send-pack: gracefully close the connection for atomic pushPatrick Steinhardt, Feb 3, 2025
  54. Junio C HamanoFeb 3, 2025

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.