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

[PATCH 2/2] connect: know that zero-ID is not a ref

From
JTJonathan Tan <jonathantanmy@google.com>
Date
Sep 2, 2016, 17:15 UTC
Message-ID
<2bea354c6218a33db3972e42baa75676fdcbc598.1472836026.git.jonathantanmy@google.com>
In-Reply-To
<cover.1472836026.git.jonathantanmy@google.com>

connect.c, when processing packfiles, treats a zero ID (with `capabilities^{}` in place of the refname) as an actual ref instead of a placeholder for a capability declaration, contrary to the specification in Reference Discovery in Documentation/technical/pack-protocol.txt. This is an issue when interacting with Git implementations that follow this specification. For example, `ls-remote` (against a git:// repository served by such a Git implementation) will report a ref when there are none, and `clone` (against something similar) will fail its checkout.

Make connect.c follow the specification with respect to this, while maintaining compatibility with existing implementations that do not serve the zero ID when a repository has no refs.

(git-daemon should probably also be changed to serve zero IDs, but such a change can be considered independently from this change; even if both the client and server changes were made in one commit, it is nearly impossible that all Git installations are updated at the same time - an updated client would still need to deal with unupdated servers and vice versa.)

The test uses JGit's daemon feature, which is specification-compliant.
Signed-off-by: Jonathan Tan <jonathantanmy@google.com>
---
 connect.c            |  7 +++++++
 t/t5512-ls-remote.sh | 22 ++++++++++++++++++++++
 2 files changed, 29 insertions(+)
diff --git a/connect.c b/connect.c
index 722dc3f..d4a58de 100644
--- a/connect.c
+++ b/connect.c
@@ -165,6 +165,13 @@ struct ref **get_remote_heads(int in, char *src_buf, size_t src_len,
 			continue;
 		}
 
+		if (is_null_oid(&old_oid)) {
+			if (strcmp(name, "capabilities^{}"))
+				warning("zero object ID received that is not accompanied by a "
+				        "capability declaration, ignoring and continuing anyway");
+			continue;
+		}
+
 		if (!check_ref(name, flags))
 			continue;
 		ref = alloc_ref(buffer + GIT_SHA1_HEXSZ + 1);
diff --git a/t/t5512-ls-remote.sh b/t/t5512-ls-remote.sh
index 819b9dd..c6f8b6f 100755
--- a/t/t5512-ls-remote.sh
+++ b/t/t5512-ls-remote.sh
@@ -207,5 +207,27 @@ test_expect_success 'ls-remote --symref omits filtered-out matches' '
 	test_cmp expect actual
 '
 
+test_lazy_prereq GIT_DAEMON '
+	test_have_prereq JGIT &&
+	test_tristate GIT_TEST_GIT_DAEMON &&
+	test "$GIT_TEST_GIT_DAEMON" != false
+'
+
+JGIT_DAEMON_PORT=${JGIT_DAEMON_PORT-${this_test#t}}
+
+# This test spawns a daemon, so run it only if the user would be OK with
+# testing with git-daemon.
+test_expect_success JGIT,GIT_DAEMON 'indicate no refs in standards-compliant empty remote' '
+	JGIT_DAEMON_PID= &&
+	git init --bare empty.git &&
+	touch empty.git/git-daemon-export-ok &&
+	{
+		jgit daemon --port="$JGIT_DAEMON_PORT" . &
+		JGIT_DAEMON_PID=$!
+	} &&
+	test_when_finished kill "$JGIT_DAEMON_PID" &&
+	sleep 1 && # allow jgit daemon some time to set up
+	test_expect_code 2 git ls-remote --exit-code git://localhost:$JGIT_DAEMON_PORT/empty.git
+'
 
 test_done
-- 
2.8.0.rc3.226.g39d4020
Previous: Jonathan TanNext: Jonathan Nieder
Message 2 of 45 in “handle empty spec-compliant remote repos correctly”
  1. 0/2 handle empty spec-compliant remote repos correctlyJonathan Tan, Sep 2, 2016
  2. 2/2 connect: know that zero-ID is not a refJonathan Tan, Sep 2, 2016
  3. Jonathan NiederSep 2, 2016
  4. Shawn PearceSep 2, 2016
  5. Stefan BellerSep 2, 2016
  6. Shawn PearceSep 2, 2016
  7. Jeff KingSep 2, 2016
  8. Jonathan TanSep 2, 2016
  9. Jeff KingSep 2, 2016
  10. Shawn PearceSep 3, 2016
  11. Jeff KingSep 3, 2016
  12. 1/2 tests: move test_lazy_prereq JGIT to test-lib.shJonathan Tan, Sep 2, 2016
  13. 0/2 handle empty spec-compliant remote repos correctlyJonathan Tan, Sep 2, 2016
  14. 1/2 tests: move test_lazy_prereq JGIT to test-lib.shJonathan Tan, Sep 2, 2016
  15. Junio C HamanoSep 7, 2016
  16. 2/2 connect: advertized capability is not a refJonathan Tan, Sep 2, 2016
  17. Jonathan NiederSep 2, 2016
  18. Jeff KingSep 2, 2016
  19. Stefan BellerSep 2, 2016
  20. Jonathan NiederSep 3, 2016
  21. Jonathan NiederSep 2, 2016
  22. Jeff KingSep 3, 2016
  23. Junio C HamanoSep 7, 2016
  24. Junio C HamanoSep 7, 2016
  25. Jonathan NiederSep 7, 2016
  26. Junio C HamanoSep 7, 2016
  27. 0/2 handle empty spec-compliant remote repos correctlyJonathan Tan, Sep 7, 2016
  28. 1/2 tests: move test_lazy_prereq JGIT to test-lib.shJonathan Tan, Sep 7, 2016
  29. 2/2 connect: advertized capability is not a refJonathan Tan, Sep 7, 2016
  30. Jonathan NiederSep 8, 2016
  31. connect: tighten check for unexpected early hang up (Re: [PATCH v3 2/2] connect: advertized capability is not a ref)Jonathan Nieder, Sep 8, 2016
  32. Jonathan NiederSep 8, 2016
  33. Jonathan NiederSep 8, 2016
  34. Junio C HamanoSep 8, 2016
  35. Stefan BellerSep 8, 2016
  36. Junio C HamanoSep 8, 2016
  37. 0/3 handle empty spec-compliant remote repos correctlyJonathan Tan, Sep 9, 2016
  38. 1/3 tests: move test_lazy_prereq JGIT to test-lib.shJonathan Tan, Sep 9, 2016
  39. Torsten BögershausenSep 10, 2016
  40. Jeff KingSep 10, 2016
  41. 3/3 connect: advertized capability is not a refJonathan Tan, Sep 9, 2016
  42. Jonathan NiederSep 9, 2016
  43. Junio C HamanoSep 9, 2016
  44. Junio C HamanoSep 9, 2016
  45. 2/3 connect: tighten check for unexpected early hang upJonathan Tan, Sep 9, 2016

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.