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

[PATCH v2 6/7] promisor-remote: lazy-fetch objects in subprocess

From
JTJonathan Tan <jonathantanmy@google.com>
Date
Aug 11, 2020, 22:52 UTC
Message-ID
<55d2e5a4cccee0ae719f4210c2cbeeb6a691cf2f.1597184949.git.jonathantanmy@google.com>
In-Reply-To
<cover.1597184948.git.jonathantanmy@google.com>

Teach Git to lazy-fetch missing objects in a subprocess instead of doing it in-process. This allows any fatal errors that occur during the fetch to be isolated and converted into an error return value, instead of causing the current command being run to terminate.

Signed-off-by: Jonathan Tan <jonathantanmy@google.com>
---
 Documentation/technical/partial-clone.txt | 13 ++-----
 promisor-remote.c                         | 46 +++++++++++------------
 t/t0410-partial-clone.sh                  |  2 +-
 t/t4067-diff-partial-clone.sh             |  8 ++--
 t/t5601-clone.sh                          |  2 +-
 5 files changed, 30 insertions(+), 41 deletions(-)
diff --git a/Documentation/technical/partial-clone.txt b/Documentation/technical/partial-clone.txt
index b9e17e7a28..0780d30cac 100644
--- a/Documentation/technical/partial-clone.txt
+++ b/Documentation/technical/partial-clone.txt
@@ -171,20 +171,13 @@ additional flag.
 Fetching Missing Objects
 ------------------------
 
-- Fetching of objects is done using the existing transport mechanism using
-  transport_fetch_refs(), setting a new transport option
-  TRANS_OPT_NO_DEPENDENTS to indicate that only the objects themselves are
-  desired, not any object that they refer to.
-+
-Because some transports invoke fetch_pack() in the same process, fetch_pack()
-has been updated to not use any object flags when the corresponding argument
-(no_dependents) is set.
+- Fetching of objects is done by invoking a "git fetch" subprocess.
 
 - The local repository sends a request with the hashes of all requested
-  objects as "want" lines, and does not perform any packfile negotiation.
+  objects, and does not perform any packfile negotiation.
   It then receives a packfile.
 
-- Because we are reusing the existing fetch-pack mechanism, fetching
+- Because we are reusing the existing fetch mechanism, fetching
   currently fetches all objects referred to by the requested objects, even
   though they are not necessary.
 
diff --git a/promisor-remote.c b/promisor-remote.c
index baaea12fd6..6e647610e9 100644
--- a/promisor-remote.c
+++ b/promisor-remote.c
@@ -3,6 +3,7 @@
 #include "promisor-remote.h"
 #include "config.h"
 #include "transport.h"
+#include "argv-array.h"
 
 static char *repository_format_partial_clone;
 static const char *core_partial_clone_filter_default;
@@ -12,39 +13,34 @@ void set_repository_format_partial_clone(char *partial_clone)
 	repository_format_partial_clone = xstrdup_or_null(partial_clone);
 }
 
-static int fetch_refs(const char *remote_name, struct ref *ref)
-{
-	struct remote *remote;
-	struct transport *transport;
-	int res;
-
-	remote = remote_get(remote_name);
-	if (!remote->url[0])
-		die(_("Remote with no URL"));
-	transport = transport_get(remote, remote->url[0]);
-
-	transport_set_option(transport, TRANS_OPT_FROM_PROMISOR, "1");
-	transport_set_option(transport, TRANS_OPT_NO_DEPENDENTS, "1");
-	res = transport_fetch_refs(transport, ref);
-
-	return res;
-}
-
 static int fetch_objects(const char *remote_name,
 			 const struct object_id *oids,
 			 int oid_nr)
 {
-	struct ref *ref = NULL;
+	struct child_process child = CHILD_PROCESS_INIT;
 	int i;
+	FILE *child_in;
+
+	child.git_cmd = 1;
+	child.in = -1;
+	argv_array_pushl(&child.args, "-c", "fetch.negotiationAlgorithm=null",
+			 "fetch", remote_name, "--no-tags",
+			 "--no-write-fetch-head", "--recurse-submodules=no",
+			 "--filter=blob:none", "--stdin", NULL);
+	if (start_command(&child))
+		die(_("promisor-remote: unable to fork off fetch subprocess"));
+	child_in = xfdopen(child.in, "w");
 
 	for (i = 0; i < oid_nr; i++) {
-		struct ref *new_ref = alloc_ref(oid_to_hex(&oids[i]));
-		oidcpy(&new_ref->old_oid, &oids[i]);
-		new_ref->exact_oid = 1;
-		new_ref->next = ref;
-		ref = new_ref;
+		if (fputs(oid_to_hex(&oids[i]), child_in) < 0)
+			die_errno(_("promisor-remote: could not write to fetch subprocess"));
+		if (fputc('\n', child_in) < 0)
+			die_errno(_("promisor-remote: could not write to fetch subprocess"));
 	}
-	return fetch_refs(remote_name, ref);
+
+	if (fclose(child_in) < 0)
+		die_errno(_("promisor-remote: could not close stdin to fetch subprocess"));
+	return finish_command(&child) ? -1 : 0;
 }
 
 static struct promisor_remote *promisors;
diff --git a/t/t0410-partial-clone.sh b/t/t0410-partial-clone.sh
index 463dc3a8be..3e454f934e 100755
--- a/t/t0410-partial-clone.sh
+++ b/t/t0410-partial-clone.sh
@@ -203,7 +203,7 @@ test_expect_success 'fetching of missing objects works with ref-in-want enabled'
 	rm -rf repo/.git/objects/* &&
 	rm -f trace &&
 	GIT_TRACE_PACKET="$(pwd)/trace" git -C repo cat-file -p "$HASH" &&
-	grep "git< fetch=.*ref-in-want" trace
+	grep "fetch< fetch=.*ref-in-want" trace
 '
 
 test_expect_success 'fetching of missing objects from another promisor remote' '
diff --git a/t/t4067-diff-partial-clone.sh b/t/t4067-diff-partial-clone.sh
index ef8e0e9cb0..804f2a82e8 100755
--- a/t/t4067-diff-partial-clone.sh
+++ b/t/t4067-diff-partial-clone.sh
@@ -20,7 +20,7 @@ test_expect_success 'git show batches blobs' '
 	# Ensure that there is exactly 1 negotiation by checking that there is
 	# only 1 "done" line sent. ("done" marks the end of negotiation.)
 	GIT_TRACE_PACKET="$(pwd)/trace" git -C client show HEAD &&
-	grep "git> done" trace >done_lines &&
+	grep "fetch> done" trace >done_lines &&
 	test_line_count = 1 done_lines
 '
 
@@ -44,7 +44,7 @@ test_expect_success 'diff batches blobs' '
 	# Ensure that there is exactly 1 negotiation by checking that there is
 	# only 1 "done" line sent. ("done" marks the end of negotiation.)
 	GIT_TRACE_PACKET="$(pwd)/trace" git -C client diff HEAD^ HEAD &&
-	grep "git> done" trace >done_lines &&
+	grep "fetch> done" trace >done_lines &&
 	test_line_count = 1 done_lines
 '
 
@@ -127,7 +127,7 @@ test_expect_success 'diff with rename detection batches blobs' '
 	# only 1 "done" line sent. ("done" marks the end of negotiation.)
 	GIT_TRACE_PACKET="$(pwd)/trace" git -C client diff --raw -M HEAD^ HEAD >out &&
 	grep ":100644 100644.*R[0-9][0-9][0-9].*b.*c" out &&
-	grep "git> done" trace >done_lines &&
+	grep "fetch> done" trace >done_lines &&
 	test_line_count = 1 done_lines
 '
 
@@ -175,7 +175,7 @@ test_expect_success 'diff --break-rewrites fetches only if necessary, and batche
 	# by checking that there is only 1 "done" line sent. ("done" marks the
 	# end of negotiation.)
 	GIT_TRACE_PACKET="$(pwd)/trace" git -C client diff --break-rewrites --raw -M HEAD^ HEAD &&
-	grep "git> done" trace >done_lines &&
+	grep "fetch> done" trace >done_lines &&
 	test_line_count = 1 done_lines
 '
 
diff --git a/t/t5601-clone.sh b/t/t5601-clone.sh
index 84ea2a3eb7..f82b0dbb5a 100755
--- a/t/t5601-clone.sh
+++ b/t/t5601-clone.sh
@@ -702,7 +702,7 @@ test_expect_success 'batch missing blob request during checkout' '
 	# Ensure that there is only one negotiation by checking that there is
 	# only "done" line sent. ("done" marks the end of negotiation.)
 	GIT_TRACE_PACKET="$(pwd)/trace" git -C client checkout HEAD^ &&
-	grep "git> done" trace >done_lines &&
+	grep "fetch> done" trace >done_lines &&
 	test_line_count = 1 done_lines
 '
 

-- 
2.28.0.236.gb10cc79966-goog
Previous: Junio C HamanoNext: Junio C Hamano
Message 42 of 57 in “Modify fetch-pack to no longer die on error?”
  1. Modify fetch-pack to no longer die on error?Jonathan Tan, Jul 24, 2020
  2. Junio C HamanoJul 24, 2020
  3. Junio C HamanoJul 24, 2020
  4. Jeff KingJul 25, 2020
  5. Junio C HamanoJul 25, 2020
  6. Jeff KingJul 27, 2020
  7. Jonathan TanJul 28, 2020
  8. Jeff KingJul 28, 2020
  9. Jonathan TanJul 29, 2020
  10. Jeff KingJul 29, 2020
  11. Junio C HamanoJul 29, 2020
  12. Jonathan TanJul 29, 2020
  13. 0/7 Lazy fetch with subprocessJonathan Tan, Aug 5, 2020
  14. 1/7 fetch-pack: allow NULL negotiator->add_tipJonathan Tan, Aug 5, 2020
  15. Junio C HamanoAug 5, 2020
  16. Jonathan TanAug 7, 2020
  17. 3/7 negotiator/null: add null fetch negotiatorJonathan Tan, Aug 5, 2020
  18. 2/7 fetch-pack: allow NULL negotiator->known_commonJonathan Tan, Aug 5, 2020
  19. Junio C HamanoAug 5, 2020
  20. Junio C HamanoAug 5, 2020
  21. Jonathan TanAug 7, 2020
  22. 4/7 fetch: --stdinJonathan Tan, Aug 5, 2020
  23. Junio C HamanoAug 5, 2020
  24. Jonathan TanAug 7, 2020
  25. Junio C HamanoAug 7, 2020
  26. Jonathan TanAug 7, 2020
  27. 5/7 fetch: submodule configJonathan Tan, Aug 5, 2020
  28. 7/7 promisor-remote: use subprocess to fetchJonathan Tan, Aug 5, 2020
  29. 6/7 fetch: only populate existing_refs if neededJonathan Tan, Aug 5, 2020
  30. 0/7 Lazy fetch with subprocessJonathan Tan, Aug 11, 2020
  31. 1/7 negotiator/null: add null fetch negotiatorJonathan Tan, Aug 11, 2020
  32. Derrick StoleeAug 12, 2020
  33. Junio C HamanoAug 12, 2020
  34. Jonathan TanAug 12, 2020
  35. 2/7 fetch: allow refspecs specified through stdinJonathan Tan, Aug 11, 2020
  36. 3/7 fetch: avoid reading submodule config until neededJonathan Tan, Aug 11, 2020
  37. Junio C HamanoAug 12, 2020
  38. 4/7 fetch: only populate existing_refs if neededJonathan Tan, Aug 11, 2020
  39. Junio C HamanoAug 12, 2020
  40. 5/7 fetch-pack: do not lazy-fetch during ref iterationJonathan Tan, Aug 11, 2020
  41. Junio C HamanoAug 12, 2020
  42. 6/7 promisor-remote: lazy-fetch objects in subprocessJonathan Tan, Aug 11, 2020
  43. Junio C HamanoAug 12, 2020
  44. 7/7 fetch-pack: remove no_dependents codeJonathan Tan, Aug 11, 2020
  45. Derrick StoleeAug 12, 2020
  46. 0/7 Lazy fetch with subprocessJonathan Tan, Aug 18, 2020
  47. 1/7 negotiator/noop: add noop fetch negotiatorJonathan Tan, Aug 18, 2020
  48. 2/7 fetch: allow refspecs specified through stdinJonathan Tan, Aug 18, 2020
  49. 4/7 fetch: only populate existing_refs if neededJonathan Tan, Aug 18, 2020
  50. 5/7 fetch-pack: do not lazy-fetch during ref iterationJonathan Tan, Aug 18, 2020
  51. 3/7 fetch: avoid reading submodule config until neededJonathan Tan, Aug 18, 2020
  52. 6/7 promisor-remote: lazy-fetch objects in subprocessJonathan Tan, Aug 18, 2020
  53. 7/7 fetch-pack: remove no_dependents codeJonathan Tan, Aug 18, 2020
  54. Junio C HamanoAug 18, 2020
  55. Junio C HamanoAug 18, 2020
  56. fixup! promisor-remote: lazy-fetch objects in subprocessJonathan Tan, Aug 18, 2020
  57. Junio C HamanoAug 18, 2020

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.