backfill: add --sparse option

One way to significantly reduce the cost of a Git clone and later fetches is to use a blobless partial clone and combine that with a sparse-checkout that reduces the paths that need to be populated in the working directory. Not only does this reduce the cost of clones and fetches, the sparse-checkout reduces the number of objects needed to download from a promisor remote. However, history investigations can be expensive as computing blob diffs will trigger promisor remote requests for one object at a time. This can be avoided by downloading the blobs needed for the given sparse-checkout using 'git backfill' and its new '--sparse' mode, at a time that the user is willing to pay that extra cost. Note that this is distinctly different from the '--filter=sparse:<oid>' option, as this assumes that the partial clone has all reachable trees and we are using client-side logic to avoid downloading blobs outside of the sparse-checkout cone. This avoids the server-side cost of walking trees while also achieving a similar goal. It also downloads in batches based on similar path names, presenting a resumable download if things are interrupted. This augments the path-walk API to have a possibly-NULL 'pl' member that may point to a 'struct pattern_list'. This could be more general than the sparse-checkout definition at HEAD, but 'git backfill --sparse' is currently the only consumer. Be sure to test this in both cone mode and not cone mode. Cone mode has the benefit that the path-walk can skip certain paths once they would expand beyond the sparse-checkout. Non-cone mode can describe the included files using both positive and negative patterns, which changes the possible return values of path_matches_pattern_list(). Test both kinds of matches for increased coverage. To test this, we can create a blobless sparse clone, expand the sparse-checkout slightly, and then run 'git backfill --sparse' to see how much data is downloaded. The general steps are 1. git clone --filter=blob:none --sparse <url> 2. git sparse-checkout set <dir1> ... <dirN> 3. git backfill --sparse For the Git repository with the 'builtin' directory in the sparse-checkout, we get these results for various batch sizes: | Batch Size | Pack Count | Pack Size | Time | |-----------------|------------|-----------|-------| | (Initial clone) | 3 | 110 MB | | | 10K | 12 | 192 MB | 17.2s | | 15K | 9 | 192 MB | 15.5s | | 20K | 8 | 192 MB | 15.5s | | 25K | 7 | 192 MB | 14.7s | This case matters less because a full clone of the Git repository from GitHub is currently at 277 MB. Using a copy of the Linux repository with the 'kernel/' directory in the sparse-checkout, we get these results: | Batch Size | Pack Count | Pack Size | Time | |-----------------|------------|-----------|------| | (Initial clone) | 2 | 1,876 MB | | | 10K | 11 | 2,187 MB | 46s | | 25K | 7 | 2,188 MB | 43s | | 50K | 5 | 2,194 MB | 44s | | 100K | 4 | 2,194 MB | 48s | This case is more meaningful because a full clone of the Linux repository is currently over 6 GB, so this is a valuable way to download a fraction of the repository and no longer need network access for all reachable objects within the sparse-checkout. Choosing a batch size will depend on a lot of factors, including the user's network speed or reliability, the repository's file structure, and how many versions there are of the file within the sparse-checkout scope. There will not be a one-size-fits-all solution. Signed-off-by: Derrick Stolee <stolee@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
author: Derrick Stolee <derrickstolee@github.com> 2025-02-03 17:11:06 +0000
committer: Junio C Hamano <gitster@pobox.com> 2025-02-03 16:12:42 -0800
commit: bff455576750bd013a3c87b15cc7086cb8c1eab0 (patch)
tree: 28c1c20f55e9da7b10e59b83d49e1ddcac775a08
parent: 6840fe9ee29ab51ffd7d924c624dc62da22c50bf (diff)
download: git-bff455576750bd013a3c87b15cc7086cb8c1eab0.tar.gz
10 files changed, 208 insertions, 15 deletions
diff --git a/Documentation/git-backfill.txt b/Documentation/git-backfill.txt
index 136a1f1d29..a28678983e 100644
--- a/Documentation/git-backfill.txt
+++ b/Documentation/git-backfill.txt
@@ -9,7 +9,7 @@ git-backfill - Download missing objects in a partial clone
 SYNOPSIS
 --------
 [synopsis]
-git backfill [--min-batch-size=<n>]
+git backfill [--min-batch-size=<n>] [--[no-]sparse]
 
 DESCRIPTION
 -----------
@@ -57,6 +57,10 @@ OPTIONS
 	blobs seen at a given path. The default minimum batch size is
 	50,000.
 
+`--[no-]sparse`::
+	Only download objects if they appear at a path that matches the
+	current sparse-checkout.
+
 SEE ALSO
 --------
 linkgit:git-clone[1].
diff --git a/Documentation/technical/api-path-walk.txt b/Documentation/technical/api-path-walk.txt
index 1fba0ce04c..3e089211fb 100644
--- a/Documentation/technical/api-path-walk.txt
+++ b/Documentation/technical/api-path-walk.txt
@@ -56,6 +56,14 @@ better off using the revision walk API instead.
 	the revision walk so that the walk emits commits marked with the
 	`UNINTERESTING` flag.
 
+`pl`::
+	This pattern list pointer allows focusing the path-walk search to
+	a set of patterns, only emitting paths that match the given
+	patterns. See linkgit:gitignore[5] or
+	linkgit:git-sparse-checkout[1] for details about pattern lists.
+	When the pattern list uses cone-mode patterns, then the path-walk
+	API can prune the set of paths it walks to improve performance.
+
 Examples
 --------
 
diff --git a/builtin/backfill.c b/builtin/backfill.c
index cfebee6e17..d7b997fd6f 100644
--- a/builtin/backfill.c
+++ b/builtin/backfill.c
@@ -4,6 +4,7 @@
 #include "parse-options.h"
 #include "repository.h"
 #include "commit.h"
+#include "dir.h"
 #include "hex.h"
 #include "tree.h"
 #include "tree-walk.h"
@@ -21,7 +22,7 @@
 #include "path-walk.h"
 
 static const char * const builtin_backfill_usage[] = {
-	N_("git backfill [--min-batch-size=<n>]"),
+	N_("git backfill [--min-batch-size=<n>] [--[no-]sparse]"),
 	NULL
 };
 
@@ -29,6 +30,7 @@ struct backfill_context {
 	struct repository *repo;
 	struct oid_array current_batch;
 	size_t min_batch_size;
+	int sparse;
 };
 
 static void backfill_context_clear(struct backfill_context *ctx)
@@ -78,6 +80,14 @@ static int do_backfill(struct backfill_context *ctx)
 	struct path_walk_info info = PATH_WALK_INFO_INIT;
 	int ret;
 
+	if (ctx->sparse) {
+		CALLOC_ARRAY(info.pl, 1);
+		if (get_sparse_checkout_patterns(info.pl)) {
+			path_walk_info_clear(&info);
+			return error(_("problem loading sparse-checkout"));
+		}
+	}
+
 	repo_init_revisions(ctx->repo, &revs, "");
 	handle_revision_arg("HEAD", &revs, 0, 0);
 
@@ -106,10 +116,13 @@ int cmd_backfill(int argc, const char **argv, const char *prefix, struct reposit
 		.repo = repo,
 		.current_batch = OID_ARRAY_INIT,
 		.min_batch_size = 50000,
+		.sparse = 0,
 	};
 	struct option options[] = {
 		OPT_INTEGER(0, "min-batch-size", &ctx.min_batch_size,
 			    N_("Minimum number of objects to request at a time")),
+		OPT_BOOL(0, "sparse", &ctx.sparse,
+			 N_("Restrict the missing objects to the current sparse-checkout")),
 		OPT_END(),
 	};
 
diff --git a/dir.c b/dir.c
index 5b2181e589..16ccfe7e4e 100644
--- a/dir.c
+++ b/dir.c
@@ -1093,10 +1093,6 @@ static void invalidate_directory(struct untracked_cache *uc,
 		dir->dirs[i]->recurse = 0;
 }
 
-static int add_patterns_from_buffer(char *buf, size_t size,
-				    const char *base, int baselen,
-				    struct pattern_list *pl);
-
 /* Flags for add_patterns() */
 #define PATTERN_NOFOLLOW (1<<0)
 
@@ -1186,9 +1182,9 @@ static int add_patterns(const char *fname, const char *base, int baselen,
 	return 0;
 }
 
-static int add_patterns_from_buffer(char *buf, size_t size,
-				    const char *base, int baselen,
-				    struct pattern_list *pl)
+int add_patterns_from_buffer(char *buf, size_t size,
+			     const char *base, int baselen,
+			     struct pattern_list *pl)
 {
 	char *orig = buf;
 	int i, lineno = 1;
diff --git a/dir.h b/dir.h
index a3a2f00f5d..6cfef5df66 100644
--- a/dir.h
+++ b/dir.h
@@ -467,6 +467,9 @@ void add_patterns_from_file(struct dir_struct *, const char *fname);
 int add_patterns_from_blob_to_list(struct object_id *oid,
 				   const char *base, int baselen,
 				   struct pattern_list *pl);
+int add_patterns_from_buffer(char *buf, size_t size,
+			     const char *base, int baselen,
+			     struct pattern_list *pl);
 void parse_path_pattern(const char **string, int *patternlen, unsigned *flags, int *nowildcardlen);
 void add_pattern(const char *string, const char *base,
 		 int baselen, struct pattern_list *pl, int srcpos);
diff --git a/path-walk.c b/path-walk.c
index 9715a5550e..341bdd2ba4 100644
--- a/path-walk.c
+++ b/path-walk.c
@@ -12,6 +12,7 @@
 #include "object.h"
 #include "oid-array.h"
 #include "prio-queue.h"
+#include "repository.h"
 #include "revision.h"
 #include "string-list.h"
 #include "strmap.h"
@@ -172,6 +173,23 @@ static int add_tree_entries(struct path_walk_context *ctx,
 		if (type == OBJ_TREE)
 			strbuf_addch(&path, '/');
 
+		if (ctx->info->pl) {
+			int dtype;
+			enum pattern_match_result match;
+			match = path_matches_pattern_list(path.buf, path.len,
+							  path.buf + base_len, &dtype,
+							  ctx->info->pl,
+							  ctx->repo->index);
+
+			if (ctx->info->pl->use_cone_patterns &&
+			    match == NOT_MATCHED)
+				continue;
+			else if (!ctx->info->pl->use_cone_patterns &&
+				 type == OBJ_BLOB &&
+				 match != MATCHED)
+				continue;
+		}
+
 		if (!(list = strmap_get(&ctx->paths_to_lists, path.buf))) {
 			CALLOC_ARRAY(list, 1);
 			list->type = type;
@@ -582,10 +600,10 @@ void path_walk_info_init(struct path_walk_info *info)
 	memcpy(info, &empty, sizeof(empty));
 }
 
-void path_walk_info_clear(struct path_walk_info *info UNUSED)
+void path_walk_info_clear(struct path_walk_info *info)
 {
-	/*
-	 * This destructor is empty for now, as info->revs
-	 * is not owned by 'struct path_walk_info'.
-	 */
+	if (info->pl) {
+		clear_pattern_list(info->pl);
+		free(info->pl);
+	}
 }
diff --git a/path-walk.h b/path-walk.h
index 414d6db23c..473ee9d361 100644
--- a/path-walk.h
+++ b/path-walk.h
@@ -6,6 +6,7 @@
 
 struct rev_info;
 struct oid_array;
+struct pattern_list;
 
 /**
  * The type of a function pointer for the method that is called on a list of
@@ -48,6 +49,16 @@ struct path_walk_info {
 	 * walk the children of such trees.
 	 */
 	int prune_all_uninteresting;
+
+	/**
+	 * Specify a sparse-checkout definition to match our paths to. Do not
+	 * walk outside of this sparse definition. If the patterns are in
+	 * cone mode, then the search may prune directories that are outside
+	 * of the cone. If not in cone mode, then all tree paths will be
+	 * explored but the path_fn will only be called when the path matches
+	 * the sparse-checkout patterns.
+	 */
+	struct pattern_list *pl;
 };
 
 #define PATH_WALK_INFO_INIT {   \
diff --git a/t/helper/test-path-walk.c b/t/helper/test-path-walk.c
index 7f2d409c5b..61e845e5ec 100644
--- a/t/helper/test-path-walk.c
+++ b/t/helper/test-path-walk.c
@@ -1,6 +1,7 @@
 #define USE_THE_REPOSITORY_VARIABLE
 
 #include "test-tool.h"
+#include "dir.h"
 #include "environment.h"
 #include "hex.h"
 #include "object-name.h"
@@ -9,6 +10,7 @@
 #include "revision.h"
 #include "setup.h"
 #include "parse-options.h"
+#include "strbuf.h"
 #include "path-walk.h"
 #include "oid-array.h"
 
@@ -65,7 +67,7 @@ static int emit_block(const char *path, struct oid_array *oids,
 
 int cmd__path_walk(int argc, const char **argv)
 {
-	int res;
+	int res, stdin_pl = 0;
 	struct rev_info revs = REV_INFO_INIT;
 	struct path_walk_info info = PATH_WALK_INFO_INIT;
 	struct path_walk_test_data data = { 0 };
@@ -80,6 +82,8 @@ int cmd__path_walk(int argc, const char **argv)
 			 N_("toggle inclusion of tree objects")),
 		OPT_BOOL(0, "prune", &info.prune_all_uninteresting,
 			 N_("toggle pruning of uninteresting paths")),
+		OPT_BOOL(0, "stdin-pl", &stdin_pl,
+			 N_("read a pattern list over stdin")),
 		OPT_END(),
 	};
 
@@ -99,6 +103,17 @@ int cmd__path_walk(int argc, const char **argv)
 	info.path_fn = emit_block;
 	info.path_fn_data = &data;
 
+	if (stdin_pl) {
+		struct strbuf in = STRBUF_INIT;
+		CALLOC_ARRAY(info.pl, 1);
+
+		info.pl->use_cone_patterns = 1;
+
+		strbuf_fread(&in, 2048, stdin);
+		add_patterns_from_buffer(in.buf, in.len, "", 0, info.pl);
+		strbuf_release(&in);
+	}
+
 	res = walk_objects_by_path(&info);
 
 	printf("commits:%" PRIuMAX "\n"
@@ -107,6 +122,11 @@ int cmd__path_walk(int argc, const char **argv)
 	       "tags:%" PRIuMAX "\n",
 	       data.commit_nr, data.tree_nr, data.blob_nr, data.tag_nr);
 
+	if (info.pl) {
+		clear_pattern_list(info.pl);
+		free(info.pl);
+	}
+
 	release_revisions(&revs);
 	return res;
 }
diff --git a/t/t5620-backfill.sh b/t/t5620-backfill.sh
index 36107a51c5..6b72e9d0e3 100755
--- a/t/t5620-backfill.sh
+++ b/t/t5620-backfill.sh
@@ -77,6 +77,94 @@ test_expect_success 'do partial clone 2, backfill min batch size' '
 	test_line_count = 0 revs2
 '
 
+test_expect_success 'backfill --sparse' '
+	git clone --sparse --filter=blob:none		\
+		--single-branch --branch=main 		\
+		"file://$(pwd)/srv.bare" backfill3 &&
+
+	# Initial checkout includes four files at root.
+	git -C backfill3 rev-list --quiet --objects --missing=print HEAD >missing &&
+	test_line_count = 44 missing &&
+
+	# Initial sparse-checkout is just the files at root, so we get the
+	# older versions of the four files at tip.
+	GIT_TRACE2_EVENT="$(pwd)/sparse-trace1" git \
+		-C backfill3 backfill --sparse &&
+	test_trace2_data promisor fetch_count 4 <sparse-trace1 &&
+	test_trace2_data path-walk paths 5 <sparse-trace1 &&
+	git -C backfill3 rev-list --quiet --objects --missing=print HEAD >missing &&
+	test_line_count = 40 missing &&
+
+	# Expand the sparse-checkout to include 'd' recursively. This
+	# engages the algorithm to skip the trees for 'a'. Note that
+	# the "sparse-checkout set" command downloads the objects at tip
+	# to satisfy the current checkout.
+	git -C backfill3 sparse-checkout set d &&
+	GIT_TRACE2_EVENT="$(pwd)/sparse-trace2" git \
+		-C backfill3 backfill --sparse &&
+	test_trace2_data promisor fetch_count 8 <sparse-trace2 &&
+	test_trace2_data path-walk paths 15 <sparse-trace2 &&
+	git -C backfill3 rev-list --quiet --objects --missing=print HEAD >missing &&
+	test_line_count = 24 missing
+'
+
+test_expect_success 'backfill --sparse without cone mode (positive)' '
+	git clone --no-checkout --filter=blob:none		\
+		--single-branch --branch=main 		\
+		"file://$(pwd)/srv.bare" backfill4 &&
+
+	# No blobs yet
+	git -C backfill4 rev-list --quiet --objects --missing=print HEAD >missing &&
+	test_line_count = 48 missing &&
+
+	# Define sparse-checkout by filename regardless of parent directory.
+	# This downloads 6 blobs to satisfy the checkout.
+	git -C backfill4 sparse-checkout set --no-cone "**/file.1.txt" &&
+	git -C backfill4 checkout main &&
+
+	# Track new blob count
+	git -C backfill4 rev-list --quiet --objects --missing=print HEAD >missing &&
+	test_line_count = 42 missing &&
+
+	GIT_TRACE2_EVENT="$(pwd)/no-cone-trace1" git \
+		-C backfill4 backfill --sparse &&
+	test_trace2_data promisor fetch_count 6 <no-cone-trace1 &&
+
+	# This walk needed to visit all directories to search for these paths.
+	test_trace2_data path-walk paths 12 <no-cone-trace1 &&
+	git -C backfill4 rev-list --quiet --objects --missing=print HEAD >missing &&
+	test_line_count = 36 missing
+'
+
+test_expect_success 'backfill --sparse without cone mode (negative)' '
+	git clone --no-checkout --filter=blob:none		\
+		--single-branch --branch=main 		\
+		"file://$(pwd)/srv.bare" backfill5 &&
+
+	# No blobs yet
+	git -C backfill5 rev-list --quiet --objects --missing=print HEAD >missing &&
+	test_line_count = 48 missing &&
+
+	# Define sparse-checkout by filename regardless of parent directory.
+	# This downloads 18 blobs to satisfy the checkout
+	git -C backfill5 sparse-checkout set --no-cone "**/file*" "!**/file.1.txt" &&
+	git -C backfill5 checkout main &&
+
+	# Track new blob count
+	git -C backfill5 rev-list --quiet --objects --missing=print HEAD >missing &&
+	test_line_count = 30 missing &&
+
+	GIT_TRACE2_EVENT="$(pwd)/no-cone-trace2" git \
+		-C backfill5 backfill --sparse &&
+	test_trace2_data promisor fetch_count 18 <no-cone-trace2 &&
+
+	# This walk needed to visit all directories to search for these paths, plus
+	# 12 extra "file.?.txt" paths than the previous test.
+	test_trace2_data path-walk paths 24 <no-cone-trace2 &&
+	git -C backfill5 rev-list --quiet --objects --missing=print HEAD >missing &&
+	test_line_count = 12 missing
+'
+
 . "$TEST_DIRECTORY"/lib-httpd.sh
 start_httpd
 
diff --git a/t/t6601-path-walk.sh b/t/t6601-path-walk.sh
index 5f04acb8a2..c89b0f1e19 100755
--- a/t/t6601-path-walk.sh
+++ b/t/t6601-path-walk.sh
@@ -176,6 +176,38 @@ test_expect_success 'branches and indexed objects mix well' '
 	test_cmp_sorted expect out
 '
 
+test_expect_success 'base & topic, sparse' '
+	cat >patterns <<-EOF &&
+	/*
+	!/*/
+	/left/
+	EOF
+
+	test-tool path-walk --stdin-pl -- base topic <patterns >out &&
+
+	cat >expect <<-EOF &&
+	0:commit::$(git rev-parse topic)
+	0:commit::$(git rev-parse base)
+	0:commit::$(git rev-parse base~1)
+	0:commit::$(git rev-parse base~2)
+	1:tree::$(git rev-parse topic^{tree})
+	1:tree::$(git rev-parse base^{tree})
+	1:tree::$(git rev-parse base~1^{tree})
+	1:tree::$(git rev-parse base~2^{tree})
+	2:blob:a:$(git rev-parse base~2:a)
+	3:tree:left/:$(git rev-parse base:left)
+	3:tree:left/:$(git rev-parse base~2:left)
+	4:blob:left/b:$(git rev-parse base~2:left/b)
+	4:blob:left/b:$(git rev-parse base:left/b)
+	blobs:3
+	commits:4
+	tags:0
+	trees:6
+	EOF
+
+	test_cmp_sorted expect out
+'
+
 test_expect_success 'topic only' '
 	test-tool path-walk -- topic >out &&
author	Derrick Stolee <derrickstolee@github.com>	2025-02-03 17:11:06 +0000
committer	Junio C Hamano <gitster@pobox.com>	2025-02-03 16:12:42 -0800
commit	bff455576750bd013a3c87b15cc7086cb8c1eab0 (patch)
tree	28c1c20f55e9da7b10e59b83d49e1ddcac775a08
parent	6840fe9ee29ab51ffd7d924c624dc62da22c50bf (diff)
download	git-bff455576750bd013a3c87b15cc7086cb8c1eab0.tar.gz