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

[PATCH v3 4/4] midx: during verify group objects by packfile to speed verification

From
JGJeff Hostetler via GitGitGadget <gitgitgadget@gmail.com>
Date
Mar 21, 2019, 19:36 UTC
Message-ID
<7e98ea005aa62ca36af7da34dc6c21a1f77b351e.1553196970.git.gitgitgadget@gmail.com>
In-Reply-To
<pull.166.v3.git.gitgitgadget@gmail.com>
From: Jeff Hostetler <jeffhost@microsoft.com>

Teach `multi-pack-index verify` to sort the set of object by packfile so that only one packfile needs to be open at a time.

This is a performance improvement. Previously, objects were verified in OID order. This essentially requires all packfiles to be open at the same time. If the number of packfiles exceeds the open file limit, packfiles would be LRU-closed and re-opened many times.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 midx.c     | 49 ++++++++++++++++++++++++++++++++++++++++++++++---
 packfile.c |  2 +-
 packfile.h |  2 ++
 3 files changed, 49 insertions(+), 4 deletions(-)
diff --git a/midx.c b/midx.c
index e3919387d9..f1cd868f8c 100644
--- a/midx.c
+++ b/midx.c
@@ -962,6 +962,20 @@ static void midx_report(const char *fmt, ...)
 	va_end(ap);
 }
 
+struct pair_pos_vs_id
+{
+	uint32_t pos;
+	uint32_t pack_int_id;
+};
+
+static int compare_pair_pos_vs_id(const void *_a, const void *_b)
+{
+	struct pair_pos_vs_id *a = (struct pair_pos_vs_id *)_a;
+	struct pair_pos_vs_id *b = (struct pair_pos_vs_id *)_b;
+
+	return b->pack_int_id - a->pack_int_id;
+}
+
 /*
  * Limit calls to display_progress() for performance reasons.
  * The interval here was arbitrarily chosen.
@@ -976,6 +990,7 @@ static void midx_report(const char *fmt, ...)
 
 int verify_midx_file(const char *object_dir)
 {
+	struct pair_pos_vs_id *pairs = NULL;
 	uint32_t i;
 	struct progress *progress;
 	struct multi_pack_index *m = load_multi_pack_index(object_dir, 1);
@@ -1019,16 +1034,42 @@ int verify_midx_file(const char *object_dir)
 	}
 	stop_progress(&progress);
 
+	/*
+	 * Create an array mapping each object to its packfile id.  Sort it
+	 * to group the objects by packfile.  Use this permutation to visit
+	 * each of the objects and only require 1 packfile to be open at a
+	 * time.
+	 */
+	ALLOC_ARRAY(pairs, m->num_objects);
+	for (i = 0; i < m->num_objects; i++) {
+		pairs[i].pos = i;
+		pairs[i].pack_int_id = nth_midxed_pack_int_id(m, i);
+	}
+
+	progress = start_sparse_progress(_("Sorting objects by packfile"),
+					 m->num_objects);
+	display_progress(progress, 0); /* TODO: Measure QSORT() progress */
+	QSORT(pairs, m->num_objects, compare_pair_pos_vs_id);
+	stop_progress(&progress);
+
 	progress = start_sparse_progress(_("Verifying object offsets"), m->num_objects);
 	for (i = 0; i < m->num_objects; i++) {
 		struct object_id oid;
 		struct pack_entry e;
 		off_t m_offset, p_offset;
 
-		nth_midxed_object_oid(&oid, m, i);
+		if (i > 0 && pairs[i-1].pack_int_id != pairs[i].pack_int_id &&
+		    m->packs[pairs[i-1].pack_int_id])
+		{
+			close_pack_fd(m->packs[pairs[i-1].pack_int_id]);
+			close_pack_index(m->packs[pairs[i-1].pack_int_id]);
+		}
+
+		nth_midxed_object_oid(&oid, m, pairs[i].pos);
+
 		if (!fill_midx_entry(&oid, &e, m)) {
 			midx_report(_("failed to load pack entry for oid[%d] = %s"),
-				    i, oid_to_hex(&oid));
+				    pairs[i].pos, oid_to_hex(&oid));
 			continue;
 		}
 
@@ -1043,11 +1084,13 @@ int verify_midx_file(const char *object_dir)
 
 		if (m_offset != p_offset)
 			midx_report(_("incorrect object offset for oid[%d] = %s: %"PRIx64" != %"PRIx64),
-				    i, oid_to_hex(&oid), m_offset, p_offset);
+				    pairs[i].pos, oid_to_hex(&oid), m_offset, p_offset);
 
 		midx_display_sparse_progress(progress, i + 1);
 	}
 	stop_progress(&progress);
 
+	free(pairs);
+
 	return verify_midx_error;
 }
diff --git a/packfile.c b/packfile.c
index 16bcb75262..d2bcb2f860 100644
--- a/packfile.c
+++ b/packfile.c
@@ -309,7 +309,7 @@ void close_pack_windows(struct packed_git *p)
 	}
 }
 
-static int close_pack_fd(struct packed_git *p)
+int close_pack_fd(struct packed_git *p)
 {
 	if (p->pack_fd < 0)
 		return 0;
diff --git a/packfile.h b/packfile.h
index d70c6d9afb..b1c18504eb 100644
--- a/packfile.h
+++ b/packfile.h
@@ -76,6 +76,8 @@ extern int open_pack_index(struct packed_git *);
  */
 extern void close_pack_index(struct packed_git *);
 
+int close_pack_fd(struct packed_git *p);
+
 extern uint32_t get_pack_fanout(struct packed_git *p, uint32_t value);
 
 extern unsigned char *use_pack(struct packed_git *, struct pack_window **, off_t, unsigned long *);
-- 
gitgitgadget
Previous: Jeff Hostetler via GitGitGadgetNext: Junio C Hamano
Message 24 of 26 in “multi-pack-index: fix verify on large repos”
  1. 0/3 multi-pack-index: fix verify on large reposJeff Hostetler via GitGitGadget, Mar 18, 2019
  2. 1/3 midx: verify: add midx packfiles to the packed_git listJeff Hostetler via GitGitGadget, Mar 18, 2019
  3. 3/3 trace2:data: add trace2 data to midxJeff Hostetler via GitGitGadget, Mar 18, 2019
  4. 2/3 midx: verify: group objects by packfile to speed up object verificationJeff Hostetler via GitGitGadget, Mar 18, 2019
  5. Ævar Arnfjörð BjarmasonMar 18, 2019
  6. Jeff HostetlerMar 18, 2019
  7. Ævar Arnfjörð BjarmasonMar 18, 2019
  8. Junio C HamanoMar 19, 2019
  9. Jeff HostetlerMar 19, 2019
  10. Ævar Arnfjörð BjarmasonMar 19, 2019
  11. 0/4 multi-pack-index: fix verify on large reposJeff Hostetler via GitGitGadget, Mar 19, 2019
  12. 1/4 progress: add sparse mode to force 100% complete messageJeff Hostetler via GitGitGadget, Mar 19, 2019
  13. Eric SunshineMar 19, 2019
  14. Jeff HostetlerMar 19, 2019
  15. Eric SunshineMar 19, 2019
  16. 3/4 midx: verify: add midx packfiles to the packed_git listJeff Hostetler via GitGitGadget, Mar 19, 2019
  17. Jeff HostetlerMar 19, 2019
  18. 2/4 trace2:data: add trace2 data to midxJeff Hostetler via GitGitGadget, Mar 19, 2019
  19. 4/4 midx: verify: group objects by packfile to speed up object verificationJeff Hostetler via GitGitGadget, Mar 19, 2019
  20. 0/4 multi-pack-index: fix verify on large reposJeff Hostetler via GitGitGadget, Mar 21, 2019
  21. 2/4 trace2:data: add trace2 data to midxJeff Hostetler via GitGitGadget, Mar 21, 2019
  22. 3/4 midx: add progress indicators in multi-pack-index verifyJeff Hostetler via GitGitGadget, Mar 21, 2019
  23. 1/4 progress: add sparse mode to force 100% complete messageJeff Hostetler via GitGitGadget, Mar 21, 2019
  24. 4/4 midx: during verify group objects by packfile to speed verificationJeff Hostetler via GitGitGadget, Mar 21, 2019
  25. Junio C HamanoMar 22, 2019
  26. Jeff HostetlerMar 25, 2019

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.