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

[PATCH 07/10] pack-check: push oid lookup into loop

From
Jeff King <peff@peff.net>
Date
Feb 24, 2020, 04:36 UTC
Message-ID
<20200224043631.GG1018190@coredump.intra.peff.net>
In-Reply-To
<20200224042625.GA1015553@coredump.intra.peff.net>

When we're checking a pack with fsck or verify-pack, we first sort the idx entries by offset, since accessing them in pack order is more efficient. To do so, we loop over them and fill in an array of structs with the offset, object_id, and index position of each, sort the result, and only then do we iterate over the sorted array and process each entry.

In order to avoid the memory cost of storing the hash of each object, we just store a pointer into the copy in the mmap'd pack index file. To keep that property even as the rest of the code converted to "struct object_id", commit 9fd750461b (Convert the verify_pack callback to struct object_id, 2017-05-06) introduced a union in order to type-pun the pointer-to-hash into an object_id struct.

But we can make this even simpler by observing that the sort operation doesn't need the object id at all! We only need them one at a time while we actually process each entry. So we can just omit the oid from the struct entirely and load it on the fly into a local variable in the second loop.

This gets rid of the type-punning, and lets us directly use the more type-safe nth_packed_object_id(), simplifying the code. And as a bonus, it saves 8 bytes of memory per object.

Note that this does mean we'll do the offset lookup for each object before the oid lookup. The oid lookup has more safety checks in it (e.g., for looking past p->num_objects) which in theory protected the offset lookup. But since violating those checks was already a BUG() condition (as described in the previous commit), it's not worth worrying about.

Signed-off-by: Jeff King <peff@peff.net>
---
 pack-check.c | 23 ++++++++++-------------
 1 file changed, 10 insertions(+), 13 deletions(-)
diff --git a/pack-check.c b/pack-check.c
index 39196ecfbc..dad6d8ae7f 100644
--- a/pack-check.c
+++ b/pack-check.c
@@ -8,10 +8,6 @@
 
 struct idx_entry {
 	off_t                offset;
-	union idx_entry_object {
-		const unsigned char *hash;
-		struct object_id *oid;
-	} oid;
 	unsigned int nr;
 };
 
@@ -97,30 +93,31 @@ static int verify_packfile(struct repository *r,
 	entries[nr_objects].offset = pack_sig_ofs;
 	/* first sort entries by pack offset, since unpacking them is more efficient that way */
 	for (i = 0; i < nr_objects; i++) {
-		entries[i].oid.hash = nth_packed_object_sha1(p, i);
-		if (!entries[i].oid.hash)
-			BUG("unable to get oid of object %lu from %s",
-			    (unsigned long)i, p->pack_name);
 		entries[i].offset = nth_packed_object_offset(p, i);
 		entries[i].nr = i;
 	}
 	QSORT(entries, nr_objects, compare_entries);
 
 	for (i = 0; i < nr_objects; i++) {
 		void *data;
+		struct object_id oid;
 		enum object_type type;
 		unsigned long size;
 		off_t curpos;
 		int data_valid;
 
+		if (nth_packed_object_id(&oid, p, entries[i].nr) < 0)
+			BUG("unable to get oid of object %lu from %s",
+			    (unsigned long)entries[i].nr, p->pack_name);
+
 		if (p->index_version > 1) {
 			off_t offset = entries[i].offset;
 			off_t len = entries[i+1].offset - offset;
 			unsigned int nr = entries[i].nr;
 			if (check_pack_crc(p, w_curs, offset, len, nr))
 				err = error("index CRC mismatch for object %s "
 					    "from %s at offset %"PRIuMAX"",
-					    oid_to_hex(entries[i].oid.oid),
+					    oid_to_hex(&oid),
 					    p->pack_name, (uintmax_t)offset);
 		}
 
@@ -143,14 +140,14 @@ static int verify_packfile(struct repository *r,
 
 		if (data_valid && !data)
 			err = error("cannot unpack %s from %s at offset %"PRIuMAX"",
-				    oid_to_hex(entries[i].oid.oid), p->pack_name,
+				    oid_to_hex(&oid), p->pack_name,
 				    (uintmax_t)entries[i].offset);
-		else if (check_object_signature(r, entries[i].oid.oid, data, size, type_name(type)))
+		else if (check_object_signature(r, &oid, data, size, type_name(type)))
 			err = error("packed %s from %s is corrupt",
-				    oid_to_hex(entries[i].oid.oid), p->pack_name);
+				    oid_to_hex(&oid), p->pack_name);
 		else if (fn) {
 			int eaten = 0;
-			err |= fn(entries[i].oid.oid, type, size, data, &eaten);
+			err |= fn(&oid, type, size, data, &eaten);
 			if (eaten)
 				data = NULL;
 		}
-- 
2.25.1.823.g95c5488cf7
Previous: Jeff KingNext: Jeff King
Message 10 of 14 in “removing nth_packed_object_sha1()”
  1. 0/10 removing nth_packed_object_sha1()Jeff King, Feb 24, 2020
  2. 01/10 nth_packed_object_oid(): use customary integer returnJeff King, Feb 24, 2020
  3. Jeff KingFeb 24, 2020
  4. 02/10 pack-objects: read delta base oid into object_id structJeff King, Feb 24, 2020
  5. brian m. carlsonFeb 25, 2020
  6. 03/10 pack-objects: convert oe_set_delta_ext() to use object_idJeff King, Feb 24, 2020
  7. 04/10 pack-objects: use object_id struct in pack-reuse codeJeff King, Feb 24, 2020
  8. 05/10 pack-bitmap: use object_id when loading on-disk bitmapsJeff King, Feb 24, 2020
  9. 06/10 pack-check: convert "internal error" die to a BUG()Jeff King, Feb 24, 2020
  10. 07/10 pack-check: push oid lookup into loopJeff King, Feb 24, 2020
  11. 08/10 packed_object_info(): use object_id for returning delta baseJeff King, Feb 24, 2020
  12. 09/10 packed_object_info(): use object_id internally for delta baseJeff King, Feb 24, 2020
  13. 10/10 packfile: drop nth_packed_object_sha1()Jeff King, Feb 24, 2020
  14. brian m. carlsonFeb 25, 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.