commit 465971eec96aef0fcae09797cf38a9f3c2e8cc6a
from: Stefan Sperling <stsp@stsp.name>
date: Tue Mar 24 15:32:54 2020 UTC

stop verifying pack file checksum in the main process during clone/fetch

Both got-fetch-pack and got-index-pack now verify the pack file checksum.
This means we can avoid reading the entire pack file in the main process
just to verify its checksum and avoid a noticable stall between fetching
and indexing on slow machines.

commit - 3abebdc26faf6005af348fa5146420b7cca61437
commit + 465971eec96aef0fcae09797cf38a9f3c2e8cc6a
blob - bc538754a1763f2c475adf6b02dd67afde0541ef
blob + 4ad71dd6e0d4b14735e7bc5eb66696a792e2d6a4
--- lib/fetch.c
+++ lib/fetch.c
@@ -344,49 +344,7 @@ done:
 	}
 	return err;
 }
-
-static const struct got_error *
-check_pack_hash(int fd, size_t sz, uint8_t *hcomp)
-{
-	SHA1_CTX ctx;
-	uint8_t hexpect[SHA1_DIGEST_LENGTH];
-	uint8_t buf[32 * 1024];
-	ssize_t n, r, nr;
 
-	if (sz < sizeof(struct got_packfile_hdr) + SHA1_DIGEST_LENGTH)
-		return got_error_msg(GOT_ERR_BAD_PACKFILE, "short packfile");
-
-	n = 0;
-	SHA1Init(&ctx);
-	while (n < sz - 20) {
-		nr = sizeof(buf);
-		if (sz - n - 20 < sizeof(buf))
-			nr = sz - n - 20;
-		r = read(fd, buf, nr);
-		if (r == -1)
-			return got_error_from_errno("read");
-		if (r != nr)
-			return got_error_msg(GOT_ERR_BAD_PACKFILE,
-			    "short pack file");
-		SHA1Update(&ctx, buf, nr);
-		n += r;
-	}
-	SHA1Final(hcomp, &ctx);
-
-	r = read(fd, hexpect, sizeof(hexpect));
-	if (r == -1)
-		return got_error_from_errno("read");
-	if (r != sizeof(hexpect))
-		return got_error_msg(GOT_ERR_BAD_PACKFILE,
-		    "short pack file");
-
-	if (memcmp(hcomp, hexpect, SHA1_DIGEST_LENGTH) != 0)
-		return got_error_msg(GOT_ERR_BAD_PACKFILE,
-		    "packfile checksum mismatch");
-
-	return NULL;
-}
-
 const struct got_error*
 got_fetch_pack(struct got_object_id **pack_hash, struct got_pathlist_head *refs,
     struct got_pathlist_head *symrefs, const char *remote_name,
@@ -738,17 +696,8 @@ got_fetch_pack(struct got_object_id **pack_hash, struc
 	 * references in our repository. The caller will take care of that.
 	 */
 	if (nobj == 0)
-		goto done;
-
-	if (lseek(packfd, 0, SEEK_SET) == -1) {
-		err = got_error_from_errno("lseek");
 		goto done;
-	}
 
-	err = check_pack_hash(packfd, packfile_size, (*pack_hash)->sha1);
-	if (err)
-		goto done;
-
 	if (lseek(packfd, 0, SEEK_SET) == -1) {
 		err = got_error_from_errno("lseek");
 		goto done;