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

Re: git-index-pack really does suck..

From
Linus Torvalds <torvalds@linux-foundation.org>
Date
Apr 3, 2007, 20:18 UTC
Message-ID
<Pine.LNX.4.64.0704031304420.6730@woody.linux-foundation.org>
In-Reply-To
<db69205d0704031227q1009eabfhdd82aa3636f25bb6@mail.gmail.com>
On Tue, 3 Apr 2007, Chris Lee wrote:
Show 21 quoted lines
> 
> There's another issue here.
> 
> I'm running git-index-pack as part of a workflow like so:
> 
> $ git-verify-pack -v .git/objects/pack/*.idx > /tmp/all-objects
> $ grep 'blob' /tmp/all-objects > /tmp/blob-objects
> $ cat /tmp/blob-objects | awk '{print $1;}' | git-pack-objects
> --delta-base-offset --all-progress --stdout > blob.pack
> $ git-index-pack -v blob.pack
> 
> Now, when I run 'git-index-pack' on blob.pack in the current
> directory, memory usage is pretty horrific (even with the applied
> patch to not leak all everything). Shawn tells me that index-pack
> should only be decompressing the object twice - once from the repo and
> once from blob.pack - iff I call git-index-pack with --stdin, which I
> am not.
> 
> If I move the blob.pack into /tmp, and run git-index-pack on it there,
> it completes much faster and the memory usage never exceeds 200MB.
> (Inside the repo, it takes up over 3G of RES according to top.)

Yeah. What happens is that inside the repo, because we do all the duplicate object checks (verifying that there are no evil hash collisions) even after fixing the memory leak, we end up keeping *track* of all those objects.

And with a large repository, it's quite the expensive operation.

That whole "verify no SHA1 hash collision" code is really pretty damn paranoid. Maybe we shouldn't have it enabled by default.

So how about this updated patch? We could certainly make "git pull" imply "--paranoid" if we want to, but even that is likely pretty unnecessary. It's not like anybody has ever shown a SHA1 collision, and if the *local* repository is corrupt (and has an object with the wrong SHA1 - that's what the testsuite checks for), then it's probably good to get the valid object from the remote..

This includes the previous one-liner, but also adds the "--paranoid" flag and fixes up the Documentation and tests to match.

Junio, your choice, but regardless which one you choose:
	Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Thanks,
		Linus
---
 Documentation/git-index-pack.txt |    3 +--
 index-pack.c                     |    6 +++++-
 t/t5300-pack-object.sh           |    4 ++--
 3 files changed, 8 insertions(+), 5 deletions(-)
diff --git a/Documentation/git-index-pack.txt b/Documentation/git-index-pack.txt
index 2229ee8..7d8d33b 100644
--- a/Documentation/git-index-pack.txt
+++ b/Documentation/git-index-pack.txt
@@ -8,8 +8,7 @@ git-index-pack - Build pack index file for an existing packed archive
 
 SYNOPSIS
 --------
-'git-index-pack' [-v] [-o <index-file>] <pack-file>
-'git-index-pack' --stdin [--fix-thin] [--keep] [-v] [-o <index-file>] [<pack-file>]
+'git-index-pack' [--stdin [--fix-thin] [--keep]] [-v] [--paranoid] [-o <index-file>] <pack-file>
 
 
 DESCRIPTION
diff --git a/index-pack.c b/index-pack.c
index 6284fe3..8a4c27a 100644
--- a/index-pack.c
+++ b/index-pack.c
@@ -45,6 +45,7 @@ static int nr_resolved_deltas;
 
 static int from_stdin;
 static int verbose;
+static int paranoid;
 
 static volatile sig_atomic_t progress_update;
 
@@ -348,7 +349,7 @@ static void sha1_object(const void *data, unsigned long size,
 			enum object_type type, unsigned char *sha1)
 {
 	hash_sha1_file(data, size, typename(type), sha1);
-	if (has_sha1_file(sha1)) {
+	if (paranoid && has_sha1_file(sha1)) {
 		void *has_data;
 		enum object_type has_type;
 		unsigned long has_size;
@@ -358,6 +359,7 @@ static void sha1_object(const void *data, unsigned long size,
 		if (size != has_size || type != has_type ||
 		    memcmp(data, has_data, size) != 0)
 			die("SHA1 COLLISION FOUND WITH %s !", sha1_to_hex(sha1));
+		free(has_data);
 	}
 }
 
@@ -839,6 +841,8 @@ int main(int argc, char **argv)
 		if (*arg == '-') {
 			if (!strcmp(arg, "--stdin")) {
 				from_stdin = 1;
+			} else if (!strcmp(arg, "--paranoid")) {
+				paranoid = 1;
 			} else if (!strcmp(arg, "--fix-thin")) {
 				fix_thin_pack = 1;
 			} else if (!strcmp(arg, "--keep")) {
diff --git a/t/t5300-pack-object.sh b/t/t5300-pack-object.sh
index 35e036a..407c71e 100755
--- a/t/t5300-pack-object.sh
+++ b/t/t5300-pack-object.sh
@@ -262,7 +262,7 @@ test_expect_success \
 		.git/objects/c8/2de19312b6c3695c0c18f70709a6c535682a67'
 
 test_expect_failure \
-    'make sure index-pack detects the SHA1 collision' \
-    'git-index-pack -o bad.idx test-3.pack'
+    'make sure index-pack detects the SHA1 collision when paranoid' \
+    'git-index-pack --paranoid -o bad.idx test-3.pack'
 
 test_done
Previous: Chris LeeNext: Nicolas Pitre
Message 8 of 58 in “git-index-pack really does suck..”
  1. Linus TorvaldsApr 3, 2007
  2. Linus TorvaldsApr 3, 2007
  3. Nicolas PitreApr 3, 2007
  4. Nicolas PitreApr 3, 2007
  5. Chris LeeApr 3, 2007
  6. Nicolas PitreApr 3, 2007
  7. Chris LeeApr 3, 2007
  8. Linus TorvaldsApr 3, 2007
  9. Nicolas PitreApr 3, 2007
  10. Junio C HamanoApr 3, 2007
  11. Linus TorvaldsApr 3, 2007
  12. Nicolas PitreApr 3, 2007
  13. Chris LeeApr 3, 2007
  14. Linus TorvaldsApr 3, 2007
  15. Linus TorvaldsApr 3, 2007
  16. Shawn O. PearceApr 3, 2007
  17. Linus TorvaldsApr 3, 2007
  18. Shawn O. PearceApr 3, 2007
  19. Linus TorvaldsApr 3, 2007
  20. Linus TorvaldsApr 3, 2007
  21. Junio C HamanoApr 3, 2007
  22. Shawn O. PearceApr 3, 2007
  23. Junio C HamanoApr 3, 2007
  24. 1/2 git-fetch--tool pick-rrefJunio C Hamano, Apr 5, 2007
  25. 2/2 git-fetch: use fetch--tool pick-rref to avoid local fetch from alternateJunio C Hamano, Apr 5, 2007
  26. Shawn O. PearceApr 5, 2007
  27. Junio C HamanoApr 5, 2007
  28. Nicolas PitreApr 3, 2007
  29. Shawn O. PearceApr 3, 2007
  30. Junio C HamanoApr 3, 2007
  31. Shawn O. PearceApr 3, 2007
  32. Jeff KingApr 3, 2007
  33. Dana HowApr 3, 2007
  34. Linus TorvaldsApr 3, 2007
  35. David LangApr 3, 2007
  36. Nicolas PitreApr 3, 2007
  37. Nicolas PitreApr 3, 2007
  38. Linus TorvaldsApr 3, 2007
  39. Nicolas PitreApr 3, 2007
  40. Shawn O. PearceApr 3, 2007
  41. Linus TorvaldsApr 3, 2007
  42. Nicolas PitreApr 3, 2007
  43. Junio C HamanoApr 3, 2007
  44. Shawn O. PearceApr 3, 2007
  45. Nicolas PitreApr 3, 2007
  46. Linus TorvaldsApr 3, 2007
  47. Nicolas PitreApr 3, 2007
  48. David LangApr 3, 2007
  49. Alex RiesenApr 4, 2007
  50. David LangApr 6, 2007
  51. Junio C HamanoApr 6, 2007
  52. Junio C HamanoApr 6, 2007
  53. David LangApr 6, 2007
  54. Junio C HamanoApr 6, 2007
  55. David LangApr 6, 2007
  56. Linus TorvaldsApr 3, 2007
  57. Junio C HamanoApr 3, 2007
  58. Nicolas PitreApr 3, 2007

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.