From: Jeff King <peff@peff.net>
To: "René Scharfe" <l.s.r@web.de>
Cc: Git List <git@vger.kernel.org>, Junio C Hamano <gitster@pobox.com>
Subject: Re: [PATCH 3/3] packfile: use oidset for bad objects
Date: Sat, 11 Sep 2021 10:26:43 -0400 [thread overview]
Message-ID: <YTy8o3qXvyVv8fNr@coredump.intra.peff.net> (raw)
In-Reply-To: <14d48124-d8bb-aa34-aad0-4203d699e17e@web.de>
On Sat, Sep 11, 2021 at 10:01:40AM +0200, René Scharfe wrote:
> Store the object ID of broken pack entries in an oidset instead of
> keeping only their hashes in an unsorted array. The resulting code is
> shorter and easier to read. It also handles the (hopefully) very rare
> case of having a high number of bad objects better.
Yay, I'm very happy to see this kind of cleanup replacing ad hoc data
structures with well-tested ones.
> @@ -303,15 +304,9 @@ static int nth_midxed_pack_entry(struct repository *r,
> if (!is_pack_valid(p))
> return 0;
>
> - if (p->num_bad_objects) {
> - uint32_t i;
> - struct object_id oid;
> - nth_midxed_object_oid(&oid, m, pos);
> - for (i = 0; i < p->num_bad_objects; i++)
> - if (hasheq(oid.hash,
> - p->bad_object_sha1 + the_hash_algo->rawsz * i))
> - return 0;
> - }
> + nth_midxed_object_oid(&oid, m, pos);
> + if (oidset_contains(&p->bad_objects, &oid))
> + return 0;
Calling nth_midxed_object_oid() implies a memcpy() under the hood. In
the old code, we'd skip that in the common case that we had no corrupt
objects, but now we'll pay the cost regardless. memcpy() isn't _that_
expensive, but I'd expect this to be a relatively hot code path.
Is it worth sticking all of this inside:
if (oidset_size(&p->bad_objects))
?
> diff --git a/packfile.c b/packfile.c
> index 04080a558b..8f6d1d6328 100644
> --- a/packfile.c
> +++ b/packfile.c
> @@ -1163,29 +1163,17 @@ int unpack_object_header(struct packed_git *p,
>
> void mark_bad_packed_object(struct packed_git *p, const struct object_id *oid)
> {
> - unsigned i;
> - const unsigned hashsz = the_hash_algo->rawsz;
> - for (i = 0; i < p->num_bad_objects; i++)
> - if (hasheq(oid->hash, p->bad_object_sha1 + hashsz * i))
> - return;
I cringed at the hasheq() and hashcpy() calls in the earlier patches.
Happy to see them go away now. :)
> - p->bad_object_sha1 = xrealloc(p->bad_object_sha1,
> - st_mult(GIT_MAX_RAWSZ,
> - st_add(p->num_bad_objects, 1)));
> - hashcpy(p->bad_object_sha1 + hashsz * p->num_bad_objects, oid->hash);
> - p->num_bad_objects++;
> + oidset_insert(&p->bad_objects, oid);
> }
So now marking a bad object is a one-liner. We _could_ just inline it
at the callers, but I like keeping the implementation abstract.
> const struct packed_git *has_packed_and_bad(struct repository *r,
> const struct object_id *oid)
> {
> struct packed_git *p;
> - unsigned i;
>
> for (p = r->objects->packed_git; p; p = p->next)
> - for (i = 0; i < p->num_bad_objects; i++)
> - if (hasheq(oid->hash,
> - p->bad_object_sha1 + the_hash_algo->rawsz * i))
> - return p;
> + if (oidset_contains(&p->bad_objects, oid))
> + return p;
> return NULL;
> }
Not related to your patch, but I noticed how terribly inefficient this
function could be in a repo with a lot of packs. But we only call it
once in the error case right before we die(), so a linear scan is no
problem.
> @@ -2016,13 +2004,8 @@ static int fill_pack_entry(const struct object_id *oid,
> {
> off_t offset;
>
> - if (p->num_bad_objects) {
> - unsigned i;
> - for (i = 0; i < p->num_bad_objects; i++)
> - if (hasheq(oid->hash,
> - p->bad_object_sha1 + the_hash_algo->rawsz * i))
> - return 0;
> - }
> + if (oidset_contains(&p->bad_objects, oid))
> + return 0;
And this one (and the previous) have the oid already, so they don't have
to worry about optimizing the is-it-empty check first.
-Peff
next prev parent reply other threads:[~2021-09-11 14:26 UTC|newest]
Thread overview: 28+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-09-11 7:50 [PATCH 0/3] packfile: use oidset for bad objects René Scharfe
2021-09-11 7:59 ` [PATCH 1/3] packfile: convert mark_bad_packed_object() to object_id René Scharfe
2021-09-11 8:00 ` [PATCH 2/3] packfile: convert has_packed_and_bad() " René Scharfe
2021-09-11 8:01 ` [PATCH 3/3] packfile: use oidset for bad objects René Scharfe
2021-09-11 14:26 ` Jeff King [this message]
2021-09-11 16:08 ` René Scharfe
2021-09-11 17:03 ` Jeff King
2021-09-11 17:16 ` René Scharfe
2021-09-11 14:27 ` [PATCH 0/3] " Jeff King
2021-09-11 16:08 ` [PATCH 4/3] midx: inline nth_midxed_pack_entry() René Scharfe
2021-09-11 17:03 ` René Scharfe
2021-09-11 17:07 ` Jeff King
2021-09-11 20:31 ` René Scharfe
2021-09-11 21:20 ` Jeff King
2021-09-11 23:39 ` René Scharfe
2021-09-11 20:31 ` [PATCH v2 0/5] packfile: use oidset for bad objects René Scharfe
2021-09-11 20:36 ` [PATCH 1/5] oidset: make oidset_size() an inline function René Scharfe
2021-09-11 20:39 ` [PATCH 2/5] midx: inline nth_midxed_pack_entry() René Scharfe
2021-09-11 20:40 ` [PATCH 3/5] packfile: convert mark_bad_packed_object() to object_id René Scharfe
2021-09-11 20:42 ` [PATCH v2 4/5] packfile: convert has_packed_and_bad() " René Scharfe
2021-09-11 20:43 ` [PATCH v2 5/5] packfile: use oidset for bad objects René Scharfe
2021-09-11 20:45 ` [PATCH v2 0/5] " René Scharfe
2021-09-11 21:22 ` Jeff King
2021-09-11 23:59 ` Derrick Stolee
2021-09-12 1:51 ` Taylor Blau
2021-09-12 2:29 ` Taylor Blau
2021-09-12 3:51 ` Derrick Stolee
2021-09-12 4:01 ` Taylor Blau
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
List information: http://vger.kernel.org/majordomo-info.html
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=YTy8o3qXvyVv8fNr@coredump.intra.peff.net \
--to=peff@peff.net \
--cc=git@vger.kernel.org \
--cc=gitster@pobox.com \
--cc=l.s.r@web.de \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
Code repositories for project(s) associated with this public inbox
https://80x24.org/mirrors/git.git
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).