From: Christian Couder <christian.couder@gmail.com>
To: Jonathan Tan <jonathantanmy@google.com>
Cc: git <git@vger.kernel.org>, Junio C Hamano <gitster@pobox.com>,
Jeff King <peff@peff.net>,
Christian Couder <chriscool@tuxfamily.org>,
Ramsay Jones <ramsay@ramsayjones.plus.com>
Subject: Re: [RFC PATCH 10/10] pack-objects: improve partial packfile reuse
Date: Fri, 11 Oct 2019 09:39:58 +0200 [thread overview]
Message-ID: <CAP8UFD0yefCapgLG78bqBY0ZNWNiifJZUxQ=MGDa8AVJJrWBFg@mail.gmail.com> (raw)
In-Reply-To: <20191010235952.174426-1-jonathantanmy@google.com>
On Fri, Oct 11, 2019 at 1:59 AM Jonathan Tan <jonathantanmy@google.com> wrote:
>
> I'm going to start with pack-bitmap.h, then builtin/pack-objects.c.
>
> > int reuse_partial_packfile_from_bitmap(struct bitmap_index *,
> > struct packed_git **packfile,
> > - uint32_t *entries, off_t *up_to);
> > + uint32_t *entries,
> > + struct bitmap **bitmap);
>
> Makes sense. The existing code determines if the given packfile is
> suitable, and if yes, the span of the packfile to use. With this patch,
> we resolve to use the given packfile, and want to compute which objects
> to use, storing the computation result in an uncompressed bitmap.
> (Resolving to use the given packfile is not that much different from
> existing behavior, as far as I know, since we currently only consider
> one packfile at most anyway.)
>
> So replacing the out param makes sense, although a more descriptive name
> than "bitmap" would be nice.
Yeah, in pack-bitmap.c this argument is called "reuse_out", so the
same name could be used in pack-bitmap.c too. I changed that on my
current version.
> > +/*
> > + * Record the offsets needed in our reused packfile chunks due to
> > + * "gaps" where we omitted some objects.
> > + */
> > +static struct reused_chunk {
> > + off_t start;
> > + off_t offset;
> > +} *reused_chunks;
> > +static int reused_chunks_nr;
> > +static int reused_chunks_alloc;
>
> This makes sense - offsets may be different when we omit objects from
> the packfile. I think this can be computed by calculating the number of
> zero bits between the current object's index and the nth object prior
> (where n is the offset) in the bitmap resulting from
> reuse_partial_packfile_from_bitmap() above, thus eliminating the need
> for this array, but I haven't tested it.
Thanks for the idea. I will let others comment on that though before
maybe trying to implement it. I think it could come as a further
optimization in a following patch if it makes things faster or reduce
memory usage.
> > + if (0) {
> > + off_t expected_size = cur - offset;
> > +
> > + if (len + ofs_len < expected_size) {
> > + unsigned max_pad = (len >= 4) ? 9 : 5;
> > + header[len - 1] |= 0x80;
> > + while (len < max_pad && len + ofs_len < expected_size)
> > + header[len++] = 0x80;
> > + header[len - 1] &= 0x7F;
> > + }
> > + }
>
> This if(0) should be deleted.
Yeah, good eyes. I removed it on my current version.
> > @@ -1002,6 +1132,10 @@ static int have_duplicate_entry(const struct object_id *oid,
> > {
> > struct object_entry *entry;
> >
> > + if (reuse_packfile_bitmap &&
> > + bitmap_walk_contains(bitmap_git, reuse_packfile_bitmap, oid))
> > + return 1;
>
> Hmm...why did we previously not need to check the reuse information, but
> we do now? I gave the code a cursory glance but couldn't find the
> answer.
>
> > @@ -2571,7 +2706,9 @@ static void ll_find_deltas(struct object_entry **list, unsigned list_size,
> >
> > static int obj_is_packed(const struct object_id *oid)
> > {
> > - return !!packlist_find(&to_pack, oid, NULL);
> > + return packlist_find(&to_pack, oid, NULL) ||
> > + (reuse_packfile_bitmap &&
> > + bitmap_walk_contains(bitmap_git, reuse_packfile_bitmap, oid));
>
> Same question here - why do we need to check the reuse information?
Maybe a reuse bitmap makes it cheap enough to check that information?
Thank you for the review,
Christian.
next prev parent reply other threads:[~2019-10-11 7:40 UTC|newest]
Thread overview: 31+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-09-13 13:02 [RFC PATCH 00/10] Rewrite packfile reuse code Christian Couder
2019-09-13 13:02 ` [RFC PATCH 01/10] builtin/pack-objects: report reused packfile objects Christian Couder
2019-09-13 13:02 ` [RFC PATCH 02/10] packfile: expose get_delta_base() Christian Couder
2019-09-13 13:02 ` [RFC PATCH 03/10] ewah/bitmap: introduce bitmap_word_alloc() Christian Couder
2019-09-13 13:02 ` [RFC PATCH 04/10] ewah/bitmap: always allocate 2 more words Christian Couder
2019-10-10 23:40 ` Jonathan Tan
2019-10-11 7:49 ` Christian Couder
2019-10-11 18:05 ` Jeff King
2019-09-13 13:02 ` [RFC PATCH 05/10] pack-bitmap: don't rely on bitmap_git->reuse_objects Christian Couder
2019-10-10 23:44 ` Jonathan Tan
2019-10-11 7:50 ` Christian Couder
2019-09-13 13:02 ` [RFC PATCH 06/10] pack-bitmap: introduce bitmap_walk_contains() Christian Couder
2019-09-13 13:02 ` [RFC PATCH 07/10] csum-file: introduce hashfile_total() Christian Couder
2019-09-13 13:02 ` [RFC PATCH 08/10] pack-objects: introduce pack.allowPackReuse Christian Couder
2019-09-13 21:37 ` Junio C Hamano
2019-09-13 13:02 ` [RFC PATCH 09/10] builtin/pack-objects: introduce obj_is_packed() Christian Couder
2019-09-13 13:02 ` [RFC PATCH 10/10] pack-objects: improve partial packfile reuse Christian Couder
2019-09-13 22:29 ` Junio C Hamano
2019-09-14 2:02 ` Jeff King
2019-09-14 3:06 ` Junio C Hamano
2019-10-02 15:57 ` Jeff King
2019-10-03 2:06 ` Junio C Hamano
2019-10-03 6:55 ` Christian Couder
2019-10-10 23:59 ` Jonathan Tan
2019-10-11 7:39 ` Christian Couder [this message]
2019-10-11 18:01 ` Jeff King
2019-10-11 21:04 ` Jonathan Tan
2019-10-12 0:04 ` Junio C Hamano
2019-10-13 7:38 ` Jeff King
2019-10-17 7:03 ` Junio C Hamano
2019-10-17 7:23 ` Jeff King
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='CAP8UFD0yefCapgLG78bqBY0ZNWNiifJZUxQ=MGDa8AVJJrWBFg@mail.gmail.com' \
--to=christian.couder@gmail.com \
--cc=chriscool@tuxfamily.org \
--cc=git@vger.kernel.org \
--cc=gitster@pobox.com \
--cc=jonathantanmy@google.com \
--cc=peff@peff.net \
--cc=ramsay@ramsayjones.plus.com \
/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).