From: Brandon Williams <bmwill@google.com>
To: Jeff King <peff@peff.net>
Cc: Michael Haggerty <mhagger@alum.mit.edu>,
Junio C Hamano <gitster@pobox.com>,
s@kazlauskas.me, git@vger.kernel.org
Subject: Re: [PATCH] files-backend: cheapen refname_available check when locking refs
Date: Thu, 17 Aug 2017 10:56:52 -0700 [thread overview]
Message-ID: <20170817175652.GB109680@google.com> (raw)
In-Reply-To: <20170817152240.coioktoqfkcvxldj@sigill.intra.peff.net>
On 08/17, Jeff King wrote:
> On Thu, Aug 17, 2017 at 05:12:50PM +0200, Michael Haggerty wrote:
>
> > I was testing this using the reporter's recipe (but fetching from a
> > local clone), and found the following surprising timing numbers:
> >
> > b05855b5bc (before the slowdown): 22.7 s
> > 524a9fdb51 (immediately after the slowdown): 13 minutes
> > 4e81f1ecf1 (after this fix): 14.5 s
> >
> > The fact that the fetch is now significantly *faster* than before the
> > slowdown seems not to have anything to do with the reference code.
>
> I bisected this (with some hackery, since the commits in the middle all
> take 13 minutes to run). The other speedup is indeed unrelated, and is
> due to Brandon's aacc5c1a81 (submodule: refactor logic to determine
> changed submodules, 2017-05-01).
>
> The commit message doesn't mention performance (it's mostly about code
> reduction). I think the speedup comes from using
> diff_tree_combined_merge() instead of manually diffing each commit
> against its parents. But I didn't do further timings to verify that (I'm
> reporting it here mostly as an interesting curiosity for submodule
> folks).
Haha always great to see an unintended improvement in performance! Yeah
that commit was mostly about removing duplicate code but I'm glad that
it ended up being a benefit to perf too.
>
> > diff --git a/refs/files-backend.c b/refs/files-backend.c
> > index e9b95592b6..f2a420c611 100644
> > --- a/refs/files-backend.c
> > +++ b/refs/files-backend.c
> > @@ -631,11 +631,11 @@ static int lock_raw_ref(struct files_ref_store *refs,
> >
> > /*
> > * If the ref did not exist and we are creating it,
> > - * make sure there is no existing ref that conflicts
> > - * with refname:
> > + * make sure there is no existing packed ref that
> > + * conflicts with refname:
> > */
> > if (refs_verify_refname_available(
> > - &refs->base, refname,
> > + refs->packed_ref_store, refname,
> > extras, skip, err))
> > goto error_return;
> > }
>
> This seems too easy to be true. :) But I think it matches what we were
> doing before 524a9fdb51 (so it's correct), and the performance numbers
> don't lie.
>
> -Peff
--
Brandon Williams
next prev parent reply other threads:[~2017-08-17 17:56 UTC|newest]
Thread overview: 6+ messages / expand[flat|nested] mbox.gz Atom feed top
2017-07-09 10:25 Fetching new refs gets progressively slower s
2017-07-09 11:29 ` Jeff King
2017-08-17 15:12 ` [PATCH] files-backend: cheapen refname_available check when locking refs Michael Haggerty
2017-08-17 15:22 ` Jeff King
2017-08-17 17:56 ` Brandon Williams [this message]
2017-08-17 21:37 ` Junio C Hamano
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=20170817175652.GB109680@google.com \
--to=bmwill@google.com \
--cc=git@vger.kernel.org \
--cc=gitster@pobox.com \
--cc=mhagger@alum.mit.edu \
--cc=peff@peff.net \
--cc=s@kazlauskas.me \
/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).