git.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Junio C Hamano <gitster@pobox.com>
To: Jeff King <peff@peff.net>
Cc: Johan Herland <johan@herland.net>,
	Michael Haggerty <mhagger@alum.mit.edu>,
	git@vger.kernel.org
Subject: Re: [PATCH 00/17] Remove assumptions about refname lifetimes
Date: Mon, 20 May 2013 11:03:33 -0700	[thread overview]
Message-ID: <7vy5b9v896.fsf@alter.siamese.dyndns.org> (raw)
In-Reply-To: <20130520165942.GB27788@sigill.intra.peff.net> (Jeff King's message of "Mon, 20 May 2013 12:59:42 -0400")

Jeff King <peff@peff.net> writes:

> But more importantly, it introduces contention between two unrelated
> refs that are being updated. Even if we reconcile the differences
> automatically (e.g., with a "merge-and-retry" strategy), that is going
> to be a serious performance regression for a busy repository, as we
> repeatedly try to reconcile the serialized updates to the refs/ root
> tree.

I think we are on the same page.

> Any transactional solution needs to have the equivalent of ref-level
> locking (i.e., row-level locking in a relational setting).

Not necessarily if we can exploit assumptions such as deletion is
far rare compared to update and creation which in turn are far rare
compared to lookup.  I've been wondering if we can find a cheap
reader-writer lock mechanism, use a single instance of it to govern
the whole repository, and have everybody but ref deleters and "git
pack-ref" take the read side of the lock.

Then only while ref deletion or ref packing is going on, everybody
need to stall, but otherwise the most common "read or update
recently touched refs (aka loose refs)" will be taking the reader
side of that cheap read-write lock and reading a single loose ref
file.

Perhaps such a "cheap on reader side" reader-write lock is hard to
come by?

  parent reply	other threads:[~2013-05-20 18:03 UTC|newest]

Thread overview: 42+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2013-05-19 20:26 [PATCH 00/17] Remove assumptions about refname lifetimes Michael Haggerty
2013-05-19 20:26 ` [PATCH 01/17] describe: make own copy of refname Michael Haggerty
2013-05-19 20:26 ` [PATCH 02/17] fetch: make own copies of refnames Michael Haggerty
2013-05-19 20:26 ` [PATCH 03/17] add_rev_cmdline(): make a copy of the name argument Michael Haggerty
2013-05-19 20:26 ` [PATCH 04/17] builtin_diff_tree(): make it obvious that function wants two entries Michael Haggerty
2013-05-21 17:27   ` Junio C Hamano
2013-05-23  7:19     ` Michael Haggerty
2013-05-19 20:27 ` [PATCH 05/17] cmd_diff(): use an object_array for holding trees Michael Haggerty
2013-05-21 17:30   ` Junio C Hamano
2013-05-23  7:21     ` Michael Haggerty
2013-05-19 20:27 ` [PATCH 06/17] cmd_diff(): rename local variable "list" -> "entry" Michael Haggerty
2013-05-19 20:27 ` [PATCH 07/17] cmd_diff(): make it obvious which cases are exclusive of each other Michael Haggerty
2013-05-19 20:27 ` [PATCH 08/17] revision: split some overly-long lines Michael Haggerty
2013-05-21 17:34   ` Junio C Hamano
2013-05-23  6:27     ` Michael Haggerty
2013-05-23 17:08       ` Junio C Hamano
2013-05-19 20:27 ` [PATCH 09/17] gc_boundary(): move the check "alloc <= nr" to caller Michael Haggerty
2013-05-21 17:49   ` Junio C Hamano
2013-05-23  7:09     ` Michael Haggerty
2013-05-23 18:02       ` Junio C Hamano
2013-05-19 20:27 ` [PATCH 10/17] get_revision_internal(): make check less mysterious Michael Haggerty
2013-05-21 17:38   ` Junio C Hamano
2013-05-23  6:39     ` Michael Haggerty
2013-05-19 20:27 ` [PATCH 11/17] object_array: add function object_array_filter() Michael Haggerty
2013-05-19 20:27 ` [PATCH 12/17] object_array_remove_duplicates(): rewrite to reduce copying Michael Haggerty
2013-05-19 20:27 ` [PATCH 13/17] fsck: don't put a void*-shaped peg in a char*-shaped hole Michael Haggerty
2013-05-19 20:27 ` [PATCH 14/17] find_first_merges(): initialize merges variable using initializer Michael Haggerty
2013-05-19 20:27 ` [PATCH 15/17] find_first_merges(): remove unnecessary code Michael Haggerty
2013-05-19 20:27 ` [RFC 16/17] object_array_entry: copy name before storing in name field Michael Haggerty
2013-05-20 10:33   ` Johan Herland
2013-05-20 14:42     ` Michael Haggerty
2013-05-20 16:44       ` Jeff King
2013-05-20 21:34         ` Michael Haggerty
2013-05-19 20:27 ` [RFC 17/17] refs: document the lifetime of the refname passed to each_ref_fn Michael Haggerty
2013-05-20 10:28 ` [PATCH 00/17] Remove assumptions about refname lifetimes Johan Herland
2013-05-20 12:15   ` Michael Haggerty
2013-05-20 16:37   ` Junio C Hamano
2013-05-20 16:59     ` Jeff King
2013-05-20 17:08       ` Johan Herland
2013-05-20 18:03       ` Junio C Hamano [this message]
2013-05-20 17:03     ` Johan Herland
2013-05-21 18:39       ` 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

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=7vy5b9v896.fsf@alter.siamese.dyndns.org \
    --to=gitster@pobox.com \
    --cc=git@vger.kernel.org \
    --cc=johan@herland.net \
    --cc=mhagger@alum.mit.edu \
    --cc=peff@peff.net \
    /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.
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 NNTP newsgroup(s).