git.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Ronnie Sahlberg <sahlberg@google.com>
To: git@vger.kernel.org
Cc: Ronnie Sahlberg <sahlberg@google.com>
Subject: [PATCH v4 3/3] refs.c: change ref_transaction_commit to run the commit loops once all work is finished
Date: Mon, 14 Apr 2014 11:29:23 -0700	[thread overview]
Message-ID: <1397500163-7617-4-git-send-email-sahlberg@google.com> (raw)
In-Reply-To: <1397500163-7617-1-git-send-email-sahlberg@google.com>

During a transaction commit we will both update and delete refs.
Since both update and delete now use the same pattern

    lock = lock_ref_sha1_basic() (or varient of)
    write_ref_sha1(lock)/delete_ref_loose(lock)
    unlock_ref(lock) | commit_ref_lock(lock)

we can now simplify ref_transaction_commit to have one loop that locks all
involved refs.
A second loop that writes or flags for deletion, but does not commit, all
the refs.
And a final third loop that commits all the refs once all the work and
preparations are complete.

This makes updating/deleting multiple refs more atomic since we will not start
the commit phase until all the preparations have completed successfully.

Signed-off-by: Ronnie Sahlberg <sahlberg@google.com>
---
 refs.c | 39 ++++++++++++++++++++++-----------------
 1 file changed, 22 insertions(+), 17 deletions(-)

diff --git a/refs.c b/refs.c
index a14addb..87193c7 100644
--- a/refs.c
+++ b/refs.c
@@ -3467,42 +3467,47 @@ int ref_transaction_commit(struct ref_transaction *transaction,
 		}
 	}
 
-	/* Perform updates first so live commits remain referenced */
+	/* Prepare all the updates/deletes */
 	for (i = 0; i < n; i++) {
 		struct ref_update *update = updates[i];
 
-		if (!is_null_sha1(update->new_sha1)) {
+		if (!is_null_sha1(update->new_sha1))
 			ret = update_ref_write(msg,
 					       update->refname,
 					       update->new_sha1,
 					       update->lock, onerr);
-			if (ret)
-				unlock_ref(update->lock);
-			else
-				commit_ref_lock(update->lock);
-			update->lock = NULL;
-			if (ret)
-				goto cleanup;
+		else {
+			delnames[delnum++] = update->refname;
+			ret = delete_ref_loose(update->lock, update->type);
 		}
+		if (ret)
+			goto cleanup;
 	}
 
-	/* Perform deletes now that updates are safely completed */
+	ret |= repack_without_refs(delnames, delnum);
+	for (i = 0; i < delnum; i++)
+		unlink_or_warn(git_path("logs/%s", delnames[i]));
+	clear_loose_ref_cache(&ref_cache);
+
+	/* Perform updates first so live commits remain referenced */
+	for (i = 0; i < n; i++) {
+		struct ref_update *update = updates[i];
+
+		if (update->lock && !update->lock->delete_ref) {
+			ret |= commit_ref_lock(update->lock);
+			update->lock = NULL;
+		}
+	}
+	/* And finally perform all deletes */
 	for (i = 0; i < n; i++) {
 		struct ref_update *update = updates[i];
 
 		if (update->lock) {
-			delnames[delnum++] = update->refname;
-			ret |= delete_ref_loose(update->lock, update->type);
 			ret |= commit_ref_lock(update->lock);
 			update->lock = NULL;
 		}
 	}
 
-	ret |= repack_without_refs(delnames, delnum);
-	for (i = 0; i < delnum; i++)
-		unlink_or_warn(git_path("logs/%s", delnames[i]));
-	clear_loose_ref_cache(&ref_cache);
-
 cleanup:
 	for (i = 0; i < n; i++)
 		if (updates[i]->lock)
-- 
1.9.1.505.gd05696d

  parent reply	other threads:[~2014-04-14 18:36 UTC|newest]

Thread overview: 16+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2014-04-14 18:29 [PATCH v4 0/3] Make update refs more atomic Ronnie Sahlberg
2014-04-14 18:29 ` [PATCH v4 1/3] refs.c: split writing and commiting a ref into two separate functions Ronnie Sahlberg
2014-04-15 11:17   ` Michael Haggerty
2014-04-14 18:29 ` [PATCH v4 2/3] refs.c: split delete_ref_loose() into a separate flag-for-deletion and commit phase Ronnie Sahlberg
2014-04-15 17:19   ` Michael Haggerty
2014-04-14 18:29 ` Ronnie Sahlberg [this message]
2014-04-14 20:24 ` [PATCH v4 0/3] Make update refs more atomic Junio C Hamano
2014-04-15 16:41   ` Ronnie Sahlberg
2014-04-15  6:36 ` Michael Haggerty
2014-04-15 16:33   ` Ronnie Sahlberg
2014-04-15 20:32     ` Michael Haggerty
2014-04-16 17:11       ` Ronnie Sahlberg
2014-04-16 19:31         ` Junio C Hamano
2014-04-16 21:31           ` Ronnie Sahlberg
2014-04-16 21:42             ` Junio C Hamano
2014-04-16 21:51           ` Michael Haggerty

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=1397500163-7617-4-git-send-email-sahlberg@google.com \
    --to=sahlberg@google.com \
    --cc=git@vger.kernel.org \
    /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).