From: Johannes Schindelin <Johannes.Schindelin@gmx.de>
To: Jeff King <peff@peff.net>
Cc: Johannes Schindelin via GitGitGadget <gitgitgadget@gmail.com>,
git@vger.kernel.org, Junio C Hamano <gitster@pobox.com>
Subject: Re: [PATCH 1/2] rebase: consolidate clean-up code before leaving reset_head()
Date: Fri, 9 Nov 2018 18:13:08 +0100 (STD) [thread overview]
Message-ID: <nycvar.QRO.7.76.6.1811091750330.39@tvgsbejvaqbjf.bet> (raw)
In-Reply-To: <20181109100357.GA7410@sigill.intra.peff.net>
Hi Peff,
On Fri, 9 Nov 2018, Jeff King wrote:
> On Fri, Nov 09, 2018 at 01:34:17AM -0800, Johannes Schindelin via GitGitGadget wrote:
>
> > diff --git a/builtin/rebase.c b/builtin/rebase.c
> > index 0ee06aa363..6f6d7de156 100644
> > --- a/builtin/rebase.c
> > +++ b/builtin/rebase.c
> > @@ -569,16 +569,13 @@ static int reset_head(struct object_id *oid, const char *action,
> > }
> >
> > if (!fill_tree_descriptor(&desc, oid)) {
> > - error(_("failed to find tree of %s"), oid_to_hex(oid));
> > - rollback_lock_file(&lock);
> > - free((void *)desc.buffer);
> > - return -1;
> > + ret = error(_("failed to find tree of %s"), oid_to_hex(oid));
> > + goto leave_reset_head;
> > }
>
> If fill_tree_descriptor() fails, what is left in desc.buffer? Looking at
> the implementation, I think it's always NULL or a valid buffer. But I
> think all code paths actually die() unless we pass a NULL oid (and in
> that case desc.buffer would be NULL, too).
>
> So I think the original here that calls free() doesn't ever do anything
> but it did not hurt. After your patch, the leave_reset_head code would
> continue to call free(), and that's OK.
Right, that was my thinking, too.
> There are a few earlier conditionals in reset_head() that do only
> rollback_lock_file() that could similarly be converted to use the goto.
> But they would need desc.buffer to be initialized to NULL. I could go
> either way on converting them or not.
Whoops. I should have checked more carefully?
> > @@ -586,10 +583,9 @@ static int reset_head(struct object_id *oid, const char *action,
> >
> > if (write_locked_index(the_repository->index, &lock, COMMIT_LOCK) < 0)
> > ret = error(_("could not write index"));
> > - free((void *)desc.buffer);
> >
> > if (ret)
> > - return ret;
> > + goto leave_reset_head;
> >
> > reflog_action = getenv(GIT_REFLOG_ACTION_ENVIRONMENT);
> > strbuf_addf(&msg, "%s: ", reflog_action ? reflog_action : "rebase");
> > @@ -622,7 +618,10 @@ static int reset_head(struct object_id *oid, const char *action,
> > UPDATE_REFS_MSG_ON_ERR);
> > }
> >
> > +leave_reset_head:
> > strbuf_release(&msg);
> > + rollback_lock_file(&lock);
> > + free((void *)desc.buffer);
> > return ret;
>
> We get here on success, too. So we may call rollback_lock_file() on an
> already-committed lock. This is explicitly documented as a no-op by the
> lock code, so that's OK.
Indeed. I did not check the documentation, but the code, and came to the
same conclusion.
> So overall looks good to me.
Thanks!
Dscho
next prev parent reply other threads:[~2018-11-09 17:13 UTC|newest]
Thread overview: 18+ messages / expand[flat|nested] mbox.gz Atom feed top
2018-11-09 9:34 [PATCH 0/2] Fix built-in rebase perf regression Johannes Schindelin via GitGitGadget
2018-11-09 9:34 ` [PATCH 1/2] rebase: consolidate clean-up code before leaving reset_head() Johannes Schindelin via GitGitGadget
2018-11-09 10:03 ` Jeff King
2018-11-09 17:13 ` Johannes Schindelin [this message]
2018-11-09 9:34 ` [PATCH 2/2] built-in rebase: reinstate `checkout -q` behavior where appropriate Johannes Schindelin via GitGitGadget
2018-11-09 10:11 ` Jeff King
2018-11-09 17:21 ` Johannes Schindelin
2018-11-09 19:59 ` Jeff King
2018-11-12 11:11 ` Johannes Schindelin
2018-11-12 4:04 ` Junio C Hamano
2018-11-12 11:42 ` Johannes Schindelin
2018-11-13 1:49 ` Junio C Hamano
2018-11-13 6:07 ` Junio C Hamano
2018-11-12 11:44 ` [PATCH v2 0/3] Fix built-in rebase perf regression Johannes Schindelin via GitGitGadget
2018-11-12 11:44 ` [PATCH v2 1/3] rebase: consolidate clean-up code before leaving reset_head() Johannes Schindelin via GitGitGadget
2018-11-12 11:44 ` [PATCH v2 2/3] rebase: prepare reset_head() for more flags Johannes Schindelin via GitGitGadget
2018-11-12 11:44 ` [PATCH v2 3/3] built-in rebase: reinstate `checkout -q` behavior where appropriate Johannes Schindelin via GitGitGadget
2018-11-12 14:26 ` [PATCH v2 0/3] Fix built-in rebase perf regression Johannes Schindelin
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=nycvar.QRO.7.76.6.1811091750330.39@tvgsbejvaqbjf.bet \
--to=johannes.schindelin@gmx.de \
--cc=git@vger.kernel.org \
--cc=gitgitgadget@gmail.com \
--cc=gitster@pobox.com \
--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.
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).