git/list[1] front-page[2] threads[3] people[4] search[5] about
 

Re: [PATCH] refs/files: skip lock files during consistency checks

From
Karthik Nayak <karthik.188@gmail.com>
Date
Apr 21, 2026, 12:45 UTC
Message-ID
<CAOLa=ZTUjqE4msQcTDcemczdOuD+dREGuxozttp1EaPgRnTNww@mail.gmail.com>
In-Reply-To
<xmqqtst5pkg3.fsf@gitster.g>
Junio C Hamano <gitster@pobox.com> writes:
Show 20 quoted lines
> Karthik Nayak <karthik.188@gmail.com> writes:
>
>> Running consistency checks on the files reference backend involves
>> iterating over all the existing files in the 'refs/' directory and
>> running all `files_fsck_ref()` on each of them.
>>
>> Unfortunately this also includes the '*.lock' files created by the files
>> backend. While the `files_fsck_refs_name()` check was skipping over such
>> lock files, all other checks still continue to validate them.
>>
>> Avoid this situation by moving the check for lock files to a layer above
>> and skipping all checks when encountering such a file.
>
> Saying "all other checks" when there is only one other check is
> highly confusing, even though it may not be telling any lies.
> files_fsck_ref() calls files_fsck_refs_name() and
> files_fsck_refs_content(), so this change moves the test from the
> former to a higher level caller so that files_fsck_ref() itself
> won't be called, the net result is that files_fsck_refs_content)( is
> not called on these *.lock files.

You're right. I think I wanted to say that moving it to the upper layer skips both files_fsck_refs_name() and files_fsck_refs_content() and any other new functions which would be added to fsck_refs_fn[]. I'll modify to explain this better with more details.

Show 10 quoted lines
>
> And this does not explain why only one of the callers of
> files_fsck_ref() is the best place to add this "*.lock files are
> exempt" knowledge to, compared to (presumably at the beginning of)
> files_fsck_ref() itself.  If we do not anticipate that we will ever
> gain new caller to the function and the only meaningful caller that
> needs this protection is files_fsck_refs_dir(), then the choice may
> be justifiable, but if we check at the beginning of the callee, we
> do not have to rely on such an assuption, do we?
>
There are two callers to files_fsck_ref():
1. files_fsck_root_ref(): where is a callback function passed to
for_each_root_ref(), so it only iterates over root refs.
2. files_fsck_refs_dir(): which iterates overs entities within the
'refs/' dir and then runs the checks provided in fsck_refs_fn[].

We don't have to worry about encountering lock files in #1 since we rely on the refs API, but in #2 since we manually iterate over the directory, we need to skip the '.lock' files.

Any new checks which are added should be added via fsck_refs_fn[], so this makes adding the check in files_fsck_refs_dir() the right place.

Will add this additional information into the commit message.
Show 90 quoted lines
>> Signed-off-by: Karthik Nayak <karthik.188@gmail.com>
>> ---
>>  refs/files-backend.c     | 22 +++++++++++-----------
>>  t/t0602-reffiles-fsck.sh | 21 +++++++++++++++++++++
>>  2 files changed, 32 insertions(+), 11 deletions(-)
>
>> diff --git a/refs/files-backend.c b/refs/files-backend.c
>> index b3b0c25f84..f1bdfbe88e 100644
>> --- a/refs/files-backend.c
>> +++ b/refs/files-backend.c
>> @@ -3864,22 +3864,12 @@ static int files_fsck_refs_content(struct ref_store *ref_store,
>>  static int files_fsck_refs_name(struct ref_store *ref_store UNUSED,
>>  				struct fsck_options *o,
>>  				const char *refname,
>> -				const char *path,
>> +				const char *path UNUSED,
>>  				int mode UNUSED)
>>  {
>>  	struct strbuf sb = STRBUF_INIT;
>> -	const char *filename;
>>  	int ret = 0;
>>
>> -	filename = basename((char *) path);
>> -
>> -	/*
>> -	 * Ignore the files ending with ".lock" as they may be lock files
>> -	 * However, do not allow bare ".lock" files.
>> -	 */
>> -	if (filename[0] != '.' && ends_with(filename, ".lock"))
>> -		goto cleanup;
>> -
>>  	if (is_root_ref(refname))
>>  		goto cleanup;
>>
>> @@ -3939,6 +3929,7 @@ static int files_fsck_refs_dir(struct ref_store *ref_store,
>>  	struct strbuf refname = STRBUF_INIT;
>>  	struct strbuf sb = STRBUF_INIT;
>>  	struct dir_iterator *iter;
>> +	const char *filename;
>>  	int iter_status;
>>  	int ret = 0;
>>
>> @@ -3962,6 +3953,15 @@ static int files_fsck_refs_dir(struct ref_store *ref_store,
>>  			strbuf_addf(&refname, "worktrees/%s/", wt->id);
>>  		strbuf_addf(&refname, "refs/%s", iter->relative_path);
>>
>> +		filename = basename((char *) iter->path.buf);
>> +
>> +		/*
>> +		 * Ignore the files ending with ".lock" as they may be lock files
>> +		 * However, do not allow bare ".lock" files.
>> +		 */
>> +		if (filename[0] != '.' && ends_with(filename, ".lock"))
>> +			continue;
>> +
>>  		if (files_fsck_ref(ref_store, o, refname.buf,
>>  				   iter->path.buf, iter->st.st_mode) < 0)
>>  			ret = -1;
>> diff --git a/t/t0602-reffiles-fsck.sh b/t/t0602-reffiles-fsck.sh
>> index 3c1f553b81..fc67bee161 100755
>> --- a/t/t0602-reffiles-fsck.sh
>> +++ b/t/t0602-reffiles-fsck.sh
>> @@ -87,6 +87,27 @@ test_expect_success 'ref name should be checked' '
>>  	)
>>  '
>>
>> +test_expect_success 'lock files should be ignored' '
>> +	test_when_finished "rm -rf repo" &&
>> +	git init repo &&
>> +	(
>> +		cd repo &&
>> +		git commit --allow-empty -m initial &&
>> +		git checkout -b branch-1 &&
>> +
>> +		touch .git/refs/heads/branch-1.lock &&
>> +		git refs verify 2>err &&
>> +		test_must_be_empty err &&
>> +
>> +		echo "foobar" >.git/refs/heads/branch-2 &&
>> +		test_must_fail git refs verify 2>err &&
>> +		cat >expect <<-EOF &&
>> +		error: refs/heads/branch-2: badRefContent: foobar
>> +		EOF
>> +		test_cmp expect err
>> +	)
>> +'
>> +
>>  test_expect_success 'ref name check should be adapted into fsck messages' '
>>  	test_when_finished "rm -rf repo" &&
>>  	git init repo &&
Previous: Junio C HamanoNext: Christian Couder
Message 3 of 10 in “refs/files: skip lock files during consistency checks”
  1. refs/files: skip lock files during consistency checksKarthik Nayak, Apr 20, 2026
  2. Junio C HamanoApr 20, 2026
  3. Karthik NayakApr 21, 2026
  4. Christian CouderApr 21, 2026
  5. Karthik NayakApr 21, 2026
  6. Karthik NayakApr 21, 2026
  7. refs/files: skip lock files during consistency checksKarthik Nayak, Apr 22, 2026
  8. Patrick SteinhardtApr 22, 2026
  9. Karthik NayakApr 22, 2026
  10. refs/files: skip lock files during consistency checksKarthik Nayak, May 17, 2026

Read the whole thread, see it on lore, or plain text.

$ cat FOOTERMessages come from the public archive at lore.kernel.org/git, fetched every hour. The front page is chosen and written each morning by an AI editor and can be wrong; the threads themselves are the record. About and API. For agents: an MCP server at https://gitlist.dev/mcp, and any thread, story or person page as Markdown by adding .md to its URL (or sending Accept: text/markdown). Details in /llms.txt.