Re: [RFC PATCH 1/2] sha1-file: fsync() loose dir entry when core.fsyncObjectFiles

From: Johannes Sixt <j6t@kdbg.org>
Date: Sep 17, 2020, 20:21 UTC
Message-ID: <64358b70-4fff-5dc8-6e63-2fc916bea6af@kdbg.org>
In-Reply-To: <20200917112830.26606-2-avarab@gmail.com>

Am 17.09.20 um 13:28 schrieb Ævar Arnfjörð Bjarmason:

Show 26 quoted lines

> Change the behavior of core.fsyncObjectFiles to also sync the
> directory entry. I don't have a case where this broke, just going by
> paranoia and the fsync(2) manual page's guarantees about its behavior.
> 
> Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
> ---
>  sha1-file.c | 19 ++++++++++++++-----
>  1 file changed, 14 insertions(+), 5 deletions(-)
> 
> diff --git a/sha1-file.c b/sha1-file.c
> index dd65bd5c68..d286346921 100644
> --- a/sha1-file.c
> +++ b/sha1-file.c
> @@ -1784,10 +1784,14 @@ int hash_object_file(const struct git_hash_algo *algo, const void *buf,
>  }
>  
>  /* Finalize a file on disk, and close it. */
> -static void close_loose_object(int fd)
> +static void close_loose_object(int fd, const struct strbuf *dirname)
>  {
> -	if (fsync_object_files)
> +	int dirfd;
> +	if (fsync_object_files) {
>  		fsync_or_die(fd, "loose object file");
> +		dirfd = xopen(dirname->buf, O_RDONLY);
> +		fsync_or_die(dirfd, "loose object directory");

Did you have the opportunity to verify that this works on Windows? Opening a directory with open(2), I mean: It's disallowed according to the docs: https://docs.microsoft.com/en-us/cpp/c-runtime-library/reference/open-wopen?view=vs-2019#return-value

> +	}
>  	if (close(fd) != 0)
>  		die_errno(_("error when closing loose object file"));
>  }

-- Hannes