Re: [PATCH 10/15] commit.c: fix a memory leak

[Junio C Hamano <gitster@pobox.com>]

Stefan Beller <sbeller@google.com> writes:

> Signed-off-by: Stefan Beller <sbeller@google.com>
> ---
>  builtin/commit.c | 6 ++++--
>  1 file changed, 4 insertions(+), 2 deletions(-)
>
> diff --git a/builtin/commit.c b/builtin/commit.c
> index 961e467..da79ac4 100644
> --- a/builtin/commit.c
> +++ b/builtin/commit.c
> @@ -229,7 +229,7 @@ static int commit_index_files(void)
>  static int list_paths(struct string_list *list, const char *with_tree,
>  		      const char *prefix, const struct pathspec *pattern)
>  {
> -	int i;
> +	int i, ret;
>  	char *m;
>  
>  	if (!pattern->nr)
> @@ -256,7 +256,9 @@ static int list_paths(struct string_list *list, const char *with_tree,
>  			item->util = item; /* better a valid pointer than a fake one */
>  	}
>  
> -	return report_path_error(m, pattern, prefix);
> +	ret = report_path_error(m, pattern, prefix);
> +	free(m);
> +	return ret;
>  }

Looks correct.

A tangent.  We may want to move report_path_error() to somewhere
more "common"-ish, like dir.c which is where we have bulk of
pathspec matching infrastructure.  Seeing the function used from
builtin/ls-files.c makes me feel dirty.

A further tangent (Duy Cc'ed for this point).  We might want to
rethink the interface to ce_path_match() and report_path_error()
so that we do not have to do a separate allocation of "has this
pathspec been used?" array.  This was a remnant from the olden days
back when pathspec were mere "const char **" where we did not have
any room to add per-item bit---these days pathspec is repreasented
as an array of "struct pathspec" and we can afford to add a bit
to the structure---which will make this kind of leak much less
likely to happen.