From: Junio C Hamano <gitster@pobox.com>
Date: Sun, 15 Mar 2026 05:05:50 GMT
Subject: Re: [PATCH v3 1/9] repository: fix repo_init() memleak due to missing _clear()
Message-ID: <xmqq7brdvg0h.fsf@gitster.g>
In-Reply-To: <20260309133739.294555-2-adrian.ratiu@collabora.com>

Adrian Ratiu <adrian.ratiu@collabora.com> writes:

> There is an old pre-existing memory leak in repo_init() due to failing
> to call clear_repository_format() in the error case.
>
> It went undetected because a specific bug is required to trigger it:
> enable a v1 extension in a repository with format v0. Obviously this
> can only happen in a development environment, so it does not trigger
> in normal usage, however the memleak is real and needs fixing.
>
> Fix it by also calling clear_repository_format() in the error case.
>
> Signed-off-by: Adrian Ratiu <adrian.ratiu@collabora.com>
> ---
>  repository.c | 1 +
>  1 file changed, 1 insertion(+)
>
> diff --git a/repository.c b/repository.c
> index 0b8f7ec200..fb4356ca55 100644
> --- a/repository.c
> +++ b/repository.c
> @@ -322,6 +322,7 @@ int repo_init(struct repository *repo,
>  	return 0;
>  
>  error:
> +	clear_repository_format(&format);
>  	repo_clear(repo);
>  	return -1;
>  }

It is arguable if the fault is on the caller, or the callee which is
read_and_verify_repository_format() that answers the caller "hey,
you do not have a valid format to work with" without releasing the
thing *it* sample-read.  As you said, this only triggers in a broken
environment, and there is just a single caller-callee involved, so I
am fine fixing it on the caller side like this patch does.

Thanks.