From: Junio C Hamano <gitster@pobox.com>
Date: Thu, 21 Aug 2008 00:16:25 GMT
Subject: Re: [PATCH] * remote.c (valid_fetch_refspec): remove useless if-before-free test
Message-ID: <7vr68jmdd2.fsf@gitster.siamese.dyndns.org>
In-Reply-To: <pI8EXbuWhkm4pM0AaaTAPQQYBeFSffOSzCiM4NfNHKomivEcfUfShA@cipher.nrlssc.navy.mil>

Brandon Casey <casey@nrlssc.navy.mil> writes:

> Maybe we should also begin the process of not leaking memory here...
>
> diff --git a/remote.c b/remote.c
> index 7f2897b..984ad1b 100644
> --- a/remote.c
> +++ b/remote.c
> @@ -449,6 +449,20 @@ static int verify_refname(char *name, int is_glob)
> 	return result;
>  }
>  
> +void free_refspecs(struct refspec *refspec, int nr_refspec)
> +{
> +	int i;
> +
> +	if (!refspec)
> +		return;
> +
> +	for (i = 0; i < nr_refspec; i++) {
> +		free(refspec[i].src);
> +		free(refspec[i].dst);
> +	}
> +	free(refspec);
> +}

Are you sure all the codepaths that stuff refspec[].{src,dst} give
freeable pointer?  E.g. if anybody splits a originally single string
"refs/heads/foo:refs/remotes/origin/foo" into two by replacing the colon
with NUL and pointing the halves, and/or such string came from argv[]
without strdup(), I'd imagine free() would not like you very much.

I didn't look, though.