Re: [Resend PATCH] Fix some memory leaks in various places

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

Mike Hommey <mh@glandium.org> writes:

> diff --git a/builtin-init-db.c b/builtin-init-db.c
> index e1393b8..df61758 100644
> --- a/builtin-init-db.c
> +++ b/builtin-init-db.c
> @@ -415,6 +415,7 @@ int cmd_init_db(int argc, const char **argv, const char *prefix)
>  	safe_create_dir(path, 1);
>  	strcpy(path+len, "/info");
>  	safe_create_dir(path, 1);
> +	free(path);
>  
>  	if (shared_repository) {
>  		char buf[10];

This is "technically correct but do we care?" category
immediately before exiting.

> diff --git a/http-walker.c b/http-walker.c
> index 2c37868..1a02f86 100644
> --- a/http-walker.c
> +++ b/http-walker.c
> @@ -231,6 +231,8 @@ static void finish_object_request(struct object_request *obj_req)
>  {
>  	struct stat st;
>  
> +	free(obj_req->url);
> +
>  	fchmod(obj_req->local, 0444);
>  	close(obj_req->local); obj_req->local = -1;
>  

I am not sure about this.  Instead of freeing this memory early,
the function's sole caller, process_object_response(), may be
the right place to do this after calling this function.

Who is responsible for reclaiming the memory for obj_req itself
that is passed to the callback function process_object_response()?
Having two different functions release_object_request() and
finish_object_request() that clean things up differently makes
me feel somewhat queasy.

> @@ -897,9 +899,17 @@ static int fetch_ref(struct walker *walker, char *ref, unsigned char *sha1)
>  static void cleanup(struct walker *walker)
>  {
>  	struct walker_data *data = walker->data;
> +	struct alt_base *prev_altbase, *altbase = data->alt;
> +	while (altbase) {
> +		free(altbase->base);
> +		prev_altbase = altbase;
> +		altbase = altbase->next;
> +		free(prev_altbase);
> +	}
>  	http_cleanup();
>  
>  	curl_slist_free_all(data->no_pragma_header);
> +	free(data);
>  }
>  
>  struct walker *get_http_walker(const char *url)

Looks Ok.

> diff --git a/walker.c b/walker.c
> index adc3e80..64fc419 100644
> --- a/walker.c
> +++ b/walker.c
> @@ -299,6 +299,7 @@ int walker_fetch(struct walker *walker, int targets, char **target,
>  			goto unlock_and_fail;
>  	}
>  	free(msg);
> +	free(sha1);
>  
>  	return 0;
>  
> @@ -306,6 +307,7 @@ unlock_and_fail:
>  	for (i = 0; i < targets; i++)
>  		if (lock[i])
>  			unlock_ref(lock[i]);
> +	free(sha1);
>  
>  	return -1;
>  }

This does not seem to free something that should not get freed,
but there is another leak, I think.  If write_ref_log_details is
set and write_ref_sha1() fails, goto unlock_and_fail will leak
msg, won't it?