Re: [PATCH] UBI: fastmap: fix a mem leak on error path

[Richard Weinberger <richard@nod.at>]

Am 27.09.2013 17:37, schrieb Richard Genoud:
> If the scan_pool() function fails, the "free" list is not freed.
> This leads to destroy_ai(ai) complaining with:
> "kmem_cache_destroy ubi_ainf_peb_slab: Slab cache still has objects"

Hmm, I *think* we need to take care about used and eba_orphans too.
Tomorrow I'll have time to investigate...

Thanks,
//richard

> Signed-off-by: Richard Genoud <richard.genoud@gmail.com>
> ---
>  drivers/mtd/ubi/fastmap.c |    4 ++++
>  1 file changed, 4 insertions(+)
> 
> diff --git a/drivers/mtd/ubi/fastmap.c b/drivers/mtd/ubi/fastmap.c
> index 05067f5..22eb527 100644
> --- a/drivers/mtd/ubi/fastmap.c
> +++ b/drivers/mtd/ubi/fastmap.c
> @@ -841,6 +841,10 @@ static int ubi_attach_fastmap(struct ubi_device *ubi,
>  fail_bad:
>  	ret = UBI_BAD_FASTMAP;
>  fail:
> +	list_for_each_entry_safe(tmp_aeb, _tmp_aeb, &free, u.list) {
> +		kmem_cache_free(ai->aeb_slab_cache, tmp_aeb);
> +		list_del(&tmp_aeb->u.list);
> +	}
>  	return ret;
>  }
>  
>