[dm-devel] [PATCH 5/7] dm-thin: use dm_kvmalloc

[Mike Snitzer]

On Fri, Jul 03 2015 at  5:00pm -0400,
Mikulas Patocka <mpatocka at redhat.com> wrote:

> Make dm-thin use dm_kvmalloc instead of kmalloc because there was a
> reported allocation failure - see
> https://bugzilla.redhat.com/show_bug.cgi?id=1225370
> 
> Signed-off-by: Mikulas Patocka <mpatocka at redhat.com>
> 
> ---
>  drivers/md/dm-thin.c |    6 +++---
>  1 file changed, 3 insertions(+), 3 deletions(-)
> 
> Index: linux-4.1/drivers/md/dm-thin.c
> ===================================================================
> --- linux-4.1.orig/drivers/md/dm-thin.c	2015-07-02 18:57:59.000000000 +0200
> +++ linux-4.1/drivers/md/dm-thin.c	2015-07-02 19:00:19.000000000 +0200
> @@ -2513,7 +2513,7 @@ static void __pool_destroy(struct pool *
>  	mempool_destroy(pool->mapping_pool);
>  	dm_deferred_set_destroy(pool->shared_read_ds);
>  	dm_deferred_set_destroy(pool->all_io_ds);
> -	kfree(pool);
> +	kvfree(pool);
>  }
>  
>  static struct kmem_cache *_new_mapping_cache;
> @@ -2535,7 +2535,7 @@ static struct pool *pool_create(struct m
>  		return (struct pool *)pmd;
>  	}
>  
> -	pool = kmalloc(sizeof(*pool), GFP_KERNEL);
> +	pool = dm_kvmalloc(sizeof(*pool), GFP_KERNEL);
>  	if (!pool) {
>  		*error = "Error allocating memory for pool";
>  		err_p = ERR_PTR(-ENOMEM);
> @@ -2630,7 +2630,7 @@ bad_wq:
>  bad_kcopyd_client:
>  	dm_bio_prison_destroy(pool->prison);
>  bad_prison:
> -	kfree(pool);
> +	kvfree(pool);
>  bad_pool:
>  	if (dm_pool_metadata_close(pmd))
>  		DMWARN("%s: dm_pool_metadata_close() failed.", __func__);

Also, this patchset is entirely too much interface churn to act as a
minimalist fix for BZ 1225370.  I've staged this fix for 4.2-rcX:
https://git.kernel.org/cgit/linux/kernel/git/device-mapper/linux-dm.git/commit/?h=for-next&id=a822c83e47d97cdef38c4352e1ef62d9f46cfe98

We'll need to rebase to use your kvmalloc() for 4.3.