Re: [PATCH 6/6] osd_client: send watch ping messages

[Mike Christie <mchristi@redhat.com>]

On 06/12/2015 10:56 AM, Douglas Fuller wrote:
>  static int ceph_oloc_decode(void **p, void *end,
>  			    struct ceph_object_locator *oloc)
>  {
> @@ -2795,6 +2889,7 @@ int ceph_osdc_init(struct ceph_osd_client *osdc, struct ceph_client *client)
>  	osdc->num_requests = 0;
>  	INIT_DELAYED_WORK(&osdc->timeout_work, handle_timeout);
>  	INIT_DELAYED_WORK(&osdc->osds_timeout_work, handle_osds_timeout);
> +	INIT_DELAYED_WORK(&osdc->linger_ping_work, handle_linger_ping);
>  	spin_lock_init(&osdc->event_lock);
>  	osdc->event_tree = RB_ROOT;
>  	osdc->event_count = 0;
> @@ -3079,12 +3174,15 @@ static struct ceph_msg *alloc_msg(struct ceph_connection *con,
>  	case CEPH_MSG_OSD_MAP:
>  	case CEPH_MSG_WATCH_NOTIFY:
>  		{
> -			struct ceph_msg *m = ceph_msg_new(type, front, GFP_NOFS, false);
> +			struct ceph_msg *m = ceph_msg_new(type, front,
> +			                                  GFP_NOFS, false);
>  			size_t len = con->in_hdr.data_len;
>  			if (len > 0) {
>  				struct page **pages;
>  				struct ceph_osd_data osd_data;
> -				pages = ceph_alloc_page_vector(calc_pages_for(0, len), GFP_KERNEL);
> +				pages = ceph_alloc_page_vector(
> +				              calc_pages_for(0, len), GFP_NOFS);
> +				WARN_ON(!pages);

Are you wanting this warn to get more info in case someone sends us a
really large buffer?

Handle the null pointer here like is done elsewhere. If you don't you
will get NULL pointer ooppses or other crashes due to there being a non
zero len but null pages pointer.