Re: [PATCH] nvme-multipath: fix possible hang in live ns resize with ANA access

[Christoph Hellwig <hch@lst.de>]

On Fri, Sep 30, 2022 at 08:29:54AM +0200, Hannes Reinecke wrote:
>>   void nvme_mpath_revalidate_paths(struct nvme_ns *ns)
>>   {
>>   	struct nvme_ns_head *head = ns->head;
>> +	struct nvme_ns *n;
>>   	sector_t capacity = get_capacity(head->disk);
>>   	int node;
>>   -	list_for_each_entry_rcu(ns, &head->list, siblings) {
>> -		if (capacity != get_capacity(ns->disk))
>> -			clear_bit(NVME_NS_READY, &ns->flags);
>> +	list_for_each_entry_rcu(n, &head->list, siblings) {
>> +		if (capacity != get_capacity(n->disk))
>> +			clear_bit(NVME_NS_READY, &n->flags);
>>   	}
>>     	for_each_node(node)
>>   		rcu_assign_pointer(head->current_path[node], NULL);
>> +	nvme_kick_requeue_lists(ns->ctrl);
>>   }
>>     static bool nvme_path_is_disabled(struct nvme_ns *ns)
>
> Hmm. I guess the real issue is that we use 'ns' as both function argument 
> _and_ iterator, so that after 'list_for_each_rcu()' 'ns' is pointing to a 
> _different_ namespace, such that we end up kicking that namespace and not 
> the one used as argument.
> But in either case, patch is fine.

BEfore this patch that wasn't an actual problem as the ns pass as
argument was only used to derived the ns_head but not used after
the loop.  It was a bit of a landmine, though.