[PATCH v4] nvme_core: scan namespaces asynchronously

[stuart hayes]

On 7/17/2024 3:46 PM, Keith Busch wrote:
> On Wed, Jul 17, 2024 at 01:55:50PM -0500, Stuart Hayes wrote:
> 
> Looks good to me. Just one minor comment below.
> 
>> @@ -3978,11 +4008,15 @@ static int nvme_scan_ns_list(struct nvme_ctrl *ctrl)
>>   	__le32 *ns_list;
>>   	u32 prev = 0;
>>   	int ret = 0, i;
>> +	ASYNC_DOMAIN(domain);
>> +	struct async_scan_info scan_info;
>>   
>>   	ns_list = kzalloc(NVME_IDENTIFY_DATA_SIZE, GFP_KERNEL);
>>   	if (!ns_list)
>>   		return -ENOMEM;
>>   
>> +	scan_info.ctrl = ctrl;
>> +	scan_info.ns_list = ns_list;
>>   	for (;;) {
>>   		struct nvme_command cmd = {
>>   			.identify.opcode	= nvme_admin_identify,
>> @@ -3998,19 +4032,23 @@ static int nvme_scan_ns_list(struct nvme_ctrl *ctrl)
>>   			goto free;
>>   		}
>>   
>> +		atomic_set(&scan_info.next_nsid, 0);
>>   		for (i = 0; i < nr_entries; i++) {
>>   			u32 nsid = le32_to_cpu(ns_list[i]);
>>   
>>   			if (!nsid)	/* end of the list? */
>>   				goto out;
>> -			nvme_scan_ns(ctrl, nsid);
>> +			async_schedule_domain(nvme_scan_ns_async, &scan_info,
>> +						&domain);
>>   			while (++prev < nsid)
>>   				nvme_ns_remove_by_nsid(ctrl, prev);
>>   		}
>> +		async_synchronize_full_domain(&domain);
>>   	}
>>    out:
>>   	nvme_remove_invalid_namespaces(ctrl, prev);
>>    free:
>> +	async_synchronize_full_domain(&domain);
> 
> A goto this "free" label appears to mean the async domain has nothing
> scheduled, so synchronizing won't be necessary. This should be moved up
> tharmlesso the "out" label, I think.
> 
> I can change that up when applying; just wanted to mention it in case
> I'm missing something.

I agree, you aren't missing anything.  Thank you very much!  If you want
me to submit a new patch with this and Thomas' changes, let me know.