[PATCH] nvme: enable ro namespace for ZNS without append

[Javier González]

> On 6 Nov 2020, at 15.14, Christoph Hellwig <hch at lst.de> wrote:
> 
> 
>> 
>> +#ifdef CONFIG_BLK_DEV_ZONED
>> +    if (blk_queue_is_zoned(disk->queue) && !ns->zoned_ns_supp)
>> +        set_disk_ro(disk, true);
>> +#endif
> 
> I think we can simplify this at bit.  Add a new NVME_NS_FORCE_RO flag
> to ns->flags, set it in nvme_update_zone_info, and just query it here
> without any ifdefs.
> 
>>    struct request_queue *q = disk->queue;
>>    struct nvme_command c = { };
>>    struct nvme_id_ns_zns *id;
>> +    bool zoned_ns_supp = true;
>>    int status;
>> 
>>    /* Driver requires zone append support */
>>    if (!(le32_to_cpu(log->iocs[nvme_cmd_zone_append]) &
>>            NVME_CMD_EFFECTS_CSUPP)) {
>> +        zoned_ns_supp = false;
>>        dev_warn(ns->ctrl->device,
>>            "append not supported for zoned namespace:%d\n",
>>            ns->head->ns_id);
>> +    } else {
>> +        /* Lazily query controller append limit for the first
>> +         * zoned namespace
>> +         */
>> +        if (!ns->ctrl->max_zone_append) {
>> +            status = nvme_set_max_append(ns->ctrl);
>> +            if (status)
>> +                return status;
>> +        }
> 
> While you're at it:  my inverse the polarity of the if check as that reads just
> a little more natural, and maybe mention in the warning that the namespace is
> forced to read-only mode?
> 
> Otherwise this looks good.

Thanks. I’ll send a V2.