[[PATCH v1] 13/37] [CIFS] SMBD: Implement SMBD protocol negotiation

[Christoph Hellwig]

> +	request = mempool_alloc(info->request_mempool, GFP_KERNEL);
> +	if (!request)
> +		return rc;

Here you do a mempool allocation to guarantee forward progress..

> +	request->sge = kzalloc(sizeof(struct ib_sge), GFP_KERNEL);
> +	if (!request->sge)
> +		goto allocate_sge_failed;

... and here it's a plain malloc in the same path, which renders the
above useless.  Also I would just embedd the sge into the containing
structure to avoid a memory allocation.

> +	request->info = info;
> +
> +	packet = (struct smbd_negotiate_req *) request->packet;
> +	packet->min_version = cpu_to_le16(0x100);
> +	packet->max_version = cpu_to_le16(0x100);

Canyou add a constant for the version code?

> +	packet->reserved = cpu_to_le16(0);

no need to byte swap 0 - you can always assign it directly.