Re: [PATCH 1/2] scsi: ufs: Add Multi-Circular Queue support

[Bart Van Assche]

On 7/19/22 00:01, Can Guo wrote:
> +/**
> + * @ucdl_base_addr: UFS Command Descriptor base address
> + * @sqe_base_addr: submission queue entry base address
> + * @sqe_shadow_addr: submission queue entry shadow address
> + * @ucdl_dma_addr: UFS Command Descriptor DMA address
> + * @sqe_dma_addr: submission queue dma address
> + * @cqe_base_addr: completion queue base address
> + * @cqe_dma_addr: completion queue dma address
> + * @lrb: array of lrb for this hardware queue
> + * @max_entries: max number of slots in this hardware queue
> + * @sq_tp_slot: current slot to which SQ tail pointer is pointing
> + * @sq_hp_slot: current slot to which SQ head pointer is pointing
> + * @cq_tp_slot: current slot to which CQ tail pointer is pointing
> + * @cq_hp_slot: current slot to which CQ head pointer is pointing
> + */
> +struct ufs_hw_queue {
> +	struct utp_transfer_cmd_desc *ucdl_base_addr;
> +	void *sqe_base_addr;
> +	struct utp_transfer_req_desc *sqe_shadow_addr;
> +	dma_addr_t ucdl_dma_addr;
> +	dma_addr_t sqe_dma_addr;
> +	struct cq_entry *cqe_base_addr;
> +	dma_addr_t cqe_dma_addr;
> +	struct ufshcd_lrb *lrb;
> +	u32 max_entries;
> +	u32 id;
> +
> +	void __iomem *mcq_sq_hp;
> +	void __iomem *mcq_sq_tp;
> +	void __iomem *mcq_cq_hp;
> +	void __iomem *mcq_cq_tp;
> +
> +	spinlock_t sq_lock;
> +	u32 sq_tp_slot;
> +	u32 sq_hp_slot;
> +	spinlock_t cq_lock;
> +	u32 cq_tp_slot;
> +	u32 cq_hp_slot;
>   };
>   

Please move all new data structures into a private header that can be 
moved into a private header. I think the above data structure can be 
moved from a public into a private header (a header that is not shared 
with the host drivers).

> +#define for_each_hw_queue(hba, i) \
> +	for ((i) = 0; (i) < (hba)->nr_hw_queues; (i) ++)

A macro like the above reduces code readability. Please remove this 
macro definition.

> +static inline bool is_mcq_enabled(struct ufs_hba *hba)
> +{
> +	return hba->use_mcq;
> +}
> +
> +static inline bool is_mcq_supported(struct ufs_hba *hba)
> +{
> +	return hba->mcq_sup;
> +}

The names of the two above functions are longer than their 
implementation so it's probably better to remove these function definitions.

> -#define ufshcd_writel(hba, val, reg)	\
> +#define ufshcd_writel(hba, val, reg)		\
>   	writel((val), (hba)->mmio_base + (reg))

Is this a whitespace-only change? If so, should that change be in this 
patch?

Thanks,

Bart.