Re: [PATCH 08/15] tty: serial: qcom-geni-serial: split out the FIFO tx code

["Ilpo Järvinen" <ilpo.jarvinen@linux.intel.com>]

On Fri, 18 Nov 2022, Bartosz Golaszewski wrote:

> From: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>
> 
> qcom_geni_serial_handle_tx() is pretty big, let's move the code that
> handles the actual writing of data to a separate function which makes
> sense in preparation for introducing a dma variant of handle_tx().
> 
> Signed-off-by: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>
> ---
>  drivers/tty/serial/qcom_geni_serial.c | 60 +++++++++++++++------------
>  1 file changed, 33 insertions(+), 27 deletions(-)
> 
> diff --git a/drivers/tty/serial/qcom_geni_serial.c b/drivers/tty/serial/qcom_geni_serial.c
> index 39041538e5d2..4b155ca0ac74 100644
> --- a/drivers/tty/serial/qcom_geni_serial.c
> +++ b/drivers/tty/serial/qcom_geni_serial.c
> @@ -701,19 +701,48 @@ static void qcom_geni_serial_start_rx(struct uart_port *uport)
>  	writel(irq_en, uport->membase + SE_GENI_M_IRQ_EN);
>  }
>  
> +static int qcom_geni_serial_send_chunk_fifo(struct uart_port *uport,
> +					    unsigned int chunk)
> +{

Some of the comments I have for this function you might want to implement 
in a different patch than this simple move to own function.

> +	struct qcom_geni_serial_port *port = to_dev_port(uport);
> +	struct circ_buf *xmit = &uport->state->xmit;
> +	size_t remaining = chunk;
> +	int i, tail = xmit->tail;

Why there's remaining and i variables, both seem to provide a chunk sized 
limit? Couldn't one be dropped?

You could drop local tail handling, it doesn't seem to add much value.

> +	for (i = 0; i < chunk; ) {
> +		unsigned int tx_bytes;
> +		u8 buf[sizeof(u32)];
> +		int c;
> +
> +		memset(buf, 0, sizeof(buf));
> +		tx_bytes = min_t(size_t, remaining, BYTES_PER_FIFO_WORD);

It's a bit confusing to mix sizeof(u32) and BYTES_PER_FIFO_WORD for the 
same purpose, no?

> +		for (c = 0; c < tx_bytes ; c++) {
> +			buf[c] = xmit->buf[tail++];
> +			tail &= UART_XMIT_SIZE - 1;

If you drop the local tail, this becomes:

			buf[c] = xmit->buf[xmit->tail];
			uart_xmit_advance(uport, 1);

> +		}
> +
> +		iowrite32_rep(uport->membase + SE_GENI_TX_FIFOn, buf, 1);
> +
> +		i += tx_bytes;
> +		uport->icount.tx += tx_bytes;

With uart_xmit_advance, this is to be dropped.

> +		remaining -= tx_bytes;
> +		port->tx_remaining -= tx_bytes;
> +	}
> +
> +	return tail;
> +}
> +
>  static void qcom_geni_serial_handle_tx(struct uart_port *uport, bool done,
>  		bool active)
>  {
>  	struct qcom_geni_serial_port *port = to_dev_port(uport);
>  	struct circ_buf *xmit = &uport->state->xmit;
>  	size_t avail;
> -	size_t remaining;
>  	size_t pending;
> -	int i;
>  	u32 status;
>  	u32 irq_en;
>  	unsigned int chunk;
> -	int tail;
>  
>  	status = readl(uport->membase + SE_GENI_TX_FIFO_STATUS);
>  
> @@ -732,7 +761,6 @@ static void qcom_geni_serial_handle_tx(struct uart_port *uport, bool done,
>  	avail = port->tx_fifo_depth - (status & TX_FIFO_WC);
>  	avail *= BYTES_PER_FIFO_WORD;
>  
> -	tail = xmit->tail;
>  	chunk = min(avail, pending);
>  	if (!chunk)
>  		goto out_write_wakeup;
> @@ -747,29 +775,7 @@ static void qcom_geni_serial_handle_tx(struct uart_port *uport, bool done,
>  					uport->membase + SE_GENI_M_IRQ_EN);
>  	}
>  
> -	remaining = chunk;
> -	for (i = 0; i < chunk; ) {
> -		unsigned int tx_bytes;
> -		u8 buf[sizeof(u32)];
> -		int c;
> -
> -		memset(buf, 0, sizeof(buf));
> -		tx_bytes = min_t(size_t, remaining, BYTES_PER_FIFO_WORD);
> -
> -		for (c = 0; c < tx_bytes ; c++) {
> -			buf[c] = xmit->buf[tail++];
> -			tail &= UART_XMIT_SIZE - 1;
> -		}
> -
> -		iowrite32_rep(uport->membase + SE_GENI_TX_FIFOn, buf, 1);
> -
> -		i += tx_bytes;
> -		uport->icount.tx += tx_bytes;
> -		remaining -= tx_bytes;
> -		port->tx_remaining -= tx_bytes;
> -	}
> -
> -	xmit->tail = tail;
> +	xmit->tail = qcom_geni_serial_send_chunk_fifo(uport, chunk);

Why you want to assign to xmit->tail here?

>  
>  	/*
>  	 * The tx fifo watermark is level triggered and latched. Though we had
> 

-- 
 i.