Re: [PATCH 07/17] spi: dw: Add Tx/Rx finish wait methods to DMA

[Mark Brown]

On Fri, May 08, 2020 at 04:29:32PM +0300, Serge Semin wrote:

> Since DMA transfers are performed asynchronously with actual SPI
> transaction, then even if DMA transfers are finished it doesn't mean
> all data is actually pushed to the SPI bus. Some data might still be

This looks like a bug fix so it should really have gone at the start of
the series so it can be sent to Linus as a bug fix rather than waiting
for the merge window.  This makes sense to me, a couple of nits below:

> +static void dw_spi_dma_wait_tx_done(struct dw_spi *dws)
> +{
> +	int retry = WAIT_RETRIES;
> +	unsigned long ns;
> +
> +	ns = (NSEC_PER_SEC / spi_get_clk(dws)) * dws->n_bytes * BITS_PER_BYTE;
> +	ns *= dw_readl(dws, DW_SPI_TXFLR);
> +
> +	while (dw_spi_dma_tx_busy(dws) && retry--)
> +		ndelay(ns);

How deep can the FIFO be with this IP - could we end up ndelay()ing for
non-trivial amounts of time?

> +static inline u32 spi_get_clk(struct dw_spi *dws)
> +{
> +	u32 div = dw_readl(dws, DW_SPI_BAUDR);
> +
> +	return div ? dws->max_freq / div : 0;

Please write normal conditional statements rather than using the ternery
operator - it helps with legibility.

Attachment: signature.asc
Description: PGP signature