Re: [PATCH v3 09/12] iio: dac: ad5686: implement new sync() op for the spi bus

[David Lechner <dlechner@baylibre.com>]

On 6/22/26 3:50 AM, Rodrigo Alencar wrote:
> On 20/06/26 11:33, David Lechner wrote:
>> On 6/16/26 3:21 AM, Rodrigo Alencar via B4 Relay wrote:
>>> From: Rodrigo Alencar <rodrigo.alencar@analog.com>
>>>
>>> Use of local SPI bus data to manage a collection of SPI transfers and
>>> flush them to the SPI platform driver with the sync() operation. This
>>> allows for faster handling of multiple channel DAC writes, avoiding kernel
>>> overhead per spi_sync() call, which will be helpful when enabling
>>> triggered buffer support.
> 
> ...
> 
>>> +/**
>>> + * struct ad5686_spi_data - SPI bus specific data
>>> + * @msg: SPI message used for transfers
>>> + * @size: number of transfers currently in the message
>>> + * @capacity: maximum number of transfers that can be added to the message
>>> + * @xfers: array of SPI transfers, allocated with the provided capacity
>>> + */
>>> +struct ad5686_spi_data {
>>> +	struct spi_message msg;
>>> +	unsigned int size;
>>> +	unsigned int capacity;
>>> +	struct spi_transfer xfers[] __counted_by(capacity);
>>> +};
>>> +
>>>  static int ad5686_spi_write(struct ad5686_state *st,
>>>  			    u8 cmd, u8 addr, u16 val)
>>>  {
>>> -	struct spi_device *spi = to_spi_device(st->dev);
>>> -	u8 tx_len, *buf;
>>> +	struct ad5686_spi_data *bus_data = st->bus_data;
>>> +	struct spi_transfer *xfer;
>>>  
>>> +	if (bus_data->size >= bus_data->capacity)
>>> +		return -E2BIG;
>>> +
>>> +	if (bus_data->size)
>>> +		bus_data->xfers[bus_data->size - 1].cs_change = 1;
>>> +	else
>>> +		spi_message_init(&bus_data->msg);
>>
>> Seems odd that spi_message_init() is called conditionally. What
>> prevents spi_message_add_tail() from growing the message unbounded
>> on repeated calls?
> 
> The "size >= capacity" check right above this.

OK, but it could still grow to capacity, adding one xfer each
time this function is called with bus_data->size > 0, right?

> 
>>
>>> +
>>> +	xfer = &bus_data->xfers[bus_data->size];
>>>  	switch (st->chip_info->regmap_type) {
>>>  	case AD5310_REGMAP:
>>> -		st->data[0].d16 = cpu_to_be16(AD5310_CMD(cmd) |
>>> -					      val);
>>> -		buf = &st->data[0].d8[0];
>>> -		tx_len = 2;
>>> +		st->data[bus_data->size].d16 =
>>> +			cpu_to_be16(AD5310_CMD(cmd) | val);
>>> +		*xfer = (struct spi_transfer) {
>>> +			.tx_buf = &st->data[bus_data->size].d16,
>>> +			.len = sizeof(st->data[bus_data->size].d16),
>>> +		};
>>>  		break;
>>>  	case AD5683_REGMAP:
>>> -		st->data[0].d32 = cpu_to_be32(AD5686_CMD(cmd) |
>>> -					      AD5683_DATA(val));
>>> -		buf = &st->data[0].d8[1];
>>> -		tx_len = 3;
>>> +		st->data[bus_data->size].d32 =
>>> +			cpu_to_be32(AD5686_CMD(cmd) | AD5683_DATA(val));
>>> +		*xfer = (struct spi_transfer) {
>>> +			.tx_buf = &st->data[bus_data->size].d8[1],
>>> +			.len = sizeof(st->data[bus_data->size].d32) - 1,
>>> +		};
>>>  		break;
>>>  	case AD5686_REGMAP:
>>> -		st->data[0].d32 = cpu_to_be32(AD5686_CMD(cmd) |
>>> -					      AD5686_ADDR(addr) |
>>> -					      val);
>>> -		buf = &st->data[0].d8[1];
>>> -		tx_len = 3;
>>> +		st->data[bus_data->size].d32 =
>>> +			cpu_to_be32(AD5686_CMD(cmd) | AD5686_ADDR(addr) | val);
>>> +		*xfer = (struct spi_transfer) {
>>> +			.tx_buf = &st->data[bus_data->size].d8[1],
>>> +			.len = sizeof(st->data[bus_data->size].d32) - 1,
>>> +		};
>>>  		break;
>>>  	default:
>>>  		return -EINVAL;
>>>  	}
>>>  
>>> -	return spi_write(spi, buf, tx_len);
>>
>> If this function no longer writes, should we change the name of
>> the function to something like ad5686_spi_write_prepare_msg()?
>>
>>> +	spi_message_add_tail(xfer, &bus_data->msg);
>>> +	bus_data->size++;
>>> +
>>> +	return 0;
>>> +}
>>> +
>>> +static int ad5686_spi_sync(struct ad5686_state *st)
>>> +{
>>> +	struct spi_device *spi = to_spi_device(st->dev);
>>> +	struct ad5686_spi_data *bus_data = st->bus_data;
>>> +
>>> +	bus_data->size = 0; /* always reset, even on sync failure */
>>> +	return spi_sync(spi, &bus_data->msg);
>>>  }
>>>  
>>>  static int ad5686_spi_read(struct ad5686_state *st, u8 addr)
>>>  {
>>> -	struct spi_transfer t[] = {
>>> -		{
>>> -			.tx_buf = &st->data[0].d8[1],
>>> -			.len = 3,
>>> -			.cs_change = 1,
>>> -		}, {
>>> -			.tx_buf = &st->data[1].d8[1],
>>> -			.rx_buf = &st->data[2].d8[1],
>>> -			.len = 3,
>>> -		},
>>> -	};
>>>  	struct spi_device *spi = to_spi_device(st->dev);
>>> +	struct ad5686_spi_data *bus_data = st->bus_data;
>>> +	struct spi_transfer *xfer = &bus_data->xfers[0];
>>>  	u8 cmd = 0;
>>>  	int ret;
>>>  
>>> @@ -85,8 +117,21 @@ static int ad5686_spi_read(struct ad5686_state *st, u8 addr)
>>>  				      AD5686_ADDR(addr));
>>>  	st->data[1].d32 = cpu_to_be32(AD5686_CMD(AD5686_CMD_NOOP));
>>>  
>>> -	ret = spi_sync_transfer(spi, t, ARRAY_SIZE(t));
>>> -	if (ret < 0)
>>> +	xfer[0] = (struct spi_transfer) {
>>> +		.tx_buf = &st->data[0].d8[1],
>>> +		.len = sizeof(st->data[0].d32) - 1,
>>
>> Would make more sense to say `sizeof(st->data[0].d8) - 1` since
>> the buffer is  &st->data[0].d8[1].
> 
> the buffer is d8[1], d8[2] and d8[3], skipping d8[0], for a len = 3.

Yes, I see that. This is why I suggested that it should be based on
sizeof(...d8) - 1 rather than sizeof(...d32) - 1. It makes more sense
to use the same field for both the size and the buffer pointer.

> 
>>> +		.cs_change = 1,
>>> +	};
>>> +	xfer[1] = (struct spi_transfer) {
>>> +		.tx_buf = &st->data[1].d8[1],
>>> +		.rx_buf = &st->data[2].d8[1],
>>> +		.len = sizeof(st->data[1].d32) - 1,
>>
>> And here.
>>
>>> +	};
>>> +
>>> +	spi_message_init_with_transfers(&bus_data->msg, xfer, 2);
>>> +
>>> +	ret = spi_sync(spi, &bus_data->msg);
>>> +	if (ret)
>>>  		return ret;
>>>  
>>>  	return be32_to_cpu(st->data[2].d32);
>>> @@ -95,12 +140,26 @@ static int ad5686_spi_read(struct ad5686_state *st, u8 addr)
>>>  static const struct ad5686_bus_ops ad5686_spi_ops = {
>>>  	.write = ad5686_spi_write,
>>>  	.read = ad5686_spi_read,
>>> +	.sync = ad5686_spi_sync,
>>>  };
>>>  
>