On 05/04/19 10:55 AM, Naga Sureshkumar Relli wrote:
> Hi Vignesh,
>
> Thanks for the review.
>
>> -----Original Message-----
>> From: Vignesh Raghavendra <vigne...@ti.com>
>> Sent: Friday, April 5, 2019 10:14 AM
>> To: Naga Sureshkumar Relli <nagas...@xilinx.com>; broo...@kernel.org;
>> bbrezil...@kernel.org
>> Cc: linux-...@vger.kernel.org; dw...@infradead.org; marek.va...@gmail.com;
>> rich...@nod.at; linux-...@lists.infradead.org; linux-kernel@vger.kernel.org;
>> Michal Simek
>> <mich...@xilinx.com>; nagasures...@gmail.com
>> Subject: Re: [LINUX PATCH v2 3/3] spi: spi-mem: Add support for Zynq QSPI
>> controller
>>
>>
>>
>> On 01/04/19 1:29 PM, Naga Sureshkumar Relli wrote:
>>> +/**
>>> + * zynq_qspi_config_op - Configure QSPI controller for specified transfer
>>> + * @xqspi: Pointer to the zynq_qspi structure
>>> + * @qspi: Pointer to the spi_device structure
>>> + *
>>> + * Sets the operational mode of QSPI controller for the next QSPI
>>> +transfer and
>>> + * sets the requested clock frequency.
>>> + *
>>> + * Return: 0 on success and -EINVAL on invalid input parameter
>>> + *
>>> + * Note: If the requested frequency is not an exact match with what
>>> +can be
>>> + * obtained using the prescalar value, the driver sets the clock
>>> +frequency which
>>> + * is lower than the requested frequency (maximum lower) for the
>>> +transfer. If
>>> + * the requested frequency is higher or lower than that is supported
>>> +by the QSPI
>>> + * controller the driver will set the highest or lowest frequency
>>> +supported by
>>> + * controller.
>>> + */
>>> +static int zynq_qspi_config_op(struct zynq_qspi *xqspi, struct
>>> +spi_device *spi) {
>>> + u32 config_reg, baud_rate_val = 0;
>>> +
>>> + /*
>>> + * Set the clock frequency
>>> + * The baud rate divisor is not a direct mapping to the value written
>>> + * into the configuration register (config_reg[5:3])
>>> + * i.e. 000 - divide by 2
>>> + * 001 - divide by 4
>>> + * ----------------
>>> + * 111 - divide by 256
>>> + */
>>> + while ((baud_rate_val < ZYNQ_QSPI_BAUD_DIV_MAX) &&
>>> + (clk_get_rate(xqspi->refclk) / (2 << baud_rate_val)) >
>>> + spi->max_speed_hz)
>>> + baud_rate_val++;
>>> +
>>
>> Instead use DIV_ROUND_UP, something like below should work(untested):
>>
>> unsigned long refclk_rate = clk_get_rate(xqspi->refclk);
>> u32 baud_rate_val = DIV_ROUND_UP(refclk_rate, spi->max_speed_hz) - 1;
>>
Oops, sorry, I had meant
u32 baud_rate_val = DIV_ROUND_UP(refclk_rate, 2 * spi->max_speed_hz) -
1;
But please ignore my comment, I see that div values goes by power of 2 and not
multiple of 2
Regards
Vignesh
> This is not just direct calculation.
> i.e. for example
> refclk_rate = 200MHz and max_speed_hx = 100MHz.
> then DIV_ROUND_UP gives a value of 2.
> But writing a value of 2 to config registers means, divide by 8. But we
> should write divide by 2 (value of zero).
> That’s why we implemented the above calculation.
>
> 000: divide by 2.
> 001: divide by 4
> 010: divide by 8
> 011: divide by 16
> 100: divide by 32
> 101: divide by 64
> 110: divide by 128
> 111: divide by 256
>
--
Regards
Vignesh
