[virtio-dev] Re: [virtio-comment] [PATCH] [PATCH v5] virtio-spi: add the device specification

Hi Huck,

On 11/29/2023 8:42 PM, Cornelia Huck wrote:
> On Wed, Nov 29 2023, Viresh Kumar <[email protected]> wrote:
> 
>> On 29-11-23, 18:31, Haixu Cui wrote:
>>> Hi Viresh,
>>>
>>>      Thank you for your helpful comments. In next patch, I will clearly point
>>> this out:
>>
>> Great, finally we are on the same page. Thanks Haixu.
>>
>>>      "For full-duplex read and write transfer, both \field{tx_buf} and
>>> \field{rx_buf} are used and the buffer size of \field{tx_buf} must be same
>>> as \field{rx_buf}."
>>
>> Suggest rewriting as:
>>
>> In full-duplex transfer mode, both \field{tx_buf} and \field{rx_buf} are sent by
>> the driver, \field{tx_buf} followed by \field{rx_buf}. The length of both the
>> buffers MUST be same.
> 
> Is that in a non-normative section? (Sorry, I've lost track here...) If
> so, I would say:
> 
> "The length of both buffers has to be the same."

This is non-normative section.
> 
>>
>>>      And in drivernormative section, I will add a requirement:
>>>      "For full-duplex transfer, Virtio SPI driver MUST guarantee the write
>>> transfer size is equal to the read transfer size"
>>
>> Maybe:
>>
>> drivernormative:
>>
>> In full-duplex transfer mode, the Virtio SPI driver MUST guarantee that the
>> length of both \field{tx_buf} and \field{rx_buf} are same.
> 
> s/are same/is the same/
> 
>>
>> devicenormative:
>>
>> In full-duplex transfer mode, the Virtio SPI device MUST verify that the length
>> of both \field{tx_buf} and \field{rx_buf} are same. In case of any mismatch, the
> 
> s/are same/is the same/
> 
>> device MUST fail the transfer and notify the driver.
> 
> "notify the driver" == "set an appropriate error value"?
> 
> Can this be covered by one of the existing error values?

I'd like to use VIRTIO_SPI_PARAM_ERR to indicates the buffer size 
inequality issue here because this is in the scope of parameter checking.

Thanks & BR

Haixu Cui

> 

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]