Re: [virtio-dev] Re: [RFC PATCH v2 3/3] SPI: Add virtio SPI driver (V10 draft specification).

[Harald Mommer <harald.mommer@opensynergy.com>]

Hello Haixu,

Thanks. This was a hard one. I knew that I did the delay settingsmost 
probably somewhat wrong but I had no idea that I did it so wrong.

Reworked, made a new function for this, currently testing. Added the 
link to the place where your E-Mail is stored

https://lore.kernel.org/all/6171c1c3-55ba-4f74-ae60-764820cf1caf@quicinc.com/

as a comment to the source code and hope this will survive the reviews. 
Another option would be to add the diagram + the explanations below as 
quoted here as comment to the code. Whether 30+ lines of comments would 
survive the reviews I don't know.

Just to say nothing in the code making live hard for people trying to 
understand the code is something I would like to avoid.

Regards
Harald


On 30.01.24 04:21, Haixu Cui wrote:
> .ÂÂ .ÂÂÂÂÂ .ÂÂÂ .ÂÂÂ .ÂÂ .ÂÂ .ÂÂ .ÂÂ .ÂÂ .
> Delay + A +ÂÂÂÂÂ + BÂ +ÂÂÂ + C + D + E + F + A +
> ÂÂÂÂÂ .ÂÂ .ÂÂÂÂÂ .ÂÂÂ .ÂÂÂ .ÂÂ .ÂÂ .ÂÂ .ÂÂ .ÂÂ .
> ÂÂ ___.ÂÂ .ÂÂÂÂÂ .ÂÂÂ .ÂÂÂ .ÂÂ .ÂÂ .___.___.ÂÂ .
> CS#ÂÂ |___.______.____.____.___.___|ÂÂ .ÂÂ |___._____________
> ÂÂÂÂÂ .ÂÂ .ÂÂÂÂÂ .ÂÂÂ .ÂÂÂ .ÂÂ .ÂÂ .ÂÂ .ÂÂ .ÂÂ .
> ÂÂÂÂÂ .ÂÂ .ÂÂÂÂÂ .ÂÂÂ .ÂÂÂ .ÂÂ .ÂÂ .ÂÂ .ÂÂ .ÂÂ .
> SCLK__.___.___NNN_____NNN__.___.___.___.___.___.___NNN_______
>
>
> NOTE: 1st transfer has two words, the delay betweent these two words 
> are 'B' in the diagram.
>
> A => struct spi_device -> cs_setup
> B => max{struct spi_transfer -> word_delay,
> ÂÂÂÂÂÂÂÂ struct spi_device -> word_delay}
> ÂÂÂ Note: spi_device and spi_transfer both have word_delay, Linux
> ÂÂÂÂÂÂÂÂ choose the bigger one, refer to _spi_xfer_word_delay_update
> ÂÂÂÂÂÂÂÂ function
> C => struct spi_transfer -> delay
> D => struct spi_device -> cs_hold
> E => struct spi_device -> cs_inactive
> F => struct spi_transfer -> cs_change_delay
>
> So the corresponding relationship:
> A <===> cs_setup_ns (after CS asserted)
> B <===> word_delay_ns (no matter with CS)
> C+D <===> cs_delay_hold_ns (before CS deasserted)
> E+F <===> cs_change_delay_inactive_ns (after CS deasserted, these two 
> values also recommend in Linux driver to be added up)