Re: [PATCH V1] spi: Add support for Double Transfer Rate (DTR) mode

[Mukesh Kumar Savaliya]

Thanks Mark !

On 3/26/2025 8:12 PM, Mark Brown wrote:
> On Wed, Mar 26, 2025 at 07:55:05PM +0530, Mukesh Kumar Savaliya wrote:
> > On 3/26/2025 6:34 PM, Mark Brown wrote:
>
> > > We should have a flag in the controller indicating if it supports this,
> > > and code in the core which returns an error if a driver attempts to use
> > > it when the controller doesn't support it.
>
> > Have added below in spi.h which can be set by client and controller driver
> > should be using it to decide mode.
>
> > + bool        dtr_mode;
>
> > since default it's false, should continue with SDR.
> > I believe for QSPI, it supports SDR or DDR, but it's not applicable to
> > standard SPI right ? So not sure in which case we should return an error ?
>
> Standard SPI is the main thing I'm thinking of here, or possibly some
> limited QSPI controller that doesn't support DTR.  It's not something
> that should actually come up really, it's more error handling if things
> aren't set up properly.

IIUC, It comes to the point of first identifying if it's in context of 
QSPI controller or SPI controller right ?

Identify if SPI/QSPI controller has this capability using dtr_caps = 
true/false. Then check if it's supporting SDR/DDR mode. Can we then 
introduce below struct to first mark capability as true/false and then 
process dtr_mode ?

if not supported (dtr_caps = false), then don't care dtr_mode.
struct spi_caps {
	bool dtr_mode;
	bool dtr_caps;
};

OR we can have an API spi_controller_dtr_caps() which returns this flag 
set by individual SPI/QSPI controller. We can use above struct OR 
separate  struct spi_controller_dtr_caps { bool dtr_caps };.

Please confirm if i am correct OR suggest more. Else you may approve one 
of the above option. (Sorry if i am missing anything ).