Re: [PATCH v1 2/2] scsi: ufs: core: Reduce the sleep before vcc can be powered on

[Bao D. Nguyen]

On 10/7/2025 9:19 AM, Bart Van Assche wrote:
> On 10/7/25 12:04 AM, Peter Wang (王信友) wrote:
> > On Fri, 2025-10-03 at 14:27 -0700, Bao D. Nguyen wrote:
> > > With the current or recent offerings of ufs devices in the market, 
> > > the requirement is 1ms. For example, the Kioxia datasheet says
> > > "Vcc shall be kept less than 0.3V for at least 1ms before it goes
> > > beyond 0.3V again". Similarly other vendors have this 1ms
> > > requirement. So I believe this indicates the worst case scenario. I 
> > > understand there may be very old devices that are upgrading the 
> > > kernel only. In that case I don't know the specifics for these old
> > > ufs parts as mentioned.
> >
> > Hi Bao,
> >
> > Please consider using module_param_cb to set the default
> > delay to 2ms(or 1ms). At the same time, we should keep the
> > flexibility for devices that may require a longer delay by
> > allowing them to extend the delay through a module parameter.
>
> Why a kernel module parameter? Why can't the default delay be set by
> ufshcd_variant_ops.init()?

I am also not sure if it is worth adding complexity to the driver using 
a kernel parameter for this particular case.
I can try Bart's suggestion to override the default value with the 
platform driver init, or drop the change trying to reduce the sleep time 
from 5ms to 2ms.

Thanks, Bao

> Thanks,
>
> Bart.