Re: [PATCH v2 3/3] usb: typec: tcpm: add get max power support

[Frank Wang]

Hi Sebastian,

On 2023/3/21 13:00, Sebastian Reichel wrote:
> Hi,
>
> On Tue, Mar 21, 2023 at 09:32:53AM +0800, Frank Wang wrote:
> > On 2023/3/21 4:31, Sebastian Reichel wrote:
> > > On Mon, Mar 20, 2023 at 06:07:11PM +0800, Frank Wang wrote:
> > > > Traverse fixed pdos to calculate the maximum power that the charger
> > > > can provide, and it can be get by POWER_SUPPLY_PROP_INPUT_POWER_LIMIT
> > > > property.
> > > >
> > > > Signed-off-by: Frank Wang <frank.wang@xxxxxxxxxxxxxx>
> > > > ---
> > > >    drivers/usb/typec/tcpm/tcpm.c | 24 ++++++++++++++++++++++++
> > > >    1 file changed, 24 insertions(+)
> > > >
> > > > diff --git a/drivers/usb/typec/tcpm/tcpm.c b/drivers/usb/typec/tcpm/tcpm.c
> > > > index 13830b5e2d09f..d6ad3cdf9e4af 100644
> > > > --- a/drivers/usb/typec/tcpm/tcpm.c
> > > > +++ b/drivers/usb/typec/tcpm/tcpm.c
> > > > @@ -6320,6 +6320,27 @@ static int tcpm_psy_get_current_now(struct tcpm_port *port,
> > > >    	return 0;
> > > >    }
> > > > +static int tcpm_psy_get_input_power_limit(struct tcpm_port *port,
> > > > +					  union power_supply_propval *val)
> > > > +{
> > > > +	unsigned int src_mv, src_ma, max_src_mw = 0;
> > > > +	unsigned int i, tmp;
> > > > +
> > > > +	for (i = 0; i < port->nr_source_caps; i++) {
> > > > +		u32 pdo = port->source_caps[i];
> > > > +
> > > > +		if (pdo_type(pdo) == PDO_TYPE_FIXED) {
> > > > +			src_mv = pdo_fixed_voltage(pdo);
> > > > +			src_ma = pdo_max_current(pdo);
> > > > +			tmp = src_mv * src_ma / 1000;
> > > > +			max_src_mw = tmp > max_src_mw ? tmp : max_src_mw;
> > > > +		}
> > > > +	}
> > > > +
> > > > +	val->intval = max_src_mw;
> > > The power-supply subsystem expects Microwatts and not Milliwatts.
> > Yes, but I see the 'power_supply_propval' member 'intval' is an integer
> > type, I worry about it may be overflowed that uses Microwatts.
> Data being encoded in Microwatts is part of the ABI. The data
> you are supplying will be interpreted in µW. If you submit your
> data in mW it is basically always wrong even without an overflow.
>
> Now regarding the overflow: A signed int can store 2^31 bit, so
> 2,147,483,648 µW = 2147 W. Looking at your code you effectively
> calculate Microwatts in an unsigned int and then divide by 1000.
> Since the intermediate value (before dividing by 1000) needs to be
> stored you gain only one bit. That raises the question: Why do you
> expect data to be between 2147 W and 4294 W when the latest released
> USB PD spec allows 5A@48V = 240W?
>
> -- Sebastian

Okay, got it, I shall delete conversion codes ( divide by 1000 ) in the 
next version.


BR.
Frank