tcpm-source-psy- does not invoke power_supply_changed API when
one of the published power supply properties is changed.
power_supply_changed needs to be called to notify
userspace clients(uevents) and kernel clients.
Fixes: f2a8aa053c176("typec: tcpm: Represent source supply through power_supply")
Signed-off-by: Badhri Jagan Sridharan <[email protected]>
Reviewed-by: Guenter Roeck <[email protected]>
Reviewed-by: Heikki Krogerus <[email protected]>
---
Changes since V1:
- Fixed commit message as per Guenter's suggestion
- Added Reviewed-by tags
- cc'ed stable
---
drivers/usb/typec/tcpm/tcpm.c | 9 ++++++++-
1 file changed, 8 insertions(+), 1 deletion(-)
diff --git a/drivers/usb/typec/tcpm/tcpm.c b/drivers/usb/typec/tcpm/tcpm.c
index 11d0c40bc47d..e8936ea17f80 100644
--- a/drivers/usb/typec/tcpm/tcpm.c
+++ b/drivers/usb/typec/tcpm/tcpm.c
@@ -945,6 +945,7 @@ static int tcpm_set_current_limit(struct tcpm_port *port, u32 max_ma, u32 mv)
port->supply_voltage = mv;
port->current_limit = max_ma;
+ power_supply_changed(port->psy);
if (port->tcpc->set_current_limit)
ret = port->tcpc->set_current_limit(port->tcpc, max_ma, mv);
@@ -2931,6 +2932,7 @@ static int tcpm_pd_select_pdo(struct tcpm_port *port, int *sink_pdo,
port->pps_data.supported = false;
port->usb_type = POWER_SUPPLY_USB_TYPE_PD;
+ power_supply_changed(port->psy);
/*
* Select the source PDO providing the most power which has a
@@ -2955,6 +2957,7 @@ static int tcpm_pd_select_pdo(struct tcpm_port *port, int *sink_pdo,
port->pps_data.supported = true;
port->usb_type =
POWER_SUPPLY_USB_TYPE_PD_PPS;
+ power_supply_changed(port->psy);
}
continue;
default:
@@ -3112,6 +3115,7 @@ static unsigned int tcpm_pd_select_pps_apdo(struct tcpm_port *port)
port->pps_data.out_volt));
port->pps_data.op_curr = min(port->pps_data.max_curr,
port->pps_data.op_curr);
+ power_supply_changed(port->psy);
}
return src_pdo;
@@ -3347,6 +3351,7 @@ static int tcpm_set_charge(struct tcpm_port *port, bool charge)
return ret;
}
port->vbus_charge = charge;
+ power_supply_changed(port->psy);
return 0;
}
@@ -3530,6 +3535,7 @@ static void tcpm_reset_port(struct tcpm_port *port)
port->try_src_count = 0;
port->try_snk_count = 0;
port->usb_type = POWER_SUPPLY_USB_TYPE_C;
+ power_supply_changed(port->psy);
port->nr_sink_caps = 0;
port->sink_cap_done = false;
if (port->tcpc->enable_frs)
@@ -5957,7 +5963,7 @@ static int tcpm_psy_set_prop(struct power_supply *psy,
ret = -EINVAL;
break;
}
-
+ power_supply_changed(port->psy);
return ret;
}
@@ -6110,6 +6116,7 @@ struct tcpm_port *tcpm_register_port(struct device *dev, struct tcpc_dev *tcpc)
err = devm_tcpm_psy_register(port);
if (err)
goto out_role_sw_put;
+ power_supply_changed(port->psy);
port->typec_port = typec_register_port(port->dev, &port->typec_caps);
if (IS_ERR(port->typec_port)) {
--
2.31.0.rc2.261.g7f71774620-goog
On 17 March 2021 18:13, Badhri Jagan Sridharan wrote:
> tcpm-source-psy- does not invoke power_supply_changed API when
> one of the published power supply properties is changed.
> power_supply_changed needs to be called to notify
> userspace clients(uevents) and kernel clients.
>
> Fixes: f2a8aa053c176("typec: tcpm: Represent source supply through
> power_supply")
> Signed-off-by: Badhri Jagan Sridharan <[email protected]>
> Reviewed-by: Guenter Roeck <[email protected]>
> Reviewed-by: Heikki Krogerus <[email protected]>
> ---
> Changes since V1:
> - Fixed commit message as per Guenter's suggestion
> - Added Reviewed-by tags
> - cc'ed stable
> ---
> drivers/usb/typec/tcpm/tcpm.c | 9 ++++++++-
> 1 file changed, 8 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/usb/typec/tcpm/tcpm.c b/drivers/usb/typec/tcpm/tcpm.c
> index 11d0c40bc47d..e8936ea17f80 100644
> --- a/drivers/usb/typec/tcpm/tcpm.c
> +++ b/drivers/usb/typec/tcpm/tcpm.c
> @@ -945,6 +945,7 @@ static int tcpm_set_current_limit(struct tcpm_port *port,
> u32 max_ma, u32 mv)
>
> port->supply_voltage = mv;
> port->current_limit = max_ma;
> + power_supply_changed(port->psy);
>
> if (port->tcpc->set_current_limit)
> ret = port->tcpc->set_current_limit(port->tcpc, max_ma, mv);
> @@ -2931,6 +2932,7 @@ static int tcpm_pd_select_pdo(struct tcpm_port *port,
> int *sink_pdo,
>
> port->pps_data.supported = false;
> port->usb_type = POWER_SUPPLY_USB_TYPE_PD;
> + power_supply_changed(port->psy);
>
> /*
> * Select the source PDO providing the most power which has a
> @@ -2955,6 +2957,7 @@ static int tcpm_pd_select_pdo(struct tcpm_port *port,
> int *sink_pdo,
> port->pps_data.supported = true;
> port->usb_type =
> POWER_SUPPLY_USB_TYPE_PD_PPS;
> + power_supply_changed(port->psy);
> }
> continue;
> default:
> @@ -3112,6 +3115,7 @@ static unsigned int tcpm_pd_select_pps_apdo(struct
> tcpm_port *port)
> port->pps_data.out_volt));
> port->pps_data.op_curr = min(port->pps_data.max_curr,
> port->pps_data.op_curr);
> + power_supply_changed(port->psy);
> }
>
> return src_pdo;
Regarding selecting PDOs or PPS APDOs, surely we should only notify of a change
when we reach SNK_READY which means a new contract has been established? Until
that point it's possible any requested change could be rejected so why inform
clients before we know the settings have taken effect? I could be missing
something here as it's been a little while since I delved into this, but this
doesn't seem to make sense to me.
> @@ -3347,6 +3351,7 @@ static int tcpm_set_charge(struct tcpm_port *port, bool
> charge)
> return ret;
> }
> port->vbus_charge = charge;
> + power_supply_changed(port->psy);
> return 0;
> }
>
> @@ -3530,6 +3535,7 @@ static void tcpm_reset_port(struct tcpm_port *port)
> port->try_src_count = 0;
> port->try_snk_count = 0;
> port->usb_type = POWER_SUPPLY_USB_TYPE_C;
> + power_supply_changed(port->psy);
This is already taken care of at the end of this function, isn't it?
> port->nr_sink_caps = 0;
> port->sink_cap_done = false;
> if (port->tcpc->enable_frs)
> @@ -5957,7 +5963,7 @@ static int tcpm_psy_set_prop(struct power_supply
> *psy,
> ret = -EINVAL;
> break;
> }
> -
> + power_supply_changed(port->psy);
> return ret;
> }
>
> @@ -6110,6 +6116,7 @@ struct tcpm_port *tcpm_register_port(struct device
> *dev, struct tcpc_dev *tcpc)
> err = devm_tcpm_psy_register(port);
> if (err)
> goto out_role_sw_put;
> + power_supply_changed(port->psy);
>
> port->typec_port = typec_register_port(port->dev, &port->typec_caps);
> if (IS_ERR(port->typec_port)) {
> --
> 2.31.0.rc2.261.g7f71774620-goog
On Thu, Mar 18, 2021 at 8:26 AM Adam Thomson
<[email protected]> wrote:
>
> On 17 March 2021 18:13, Badhri Jagan Sridharan wrote:
>
> > tcpm-source-psy- does not invoke power_supply_changed API when
> > one of the published power supply properties is changed.
> > power_supply_changed needs to be called to notify
> > userspace clients(uevents) and kernel clients.
> >
> > Fixes: f2a8aa053c176("typec: tcpm: Represent source supply through
> > power_supply")
> > Signed-off-by: Badhri Jagan Sridharan <[email protected]>
> > Reviewed-by: Guenter Roeck <[email protected]>
> > Reviewed-by: Heikki Krogerus <[email protected]>
> > ---
> > Changes since V1:
> > - Fixed commit message as per Guenter's suggestion
> > - Added Reviewed-by tags
> > - cc'ed stable
> > ---
> > drivers/usb/typec/tcpm/tcpm.c | 9 ++++++++-
> > 1 file changed, 8 insertions(+), 1 deletion(-)
> >
> > diff --git a/drivers/usb/typec/tcpm/tcpm.c b/drivers/usb/typec/tcpm/tcpm.c
> > index 11d0c40bc47d..e8936ea17f80 100644
> > --- a/drivers/usb/typec/tcpm/tcpm.c
> > +++ b/drivers/usb/typec/tcpm/tcpm.c
> > @@ -945,6 +945,7 @@ static int tcpm_set_current_limit(struct tcpm_port *port,
> > u32 max_ma, u32 mv)
> >
> > port->supply_voltage = mv;
> > port->current_limit = max_ma;
> > + power_supply_changed(port->psy);
> >
> > if (port->tcpc->set_current_limit)
> > ret = port->tcpc->set_current_limit(port->tcpc, max_ma, mv);
> > @@ -2931,6 +2932,7 @@ static int tcpm_pd_select_pdo(struct tcpm_port *port,
> > int *sink_pdo,
> >
> > port->pps_data.supported = false;
> > port->usb_type = POWER_SUPPLY_USB_TYPE_PD;
> > + power_supply_changed(port->psy);
> >
> > /*
> > * Select the source PDO providing the most power which has a
> > @@ -2955,6 +2957,7 @@ static int tcpm_pd_select_pdo(struct tcpm_port *port,
> > int *sink_pdo,
> > port->pps_data.supported = true;
> > port->usb_type =
> > POWER_SUPPLY_USB_TYPE_PD_PPS;
> > + power_supply_changed(port->psy);
> > }
> > continue;
> > default:
> > @@ -3112,6 +3115,7 @@ static unsigned int tcpm_pd_select_pps_apdo(struct
> > tcpm_port *port)
> > port->pps_data.out_volt));
> > port->pps_data.op_curr = min(port->pps_data.max_curr,
> > port->pps_data.op_curr);
> > + power_supply_changed(port->psy);
> > }
> >
> > return src_pdo;
>
> Regarding selecting PDOs or PPS APDOs, surely we should only notify of a change
> when we reach SNK_READY which means a new contract has been established? Until
> that point it's possible any requested change could be rejected so why inform
> clients before we know the settings have taken effect? I could be missing
> something here as it's been a little while since I delved into this, but this
> doesn't seem to make sense to me.
I was trying to keep the power_supply_changed call close to the
variables which are used to infer the power supply property values.
Since port->pps_data.max_curr is already updated here and that's used
to infer the CURRENT_MAX a client could still read this before the
request goes through right ?
>
> > @@ -3347,6 +3351,7 @@ static int tcpm_set_charge(struct tcpm_port *port, bool
> > charge)
> > return ret;
> > }
> > port->vbus_charge = charge;
> > + power_supply_changed(port->psy);
> > return 0;
> > }
> >
> > @@ -3530,6 +3535,7 @@ static void tcpm_reset_port(struct tcpm_port *port)
> > port->try_src_count = 0;
> > port->try_snk_count = 0;
> > port->usb_type = POWER_SUPPLY_USB_TYPE_C;
> > + power_supply_changed(port->psy);
>
> This is already taken care of at the end of this function, isn't it?
I thought I deleted that. Looks like I didn't. Will send in a patch to
remove that.
Will wait for what we decide for the PPS case so that I can send in
both in the same patch.
Thanks,
Badhri
>
> > port->nr_sink_caps = 0;
> > port->sink_cap_done = false;
> > if (port->tcpc->enable_frs)
> > @@ -5957,7 +5963,7 @@ static int tcpm_psy_set_prop(struct power_supply
> > *psy,
> > ret = -EINVAL;
> > break;
> > }
> > -
> > + power_supply_changed(port->psy);
> > return ret;
> > }
> >
> > @@ -6110,6 +6116,7 @@ struct tcpm_port *tcpm_register_port(struct device
> > *dev, struct tcpc_dev *tcpc)
> > err = devm_tcpm_psy_register(port);
> > if (err)
> > goto out_role_sw_put;
> > + power_supply_changed(port->psy);
> >
> > port->typec_port = typec_register_port(port->dev, &port->typec_caps);
> > if (IS_ERR(port->typec_port)) {
> > --
> > 2.31.0.rc2.261.g7f71774620-goog
>
On 18 March 2021 20:40, Badhri Jagan Sridharan wrote:
> > Regarding selecting PDOs or PPS APDOs, surely we should only notify of a
> change
> > when we reach SNK_READY which means a new contract has been established?
> Until
> > that point it's possible any requested change could be rejected so why inform
> > clients before we know the settings have taken effect? I could be missing
> > something here as it's been a little while since I delved into this, but this
> > doesn't seem to make sense to me.
>
> I was trying to keep the power_supply_changed call close to the
> variables which are used to infer the power supply property values.
> Since port->pps_data.max_curr is already updated here and that's used
> to infer the CURRENT_MAX a client could still read this before the
> request goes through right ?
Actually that's fair but I think the problem here relates to 'max_curr' not
being reset if the SRC rejects our request when we're swapping between one PPS
APDO and another PPS APDO. I think the 'max_curr' value should be reverted back
to the value for the existing PPS APDO we were already using. I suspect the same
might be true of 'min_volt' and 'max_volt' as well, now I look at it. It might
actually be prudent to have pending PPS data based on a request, which is only
committed as active once ACCEPT has been received.
Regarding power_supply_changed() though, I still think we should only notify of
a change when the requested change has been accepted by the source, in relation
to these values as they should reflect the real, in-use voltage and current
values.
Hi Adam,
Just sent out a patch stack
https://patchwork.kernel.org/project/linux-usb/list/?series=461087
to address the issue that you mentioned here.
Thanks,
Badhri
On Mon, Apr 5, 2021 at 6:43 PM Badhri Jagan Sridharan <[email protected]> wrote:
>
> Hi Adam,
>
> Just sent out a patch stack https://patchwork.kernel.org/project/linux-usb/list/?series=461087
> to address the issue that you mentioned here.
>
> Thanks,
> Badhri
>
> On Fri, Mar 19, 2021 at 9:32 AM Adam Thomson <[email protected]> wrote:
>>
>> On 18 March 2021 20:40, Badhri Jagan Sridharan wrote:
>>
>> > > Regarding selecting PDOs or PPS APDOs, surely we should only notify of a
>> > change
>> > > when we reach SNK_READY which means a new contract has been established?
>> > Until
>> > > that point it's possible any requested change could be rejected so why inform
>> > > clients before we know the settings have taken effect? I could be missing
>> > > something here as it's been a little while since I delved into this, but this
>> > > doesn't seem to make sense to me.
>> >
>> > I was trying to keep the power_supply_changed call close to the
>> > variables which are used to infer the power supply property values.
>> > Since port->pps_data.max_curr is already updated here and that's used
>> > to infer the CURRENT_MAX a client could still read this before the
>> > request goes through right ?
>>
>> Actually that's fair but I think the problem here relates to 'max_curr' not
>> being reset if the SRC rejects our request when we're swapping between one PPS
>> APDO and another PPS APDO. I think the 'max_curr' value should be reverted back
>> to the value for the existing PPS APDO we were already using. I suspect the same
>> might be true of 'min_volt' and 'max_volt' as well, now I look at it. It might
>> actually be prudent to have pending PPS data based on a request, which is only
>> committed as active once ACCEPT has been received.
>>
>> Regarding power_supply_changed() though, I still think we should only notify of
>> a change when the requested change has been accepted by the source, in relation
>> to these values as they should reflect the real, in-use voltage and current
>> values.
>>