On Thu, 13 Jun 2024 at 10:30, Ilpo Järvinen
<ilpo.jarvinen@xxxxxxxxxxxxxxx> wrote:
>
> On Wed, 12 Jun 2024, Dmitry Baryshkov wrote:
>
> > The Lenovo Yoga C630 WOS laptop provides implements UCSI interface in
> > the onboard EC. Add glue driver to interface the platform's UCSI
> > implementation.
> >
> > Reviewed-by: Bryan O'Donoghue <bryan.odonoghue@xxxxxxxxxx>
> > Reviewed-by: Heikki Krogerus <heikki.krogerus@xxxxxxxxxxxxxxx>
> > Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@xxxxxxxxxx>
> > ---
>
> > +static int yoga_c630_ucsi_read(struct ucsi *ucsi, unsigned int offset,
> > + void *val, size_t val_len)
> > +{
> > + struct yoga_c630_ucsi *uec = ucsi_get_drvdata(ucsi);
> > + u8 buf[YOGA_C630_UCSI_READ_SIZE];
> > + int ret;
> > +
> > + ret = yoga_c630_ec_ucsi_read(uec->ec, buf);
> > + if (ret)
> > + return ret;
> > +
> > + if (offset == UCSI_VERSION) {
> > + memcpy(val, &uec->version, min(val_len, sizeof(uec->version)));
> > + return 0;
> > + }
> > +
> > + if (offset == UCSI_CCI)
> > + memcpy(val, buf, min(val_len, YOGA_C630_UCSI_CCI_SIZE));
> > + else if (offset == UCSI_MESSAGE_IN)
> > + memcpy(val, buf + YOGA_C630_UCSI_CCI_SIZE,
> > + min(val_len, YOGA_C630_UCSI_DATA_SIZE));
> > + else
> > + return -EINVAL;
> > +
> > + return 0;
>
> Hmm, the inconsistency when to do return 0 is a bit odd. Also, using
> switch (offset) would probably be better here anyway to replace all the
> ifs.
I'll see if I can improve this bit.
>
> > +}
> > +
> > +static int yoga_c630_ucsi_async_write(struct ucsi *ucsi, unsigned int offset,
> > + const void *val, size_t val_len)
> > +{
> > + struct yoga_c630_ucsi *uec = ucsi_get_drvdata(ucsi);
> > +
> > + if (offset != UCSI_CONTROL ||
> > + val_len != YOGA_C630_UCSI_WRITE_SIZE)
> > + return -EINVAL;
> > +
> > + return yoga_c630_ec_ucsi_write(uec->ec, val);
> > +}
> > +
> > +static int yoga_c630_ucsi_sync_write(struct ucsi *ucsi, unsigned int offset,
> > + const void *val, size_t val_len)
> > +{
> > + struct yoga_c630_ucsi *uec = ucsi_get_drvdata(ucsi);
> > + bool ack = UCSI_COMMAND(*(u64 *)val) == UCSI_ACK_CC_CI;
> > + int ret;
> > +
> > + if (ack)
> > + set_bit(UCSI_C630_ACK_PENDING, &uec->flags);
> > + else
> > + set_bit(UCSI_C630_COMMAND_PENDING, &uec->flags);
> > +
> > + reinit_completion(&uec->complete);
> > +
> > + ret = yoga_c630_ucsi_async_write(ucsi, offset, val, val_len);
> > + if (ret)
> > + goto out_clear_bit;
> > +
> > + if (!wait_for_completion_timeout(&uec->complete, 5 * HZ))
> > + ret = -ETIMEDOUT;
> > +
> > +out_clear_bit:
> > + if (ack)
> > + clear_bit(UCSI_C630_ACK_PENDING, &uec->flags);
> > + else
> > + clear_bit(UCSI_C630_COMMAND_PENDING, &uec->flags);
> > +
> > + return ret;
> > +}
> > +
> > +const struct ucsi_operations yoga_c630_ucsi_ops = {
> > + .read = yoga_c630_ucsi_read,
> > + .sync_write = yoga_c630_ucsi_sync_write,
> > + .async_write = yoga_c630_ucsi_async_write,
> > +};
> > +
> > +static void yoga_c630_ucsi_notify_ucsi(struct yoga_c630_ucsi *uec, u32 cci)
> > +{
> > + if (UCSI_CCI_CONNECTOR(cci))
> > + ucsi_connector_change(uec->ucsi, UCSI_CCI_CONNECTOR(cci));
> > +
> > + if (cci & UCSI_CCI_ACK_COMPLETE &&
> > + test_bit(UCSI_C630_ACK_PENDING, &uec->flags))
> > + complete(&uec->complete);
> > +
> > + if (cci & UCSI_CCI_COMMAND_COMPLETE &&
> > + test_bit(UCSI_C630_COMMAND_PENDING, &uec->flags))
> > + complete(&uec->complete);
>
> Is this racy? Can another command start after an ACK in between these two
> ifs and complete() is called prematurely for the new command? (Or will
> different value in cci protect against that?)
No, there is no race. The UCSI is locked for the duration of the command.
>
> > +}
> > +
> > +static int yoga_c630_ucsi_notify(struct notifier_block *nb,
> > + unsigned long action, void *data)
> > +{
> > + struct yoga_c630_ucsi *uec = container_of(nb, struct yoga_c630_ucsi, nb);
> > + u32 cci;
> > + int ret;
> > +
> > + switch (action) {
> > + case LENOVO_EC_EVENT_USB:
> > + case LENOVO_EC_EVENT_HPD:
> > + ucsi_connector_change(uec->ucsi, 1);
> > + return NOTIFY_OK;
> > +
> > + case LENOVO_EC_EVENT_UCSI:
> > + ret = uec->ucsi->ops->read(uec->ucsi, UCSI_CCI, &cci, sizeof(cci));
> > + if (ret)
> > + return NOTIFY_DONE;
> > +
> > + yoga_c630_ucsi_notify_ucsi(uec, cci);
> > +
> > + return NOTIFY_OK;
> > +
> > + default:
> > + return NOTIFY_DONE;
> > + }
> > +}
> > +
> > +static int yoga_c630_ucsi_probe(struct auxiliary_device *adev,
> > + const struct auxiliary_device_id *id)
> > +{
> > + struct yoga_c630_ec *ec = adev->dev.platform_data;
> > + struct yoga_c630_ucsi *uec;
> > + int ret;
> > +
> > + uec = devm_kzalloc(&adev->dev, sizeof(*uec), GFP_KERNEL);
> > + if (!uec)
> > + return -ENOMEM;
> > +
> > + uec->ec = ec;
> > + init_completion(&uec->complete);
> > + uec->nb.notifier_call = yoga_c630_ucsi_notify;
> > +
> > + uec->ucsi = ucsi_create(&adev->dev, &yoga_c630_ucsi_ops);
> > + if (IS_ERR(uec->ucsi))
> > + return PTR_ERR(uec->ucsi);
> > +
> > + ucsi_set_drvdata(uec->ucsi, uec);
> > +
> > + uec->version = yoga_c630_ec_ucsi_get_version(uec->ec);
> > +
> > + auxiliary_set_drvdata(adev, uec);
> > +
> > + ret = yoga_c630_ec_register_notify(ec, &uec->nb);
> > + if (ret)
> > + return ret;
> > +
> > + return ucsi_register(uec->ucsi);
> > +}
> > +
> > +static void yoga_c630_ucsi_remove(struct auxiliary_device *adev)
> > +{
> > + struct yoga_c630_ucsi *uec = auxiliary_get_drvdata(adev);
> > +
> > + yoga_c630_ec_unregister_notify(uec->ec, &uec->nb);
> > + ucsi_unregister(uec->ucsi);
>
> Usually, the remove should tear down in reverse order than the probe side.
> Is the divergence from that here intentional?
Yes, it's intentional, so that the driver doesn't get a notification
while UCSI is being torn down. Consider it to be paired with
ucsi_create().
--
With best wishes
Dmitry
