Re: [PATCH 1/2] usb: typec: ucsi: fix UCSI on buggy Qualcomm devices

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Tue, Oct 24, 2023 at 11:13:55PM +0300, Dmitry Baryshkov wrote:
> On Tue, 24 Oct 2023 at 19:42, Bjorn Andersson <quic_bjorande@xxxxxxxxxxx> wrote:
> >
> > On Tue, Oct 24, 2023 at 02:08:33AM +0300, Dmitry Baryshkov wrote:
> > > On Tue, 24 Oct 2023 at 01:47, Bjorn Andersson <quic_bjorande@xxxxxxxxxxx> wrote:
> > > >
> > > > On Tue, Oct 24, 2023 at 12:47:26AM +0300, Dmitry Baryshkov wrote:
> > > > > On sevral Qualcomm platforms (SC8180X, SM8350, SC8280XP) a call to
> > > > > UCSI_GET_PDOS for non-PD partners will cause a firmware crash with no
> > > > > easy way to recover from it. Since we have no easy way to determine
> > > > > whether the partner really has PD support, shortcut UCSI_GET_PDOS on
> > > > > such platforms. This allows us to enable UCSI support on such devices.
> > > > >
> > > >
> > > > Really nice to see this. Thanks.
> > > >
> > > > > Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@xxxxxxxxxx>
> > > > > ---
> > > > >  drivers/usb/typec/ucsi/ucsi.c       | 3 +++
> > > > >  drivers/usb/typec/ucsi/ucsi.h       | 3 +++
> > > > >  drivers/usb/typec/ucsi/ucsi_glink.c | 3 +++
> > > > >  3 files changed, 9 insertions(+)
> > > > >
> > > > > diff --git a/drivers/usb/typec/ucsi/ucsi.c b/drivers/usb/typec/ucsi/ucsi.c
> > > > > index 61b64558f96c..5392ec698959 100644
> > > > > --- a/drivers/usb/typec/ucsi/ucsi.c
> > > > > +++ b/drivers/usb/typec/ucsi/ucsi.c
> > > > > @@ -578,6 +578,9 @@ static int ucsi_read_pdos(struct ucsi_connector *con,
> > > > >       u64 command;
> > > > >       int ret;
> > > > >
> > > > > +     if (ucsi->quirks & UCSI_NO_PARTNER_PDOS)
> > > > > +             return 0;
> > > > > +
> > > > >       command = UCSI_COMMAND(UCSI_GET_PDOS) | UCSI_CONNECTOR_NUMBER(con->num);
> > > > >       command |= UCSI_GET_PDOS_PARTNER_PDO(is_partner);
> > > > >       command |= UCSI_GET_PDOS_PDO_OFFSET(offset);
> > > > > diff --git a/drivers/usb/typec/ucsi/ucsi.h b/drivers/usb/typec/ucsi/ucsi.h
> > > > > index 474315a72c77..6478016d5cb8 100644
> > > > > --- a/drivers/usb/typec/ucsi/ucsi.h
> > > > > +++ b/drivers/usb/typec/ucsi/ucsi.h
> > > > > @@ -317,6 +317,9 @@ struct ucsi {
> > > > >  #define EVENT_PENDING        0
> > > > >  #define COMMAND_PENDING      1
> > > > >  #define ACK_PENDING  2
> > > > > +
> > > > > +     unsigned long quirks;
> > > > > +#define UCSI_NO_PARTNER_PDOS BIT(0)  /* Don't read partner's PDOs */
> > > > >  };
> > > > >
> > > > >  #define UCSI_MAX_SVID                5
> > > > > diff --git a/drivers/usb/typec/ucsi/ucsi_glink.c b/drivers/usb/typec/ucsi/ucsi_glink.c
> > > > > index db6e248f8208..5c159e7b2b65 100644
> > > > > --- a/drivers/usb/typec/ucsi/ucsi_glink.c
> > > > > +++ b/drivers/usb/typec/ucsi/ucsi_glink.c
> > > > > @@ -327,6 +327,8 @@ static int pmic_glink_ucsi_probe(struct auxiliary_device *adev,
> > > > >       if (ret)
> > > > >               return ret;
> > > > >
> > > > > +     ucsi->ucsi->quirks = id->driver_data;
> > > > > +
> > > > >       ucsi_set_drvdata(ucsi->ucsi, ucsi);
> > > > >
> > > > >       device_for_each_child_node(dev, fwnode) {
> > > > > @@ -379,6 +381,7 @@ static void pmic_glink_ucsi_remove(struct auxiliary_device *adev)
> > > > >
> > > > >  static const struct auxiliary_device_id pmic_glink_ucsi_id_table[] = {
> > > > >       { .name = "pmic_glink.ucsi", },
> > > > > +     { .name = "pmic_glink.ucsi-no-pdos", .driver_data = UCSI_NO_PARTNER_PDOS, },
> > > >
> > > > In altmode and battmgr drivers we apply quirks based on the compatible
> > > > of the pmic_glink of_node.
> > >
> > > ... and I can't say that I like that. In typical drivers we perform
> > > driver tuning by looking at the device's data (e.g. by using
> > > of_device_is_compatible or by of_device_get_match_data. Checking the
> > > parent device seems like breaking the layering.
> >
> > It felt like it was the cleaner option of the two when I did it. I think
> > there was some variation of quirks which made me feel this would grow
> > large - but I might misremember things now.
> >
> > > But if you insist, I can follow that approach.
> >
> > I insist that we should use the same mechanism of dealing with the
> > quirks across the three parts, and following the existing approach
> > doesn't seem too unreasonable...
> 
> The problem with the current approach is that it adds dependency
> between patches. We can not apply patch2 without patch1 being in
> place, since applying will enable buggy UCSI.
> 

Good point. Please describe this dependency when you respin the patches,
and we can take them together through the USB tree.

Regards,
Bjorn




[Index of Archives]     [Linux Media]     [Linux Input]     [Linux Audio Users]     [Yosemite News]     [Linux Kernel]     [Linux SCSI]     [Old Linux USB Devel Archive]

  Powered by Linux