The interface between UCSI and the glue driver is very low-level. It
allows reading the UCSI data from any offset (but in reality the UCSI
driver reads only VERSION, CCI an MESSAGE_IN data). All event handling
is to be done by the glue driver (which already resulted in several
similar-but-slightly different implementations). It leaves no place to
optimize the write-read-read sequence for the command execution (which
might be beneficial for some of the drivers), etc.

The patchseries attempts to restructure the UCSI glue driver interface
in order to provide sensible operations instead of a low-level read /
write calls.

If this approach is found to be acceptable, I plan to further rework the
command interface, moving reading CCI and MESSAGE_IN to the common
control code, which should simplify driver's implementation and remove
necessity to split quirks between sync_control and read_message_in e.g.
as implemented in the ucsi_ccg.c.

Note, the series was tested only on the ucsi_glink platforms. Further
testing is appreciated.

Depends: [1], [2]

[1] https://lore.kernel.org/linux-usb/20240612124656.2305603-1-fabrice.gasnier@foss.st.com/

[2] https://lore.kernel.org/linux-usb/20240621-ucsi-yoga-ec-driver-v8-1-e03f3536b8c6@linaro.org/

Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
---
Changes in v3:
- Rebased on top of the next to resolve conflicts with the merge LG Gram
  patch (Heikki)
- Link to v2: https://lore.kernel.org/r/20240621-ucsi-rework-interface-v2-0-a399ff96bf88@linaro.org

Changes in v2:
- Dropped the RFC prefix
- Rebased on top of the fixed STM32 patch
- Included the pending Yoga C630 driver into the cleanup.
- Link to v1: https://lore.kernel.org/r/20240603-ucsi-rework-interface-v1-0-99a6d544cec8@linaro.org

---
Dmitry Baryshkov (7):
      usb: typec: ucsi: move ucsi_acknowledge() from ucsi_read_error()
      usb: typec: ucsi: simplify command sending API
      usb: typec: ucsi: split read operation
      usb: typec: ucsi: rework command execution functions
      usb: typec: ucsi: inline ucsi_read_message_in
      usb: typec: ucsi: extract common code for command handling
      usb: typec: ucsi: reorder operations in ucsi_run_command()

 drivers/usb/typec/ucsi/ucsi.c           | 215 +++++++++++++++++---------------
 drivers/usb/typec/ucsi/ucsi.h           |  26 ++--
 drivers/usb/typec/ucsi/ucsi_acpi.c      | 127 ++++++++++---------
 drivers/usb/typec/ucsi/ucsi_ccg.c       | 103 +++++++--------
 drivers/usb/typec/ucsi/ucsi_glink.c     |  74 ++++-------
 drivers/usb/typec/ucsi/ucsi_stm32g0.c   |  79 ++++--------
 drivers/usb/typec/ucsi/ucsi_yoga_c630.c | 104 +++++----------
 7 files changed, 322 insertions(+), 406 deletions(-)
---
base-commit: c0068dfdd1aaf0fa9d60367b0f392ce145ee27c3
change-id: 20240525-ucsi-rework-interface-5ff2264f6aec

Best regards,
-- 
Dmitry Baryshkov <dmitry.baryshkov@linaro.org>

On Tue, Jun 25, 2024 at 05:54:25PM +0300, Dmitry Baryshkov wrote:
> The interface between UCSI and the glue driver is very low-level. It
> allows reading the UCSI data from any offset (but in reality the UCSI
> driver reads only VERSION, CCI an MESSAGE_IN data). All event handling
> is to be done by the glue driver (which already resulted in several
> similar-but-slightly different implementations). It leaves no place to
> optimize the write-read-read sequence for the command execution (which
> might be beneficial for some of the drivers), etc.
> 
> The patchseries attempts to restructure the UCSI glue driver interface
> in order to provide sensible operations instead of a low-level read /
> write calls.
> 
> If this approach is found to be acceptable, I plan to further rework the
> command interface, moving reading CCI and MESSAGE_IN to the common
> control code, which should simplify driver's implementation and remove
> necessity to split quirks between sync_control and read_message_in e.g.
> as implemented in the ucsi_ccg.c.
> 
> Note, the series was tested only on the ucsi_glink platforms. Further
> testing is appreciated.

I tested these on couple of systems that use the acpi mailbox, and
didn't see any problems. I'll be away for most of July, so if there's
nothing else, for the series:

Tested-by: Heikki Krogerus <heikki.krogerus@linux.intel.com>
Reviewed-by: Heikki Krogerus <heikki.krogerus@linux.intel.com>

thanks,

-- 
heikki

On Wed, Jun 26, 2024 at 05:24:22PM GMT, Heikki Krogerus wrote:
> On Tue, Jun 25, 2024 at 05:54:25PM +0300, Dmitry Baryshkov wrote:
> > The interface between UCSI and the glue driver is very low-level. It
> > allows reading the UCSI data from any offset (but in reality the UCSI
> > driver reads only VERSION, CCI an MESSAGE_IN data). All event handling
> > is to be done by the glue driver (which already resulted in several
> > similar-but-slightly different implementations). It leaves no place to
> > optimize the write-read-read sequence for the command execution (which
> > might be beneficial for some of the drivers), etc.
> > 
> > The patchseries attempts to restructure the UCSI glue driver interface
> > in order to provide sensible operations instead of a low-level read /
> > write calls.
> > 
> > If this approach is found to be acceptable, I plan to further rework the
> > command interface, moving reading CCI and MESSAGE_IN to the common
> > control code, which should simplify driver's implementation and remove
> > necessity to split quirks between sync_control and read_message_in e.g.
> > as implemented in the ucsi_ccg.c.
> > 
> > Note, the series was tested only on the ucsi_glink platforms. Further
> > testing is appreciated.
> 
> I tested these on couple of systems that use the acpi mailbox, and
> didn't see any problems. I'll be away for most of July, so if there's
> nothing else, for the series:
> 
> Tested-by: Heikki Krogerus <heikki.krogerus@linux.intel.com>
> Reviewed-by: Heikki Krogerus <heikki.krogerus@linux.intel.com>
> 

Thanks!

-- 
With best wishes
Dmitry

On Tue, Jun 25, 2024 at 05:54:25PM +0300, Dmitry Baryshkov wrote:
> The interface between UCSI and the glue driver is very low-level. It
> allows reading the UCSI data from any offset (but in reality the UCSI
> driver reads only VERSION, CCI an MESSAGE_IN data). All event handling
> is to be done by the glue driver (which already resulted in several
> similar-but-slightly different implementations). It leaves no place to
> optimize the write-read-read sequence for the command execution (which
> might be beneficial for some of the drivers), etc.
> 
> The patchseries attempts to restructure the UCSI glue driver interface
> in order to provide sensible operations instead of a low-level read /
> write calls.
> 
> If this approach is found to be acceptable, I plan to further rework the
> command interface, moving reading CCI and MESSAGE_IN to the common
> control code, which should simplify driver's implementation and remove
> necessity to split quirks between sync_control and read_message_in e.g.
> as implemented in the ucsi_ccg.c.
> 
> Note, the series was tested only on the ucsi_glink platforms. Further
> testing is appreciated.
> 
> Depends: [1], [2]
> 
> [1] https://lore.kernel.org/linux-usb/20240612124656.2305603-1-fabrice.gasnier@foss.st.com/
> 
> [2] https://lore.kernel.org/linux-usb/20240621-ucsi-yoga-ec-driver-v8-1-e03f3536b8c6@linaro.org/
> 
> Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>

This doesn't apply to my tree at all, can you rebase and add Heikki's
review and resend?

thanks,

greg k-h