Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.35 as permitted sender) client-ip=23.128.96.35;
From:   Luca Weiss <luca.weiss@fairphone.com>
Subject: [PATCH v2 0/3] Handle reversed SBU orientation for FSA4480
Date:   Fri, 20 Oct 2023 11:33:17 +0200
Message-Id: <20231020-fsa4480-swap-v2-0-9a7f9bb59873@fairphone.com>
MIME-Version: 1.0
Content-Type: text/plain; charset="utf-8"
Content-Transfer-Encoding: 7bit
To:     Heikki Krogerus <heikki.krogerus@linux.intel.com>,
        Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
        Neil Armstrong <neil.armstrong@linaro.org>,
        Rob Herring <robh+dt@kernel.org>,
        Krzysztof Kozlowski <krzysztof.kozlowski+dt@linaro.org>,
        Conor Dooley <conor+dt@kernel.org>,
        Bjorn Andersson <andersson@kernel.org>
Cc:     ~postmarketos/upstreaming@lists.sr.ht, phone-devel@vger.kernel.org,
        linux-usb@vger.kernel.org, linux-kernel@vger.kernel.org,
        linux-arm-msm@vger.kernel.org, devicetree@vger.kernel.org,
        Luca Weiss <luca.weiss@fairphone.com>
Precedence: bulk

Short reason:
Without swapping the SBU lanes, on QCM6490 Fairphone 5 the
DisplayPort-over-USB-C doesn't work.

The Orient-Chip OCP96011 used in this phone is generally compatible with
FSA4480 but has a difference how AUX+/- should be connected to SBU1/2.

Long explanation, with my current understanding:
* FSA4480 block diagram shows AUX+ connected to SBU2 and AUX- to SBU1.
* OCP96011 block diagram shows AUX+ connected to SBU1 and AUX- to SBU2
  (it's not 100% clear though in the picture but makes sense with the
  observed behavior)
* Fairphone 5 schematics have AUX+ connected to SBU2 and AUX- to SBU1,
  which would be correct for FSA4480 but since OCP96011 is used (which
  expects it to be the other way around) the Linux driver needs to
  reverse it.
  If AUX+ would be connected to SBU1 and AUX- to SBU2 as shown in the
  OCP96011 block diagram, then no driver/dts change would be needed.

Not sure if I've implemented the best solution in this patch. Other
solutions I could think of are:
* Add some custom boolean property to the node, e.g. 'fsa,swap-sbu'
* Reverse when ocs,ocp96011 compatible is used. This would be incorrect
  since when following the OCP96011 block diagram no reversing would be
  needed, as explained above.

However I think the current solution with data-lanes in the endpoint is
the best fit and is also already used for a similar purpose in another
USB mux driver.

Signed-off-by: Luca Weiss <luca.weiss@fairphone.com>
---
Changes in v2:
- Expand commit message for first dt-bindings patch incl. ASCII art
- Use fwnode_* instead of_* APIs
- Drop manual check for data-lanes property length, read_u32_array will
  already error on too short or too long property.
- Drop dev_info for reversed data-lanes mapping
- Link to v1: https://lore.kernel.org/r/20231013-fsa4480-swap-v1-0-b877f62046cc@fairphone.com

---
Luca Weiss (3):
      dt-bindings: usb: fsa4480: Add data-lanes property to endpoint
      usb: typec: fsa4480: Add support to swap SBU orientation
      dt-bindings: usb: fsa4480: Add compatible for OCP96011

 .../devicetree/bindings/usb/fcs,fsa4480.yaml       | 43 ++++++++++++-
 drivers/usb/typec/mux/fsa4480.c                    | 71 ++++++++++++++++++++++
 2 files changed, 111 insertions(+), 3 deletions(-)
---
base-commit: e3b18f7200f45d66f7141136c25554ac1e82009b
change-id: 20231013-fsa4480-swap-9b0f76d73c19

Best regards,
-- 
Luca Weiss <luca.weiss@fairphone.com>