This patch set is based on Felipe's usb.git / testing/next branch (commit id = 47265c067c0d129f3a0e94bc221293a780af9d78). I still marked this patch set as "RFC". I would like to know whether this way is good or not. About last discusstion with Heikki: https://patchwork.kernel.org/patch/10397635/ Changes from RFC v3: - Rebase latest usb.git / testing/next branch. - Add graph parse into device_connection_find_match(). - Use workqueue to call _usb3_set_mode() in patch 3. (I realized renesas_usb3_role_switch_set() cannot run on atomic because device_attach() might sleep.) Changes from RFC v2: - Add registering usb role switch into drivers/usb/gadget/udc/renesas_usb3 because hardware resource (a register) is shared and remove individual usb role switch driver/dt-bindings for R-Car. - Remove "usb_role_switch_get_by_graph" API because the renesas_usb3 driver doesn't need such API now. Changes from RFC: - Remove "device-connection-id" and "usb role switch driver" dt-bingings. - Remove drivers/of code. - Add a new API for find the connection by using graph on devcon.c and roles.c. - Use each new API on the rcar usb role switch and renesas_usb3 drivers. - Update the dtsi file for r8a7795. Yoshihiro Shimoda (4): base: devcon: add graph parse in device_connection_find_match() usb: gadget: udc: renesas_usb3: Add register of usb role switch usb: gadget: udc: renesas_usb3: use usb role switch API arm64: dts: renesas: r8a7795: add OF graph for usb role switch .../devicetree/bindings/usb/renesas_usb3.txt | 15 +++ Documentation/driver-api/device_connection.rst | 2 +- arch/arm64/boot/dts/renesas/r8a7795.dtsi | 12 +++ drivers/base/devcon.c | 15 +++ drivers/usb/gadget/udc/Kconfig | 1 + drivers/usb/gadget/udc/renesas_usb3.c | 105 ++++++++++++++++++++- include/linux/device.h | 2 + 7 files changed, 150 insertions(+), 2 deletions(-) -- 1.9.1
