On Mon, 09 May 2022 16:59:31 +0200, Daniel Kaehn wrote:
Generic serial MIDI driver adding support for using serial devices compatible with the serial bus as raw MIDI devices, allowing using additional serial devices not compatible with the existing serial-u16550 driver. Supports only setting standard serial baudrates on the underlying serial device; however, the underlying serial device can be configured so that a requested 38.4 kBaud is actually the standard MIDI 31.25 kBaud. Supports DeviceTree configuration.
Changes in v7:
- Separate examples in dt-binding to remove need to specify unit name (fixing dt_binding_check error)
Changes in v6:
- Change compatible "serialmidi" -> serial-midi"
- Default current-speed to 38400 (closest baud to MIDI standard speed)
- Appropriately stop reading or writing MIDI if input or output _trigger() is called with a parameter of zero, respectively
- Zero out corresponding triggered state on close to ensure input and output closing results in the serial port being closed
- Fix order of operations in _probe()
- Remove "DEBUG" literal from debug messages
- Remove unused dt-parsing patch checking for existence of node
- Whitespace / tabbing fixes / improvements
Changes in v5:
- Reword description in dt-binding for clarity
- Change 'speed' dt property to standard 'current-speed'
- Move MIDI output loop onto workqueue (since this could loop quite a while, if ALSA provides a continuous stream of bytes)
- Add tx_state bit flags to snd_serial_generic struct
- Safegard critical section in tx_work with atomic bit ops on tx_state
- Switch operations on filemode to use atomic bit ops
Changes in v4:
- Fix regressed typo - Correct 3.84 kBaud -> 38.4 kBaud in DT & Kconfig (sorry about spam - noticed after sending v3 and didn't want to let the error sit around for too long)
Changes in v3:
- Replace use of snd_printk() with dev_* alternatives
- Removed unnecessary initialization of err variables
- Replaced instances of `== SERIAL_MODE_NOT_OPENED` with zero check
- Loop on output_write to completely fill output buffer if data available
- Depend on CONFIG_OF in Kconfig
- Replace use of devm_kzalloc() with extra_size allocation in snd_devm_card_new()
- Use module_serdev_device_driver() instead of module_init() and module_exit(0)
Changes in v2:
- Fix 'snd_serial_generic_write_wakeup' missing static keyword
- Correct 3.125 kBaud > 31.25 kBaud in documentation for MIDI
The need for this driver arose from a project using a Raspberry Pi4 which needed to receive and send raw MIDI with low latency. The pl011 UART used is not compatible with the existing serial MIDI driver made for u16550-style devices. Using a userspace program such as ttymidi to feed input from the TTY device to a virtual ALSA MIDI device was functional, but not ideal.
I am not sure if a MIDI driver needing the mentioned 'hack' to clock 38.4 kBaud down to the standard MIDI baud is permissible in the mainline kernel, but am submitting nevertheless in case it is useful. To my knowledge, it doesn't seem that there would be any way for this driver to manually configure a serial port to 31.25 kBaud using the serial bus API (please correct me f I'm wrong). In my use case, I am actually configuring one port to run at 115.2 kBaud for faster communication with a custom onboard MIDI controller.
Daniel Kaehn (2): dt-bindings: sound: Add generic serial MIDI device Add generic serial MIDI driver using serial bus API
Now applied both patches to for-next branch.
thanks,
Takashi