Collection of various issues related to I2C API

[zephyrbot]

Reported by Piotr Mienkowski:

I2C API should be updated to fix all linked issues.

• I2C Slave Driver #2956: I2C Slave Driver (PR Add support for I2C Slave Drivers #5224)
• i2c: Update API to support arbitrary bus speed values #4039: Update I2C API to support arbitrary bus speed values
• I2C API is mixing two incompatible definitions of bit-fields #4009: I2C API is mixing two incompatible definitions of bit-fields
• I2C: move "Use 10-bit addressing" option to i2c_msg #3806: I2C: move "Use 10-bit addressing" option to i2c_msg (PR drivers: i2c: Move "Use 10-bit addressing" option to i2c_msg #9334)
• Add support in I2C shim driver for DMA transfer mode #2001: Add support in I2C shim driver for DMA transfer mode
• driver: i2c_mcux: unable to perform more than one write transfer like i2c_burst_write #8684: unable to perform more than one write transfer like i2c_burst_write
• I2C slave DTS support #9920: I2C slave DTS support (PR [WIP] drivers: i2c: Add dts support for i2c slaves. #9943)
• i2c_burst_write on nrf51 is not a burst write #11612: i2c_burst_write on nrf51 is not a burst write
• i2c: API should define fine-grained error responses #12277: i2c should define fine-grained error responses
• Improve caching configuration and move it to be cross architecture #26760: i2c: complete support for 10-bit addressing
• i2c: complete and stabilize support for target i2c devices #27675: i2c: complete and stabilize support for "slave" I2C devices
• i2c: consistent support for bus timeouts and recovery #27715: i2c: consistent support for bus timeouts and recovery

(Imported from Jira ZEP-2612)

[nashif]

@MaureenHelm to submit a proposal for #4039 and get comments in a PR

[endian-benjamin]

What's the status on async I2C? Asking because of #1387 mainly

[mnkp]

There are a few more changes to the I2C API which might be considered

• Despite the fact that PR drivers: i2c: Move "Use 10-bit addressing" option to i2c_msg #9334 moved 10-bit addressing flag to struct i2c_msg - this allows to use 10-bit addressing on a per message basis - the only function that takes struct i2c_msg as a parameter is i2c_transfer. No other function e.g. i2c_write, i2c_reg_read_byte allows to use 10-bit addressing. As such 10-bit addressing is not really supported by the API. added as i2c: complete support for 10-bit addressing #27670
• while most of the functions follow the pattern int i2c_write_read(struct device *dev, uint16_t addr, where the second parameter is the device address on the i2c bus the i2c_write, i2c_read functions take it as the last parameter. This is quite confusing when reading the application code.

[pabigot]

API 2020-07-07: Moved from triage to in-progress, as there has been recent activity.

[carlescufi]

@teburd assigned to you as per @nashif

[nashif]

closing, the remaining issues have their own entries, so no need to keep this open