Sorry for the delay, I went on a long vacation prior to receiving feedback and got back in the middle of a hardware bring up that consumed all of my attention for an extended period of time. I will try to plan upstream submissions around my other responsibilities better in the future. Addressed comments from: - Vladimir in: https://www.spinics.net/lists/linux-i2c/msg27387.html and: https://www.spinics.net/lists/linux-i2c/msg27386.html - Wolfram in: https://www.spinics.net/lists/linux-i2c/msg27476.html and: https://www.spinics.net/lists/linux-i2c/msg27483.html Changes since previous update: - No longer arbitrarily restrict bus to be slave xor master. - Pulled out "struct aspeed_i2c_controller" as a interrupt controller. - Pulled out slave support into its own commit. - Rewrote code that sets clock divider register because the original version set it incorrectly. - Discovered and fixed issue in implementation that caused certain slave devices to misbehave; the cause was that the master IRQ handler would return control to the requesting thread after the last RX or TX command was handled such that the requesting thread would issue either a repeated start or stop. This was incorrect because the time taken to complete the completion was too great. I fixed this by rewriting the master IRQ handler so that it now manages the entire transaction only returning control to the requesting thread once the entire transaction is complete. - Rewrote the aspeed_i2c_master_irq handler because the old method of completing a completion in between restarts was too slow causing devices to misbehave. - Added support for I2C_M_RECV_LEN which I had incorrectly said was supported before. - Addressed other comments from Vladimir. Changes have been tested on the Aspeed 2500 evaluation board, as before, and now on a real platform with an Aspeed 2520. -- To unsubscribe from this list: send the line "unsubscribe devicetree" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
