On 7/15/2026 5:15 PM, Dmitry Guzman wrote:
On Wed, 15 Jul 2026 15:48:05 +0530 Aniket Randive aniket.randive@oss.qualcomm.com wrote:
This replaces the fixed 1-second timeout with a transfer-specific timeout while preserving sufficient margin for software overheads and bus-level delays.
The dynamic timeout may be useful for any I2C bus controller, not only for qcom-geni. Structure i2c_adapter already has field "timeout". Isn't
This is good, thanks for pointing. For geni driver, please review gi2c->adap.timeout (client->adapter->timeout) can be used and you may explore retries also in case you want to retry.
it worth to move the timeout calculation to i2c-core, so that the core updates the timeout field in i2c_adapter structure, and the controller driver just uses this value, instead of duplicating the calculation in every driver that will use this feature?
Also, such parameters as I2C_TIMEOUT_SAFETY_COEFFICIENT and I2C_TIMEOUT_MIN_USEC may be configurable by user (for example, in device tree) for more flexibility.
I think it's good to keep it within driver, not to expose as configurable. This would keep changing per user/system load, but ideal timing is theoretical + some offset, hence can be better managed by driver.
Best regards,