On Mon, Nov 25, 2024 at 05:26 PM, William Breathitt Gray wrote: > Hello Rafael, > > I wonder whether a new enum option is actually needed in this case. > Wouldn't the Count "enable" extension already provide a way for users to > stop the counting? I imagine the driver would determine implicitly that > the device can enter a power-saving state in that scenario when counting > has been disabled by the user. > > William Breathitt Gray Hello, William! Thanks a lot for the feedback! You raised an important point. I would like to explain my reasoning behind the implementation so you can see if it makes sense or if I misunderstood. Basically, Texas drivers are described as user-operated through the sysfs file system, echoing some directories in /dev/. The requirement of my application was to communicate at the kernel space level with another driver, and to do so, I retrieved the pdev with platform_get_drvdata(), which was allocated with platform_set_drvdata in the texas driver probe. To apply the required settings, I used the pdev I retrieved pointing to the fops, such as ti_eqep_function_write(). I made this small change in the module for this enum value and a patch in the ti eqep driver to handle this in order to establish a safe shutdown. In my application it worked perfectly, and so I had the idea of applying a patch to kernel.org to see if my changes could contribute something to the Texas driver infrastructure. Best regards
