Hey Simon, > Since the uclass now knows both the requested value and the > driver-reported maximum, would it be cleaner to have the uclass clamp > timeout_ms before calling ops->start(), rather than reporting a value > the driver may or may not honour?
Agreed, will be done in v3. > The sandbox driver has had to grow > its own clamp in patch 4 purely so the print path is testable. If the > uclass clamped, real drivers would never see an out-of-range value and > the sandbox change in patch 4 would not be needed. What do you think? Sandbox will still need to set max_timeout_ms in probe so the uclass knows the limit, but v3 will get rid of the extra clamping in sandbox_wdt_start(). Best, Juuso
