search for: nbdkit_sleep

...USR1 to the thread stuck in nanosleep, I confirmed that nanosleep() indeed returns early with EINTR, and we end up with insufficient delay. But getting to that point in my testing required that I first diagnosed why my attempt at a 1000-second sleep acted like no delay at all. I plan on adding an nbdkit_sleep() wrapper function, which will resume the sleep on unrelated EINTR (fixing the theoretical problem of not sleeping long enough - theoretical since it only matters if we handle a signal but want to continue execution, but all our current handled signals instead request process termination) as well a...