On Fri, Jan 10, 2025 at 09:51:18AM +0000, John Garry wrote: > ok, and that just comes down to the behavior of queue_var_store(), which > mimics sysfs_ops.store > > I will note that queue_var_store and queue_var_store64 differ in behaviour > here :( > >> ->store_limits uses >> the simpler and harder to get wrong convention of returning 0 on >> success. >> > > understood, so any reason why not to change the rest (apart from being > busy)? Not real urge. The idea here was to get it right for the new one. Changing the existing would be a lot of churn for a relatively small improvement. For me that's only worth it when touching the area anyway. Which might or might not happen when trying to remove the sysfs_lock around ->store.
