Hello, On (12/08/17 15:00), Petr Mladek wrote: [..] > > However, now that cross-release was introduces, lockdep can be applied > > to semaphore operations. Actually, I have a plan to do that. I think it > > would be better to make semaphore tracked with lockdep and remove all > > these manual acquire() and release() here. What do you think about it? > > IMHO, it would be great to add lockdep annotations into semaphore > operations. certain types of locks have no guaranteed lock-unlock ordering. e.g. readers-writer locks, semaphores, etc. for readers-writer lock we can easily have CPU0 CPU1 CPU2 CPU3 CPU4 read_lock write_lock // sleep because // of CPU0 read_lock read_unlock read_lock read_unlock read_lock read_unlock read_unlock // wake up CPU1 so for CPU1 the lock was "locked" by CPU0 and "unlocked" by CPU4. semaphore not necessarily has the mutual-exclusion property, because its ->count is not required to be set to 1. in printk we use semaphore with ->count == 1, but that's just an accident. -ss p.s. frankly, I don't see any "locking issues" in Steven's patch. -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@xxxxxxxxx. For more info on Linux MM, see: http://www.linux-mm.org/ . Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>
