This list_empty_careful() (if u observed carefully, the entire kernel source code only used it ONCE - only wait->task_list - which matched exactly the prerequisite as stated in finish_wait()). It belongs to the category of sync mechanism called "lock-free/wait-free" synchronization mechanism. As stated in the comments in finish_wait(), the lock-free mechanism is implemented via using TWO variable - testing one after another, and this is effectively implementing the sync mechanism (other methods of lock-free methods using atomic instruction exists): I cannot do justice to this big topic, please check this video out: Advanced Topics in Programming Languages: A Lock-Free Hash Table http://video.google.com/videoplay?docid=2139967204534450862 http://www.audiomulch.com/~rossb/code/lockfree/ And many other references on the internet. On 10/29/07, Bahadir Balban <bahadir.balban@xxxxxxxxx> wrote: > Hi, > > For what operations do linked lists require locking in an smp-enabled kernel? > > In comments for list_empty_careful() it says it only protects against > a list_del_init(). Also all the safe traversal macros, do they have > any issue with list_add() during traversal? IOW, what can you *not* do > with these safe versions and need a lock? > > Thanks, > Bahadir > > -- > To unsubscribe from this list: send an email with > "unsubscribe kernelnewbies" to ecartis@xxxxxxxxxxxx > Please read the FAQ at http://kernelnewbies.org/FAQ > > -- Peter Teoh HP: 96809281 -- To unsubscribe from this list: send an email with "unsubscribe kernelnewbies" to ecartis@xxxxxxxxxxxx Please read the FAQ at http://kernelnewbies.org/FAQ
