[EMAIL PROTECTED] writes: > We've got the situation where a certain codepath needs two locks in the > filesystem space quite regularly (think: rename locking source and target > directories), and there's no single golden path how one prevents deadlocks > in such situations - have seen both trylock() as well as state-dependent > locking protocols (say, lock the alphabetically-lower pathname first, or > lock the one with the higher kernel virtual address first). You're right, > it's tricky.
All you really need is something that evaluates in the usual staircase order, and some design-level guarantee that everyone takes the locks in exactly the same way. We do this in several other places -- the ill_lock_ills() function sorts ill_t pointers into descending address order using a bubble sort, and in strlock() by sorting the struct stdata pointers for mated streams in descending order. Using the same here (if it's really needed) should be sufficient. (Something about trylock just seems inherently evil to me ...) -- James Carlson, Solaris Networking <[EMAIL PROTECTED]> Sun Microsystems / 35 Network Drive 71.232W Vox +1 781 442 2084 MS UBUR02-212 / Burlington MA 01803-2757 42.496N Fax +1 781 442 1677 _______________________________________________ opensolaris-code mailing list opensolaris-code@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/opensolaris-code
