Couple of heads up if you are playing with the tmpfs diff from pedro@. NetBSD does the sanity checking for hardlinks (not across mountpoints, not on directories) at the syscall layer and KASSERTs this in the filesystems.
Also, lockstatus() needs to be able to report if the lock is exclusive or shared (for naked VOP_ISLOCKED() == LK_EXCLUSIVE) checks. I've fixed both issues here: https://www.bitrig.org/viewgit/?a=shortlog&p=bitrig&h=refs/heads/tmpfs On Wed, May 1, 2013 at 7:13 PM, Ted Unangst <t...@cvs.openbsd.org> wrote: > CVSROOT: /cvs > Module name: src > Changes by: t...@cvs.openbsd.org 2013/05/01 11:13:05 > > Modified files: > sys/kern : kern_lock.c kern_rwlock.c > sys/sys : lock.h rwlock.h > > Log message: > exorcise lockmgr. the api remains, but is now backed by recursive rwlocks. > originally by thib. > ok deraadt jsing and anyone who tested
