Hi Jeff, > I have a patch at http://www.chesapeake.net/~jroberson/vgone.diff that > greatly simplifies the vnode teardown code, and should fix some of the > races that have been seen with smp vfs enabled. I would appreciate any > testing people can offer. See the thread on arch_at_ for details. I applied your patch yesterday, and I got a panic during the night. My ssh session froze at about 3:02am, so I guess it's happened during daily periodic(8). This is hand written: %%% panic: lockmgr: unknown locktype request 0 KDB: stack backtrace: kdb_backtrace(c0707071,c0763000,c0705327,d50d2c0c,100) at kdb_backtrace+0x2e panic(c0705327,0,c07051f4,1d2,c18fc320) lockmgr(c1928164,10000,c1928188,c18fc320,d50d2c5c) at lockmgr+0x563 vop_stdlock(d50d2cc4,0,d50d2c94,c1928110,d50d2c7c) at vop_stdlock+0x2f VOP_LOCK_APV(c074d780,d50d2cc4,c070df3b,d50d2c94,d50d2cc4) at VOP_LOCK_APV+0xbc ffs_lock(d50d2cc4,c1928110,28,c1928110,d50d2ce0) at ffs_lock+0x90 VOP_LOCK_APV(c074d180,d50d2cc4,c070df3b,235,c18fc320) at VOP_LOCK_APV+0xbc vlrureclaim(c184e800,10,c0768ce0,c18fc320,3e8) at vlrureclaim+0x1c1 vnlru_proc(0,d50d2d48,c0704615,30e,0) at vnlru_proc+0x19f fork_exit(c0599db0,0,d50d2d48) at fork_exit+0xc6 fork_trampoline() at fork_trampoline+0x8 --- trap 0x1, eip = 0, esp = 0xd50d2d7c, ebp = 0 --- KDB: enter: panic [thread pid 45 tid 100062 ] Stopped at kdb_enter+Xà30: leave db> show lockedvnods Locked vnodes 0xc348880: tag ufs, type VDIR usecount 2, writecount 0, refcount 1 mountedhere 0 flags () v_object 0xc34b3ce4 ref 0 pages 1 lock type ufs: EXCL (count 1) by thread 0xc2056000 (pid 2137) ino 34539, on dev ad0s1e db> show all procs [ pid 2137 is find(1) ] %%% A dump is available, please contact me off-list to get it. I hope this will help. Thanks for your work. Regards, -- Jeremie Le Hen < jeremie at le-hen dot org >< ttz at chchile dot org >Received on Fri Mar 11 2005 - 09:29:57 UTC
This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:38:29 UTC