[00:58:18] --- Russ has left: Disconnected [06:04:33] --- Simon has become available [06:22:08] --- Simon has left [06:33:46] --- Jeffrey Altman has left [08:10:13] --- jaltman/FrogsLeap has left: Disconnected [08:14:14] --- jaltman/FrogsLeap has become available [09:27:25] --- deason has become available [09:34:27] jakllsch: just inc/dec the refcount on the vnode? [10:00:33] --- jaltman/FrogsLeap has left: Disconnected [10:04:59] --- jaltman/FrogsLeap has become available [11:04:25] --- jaltman/FrogsLeap has left: Disconnected [11:05:39] i'm not sure it's that simple, but it might be [11:06:07] there's the option of vget() or vref() [11:06:47] at the moment, my biggest problem seems to be in the osi_TryEvictVCache() area [11:07:36] it tries to vgone() a node on the freelist (thus 0 usecount), which the kernel says i can't do [11:07:57] i try to keep them off the freelist and other problems arise [11:08:27] (i'm using vget() for now, it seems to be slightly more correct) [11:11:38] there's also the vhold() and holdrele() pair, but those seem to be for a different kind of refcount [11:12:47] --- jaltman/FrogsLeap has become available [11:13:06] it might help if i had some idea of what the lifecycle of a vnode in the unix cachmanager looked like [11:13:51] though i suppose i'll get an idea of that myself eventually [11:26:13] i find the FBSD osi_TryEvictVCache kind of interesting [11:26:25] it doesn't attempt to use the kernel vnode layer [11:39:57] and it seems to work better if i do it like that [11:41:25] we keep our own list of vnode structures; we don't allocate them from the kernel [11:41:25] --- jaltman/FrogsLeap has left: Disconnected [11:41:42] according to the fbsd comments, vgone calls afs_vop_reclaim, which calls our afs_FlushVCache code, which puts the vnode on our free list [11:42:47] so if the only problem is the count being 0, just try inc'ing it like fbsd does [11:43:17] --- jaltman/FrogsLeap has become available [11:43:26] what's in fbsd makes sense to me; I'm not familiar with the *bsd vfs layer or anything, but assuming the comments are right, it sounds right [11:58:55] --- jaltman/FrogsLeap has left: Disconnected [12:02:53] --- jaltman/FrogsLeap has become available [13:08:07] --- Russ has become available [13:17:54] --- Simon Wilkinson has become available [13:18:52] --- Simon Wilkinson has left [13:19:07] --- Simon Wilkinson has become available [15:10:27] --- Simon Wilkinson has left [22:06:34] --- deason has left [22:58:23] --- reuteras has become available [22:59:08] --- lars.malinowsky has left