[00:03:54] --- SecureEndpoints has left [00:13:34] --- thomas.kula@gmail.com has left [00:19:10] --- dragos.tatulea has become available [00:19:22] good morning [00:20:16] hi [00:22:38] --- thomas.kula@gmail.com has become available [00:52:47] --- SecureEndpoints has become available [00:52:55] gm [01:14:21] --- dev-zero@jabber.org has become available [01:33:57] --- dev-zero@jabber.org has left [02:29:33] SecureEndpoints: Up so early? [02:29:40] Or not in NY? [03:31:47] hmm [03:32:14] What is the easiest way to increment a vnode refcount [03:32:22] in a platform independet way. [04:12:57] --- dev-zero@jabber.org has become available [05:18:52] --- dev-zero@jabber.org has left [05:26:10] do you mean the VN_HOLD() / VN_RELE() macros? [05:53:31] --- Simon Wilkinson has become available [06:00:27] --- Simon Wilkinson has left: Replaced by new connection [06:00:29] --- Simon Wilkinson has become available [06:06:06] --- Simon Wilkinson has left: Replaced by new connection [06:06:06] --- Simon Wilkinson has become available [06:06:51] The common use case seems to be to call AFS_FAST_RELE to release, but to use osi_vnhold (which calls VN_HOLD on most platforms) to hold a reference count. [06:12:00] ... and then to do additional magic with vnode_ref and vnode_get on Darwin. In fact, I don't think there is a clear abstraction function to get a vnode any more. [06:13:02] For example - in afs_callback,c [06:13:05] #if defined(AFS_SGI_ENV) || defined(AFS_OSF_ENV) || defined(AFS_SUN5_ENV) || defined(AFS_HPUX_ENV) || defined(AFS_LINUX20_ENV) VN_HOLD(AFSTOV(tvc)); #else #ifdef AFS_DARWIN80_ENV vp = AFSTOV(tvc); if (vnode_get(vp)) continue; if (vnode_ref(vp)) { AFS_GUNLOCK(); vnode_put(vp); AFS_GLOCK(); continue; } #else #if defined(AFS_DARWIN_ENV) || defined(AFS_XBSD_ENV) osi_vnhold(tvc, 0); #else VREFCOUNT_INC(tvc); /* AIX, apparently */ #endif #endif #endif [06:13:31] ... and in afs_vcache.c [06:14:20] it's even more exciting in FindVCache, but in other places we just call osi_vnhold() [06:14:45] Anyone got any inspiration as to which set is correct.... [06:23:24] --- Simon Wilkinson has left [07:23:43] oh darn [07:24:09] he just left [07:24:12] oh well [07:37:31] oh, the darwin fun is because of iocount versus usecount [07:38:07] however, only the first reference needs to be done that way (for darwin), essentially. [08:02:29] iocount? usecount? [08:04:04] darwin's vnodes don't have a v_usecount; it's a black box but the internal implementation separates reference from "doing i/o on it" [10:30:14] --- summatusmentis has left [11:36:10] --- dragos.tatulea has left [11:43:33] --- Simon Wilkinson has become available [11:49:17] --- Simon Wilkinson has left: Replaced by new connection [11:49:18] --- Simon Wilkinson has become available [11:52:52] --- summatusmentis has become available [11:53:01] --- summatusmentis has left [11:53:10] --- summatusmentis has become available [12:00:15] --- dragos.tatulea has become available [12:00:59] I did a dist ugrade on my development machine (now it's debian lenny) and I got a 2.6.26. [12:01:54] And now I can't compile the 1.5.x devel. [12:02:12] It compiles only for the old version. MODLOAD-2.6.18... [12:08:05] "the 1.5.x devel" [12:08:34] yeah [12:08:42] which one [12:12:10] lates [12:12:19] latest [12:15:35] i suppose you'd like me to guess the problem, too [12:16:14] no [12:16:21] but i'm confused [12:16:27] i did clean [12:16:29] so make with the details [12:16:34] ran regen.sh [12:17:05] ./configure --enable-disconnected --enable-warnings --enable-debug --with-krb5-conf=/path/to/krb5-conf [12:17:09] and then make [12:17:17] compiling works [12:17:35] and loading the modules fails, i take it. the system messages might tell you why [12:17:40] but it compiles under src/libafs/MODLOAD-2.6.18 [12:17:54] then it found the wrong kernel headers. tell it where the right ones are [12:17:54] so it's clearly not for 2.6.26 [12:18:01] oh, okay [12:18:03] I will [12:19:06] thanks [12:19:50] --- Simon Wilkinson has left [12:38:03] --- summatusmentis has left [13:27:09] --- summatusmentis has become available [13:29:05] --- summatusmentis has left [13:34:52] --- dragos.tatulea has left [14:22:47] --- summatusmentis has become available [23:06:52] --- summatusmentis has left [23:48:04] --- summatusmentis has become available