[00:20:42] --- reuteras has left [00:20:55] --- reuteras has become available [00:53:06] --- lama has become available [01:43:45] --- Russ has left: Disconnected [02:06:11] --- Simon Wilkinson has become available [03:43:25] --- reuteras has left [03:59:56] --- jaltman/FrogsLeap has left: Disconnected [04:58:09] --- Simon Wilkinson has left [05:23:08] --- jaltman/FrogsLeap has become available [05:30:42] --- haba has become available [05:35:22] How near is a 1.6.0pre5 to appear (for example in the candidate/1.6.0pre5/openafs-1.6.0pre5-src.tar.gz form)? [05:35:23] --- shadow@gmail.com/barnowl1CD2CB82 has left: Lost connection [05:36:37] --- shadow@gmail.com/barnowl1CD2CB82 has become available [06:04:59] --- meffie has become available [07:48:04] --- jaltman/FrogsLeap has left: Disconnected [07:48:17] --- jaltman/FrogsLeap has become available [08:05:07] --- deason has become available [08:14:45] --- lama has left [08:42:48] Hm. I'm testing 1.6.0pre4. I hauled in the exportfs kernel module, but I still get complains about libafs: Unknown symbol find_exported_dentry libafs: Unknown symbol export_op_default This is a centos kernel. have they made stuff gnu-only or something like that? [08:43:27] modprobe exportfs [08:43:38] done [08:43:48] ok, now in... [08:43:50] well, wait [08:43:51] # lsmod | grep exportfs exportfs 39168 0 [08:43:54] lsmod show it? [08:43:59] ok [08:44:16] grep export /proc/ksyms [08:44:30] (don't paste the entire result please) [08:44:49] :-) [08:45:30] I have it in /proc/kallsyms (no ksyms) [08:45:53] maybe they made it gplonly then. odd [08:45:57] # grep find_exported_dentry /proc/kallsyms ffffffff8adbe860 r __ksymtab_find_exported_dentry [exportfs] ffffffff8adbe8c0 r __kstrtab_find_exported_dentry [exportfs] ffffffff8adbe888 r __kcrctab_find_exported_dentry [exportfs] ffffffff8adbe522 t find_exported_dentry [exportfs] 000000000f7ef68b a __crc_find_exported_dentry [exportfs] [08:46:44] This is 2.6.18-194.17.4.el5.pdc1 which is almost like the 2.6.18-194.17.4 that comes with centos [08:47:34] our modifications are only in another corner, but I can talk to the guy who did the kernel build [08:48:15] nah, he went home [08:49:28] could see if it's a gplonly issue if changing MODULE_LICENSE changes anything [08:49:37] I thought that gave a different error, though [09:06:52] "Funny": Both nfsd.ko and libafs.ko state that they need find_exported_dentry, but nfsd loads fine and libafs does not [09:07:58] maybe find_exported_dentry is not the problem, but the other one that is reported as well: export_op_default [09:08:11] what is that? I can't find export_op_default anywhere. [09:13:31] if you have the source for that kernel, it should be in fs/exportfs/expfs.c, the same file find_exported_dentry is exported from [09:14:22] we only use export_op_default depending on a configure test, though; you could look at the reason for the config test "if kernel uses new export ops" is failing, if it looks like it's failing for the wrong reason or something [09:14:41] I will dig into that direction, but after dinner. [09:14:47] thanks [09:15:19] (6:15pm here) [09:16:02] --- haba has left [09:35:53] --- ksumner has become available [09:38:27] --- rra has become available [09:46:32] --- lars.malinowsky has become available [10:36:52] --- haba has become available [12:08:25] #if !defined(NEW_EXPORT_OPS) extern struct export_operations export_op_default; #endif Which one is the new one and which one the old? [12:11:46] * haba puzzled [12:11:48] --- haba has left [12:13:02] fh_to_parent and fh_to_dentry it looks like are the ones we expect to be in export_operations for NEW_EXPORT_OPS [12:40:35] --- haba has become available [12:41:43] So I think my kernel actually should be "NEW" as it does not have extern struct export_operations export_op_default [12:42:16] * haba thinks that defines with the substring "NEW" are not the best choice. [12:42:18] what does the definition of struct export_operations look like? [12:42:58] I have to look that I have all the includes at the right places. [12:51:26] I found the struct export_operations (I think) in fs.h. It contains func pointers for decode_fh, encode_fh, get_name, get_parent, get_dentry, find_exported_dentry. [12:55:40] --- lars.malinowsky has left [12:55:45] --- lars.malinowsky has become available [12:56:35] So there is no fh_to_parent [13:02:52] but my kernel does not have the export_op_default struct neiter. So currently I loose. [13:02:55] can you see how fs/nfsd/nfsfh.c calls decode_fh? in the vanilla 2.6.18 it's by the CALL macro, which calls into export_op_default if the relevant function pointer is null [13:03:46] Somewhere we have the full kernel source lying around. Will find and look. [13:21:54] Yes, it's that CALL macro here, too. But how/where is the export_op_default hidden. [13:30:30] * haba still more puzzled ;-/ [13:40:40] er, the CALL is defined to use export_op_default? does it have a decl for export_op_default? [13:43:06] CALL uses export_op_default. Just pwr cycling the box, brb. [13:44:50] you could also see if export_op_default is EXPORT_SYMBOL_GPL while you have the source... [13:46:14] # grep export_op_default expfs.c struct export_operations export_op_default; #define CALL(ops,fun) ((ops->fun)?(ops->fun):export_op_default.fun) struct export_operations export_op_default = { EXPORT_SYMBOL(export_op_default); [13:46:49] The same #define CALL is in the nfs module [13:48:52] So as far as I understand this, export_op_default should be in the exportfs module _and_ be exported. [13:49:37] Same goes for find_exported_dentry [13:50:38] HM. MODULE_LICENSE("GPL"); in expfs.c [13:56:58] But now I say MODULE_LICENSE("GPL") in libafs.ko as well, so that should not matter either. [14:06:45] so yeah, so the symbols are there, that shouldn't be a problem [14:10:02] * haba tired. It's late here. [14:10:08] ->tomorrow [14:10:22] *wave* [14:17:51] --- matt has become available [15:23:08] --- matt has left [15:47:25] --- deason has left [15:49:40] --- ksumner has left [16:22:09] --- steven.jenkins has left [16:23:09] --- steven.jenkins has become available [16:23:15] --- shadow@gmail.com/barnowl1CD2CB82 has left [16:23:24] --- shadow@gmail.com/barnowl1CD2CB82 has become available [16:23:42] --- kula has left [17:25:42] --- kula has become available [18:03:03] --- phalenor has left [18:03:09] --- phalenor has become available [18:32:20] --- lars.malinowsky has left: Lost connection [18:33:36] --- lars.malinowsky has become available [18:35:11] --- jakllsch has left [18:35:47] --- steven.jenkins has left [18:36:45] --- steven.jenkins has become available [18:38:25] --- kula has left [18:39:20] --- shadow@gmail.com/barnowl1CD2CB82 has left [18:39:31] --- shadow@gmail.com/barnowl1CD2CB82 has become available [18:40:32] --- lars.malinowsky has left [18:40:37] --- lars.malinowsky has become available [19:55:49] --- Russ has become available [22:44:19] --- reuteras has become available