[00:09:04] --- dev-zero@jabber.org has left [00:18:21] --- Russ has left: Disconnected [00:43:50] --- dev-zero@jabber.org has become available [01:16:42] --- haba has become available [01:51:40] --- Claudio Bisegni has left [02:24:26] --- jaltman has become available [02:54:51] --- haba has left [03:29:09] --- haba has become available [03:56:55] --- mmeffie has left [04:03:12] the PFsCmd implementation required for VIOC_FS_CMD is present in the tree because it was part of Hartmut's 2001 contribution of MR-AFS support. This pioctl interface requires that the fs process be aware of AFS3 wire protocol data structures because the pioctl is simply a pass-through to the cache manager calling RXAFS_FsCmd (formerly RXAFS_ResidencyCmd). I really wish this had been implemented by having the fs command obtain a token from the cache manager and perform the RPCs directly. [04:04:46] In fact, why does afsio even need this pioctl call? It is already communicating directly with the file server. It should just perform the RXAFS_FetchStatus call directly [04:08:47] i should look but my guess would be path translation [04:08:56] since, well, afscp had the same problem [04:13:15] the path translation is easy. call the GetFid pioctl [04:14:05] ideally, i'd want to run with no cm at all [04:17:16] sure. [04:17:31] in that vein somewhere there's an afscp we can steal code from [04:29:38] the RXAFS_FsCmd takes as input a "command" to be executed by the file and a set of matching inputs from the client. It turns out that Hartmut instead of leaving the Residency command values alone and allocating new ones for OSD simply started over counting from 1. The CMD_ values in the OpenAFS RT and the ones in use by OSD are now in conflict: %#define OSD_WRITING 1 %#define OSD_WAIT 2 %#define FS_OSD_COMMAND 4 %#define OSD_READ_END_NOTIFY 8 %#define CMD_WIPEFILE 7 %#define CMD_FETCHSTATUS 9 %#define CMD_SERVERPATH 10 %#define CMD_LISTOFFLINEVOLUMES 11 %#define CMD_LISTLOCKEDVNODES 12 %#define CMD_LISTDISKVNODE 13 %#define CMD_LISTFSYNCTRACE 14 %#define CMD_SHOWTHREADS 15 %#define CMD_INVERSELOOKUP 16 %#define CMD_OSD_ARCHIVE 17 %#define CMD_OSD_MD5SUM 18 %#define CMD_SET_OSD_MD5SUM 19 %#define CMD_STRIPED_OSD_FILE 20 %#define CMD_REPLACE_OSD 21 %#define CMD_GET_ARCH_OSDS 22 %#define CMD_LIST_OSDS 23 %#define CMD_SET_POLICY 24 %#define CMD_GET_POLICIES 25 [04:31:50] that rendered very badly in owl in an xterm [04:32:28] however, "damnit" [05:00:27] --- matt has become available [05:33:32] --- mmeffie has become available [05:48:23] --- mmeffie has left [06:07:26] --- mmeffie has become available [06:59:17] --- deason has become available [07:16:00] --- mho has become available [07:45:28] --- jaltman has left: Disconnected [07:47:37] --- haba has left [08:25:12] --- haba has become available [08:26:26] --- jaltman has become available [08:47:50] --- jaltman has left: Disconnected [08:48:58] > fs command obtain a token from the cache manager and perform the RPC Yeah, the fs command doesn't make RPC's, by design. [08:50:13] Ew. [08:50:45] your client needs a flag which precludes you from sending until you've caught up [08:51:00] hm. and i still have commit access... [08:53:17] I don't know why you keep claiming I'm "not caught up" when I say something that is not rendered pointless by something later than what I'm replying to. [08:53:34] Jaltman wishes it had made an RPC. I don't. fs doesn't do that. [08:54:09] afsio isn't fs. [09:02:32] --- cg2v has become available [09:18:25] --- Russ has become available [09:47:18] --- mmeffie has left [10:12:44] --- jaltman has become available [10:20:52] s/fs/afsio which is what I was actually talking about [10:22:08] OK, well, reading later messages didn't make that clear to me. [10:25:37] given the overall context of the work I have been submitting to gerrit plus the discussions from yesterday on the subject and the reference to PFsCmd it might have been obvious to someone that knows the contents of fs.c since there are no such calls there to replace [10:26:28] but thats ok. its really not a big deal [12:59:55] --- Derrick Brashear has become available [13:03:17] --- dev-zero@jabber.org has left [13:24:40] --- matt has left [13:47:33] --- dev-zero@jabber.org has become available [14:50:07] --- Derrick Brashear has left [14:55:17] --- dev-zero@jabber.org has left [15:22:28] --- matt has become available [15:22:38] --- matt has left [15:57:00] --- deason has left [17:04:44] --- deason has become available [18:15:37] --- jaltman has left: Replaced by new connection [18:15:38] --- jaltman has become available [19:39:24] --- Derrick Brashear has become available [19:44:07] --- jaltman has left: Replaced by new connection [19:44:09] --- jaltman has become available [20:01:00] --- deason has left [20:01:39] --- deason has become available [21:40:09] --- deason has left [22:13:18] --- dev-zero@jabber.org has become available [22:24:07] --- shadow@gmail.com/owl8414F608 has left [23:01:00] --- Russ has left: Disconnected [23:36:11] --- haba has left