[deliverable/linux.git] / fs / nfs / callback_proc.c

/*
 * linux/fs/nfs/callback_proc.c
 *
 * Copyright (C) 2004 Trond Myklebust
 *
 * NFSv4 callback procedures
 */
#include <linux/nfs4.h>
#include <linux/nfs_fs.h>
#include <linux/slab.h>
#include <linux/rcupdate.h>
#include "nfs4_fs.h"
#include "callback.h"
#include "delegation.h"
#include "internal.h"
#include "pnfs.h"
#include "nfs4session.h"
#include "nfs4trace.h"

#ifdef NFS_DEBUG
#define NFSDBG_FACILITY NFSDBG_CALLBACK
#endif

__be32 nfs4_callback_getattr(struct cb_getattrargs *args,
			     struct cb_getattrres *res,
			     struct cb_process_state *cps)
{
	struct nfs_delegation *delegation;
	struct nfs_inode *nfsi;
	struct inode *inode;

	res->status = htonl(NFS4ERR_OP_NOT_IN_SESSION);
	if (!cps->clp) /* Always set for v4.0. Set in cb_sequence for v4.1 */
		goto out;

	res->bitmap[0] = res->bitmap[1] = 0;
	res->status = htonl(NFS4ERR_BADHANDLE);

	dprintk_rcu("NFS: GETATTR callback request from %s\n",
		rpc_peeraddr2str(cps->clp->cl_rpcclient, RPC_DISPLAY_ADDR));

	inode = nfs_delegation_find_inode(cps->clp, &args->fh);
	if (inode == NULL)
		goto out;
	nfsi = NFS_I(inode);
	rcu_read_lock();
	delegation = rcu_dereference(nfsi->delegation);
	if (delegation == NULL || (delegation->type & FMODE_WRITE) == 0)
		goto out_iput;
	res->size = i_size_read(inode);
	res->change_attr = delegation->change_attr;
	if (nfsi->nrequests != 0)
		res->change_attr++;
	res->ctime = inode->i_ctime;
	res->mtime = inode->i_mtime;
	res->bitmap[0] = (FATTR4_WORD0_CHANGE|FATTR4_WORD0_SIZE) &
		args->bitmap[0];
	res->bitmap[1] = (FATTR4_WORD1_TIME_METADATA|FATTR4_WORD1_TIME_MODIFY) &
		args->bitmap[1];
	res->status = 0;
out_iput:
	rcu_read_unlock();
	iput(inode);
out:
	dprintk("%s: exit with status = %d\n", __func__, ntohl(res->status));
	return res->status;
}

__be32 nfs4_callback_recall(struct cb_recallargs *args, void *dummy,
			    struct cb_process_state *cps)
{
	struct inode *inode;
	__be32 res;
	
	res = htonl(NFS4ERR_OP_NOT_IN_SESSION);
	if (!cps->clp) /* Always set for v4.0. Set in cb_sequence for v4.1 */
		goto out;

	dprintk_rcu("NFS: RECALL callback request from %s\n",
		rpc_peeraddr2str(cps->clp->cl_rpcclient, RPC_DISPLAY_ADDR));

	res = htonl(NFS4ERR_BADHANDLE);
	inode = nfs_delegation_find_inode(cps->clp, &args->fh);
	if (inode == NULL)
		goto out;
	/* Set up a helper thread to actually return the delegation */
	switch (nfs_async_inode_return_delegation(inode, &args->stateid)) {
	case 0:
		res = 0;
		break;
	case -ENOENT:
		res = htonl(NFS4ERR_BAD_STATEID);
		break;
	default:
		res = htonl(NFS4ERR_RESOURCE);
	}
	trace_nfs4_recall_delegation(inode, -ntohl(res));
	iput(inode);
out:
	dprintk("%s: exit with status = %d\n", __func__, ntohl(res));
	return res;
}

#if defined(CONFIG_NFS_V4_1)

/*
 * Lookup a layout by filehandle.
 *
 * Note: gets a refcount on the layout hdr and on its respective inode.
 * Caller must put the layout hdr and the inode.
 *
 * TODO: keep track of all layouts (and delegations) in a hash table
 * hashed by filehandle.
 */
static struct pnfs_layout_hdr * get_layout_by_fh_locked(struct nfs_client *clp,
		struct nfs_fh *fh, nfs4_stateid *stateid)
{
	struct nfs_server *server;
	struct inode *ino;
	struct pnfs_layout_hdr *lo;

	list_for_each_entry_rcu(server, &clp->cl_superblocks, client_link) {
		list_for_each_entry(lo, &server->layouts, plh_layouts) {
			if (!nfs4_stateid_match_other(&lo->plh_stateid, stateid))
				continue;
			if (nfs_compare_fh(fh, &NFS_I(lo->plh_inode)->fh))
				continue;
			ino = igrab(lo->plh_inode);
			if (!ino)
				break;
			spin_lock(&ino->i_lock);
			/* Is this layout in the process of being freed? */
			if (NFS_I(ino)->layout != lo) {
				spin_unlock(&ino->i_lock);
				iput(ino);
				break;
			}
			pnfs_get_layout_hdr(lo);
			spin_unlock(&ino->i_lock);
			return lo;
		}
	}

	return NULL;
}

static struct pnfs_layout_hdr * get_layout_by_fh(struct nfs_client *clp,
		struct nfs_fh *fh, nfs4_stateid *stateid)
{
	struct pnfs_layout_hdr *lo;

	spin_lock(&clp->cl_lock);
	rcu_read_lock();
	lo = get_layout_by_fh_locked(clp, fh, stateid);
	rcu_read_unlock();
	spin_unlock(&clp->cl_lock);

	return lo;
}

static u32 initiate_file_draining(struct nfs_client *clp,
				  struct cb_layoutrecallargs *args)
{
	struct inode *ino;
	struct pnfs_layout_hdr *lo;
	u32 rv = NFS4ERR_NOMATCHING_LAYOUT;
	LIST_HEAD(free_me_list);

	lo = get_layout_by_fh(clp, &args->cbl_fh, &args->cbl_stateid);
	if (!lo)
		goto out;

	ino = lo->plh_inode;

	spin_lock(&ino->i_lock);
	pnfs_set_layout_stateid(lo, &args->cbl_stateid, true);
	spin_unlock(&ino->i_lock);

	pnfs_layoutcommit_inode(ino, false);

	spin_lock(&ino->i_lock);
	if (test_bit(NFS_LAYOUT_BULK_RECALL, &lo->plh_flags) ||
	    pnfs_mark_matching_lsegs_invalid(lo, &free_me_list,
					&args->cbl_range)) {
		rv = NFS4ERR_DELAY;
		goto unlock;
	}

	if (NFS_SERVER(ino)->pnfs_curr_ld->return_range) {
		NFS_SERVER(ino)->pnfs_curr_ld->return_range(lo,
			&args->cbl_range);
	}
unlock:
	spin_unlock(&ino->i_lock);
	pnfs_free_lseg_list(&free_me_list);
	pnfs_put_layout_hdr(lo);
	iput(ino);
out:
	return rv;
}

static u32 initiate_bulk_draining(struct nfs_client *clp,
				  struct cb_layoutrecallargs *args)
{
	int stat;

	if (args->cbl_recall_type == RETURN_FSID)
		stat = pnfs_destroy_layouts_byfsid(clp, &args->cbl_fsid, true);
	else
		stat = pnfs_destroy_layouts_byclid(clp, true);
	if (stat != 0)
		return NFS4ERR_DELAY;
	return NFS4ERR_NOMATCHING_LAYOUT;
}

static u32 do_callback_layoutrecall(struct nfs_client *clp,
				    struct cb_layoutrecallargs *args)
{
	u32 res;

	dprintk("%s enter, type=%i\n", __func__, args->cbl_recall_type);
	if (args->cbl_recall_type == RETURN_FILE)
		res = initiate_file_draining(clp, args);
	else
		res = initiate_bulk_draining(clp, args);
	dprintk("%s returning %i\n", __func__, res);
	return res;

}

__be32 nfs4_callback_layoutrecall(struct cb_layoutrecallargs *args,
				  void *dummy, struct cb_process_state *cps)
{
	u32 res;

	dprintk("%s: -->\n", __func__);

	if (cps->clp)
		res = do_callback_layoutrecall(cps->clp, args);
	else
		res = NFS4ERR_OP_NOT_IN_SESSION;

	dprintk("%s: exit with status = %d\n", __func__, res);
	return cpu_to_be32(res);
}

static void pnfs_recall_all_layouts(struct nfs_client *clp)
{
	struct cb_layoutrecallargs args;

	/* Pretend we got a CB_LAYOUTRECALL(ALL) */
	memset(&args, 0, sizeof(args));
	args.cbl_recall_type = RETURN_ALL;
	/* FIXME we ignore errors, what should we do? */
	do_callback_layoutrecall(clp, &args);
}

__be32 nfs4_callback_devicenotify(struct cb_devicenotifyargs *args,
				  void *dummy, struct cb_process_state *cps)
{
	int i;
	__be32 res = 0;
	struct nfs_client *clp = cps->clp;
	struct nfs_server *server = NULL;

	dprintk("%s: -->\n", __func__);

	if (!clp) {
		res = cpu_to_be32(NFS4ERR_OP_NOT_IN_SESSION);
		goto out;
	}

	for (i = 0; i < args->ndevs; i++) {
		struct cb_devicenotifyitem *dev = &args->devs[i];

		if (!server ||
		    server->pnfs_curr_ld->id != dev->cbd_layout_type) {
			rcu_read_lock();
			list_for_each_entry_rcu(server, &clp->cl_superblocks, client_link)
				if (server->pnfs_curr_ld &&
				    server->pnfs_curr_ld->id == dev->cbd_layout_type) {
					rcu_read_unlock();
					goto found;
				}
			rcu_read_unlock();
			dprintk("%s: layout type %u not found\n",
				__func__, dev->cbd_layout_type);
			continue;
		}

	found:
		nfs4_delete_deviceid(server->pnfs_curr_ld, clp, &dev->cbd_dev_id);
	}

out:
	kfree(args->devs);
	dprintk("%s: exit with status = %u\n",
		__func__, be32_to_cpu(res));
	return res;
}

/*
 * Validate the sequenceID sent by the server.
 * Return success if the sequenceID is one more than what we last saw on
 * this slot, accounting for wraparound.  Increments the slot's sequence.
 *
 * We don't yet implement a duplicate request cache, instead we set the
 * back channel ca_maxresponsesize_cached to zero. This is OK for now
 * since we only currently implement idempotent callbacks anyway.
 *
 * We have a single slot backchannel at this time, so we don't bother
 * checking the used_slots bit array on the table.  The lower layer guarantees
 * a single outstanding callback request at a time.
 */
static __be32
validate_seqid(struct nfs4_slot_table *tbl, struct cb_sequenceargs * args)
{
	struct nfs4_slot *slot;

	dprintk("%s enter. slotid %u seqid %u\n",
		__func__, args->csa_slotid, args->csa_sequenceid);

	if (args->csa_slotid >= NFS41_BC_MAX_CALLBACKS)
		return htonl(NFS4ERR_BADSLOT);

	slot = tbl->slots + args->csa_slotid;
	dprintk("%s slot table seqid: %u\n", __func__, slot->seq_nr);

	/* Normal */
	if (likely(args->csa_sequenceid == slot->seq_nr + 1))
		goto out_ok;

	/* Replay */
	if (args->csa_sequenceid == slot->seq_nr) {
		dprintk("%s seqid %u is a replay\n",
			__func__, args->csa_sequenceid);
		/* Signal process_op to set this error on next op */
		if (args->csa_cachethis == 0)
			return htonl(NFS4ERR_RETRY_UNCACHED_REP);

		/* The ca_maxresponsesize_cached is 0 with no DRC */
		else if (args->csa_cachethis == 1)
			return htonl(NFS4ERR_REP_TOO_BIG_TO_CACHE);
	}

	/* Wraparound */
	if (args->csa_sequenceid == 1 && (slot->seq_nr + 1) == 0) {
		slot->seq_nr = 1;
		goto out_ok;
	}

	/* Misordered request */
	return htonl(NFS4ERR_SEQ_MISORDERED);
out_ok:
	tbl->highest_used_slotid = args->csa_slotid;
	return htonl(NFS4_OK);
}

/*
 * For each referring call triple, check the session's slot table for
 * a match.  If the slot is in use and the sequence numbers match, the
 * client is still waiting for a response to the original request.
 */
static bool referring_call_exists(struct nfs_client *clp,
				  uint32_t nrclists,
				  struct referring_call_list *rclists)
{
	bool status = 0;
	int i, j;
	struct nfs4_session *session;
	struct nfs4_slot_table *tbl;
	struct referring_call_list *rclist;
	struct referring_call *ref;

	/*
	 * XXX When client trunking is implemented, this becomes
	 * a session lookup from within the loop
	 */
	session = clp->cl_session;
	tbl = &session->fc_slot_table;

	for (i = 0; i < nrclists; i++) {
		rclist = &rclists[i];
		if (memcmp(session->sess_id.data,
			   rclist->rcl_sessionid.data,
			   NFS4_MAX_SESSIONID_LEN) != 0)
			continue;

		for (j = 0; j < rclist->rcl_nrefcalls; j++) {
			ref = &rclist->rcl_refcalls[j];

			dprintk("%s: sessionid %x:%x:%x:%x sequenceid %u "
				"slotid %u\n", __func__,
				((u32 *)&rclist->rcl_sessionid.data)[0],
				((u32 *)&rclist->rcl_sessionid.data)[1],
				((u32 *)&rclist->rcl_sessionid.data)[2],
				((u32 *)&rclist->rcl_sessionid.data)[3],
				ref->rc_sequenceid, ref->rc_slotid);

			spin_lock(&tbl->slot_tbl_lock);
			status = (test_bit(ref->rc_slotid, tbl->used_slots) &&
				  tbl->slots[ref->rc_slotid].seq_nr ==
					ref->rc_sequenceid);
			spin_unlock(&tbl->slot_tbl_lock);
			if (status)
				goto out;
		}
	}

out:
	return status;
}

__be32 nfs4_callback_sequence(struct cb_sequenceargs *args,
			      struct cb_sequenceres *res,
			      struct cb_process_state *cps)
{
	struct nfs4_slot_table *tbl;
	struct nfs4_slot *slot;
	struct nfs_client *clp;
	int i;
	__be32 status = htonl(NFS4ERR_BADSESSION);

	clp = nfs4_find_client_sessionid(cps->net, args->csa_addr,
					 &args->csa_sessionid, cps->minorversion);
	if (clp == NULL)
		goto out;

	if (!(clp->cl_session->flags & SESSION4_BACK_CHAN))
		goto out;

	tbl = &clp->cl_session->bc_slot_table;
	slot = tbl->slots + args->csa_slotid;

	spin_lock(&tbl->slot_tbl_lock);
	/* state manager is resetting the session */
	if (test_bit(NFS4_SLOT_TBL_DRAINING, &tbl->slot_tbl_state)) {
		status = htonl(NFS4ERR_DELAY);
		/* Return NFS4ERR_BADSESSION if we're draining the session
		 * in order to reset it.
		 */
		if (test_bit(NFS4CLNT_SESSION_RESET, &clp->cl_state))
			status = htonl(NFS4ERR_BADSESSION);
		goto out_unlock;
	}

	memcpy(&res->csr_sessionid, &args->csa_sessionid,
	       sizeof(res->csr_sessionid));
	res->csr_sequenceid = args->csa_sequenceid;
	res->csr_slotid = args->csa_slotid;
	res->csr_highestslotid = NFS41_BC_MAX_CALLBACKS - 1;
	res->csr_target_highestslotid = NFS41_BC_MAX_CALLBACKS - 1;

	status = validate_seqid(tbl, args);
	if (status)
		goto out_unlock;

	cps->slotid = args->csa_slotid;

	/*
	 * Check for pending referring calls.  If a match is found, a
	 * related callback was received before the response to the original
	 * call.
	 */
	if (referring_call_exists(clp, args->csa_nrclists, args->csa_rclists)) {
		status = htonl(NFS4ERR_DELAY);
		goto out_unlock;
	}

	/*
	 * RFC5661 20.9.3
	 * If CB_SEQUENCE returns an error, then the state of the slot
	 * (sequence ID, cached reply) MUST NOT change.
	 */
	slot->seq_nr++;
out_unlock:
	spin_unlock(&tbl->slot_tbl_lock);

out:
	cps->clp = clp; /* put in nfs4_callback_compound */
	for (i = 0; i < args->csa_nrclists; i++)
		kfree(args->csa_rclists[i].rcl_refcalls);
	kfree(args->csa_rclists);

	if (status == htonl(NFS4ERR_RETRY_UNCACHED_REP)) {
		cps->drc_status = status;
		status = 0;
	} else
		res->csr_status = status;

	trace_nfs4_cb_sequence(args, res, status);
	dprintk("%s: exit with status = %d res->csr_status %d\n", __func__,
		ntohl(status), ntohl(res->csr_status));
	return status;
}

static bool
validate_bitmap_values(unsigned long mask)
{
	return (mask & ~RCA4_TYPE_MASK_ALL) == 0;
}

__be32 nfs4_callback_recallany(struct cb_recallanyargs *args, void *dummy,
			       struct cb_process_state *cps)
{
	__be32 status;
	fmode_t flags = 0;

	status = cpu_to_be32(NFS4ERR_OP_NOT_IN_SESSION);
	if (!cps->clp) /* set in cb_sequence */
		goto out;

	dprintk_rcu("NFS: RECALL_ANY callback request from %s\n",
		rpc_peeraddr2str(cps->clp->cl_rpcclient, RPC_DISPLAY_ADDR));

	status = cpu_to_be32(NFS4ERR_INVAL);
	if (!validate_bitmap_values(args->craa_type_mask))
		goto out;

	status = cpu_to_be32(NFS4_OK);
	if (test_bit(RCA4_TYPE_MASK_RDATA_DLG, (const unsigned long *)
		     &args->craa_type_mask))
		flags = FMODE_READ;
	if (test_bit(RCA4_TYPE_MASK_WDATA_DLG, (const unsigned long *)
		     &args->craa_type_mask))
		flags |= FMODE_WRITE;
	if (test_bit(RCA4_TYPE_MASK_FILE_LAYOUT, (const unsigned long *)
		     &args->craa_type_mask))
		pnfs_recall_all_layouts(cps->clp);
	if (flags)
		nfs_expire_unused_delegation_types(cps->clp, flags);
out:
	dprintk("%s: exit with status = %d\n", __func__, ntohl(status));
	return status;
}

/* Reduce the fore channel's max_slots to the target value */
__be32 nfs4_callback_recallslot(struct cb_recallslotargs *args, void *dummy,
				struct cb_process_state *cps)
{
	struct nfs4_slot_table *fc_tbl;
	__be32 status;

	status = htonl(NFS4ERR_OP_NOT_IN_SESSION);
	if (!cps->clp) /* set in cb_sequence */
		goto out;

	dprintk_rcu("NFS: CB_RECALL_SLOT request from %s target highest slotid %u\n",
		rpc_peeraddr2str(cps->clp->cl_rpcclient, RPC_DISPLAY_ADDR),
		args->crsa_target_highest_slotid);

	fc_tbl = &cps->clp->cl_session->fc_slot_table;

	status = htonl(NFS4_OK);

	nfs41_set_target_slotid(fc_tbl, args->crsa_target_highest_slotid);
	nfs41_server_notify_target_slotid_update(cps->clp);
out:
	dprintk("%s: exit with status = %d\n", __func__, ntohl(status));
	return status;
}
#endif /* CONFIG_NFS_V4_1 */
Commit	Line	Data
1da177e4 LT	1	/*
	2	* linux/fs/nfs/callback_proc.c
	3	*
	4	* Copyright (C) 2004 Trond Myklebust
	5	*
	6	* NFSv4 callback procedures
	7	*/
1da177e4 LT	8	#include <linux/nfs4.h>
1da177e4 LT	9	#include <linux/nfs_fs.h>
5a0e3ad6	10	#include <linux/slab.h>
2446ab60	11	#include <linux/rcupdate.h>
4ce79717	12	#include "nfs4_fs.h"
1da177e4 LT	13	#include "callback.h"
1da177e4 LT	14	#include "delegation.h"
24c8dbbb	15	#include "internal.h"
43f1b3da	16	#include "pnfs.h"
73e39aaa	17	#include "nfs4session.h"
ca8acf8d	18	#include "nfs4trace.h"
1da177e4	19
1d98fe67	20	#ifdef NFS_DEBUG
1da177e4	21	#define NFSDBG_FACILITY NFSDBG_CALLBACK
1d98fe67	22	#endif
c36fca52 AA	23
	24	__be32 nfs4_callback_getattr(struct cb_getattrargs *args,
	25	struct cb_getattrres *res,
	26	struct cb_process_state *cps)
1da177e4	27	{
1da177e4 LT	28	struct nfs_delegation *delegation;
	29	struct nfs_inode *nfsi;
	30	struct inode *inode;
1d98fe67	31
c36fca52 AA	32	res->status = htonl(NFS4ERR_OP_NOT_IN_SESSION);
	33	if (!cps->clp) /* Always set for v4.0. Set in cb_sequence for v4.1 */
	34	goto out;
	35
1da177e4 LT	36	res->bitmap[0] = res->bitmap[1] = 0;
1da177e4 LT	37	res->status = htonl(NFS4ERR_BADHANDLE);
1d98fe67	38
2446ab60	39	dprintk_rcu("NFS: GETATTR callback request from %s\n",
c36fca52	40	rpc_peeraddr2str(cps->clp->cl_rpcclient, RPC_DISPLAY_ADDR));
1d98fe67	41
c36fca52	42	inode = nfs_delegation_find_inode(cps->clp, &args->fh);
1da177e4	43	if (inode == NULL)
c36fca52	44	goto out;
1da177e4	45	nfsi = NFS_I(inode);
761fe93c TM	46	rcu_read_lock();
761fe93c TM	47	delegation = rcu_dereference(nfsi->delegation);
1da177e4 LT	48	if (delegation == NULL \|\| (delegation->type & FMODE_WRITE) == 0)
	49	goto out_iput;
	50	res->size = i_size_read(inode);
beb2a5ec	51	res->change_attr = delegation->change_attr;
cb1410c7	52	if (nfsi->nrequests != 0)
beb2a5ec	53	res->change_attr++;
1da177e4 LT	54	res->ctime = inode->i_ctime;
	55	res->mtime = inode->i_mtime;
	56	res->bitmap[0] = (FATTR4_WORD0_CHANGE\|FATTR4_WORD0_SIZE) &
	57	args->bitmap[0];
	58	res->bitmap[1] = (FATTR4_WORD1_TIME_METADATA\|FATTR4_WORD1_TIME_MODIFY) &
	59	args->bitmap[1];
	60	res->status = 0;
	61	out_iput:
761fe93c	62	rcu_read_unlock();
1da177e4	63	iput(inode);
1da177e4	64	out:
3110ff80	65	dprintk("%s: exit with status = %d\n", __func__, ntohl(res->status));
1da177e4 LT	66	return res->status;
	67	}
	68
c36fca52 AA	69	__be32 nfs4_callback_recall(struct cb_recallargs args, void dummy,
c36fca52 AA	70	struct cb_process_state *cps)
1da177e4	71	{
1da177e4	72	struct inode *inode;
e6f684f6	73	__be32 res;
1da177e4	74
c36fca52 AA	75	res = htonl(NFS4ERR_OP_NOT_IN_SESSION);
c36fca52 AA	76	if (!cps->clp) /* Always set for v4.0. Set in cb_sequence for v4.1 */
1da177e4	77	goto out;
1d98fe67	78
2446ab60	79	dprintk_rcu("NFS: RECALL callback request from %s\n",
c36fca52 AA	80	rpc_peeraddr2str(cps->clp->cl_rpcclient, RPC_DISPLAY_ADDR));
	81
	82	res = htonl(NFS4ERR_BADHANDLE);
	83	inode = nfs_delegation_find_inode(cps->clp, &args->fh);
	84	if (inode == NULL)
	85	goto out;
	86	/* Set up a helper thread to actually return the delegation */
	87	switch (nfs_async_inode_return_delegation(inode, &args->stateid)) {
	88	case 0:
	89	res = 0;
	90	break;
	91	case -ENOENT:
8e663f0e	92	res = htonl(NFS4ERR_BAD_STATEID);
c36fca52 AA	93	break;
	94	default:
	95	res = htonl(NFS4ERR_RESOURCE);
	96	}
ca8acf8d	97	trace_nfs4_recall_delegation(inode, -ntohl(res));
c36fca52	98	iput(inode);
1da177e4	99	out:
3110ff80	100	dprintk("%s: exit with status = %d\n", __func__, ntohl(res));
1da177e4 LT	101	return res;
1da177e4 LT	102	}
d49433e1 BH	103
	104	#if defined(CONFIG_NFS_V4_1)
	105
d36b7cf7 BH	106	/*
	107	* Lookup a layout by filehandle.
	108	*
	109	* Note: gets a refcount on the layout hdr and on its respective inode.
	110	* Caller must put the layout hdr and the inode.
	111	*
	112	* TODO: keep track of all layouts (and delegations) in a hash table
	113	* hashed by filehandle.
	114	*/
27999f25 TM	115	static struct pnfs_layout_hdr * get_layout_by_fh_locked(struct nfs_client *clp,
27999f25 TM	116	struct nfs_fh fh, nfs4_stateid stateid)
43f1b3da	117	{
6382a441	118	struct nfs_server *server;
43f1b3da	119	struct inode *ino;
d36b7cf7	120	struct pnfs_layout_hdr *lo;
43f1b3da	121
6382a441 WAA	122	list_for_each_entry_rcu(server, &clp->cl_superblocks, client_link) {
6382a441 WAA	123	list_for_each_entry(lo, &server->layouts, plh_layouts) {
27999f25 TM	124	if (!nfs4_stateid_match_other(&lo->plh_stateid, stateid))
27999f25 TM	125	continue;
d36b7cf7	126	if (nfs_compare_fh(fh, &NFS_I(lo->plh_inode)->fh))
6382a441 WAA	127	continue;
	128	ino = igrab(lo->plh_inode);
	129	if (!ino)
9a7fe9e8	130	break;
9c626381 TM	131	spin_lock(&ino->i_lock);
	132	/* Is this layout in the process of being freed? */
	133	if (NFS_I(ino)->layout != lo) {
	134	spin_unlock(&ino->i_lock);
	135	iput(ino);
9a7fe9e8	136	break;
9c626381	137	}
70c3bd2b	138	pnfs_get_layout_hdr(lo);
9c626381	139	spin_unlock(&ino->i_lock);
d36b7cf7	140	return lo;
6382a441	141	}
43f1b3da	142	}
d36b7cf7 BH	143
	144	return NULL;
	145	}
	146
27999f25 TM	147	static struct pnfs_layout_hdr * get_layout_by_fh(struct nfs_client *clp,
27999f25 TM	148	struct nfs_fh fh, nfs4_stateid stateid)
d36b7cf7 BH	149	{
	150	struct pnfs_layout_hdr *lo;
	151
	152	spin_lock(&clp->cl_lock);
	153	rcu_read_lock();
27999f25	154	lo = get_layout_by_fh_locked(clp, fh, stateid);
6382a441	155	rcu_read_unlock();
43f1b3da	156	spin_unlock(&clp->cl_lock);
6382a441	157
d36b7cf7 BH	158	return lo;
	159	}
	160
	161	static u32 initiate_file_draining(struct nfs_client *clp,
	162	struct cb_layoutrecallargs *args)
	163	{
	164	struct inode *ino;
	165	struct pnfs_layout_hdr *lo;
	166	u32 rv = NFS4ERR_NOMATCHING_LAYOUT;
	167	LIST_HEAD(free_me_list);
	168
27999f25	169	lo = get_layout_by_fh(clp, &args->cbl_fh, &args->cbl_stateid);
d36b7cf7	170	if (!lo)
27999f25	171	goto out;
43f1b3da	172
d36b7cf7	173	ino = lo->plh_inode;
7c5d1875 CH	174
	175	spin_lock(&ino->i_lock);
	176	pnfs_set_layout_stateid(lo, &args->cbl_stateid, true);
	177	spin_unlock(&ino->i_lock);
	178
	179	pnfs_layoutcommit_inode(ino, false);
	180
43f1b3da FI	181	spin_lock(&ino->i_lock);
43f1b3da FI	182	if (test_bit(NFS_LAYOUT_BULK_RECALL, &lo->plh_flags) \|\|
49a85061	183	pnfs_mark_matching_lsegs_invalid(lo, &free_me_list,
c88953d8	184	&args->cbl_range)) {
43f1b3da	185	rv = NFS4ERR_DELAY;
c88953d8 CH	186	goto unlock;
	187	}
	188
	189	if (NFS_SERVER(ino)->pnfs_curr_ld->return_range) {
	190	NFS_SERVER(ino)->pnfs_curr_ld->return_range(lo,
	191	&args->cbl_range);
	192	}
	193	unlock:
43f1b3da FI	194	spin_unlock(&ino->i_lock);
43f1b3da FI	195	pnfs_free_lseg_list(&free_me_list);
70c3bd2b	196	pnfs_put_layout_hdr(lo);
43f1b3da	197	iput(ino);
27999f25	198	out:
43f1b3da FI	199	return rv;
	200	}
	201
	202	static u32 initiate_bulk_draining(struct nfs_client *clp,
	203	struct cb_layoutrecallargs *args)
	204	{
fd9a8d71	205	int stat;
6382a441	206
fd9a8d71 TM	207	if (args->cbl_recall_type == RETURN_FSID)
	208	stat = pnfs_destroy_layouts_byfsid(clp, &args->cbl_fsid, true);
	209	else
	210	stat = pnfs_destroy_layouts_byclid(clp, true);
	211	if (stat != 0)
	212	return NFS4ERR_DELAY;
	213	return NFS4ERR_NOMATCHING_LAYOUT;
43f1b3da FI	214	}
	215
	216	static u32 do_callback_layoutrecall(struct nfs_client *clp,
	217	struct cb_layoutrecallargs *args)
	218	{
0cb3284b	219	u32 res;
43f1b3da FI	220
43f1b3da FI	221	dprintk("%s enter, type=%i\n", __func__, args->cbl_recall_type);
43f1b3da FI	222	if (args->cbl_recall_type == RETURN_FILE)
	223	res = initiate_file_draining(clp, args);
	224	else
	225	res = initiate_bulk_draining(clp, args);
43f1b3da FI	226	dprintk("%s returning %i\n", __func__, res);
	227	return res;
	228
	229	}
	230
f2a62561 FI	231	__be32 nfs4_callback_layoutrecall(struct cb_layoutrecallargs *args,
	232	void dummy, struct cb_process_state cps)
	233	{
43f1b3da FI	234	u32 res;
	235
	236	dprintk("%s: -->\n", __func__);
	237
	238	if (cps->clp)
	239	res = do_callback_layoutrecall(cps->clp, args);
	240	else
	241	res = NFS4ERR_OP_NOT_IN_SESSION;
	242
	243	dprintk("%s: exit with status = %d\n", __func__, res);
	244	return cpu_to_be32(res);
f2a62561 FI	245	}
f2a62561 FI	246
36840370 AB	247	static void pnfs_recall_all_layouts(struct nfs_client *clp)
	248	{
	249	struct cb_layoutrecallargs args;
	250
	251	/* Pretend we got a CB_LAYOUTRECALL(ALL) */
	252	memset(&args, 0, sizeof(args));
	253	args.cbl_recall_type = RETURN_ALL;
	254	/* FIXME we ignore errors, what should we do? */
	255	do_callback_layoutrecall(clp, &args);
	256	}
	257
1be5683b ME	258	__be32 nfs4_callback_devicenotify(struct cb_devicenotifyargs *args,
	259	void dummy, struct cb_process_state cps)
	260	{
	261	int i;
	262	__be32 res = 0;
	263	struct nfs_client *clp = cps->clp;
	264	struct nfs_server *server = NULL;
	265
	266	dprintk("%s: -->\n", __func__);
	267
	268	if (!clp) {
	269	res = cpu_to_be32(NFS4ERR_OP_NOT_IN_SESSION);
	270	goto out;
	271	}
	272
	273	for (i = 0; i < args->ndevs; i++) {
	274	struct cb_devicenotifyitem *dev = &args->devs[i];
	275
	276	if (!server \|\|
	277	server->pnfs_curr_ld->id != dev->cbd_layout_type) {
	278	rcu_read_lock();
	279	list_for_each_entry_rcu(server, &clp->cl_superblocks, client_link)
	280	if (server->pnfs_curr_ld &&
	281	server->pnfs_curr_ld->id == dev->cbd_layout_type) {
	282	rcu_read_unlock();
	283	goto found;
	284	}
	285	rcu_read_unlock();
	286	dprintk("%s: layout type %u not found\n",
	287	__func__, dev->cbd_layout_type);
	288	continue;
	289	}
	290
	291	found:
35c8bb54	292	nfs4_delete_deviceid(server->pnfs_curr_ld, clp, &dev->cbd_dev_id);
1be5683b ME	293	}
	294
	295	out:
	296	kfree(args->devs);
	297	dprintk("%s: exit with status = %u\n",
	298	__func__, be32_to_cpu(res));
	299	return res;
	300	}
	301
68f3f901 RL	302	/*
	303	* Validate the sequenceID sent by the server.
	304	* Return success if the sequenceID is one more than what we last saw on
	305	* this slot, accounting for wraparound. Increments the slot's sequence.
	306	*
4911096f AA	307	* We don't yet implement a duplicate request cache, instead we set the
4911096f AA	308	* back channel ca_maxresponsesize_cached to zero. This is OK for now
68f3f901 RL	309	* since we only currently implement idempotent callbacks anyway.
	310	*
	311	* We have a single slot backchannel at this time, so we don't bother
	312	* checking the used_slots bit array on the table. The lower layer guarantees
	313	* a single outstanding callback request at a time.
	314	*/
9733f0d9	315	static __be32
b2f28bd7	316	validate_seqid(struct nfs4_slot_table tbl, struct cb_sequenceargs args)
68f3f901 RL	317	{
	318	struct nfs4_slot *slot;
	319
e8d92382	320	dprintk("%s enter. slotid %u seqid %u\n",
b2f28bd7	321	__func__, args->csa_slotid, args->csa_sequenceid);
68f3f901	322
61f2e510	323	if (args->csa_slotid >= NFS41_BC_MAX_CALLBACKS)
68f3f901 RL	324	return htonl(NFS4ERR_BADSLOT);
68f3f901 RL	325
b2f28bd7	326	slot = tbl->slots + args->csa_slotid;
e8d92382	327	dprintk("%s slot table seqid: %u\n", __func__, slot->seq_nr);
68f3f901 RL	328
68f3f901 RL	329	/* Normal */
e937ee71	330	if (likely(args->csa_sequenceid == slot->seq_nr + 1))
55a67399	331	goto out_ok;
68f3f901 RL	332
68f3f901 RL	333	/* Replay */
b2f28bd7	334	if (args->csa_sequenceid == slot->seq_nr) {
e8d92382	335	dprintk("%s seqid %u is a replay\n",
b2f28bd7	336	__func__, args->csa_sequenceid);
4911096f AA	337	/* Signal process_op to set this error on next op */
	338	if (args->csa_cachethis == 0)
	339	return htonl(NFS4ERR_RETRY_UNCACHED_REP);
	340
	341	/* The ca_maxresponsesize_cached is 0 with no DRC */
	342	else if (args->csa_cachethis == 1)
	343	return htonl(NFS4ERR_REP_TOO_BIG_TO_CACHE);
68f3f901 RL	344	}
	345
	346	/* Wraparound */
b2f28bd7	347	if (args->csa_sequenceid == 1 && (slot->seq_nr + 1) == 0) {
68f3f901	348	slot->seq_nr = 1;
55a67399	349	goto out_ok;
68f3f901 RL	350	}
	351
	352	/* Misordered request */
	353	return htonl(NFS4ERR_SEQ_MISORDERED);
55a67399 TM	354	out_ok:
	355	tbl->highest_used_slotid = args->csa_slotid;
	356	return htonl(NFS4_OK);
68f3f901 RL	357	}
68f3f901 RL	358
a7989c3e MS	359	/*
	360	* For each referring call triple, check the session's slot table for
	361	* a match. If the slot is in use and the sequence numbers match, the
	362	* client is still waiting for a response to the original request.
	363	*/
	364	static bool referring_call_exists(struct nfs_client *clp,
	365	uint32_t nrclists,
	366	struct referring_call_list *rclists)
	367	{
	368	bool status = 0;
	369	int i, j;
	370	struct nfs4_session *session;
	371	struct nfs4_slot_table *tbl;
	372	struct referring_call_list *rclist;
	373	struct referring_call *ref;
	374
	375	/*
	376	* XXX When client trunking is implemented, this becomes
	377	* a session lookup from within the loop
	378	*/
	379	session = clp->cl_session;
	380	tbl = &session->fc_slot_table;
	381
	382	for (i = 0; i < nrclists; i++) {
	383	rclist = &rclists[i];
	384	if (memcmp(session->sess_id.data,
	385	rclist->rcl_sessionid.data,
	386	NFS4_MAX_SESSIONID_LEN) != 0)
	387	continue;
	388
	389	for (j = 0; j < rclist->rcl_nrefcalls; j++) {
	390	ref = &rclist->rcl_refcalls[j];
	391
	392	dprintk("%s: sessionid %x:%x:%x:%x sequenceid %u "
	393	"slotid %u\n", __func__,
	394	((u32 *)&rclist->rcl_sessionid.data)[0],
	395	((u32 *)&rclist->rcl_sessionid.data)[1],
	396	((u32 *)&rclist->rcl_sessionid.data)[2],
	397	((u32 *)&rclist->rcl_sessionid.data)[3],
	398	ref->rc_sequenceid, ref->rc_slotid);
	399
	400	spin_lock(&tbl->slot_tbl_lock);
	401	status = (test_bit(ref->rc_slotid, tbl->used_slots) &&
	402	tbl->slots[ref->rc_slotid].seq_nr ==
	403	ref->rc_sequenceid);
	404	spin_unlock(&tbl->slot_tbl_lock);
	405	if (status)
	406	goto out;
	407	}
	408	}
	409
	410	out:
	411	return status;
	412	}
	413
9733f0d9	414	__be32 nfs4_callback_sequence(struct cb_sequenceargs *args,
c36fca52 AA	415	struct cb_sequenceres *res,
c36fca52 AA	416	struct cb_process_state *cps)
d49433e1	417	{
55a67399	418	struct nfs4_slot_table *tbl;
e937ee71	419	struct nfs4_slot *slot;
68f3f901	420	struct nfs_client *clp;
9733f0d9	421	int i;
778be232	422	__be32 status = htonl(NFS4ERR_BADSESSION);
d49433e1	423
459de2ed BS	424	clp = nfs4_find_client_sessionid(cps->net, args->csa_addr,
459de2ed BS	425	&args->csa_sessionid, cps->minorversion);
68f3f901 RL	426	if (clp == NULL)
	427	goto out;
	428
b1c0df5f TM	429	if (!(clp->cl_session->flags & SESSION4_BACK_CHAN))
b1c0df5f TM	430	goto out;
e937ee71	431
55a67399	432	tbl = &clp->cl_session->bc_slot_table;
e937ee71	433	slot = tbl->slots + args->csa_slotid;
55a67399 TM	434
55a67399 TM	435	spin_lock(&tbl->slot_tbl_lock);
42acd021	436	/* state manager is resetting the session */
774d5f14	437	if (test_bit(NFS4_SLOT_TBL_DRAINING, &tbl->slot_tbl_state)) {
55a67399	438	status = htonl(NFS4ERR_DELAY);
910ac68a TM	439	/* Return NFS4ERR_BADSESSION if we're draining the session
	440	* in order to reset it.
	441	*/
	442	if (test_bit(NFS4CLNT_SESSION_RESET, &clp->cl_state))
	443	status = htonl(NFS4ERR_BADSESSION);
4e54ab8d	444	goto out_unlock;
42acd021 AA	445	}
42acd021 AA	446
0579c8d2 KM	447	memcpy(&res->csr_sessionid, &args->csa_sessionid,
	448	sizeof(res->csr_sessionid));
	449	res->csr_sequenceid = args->csa_sequenceid;
	450	res->csr_slotid = args->csa_slotid;
	451	res->csr_highestslotid = NFS41_BC_MAX_CALLBACKS - 1;
	452	res->csr_target_highestslotid = NFS41_BC_MAX_CALLBACKS - 1;
	453
e937ee71	454	status = validate_seqid(tbl, args);
68f3f901	455	if (status)
4e54ab8d	456	goto out_unlock;
68f3f901	457
55a67399 TM	458	cps->slotid = args->csa_slotid;
55a67399 TM	459
72ce2b3c MS	460	/*
	461	* Check for pending referring calls. If a match is found, a
	462	* related callback was received before the response to the original
	463	* call.
	464	*/
	465	if (referring_call_exists(clp, args->csa_nrclists, args->csa_rclists)) {
	466	status = htonl(NFS4ERR_DELAY);
4e54ab8d	467	goto out_unlock;
72ce2b3c MS	468	}
72ce2b3c MS	469
e937ee71 KM	470	/*
	471	* RFC5661 20.9.3
	472	* If CB_SEQUENCE returns an error, then the state of the slot
	473	* (sequence ID, cached reply) MUST NOT change.
	474	*/
	475	slot->seq_nr++;
4e54ab8d TM	476	out_unlock:
4e54ab8d TM	477	spin_unlock(&tbl->slot_tbl_lock);
e937ee71	478
68f3f901	479	out:
2c4cdf8f	480	cps->clp = clp; /* put in nfs4_callback_compound */
72ce2b3c MS	481	for (i = 0; i < args->csa_nrclists; i++)
	482	kfree(args->csa_rclists[i].rcl_refcalls);
	483	kfree(args->csa_rclists);
	484
c36fca52 AA	485	if (status == htonl(NFS4ERR_RETRY_UNCACHED_REP)) {
	486	cps->drc_status = status;
	487	status = 0;
	488	} else
4911096f	489	res->csr_status = status;
c36fca52	490
2f92ae34	491	trace_nfs4_cb_sequence(args, res, status);
4911096f AA	492	dprintk("%s: exit with status = %d res->csr_status %d\n", __func__,
	493	ntohl(status), ntohl(res->csr_status));
	494	return status;
d49433e1 BH	495	}
d49433e1 BH	496
36840370 AB	497	static bool
	498	validate_bitmap_values(unsigned long mask)
	499	{
	500	return (mask & ~RCA4_TYPE_MASK_ALL) == 0;
	501	}
	502
c36fca52 AA	503	__be32 nfs4_callback_recallany(struct cb_recallanyargs args, void dummy,
c36fca52 AA	504	struct cb_process_state *cps)
31f09607	505	{
9733f0d9	506	__be32 status;
31f09607 AB	507	fmode_t flags = 0;
31f09607 AB	508
36840370	509	status = cpu_to_be32(NFS4ERR_OP_NOT_IN_SESSION);
c36fca52	510	if (!cps->clp) /* set in cb_sequence */
31f09607 AB	511	goto out;
31f09607 AB	512
2446ab60	513	dprintk_rcu("NFS: RECALL_ANY callback request from %s\n",
c36fca52	514	rpc_peeraddr2str(cps->clp->cl_rpcclient, RPC_DISPLAY_ADDR));
31f09607	515
36840370 AB	516	status = cpu_to_be32(NFS4ERR_INVAL);
	517	if (!validate_bitmap_values(args->craa_type_mask))
	518	goto out;
	519
	520	status = cpu_to_be32(NFS4_OK);
31f09607 AB	521	if (test_bit(RCA4_TYPE_MASK_RDATA_DLG, (const unsigned long *)
	522	&args->craa_type_mask))
	523	flags = FMODE_READ;
	524	if (test_bit(RCA4_TYPE_MASK_WDATA_DLG, (const unsigned long *)
	525	&args->craa_type_mask))
	526	flags \|= FMODE_WRITE;
36840370 AB	527	if (test_bit(RCA4_TYPE_MASK_FILE_LAYOUT, (const unsigned long *)
	528	&args->craa_type_mask))
	529	pnfs_recall_all_layouts(cps->clp);
31f09607	530	if (flags)
826e0013	531	nfs_expire_unused_delegation_types(cps->clp, flags);
31f09607 AB	532	out:
	533	dprintk("%s: exit with status = %d\n", __func__, ntohl(status));
	534	return status;
	535	}
b9efa1b2 AA	536
b9efa1b2 AA	537	/* Reduce the fore channel's max_slots to the target value */
c36fca52 AA	538	__be32 nfs4_callback_recallslot(struct cb_recallslotargs args, void dummy,
c36fca52 AA	539	struct cb_process_state *cps)
b9efa1b2	540	{
b9efa1b2	541	struct nfs4_slot_table *fc_tbl;
9733f0d9	542	__be32 status;
b9efa1b2 AA	543
b9efa1b2 AA	544	status = htonl(NFS4ERR_OP_NOT_IN_SESSION);
c36fca52	545	if (!cps->clp) /* set in cb_sequence */
b9efa1b2 AA	546	goto out;
b9efa1b2 AA	547
e8d92382	548	dprintk_rcu("NFS: CB_RECALL_SLOT request from %s target highest slotid %u\n",
c36fca52	549	rpc_peeraddr2str(cps->clp->cl_rpcclient, RPC_DISPLAY_ADDR),
d5fb4ce3	550	args->crsa_target_highest_slotid);
b9efa1b2	551
c36fca52	552	fc_tbl = &cps->clp->cl_session->fc_slot_table;
b9efa1b2	553
bae0ac0e	554	status = htonl(NFS4_OK);
b9efa1b2	555
d5fb4ce3	556	nfs41_set_target_slotid(fc_tbl, args->crsa_target_highest_slotid);
ac074835	557	nfs41_server_notify_target_slotid_update(cps->clp);
b9efa1b2 AA	558	out:
	559	dprintk("%s: exit with status = %d\n", __func__, ntohl(status));
	560	return status;
	561	}
d49433e1	562	#endif /* CONFIG_NFS_V4_1 */