[linux-2.6-block.git] / fs / nfsd / nfssvc.c

/*
 * linux/fs/nfsd/nfssvc.c
 *
 * Central processing for nfsd.
 *
 * Authors:	Olaf Kirch (okir@monad.swb.de)
 *
 * Copyright (C) 1995, 1996, 1997 Olaf Kirch <okir@monad.swb.de>
 */

#include <linux/module.h>
#include <linux/sched.h>
#include <linux/time.h>
#include <linux/errno.h>
#include <linux/nfs.h>
#include <linux/in.h>
#include <linux/uio.h>
#include <linux/unistd.h>
#include <linux/slab.h>
#include <linux/smp.h>
#include <linux/smp_lock.h>
#include <linux/freezer.h>
#include <linux/fs_struct.h>
#include <linux/kthread.h>

#include <linux/sunrpc/types.h>
#include <linux/sunrpc/stats.h>
#include <linux/sunrpc/svc.h>
#include <linux/sunrpc/svcsock.h>
#include <linux/sunrpc/cache.h>
#include <linux/nfsd/nfsd.h>
#include <linux/nfsd/stats.h>
#include <linux/nfsd/cache.h>
#include <linux/nfsd/syscall.h>
#include <linux/lockd/bind.h>
#include <linux/nfsacl.h>

#define NFSDDBG_FACILITY	NFSDDBG_SVC

/* these signals will be delivered to an nfsd thread 
 * when handling a request
 */
#define ALLOWED_SIGS	(sigmask(SIGKILL))
/* these signals will be delivered to an nfsd thread
 * when not handling a request. i.e. when waiting
 */
#define SHUTDOWN_SIGS	(sigmask(SIGKILL) | sigmask(SIGHUP) | sigmask(SIGINT) | sigmask(SIGQUIT))

extern struct svc_program	nfsd_program;
static int			nfsd(void *vrqstp);
struct timeval			nfssvc_boot;
static atomic_t			nfsd_busy;
static unsigned long		nfsd_last_call;
static DEFINE_SPINLOCK(nfsd_call_lock);

/*
 * nfsd_mutex protects nfsd_serv -- both the pointer itself and the members
 * of the svc_serv struct. In particular, ->sv_nrthreads but also to some
 * extent ->sv_temp_socks and ->sv_permsocks. It also protects nfsdstats.th_cnt
 *
 * If (out side the lock) nfsd_serv is non-NULL, then it must point to a
 * properly initialised 'struct svc_serv' with ->sv_nrthreads > 0. That number
 * of nfsd threads must exist and each must listed in ->sp_all_threads in each
 * entry of ->sv_pools[].
 *
 * Transitions of the thread count between zero and non-zero are of particular
 * interest since the svc_serv needs to be created and initialized at that
 * point, or freed.
 *
 * Finally, the nfsd_mutex also protects some of the global variables that are
 * accessed when nfsd starts and that are settable via the write_* routines in
 * nfsctl.c. In particular:
 *
 *	user_recovery_dirname
 *	user_lease_time
 *	nfsd_versions
 */
DEFINE_MUTEX(nfsd_mutex);
struct svc_serv 		*nfsd_serv;

#if defined(CONFIG_NFSD_V2_ACL) || defined(CONFIG_NFSD_V3_ACL)
static struct svc_stat	nfsd_acl_svcstats;
static struct svc_version *	nfsd_acl_version[] = {
	[2] = &nfsd_acl_version2,
	[3] = &nfsd_acl_version3,
};

#define NFSD_ACL_MINVERS            2
#define NFSD_ACL_NRVERS		ARRAY_SIZE(nfsd_acl_version)
static struct svc_version *nfsd_acl_versions[NFSD_ACL_NRVERS];

static struct svc_program	nfsd_acl_program = {
	.pg_prog		= NFS_ACL_PROGRAM,
	.pg_nvers		= NFSD_ACL_NRVERS,
	.pg_vers		= nfsd_acl_versions,
	.pg_name		= "nfsacl",
	.pg_class		= "nfsd",
	.pg_stats		= &nfsd_acl_svcstats,
	.pg_authenticate	= &svc_set_client,
};

static struct svc_stat	nfsd_acl_svcstats = {
	.program	= &nfsd_acl_program,
};
#endif /* defined(CONFIG_NFSD_V2_ACL) || defined(CONFIG_NFSD_V3_ACL) */

static struct svc_version *	nfsd_version[] = {
	[2] = &nfsd_version2,
#if defined(CONFIG_NFSD_V3)
	[3] = &nfsd_version3,
#endif
#if defined(CONFIG_NFSD_V4)
	[4] = &nfsd_version4,
#endif
};

#define NFSD_MINVERS    	2
#define NFSD_NRVERS		ARRAY_SIZE(nfsd_version)
static struct svc_version *nfsd_versions[NFSD_NRVERS];

struct svc_program		nfsd_program = {
#if defined(CONFIG_NFSD_V2_ACL) || defined(CONFIG_NFSD_V3_ACL)
	.pg_next		= &nfsd_acl_program,
#endif
	.pg_prog		= NFS_PROGRAM,		/* program number */
	.pg_nvers		= NFSD_NRVERS,		/* nr of entries in nfsd_version */
	.pg_vers		= nfsd_versions,	/* version table */
	.pg_name		= "nfsd",		/* program name */
	.pg_class		= "nfsd",		/* authentication class */
	.pg_stats		= &nfsd_svcstats,	/* version table */
	.pg_authenticate	= &svc_set_client,	/* export authentication */

};

int nfsd_vers(int vers, enum vers_op change)
{
	if (vers < NFSD_MINVERS || vers >= NFSD_NRVERS)
		return -1;
	switch(change) {
	case NFSD_SET:
		nfsd_versions[vers] = nfsd_version[vers];
#if defined(CONFIG_NFSD_V2_ACL) || defined(CONFIG_NFSD_V3_ACL)
		if (vers < NFSD_ACL_NRVERS)
			nfsd_acl_versions[vers] = nfsd_acl_version[vers];
#endif
		break;
	case NFSD_CLEAR:
		nfsd_versions[vers] = NULL;
#if defined(CONFIG_NFSD_V2_ACL) || defined(CONFIG_NFSD_V3_ACL)
		if (vers < NFSD_ACL_NRVERS)
			nfsd_acl_versions[vers] = NULL;
#endif
		break;
	case NFSD_TEST:
		return nfsd_versions[vers] != NULL;
	case NFSD_AVAIL:
		return nfsd_version[vers] != NULL;
	}
	return 0;
}
/*
 * Maximum number of nfsd processes
 */
#define	NFSD_MAXSERVS		8192

int nfsd_nrthreads(void)
{
	int rv = 0;
	mutex_lock(&nfsd_mutex);
	if (nfsd_serv)
		rv = nfsd_serv->sv_nrthreads;
	mutex_unlock(&nfsd_mutex);
	return rv;
}

static void nfsd_last_thread(struct svc_serv *serv)
{
	/* When last nfsd thread exits we need to do some clean-up */
	struct svc_xprt *xprt;
	list_for_each_entry(xprt, &serv->sv_permsocks, xpt_list)
		lockd_down();
	nfsd_serv = NULL;
	nfsd_racache_shutdown();
	nfs4_state_shutdown();

	printk(KERN_WARNING "nfsd: last server has exited, flushing export "
			    "cache\n");
	nfsd_export_flush();
}

void nfsd_reset_versions(void)
{
	int found_one = 0;
	int i;

	for (i = NFSD_MINVERS; i < NFSD_NRVERS; i++) {
		if (nfsd_program.pg_vers[i])
			found_one = 1;
	}

	if (!found_one) {
		for (i = NFSD_MINVERS; i < NFSD_NRVERS; i++)
			nfsd_program.pg_vers[i] = nfsd_version[i];
#if defined(CONFIG_NFSD_V2_ACL) || defined(CONFIG_NFSD_V3_ACL)
		for (i = NFSD_ACL_MINVERS; i < NFSD_ACL_NRVERS; i++)
			nfsd_acl_program.pg_vers[i] =
				nfsd_acl_version[i];
#endif
	}
}


int nfsd_create_serv(void)
{
	int err = 0;

	WARN_ON(!mutex_is_locked(&nfsd_mutex));
	if (nfsd_serv) {
		svc_get(nfsd_serv);
		return 0;
	}
	if (nfsd_max_blksize == 0) {
		/* choose a suitable default */
		struct sysinfo i;
		si_meminfo(&i);
		/* Aim for 1/4096 of memory per thread
		 * This gives 1MB on 4Gig machines
		 * But only uses 32K on 128M machines.
		 * Bottom out at 8K on 32M and smaller.
		 * Of course, this is only a default.
		 */
		nfsd_max_blksize = NFSSVC_MAXBLKSIZE;
		i.totalram <<= PAGE_SHIFT - 12;
		while (nfsd_max_blksize > i.totalram &&
		       nfsd_max_blksize >= 8*1024*2)
			nfsd_max_blksize /= 2;
	}

	atomic_set(&nfsd_busy, 0);
	nfsd_serv = svc_create_pooled(&nfsd_program, nfsd_max_blksize,
				      nfsd_last_thread, nfsd, THIS_MODULE);
	if (nfsd_serv == NULL)
		err = -ENOMEM;

	do_gettimeofday(&nfssvc_boot);		/* record boot time */
	return err;
}

static int nfsd_init_socks(int port)
{
	int error;
	if (!list_empty(&nfsd_serv->sv_permsocks))
		return 0;

	error = lockd_up(IPPROTO_UDP);
	if (error >= 0) {
		error = svc_create_xprt(nfsd_serv, "udp", port,
					SVC_SOCK_DEFAULTS);
		if (error < 0)
			lockd_down();
	}
	if (error < 0)
		return error;

	error = lockd_up(IPPROTO_TCP);
	if (error >= 0) {
		error = svc_create_xprt(nfsd_serv, "tcp", port,
					SVC_SOCK_DEFAULTS);
		if (error < 0)
			lockd_down();
	}
	if (error < 0)
		return error;
	return 0;
}

int nfsd_nrpools(void)
{
	if (nfsd_serv == NULL)
		return 0;
	else
		return nfsd_serv->sv_nrpools;
}

int nfsd_get_nrthreads(int n, int *nthreads)
{
	int i = 0;

	if (nfsd_serv != NULL) {
		for (i = 0; i < nfsd_serv->sv_nrpools && i < n; i++)
			nthreads[i] = nfsd_serv->sv_pools[i].sp_nrthreads;
	}

	return 0;
}

int nfsd_set_nrthreads(int n, int *nthreads)
{
	int i = 0;
	int tot = 0;
	int err = 0;

	WARN_ON(!mutex_is_locked(&nfsd_mutex));

	if (nfsd_serv == NULL || n <= 0)
		return 0;

	if (n > nfsd_serv->sv_nrpools)
		n = nfsd_serv->sv_nrpools;

	/* enforce a global maximum number of threads */
	tot = 0;
	for (i = 0; i < n; i++) {
		if (nthreads[i] > NFSD_MAXSERVS)
			nthreads[i] = NFSD_MAXSERVS;
		tot += nthreads[i];
	}
	if (tot > NFSD_MAXSERVS) {
		/* total too large: scale down requested numbers */
		for (i = 0; i < n && tot > 0; i++) {
		    	int new = nthreads[i] * NFSD_MAXSERVS / tot;
			tot -= (nthreads[i] - new);
			nthreads[i] = new;
		}
		for (i = 0; i < n && tot > 0; i++) {
			nthreads[i]--;
			tot--;
		}
	}

	/*
	 * There must always be a thread in pool 0; the admin
	 * can't shut down NFS completely using pool_threads.
	 */
	if (nthreads[0] == 0)
		nthreads[0] = 1;

	/* apply the new numbers */
	svc_get(nfsd_serv);
	for (i = 0; i < n; i++) {
		err = svc_set_num_threads(nfsd_serv, &nfsd_serv->sv_pools[i],
				    	  nthreads[i]);
		if (err)
			break;
	}
	svc_destroy(nfsd_serv);

	return err;
}

int
nfsd_svc(unsigned short port, int nrservs)
{
	int	error;

	mutex_lock(&nfsd_mutex);
	dprintk("nfsd: creating service\n");
	error = -EINVAL;
	if (nrservs <= 0)
		nrservs = 0;
	if (nrservs > NFSD_MAXSERVS)
		nrservs = NFSD_MAXSERVS;
	
	/* Readahead param cache - will no-op if it already exists */
	error =	nfsd_racache_init(2*nrservs);
	if (error<0)
		goto out;
	nfs4_state_start();

	nfsd_reset_versions();

	error = nfsd_create_serv();

	if (error)
		goto out;
	error = nfsd_init_socks(port);
	if (error)
		goto failure;

	error = svc_set_num_threads(nfsd_serv, NULL, nrservs);
 failure:
	svc_destroy(nfsd_serv);		/* Release server */
 out:
	mutex_unlock(&nfsd_mutex);
	return error;
}

static inline void
update_thread_usage(int busy_threads)
{
	unsigned long prev_call;
	unsigned long diff;
	int decile;

	spin_lock(&nfsd_call_lock);
	prev_call = nfsd_last_call;
	nfsd_last_call = jiffies;
	decile = busy_threads*10/nfsdstats.th_cnt;
	if (decile>0 && decile <= 10) {
		diff = nfsd_last_call - prev_call;
		if ( (nfsdstats.th_usage[decile-1] += diff) >= NFSD_USAGE_WRAP)
			nfsdstats.th_usage[decile-1] -= NFSD_USAGE_WRAP;
		if (decile == 10)
			nfsdstats.th_fullcnt++;
	}
	spin_unlock(&nfsd_call_lock);
}

/*
 * This is the NFS server kernel thread
 */
static int
nfsd(void *vrqstp)
{
	struct svc_rqst *rqstp = (struct svc_rqst *) vrqstp;
	struct fs_struct *fsp;
	sigset_t shutdown_mask, allowed_mask;
	int err, preverr = 0;
	unsigned int signo;

	/* Lock module and set up kernel thread */
	mutex_lock(&nfsd_mutex);

	/* At this point, the thread shares current->fs
	 * with the init process. We need to create files with a
	 * umask of 0 instead of init's umask. */
	fsp = copy_fs_struct(current->fs);
	if (!fsp) {
		printk("Unable to start nfsd thread: out of memory\n");
		goto out;
	}
	exit_fs(current);
	current->fs = fsp;
	current->fs->umask = 0;

	siginitsetinv(&shutdown_mask, SHUTDOWN_SIGS);
	siginitsetinv(&allowed_mask, ALLOWED_SIGS);

	/*
	 * thread is spawned with all signals set to SIG_IGN, re-enable
	 * the ones that matter
	 */
	for (signo = 1; signo <= _NSIG; signo++) {
		if (!sigismember(&shutdown_mask, signo))
			allow_signal(signo);
	}

	nfsdstats.th_cnt++;
	mutex_unlock(&nfsd_mutex);

	/*
	 * We want less throttling in balance_dirty_pages() so that nfs to
	 * localhost doesn't cause nfsd to lock up due to all the client's
	 * dirty pages.
	 */
	current->flags |= PF_LESS_THROTTLE;
	set_freezable();

	/*
	 * The main request loop
	 */
	for (;;) {
		/* Block all but the shutdown signals */
		sigprocmask(SIG_SETMASK, &shutdown_mask, NULL);

		/*
		 * Find a socket with data available and call its
		 * recvfrom routine.
		 */
		while ((err = svc_recv(rqstp, 60*60*HZ)) == -EAGAIN)
			;
		if (err == -EINTR)
			break;
		else if (err < 0) {
			if (err != preverr) {
				printk(KERN_WARNING "%s: unexpected error "
					"from svc_recv (%d)\n", __func__, -err);
				preverr = err;
			}
			schedule_timeout_uninterruptible(HZ);
			continue;
		}

		update_thread_usage(atomic_read(&nfsd_busy));
		atomic_inc(&nfsd_busy);

		/* Lock the export hash tables for reading. */
		exp_readlock();

		/* Process request with signals blocked. */
		sigprocmask(SIG_SETMASK, &allowed_mask, NULL);

		svc_process(rqstp);

		/* Unlock export hash tables */
		exp_readunlock();
		update_thread_usage(atomic_read(&nfsd_busy));
		atomic_dec(&nfsd_busy);
	}

	/* Clear signals before calling svc_exit_thread() */
	flush_signals(current);

	mutex_lock(&nfsd_mutex);
	nfsdstats.th_cnt --;

out:
	/* Release the thread */
	svc_exit_thread(rqstp);

	/* Release module */
	mutex_unlock(&nfsd_mutex);
	module_put_and_exit(0);
	return 0;
}

static __be32 map_new_errors(u32 vers, __be32 nfserr)
{
	if (nfserr == nfserr_jukebox && vers == 2)
		return nfserr_dropit;
	if (nfserr == nfserr_wrongsec && vers < 4)
		return nfserr_acces;
	return nfserr;
}

int
nfsd_dispatch(struct svc_rqst *rqstp, __be32 *statp)
{
	struct svc_procedure	*proc;
	kxdrproc_t		xdr;
	__be32			nfserr;
	__be32			*nfserrp;

	dprintk("nfsd_dispatch: vers %d proc %d\n",
				rqstp->rq_vers, rqstp->rq_proc);
	proc = rqstp->rq_procinfo;

	/* Check whether we have this call in the cache. */
	switch (nfsd_cache_lookup(rqstp, proc->pc_cachetype)) {
	case RC_INTR:
	case RC_DROPIT:
		return 0;
	case RC_REPLY:
		return 1;
	case RC_DOIT:;
		/* do it */
	}

	/* Decode arguments */
	xdr = proc->pc_decode;
	if (xdr && !xdr(rqstp, (__be32*)rqstp->rq_arg.head[0].iov_base,
			rqstp->rq_argp)) {
		dprintk("nfsd: failed to decode arguments!\n");
		nfsd_cache_update(rqstp, RC_NOCACHE, NULL);
		*statp = rpc_garbage_args;
		return 1;
	}

	/* need to grab the location to store the status, as
	 * nfsv4 does some encoding while processing 
	 */
	nfserrp = rqstp->rq_res.head[0].iov_base
		+ rqstp->rq_res.head[0].iov_len;
	rqstp->rq_res.head[0].iov_len += sizeof(__be32);

	/* Now call the procedure handler, and encode NFS status. */
	nfserr = proc->pc_func(rqstp, rqstp->rq_argp, rqstp->rq_resp);
	nfserr = map_new_errors(rqstp->rq_vers, nfserr);
	if (nfserr == nfserr_dropit) {
		dprintk("nfsd: Dropping request; may be revisited later\n");
		nfsd_cache_update(rqstp, RC_NOCACHE, NULL);
		return 0;
	}

	if (rqstp->rq_proc != 0)
		*nfserrp++ = nfserr;

	/* Encode result.
	 * For NFSv2, additional info is never returned in case of an error.
	 */
	if (!(nfserr && rqstp->rq_vers == 2)) {
		xdr = proc->pc_encode;
		if (xdr && !xdr(rqstp, nfserrp,
				rqstp->rq_resp)) {
			/* Failed to encode result. Release cache entry */
			dprintk("nfsd: failed to encode result!\n");
			nfsd_cache_update(rqstp, RC_NOCACHE, NULL);
			*statp = rpc_system_err;
			return 1;
		}
	}

	/* Store reply in cache. */
	nfsd_cache_update(rqstp, proc->pc_cachetype, statp + 1);
	return 1;
}
Commit	Line	Data
1da177e4 LT	1	/*
	2	* linux/fs/nfsd/nfssvc.c
	3	*
	4	* Central processing for nfsd.
	5	*
	6	* Authors: Olaf Kirch (okir@monad.swb.de)
	7	*
	8	* Copyright (C) 1995, 1996, 1997 Olaf Kirch <okir@monad.swb.de>
	9	*/
	10
1da177e4	11	#include <linux/module.h>
e8edc6e0	12	#include <linux/sched.h>
1da177e4 LT	13	#include <linux/time.h>
	14	#include <linux/errno.h>
	15	#include <linux/nfs.h>
	16	#include <linux/in.h>
	17	#include <linux/uio.h>
	18	#include <linux/unistd.h>
	19	#include <linux/slab.h>
	20	#include <linux/smp.h>
	21	#include <linux/smp_lock.h>
83144186	22	#include <linux/freezer.h>
1da177e4	23	#include <linux/fs_struct.h>
9867d76c	24	#include <linux/kthread.h>
1da177e4 LT	25
	26	#include <linux/sunrpc/types.h>
	27	#include <linux/sunrpc/stats.h>
	28	#include <linux/sunrpc/svc.h>
	29	#include <linux/sunrpc/svcsock.h>
	30	#include <linux/sunrpc/cache.h>
	31	#include <linux/nfsd/nfsd.h>
	32	#include <linux/nfsd/stats.h>
	33	#include <linux/nfsd/cache.h>
70c3b76c	34	#include <linux/nfsd/syscall.h>
1da177e4	35	#include <linux/lockd/bind.h>
a257cdd0	36	#include <linux/nfsacl.h>
1da177e4 LT	37
	38	#define NFSDDBG_FACILITY NFSDDBG_SVC
	39
	40	/* these signals will be delivered to an nfsd thread
	41	* when handling a request
	42	*/
	43	#define ALLOWED_SIGS (sigmask(SIGKILL))
	44	/* these signals will be delivered to an nfsd thread
	45	* when not handling a request. i.e. when waiting
	46	*/
	47	#define SHUTDOWN_SIGS (sigmask(SIGKILL) \| sigmask(SIGHUP) \| sigmask(SIGINT) \| sigmask(SIGQUIT))
1da177e4 LT	48
1da177e4 LT	49	extern struct svc_program nfsd_program;
9867d76c	50	static int nfsd(void *vrqstp);
1da177e4	51	struct timeval nfssvc_boot;
1da177e4 LT	52	static atomic_t nfsd_busy;
	53	static unsigned long nfsd_last_call;
	54	static DEFINE_SPINLOCK(nfsd_call_lock);
	55
bedbdd8b NB	56	/*
	57	* nfsd_mutex protects nfsd_serv -- both the pointer itself and the members
	58	* of the svc_serv struct. In particular, ->sv_nrthreads but also to some
	59	* extent ->sv_temp_socks and ->sv_permsocks. It also protects nfsdstats.th_cnt
	60	*
	61	* If (out side the lock) nfsd_serv is non-NULL, then it must point to a
	62	* properly initialised 'struct svc_serv' with ->sv_nrthreads > 0. That number
	63	* of nfsd threads must exist and each must listed in ->sp_all_threads in each
	64	* entry of ->sv_pools[].
	65	*
	66	* Transitions of the thread count between zero and non-zero are of particular
	67	* interest since the svc_serv needs to be created and initialized at that
	68	* point, or freed.
3dd98a3b JL	69	*
	70	* Finally, the nfsd_mutex also protects some of the global variables that are
	71	* accessed when nfsd starts and that are settable via the write_* routines in
	72	* nfsctl.c. In particular:
	73	*
	74	* user_recovery_dirname
	75	* user_lease_time
	76	* nfsd_versions
bedbdd8b NB	77	*/
	78	DEFINE_MUTEX(nfsd_mutex);
	79	struct svc_serv *nfsd_serv;
	80
3fb803a9 AG	81	#if defined(CONFIG_NFSD_V2_ACL) \|\| defined(CONFIG_NFSD_V3_ACL)
	82	static struct svc_stat nfsd_acl_svcstats;
	83	static struct svc_version * nfsd_acl_version[] = {
	84	[2] = &nfsd_acl_version2,
	85	[3] = &nfsd_acl_version3,
	86	};
	87
	88	#define NFSD_ACL_MINVERS 2
e8c96f8c	89	#define NFSD_ACL_NRVERS ARRAY_SIZE(nfsd_acl_version)
3fb803a9 AG	90	static struct svc_version *nfsd_acl_versions[NFSD_ACL_NRVERS];
	91
	92	static struct svc_program nfsd_acl_program = {
	93	.pg_prog = NFS_ACL_PROGRAM,
	94	.pg_nvers = NFSD_ACL_NRVERS,
	95	.pg_vers = nfsd_acl_versions,
1a8eff6d	96	.pg_name = "nfsacl",
3fb803a9 AG	97	.pg_class = "nfsd",
	98	.pg_stats = &nfsd_acl_svcstats,
	99	.pg_authenticate = &svc_set_client,
	100	};
	101
	102	static struct svc_stat nfsd_acl_svcstats = {
	103	.program = &nfsd_acl_program,
	104	};
	105	#endif /* defined(CONFIG_NFSD_V2_ACL) \|\| defined(CONFIG_NFSD_V3_ACL) */
	106
70c3b76c N	107	static struct svc_version * nfsd_version[] = {
	108	[2] = &nfsd_version2,
	109	#if defined(CONFIG_NFSD_V3)
	110	[3] = &nfsd_version3,
	111	#endif
	112	#if defined(CONFIG_NFSD_V4)
	113	[4] = &nfsd_version4,
	114	#endif
	115	};
	116
	117	#define NFSD_MINVERS 2
e8c96f8c	118	#define NFSD_NRVERS ARRAY_SIZE(nfsd_version)
70c3b76c N	119	static struct svc_version *nfsd_versions[NFSD_NRVERS];
	120
	121	struct svc_program nfsd_program = {
3fb803a9 AG	122	#if defined(CONFIG_NFSD_V2_ACL) \|\| defined(CONFIG_NFSD_V3_ACL)
	123	.pg_next = &nfsd_acl_program,
	124	#endif
70c3b76c N	125	.pg_prog = NFS_PROGRAM, /* program number */
	126	.pg_nvers = NFSD_NRVERS, /* nr of entries in nfsd_version */
	127	.pg_vers = nfsd_versions, /* version table */
	128	.pg_name = "nfsd", /* program name */
	129	.pg_class = "nfsd", /* authentication class */
	130	.pg_stats = &nfsd_svcstats, /* version table */
	131	.pg_authenticate = &svc_set_client, /* export authentication */
	132
	133	};
	134
6658d3a7 N	135	int nfsd_vers(int vers, enum vers_op change)
	136	{
	137	if (vers < NFSD_MINVERS \|\| vers >= NFSD_NRVERS)
	138	return -1;
	139	switch(change) {
	140	case NFSD_SET:
	141	nfsd_versions[vers] = nfsd_version[vers];
6658d3a7 N	142	#if defined(CONFIG_NFSD_V2_ACL) \|\| defined(CONFIG_NFSD_V3_ACL)
6658d3a7 N	143	if (vers < NFSD_ACL_NRVERS)
1a8eff6d	144	nfsd_acl_versions[vers] = nfsd_acl_version[vers];
6658d3a7	145	#endif
1a8eff6d	146	break;
6658d3a7 N	147	case NFSD_CLEAR:
	148	nfsd_versions[vers] = NULL;
	149	#if defined(CONFIG_NFSD_V2_ACL) \|\| defined(CONFIG_NFSD_V3_ACL)
	150	if (vers < NFSD_ACL_NRVERS)
1a8eff6d	151	nfsd_acl_versions[vers] = NULL;
6658d3a7 N	152	#endif
	153	break;
	154	case NFSD_TEST:
	155	return nfsd_versions[vers] != NULL;
	156	case NFSD_AVAIL:
	157	return nfsd_version[vers] != NULL;
	158	}
	159	return 0;
	160	}
1da177e4 LT	161	/*
	162	* Maximum number of nfsd processes
	163	*/
	164	#define NFSD_MAXSERVS 8192
	165
	166	int nfsd_nrthreads(void)
	167	{
c7d106c9 NB	168	int rv = 0;
	169	mutex_lock(&nfsd_mutex);
	170	if (nfsd_serv)
	171	rv = nfsd_serv->sv_nrthreads;
	172	mutex_unlock(&nfsd_mutex);
	173	return rv;
1da177e4 LT	174	}
1da177e4 LT	175
bc591ccf N	176	static void nfsd_last_thread(struct svc_serv *serv)
	177	{
	178	/* When last nfsd thread exits we need to do some clean-up */
7a182083 TT	179	struct svc_xprt *xprt;
7a182083 TT	180	list_for_each_entry(xprt, &serv->sv_permsocks, xpt_list)
24e36663	181	lockd_down();
bc591ccf N	182	nfsd_serv = NULL;
	183	nfsd_racache_shutdown();
	184	nfs4_state_shutdown();
	185
e096bbc6 JL	186	printk(KERN_WARNING "nfsd: last server has exited, flushing export "
	187	"cache\n");
	188	nfsd_export_flush();
bc591ccf	189	}
6658d3a7 N	190
	191	void nfsd_reset_versions(void)
	192	{
	193	int found_one = 0;
	194	int i;
	195
	196	for (i = NFSD_MINVERS; i < NFSD_NRVERS; i++) {
	197	if (nfsd_program.pg_vers[i])
	198	found_one = 1;
	199	}
	200
	201	if (!found_one) {
	202	for (i = NFSD_MINVERS; i < NFSD_NRVERS; i++)
	203	nfsd_program.pg_vers[i] = nfsd_version[i];
	204	#if defined(CONFIG_NFSD_V2_ACL) \|\| defined(CONFIG_NFSD_V3_ACL)
	205	for (i = NFSD_ACL_MINVERS; i < NFSD_ACL_NRVERS; i++)
	206	nfsd_acl_program.pg_vers[i] =
	207	nfsd_acl_version[i];
	208	#endif
	209	}
	210	}
	211
bedbdd8b	212
b41b66d6	213	int nfsd_create_serv(void)
02a375f0 N	214	{
02a375f0 N	215	int err = 0;
bedbdd8b NB	216
bedbdd8b NB	217	WARN_ON(!mutex_is_locked(&nfsd_mutex));
02a375f0	218	if (nfsd_serv) {
9a24ab57	219	svc_get(nfsd_serv);
02a375f0 N	220	return 0;
02a375f0 N	221	}
596bbe53 N	222	if (nfsd_max_blksize == 0) {
	223	/* choose a suitable default */
	224	struct sysinfo i;
	225	si_meminfo(&i);
	226	/* Aim for 1/4096 of memory per thread
	227	* This gives 1MB on 4Gig machines
	228	* But only uses 32K on 128M machines.
	229	* Bottom out at 8K on 32M and smaller.
	230	* Of course, this is only a default.
	231	*/
	232	nfsd_max_blksize = NFSSVC_MAXBLKSIZE;
44c55600	233	i.totalram <<= PAGE_SHIFT - 12;
596bbe53 N	234	while (nfsd_max_blksize > i.totalram &&
	235	nfsd_max_blksize >= 810242)
	236	nfsd_max_blksize /= 2;
	237	}
02a375f0 N	238
02a375f0 N	239	atomic_set(&nfsd_busy, 0);
e096bbc6	240	nfsd_serv = svc_create_pooled(&nfsd_program, nfsd_max_blksize,
a75c5d01	241	nfsd_last_thread, nfsd, THIS_MODULE);
02a375f0 N	242	if (nfsd_serv == NULL)
02a375f0 N	243	err = -ENOMEM;
bedbdd8b	244
02a375f0 N	245	do_gettimeofday(&nfssvc_boot); /* record boot time */
	246	return err;
	247	}
	248
	249	static int nfsd_init_socks(int port)
	250	{
	251	int error;
	252	if (!list_empty(&nfsd_serv->sv_permsocks))
	253	return 0;
	254
02a375f0	255	error = lockd_up(IPPROTO_UDP);
4a3ae42d	256	if (error >= 0) {
d7c9f1ed	257	error = svc_create_xprt(nfsd_serv, "udp", port,
482fb94e	258	SVC_SOCK_DEFAULTS);
4a3ae42d N	259	if (error < 0)
	260	lockd_down();
	261	}
02a375f0 N	262	if (error < 0)
	263	return error;
	264
02a375f0	265	error = lockd_up(IPPROTO_TCP);
4a3ae42d	266	if (error >= 0) {
d7c9f1ed	267	error = svc_create_xprt(nfsd_serv, "tcp", port,
482fb94e	268	SVC_SOCK_DEFAULTS);
4a3ae42d N	269	if (error < 0)
	270	lockd_down();
	271	}
02a375f0 N	272	if (error < 0)
02a375f0 N	273	return error;
02a375f0 N	274	return 0;
	275	}
	276
eed2965a GB	277	int nfsd_nrpools(void)
	278	{
	279	if (nfsd_serv == NULL)
	280	return 0;
	281	else
	282	return nfsd_serv->sv_nrpools;
	283	}
	284
	285	int nfsd_get_nrthreads(int n, int *nthreads)
	286	{
	287	int i = 0;
	288
	289	if (nfsd_serv != NULL) {
	290	for (i = 0; i < nfsd_serv->sv_nrpools && i < n; i++)
	291	nthreads[i] = nfsd_serv->sv_pools[i].sp_nrthreads;
	292	}
	293
	294	return 0;
	295	}
	296
	297	int nfsd_set_nrthreads(int n, int *nthreads)
	298	{
	299	int i = 0;
	300	int tot = 0;
	301	int err = 0;
	302
bedbdd8b NB	303	WARN_ON(!mutex_is_locked(&nfsd_mutex));
bedbdd8b NB	304
eed2965a GB	305	if (nfsd_serv == NULL \|\| n <= 0)
	306	return 0;
	307
	308	if (n > nfsd_serv->sv_nrpools)
	309	n = nfsd_serv->sv_nrpools;
	310
	311	/* enforce a global maximum number of threads */
	312	tot = 0;
	313	for (i = 0; i < n; i++) {
	314	if (nthreads[i] > NFSD_MAXSERVS)
	315	nthreads[i] = NFSD_MAXSERVS;
	316	tot += nthreads[i];
	317	}
	318	if (tot > NFSD_MAXSERVS) {
	319	/* total too large: scale down requested numbers */
	320	for (i = 0; i < n && tot > 0; i++) {
	321	int new = nthreads[i] * NFSD_MAXSERVS / tot;
	322	tot -= (nthreads[i] - new);
	323	nthreads[i] = new;
	324	}
	325	for (i = 0; i < n && tot > 0; i++) {
	326	nthreads[i]--;
	327	tot--;
	328	}
	329	}
	330
	331	/*
	332	* There must always be a thread in pool 0; the admin
	333	* can't shut down NFS completely using pool_threads.
	334	*/
	335	if (nthreads[0] == 0)
	336	nthreads[0] = 1;
	337
	338	/* apply the new numbers */
eed2965a GB	339	svc_get(nfsd_serv);
	340	for (i = 0; i < n; i++) {
	341	err = svc_set_num_threads(nfsd_serv, &nfsd_serv->sv_pools[i],
	342	nthreads[i]);
	343	if (err)
	344	break;
	345	}
	346	svc_destroy(nfsd_serv);
eed2965a GB	347
	348	return err;
	349	}
	350
1da177e4 LT	351	int
	352	nfsd_svc(unsigned short port, int nrservs)
	353	{
	354	int error;
bedbdd8b NB	355
bedbdd8b NB	356	mutex_lock(&nfsd_mutex);
6658d3a7	357	dprintk("nfsd: creating service\n");
1da177e4 LT	358	error = -EINVAL;
	359	if (nrservs <= 0)
	360	nrservs = 0;
	361	if (nrservs > NFSD_MAXSERVS)
	362	nrservs = NFSD_MAXSERVS;
	363
	364	/* Readahead param cache - will no-op if it already exists */
	365	error = nfsd_racache_init(2*nrservs);
	366	if (error<0)
	367	goto out;
e8ff2a84	368	nfs4_state_start();
02a375f0 N	369
	370	nfsd_reset_versions();
	371
	372	error = nfsd_create_serv();
	373
	374	if (error)
	375	goto out;
	376	error = nfsd_init_socks(port);
	377	if (error)
	378	goto failure;
	379
eec09661	380	error = svc_set_num_threads(nfsd_serv, NULL, nrservs);
1da177e4	381	failure:
1da177e4	382	svc_destroy(nfsd_serv); /* Release server */
1da177e4	383	out:
bedbdd8b	384	mutex_unlock(&nfsd_mutex);
1da177e4 LT	385	return error;
	386	}
	387
	388	static inline void
	389	update_thread_usage(int busy_threads)
	390	{
	391	unsigned long prev_call;
	392	unsigned long diff;
	393	int decile;
	394
	395	spin_lock(&nfsd_call_lock);
	396	prev_call = nfsd_last_call;
	397	nfsd_last_call = jiffies;
	398	decile = busy_threads*10/nfsdstats.th_cnt;
	399	if (decile>0 && decile <= 10) {
	400	diff = nfsd_last_call - prev_call;
	401	if ( (nfsdstats.th_usage[decile-1] += diff) >= NFSD_USAGE_WRAP)
	402	nfsdstats.th_usage[decile-1] -= NFSD_USAGE_WRAP;
	403	if (decile == 10)
	404	nfsdstats.th_fullcnt++;
	405	}
	406	spin_unlock(&nfsd_call_lock);
	407	}
	408
	409	/*
	410	* This is the NFS server kernel thread
	411	*/
9867d76c JL	412	static int
9867d76c JL	413	nfsd(void *vrqstp)
1da177e4	414	{
9867d76c	415	struct svc_rqst rqstp = (struct svc_rqst ) vrqstp;
1da177e4	416	struct fs_struct *fsp;
1da177e4	417	sigset_t shutdown_mask, allowed_mask;
9867d76c JL	418	int err, preverr = 0;
9867d76c JL	419	unsigned int signo;
1da177e4 LT	420
1da177e4 LT	421	/* Lock module and set up kernel thread */
bedbdd8b	422	mutex_lock(&nfsd_mutex);
1da177e4	423
9867d76c	424	/* At this point, the thread shares current->fs
1da177e4 LT	425	* with the init process. We need to create files with a
	426	* umask of 0 instead of init's umask. */
	427	fsp = copy_fs_struct(current->fs);
	428	if (!fsp) {
	429	printk("Unable to start nfsd thread: out of memory\n");
	430	goto out;
	431	}
	432	exit_fs(current);
	433	current->fs = fsp;
	434	current->fs->umask = 0;
	435
	436	siginitsetinv(&shutdown_mask, SHUTDOWN_SIGS);
	437	siginitsetinv(&allowed_mask, ALLOWED_SIGS);
	438
9867d76c JL	439	/*
	440	* thread is spawned with all signals set to SIG_IGN, re-enable
	441	* the ones that matter
	442	*/
	443	for (signo = 1; signo <= _NSIG; signo++) {
	444	if (!sigismember(&shutdown_mask, signo))
	445	allow_signal(signo);
	446	}
bedbdd8b	447
1da177e4	448	nfsdstats.th_cnt++;
bedbdd8b NB	449	mutex_unlock(&nfsd_mutex);
bedbdd8b NB	450
1da177e4 LT	451	/*
	452	* We want less throttling in balance_dirty_pages() so that nfs to
	453	* localhost doesn't cause nfsd to lock up due to all the client's
	454	* dirty pages.
	455	*/
	456	current->flags \|= PF_LESS_THROTTLE;
83144186	457	set_freezable();
1da177e4 LT	458
	459	/*
	460	* The main request loop
	461	*/
	462	for (;;) {
	463	/* Block all but the shutdown signals */
	464	sigprocmask(SIG_SETMASK, &shutdown_mask, NULL);
	465
	466	/*
	467	* Find a socket with data available and call its
	468	* recvfrom routine.
	469	*/
6fb2b47f	470	while ((err = svc_recv(rqstp, 6060HZ)) == -EAGAIN)
1da177e4	471	;
9867d76c	472	if (err == -EINTR)
1da177e4	473	break;
9867d76c JL	474	else if (err < 0) {
	475	if (err != preverr) {
	476	printk(KERN_WARNING "%s: unexpected error "
	477	"from svc_recv (%d)\n", __func__, -err);
	478	preverr = err;
	479	}
	480	schedule_timeout_uninterruptible(HZ);
	481	continue;
	482	}
	483
1da177e4 LT	484	update_thread_usage(atomic_read(&nfsd_busy));
	485	atomic_inc(&nfsd_busy);
	486
	487	/* Lock the export hash tables for reading. */
	488	exp_readlock();
	489
9867d76c	490	/* Process request with signals blocked. */
1da177e4 LT	491	sigprocmask(SIG_SETMASK, &allowed_mask, NULL);
1da177e4 LT	492
6fb2b47f	493	svc_process(rqstp);
1da177e4 LT	494
	495	/* Unlock export hash tables */
	496	exp_readunlock();
	497	update_thread_usage(atomic_read(&nfsd_busy));
	498	atomic_dec(&nfsd_busy);
	499	}
	500
24e36663	501	/* Clear signals before calling svc_exit_thread() */
9e416052	502	flush_signals(current);
1da177e4	503
bedbdd8b	504	mutex_lock(&nfsd_mutex);
1da177e4 LT	505	nfsdstats.th_cnt --;
	506
	507	out:
	508	/* Release the thread */
	509	svc_exit_thread(rqstp);
	510
	511	/* Release module */
bedbdd8b	512	mutex_unlock(&nfsd_mutex);
1da177e4	513	module_put_and_exit(0);
9867d76c	514	return 0;
1da177e4 LT	515	}
1da177e4 LT	516
32c1eb0c AA	517	static __be32 map_new_errors(u32 vers, __be32 nfserr)
	518	{
	519	if (nfserr == nfserr_jukebox && vers == 2)
	520	return nfserr_dropit;
	521	if (nfserr == nfserr_wrongsec && vers < 4)
	522	return nfserr_acces;
	523	return nfserr;
	524	}
	525
1da177e4	526	int
c7afef1f	527	nfsd_dispatch(struct svc_rqst rqstp, __be32 statp)
1da177e4 LT	528	{
	529	struct svc_procedure *proc;
	530	kxdrproc_t xdr;
ad451d38 AV	531	__be32 nfserr;
ad451d38 AV	532	__be32 *nfserrp;
1da177e4 LT	533
	534	dprintk("nfsd_dispatch: vers %d proc %d\n",
	535	rqstp->rq_vers, rqstp->rq_proc);
	536	proc = rqstp->rq_procinfo;
	537
	538	/* Check whether we have this call in the cache. */
	539	switch (nfsd_cache_lookup(rqstp, proc->pc_cachetype)) {
	540	case RC_INTR:
	541	case RC_DROPIT:
	542	return 0;
	543	case RC_REPLY:
	544	return 1;
	545	case RC_DOIT:;
	546	/* do it */
	547	}
	548
	549	/* Decode arguments */
	550	xdr = proc->pc_decode;
ad451d38	551	if (xdr && !xdr(rqstp, (__be32*)rqstp->rq_arg.head[0].iov_base,
1da177e4 LT	552	rqstp->rq_argp)) {
	553	dprintk("nfsd: failed to decode arguments!\n");
	554	nfsd_cache_update(rqstp, RC_NOCACHE, NULL);
	555	*statp = rpc_garbage_args;
	556	return 1;
	557	}
	558
	559	/* need to grab the location to store the status, as
	560	* nfsv4 does some encoding while processing
	561	*/
	562	nfserrp = rqstp->rq_res.head[0].iov_base
	563	+ rqstp->rq_res.head[0].iov_len;
ad451d38	564	rqstp->rq_res.head[0].iov_len += sizeof(__be32);
1da177e4 LT	565
	566	/* Now call the procedure handler, and encode NFS status. */
	567	nfserr = proc->pc_func(rqstp, rqstp->rq_argp, rqstp->rq_resp);
32c1eb0c	568	nfserr = map_new_errors(rqstp->rq_vers, nfserr);
1da177e4	569	if (nfserr == nfserr_dropit) {
45457e09	570	dprintk("nfsd: Dropping request; may be revisited later\n");
1da177e4 LT	571	nfsd_cache_update(rqstp, RC_NOCACHE, NULL);
	572	return 0;
	573	}
	574
	575	if (rqstp->rq_proc != 0)
	576	*nfserrp++ = nfserr;
	577
	578	/* Encode result.
	579	* For NFSv2, additional info is never returned in case of an error.
	580	*/
	581	if (!(nfserr && rqstp->rq_vers == 2)) {
	582	xdr = proc->pc_encode;
	583	if (xdr && !xdr(rqstp, nfserrp,
	584	rqstp->rq_resp)) {
	585	/* Failed to encode result. Release cache entry */
	586	dprintk("nfsd: failed to encode result!\n");
	587	nfsd_cache_update(rqstp, RC_NOCACHE, NULL);
	588	*statp = rpc_system_err;
	589	return 1;
	590	}
	591	}
	592
	593	/* Store reply in cache. */
	594	nfsd_cache_update(rqstp, proc->pc_cachetype, statp + 1);
	595	return 1;
	596	}