[net-next-2.6.git] / fs / afs / volume.c

/* AFS volume management
 *
 * Copyright (C) 2002 Red Hat, Inc. All Rights Reserved.
 * Written by David Howells (dhowells@redhat.com)
 *
 * This program is free software; you can redistribute it and/or
 * modify it under the terms of the GNU General Public License
 * as published by the Free Software Foundation; either version
 * 2 of the License, or (at your option) any later version.
 */

#include <linux/kernel.h>
#include <linux/module.h>
#include <linux/init.h>
#include <linux/slab.h>
#include <linux/fs.h>
#include <linux/pagemap.h>
#include "volume.h"
#include "vnode.h"
#include "cell.h"
#include "cache.h"
#include "cmservice.h"
#include "fsclient.h"
#include "vlclient.h"
#include "internal.h"

#ifdef __KDEBUG
static const char *afs_voltypes[] = { "R/W", "R/O", "BAK" };
#endif

#ifdef AFS_CACHING_SUPPORT
static cachefs_match_val_t afs_volume_cache_match(void *target,
						  const void *entry);
static void afs_volume_cache_update(void *source, void *entry);

struct cachefs_index_def afs_volume_cache_index_def = {
	.name		= "volume",
	.data_size	= sizeof(struct afs_cache_vhash),
	.keys[0]	= { CACHEFS_INDEX_KEYS_BIN, 1 },
	.keys[1]	= { CACHEFS_INDEX_KEYS_BIN, 1 },
	.match		= afs_volume_cache_match,
	.update		= afs_volume_cache_update,
};
#endif

/*
 * lookup a volume by name
 * - this can be one of the following:
 *	"%[cell:]volume[.]"		R/W volume
 *	"#[cell:]volume[.]"		R/O or R/W volume (rwparent=0),
 *					 or R/W (rwparent=1) volume
 *	"%[cell:]volume.readonly"	R/O volume
 *	"#[cell:]volume.readonly"	R/O volume
 *	"%[cell:]volume.backup"		Backup volume
 *	"#[cell:]volume.backup"		Backup volume
 *
 * The cell name is optional, and defaults to the current cell.
 *
 * See "The Rules of Mount Point Traversal" in Chapter 5 of the AFS SysAdmin
 * Guide
 * - Rule 1: Explicit type suffix forces access of that type or nothing
 *           (no suffix, then use Rule 2 & 3)
 * - Rule 2: If parent volume is R/O, then mount R/O volume by preference, R/W
 *           if not available
 * - Rule 3: If parent volume is R/W, then only mount R/W volume unless
 *           explicitly told otherwise
 */
int afs_volume_lookup(const char *name, struct afs_cell *cell, int rwpath,
		      struct afs_volume **_volume)
{
	struct afs_vlocation *vlocation = NULL;
	struct afs_volume *volume = NULL;
	afs_voltype_t type;
	const char *cellname, *volname, *suffix;
	char srvtmask;
	int force, ret, loop, cellnamesz, volnamesz;

	_enter("%s,,%d,", name, rwpath);

	if (!name || (name[0] != '%' && name[0] != '#') || !name[1]) {
		printk("kAFS: unparsable volume name\n");
		return -EINVAL;
	}

	/* determine the type of volume we're looking for */
	force = 0;
	type = AFSVL_ROVOL;

	if (rwpath || name[0] == '%') {
		type = AFSVL_RWVOL;
		force = 1;
	}

	suffix = strrchr(name, '.');
	if (suffix) {
		if (strcmp(suffix, ".readonly") == 0) {
			type = AFSVL_ROVOL;
			force = 1;
		} else if (strcmp(suffix, ".backup") == 0) {
			type = AFSVL_BACKVOL;
			force = 1;
		} else if (suffix[1] == 0) {
		} else {
			suffix = NULL;
		}
	}

	/* split the cell and volume names */
	name++;
	volname = strchr(name, ':');
	if (volname) {
		cellname = name;
		cellnamesz = volname - name;
		volname++;
	} else {
		volname = name;
		cellname = NULL;
		cellnamesz = 0;
	}

	volnamesz = suffix ? suffix - volname : strlen(volname);

	_debug("CELL:%*.*s [%p] VOLUME:%*.*s SUFFIX:%s TYPE:%d%s",
	       cellnamesz, cellnamesz, cellname ?: "", cell,
	       volnamesz, volnamesz, volname, suffix ?: "-",
	       type,
	       force ? " FORCE" : "");

	/* lookup the cell record */
	if (cellname || !cell) {
		ret = afs_cell_lookup(cellname, cellnamesz, &cell);
		if (ret<0) {
			printk("kAFS: unable to lookup cell '%s'\n",
			       cellname ?: "");
			goto error;
		}
	} else {
		afs_get_cell(cell);
	}

	/* lookup the volume location record */
	ret = afs_vlocation_lookup(cell, volname, volnamesz, &vlocation);
	if (ret < 0)
		goto error;

	/* make the final decision on the type we want */
	ret = -ENOMEDIUM;
	if (force && !(vlocation->vldb.vidmask & (1 << type)))
		goto error;

	srvtmask = 0;
	for (loop = 0; loop < vlocation->vldb.nservers; loop++)
		srvtmask |= vlocation->vldb.srvtmask[loop];

	if (force) {
		if (!(srvtmask & (1 << type)))
			goto error;
	} else if (srvtmask & AFS_VOL_VTM_RO) {
		type = AFSVL_ROVOL;
	} else if (srvtmask & AFS_VOL_VTM_RW) {
		type = AFSVL_RWVOL;
	} else {
		goto error;
	}

	down_write(&cell->vl_sem);

	/* is the volume already active? */
	if (vlocation->vols[type]) {
		/* yes - re-use it */
		volume = vlocation->vols[type];
		afs_get_volume(volume);
		goto success;
	}

	/* create a new volume record */
	_debug("creating new volume record");

	ret = -ENOMEM;
	volume = kzalloc(sizeof(struct afs_volume), GFP_KERNEL);
	if (!volume)
		goto error_up;

	atomic_set(&volume->usage, 1);
	volume->type		= type;
	volume->type_force	= force;
	volume->cell		= cell;
	volume->vid		= vlocation->vldb.vid[type];

	init_rwsem(&volume->server_sem);

	/* look up all the applicable server records */
	for (loop = 0; loop < 8; loop++) {
		if (vlocation->vldb.srvtmask[loop] & (1 << volume->type)) {
			ret = afs_server_lookup(
				volume->cell,
				&vlocation->vldb.servers[loop],
				&volume->servers[volume->nservers]);
			if (ret < 0)
				goto error_discard;

			volume->nservers++;
		}
	}

	/* attach the cache and volume location */
#ifdef AFS_CACHING_SUPPORT
	cachefs_acquire_cookie(vlocation->cache,
			       &afs_vnode_cache_index_def,
			       volume,
			       &volume->cache);
#endif

	afs_get_vlocation(vlocation);
	volume->vlocation = vlocation;

	vlocation->vols[type] = volume;

success:
	_debug("kAFS selected %s volume %08x",
	       afs_voltypes[volume->type], volume->vid);
	*_volume = volume;
	ret = 0;

	/* clean up */
error_up:
	up_write(&cell->vl_sem);
error:
	afs_put_vlocation(vlocation);
	afs_put_cell(cell);

	_leave(" = %d (%p)", ret, volume);
	return ret;

error_discard:
	up_write(&cell->vl_sem);

	for (loop = volume->nservers - 1; loop >= 0; loop--)
		afs_put_server(volume->servers[loop]);

	kfree(volume);
	goto error;
}

/*
 * destroy a volume record
 */
void afs_put_volume(struct afs_volume *volume)
{
	struct afs_vlocation *vlocation;
	int loop;

	if (!volume)
		return;

	_enter("%p", volume);

	vlocation = volume->vlocation;

	/* sanity check */
	BUG_ON(atomic_read(&volume->usage) <= 0);

	/* to prevent a race, the decrement and the dequeue must be effectively
	 * atomic */
	down_write(&vlocation->cell->vl_sem);

	if (likely(!atomic_dec_and_test(&volume->usage))) {
		up_write(&vlocation->cell->vl_sem);
		_leave("");
		return;
	}

	vlocation->vols[volume->type] = NULL;

	up_write(&vlocation->cell->vl_sem);

	/* finish cleaning up the volume */
#ifdef AFS_CACHING_SUPPORT
	cachefs_relinquish_cookie(volume->cache, 0);
#endif
	afs_put_vlocation(vlocation);

	for (loop = volume->nservers - 1; loop >= 0; loop--)
		afs_put_server(volume->servers[loop]);

	kfree(volume);

	_leave(" [destroyed]");
}

/*
 * pick a server to use to try accessing this volume
 * - returns with an elevated usage count on the server chosen
 */
int afs_volume_pick_fileserver(struct afs_volume *volume,
			       struct afs_server **_server)
{
	struct afs_server *server;
	int ret, state, loop;

	_enter("%s", volume->vlocation->vldb.name);

	down_read(&volume->server_sem);

	/* handle the no-server case */
	if (volume->nservers == 0) {
		ret = volume->rjservers ? -ENOMEDIUM : -ESTALE;
		up_read(&volume->server_sem);
		_leave(" = %d [no servers]", ret);
		return ret;
	}

	/* basically, just search the list for the first live server and use
	 * that */
	ret = 0;
	for (loop = 0; loop < volume->nservers; loop++) {
		server = volume->servers[loop];
		state = server->fs_state;

		switch (state) {
			/* found an apparently healthy server */
		case 0:
			afs_get_server(server);
			up_read(&volume->server_sem);
			*_server = server;
			_leave(" = 0 (picked %08x)",
			       ntohl(server->addr.s_addr));
			return 0;

		case -ENETUNREACH:
			if (ret == 0)
				ret = state;
			break;

		case -EHOSTUNREACH:
			if (ret == 0 ||
			    ret == -ENETUNREACH)
				ret = state;
			break;

		case -ECONNREFUSED:
			if (ret == 0 ||
			    ret == -ENETUNREACH ||
			    ret == -EHOSTUNREACH)
				ret = state;
			break;

		default:
		case -EREMOTEIO:
			if (ret == 0 ||
			    ret == -ENETUNREACH ||
			    ret == -EHOSTUNREACH ||
			    ret == -ECONNREFUSED)
				ret = state;
			break;
		}
	}

	/* no available servers
	 * - TODO: handle the no active servers case better
	 */
	up_read(&volume->server_sem);
	_leave(" = %d", ret);
	return ret;
}

/*
 * release a server after use
 * - releases the ref on the server struct that was acquired by picking
 * - records result of using a particular server to access a volume
 * - return 0 to try again, 1 if okay or to issue error
 */
int afs_volume_release_fileserver(struct afs_volume *volume,
				  struct afs_server *server,
				  int result)
{
	unsigned loop;

	_enter("%s,%08x,%d",
	       volume->vlocation->vldb.name, ntohl(server->addr.s_addr),
	       result);

	switch (result) {
		/* success */
	case 0:
		server->fs_act_jif = jiffies;
		break;

		/* the fileserver denied all knowledge of the volume */
	case -ENOMEDIUM:
		server->fs_act_jif = jiffies;
		down_write(&volume->server_sem);

		/* first, find where the server is in the active list (if it
		 * is) */
		for (loop = 0; loop < volume->nservers; loop++)
			if (volume->servers[loop] == server)
				goto present;

		/* no longer there - may have been discarded by another op */
		goto try_next_server_upw;

	present:
		volume->nservers--;
		memmove(&volume->servers[loop],
			&volume->servers[loop + 1],
			sizeof(volume->servers[loop]) *
			(volume->nservers - loop));
		volume->servers[volume->nservers] = NULL;
		afs_put_server(server);
		volume->rjservers++;

		if (volume->nservers > 0)
			/* another server might acknowledge its existence */
			goto try_next_server_upw;

		/* handle the case where all the fileservers have rejected the
		 * volume
		 * - TODO: try asking the fileservers for volume information
		 * - TODO: contact the VL server again to see if the volume is
		 *         no longer registered
		 */
		up_write(&volume->server_sem);
		afs_put_server(server);
		_leave(" [completely rejected]");
		return 1;

		/* problem reaching the server */
	case -ENETUNREACH:
	case -EHOSTUNREACH:
	case -ECONNREFUSED:
	case -ETIMEDOUT:
	case -EREMOTEIO:
		/* mark the server as dead
		 * TODO: vary dead timeout depending on error
		 */
		spin_lock(&server->fs_lock);
		if (!server->fs_state) {
			server->fs_dead_jif = jiffies + HZ * 10;
			server->fs_state = result;
			printk("kAFS: SERVER DEAD state=%d\n", result);
		}
		spin_unlock(&server->fs_lock);
		goto try_next_server;

		/* miscellaneous error */
	default:
		server->fs_act_jif = jiffies;
	case -ENOMEM:
	case -ENONET:
		break;
	}

	/* tell the caller to accept the result */
	afs_put_server(server);
	_leave("");
	return 1;

	/* tell the caller to loop around and try the next server */
try_next_server_upw:
	up_write(&volume->server_sem);
try_next_server:
	afs_put_server(server);
	_leave(" [try next server]");
	return 0;
}

/*
 * match a volume hash record stored in the cache
 */
#ifdef AFS_CACHING_SUPPORT
static cachefs_match_val_t afs_volume_cache_match(void *target,
						  const void *entry)
{
	const struct afs_cache_vhash *vhash = entry;
	struct afs_volume *volume = target;

	_enter("{%u},{%u}", volume->type, vhash->vtype);

	if (volume->type == vhash->vtype) {
		_leave(" = SUCCESS");
		return CACHEFS_MATCH_SUCCESS;
	}

	_leave(" = FAILED");
	return CACHEFS_MATCH_FAILED;
}
#endif

/*
 * update a volume hash record stored in the cache
 */
#ifdef AFS_CACHING_SUPPORT
static void afs_volume_cache_update(void *source, void *entry)
{
	struct afs_cache_vhash *vhash = entry;
	struct afs_volume *volume = source;

	_enter("");

	vhash->vtype = volume->type;
}
#endif
Commit	Line	Data
ec26815a	1	/* AFS volume management
1da177e4 LT	2	*
	3	* Copyright (C) 2002 Red Hat, Inc. All Rights Reserved.
	4	* Written by David Howells (dhowells@redhat.com)
	5	*
	6	* This program is free software; you can redistribute it and/or
	7	* modify it under the terms of the GNU General Public License
	8	* as published by the Free Software Foundation; either version
	9	* 2 of the License, or (at your option) any later version.
	10	*/
	11
	12	#include <linux/kernel.h>
	13	#include <linux/module.h>
	14	#include <linux/init.h>
	15	#include <linux/slab.h>
	16	#include <linux/fs.h>
	17	#include <linux/pagemap.h>
	18	#include "volume.h"
	19	#include "vnode.h"
	20	#include "cell.h"
	21	#include "cache.h"
	22	#include "cmservice.h"
	23	#include "fsclient.h"
	24	#include "vlclient.h"
	25	#include "internal.h"
	26
	27	#ifdef __KDEBUG
	28	static const char *afs_voltypes[] = { "R/W", "R/O", "BAK" };
	29	#endif
	30
	31	#ifdef AFS_CACHING_SUPPORT
	32	static cachefs_match_val_t afs_volume_cache_match(void *target,
	33	const void *entry);
	34	static void afs_volume_cache_update(void source, void entry);
	35
	36	struct cachefs_index_def afs_volume_cache_index_def = {
	37	.name = "volume",
	38	.data_size = sizeof(struct afs_cache_vhash),
	39	.keys[0] = { CACHEFS_INDEX_KEYS_BIN, 1 },
	40	.keys[1] = { CACHEFS_INDEX_KEYS_BIN, 1 },
	41	.match = afs_volume_cache_match,
	42	.update = afs_volume_cache_update,
	43	};
	44	#endif
	45
1da177e4 LT	46	/*
	47	* lookup a volume by name
	48	* - this can be one of the following:
	49	* "%[cell:]volume[.]" R/W volume
	50	* "#[cell:]volume[.]" R/O or R/W volume (rwparent=0),
	51	* or R/W (rwparent=1) volume
	52	* "%[cell:]volume.readonly" R/O volume
	53	* "#[cell:]volume.readonly" R/O volume
	54	* "%[cell:]volume.backup" Backup volume
	55	* "#[cell:]volume.backup" Backup volume
	56	*
	57	* The cell name is optional, and defaults to the current cell.
	58	*
	59	* See "The Rules of Mount Point Traversal" in Chapter 5 of the AFS SysAdmin
	60	* Guide
	61	* - Rule 1: Explicit type suffix forces access of that type or nothing
	62	* (no suffix, then use Rule 2 & 3)
	63	* - Rule 2: If parent volume is R/O, then mount R/O volume by preference, R/W
	64	* if not available
	65	* - Rule 3: If parent volume is R/W, then only mount R/W volume unless
	66	* explicitly told otherwise
	67	*/
	68	int afs_volume_lookup(const char name, struct afs_cell cell, int rwpath,
	69	struct afs_volume **_volume)
	70	{
	71	struct afs_vlocation *vlocation = NULL;
	72	struct afs_volume *volume = NULL;
	73	afs_voltype_t type;
	74	const char cellname, volname, *suffix;
	75	char srvtmask;
	76	int force, ret, loop, cellnamesz, volnamesz;
	77
	78	_enter("%s,,%d,", name, rwpath);
	79
	80	if (!name \|\| (name[0] != '%' && name[0] != '#') \|\| !name[1]) {
	81	printk("kAFS: unparsable volume name\n");
	82	return -EINVAL;
	83	}
	84
	85	/* determine the type of volume we're looking for */
	86	force = 0;
	87	type = AFSVL_ROVOL;
	88
	89	if (rwpath \|\| name[0] == '%') {
	90	type = AFSVL_RWVOL;
	91	force = 1;
	92	}
	93
	94	suffix = strrchr(name, '.');
	95	if (suffix) {
	96	if (strcmp(suffix, ".readonly") == 0) {
	97	type = AFSVL_ROVOL;
	98	force = 1;
ec26815a	99	} else if (strcmp(suffix, ".backup") == 0) {
1da177e4 LT	100	type = AFSVL_BACKVOL;
1da177e4 LT	101	force = 1;
ec26815a DH	102	} else if (suffix[1] == 0) {
ec26815a DH	103	} else {
1da177e4 LT	104	suffix = NULL;
	105	}
	106	}
	107
	108	/* split the cell and volume names */
	109	name++;
	110	volname = strchr(name, ':');
	111	if (volname) {
	112	cellname = name;
	113	cellnamesz = volname - name;
	114	volname++;
ec26815a	115	} else {
1da177e4 LT	116	volname = name;
	117	cellname = NULL;
	118	cellnamesz = 0;
	119	}
	120
	121	volnamesz = suffix ? suffix - volname : strlen(volname);
	122
	123	_debug("CELL:%.s [%p] VOLUME:%.s SUFFIX:%s TYPE:%d%s",
	124	cellnamesz, cellnamesz, cellname ?: "", cell,
	125	volnamesz, volnamesz, volname, suffix ?: "-",
	126	type,
	127	force ? " FORCE" : "");
	128
	129	/* lookup the cell record */
	130	if (cellname \|\| !cell) {
	131	ret = afs_cell_lookup(cellname, cellnamesz, &cell);
	132	if (ret<0) {
	133	printk("kAFS: unable to lookup cell '%s'\n",
	134	cellname ?: "");
	135	goto error;
	136	}
ec26815a	137	} else {
1da177e4 LT	138	afs_get_cell(cell);
	139	}
	140
	141	/* lookup the volume location record */
	142	ret = afs_vlocation_lookup(cell, volname, volnamesz, &vlocation);
	143	if (ret < 0)
	144	goto error;
	145
	146	/* make the final decision on the type we want */
	147	ret = -ENOMEDIUM;
	148	if (force && !(vlocation->vldb.vidmask & (1 << type)))
	149	goto error;
	150
	151	srvtmask = 0;
	152	for (loop = 0; loop < vlocation->vldb.nservers; loop++)
	153	srvtmask \|= vlocation->vldb.srvtmask[loop];
	154
	155	if (force) {
	156	if (!(srvtmask & (1 << type)))
	157	goto error;
ec26815a	158	} else if (srvtmask & AFS_VOL_VTM_RO) {
1da177e4	159	type = AFSVL_ROVOL;
ec26815a	160	} else if (srvtmask & AFS_VOL_VTM_RW) {
1da177e4	161	type = AFSVL_RWVOL;
ec26815a	162	} else {
1da177e4 LT	163	goto error;
	164	}
	165
	166	down_write(&cell->vl_sem);
	167
	168	/* is the volume already active? */
	169	if (vlocation->vols[type]) {
	170	/* yes - re-use it */
	171	volume = vlocation->vols[type];
	172	afs_get_volume(volume);
	173	goto success;
	174	}
	175
	176	/* create a new volume record */
	177	_debug("creating new volume record");
	178
	179	ret = -ENOMEM;
f8314dc6	180	volume = kzalloc(sizeof(struct afs_volume), GFP_KERNEL);
1da177e4 LT	181	if (!volume)
	182	goto error_up;
	183
1da177e4 LT	184	atomic_set(&volume->usage, 1);
	185	volume->type = type;
	186	volume->type_force = force;
	187	volume->cell = cell;
	188	volume->vid = vlocation->vldb.vid[type];
	189
	190	init_rwsem(&volume->server_sem);
	191
	192	/* look up all the applicable server records */
	193	for (loop = 0; loop < 8; loop++) {
	194	if (vlocation->vldb.srvtmask[loop] & (1 << volume->type)) {
	195	ret = afs_server_lookup(
	196	volume->cell,
	197	&vlocation->vldb.servers[loop],
	198	&volume->servers[volume->nservers]);
	199	if (ret < 0)
	200	goto error_discard;
	201
	202	volume->nservers++;
	203	}
	204	}
	205
	206	/* attach the cache and volume location */
	207	#ifdef AFS_CACHING_SUPPORT
	208	cachefs_acquire_cookie(vlocation->cache,
	209	&afs_vnode_cache_index_def,
	210	volume,
	211	&volume->cache);
	212	#endif
	213
	214	afs_get_vlocation(vlocation);
	215	volume->vlocation = vlocation;
	216
	217	vlocation->vols[type] = volume;
	218
ec26815a	219	success:
1da177e4 LT	220	_debug("kAFS selected %s volume %08x",
	221	afs_voltypes[volume->type], volume->vid);
	222	*_volume = volume;
	223	ret = 0;
	224
	225	/* clean up */
ec26815a	226	error_up:
1da177e4	227	up_write(&cell->vl_sem);
ec26815a	228	error:
1da177e4 LT	229	afs_put_vlocation(vlocation);
	230	afs_put_cell(cell);
	231
	232	_leave(" = %d (%p)", ret, volume);
	233	return ret;
	234
ec26815a	235	error_discard:
1da177e4 LT	236	up_write(&cell->vl_sem);
	237
	238	for (loop = volume->nservers - 1; loop >= 0; loop--)
	239	afs_put_server(volume->servers[loop]);
	240
	241	kfree(volume);
	242	goto error;
ec26815a	243	}
1da177e4	244
1da177e4 LT	245	/*
	246	* destroy a volume record
	247	*/
	248	void afs_put_volume(struct afs_volume *volume)
	249	{
	250	struct afs_vlocation *vlocation;
	251	int loop;
	252
	253	if (!volume)
	254	return;
	255
	256	_enter("%p", volume);
	257
	258	vlocation = volume->vlocation;
	259
	260	/* sanity check */
	261	BUG_ON(atomic_read(&volume->usage) <= 0);
	262
	263	/* to prevent a race, the decrement and the dequeue must be effectively
	264	* atomic */
	265	down_write(&vlocation->cell->vl_sem);
	266
	267	if (likely(!atomic_dec_and_test(&volume->usage))) {
	268	up_write(&vlocation->cell->vl_sem);
	269	_leave("");
	270	return;
	271	}
	272
	273	vlocation->vols[volume->type] = NULL;
	274
	275	up_write(&vlocation->cell->vl_sem);
	276
	277	/* finish cleaning up the volume */
	278	#ifdef AFS_CACHING_SUPPORT
	279	cachefs_relinquish_cookie(volume->cache, 0);
	280	#endif
	281	afs_put_vlocation(vlocation);
	282
	283	for (loop = volume->nservers - 1; loop >= 0; loop--)
	284	afs_put_server(volume->servers[loop]);
	285
	286	kfree(volume);
	287
	288	_leave(" [destroyed]");
ec26815a	289	}
1da177e4	290
1da177e4 LT	291	/*
	292	* pick a server to use to try accessing this volume
	293	* - returns with an elevated usage count on the server chosen
	294	*/
	295	int afs_volume_pick_fileserver(struct afs_volume *volume,
	296	struct afs_server **_server)
	297	{
	298	struct afs_server *server;
	299	int ret, state, loop;
	300
	301	_enter("%s", volume->vlocation->vldb.name);
	302
	303	down_read(&volume->server_sem);
	304
	305	/* handle the no-server case */
	306	if (volume->nservers == 0) {
	307	ret = volume->rjservers ? -ENOMEDIUM : -ESTALE;
	308	up_read(&volume->server_sem);
	309	_leave(" = %d [no servers]", ret);
	310	return ret;
	311	}
	312
	313	/* basically, just search the list for the first live server and use
	314	* that */
	315	ret = 0;
	316	for (loop = 0; loop < volume->nservers; loop++) {
	317	server = volume->servers[loop];
	318	state = server->fs_state;
	319
	320	switch (state) {
	321	/* found an apparently healthy server */
	322	case 0:
	323	afs_get_server(server);
	324	up_read(&volume->server_sem);
	325	*_server = server;
	326	_leave(" = 0 (picked %08x)",
	327	ntohl(server->addr.s_addr));
	328	return 0;
	329
	330	case -ENETUNREACH:
	331	if (ret == 0)
	332	ret = state;
	333	break;
	334
	335	case -EHOSTUNREACH:
	336	if (ret == 0 \|\|
	337	ret == -ENETUNREACH)
	338	ret = state;
	339	break;
	340
	341	case -ECONNREFUSED:
	342	if (ret == 0 \|\|
	343	ret == -ENETUNREACH \|\|
	344	ret == -EHOSTUNREACH)
	345	ret = state;
	346	break;
	347
	348	default:
	349	case -EREMOTEIO:
	350	if (ret == 0 \|\|
	351	ret == -ENETUNREACH \|\|
	352	ret == -EHOSTUNREACH \|\|
	353	ret == -ECONNREFUSED)
	354	ret = state;
355	break;
356	}
357	}
358
359	/* no available servers
360	* - TODO: handle the no active servers case better
361	*/
362	up_read(&volume->server_sem);
363	_leave(" = %d", ret);
364	return ret;
ec26815a	365	}
1da177e4	366
1da177e4 LT	367	/*
	368	* release a server after use
	369	* - releases the ref on the server struct that was acquired by picking
	370	* - records result of using a particular server to access a volume
	371	* - return 0 to try again, 1 if okay or to issue error
	372	*/
	373	int afs_volume_release_fileserver(struct afs_volume *volume,
	374	struct afs_server *server,
	375	int result)
	376	{
	377	unsigned loop;
	378
	379	_enter("%s,%08x,%d",
	380	volume->vlocation->vldb.name, ntohl(server->addr.s_addr),
	381	result);
	382
	383	switch (result) {
	384	/* success */
	385	case 0:
	386	server->fs_act_jif = jiffies;
	387	break;
	388
	389	/* the fileserver denied all knowledge of the volume */
	390	case -ENOMEDIUM:
	391	server->fs_act_jif = jiffies;
	392	down_write(&volume->server_sem);
	393
	394	/* first, find where the server is in the active list (if it
	395	* is) */
	396	for (loop = 0; loop < volume->nservers; loop++)
	397	if (volume->servers[loop] == server)
	398	goto present;
	399
	400	/* no longer there - may have been discarded by another op */
	401	goto try_next_server_upw;
	402
	403	present:
	404	volume->nservers--;
	405	memmove(&volume->servers[loop],
	406	&volume->servers[loop + 1],
	407	sizeof(volume->servers[loop]) *
	408	(volume->nservers - loop));
	409	volume->servers[volume->nservers] = NULL;
	410	afs_put_server(server);
	411	volume->rjservers++;
	412
	413	if (volume->nservers > 0)
	414	/* another server might acknowledge its existence */
	415	goto try_next_server_upw;
	416
	417	/* handle the case where all the fileservers have rejected the
	418	* volume
	419	* - TODO: try asking the fileservers for volume information
	420	* - TODO: contact the VL server again to see if the volume is
	421	* no longer registered
	422	*/
	423	up_write(&volume->server_sem);
	424	afs_put_server(server);
	425	_leave(" [completely rejected]");
	426	return 1;
	427
	428	/* problem reaching the server */
	429	case -ENETUNREACH:
	430	case -EHOSTUNREACH:
431	case -ECONNREFUSED:
432	case -ETIMEDOUT:
433	case -EREMOTEIO:
434	/* mark the server as dead
435	* TODO: vary dead timeout depending on error
436	*/
437	spin_lock(&server->fs_lock);
438	if (!server->fs_state) {
439	server->fs_dead_jif = jiffies + HZ * 10;
440	server->fs_state = result;
441	printk("kAFS: SERVER DEAD state=%d\n", result);
442	}
443	spin_unlock(&server->fs_lock);
444	goto try_next_server;
445
446	/* miscellaneous error */
447	default:
448	server->fs_act_jif = jiffies;
449	case -ENOMEM:
450	case -ENONET:
451	break;
452	}
453
454	/* tell the caller to accept the result */
455	afs_put_server(server);
456	_leave("");
457	return 1;
458
459	/* tell the caller to loop around and try the next server */
ec26815a	460	try_next_server_upw:
1da177e4	461	up_write(&volume->server_sem);
ec26815a	462	try_next_server:
1da177e4 LT	463	afs_put_server(server);
	464	_leave(" [try next server]");
	465	return 0;
ec26815a	466	}
1da177e4	467
1da177e4 LT	468	/*
	469	* match a volume hash record stored in the cache
	470	*/
	471	#ifdef AFS_CACHING_SUPPORT
	472	static cachefs_match_val_t afs_volume_cache_match(void *target,
	473	const void *entry)
	474	{
	475	const struct afs_cache_vhash *vhash = entry;
	476	struct afs_volume *volume = target;
	477
	478	_enter("{%u},{%u}", volume->type, vhash->vtype);
	479
	480	if (volume->type == vhash->vtype) {
	481	_leave(" = SUCCESS");
	482	return CACHEFS_MATCH_SUCCESS;
	483	}
	484
	485	_leave(" = FAILED");
	486	return CACHEFS_MATCH_FAILED;
ec26815a	487	}
1da177e4 LT	488	#endif
1da177e4 LT	489
1da177e4 LT	490	/*
	491	* update a volume hash record stored in the cache
	492	*/
	493	#ifdef AFS_CACHING_SUPPORT
	494	static void afs_volume_cache_update(void source, void entry)
	495	{
	496	struct afs_cache_vhash *vhash = entry;
	497	struct afs_volume *volume = source;
	498
	499	_enter("");
	500
	501	vhash->vtype = volume->type;
ec26815a	502	}
1da177e4	503	#endif