[linux-2.6-block.git] / drivers / staging / lustre / lustre / llite / xattr.c

/*
 * GPL HEADER START
 *
 * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
 *
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License version 2 only,
 * as published by the Free Software Foundation.
 *
 * This program is distributed in the hope that it will be useful, but
 * WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * General Public License version 2 for more details (a copy is included
 * in the LICENSE file that accompanied this code).
 *
 * You should have received a copy of the GNU General Public License
 * version 2 along with this program; If not, see
 * http://www.sun.com/software/products/lustre/docs/GPLv2.pdf
 *
 * Please contact Sun Microsystems, Inc., 4150 Network Circle, Santa Clara,
 * CA 95054 USA or visit www.sun.com if you need additional information or
 * have any questions.
 *
 * GPL HEADER END
 */
/*
 * Copyright (c) 2007, 2010, Oracle and/or its affiliates. All rights reserved.
 * Use is subject to license terms.
 *
 * Copyright (c) 2011, 2012, Intel Corporation.
 */
/*
 * This file is part of Lustre, http://www.lustre.org/
 * Lustre is a trademark of Sun Microsystems, Inc.
 */

#include <linux/fs.h>
#include <linux/sched.h>
#include <linux/mm.h>
#include <linux/selinux.h>

#define DEBUG_SUBSYSTEM S_LLITE

#include <obd_support.h>
#include <lustre_lite.h>
#include <lustre_dlm.h>
#include <lustre_ver.h>
#include <lustre_eacl.h>

#include "llite_internal.h"

#define XATTR_USER_T	    (1)
#define XATTR_TRUSTED_T	 (2)
#define XATTR_SECURITY_T	(3)
#define XATTR_ACL_ACCESS_T      (4)
#define XATTR_ACL_DEFAULT_T     (5)
#define XATTR_LUSTRE_T	  (6)
#define XATTR_OTHER_T	   (7)

static
int get_xattr_type(const char *name)
{
	if (!strcmp(name, POSIX_ACL_XATTR_ACCESS))
		return XATTR_ACL_ACCESS_T;

	if (!strcmp(name, POSIX_ACL_XATTR_DEFAULT))
		return XATTR_ACL_DEFAULT_T;

	if (!strncmp(name, XATTR_USER_PREFIX,
		     sizeof(XATTR_USER_PREFIX) - 1))
		return XATTR_USER_T;

	if (!strncmp(name, XATTR_TRUSTED_PREFIX,
		     sizeof(XATTR_TRUSTED_PREFIX) - 1))
		return XATTR_TRUSTED_T;

	if (!strncmp(name, XATTR_SECURITY_PREFIX,
		     sizeof(XATTR_SECURITY_PREFIX) - 1))
		return XATTR_SECURITY_T;

	if (!strncmp(name, XATTR_LUSTRE_PREFIX,
		     sizeof(XATTR_LUSTRE_PREFIX) - 1))
		return XATTR_LUSTRE_T;

	return XATTR_OTHER_T;
}

static
int xattr_type_filter(struct ll_sb_info *sbi, int xattr_type)
{
	if ((xattr_type == XATTR_ACL_ACCESS_T ||
	     xattr_type == XATTR_ACL_DEFAULT_T) &&
	   !(sbi->ll_flags & LL_SBI_ACL))
		return -EOPNOTSUPP;

	if (xattr_type == XATTR_USER_T && !(sbi->ll_flags & LL_SBI_USER_XATTR))
		return -EOPNOTSUPP;
	if (xattr_type == XATTR_TRUSTED_T && !cfs_capable(CFS_CAP_SYS_ADMIN))
		return -EPERM;
	if (xattr_type == XATTR_OTHER_T)
		return -EOPNOTSUPP;

	return 0;
}

static
int ll_setxattr_common(struct inode *inode, const char *name,
		       const void *value, size_t size,
		       int flags, __u64 valid)
{
	struct ll_sb_info *sbi = ll_i2sbi(inode);
	struct ptlrpc_request *req;
	int xattr_type, rc;
	struct obd_capa *oc;
#ifdef CONFIG_FS_POSIX_ACL
	posix_acl_xattr_header *new_value = NULL;
	struct rmtacl_ctl_entry *rce = NULL;
	ext_acl_xattr_header *acl = NULL;
#endif
	const char *pv = value;

	xattr_type = get_xattr_type(name);
	rc = xattr_type_filter(sbi, xattr_type);
	if (rc)
		RETURN(rc);

	/* b10667: ignore lustre special xattr for now */
	if ((xattr_type == XATTR_TRUSTED_T && strcmp(name, "trusted.lov") == 0) ||
	    (xattr_type == XATTR_LUSTRE_T && strcmp(name, "lustre.lov") == 0))
		RETURN(0);

	/* b15587: ignore security.capability xattr for now */
	if ((xattr_type == XATTR_SECURITY_T &&
	    strcmp(name, "security.capability") == 0))
		RETURN(0);

	/* LU-549:  Disable security.selinux when selinux is disabled */
	if (xattr_type == XATTR_SECURITY_T && !selinux_is_enabled() &&
	    strcmp(name, "security.selinux") == 0)
		RETURN(-EOPNOTSUPP);

#ifdef CONFIG_FS_POSIX_ACL
	if (sbi->ll_flags & LL_SBI_RMT_CLIENT &&
	    (xattr_type == XATTR_ACL_ACCESS_T ||
	    xattr_type == XATTR_ACL_DEFAULT_T)) {
		rce = rct_search(&sbi->ll_rct, current_pid());
		if (rce == NULL ||
		    (rce->rce_ops != RMT_LSETFACL &&
		    rce->rce_ops != RMT_RSETFACL))
			RETURN(-EOPNOTSUPP);

		if (rce->rce_ops == RMT_LSETFACL) {
			struct eacl_entry *ee;

			ee = et_search_del(&sbi->ll_et, current_pid(),
					   ll_inode2fid(inode), xattr_type);
			LASSERT(ee != NULL);
			if (valid & OBD_MD_FLXATTR) {
				acl = lustre_acl_xattr_merge2ext(
						(posix_acl_xattr_header *)value,
						size, ee->ee_acl);
				if (IS_ERR(acl)) {
					ee_free(ee);
					RETURN(PTR_ERR(acl));
				}
				size =  CFS_ACL_XATTR_SIZE(\
						le32_to_cpu(acl->a_count), \
						ext_acl_xattr);
				pv = (const char *)acl;
			}
			ee_free(ee);
		} else if (rce->rce_ops == RMT_RSETFACL) {
			size = lustre_posix_acl_xattr_filter(
						(posix_acl_xattr_header *)value,
						size, &new_value);
			if (unlikely(size < 0))
				RETURN(size);

			pv = (const char *)new_value;
		} else
			RETURN(-EOPNOTSUPP);

		valid |= rce_ops2valid(rce->rce_ops);
	}
#endif
	oc = ll_mdscapa_get(inode);
	rc = md_setxattr(sbi->ll_md_exp, ll_inode2fid(inode), oc,
			 valid, name, pv, size, 0, flags, ll_i2suppgid(inode),
			 &req);
	capa_put(oc);
#ifdef CONFIG_FS_POSIX_ACL
	if (new_value != NULL)
		lustre_posix_acl_xattr_free(new_value, size);
	if (acl != NULL)
		lustre_ext_acl_xattr_free(acl);
#endif
	if (rc) {
		if (rc == -EOPNOTSUPP && xattr_type == XATTR_USER_T) {
			LCONSOLE_INFO("Disabling user_xattr feature because "
				      "it is not supported on the server\n");
			sbi->ll_flags &= ~LL_SBI_USER_XATTR;
		}
		RETURN(rc);
	}

	ptlrpc_req_finished(req);
	RETURN(0);
}

int ll_setxattr(struct dentry *dentry, const char *name,
		const void *value, size_t size, int flags)
{
	struct inode *inode = dentry->d_inode;

	LASSERT(inode);
	LASSERT(name);

	CDEBUG(D_VFSTRACE, "VFS Op:inode=%lu/%u(%p), xattr %s\n",
	       inode->i_ino, inode->i_generation, inode, name);

	ll_stats_ops_tally(ll_i2sbi(inode), LPROC_LL_SETXATTR, 1);

	if ((strncmp(name, XATTR_TRUSTED_PREFIX,
		     sizeof(XATTR_TRUSTED_PREFIX) - 1) == 0 &&
	     strcmp(name + sizeof(XATTR_TRUSTED_PREFIX) - 1, "lov") == 0) ||
	    (strncmp(name, XATTR_LUSTRE_PREFIX,
		     sizeof(XATTR_LUSTRE_PREFIX) - 1) == 0 &&
	     strcmp(name + sizeof(XATTR_LUSTRE_PREFIX) - 1, "lov") == 0)) {
		struct lov_user_md *lump = (struct lov_user_md *)value;
		int rc = 0;

		/* Attributes that are saved via getxattr will always have
		 * the stripe_offset as 0.  Instead, the MDS should be
		 * allowed to pick the starting OST index.   b=17846 */
		if (lump != NULL && lump->lmm_stripe_offset == 0)
			lump->lmm_stripe_offset = -1;

		if (lump != NULL && S_ISREG(inode->i_mode)) {
			struct file f;
			int flags = FMODE_WRITE;
			int lum_size = (lump->lmm_magic == LOV_USER_MAGIC_V1) ?
				sizeof(*lump) : sizeof(struct lov_user_md_v3);

			f.f_dentry = dentry;
			rc = ll_lov_setstripe_ea_info(inode, &f, flags, lump,
						      lum_size);
			/* b10667: rc always be 0 here for now */
			rc = 0;
		} else if (S_ISDIR(inode->i_mode)) {
			rc = ll_dir_setstripe(inode, lump, 0);
		}

		return rc;

	} else if (strcmp(name, XATTR_NAME_LMA) == 0 ||
		   strcmp(name, XATTR_NAME_LINK) == 0)
		return 0;

	return ll_setxattr_common(inode, name, value, size, flags,
				  OBD_MD_FLXATTR);
}

int ll_removexattr(struct dentry *dentry, const char *name)
{
	struct inode *inode = dentry->d_inode;

	LASSERT(inode);
	LASSERT(name);

	CDEBUG(D_VFSTRACE, "VFS Op:inode=%lu/%u(%p), xattr %s\n",
	       inode->i_ino, inode->i_generation, inode, name);

	ll_stats_ops_tally(ll_i2sbi(inode), LPROC_LL_REMOVEXATTR, 1);
	return ll_setxattr_common(inode, name, NULL, 0, 0,
				  OBD_MD_FLXATTRRM);
}

static
int ll_getxattr_common(struct inode *inode, const char *name,
		       void *buffer, size_t size, __u64 valid)
{
	struct ll_sb_info *sbi = ll_i2sbi(inode);
	struct ptlrpc_request *req = NULL;
	struct mdt_body *body;
	int xattr_type, rc;
	void *xdata;
	struct obd_capa *oc;
	struct rmtacl_ctl_entry *rce = NULL;

	CDEBUG(D_VFSTRACE, "VFS Op:inode=%lu/%u(%p)\n",
	       inode->i_ino, inode->i_generation, inode);

	/* listxattr have slightly different behavior from of ext3:
	 * without 'user_xattr' ext3 will list all xattr names but
	 * filtered out "^user..*"; we list them all for simplicity.
	 */
	if (!name) {
		xattr_type = XATTR_OTHER_T;
		goto do_getxattr;
	}

	xattr_type = get_xattr_type(name);
	rc = xattr_type_filter(sbi, xattr_type);
	if (rc)
		RETURN(rc);

	/* b15587: ignore security.capability xattr for now */
	if ((xattr_type == XATTR_SECURITY_T &&
	    strcmp(name, "security.capability") == 0))
		RETURN(-ENODATA);

	/* LU-549:  Disable security.selinux when selinux is disabled */
	if (xattr_type == XATTR_SECURITY_T && !selinux_is_enabled() &&
	    strcmp(name, "security.selinux") == 0)
		RETURN(-EOPNOTSUPP);

#ifdef CONFIG_FS_POSIX_ACL
	if (sbi->ll_flags & LL_SBI_RMT_CLIENT &&
	    (xattr_type == XATTR_ACL_ACCESS_T ||
	    xattr_type == XATTR_ACL_DEFAULT_T)) {
		rce = rct_search(&sbi->ll_rct, current_pid());
		if (rce == NULL ||
		    (rce->rce_ops != RMT_LSETFACL &&
		    rce->rce_ops != RMT_LGETFACL &&
		    rce->rce_ops != RMT_RSETFACL &&
		    rce->rce_ops != RMT_RGETFACL))
			RETURN(-EOPNOTSUPP);
	}

	/* posix acl is under protection of LOOKUP lock. when calling to this,
	 * we just have path resolution to the target inode, so we have great
	 * chance that cached ACL is uptodate.
	 */
	if (xattr_type == XATTR_ACL_ACCESS_T &&
	    !(sbi->ll_flags & LL_SBI_RMT_CLIENT)) {
		struct ll_inode_info *lli = ll_i2info(inode);
		struct posix_acl *acl;

		spin_lock(&lli->lli_lock);
		acl = posix_acl_dup(lli->lli_posix_acl);
		spin_unlock(&lli->lli_lock);

		if (!acl)
			RETURN(-ENODATA);

		rc = posix_acl_to_xattr(&init_user_ns, acl, buffer, size);
		posix_acl_release(acl);
		RETURN(rc);
	}
	if (xattr_type == XATTR_ACL_DEFAULT_T && !S_ISDIR(inode->i_mode))
		RETURN(-ENODATA);
#endif

do_getxattr:
	oc = ll_mdscapa_get(inode);
	rc = md_getxattr(sbi->ll_md_exp, ll_inode2fid(inode), oc,
			 valid | (rce ? rce_ops2valid(rce->rce_ops) : 0),
			 name, NULL, 0, size, 0, &req);
	capa_put(oc);
	if (rc) {
		if (rc == -EOPNOTSUPP && xattr_type == XATTR_USER_T) {
			LCONSOLE_INFO("Disabling user_xattr feature because "
				      "it is not supported on the server\n");
			sbi->ll_flags &= ~LL_SBI_USER_XATTR;
		}
		RETURN(rc);
	}

	body = req_capsule_server_get(&req->rq_pill, &RMF_MDT_BODY);
	LASSERT(body);

	/* only detect the xattr size */
	if (size == 0)
		GOTO(out, rc = body->eadatasize);

	if (size < body->eadatasize) {
		CERROR("server bug: replied size %u > %u\n",
		       body->eadatasize, (int)size);
		GOTO(out, rc = -ERANGE);
	}

	if (body->eadatasize == 0)
		GOTO(out, rc = -ENODATA);

	/* do not need swab xattr data */
	xdata = req_capsule_server_sized_get(&req->rq_pill, &RMF_EADATA,
					     body->eadatasize);
	if (!xdata)
		GOTO(out, rc = -EFAULT);

#ifdef CONFIG_FS_POSIX_ACL
	if (body->eadatasize >= 0 && rce && rce->rce_ops == RMT_LSETFACL) {
		ext_acl_xattr_header *acl;

		acl = lustre_posix_acl_xattr_2ext((posix_acl_xattr_header *)xdata,
						  body->eadatasize);
		if (IS_ERR(acl))
			GOTO(out, rc = PTR_ERR(acl));

		rc = ee_add(&sbi->ll_et, current_pid(), ll_inode2fid(inode),
			    xattr_type, acl);
		if (unlikely(rc < 0)) {
			lustre_ext_acl_xattr_free(acl);
			GOTO(out, rc);
		}
	}
#endif

	if (body->eadatasize == 0) {
		rc = -ENODATA;
	} else {
		LASSERT(buffer);
		memcpy(buffer, xdata, body->eadatasize);
		rc = body->eadatasize;
	}
out:
	ptlrpc_req_finished(req);
	return rc;
}

ssize_t ll_getxattr(struct dentry *dentry, const char *name,
		    void *buffer, size_t size)
{
	struct inode *inode = dentry->d_inode;

	LASSERT(inode);
	LASSERT(name);

	CDEBUG(D_VFSTRACE, "VFS Op:inode=%lu/%u(%p), xattr %s\n",
	       inode->i_ino, inode->i_generation, inode, name);

	ll_stats_ops_tally(ll_i2sbi(inode), LPROC_LL_GETXATTR, 1);

	if ((strncmp(name, XATTR_TRUSTED_PREFIX,
		     sizeof(XATTR_TRUSTED_PREFIX) - 1) == 0 &&
	     strcmp(name + sizeof(XATTR_TRUSTED_PREFIX) - 1, "lov") == 0) ||
	    (strncmp(name, XATTR_LUSTRE_PREFIX,
		     sizeof(XATTR_LUSTRE_PREFIX) - 1) == 0 &&
	     strcmp(name + sizeof(XATTR_LUSTRE_PREFIX) - 1, "lov") == 0)) {
		struct lov_stripe_md *lsm;
		struct lov_user_md *lump;
		struct lov_mds_md *lmm = NULL;
		struct ptlrpc_request *request = NULL;
		int rc = 0, lmmsize = 0;

		if (!S_ISREG(inode->i_mode) && !S_ISDIR(inode->i_mode))
			return -ENODATA;

		if (size == 0 && S_ISDIR(inode->i_mode)) {
			/* XXX directory EA is fix for now, optimize to save
			 * RPC transfer */
			GOTO(out, rc = sizeof(struct lov_user_md));
		}

		lsm = ccc_inode_lsm_get(inode);
		if (lsm == NULL) {
			if (S_ISDIR(inode->i_mode)) {
				rc = ll_dir_getstripe(inode, &lmm,
						      &lmmsize, &request);
			} else {
				rc = -ENODATA;
			}
		} else {
			/* LSM is present already after lookup/getattr call.
			 * we need to grab layout lock once it is implemented */
			rc = obd_packmd(ll_i2dtexp(inode), &lmm, lsm);
			lmmsize = rc;
		}
		ccc_inode_lsm_put(inode, lsm);

		if (rc < 0)
		       GOTO(out, rc);

		if (size == 0) {
			/* used to call ll_get_max_mdsize() forward to get
			 * the maximum buffer size, while some apps (such as
			 * rsync 3.0.x) care much about the exact xattr value
			 * size */
			rc = lmmsize;
			GOTO(out, rc);
		}

		if (size < lmmsize) {
			CERROR("server bug: replied size %d > %d for %s (%s)\n",
			       lmmsize, (int)size, dentry->d_name.name, name);
			GOTO(out, rc = -ERANGE);
		}

		lump = (struct lov_user_md *)buffer;
		memcpy(lump, lmm, lmmsize);
		/* do not return layout gen for getxattr otherwise it would
		 * confuse tar --xattr by recognizing layout gen as stripe
		 * offset when the file is restored. See LU-2809. */
		lump->lmm_layout_gen = 0;

		rc = lmmsize;
out:
		if (request)
			ptlrpc_req_finished(request);
		else if (lmm)
			obd_free_diskmd(ll_i2dtexp(inode), &lmm);
		return(rc);
	}

	return ll_getxattr_common(inode, name, buffer, size, OBD_MD_FLXATTR);
}

ssize_t ll_listxattr(struct dentry *dentry, char *buffer, size_t size)
{
	struct inode *inode = dentry->d_inode;
	int rc = 0, rc2 = 0;
	struct lov_mds_md *lmm = NULL;
	struct ptlrpc_request *request = NULL;
	int lmmsize;

	LASSERT(inode);

	CDEBUG(D_VFSTRACE, "VFS Op:inode=%lu/%u(%p)\n",
	       inode->i_ino, inode->i_generation, inode);

	ll_stats_ops_tally(ll_i2sbi(inode), LPROC_LL_LISTXATTR, 1);

	rc = ll_getxattr_common(inode, NULL, buffer, size, OBD_MD_FLXATTRLS);
	if (rc < 0)
		GOTO(out, rc);

	if (buffer != NULL) {
		struct ll_sb_info *sbi = ll_i2sbi(inode);
		char *xattr_name = buffer;
		int xlen, rem = rc;

		while (rem > 0) {
			xlen = strnlen(xattr_name, rem - 1) + 1;
			rem -= xlen;
			if (xattr_type_filter(sbi,
					get_xattr_type(xattr_name)) == 0) {
				/* skip OK xattr type
				 * leave it in buffer
				 */
				xattr_name += xlen;
				continue;
			}
			/* move up remaining xattrs in buffer
			 * removing the xattr that is not OK
			 */
			memmove(xattr_name, xattr_name + xlen, rem);
			rc -= xlen;
		}
	}
	if (S_ISREG(inode->i_mode)) {
		if (!ll_i2info(inode)->lli_has_smd)
			rc2 = -1;
	} else if (S_ISDIR(inode->i_mode)) {
		rc2 = ll_dir_getstripe(inode, &lmm, &lmmsize, &request);
	}

	if (rc2 < 0) {
		GOTO(out, rc2 = 0);
	} else if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode)) {
		const int prefix_len = sizeof(XATTR_LUSTRE_PREFIX) - 1;
		const size_t name_len   = sizeof("lov") - 1;
		const size_t total_len  = prefix_len + name_len + 1;

		if (((rc + total_len) > size) && (buffer != NULL)) {
			ptlrpc_req_finished(request);
			return -ERANGE;
		}

		if (buffer != NULL) {
			buffer += rc;
			memcpy(buffer, XATTR_LUSTRE_PREFIX, prefix_len);
			memcpy(buffer + prefix_len, "lov", name_len);
			buffer[prefix_len + name_len] = '\0';
		}
		rc2 = total_len;
	}
out:
	ptlrpc_req_finished(request);
	rc = rc + rc2;

	return rc;
}
Commit	Line	Data
d7e09d03 PT	1	/*
	2	* GPL HEADER START
	3	*
	4	* DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
	5	*
	6	* This program is free software; you can redistribute it and/or modify
	7	* it under the terms of the GNU General Public License version 2 only,
	8	* as published by the Free Software Foundation.
	9	*
	10	* This program is distributed in the hope that it will be useful, but
	11	* WITHOUT ANY WARRANTY; without even the implied warranty of
	12	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
	13	* General Public License version 2 for more details (a copy is included
	14	* in the LICENSE file that accompanied this code).
	15	*
	16	* You should have received a copy of the GNU General Public License
	17	* version 2 along with this program; If not, see
	18	* http://www.sun.com/software/products/lustre/docs/GPLv2.pdf
	19	*
	20	* Please contact Sun Microsystems, Inc., 4150 Network Circle, Santa Clara,
	21	* CA 95054 USA or visit www.sun.com if you need additional information or
	22	* have any questions.
	23	*
	24	* GPL HEADER END
	25	*/
	26	/*
	27	* Copyright (c) 2007, 2010, Oracle and/or its affiliates. All rights reserved.
	28	* Use is subject to license terms.
	29	*
	30	* Copyright (c) 2011, 2012, Intel Corporation.
	31	*/
	32	/*
	33	* This file is part of Lustre, http://www.lustre.org/
	34	* Lustre is a trademark of Sun Microsystems, Inc.
	35	*/
	36
	37	#include <linux/fs.h>
	38	#include <linux/sched.h>
	39	#include <linux/mm.h>
	40	#include <linux/selinux.h>
	41
	42	#define DEBUG_SUBSYSTEM S_LLITE
	43
	44	#include <obd_support.h>
	45	#include <lustre_lite.h>
	46	#include <lustre_dlm.h>
	47	#include <lustre_ver.h>
	48	#include <lustre_eacl.h>
	49
	50	#include "llite_internal.h"
	51
	52	#define XATTR_USER_T (1)
	53	#define XATTR_TRUSTED_T (2)
	54	#define XATTR_SECURITY_T (3)
	55	#define XATTR_ACL_ACCESS_T (4)
	56	#define XATTR_ACL_DEFAULT_T (5)
	57	#define XATTR_LUSTRE_T (6)
	58	#define XATTR_OTHER_T (7)
	59
	60	static
	61	int get_xattr_type(const char *name)
	62	{
	63	if (!strcmp(name, POSIX_ACL_XATTR_ACCESS))
	64	return XATTR_ACL_ACCESS_T;
65
66	if (!strcmp(name, POSIX_ACL_XATTR_DEFAULT))
67	return XATTR_ACL_DEFAULT_T;
68
69	if (!strncmp(name, XATTR_USER_PREFIX,
70	sizeof(XATTR_USER_PREFIX) - 1))
71	return XATTR_USER_T;
72
73	if (!strncmp(name, XATTR_TRUSTED_PREFIX,
74	sizeof(XATTR_TRUSTED_PREFIX) - 1))
75	return XATTR_TRUSTED_T;
76
77	if (!strncmp(name, XATTR_SECURITY_PREFIX,
78	sizeof(XATTR_SECURITY_PREFIX) - 1))
79	return XATTR_SECURITY_T;
80
81	if (!strncmp(name, XATTR_LUSTRE_PREFIX,
82	sizeof(XATTR_LUSTRE_PREFIX) - 1))
83	return XATTR_LUSTRE_T;
84
85	return XATTR_OTHER_T;
86	}
87
88	static
89	int xattr_type_filter(struct ll_sb_info *sbi, int xattr_type)
90	{
91	if ((xattr_type == XATTR_ACL_ACCESS_T \|\|
92	xattr_type == XATTR_ACL_DEFAULT_T) &&
93	!(sbi->ll_flags & LL_SBI_ACL))
94	return -EOPNOTSUPP;
95
96	if (xattr_type == XATTR_USER_T && !(sbi->ll_flags & LL_SBI_USER_XATTR))
97	return -EOPNOTSUPP;
98	if (xattr_type == XATTR_TRUSTED_T && !cfs_capable(CFS_CAP_SYS_ADMIN))
99	return -EPERM;
100	if (xattr_type == XATTR_OTHER_T)
101	return -EOPNOTSUPP;
102
103	return 0;
104	}
105
106	static
107	int ll_setxattr_common(struct inode inode, const char name,
108	const void *value, size_t size,
109	int flags, __u64 valid)
110	{
111	struct ll_sb_info *sbi = ll_i2sbi(inode);
112	struct ptlrpc_request *req;
113	int xattr_type, rc;
114	struct obd_capa *oc;
9147dc8d	115	#ifdef CONFIG_FS_POSIX_ACL
d7e09d03 PT	116	posix_acl_xattr_header *new_value = NULL;
	117	struct rmtacl_ctl_entry *rce = NULL;
	118	ext_acl_xattr_header *acl = NULL;
9147dc8d	119	#endif
d7e09d03	120	const char *pv = value;
d7e09d03 PT	121
	122	xattr_type = get_xattr_type(name);
	123	rc = xattr_type_filter(sbi, xattr_type);
	124	if (rc)
	125	RETURN(rc);
	126
	127	/* b10667: ignore lustre special xattr for now */
	128	if ((xattr_type == XATTR_TRUSTED_T && strcmp(name, "trusted.lov") == 0) \|\|
	129	(xattr_type == XATTR_LUSTRE_T && strcmp(name, "lustre.lov") == 0))
	130	RETURN(0);
	131
	132	/* b15587: ignore security.capability xattr for now */
	133	if ((xattr_type == XATTR_SECURITY_T &&
	134	strcmp(name, "security.capability") == 0))
	135	RETURN(0);
	136
	137	/* LU-549: Disable security.selinux when selinux is disabled */
	138	if (xattr_type == XATTR_SECURITY_T && !selinux_is_enabled() &&
	139	strcmp(name, "security.selinux") == 0)
	140	RETURN(-EOPNOTSUPP);
	141
	142	#ifdef CONFIG_FS_POSIX_ACL
	143	if (sbi->ll_flags & LL_SBI_RMT_CLIENT &&
	144	(xattr_type == XATTR_ACL_ACCESS_T \|\|
	145	xattr_type == XATTR_ACL_DEFAULT_T)) {
	146	rce = rct_search(&sbi->ll_rct, current_pid());
	147	if (rce == NULL \|\|
	148	(rce->rce_ops != RMT_LSETFACL &&
	149	rce->rce_ops != RMT_RSETFACL))
	150	RETURN(-EOPNOTSUPP);
	151
	152	if (rce->rce_ops == RMT_LSETFACL) {
	153	struct eacl_entry *ee;
	154
	155	ee = et_search_del(&sbi->ll_et, current_pid(),
	156	ll_inode2fid(inode), xattr_type);
	157	LASSERT(ee != NULL);
	158	if (valid & OBD_MD_FLXATTR) {
	159	acl = lustre_acl_xattr_merge2ext(
	160	(posix_acl_xattr_header *)value,
	161	size, ee->ee_acl);
	162	if (IS_ERR(acl)) {
	163	ee_free(ee);
	164	RETURN(PTR_ERR(acl));
	165	}
	166	size = CFS_ACL_XATTR_SIZE(\
	167	le32_to_cpu(acl->a_count), \
	168	ext_acl_xattr);
	169	pv = (const char *)acl;
	170	}
	171	ee_free(ee);
	172	} else if (rce->rce_ops == RMT_RSETFACL) {
	173	size = lustre_posix_acl_xattr_filter(
	174	(posix_acl_xattr_header *)value,
	175	size, &new_value);
	176	if (unlikely(size < 0))
	177	RETURN(size);
	178
	179	pv = (const char *)new_value;
	180	} else
	181	RETURN(-EOPNOTSUPP);
	182
	183	valid \|= rce_ops2valid(rce->rce_ops);
	184	}
185	#endif
186	oc = ll_mdscapa_get(inode);
187	rc = md_setxattr(sbi->ll_md_exp, ll_inode2fid(inode), oc,
188	valid, name, pv, size, 0, flags, ll_i2suppgid(inode),
189	&req);
190	capa_put(oc);
191	#ifdef CONFIG_FS_POSIX_ACL
192	if (new_value != NULL)
193	lustre_posix_acl_xattr_free(new_value, size);
194	if (acl != NULL)
195	lustre_ext_acl_xattr_free(acl);
196	#endif
197	if (rc) {
198	if (rc == -EOPNOTSUPP && xattr_type == XATTR_USER_T) {
199	LCONSOLE_INFO("Disabling user_xattr feature because "
200	"it is not supported on the server\n");
201	sbi->ll_flags &= ~LL_SBI_USER_XATTR;
202	}
203	RETURN(rc);
204	}
205
206	ptlrpc_req_finished(req);
207	RETURN(0);
208	}
209
210	int ll_setxattr(struct dentry dentry, const char name,
211	const void *value, size_t size, int flags)
212	{
213	struct inode *inode = dentry->d_inode;
214
215	LASSERT(inode);
216	LASSERT(name);
217
218	CDEBUG(D_VFSTRACE, "VFS Op:inode=%lu/%u(%p), xattr %s\n",
219	inode->i_ino, inode->i_generation, inode, name);
220
221	ll_stats_ops_tally(ll_i2sbi(inode), LPROC_LL_SETXATTR, 1);
222
223	if ((strncmp(name, XATTR_TRUSTED_PREFIX,
224	sizeof(XATTR_TRUSTED_PREFIX) - 1) == 0 &&
225	strcmp(name + sizeof(XATTR_TRUSTED_PREFIX) - 1, "lov") == 0) \|\|
226	(strncmp(name, XATTR_LUSTRE_PREFIX,
227	sizeof(XATTR_LUSTRE_PREFIX) - 1) == 0 &&
228	strcmp(name + sizeof(XATTR_LUSTRE_PREFIX) - 1, "lov") == 0)) {
229	struct lov_user_md lump = (struct lov_user_md )value;
230	int rc = 0;
231
232	/* Attributes that are saved via getxattr will always have
233	* the stripe_offset as 0. Instead, the MDS should be
234	* allowed to pick the starting OST index. b=17846 */
235	if (lump != NULL && lump->lmm_stripe_offset == 0)
236	lump->lmm_stripe_offset = -1;
237
238	if (lump != NULL && S_ISREG(inode->i_mode)) {
239	struct file f;
240	int flags = FMODE_WRITE;
241	int lum_size = (lump->lmm_magic == LOV_USER_MAGIC_V1) ?
242	sizeof(*lump) : sizeof(struct lov_user_md_v3);
243
244	f.f_dentry = dentry;
245	rc = ll_lov_setstripe_ea_info(inode, &f, flags, lump,
246	lum_size);
247	/* b10667: rc always be 0 here for now */
248	rc = 0;
249	} else if (S_ISDIR(inode->i_mode)) {
250	rc = ll_dir_setstripe(inode, lump, 0);
251	}
252
253	return rc;
254
255	} else if (strcmp(name, XATTR_NAME_LMA) == 0 \|\|
256	strcmp(name, XATTR_NAME_LINK) == 0)
257	return 0;
258
259	return ll_setxattr_common(inode, name, value, size, flags,
260	OBD_MD_FLXATTR);
261	}
262
263	int ll_removexattr(struct dentry dentry, const char name)
264	{
265	struct inode *inode = dentry->d_inode;
266
267	LASSERT(inode);
268	LASSERT(name);
269
270	CDEBUG(D_VFSTRACE, "VFS Op:inode=%lu/%u(%p), xattr %s\n",
271	inode->i_ino, inode->i_generation, inode, name);
272
273	ll_stats_ops_tally(ll_i2sbi(inode), LPROC_LL_REMOVEXATTR, 1);
274	return ll_setxattr_common(inode, name, NULL, 0, 0,
275	OBD_MD_FLXATTRRM);
276	}
277
278	static
279	int ll_getxattr_common(struct inode inode, const char name,
280	void *buffer, size_t size, __u64 valid)
281	{
282	struct ll_sb_info *sbi = ll_i2sbi(inode);
283	struct ptlrpc_request *req = NULL;
284	struct mdt_body *body;
285	int xattr_type, rc;
286	void *xdata;
287	struct obd_capa *oc;
288	struct rmtacl_ctl_entry *rce = NULL;
d7e09d03 PT	289
	290	CDEBUG(D_VFSTRACE, "VFS Op:inode=%lu/%u(%p)\n",
	291	inode->i_ino, inode->i_generation, inode);
	292
	293	/* listxattr have slightly different behavior from of ext3:
	294	* without 'user_xattr' ext3 will list all xattr names but
	295	* filtered out "^user..*"; we list them all for simplicity.
	296	*/
	297	if (!name) {
	298	xattr_type = XATTR_OTHER_T;
	299	goto do_getxattr;
	300	}
	301
	302	xattr_type = get_xattr_type(name);
	303	rc = xattr_type_filter(sbi, xattr_type);
	304	if (rc)
	305	RETURN(rc);
	306
	307	/* b15587: ignore security.capability xattr for now */
	308	if ((xattr_type == XATTR_SECURITY_T &&
	309	strcmp(name, "security.capability") == 0))
	310	RETURN(-ENODATA);
	311
	312	/* LU-549: Disable security.selinux when selinux is disabled */
	313	if (xattr_type == XATTR_SECURITY_T && !selinux_is_enabled() &&
	314	strcmp(name, "security.selinux") == 0)
	315	RETURN(-EOPNOTSUPP);
	316
	317	#ifdef CONFIG_FS_POSIX_ACL
	318	if (sbi->ll_flags & LL_SBI_RMT_CLIENT &&
	319	(xattr_type == XATTR_ACL_ACCESS_T \|\|
	320	xattr_type == XATTR_ACL_DEFAULT_T)) {
	321	rce = rct_search(&sbi->ll_rct, current_pid());
	322	if (rce == NULL \|\|
	323	(rce->rce_ops != RMT_LSETFACL &&
	324	rce->rce_ops != RMT_LGETFACL &&
	325	rce->rce_ops != RMT_RSETFACL &&
	326	rce->rce_ops != RMT_RGETFACL))
	327	RETURN(-EOPNOTSUPP);
	328	}
	329
	330	/* posix acl is under protection of LOOKUP lock. when calling to this,
	331	* we just have path resolution to the target inode, so we have great
	332	* chance that cached ACL is uptodate.
	333	*/
	334	if (xattr_type == XATTR_ACL_ACCESS_T &&
	335	!(sbi->ll_flags & LL_SBI_RMT_CLIENT)) {
	336	struct ll_inode_info *lli = ll_i2info(inode);
	337	struct posix_acl *acl;
	338
	339	spin_lock(&lli->lli_lock);
	340	acl = posix_acl_dup(lli->lli_posix_acl);
	341	spin_unlock(&lli->lli_lock);
	342
	343	if (!acl)
	344	RETURN(-ENODATA);
	345
	346	rc = posix_acl_to_xattr(&init_user_ns, acl, buffer, size);
	347	posix_acl_release(acl);
	348	RETURN(rc);
	349	}
	350	if (xattr_type == XATTR_ACL_DEFAULT_T && !S_ISDIR(inode->i_mode))
	351	RETURN(-ENODATA);
	352	#endif
353
354	do_getxattr:
355	oc = ll_mdscapa_get(inode);
356	rc = md_getxattr(sbi->ll_md_exp, ll_inode2fid(inode), oc,
357	valid \| (rce ? rce_ops2valid(rce->rce_ops) : 0),
358	name, NULL, 0, size, 0, &req);
359	capa_put(oc);
360	if (rc) {
361	if (rc == -EOPNOTSUPP && xattr_type == XATTR_USER_T) {
362	LCONSOLE_INFO("Disabling user_xattr feature because "
363	"it is not supported on the server\n");
364	sbi->ll_flags &= ~LL_SBI_USER_XATTR;
365	}
366	RETURN(rc);
367	}
368
369	body = req_capsule_server_get(&req->rq_pill, &RMF_MDT_BODY);
370	LASSERT(body);
371
372	/* only detect the xattr size */
373	if (size == 0)
374	GOTO(out, rc = body->eadatasize);
375
376	if (size < body->eadatasize) {
377	CERROR("server bug: replied size %u > %u\n",
378	body->eadatasize, (int)size);
379	GOTO(out, rc = -ERANGE);
380	}
381
382	if (body->eadatasize == 0)
383	GOTO(out, rc = -ENODATA);
384
385	/* do not need swab xattr data */
386	xdata = req_capsule_server_sized_get(&req->rq_pill, &RMF_EADATA,
387	body->eadatasize);
388	if (!xdata)
389	GOTO(out, rc = -EFAULT);
390
391	#ifdef CONFIG_FS_POSIX_ACL
392	if (body->eadatasize >= 0 && rce && rce->rce_ops == RMT_LSETFACL) {
393	ext_acl_xattr_header *acl;
394
395	acl = lustre_posix_acl_xattr_2ext((posix_acl_xattr_header *)xdata,
396	body->eadatasize);
397	if (IS_ERR(acl))
398	GOTO(out, rc = PTR_ERR(acl));
399
400	rc = ee_add(&sbi->ll_et, current_pid(), ll_inode2fid(inode),
401	xattr_type, acl);
402	if (unlikely(rc < 0)) {
403	lustre_ext_acl_xattr_free(acl);
404	GOTO(out, rc);
405	}
406	}
407	#endif
408
409	if (body->eadatasize == 0) {
410	rc = -ENODATA;
411	} else {
412	LASSERT(buffer);
413	memcpy(buffer, xdata, body->eadatasize);
414	rc = body->eadatasize;
415	}
d7e09d03 PT	416	out:
	417	ptlrpc_req_finished(req);
	418	return rc;
	419	}
	420
	421	ssize_t ll_getxattr(struct dentry dentry, const char name,
	422	void *buffer, size_t size)
	423	{
	424	struct inode *inode = dentry->d_inode;
	425
	426	LASSERT(inode);
	427	LASSERT(name);
	428
	429	CDEBUG(D_VFSTRACE, "VFS Op:inode=%lu/%u(%p), xattr %s\n",
	430	inode->i_ino, inode->i_generation, inode, name);
	431
	432	ll_stats_ops_tally(ll_i2sbi(inode), LPROC_LL_GETXATTR, 1);
	433
	434	if ((strncmp(name, XATTR_TRUSTED_PREFIX,
	435	sizeof(XATTR_TRUSTED_PREFIX) - 1) == 0 &&
	436	strcmp(name + sizeof(XATTR_TRUSTED_PREFIX) - 1, "lov") == 0) \|\|
	437	(strncmp(name, XATTR_LUSTRE_PREFIX,
	438	sizeof(XATTR_LUSTRE_PREFIX) - 1) == 0 &&
	439	strcmp(name + sizeof(XATTR_LUSTRE_PREFIX) - 1, "lov") == 0)) {
	440	struct lov_stripe_md *lsm;
	441	struct lov_user_md *lump;
	442	struct lov_mds_md *lmm = NULL;
	443	struct ptlrpc_request *request = NULL;
	444	int rc = 0, lmmsize = 0;
	445
	446	if (!S_ISREG(inode->i_mode) && !S_ISDIR(inode->i_mode))
	447	return -ENODATA;
	448
	449	if (size == 0 && S_ISDIR(inode->i_mode)) {
	450	/* XXX directory EA is fix for now, optimize to save
	451	* RPC transfer */
	452	GOTO(out, rc = sizeof(struct lov_user_md));
	453	}
	454
	455	lsm = ccc_inode_lsm_get(inode);
	456	if (lsm == NULL) {
	457	if (S_ISDIR(inode->i_mode)) {
	458	rc = ll_dir_getstripe(inode, &lmm,
	459	&lmmsize, &request);
	460	} else {
	461	rc = -ENODATA;
	462	}
	463	} else {
	464	/* LSM is present already after lookup/getattr call.
	465	* we need to grab layout lock once it is implemented */
	466	rc = obd_packmd(ll_i2dtexp(inode), &lmm, lsm);
	467	lmmsize = rc;
	468	}
	469	ccc_inode_lsm_put(inode, lsm);
	470
	471	if (rc < 0)
	472	GOTO(out, rc);
	473
	474	if (size == 0) {
	475	/* used to call ll_get_max_mdsize() forward to get
	476	* the maximum buffer size, while some apps (such as
	477	* rsync 3.0.x) care much about the exact xattr value
	478	* size */
	479	rc = lmmsize;
480	GOTO(out, rc);
481	}
482
483	if (size < lmmsize) {
484	CERROR("server bug: replied size %d > %d for %s (%s)\n",
485	lmmsize, (int)size, dentry->d_name.name, name);
486	GOTO(out, rc = -ERANGE);
487	}
488
489	lump = (struct lov_user_md *)buffer;
490	memcpy(lump, lmm, lmmsize);
491	/* do not return layout gen for getxattr otherwise it would
492	* confuse tar --xattr by recognizing layout gen as stripe
493	* offset when the file is restored. See LU-2809. */
494	lump->lmm_layout_gen = 0;
495
496	rc = lmmsize;
497	out:
498	if (request)
499	ptlrpc_req_finished(request);
500	else if (lmm)
501	obd_free_diskmd(ll_i2dtexp(inode), &lmm);
502	return(rc);
503	}
504
505	return ll_getxattr_common(inode, name, buffer, size, OBD_MD_FLXATTR);
506	}
507
508	ssize_t ll_listxattr(struct dentry dentry, char buffer, size_t size)
509	{
510	struct inode *inode = dentry->d_inode;
511	int rc = 0, rc2 = 0;
512	struct lov_mds_md *lmm = NULL;
513	struct ptlrpc_request *request = NULL;
514	int lmmsize;
515
516	LASSERT(inode);
517
518	CDEBUG(D_VFSTRACE, "VFS Op:inode=%lu/%u(%p)\n",
519	inode->i_ino, inode->i_generation, inode);
520
521	ll_stats_ops_tally(ll_i2sbi(inode), LPROC_LL_LISTXATTR, 1);
522
523	rc = ll_getxattr_common(inode, NULL, buffer, size, OBD_MD_FLXATTRLS);
524	if (rc < 0)
525	GOTO(out, rc);
526
527	if (buffer != NULL) {
528	struct ll_sb_info *sbi = ll_i2sbi(inode);
529	char *xattr_name = buffer;
530	int xlen, rem = rc;
531
532	while (rem > 0) {
533	xlen = strnlen(xattr_name, rem - 1) + 1;
534	rem -= xlen;
535	if (xattr_type_filter(sbi,
536	get_xattr_type(xattr_name)) == 0) {
537	/* skip OK xattr type
538	* leave it in buffer
539	*/
540	xattr_name += xlen;
541	continue;
542	}
543	/* move up remaining xattrs in buffer
544	* removing the xattr that is not OK
545	*/
546	memmove(xattr_name, xattr_name + xlen, rem);
547	rc -= xlen;
548	}
549	}
550	if (S_ISREG(inode->i_mode)) {
551	if (!ll_i2info(inode)->lli_has_smd)
552	rc2 = -1;
553	} else if (S_ISDIR(inode->i_mode)) {
554	rc2 = ll_dir_getstripe(inode, &lmm, &lmmsize, &request);
555	}
556
557	if (rc2 < 0) {
558	GOTO(out, rc2 = 0);
559	} else if (S_ISREG(inode->i_mode) \|\| S_ISDIR(inode->i_mode)) {
560	const int prefix_len = sizeof(XATTR_LUSTRE_PREFIX) - 1;
561	const size_t name_len = sizeof("lov") - 1;
562	const size_t total_len = prefix_len + name_len + 1;
563
cb6a2db6 KM	564	if (((rc + total_len) > size) && (buffer != NULL)) {
	565	ptlrpc_req_finished(request);
	566	return -ERANGE;
	567	}
	568
	569	if (buffer != NULL) {
d7e09d03 PT	570	buffer += rc;
	571	memcpy(buffer, XATTR_LUSTRE_PREFIX, prefix_len);
	572	memcpy(buffer + prefix_len, "lov", name_len);
	573	buffer[prefix_len + name_len] = '\0';
	574	}
	575	rc2 = total_len;
	576	}
	577	out:
	578	ptlrpc_req_finished(request);
	579	rc = rc + rc2;
	580
	581	return rc;
	582	}