* Copyright (c) 2002, 2010, Oracle and/or its affiliates. All rights reserved.
* Use is subject to license terms.
*
- * Copyright (c) 2011, 2013, Intel Corporation.
+ * Copyright (c) 2011, 2015, Intel Corporation.
*/
/*
* This file is part of Lustre, http://www.lustre.org/
#include <linux/highmem.h>
#include <linux/pagemap.h>
#include <linux/security.h>
+#include <linux/user_namespace.h>
+#ifdef HAVE_UIDGID_HEADER
+# include <linux/uidgid.h>
+#endif
#define DEBUG_SUBSYSTEM S_LLITE
#include <obd_support.h>
#include <lustre_fid.h>
-#include <lustre_lite.h>
#include <lustre_dlm.h>
#include <lustre_ver.h>
#include "llite_internal.h"
-static int ll_create_it(struct inode *, struct dentry *,
- int, struct lookup_intent *);
-
-/*
- * Check if we have something mounted at the named dchild.
- * In such a case there would always be dentry present.
- */
-static int ll_d_mountpoint(struct dentry *dparent, struct dentry *dchild,
- struct qstr *name)
-{
- int mounted = 0;
-
- if (unlikely(dchild)) {
- mounted = d_mountpoint(dchild);
- } else if (dparent) {
- dchild = d_lookup(dparent, name);
- if (dchild) {
- mounted = d_mountpoint(dchild);
- dput(dchild);
- }
- }
- return mounted;
-}
+static int ll_create_it(struct inode *dir, struct dentry *dentry,
+ struct lookup_intent *it);
/* called from iget5_locked->find_inode() under inode_lock spinlock */
static int ll_test_inode(struct inode *inode, void *opaque)
{
- struct ll_inode_info *lli = ll_i2info(inode);
- struct lustre_md *md = opaque;
+ struct ll_inode_info *lli = ll_i2info(inode);
+ struct lustre_md *md = opaque;
- if (unlikely(!(md->body->valid & OBD_MD_FLID))) {
- CERROR("MDS body missing FID\n");
- return 0;
- }
+ if (unlikely(!(md->body->mbo_valid & OBD_MD_FLID))) {
+ CERROR("MDS body missing FID\n");
+ return 0;
+ }
- if (!lu_fid_eq(&lli->lli_fid, &md->body->fid1))
- return 0;
+ if (!lu_fid_eq(&lli->lli_fid, &md->body->mbo_fid1))
+ return 0;
- return 1;
+ return 1;
}
static int ll_set_inode(struct inode *inode, void *opaque)
{
- struct ll_inode_info *lli = ll_i2info(inode);
- struct mdt_body *body = ((struct lustre_md *)opaque)->body;
+ struct ll_inode_info *lli = ll_i2info(inode);
+ struct mdt_body *body = ((struct lustre_md *)opaque)->body;
- if (unlikely(!(body->valid & OBD_MD_FLID))) {
- CERROR("MDS body missing FID\n");
- return -EINVAL;
- }
+ if (unlikely(!(body->mbo_valid & OBD_MD_FLID))) {
+ CERROR("MDS body missing FID\n");
+ return -EINVAL;
+ }
- lli->lli_fid = body->fid1;
- if (unlikely(!(body->valid & OBD_MD_FLTYPE))) {
- CERROR("Can not initialize inode "DFID" without object type: "
- "valid = "LPX64"\n", PFID(&lli->lli_fid), body->valid);
- return -EINVAL;
- }
+ lli->lli_fid = body->mbo_fid1;
+ if (unlikely(!(body->mbo_valid & OBD_MD_FLTYPE))) {
+ CERROR("Can not initialize inode "DFID" without object type: "
+ "valid = %#llx\n",
+ PFID(&lli->lli_fid), body->mbo_valid);
+ return -EINVAL;
+ }
- inode->i_mode = (inode->i_mode & ~S_IFMT) | (body->mode & S_IFMT);
- if (unlikely(inode->i_mode == 0)) {
- CERROR("Invalid inode "DFID" type\n", PFID(&lli->lli_fid));
- return -EINVAL;
- }
+ inode->i_mode = (inode->i_mode & ~S_IFMT) | (body->mbo_mode & S_IFMT);
+ if (unlikely(inode->i_mode == 0)) {
+ CERROR("Invalid inode "DFID" type\n", PFID(&lli->lli_fid));
+ return -EINVAL;
+ }
- ll_lli_init(lli);
+ ll_lli_init(lli);
- return 0;
+ return 0;
}
if (inode->i_state & I_NEW) {
rc = ll_read_inode2(inode, md);
if (rc == 0 && S_ISREG(inode->i_mode) &&
- ll_i2info(inode)->lli_clob == NULL) {
- CDEBUG(D_INODE, "%s: apply lsm %p to inode "DFID"\n",
- ll_get_fsname(sb, NULL, 0), md->lsm,
- PFID(ll_inode2fid(inode)));
+ ll_i2info(inode)->lli_clob == NULL)
rc = cl_file_inode_init(inode, md);
- }
+
if (rc != 0) {
+ /* Let's clear directory lsm here, otherwise
+ * make_bad_inode() will reset the inode mode
+ * to regular, then ll_clear_inode will not
+ * be able to clear lsm_md */
+ if (S_ISDIR(inode->i_mode))
+ ll_dir_clear_lsm_md(inode);
make_bad_inode(inode);
unlock_new_inode(inode);
iput(inode);
} else if (!(inode->i_state & (I_FREEING | I_CLEAR))) {
rc = ll_update_inode(inode, md);
CDEBUG(D_VFSTRACE, "got inode: "DFID"(%p): rc = %d\n",
- PFID(&md->body->fid1), inode, rc);
+ PFID(&md->body->mbo_fid1), inode, rc);
if (rc != 0) {
- make_bad_inode(inode);
+ if (S_ISDIR(inode->i_mode))
+ ll_dir_clear_lsm_md(inode);
iput(inode);
inode = ERR_PTR(rc);
}
DECLARE_LL_D_HLIST_NODE_PTR(p);
ll_lock_dcache(dir);
- ll_d_hlist_for_each_entry(dentry, p, &dir->i_dentry, d_alias) {
+ ll_d_hlist_for_each_entry(dentry, p, &dir->i_dentry) {
spin_lock(&dentry->d_lock);
if (!list_empty(&dentry->d_subdirs)) {
struct dentry *child;
list_for_each_entry_safe(child, tmp_subdir,
&dentry->d_subdirs,
- d_u.d_child) {
+ d_child) {
if (child->d_inode == NULL)
d_lustre_invalidate(child, 1);
}
}
if (bits & MDS_INODELOCK_XATTR) {
+ if (S_ISDIR(inode->i_mode))
+ ll_i2info(inode)->lli_def_stripe_offset = -1;
ll_xattr_cache_destroy(inode);
bits &= ~MDS_INODELOCK_XATTR;
}
struct ll_inode_info *lli = ll_i2info(inode);
spin_lock(&lli->lli_lock);
- lli->lli_flags &= ~LLIF_MDS_SIZE_LOCK;
+ LTIME_S(inode->i_mtime) = 0;
+ LTIME_S(inode->i_atime) = 0;
+ LTIME_S(inode->i_ctime) = 0;
spin_unlock(&lli->lli_lock);
}
hash = cl_fid_build_ino(&lli->lli_pfid,
ll_need_32bit_api(ll_i2sbi(inode)));
- master_inode = ilookup5(inode->i_sb, hash,
- ll_test_inode_by_fid,
+ /* Do not lookup the inode with ilookup5,
+ * otherwise it will cause dead lock,
+ *
+ * 1. Client1 send chmod req to the MDT0, then
+ * on MDT0, it enqueues master and all of its
+ * slaves lock, (mdt_attr_set() ->
+ * mdt_lock_slaves()), after gets master and
+ * stripe0 lock, it will send the enqueue req
+ * (for stripe1) to MDT1, then MDT1 finds the
+ * lock has been granted to client2. Then MDT1
+ * sends blocking ast to client2.
+ *
+ * 2. At the same time, client2 tries to unlink
+ * the striped dir (rm -rf striped_dir), and
+ * during lookup, it will hold the master inode
+ * of the striped directory, whose inode state
+ * is NEW, then tries to revalidate all of its
+ * slaves, (ll_prep_inode()->ll_iget()->
+ * ll_read_inode2()-> ll_update_inode().). And
+ * it will be blocked on the server side because
+ * of 1.
+ *
+ * 3. Then the client get the blocking_ast req,
+ * cancel the lock, but being blocked if using
+ * ->ilookup5()), because master inode state is
+ * NEW. */
+ master_inode = ilookup5_nowait(inode->i_sb,
+ hash, ll_test_inode_by_fid,
(void *)&lli->lli_pfid);
if (master_inode != NULL &&
!IS_ERR(master_inode)) {
discon_alias = invalid_alias = NULL;
ll_lock_dcache(inode);
- ll_d_hlist_for_each_entry(alias, p, &inode->i_dentry, d_alias) {
+ ll_d_hlist_for_each_entry(alias, p, &inode->i_dentry) {
LASSERT(alias != dentry);
spin_lock(&alias->d_lock);
- if (alias->d_flags & DCACHE_DISCONNECTED)
+ if ((alias->d_flags & DCACHE_DISCONNECTED) &&
+ S_ISDIR(inode->i_mode))
/* LASSERT(last_discon == NULL); LU-405, bz 20055 */
discon_alias = alias;
else if (alias->d_parent == dentry->d_parent &&
iput(inode);
CDEBUG(D_DENTRY,
"Reuse dentry %p inode %p refc %d flags %#x\n",
- new, new->d_inode, d_count(new), new->d_flags);
+ new, new->d_inode, ll_d_count(new), new->d_flags);
return new;
}
}
return ERR_PTR(rc);
d_add(de, inode);
CDEBUG(D_DENTRY, "Add dentry %p inode %p refc %d flags %#x\n",
- de, de->d_inode, d_count(de), de->d_flags);
+ de, de->d_inode, ll_d_count(de), de->d_flags);
return de;
}
struct inode *inode = NULL;
__u64 bits = 0;
int rc;
+ struct dentry *alias;
ENTRY;
/* NB 1 request reference will be taken away by ll_intent_lock()
* when I return */
CDEBUG(D_DENTRY, "it %p it_disposition %x\n", it,
- it->d.lustre.it_disposition);
+ it->it_disposition);
if (!it_disposition(it, DISP_LOOKUP_NEG)) {
rc = ll_prep_inode(&inode, request, (*de)->d_sb, it);
if (rc)
/* Only hash *de if it is unhashed (new dentry).
* Atoimc_open may passin hashed dentries for open.
*/
- if (d_unhashed(*de)) {
- struct dentry *alias;
-
- alias = ll_splice_alias(inode, *de);
- if (IS_ERR(alias))
- RETURN(PTR_ERR(alias));
- *de = alias;
- } else if (!it_disposition(it, DISP_LOOKUP_NEG) &&
- !it_disposition(it, DISP_OPEN_CREATE)) {
- /* With DISP_OPEN_CREATE dentry will
- instantiated in ll_create_it. */
- LASSERT((*de)->d_inode == NULL);
- d_instantiate(*de, inode);
- }
+ alias = ll_splice_alias(inode, *de);
+ if (IS_ERR(alias))
+ GOTO(out, rc = PTR_ERR(alias));
+
+ *de = alias;
if (!it_disposition(it, DISP_LOOKUP_NEG)) {
/* we have lookup look - unhide dentry */
/* Check that parent has UPDATE lock. */
struct lookup_intent parent_it = {
.it_op = IT_GETATTR,
- .d.lustre.it_lock_handle = 0 };
+ .it_lock_handle = 0 };
struct lu_fid fid = ll_i2info(parent)->lli_fid;
/* If it is striped directory, get the real stripe parent */
(*de)->d_name.name,
(*de)->d_name.len, &fid);
if (rc != 0)
- RETURN(rc);
+ GOTO(out, rc);
}
if (md_revalidate_lock(ll_i2mdexp(parent), &parent_it, &fid,
d_lustre_revalidate(*de);
ll_intent_release(&parent_it);
}
- }
+ }
+
+ GOTO(out, rc = 0);
+
+out:
+ if (rc != 0 && it->it_op & IT_OPEN)
+ ll_open_cleanup((*de)->d_sb, request);
- RETURN(0);
+ return rc;
}
static struct dentry *ll_lookup_it(struct inode *parent, struct dentry *dentry,
- struct lookup_intent *it, int lookup_flags)
+ struct lookup_intent *it)
{
- struct lookup_intent lookup_it = { .it_op = IT_LOOKUP };
- struct dentry *save = dentry, *retval;
- struct ptlrpc_request *req = NULL;
- struct md_op_data *op_data;
+ struct lookup_intent lookup_it = { .it_op = IT_LOOKUP };
+ struct dentry *save = dentry, *retval;
+ struct ptlrpc_request *req = NULL;
+ struct md_op_data *op_data = NULL;
__u32 opc;
int rc;
ENTRY;
if (it == NULL || it->it_op == IT_GETXATTR)
it = &lookup_it;
- if (it->it_op == IT_GETATTR) {
- rc = ll_statahead_enter(parent, &dentry, 0);
- if (rc == 1) {
- if (dentry == save)
- GOTO(out, retval = NULL);
- GOTO(out, retval = dentry);
- }
- }
+ if (it->it_op == IT_GETATTR && dentry_may_statahead(parent, dentry)) {
+ rc = ll_statahead(parent, &dentry, 0);
+ if (rc == 1)
+ RETURN(dentry == save ? NULL : dentry);
+ }
+
+ if (it->it_op & IT_OPEN && it->it_flags & FMODE_WRITE &&
+ dentry->d_sb->s_flags & MS_RDONLY)
+ RETURN(ERR_PTR(-EROFS));
if (it->it_op & IT_CREAT)
opc = LUSTRE_OPC_CREATE;
else
opc = LUSTRE_OPC_ANY;
- op_data = ll_prep_md_op_data(NULL, parent, NULL, dentry->d_name.name,
- dentry->d_name.len, lookup_flags, opc,
- NULL);
- if (IS_ERR(op_data))
- RETURN((void *)op_data);
+ op_data = ll_prep_md_op_data(NULL, parent, NULL, dentry->d_name.name,
+ dentry->d_name.len, 0, opc, NULL);
+ if (IS_ERR(op_data))
+ GOTO(out, retval = ERR_CAST(op_data));
- /* enforce umask if acl disabled or MDS doesn't support umask */
- if (!IS_POSIXACL(parent) || !exp_connect_umask(ll_i2mdexp(parent)))
+ /* enforce umask if acl disabled or MDS doesn't support umask */
+ if (!IS_POSIXACL(parent) || !exp_connect_umask(ll_i2mdexp(parent)))
it->it_create_mode &= ~current_umask();
+ if (it->it_op & IT_CREAT &&
+ ll_i2sbi(parent)->ll_flags & LL_SBI_FILE_SECCTX) {
+ rc = ll_dentry_init_security(dentry, it->it_create_mode,
+ &dentry->d_name,
+ &op_data->op_file_secctx_name,
+ &op_data->op_file_secctx,
+ &op_data->op_file_secctx_size);
+ if (rc < 0)
+ GOTO(out, retval = ERR_PTR(rc));
+ }
+
rc = md_intent_lock(ll_i2mdexp(parent), op_data, it, &req,
&ll_md_blocking_ast, 0);
- ll_finish_md_op_data(op_data);
- if (rc < 0)
- GOTO(out, retval = ERR_PTR(rc));
+ /* If the MDS allows the client to chgrp (CFS_SETGRP_PERM), but the
+ * client does not know which suppgid should be sent to the MDS, or
+ * some other(s) changed the target file's GID after this RPC sent
+ * to the MDS with the suppgid as the original GID, then we should
+ * try again with right suppgid. */
+ if (rc == -EACCES && it->it_op & IT_OPEN &&
+ it_disposition(it, DISP_OPEN_DENY)) {
+ struct mdt_body *body;
+
+ LASSERT(req != NULL);
+
+ body = req_capsule_server_get(&req->rq_pill, &RMF_MDT_BODY);
+ if (op_data->op_suppgids[0] == body->mbo_gid ||
+ op_data->op_suppgids[1] == body->mbo_gid ||
+ !in_group_p(make_kgid(&init_user_ns, body->mbo_gid)))
+ GOTO(out, retval = ERR_PTR(-EACCES));
+
+ fid_zero(&op_data->op_fid2);
+ op_data->op_suppgids[1] = body->mbo_gid;
+ ptlrpc_req_finished(req);
+ req = NULL;
+ ll_intent_release(it);
+ rc = md_intent_lock(ll_i2mdexp(parent), op_data, it, &req,
+ &ll_md_blocking_ast, 0);
+ }
+
+ if (rc < 0)
+ GOTO(out, retval = ERR_PTR(rc));
rc = ll_lookup_it_finish(req, it, parent, &dentry);
if (rc != 0) {
}
ll_lookup_finish_locks(it, dentry);
- if (dentry == save)
- GOTO(out, retval = NULL);
- else
- GOTO(out, retval = dentry);
- out:
- if (req)
- ptlrpc_req_finished(req);
- if (it->it_op == IT_GETATTR && (retval == NULL || retval == dentry))
- ll_statahead_mark(parent, dentry);
- return retval;
+ GOTO(out, retval = (dentry == save) ? NULL : dentry);
+
+out:
+ if (op_data != NULL && !IS_ERR(op_data))
+ ll_finish_md_op_data(op_data);
+
+ ptlrpc_req_finished(req);
+ return retval;
}
#ifdef HAVE_IOP_ATOMIC_OPEN
dentry->d_name.len, dentry->d_name.name,
PFID(ll_inode2fid(parent)), parent, flags);
- /* Optimize away (CREATE && !OPEN). Let .create handle the race. */
- if ((flags & LOOKUP_CREATE) && !(flags & LOOKUP_OPEN))
+ /*
+ * Optimize away (CREATE && !OPEN). Let .create handle the race.
+ * but only if we have write permissions there, otherwise we need
+ * to proceed with lookup. LU-4185
+ */
+ if ((flags & LOOKUP_CREATE) && !(flags & LOOKUP_OPEN) &&
+ (inode_permission(parent, MAY_WRITE | MAY_EXEC) == 0))
return NULL;
if (flags & (LOOKUP_PARENT|LOOKUP_OPEN|LOOKUP_CREATE))
itp = NULL;
else
itp = ⁢
- de = ll_lookup_it(parent, dentry, itp, 0);
+ de = ll_lookup_it(parent, dentry, itp);
if (itp != NULL)
ll_intent_release(itp);
dentry->d_name.len, dentry->d_name.name,
PFID(ll_inode2fid(dir)), dir, file, open_flags, mode, *opened);
+ /* Only negative dentries enter here */
+ LASSERT(dentry->d_inode == NULL);
+
+ if (!d_unhashed(dentry)) {
+ /* A valid negative dentry that just passed revalidation,
+ * there's little point to try and open it server-side,
+ * even though there's a minuscule chance it might succeed.
+ * Either way it's a valid race to just return -ENOENT here.
+ */
+ if (!(open_flags & O_CREAT))
+ return -ENOENT;
+
+ /* Otherwise we just unhash it to be rehashed afresh via
+ * lookup if necessary
+ */
+ d_drop(dentry);
+ }
+
OBD_ALLOC(it, sizeof(*it));
if (!it)
RETURN(-ENOMEM);
}
it->it_create_mode = (mode & S_IALLUGO) | S_IFREG;
it->it_flags = (open_flags & ~O_ACCMODE) | OPEN_FMODE(open_flags);
+ it->it_flags &= ~MDS_OPEN_FL_INTERNAL;
/* Dentry added to dcache tree in ll_lookup_it */
- de = ll_lookup_it(dir, dentry, it, lookup_flags);
+ de = ll_lookup_it(dir, dentry, it);
if (IS_ERR(de))
rc = PTR_ERR(de);
else if (de != NULL)
dentry = de;
+ CFS_FAIL_TIMEOUT(OBD_FAIL_LLITE_CREATE_FILE_PAUSE, cfs_fail_val);
+
if (!rc) {
if (it_disposition(it, DISP_OPEN_CREATE)) {
/* Dentry instantiated in ll_create_it. */
- rc = ll_create_it(dir, dentry, mode, it);
+ rc = ll_create_it(dir, dentry, it);
if (rc) {
/* We dget in ll_splice_alias. */
if (de != NULL)
if (dentry->d_inode && it_disposition(it, DISP_OPEN_OPEN)) {
/* Open dentry. */
if (S_ISFIFO(dentry->d_inode->i_mode)) {
- /* We cannot call open here as it would
- * deadlock.
- */
- if (it_disposition(it, DISP_ENQ_OPEN_REF))
- ptlrpc_req_finished(
- (struct ptlrpc_request *)
- it->d.lustre.it_data);
+ /* We cannot call open here as it might
+ * deadlock. This case is unreachable in
+ * practice because of OBD_CONNECT_NODEVOH. */
rc = finish_no_open(file, de);
} else {
file->private_data = it;
{
struct lookup_intent *it;
- OBD_ALLOC(it, sizeof(*it));
+ OBD_ALLOC_PTR(it);
if (!it)
return ERR_PTR(-ENOMEM);
it->it_op |= IT_CREAT;
it->it_create_mode = (oit->create_mode & S_IALLUGO) | S_IFREG;
it->it_flags = ll_namei_to_lookup_intent_flag(oit->flags);
+ it->it_flags &= ~MDS_OPEN_FL_INTERNAL;
} else {
it->it_op = IT_GETATTR;
}
it = ll_d2d(dentry)->lld_it;
ll_d2d(dentry)->lld_it = NULL;
} else {
+ /*
+ * Optimize away (CREATE && !OPEN). Let .create handle
+ * the race. But only if we have write permissions
+ * there, otherwise we need to proceed with lookup.
+ * LU-4185
+ */
if ((nd->flags & LOOKUP_CREATE) &&
- !(nd->flags & LOOKUP_OPEN))
- RETURN(NULL);
+ !(nd->flags & LOOKUP_OPEN) &&
+ (inode_permission(parent,
+ MAY_WRITE | MAY_EXEC) == 0))
+ RETURN(NULL);
it = ll_convert_intent(&nd->intent.open, nd->flags);
if (IS_ERR(it))
RETURN((struct dentry *)it);
}
- de = ll_lookup_it(parent, dentry, it, nd->flags);
+ de = ll_lookup_it(parent, dentry, it);
if (de)
dentry = de;
if ((nd->flags & LOOKUP_OPEN) && !IS_ERR(dentry)) { /* Open */
if (dentry->d_inode &&
it_disposition(it, DISP_OPEN_OPEN)) { /* nocreate */
if (S_ISFIFO(dentry->d_inode->i_mode)) {
- // We cannot call open here as it would
- // deadlock.
- ptlrpc_req_finished(
- (struct ptlrpc_request *)
- it->d.lustre.it_data);
+ /* We cannot call open here as it might
+ * deadlock. This case is unreachable in
+ * practice because of
+ * OBD_CONNECT_NODEVOH. */
} else {
struct file *filp;
OBD_FREE(it, sizeof(*it));
}
} else {
- de = ll_lookup_it(parent, dentry, NULL, 0);
- }
+ de = ll_lookup_it(parent, dentry, NULL);
+ }
- RETURN(de);
+ RETURN(de);
}
#endif /* HAVE_IOP_ATOMIC_OPEN */
int rc;
ENTRY;
- LASSERT(it && it->d.lustre.it_disposition);
+ LASSERT(it && it->it_disposition);
- LASSERT(it_disposition(it, DISP_ENQ_CREATE_REF));
- request = it->d.lustre.it_data;
+ LASSERT(it_disposition(it, DISP_ENQ_CREATE_REF));
+ request = it->it_request;
it_clear_disposition(it, DISP_ENQ_CREATE_REF);
rc = ll_prep_inode(&inode, request, dir->i_sb, it);
if (rc)
* If the create succeeds, we fill in the inode information
* with d_instantiate().
*/
-static int ll_create_it(struct inode *dir, struct dentry *dentry, int mode,
+static int ll_create_it(struct inode *dir, struct dentry *dentry,
struct lookup_intent *it)
{
struct inode *inode;
RETURN(PTR_ERR(inode));
d_instantiate(dentry, inode);
+
+ if (!(ll_i2sbi(inode)->ll_flags & LL_SBI_FILE_SECCTX)) {
+ rc = ll_inode_init_security(dentry, inode, dir);
+ if (rc)
+ RETURN(rc);
+ }
+
RETURN(0);
}
void ll_update_times(struct ptlrpc_request *request, struct inode *inode)
{
- struct mdt_body *body = req_capsule_server_get(&request->rq_pill,
- &RMF_MDT_BODY);
-
- LASSERT(body);
- if (body->valid & OBD_MD_FLMTIME &&
- body->mtime > LTIME_S(inode->i_mtime)) {
- CDEBUG(D_INODE, "setting fid "DFID" mtime from %lu to "LPU64
- "\n", PFID(ll_inode2fid(inode)),
- LTIME_S(inode->i_mtime), body->mtime);
- LTIME_S(inode->i_mtime) = body->mtime;
- }
- if (body->valid & OBD_MD_FLCTIME &&
- body->ctime > LTIME_S(inode->i_ctime))
- LTIME_S(inode->i_ctime) = body->ctime;
+ struct mdt_body *body = req_capsule_server_get(&request->rq_pill,
+ &RMF_MDT_BODY);
+
+ LASSERT(body);
+ if (body->mbo_valid & OBD_MD_FLMTIME &&
+ body->mbo_mtime > LTIME_S(inode->i_mtime)) {
+ CDEBUG(D_INODE, "setting fid "DFID" mtime from %lu to %llu"
+ "\n", PFID(ll_inode2fid(inode)),
+ LTIME_S(inode->i_mtime), body->mbo_mtime);
+ LTIME_S(inode->i_mtime) = body->mbo_mtime;
+ }
+
+ if (body->mbo_valid & OBD_MD_FLCTIME &&
+ body->mbo_ctime > LTIME_S(inode->i_ctime))
+ LTIME_S(inode->i_ctime) = body->mbo_ctime;
}
-static int ll_new_node(struct inode *dir, struct qstr *name,
- const char *tgt, int mode, int rdev,
- struct dentry *dchild, __u32 opc)
+static int ll_new_node(struct inode *dir, struct dentry *dchild,
+ const char *tgt, umode_t mode, int rdev, __u32 opc)
{
+ struct qstr *name = &dchild->d_name;
struct ptlrpc_request *request = NULL;
struct md_op_data *op_data;
struct inode *inode = NULL;
if (unlikely(tgt != NULL))
tgt_len = strlen(tgt) + 1;
- op_data = ll_prep_md_op_data(NULL, dir, NULL, name->name,
- name->len, 0, opc, NULL);
- if (IS_ERR(op_data))
- GOTO(err_exit, err = PTR_ERR(op_data));
+again:
+ op_data = ll_prep_md_op_data(NULL, dir, NULL, name->name,
+ name->len, 0, opc, NULL);
+ if (IS_ERR(op_data))
+ GOTO(err_exit, err = PTR_ERR(op_data));
+
+ if (sbi->ll_flags & LL_SBI_FILE_SECCTX) {
+ err = ll_dentry_init_security(dchild, mode, &dchild->d_name,
+ &op_data->op_file_secctx_name,
+ &op_data->op_file_secctx,
+ &op_data->op_file_secctx_size);
+ if (err < 0)
+ GOTO(err_exit, err);
+ }
err = md_create(sbi->ll_md_exp, op_data, tgt, tgt_len, mode,
from_kuid(&init_user_ns, current_fsuid()),
from_kgid(&init_user_ns, current_fsgid()),
cfs_curproc_cap_pack(), rdev, &request);
ll_finish_md_op_data(op_data);
- if (err)
- GOTO(err_exit, err);
+ op_data = NULL;
+ if (err < 0 && err != -EREMOTE)
+ GOTO(err_exit, err);
+
+ /* If the client doesn't know where to create a subdirectory (or
+ * in case of a race that sends the RPC to the wrong MDS), the
+ * MDS will return -EREMOTE and the client will fetch the layout
+ * of the directory, then create the directory on the right MDT. */
+ if (unlikely(err == -EREMOTE)) {
+ struct ll_inode_info *lli = ll_i2info(dir);
+ struct lmv_user_md *lum;
+ int lumsize;
+ int err2;
+
+ ptlrpc_req_finished(request);
+ request = NULL;
+
+ err2 = ll_dir_getstripe(dir, (void **)&lum, &lumsize, &request,
+ OBD_MD_DEFAULT_MEA);
+ if (err2 == 0) {
+ /* Update stripe_offset and retry */
+ lli->lli_def_stripe_offset = lum->lum_stripe_offset;
+ } else if (err2 == -ENODATA &&
+ lli->lli_def_stripe_offset != -1) {
+ /* If there are no default stripe EA on the MDT, but the
+ * client has default stripe, then it probably means
+ * default stripe EA has just been deleted. */
+ lli->lli_def_stripe_offset = -1;
+ } else {
+ GOTO(err_exit, err);
+ }
- ll_update_times(request, dir);
+ ptlrpc_req_finished(request);
+ request = NULL;
+ goto again;
+ }
- if (dchild) {
- err = ll_prep_inode(&inode, request, dchild->d_sb, NULL);
- if (err)
- GOTO(err_exit, err);
+ ll_update_times(request, dir);
- d_instantiate(dchild, inode);
- }
- EXIT;
+ CFS_FAIL_TIMEOUT(OBD_FAIL_LLITE_NEWNODE_PAUSE, cfs_fail_val);
+
+ err = ll_prep_inode(&inode, request, dchild->d_sb, NULL);
+ if (err)
+ GOTO(err_exit, err);
+
+ d_instantiate(dchild, inode);
+
+ if (!(sbi->ll_flags & LL_SBI_FILE_SECCTX)) {
+ err = ll_inode_init_security(dchild, inode, dir);
+ if (err)
+ GOTO(err_exit, err);
+ }
+
+ EXIT;
err_exit:
- ptlrpc_req_finished(request);
+ if (request != NULL)
+ ptlrpc_req_finished(request);
- return err;
+ if (!IS_ERR_OR_NULL(op_data))
+ ll_finish_md_op_data(op_data);
+
+ return err;
}
-static int ll_mknod_generic(struct inode *dir, struct qstr *name, int mode,
- unsigned rdev, struct dentry *dchild)
+static int ll_mknod(struct inode *dir, struct dentry *dchild, ll_umode_t mode,
+ dev_t rdev)
{
- int err;
+ struct qstr *name = &dchild->d_name;
+ int err;
ENTRY;
CDEBUG(D_VFSTRACE, "VFS Op:name=%.*s, dir="DFID"(%p) mode %o dev %x\n",
case S_IFBLK:
case S_IFIFO:
case S_IFSOCK:
- err = ll_new_node(dir, name, NULL, mode, rdev, dchild,
- LUSTRE_OPC_MKNOD);
+ err = ll_new_node(dir, dchild, NULL, mode, old_encode_dev(rdev),
+ LUSTRE_OPC_MKNOD);
break;
case S_IFDIR:
err = -EPERM;
{
int rc;
+ CFS_FAIL_TIMEOUT(OBD_FAIL_LLITE_CREATE_FILE_PAUSE, cfs_fail_val);
+
CDEBUG(D_VFSTRACE, "VFS Op:name=%.*s, dir="DFID"(%p), "
"flags=%u, excl=%d\n", dentry->d_name.len,
dentry->d_name.name, PFID(ll_inode2fid(dir)),
dir, mode, want_excl);
- rc = ll_mknod_generic(dir, &dentry->d_name, mode, 0, dentry);
+ rc = ll_mknod(dir, dentry, mode, 0);
ll_stats_ops_tally(ll_i2sbi(dir), LPROC_LL_CREATE, 1);
{
struct ll_dentry_data *lld = ll_d2d(dentry);
struct lookup_intent *it = NULL;
- int rc;
+ int rc;
+
+ CFS_FAIL_TIMEOUT(OBD_FAIL_LLITE_CREATE_FILE_PAUSE, cfs_fail_val);
if (lld != NULL)
it = lld->lld_it;
if (!it)
- return ll_mknod_generic(dir, &dentry->d_name, mode, 0, dentry);
+ return ll_mknod(dir, dentry, mode, 0);
lld->lld_it = NULL;
/* Was there an error? Propagate it! */
- if (it->d.lustre.it_status) {
- rc = it->d.lustre.it_status;
+ if (it->it_status) {
+ rc = it->it_status;
goto out;
}
- rc = ll_create_it(dir, dentry, mode, it);
+ rc = ll_create_it(dir, dentry, it);
if (nd && (nd->flags & LOOKUP_OPEN) && dentry->d_inode) { /* Open */
struct file *filp;
}
#endif /* HAVE_IOP_ATOMIC_OPEN */
-static int ll_symlink_generic(struct inode *dir, struct qstr *name,
- const char *tgt, struct dentry *dchild)
+static int ll_symlink(struct inode *dir, struct dentry *dchild,
+ const char *oldpath)
{
- int err;
- ENTRY;
+ struct qstr *name = &dchild->d_name;
+ int err;
+ ENTRY;
CDEBUG(D_VFSTRACE, "VFS Op:name=%.*s, dir="DFID"(%p), target=%.*s\n",
name->len, name->name, PFID(ll_inode2fid(dir)),
- dir, 3000, tgt);
+ dir, 3000, oldpath);
- err = ll_new_node(dir, name, (char *)tgt, S_IFLNK | S_IRWXUGO,
- 0, dchild, LUSTRE_OPC_SYMLINK);
+ err = ll_new_node(dir, dchild, oldpath, S_IFLNK | S_IRWXUGO, 0,
+ LUSTRE_OPC_SYMLINK);
if (!err)
ll_stats_ops_tally(ll_i2sbi(dir), LPROC_LL_SYMLINK, 1);
RETURN(err);
}
-static int ll_link_generic(struct inode *src, struct inode *dir,
- struct qstr *name, struct dentry *dchild)
+static int ll_link(struct dentry *old_dentry, struct inode *dir,
+ struct dentry *new_dentry)
{
- struct ll_sb_info *sbi = ll_i2sbi(dir);
- struct ptlrpc_request *request = NULL;
- struct md_op_data *op_data;
- int err;
+ struct inode *src = old_dentry->d_inode;
+ struct qstr *name = &new_dentry->d_name;
+ struct ll_sb_info *sbi = ll_i2sbi(dir);
+ struct ptlrpc_request *request = NULL;
+ struct md_op_data *op_data;
+ int err;
- ENTRY;
- CDEBUG(D_VFSTRACE, "VFS Op: inode="DFID"(%p), dir="DFID
- "(%p), target=%.*s\n", PFID(ll_inode2fid(src)), src,
+ ENTRY;
+ CDEBUG(D_VFSTRACE, "VFS Op: inode="DFID"(%p), dir="DFID"(%p), "
+ "target=%.*s\n", PFID(ll_inode2fid(src)), src,
PFID(ll_inode2fid(dir)), dir, name->len, name->name);
op_data = ll_prep_md_op_data(NULL, src, dir, name->name, name->len,
RETURN(err);
}
-static int ll_mkdir_generic(struct inode *dir, struct qstr *name,
- int mode, struct dentry *dchild)
-
+static int ll_mkdir(struct inode *dir, struct dentry *dchild, ll_umode_t mode)
{
+ struct qstr *name = &dchild->d_name;
int err;
ENTRY;
if (!IS_POSIXACL(dir) || !exp_connect_umask(ll_i2mdexp(dir)))
mode &= ~current_umask();
+
mode = (mode & (S_IRWXUGO|S_ISVTX)) | S_IFDIR;
- err = ll_new_node(dir, name, NULL, mode, 0, dchild, LUSTRE_OPC_MKDIR);
- if (!err)
- ll_stats_ops_tally(ll_i2sbi(dir), LPROC_LL_MKDIR, 1);
+ err = ll_new_node(dir, dchild, NULL, mode, 0, LUSTRE_OPC_MKDIR);
+ if (err == 0)
+ ll_stats_ops_tally(ll_i2sbi(dir), LPROC_LL_MKDIR, 1);
- RETURN(err);
+ RETURN(err);
}
-static int ll_rmdir_generic(struct inode *dir, struct dentry *dparent,
- struct dentry *dchild, struct qstr *name)
+static int ll_rmdir(struct inode *dir, struct dentry *dchild)
{
+ struct qstr *name = &dchild->d_name;
struct ptlrpc_request *request = NULL;
struct md_op_data *op_data;
int rc;
CDEBUG(D_VFSTRACE, "VFS Op:name=%.*s, dir="DFID"(%p)\n",
name->len, name->name, PFID(ll_inode2fid(dir)), dir);
- if (unlikely(ll_d_mountpoint(dparent, dchild, name)))
+ if (unlikely(d_mountpoint(dchild)))
RETURN(-EBUSY);
op_data = ll_prep_md_op_data(NULL, dir, NULL, name->name, name->len,
if (IS_ERR(op_data))
RETURN(PTR_ERR(op_data));
- if (dchild != NULL && dchild->d_inode != NULL)
+ if (dchild->d_inode != NULL)
op_data->op_fid3 = *ll_inode2fid(dchild->d_inode);
+
op_data->op_fid2 = op_data->op_fid3;
rc = md_unlink(ll_i2sbi(dir)->ll_md_exp, op_data, &request);
ll_finish_md_op_data(op_data);
RETURN(rc);
}
-int ll_objects_destroy(struct ptlrpc_request *request, struct inode *dir)
-{
- struct mdt_body *body;
- struct lov_mds_md *eadata;
- struct lov_stripe_md *lsm = NULL;
- struct obd_trans_info oti = { 0 };
- struct obdo *oa;
- struct obd_capa *oc = NULL;
- int rc;
- ENTRY;
-
- /* req is swabbed so this is safe */
- body = req_capsule_server_get(&request->rq_pill, &RMF_MDT_BODY);
- if (!(body->valid & OBD_MD_FLEASIZE))
- RETURN(0);
-
- if (body->eadatasize == 0) {
- CERROR("OBD_MD_FLEASIZE set but eadatasize zero\n");
- GOTO(out, rc = -EPROTO);
- }
-
- /* The MDS sent back the EA because we unlinked the last reference
- * to this file. Use this EA to unlink the objects on the OST.
- * It's opaque so we don't swab here; we leave it to obd_unpackmd() to
- * check it is complete and sensible. */
- eadata = req_capsule_server_sized_get(&request->rq_pill, &RMF_MDT_MD,
- body->eadatasize);
- LASSERT(eadata != NULL);
-
- rc = obd_unpackmd(ll_i2dtexp(dir), &lsm, eadata, body->eadatasize);
- if (rc < 0) {
- CERROR("obd_unpackmd: %d\n", rc);
- GOTO(out, rc);
- }
- LASSERT(rc >= sizeof(*lsm));
-
- OBDO_ALLOC(oa);
- if (oa == NULL)
- GOTO(out_free_memmd, rc = -ENOMEM);
-
- oa->o_oi = lsm->lsm_oi;
- oa->o_mode = body->mode & S_IFMT;
- oa->o_valid = OBD_MD_FLID | OBD_MD_FLTYPE | OBD_MD_FLGROUP;
-
- if (body->valid & OBD_MD_FLCOOKIE) {
- oa->o_valid |= OBD_MD_FLCOOKIE;
- oti.oti_logcookies =
- req_capsule_server_sized_get(&request->rq_pill,
- &RMF_LOGCOOKIES,
- sizeof(struct llog_cookie) *
- lsm->lsm_stripe_count);
- if (oti.oti_logcookies == NULL) {
- oa->o_valid &= ~OBD_MD_FLCOOKIE;
- body->valid &= ~OBD_MD_FLCOOKIE;
- }
- }
-
- if (body->valid & OBD_MD_FLOSSCAPA) {
- rc = md_unpack_capa(ll_i2mdexp(dir), request, &RMF_CAPA2, &oc);
- if (rc)
- GOTO(out_free_memmd, rc);
- }
-
- rc = obd_destroy(NULL, ll_i2dtexp(dir), oa, lsm, &oti,
- ll_i2mdexp(dir), oc);
- capa_put(oc);
- if (rc)
- CERROR("obd destroy objid "DOSTID" error %d\n",
- POSTID(&lsm->lsm_oi), rc);
-out_free_memmd:
- obd_free_memmd(ll_i2dtexp(dir), &lsm);
- OBDO_FREE(oa);
-out:
- return rc;
-}
-
-/* ll_unlink_generic() doesn't update the inode with the new link count.
- * Instead, ll_ddelete() and ll_d_iput() will update it based upon if there
- * is any lock existing. They will recycle dentries and inodes based upon locks
- * too. b=20433 */
-static int ll_unlink_generic(struct inode *dir, struct dentry *dparent,
- struct dentry *dchild, struct qstr *name)
+static int ll_unlink(struct inode *dir, struct dentry *dchild)
{
+ struct qstr *name = &dchild->d_name;
struct ptlrpc_request *request = NULL;
struct md_op_data *op_data;
int rc;
* XXX: unlink bind mountpoint maybe call to here,
* just check it as vfs_unlink does.
*/
- if (unlikely(ll_d_mountpoint(dparent, dchild, name)))
- RETURN(-EBUSY);
+ if (unlikely(d_mountpoint(dchild)))
+ RETURN(-EBUSY);
- op_data = ll_prep_md_op_data(NULL, dir, NULL, name->name,
- name->len, 0, LUSTRE_OPC_ANY, NULL);
- if (IS_ERR(op_data))
- RETURN(PTR_ERR(op_data));
+ op_data = ll_prep_md_op_data(NULL, dir, NULL, name->name, name->len, 0,
+ LUSTRE_OPC_ANY, NULL);
+ if (IS_ERR(op_data))
+ RETURN(PTR_ERR(op_data));
- if (dchild != NULL && dchild->d_inode != NULL)
+ if (dchild->d_inode != NULL)
op_data->op_fid3 = *ll_inode2fid(dchild->d_inode);
op_data->op_fid2 = op_data->op_fid3;
ll_update_times(request, dir);
ll_stats_ops_tally(ll_i2sbi(dir), LPROC_LL_UNLINK, 1);
- rc = ll_objects_destroy(request, dir);
out:
ptlrpc_req_finished(request);
RETURN(rc);
}
-static int ll_rename_generic(struct inode *src, struct dentry *src_dparent,
- struct dentry *src_dchild, struct qstr *src_name,
- struct inode *tgt, struct dentry *tgt_dparent,
- struct dentry *tgt_dchild, struct qstr *tgt_name)
+static int ll_rename(struct inode *src, struct dentry *src_dchild,
+ struct inode *tgt, struct dentry *tgt_dchild)
{
+ struct qstr *src_name = &src_dchild->d_name;
+ struct qstr *tgt_name = &tgt_dchild->d_name;
struct ptlrpc_request *request = NULL;
struct ll_sb_info *sbi = ll_i2sbi(src);
struct md_op_data *op_data;
PFID(ll_inode2fid(src)), src, tgt_name->len,
tgt_name->name, PFID(ll_inode2fid(tgt)), tgt);
- if (unlikely(ll_d_mountpoint(src_dparent, src_dchild, src_name) ||
- ll_d_mountpoint(tgt_dparent, tgt_dchild, tgt_name)))
- RETURN(-EBUSY);
+ if (unlikely(d_mountpoint(src_dchild) || d_mountpoint(tgt_dchild)))
+ RETURN(-EBUSY);
- op_data = ll_prep_md_op_data(NULL, src, tgt, NULL, 0, 0,
- LUSTRE_OPC_ANY, NULL);
- if (IS_ERR(op_data))
- RETURN(PTR_ERR(op_data));
+ op_data = ll_prep_md_op_data(NULL, src, tgt, NULL, 0, 0,
+ LUSTRE_OPC_ANY, NULL);
+ if (IS_ERR(op_data))
+ RETURN(PTR_ERR(op_data));
- if (src_dchild != NULL && src_dchild->d_inode != NULL)
+ if (src_dchild->d_inode != NULL)
op_data->op_fid3 = *ll_inode2fid(src_dchild->d_inode);
- if (tgt_dchild != NULL && tgt_dchild->d_inode != NULL)
+
+ if (tgt_dchild->d_inode != NULL)
op_data->op_fid4 = *ll_inode2fid(tgt_dchild->d_inode);
err = md_rename(sbi->ll_md_exp, op_data,
ll_update_times(request, src);
ll_update_times(request, tgt);
ll_stats_ops_tally(sbi, LPROC_LL_RENAME, 1);
- err = ll_objects_destroy(request, src);
}
ptlrpc_req_finished(request);
- RETURN(err);
-}
-
-static int ll_mknod(struct inode *dir, struct dentry *dchild, ll_umode_t mode,
- dev_t rdev)
-{
- return ll_mknod_generic(dir, &dchild->d_name, mode,
- old_encode_dev(rdev), dchild);
-}
-
-static int ll_unlink(struct inode * dir, struct dentry *dentry)
-{
- return ll_unlink_generic(dir, NULL, dentry, &dentry->d_name);
-}
-
-static int ll_mkdir(struct inode *dir, struct dentry *dentry, ll_umode_t mode)
-{
- return ll_mkdir_generic(dir, &dentry->d_name, mode, dentry);
-}
-
-static int ll_rmdir(struct inode *dir, struct dentry *dentry)
-{
- return ll_rmdir_generic(dir, NULL, dentry, &dentry->d_name);
-}
-
-static int ll_symlink(struct inode *dir, struct dentry *dentry,
- const char *oldname)
-{
- return ll_symlink_generic(dir, &dentry->d_name, oldname, dentry);
-}
-
-static int ll_link(struct dentry *old_dentry, struct inode *dir,
- struct dentry *new_dentry)
-{
- return ll_link_generic(old_dentry->d_inode, dir, &new_dentry->d_name,
- new_dentry);
-}
+ if (err == 0)
+ d_move(src_dchild, tgt_dchild);
-static int ll_rename(struct inode *old_dir, struct dentry *old_dentry,
- struct inode *new_dir, struct dentry *new_dentry)
-{
- int err;
- err = ll_rename_generic(old_dir, NULL,
- old_dentry, &old_dentry->d_name,
- new_dir, NULL, new_dentry,
- &new_dentry->d_name);
- if (!err)
- d_move(old_dentry, new_dentry);
- return err;
+ RETURN(err);
}
const struct inode_operations ll_dir_inode_operations = {