* 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/
#define DEBUG_SUBSYSTEM S_LLITE
#include <lustre_dlm.h>
-#include <lustre_lite.h>
#include <linux/pagemap.h>
#include <linux/file.h>
#include <linux/sched.h>
-#include "llite_internal.h"
+#include <linux/user_namespace.h>
+#ifdef HAVE_UIDGID_HEADER
+# include <linux/uidgid.h>
+#endif
#include <lustre/ll_fiemap.h>
+
#include <lustre_ioctl.h>
+#include <lustre_swab.h>
#include "cl_object.h"
+#include "llite_internal.h"
+#include "vvp_internal.h"
static int
ll_put_grouplock(struct inode *inode, struct file *file, unsigned long arg);
OBD_SLAB_FREE_PTR(fd, ll_file_data_slab);
}
-void ll_pack_inode2opdata(struct inode *inode, struct md_op_data *op_data,
- struct lustre_handle *fh)
-{
- op_data->op_fid1 = ll_i2info(inode)->lli_fid;
- op_data->op_attr.ia_mode = inode->i_mode;
- op_data->op_attr.ia_atime = inode->i_atime;
- op_data->op_attr.ia_mtime = inode->i_mtime;
- op_data->op_attr.ia_ctime = inode->i_ctime;
- op_data->op_attr.ia_size = i_size_read(inode);
- op_data->op_attr_blocks = inode->i_blocks;
- ((struct ll_iattr *)&op_data->op_attr)->ia_attr_flags =
- ll_inode_to_ext_flags(inode->i_flags);
- op_data->op_ioepoch = ll_i2info(inode)->lli_ioepoch;
- if (fh)
- op_data->op_handle = *fh;
- op_data->op_capa1 = ll_mdscapa_get(inode);
-
- if (LLIF_DATA_MODIFIED & ll_i2info(inode)->lli_flags)
- op_data->op_bias |= MDS_DATA_MODIFIED;
-}
-
/**
- * Closes the IO epoch and packs all the attributes into @op_data for
- * the CLOSE rpc.
+ * Packs all the attributes into @op_data for the CLOSE rpc.
*/
static void ll_prepare_close(struct inode *inode, struct md_op_data *op_data,
struct obd_client_handle *och)
{
- ENTRY;
-
- op_data->op_attr.ia_valid = ATTR_MODE | ATTR_ATIME | ATTR_ATIME_SET |
- ATTR_MTIME | ATTR_MTIME_SET |
- ATTR_CTIME | ATTR_CTIME_SET;
-
- if (!(och->och_flags & FMODE_WRITE))
- goto out;
+ ENTRY;
- if (!exp_connect_som(ll_i2mdexp(inode)) || !S_ISREG(inode->i_mode))
- op_data->op_attr.ia_valid |= ATTR_SIZE | ATTR_BLOCKS;
- else
- ll_ioepoch_close(inode, op_data, &och, 0);
+ ll_prep_md_op_data(op_data, inode, NULL, NULL,
+ 0, 0, LUSTRE_OPC_ANY, NULL);
+
+ op_data->op_attr.ia_mode = inode->i_mode;
+ op_data->op_attr.ia_atime = inode->i_atime;
+ op_data->op_attr.ia_mtime = inode->i_mtime;
+ op_data->op_attr.ia_ctime = inode->i_ctime;
+ op_data->op_attr.ia_size = i_size_read(inode);
+ op_data->op_attr.ia_valid |= ATTR_MODE | ATTR_ATIME | ATTR_ATIME_SET |
+ ATTR_MTIME | ATTR_MTIME_SET |
+ ATTR_CTIME | ATTR_CTIME_SET;
+ op_data->op_attr_blocks = inode->i_blocks;
+ op_data->op_attr_flags = ll_inode_to_ext_flags(inode->i_flags);
+ op_data->op_handle = och->och_fh;
+
+ if (och->och_flags & FMODE_WRITE &&
+ ll_file_test_and_clear_flag(ll_i2info(inode), LLIF_DATA_MODIFIED))
+ /* For HSM: if inode data has been modified, pack it so that
+ * MDT can set data dirty flag in the archive. */
+ op_data->op_bias |= MDS_DATA_MODIFIED;
-out:
- ll_pack_inode2opdata(inode, op_data, &och->och_fh);
- ll_prep_md_op_data(op_data, inode, NULL, NULL,
- 0, 0, LUSTRE_OPC_ANY, NULL);
- EXIT;
+ EXIT;
}
-static int ll_close_inode_openhandle(struct obd_export *md_exp,
- struct inode *inode,
+/**
+ * Perform a close, possibly with a bias.
+ * The meaning of "data" depends on the value of "bias".
+ *
+ * If \a bias is MDS_HSM_RELEASE then \a data is a pointer to the data version.
+ * If \a bias is MDS_CLOSE_LAYOUT_SWAP then \a data is a pointer to the inode to
+ * swap layouts with.
+ */
+static int ll_close_inode_openhandle(struct inode *inode,
struct obd_client_handle *och,
- const __u64 *data_version)
+ enum mds_op_bias bias, void *data)
{
- struct obd_export *exp = ll_i2mdexp(inode);
- struct md_op_data *op_data;
- struct ptlrpc_request *req = NULL;
- struct obd_device *obd = class_exp2obd(exp);
- int epoch_close = 1;
- int rc;
- ENTRY;
+ struct obd_export *md_exp = ll_i2mdexp(inode);
+ const struct ll_inode_info *lli = ll_i2info(inode);
+ struct md_op_data *op_data;
+ struct ptlrpc_request *req = NULL;
+ int rc;
+ ENTRY;
- if (obd == NULL) {
- /*
- * XXX: in case of LMV, is this correct to access
- * ->exp_handle?
- */
- CERROR("Invalid MDC connection handle "LPX64"\n",
- ll_i2mdexp(inode)->exp_handle.h_cookie);
- GOTO(out, rc = 0);
- }
+ if (class_exp2obd(md_exp) == NULL) {
+ CERROR("%s: invalid MDC connection handle closing "DFID"\n",
+ ll_get_fsname(inode->i_sb, NULL, 0),
+ PFID(&lli->lli_fid));
+ GOTO(out, rc = 0);
+ }
- OBD_ALLOC_PTR(op_data);
- if (op_data == NULL)
- GOTO(out, rc = -ENOMEM); // XXX We leak openhandle and request here.
+ OBD_ALLOC_PTR(op_data);
+ /* We leak openhandle and request here on error, but not much to be
+ * done in OOM case since app won't retry close on error either. */
+ if (op_data == NULL)
+ GOTO(out, rc = -ENOMEM);
ll_prepare_close(inode, op_data, och);
- if (data_version != NULL) {
- /* Pass in data_version implies release. */
+ switch (bias) {
+ case MDS_CLOSE_LAYOUT_SWAP:
+ LASSERT(data != NULL);
+ op_data->op_bias |= MDS_CLOSE_LAYOUT_SWAP;
+ op_data->op_data_version = 0;
+ op_data->op_lease_handle = och->och_lease_handle;
+ op_data->op_fid2 = *ll_inode2fid(data);
+ break;
+
+ case MDS_HSM_RELEASE:
+ LASSERT(data != NULL);
op_data->op_bias |= MDS_HSM_RELEASE;
- op_data->op_data_version = *data_version;
+ op_data->op_data_version = *(__u64 *)data;
op_data->op_lease_handle = och->och_lease_handle;
op_data->op_attr.ia_valid |= ATTR_SIZE | ATTR_BLOCKS;
- }
- epoch_close = (op_data->op_flags & MF_EPOCH_CLOSE);
- rc = md_close(md_exp, op_data, och->och_mod, &req);
- if (rc == -EAGAIN) {
- /* This close must have the epoch closed. */
- LASSERT(epoch_close);
- /* MDS has instructed us to obtain Size-on-MDS attribute from
- * OSTs and send setattr to back to MDS. */
- rc = ll_som_update(inode, op_data);
- if (rc) {
- CERROR("%s: inode "DFID" mdc Size-on-MDS update"
- " failed: rc = %d\n",
- ll_i2mdexp(inode)->exp_obd->obd_name,
- PFID(ll_inode2fid(inode)), rc);
- rc = 0;
- }
- } else if (rc) {
- CERROR("%s: inode "DFID" mdc close failed: rc = %d\n",
- ll_i2mdexp(inode)->exp_obd->obd_name,
- PFID(ll_inode2fid(inode)), rc);
- }
-
- /* DATA_MODIFIED flag was successfully sent on close, cancel data
- * modification flag. */
- if (rc == 0 && (op_data->op_bias & MDS_DATA_MODIFIED)) {
- struct ll_inode_info *lli = ll_i2info(inode);
+ break;
- spin_lock(&lli->lli_lock);
- lli->lli_flags &= ~LLIF_DATA_MODIFIED;
- spin_unlock(&lli->lli_lock);
+ default:
+ LASSERT(data == NULL);
+ break;
}
- if (rc == 0) {
- rc = ll_objects_destroy(req, inode);
- if (rc)
- CERROR("%s: inode "DFID
- " ll_objects destroy: rc = %d\n",
- ll_i2mdexp(inode)->exp_obd->obd_name,
- PFID(ll_inode2fid(inode)), rc);
- }
+ rc = md_close(md_exp, op_data, och->och_mod, &req);
+ if (rc != 0 && rc != -EINTR)
+ CERROR("%s: inode "DFID" mdc close failed: rc = %d\n",
+ md_exp->exp_obd->obd_name, PFID(&lli->lli_fid), rc);
- if (rc == 0 && op_data->op_bias & MDS_HSM_RELEASE) {
+ if (rc == 0 &&
+ op_data->op_bias & (MDS_HSM_RELEASE | MDS_CLOSE_LAYOUT_SWAP)) {
struct mdt_body *body;
+
body = req_capsule_server_get(&req->rq_pill, &RMF_MDT_BODY);
- if (!(body->mbo_valid & OBD_MD_FLRELEASED))
+ if (!(body->mbo_valid & OBD_MD_CLOSE_INTENT_EXECED))
rc = -EBUSY;
}
- ll_finish_md_op_data(op_data);
- EXIT;
+ ll_finish_md_op_data(op_data);
+ EXIT;
out:
- if (exp_connect_som(exp) && !epoch_close &&
- S_ISREG(inode->i_mode) && (och->och_flags & FMODE_WRITE)) {
- ll_queue_done_writing(inode, LLIF_DONE_WRITING);
- } else {
- md_clear_open_replay_data(md_exp, och);
- /* Free @och if it is not waiting for DONE_WRITING. */
- och->och_fh.cookie = DEAD_HANDLE_MAGIC;
- OBD_FREE_PTR(och);
- }
- if (req) /* This is close request */
- ptlrpc_req_finished(req);
- return rc;
+ md_clear_open_replay_data(md_exp, och);
+ och->och_fh.cookie = DEAD_HANDLE_MAGIC;
+ OBD_FREE_PTR(och);
+
+ ptlrpc_req_finished(req); /* This is close request */
+ return rc;
}
int ll_md_real_close(struct inode *inode, fmode_t fmode)
if (och != NULL) {
/* There might be a race and this handle may already
* be closed. */
- rc = ll_close_inode_openhandle(ll_i2sbi(inode)->ll_md_exp,
- inode, och, NULL);
+ rc = ll_close_inode_openhandle(inode, och, 0, NULL);
}
RETURN(rc);
}
-static int ll_md_close(struct obd_export *md_exp, struct inode *inode,
- struct file *file)
+static int ll_md_close(struct inode *inode, struct file *file)
{
- struct ll_file_data *fd = LUSTRE_FPRIVATE(file);
- struct ll_inode_info *lli = ll_i2info(inode);
- int rc = 0;
- ENTRY;
+ union ldlm_policy_data policy = {
+ .l_inodebits = { MDS_INODELOCK_OPEN },
+ };
+ __u64 flags = LDLM_FL_BLOCK_GRANTED | LDLM_FL_TEST_LOCK;
+ struct ll_file_data *fd = LUSTRE_FPRIVATE(file);
+ struct ll_inode_info *lli = ll_i2info(inode);
+ struct lustre_handle lockh;
+ enum ldlm_mode lockmode;
+ int rc = 0;
+ ENTRY;
- /* clear group lock, if present */
- if (unlikely(fd->fd_flags & LL_FILE_GROUP_LOCKED))
- ll_put_grouplock(inode, file, fd->fd_grouplock.cg_gid);
+ /* clear group lock, if present */
+ if (unlikely(fd->fd_flags & LL_FILE_GROUP_LOCKED))
+ ll_put_grouplock(inode, file, fd->fd_grouplock.lg_gid);
if (fd->fd_lease_och != NULL) {
bool lease_broken;
}
if (fd->fd_och != NULL) {
- rc = ll_close_inode_openhandle(md_exp, inode, fd->fd_och, NULL);
+ rc = ll_close_inode_openhandle(inode, fd->fd_och, 0, NULL);
fd->fd_och = NULL;
GOTO(out, rc);
}
/* Let's see if we have good enough OPEN lock on the file and if
we can skip talking to MDS */
- if (file->f_dentry->d_inode) { /* Can this ever be false? */
- int lockmode;
- __u64 flags = LDLM_FL_BLOCK_GRANTED | LDLM_FL_TEST_LOCK;
- struct lustre_handle lockh;
- struct inode *inode = file->f_dentry->d_inode;
- ldlm_policy_data_t policy = {.l_inodebits={MDS_INODELOCK_OPEN}};
-
- mutex_lock(&lli->lli_och_mutex);
- if (fd->fd_omode & FMODE_WRITE) {
- lockmode = LCK_CW;
- LASSERT(lli->lli_open_fd_write_count);
- lli->lli_open_fd_write_count--;
- } else if (fd->fd_omode & FMODE_EXEC) {
- lockmode = LCK_PR;
- LASSERT(lli->lli_open_fd_exec_count);
- lli->lli_open_fd_exec_count--;
- } else {
- lockmode = LCK_CR;
- LASSERT(lli->lli_open_fd_read_count);
- lli->lli_open_fd_read_count--;
- }
- mutex_unlock(&lli->lli_och_mutex);
-
- if (!md_lock_match(md_exp, flags, ll_inode2fid(inode),
- LDLM_IBITS, &policy, lockmode,
- &lockh)) {
- rc = ll_md_real_close(file->f_dentry->d_inode,
- fd->fd_omode);
- }
+ mutex_lock(&lli->lli_och_mutex);
+ if (fd->fd_omode & FMODE_WRITE) {
+ lockmode = LCK_CW;
+ LASSERT(lli->lli_open_fd_write_count);
+ lli->lli_open_fd_write_count--;
+ } else if (fd->fd_omode & FMODE_EXEC) {
+ lockmode = LCK_PR;
+ LASSERT(lli->lli_open_fd_exec_count);
+ lli->lli_open_fd_exec_count--;
} else {
- CERROR("released file has negative dentry: file = %p, "
- "dentry = %p, name = %s\n",
- file, file->f_dentry, file->f_dentry->d_name.name);
+ lockmode = LCK_CR;
+ LASSERT(lli->lli_open_fd_read_count);
+ lli->lli_open_fd_read_count--;
}
+ mutex_unlock(&lli->lli_och_mutex);
+
+ if (!md_lock_match(ll_i2mdexp(inode), flags, ll_inode2fid(inode),
+ LDLM_IBITS, &policy, lockmode, &lockh))
+ rc = ll_md_real_close(inode, fd->fd_omode);
out:
LUSTRE_FPRIVATE(file) = NULL;
ll_file_data_put(fd);
- ll_capa_close(inode);
RETURN(rc);
}
CDEBUG(D_VFSTRACE, "VFS Op:inode="DFID"(%p)\n",
PFID(ll_inode2fid(inode)), inode);
-#ifdef CONFIG_FS_POSIX_ACL
- if (sbi->ll_flags & LL_SBI_RMT_CLIENT &&
- inode == inode->i_sb->s_root->d_inode) {
- struct ll_file_data *fd = LUSTRE_FPRIVATE(file);
-
- LASSERT(fd != NULL);
- if (unlikely(fd->fd_flags & LL_FILE_RMTACL)) {
- fd->fd_flags &= ~LL_FILE_RMTACL;
- rct_del(&sbi->ll_rct, current_pid());
- et_search_free(&sbi->ll_et, current_pid());
- }
- }
-#endif
-
- if (inode->i_sb->s_root != file->f_dentry)
+ if (inode->i_sb->s_root != file_dentry(file))
ll_stats_ops_tally(sbi, LPROC_LL_RELEASE, 1);
fd = LUSTRE_FPRIVATE(file);
LASSERT(fd != NULL);
if (S_ISDIR(inode->i_mode) && lli->lli_opendir_key == fd)
ll_deauthorize_statahead(inode, fd);
- if (inode->i_sb->s_root == file->f_dentry) {
- LUSTRE_FPRIVATE(file) = NULL;
- ll_file_data_put(fd);
- RETURN(0);
- }
+ if (inode->i_sb->s_root == file_dentry(file)) {
+ LUSTRE_FPRIVATE(file) = NULL;
+ ll_file_data_put(fd);
+ RETURN(0);
+ }
- if (!S_ISDIR(inode->i_mode)) {
+ if (!S_ISDIR(inode->i_mode)) {
if (lli->lli_clob != NULL)
lov_read_and_clear_async_rc(lli->lli_clob);
- lli->lli_async_rc = 0;
- }
+ lli->lli_async_rc = 0;
+ }
- rc = ll_md_close(sbi->ll_md_exp, inode, file);
+ rc = ll_md_close(inode, file);
- if (CFS_FAIL_TIMEOUT_MS(OBD_FAIL_PTLRPC_DUMP_LOG, cfs_fail_val))
- libcfs_debug_dumplog();
+ if (CFS_FAIL_TIMEOUT_MS(OBD_FAIL_PTLRPC_DUMP_LOG, cfs_fail_val))
+ libcfs_debug_dumplog();
- RETURN(rc);
+ RETURN(rc);
}
static int ll_intent_file_open(struct file *file, void *lmm, int lmmsize,
struct lookup_intent *itp)
{
- struct dentry *de = file->f_dentry;
+ struct dentry *de = file_dentry(file);
struct ll_sb_info *sbi = ll_i2sbi(de->d_inode);
struct dentry *parent = de->d_parent;
const char *name = NULL;
}
rc = ll_prep_inode(&de->d_inode, req, NULL, itp);
- if (!rc && itp->d.lustre.it_lock_mode)
+ if (!rc && itp->it_lock_mode)
ll_set_lock_data(sbi->ll_md_exp, de->d_inode, itp, NULL);
out:
ptlrpc_req_finished(req);
ll_intent_drop_lock(itp);
- RETURN(rc);
-}
+ /* We did open by fid, but by the time we got to the server,
+ * the object disappeared. If this is a create, we cannot really
+ * tell the userspace that the file it was trying to create
+ * does not exist. Instead let's return -ESTALE, and the VFS will
+ * retry the create with LOOKUP_REVAL that we are going to catch
+ * in ll_revalidate_dentry() and use lookup then.
+ */
+ if (rc == -ENOENT && itp->it_op & IT_CREAT)
+ rc = -ESTALE;
-/**
- * Assign an obtained @ioepoch to client's inode. No lock is needed, MDS does
- * not believe attributes if a few ioepoch holders exist. Attributes for
- * previous ioepoch if new one is opened are also skipped by MDS.
- */
-void ll_ioepoch_open(struct ll_inode_info *lli, __u64 ioepoch)
-{
- if (ioepoch && lli->lli_ioepoch != ioepoch) {
- lli->lli_ioepoch = ioepoch;
- CDEBUG(D_INODE, "Epoch "LPU64" opened on "DFID"\n",
- ioepoch, PFID(&lli->lli_fid));
- }
+ RETURN(rc);
}
static int ll_och_fill(struct obd_export *md_exp, struct lookup_intent *it,
struct obd_client_handle *och)
{
- struct ptlrpc_request *req = it->d.lustre.it_data;
struct mdt_body *body;
- body = req_capsule_server_get(&req->rq_pill, &RMF_MDT_BODY);
+ body = req_capsule_server_get(&it->it_request->rq_pill, &RMF_MDT_BODY);
och->och_fh = body->mbo_handle;
och->och_fid = body->mbo_fid1;
- och->och_lease_handle.cookie = it->d.lustre.it_lock_handle;
+ och->och_lease_handle.cookie = it->it_lock_handle;
och->och_magic = OBD_CLIENT_HANDLE_MAGIC;
och->och_flags = it->it_flags;
static int ll_local_open(struct file *file, struct lookup_intent *it,
struct ll_file_data *fd, struct obd_client_handle *och)
{
- struct inode *inode = file->f_dentry->d_inode;
- struct ll_inode_info *lli = ll_i2info(inode);
- ENTRY;
+ struct inode *inode = file_inode(file);
+ ENTRY;
- LASSERT(!LUSTRE_FPRIVATE(file));
+ LASSERT(!LUSTRE_FPRIVATE(file));
- LASSERT(fd != NULL);
+ LASSERT(fd != NULL);
- if (och) {
- struct ptlrpc_request *req = it->d.lustre.it_data;
- struct mdt_body *body;
- int rc;
+ if (och) {
+ int rc;
rc = ll_och_fill(ll_i2sbi(inode)->ll_md_exp, it, och);
if (rc != 0)
RETURN(rc);
-
- body = req_capsule_server_get(&req->rq_pill, &RMF_MDT_BODY);
- ll_ioepoch_open(lli, body->mbo_ioepoch);
}
LUSTRE_FPRIVATE(file) = fd;
if (S_ISDIR(inode->i_mode))
ll_authorize_statahead(inode, fd);
- if (inode->i_sb->s_root == file->f_dentry) {
+ if (inode->i_sb->s_root == file_dentry(file)) {
LUSTRE_FPRIVATE(file) = fd;
RETURN(0);
}
- if (!it || !it->d.lustre.it_disposition) {
+ if (!it || !it->it_disposition) {
/* Convert f_flags into access mode. We cannot use file->f_mode,
* because everything but O_ACCMODE mask was stripped from
* there */
GOTO(out_openerr, rc);
}
- ll_release_openhandle(file->f_dentry, it);
+ ll_release_openhandle(file_dentry(file), it);
}
(*och_usecount)++;
}
} else {
LASSERT(*och_usecount == 0);
- if (!it->d.lustre.it_disposition) {
+ if (!it->it_disposition) {
+ struct ll_dentry_data *ldd = ll_d2d(file->f_path.dentry);
/* We cannot just request lock handle now, new ELC code
means that one of other OPEN locks for this file
could be cancelled, and since blocking ast handler
* handle to be returned from LOOKUP|OPEN request,
* for example if the target entry was a symlink.
*
- * Always fetch MDS_OPEN_LOCK if this is not setstripe.
+ * Only fetch MDS_OPEN_LOCK if this is in NFS path,
+ * marked by a bit set in ll_iget_for_nfs. Clear the
+ * bit so that it's not confusing later callers.
*
+ * NB; when ldd is NULL, it must have come via normal
+ * lookup path only, since ll_iget_for_nfs always calls
+ * ll_d_init().
+ */
+ if (ldd && ldd->lld_nfs_dentry) {
+ ldd->lld_nfs_dentry = 0;
+ it->it_flags |= MDS_OPEN_LOCK;
+ }
+
+ /*
* Always specify MDS_OPEN_BY_FID because we don't want
* to get file with different fid.
*/
- it->it_flags |= MDS_OPEN_LOCK | MDS_OPEN_BY_FID;
+ it->it_flags |= MDS_OPEN_BY_FID;
rc = ll_intent_file_open(file, NULL, 0, it);
if (rc)
GOTO(out_openerr, rc);
LASSERTF(it_disposition(it, DISP_ENQ_OPEN_REF),
"inode %p: disposition %x, status %d\n", inode,
- it_disposition(it, ~0), it->d.lustre.it_status);
+ it_disposition(it, ~0), it->it_status);
rc = ll_local_open(file, it, fd, *och_p);
if (rc)
if (!S_ISREG(inode->i_mode))
GOTO(out_och_free, rc);
- ll_capa_open(inode);
-
- if (!lli->lli_has_smd &&
- (cl_is_lov_delay_create(file->f_flags) ||
- (file->f_mode & FMODE_WRITE) == 0)) {
- CDEBUG(D_INODE, "object creation was delayed\n");
- GOTO(out_och_free, rc);
- }
cl_lov_delay_create_clear(&file->f_flags);
GOTO(out_och_free, rc);
}
if (it && it_disposition(it, DISP_ENQ_OPEN_REF)) {
- ptlrpc_req_finished(it->d.lustre.it_data);
+ ptlrpc_req_finished(it->it_request);
it_clear_disposition(it, DISP_ENQ_OPEN_REF);
}
}
/**
+ * When setting a lease on a file, we take ownership of the lli_mds_*_och
+ * and save it as fd->fd_och so as to force client to reopen the file even
+ * if it has an open lock in cache already.
+ */
+static int ll_lease_och_acquire(struct inode *inode, struct file *file,
+ struct lustre_handle *old_handle)
+{
+ struct ll_inode_info *lli = ll_i2info(inode);
+ struct ll_file_data *fd = LUSTRE_FPRIVATE(file);
+ struct obd_client_handle **och_p;
+ __u64 *och_usecount;
+ int rc = 0;
+ ENTRY;
+
+ /* Get the openhandle of the file */
+ mutex_lock(&lli->lli_och_mutex);
+ if (fd->fd_lease_och != NULL)
+ GOTO(out_unlock, rc = -EBUSY);
+
+ if (fd->fd_och == NULL) {
+ if (file->f_mode & FMODE_WRITE) {
+ LASSERT(lli->lli_mds_write_och != NULL);
+ och_p = &lli->lli_mds_write_och;
+ och_usecount = &lli->lli_open_fd_write_count;
+ } else {
+ LASSERT(lli->lli_mds_read_och != NULL);
+ och_p = &lli->lli_mds_read_och;
+ och_usecount = &lli->lli_open_fd_read_count;
+ }
+
+ if (*och_usecount > 1)
+ GOTO(out_unlock, rc = -EBUSY);
+
+ fd->fd_och = *och_p;
+ *och_usecount = 0;
+ *och_p = NULL;
+ }
+
+ *old_handle = fd->fd_och->och_fh;
+
+ EXIT;
+out_unlock:
+ mutex_unlock(&lli->lli_och_mutex);
+ return rc;
+}
+
+/**
+ * Release ownership on lli_mds_*_och when putting back a file lease.
+ */
+static int ll_lease_och_release(struct inode *inode, struct file *file)
+{
+ struct ll_inode_info *lli = ll_i2info(inode);
+ struct ll_file_data *fd = LUSTRE_FPRIVATE(file);
+ struct obd_client_handle **och_p;
+ struct obd_client_handle *old_och = NULL;
+ __u64 *och_usecount;
+ int rc = 0;
+ ENTRY;
+
+ mutex_lock(&lli->lli_och_mutex);
+ if (file->f_mode & FMODE_WRITE) {
+ och_p = &lli->lli_mds_write_och;
+ och_usecount = &lli->lli_open_fd_write_count;
+ } else {
+ och_p = &lli->lli_mds_read_och;
+ och_usecount = &lli->lli_open_fd_read_count;
+ }
+
+ /* The file may have been open by another process (broken lease) so
+ * *och_p is not NULL. In this case we should simply increase usecount
+ * and close fd_och.
+ */
+ if (*och_p != NULL) {
+ old_och = fd->fd_och;
+ (*och_usecount)++;
+ } else {
+ *och_p = fd->fd_och;
+ *och_usecount = 1;
+ }
+ fd->fd_och = NULL;
+ mutex_unlock(&lli->lli_och_mutex);
+
+ if (old_och != NULL)
+ rc = ll_close_inode_openhandle(inode, old_och, 0, NULL);
+
+ RETURN(rc);
+}
+
+/**
* Acquire a lease and open the file.
*/
static struct obd_client_handle *
RETURN(ERR_PTR(-EINVAL));
if (file != NULL) {
- struct ll_inode_info *lli = ll_i2info(inode);
- struct ll_file_data *fd = LUSTRE_FPRIVATE(file);
- struct obd_client_handle **och_p;
- __u64 *och_usecount;
-
if (!(fmode & file->f_mode) || (file->f_mode & FMODE_EXEC))
RETURN(ERR_PTR(-EPERM));
- /* Get the openhandle of the file */
- rc = -EBUSY;
- mutex_lock(&lli->lli_och_mutex);
- if (fd->fd_lease_och != NULL) {
- mutex_unlock(&lli->lli_och_mutex);
- RETURN(ERR_PTR(rc));
- }
-
- if (fd->fd_och == NULL) {
- if (file->f_mode & FMODE_WRITE) {
- LASSERT(lli->lli_mds_write_och != NULL);
- och_p = &lli->lli_mds_write_och;
- och_usecount = &lli->lli_open_fd_write_count;
- } else {
- LASSERT(lli->lli_mds_read_och != NULL);
- och_p = &lli->lli_mds_read_och;
- och_usecount = &lli->lli_open_fd_read_count;
- }
- if (*och_usecount == 1) {
- fd->fd_och = *och_p;
- *och_p = NULL;
- *och_usecount = 0;
- rc = 0;
- }
- }
- mutex_unlock(&lli->lli_och_mutex);
- if (rc < 0) /* more than 1 opener */
+ rc = ll_lease_och_acquire(inode, file, &old_handle);
+ if (rc)
RETURN(ERR_PTR(rc));
-
- LASSERT(fd->fd_och != NULL);
- old_handle = fd->fd_och->och_fh;
}
OBD_ALLOC_PTR(och);
/* already get lease, handle lease lock */
ll_set_lock_data(sbi->ll_md_exp, inode, &it, NULL);
- if (it.d.lustre.it_lock_mode == 0 ||
- it.d.lustre.it_lock_bits != MDS_INODELOCK_OPEN) {
+ if (it.it_lock_mode == 0 ||
+ it.it_lock_bits != MDS_INODELOCK_OPEN) {
/* open lock must return for lease */
- CERROR(DFID "lease granted but no open lock, %d/"LPU64".\n",
- PFID(ll_inode2fid(inode)), it.d.lustre.it_lock_mode,
- it.d.lustre.it_lock_bits);
+ CERROR(DFID "lease granted but no open lock, %d/%llu.\n",
+ PFID(ll_inode2fid(inode)), it.it_lock_mode,
+ it.it_lock_bits);
GOTO(out_close, rc = -EPROTO);
}
out_close:
/* Cancel open lock */
- if (it.d.lustre.it_lock_mode != 0) {
+ if (it.it_lock_mode != 0) {
ldlm_lock_decref_and_cancel(&och->och_lease_handle,
- it.d.lustre.it_lock_mode);
- it.d.lustre.it_lock_mode = 0;
+ it.it_lock_mode);
+ it.it_lock_mode = 0;
och->och_lease_handle.cookie = 0ULL;
}
- rc2 = ll_close_inode_openhandle(sbi->ll_md_exp, inode, och, NULL);
+ rc2 = ll_close_inode_openhandle(inode, och, 0, NULL);
if (rc2 < 0)
CERROR("%s: error closing file "DFID": %d\n",
ll_get_fsname(inode->i_sb, NULL, 0),
}
/**
+ * Check whether a layout swap can be done between two inodes.
+ *
+ * \param[in] inode1 First inode to check
+ * \param[in] inode2 Second inode to check
+ *
+ * \retval 0 on success, layout swap can be performed between both inodes
+ * \retval negative error code if requirements are not met
+ */
+static int ll_check_swap_layouts_validity(struct inode *inode1,
+ struct inode *inode2)
+{
+ if (!S_ISREG(inode1->i_mode) || !S_ISREG(inode2->i_mode))
+ return -EINVAL;
+
+ if (inode_permission(inode1, MAY_WRITE) ||
+ inode_permission(inode2, MAY_WRITE))
+ return -EPERM;
+
+ if (inode1->i_sb != inode2->i_sb)
+ return -EXDEV;
+
+ return 0;
+}
+
+static int ll_swap_layouts_close(struct obd_client_handle *och,
+ struct inode *inode, struct inode *inode2)
+{
+ const struct lu_fid *fid1 = ll_inode2fid(inode);
+ const struct lu_fid *fid2;
+ int rc;
+ ENTRY;
+
+ CDEBUG(D_INODE, "%s: biased close of file "DFID"\n",
+ ll_get_fsname(inode->i_sb, NULL, 0), PFID(fid1));
+
+ rc = ll_check_swap_layouts_validity(inode, inode2);
+ if (rc < 0)
+ GOTO(out_free_och, rc);
+
+ /* We now know that inode2 is a lustre inode */
+ fid2 = ll_inode2fid(inode2);
+
+ rc = lu_fid_cmp(fid1, fid2);
+ if (rc == 0)
+ GOTO(out_free_och, rc = -EINVAL);
+
+ /* Close the file and swap layouts between inode & inode2.
+ * NB: lease lock handle is released in mdc_close_layout_swap_pack()
+ * because we still need it to pack l_remote_handle to MDT. */
+ rc = ll_close_inode_openhandle(inode, och, MDS_CLOSE_LAYOUT_SWAP,
+ inode2);
+
+ och = NULL; /* freed in ll_close_inode_openhandle() */
+
+out_free_och:
+ if (och != NULL)
+ OBD_FREE_PTR(och);
+
+ RETURN(rc);
+}
+
+/**
* Release lease and close the file.
* It will check if the lease has ever broken.
*/
}
CDEBUG(D_INODE, "lease for "DFID" broken? %d\n",
- PFID(&ll_i2info(inode)->lli_fid), cancelled);
+ PFID(&ll_i2info(inode)->lli_fid), cancelled);
if (!cancelled)
ldlm_cli_cancel(&och->och_lease_handle, 0);
+
if (lease_broken != NULL)
*lease_broken = cancelled;
- rc = ll_close_inode_openhandle(ll_i2sbi(inode)->ll_md_exp, inode, och,
- NULL);
+ rc = ll_close_inode_openhandle(inode, och, 0, NULL);
RETURN(rc);
}
-/* Fills the obdo with the attributes for the lsm */
-static int ll_lsm_getattr(struct lov_stripe_md *lsm, struct obd_export *exp,
- struct obd_capa *capa, struct obdo *obdo,
- __u64 ioepoch, int dv_flags)
-{
- struct ptlrpc_request_set *set;
- struct obd_info oinfo = { { { 0 } } };
- int rc;
-
- ENTRY;
-
- LASSERT(lsm != NULL);
-
- oinfo.oi_md = lsm;
- oinfo.oi_oa = obdo;
- oinfo.oi_oa->o_oi = lsm->lsm_oi;
- oinfo.oi_oa->o_mode = S_IFREG;
- oinfo.oi_oa->o_ioepoch = ioepoch;
- oinfo.oi_oa->o_valid = OBD_MD_FLID | OBD_MD_FLTYPE |
- OBD_MD_FLSIZE | OBD_MD_FLBLOCKS |
- OBD_MD_FLBLKSZ | OBD_MD_FLATIME |
- OBD_MD_FLMTIME | OBD_MD_FLCTIME |
- OBD_MD_FLGROUP | OBD_MD_FLEPOCH |
- OBD_MD_FLDATAVERSION;
- oinfo.oi_capa = capa;
- if (dv_flags & (LL_DV_WR_FLUSH | LL_DV_RD_FLUSH)) {
- oinfo.oi_oa->o_valid |= OBD_MD_FLFLAGS;
- oinfo.oi_oa->o_flags |= OBD_FL_SRVLOCK;
- if (dv_flags & LL_DV_WR_FLUSH)
- oinfo.oi_oa->o_flags |= OBD_FL_FLUSH;
- }
-
- set = ptlrpc_prep_set();
- if (set == NULL) {
- CERROR("cannot allocate ptlrpc set: rc = %d\n", -ENOMEM);
- rc = -ENOMEM;
- } else {
- rc = obd_getattr_async(exp, &oinfo, set);
- if (rc == 0)
- rc = ptlrpc_set_wait(set);
- ptlrpc_set_destroy(set);
- }
- if (rc == 0) {
- oinfo.oi_oa->o_valid &= (OBD_MD_FLBLOCKS | OBD_MD_FLBLKSZ |
- OBD_MD_FLATIME | OBD_MD_FLMTIME |
- OBD_MD_FLCTIME | OBD_MD_FLSIZE |
- OBD_MD_FLDATAVERSION | OBD_MD_FLFLAGS);
- if (dv_flags & LL_DV_WR_FLUSH &&
- !(oinfo.oi_oa->o_valid & OBD_MD_FLFLAGS &&
- oinfo.oi_oa->o_flags & OBD_FL_FLUSH))
- RETURN(-ENOTSUPP);
- }
- RETURN(rc);
-}
-
-/**
- * Performs the getattr on the inode and updates its fields.
- * If @sync != 0, perform the getattr under the server-side lock.
- */
-int ll_inode_getattr(struct inode *inode, struct obdo *obdo,
- __u64 ioepoch, int sync)
-{
- struct obd_capa *capa = ll_mdscapa_get(inode);
- struct lov_stripe_md *lsm;
- int rc;
- ENTRY;
-
- lsm = ccc_inode_lsm_get(inode);
- rc = ll_lsm_getattr(lsm, ll_i2dtexp(inode),
- capa, obdo, ioepoch, sync ? LL_DV_RD_FLUSH : 0);
- capa_put(capa);
- if (rc == 0) {
- struct ost_id *oi = lsm ? &lsm->lsm_oi : &obdo->o_oi;
-
- obdo_refresh_inode(inode, obdo, obdo->o_valid);
- CDEBUG(D_INODE, "objid "DOSTID" size %llu, blocks %llu,"
- " blksize %lu\n", POSTID(oi), i_size_read(inode),
- (unsigned long long)inode->i_blocks,
- 1UL << inode->i_blkbits);
- }
- ccc_inode_lsm_put(inode, lsm);
- RETURN(rc);
-}
-
-int ll_merge_lvb(const struct lu_env *env, struct inode *inode)
+int ll_merge_attr(const struct lu_env *env, struct inode *inode)
{
struct ll_inode_info *lli = ll_i2info(inode);
struct cl_object *obj = lli->lli_clob;
- struct cl_attr *attr = ccc_env_thread_attr(env);
- struct ost_lvb lvb;
+ struct cl_attr *attr = vvp_env_thread_attr(env);
+ s64 atime;
+ s64 mtime;
+ s64 ctime;
int rc = 0;
ENTRY;
ll_inode_size_lock(inode);
- /* merge timestamps the most recently obtained from mds with
- timestamps obtained from osts */
- LTIME_S(inode->i_atime) = lli->lli_lvb.lvb_atime;
- LTIME_S(inode->i_mtime) = lli->lli_lvb.lvb_mtime;
- LTIME_S(inode->i_ctime) = lli->lli_lvb.lvb_ctime;
-
- lvb.lvb_size = i_size_read(inode);
- lvb.lvb_blocks = inode->i_blocks;
- lvb.lvb_mtime = LTIME_S(inode->i_mtime);
- lvb.lvb_atime = LTIME_S(inode->i_atime);
- lvb.lvb_ctime = LTIME_S(inode->i_ctime);
+
+ /* Merge timestamps the most recently obtained from MDS with
+ * timestamps obtained from OSTs.
+ *
+ * Do not overwrite atime of inode because it may be refreshed
+ * by file_accessed() function. If the read was served by cache
+ * data, there is no RPC to be sent so that atime may not be
+ * transferred to OSTs at all. MDT only updates atime at close time
+ * if it's at least 'mdd.*.atime_diff' older.
+ * All in all, the atime in Lustre does not strictly comply with
+ * POSIX. Solving this problem needs to send an RPC to MDT for each
+ * read, this will hurt performance. */
+ if (LTIME_S(inode->i_atime) < lli->lli_atime)
+ LTIME_S(inode->i_atime) = lli->lli_atime;
+ LTIME_S(inode->i_mtime) = lli->lli_mtime;
+ LTIME_S(inode->i_ctime) = lli->lli_ctime;
+
+ atime = LTIME_S(inode->i_atime);
+ mtime = LTIME_S(inode->i_mtime);
+ ctime = LTIME_S(inode->i_ctime);
cl_object_attr_lock(obj);
rc = cl_object_attr_get(env, obj, attr);
cl_object_attr_unlock(obj);
- if (rc == 0) {
- if (lvb.lvb_atime < attr->cat_atime)
- lvb.lvb_atime = attr->cat_atime;
- if (lvb.lvb_ctime < attr->cat_ctime)
- lvb.lvb_ctime = attr->cat_ctime;
- if (lvb.lvb_mtime < attr->cat_mtime)
- lvb.lvb_mtime = attr->cat_mtime;
-
- CDEBUG(D_VFSTRACE, DFID" updating i_size "LPU64"\n",
- PFID(&lli->lli_fid), attr->cat_size);
- cl_isize_write_nolock(inode, attr->cat_size);
-
- inode->i_blocks = attr->cat_blocks;
-
- LTIME_S(inode->i_mtime) = lvb.lvb_mtime;
- LTIME_S(inode->i_atime) = lvb.lvb_atime;
- LTIME_S(inode->i_ctime) = lvb.lvb_ctime;
- }
- ll_inode_size_unlock(inode);
+ if (rc != 0)
+ GOTO(out_size_unlock, rc);
- RETURN(rc);
-}
+ if (atime < attr->cat_atime)
+ atime = attr->cat_atime;
-int ll_glimpse_ioctl(struct ll_sb_info *sbi, struct lov_stripe_md *lsm,
- lstat_t *st)
-{
- struct obdo obdo = { 0 };
- int rc;
+ if (ctime < attr->cat_ctime)
+ ctime = attr->cat_ctime;
- rc = ll_lsm_getattr(lsm, sbi->ll_dt_exp, NULL, &obdo, 0, 0);
- if (rc == 0) {
- st->st_size = obdo.o_size;
- st->st_blocks = obdo.o_blocks;
- st->st_mtime = obdo.o_mtime;
- st->st_atime = obdo.o_atime;
- st->st_ctime = obdo.o_ctime;
- }
- return rc;
+ if (mtime < attr->cat_mtime)
+ mtime = attr->cat_mtime;
+
+ CDEBUG(D_VFSTRACE, DFID" updating i_size %llu\n",
+ PFID(&lli->lli_fid), attr->cat_size);
+
+ i_size_write(inode, attr->cat_size);
+ inode->i_blocks = attr->cat_blocks;
+
+ LTIME_S(inode->i_atime) = atime;
+ LTIME_S(inode->i_mtime) = mtime;
+ LTIME_S(inode->i_ctime) = ctime;
+
+out_size_unlock:
+ ll_inode_size_unlock(inode);
+
+ RETURN(rc);
}
static bool file_is_noatime(const struct file *file)
{
const struct vfsmount *mnt = file->f_path.mnt;
- const struct inode *inode = file->f_path.dentry->d_inode;
+ const struct inode *inode = file_inode((struct file *)file);
/* Adapted from file_accessed() and touch_atime().*/
if (file->f_flags & O_NOATIME)
static void ll_io_init(struct cl_io *io, const struct file *file, int write)
{
- struct inode *inode = file->f_dentry->d_inode;
+ struct inode *inode = file_inode((struct file *)file);
io->u.ci_rw.crw_nonblock = file->f_flags & O_NONBLOCK;
if (write) {
struct file *file, enum cl_io_type iot,
loff_t *ppos, size_t count)
{
- struct ll_inode_info *lli = ll_i2info(file->f_dentry->d_inode);
- struct ll_file_data *fd = LUSTRE_FPRIVATE(file);
- struct cl_io *io;
- ssize_t result;
- struct range_lock range;
+ struct vvp_io *vio = vvp_env_io(env);
+ struct inode *inode = file_inode(file);
+ struct ll_inode_info *lli = ll_i2info(inode);
+ struct ll_file_data *fd = LUSTRE_FPRIVATE(file);
+ struct cl_io *io;
+ ssize_t result = 0;
+ int rc = 0;
+ struct range_lock range;
+
ENTRY;
- CDEBUG(D_VFSTRACE, "file: %s, type: %d ppos: "LPU64", count: %zd\n",
- file->f_dentry->d_name.name, iot, *ppos, count);
+ CDEBUG(D_VFSTRACE, "file: %s, type: %d ppos: %llu, count: %zu\n",
+ file_dentry(file)->d_name.name, iot, *ppos, count);
restart:
- io = ccc_env_thread_io(env);
- ll_io_init(io, file, iot == CIT_WRITE);
+ io = vvp_env_thread_io(env);
+ ll_io_init(io, file, iot == CIT_WRITE);
- if (cl_io_rw_init(env, io, iot, *ppos, count) == 0) {
- struct vvp_io *vio = vvp_env_io(env);
- struct ccc_io *cio = ccc_env_io(env);
+ if (cl_io_rw_init(env, io, iot, *ppos, count) == 0) {
bool range_locked = false;
if (file->f_flags & O_APPEND)
range_lock_init(&range, 0, LUSTRE_EOF);
else
range_lock_init(&range, *ppos, *ppos + count - 1);
- cio->cui_fd = LUSTRE_FPRIVATE(file);
- vio->cui_io_subtype = args->via_io_subtype;
-
- switch (vio->cui_io_subtype) {
- case IO_NORMAL:
- cio->cui_iov = args->u.normal.via_iov;
- cio->cui_nrsegs = args->u.normal.via_nrsegs;
- cio->cui_tot_nrsegs = cio->cui_nrsegs;
- cio->cui_iocb = args->u.normal.via_iocb;
- if ((iot == CIT_WRITE) &&
- !(cio->cui_fd->fd_flags & LL_FILE_GROUP_LOCKED)) {
+
+ vio->vui_fd = LUSTRE_FPRIVATE(file);
+ vio->vui_io_subtype = args->via_io_subtype;
+
+ switch (vio->vui_io_subtype) {
+ case IO_NORMAL:
+ vio->vui_iter = args->u.normal.via_iter;
+ vio->vui_iocb = args->u.normal.via_iocb;
+ /* Direct IO reads must also take range lock,
+ * or multiple reads will try to work on the same pages
+ * See LU-6227 for details. */
+ if (((iot == CIT_WRITE) ||
+ (iot == CIT_READ && (file->f_flags & O_DIRECT))) &&
+ !(vio->vui_fd->fd_flags & LL_FILE_GROUP_LOCKED)) {
CDEBUG(D_VFSTRACE, "Range lock "RL_FMT"\n",
RL_PARA(&range));
- result = range_lock(&lli->lli_write_tree,
- &range);
- if (result < 0)
- GOTO(out, result);
+ rc = range_lock(&lli->lli_write_tree, &range);
+ if (rc < 0)
+ GOTO(out, rc);
range_locked = true;
}
- down_read(&lli->lli_trunc_sem);
- break;
- case IO_SPLICE:
- vio->u.splice.cui_pipe = args->u.splice.via_pipe;
- vio->u.splice.cui_flags = args->u.splice.via_flags;
- break;
- default:
- CERROR("Unknow IO type - %u\n", vio->cui_io_subtype);
- LBUG();
- }
+ break;
+ case IO_SPLICE:
+ vio->u.splice.vui_pipe = args->u.splice.via_pipe;
+ vio->u.splice.vui_flags = args->u.splice.via_flags;
+ break;
+ default:
+ CERROR("unknown IO subtype %u\n", vio->vui_io_subtype);
+ LBUG();
+ }
- ll_cl_add(file, env, io);
- result = cl_io_loop(env, io);
+ ll_cl_add(file, env, io, LCC_RW);
+ rc = cl_io_loop(env, io);
ll_cl_remove(file, env);
- if (args->via_io_subtype == IO_NORMAL)
- up_read(&lli->lli_trunc_sem);
if (range_locked) {
CDEBUG(D_VFSTRACE, "Range unlock "RL_FMT"\n",
RL_PARA(&range));
range_unlock(&lli->lli_write_tree, &range);
}
- } else {
- /* cl_io_rw_init() handled IO */
- result = io->ci_result;
- }
+ } else {
+ /* cl_io_rw_init() handled IO */
+ rc = io->ci_result;
+ }
- if (io->ci_nob > 0) {
- result = io->ci_nob;
- *ppos = io->u.ci_wr.wr.crw_pos;
- }
- GOTO(out, result);
+ if (io->ci_nob > 0) {
+ result += io->ci_nob;
+ count -= io->ci_nob;
+ *ppos = io->u.ci_wr.wr.crw_pos; /* for splice */
+
+ /* prepare IO restart */
+ if (count > 0 && args->via_io_subtype == IO_NORMAL)
+ args->u.normal.via_iter = vio->vui_iter;
+ }
+ GOTO(out, rc);
out:
- cl_io_fini(env, io);
- /* If any bit been read/written (result != 0), we just return
- * short read/write instead of restart io. */
- if ((result == 0 || result == -ENODATA) && io->ci_need_restart) {
- CDEBUG(D_VFSTRACE, "Restart %s on %s from %lld, count:%zd\n",
+ cl_io_fini(env, io);
+
+ if ((rc == 0 || rc == -ENODATA) && count > 0 && io->ci_need_restart) {
+ CDEBUG(D_VFSTRACE,
+ "%s: restart %s from %lld, count:%zu, result: %zd\n",
+ file_dentry(file)->d_name.name,
iot == CIT_READ ? "read" : "write",
- file->f_dentry->d_name.name, *ppos, count);
- LASSERTF(io->ci_nob == 0, "%zd\n", io->ci_nob);
+ *ppos, count, result);
goto restart;
}
- if (iot == CIT_READ) {
- if (result >= 0)
- ll_stats_ops_tally(ll_i2sbi(file->f_dentry->d_inode),
- LPROC_LL_READ_BYTES, result);
- } else if (iot == CIT_WRITE) {
- if (result >= 0) {
- ll_stats_ops_tally(ll_i2sbi(file->f_dentry->d_inode),
- LPROC_LL_WRITE_BYTES, result);
+ if (iot == CIT_READ) {
+ if (result > 0)
+ ll_stats_ops_tally(ll_i2sbi(inode),
+ LPROC_LL_READ_BYTES, result);
+ } else if (iot == CIT_WRITE) {
+ if (result > 0) {
+ ll_stats_ops_tally(ll_i2sbi(inode),
+ LPROC_LL_WRITE_BYTES, result);
fd->fd_write_failed = false;
- } else if (result != -ERESTARTSYS) {
+ } else if (result == 0 && rc == 0) {
+ rc = io->ci_result;
+ if (rc < 0)
+ fd->fd_write_failed = true;
+ else
+ fd->fd_write_failed = false;
+ } else if (rc != -ERESTARTSYS) {
fd->fd_write_failed = true;
}
}
+
CDEBUG(D_VFSTRACE, "iot: %d, result: %zd\n", iot, result);
+ return result > 0 ? result : rc;
+}
+
+/**
+ * The purpose of fast read is to overcome per I/O overhead and improve IOPS
+ * especially for small I/O.
+ *
+ * To serve a read request, CLIO has to create and initialize a cl_io and
+ * then request DLM lock. This has turned out to have siginificant overhead
+ * and affects the performance of small I/O dramatically.
+ *
+ * It's not necessary to create a cl_io for each I/O. Under the help of read
+ * ahead, most of the pages being read are already in memory cache and we can
+ * read those pages directly because if the pages exist, the corresponding DLM
+ * lock must exist so that page content must be valid.
+ *
+ * In fast read implementation, the llite speculatively finds and reads pages
+ * in memory cache. There are three scenarios for fast read:
+ * - If the page exists and is uptodate, kernel VM will provide the data and
+ * CLIO won't be intervened;
+ * - If the page was brought into memory by read ahead, it will be exported
+ * and read ahead parameters will be updated;
+ * - Otherwise the page is not in memory, we can't do fast read. Therefore,
+ * it will go back and invoke normal read, i.e., a cl_io will be created
+ * and DLM lock will be requested.
+ *
+ * POSIX compliance: posix standard states that read is intended to be atomic.
+ * Lustre read implementation is in line with Linux kernel read implementation
+ * and neither of them complies with POSIX standard in this matter. Fast read
+ * doesn't make the situation worse on single node but it may interleave write
+ * results from multiple nodes due to short read handling in ll_file_aio_read().
+ *
+ * \param env - lu_env
+ * \param iocb - kiocb from kernel
+ * \param iter - user space buffers where the data will be copied
+ *
+ * \retval - number of bytes have been read, or error code if error occurred.
+ */
+static ssize_t
+ll_do_fast_read(const struct lu_env *env, struct kiocb *iocb,
+ struct iov_iter *iter)
+{
+ ssize_t result;
+
+ if (!ll_sbi_has_fast_read(ll_i2sbi(file_inode(iocb->ki_filp))))
+ return 0;
+
+ /* NB: we can't do direct IO for fast read because it will need a lock
+ * to make IO engine happy. */
+ if (iocb->ki_filp->f_flags & O_DIRECT)
+ return 0;
+
+ ll_cl_add(iocb->ki_filp, env, NULL, LCC_RW);
+ result = generic_file_read_iter(iocb, iter);
+ ll_cl_remove(iocb->ki_filp, env);
+
+ /* If the first page is not in cache, generic_file_aio_read() will be
+ * returned with -ENODATA.
+ * See corresponding code in ll_readpage(). */
+ if (result == -ENODATA)
+ result = 0;
+
+ if (result > 0)
+ ll_stats_ops_tally(ll_i2sbi(file_inode(iocb->ki_filp)),
+ LPROC_LL_READ_BYTES, result);
+
return result;
}
+/*
+ * Read from a file (through the page cache).
+ */
+static ssize_t ll_file_read_iter(struct kiocb *iocb, struct iov_iter *to)
+{
+ struct lu_env *env;
+ struct vvp_io_args *args;
+ ssize_t result;
+ ssize_t rc2;
+ __u16 refcheck;
+
+ env = cl_env_get(&refcheck);
+ if (IS_ERR(env))
+ return PTR_ERR(env);
+
+ result = ll_do_fast_read(env, iocb, to);
+ if (result < 0 || iov_iter_count(to) == 0)
+ GOTO(out, result);
+
+ args = ll_env_args(env, IO_NORMAL);
+ args->u.normal.via_iter = to;
+ args->u.normal.via_iocb = iocb;
+
+ rc2 = ll_file_io_generic(env, args, iocb->ki_filp, CIT_READ,
+ &iocb->ki_pos, iov_iter_count(to));
+ if (rc2 > 0)
+ result += rc2;
+ else if (result == 0)
+ result = rc2;
+
+out:
+ cl_env_put(env, &refcheck);
+ return result;
+}
+
+/*
+ * Write to a file (through the page cache).
+ */
+static ssize_t ll_file_write_iter(struct kiocb *iocb, struct iov_iter *from)
+{
+ struct vvp_io_args *args;
+ struct lu_env *env;
+ ssize_t result;
+ __u16 refcheck;
+
+ env = cl_env_get(&refcheck);
+ if (IS_ERR(env))
+ return PTR_ERR(env);
+
+ args = ll_env_args(env, IO_NORMAL);
+ args->u.normal.via_iter = from;
+ args->u.normal.via_iocb = iocb;
+
+ result = ll_file_io_generic(env, args, iocb->ki_filp, CIT_WRITE,
+ &iocb->ki_pos, iov_iter_count(from));
+ cl_env_put(env, &refcheck);
+ return result;
+}
+#ifndef HAVE_FILE_OPERATIONS_READ_WRITE_ITER
/*
* XXX: exact copy from kernel code (__generic_file_aio_write_nolock)
*/
static int ll_file_get_iov_count(const struct iovec *iov,
- unsigned long *nr_segs, size_t *count)
+ unsigned long *nr_segs, size_t *count)
{
- size_t cnt = 0;
- unsigned long seg;
+ size_t cnt = 0;
+ unsigned long seg;
- for (seg = 0; seg < *nr_segs; seg++) {
- const struct iovec *iv = &iov[seg];
+ for (seg = 0; seg < *nr_segs; seg++) {
+ const struct iovec *iv = &iov[seg];
- /*
- * If any segment has a negative length, or the cumulative
- * length ever wraps negative then return -EINVAL.
- */
- cnt += iv->iov_len;
- if (unlikely((ssize_t)(cnt|iv->iov_len) < 0))
- return -EINVAL;
- if (access_ok(VERIFY_READ, iv->iov_base, iv->iov_len))
- continue;
- if (seg == 0)
- return -EFAULT;
- *nr_segs = seg;
- cnt -= iv->iov_len; /* This segment is no good */
- break;
- }
- *count = cnt;
- return 0;
+ /*
+ * If any segment has a negative length, or the cumulative
+ * length ever wraps negative then return -EINVAL.
+ */
+ cnt += iv->iov_len;
+ if (unlikely((ssize_t)(cnt|iv->iov_len) < 0))
+ return -EINVAL;
+ if (access_ok(VERIFY_READ, iv->iov_base, iv->iov_len))
+ continue;
+ if (seg == 0)
+ return -EFAULT;
+ *nr_segs = seg;
+ cnt -= iv->iov_len; /* This segment is no good */
+ break;
+ }
+ *count = cnt;
+ return 0;
}
static ssize_t ll_file_aio_read(struct kiocb *iocb, const struct iovec *iov,
- unsigned long nr_segs, loff_t pos)
+ unsigned long nr_segs, loff_t pos)
{
- struct lu_env *env;
- struct vvp_io_args *args;
- size_t count;
- ssize_t result;
- int refcheck;
- ENTRY;
+ struct iov_iter to;
+ size_t iov_count;
+ ssize_t result;
+ ENTRY;
- result = ll_file_get_iov_count(iov, &nr_segs, &count);
- if (result)
- RETURN(result);
+ result = ll_file_get_iov_count(iov, &nr_segs, &iov_count);
+ if (result)
+ RETURN(result);
- env = cl_env_get(&refcheck);
- if (IS_ERR(env))
- RETURN(PTR_ERR(env));
+# ifdef HAVE_IOV_ITER_INIT_DIRECTION
+ iov_iter_init(&to, READ, iov, nr_segs, iov_count);
+# else /* !HAVE_IOV_ITER_INIT_DIRECTION */
+ iov_iter_init(&to, iov, nr_segs, iov_count, 0);
+# endif /* HAVE_IOV_ITER_INIT_DIRECTION */
- args = vvp_env_args(env, IO_NORMAL);
- args->u.normal.via_iov = (struct iovec *)iov;
- args->u.normal.via_nrsegs = nr_segs;
- args->u.normal.via_iocb = iocb;
+ result = ll_file_read_iter(iocb, &to);
- result = ll_file_io_generic(env, args, iocb->ki_filp, CIT_READ,
- &iocb->ki_pos, count);
- cl_env_put(env, &refcheck);
- RETURN(result);
+ RETURN(result);
}
-static ssize_t ll_file_read(struct file *file, char *buf, size_t count,
- loff_t *ppos)
+static ssize_t ll_file_read(struct file *file, char __user *buf, size_t count,
+ loff_t *ppos)
{
- struct lu_env *env;
- struct iovec *local_iov;
+ struct iovec iov = { .iov_base = buf, .iov_len = count };
struct kiocb *kiocb;
ssize_t result;
- int refcheck;
ENTRY;
- env = cl_env_get(&refcheck);
- if (IS_ERR(env))
- RETURN(PTR_ERR(env));
+ OBD_ALLOC_PTR(kiocb);
+ if (kiocb == NULL)
+ RETURN(-ENOMEM);
- local_iov = &vvp_env_info(env)->vti_local_iov;
- kiocb = &vvp_env_info(env)->vti_kiocb;
- local_iov->iov_base = (void __user *)buf;
- local_iov->iov_len = count;
init_sync_kiocb(kiocb, file);
kiocb->ki_pos = *ppos;
#ifdef HAVE_KIOCB_KI_LEFT
- kiocb->ki_left = count;
-#else
- kiocb->ki_nbytes = count;
+ kiocb->ki_left = count;
+#elif defined(HAVE_KI_NBYTES)
+ kiocb->ki_nbytes = count;
#endif
- result = ll_file_aio_read(kiocb, local_iov, 1, kiocb->ki_pos);
- *ppos = kiocb->ki_pos;
+ result = ll_file_aio_read(kiocb, &iov, 1, kiocb->ki_pos);
+ *ppos = kiocb->ki_pos;
- cl_env_put(env, &refcheck);
- RETURN(result);
+ OBD_FREE_PTR(kiocb);
+ RETURN(result);
}
/*
* AIO stuff
*/
static ssize_t ll_file_aio_write(struct kiocb *iocb, const struct iovec *iov,
- unsigned long nr_segs, loff_t pos)
+ unsigned long nr_segs, loff_t pos)
{
- struct lu_env *env;
- struct vvp_io_args *args;
- size_t count;
- ssize_t result;
- int refcheck;
- ENTRY;
+ struct iov_iter from;
+ size_t iov_count;
+ ssize_t result;
+ ENTRY;
- result = ll_file_get_iov_count(iov, &nr_segs, &count);
- if (result)
- RETURN(result);
+ result = ll_file_get_iov_count(iov, &nr_segs, &iov_count);
+ if (result)
+ RETURN(result);
- env = cl_env_get(&refcheck);
- if (IS_ERR(env))
- RETURN(PTR_ERR(env));
+# ifdef HAVE_IOV_ITER_INIT_DIRECTION
+ iov_iter_init(&from, WRITE, iov, nr_segs, iov_count);
+# else /* !HAVE_IOV_ITER_INIT_DIRECTION */
+ iov_iter_init(&from, iov, nr_segs, iov_count, 0);
+# endif /* HAVE_IOV_ITER_INIT_DIRECTION */
- args = vvp_env_args(env, IO_NORMAL);
- args->u.normal.via_iov = (struct iovec *)iov;
- args->u.normal.via_nrsegs = nr_segs;
- args->u.normal.via_iocb = iocb;
+ result = ll_file_write_iter(iocb, &from);
- result = ll_file_io_generic(env, args, iocb->ki_filp, CIT_WRITE,
- &iocb->ki_pos, count);
- cl_env_put(env, &refcheck);
- RETURN(result);
+ RETURN(result);
}
-static ssize_t ll_file_write(struct file *file, const char *buf, size_t count,
- loff_t *ppos)
+static ssize_t ll_file_write(struct file *file, const char __user *buf,
+ size_t count, loff_t *ppos)
{
- struct lu_env *env;
- struct iovec *local_iov;
+ struct lu_env *env;
+ struct iovec iov = { .iov_base = (void __user *)buf,
+ .iov_len = count };
struct kiocb *kiocb;
ssize_t result;
- int refcheck;
+ __u16 refcheck;
ENTRY;
env = cl_env_get(&refcheck);
if (IS_ERR(env))
RETURN(PTR_ERR(env));
- local_iov = &vvp_env_info(env)->vti_local_iov;
- kiocb = &vvp_env_info(env)->vti_kiocb;
- local_iov->iov_base = (void __user *)buf;
- local_iov->iov_len = count;
+ kiocb = &ll_env_info(env)->lti_kiocb;
init_sync_kiocb(kiocb, file);
kiocb->ki_pos = *ppos;
#ifdef HAVE_KIOCB_KI_LEFT
- kiocb->ki_left = count;
-#else
- kiocb->ki_nbytes = count;
+ kiocb->ki_left = count;
+#elif defined(HAVE_KI_NBYTES)
+ kiocb->ki_nbytes = count;
#endif
- result = ll_file_aio_write(kiocb, local_iov, 1, kiocb->ki_pos);
- *ppos = kiocb->ki_pos;
+ result = ll_file_aio_write(kiocb, &iov, 1, kiocb->ki_pos);
+ *ppos = kiocb->ki_pos;
- cl_env_put(env, &refcheck);
- RETURN(result);
+ cl_env_put(env, &refcheck);
+ RETURN(result);
}
+#endif /* !HAVE_FILE_OPERATIONS_READ_WRITE_ITER */
/*
* Send file content (through pagecache) somewhere with helper
struct lu_env *env;
struct vvp_io_args *args;
ssize_t result;
- int refcheck;
+ __u16 refcheck;
ENTRY;
env = cl_env_get(&refcheck);
if (IS_ERR(env))
RETURN(PTR_ERR(env));
- args = vvp_env_args(env, IO_SPLICE);
+ args = ll_env_args(env, IO_SPLICE);
args->u.splice.via_pipe = pipe;
args->u.splice.via_flags = flags;
RETURN(result);
}
-static int ll_lov_recreate(struct inode *inode, struct ost_id *oi,
- obd_count ost_idx)
-{
- struct obd_export *exp = ll_i2dtexp(inode);
- struct obd_trans_info oti = { 0 };
- struct obdo *oa = NULL;
- int lsm_size;
- int rc = 0;
- struct lov_stripe_md *lsm = NULL, *lsm2;
- ENTRY;
-
- OBDO_ALLOC(oa);
- if (oa == NULL)
- RETURN(-ENOMEM);
-
- lsm = ccc_inode_lsm_get(inode);
- if (!lsm_has_objects(lsm))
- GOTO(out, rc = -ENOENT);
-
- lsm_size = sizeof(*lsm) + (sizeof(struct lov_oinfo) *
- (lsm->lsm_stripe_count));
-
- OBD_ALLOC_LARGE(lsm2, lsm_size);
- if (lsm2 == NULL)
- GOTO(out, rc = -ENOMEM);
-
- oa->o_oi = *oi;
- oa->o_nlink = ost_idx;
- oa->o_flags |= OBD_FL_RECREATE_OBJS;
- oa->o_valid = OBD_MD_FLID | OBD_MD_FLFLAGS | OBD_MD_FLGROUP;
- obdo_from_inode(oa, inode, OBD_MD_FLTYPE | OBD_MD_FLATIME |
- OBD_MD_FLMTIME | OBD_MD_FLCTIME);
- obdo_set_parent_fid(oa, &ll_i2info(inode)->lli_fid);
- memcpy(lsm2, lsm, lsm_size);
- ll_inode_size_lock(inode);
- rc = obd_create(NULL, exp, oa, &lsm2, &oti);
- ll_inode_size_unlock(inode);
-
- OBD_FREE_LARGE(lsm2, lsm_size);
- GOTO(out, rc);
-out:
- ccc_inode_lsm_put(inode, lsm);
- OBDO_FREE(oa);
- return rc;
-}
-
-static int ll_lov_recreate_obj(struct inode *inode, unsigned long arg)
-{
- struct ll_recreate_obj ucreat;
- struct ost_id oi;
- ENTRY;
-
- if (!cfs_capable(CFS_CAP_SYS_ADMIN))
- RETURN(-EPERM);
-
- if (copy_from_user(&ucreat, (struct ll_recreate_obj *)arg,
- sizeof(ucreat)))
- RETURN(-EFAULT);
-
- ostid_set_seq_mdt0(&oi);
- ostid_set_id(&oi, ucreat.lrc_id);
- RETURN(ll_lov_recreate(inode, &oi, ucreat.lrc_ost_idx));
-}
-
-static int ll_lov_recreate_fid(struct inode *inode, unsigned long arg)
-{
- struct lu_fid fid;
- struct ost_id oi;
- obd_count ost_idx;
- ENTRY;
-
- if (!cfs_capable(CFS_CAP_SYS_ADMIN))
- RETURN(-EPERM);
-
- if (copy_from_user(&fid, (struct lu_fid *)arg, sizeof(fid)))
- RETURN(-EFAULT);
-
- fid_to_ostid(&fid, &oi);
- ost_idx = (fid_seq(&fid) >> 16) & 0xffff;
- RETURN(ll_lov_recreate(inode, &oi, ost_idx));
-}
-
int ll_lov_setstripe_ea_info(struct inode *inode, struct file *file,
__u64 flags, struct lov_user_md *lum,
int lum_size)
{
- struct lov_stripe_md *lsm = NULL;
- struct lookup_intent oit = {.it_op = IT_OPEN, .it_flags = flags};
- int rc = 0;
+ struct lookup_intent oit = {
+ .it_op = IT_OPEN,
+ .it_flags = flags | MDS_OPEN_BY_FID,
+ };
+ int rc;
ENTRY;
- lsm = ccc_inode_lsm_get(inode);
- if (lsm != NULL) {
- ccc_inode_lsm_put(inode, lsm);
- CDEBUG(D_IOCTL, "stripe already exists for inode "DFID"\n",
- PFID(ll_inode2fid(inode)));
- GOTO(out, rc = -EEXIST);
- }
-
ll_inode_size_lock(inode);
- oit.it_flags |= MDS_OPEN_BY_FID;
rc = ll_intent_file_open(file, lum, lum_size, &oit);
- if (rc)
- GOTO(out_unlock, rc);
- rc = oit.d.lustre.it_status;
if (rc < 0)
- GOTO(out_req_free, rc);
+ GOTO(out_unlock, rc);
- ll_release_openhandle(file->f_dentry, &oit);
+ ll_release_openhandle(file_dentry(file), &oit);
out_unlock:
ll_inode_size_unlock(inode);
ll_intent_release(&oit);
- ccc_inode_lsm_put(inode, lsm);
-out:
cl_lov_delay_create_clear(&file->f_flags);
+
RETURN(rc);
-out_req_free:
- ptlrpc_req_finished((struct ptlrpc_request *) oit.d.lustre.it_data);
- goto out;
}
int ll_lov_getstripe_ea_info(struct inode *inode, const char *filename,
if (lump == NULL)
RETURN(-ENOMEM);
- if (copy_from_user(lump, (struct lov_user_md *)arg, lum_size)) {
- OBD_FREE_LARGE(lump, lum_size);
- RETURN(-EFAULT);
- }
+ if (copy_from_user(lump, (struct lov_user_md __user *)arg, lum_size))
+ GOTO(out_lump, rc = -EFAULT);
rc = ll_lov_setstripe_ea_info(inode, file, flags, lump, lum_size);
+out_lump:
OBD_FREE_LARGE(lump, lum_size);
RETURN(rc);
}
+static int ll_file_getstripe(struct inode *inode,
+ struct lov_user_md __user *lum)
+{
+ struct lu_env *env;
+ __u16 refcheck;
+ int rc;
+ ENTRY;
+
+ env = cl_env_get(&refcheck);
+ if (IS_ERR(env))
+ RETURN(PTR_ERR(env));
+
+ rc = cl_object_getstripe(env, ll_i2info(inode)->lli_clob, lum);
+ cl_env_put(env, &refcheck);
+ RETURN(rc);
+}
+
static int ll_lov_setstripe(struct inode *inode, struct file *file,
unsigned long arg)
{
- struct lov_user_md_v3 lumv3;
- struct lov_user_md_v1 *lumv1 = (struct lov_user_md_v1 *)&lumv3;
- struct lov_user_md_v1 *lumv1p = (struct lov_user_md_v1 *)arg;
- struct lov_user_md_v3 *lumv3p = (struct lov_user_md_v3 *)arg;
- int lum_size, rc;
- __u64 flags = FMODE_WRITE;
+ struct lov_user_md __user *lum = (struct lov_user_md __user *)arg;
+ struct lov_user_md *klum;
+ int lum_size, rc;
+ __u64 flags = FMODE_WRITE;
ENTRY;
- /* first try with v1 which is smaller than v3 */
- lum_size = sizeof(struct lov_user_md_v1);
- if (copy_from_user(lumv1, lumv1p, lum_size))
- RETURN(-EFAULT);
-
- if (lumv1->lmm_magic == LOV_USER_MAGIC_V3) {
- lum_size = sizeof(struct lov_user_md_v3);
- if (copy_from_user(&lumv3, lumv3p, lum_size))
- RETURN(-EFAULT);
- }
+ rc = ll_copy_user_md(lum, &klum);
+ if (rc < 0)
+ RETURN(rc);
- rc = ll_lov_setstripe_ea_info(inode, file, flags, lumv1, lum_size);
+ lum_size = rc;
+ rc = ll_lov_setstripe_ea_info(inode, file, flags, klum, lum_size);
if (rc == 0) {
- struct lov_stripe_md *lsm;
__u32 gen;
- put_user(0, &lumv1p->lmm_stripe_count);
+ put_user(0, &lum->lmm_stripe_count);
ll_layout_refresh(inode, &gen);
- lsm = ccc_inode_lsm_get(inode);
- rc = obd_iocontrol(LL_IOC_LOV_GETSTRIPE, ll_i2dtexp(inode),
- 0, lsm, (void *)arg);
- ccc_inode_lsm_put(inode, lsm);
+ rc = ll_file_getstripe(inode, (struct lov_user_md __user *)arg);
}
- RETURN(rc);
-}
-
-static int ll_lov_getstripe(struct inode *inode, unsigned long arg)
-{
- struct lov_stripe_md *lsm;
- int rc = -ENODATA;
- ENTRY;
- lsm = ccc_inode_lsm_get(inode);
- if (lsm != NULL)
- rc = obd_iocontrol(LL_IOC_LOV_GETSTRIPE, ll_i2dtexp(inode), 0,
- lsm, (void *)arg);
- ccc_inode_lsm_put(inode, lsm);
+ OBD_FREE(klum, lum_size);
RETURN(rc);
}
{
struct ll_inode_info *lli = ll_i2info(inode);
struct ll_file_data *fd = LUSTRE_FPRIVATE(file);
- struct ccc_grouplock grouplock;
+ struct ll_grouplock grouplock;
int rc;
ENTRY;
+ if (arg == 0) {
+ CWARN("group id for group lock must not be 0\n");
+ RETURN(-EINVAL);
+ }
+
if (ll_file_nolock(file))
RETURN(-EOPNOTSUPP);
spin_lock(&lli->lli_lock);
if (fd->fd_flags & LL_FILE_GROUP_LOCKED) {
CWARN("group lock already existed with gid %lu\n",
- fd->fd_grouplock.cg_gid);
+ fd->fd_grouplock.lg_gid);
spin_unlock(&lli->lli_lock);
RETURN(-EINVAL);
}
- LASSERT(fd->fd_grouplock.cg_lock == NULL);
+ LASSERT(fd->fd_grouplock.lg_lock == NULL);
spin_unlock(&lli->lli_lock);
- rc = cl_get_grouplock(cl_i2info(inode)->lli_clob,
+ rc = cl_get_grouplock(ll_i2info(inode)->lli_clob,
arg, (file->f_flags & O_NONBLOCK), &grouplock);
if (rc)
RETURN(rc);
RETURN(0);
}
-int ll_put_grouplock(struct inode *inode, struct file *file, unsigned long arg)
+static int ll_put_grouplock(struct inode *inode, struct file *file,
+ unsigned long arg)
{
struct ll_inode_info *lli = ll_i2info(inode);
struct ll_file_data *fd = LUSTRE_FPRIVATE(file);
- struct ccc_grouplock grouplock;
+ struct ll_grouplock grouplock;
ENTRY;
spin_lock(&lli->lli_lock);
CWARN("no group lock held\n");
RETURN(-EINVAL);
}
- LASSERT(fd->fd_grouplock.cg_lock != NULL);
- if (fd->fd_grouplock.cg_gid != arg) {
- CWARN("group lock %lu doesn't match current id %lu\n",
- arg, fd->fd_grouplock.cg_gid);
+ LASSERT(fd->fd_grouplock.lg_lock != NULL);
+
+ if (fd->fd_grouplock.lg_gid != arg) {
+ CWARN("group lock %lu doesn't match current id %lu\n",
+ arg, fd->fd_grouplock.lg_gid);
spin_unlock(&lli->lli_lock);
RETURN(-EINVAL);
}
ll_och_fill(ll_i2sbi(inode)->ll_md_exp, it, och);
- rc = ll_close_inode_openhandle(ll_i2sbi(inode)->ll_md_exp,
- inode, och, NULL);
+ rc = ll_close_inode_openhandle(inode, och, 0, NULL);
out:
/* this one is in place of ll_file_open */
if (it_disposition(it, DISP_ENQ_OPEN_REF)) {
- ptlrpc_req_finished(it->d.lustre.it_data);
+ ptlrpc_req_finished(it->it_request);
it_clear_disposition(it, DISP_ENQ_OPEN_REF);
}
RETURN(rc);
/**
* Get size for inode for which FIEMAP mapping is requested.
* Make the FIEMAP get_info call and returns the result.
+ * \param fiemap kernel buffer to hold extens
+ * \param num_bytes kernel buffer size
*/
-static int ll_do_fiemap(struct inode *inode, struct ll_user_fiemap *fiemap,
+static int ll_do_fiemap(struct inode *inode, struct fiemap *fiemap,
size_t num_bytes)
{
- struct obd_export *exp = ll_i2dtexp(inode);
- struct lov_stripe_md *lsm = NULL;
- struct ll_fiemap_info_key fm_key = { .name = KEY_FIEMAP, };
- __u32 vallen = num_bytes;
- int rc;
- ENTRY;
+ struct lu_env *env;
+ __u16 refcheck;
+ int rc = 0;
+ struct ll_fiemap_info_key fmkey = { .lfik_name = KEY_FIEMAP, };
+ ENTRY;
- /* Checks for fiemap flags */
- if (fiemap->fm_flags & ~LUSTRE_FIEMAP_FLAGS_COMPAT) {
- fiemap->fm_flags &= ~LUSTRE_FIEMAP_FLAGS_COMPAT;
- return -EBADR;
- }
+ /* Checks for fiemap flags */
+ if (fiemap->fm_flags & ~LUSTRE_FIEMAP_FLAGS_COMPAT) {
+ fiemap->fm_flags &= ~LUSTRE_FIEMAP_FLAGS_COMPAT;
+ return -EBADR;
+ }
- /* Check for FIEMAP_FLAG_SYNC */
- if (fiemap->fm_flags & FIEMAP_FLAG_SYNC) {
- rc = filemap_fdatawrite(inode->i_mapping);
- if (rc)
- return rc;
- }
+ /* Check for FIEMAP_FLAG_SYNC */
+ if (fiemap->fm_flags & FIEMAP_FLAG_SYNC) {
+ rc = filemap_fdatawrite(inode->i_mapping);
+ if (rc)
+ return rc;
+ }
- lsm = ccc_inode_lsm_get(inode);
- if (lsm == NULL)
- return -ENOENT;
+ env = cl_env_get(&refcheck);
+ if (IS_ERR(env))
+ RETURN(PTR_ERR(env));
- /* If the stripe_count > 1 and the application does not understand
- * DEVICE_ORDER flag, then it cannot interpret the extents correctly.
- */
- if (lsm->lsm_stripe_count > 1 &&
- !(fiemap->fm_flags & FIEMAP_FLAG_DEVICE_ORDER))
- GOTO(out, rc = -EOPNOTSUPP);
-
- fm_key.oa.o_oi = lsm->lsm_oi;
- fm_key.oa.o_valid = OBD_MD_FLID | OBD_MD_FLGROUP;
-
- obdo_from_inode(&fm_key.oa, inode, OBD_MD_FLSIZE);
- obdo_set_parent_fid(&fm_key.oa, &ll_i2info(inode)->lli_fid);
- /* If filesize is 0, then there would be no objects for mapping */
- if (fm_key.oa.o_size == 0) {
- fiemap->fm_mapped_extents = 0;
- GOTO(out, rc = 0);
- }
+ if (i_size_read(inode) == 0) {
+ rc = ll_glimpse_size(inode);
+ if (rc)
+ GOTO(out, rc);
+ }
+
+ fmkey.lfik_oa.o_valid = OBD_MD_FLID | OBD_MD_FLGROUP;
+ obdo_from_inode(&fmkey.lfik_oa, inode, OBD_MD_FLSIZE);
+ obdo_set_parent_fid(&fmkey.lfik_oa, &ll_i2info(inode)->lli_fid);
- memcpy(&fm_key.fiemap, fiemap, sizeof(*fiemap));
+ /* If filesize is 0, then there would be no objects for mapping */
+ if (fmkey.lfik_oa.o_size == 0) {
+ fiemap->fm_mapped_extents = 0;
+ GOTO(out, rc = 0);
+ }
- rc = obd_get_info(NULL, exp, sizeof(fm_key), &fm_key, &vallen,
- fiemap, lsm);
- if (rc)
- CERROR("obd_get_info failed: rc = %d\n", rc);
+ fmkey.lfik_fiemap = *fiemap;
+ rc = cl_object_fiemap(env, ll_i2info(inode)->lli_clob,
+ &fmkey, fiemap, &num_bytes);
out:
- ccc_inode_lsm_put(inode, lsm);
+ cl_env_put(env, &refcheck);
RETURN(rc);
}
if (copy_from_user(gfout, arg, sizeof(*gfout)))
GOTO(gf_free, rc = -EFAULT);
+ /* append root FID after gfout to let MDT know the root FID so that it
+ * can lookup the correct path, this is mainly for fileset.
+ * old server without fileset mount support will ignore this. */
+ *gfout->gf_u.gf_root_fid = *ll_inode2fid(inode);
/* Call mdc_iocontrol */
rc = obd_iocontrol(OBD_IOC_FID2PATH, exp, outsize, gfout, NULL);
RETURN(rc);
}
-static int ll_ioctl_fiemap(struct inode *inode, unsigned long arg)
-{
- struct ll_user_fiemap *fiemap_s;
- size_t num_bytes, ret_bytes;
- unsigned int extent_count;
- int rc = 0;
-
- /* Get the extent count so we can calculate the size of
- * required fiemap buffer */
- if (get_user(extent_count,
- &((struct ll_user_fiemap __user *)arg)->fm_extent_count))
- RETURN(-EFAULT);
-
- if (extent_count >=
- (SIZE_MAX - sizeof(*fiemap_s)) / sizeof(struct ll_fiemap_extent))
- RETURN(-EINVAL);
- num_bytes = sizeof(*fiemap_s) + (extent_count *
- sizeof(struct ll_fiemap_extent));
-
- OBD_ALLOC_LARGE(fiemap_s, num_bytes);
- if (fiemap_s == NULL)
- RETURN(-ENOMEM);
-
- /* get the fiemap value */
- if (copy_from_user(fiemap_s, (struct ll_user_fiemap __user *)arg,
- sizeof(*fiemap_s)))
- GOTO(error, rc = -EFAULT);
-
- /* If fm_extent_count is non-zero, read the first extent since
- * it is used to calculate end_offset and device from previous
- * fiemap call. */
- if (extent_count) {
- if (copy_from_user(&fiemap_s->fm_extents[0],
- (char __user *)arg + sizeof(*fiemap_s),
- sizeof(struct ll_fiemap_extent)))
- GOTO(error, rc = -EFAULT);
- }
-
- rc = ll_do_fiemap(inode, fiemap_s, num_bytes);
- if (rc)
- GOTO(error, rc);
-
- ret_bytes = sizeof(struct ll_user_fiemap);
-
- if (extent_count != 0)
- ret_bytes += (fiemap_s->fm_mapped_extents *
- sizeof(struct ll_fiemap_extent));
-
- if (copy_to_user((void *)arg, fiemap_s, ret_bytes))
- rc = -EFAULT;
-
-error:
- OBD_FREE_LARGE(fiemap_s, num_bytes);
- RETURN(rc);
-}
-
/*
* Read the data_version for inode.
*
* This value is computed using stripe object version on OST.
* Version is computed using server side locking.
*
- * @param sync if do sync on the OST side;
+ * @param flags if do sync on the OST side;
* 0: no sync
* LL_DV_RD_FLUSH: flush dirty pages, LCK_PR on OSTs
* LL_DV_WR_FLUSH: drop all caching pages, LCK_PW on OSTs
*/
int ll_data_version(struct inode *inode, __u64 *data_version, int flags)
{
- struct lov_stripe_md *lsm = NULL;
- struct ll_sb_info *sbi = ll_i2sbi(inode);
- struct obdo *obdo = NULL;
- int rc;
+ struct cl_object *obj = ll_i2info(inode)->lli_clob;
+ struct lu_env *env;
+ struct cl_io *io;
+ __u16 refcheck;
+ int result;
+
ENTRY;
- /* If no stripe, we consider version is 0. */
- lsm = ccc_inode_lsm_get(inode);
- if (!lsm_has_objects(lsm)) {
+ /* If no file object initialized, we consider its version is 0. */
+ if (obj == NULL) {
*data_version = 0;
- CDEBUG(D_INODE, "No object for inode\n");
- GOTO(out, rc = 0);
+ RETURN(0);
}
- OBD_ALLOC_PTR(obdo);
- if (obdo == NULL)
- GOTO(out, rc = -ENOMEM);
+ env = cl_env_get(&refcheck);
+ if (IS_ERR(env))
+ RETURN(PTR_ERR(env));
- rc = ll_lsm_getattr(lsm, sbi->ll_dt_exp, NULL, obdo, 0, flags);
- if (rc == 0) {
- if (!(obdo->o_valid & OBD_MD_FLDATAVERSION))
- rc = -EOPNOTSUPP;
- else
- *data_version = obdo->o_data_version;
- }
+ io = vvp_env_thread_io(env);
+ io->ci_obj = obj;
+ io->u.ci_data_version.dv_data_version = 0;
+ io->u.ci_data_version.dv_flags = flags;
- OBD_FREE_PTR(obdo);
- EXIT;
-out:
- ccc_inode_lsm_put(inode, lsm);
- RETURN(rc);
+restart:
+ if (cl_io_init(env, io, CIT_DATA_VERSION, io->ci_obj) == 0)
+ result = cl_io_loop(env, io);
+ else
+ result = io->ci_result;
+
+ *data_version = io->u.ci_data_version.dv_data_version;
+
+ cl_io_fini(env, io);
+
+ if (unlikely(io->ci_need_restart))
+ goto restart;
+
+ cl_env_put(env, &refcheck);
+
+ RETURN(result);
}
/*
*/
int ll_hsm_release(struct inode *inode)
{
- struct cl_env_nest nest;
struct lu_env *env;
struct obd_client_handle *och = NULL;
__u64 data_version = 0;
int rc;
+ __u16 refcheck;
ENTRY;
CDEBUG(D_INODE, "%s: Releasing file "DFID".\n",
if (rc != 0)
GOTO(out, rc);
- env = cl_env_nested_get(&nest);
+ env = cl_env_get(&refcheck);
if (IS_ERR(env))
GOTO(out, rc = PTR_ERR(env));
- ll_merge_lvb(env, inode);
- cl_env_nested_put(&nest, env);
+ ll_merge_attr(env, inode);
+ cl_env_put(env, &refcheck);
/* Release the file.
* NB: lease lock handle is released in mdc_hsm_release_pack() because
* we still need it to pack l_remote_handle to MDT. */
- rc = ll_close_inode_openhandle(ll_i2sbi(inode)->ll_md_exp, inode, och,
+ rc = ll_close_inode_openhandle(inode, och, MDS_HSM_RELEASE,
&data_version);
och = NULL;
}
struct ll_swap_stack {
- struct iattr ia1, ia2;
- __u64 dv1, dv2;
- struct inode *inode1, *inode2;
- bool check_dv1, check_dv2;
+ __u64 dv1;
+ __u64 dv2;
+ struct inode *inode1;
+ struct inode *inode2;
+ bool check_dv1;
+ bool check_dv2;
};
static int ll_swap_layouts(struct file *file1, struct file *file2,
if (llss == NULL)
RETURN(-ENOMEM);
- llss->inode1 = file1->f_dentry->d_inode;
- llss->inode2 = file2->f_dentry->d_inode;
-
- if (!S_ISREG(llss->inode2->i_mode))
- GOTO(free, rc = -EINVAL);
-
- if (inode_permission(llss->inode1, MAY_WRITE) ||
- inode_permission(llss->inode2, MAY_WRITE))
- GOTO(free, rc = -EPERM);
+ llss->inode1 = file_inode(file1);
+ llss->inode2 = file_inode(file2);
- if (llss->inode2->i_sb != llss->inode1->i_sb)
- GOTO(free, rc = -EXDEV);
+ rc = ll_check_swap_layouts_validity(llss->inode1, llss->inode2);
+ if (rc < 0)
+ GOTO(free, rc);
/* we use 2 bool because it is easier to swap than 2 bits */
if (lsl->sl_flags & SWAP_LAYOUTS_CHECK_DV1)
rc = lu_fid_cmp(ll_inode2fid(llss->inode1), ll_inode2fid(llss->inode2));
if (rc == 0) /* same file, done! */
- GOTO(free, rc = 0);
+ GOTO(free, rc);
if (rc < 0) { /* sequentialize it */
swap(llss->inode1, llss->inode2);
}
}
- /* to be able to restore mtime and atime after swap
- * we need to first save them */
- if (lsl->sl_flags &
- (SWAP_LAYOUTS_KEEP_MTIME | SWAP_LAYOUTS_KEEP_ATIME)) {
- llss->ia1.ia_mtime = llss->inode1->i_mtime;
- llss->ia1.ia_atime = llss->inode1->i_atime;
- llss->ia1.ia_valid = ATTR_MTIME | ATTR_ATIME;
- llss->ia2.ia_mtime = llss->inode2->i_mtime;
- llss->ia2.ia_atime = llss->inode2->i_atime;
- llss->ia2.ia_valid = ATTR_MTIME | ATTR_ATIME;
- }
-
/* ultimate check, before swaping the layouts we check if
* dataversion has changed (if requested) */
if (llss->check_dv1) {
sizeof(*op_data), op_data, NULL);
ll_finish_md_op_data(op_data);
+ if (rc < 0)
+ GOTO(putgl, rc);
+
putgl:
if (gid != 0) {
ll_put_grouplock(llss->inode2, file2, gid);
ll_put_grouplock(llss->inode1, file1, gid);
}
- /* rc can be set from obd_iocontrol() or from a GOTO(putgl, ...) */
- if (rc != 0)
- GOTO(free, rc);
-
- /* clear useless flags */
- if (!(lsl->sl_flags & SWAP_LAYOUTS_KEEP_MTIME)) {
- llss->ia1.ia_valid &= ~ATTR_MTIME;
- llss->ia2.ia_valid &= ~ATTR_MTIME;
- }
-
- if (!(lsl->sl_flags & SWAP_LAYOUTS_KEEP_ATIME)) {
- llss->ia1.ia_valid &= ~ATTR_ATIME;
- llss->ia2.ia_valid &= ~ATTR_ATIME;
- }
-
- /* update time if requested */
- rc = 0;
- if (llss->ia2.ia_valid != 0) {
- mutex_lock(&llss->inode1->i_mutex);
- rc = ll_setattr(file1->f_dentry, &llss->ia2);
- mutex_unlock(&llss->inode1->i_mutex);
- }
-
- if (llss->ia1.ia_valid != 0) {
- int rc1;
-
- mutex_lock(&llss->inode2->i_mutex);
- rc1 = ll_setattr(file2->f_dentry, &llss->ia1);
- mutex_unlock(&llss->inode2->i_mutex);
- if (rc == 0)
- rc = rc1;
- }
-
free:
if (llss != NULL)
OBD_FREE_PTR(llss);
RETURN(rc);
}
-static int ll_hsm_state_set(struct inode *inode, struct hsm_state_set *hss)
+int ll_hsm_state_set(struct inode *inode, struct hsm_state_set *hss)
{
struct md_op_data *op_data;
int rc;
+ ENTRY;
+
+ /* Detect out-of range masks */
+ if ((hss->hss_setmask | hss->hss_clearmask) & ~HSM_FLAGS_MASK)
+ RETURN(-EINVAL);
/* Non-root users are forbidden to set or clear flags which are
* NOT defined in HSM_USER_MASK. */
!cfs_capable(CFS_CAP_SYS_ADMIN))
RETURN(-EPERM);
+ /* Detect out-of range archive id */
+ if ((hss->hss_valid & HSS_ARCHIVE_ID) &&
+ (hss->hss_archive_id > LL_HSM_MAX_ARCHIVE))
+ RETURN(-EINVAL);
+
op_data = ll_prep_md_op_data(NULL, inode, NULL, NULL, 0, 0,
LUSTRE_OPC_ANY, hss);
if (IS_ERR(op_data))
ATTR_MTIME | ATTR_MTIME_SET |
ATTR_ATIME | ATTR_ATIME_SET;
- mutex_lock(&inode->i_mutex);
+ inode_lock(inode);
- rc = ll_setattr_raw(file->f_dentry, attr, true);
+ rc = ll_setattr_raw(file_dentry(file), attr, true);
if (rc == -ENODATA)
rc = 0;
- mutex_unlock(&inode->i_mutex);
+ inode_unlock(inode);
out:
if (hss != NULL)
((fmode & FMODE_WRITE) ? LL_LEASE_WRLCK : 0);
}
+static int ll_file_futimes_3(struct file *file, const struct ll_futimes_3 *lfu)
+{
+ struct inode *inode = file_inode(file);
+ struct iattr ia = {
+ .ia_valid = ATTR_ATIME | ATTR_ATIME_SET |
+ ATTR_MTIME | ATTR_MTIME_SET |
+ ATTR_CTIME | ATTR_CTIME_SET,
+ .ia_atime = {
+ .tv_sec = lfu->lfu_atime_sec,
+ .tv_nsec = lfu->lfu_atime_nsec,
+ },
+ .ia_mtime = {
+ .tv_sec = lfu->lfu_mtime_sec,
+ .tv_nsec = lfu->lfu_mtime_nsec,
+ },
+ .ia_ctime = {
+ .tv_sec = lfu->lfu_ctime_sec,
+ .tv_nsec = lfu->lfu_ctime_nsec,
+ },
+ };
+ int rc;
+ ENTRY;
+
+ if (!capable(CAP_SYS_ADMIN))
+ RETURN(-EPERM);
+
+ if (!S_ISREG(inode->i_mode))
+ RETURN(-EINVAL);
+
+ inode_lock(inode);
+ rc = ll_setattr_raw(file_dentry(file), &ia, false);
+ inode_unlock(inode);
+
+ RETURN(rc);
+}
+
+/*
+ * Give file access advices
+ *
+ * The ladvise interface is similar to Linux fadvise() system call, except it
+ * forwards the advices directly from Lustre client to server. The server side
+ * codes will apply appropriate read-ahead and caching techniques for the
+ * corresponding files.
+ *
+ * A typical workload for ladvise is e.g. a bunch of different clients are
+ * doing small random reads of a file, so prefetching pages into OSS cache
+ * with big linear reads before the random IO is a net benefit. Fetching
+ * all that data into each client cache with fadvise() may not be, due to
+ * much more data being sent to the client.
+ */
+static int ll_ladvise(struct inode *inode, struct file *file, __u64 flags,
+ struct llapi_lu_ladvise *ladvise)
+{
+ struct lu_env *env;
+ struct cl_io *io;
+ struct cl_ladvise_io *lio;
+ int rc;
+ __u16 refcheck;
+ ENTRY;
+
+ env = cl_env_get(&refcheck);
+ if (IS_ERR(env))
+ RETURN(PTR_ERR(env));
+
+ io = vvp_env_thread_io(env);
+ io->ci_obj = ll_i2info(inode)->lli_clob;
+
+ /* initialize parameters for ladvise */
+ lio = &io->u.ci_ladvise;
+ lio->li_start = ladvise->lla_start;
+ lio->li_end = ladvise->lla_end;
+ lio->li_fid = ll_inode2fid(inode);
+ lio->li_advice = ladvise->lla_advice;
+ lio->li_flags = flags;
+
+ if (cl_io_init(env, io, CIT_LADVISE, io->ci_obj) == 0)
+ rc = cl_io_loop(env, io);
+ else
+ rc = io->ci_result;
+
+ cl_io_fini(env, io);
+ cl_env_put(env, &refcheck);
+ RETURN(rc);
+}
+
static long
ll_file_ioctl(struct file *file, unsigned int cmd, unsigned long arg)
{
- struct inode *inode = file->f_dentry->d_inode;
+ struct inode *inode = file_inode(file);
struct ll_file_data *fd = LUSTRE_FPRIVATE(file);
int flags, rc;
ENTRY;
switch(cmd) {
case LL_IOC_GETFLAGS:
/* Get the current value of the file flags */
- return put_user(fd->fd_flags, (int *)arg);
+ return put_user(fd->fd_flags, (int __user *)arg);
case LL_IOC_SETFLAGS:
case LL_IOC_CLRFLAGS:
/* Set or clear specific file flags */
/* XXX This probably needs checks to ensure the flags are
* not abused, and to handle any flag side effects.
*/
- if (get_user(flags, (int *) arg))
+ if (get_user(flags, (int __user *) arg))
RETURN(-EFAULT);
if (cmd == LL_IOC_SETFLAGS) {
struct file *file2;
struct lustre_swap_layouts lsl;
- if (copy_from_user(&lsl, (char *)arg,
+ if (copy_from_user(&lsl, (char __user *)arg,
sizeof(struct lustre_swap_layouts)))
RETURN(-EFAULT);
- if ((file->f_flags & O_ACCMODE) == 0) /* O_RDONLY */
+ if ((file->f_flags & O_ACCMODE) == O_RDONLY)
RETURN(-EPERM);
file2 = fget(lsl.sl_fd);
if (file2 == NULL)
RETURN(-EBADF);
- rc = -EPERM;
- if ((file2->f_flags & O_ACCMODE) != 0) /* O_WRONLY or O_RDWR */
+ /* O_WRONLY or O_RDWR */
+ if ((file2->f_flags & O_ACCMODE) == O_RDONLY)
+ GOTO(out, rc = -EPERM);
+
+ if (lsl.sl_flags & SWAP_LAYOUTS_CLOSE) {
+ struct inode *inode2;
+ struct ll_inode_info *lli;
+ struct obd_client_handle *och = NULL;
+
+ if (lsl.sl_flags != SWAP_LAYOUTS_CLOSE)
+ GOTO(out, rc = -EINVAL);
+
+ lli = ll_i2info(inode);
+ mutex_lock(&lli->lli_och_mutex);
+ if (fd->fd_lease_och != NULL) {
+ och = fd->fd_lease_och;
+ fd->fd_lease_och = NULL;
+ }
+ mutex_unlock(&lli->lli_och_mutex);
+ if (och == NULL)
+ GOTO(out, rc = -ENOLCK);
+ inode2 = file_inode(file2);
+ rc = ll_swap_layouts_close(och, inode, inode2);
+ } else {
rc = ll_swap_layouts(file, file2, &lsl);
+ }
+out:
fput(file2);
RETURN(rc);
}
- case LL_IOC_LOV_GETSTRIPE:
- RETURN(ll_lov_getstripe(inode, arg));
- case LL_IOC_RECREATE_OBJ:
- RETURN(ll_lov_recreate_obj(inode, arg));
- case LL_IOC_RECREATE_FID:
- RETURN(ll_lov_recreate_fid(inode, arg));
- case FSFILT_IOC_FIEMAP:
- RETURN(ll_ioctl_fiemap(inode, arg));
+ case LL_IOC_LOV_GETSTRIPE:
+ RETURN(ll_file_getstripe(inode,
+ (struct lov_user_md __user *)arg));
case FSFILT_IOC_GETFLAGS:
case FSFILT_IOC_SETFLAGS:
RETURN(ll_iocontrol(inode, file, cmd, arg));
case FSFILT_IOC_GETVERSION_OLD:
case FSFILT_IOC_GETVERSION:
- RETURN(put_user(inode->i_generation, (int *)arg));
+ RETURN(put_user(inode->i_generation, (int __user *)arg));
case LL_IOC_GROUP_LOCK:
RETURN(ll_get_grouplock(inode, file, arg));
case LL_IOC_GROUP_UNLOCK:
RETURN(ll_put_grouplock(inode, file, arg));
case IOC_OBD_STATFS:
- RETURN(ll_obd_statfs(inode, (void *)arg));
+ RETURN(ll_obd_statfs(inode, (void __user *)arg));
/* We need to special case any other ioctls we want to handle,
* to send them to the MDS/OST as appropriate and to properly
case LL_IOC_FLUSHCTX:
RETURN(ll_flush_ctx(inode));
case LL_IOC_PATH2FID: {
- if (copy_to_user((void *)arg, ll_inode2fid(inode),
+ if (copy_to_user((void __user *)arg, ll_inode2fid(inode),
sizeof(struct lu_fid)))
RETURN(-EFAULT);
RETURN(0);
}
+ case LL_IOC_GETPARENT:
+ RETURN(ll_getparent(file, (struct getparent __user *)arg));
+
case OBD_IOC_FID2PATH:
- RETURN(ll_fid2path(inode, (void *)arg));
+ RETURN(ll_fid2path(inode, (void __user *)arg));
case LL_IOC_DATA_VERSION: {
struct ioc_data_version idv;
int rc;
- if (copy_from_user(&idv, (char *)arg, sizeof(idv)))
+ if (copy_from_user(&idv, (char __user *)arg, sizeof(idv)))
RETURN(-EFAULT);
idv.idv_flags &= LL_DV_RD_FLUSH | LL_DV_WR_FLUSH;
rc = ll_data_version(inode, &idv.idv_version, idv.idv_flags);
- if (rc == 0 && copy_to_user((char *) arg, &idv, sizeof(idv)))
+ if (rc == 0 &&
+ copy_to_user((char __user *)arg, &idv, sizeof(idv)))
RETURN(-EFAULT);
RETURN(rc);
if (mdtidx < 0)
RETURN(mdtidx);
- if (put_user((int)mdtidx, (int*)arg))
+ if (put_user((int)mdtidx, (int __user *)arg))
RETURN(-EFAULT);
RETURN(0);
rc = obd_iocontrol(cmd, ll_i2mdexp(inode), sizeof(*op_data),
op_data, NULL);
- if (copy_to_user((void *)arg, hus, sizeof(*hus)))
+ if (copy_to_user((void __user *)arg, hus, sizeof(*hus)))
rc = -EFAULT;
ll_finish_md_op_data(op_data);
if (hss == NULL)
RETURN(-ENOMEM);
- if (copy_from_user(hss, (char *)arg, sizeof(*hss))) {
+ if (copy_from_user(hss, (char __user *)arg, sizeof(*hss))) {
OBD_FREE_PTR(hss);
RETURN(-EFAULT);
}
rc = obd_iocontrol(cmd, ll_i2mdexp(inode), sizeof(*op_data),
op_data, NULL);
- if (copy_to_user((char *)arg, hca, sizeof(*hca)))
+ if (copy_to_user((char __user *)arg, hca, sizeof(*hca)))
rc = -EFAULT;
ll_finish_md_op_data(op_data);
if (rc < 0)
RETURN(rc);
+ rc = ll_lease_och_release(inode, file);
+ if (rc < 0)
+ RETURN(rc);
+
if (lease_broken)
fmode = 0;
if (hui == NULL)
RETURN(-ENOMEM);
- if (copy_from_user(hui, (void *)arg, sizeof(*hui))) {
+ if (copy_from_user(hui, (void __user *)arg, sizeof(*hui))) {
OBD_FREE_PTR(hui);
RETURN(-EFAULT);
}
OBD_FREE_PTR(hui);
RETURN(rc);
}
+ case LL_IOC_FUTIMES_3: {
+ struct ll_futimes_3 lfu;
+
+ if (copy_from_user(&lfu,
+ (const struct ll_futimes_3 __user *)arg,
+ sizeof(lfu)))
+ RETURN(-EFAULT);
+
+ RETURN(ll_file_futimes_3(file, &lfu));
+ }
+ case LL_IOC_LADVISE: {
+ struct llapi_ladvise_hdr *ladvise_hdr;
+ int i;
+ int num_advise;
+ int alloc_size = sizeof(*ladvise_hdr);
+
+ rc = 0;
+ OBD_ALLOC_PTR(ladvise_hdr);
+ if (ladvise_hdr == NULL)
+ RETURN(-ENOMEM);
+
+ if (copy_from_user(ladvise_hdr,
+ (const struct llapi_ladvise_hdr __user *)arg,
+ alloc_size))
+ GOTO(out_ladvise, rc = -EFAULT);
+
+ if (ladvise_hdr->lah_magic != LADVISE_MAGIC ||
+ ladvise_hdr->lah_count < 1)
+ GOTO(out_ladvise, rc = -EINVAL);
+
+ num_advise = ladvise_hdr->lah_count;
+ if (num_advise >= LAH_COUNT_MAX)
+ GOTO(out_ladvise, rc = -EFBIG);
+
+ OBD_FREE_PTR(ladvise_hdr);
+ alloc_size = offsetof(typeof(*ladvise_hdr),
+ lah_advise[num_advise]);
+ OBD_ALLOC(ladvise_hdr, alloc_size);
+ if (ladvise_hdr == NULL)
+ RETURN(-ENOMEM);
+
+ /*
+ * TODO: submit multiple advices to one server in a single RPC
+ */
+ if (copy_from_user(ladvise_hdr,
+ (const struct llapi_ladvise_hdr __user *)arg,
+ alloc_size))
+ GOTO(out_ladvise, rc = -EFAULT);
+
+ for (i = 0; i < num_advise; i++) {
+ rc = ll_ladvise(inode, file, ladvise_hdr->lah_flags,
+ &ladvise_hdr->lah_advise[i]);
+ if (rc)
+ break;
+ }
+out_ladvise:
+ OBD_FREE(ladvise_hdr, alloc_size);
+ RETURN(rc);
+ }
default: {
int err;
RETURN(err);
RETURN(obd_iocontrol(cmd, ll_i2dtexp(inode), 0, NULL,
- (void *)arg));
+ (void __user *)arg));
}
}
}
generic_file_llseek_size(struct file *file, loff_t offset, int origin,
loff_t maxsize, loff_t eof)
{
- struct inode *inode = file->f_dentry->d_inode;
+ struct inode *inode = file_inode(file);
switch (origin) {
case SEEK_END:
* SEEK_CURs. Note that parallel writes and reads behave
* like SEEK_SET.
*/
- mutex_lock(&inode->i_mutex);
+ inode_lock(inode);
offset = llseek_execute(file, file->f_pos + offset, maxsize);
- mutex_unlock(&inode->i_mutex);
+ inode_unlock(inode);
return offset;
case SEEK_DATA:
/*
static loff_t ll_file_seek(struct file *file, loff_t offset, int origin)
{
- struct inode *inode = file->f_dentry->d_inode;
+ struct inode *inode = file_inode(file);
loff_t retval, eof = 0;
ENTRY;
static int ll_flush(struct file *file, fl_owner_t id)
{
- struct inode *inode = file->f_dentry->d_inode;
+ struct inode *inode = file_inode(file);
struct ll_inode_info *lli = ll_i2info(inode);
struct ll_file_data *fd = LUSTRE_FPRIVATE(file);
int rc, err;
int cl_sync_file_range(struct inode *inode, loff_t start, loff_t end,
enum cl_fsync_mode mode, int ignore_layout)
{
- struct cl_env_nest nest;
struct lu_env *env;
struct cl_io *io;
- struct obd_capa *capa = NULL;
struct cl_fsync_io *fio;
int result;
+ __u16 refcheck;
ENTRY;
if (mode != CL_FSYNC_NONE && mode != CL_FSYNC_LOCAL &&
mode != CL_FSYNC_DISCARD && mode != CL_FSYNC_ALL)
RETURN(-EINVAL);
- env = cl_env_nested_get(&nest);
+ env = cl_env_get(&refcheck);
if (IS_ERR(env))
RETURN(PTR_ERR(env));
- capa = ll_osscapa_get(inode, CAPA_OPC_OSS_WRITE);
-
- io = ccc_env_thread_io(env);
- io->ci_obj = cl_i2info(inode)->lli_clob;
+ io = vvp_env_thread_io(env);
+ io->ci_obj = ll_i2info(inode)->lli_clob;
io->ci_ignore_layout = ignore_layout;
/* initialize parameters for sync */
fio = &io->u.ci_fsync;
- fio->fi_capa = capa;
fio->fi_start = start;
fio->fi_end = end;
fio->fi_fid = ll_inode2fid(inode);
if (result == 0)
result = fio->fi_nr_written;
cl_io_fini(env, io);
- cl_env_nested_put(&nest, env);
-
- capa_put(capa);
+ cl_env_put(env, &refcheck);
RETURN(result);
}
/*
- * When dentry is provided (the 'else' case), *file->f_dentry may be
+ * When dentry is provided (the 'else' case), file_dentry() may be
* null and dentry must be used directly rather than pulled from
- * *file->f_dentry as is done otherwise.
+ * file_dentry() as is done otherwise.
*/
#ifdef HAVE_FILE_FSYNC_4ARGS
int ll_fsync(struct file *file, loff_t start, loff_t end, int datasync)
{
- struct dentry *dentry = file->f_dentry;
+ struct dentry *dentry = file_dentry(file);
#elif defined(HAVE_FILE_FSYNC_2ARGS)
int ll_fsync(struct file *file, int datasync)
{
- struct dentry *dentry = file->f_dentry;
+ struct dentry *dentry = file_dentry(file);
loff_t start = 0;
loff_t end = LLONG_MAX;
#else
struct inode *inode = dentry->d_inode;
struct ll_inode_info *lli = ll_i2info(inode);
struct ptlrpc_request *req;
- struct obd_capa *oc;
int rc, err;
ENTRY;
#ifdef HAVE_FILE_FSYNC_4ARGS
rc = filemap_write_and_wait_range(inode->i_mapping, start, end);
- mutex_lock(&inode->i_mutex);
+ inode_lock(inode);
#else
/* fsync's caller has already called _fdata{sync,write}, we want
* that IO to finish before calling the osc and mdc sync methods */
rc = err;
}
- oc = ll_mdscapa_get(inode);
- err = md_fsync(ll_i2sbi(inode)->ll_md_exp, ll_inode2fid(inode), oc,
- &req);
- capa_put(oc);
+ err = md_fsync(ll_i2sbi(inode)->ll_md_exp, ll_inode2fid(inode), &req);
if (!rc)
rc = err;
if (!err)
}
#ifdef HAVE_FILE_FSYNC_4ARGS
- mutex_unlock(&inode->i_mutex);
+ inode_unlock(inode);
#endif
RETURN(rc);
}
static int
ll_file_flock(struct file *file, int cmd, struct file_lock *file_lock)
{
- struct inode *inode = file->f_dentry->d_inode;
+ struct inode *inode = file_inode(file);
struct ll_sb_info *sbi = ll_i2sbi(inode);
struct ldlm_enqueue_info einfo = {
.ei_type = LDLM_FLOCK,
.ei_cbdata = file_lock,
};
struct md_op_data *op_data;
- struct lustre_handle lockh = {0};
- ldlm_policy_data_t flock = {{0}};
+ struct lustre_handle lockh = { 0 };
+ union ldlm_policy_data flock = { { 0 } };
int fl_type = file_lock->fl_type;
__u64 flags = 0;
int rc;
if (IS_ERR(op_data))
RETURN(PTR_ERR(op_data));
- CDEBUG(D_DLMTRACE, "inode="DFID", pid=%u, flags="LPX64", mode=%u, "
- "start="LPU64", end="LPU64"\n", PFID(ll_inode2fid(inode)),
+ CDEBUG(D_DLMTRACE, "inode="DFID", pid=%u, flags=%#llx, mode=%u, "
+ "start=%llu, end=%llu\n", PFID(ll_inode2fid(inode)),
flock.l_flock.pid, flags, einfo.ei_mode,
flock.l_flock.start, flock.l_flock.end);
if (!(flags & LDLM_FL_TEST_LOCK))
file_lock->fl_type = fl_type;
+#ifdef HAVE_LOCKS_LOCK_FILE_WAIT
+ if ((rc == 0 || file_lock->fl_type == F_UNLCK) &&
+ !(flags & LDLM_FL_TEST_LOCK))
+ rc2 = locks_lock_file_wait(file, file_lock);
+#else
if ((file_lock->fl_flags & FL_FLOCK) &&
(rc == 0 || file_lock->fl_type == F_UNLCK))
rc2 = flock_lock_file_wait(file, file_lock);
(rc == 0 || file_lock->fl_type == F_UNLCK) &&
!(flags & LDLM_FL_TEST_LOCK))
rc2 = posix_lock_file_wait(file, file_lock);
+#endif /* HAVE_LOCKS_LOCK_FILE_WAIT */
if (rc2 && file_lock->fl_type != F_UNLCK) {
einfo.ei_mode = LCK_NL;
}
int ll_get_fid_by_name(struct inode *parent, const char *name,
- int namelen, struct lu_fid *fid)
+ int namelen, struct lu_fid *fid,
+ struct inode **inode)
{
struct md_op_data *op_data = NULL;
struct mdt_body *body;
if (IS_ERR(op_data))
RETURN(PTR_ERR(op_data));
- op_data->op_valid = OBD_MD_FLID;
+ op_data->op_valid = OBD_MD_FLID | OBD_MD_FLTYPE;
rc = md_getattr_name(ll_i2sbi(parent)->ll_md_exp, op_data, &req);
ll_finish_md_op_data(op_data);
if (rc < 0)
GOTO(out_req, rc = -EFAULT);
if (fid != NULL)
*fid = body->mbo_fid1;
+
+ if (inode != NULL)
+ rc = ll_prep_inode(inode, req, parent->i_sb, NULL);
out_req:
ptlrpc_req_finished(req);
RETURN(rc);
struct inode *child_inode = NULL;
struct md_op_data *op_data;
struct ptlrpc_request *request = NULL;
+ struct obd_client_handle *och = NULL;
struct qstr qstr;
+ struct mdt_body *body;
int rc;
+ __u64 data_version = 0;
ENTRY;
CDEBUG(D_VFSTRACE, "migrate %s under "DFID" to MDT%04x\n",
qstr.hash = full_name_hash(name, namelen);
qstr.name = name;
qstr.len = namelen;
- dchild = d_lookup(file->f_dentry, &qstr);
- if (dchild != NULL && dchild->d_inode != NULL) {
- op_data->op_fid3 = *ll_inode2fid(dchild->d_inode);
- if (dchild->d_inode != NULL) {
+ dchild = d_lookup(file_dentry(file), &qstr);
+ if (dchild != NULL) {
+ if (dchild->d_inode != NULL)
child_inode = igrab(dchild->d_inode);
- ll_invalidate_aliases(child_inode);
- }
dput(dchild);
- } else {
+ }
+
+ if (child_inode == NULL) {
rc = ll_get_fid_by_name(parent, name, namelen,
- &op_data->op_fid3);
+ &op_data->op_fid3, &child_inode);
if (rc != 0)
GOTO(out_free, rc);
}
+ if (child_inode == NULL)
+ GOTO(out_free, rc = -EINVAL);
+
+ /*
+ * lfs migrate command needs to be blocked on the client
+ * by checking the migrate FID against the FID of the
+ * filesystem root.
+ */
+ if (child_inode == parent->i_sb->s_root->d_inode)
+ GOTO(out_iput, rc = -EINVAL);
+
+ inode_lock(child_inode);
+ op_data->op_fid3 = *ll_inode2fid(child_inode);
if (!fid_is_sane(&op_data->op_fid3)) {
- CERROR("%s: migrate %s , but fid "DFID" is insane\n",
+ CERROR("%s: migrate %s, but FID "DFID" is insane\n",
ll_get_fsname(parent->i_sb, NULL, 0), name,
PFID(&op_data->op_fid3));
- GOTO(out_free, rc = -EINVAL);
+ GOTO(out_unlock, rc = -EINVAL);
}
rc = ll_get_mdt_idx_by_fid(ll_i2sbi(parent), &op_data->op_fid3);
if (rc < 0)
- GOTO(out_free, rc);
+ GOTO(out_unlock, rc);
if (rc == mdtidx) {
- CDEBUG(D_INFO, "%s:"DFID" is already on MDT%d.\n", name,
+ CDEBUG(D_INFO, "%s: "DFID" is already on MDT%04x\n", name,
PFID(&op_data->op_fid3), mdtidx);
- GOTO(out_free, rc = 0);
+ GOTO(out_unlock, rc = 0);
+ }
+again:
+ if (S_ISREG(child_inode->i_mode)) {
+ och = ll_lease_open(child_inode, NULL, FMODE_WRITE, 0);
+ if (IS_ERR(och)) {
+ rc = PTR_ERR(och);
+ och = NULL;
+ GOTO(out_unlock, rc);
+ }
+
+ rc = ll_data_version(child_inode, &data_version,
+ LL_DV_WR_FLUSH);
+ if (rc != 0)
+ GOTO(out_close, rc);
+
+ op_data->op_handle = och->och_fh;
+ op_data->op_data = och->och_mod;
+ op_data->op_data_version = data_version;
+ op_data->op_lease_handle = och->och_lease_handle;
+ op_data->op_bias |= MDS_RENAME_MIGRATE;
}
op_data->op_mds = mdtidx;
if (rc == 0)
ll_update_times(request, parent);
- ptlrpc_req_finished(request);
- if (rc != 0)
- GOTO(out_free, rc);
+ if (request != NULL) {
+ body = req_capsule_server_get(&request->rq_pill, &RMF_MDT_BODY);
+ if (body == NULL) {
+ ptlrpc_req_finished(request);
+ GOTO(out_close, rc = -EPROTO);
+ }
-out_free:
- if (child_inode != NULL) {
- clear_nlink(child_inode);
- iput(child_inode);
+ /* If the server does release layout lock, then we cleanup
+ * the client och here, otherwise release it in out_close: */
+ if (och != NULL &&
+ body->mbo_valid & OBD_MD_CLOSE_INTENT_EXECED) {
+ obd_mod_put(och->och_mod);
+ md_clear_open_replay_data(ll_i2sbi(parent)->ll_md_exp,
+ och);
+ och->och_fh.cookie = DEAD_HANDLE_MAGIC;
+ OBD_FREE_PTR(och);
+ och = NULL;
+ }
+ ptlrpc_req_finished(request);
}
+ /* Try again if the file layout has changed. */
+ if (rc == -EAGAIN && S_ISREG(child_inode->i_mode)) {
+ request = NULL;
+ goto again;
+ }
+out_close:
+ if (och != NULL) /* close the file */
+ ll_lease_close(och, child_inode, NULL);
+ if (rc == 0)
+ clear_nlink(child_inode);
+out_unlock:
+ inode_unlock(child_inode);
+out_iput:
+ iput(child_inode);
+out_free:
ll_finish_md_op_data(op_data);
RETURN(rc);
}
* \param l_req_mode [IN] searched lock mode
* \retval boolean, true iff all bits are found
*/
-int ll_have_md_lock(struct inode *inode, __u64 *bits, ldlm_mode_t l_req_mode)
+int ll_have_md_lock(struct inode *inode, __u64 *bits, enum ldlm_mode l_req_mode)
{
- struct lustre_handle lockh;
- ldlm_policy_data_t policy;
- ldlm_mode_t mode = (l_req_mode == LCK_MINMODE) ?
- (LCK_CR|LCK_CW|LCK_PR|LCK_PW) : l_req_mode;
- struct lu_fid *fid;
+ struct lustre_handle lockh;
+ union ldlm_policy_data policy;
+ enum ldlm_mode mode = (l_req_mode == LCK_MINMODE) ?
+ (LCK_CR | LCK_CW | LCK_PR | LCK_PW) : l_req_mode;
+ struct lu_fid *fid;
__u64 flags;
- int i;
- ENTRY;
+ int i;
+ ENTRY;
if (!inode)
RETURN(0);
RETURN(*bits == 0);
}
-ldlm_mode_t ll_take_md_lock(struct inode *inode, __u64 bits,
- struct lustre_handle *lockh, __u64 flags,
- ldlm_mode_t mode)
+enum ldlm_mode ll_take_md_lock(struct inode *inode, __u64 bits,
+ struct lustre_handle *lockh, __u64 flags,
+ enum ldlm_mode mode)
{
- ldlm_policy_data_t policy = { .l_inodebits = {bits}};
- struct lu_fid *fid;
- ldlm_mode_t rc;
- ENTRY;
+ union ldlm_policy_data policy = { .l_inodebits = { bits } };
+ struct lu_fid *fid;
+ enum ldlm_mode rc;
+ ENTRY;
- fid = &ll_i2info(inode)->lli_fid;
- CDEBUG(D_INFO, "trying to match res "DFID"\n", PFID(fid));
+ fid = &ll_i2info(inode)->lli_fid;
+ CDEBUG(D_INFO, "trying to match res "DFID"\n", PFID(fid));
rc = md_lock_match(ll_i2mdexp(inode), LDLM_FL_BLOCK_GRANTED|flags,
fid, LDLM_IBITS, &policy, mode, lockh);
/* Already unlinked. Just update nlink and return success */
if (rc == -ENOENT) {
clear_nlink(inode);
+ /* If it is striped directory, and there is bad stripe
+ * Let's revalidate the dentry again, instead of returning
+ * error */
+ if (S_ISDIR(inode->i_mode) &&
+ ll_i2info(inode)->lli_lsm_md != NULL)
+ return 0;
+
/* This path cannot be hit for regular files unless in
* case of obscure races, so no need to to validate
* size. */
do_lookup() -> ll_revalidate_it(). We cannot use d_drop
here to preserve get_cwd functionality on 2.6.
Bug 10503 */
- if (!dentry->d_inode->i_nlink)
+ if (!dentry->d_inode->i_nlink) {
+ ll_lock_dcache(inode);
d_lustre_invalidate(dentry, 0);
+ ll_unlock_dcache(inode);
+ }
ll_lookup_finish_locks(&oit, dentry);
} else if (!ll_have_md_lock(dentry->d_inode, &ibits, LCK_MINMODE)) {
- struct ll_sb_info *sbi = ll_i2sbi(dentry->d_inode);
- obd_valid valid = OBD_MD_FLGETATTR;
- struct md_op_data *op_data;
- int ealen = 0;
+ struct ll_sb_info *sbi = ll_i2sbi(dentry->d_inode);
+ u64 valid = OBD_MD_FLGETATTR;
+ struct md_op_data *op_data;
+ int ealen = 0;
if (S_ISREG(inode->i_mode)) {
rc = ll_get_default_mdsize(sbi, &ealen);
RETURN(PTR_ERR(op_data));
op_data->op_valid = valid;
- /* Once OBD_CONNECT_ATTRFID is not supported, we can't find one
- * capa for this inode. Because we only keep capas of dirs
- * fresh. */
rc = md_getattr(sbi->ll_md_exp, op_data, &req);
ll_finish_md_op_data(op_data);
if (rc) {
LASSERT(ll_i2info(inode)->lli_lsm_md != NULL);
rc = md_merge_attr(ll_i2mdexp(inode), ll_i2info(inode)->lli_lsm_md,
- &attr);
+ &attr, ll_md_blocking_ast);
if (rc != 0)
RETURN(rc);
- ll_i2info(inode)->lli_stripe_dir_size = attr.cat_size;
- ll_i2info(inode)->lli_stripe_dir_nlink = attr.cat_nlink;
+ set_nlink(inode, attr.cat_nlink);
+ inode->i_blocks = attr.cat_blocks;
+ i_size_write(inode, attr.cat_size);
- ll_i2info(inode)->lli_lvb.lvb_atime = attr.cat_atime;
- ll_i2info(inode)->lli_lvb.lvb_mtime = attr.cat_mtime;
- ll_i2info(inode)->lli_lvb.lvb_ctime = attr.cat_ctime;
+ ll_i2info(inode)->lli_atime = attr.cat_atime;
+ ll_i2info(inode)->lli_mtime = attr.cat_mtime;
+ ll_i2info(inode)->lli_ctime = attr.cat_ctime;
RETURN(0);
}
RETURN(rc);
}
- LTIME_S(inode->i_atime) = ll_i2info(inode)->lli_lvb.lvb_atime;
- LTIME_S(inode->i_mtime) = ll_i2info(inode)->lli_lvb.lvb_mtime;
- LTIME_S(inode->i_ctime) = ll_i2info(inode)->lli_lvb.lvb_ctime;
+ LTIME_S(inode->i_atime) = ll_i2info(inode)->lli_atime;
+ LTIME_S(inode->i_mtime) = ll_i2info(inode)->lli_mtime;
+ LTIME_S(inode->i_ctime) = ll_i2info(inode)->lli_ctime;
} else {
/* In case of restore, the MDT has the right size and has
* already send it back without granting the layout lock,
* restore the MDT holds the layout lock so the glimpse will
* block up to the end of restore (getattr will block)
*/
- if (!(ll_i2info(inode)->lli_flags & LLIF_FILE_RESTORING))
+ if (!ll_file_test_flag(ll_i2info(inode), LLIF_FILE_RESTORING))
rc = ll_glimpse_size(inode);
}
RETURN(rc);
if (res)
return res;
+ OBD_FAIL_TIMEOUT(OBD_FAIL_GETATTR_DELAY, 30);
+
stat->dev = inode->i_sb->s_dev;
if (ll_need_32bit_api(sbi))
stat->ino = cl_fid_build_ino(&lli->lli_fid, 1);
stat->mtime = inode->i_mtime;
stat->ctime = inode->i_ctime;
stat->blksize = 1 << inode->i_blkbits;
- stat->blocks = inode->i_blocks;
- if (S_ISDIR(inode->i_mode) &&
- ll_i2info(inode)->lli_lsm_md != NULL) {
- stat->nlink = lli->lli_stripe_dir_nlink;
- stat->size = lli->lli_stripe_dir_size;
- } else {
- stat->nlink = inode->i_nlink;
- stat->size = i_size_read(inode);
- }
+ stat->nlink = inode->i_nlink;
+ stat->size = i_size_read(inode);
+ stat->blocks = inode->i_blocks;
return 0;
}
static int ll_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
__u64 start, __u64 len)
{
- int rc;
- size_t num_bytes;
- struct ll_user_fiemap *fiemap;
- unsigned int extent_count = fieinfo->fi_extents_max;
+ int rc;
+ size_t num_bytes;
+ struct fiemap *fiemap;
+ unsigned int extent_count = fieinfo->fi_extents_max;
- num_bytes = sizeof(*fiemap) + (extent_count *
- sizeof(struct ll_fiemap_extent));
- OBD_ALLOC_LARGE(fiemap, num_bytes);
+ num_bytes = sizeof(*fiemap) + (extent_count *
+ sizeof(struct fiemap_extent));
+ OBD_ALLOC_LARGE(fiemap, num_bytes);
- if (fiemap == NULL)
- RETURN(-ENOMEM);
+ if (fiemap == NULL)
+ RETURN(-ENOMEM);
- fiemap->fm_flags = fieinfo->fi_flags;
- fiemap->fm_extent_count = fieinfo->fi_extents_max;
- fiemap->fm_start = start;
- fiemap->fm_length = len;
- if (extent_count > 0)
- memcpy(&fiemap->fm_extents[0], fieinfo->fi_extents_start,
- sizeof(struct ll_fiemap_extent));
+ fiemap->fm_flags = fieinfo->fi_flags;
+ fiemap->fm_extent_count = fieinfo->fi_extents_max;
+ fiemap->fm_start = start;
+ fiemap->fm_length = len;
+ if (extent_count > 0 &&
+ copy_from_user(&fiemap->fm_extents[0], fieinfo->fi_extents_start,
+ sizeof(struct fiemap_extent)) != 0)
+ GOTO(out, rc = -EFAULT);
rc = ll_do_fiemap(inode, fiemap, num_bytes);
fieinfo->fi_flags = fiemap->fm_flags;
fieinfo->fi_extents_mapped = fiemap->fm_mapped_extents;
- if (extent_count > 0)
- memcpy(fieinfo->fi_extents_start, &fiemap->fm_extents[0],
- fiemap->fm_mapped_extents *
- sizeof(struct ll_fiemap_extent));
-
+ if (extent_count > 0 &&
+ copy_to_user(fieinfo->fi_extents_start, &fiemap->fm_extents[0],
+ fiemap->fm_mapped_extents *
+ sizeof(struct fiemap_extent)) != 0)
+ GOTO(out, rc = -EFAULT);
+out:
OBD_FREE_LARGE(fiemap, num_bytes);
return rc;
}
}
ll_stats_ops_tally(sbi, LPROC_LL_INODE_PERM, 1);
-
- if (sbi->ll_flags & LL_SBI_RMT_CLIENT)
- rc = lustre_check_remote_perm(inode, mask);
- else
- rc = ll_generic_permission(inode, mask, flags, ll_check_acl);
-
+ rc = ll_generic_permission(inode, mask, flags, ll_check_acl);
/* restore current process's credentials and FS capability */
if (squash_id) {
revert_creds(old_cred);
/* -o localflock - only provides locally consistent flock locks */
struct file_operations ll_file_operations = {
- .read = ll_file_read,
- .aio_read = ll_file_aio_read,
- .write = ll_file_write,
- .aio_write = ll_file_aio_write,
- .unlocked_ioctl = ll_file_ioctl,
- .open = ll_file_open,
- .release = ll_file_release,
- .mmap = ll_file_mmap,
- .llseek = ll_file_seek,
- .splice_read = ll_file_splice_read,
- .fsync = ll_fsync,
- .flush = ll_flush
+#ifdef HAVE_FILE_OPERATIONS_READ_WRITE_ITER
+# ifdef HAVE_SYNC_READ_WRITE
+ .read = new_sync_read,
+ .write = new_sync_write,
+# endif
+ .read_iter = ll_file_read_iter,
+ .write_iter = ll_file_write_iter,
+#else /* !HAVE_FILE_OPERATIONS_READ_WRITE_ITER */
+ .read = ll_file_read,
+ .aio_read = ll_file_aio_read,
+ .write = ll_file_write,
+ .aio_write = ll_file_aio_write,
+#endif /* HAVE_FILE_OPERATIONS_READ_WRITE_ITER */
+ .unlocked_ioctl = ll_file_ioctl,
+ .open = ll_file_open,
+ .release = ll_file_release,
+ .mmap = ll_file_mmap,
+ .llseek = ll_file_seek,
+ .splice_read = ll_file_splice_read,
+ .fsync = ll_fsync,
+ .flush = ll_flush
};
struct file_operations ll_file_operations_flock = {
- .read = ll_file_read,
- .aio_read = ll_file_aio_read,
- .write = ll_file_write,
- .aio_write = ll_file_aio_write,
- .unlocked_ioctl = ll_file_ioctl,
- .open = ll_file_open,
- .release = ll_file_release,
- .mmap = ll_file_mmap,
- .llseek = ll_file_seek,
- .splice_read = ll_file_splice_read,
- .fsync = ll_fsync,
- .flush = ll_flush,
- .flock = ll_file_flock,
- .lock = ll_file_flock
+#ifdef HAVE_FILE_OPERATIONS_READ_WRITE_ITER
+# ifdef HAVE_SYNC_READ_WRITE
+ .read = new_sync_read,
+ .write = new_sync_write,
+# endif /* HAVE_SYNC_READ_WRITE */
+ .read_iter = ll_file_read_iter,
+ .write_iter = ll_file_write_iter,
+#else /* !HAVE_FILE_OPERATIONS_READ_WRITE_ITER */
+ .read = ll_file_read,
+ .aio_read = ll_file_aio_read,
+ .write = ll_file_write,
+ .aio_write = ll_file_aio_write,
+#endif /* HAVE_FILE_OPERATIONS_READ_WRITE_ITER */
+ .unlocked_ioctl = ll_file_ioctl,
+ .open = ll_file_open,
+ .release = ll_file_release,
+ .mmap = ll_file_mmap,
+ .llseek = ll_file_seek,
+ .splice_read = ll_file_splice_read,
+ .fsync = ll_fsync,
+ .flush = ll_flush,
+ .flock = ll_file_flock,
+ .lock = ll_file_flock
};
/* These are for -o noflock - to return ENOSYS on flock calls */
struct file_operations ll_file_operations_noflock = {
- .read = ll_file_read,
- .aio_read = ll_file_aio_read,
- .write = ll_file_write,
- .aio_write = ll_file_aio_write,
- .unlocked_ioctl = ll_file_ioctl,
- .open = ll_file_open,
- .release = ll_file_release,
- .mmap = ll_file_mmap,
- .llseek = ll_file_seek,
- .splice_read = ll_file_splice_read,
- .fsync = ll_fsync,
- .flush = ll_flush,
- .flock = ll_file_noflock,
- .lock = ll_file_noflock
+#ifdef HAVE_FILE_OPERATIONS_READ_WRITE_ITER
+# ifdef HAVE_SYNC_READ_WRITE
+ .read = new_sync_read,
+ .write = new_sync_write,
+# endif /* HAVE_SYNC_READ_WRITE */
+ .read_iter = ll_file_read_iter,
+ .write_iter = ll_file_write_iter,
+#else /* !HAVE_FILE_OPERATIONS_READ_WRITE_ITER */
+ .read = ll_file_read,
+ .aio_read = ll_file_aio_read,
+ .write = ll_file_write,
+ .aio_write = ll_file_aio_write,
+#endif /* HAVE_FILE_OPERATIONS_READ_WRITE_ITER */
+ .unlocked_ioctl = ll_file_ioctl,
+ .open = ll_file_open,
+ .release = ll_file_release,
+ .mmap = ll_file_mmap,
+ .llseek = ll_file_seek,
+ .splice_read = ll_file_splice_read,
+ .fsync = ll_fsync,
+ .flush = ll_flush,
+ .flock = ll_file_noflock,
+ .lock = ll_file_noflock
};
struct inode_operations ll_file_inode_operations = {
int ll_layout_conf(struct inode *inode, const struct cl_object_conf *conf)
{
struct ll_inode_info *lli = ll_i2info(inode);
- struct cl_env_nest nest;
+ struct cl_object *obj = lli->lli_clob;
struct lu_env *env;
- int result;
+ int rc;
+ __u16 refcheck;
ENTRY;
- if (lli->lli_clob == NULL)
+ if (obj == NULL)
RETURN(0);
- env = cl_env_nested_get(&nest);
+ env = cl_env_get(&refcheck);
if (IS_ERR(env))
RETURN(PTR_ERR(env));
- result = cl_conf_set(env, lli->lli_clob, conf);
- cl_env_nested_put(&nest, env);
+ rc = cl_conf_set(env, lli->lli_clob, conf);
+ if (rc < 0)
+ GOTO(out, rc);
if (conf->coc_opc == OBJECT_CONF_SET) {
struct ldlm_lock *lock = conf->coc_lock;
+ struct cl_layout cl = {
+ .cl_layout_gen = 0,
+ };
LASSERT(lock != NULL);
LASSERT(ldlm_has_layout(lock));
- if (result == 0) {
- struct lustre_md *md = conf->u.coc_md;
- __u32 gen = LL_LAYOUT_GEN_EMPTY;
-
- /* it can only be allowed to match after layout is
- * applied to inode otherwise false layout would be
- * seen. Applying layout shoud happen before dropping
- * the intent lock. */
- ldlm_lock_allow_match(lock);
-
- lli->lli_has_smd = lsm_has_objects(md->lsm);
- if (md->lsm != NULL)
- gen = md->lsm->lsm_layout_gen;
-
- CDEBUG(D_VFSTRACE,
- DFID ": layout version change: %u -> %u\n",
- PFID(&lli->lli_fid), ll_layout_version_get(lli),
- gen);
- ll_layout_version_set(lli, gen);
- }
+
+ /* it can only be allowed to match after layout is
+ * applied to inode otherwise false layout would be
+ * seen. Applying layout shoud happen before dropping
+ * the intent lock. */
+ ldlm_lock_allow_match(lock);
+
+ rc = cl_object_layout_get(env, obj, &cl);
+ if (rc < 0)
+ GOTO(out, rc);
+
+ CDEBUG(D_VFSTRACE,
+ DFID": layout version change: %u -> %u\n",
+ PFID(&lli->lli_fid), ll_layout_version_get(lli),
+ cl.cl_layout_gen);
+ ll_layout_version_set(lli, cl.cl_layout_gen);
}
- RETURN(result);
+
+out:
+ cl_env_put(env, &refcheck);
+
+ RETURN(rc);
}
/* Fetch layout from MDT with getxattr request, if it's not ready yet */
{
struct ll_sb_info *sbi = ll_i2sbi(inode);
- struct obd_capa *oc;
struct ptlrpc_request *req;
struct mdt_body *body;
void *lvbdata;
PFID(ll_inode2fid(inode)), ldlm_is_lvb_ready(lock),
lock->l_lvb_data, lock->l_lvb_len);
- if ((lock->l_lvb_data != NULL) && ldlm_is_lvb_ready(lock))
+ if (lock->l_lvb_data != NULL)
RETURN(0);
/* if layout lock was granted right away, the layout is returned
* blocked and then granted via completion ast, we have to fetch
* layout here. Please note that we can't use the LVB buffer in
* completion AST because it doesn't have a large enough buffer */
- oc = ll_mdscapa_get(inode);
rc = ll_get_default_mdsize(sbi, &lmmsize);
if (rc == 0)
- rc = md_getxattr(sbi->ll_md_exp, ll_inode2fid(inode), oc,
+ rc = md_getxattr(sbi->ll_md_exp, ll_inode2fid(inode),
OBD_MD_FLXATTR, XATTR_NAME_LOV, NULL, 0,
lmmsize, 0, &req);
- capa_put(oc);
if (rc < 0)
RETURN(rc);
memcpy(lvbdata, lmm, lmmsize);
lock_res_and_lock(lock);
- if (lock->l_lvb_data != NULL)
- OBD_FREE_LARGE(lock->l_lvb_data, lock->l_lvb_len);
-
- lock->l_lvb_data = lvbdata;
- lock->l_lvb_len = lmmsize;
+ if (unlikely(lock->l_lvb_data == NULL)) {
+ lock->l_lvb_type = LVB_T_LAYOUT;
+ lock->l_lvb_data = lvbdata;
+ lock->l_lvb_len = lmmsize;
+ lvbdata = NULL;
+ }
unlock_res_and_lock(lock);
+ if (lvbdata)
+ OBD_FREE_LARGE(lvbdata, lmmsize);
+
EXIT;
out:
* Apply the layout to the inode. Layout lock is held and will be released
* in this function.
*/
-static int ll_layout_lock_set(struct lustre_handle *lockh, ldlm_mode_t mode,
- struct inode *inode, __u32 *gen, bool reconf)
+static int ll_layout_lock_set(struct lustre_handle *lockh, enum ldlm_mode mode,
+ struct inode *inode)
{
struct ll_inode_info *lli = ll_i2info(inode);
struct ll_sb_info *sbi = ll_i2sbi(inode);
struct ldlm_lock *lock;
- struct lustre_md md = { NULL };
struct cl_object_conf conf;
int rc = 0;
bool lvb_ready;
LASSERT(lock != NULL);
LASSERT(ldlm_has_layout(lock));
- LDLM_DEBUG(lock, "file "DFID"(%p) being reconfigured: %d",
- PFID(&lli->lli_fid), inode, reconf);
+ LDLM_DEBUG(lock, "file "DFID"(%p) being reconfigured",
+ PFID(&lli->lli_fid), inode);
/* in case this is a caching lock and reinstate with new inode */
- md_set_lock_data(sbi->ll_md_exp, &lockh->cookie, inode, NULL);
+ md_set_lock_data(sbi->ll_md_exp, lockh, inode, NULL);
lock_res_and_lock(lock);
lvb_ready = ldlm_is_lvb_ready(lock);
/* checking lvb_ready is racy but this is okay. The worst case is
* that multi processes may configure the file on the same time. */
- if (lvb_ready || !reconf) {
- rc = -ENODATA;
- if (lvb_ready) {
- /* layout_gen must be valid if layout lock is not
- * cancelled and stripe has already set */
- *gen = ll_layout_version_get(lli);
- rc = 0;
- }
- GOTO(out, rc);
- }
+ if (lvb_ready)
+ GOTO(out, rc = 0);
rc = ll_layout_fetch(inode, lock);
if (rc < 0)
GOTO(out, rc);
- /* for layout lock, lmm is returned in lock's lvb.
+ /* for layout lock, lmm is stored in lock's lvb.
* lvb_data is immutable if the lock is held so it's safe to access it
- * without res lock. See the description in ldlm_lock_decref_internal()
- * for the condition to free lvb_data of layout lock */
- if (lock->l_lvb_data != NULL) {
- rc = obd_unpackmd(sbi->ll_dt_exp, &md.lsm,
- lock->l_lvb_data, lock->l_lvb_len);
- if (rc >= 0) {
- *gen = LL_LAYOUT_GEN_EMPTY;
- if (md.lsm != NULL)
- *gen = md.lsm->lsm_layout_gen;
- rc = 0;
- } else {
- CERROR("%s: file "DFID" unpackmd error: %d\n",
- ll_get_fsname(inode->i_sb, NULL, 0),
- PFID(&lli->lli_fid), rc);
- }
- }
- if (rc < 0)
- GOTO(out, rc);
-
- /* set layout to file. Unlikely this will fail as old layout was
+ * without res lock.
+ *
+ * set layout to file. Unlikely this will fail as old layout was
* surely eliminated */
memset(&conf, 0, sizeof conf);
conf.coc_opc = OBJECT_CONF_SET;
conf.coc_inode = inode;
conf.coc_lock = lock;
- conf.u.coc_md = &md;
+ conf.u.coc_layout.lb_buf = lock->l_lvb_data;
+ conf.u.coc_layout.lb_len = lock->l_lvb_len;
rc = ll_layout_conf(inode, &conf);
- if (md.lsm != NULL)
- obd_free_memmd(sbi->ll_dt_exp, &md.lsm);
-
/* refresh layout failed, need to wait */
wait_layout = rc == -EBUSY;
EXIT;
RETURN(rc);
}
-/**
- * This function checks if there exists a LAYOUT lock on the client side,
- * or enqueues it if it doesn't have one in cache.
- *
- * This function will not hold layout lock so it may be revoked any time after
- * this function returns. Any operations depend on layout should be redone
- * in that case.
- *
- * This function should be called before lov_io_init() to get an uptodate
- * layout version, the caller should save the version number and after IO
- * is finished, this function should be called again to verify that layout
- * is not changed during IO time.
- */
-int ll_layout_refresh(struct inode *inode, __u32 *gen)
+static int ll_layout_refresh_locked(struct inode *inode)
{
struct ll_inode_info *lli = ll_i2info(inode);
struct ll_sb_info *sbi = ll_i2sbi(inode);
struct md_op_data *op_data;
- struct lookup_intent it;
- struct lustre_handle lockh;
- ldlm_mode_t mode;
+ struct lookup_intent it;
+ struct lustre_handle lockh;
+ enum ldlm_mode mode;
struct ldlm_enqueue_info einfo = {
.ei_type = LDLM_IBITS,
.ei_mode = LCK_CR,
int rc;
ENTRY;
- *gen = ll_layout_version_get(lli);
- if (!(sbi->ll_flags & LL_SBI_LAYOUT_LOCK) || *gen != LL_LAYOUT_GEN_NONE)
- RETURN(0);
-
- /* sanity checks */
- LASSERT(fid_is_sane(ll_inode2fid(inode)));
- LASSERT(S_ISREG(inode->i_mode));
-
- /* take layout lock mutex to enqueue layout lock exclusively. */
- mutex_lock(&lli->lli_layout_mutex);
-
again:
/* mostly layout lock is caching on the local side, so try to match
* it before grabbing layout lock mutex. */
mode = ll_take_md_lock(inode, MDS_INODELOCK_LAYOUT, &lockh, 0,
LCK_CR | LCK_CW | LCK_PR | LCK_PW);
if (mode != 0) { /* hit cached lock */
- rc = ll_layout_lock_set(&lockh, mode, inode, gen, true);
+ rc = ll_layout_lock_set(&lockh, mode, inode);
if (rc == -EAGAIN)
goto again;
- mutex_unlock(&lli->lli_layout_mutex);
RETURN(rc);
}
op_data = ll_prep_md_op_data(NULL, inode, inode, NULL,
0, 0, LUSTRE_OPC_ANY, NULL);
- if (IS_ERR(op_data)) {
- mutex_unlock(&lli->lli_layout_mutex);
+ if (IS_ERR(op_data))
RETURN(PTR_ERR(op_data));
- }
/* have to enqueue one */
memset(&it, 0, sizeof(it));
PFID(&lli->lli_fid), inode);
rc = md_enqueue(sbi->ll_md_exp, &einfo, NULL, &it, op_data, &lockh, 0);
- if (it.d.lustre.it_data != NULL)
- ptlrpc_req_finished(it.d.lustre.it_data);
- it.d.lustre.it_data = NULL;
+ if (it.it_request != NULL)
+ ptlrpc_req_finished(it.it_request);
+ it.it_request = NULL;
ll_finish_md_op_data(op_data);
- mode = it.d.lustre.it_lock_mode;
- it.d.lustre.it_lock_mode = 0;
+ mode = it.it_lock_mode;
+ it.it_lock_mode = 0;
ll_intent_drop_lock(&it);
if (rc == 0) {
/* set lock data in case this is a new lock */
ll_set_lock_data(sbi->ll_md_exp, inode, &it, NULL);
- rc = ll_layout_lock_set(&lockh, mode, inode, gen, true);
+ rc = ll_layout_lock_set(&lockh, mode, inode);
if (rc == -EAGAIN)
goto again;
}
+
+ RETURN(rc);
+}
+
+/**
+ * This function checks if there exists a LAYOUT lock on the client side,
+ * or enqueues it if it doesn't have one in cache.
+ *
+ * This function will not hold layout lock so it may be revoked any time after
+ * this function returns. Any operations depend on layout should be redone
+ * in that case.
+ *
+ * This function should be called before lov_io_init() to get an uptodate
+ * layout version, the caller should save the version number and after IO
+ * is finished, this function should be called again to verify that layout
+ * is not changed during IO time.
+ */
+int ll_layout_refresh(struct inode *inode, __u32 *gen)
+{
+ struct ll_inode_info *lli = ll_i2info(inode);
+ struct ll_sb_info *sbi = ll_i2sbi(inode);
+ int rc;
+ ENTRY;
+
+ *gen = ll_layout_version_get(lli);
+ if (!(sbi->ll_flags & LL_SBI_LAYOUT_LOCK) || *gen != CL_LAYOUT_GEN_NONE)
+ RETURN(0);
+
+ /* sanity checks */
+ LASSERT(fid_is_sane(ll_inode2fid(inode)));
+ LASSERT(S_ISREG(inode->i_mode));
+
+ /* take layout lock mutex to enqueue layout lock exclusively. */
+ mutex_lock(&lli->lli_layout_mutex);
+
+ rc = ll_layout_refresh_locked(inode);
+ if (rc < 0)
+ GOTO(out, rc);
+
+ *gen = ll_layout_version_get(lli);
+out:
mutex_unlock(&lli->lli_layout_mutex);
RETURN(rc);
hur->hur_user_item[0].hui_extent.offset = offset;
hur->hur_user_item[0].hui_extent.length = length;
hur->hur_request.hr_itemcount = 1;
- rc = obd_iocontrol(LL_IOC_HSM_REQUEST, cl_i2sbi(inode)->ll_md_exp,
+ rc = obd_iocontrol(LL_IOC_HSM_REQUEST, ll_i2sbi(inode)->ll_md_exp,
len, hur, NULL);
OBD_FREE(hur, len);
RETURN(rc);
}
-
git://git.whamcloud.com / fs / lustre-release.git / blobdiff