* Copyright (c) 2008, 2010, Oracle and/or its affiliates. All rights reserved.
* Use is subject to license terms.
*
- * Copyright (c) 2011, 2014, Intel Corporation.
+ * Copyright (c) 2011, 2015, Intel Corporation.
*/
/*
* This file is part of Lustre, http://www.lustre.org/
case CIT_WRITE:
/* don't need lock here to check lli_layout_gen as we have held
* extent lock and GROUP lock has to hold to swap layout */
- if (ll_layout_version_get(lli) != vio->vui_layout_gen) {
+ if (ll_layout_version_get(lli) != vio->vui_layout_gen ||
+ OBD_FAIL_CHECK_RESET(OBD_FAIL_LLITE_LOST_LAYOUT, 0)) {
io->ci_need_restart = 1;
- /* this will return application a short read/write */
+ /* this will cause a short read/write */
io->ci_continue = 0;
rc = false;
}
if (vio->vui_fd && (vio->vui_fd->fd_flags & LL_FILE_GROUP_LOCKED)) {
descr->cld_mode = CLM_GROUP;
descr->cld_gid = vio->vui_fd->fd_grouplock.lg_gid;
+ enqflags |= CEF_LOCK_MATCH;
} else {
descr->cld_mode = mode;
}
struct vvp_io *vio = cl2vvp_io(env, ios);
struct inode *inode = vvp_object_inode(ios->cis_obj);
- LASSERT(inode == vio->vui_fd->fd_file->f_dentry->d_inode);
+ LASSERT(inode == vio->vui_fd->fd_file->f_path.dentry->d_inode);
vio->u.fault.ft_mtime = LTIME_S(inode->i_mtime);
return 0;
io->ci_ignore_layout, io->ci_verify_layout,
vio->vui_layout_gen, io->ci_restore_needed);
- if (io->ci_restore_needed == 1) {
+ if (io->ci_restore_needed) {
int rc;
/* file was detected release, we need to restore it
/* if restore registration failed, no restart,
* we will return -ENODATA */
/* The layout will change after restore, so we need to
- * block on layout lock hold by the MDT
+ * block on layout lock held by the MDT
* as MDT will not send new layout in lvb (see LU-3124)
* we have to explicitly fetch it, all this will be done
* by ll_layout_refresh()
/* today successful restore is the only possible
* case */
/* restore was done, clear restoring state */
- ll_i2info(vvp_object_inode(obj))->lli_flags &=
- ~LLIF_FILE_RESTORING;
+ ll_file_clear_flag(ll_i2info(vvp_object_inode(obj)),
+ LLIF_FILE_RESTORING);
}
}
}
struct vvp_io *vio, struct cl_io *io)
{
struct vvp_thread_info *vti = vvp_env_info(env);
- struct mm_struct *mm = current->mm;
- struct vm_area_struct *vma;
- struct cl_lock_descr *descr = &vti->vti_descr;
- ldlm_policy_data_t policy;
- unsigned long addr;
- unsigned long seg;
- ssize_t count;
- int result = 0;
- ENTRY;
-
- LASSERT(io->ci_type == CIT_READ || io->ci_type == CIT_WRITE);
-
- if (!cl_is_normalio(env, io))
- RETURN(0);
-
- if (vio->vui_iov == NULL) /* nfs or loop back device write */
+ struct mm_struct *mm = current->mm;
+ struct vm_area_struct *vma;
+ struct cl_lock_descr *descr = &vti->vti_descr;
+ union ldlm_policy_data policy;
+#ifdef HAVE_FILE_OPERATIONS_READ_WRITE_ITER
+ struct iovec iov;
+ struct iov_iter i;
+#else
+ unsigned long seg;
+#endif
+ int result = 0;
+ ENTRY;
+
+ LASSERT(io->ci_type == CIT_READ || io->ci_type == CIT_WRITE);
+
+ if (!cl_is_normalio(env, io))
+ RETURN(0);
+
+ /* nfs or loop back device write */
+ if (vio->vui_iter == NULL)
RETURN(0);
/* No MM (e.g. NFS)? No vmas too. */
if (mm == NULL)
RETURN(0);
- for (seg = 0; seg < vio->vui_nrsegs; seg++) {
- const struct iovec *iv = &vio->vui_iov[seg];
+#ifdef HAVE_FILE_OPERATIONS_READ_WRITE_ITER
+ iov_for_each(iov, i, *(vio->vui_iter)) {
+#else /* !HAVE_FILE_OPERATIONS_READ_WRITE_ITER */
+ for (seg = 0; seg < vio->vui_iter->nr_segs; seg++) {
+ const struct iovec iov = vio->vui_iter->iov[seg];
+#endif /* HAVE_FILE_OPERATIONS_READ_WRITE_ITER */
+ unsigned long addr = (unsigned long)iov.iov_base;
+ size_t count = iov.iov_len;
- addr = (unsigned long)iv->iov_base;
- count = iv->iov_len;
if (count == 0)
continue;
down_read(&mm->mmap_sem);
while((vma = our_vma(mm, addr, count)) != NULL) {
- struct inode *inode = vma->vm_file->f_dentry->d_inode;
+ struct dentry *de = vma->vm_file->f_path.dentry;
+ struct inode *inode = de->d_inode;
int flags = CEF_MUST;
if (ll_file_nolock(vma->vm_file)) {
struct vvp_io *vio = cl2vvp_io(env, ios);
struct cl_io *io = ios->cis_io;
struct cl_object *obj = ios->cis_io->ci_obj;
-
+#ifndef HAVE_FILE_OPERATIONS_READ_WRITE_ITER
+ struct iovec *iov;
+#endif
CLOBINVRNT(env, obj, vvp_object_invariant(obj));
if (!cl_is_normalio(env, io))
return;
- LASSERT(vio->vui_tot_nrsegs >= vio->vui_nrsegs);
+#ifdef HAVE_FILE_OPERATIONS_READ_WRITE_ITER
+ vio->vui_tot_count -= nob;
+ iov_iter_reexpand(vio->vui_iter, vio->vui_tot_count);
+#else /* !HAVE_FILE_OPERATIONS_READ_WRITE_ITER */
+ LASSERT(vio->vui_tot_nrsegs >= vio->vui_iter->nr_segs);
LASSERT(vio->vui_tot_count >= nob);
- vio->vui_iov += vio->vui_nrsegs;
- vio->vui_tot_nrsegs -= vio->vui_nrsegs;
- vio->vui_tot_count -= nob;
-
- /* update the iov */
+ /* Restore the iov changed in vvp_io_update_iov() */
if (vio->vui_iov_olen > 0) {
- struct iovec *iv;
-
- vio->vui_iov--;
- vio->vui_tot_nrsegs++;
- iv = &vio->vui_iov[0];
- if (io->ci_continue) {
- iv->iov_base += iv->iov_len;
- LASSERT(vio->vui_iov_olen > iv->iov_len);
- iv->iov_len = vio->vui_iov_olen - iv->iov_len;
- } else {
- /* restore the iov_len, in case of restart io. */
- iv->iov_len = vio->vui_iov_olen;
- }
+ unsigned long idx = vio->vui_iter->nr_segs - 1;
+
+ /* In the latest kernels iov is const so that
+ * changes are done using iter helpers. In older
+ * kernels those helpers don't exist we lustre
+ * has to do some of the management of the iter
+ * itself. */
+ iov = (struct iovec *)&vio->vui_iter->iov[idx];
+ iov->iov_len = vio->vui_iov_olen;
vio->vui_iov_olen = 0;
}
+
+ /* In the latest kernels special helpers exist to help
+ * advance the iov but we don't have that in older kernels
+ * so we need to do the book keeping ourselves. */
+ iov = (struct iovec *)vio->vui_iter->iov;
+ while (nob > 0) {
+ if (iov->iov_len > nob) {
+ iov->iov_len -= nob;
+ iov->iov_base += nob;
+ break;
+ }
+
+ nob -= iov->iov_len;
+ iov++;
+ vio->vui_tot_nrsegs--;
+ }
+
+ vio->vui_iter->iov = iov;
+ vio->vui_tot_count -= nob;
+#endif /* HAVE_FILE_OPERATIONS_READ_WRITE_ITER */
}
static void vvp_io_update_iov(const struct lu_env *env,
struct vvp_io *vio, struct cl_io *io)
{
- int i;
size_t size = io->u.ci_rw.crw_count;
+#ifdef HAVE_FILE_OPERATIONS_READ_WRITE_ITER
+ if (!cl_is_normalio(env, io) || vio->vui_iter == NULL)
+ return;
+
+ iov_iter_truncate(vio->vui_iter, size);
+#else /* !HAVE_FILE_OPERATIONS_READ_WRITE_ITER */
+ unsigned long i;
vio->vui_iov_olen = 0;
if (!cl_is_normalio(env, io) || vio->vui_tot_nrsegs == 0)
return;
for (i = 0; i < vio->vui_tot_nrsegs; i++) {
- struct iovec *iv = &vio->vui_iov[i];
+ struct iovec *iv = (struct iovec *) &vio->vui_iter->iov[i];
if (iv->iov_len < size) {
size -= iv->iov_len;
}
}
- vio->vui_nrsegs = i + 1;
- LASSERTF(vio->vui_tot_nrsegs >= vio->vui_nrsegs,
+ vio->vui_iter->nr_segs = i + 1;
+ LASSERTF(vio->vui_tot_nrsegs >= vio->vui_iter->nr_segs,
"tot_nrsegs: %lu, nrsegs: %lu\n",
- vio->vui_tot_nrsegs, vio->vui_nrsegs);
+ vio->vui_tot_nrsegs, vio->vui_iter->nr_segs);
+#endif /* !HAVE_FILE_OPERATIONS_READ_WRITE_ITER */
}
static int vvp_io_rw_lock(const struct lu_env *env, struct cl_io *io,
return result;
}
-static int vvp_io_setattr_trunc(const struct lu_env *env,
- const struct cl_io_slice *ios,
- struct inode *inode, loff_t size)
-{
- inode_dio_wait(inode);
- return 0;
-}
-
static int vvp_io_setattr_time(const struct lu_env *env,
const struct cl_io_slice *ios)
{
static int vvp_io_setattr_start(const struct lu_env *env,
const struct cl_io_slice *ios)
{
- struct cl_io *io = ios->cis_io;
- struct inode *inode = vvp_object_inode(io->ci_obj);
- int result = 0;
+ struct cl_io *io = ios->cis_io;
+ struct inode *inode = vvp_object_inode(io->ci_obj);
+ struct ll_inode_info *lli = ll_i2info(inode);
mutex_lock(&inode->i_mutex);
- if (cl_io_is_trunc(io))
- result = vvp_io_setattr_trunc(env, ios, inode,
- io->u.ci_setattr.sa_attr.lvb_size);
- if (result == 0 && io->u.ci_setattr.sa_valid & TIMES_SET_FLAGS)
- result = vvp_io_setattr_time(env, ios);
- return result;
+ if (cl_io_is_trunc(io)) {
+ down_write(&lli->lli_trunc_sem);
+ inode_dio_wait(inode);
+ }
+
+ if (io->u.ci_setattr.sa_valid & TIMES_SET_FLAGS)
+ return vvp_io_setattr_time(env, ios);
+
+ return 0;
}
static void vvp_io_setattr_end(const struct lu_env *env,
const struct cl_io_slice *ios)
{
- struct cl_io *io = ios->cis_io;
- struct inode *inode = vvp_object_inode(io->ci_obj);
+ struct cl_io *io = ios->cis_io;
+ struct inode *inode = vvp_object_inode(io->ci_obj);
+ struct ll_inode_info *lli = ll_i2info(inode);
if (cl_io_is_trunc(io)) {
/* Truncate in memory pages - they must be clean pages
* because osc has already notified to destroy osc_extents. */
vvp_do_vmtruncate(inode, io->u.ci_setattr.sa_attr.lvb_size);
inode_dio_write_done(inode);
+ up_write(&lli->lli_trunc_sem);
}
mutex_unlock(&inode->i_mutex);
}
static void vvp_io_setattr_fini(const struct lu_env *env,
const struct cl_io_slice *ios)
{
+ bool restore_needed = ios->cis_io->ci_restore_needed;
+ struct inode *inode = vvp_object_inode(ios->cis_obj);
+
vvp_io_fini(env, ios);
+
+ if (restore_needed && !ios->cis_io->ci_restore_needed) {
+ /* restore finished, set data modified flag for HSM */
+ ll_file_set_flag(ll_i2info(inode), LLIF_DATA_MODIFIED);
+ }
}
static int vvp_io_read_start(const struct lu_env *env,
const struct cl_io_slice *ios)
{
- struct vvp_io *vio = cl2vvp_io(env, ios);
- struct cl_io *io = ios->cis_io;
- struct cl_object *obj = io->ci_obj;
- struct inode *inode = vvp_object_inode(obj);
- struct file *file = vio->vui_fd->fd_file;
+ struct vvp_io *vio = cl2vvp_io(env, ios);
+ struct cl_io *io = ios->cis_io;
+ struct cl_object *obj = io->ci_obj;
+ struct inode *inode = vvp_object_inode(obj);
+ struct ll_inode_info *lli = ll_i2info(inode);
+ struct file *file = vio->vui_fd->fd_file;
int result;
loff_t pos = io->u.ci_rd.rd.crw_pos;
CLOBINVRNT(env, obj, vvp_object_invariant(obj));
- CDEBUG(D_VFSTRACE, "read: -> [%lli, %lli)\n", pos, pos + cnt);
+ CDEBUG(D_VFSTRACE, "read: -> [%lli, %lli)\n", pos, pos + cnt);
+
+ if (vio->vui_io_subtype == IO_NORMAL)
+ down_read(&lli->lli_trunc_sem);
if (!can_populate_pages(env, io, inode))
return 0;
result = vvp_prep_size(env, obj, io, pos, tot, &exceed);
- if (result != 0)
- return result;
- else if (exceed != 0)
- goto out;
+ if (result != 0)
+ return result;
+ else if (exceed != 0)
+ goto out;
- LU_OBJECT_HEADER(D_INODE, env, &obj->co_lu,
- "Read ino %lu, %lu bytes, offset %lld, size %llu\n",
- inode->i_ino, cnt, pos, i_size_read(inode));
+ LU_OBJECT_HEADER(D_INODE, env, &obj->co_lu,
+ "Read ino %lu, %lu bytes, offset %lld, size %llu\n",
+ inode->i_ino, cnt, pos, i_size_read(inode));
/* turn off the kernel's read-ahead */
vio->vui_fd->fd_file->f_ra.ra_pages = 0;
switch (vio->vui_io_subtype) {
case IO_NORMAL:
LASSERT(vio->vui_iocb->ki_pos == pos);
+#ifdef HAVE_FILE_OPERATIONS_READ_WRITE_ITER
+ result = generic_file_read_iter(vio->vui_iocb, vio->vui_iter);
+#else
result = generic_file_aio_read(vio->vui_iocb,
- vio->vui_iov, vio->vui_nrsegs,
+ vio->vui_iter->iov,
+ vio->vui_iter->nr_segs,
vio->vui_iocb->ki_pos);
+#endif
break;
case IO_SPLICE:
result = generic_file_splice_read(file, &pos,
static int vvp_io_write_start(const struct lu_env *env,
const struct cl_io_slice *ios)
{
- struct vvp_io *vio = cl2vvp_io(env, ios);
- struct cl_io *io = ios->cis_io;
- struct cl_object *obj = io->ci_obj;
- struct inode *inode = vvp_object_inode(obj);
- ssize_t result = 0;
- loff_t pos = io->u.ci_wr.wr.crw_pos;
- size_t cnt = io->u.ci_wr.wr.crw_count;
+ struct vvp_io *vio = cl2vvp_io(env, ios);
+ struct cl_io *io = ios->cis_io;
+ struct cl_object *obj = io->ci_obj;
+ struct inode *inode = vvp_object_inode(obj);
+ struct ll_inode_info *lli = ll_i2info(inode);
+ ssize_t result = 0;
+ loff_t pos = io->u.ci_wr.wr.crw_pos;
+ size_t cnt = io->u.ci_wr.wr.crw_count;
+
+ ENTRY;
- ENTRY;
+ if (vio->vui_io_subtype == IO_NORMAL)
+ down_read(&lli->lli_trunc_sem);
if (!can_populate_pages(env, io, inode))
RETURN(0);
CDEBUG(D_VFSTRACE, "write: [%lli, %lli)\n", pos, pos + (long long)cnt);
- if (vio->vui_iov == NULL) {
+ /* The maximum Lustre file size is variable, based on the OST maximum
+ * object size and number of stripes. This needs another check in
+ * addition to the VFS checks earlier. */
+ if (pos + cnt > ll_file_maxbytes(inode)) {
+ CDEBUG(D_INODE,
+ "%s: file "DFID" offset %llu > maxbytes "LPU64"\n",
+ ll_get_fsname(inode->i_sb, NULL, 0),
+ PFID(ll_inode2fid(inode)), pos + cnt,
+ ll_file_maxbytes(inode));
+ RETURN(-EFBIG);
+ }
+
+ if (vio->vui_iter == NULL) {
/* from a temp io in ll_cl_init(). */
result = 0;
} else {
* consistency, proper locking to protect against writes,
* trucates, etc. is handled in the higher layers of lustre.
*/
+#ifdef HAVE_FILE_OPERATIONS_READ_WRITE_ITER
+ result = generic_file_write_iter(vio->vui_iocb, vio->vui_iter);
+#else
result = __generic_file_aio_write(vio->vui_iocb,
- vio->vui_iov, vio->vui_nrsegs,
+ vio->vui_iter->iov,
+ vio->vui_iter->nr_segs,
&vio->vui_iocb->ki_pos);
+#endif
if (result > 0 || result == -EIOCBQUEUED) {
ssize_t err;
}
}
if (result > 0) {
- struct ll_inode_info *lli = ll_i2info(inode);
-
- spin_lock(&lli->lli_lock);
- lli->lli_flags |= LLIF_DATA_MODIFIED;
- spin_unlock(&lli->lli_lock);
+ ll_file_set_flag(ll_i2info(inode), LLIF_DATA_MODIFIED);
if (result < cnt)
io->ci_continue = 0;
RETURN(result);
}
+static void vvp_io_rw_end(const struct lu_env *env,
+ const struct cl_io_slice *ios)
+{
+ struct vvp_io *vio = cl2vvp_io(env, ios);
+ struct inode *inode = vvp_object_inode(ios->cis_obj);
+ struct ll_inode_info *lli = ll_i2info(inode);
+
+ if (vio->vui_io_subtype == IO_NORMAL)
+ up_read(&lli->lli_trunc_sem);
+}
+
static int vvp_io_kernel_fault(struct vvp_fault_io *cfio)
{
struct vm_fault *vmf = cfio->ft_vmf;
static int vvp_io_fault_start(const struct lu_env *env,
const struct cl_io_slice *ios)
{
- struct vvp_io *vio = cl2vvp_io(env, ios);
- struct cl_io *io = ios->cis_io;
- struct cl_object *obj = io->ci_obj;
- struct inode *inode = vvp_object_inode(obj);
- struct cl_fault_io *fio = &io->u.ci_fault;
- struct vvp_fault_io *cfio = &vio->u.fault;
- loff_t offset;
- int result = 0;
- struct page *vmpage = NULL;
- struct cl_page *page;
- loff_t size;
- pgoff_t last_index;
+ struct vvp_io *vio = cl2vvp_io(env, ios);
+ struct cl_io *io = ios->cis_io;
+ struct cl_object *obj = io->ci_obj;
+ struct inode *inode = vvp_object_inode(obj);
+ struct ll_inode_info *lli = ll_i2info(inode);
+ struct cl_fault_io *fio = &io->u.ci_fault;
+ struct vvp_fault_io *cfio = &vio->u.fault;
+ loff_t offset;
+ int result = 0;
+ struct page *vmpage = NULL;
+ struct cl_page *page;
+ loff_t size;
+ pgoff_t last_index;
ENTRY;
- if (fio->ft_executable &&
- LTIME_S(inode->i_mtime) != vio->u.fault.ft_mtime)
- CWARN("binary "DFID
- " changed while waiting for the page fault lock\n",
- PFID(lu_object_fid(&obj->co_lu)));
+ down_read(&lli->lli_trunc_sem);
/* offset of the last byte on the page */
offset = cl_offset(obj, fio->ft_index + 1) - 1;
return result;
}
+static void vvp_io_fault_end(const struct lu_env *env,
+ const struct cl_io_slice *ios)
+{
+ struct inode *inode = vvp_object_inode(ios->cis_obj);
+ struct ll_inode_info *lli = ll_i2info(inode);
+
+ CLOBINVRNT(env, ios->cis_io->ci_obj,
+ vvp_object_invariant(ios->cis_io->ci_obj));
+ up_read(&lli->lli_trunc_sem);
+}
+
static int vvp_io_fsync_start(const struct lu_env *env,
const struct cl_io_slice *ios)
{
RETURN(result);
}
-static void vvp_io_end(const struct lu_env *env, const struct cl_io_slice *ios)
-{
- CLOBINVRNT(env, ios->cis_io->ci_obj,
- vvp_object_invariant(ios->cis_io->ci_obj));
-}
-
static const struct cl_io_operations vvp_io_ops = {
.op = {
[CIT_READ] = {
.cio_fini = vvp_io_fini,
.cio_lock = vvp_io_read_lock,
.cio_start = vvp_io_read_start,
+ .cio_end = vvp_io_rw_end,
.cio_advance = vvp_io_advance,
},
[CIT_WRITE] = {
.cio_iter_fini = vvp_io_write_iter_fini,
.cio_lock = vvp_io_write_lock,
.cio_start = vvp_io_write_start,
+ .cio_end = vvp_io_rw_end,
.cio_advance = vvp_io_advance,
},
[CIT_SETATTR] = {
.cio_iter_init = vvp_io_fault_iter_init,
.cio_lock = vvp_io_fault_lock,
.cio_start = vvp_io_fault_start,
- .cio_end = vvp_io_end,
+ .cio_end = vvp_io_fault_end,
},
[CIT_FSYNC] = {
- .cio_start = vvp_io_fsync_start,
- .cio_fini = vvp_io_fini
+ .cio_start = vvp_io_fsync_start,
+ .cio_fini = vvp_io_fini
+ },
+ [CIT_MISC] = {
+ .cio_fini = vvp_io_fini
+ },
+ [CIT_LADVISE] = {
+ .cio_fini = vvp_io_fini
},
- [CIT_MISC] = {
- .cio_fini = vvp_io_fini
- }
- },
+ },
.cio_read_ahead = vvp_io_read_ahead
};
result = 1;
else {
vio->vui_tot_count = count;
+#ifndef HAVE_FILE_OPERATIONS_READ_WRITE_ITER
vio->vui_tot_nrsegs = 0;
+#endif
}
/* for read/write, we store the jobid in the inode, and
git://git.whamcloud.com / fs / lustre-release.git / blobdiff