struct ceph_mount_options *opt = mdsc->fsc->mount_options;
ci->i_hold_caps_max = round_jiffies(jiffies +
opt->caps_wanted_delay_max * HZ);
- dout("__cap_set_timeouts %p %lu\n", &ci->vfs_inode,
+ dout("__cap_set_timeouts %p %lu\n", &ci->netfs.inode,
ci->i_hold_caps_max - jiffies);
}
static void __cap_delay_requeue(struct ceph_mds_client *mdsc,
struct ceph_inode_info *ci)
{
- dout("__cap_delay_requeue %p flags 0x%lx at %lu\n", &ci->vfs_inode,
+ dout("__cap_delay_requeue %p flags 0x%lx at %lu\n", &ci->netfs.inode,
ci->i_ceph_flags, ci->i_hold_caps_max);
if (!mdsc->stopping) {
spin_lock(&mdsc->cap_delay_lock);
static void __cap_delay_requeue_front(struct ceph_mds_client *mdsc,
struct ceph_inode_info *ci)
{
- dout("__cap_delay_requeue_front %p\n", &ci->vfs_inode);
+ dout("__cap_delay_requeue_front %p\n", &ci->netfs.inode);
spin_lock(&mdsc->cap_delay_lock);
ci->i_ceph_flags |= CEPH_I_FLUSH;
if (!list_empty(&ci->i_cap_delay_list))
static void __cap_delay_cancel(struct ceph_mds_client *mdsc,
struct ceph_inode_info *ci)
{
- dout("__cap_delay_cancel %p\n", &ci->vfs_inode);
+ dout("__cap_delay_cancel %p\n", &ci->netfs.inode);
if (list_empty(&ci->i_cap_delay_list))
return;
spin_lock(&mdsc->cap_delay_lock);
* Each time we receive FILE_CACHE anew, we increment
* i_rdcache_gen.
*/
- if (S_ISREG(ci->vfs_inode.i_mode) &&
+ if (S_ISREG(ci->netfs.inode.i_mode) &&
(issued & (CEPH_CAP_FILE_CACHE|CEPH_CAP_FILE_LAZYIO)) &&
(had & (CEPH_CAP_FILE_CACHE|CEPH_CAP_FILE_LAZYIO)) == 0) {
ci->i_rdcache_gen++;
if ((issued & CEPH_CAP_FILE_SHARED) != (had & CEPH_CAP_FILE_SHARED)) {
if (issued & CEPH_CAP_FILE_SHARED)
atomic_inc(&ci->i_shared_gen);
- if (S_ISDIR(ci->vfs_inode.i_mode)) {
- dout(" marking %p NOT complete\n", &ci->vfs_inode);
+ if (S_ISDIR(ci->netfs.inode.i_mode)) {
+ dout(" marking %p NOT complete\n", &ci->netfs.inode);
__ceph_dir_clear_complete(ci);
}
}
/* Wipe saved layout if we're losing DIR_CREATE caps */
- if (S_ISDIR(ci->vfs_inode.i_mode) && (had & CEPH_CAP_DIR_CREATE) &&
+ if (S_ISDIR(ci->netfs.inode.i_mode) && (had & CEPH_CAP_DIR_CREATE) &&
!(issued & CEPH_CAP_DIR_CREATE)) {
ceph_put_string(rcu_dereference_raw(ci->i_cached_layout.pool_ns));
memset(&ci->i_cached_layout, 0, sizeof(ci->i_cached_layout));
if (cap->cap_gen < gen || time_after_eq(jiffies, ttl)) {
dout("__cap_is_valid %p cap %p issued %s "
- "but STALE (gen %u vs %u)\n", &cap->ci->vfs_inode,
+ "but STALE (gen %u vs %u)\n", &cap->ci->netfs.inode,
cap, ceph_cap_string(cap->issued), cap->cap_gen, gen);
return 0;
}
if (!__cap_is_valid(cap))
continue;
dout("__ceph_caps_issued %p cap %p issued %s\n",
- &ci->vfs_inode, cap, ceph_cap_string(cap->issued));
+ &ci->netfs.inode, cap, ceph_cap_string(cap->issued));
have |= cap->issued;
if (implemented)
*implemented |= cap->implemented;
spin_lock(&s->s_cap_lock);
if (!s->s_cap_iterator) {
- dout("__touch_cap %p cap %p mds%d\n", &cap->ci->vfs_inode, cap,
+ dout("__touch_cap %p cap %p mds%d\n", &cap->ci->netfs.inode, cap,
s->s_mds);
list_move_tail(&cap->session_caps, &s->s_caps);
} else {
dout("__touch_cap %p cap %p mds%d NOP, iterating over caps\n",
- &cap->ci->vfs_inode, cap, s->s_mds);
+ &cap->ci->netfs.inode, cap, s->s_mds);
}
spin_unlock(&s->s_cap_lock);
}
if ((have & mask) == mask) {
dout("__ceph_caps_issued_mask ino 0x%llx snap issued %s"
- " (mask %s)\n", ceph_ino(&ci->vfs_inode),
+ " (mask %s)\n", ceph_ino(&ci->netfs.inode),
ceph_cap_string(have),
ceph_cap_string(mask));
return 1;
continue;
if ((cap->issued & mask) == mask) {
dout("__ceph_caps_issued_mask ino 0x%llx cap %p issued %s"
- " (mask %s)\n", ceph_ino(&ci->vfs_inode), cap,
+ " (mask %s)\n", ceph_ino(&ci->netfs.inode), cap,
ceph_cap_string(cap->issued),
ceph_cap_string(mask));
if (touch)
have |= cap->issued;
if ((have & mask) == mask) {
dout("__ceph_caps_issued_mask ino 0x%llx combo issued %s"
- " (mask %s)\n", ceph_ino(&ci->vfs_inode),
+ " (mask %s)\n", ceph_ino(&ci->netfs.inode),
ceph_cap_string(cap->issued),
ceph_cap_string(mask));
if (touch) {
int __ceph_caps_issued_mask_metric(struct ceph_inode_info *ci, int mask,
int touch)
{
- struct ceph_fs_client *fsc = ceph_sb_to_client(ci->vfs_inode.i_sb);
+ struct ceph_fs_client *fsc = ceph_sb_to_client(ci->netfs.inode.i_sb);
int r;
r = __ceph_caps_issued_mask(ci, mask, touch);
int ceph_caps_revoking(struct ceph_inode_info *ci, int mask)
{
- struct inode *inode = &ci->vfs_inode;
+ struct inode *inode = &ci->netfs.inode;
int ret;
spin_lock(&ci->i_ceph_lock);
if (ci->i_rd_ref)
used |= CEPH_CAP_FILE_RD;
if (ci->i_rdcache_ref ||
- (S_ISREG(ci->vfs_inode.i_mode) &&
- ci->vfs_inode.i_data.nrpages))
+ (S_ISREG(ci->netfs.inode.i_mode) &&
+ ci->netfs.inode.i_data.nrpages))
used |= CEPH_CAP_FILE_CACHE;
if (ci->i_wr_ref)
used |= CEPH_CAP_FILE_WR;
const int WR_SHIFT = ffs(CEPH_FILE_MODE_WR);
const int LAZY_SHIFT = ffs(CEPH_FILE_MODE_LAZY);
struct ceph_mount_options *opt =
- ceph_inode_to_client(&ci->vfs_inode)->mount_options;
+ ceph_inode_to_client(&ci->netfs.inode)->mount_options;
unsigned long used_cutoff = jiffies - opt->caps_wanted_delay_max * HZ;
unsigned long idle_cutoff = jiffies - opt->caps_wanted_delay_min * HZ;
- if (S_ISDIR(ci->vfs_inode.i_mode)) {
+ if (S_ISDIR(ci->netfs.inode.i_mode)) {
int want = 0;
/* use used_cutoff here, to keep dir's wanted caps longer */
int __ceph_caps_wanted(struct ceph_inode_info *ci)
{
int w = __ceph_caps_file_wanted(ci) | __ceph_caps_used(ci);
- if (S_ISDIR(ci->vfs_inode.i_mode)) {
+ if (S_ISDIR(ci->netfs.inode.i_mode)) {
/* we want EXCL if holding caps of dir ops */
if (w & CEPH_CAP_ANY_DIR_OPS)
w |= CEPH_CAP_FILE_EXCL;
lockdep_assert_held(&ci->i_ceph_lock);
- dout("__ceph_remove_cap %p from %p\n", cap, &ci->vfs_inode);
+ dout("__ceph_remove_cap %p from %p\n", cap, &ci->netfs.inode);
- mdsc = ceph_inode_to_client(&ci->vfs_inode)->mdsc;
+ mdsc = ceph_inode_to_client(&ci->netfs.inode)->mdsc;
/* remove from inode's cap rbtree, and clear auth cap */
rb_erase(&cap->ci_node, &ci->i_caps);
* keep i_snap_realm.
*/
if (ci->i_wr_ref == 0 && ci->i_snap_realm)
- ceph_change_snap_realm(&ci->vfs_inode, NULL);
+ ceph_change_snap_realm(&ci->netfs.inode, NULL);
__cap_delay_cancel(mdsc, ci);
}
lockdep_assert_held(&ci->i_ceph_lock);
- fsc = ceph_inode_to_client(&ci->vfs_inode);
+ fsc = ceph_inode_to_client(&ci->netfs.inode);
WARN_ON_ONCE(ci->i_auth_cap == cap &&
!list_empty(&ci->i_dirty_item) &&
!fsc->blocklisted &&
- !ceph_inode_is_shutdown(&ci->vfs_inode));
+ !ceph_inode_is_shutdown(&ci->netfs.inode));
__ceph_remove_cap(cap, queue_release);
}
int flushing, u64 flush_tid, u64 oldest_flush_tid)
{
struct ceph_inode_info *ci = cap->ci;
- struct inode *inode = &ci->vfs_inode;
+ struct inode *inode = &ci->netfs.inode;
int held, revoking;
lockdep_assert_held(&ci->i_ceph_lock);
static void __send_cap(struct cap_msg_args *arg, struct ceph_inode_info *ci)
{
struct ceph_msg *msg;
- struct inode *inode = &ci->vfs_inode;
+ struct inode *inode = &ci->netfs.inode;
msg = ceph_msg_new(CEPH_MSG_CLIENT_CAPS, CAP_MSG_SIZE, GFP_NOFS, false);
if (!msg) {
__releases(ci->i_ceph_lock)
__acquires(ci->i_ceph_lock)
{
- struct inode *inode = &ci->vfs_inode;
+ struct inode *inode = &ci->netfs.inode;
struct ceph_mds_client *mdsc = session->s_mdsc;
struct ceph_cap_snap *capsnap;
u64 oldest_flush_tid = 0;
while (first_tid <= last_tid) {
struct ceph_cap *cap = ci->i_auth_cap;
- struct ceph_cap_flush *cf;
+ struct ceph_cap_flush *cf = NULL, *iter;
int ret;
if (!(cap && cap->session == session)) {
}
ret = -ENOENT;
- list_for_each_entry(cf, &ci->i_cap_flush_list, i_list) {
- if (cf->tid >= first_tid) {
+ list_for_each_entry(iter, &ci->i_cap_flush_list, i_list) {
+ if (iter->tid >= first_tid) {
+ cf = iter;
ret = 0;
break;
}
void ceph_flush_snaps(struct ceph_inode_info *ci,
struct ceph_mds_session **psession)
{
- struct inode *inode = &ci->vfs_inode;
+ struct inode *inode = &ci->netfs.inode;
struct ceph_mds_client *mdsc = ceph_inode_to_client(inode)->mdsc;
struct ceph_mds_session *session = NULL;
int mds;
struct ceph_cap_flush **pcf)
{
struct ceph_mds_client *mdsc =
- ceph_sb_to_client(ci->vfs_inode.i_sb)->mdsc;
- struct inode *inode = &ci->vfs_inode;
+ ceph_sb_to_client(ci->netfs.inode.i_sb)->mdsc;
+ struct inode *inode = &ci->netfs.inode;
int was = ci->i_dirty_caps;
int dirty = 0;
return 0;
}
- dout("__mark_dirty_caps %p %s dirty %s -> %s\n", &ci->vfs_inode,
+ dout("__mark_dirty_caps %p %s dirty %s -> %s\n", &ci->netfs.inode,
ceph_cap_string(mask), ceph_cap_string(was),
ceph_cap_string(was | mask));
ci->i_dirty_caps |= mask;
ci->i_snap_realm->cached_context);
}
dout(" inode %p now dirty snapc %p auth cap %p\n",
- &ci->vfs_inode, ci->i_head_snapc, ci->i_auth_cap);
+ &ci->netfs.inode, ci->i_head_snapc, ci->i_auth_cap);
BUG_ON(!list_empty(&ci->i_dirty_item));
spin_lock(&mdsc->cap_dirty_lock);
list_add(&ci->i_dirty_item, &session->s_cap_dirty);
bool __ceph_should_report_size(struct ceph_inode_info *ci)
{
- loff_t size = i_size_read(&ci->vfs_inode);
+ loff_t size = i_size_read(&ci->netfs.inode);
/* mds will adjust max size according to the reported size */
if (ci->i_flushing_caps & CEPH_CAP_FILE_WR)
return false;
void ceph_check_caps(struct ceph_inode_info *ci, int flags,
struct ceph_mds_session *session)
{
- struct inode *inode = &ci->vfs_inode;
+ struct inode *inode = &ci->netfs.inode;
struct ceph_mds_client *mdsc = ceph_sb_to_mdsc(inode->i_sb);
struct ceph_cap *cap;
u64 flush_tid, oldest_flush_tid;
struct rb_node *p;
bool queue_invalidate = false;
bool tried_invalidate = false;
+ bool queue_writeback = false;
if (session)
ceph_get_mds_session(session);
}
/* completed revocation? going down and there are no caps? */
- if (revoking && (revoking & cap_used) == 0) {
- dout("completed revocation of %s\n",
- ceph_cap_string(cap->implemented & ~cap->issued));
- goto ack;
+ if (revoking) {
+ if ((revoking & cap_used) == 0) {
+ dout("completed revocation of %s\n",
+ ceph_cap_string(cap->implemented & ~cap->issued));
+ goto ack;
+ }
+
+ /*
+ * If the "i_wrbuffer_ref" was increased by mmap or generic
+ * cache write just before the ceph_check_caps() is called,
+ * the Fb capability revoking will fail this time. Then we
+ * must wait for the BDI's delayed work to flush the dirty
+ * pages and to release the "i_wrbuffer_ref", which will cost
+ * at most 5 seconds. That means the MDS needs to wait at
+ * most 5 seconds to finished the Fb capability's revocation.
+ *
+ * Let's queue a writeback for it.
+ */
+ if (S_ISREG(inode->i_mode) && ci->i_wrbuffer_ref &&
+ (revoking & CEPH_CAP_FILE_BUFFER))
+ queue_writeback = true;
}
/* want more caps from mds? */
spin_unlock(&ci->i_ceph_lock);
ceph_put_mds_session(session);
+ if (queue_writeback)
+ ceph_queue_writeback(inode);
if (queue_invalidate)
ceph_queue_invalidate(inode);
}
}
/*
- * wait for any unsafe requests to complete.
+ * flush the mdlog and wait for any unsafe requests to complete.
*/
-static int unsafe_request_wait(struct inode *inode)
+static int flush_mdlog_and_wait_inode_unsafe_requests(struct inode *inode)
{
struct ceph_mds_client *mdsc = ceph_sb_to_client(inode->i_sb)->mdsc;
struct ceph_inode_info *ci = ceph_inode(inode);
kfree(sessions);
}
- dout("unsafe_request_wait %p wait on tid %llu %llu\n",
+ dout("%s %p wait on tid %llu %llu\n", __func__,
inode, req1 ? req1->r_tid : 0ULL, req2 ? req2->r_tid : 0ULL);
if (req1) {
ret = !wait_for_completion_timeout(&req1->r_safe_completion,
dirty = try_flush_caps(inode, &flush_tid);
dout("fsync dirty caps are %s\n", ceph_cap_string(dirty));
- err = unsafe_request_wait(inode);
+ err = flush_mdlog_and_wait_inode_unsafe_requests(inode);
/*
* only wait on non-file metadata writeback (the mds
__releases(ci->i_ceph_lock)
__acquires(ci->i_ceph_lock)
{
- struct inode *inode = &ci->vfs_inode;
+ struct inode *inode = &ci->netfs.inode;
struct ceph_cap *cap;
struct ceph_cap_flush *cf;
int ret;
cap = ci->i_auth_cap;
if (!(cap && cap->session == session)) {
pr_err("%p auth cap %p not mds%d ???\n",
- &ci->vfs_inode, cap, session->s_mds);
+ &ci->netfs.inode, cap, session->s_mds);
spin_unlock(&ci->i_ceph_lock);
continue;
}
cap = ci->i_auth_cap;
if (!(cap && cap->session == session)) {
pr_err("%p auth cap %p not mds%d ???\n",
- &ci->vfs_inode, cap, session->s_mds);
+ &ci->netfs.inode, cap, session->s_mds);
spin_unlock(&ci->i_ceph_lock);
continue;
}
lockdep_assert_held(&ci->i_ceph_lock);
- dout("%s %p flushing %s\n", __func__, &ci->vfs_inode,
+ dout("%s %p flushing %s\n", __func__, &ci->netfs.inode,
ceph_cap_string(ci->i_flushing_caps));
if (!list_empty(&ci->i_cap_flush_list)) {
}
if (got & CEPH_CAP_FILE_BUFFER) {
if (ci->i_wb_ref == 0)
- ihold(&ci->vfs_inode);
+ ihold(&ci->netfs.inode);
ci->i_wb_ref++;
dout("%s %p wb %d -> %d (?)\n", __func__,
- &ci->vfs_inode, ci->i_wb_ref-1, ci->i_wb_ref);
+ &ci->netfs.inode, ci->i_wb_ref-1, ci->i_wb_ref);
}
}
return ret;
}
- if (S_ISREG(ci->vfs_inode.i_mode) &&
+ if (S_ISREG(ci->netfs.inode.i_mode) &&
ci->i_inline_version != CEPH_INLINE_NONE &&
(_got & (CEPH_CAP_FILE_CACHE|CEPH_CAP_FILE_LAZYIO)) &&
i_size_read(inode) > 0) {
static void __ceph_put_cap_refs(struct ceph_inode_info *ci, int had,
enum put_cap_refs_mode mode)
{
- struct inode *inode = &ci->vfs_inode;
+ struct inode *inode = &ci->netfs.inode;
int last = 0, put = 0, flushsnaps = 0, wake = 0;
bool check_flushsnaps = false;
void ceph_put_wrbuffer_cap_refs(struct ceph_inode_info *ci, int nr,
struct ceph_snap_context *snapc)
{
- struct inode *inode = &ci->vfs_inode;
- struct ceph_cap_snap *capsnap = NULL;
+ struct inode *inode = &ci->netfs.inode;
+ struct ceph_cap_snap *capsnap = NULL, *iter;
int put = 0;
bool last = false;
- bool found = false;
bool flush_snaps = false;
bool complete_capsnap = false;
ci->i_wrbuffer_ref, ci->i_wrbuffer_ref_head,
last ? " LAST" : "");
} else {
- list_for_each_entry(capsnap, &ci->i_cap_snaps, ci_item) {
- if (capsnap->context == snapc) {
- found = true;
+ list_for_each_entry(iter, &ci->i_cap_snaps, ci_item) {
+ if (iter->context == snapc) {
+ capsnap = iter;
break;
}
}
- if (!found) {
+ if (!capsnap) {
/*
* The capsnap should already be removed when removing
* auth cap in the case of a forced unmount.
session->s_mds,
&list_first_entry(&session->s_cap_flushing,
struct ceph_inode_info,
- i_flushing_item)->vfs_inode);
+ i_flushing_item)->netfs.inode);
}
}
mdsc->num_cap_flushing--;
struct ceph_inode_info *ci = ceph_inode(inode);
struct ceph_mds_client *mdsc = ceph_sb_to_client(inode->i_sb)->mdsc;
u64 follows = le64_to_cpu(m->snap_follows);
- struct ceph_cap_snap *capsnap;
- bool flushed = false;
+ struct ceph_cap_snap *capsnap = NULL, *iter;
bool wake_ci = false;
bool wake_mdsc = false;
inode, ci, session->s_mds, follows);
spin_lock(&ci->i_ceph_lock);
- list_for_each_entry(capsnap, &ci->i_cap_snaps, ci_item) {
- if (capsnap->follows == follows) {
- if (capsnap->cap_flush.tid != flush_tid) {
+ list_for_each_entry(iter, &ci->i_cap_snaps, ci_item) {
+ if (iter->follows == follows) {
+ if (iter->cap_flush.tid != flush_tid) {
dout(" cap_snap %p follows %lld tid %lld !="
- " %lld\n", capsnap, follows,
- flush_tid, capsnap->cap_flush.tid);
+ " %lld\n", iter, follows,
+ flush_tid, iter->cap_flush.tid);
break;
}
- flushed = true;
+ capsnap = iter;
break;
} else {
dout(" skipping cap_snap %p follows %lld\n",
- capsnap, capsnap->follows);
+ iter, iter->follows);
}
}
- if (flushed)
+ if (capsnap)
ceph_remove_capsnap(inode, capsnap, &wake_ci, &wake_mdsc);
spin_unlock(&ci->i_ceph_lock);
- if (flushed) {
+ if (capsnap) {
ceph_put_snap_context(capsnap->context);
ceph_put_cap_snap(capsnap);
if (wake_ci)
break;
list_del_init(&ci->i_cap_delay_list);
- inode = igrab(&ci->vfs_inode);
+ inode = igrab(&ci->netfs.inode);
if (inode) {
spin_unlock(&mdsc->cap_delay_lock);
dout("check_delayed_caps on %p\n", inode);
while (!list_empty(&s->s_cap_dirty)) {
ci = list_first_entry(&s->s_cap_dirty, struct ceph_inode_info,
i_dirty_item);
- inode = &ci->vfs_inode;
+ inode = &ci->netfs.inode;
ihold(inode);
dout("flush_dirty_caps %llx.%llx\n", ceph_vinop(inode));
spin_unlock(&mdsc->cap_dirty_lock);
void ceph_get_fmode(struct ceph_inode_info *ci, int fmode, int count)
{
- struct ceph_mds_client *mdsc = ceph_sb_to_mdsc(ci->vfs_inode.i_sb);
+ struct ceph_mds_client *mdsc = ceph_sb_to_mdsc(ci->netfs.inode.i_sb);
int bits = (fmode << 1) | 1;
bool already_opened = false;
int i;
*/
void ceph_put_fmode(struct ceph_inode_info *ci, int fmode, int count)
{
- struct ceph_mds_client *mdsc = ceph_sb_to_mdsc(ci->vfs_inode.i_sb);
+ struct ceph_mds_client *mdsc = ceph_sb_to_mdsc(ci->netfs.inode.i_sb);
int bits = (fmode << 1) | 1;
bool is_closed = true;
int i;
lockdep_assert_held(&ci->i_ceph_lock);
dout("removing cap %p, ci is %p, inode is %p\n",
- cap, ci, &ci->vfs_inode);
+ cap, ci, &ci->netfs.inode);
is_auth = (cap == ci->i_auth_cap);
__ceph_remove_cap(cap, false);