diff options
author | Linus Torvalds <torvalds@linux-foundation.org> | 2022-12-12 20:41:50 -0800 |
---|---|---|
committer | Linus Torvalds <torvalds@linux-foundation.org> | 2022-12-12 20:41:50 -0800 |
commit | 97971df811b8854882c0f6c6631e23ab8cdcc44f (patch) | |
tree | 2eaf96eda6b88342d88047521255c110600e1410 /fs | |
parent | 56c003e4db4adb1d57e599fb4c7d39f9130f9db1 (diff) | |
parent | 7a5e9f1f83e3271a9f05933a80b870fe55ebbb3d (diff) |
Merge tag 'dlm-6.2' of git://git.kernel.org/pub/scm/linux/kernel/git/teigland/linux-dlm
Pull dlm updates from David Teigland:
"These patches include the usual cleanups and minor fixes, the removal
of code that is no longer needed due to recent improvements, and
improvements to processing large volumes of messages during heavy
locking activity.
Summary:
- Misc code cleanup
- Fix a couple of socket handling bugs: a double release on an error
path and a data-ready race in an accept loop
- Remove code for resending dir-remove messages. This code is no
longer needed since the midcomms layer now ensures the messages are
resent if needed
- Add tracepoints for dlm messages
- Improve callback queueing by replacing the fixed array with a list
- Simplify the handling of a remove message followed by a lookup
message by sending both without releasing a spinlock in between
- Improve the concurrency of sending and receiving messages by
holding locks for a shorter time, and changing how workqueues are
used
- Remove old code for shutting down sockets, which is no longer
needed with the reliable connection handling that was recently
added"
* tag 'dlm-6.2' of git://git.kernel.org/pub/scm/linux/kernel/git/teigland/linux-dlm: (37 commits)
fs: dlm: fix building without lockdep
fs: dlm: parallelize lowcomms socket handling
fs: dlm: don't init error value
fs: dlm: use saved sk_error_report()
fs: dlm: use sock2con without checking null
fs: dlm: remove dlm_node_addrs lookup list
fs: dlm: don't put dlm_local_addrs on heap
fs: dlm: cleanup listen sock handling
fs: dlm: remove socket shutdown handling
fs: dlm: use listen sock as dlm running indicator
fs: dlm: use list_first_entry_or_null
fs: dlm: remove twice INIT_WORK
fs: dlm: add midcomms init/start functions
fs: dlm: add dst nodeid for msg tracing
fs: dlm: rename seq to h_seq for msg tracing
fs: dlm: rename DLM_IFL_NEED_SCHED to DLM_IFL_CB_PENDING
fs: dlm: ast do WARN_ON_ONCE() on hotpath
fs: dlm: drop lkb ref in bug case
fs: dlm: avoid false-positive checker warning
fs: dlm: use WARN_ON_ONCE() instead of WARN_ON()
...
Diffstat (limited to 'fs')
-rw-r--r-- | fs/dlm/ast.c | 322 | ||||
-rw-r--r-- | fs/dlm/ast.h | 17 | ||||
-rw-r--r-- | fs/dlm/config.c | 4 | ||||
-rw-r--r-- | fs/dlm/debug_fs.c | 2 | ||||
-rw-r--r-- | fs/dlm/dlm_internal.h | 25 | ||||
-rw-r--r-- | fs/dlm/lock.c | 190 | ||||
-rw-r--r-- | fs/dlm/lockspace.c | 14 | ||||
-rw-r--r-- | fs/dlm/lowcomms.c | 1538 | ||||
-rw-r--r-- | fs/dlm/lowcomms.h | 6 | ||||
-rw-r--r-- | fs/dlm/main.c | 7 | ||||
-rw-r--r-- | fs/dlm/member.c | 5 | ||||
-rw-r--r-- | fs/dlm/memory.c | 30 | ||||
-rw-r--r-- | fs/dlm/memory.h | 4 | ||||
-rw-r--r-- | fs/dlm/midcomms.c | 141 | ||||
-rw-r--r-- | fs/dlm/midcomms.h | 7 | ||||
-rw-r--r-- | fs/dlm/rcom.c | 4 | ||||
-rw-r--r-- | fs/dlm/requestqueue.c | 3 | ||||
-rw-r--r-- | fs/dlm/user.c | 74 | ||||
-rw-r--r-- | fs/dlm/user.h | 2 |
19 files changed, 1152 insertions, 1243 deletions
diff --git a/fs/dlm/ast.c b/fs/dlm/ast.c index d60a8d8f109d..26fef9945cc9 100644 --- a/fs/dlm/ast.c +++ b/fs/dlm/ast.c @@ -12,55 +12,67 @@ #include <trace/events/dlm.h> #include "dlm_internal.h" +#include "memory.h" #include "lock.h" #include "user.h" #include "ast.h" -static uint64_t dlm_cb_seq; -static DEFINE_SPINLOCK(dlm_cb_seq_spin); +void dlm_release_callback(struct kref *ref) +{ + struct dlm_callback *cb = container_of(ref, struct dlm_callback, ref); + + dlm_free_cb(cb); +} + +void dlm_callback_set_last_ptr(struct dlm_callback **from, + struct dlm_callback *to) +{ + if (*from) + kref_put(&(*from)->ref, dlm_release_callback); + + if (to) + kref_get(&to->ref); + + *from = to; +} -static void dlm_dump_lkb_callbacks(struct dlm_lkb *lkb) +void dlm_purge_lkb_callbacks(struct dlm_lkb *lkb) { - int i; - - log_print("last_bast %x %llu flags %x mode %d sb %d %x", - lkb->lkb_id, - (unsigned long long)lkb->lkb_last_bast.seq, - lkb->lkb_last_bast.flags, - lkb->lkb_last_bast.mode, - lkb->lkb_last_bast.sb_status, - lkb->lkb_last_bast.sb_flags); - - log_print("last_cast %x %llu flags %x mode %d sb %d %x", - lkb->lkb_id, - (unsigned long long)lkb->lkb_last_cast.seq, - lkb->lkb_last_cast.flags, - lkb->lkb_last_cast.mode, - lkb->lkb_last_cast.sb_status, - lkb->lkb_last_cast.sb_flags); - - for (i = 0; i < DLM_CALLBACKS_SIZE; i++) { - log_print("cb %x %llu flags %x mode %d sb %d %x", - lkb->lkb_id, - (unsigned long long)lkb->lkb_callbacks[i].seq, - lkb->lkb_callbacks[i].flags, - lkb->lkb_callbacks[i].mode, - lkb->lkb_callbacks[i].sb_status, - lkb->lkb_callbacks[i].sb_flags); + struct dlm_callback *cb, *safe; + + list_for_each_entry_safe(cb, safe, &lkb->lkb_callbacks, list) { + list_del(&cb->list); + kref_put(&cb->ref, dlm_release_callback); } + + lkb->lkb_flags &= ~DLM_IFL_CB_PENDING; + + /* invalidate */ + dlm_callback_set_last_ptr(&lkb->lkb_last_cast, NULL); + dlm_callback_set_last_ptr(&lkb->lkb_last_cb, NULL); + lkb->lkb_last_bast_mode = -1; } -int dlm_add_lkb_callback(struct dlm_lkb *lkb, uint32_t flags, int mode, - int status, uint32_t sbflags, uint64_t seq) +int dlm_enqueue_lkb_callback(struct dlm_lkb *lkb, uint32_t flags, int mode, + int status, uint32_t sbflags) { struct dlm_ls *ls = lkb->lkb_resource->res_ls; - uint64_t prev_seq; + int rv = DLM_ENQUEUE_CALLBACK_SUCCESS; + struct dlm_callback *cb; int prev_mode; - int i, rv; - for (i = 0; i < DLM_CALLBACKS_SIZE; i++) { - if (lkb->lkb_callbacks[i].seq) - continue; + if (flags & DLM_CB_BAST) { + /* if cb is a bast, it should be skipped if the blocking mode is + * compatible with the last granted mode + */ + if (lkb->lkb_last_cast) { + if (dlm_modes_compat(mode, lkb->lkb_last_cast->mode)) { + log_debug(ls, "skip %x bast mode %d for cast mode %d", + lkb->lkb_id, mode, + lkb->lkb_last_cast->mode); + goto out; + } + } /* * Suppress some redundant basts here, do more on removal. @@ -68,148 +80,95 @@ int dlm_add_lkb_callback(struct dlm_lkb *lkb, uint32_t flags, int mode, * is a bast for the same mode or a more restrictive mode. * (the addional > PR check is needed for PR/CW inversion) */ - - if ((i > 0) && (flags & DLM_CB_BAST) && - (lkb->lkb_callbacks[i-1].flags & DLM_CB_BAST)) { - - prev_seq = lkb->lkb_callbacks[i-1].seq; - prev_mode = lkb->lkb_callbacks[i-1].mode; + if (lkb->lkb_last_cb && lkb->lkb_last_cb->flags & DLM_CB_BAST) { + prev_mode = lkb->lkb_last_cb->mode; if ((prev_mode == mode) || (prev_mode > mode && prev_mode > DLM_LOCK_PR)) { - - log_debug(ls, "skip %x add bast %llu mode %d " - "for bast %llu mode %d", - lkb->lkb_id, - (unsigned long long)seq, - mode, - (unsigned long long)prev_seq, - prev_mode); - rv = 0; + log_debug(ls, "skip %x add bast mode %d for bast mode %d", + lkb->lkb_id, mode, prev_mode); goto out; } } - - lkb->lkb_callbacks[i].seq = seq; - lkb->lkb_callbacks[i].flags = flags; - lkb->lkb_callbacks[i].mode = mode; - lkb->lkb_callbacks[i].sb_status = status; - lkb->lkb_callbacks[i].sb_flags = (sbflags & 0x000000FF); - rv = 0; - break; } - if (i == DLM_CALLBACKS_SIZE) { - log_error(ls, "no callbacks %x %llu flags %x mode %d sb %d %x", - lkb->lkb_id, (unsigned long long)seq, - flags, mode, status, sbflags); - dlm_dump_lkb_callbacks(lkb); - rv = -1; + cb = dlm_allocate_cb(); + if (!cb) { + rv = DLM_ENQUEUE_CALLBACK_FAILURE; goto out; } - out: - return rv; -} - -int dlm_rem_lkb_callback(struct dlm_ls *ls, struct dlm_lkb *lkb, - struct dlm_callback *cb, int *resid) -{ - int i, rv; - - *resid = 0; - - if (!lkb->lkb_callbacks[0].seq) { - rv = -ENOENT; - goto out; - } - - /* oldest undelivered cb is callbacks[0] */ - - memcpy(cb, &lkb->lkb_callbacks[0], sizeof(struct dlm_callback)); - memset(&lkb->lkb_callbacks[0], 0, sizeof(struct dlm_callback)); - /* shift others down */ - - for (i = 1; i < DLM_CALLBACKS_SIZE; i++) { - if (!lkb->lkb_callbacks[i].seq) - break; - memcpy(&lkb->lkb_callbacks[i-1], &lkb->lkb_callbacks[i], - sizeof(struct dlm_callback)); - memset(&lkb->lkb_callbacks[i], 0, sizeof(struct dlm_callback)); - (*resid)++; + cb->flags = flags; + cb->mode = mode; + cb->sb_status = status; + cb->sb_flags = (sbflags & 0x000000FF); + kref_init(&cb->ref); + if (!(lkb->lkb_flags & DLM_IFL_CB_PENDING)) { + lkb->lkb_flags |= DLM_IFL_CB_PENDING; + rv = DLM_ENQUEUE_CALLBACK_NEED_SCHED; } + list_add_tail(&cb->list, &lkb->lkb_callbacks); - /* if cb is a bast, it should be skipped if the blocking mode is - compatible with the last granted mode */ - - if ((cb->flags & DLM_CB_BAST) && lkb->lkb_last_cast.seq) { - if (dlm_modes_compat(cb->mode, lkb->lkb_last_cast.mode)) { - cb->flags |= DLM_CB_SKIP; - - log_debug(ls, "skip %x bast %llu mode %d " - "for cast %llu mode %d", - lkb->lkb_id, - (unsigned long long)cb->seq, - cb->mode, - (unsigned long long)lkb->lkb_last_cast.seq, - lkb->lkb_last_cast.mode); - rv = 0; - goto out; - } - } + if (flags & DLM_CB_CAST) + dlm_callback_set_last_ptr(&lkb->lkb_last_cast, cb); - if (cb->flags & DLM_CB_CAST) { - memcpy(&lkb->lkb_last_cast, cb, sizeof(struct dlm_callback)); - lkb->lkb_last_cast_time = ktime_get(); - } + dlm_callback_set_last_ptr(&lkb->lkb_last_cb, cb); - if (cb->flags & DLM_CB_BAST) { - memcpy(&lkb->lkb_last_bast, cb, sizeof(struct dlm_callback)); - lkb->lkb_last_bast_time = ktime_get(); - } - rv = 0; out: return rv; } +int dlm_dequeue_lkb_callback(struct dlm_lkb *lkb, struct dlm_callback **cb) +{ + /* oldest undelivered cb is callbacks first entry */ + *cb = list_first_entry_or_null(&lkb->lkb_callbacks, + struct dlm_callback, list); + if (!*cb) + return DLM_DEQUEUE_CALLBACK_EMPTY; + + /* remove it from callbacks so shift others down */ + list_del(&(*cb)->list); + if (list_empty(&lkb->lkb_callbacks)) + return DLM_DEQUEUE_CALLBACK_LAST; + + return DLM_DEQUEUE_CALLBACK_SUCCESS; +} + void dlm_add_cb(struct dlm_lkb *lkb, uint32_t flags, int mode, int status, uint32_t sbflags) { struct dlm_ls *ls = lkb->lkb_resource->res_ls; - uint64_t new_seq, prev_seq; int rv; - spin_lock(&dlm_cb_seq_spin); - new_seq = ++dlm_cb_seq; - if (!dlm_cb_seq) - new_seq = ++dlm_cb_seq; - spin_unlock(&dlm_cb_seq_spin); - if (lkb->lkb_flags & DLM_IFL_USER) { - dlm_user_add_ast(lkb, flags, mode, status, sbflags, new_seq); + dlm_user_add_ast(lkb, flags, mode, status, sbflags); return; } - mutex_lock(&lkb->lkb_cb_mutex); - prev_seq = lkb->lkb_callbacks[0].seq; - - rv = dlm_add_lkb_callback(lkb, flags, mode, status, sbflags, new_seq); - if (rv < 0) - goto out; - - if (!prev_seq) { + spin_lock(&lkb->lkb_cb_lock); + rv = dlm_enqueue_lkb_callback(lkb, flags, mode, status, sbflags); + switch (rv) { + case DLM_ENQUEUE_CALLBACK_NEED_SCHED: kref_get(&lkb->lkb_ref); - mutex_lock(&ls->ls_cb_mutex); + spin_lock(&ls->ls_cb_lock); if (test_bit(LSFL_CB_DELAY, &ls->ls_flags)) { list_add(&lkb->lkb_cb_list, &ls->ls_cb_delay); } else { queue_work(ls->ls_callback_wq, &lkb->lkb_cb_work); } - mutex_unlock(&ls->ls_cb_mutex); + spin_unlock(&ls->ls_cb_lock); + break; + case DLM_ENQUEUE_CALLBACK_FAILURE: + WARN_ON_ONCE(1); + break; + case DLM_ENQUEUE_CALLBACK_SUCCESS: + break; + default: + WARN_ON_ONCE(1); + break; } - out: - mutex_unlock(&lkb->lkb_cb_mutex); + spin_unlock(&lkb->lkb_cb_lock); } void dlm_callback_work(struct work_struct *work) @@ -218,53 +177,46 @@ void dlm_callback_work(struct work_struct *work) struct dlm_ls *ls = lkb->lkb_resource->res_ls; void (*castfn) (void *astparam); void (*bastfn) (void *astparam, int mode); - struct dlm_callback callbacks[DLM_CALLBACKS_SIZE]; - int i, rv, resid; - - memset(&callbacks, 0, sizeof(callbacks)); + struct dlm_callback *cb; + int rv; - mutex_lock(&lkb->lkb_cb_mutex); - if (!lkb->lkb_callbacks[0].seq) { - /* no callback work exists, shouldn't happen */ - log_error(ls, "dlm_callback_work %x no work", lkb->lkb_id); - dlm_print_lkb(lkb); - dlm_dump_lkb_callbacks(lkb); - } + spin_lock(&lkb->lkb_cb_lock); + rv = dlm_dequeue_lkb_callback(lkb, &cb); + spin_unlock(&lkb->lkb_cb_lock); - for (i = 0; i < DLM_CALLBACKS_SIZE; i++) { - rv = dlm_rem_lkb_callback(ls, lkb, &callbacks[i], &resid); - if (rv < 0) - break; - } + if (WARN_ON_ONCE(rv == DLM_DEQUEUE_CALLBACK_EMPTY)) + goto out; - if (resid) { - /* cbs remain, loop should have removed all, shouldn't happen */ - log_error(ls, "dlm_callback_work %x resid %d", lkb->lkb_id, - resid); - dlm_print_lkb(lkb); - dlm_dump_lkb_callbacks(lkb); - } - mutex_unlock(&lkb->lkb_cb_mutex); + for (;;) { + castfn = lkb->lkb_astfn; + bastfn = lkb->lkb_bastfn; + + if (cb->flags & DLM_CB_BAST) { + trace_dlm_bast(ls, lkb, cb->mode); + lkb->lkb_last_bast_time = ktime_get(); + lkb->lkb_last_bast_mode = cb->mode; + bastfn(lkb->lkb_astparam, cb->mode); + } else if (cb->flags & DLM_CB_CAST) { + lkb->lkb_lksb->sb_status = cb->sb_status; + lkb->lkb_lksb->sb_flags = cb->sb_flags; + trace_dlm_ast(ls, lkb); + lkb->lkb_last_cast_time = ktime_get(); + castfn(lkb->lkb_astparam); + } - castfn = lkb->lkb_astfn; - bastfn = lkb->lkb_bastfn; + kref_put(&cb->ref, dlm_release_callback); - for (i = 0; i < DLM_CALLBACKS_SIZE; i++) { - if (!callbacks[i].seq) + spin_lock(&lkb->lkb_cb_lock); + rv = dlm_dequeue_lkb_callback(lkb, &cb); + if (rv == DLM_DEQUEUE_CALLBACK_EMPTY) { + lkb->lkb_flags &= ~DLM_IFL_CB_PENDING; + spin_unlock(&lkb->lkb_cb_lock); break; - if (callbacks[i].flags & DLM_CB_SKIP) { - continue; - } else if (callbacks[i].flags & DLM_CB_BAST) { - trace_dlm_bast(ls, lkb, callbacks[i].mode); - bastfn(lkb->lkb_astparam, callbacks[i].mode); - } else if (callbacks[i].flags & DLM_CB_CAST) { - lkb->lkb_lksb->sb_status = callbacks[i].sb_status; - lkb->lkb_lksb->sb_flags = callbacks[i].sb_flags; - trace_dlm_ast(ls, lkb); - castfn(lkb->lkb_astparam); } + spin_unlock(&lkb->lkb_cb_lock); } +out: /* undo kref_get from dlm_add_callback, may cause lkb to be freed */ dlm_put_lkb(lkb); } @@ -289,9 +241,9 @@ void dlm_callback_stop(struct dlm_ls *ls) void dlm_callback_suspend(struct dlm_ls *ls) { if (ls->ls_callback_wq) { - mutex_lock(&ls->ls_cb_mutex); + spin_lock(&ls->ls_cb_lock); set_bit(LSFL_CB_DELAY, &ls->ls_flags); - mutex_unlock(&ls->ls_cb_mutex); + spin_unlock(&ls->ls_cb_lock); flush_workqueue(ls->ls_callback_wq); } @@ -308,10 +260,8 @@ void dlm_callback_resume(struct dlm_ls *ls) if (!ls->ls_callback_wq) return; - clear_bit(LSFL_CB_DELAY, &ls->ls_flags); - more: - mutex_lock(&ls->ls_cb_mutex); + spin_lock(&ls->ls_cb_lock); list_for_each_entry_safe(lkb, safe, &ls->ls_cb_delay, lkb_cb_list) { list_del_init(&lkb->lkb_cb_list); queue_work(ls->ls_callback_wq, &lkb->lkb_cb_work); @@ -320,7 +270,9 @@ more: break; } empty = list_empty(&ls->ls_cb_delay); - mutex_unlock(&ls->ls_cb_mutex); + if (empty) + clear_bit(LSFL_CB_DELAY, &ls->ls_flags); + spin_unlock(&ls->ls_cb_lock); sum += count; if (!empty) { diff --git a/fs/dlm/ast.h b/fs/dlm/ast.h index e5e05fcc5813..880b11882495 100644 --- a/fs/dlm/ast.h +++ b/fs/dlm/ast.h @@ -11,13 +11,22 @@ #ifndef __ASTD_DOT_H__ #define __ASTD_DOT_H__ -int dlm_add_lkb_callback(struct dlm_lkb *lkb, uint32_t flags, int mode, - int status, uint32_t sbflags, uint64_t seq); -int dlm_rem_lkb_callback(struct dlm_ls *ls, struct dlm_lkb *lkb, - struct dlm_callback *cb, int *resid); +#define DLM_ENQUEUE_CALLBACK_NEED_SCHED 1 +#define DLM_ENQUEUE_CALLBACK_SUCCESS 0 +#define DLM_ENQUEUE_CALLBACK_FAILURE -1 +int dlm_enqueue_lkb_callback(struct dlm_lkb *lkb, uint32_t flags, int mode, + int status, uint32_t sbflags); +#define DLM_DEQUEUE_CALLBACK_EMPTY 2 +#define DLM_DEQUEUE_CALLBACK_LAST 1 +#define DLM_DEQUEUE_CALLBACK_SUCCESS 0 +int dlm_dequeue_lkb_callback(struct dlm_lkb *lkb, struct dlm_callback **cb); void dlm_add_cb(struct dlm_lkb *lkb, uint32_t flags, int mode, int status, uint32_t sbflags); +void dlm_callback_set_last_ptr(struct dlm_callback **from, + struct dlm_callback *to); +void dlm_release_callback(struct kref *ref); +void dlm_purge_lkb_callbacks(struct dlm_lkb *lkb); void dlm_callback_work(struct work_struct *work); int dlm_callback_start(struct dlm_ls *ls); void dlm_callback_stop(struct dlm_ls *ls); diff --git a/fs/dlm/config.c b/fs/dlm/config.c index ac8b62106ce0..20b60709eccf 100644 --- a/fs/dlm/config.c +++ b/fs/dlm/config.c @@ -183,7 +183,7 @@ static int dlm_check_protocol_and_dlm_running(unsigned int x) return -EINVAL; } - if (dlm_allow_conn) + if (dlm_lowcomms_is_running()) return -EBUSY; return 0; @@ -194,7 +194,7 @@ static int dlm_check_zero_and_dlm_running(unsigned int x) if (!x) return -EINVAL; - if (dlm_allow_conn) + if (dlm_lowcomms_is_running()) return -EBUSY; return 0; diff --git a/fs/dlm/debug_fs.c b/fs/dlm/debug_fs.c index 8fb04ebbafb5..8a0e1b1f74ad 100644 --- a/fs/dlm/debug_fs.c +++ b/fs/dlm/debug_fs.c @@ -246,7 +246,7 @@ static void print_format3_lock(struct seq_file *s, struct dlm_lkb *lkb, lkb->lkb_status, lkb->lkb_grmode, lkb->lkb_rqmode, - lkb->lkb_last_bast.mode, + lkb->lkb_last_bast_mode, rsb_lookup, lkb->lkb_wait_type, lkb->lkb_lvbseq, diff --git a/fs/dlm/dlm_internal.h b/fs/dlm/dlm_internal.h index e34c3d2639a5..ab1a55337a6e 100644 --- a/fs/dlm/dlm_internal.h +++ b/fs/dlm/dlm_internal.h @@ -211,6 +211,7 @@ struct dlm_args { #endif #define DLM_IFL_DEADLOCK_CANCEL 0x01000000 #define DLM_IFL_STUB_MS 0x02000000 /* magic number for m_flags */ +#define DLM_IFL_CB_PENDING 0x04000000 /* least significant 2 bytes are message changed, they are full transmitted * but at receive side only the 2 bytes LSB will be set. * @@ -222,18 +223,17 @@ struct dlm_args { #define DLM_IFL_USER 0x00000001 #define DLM_IFL_ORPHAN 0x00000002 -#define DLM_CALLBACKS_SIZE 6 - #define DLM_CB_CAST 0x00000001 #define DLM_CB_BAST 0x00000002 -#define DLM_CB_SKIP 0x00000004 struct dlm_callback { - uint64_t seq; uint32_t flags; /* DLM_CBF_ */ int sb_status; /* copy to lksb status */ uint8_t sb_flags; /* copy to lksb flags */ int8_t mode; /* rq mode of bast, gr mode of cast */ + + struct list_head list; + struct kref ref; }; struct dlm_lkb { @@ -268,12 +268,13 @@ struct dlm_lkb { unsigned long lkb_timeout_cs; #endif - struct mutex lkb_cb_mutex; + spinlock_t lkb_cb_lock; struct work_struct lkb_cb_work; struct list_head lkb_cb_list; /* for ls_cb_delay or proc->asts */ - struct dlm_callback lkb_callbacks[DLM_CALLBACKS_SIZE]; - struct dlm_callback lkb_last_cast; - struct dlm_callback lkb_last_bast; + struct list_head lkb_callbacks; + struct dlm_callback *lkb_last_cast; + struct dlm_callback *lkb_last_cb; + int lkb_last_bast_mode; ktime_t lkb_last_cast_time; /* for debugging */ ktime_t lkb_last_bast_time; /* for debugging */ @@ -591,11 +592,7 @@ struct dlm_ls { int ls_new_rsb_count; struct list_head ls_new_rsb; /* new rsb structs */ - spinlock_t ls_remove_spin; - wait_queue_head_t ls_remove_wait; - char ls_remove_name[DLM_RESNAME_MAXLEN+1]; char *ls_remove_names[DLM_REMOVE_NAMES_MAX]; - int ls_remove_len; int ls_remove_lens[DLM_REMOVE_NAMES_MAX]; struct list_head ls_nodes; /* current nodes in ls */ @@ -631,7 +628,7 @@ struct dlm_ls { /* recovery related */ - struct mutex ls_cb_mutex; + spinlock_t ls_cb_lock; struct list_head ls_cb_delay; /* save for queue_work later */ struct timer_list ls_timer; struct task_struct *ls_recoverd_task; @@ -670,7 +667,7 @@ struct dlm_ls { void *ls_ops_arg; int ls_namelen; - char ls_name[1]; + char ls_name[DLM_LOCKSPACE_LEN + 1]; }; /* diff --git a/fs/dlm/lock.c b/fs/dlm/lock.c index 94a72ede5764..e1adfa5aed05 100644 --- a/fs/dlm/lock.c +++ b/fs/dlm/lock.c @@ -1209,6 +1209,7 @@ static int _create_lkb(struct dlm_ls *ls, struct dlm_lkb **lkb_ret, if (!lkb) return -ENOMEM; + lkb->lkb_last_bast_mode = -1; lkb->lkb_nodeid = -1; lkb->lkb_grmode = DLM_LOCK_IV; kref_init(&lkb->lkb_ref); @@ -1218,7 +1219,8 @@ static int _create_lkb(struct dlm_ls *ls, struct dlm_lkb **lkb_ret, INIT_LIST_HEAD(&lkb->lkb_time_list); #endif INIT_LIST_HEAD(&lkb->lkb_cb_list); - mutex_init(&lkb->lkb_cb_mutex); + INIT_LIST_HEAD(&lkb->lkb_callbacks); + spin_lock_init(&lkb->lkb_cb_lock); INIT_WORK(&lkb->lkb_cb_work, dlm_callback_work); idr_preload(GFP_NOFS); @@ -1587,37 +1589,6 @@ static int remove_from_waiters_ms(struct dlm_lkb *lkb, struct dlm_message *ms) return error; } -/* If there's an rsb for the same resource being removed, ensure - * that the remove message is sent before the new lookup message. - */ - -#define DLM_WAIT_PENDING_COND(ls, r) \ - (ls->ls_remove_len && \ - !rsb_cmp(r, ls->ls_remove_name, \ - ls->ls_remove_len)) - -static void wait_pending_remove(struct dlm_rsb *r) -{ - struct dlm_ls *ls = r->res_ls; - restart: - spin_lock(&ls->ls_remove_spin); - if (DLM_WAIT_PENDING_COND(ls, r)) { - log_debug(ls, "delay lookup for remove dir %d %s", - r->res_dir_nodeid, r->res_name); - spin_unlock(&ls->ls_remove_spin); - wait_event(ls->ls_remove_wait, !DLM_WAIT_PENDING_COND(ls, r)); - goto restart; - } - spin_unlock(&ls->ls_remove_spin); -} - -/* - * ls_remove_spin protects ls_remove_name and ls_remove_len which are - * read by other threads in wait_pending_remove. ls_remove_names - * and ls_remove_lens are only used by the scan thread, so they do - * not need protection. - */ - static void shrink_bucket(struct dlm_ls *ls, int b) { struct rb_node *n, *next; @@ -1699,11 +1670,6 @@ static void shrink_bucket(struct dlm_ls *ls, int b) * list and sending the removal. Keeping this gap small is * important to keep us (the master node) from being out of sync * with the remote dir node for very long. - * - * From the time the rsb is removed from toss until just after - * send_remove, the rsb name is saved in ls_remove_name. A new - * lookup checks this to ensure that a new lookup message for the - * same resource name is not sent just before the remove message. */ for (i = 0; i < remote_count; i++) { @@ -1750,22 +1716,8 @@ static void shrink_bucket(struct dlm_ls *ls, int b) } rb_erase(&r->res_hashnode, &ls->ls_rsbtbl[b].toss); - - /* block lookup of same name until we've sent remove */ - spin_lock(&ls->ls_remove_spin); - ls->ls_remove_len = len; - memcpy(ls->ls_remove_name, name, DLM_RESNAME_MAXLEN); - spin_unlock(&ls->ls_remove_spin); - spin_unlock(&ls->ls_rsbtbl[b].lock); - send_remove(r); - - /* allow lookup of name again */ - spin_lock(&ls->ls_remove_spin); - ls->ls_remove_len = 0; - memset(ls->ls_remove_name, 0, DLM_RESNAME_MAXLEN); - spin_unlock(&ls->ls_remove_spin); - wake_up(&ls->ls_remove_wait); + spin_unlock(&ls->ls_rsbtbl[b].lock); dlm_free_rsb(r); } @@ -2716,8 +2668,6 @@ static int set_master(struct dlm_rsb *r, struct dlm_lkb *lkb) return 0; } - wait_pending_remove(r); - r->res_first_lkid = lkb->lkb_id; send_lookup(r, lkb); return 1; @@ -3552,7 +3502,8 @@ int dlm_unlock(dlm_lockspace_t *lockspace, static int _create_message(struct dlm_ls *ls, int mb_len, int to_nodeid, int mstype, struct dlm_message **ms_ret, - struct dlm_mhandle **mh_ret) + struct dlm_mhandle **mh_ret, + gfp_t allocation) { struct dlm_message *ms; struct dlm_mhandle *mh; @@ -3562,7 +3513,7 @@ static int _create_message(struct dlm_ls *ls, int mb_len, pass into midcomms_commit and a message buffer (mb) that we write our data into */ - mh = dlm_midcomms_get_mhandle(to_nodeid, mb_len, GFP_NOFS, &mb); + mh = dlm_midcomms_get_mhandle(to_nodeid, mb_len, allocation, &mb); if (!mh) return -ENOBUFS; @@ -3584,7 +3535,8 @@ static int _create_message(struct dlm_ls *ls, int mb_len, static int create_message(struct dlm_rsb *r, struct dlm_lkb *lkb, int to_nodeid, int mstype, struct dlm_message **ms_ret, - struct dlm_mhandle **mh_ret) + struct dlm_mhandle **mh_ret, + gfp_t allocation) { int mb_len = sizeof(struct dlm_message); @@ -3605,15 +3557,16 @@ static int create_message(struct dlm_rsb *r, struct dlm_lkb *lkb, } return _create_message(r->res_ls, mb_len, to_nodeid, mstype, - ms_ret, mh_ret); + ms_ret, mh_ret, allocation); } /* further lowcomms enhancements or alternate implementations may make the return value from this function useful at some point */ -static int send_message(struct dlm_mhandle *mh, struct dlm_message *ms) +static int send_message(struct dlm_mhandle *mh, struct dlm_message *ms, + const void *name, int namelen) { - dlm_midcomms_commit_mhandle(mh); + dlm_midcomms_commit_mhandle(mh, name, namelen); return 0; } @@ -3673,13 +3626,13 @@ static int send_common(struct dlm_rsb *r, struct dlm_lkb *lkb, int mstype) if (error) return error; - error = create_message(r, lkb, to_nodeid, mstype, &ms, &mh); + error = create_message(r, lkb, to_nodeid, mstype, &ms, &mh, GFP_NOFS); if (error) goto fail; send_args(r, lkb, ms); - error = send_message(mh, ms); + error = send_message(mh, ms, r->res_name, r->res_length); if (error) goto fail; return 0; @@ -3734,7 +3687,8 @@ static int send_grant(struct dlm_rsb *r, struct dlm_lkb *lkb) to_nodeid = lkb->lkb_nodeid; - error = create_message(r, lkb, to_nodeid, DLM_MSG_GRANT, &ms, &mh); + error = create_message(r, lkb, to_nodeid, DLM_MSG_GRANT, &ms, &mh, + GFP_NOFS); if (error) goto out; @@ -3742,7 +3696,7 @@ static int send_grant(struct dlm_rsb *r, struct dlm_lkb *lkb) ms->m_result = 0; - error = send_message(mh, ms); + error = send_message(mh, ms, r->res_name, r->res_length); out: return error; } @@ -3755,7 +3709,8 @@ static int send_bast(struct dlm_rsb *r, struct dlm_lkb *lkb, int mode) to_nodeid = lkb->lkb_nodeid; - error = create_message(r, NULL, to_nodeid, DLM_MSG_BAST, &ms, &mh); + error = create_message(r, NULL, to_nodeid, DLM_MSG_BAST, &ms, &mh, + GFP_NOFS); if (error) goto out; @@ -3763,7 +3718,7 @@ static int send_bast(struct dlm_rsb *r, struct dlm_lkb *lkb, int mode) ms->m_bastmode = cpu_to_le32(mode); - error = send_message(mh, ms); + error = send_message(mh, ms, r->res_name, r->res_length); out: return error; } @@ -3780,13 +3735,14 @@ static int send_lookup(struct dlm_rsb *r, struct dlm_lkb *lkb) if (error) return error; - error = create_message(r, NULL, to_nodeid, DLM_MSG_LOOKUP, &ms, &mh); + error = create_message(r, NULL, to_nodeid, DLM_MSG_LOOKUP, &ms, &mh, + GFP_NOFS); if (error) goto fail; send_args(r, lkb, ms); - error = send_message(mh, ms); + error = send_message(mh, ms, r->res_name, r->res_length); if (error) goto fail; return 0; @@ -3804,14 +3760,15 @@ static int send_remove(struct dlm_rsb *r) to_nodeid = dlm_dir_nodeid(r); - error = create_message(r, NULL, to_nodeid, DLM_MSG_REMOVE, &ms, &mh); + error = create_message(r, NULL, to_nodeid, DLM_MSG_REMOVE, &ms, &mh, + GFP_ATOMIC); if (error) goto out; memcpy(ms->m_extra, r->res_name, r->res_length); ms->m_hash = cpu_to_le32(r->res_hash); - error = send_message(mh, ms); + error = send_message(mh, ms, r->res_name, r->res_length); out: return error; } @@ -3825,7 +3782,7 @@ static int send_common_reply(struct dlm_rsb *r, struct dlm_lkb *lkb, to_nodeid = lkb->lkb_nodeid; - error = create_message(r, lkb, to_nodeid, mstype, &ms, &mh); + error = create_message(r, lkb, to_nodeid, mstype, &ms, &mh, GFP_NOFS); if (error) goto out; @@ -3833,7 +3790,7 @@ static int send_common_reply(struct dlm_rsb *r, struct dlm_lkb *lkb, ms->m_result = cpu_to_le32(to_dlm_errno(rv)); - error = send_message(mh, ms); + error = send_message(mh, ms, r->res_name, r->res_length); out: return error; } @@ -3866,7 +3823,8 @@ static int send_lookup_reply(struct dlm_ls *ls, struct dlm_message *ms_in, struct dlm_mhandle *mh; int error, nodeid = le32_to_cpu(ms_in->m_header.h_nodeid); - error = create_message(r, NULL, nodeid, DLM_MSG_LOOKUP_REPLY, &ms, &mh); + error = create_message(r, NULL, nodeid, DLM_MSG_LOOKUP_REPLY, &ms, &mh, + GFP_NOFS); if (error) goto out; @@ -3874,7 +3832,7 @@ static int send_lookup_reply(struct dlm_ls *ls, struct dlm_message *ms_in, ms->m_result = cpu_to_le32(to_dlm_errno(rv)); ms->m_nodeid = cpu_to_le32(ret_nodeid); - error = send_message(mh, ms); + error = send_message(mh, ms, ms_in->m_extra, receive_extralen(ms_in)); out: return error; } @@ -4044,66 +4002,6 @@ out: return error; } -static void send_repeat_remove(struct dlm_ls *ls, char *ms_name, int len) -{ - char name[DLM_RESNAME_MAXLEN + 1]; - struct dlm_message *ms; - struct dlm_mhandle *mh; - struct dlm_rsb *r; - uint32_t hash, b; - int rv, dir_nodeid; - - memset(name, 0, sizeof(name)); - memcpy(name, ms_name, len); - - hash = jhash(name, len, 0); - b = hash & (ls->ls_rsbtbl_size - 1); - - dir_nodeid = dlm_hash2nodeid(ls, hash); - - log_error(ls, "send_repeat_remove dir %d %s", dir_nodeid, name); - - spin_lock(&ls->ls_rsbtbl[b].lock); - rv = dlm_search_rsb_tree(&ls->ls_rsbtbl[b].keep, name, len, &r); - if (!rv) { - spin_unlock(&ls->ls_rsbtbl[b].lock); - log_error(ls, "repeat_remove on keep %s", name); - return; - } - - rv = dlm_search_rsb_tree(&ls->ls_rsbtbl[b].toss, name, len, &r); - if (!rv) { - spin_unlock(&ls->ls_rsbtbl[b].lock); - log_error(ls, "repeat_remove on toss %s", name); - return; - } - - /* use ls->remove_name2 to avoid conflict with shrink? */ - - spin_lock(&ls->ls_remove_spin); - ls->ls_remove_len = len; - memcpy(ls->ls_remove_name, name, DLM_RESNAME_MAXLEN); - spin_unlock(&ls->ls_remove_spin); - spin_unlock(&ls->ls_rsbtbl[b].lock); - - rv = _create_message(ls, sizeof(struct dlm_message) + len, - dir_nodeid, DLM_MSG_REMOVE, &ms, &mh); - if (rv) - goto out; - - memcpy(ms->m_extra, name, len); - ms->m_hash = cpu_to_le32(hash); - - send_message(mh, ms); - -out: - spin_lock(&ls->ls_remove_spin); - ls->ls_remove_len = 0; - memset(ls->ls_remove_name, 0, DLM_RESNAME_MAXLEN); - spin_unlock(&ls->ls_remove_spin); - wake_up(&ls->ls_remove_wait); -} - static int receive_request(struct dlm_ls *ls, struct dlm_message *ms) { struct dlm_lkb *lkb; @@ -4173,25 +4071,11 @@ static int receive_request(struct dlm_ls *ls, struct dlm_message *ms) ENOTBLK request failures when the lookup reply designating us as master is delayed. */ - /* We could repeatedly return -EBADR here if our send_remove() is - delayed in being sent/arriving/being processed on the dir node. - Another node would repeatedly lookup up the master, and the dir - node would continue returning our nodeid until our send_remove - took effect. - - We send another remove message in case our previous send_remove - was lost/ignored/missed somehow. */ - if (error != -ENOTBLK) { log_limit(ls, "receive_request %x from %d %d", le32_to_cpu(ms->m_lkid), from_nodeid, error); } - if (namelen && error == -EBADR) { - send_repeat_remove(ls, ms->m_extra, namelen); - msleep(1000); - } - setup_stub_lkb(ls, ms); send_request_reply(&ls->ls_stub_rsb, &ls->ls_stub_lkb, error); return error; @@ -6294,8 +6178,7 @@ void dlm_clear_proc_locks(struct dlm_ls *ls, struct dlm_user_proc *proc) } list_for_each_entry_safe(lkb, safe, &proc->asts, lkb_cb_list) { - memset(&lkb->lkb_callbacks, 0, - sizeof(struct dlm_callback) * DLM_CALLBACKS_SIZE); + dlm_purge_lkb_callbacks(lkb); list_del_init(&lkb->lkb_cb_list); dlm_put_lkb(lkb); } @@ -6336,8 +6219,7 @@ static void purge_proc_locks(struct dlm_ls *ls, struct dlm_user_proc *proc) spin_lock(&proc->asts_spin); list_for_each_entry_safe(lkb, safe, &proc->asts, lkb_cb_list) { - memset(&lkb->lkb_callbacks, 0, - sizeof(struct dlm_callback) * DLM_CALLBACKS_SIZE); + dlm_purge_lkb_callbacks(lkb); list_del_init(&lkb->lkb_cb_list); dlm_put_lkb(lkb); } @@ -6368,13 +6250,13 @@ static int send_purge(struct dlm_ls *ls, int nodeid, int pid) int error; error = _create_message(ls, sizeof(struct dlm_message), nodeid, - DLM_MSG_PURGE, &ms, &mh); + DLM_MSG_PURGE, &ms, &mh, GFP_NOFS); if (error) return error; ms->m_nodeid = cpu_to_le32(nodeid); ms->m_pid = cpu_to_le32(pid); - return send_message(mh, ms); + return send_message(mh, ms, NULL, 0); } int dlm_user_purge(struct dlm_ls *ls, struct dlm_user_proc *proc, diff --git a/fs/dlm/lockspace.c b/fs/dlm/lockspace.c index bae050df7abf..d0b4e2181a5f 100644 --- a/fs/dlm/lockspace.c +++ b/fs/dlm/lockspace.c @@ -17,7 +17,6 @@ #include "recoverd.h" #include "dir.h" #include "midcomms.h" -#include "lowcomms.h" #include "config.h" #include "memory.h" #include "lock.h" @@ -391,7 +390,7 @@ static int threads_start(void) /* Thread for sending/receiving messages for all lockspace's */ error = dlm_midcomms_start(); if (error) { - log_print("cannot start dlm lowcomms %d", error); + log_print("cannot start dlm midcomms %d", error); goto scand_fail; } @@ -473,7 +472,7 @@ static int new_lockspace(const char *name, const char *cluster, error = -ENOMEM; - ls = kzalloc(sizeof(struct dlm_ls) + namelen, GFP_NOFS); + ls = kzalloc(sizeof(*ls), GFP_NOFS); if (!ls) goto out; memcpy(ls->ls_name, name, namelen); @@ -524,9 +523,6 @@ static int new_lockspace(const char *name, const char *cluster, spin_lock_init(&ls->ls_rsbtbl[i].lock); } - spin_lock_init(&ls->ls_remove_spin); - init_waitqueue_head(&ls->ls_remove_wait); - for (i = 0; i < DLM_REMOVE_NAMES_MAX; i++) { ls->ls_remove_names[i] = kzalloc(DLM_RESNAME_MAXLEN+1, GFP_KERNEL); @@ -567,7 +563,7 @@ static int new_lockspace(const char *name, const char *cluster, init_completion(&ls->ls_recovery_done); ls->ls_recovery_result = -1; - mutex_init(&ls->ls_cb_mutex); + spin_lock_init(&ls->ls_cb_lock); INIT_LIST_HEAD(&ls->ls_cb_delay); ls->ls_recoverd_task = NULL; @@ -726,7 +722,7 @@ static int __dlm_new_lockspace(const char *name, const char *cluster, if (!ls_count) { dlm_scand_stop(); dlm_midcomms_shutdown(); - dlm_lowcomms_stop(); + dlm_midcomms_stop(); } out: mutex_unlock(&ls_lock); @@ -929,7 +925,7 @@ int dlm_release_lockspace(void *lockspace, int force) if (!error) ls_count--; if (!ls_count) - dlm_lowcomms_stop(); + dlm_midcomms_stop(); mutex_unlock(&ls_lock); return error; diff --git a/fs/dlm/lowcomms.c b/fs/dlm/lowcomms.c index 59f64c596233..8b80ca0cd65f 100644 --- a/fs/dlm/lowcomms.c +++ b/fs/dlm/lowcomms.c @@ -63,41 +63,49 @@ #define NEEDED_RMEM (4*1024*1024) -/* Number of messages to send before rescheduling */ -#define MAX_SEND_MSG_COUNT 25 -#define DLM_SHUTDOWN_WAIT_TIMEOUT msecs_to_jiffies(10000) - struct connection { struct socket *sock; /* NULL if not connected */ uint32_t nodeid; /* So we know who we are in the list */ - struct mutex sock_mutex; + /* this semaphore is used to allow parallel recv/send in read + * lock mode. When we release a sock we need to held the write lock. + * + * However this is locking code and not nice. When we remove the + * othercon handling we can look into other mechanism to synchronize + * io handling to call sock_release() at the right time. + */ + struct rw_semaphore sock_lock; unsigned long flags; -#define CF_READ_PENDING 1 -#define CF_WRITE_PENDING 2 -#define CF_INIT_PENDING 4 +#define CF_APP_LIMITED 0 +#define CF_RECV_PENDING 1 +#define CF_SEND_PENDING 2 +#define CF_RECV_INTR 3 +#define CF_IO_STOP 4 #define CF_IS_OTHERCON 5 -#define CF_CLOSE 6 -#define CF_APP_LIMITED 7 -#define CF_CLOSING 8 -#define CF_SHUTDOWN 9 -#define CF_CONNECTED 10 -#define CF_RECONNECT 11 -#define CF_DELAY_CONNECT 12 -#define CF_EOF 13 struct list_head writequeue; /* List of outgoing writequeue_entries */ spinlock_t writequeue_lock; - atomic_t writequeue_cnt; int retries; -#define MAX_CONNECT_RETRIES 3 struct hlist_node list; + /* due some connect()/accept() races we currently have this cross over + * connection attempt second connection for one node. + * + * There is a solution to avoid the race by introducing a connect + * rule as e.g. our_nodeid > nodeid_to_connect who is allowed to + * connect. Otherside can connect but will only be considered that + * the other side wants to have a reconnect. + * + * However changing to this behaviour will break backwards compatible. + * In a DLM protocol major version upgrade we should remove this! + */ struct connection *othercon; - struct connection *sendcon; - struct work_struct rwork; /* Receive workqueue */ - struct work_struct swork; /* Send workqueue */ - wait_queue_head_t shutdown_wait; /* wait for graceful shutdown */ - unsigned char *rx_buf; - int rx_buflen; + struct work_struct rwork; /* receive worker */ + struct work_struct swork; /* send worker */ + unsigned char rx_leftover_buf[DLM_MAX_SOCKET_BUFSIZE]; int rx_leftover; + int mark; + int addr_count; + int curr_addr_index; + struct sockaddr_storage addr[DLM_MAX_ADDR_COUNT]; + spinlock_t addrs_lock; struct rcu_head rcu; }; #define sock2con(x) ((struct connection *)(x)->sk_user_data) @@ -136,13 +144,12 @@ struct dlm_msg { struct kref ref; }; -struct dlm_node_addr { - struct list_head list; +struct processqueue_entry { + unsigned char *buf; int nodeid; - int mark; - int addr_count; - int curr_addr_index; - struct sockaddr_storage *addr[DLM_MAX_ADDR_COUNT]; + int buflen; + + struct list_head list; }; struct dlm_proto_ops { @@ -157,10 +164,6 @@ struct dlm_proto_ops { int (*listen_validate)(void); void (*listen_sockopts)(struct socket *sock); int (*listen_bind)(struct socket *sock); - /* What to do to shutdown */ - void (*shutdown_action)(struct connection *con); - /* What to do to eof check */ - bool (*eof_condition)(struct connection *con); }; static struct listen_sock_callbacks { @@ -170,17 +173,13 @@ static struct listen_sock_callbacks { void (*sk_write_space)(struct sock *); } listen_sock; -static LIST_HEAD(dlm_node_addrs); -static DEFINE_SPINLOCK(dlm_node_addrs_spin); - static struct listen_connection listen_con; -static struct sockaddr_storage *dlm_local_addr[DLM_MAX_ADDR_COUNT]; +static struct sockaddr_storage dlm_local_addr[DLM_MAX_ADDR_COUNT]; static int dlm_local_count; -int dlm_allow_conn; /* Work queues */ -static struct workqueue_struct *recv_workqueue; -static struct workqueue_struct *send_workqueue; +static struct workqueue_struct *io_workqueue; +static struct workqueue_struct *process_workqueue; static struct hlist_head connection_hash[CONN_HASH_SIZE]; static DEFINE_SPINLOCK(connections_lock); @@ -188,8 +187,45 @@ DEFINE_STATIC_SRCU(connections_srcu); static const struct dlm_proto_ops *dlm_proto_ops; +#define DLM_IO_SUCCESS 0 +#define DLM_IO_END 1 +#define DLM_IO_EOF 2 +#define DLM_IO_RESCHED 3 + static void process_recv_sockets(struct work_struct *work); static void process_send_sockets(struct work_struct *work); +static void process_dlm_messages(struct work_struct *work); + +static DECLARE_WORK(process_work, process_dlm_messages); +static DEFINE_SPINLOCK(processqueue_lock); +static bool process_dlm_messages_pending; +static LIST_HEAD(processqueue); + +bool dlm_lowcomms_is_running(void) +{ + return !!listen_con.sock; +} + +static void lowcomms_queue_swork(struct connection *con) +{ + assert_spin_locked(&con->writequeue_lock); + + if (!test_bit(CF_IO_STOP, &con->flags) && + !test_bit(CF_APP_LIMITED, &con->flags) && + !test_and_set_bit(CF_SEND_PENDING, &con->flags)) + queue_work(io_workqueue, &con->swork); +} + +static void lowcomms_queue_rwork(struct connection *con) +{ +#ifdef CONFIG_LOCKDEP + WARN_ON_ONCE(!lockdep_sock_is_held(con->sock->sk)); +#endif + + if (!test_bit(CF_IO_STOP, &con->flags) && + !test_and_set_bit(CF_RECV_PENDING, &con->flags)) + queue_work(io_workqueue, &con->rwork); +} static void writequeue_entry_ctor(void *data) { @@ -214,15 +250,12 @@ static struct writequeue_entry *con_next_wq(struct connection *con) { struct writequeue_entry *e; - if (list_empty(&con->writequeue)) - return NULL; - - e = list_first_entry(&con->writequeue, struct writequeue_entry, - list); + e = list_first_entry_or_null(&con->writequeue, struct writequeue_entry, + list); /* if len is zero nothing is to send, if there are users filling * buffers we wait until the users are done so we can send more. */ - if (e->users || e->len == 0) + if (!e || e->users || e->len == 0) return NULL; return e; @@ -240,28 +273,15 @@ static struct connection *__find_con(int nodeid, int r) return NULL; } -static bool tcp_eof_condition(struct connection *con) -{ - return atomic_read(&con->writequeue_cnt); -} - -static int dlm_con_init(struct connection *con, int nodeid) +static void dlm_con_init(struct connection *con, int nodeid) { - con->rx_buflen = dlm_config.ci_buffer_size; - con->rx_buf = kmalloc(con->rx_buflen, GFP_NOFS); - if (!con->rx_buf) - return -ENOMEM; - con->nodeid = nodeid; - mutex_init(&con->sock_mutex); + init_rwsem(&con->sock_lock); INIT_LIST_HEAD(&con->writequeue); spin_lock_init(&con->writequeue_lock); - atomic_set(&con->writequeue_cnt, 0); INIT_WORK(&con->swork, process_send_sockets); INIT_WORK(&con->rwork, process_recv_sockets); - init_waitqueue_head(&con->shutdown_wait); - - return 0; + spin_lock_init(&con->addrs_lock); } /* @@ -271,7 +291,7 @@ static int dlm_con_init(struct connection *con, int nodeid) static struct connection *nodeid2con(int nodeid, gfp_t alloc) { struct connection *con, *tmp; - int r, ret; + int r; r = nodeid_hash(nodeid); con = __find_con(nodeid, r); @@ -282,11 +302,7 @@ static struct connection *nodeid2con(int nodeid, gfp_t alloc) if (!con) return NULL; - ret = dlm_con_init(con, nodeid); - if (ret) { - kfree(con); - return NULL; - } + dlm_con_init(con, nodeid); spin_lock(&connections_lock); /* Because multiple workqueues/threads calls this function it can @@ -298,7 +314,6 @@ static struct connection *nodeid2con(int nodeid, gfp_t alloc) tmp = __find_con(nodeid, r); if (tmp) { spin_unlock(&connections_lock); - kfree(con->rx_buf); kfree(con); return tmp; } @@ -309,29 +324,6 @@ static struct connection *nodeid2con(int nodeid, gfp_t alloc) return con; } -/* Loop round all connections */ -static void foreach_conn(void (*conn_func)(struct connection *c)) -{ - int i; - struct connection *con; - - for (i = 0; i < CONN_HASH_SIZE; i++) { - hlist_for_each_entry_rcu(con, &connection_hash[i], list) - conn_func(con); - } -} - -static struct dlm_node_addr *find_node_addr(int nodeid) -{ - struct dlm_node_addr *na; - - list_for_each_entry(na, &dlm_node_addrs, list) { - if (na->nodeid == nodeid) - return na; - } - return NULL; -} - static int addr_compare(const struct sockaddr_storage *x, const struct sockaddr_storage *y) { @@ -365,40 +357,47 @@ static int nodeid_to_addr(int nodeid, struct sockaddr_storage *sas_out, unsigned int *mark) { struct sockaddr_storage sas; - struct dlm_node_addr *na; + struct connection *con; + int idx; if (!dlm_local_count) return -1; - spin_lock(&dlm_node_addrs_spin); - na = find_node_addr(nodeid); - if (na && na->addr_count) { - memcpy(&sas, na->addr[na->curr_addr_index], - sizeof(struct sockaddr_storage)); + idx = srcu_read_lock(&connections_srcu); + con = nodeid2con(nodeid, 0); + if (!con) { + srcu_read_unlock(&connections_srcu, idx); + return -ENOENT; + } - if (try_new_addr) { - na->curr_addr_index++; - if (na->curr_addr_index == na->addr_count) - na->curr_addr_index = 0; - } + spin_lock(&con->addrs_lock); + if (!con->addr_count) { + spin_unlock(&con->addrs_lock); + srcu_read_unlock(&connections_srcu, idx); + return -ENOENT; } - spin_unlock(&dlm_node_addrs_spin); - if (!na) - return -EEXIST; + memcpy(&sas, &con->addr[con->curr_addr_index], + sizeof(struct sockaddr_storage)); - if (!na->addr_count) - return -ENOENT; + if (try_new_addr) { + con->curr_addr_index++; + if (con->curr_addr_index == con->addr_count) + con->curr_addr_index = 0; + } - *mark = na->mark; + *mark = con->mark; + spin_unlock(&con->addrs_lock); if (sas_out) memcpy(sas_out, &sas, sizeof(struct sockaddr_storage)); - if (!sa_out) + if (!sa_out) { + srcu_read_unlock(&connections_srcu, idx); return 0; + } - if (dlm_local_addr[0]->ss_family == AF_INET) { + if (dlm_local_addr[0].ss_family == AF_INET) { struct sockaddr_in *in4 = (struct sockaddr_in *) &sas; struct sockaddr_in *ret4 = (struct sockaddr_in *) sa_out; ret4->sin_addr.s_addr = in4->sin_addr.s_addr; @@ -408,43 +407,46 @@ static int nodeid_to_addr(int nodeid, struct sockaddr_storage *sas_out, ret6->sin6_addr = in6->sin6_addr; } + srcu_read_unlock(&connections_srcu, idx); return 0; } static int addr_to_nodeid(struct sockaddr_storage *addr, int *nodeid, unsigned int *mark) { - struct dlm_node_addr *na; - int rv = -EEXIST; - int addr_i; - - spin_lock(&dlm_node_addrs_spin); - list_for_each_entry(na, &dlm_node_addrs, list) { - if (!na->addr_count) - continue; - - for (addr_i = 0; addr_i < na->addr_count; addr_i++) { - if (addr_compare(na->addr[addr_i], addr)) { - *nodeid = na->nodeid; - *mark = na->mark; - rv = 0; - goto unlock; + struct connection *con; + int i, idx, addr_i; + + idx = srcu_read_lock(&connections_srcu); + for (i = 0; i < CONN_HASH_SIZE; i++) { + hlist_for_each_entry_rcu(con, &connection_hash[i], list) { + WARN_ON_ONCE(!con->addr_count); + + spin_lock(&con->addrs_lock); + for (addr_i = 0; addr_i < con->addr_count; addr_i++) { + if (addr_compare(&con->addr[addr_i], addr)) { + *nodeid = con->nodeid; + *mark = con->mark; + spin_unlock(&con->addrs_lock); + srcu_read_unlock(&connections_srcu, idx); + return 0; + } } + spin_unlock(&con->addrs_lock); } } -unlock: - spin_unlock(&dlm_node_addrs_spin); - return rv; + srcu_read_unlock(&connections_srcu, idx); + + return -ENOENT; } -/* caller need to held dlm_node_addrs_spin lock */ -static bool dlm_lowcomms_na_has_addr(const struct dlm_node_addr *na, - const struct sockaddr_storage *addr) +static bool dlm_lowcomms_con_has_addr(const struct connection *con, + const struct sockaddr_storage *addr) { int i; - for (i = 0; i < na->addr_count; i++) { - if (addr_compare(na->addr[i], addr)) + for (i = 0; i < con->addr_count; i++) { + if (addr_compare(&con->addr[i], addr)) return true; } @@ -453,118 +455,82 @@ static bool dlm_lowcomms_na_has_addr(const struct dlm_node_addr *na, int dlm_lowcomms_addr(int nodeid, struct sockaddr_storage *addr, int len) { - struct sockaddr_storage *new_addr; - struct dlm_node_addr *new_node, *na; - bool ret; - - new_node = kzalloc(sizeof(struct dlm_node_addr), GFP_NOFS); - if (!new_node) - return -ENOMEM; + struct connection *con; + bool ret, idx; - new_addr = kzalloc(sizeof(struct sockaddr_storage), GFP_NOFS); - if (!new_addr) { - kfree(new_node); + idx = srcu_read_lock(&connections_srcu); + con = nodeid2con(nodeid, GFP_NOFS); + if (!con) { + srcu_read_unlock(&connections_srcu, idx); return -ENOMEM; } - memcpy(new_addr, addr, len); - - spin_lock(&dlm_node_addrs_spin); - na = find_node_addr(nodeid); - if (!na) { - new_node->nodeid = nodeid; - new_node->addr[0] = new_addr; - new_node->addr_count = 1; - new_node->mark = dlm_config.ci_mark; - list_add(&new_node->list, &dlm_node_addrs); - spin_unlock(&dlm_node_addrs_spin); + spin_lock(&con->addrs_lock); + if (!con->addr_count) { + memcpy(&con->addr[0], addr, sizeof(*addr)); + con->addr_count = 1; + con->mark = dlm_config.ci_mark; + spin_unlock(&con->addrs_lock); + srcu_read_unlock(&connections_srcu, idx); return 0; } - ret = dlm_lowcomms_na_has_addr(na, addr); + ret = dlm_lowcomms_con_has_addr(con, addr); if (ret) { - spin_unlock(&dlm_node_addrs_spin); - kfree(new_addr); - kfree(new_node); + spin_unlock(&con->addrs_lock); + srcu_read_unlock(&connections_srcu, idx); return -EEXIST; } - if (na->addr_count >= DLM_MAX_ADDR_COUNT) { - spin_unlock(&dlm_node_addrs_spin); - kfree(new_addr); - kfree(new_node); + if (con->addr_count >= DLM_MAX_ADDR_COUNT) { + spin_unlock(&con->addrs_lock); + srcu_read_unlock(&connections_srcu, idx); return -ENOSPC; } - na->addr[na->addr_count++] = new_addr; - spin_unlock(&dlm_node_addrs_spin); - kfree(new_node); + memcpy(&con->addr[con->addr_count++], addr, sizeof(*addr)); + srcu_read_unlock(&connections_srcu, idx); + spin_unlock(&con->addrs_lock); return 0; } /* Data available on socket or listen socket received a connect */ static void lowcomms_data_ready(struct sock *sk) { - struct connection *con; - - con = sock2con(sk); - if (con && !test_and_set_bit(CF_READ_PENDING, &con->flags)) - queue_work(recv_workqueue, &con->rwork); -} - -static void lowcomms_listen_data_ready(struct sock *sk) -{ - if (!dlm_allow_conn) - return; + struct connection *con = sock2con(sk); - queue_work(recv_workqueue, &listen_con.rwork); + set_bit(CF_RECV_INTR, &con->flags); + lowcomms_queue_rwork(con); } static void lowcomms_write_space(struct sock *sk) { - struct connection *con; - - con = sock2con(sk); - if (!con) - return; - - if (!test_and_set_bit(CF_CONNECTED, &con->flags)) { - log_print("connected to node %d", con->nodeid); - queue_work(send_workqueue, &con->swork); - return; - } + struct connection *con = sock2con(sk); clear_bit(SOCK_NOSPACE, &con->sock->flags); + spin_lock_bh(&con->writequeue_lock); if (test_and_clear_bit(CF_APP_LIMITED, &con->flags)) { con->sock->sk->sk_write_pending--; clear_bit(SOCKWQ_ASYNC_NOSPACE, &con->sock->flags); } - queue_work(send_workqueue, &con->swork); -} - -static inline void lowcomms_connect_sock(struct connection *con) -{ - if (test_bit(CF_CLOSE, &con->flags)) - return; - queue_work(send_workqueue, &con->swork); - cond_resched(); + lowcomms_queue_swork(con); + spin_unlock_bh(&con->writequeue_lock); } static void lowcomms_state_change(struct sock *sk) { /* SCTP layer is not calling sk_data_ready when the connection - * is done, so we catch the signal through here. Also, it - * doesn't switch socket state when entering shutdown, so we - * skip the write in that case. + * is done, so we catch the signal through here. */ - if (sk->sk_shutdown) { - if (sk->sk_shutdown == RCV_SHUTDOWN) - lowcomms_data_ready(sk); - } else if (sk->sk_state == TCP_ESTABLISHED) { - lowcomms_write_space(sk); - } + if (sk->sk_shutdown == RCV_SHUTDOWN) + lowcomms_data_ready(sk); +} + +static void lowcomms_listen_data_ready(struct sock *sk) +{ + queue_work(io_workqueue, &listen_con.rwork); } int dlm_lowcomms_connect_node(int nodeid) @@ -576,47 +542,49 @@ int dlm_lowcomms_connect_node(int nodeid) return 0; idx = srcu_read_lock(&connections_srcu); - con = nodeid2con(nodeid, GFP_NOFS); - if (!con) { + con = nodeid2con(nodeid, 0); + if (WARN_ON_ONCE(!con)) { srcu_read_unlock(&connections_srcu, idx); - return -ENOMEM; + return -ENOENT; } - lowcomms_connect_sock(con); + down_read(&con->sock_lock); + if (!con->sock) { + spin_lock_bh(&con->writequeue_lock); + lowcomms_queue_swork(con); + spin_unlock_bh(&con->writequeue_lock); + } + up_read(&con->sock_lock); srcu_read_unlock(&connections_srcu, idx); + cond_resched(); return 0; } int dlm_lowcomms_nodes_set_mark(int nodeid, unsigned int mark) { - struct dlm_node_addr *na; + struct connection *con; + int idx; - spin_lock(&dlm_node_addrs_spin); - na = find_node_addr(nodeid); - if (!na) { - spin_unlock(&dlm_node_addrs_spin); + idx = srcu_read_lock(&connections_srcu); + con = nodeid2con(nodeid, 0); + if (!con) { + srcu_read_unlock(&connections_srcu, idx); return -ENOENT; } - na->mark = mark; - spin_unlock(&dlm_node_addrs_spin); - + spin_lock(&con->addrs_lock); + con->mark = mark; + spin_unlock(&con->addrs_lock); + srcu_read_unlock(&connections_srcu, idx); return 0; } static void lowcomms_error_report(struct sock *sk) { - struct connection *con; - void (*orig_report)(struct sock *) = NULL; + struct connection *con = sock2con(sk); struct inet_sock *inet; - con = sock2con(sk); - if (con == NULL) - goto out; - - orig_report = listen_sock.sk_error_report; - inet = inet_sk(sk); switch (sk->sk_family) { case AF_INET: @@ -642,66 +610,25 @@ static void lowcomms_error_report(struct sock *sk) "invalid socket family %d set, " "sk_err=%d/%d\n", dlm_our_nodeid(), sk->sk_family, sk->sk_err, sk->sk_err_soft); - goto out; - } - - /* below sendcon only handling */ - if (test_bit(CF_IS_OTHERCON, &con->flags)) - con = con->sendcon; - - switch (sk->sk_err) { - case ECONNREFUSED: - set_bit(CF_DELAY_CONNECT, &con->flags); - break; - default: break; } - if (!test_and_set_bit(CF_RECONNECT, &con->flags)) - queue_work(send_workqueue, &con->swork); + dlm_midcomms_unack_msg_resend(con->nodeid); -out: - if (orig_report) - orig_report(sk); + listen_sock.sk_error_report(sk); } -/* Note: sk_callback_lock must be locked before calling this function. */ -static void save_listen_callbacks(struct socket *sock) +static void restore_callbacks(struct sock *sk) { - struct sock *sk = sock->sk; - - listen_sock.sk_data_ready = sk->sk_data_ready; - listen_sock.sk_state_change = sk->sk_state_change; - listen_sock.sk_write_space = sk->sk_write_space; - listen_sock.sk_error_report = sk->sk_error_report; -} - -static void restore_callbacks(struct socket *sock) -{ - struct sock *sk = sock->sk; +#ifdef CONFIG_LOCKDEP + WARN_ON_ONCE(!lockdep_sock_is_held(sk)); +#endif - lock_sock(sk); sk->sk_user_data = NULL; sk->sk_data_ready = listen_sock.sk_data_ready; sk->sk_state_change = listen_sock.sk_state_change; sk->sk_write_space = listen_sock.sk_write_space; sk->sk_error_report = listen_sock.sk_error_report; - release_sock(sk); -} - -static void add_listen_sock(struct socket *sock, struct listen_connection *con) -{ - struct sock *sk = sock->sk; - - lock_sock(sk); - save_listen_callbacks(sock); - con->sock = sock; - - sk->sk_user_data = con; - sk->sk_allocation = GFP_NOFS; - /* Install a data_ready callback */ - sk->sk_data_ready = lowcomms_listen_data_ready; - release_sock(sk); } /* Make a socket active */ @@ -713,10 +640,10 @@ static void add_sock(struct socket *sock, struct connection *con) con->sock = sock; sk->sk_user_data = con; - /* Install a data_ready callback */ sk->sk_data_ready = lowcomms_data_ready; sk->sk_write_space = lowcomms_write_space; - sk->sk_state_change = lowcomms_state_change; + if (dlm_config.ci_protocol == DLM_PROTO_SCTP) + sk->sk_state_change = lowcomms_state_change; sk->sk_allocation = GFP_NOFS; sk->sk_error_report = lowcomms_error_report; release_sock(sk); @@ -727,7 +654,7 @@ static void add_sock(struct socket *sock, struct connection *con) static void make_sockaddr(struct sockaddr_storage *saddr, uint16_t port, int *addr_len) { - saddr->ss_family = dlm_local_addr[0]->ss_family; + saddr->ss_family = dlm_local_addr[0].ss_family; if (saddr->ss_family == AF_INET) { struct sockaddr_in *in4_addr = (struct sockaddr_in *)saddr; in4_addr->sin_port = cpu_to_be16(port); @@ -773,43 +700,67 @@ static void free_entry(struct writequeue_entry *e) } list_del(&e->list); - atomic_dec(&e->con->writequeue_cnt); kref_put(&e->ref, dlm_page_release); } static void dlm_close_sock(struct socket **sock) { - if (*sock) { - restore_callbacks(*sock); - sock_release(*sock); - *sock = NULL; + lock_sock((*sock)->sk); + restore_callbacks((*sock)->sk); + release_sock((*sock)->sk); + + sock_release(*sock); + *sock = NULL; +} + +static void allow_connection_io(struct connection *con) +{ + if (con->othercon) + clear_bit(CF_IO_STOP, &con->othercon->flags); + clear_bit(CF_IO_STOP, &con->flags); +} + +static void stop_connection_io(struct connection *con) +{ + if (con->othercon) + stop_connection_io(con->othercon); + + down_write(&con->sock_lock); + if (con->sock) { + lock_sock(con->sock->sk); + restore_callbacks(con->sock->sk); + + spin_lock_bh(&con->writequeue_lock); + set_bit(CF_IO_STOP, &con->flags); + spin_unlock_bh(&con->writequeue_lock); + release_sock(con->sock->sk); + } else { + spin_lock_bh(&con->writequeue_lock); + set_bit(CF_IO_STOP, &con->flags); + spin_unlock_bh(&con->writequeue_lock); } + up_write(&con->sock_lock); + + cancel_work_sync(&con->swork); + cancel_work_sync(&con->rwork); } /* Close a remote connection and tidy up */ -static void close_connection(struct connection *con, bool and_other, - bool tx, bool rx) +static void close_connection(struct connection *con, bool and_other) { - bool closing = test_and_set_bit(CF_CLOSING, &con->flags); struct writequeue_entry *e; - if (tx && !closing && cancel_work_sync(&con->swork)) { - log_print("canceled swork for node %d", con->nodeid); - clear_bit(CF_WRITE_PENDING, &con->flags); - } - if (rx && !closing && cancel_work_sync(&con->rwork)) { - log_print("canceled rwork for node %d", con->nodeid); - clear_bit(CF_READ_PENDING, &con->flags); + if (con->othercon && and_other) + close_connection(con->othercon, false); + + down_write(&con->sock_lock); + if (!con->sock) { + up_write(&con->sock_lock); + return; } - mutex_lock(&con->sock_mutex); dlm_close_sock(&con->sock); - if (con->othercon && and_other) { - /* Will only re-enter once. */ - close_connection(con->othercon, false, tx, rx); - } - /* if we send a writequeue entry only a half way, we drop the * whole entry because reconnection and that we not start of the * middle of a msg which will confuse the other end. @@ -821,200 +772,209 @@ static void close_connection(struct connection *con, bool and_other, * our policy is to start on a clean state when disconnects, we don't * know what's send/received on transport layer in this case. */ - spin_lock(&con->writequeue_lock); + spin_lock_bh(&con->writequeue_lock); if (!list_empty(&con->writequeue)) { e = list_first_entry(&con->writequeue, struct writequeue_entry, list); if (e->dirty) free_entry(e); } - spin_unlock(&con->writequeue_lock); + spin_unlock_bh(&con->writequeue_lock); con->rx_leftover = 0; con->retries = 0; clear_bit(CF_APP_LIMITED, &con->flags); - clear_bit(CF_CONNECTED, &con->flags); - clear_bit(CF_DELAY_CONNECT, &con->flags); - clear_bit(CF_RECONNECT, &con->flags); - clear_bit(CF_EOF, &con->flags); - mutex_unlock(&con->sock_mutex); - clear_bit(CF_CLOSING, &con->flags); + clear_bit(CF_RECV_PENDING, &con->flags); + clear_bit(CF_SEND_PENDING, &con->flags); + up_write(&con->sock_lock); } -static void shutdown_connection(struct connection *con) +static struct processqueue_entry *new_processqueue_entry(int nodeid, + int buflen) { - int ret; - - flush_work(&con->swork); + struct processqueue_entry *pentry; - mutex_lock(&con->sock_mutex); - /* nothing to shutdown */ - if (!con->sock) { - mutex_unlock(&con->sock_mutex); - return; - } + pentry = kmalloc(sizeof(*pentry), GFP_NOFS); + if (!pentry) + return NULL; - set_bit(CF_SHUTDOWN, &con->flags); - ret = kernel_sock_shutdown(con->sock, SHUT_WR); - mutex_unlock(&con->sock_mutex); - if (ret) { - log_print("Connection %p failed to shutdown: %d will force close", - con, ret); - goto force_close; - } else { - ret = wait_event_timeout(con->shutdown_wait, - !test_bit(CF_SHUTDOWN, &con->flags), - DLM_SHUTDOWN_WAIT_TIMEOUT); - if (ret == 0) { - log_print("Connection %p shutdown timed out, will force close", - con); - goto force_close; - } + pentry->buf = kmalloc(buflen, GFP_NOFS); + if (!pentry->buf) { + kfree(pentry); + return NULL; } - return; + pentry->nodeid = nodeid; + return pentry; +} -force_close: - clear_bit(CF_SHUTDOWN, &con->flags); - close_connection(con, false, true, true); +static void free_processqueue_entry(struct processqueue_entry *pentry) +{ + kfree(pentry->buf); + kfree(pentry); } -static void dlm_tcp_shutdown(struct connection *con) +struct dlm_processed_nodes { + int nodeid; + + struct list_head list; +}; + +static void add_processed_node(int nodeid, struct list_head *processed_nodes) { - if (con->othercon) - shutdown_connection(con->othercon); - shutdown_connection(con); + struct dlm_processed_nodes *n; + + list_for_each_entry(n, processed_nodes, list) { + /* we already remembered this node */ + if (n->nodeid == nodeid) + return; + } + + /* if it's fails in worst case we simple don't send an ack back. + * We try it next time. + */ + n = kmalloc(sizeof(*n), GFP_NOFS); + if (!n) + return; + + n->nodeid = nodeid; + list_add(&n->list, processed_nodes); } -static int con_realloc_receive_buf(struct connection *con, int newlen) +static void process_dlm_messages(struct work_struct *work) { - unsigned char *newbuf; + struct dlm_processed_nodes *n, *n_tmp; + struct processqueue_entry *pentry; + LIST_HEAD(processed_nodes); - newbuf = kmalloc(newlen, GFP_NOFS); - if (!newbuf) - return -ENOMEM; + spin_lock(&processqueue_lock); + pentry = list_first_entry_or_null(&processqueue, + struct processqueue_entry, list); + if (WARN_ON_ONCE(!pentry)) { + spin_unlock(&processqueue_lock); + return; + } - /* copy any leftover from last receive */ - if (con->rx_leftover) - memmove(newbuf, con->rx_buf, con->rx_leftover); + list_del(&pentry->list); + spin_unlock(&processqueue_lock); - /* swap to new buffer space */ - kfree(con->rx_buf); - con->rx_buflen = newlen; - con->rx_buf = newbuf; + for (;;) { + dlm_process_incoming_buffer(pentry->nodeid, pentry->buf, + pentry->buflen); + add_processed_node(pentry->nodeid, &processed_nodes); + free_processqueue_entry(pentry); + + spin_lock(&processqueue_lock); + pentry = list_first_entry_or_null(&processqueue, + struct processqueue_entry, list); + if (!pentry) { + process_dlm_messages_pending = false; + spin_unlock(&processqueue_lock); + break; + } - return 0; + list_del(&pentry->list); + spin_unlock(&processqueue_lock); + } + + /* send ack back after we processed couple of messages */ + list_for_each_entry_safe(n, n_tmp, &processed_nodes, list) { + list_del(&n->list); + dlm_midcomms_receive_done(n->nodeid); + kfree(n); + } } /* Data received from remote end */ -static int receive_from_sock(struct connection *con) +static int receive_from_sock(struct connection *con, int buflen) { + struct processqueue_entry *pentry; + int ret, buflen_real; struct msghdr msg; struct kvec iov; - int ret, buflen; - mutex_lock(&con->sock_mutex); + pentry = new_processqueue_entry(con->nodeid, buflen); + if (!pentry) + return DLM_IO_RESCHED; - if (con->sock == NULL) { - ret = -EAGAIN; - goto out_close; - } - - /* realloc if we get new buffer size to read out */ - buflen = dlm_config.ci_buffer_size; - if (con->rx_buflen != buflen && con->rx_leftover <= buflen) { - ret = con_realloc_receive_buf(con, buflen); - if (ret < 0) - goto out_resched; - } + memcpy(pentry->buf, con->rx_leftover_buf, con->rx_leftover); - for (;;) { - /* calculate new buffer parameter regarding last receive and - * possible leftover bytes - */ - iov.iov_base = con->rx_buf + con->rx_leftover; - iov.iov_len = con->rx_buflen - con->rx_leftover; - - memset(&msg, 0, sizeof(msg)); - msg.msg_flags = MSG_DONTWAIT | MSG_NOSIGNAL; - ret = kernel_recvmsg(con->sock, &msg, &iov, 1, iov.iov_len, - msg.msg_flags); - trace_dlm_recv(con->nodeid, ret); - if (ret == -EAGAIN) - break; - else if (ret <= 0) - goto out_close; - - /* new buflen according readed bytes and leftover from last receive */ - buflen = ret + con->rx_leftover; - ret = dlm_process_incoming_buffer(con->nodeid, con->rx_buf, buflen); - if (ret < 0) - goto out_close; - - /* calculate leftover bytes from process and put it into begin of - * the receive buffer, so next receive we have the full message - * at the start address of the receive buffer. - */ - con->rx_leftover = buflen - ret; - if (con->rx_leftover) { - memmove(con->rx_buf, con->rx_buf + ret, - con->rx_leftover); + /* calculate new buffer parameter regarding last receive and + * possible leftover bytes + */ + iov.iov_base = pentry->buf + con->rx_leftover; + iov.iov_len = buflen - con->rx_leftover; + + memset(&msg, 0, sizeof(msg)); + msg.msg_flags = MSG_DONTWAIT | MSG_NOSIGNAL; + clear_bit(CF_RECV_INTR, &con->flags); +again: + ret = kernel_recvmsg(con->sock, &msg, &iov, 1, iov.iov_len, + msg.msg_flags); + trace_dlm_recv(con->nodeid, ret); + if (ret == -EAGAIN) { + lock_sock(con->sock->sk); + if (test_and_clear_bit(CF_RECV_INTR, &con->flags)) { + release_sock(con->sock->sk); + goto again; } + + clear_bit(CF_RECV_PENDING, &con->flags); + release_sock(con->sock->sk); + free_processqueue_entry(pentry); + return DLM_IO_END; + } else if (ret == 0) { + /* close will clear CF_RECV_PENDING */ + free_processqueue_entry(pentry); + return DLM_IO_EOF; + } else if (ret < 0) { + free_processqueue_entry(pentry); + return ret; } - dlm_midcomms_receive_done(con->nodeid); - mutex_unlock(&con->sock_mutex); - return 0; + /* new buflen according readed bytes and leftover from last receive */ + buflen_real = ret + con->rx_leftover; + ret = dlm_validate_incoming_buffer(con->nodeid, pentry->buf, + buflen_real); + if (ret < 0) { + free_processqueue_entry(pentry); + return ret; + } -out_resched: - if (!test_and_set_bit(CF_READ_PENDING, &con->flags)) - queue_work(recv_workqueue, &con->rwork); - mutex_unlock(&con->sock_mutex); - return -EAGAIN; - -out_close: - if (ret == 0) { - log_print("connection %p got EOF from %d", - con, con->nodeid); - - if (dlm_proto_ops->eof_condition && - dlm_proto_ops->eof_condition(con)) { - set_bit(CF_EOF, &con->flags); - mutex_unlock(&con->sock_mutex); - } else { - mutex_unlock(&con->sock_mutex); - close_connection(con, false, true, false); + pentry->buflen = ret; - /* handling for tcp shutdown */ - clear_bit(CF_SHUTDOWN, &con->flags); - wake_up(&con->shutdown_wait); - } + /* calculate leftover bytes from process and put it into begin of + * the receive buffer, so next receive we have the full message + * at the start address of the receive buffer. + */ + con->rx_leftover = buflen_real - ret; + memmove(con->rx_leftover_buf, pentry->buf + ret, + con->rx_leftover); - /* signal to breaking receive worker */ - ret = -1; - } else { - mutex_unlock(&con->sock_mutex); + spin_lock(&processqueue_lock); + list_add_tail(&pentry->list, &processqueue); + if (!process_dlm_messages_pending) { + process_dlm_messages_pending = true; + queue_work(process_workqueue, &process_work); } - return ret; + spin_unlock(&processqueue_lock); + + return DLM_IO_SUCCESS; } /* Listening socket is busy, accept a connection */ -static int accept_from_sock(struct listen_connection *con) +static int accept_from_sock(void) { - int result; struct sockaddr_storage peeraddr; - struct socket *newsock; - int len, idx; - int nodeid; + int len, idx, result, nodeid; struct connection *newcon; - struct connection *addcon; + struct socket *newsock; unsigned int mark; - if (!con->sock) - return -ENOTCONN; - - result = kernel_accept(con->sock, &newsock, O_NONBLOCK); - if (result < 0) + result = kernel_accept(listen_con.sock, &newsock, O_NONBLOCK); + if (result == -EAGAIN) + return DLM_IO_END; + else if (result < 0) goto accept_err; /* Get the connected socket's peer */ @@ -1062,16 +1022,16 @@ static int accept_from_sock(struct listen_connection *con) * In this case we store the incoming one in "othercon" */ idx = srcu_read_lock(&connections_srcu); - newcon = nodeid2con(nodeid, GFP_NOFS); - if (!newcon) { + newcon = nodeid2con(nodeid, 0); + if (WARN_ON_ONCE(!newcon)) { srcu_read_unlock(&connections_srcu, idx); - result = -ENOMEM; + result = -ENOENT; goto accept_err; } sock_set_mark(newsock->sk, mark); - mutex_lock(&newcon->sock_mutex); + down_write(&newcon->sock_lock); if (newcon->sock) { struct connection *othercon = newcon->othercon; @@ -1079,63 +1039,50 @@ static int accept_from_sock(struct listen_connection *con) othercon = kzalloc(sizeof(*othercon), GFP_NOFS); if (!othercon) { log_print("failed to allocate incoming socket"); - mutex_unlock(&newcon->sock_mutex); + up_write(&newcon->sock_lock); srcu_read_unlock(&connections_srcu, idx); result = -ENOMEM; goto accept_err; } - result = dlm_con_init(othercon, nodeid); - if (result < 0) { - kfree(othercon); - mutex_unlock(&newcon->sock_mutex); - srcu_read_unlock(&connections_srcu, idx); - goto accept_err; - } - - lockdep_set_subclass(&othercon->sock_mutex, 1); - set_bit(CF_IS_OTHERCON, &othercon->flags); + dlm_con_init(othercon, nodeid); + lockdep_set_subclass(&othercon->sock_lock, 1); newcon->othercon = othercon; - othercon->sendcon = newcon; + set_bit(CF_IS_OTHERCON, &othercon->flags); } else { /* close other sock con if we have something new */ - close_connection(othercon, false, true, false); + close_connection(othercon, false); } - mutex_lock(&othercon->sock_mutex); + down_write(&othercon->sock_lock); add_sock(newsock, othercon); - addcon = othercon; - mutex_unlock(&othercon->sock_mutex); + + /* check if we receved something while adding */ + lock_sock(othercon->sock->sk); + lowcomms_queue_rwork(othercon); + release_sock(othercon->sock->sk); + up_write(&othercon->sock_lock); } else { /* accept copies the sk after we've saved the callbacks, so we don't want to save them a second time or comm errors will result in calling sk_error_report recursively. */ add_sock(newsock, newcon); - addcon = newcon; - } - - set_bit(CF_CONNECTED, &addcon->flags); - mutex_unlock(&newcon->sock_mutex); - - /* - * Add it to the active queue in case we got data - * between processing the accept adding the socket - * to the read_sockets list - */ - if (!test_and_set_bit(CF_READ_PENDING, &addcon->flags)) - queue_work(recv_workqueue, &addcon->rwork); + /* check if we receved something while adding */ + lock_sock(newcon->sock->sk); + lowcomms_queue_rwork(newcon); + release_sock(newcon->sock->sk); + } + up_write(&newcon->sock_lock); srcu_read_unlock(&connections_srcu, idx); - return 0; + return DLM_IO_SUCCESS; accept_err: if (newsock) sock_release(newsock); - if (result != -EAGAIN) - log_print("error accepting connection from node: %d", result); return result; } @@ -1167,7 +1114,7 @@ static int sctp_bind_addrs(struct socket *sock, uint16_t port) int i, addr_len, result = 0; for (i = 0; i < dlm_local_count; i++) { - memcpy(&localaddr, dlm_local_addr[i], sizeof(localaddr)); + memcpy(&localaddr, &dlm_local_addr[i], sizeof(localaddr)); make_sockaddr(&localaddr, port, &addr_len); if (!i) @@ -1187,7 +1134,7 @@ static int sctp_bind_addrs(struct socket *sock, uint16_t port) /* Get local addresses */ static void init_local(void) { - struct sockaddr_storage sas, *addr; + struct sockaddr_storage sas; int i; dlm_local_count = 0; @@ -1195,21 +1142,10 @@ static void init_local(void) if (dlm_our_addr(&sas, i)) break; - addr = kmemdup(&sas, sizeof(*addr), GFP_NOFS); - if (!addr) - break; - dlm_local_addr[dlm_local_count++] = addr; + memcpy(&dlm_local_addr[dlm_local_count++], &sas, sizeof(sas)); } } -static void deinit_local(void) -{ - int i; - - for (i = 0; i < dlm_local_count; i++) - kfree(dlm_local_addr[i]); -} - static struct writequeue_entry *new_writequeue_entry(struct connection *con) { struct writequeue_entry *entry; @@ -1240,7 +1176,7 @@ static struct writequeue_entry *new_wq_entry(struct connection *con, int len, { struct writequeue_entry *e; - spin_lock(&con->writequeue_lock); + spin_lock_bh(&con->writequeue_lock); if (!list_empty(&con->writequeue)) { e = list_last_entry(&con->writequeue, struct writequeue_entry, list); if (DLM_WQ_REMAIN_BYTES(e) >= len) { @@ -1263,14 +1199,13 @@ static struct writequeue_entry *new_wq_entry(struct connection *con, int len, kref_get(&e->ref); *ppc = page_address(e->page); e->end += len; - atomic_inc(&con->writequeue_cnt); if (cb) cb(data); list_add_tail(&e->list, &con->writequeue); out: - spin_unlock(&con->writequeue_lock); + spin_unlock_bh(&con->writequeue_lock); return e; }; @@ -1319,13 +1254,13 @@ struct dlm_msg *dlm_lowcomms_new_msg(int nodeid, int len, gfp_t allocation, len < sizeof(struct dlm_header)) { BUILD_BUG_ON(PAGE_SIZE < DLM_MAX_SOCKET_BUFSIZE); log_print("failed to allocate a buffer of size %d", len); - WARN_ON(1); + WARN_ON_ONCE(1); return NULL; } idx = srcu_read_lock(&connections_srcu); - con = nodeid2con(nodeid, allocation); - if (!con) { + con = nodeid2con(nodeid, 0); + if (WARN_ON_ONCE(!con)) { srcu_read_unlock(&connections_srcu, idx); return NULL; } @@ -1350,7 +1285,7 @@ static void _dlm_lowcomms_commit_msg(struct dlm_msg *msg) struct connection *con = e->con; int users; - spin_lock(&con->writequeue_lock); + spin_lock_bh(&con->writequeue_lock); kref_get(&msg->ref); list_add(&msg->list, &e->msgs); @@ -1359,13 +1294,11 @@ static void _dlm_lowcomms_commit_msg(struct dlm_msg *msg) goto out; e->len = DLM_WQ_LENGTH_BYTES(e); - spin_unlock(&con->writequeue_lock); - queue_work(send_workqueue, &con->swork); - return; + lowcomms_queue_swork(con); out: - spin_unlock(&con->writequeue_lock); + spin_unlock_bh(&con->writequeue_lock); return; } @@ -1387,7 +1320,7 @@ void dlm_lowcomms_put_msg(struct dlm_msg *msg) kref_put(&msg->ref, dlm_msg_release); } -/* does not held connections_srcu, usage workqueue only */ +/* does not held connections_srcu, usage lowcomms_error_report only */ int dlm_lowcomms_resend_msg(struct dlm_msg *msg) { struct dlm_msg *msg_resend; @@ -1413,90 +1346,79 @@ int dlm_lowcomms_resend_msg(struct dlm_msg *msg) } /* Send a message */ -static void send_to_sock(struct connection *con) +static int send_to_sock(struct connection *con) { const int msg_flags = MSG_DONTWAIT | MSG_NOSIGNAL; struct writequeue_entry *e; int len, offset, ret; - int count = 0; - mutex_lock(&con->sock_mutex); - if (con->sock == NULL) - goto out_connect; - - spin_lock(&con->writequeue_lock); - for (;;) { - e = con_next_wq(con); - if (!e) - break; + spin_lock_bh(&con->writequeue_lock); + e = con_next_wq(con); + if (!e) { + clear_bit(CF_SEND_PENDING, &con->flags); + spin_unlock_bh(&con->writequeue_lock); + return DLM_IO_END; + } - len = e->len; - offset = e->offset; - BUG_ON(len == 0 && e->users == 0); - spin_unlock(&con->writequeue_lock); - - ret = kernel_sendpage(con->sock, e->page, offset, len, - msg_flags); - trace_dlm_send(con->nodeid, ret); - if (ret == -EAGAIN || ret == 0) { - if (ret == -EAGAIN && - test_bit(SOCKWQ_ASYNC_NOSPACE, &con->sock->flags) && - !test_and_set_bit(CF_APP_LIMITED, &con->flags)) { - /* Notify TCP that we're limited by the - * application window size. - */ - set_bit(SOCK_NOSPACE, &con->sock->flags); - con->sock->sk->sk_write_pending++; - } - cond_resched(); - goto out; - } else if (ret < 0) - goto out; + len = e->len; + offset = e->offset; + WARN_ON_ONCE(len == 0 && e->users == 0); + spin_unlock_bh(&con->writequeue_lock); - /* Don't starve people filling buffers */ - if (++count >= MAX_SEND_MSG_COUNT) { - cond_resched(); - count = 0; + ret = kernel_sendpage(con->sock, e->page, offset, len, + msg_flags); + trace_dlm_send(con->nodeid, ret); + if (ret == -EAGAIN || ret == 0) { + lock_sock(con->sock->sk); + spin_lock_bh(&con->writequeue_lock); + if (test_bit(SOCKWQ_ASYNC_NOSPACE, &con->sock->flags) && + !test_and_set_bit(CF_APP_LIMITED, &con->flags)) { + /* Notify TCP that we're limited by the + * application window size. + */ + set_bit(SOCK_NOSPACE, &con->sock->sk->sk_socket->flags); + con->sock->sk->sk_write_pending++; + + clear_bit(CF_SEND_PENDING, &con->flags); + spin_unlock_bh(&con->writequeue_lock); + release_sock(con->sock->sk); + + /* wait for write_space() event */ + return DLM_IO_END; } + spin_unlock_bh(&con->writequeue_lock); + release_sock(con->sock->sk); - spin_lock(&con->writequeue_lock); - writequeue_entry_complete(e, ret); - } - spin_unlock(&con->writequeue_lock); - - /* close if we got EOF */ - if (test_and_clear_bit(CF_EOF, &con->flags)) { - mutex_unlock(&con->sock_mutex); - close_connection(con, false, false, true); - - /* handling for tcp shutdown */ - clear_bit(CF_SHUTDOWN, &con->flags); - wake_up(&con->shutdown_wait); - } else { - mutex_unlock(&con->sock_mutex); + return DLM_IO_RESCHED; + } else if (ret < 0) { + return ret; } - return; - -out: - mutex_unlock(&con->sock_mutex); - return; + spin_lock_bh(&con->writequeue_lock); + writequeue_entry_complete(e, ret); + spin_unlock_bh(&con->writequeue_lock); -out_connect: - mutex_unlock(&con->sock_mutex); - queue_work(send_workqueue, &con->swork); - cond_resched(); + return DLM_IO_SUCCESS; } static void clean_one_writequeue(struct connection *con) { struct writequeue_entry *e, *safe; - spin_lock(&con->writequeue_lock); + spin_lock_bh(&con->writequeue_lock); list_for_each_entry_safe(e, safe, &con->writequeue, list) { free_entry(e); } - spin_unlock(&con->writequeue_lock); + spin_unlock_bh(&con->writequeue_lock); +} + +static void connection_release(struct rcu_head *rcu) +{ + struct connection *con = container_of(rcu, struct connection, rcu); + + WARN_ON_ONCE(!list_empty(&con->writequeue)); + WARN_ON_ONCE(con->sock); + kfree(con); } /* Called from recovery when it knows that a node has @@ -1504,286 +1426,311 @@ static void clean_one_writequeue(struct connection *con) int dlm_lowcomms_close(int nodeid) { struct connection *con; - struct dlm_node_addr *na; int idx; log_print("closing connection to node %d", nodeid); + idx = srcu_read_lock(&connections_srcu); con = nodeid2con(nodeid, 0); - if (con) { - set_bit(CF_CLOSE, &con->flags); - close_connection(con, true, true, true); - clean_one_writequeue(con); + if (WARN_ON_ONCE(!con)) { + srcu_read_unlock(&connections_srcu, idx); + return -ENOENT; + } + + stop_connection_io(con); + log_print("io handling for node: %d stopped", nodeid); + close_connection(con, true); + + spin_lock(&connections_lock); + hlist_del_rcu(&con->list); + spin_unlock(&connections_lock); + + clean_one_writequeue(con); + call_srcu(&connections_srcu, &con->rcu, connection_release); + if (con->othercon) { + clean_one_writequeue(con->othercon); if (con->othercon) - clean_one_writequeue(con->othercon); + call_srcu(&connections_srcu, &con->othercon->rcu, connection_release); } srcu_read_unlock(&connections_srcu, idx); - spin_lock(&dlm_node_addrs_spin); - na = find_node_addr(nodeid); - if (na) { - list_del(&na->list); - while (na->addr_count--) - kfree(na->addr[na->addr_count]); - kfree(na); - } - spin_unlock(&dlm_node_addrs_spin); + /* for debugging we print when we are done to compare with other + * messages in between. This function need to be correctly synchronized + * with io handling + */ + log_print("closing connection to node %d done", nodeid); return 0; } -/* Receive workqueue function */ +/* Receive worker function */ static void process_recv_sockets(struct work_struct *work) { struct connection *con = container_of(work, struct connection, rwork); + int ret, buflen; + + down_read(&con->sock_lock); + if (!con->sock) { + up_read(&con->sock_lock); + return; + } + + buflen = READ_ONCE(dlm_config.ci_buffer_size); + do { + ret = receive_from_sock(con, buflen); + } while (ret == DLM_IO_SUCCESS); + up_read(&con->sock_lock); - clear_bit(CF_READ_PENDING, &con->flags); - receive_from_sock(con); + switch (ret) { + case DLM_IO_END: + /* CF_RECV_PENDING cleared */ + break; + case DLM_IO_EOF: + close_connection(con, false); + /* CF_RECV_PENDING cleared */ + break; + case DLM_IO_RESCHED: + cond_resched(); + queue_work(io_workqueue, &con->rwork); + /* CF_RECV_PENDING not cleared */ + break; + default: + if (ret < 0) { + if (test_bit(CF_IS_OTHERCON, &con->flags)) { + close_connection(con, false); + } else { + spin_lock_bh(&con->writequeue_lock); + lowcomms_queue_swork(con); + spin_unlock_bh(&con->writequeue_lock); + } + + /* CF_RECV_PENDING cleared for othercon + * we trigger send queue if not already done + * and process_send_sockets will handle it + */ + break; + } + + WARN_ON_ONCE(1); + break; + } } static void process_listen_recv_socket(struct work_struct *work) { - accept_from_sock(&listen_con); + int ret; + + if (WARN_ON_ONCE(!listen_con.sock)) + return; + + do { + ret = accept_from_sock(); + } while (ret == DLM_IO_SUCCESS); + + if (ret < 0) + log_print("critical error accepting connection: %d", ret); } -static void dlm_connect(struct connection *con) +static int dlm_connect(struct connection *con) { struct sockaddr_storage addr; int result, addr_len; struct socket *sock; unsigned int mark; - /* Some odd races can cause double-connects, ignore them */ - if (con->retries++ > MAX_CONNECT_RETRIES) - return; - - if (con->sock) { - log_print("node %d already connected.", con->nodeid); - return; - } - memset(&addr, 0, sizeof(addr)); result = nodeid_to_addr(con->nodeid, &addr, NULL, dlm_proto_ops->try_new_addr, &mark); if (result < 0) { log_print("no address for nodeid %d", con->nodeid); - return; + return result; } /* Create a socket to communicate with */ - result = sock_create_kern(&init_net, dlm_local_addr[0]->ss_family, + result = sock_create_kern(&init_net, dlm_local_addr[0].ss_family, SOCK_STREAM, dlm_proto_ops->proto, &sock); if (result < 0) - goto socket_err; + return result; sock_set_mark(sock->sk, mark); dlm_proto_ops->sockopts(sock); - add_sock(sock, con); - result = dlm_proto_ops->bind(sock); - if (result < 0) - goto add_sock_err; + if (result < 0) { + sock_release(sock); + return result; + } + + add_sock(sock, con); log_print_ratelimited("connecting to %d", con->nodeid); make_sockaddr(&addr, dlm_config.ci_tcp_port, &addr_len); result = dlm_proto_ops->connect(con, sock, (struct sockaddr *)&addr, addr_len); - if (result < 0) - goto add_sock_err; - - return; - -add_sock_err: - dlm_close_sock(&con->sock); + switch (result) { + case -EINPROGRESS: + /* not an error */ + fallthrough; + case 0: + break; + default: + if (result < 0) + dlm_close_sock(&con->sock); -socket_err: - /* - * Some errors are fatal and this list might need adjusting. For other - * errors we try again until the max number of retries is reached. - */ - if (result != -EHOSTUNREACH && - result != -ENETUNREACH && - result != -ENETDOWN && - result != -EINVAL && - result != -EPROTONOSUPPORT) { - log_print("connect %d try %d error %d", con->nodeid, - con->retries, result); - msleep(1000); - lowcomms_connect_sock(con); + break; } + + return result; } -/* Send workqueue function */ +/* Send worker function */ static void process_send_sockets(struct work_struct *work) { struct connection *con = container_of(work, struct connection, swork); + int ret; - WARN_ON(test_bit(CF_IS_OTHERCON, &con->flags)); - - clear_bit(CF_WRITE_PENDING, &con->flags); + WARN_ON_ONCE(test_bit(CF_IS_OTHERCON, &con->flags)); - if (test_and_clear_bit(CF_RECONNECT, &con->flags)) { - close_connection(con, false, false, true); - dlm_midcomms_unack_msg_resend(con->nodeid); + down_read(&con->sock_lock); + if (!con->sock) { + up_read(&con->sock_lock); + down_write(&con->sock_lock); + if (!con->sock) { + ret = dlm_connect(con); + switch (ret) { + case 0: + break; + case -EINPROGRESS: + /* avoid spamming resched on connection + * we might can switch to a state_change + * event based mechanism if established + */ + msleep(100); + break; + default: + /* CF_SEND_PENDING not cleared */ + up_write(&con->sock_lock); + log_print("connect to node %d try %d error %d", + con->nodeid, con->retries++, ret); + msleep(1000); + /* For now we try forever to reconnect. In + * future we should send a event to cluster + * manager to fence itself after certain amount + * of retries. + */ + queue_work(io_workqueue, &con->swork); + return; + } + } + downgrade_write(&con->sock_lock); } - if (con->sock == NULL) { - if (test_and_clear_bit(CF_DELAY_CONNECT, &con->flags)) - msleep(1000); + do { + ret = send_to_sock(con); + } while (ret == DLM_IO_SUCCESS); + up_read(&con->sock_lock); - mutex_lock(&con->sock_mutex); - dlm_connect(con); - mutex_unlock(&con->sock_mutex); - } + switch (ret) { + case DLM_IO_END: + /* CF_SEND_PENDING cleared */ + break; + case DLM_IO_RESCHED: + /* CF_SEND_PENDING not cleared */ + cond_resched(); + queue_work(io_workqueue, &con->swork); + break; + default: + if (ret < 0) { + close_connection(con, false); + + /* CF_SEND_PENDING cleared */ + spin_lock_bh(&con->writequeue_lock); + lowcomms_queue_swork(con); + spin_unlock_bh(&con->writequeue_lock); + break; + } - if (!list_empty(&con->writequeue)) - send_to_sock(con); + WARN_ON_ONCE(1); + break; + } } static void work_stop(void) { - if (recv_workqueue) { - destroy_workqueue(recv_workqueue); - recv_workqueue = NULL; + if (io_workqueue) { + destroy_workqueue(io_workqueue); + io_workqueue = NULL; } - if (send_workqueue) { - destroy_workqueue(send_workqueue); - send_workqueue = NULL; + if (process_workqueue) { + destroy_workqueue(process_workqueue); + process_workqueue = NULL; } } static int work_start(void) { - recv_workqueue = alloc_ordered_workqueue("dlm_recv", WQ_MEM_RECLAIM); - if (!recv_workqueue) { - log_print("can't start dlm_recv"); + io_workqueue = alloc_workqueue("dlm_io", WQ_HIGHPRI | WQ_MEM_RECLAIM, + 0); + if (!io_workqueue) { + log_print("can't start dlm_io"); return -ENOMEM; } - send_workqueue = alloc_ordered_workqueue("dlm_send", WQ_MEM_RECLAIM); - if (!send_workqueue) { - log_print("can't start dlm_send"); - destroy_workqueue(recv_workqueue); - recv_workqueue = NULL; + /* ordered dlm message process queue, + * should be converted to a tasklet + */ + process_workqueue = alloc_ordered_workqueue("dlm_process", + WQ_HIGHPRI | WQ_MEM_RECLAIM); + if (!process_workqueue) { + log_print("can't start dlm_process"); + destroy_workqueue(io_workqueue); + io_workqueue = NULL; return -ENOMEM; } return 0; } -static void shutdown_conn(struct connection *con) -{ - if (dlm_proto_ops->shutdown_action) - dlm_proto_ops->shutdown_action(con); -} - void dlm_lowcomms_shutdown(void) { - int idx; - - /* Set all the flags to prevent any - * socket activity. - */ - dlm_allow_conn = 0; - - if (recv_workqueue) - flush_workqueue(recv_workqueue); - if (send_workqueue) - flush_workqueue(send_workqueue); + /* stop lowcomms_listen_data_ready calls */ + lock_sock(listen_con.sock->sk); + listen_con.sock->sk->sk_data_ready = listen_sock.sk_data_ready; + release_sock(listen_con.sock->sk); + cancel_work_sync(&listen_con.rwork); dlm_close_sock(&listen_con.sock); - idx = srcu_read_lock(&connections_srcu); - foreach_conn(shutdown_conn); - srcu_read_unlock(&connections_srcu, idx); -} - -static void _stop_conn(struct connection *con, bool and_other) -{ - mutex_lock(&con->sock_mutex); - set_bit(CF_CLOSE, &con->flags); - set_bit(CF_READ_PENDING, &con->flags); - set_bit(CF_WRITE_PENDING, &con->flags); - if (con->sock && con->sock->sk) { - lock_sock(con->sock->sk); - con->sock->sk->sk_user_data = NULL; - release_sock(con->sock->sk); - } - if (con->othercon && and_other) - _stop_conn(con->othercon, false); - mutex_unlock(&con->sock_mutex); -} - -static void stop_conn(struct connection *con) -{ - _stop_conn(con, true); + flush_workqueue(process_workqueue); } -static void connection_release(struct rcu_head *rcu) +void dlm_lowcomms_shutdown_node(int nodeid, bool force) { - struct connection *con = container_of(rcu, struct connection, rcu); - - kfree(con->rx_buf); - kfree(con); -} + struct connection *con; + int idx; -static void free_conn(struct connection *con) -{ - close_connection(con, true, true, true); - spin_lock(&connections_lock); - hlist_del_rcu(&con->list); - spin_unlock(&connections_lock); - if (con->othercon) { - clean_one_writequeue(con->othercon); - call_srcu(&connections_srcu, &con->othercon->rcu, - connection_release); + idx = srcu_read_lock(&connections_srcu); + con = nodeid2con(nodeid, 0); + if (WARN_ON_ONCE(!con)) { + srcu_read_unlock(&connections_srcu, idx); + return; } - clean_one_writequeue(con); - call_srcu(&connections_srcu, &con->rcu, connection_release); -} -static void work_flush(void) -{ - int ok; - int i; - struct connection *con; - - do { - ok = 1; - foreach_conn(stop_conn); - if (recv_workqueue) - flush_workqueue(recv_workqueue); - if (send_workqueue) - flush_workqueue(send_workqueue); - for (i = 0; i < CONN_HASH_SIZE && ok; i++) { - hlist_for_each_entry_rcu(con, &connection_hash[i], - list) { - ok &= test_bit(CF_READ_PENDING, &con->flags); - ok &= test_bit(CF_WRITE_PENDING, &con->flags); - if (con->othercon) { - ok &= test_bit(CF_READ_PENDING, - &con->othercon->flags); - ok &= test_bit(CF_WRITE_PENDING, - &con->othercon->flags); - } - } - } - } while (!ok); + flush_work(&con->swork); + stop_connection_io(con); + WARN_ON_ONCE(!force && !list_empty(&con->writequeue)); + close_connection(con, true); + clean_one_writequeue(con); + if (con->othercon) + clean_one_writequeue(con->othercon); + allow_connection_io(con); + srcu_read_unlock(&connections_srcu, idx); } void dlm_lowcomms_stop(void) { - int idx; - - idx = srcu_read_lock(&connections_srcu); - work_flush(); - foreach_conn(free_conn); - srcu_read_unlock(&connections_srcu, idx); work_stop(); - deinit_local(); - dlm_proto_ops = NULL; } @@ -1799,7 +1746,7 @@ static int dlm_listen_for_all(void) if (result < 0) return result; - result = sock_create_kern(&init_net, dlm_local_addr[0]->ss_family, + result = sock_create_kern(&init_net, dlm_local_addr[0].ss_family, SOCK_STREAM, dlm_proto_ops->proto, &sock); if (result < 0) { log_print("Can't create comms socket: %d", result); @@ -1813,14 +1760,22 @@ static int dlm_listen_for_all(void) if (result < 0) goto out; - save_listen_callbacks(sock); - add_listen_sock(sock, &listen_con); + lock_sock(sock->sk); + listen_sock.sk_data_ready = sock->sk->sk_data_ready; + listen_sock.sk_write_space = sock->sk->sk_write_space; + listen_sock.sk_error_report = sock->sk->sk_error_report; + listen_sock.sk_state_change = sock->sk->sk_state_change; + + listen_con.sock = sock; + + sock->sk->sk_allocation = GFP_NOFS; + sock->sk->sk_data_ready = lowcomms_listen_data_ready; + release_sock(sock->sk); - INIT_WORK(&listen_con.rwork, process_listen_recv_socket); result = sock->ops->listen(sock, 5); if (result < 0) { dlm_close_sock(&listen_con.sock); - goto out; + return result; } return 0; @@ -1838,7 +1793,7 @@ static int dlm_tcp_bind(struct socket *sock) /* Bind to our cluster-known address connecting to avoid * routing problems. */ - memcpy(&src_addr, dlm_local_addr[0], sizeof(src_addr)); + memcpy(&src_addr, &dlm_local_addr[0], sizeof(src_addr)); make_sockaddr(&src_addr, 0, &addr_len); result = sock->ops->bind(sock, (struct sockaddr *)&src_addr, @@ -1854,17 +1809,7 @@ static int dlm_tcp_bind(struct socket *sock) static int dlm_tcp_connect(struct connection *con, struct socket *sock, struct sockaddr *addr, int addr_len) { - int ret; - - ret = sock->ops->connect(sock, addr, addr_len, O_NONBLOCK); - switch (ret) { - case -EINPROGRESS: - fallthrough; - case 0: - return 0; - } - - return ret; + return sock->ops->connect(sock, addr, addr_len, O_NONBLOCK); } static int dlm_tcp_listen_validate(void) @@ -1895,8 +1840,8 @@ static int dlm_tcp_listen_bind(struct socket *sock) int addr_len; /* Bind to our port */ - make_sockaddr(dlm_local_addr[0], dlm_config.ci_tcp_port, &addr_len); - return sock->ops->bind(sock, (struct sockaddr *)dlm_local_addr[0], + make_sockaddr(&dlm_local_addr[0], dlm_config.ci_tcp_port, &addr_len); + return sock->ops->bind(sock, (struct sockaddr *)&dlm_local_addr[0], addr_len); } @@ -1909,8 +1854,6 @@ static const struct dlm_proto_ops dlm_tcp_ops = { .listen_validate = dlm_tcp_listen_validate, .listen_sockopts = dlm_tcp_listen_sockopts, .listen_bind = dlm_tcp_listen_bind, - .shutdown_action = dlm_tcp_shutdown, - .eof_condition = tcp_eof_condition, }; static int dlm_sctp_bind(struct socket *sock) @@ -1931,13 +1874,7 @@ static int dlm_sctp_connect(struct connection *con, struct socket *sock, sock_set_sndtimeo(sock->sk, 5); ret = sock->ops->connect(sock, addr, addr_len, 0); sock_set_sndtimeo(sock->sk, 0); - if (ret < 0) - return ret; - - if (!test_and_set_bit(CF_CONNECTED, &con->flags)) - log_print("connected to node %d", con->nodeid); - - return 0; + return ret; } static int dlm_sctp_listen_validate(void) @@ -1977,11 +1914,7 @@ static const struct dlm_proto_ops dlm_sctp_ops = { int dlm_lowcomms_start(void) { - int error = -EINVAL; - int i; - - for (i = 0; i < CONN_HASH_SIZE; i++) - INIT_HLIST_HEAD(&connection_hash[i]); + int error; init_local(); if (!dlm_local_count) { @@ -1990,13 +1923,9 @@ int dlm_lowcomms_start(void) goto fail; } - INIT_WORK(&listen_con.rwork, process_listen_recv_socket); - error = work_start(); if (error) - goto fail_local; - - dlm_allow_conn = 1; + goto fail; /* Start listening */ switch (dlm_config.ci_protocol) { @@ -2022,25 +1951,38 @@ int dlm_lowcomms_start(void) fail_listen: dlm_proto_ops = NULL; fail_proto_ops: - dlm_allow_conn = 0; - dlm_close_sock(&listen_con.sock); work_stop(); -fail_local: - deinit_local(); fail: return error; } +void dlm_lowcomms_init(void) +{ + int i; + + for (i = 0; i < CONN_HASH_SIZE; i++) + INIT_HLIST_HEAD(&connection_hash[i]); + + INIT_WORK(&listen_con.rwork, process_listen_recv_socket); +} + void dlm_lowcomms_exit(void) { - struct dlm_node_addr *na, *safe; + struct connection *con; + int i, idx; - spin_lock(&dlm_node_addrs_spin); - list_for_each_entry_safe(na, safe, &dlm_node_addrs, list) { - list_del(&na->list); - while (na->addr_count--) - kfree(na->addr[na->addr_count]); - kfree(na); + idx = srcu_read_lock(&connections_srcu); + for (i = 0; i < CONN_HASH_SIZE; i++) { + hlist_for_each_entry_rcu(con, &connection_hash[i], list) { + spin_lock(&connections_lock); + hlist_del_rcu(&con->list); + spin_unlock(&connections_lock); + + if (con->othercon) + call_srcu(&connections_srcu, &con->othercon->rcu, + connection_release); + call_srcu(&connections_srcu, &con->rcu, connection_release); + } } - spin_unlock(&dlm_node_addrs_spin); + srcu_read_unlock(&connections_srcu, idx); } diff --git a/fs/dlm/lowcomms.h b/fs/dlm/lowcomms.h index 29369feea991..3e8dca66183b 100644 --- a/fs/dlm/lowcomms.h +++ b/fs/dlm/lowcomms.h @@ -29,12 +29,14 @@ static inline int nodeid_hash(int nodeid) return nodeid & (CONN_HASH_SIZE-1); } -/* switch to check if dlm is running */ -extern int dlm_allow_conn; +/* check if dlm is running */ +bool dlm_lowcomms_is_running(void); int dlm_lowcomms_start(void); void dlm_lowcomms_shutdown(void); +void dlm_lowcomms_shutdown_node(int nodeid, bool force); void dlm_lowcomms_stop(void); +void dlm_lowcomms_init(void); void dlm_lowcomms_exit(void); int dlm_lowcomms_close(int nodeid); struct dlm_msg *dlm_lowcomms_new_msg(int nodeid, int len, gfp_t allocation, diff --git a/fs/dlm/main.c b/fs/dlm/main.c index 1c5be4b70ac1..a77338be3237 100644 --- a/fs/dlm/main.c +++ b/fs/dlm/main.c @@ -17,7 +17,7 @@ #include "user.h" #include "memory.h" #include "config.h" -#include "lowcomms.h" +#include "midcomms.h" #define CREATE_TRACE_POINTS #include <trace/events/dlm.h> @@ -30,6 +30,8 @@ static int __init init_dlm(void) if (error) goto out; + dlm_midcomms_init(); + error = dlm_lockspace_init(); if (error) goto out_mem; @@ -66,6 +68,7 @@ static int __init init_dlm(void) out_lockspace: dlm_lockspace_exit(); out_mem: + dlm_midcomms_exit(); dlm_memory_exit(); out: return error; @@ -79,7 +82,7 @@ static void __exit exit_dlm(void) dlm_config_exit(); dlm_memory_exit(); dlm_lockspace_exit(); - dlm_lowcomms_exit(); + dlm_midcomms_exit(); dlm_unregister_debugfs(); } diff --git a/fs/dlm/member.c b/fs/dlm/member.c index 2af2ccfe43a9..923c01a8a0aa 100644 --- a/fs/dlm/member.c +++ b/fs/dlm/member.c @@ -573,7 +573,10 @@ int dlm_recover_members(struct dlm_ls *ls, struct dlm_recover *rv, int *neg_out) node = &rv->nodes[i]; if (dlm_is_member(ls, node->nodeid)) continue; - dlm_add_member(ls, node); + error = dlm_add_member(ls, node); + if (error) + return error; + log_rinfo(ls, "add member %d", node->nodeid); } diff --git a/fs/dlm/memory.c b/fs/dlm/memory.c index ce35c3c19aeb..eb7a08641fcf 100644 --- a/fs/dlm/memory.c +++ b/fs/dlm/memory.c @@ -14,12 +14,14 @@ #include "lowcomms.h" #include "config.h" #include "memory.h" +#include "ast.h" static struct kmem_cache *writequeue_cache; static struct kmem_cache *mhandle_cache; static struct kmem_cache *msg_cache; static struct kmem_cache *lkb_cache; static struct kmem_cache *rsb_cache; +static struct kmem_cache *cb_cache; int __init dlm_memory_init(void) @@ -46,8 +48,16 @@ int __init dlm_memory_init(void) if (!rsb_cache) goto rsb; + cb_cache = kmem_cache_create("dlm_cb", sizeof(struct dlm_callback), + __alignof__(struct dlm_callback), 0, + NULL); + if (!rsb_cache) + goto cb; + return 0; +cb: + kmem_cache_destroy(rsb_cache); rsb: kmem_cache_destroy(msg_cache); msg: @@ -67,6 +77,7 @@ void dlm_memory_exit(void) kmem_cache_destroy(msg_cache); kmem_cache_destroy(lkb_cache); kmem_cache_destroy(rsb_cache); + kmem_cache_destroy(cb_cache); } char *dlm_allocate_lvb(struct dlm_ls *ls) @@ -115,12 +126,17 @@ void dlm_free_lkb(struct dlm_lkb *lkb) kfree(ua); } } + + /* drop references if they are set */ + dlm_callback_set_last_ptr(&lkb->lkb_last_cast, NULL); + dlm_callback_set_last_ptr(&lkb->lkb_last_cb, NULL); + kmem_cache_free(lkb_cache, lkb); } -struct dlm_mhandle *dlm_allocate_mhandle(void) +struct dlm_mhandle *dlm_allocate_mhandle(gfp_t allocation) { - return kmem_cache_alloc(mhandle_cache, GFP_NOFS); + return kmem_cache_alloc(mhandle_cache, allocation); } void dlm_free_mhandle(struct dlm_mhandle *mhandle) @@ -147,3 +163,13 @@ void dlm_free_msg(struct dlm_msg *msg) { kmem_cache_free(msg_cache, msg); } + +struct dlm_callback *dlm_allocate_cb(void) +{ + return kmem_cache_alloc(cb_cache, GFP_ATOMIC); +} + +void dlm_free_cb(struct dlm_callback *cb) +{ + kmem_cache_free(cb_cache, cb); +} diff --git a/fs/dlm/memory.h b/fs/dlm/memory.h index 7bd3f1a391ca..6b29563d24f7 100644 --- a/fs/dlm/memory.h +++ b/fs/dlm/memory.h @@ -20,12 +20,14 @@ struct dlm_lkb *dlm_allocate_lkb(struct dlm_ls *ls); void dlm_free_lkb(struct dlm_lkb *l); char *dlm_allocate_lvb(struct dlm_ls *ls); void dlm_free_lvb(char *l); -struct dlm_mhandle *dlm_allocate_mhandle(void); +struct dlm_mhandle *dlm_allocate_mhandle(gfp_t allocation); void dlm_free_mhandle(struct dlm_mhandle *mhandle); struct writequeue_entry *dlm_allocate_writequeue(void); void dlm_free_writequeue(struct writequeue_entry *writequeue); struct dlm_msg *dlm_allocate_msg(gfp_t allocation); void dlm_free_msg(struct dlm_msg *msg); +struct dlm_callback *dlm_allocate_cb(void); +void dlm_free_cb(struct dlm_callback *cb); #endif /* __MEMORY_DOT_H__ */ diff --git a/fs/dlm/midcomms.c b/fs/dlm/midcomms.c index 6489bc22ad61..fc015a6abe17 100644 --- a/fs/dlm/midcomms.c +++ b/fs/dlm/midcomms.c @@ -132,6 +132,7 @@ */ #define DLM_DEBUG_FENCE_TERMINATION 0 +#include <trace/events/dlm.h> #include <net/tcp.h> #include "dlm_internal.h" @@ -194,7 +195,7 @@ struct midcomms_node { }; struct dlm_mhandle { - const struct dlm_header *inner_hd; + const union dlm_packet *inner_p; struct midcomms_node *node; struct dlm_opts *opts; struct dlm_msg *msg; @@ -305,11 +306,11 @@ static void dlm_send_queue_flush(struct midcomms_node *node) pr_debug("flush midcomms send queue of node %d\n", node->nodeid); rcu_read_lock(); - spin_lock(&node->send_queue_lock); + spin_lock_bh(&node->send_queue_lock); list_for_each_entry_rcu(mh, &node->send_queue, list) { dlm_mhandle_delete(node, mh); } - spin_unlock(&node->send_queue_lock); + spin_unlock_bh(&node->send_queue_lock); rcu_read_unlock(); } @@ -415,7 +416,7 @@ static int dlm_send_fin(struct midcomms_node *node, m_header->h_cmd = DLM_FIN; pr_debug("sending fin msg to node %d\n", node->nodeid); - dlm_midcomms_commit_mhandle(mh); + dlm_midcomms_commit_mhandle(mh, NULL, 0); set_bit(DLM_NODE_FLAG_STOP_TX, &node->flags); return 0; @@ -436,7 +437,7 @@ static void dlm_receive_ack(struct midcomms_node *node, uint32_t seq) } } - spin_lock(&node->send_queue_lock); + spin_lock_bh(&node->send_queue_lock); list_for_each_entry_rcu(mh, &node->send_queue, list) { if (before(mh->seq, seq)) { dlm_mhandle_delete(node, mh); @@ -445,7 +446,7 @@ static void dlm_receive_ack(struct midcomms_node *node, uint32_t seq) break; } } - spin_unlock(&node->send_queue_lock); + spin_unlock_bh(&node->send_queue_lock); rcu_read_unlock(); } @@ -468,12 +469,26 @@ static void dlm_pas_fin_ack_rcv(struct midcomms_node *node) spin_unlock(&node->state_lock); log_print("%s: unexpected state: %d\n", __func__, node->state); - WARN_ON(1); + WARN_ON_ONCE(1); return; } spin_unlock(&node->state_lock); } +static void dlm_receive_buffer_3_2_trace(uint32_t seq, union dlm_packet *p) +{ + switch (p->header.h_cmd) { + case DLM_MSG: + trace_dlm_recv_message(dlm_our_nodeid(), seq, &p->message); + break; + case DLM_RCOM: + trace_dlm_recv_rcom(dlm_our_nodeid(), seq, &p->rcom); + break; + default: + break; + } +} + static void dlm_midcomms_receive_buffer(union dlm_packet *p, struct midcomms_node *node, uint32_t seq) @@ -525,7 +540,7 @@ static void dlm_midcomms_receive_buffer(union dlm_packet *p, spin_unlock(&node->state_lock); log_print("%s: unexpected state: %d\n", __func__, node->state); - WARN_ON(1); + WARN_ON_ONCE(1); return; } spin_unlock(&node->state_lock); @@ -533,7 +548,8 @@ static void dlm_midcomms_receive_buffer(union dlm_packet *p, set_bit(DLM_NODE_FLAG_STOP_RX, &node->flags); break; default: - WARN_ON(test_bit(DLM_NODE_FLAG_STOP_RX, &node->flags)); + WARN_ON_ONCE(test_bit(DLM_NODE_FLAG_STOP_RX, &node->flags)); + dlm_receive_buffer_3_2_trace(seq, p); dlm_receive_buffer(p, node->nodeid); set_bit(DLM_NODE_ULP_DELIVERED, &node->flags); break; @@ -754,7 +770,7 @@ static void dlm_midcomms_receive_buffer_3_2(union dlm_packet *p, int nodeid) goto out; } - WARN_ON(test_bit(DLM_NODE_FLAG_STOP_RX, &node->flags)); + WARN_ON_ONCE(test_bit(DLM_NODE_FLAG_STOP_RX, &node->flags)); dlm_receive_buffer(p, nodeid); break; case DLM_OPTS: @@ -874,12 +890,7 @@ static void dlm_midcomms_receive_buffer_3_1(union dlm_packet *p, int nodeid) dlm_receive_buffer(p, nodeid); } -/* - * Called from the low-level comms layer to process a buffer of - * commands. - */ - -int dlm_process_incoming_buffer(int nodeid, unsigned char *buf, int len) +int dlm_validate_incoming_buffer(int nodeid, unsigned char *buf, int len) { const unsigned char *ptr = buf; const struct dlm_header *hd; @@ -914,6 +925,32 @@ int dlm_process_incoming_buffer(int nodeid, unsigned char *buf, int len) if (msglen > len) break; + ret += msglen; + len -= msglen; + ptr += msglen; + } + + return ret; +} + +/* + * Called from the low-level comms layer to process a buffer of + * commands. + */ +int dlm_process_incoming_buffer(int nodeid, unsigned char *buf, int len) +{ + const unsigned char *ptr = buf; + const struct dlm_header *hd; + uint16_t msglen; + int ret = 0; + + while (len >= sizeof(struct dlm_header)) { + hd = (struct dlm_header *)ptr; + + msglen = le16_to_cpu(hd->h_length); + if (msglen > len) + break; + switch (hd->h_version) { case cpu_to_le32(DLM_VERSION_3_1): dlm_midcomms_receive_buffer_3_1((union dlm_packet *)ptr, nodeid); @@ -1030,9 +1067,9 @@ static void midcomms_new_msg_cb(void *data) atomic_inc(&mh->node->send_queue_cnt); - spin_lock(&mh->node->send_queue_lock); + spin_lock_bh(&mh->node->send_queue_lock); list_add_tail_rcu(&mh->list, &mh->node->send_queue); - spin_unlock(&mh->node->send_queue_lock); + spin_unlock_bh(&mh->node->send_queue_lock); mh->seq = mh->node->seq_send++; } @@ -1055,7 +1092,7 @@ static struct dlm_msg *dlm_midcomms_get_msg_3_2(struct dlm_mhandle *mh, int node dlm_fill_opts_header(opts, len, mh->seq); *ppc += sizeof(*opts); - mh->inner_hd = (const struct dlm_header *)*ppc; + mh->inner_p = (const union dlm_packet *)*ppc; return msg; } @@ -1079,9 +1116,9 @@ struct dlm_mhandle *dlm_midcomms_get_mhandle(int nodeid, int len, } /* this is a bug, however we going on and hope it will be resolved */ - WARN_ON(test_bit(DLM_NODE_FLAG_STOP_TX, &node->flags)); + WARN_ON_ONCE(test_bit(DLM_NODE_FLAG_STOP_TX, &node->flags)); - mh = dlm_allocate_mhandle(); + mh = dlm_allocate_mhandle(allocation); if (!mh) goto err; @@ -1111,7 +1148,7 @@ struct dlm_mhandle *dlm_midcomms_get_mhandle(int nodeid, int len, break; default: dlm_free_mhandle(mh); - WARN_ON(1); + WARN_ON_ONCE(1); goto err; } @@ -1130,11 +1167,32 @@ err: } #endif -static void dlm_midcomms_commit_msg_3_2(struct dlm_mhandle *mh) +static void dlm_midcomms_commit_msg_3_2_trace(const struct dlm_mhandle *mh, + const void *name, int namelen) +{ + switch (mh->inner_p->header.h_cmd) { + case DLM_MSG: + trace_dlm_send_message(mh->node->nodeid, mh->seq, + &mh->inner_p->message, + name, namelen); + break; + case DLM_RCOM: + trace_dlm_send_rcom(mh->node->nodeid, mh->seq, + &mh->inner_p->rcom); + break; + default: + /* nothing to trace */ + break; + } +} + +static void dlm_midcomms_commit_msg_3_2(struct dlm_mhandle *mh, + const void *name, int namelen) { /* nexthdr chain for fast lookup */ - mh->opts->o_nextcmd = mh->inner_hd->h_cmd; + mh->opts->o_nextcmd = mh->inner_p->header.h_cmd; mh->committed = true; + dlm_midcomms_commit_msg_3_2_trace(mh, name, namelen); dlm_lowcomms_commit_msg(mh->msg); } @@ -1142,8 +1200,10 @@ static void dlm_midcomms_commit_msg_3_2(struct dlm_mhandle *mh) * dlm_midcomms_get_mhandle */ #ifndef __CHECKER__ -void dlm_midcomms_commit_mhandle(struct dlm_mhandle *mh) +void dlm_midcomms_commit_mhandle(struct dlm_mhandle *mh, + const void *name, int namelen) { + switch (mh->node->version) { case DLM_VERSION_3_1: srcu_read_unlock(&nodes_srcu, mh->idx); @@ -1154,12 +1214,12 @@ void dlm_midcomms_commit_mhandle(struct dlm_mhandle *mh) dlm_free_mhandle(mh); break; case DLM_VERSION_3_2: - dlm_midcomms_commit_msg_3_2(mh); + dlm_midcomms_commit_msg_3_2(mh, name, namelen); srcu_read_unlock(&nodes_srcu, mh->idx); break; default: srcu_read_unlock(&nodes_srcu, mh->idx); - WARN_ON(1); + WARN_ON_ONCE(1); break; } } @@ -1167,12 +1227,27 @@ void dlm_midcomms_commit_mhandle(struct dlm_mhandle *mh) int dlm_midcomms_start(void) { + return dlm_lowcomms_start(); +} + +void dlm_midcomms_stop(void) +{ + dlm_lowcomms_stop(); +} + +void dlm_midcomms_init(void) +{ int i; for (i = 0; i < CONN_HASH_SIZE; i++) INIT_HLIST_HEAD(&node_hash[i]); - return dlm_lowcomms_start(); + dlm_lowcomms_init(); +} + +void dlm_midcomms_exit(void) +{ + dlm_lowcomms_exit(); } static void dlm_act_fin_ack_rcv(struct midcomms_node *node) @@ -1201,7 +1276,7 @@ static void dlm_act_fin_ack_rcv(struct midcomms_node *node) spin_unlock(&node->state_lock); log_print("%s: unexpected state: %d\n", __func__, node->state); - WARN_ON(1); + WARN_ON_ONCE(1); return; } spin_unlock(&node->state_lock); @@ -1319,7 +1394,7 @@ static void midcomms_node_release(struct rcu_head *rcu) { struct midcomms_node *node = container_of(rcu, struct midcomms_node, rcu); - WARN_ON(atomic_read(&node->send_queue_cnt)); + WARN_ON_ONCE(atomic_read(&node->send_queue_cnt)); kfree(node); } @@ -1372,11 +1447,13 @@ static void midcomms_shutdown(struct midcomms_node *node) pr_debug("active shutdown timed out for node %d with state %s\n", node->nodeid, dlm_state_str(node->state)); midcomms_node_reset(node); + dlm_lowcomms_shutdown_node(node->nodeid, true); return; } pr_debug("active shutdown done for node %d with state %s\n", node->nodeid, dlm_state_str(node->state)); + dlm_lowcomms_shutdown_node(node->nodeid, false); } void dlm_midcomms_shutdown(void) @@ -1384,6 +1461,8 @@ void dlm_midcomms_shutdown(void) struct midcomms_node *node; int i, idx; + dlm_lowcomms_shutdown(); + mutex_lock(&close_lock); idx = srcu_read_lock(&nodes_srcu); for (i = 0; i < CONN_HASH_SIZE; i++) { @@ -1401,8 +1480,6 @@ void dlm_midcomms_shutdown(void) } srcu_read_unlock(&nodes_srcu, idx); mutex_unlock(&close_lock); - - dlm_lowcomms_shutdown(); } int dlm_midcomms_close(int nodeid) diff --git a/fs/dlm/midcomms.h b/fs/dlm/midcomms.h index 82bcd9661922..bea1cee4279c 100644 --- a/fs/dlm/midcomms.h +++ b/fs/dlm/midcomms.h @@ -14,12 +14,17 @@ struct midcomms_node; +int dlm_validate_incoming_buffer(int nodeid, unsigned char *buf, int len); int dlm_process_incoming_buffer(int nodeid, unsigned char *buf, int buflen); struct dlm_mhandle *dlm_midcomms_get_mhandle(int nodeid, int len, gfp_t allocation, char **ppc); -void dlm_midcomms_commit_mhandle(struct dlm_mhandle *mh); +void dlm_midcomms_commit_mhandle(struct dlm_mhandle *mh, const void *name, + int namelen); int dlm_midcomms_close(int nodeid); int dlm_midcomms_start(void); +void dlm_midcomms_stop(void); +void dlm_midcomms_init(void); +void dlm_midcomms_exit(void); void dlm_midcomms_shutdown(void); void dlm_midcomms_add_member(int nodeid); void dlm_midcomms_remove_member(int nodeid); diff --git a/fs/dlm/rcom.c b/fs/dlm/rcom.c index f19860315043..b76d52e2f6bd 100644 --- a/fs/dlm/rcom.c +++ b/fs/dlm/rcom.c @@ -91,7 +91,7 @@ static int create_rcom_stateless(struct dlm_ls *ls, int to_nodeid, int type, static void send_rcom(struct dlm_mhandle *mh, struct dlm_rcom *rc) { - dlm_midcomms_commit_mhandle(mh); + dlm_midcomms_commit_mhandle(mh, NULL, 0); } static void send_rcom_stateless(struct dlm_msg *msg, struct dlm_rcom *rc) @@ -516,7 +516,7 @@ int dlm_send_ls_not_ready(int nodeid, struct dlm_rcom *rc_in) rf = (struct rcom_config *) rc->rc_buf; rf->rf_lvblen = cpu_to_le32(~0U); - dlm_midcomms_commit_mhandle(mh); + dlm_midcomms_commit_mhandle(mh, NULL, 0); return 0; } diff --git a/fs/dlm/requestqueue.c b/fs/dlm/requestqueue.c index 036a9a0078f6..8be2893ad15b 100644 --- a/fs/dlm/requestqueue.c +++ b/fs/dlm/requestqueue.c @@ -44,7 +44,8 @@ void dlm_add_requestqueue(struct dlm_ls *ls, int nodeid, struct dlm_message *ms) e->recover_seq = ls->ls_recover_seq & 0xFFFFFFFF; e->nodeid = nodeid; - memcpy(&e->request, ms, le16_to_cpu(ms->m_header.h_length)); + memcpy(&e->request, ms, sizeof(*ms)); + memcpy(&e->request.m_extra, ms->m_extra, length); atomic_inc(&ls->ls_requestqueue_cnt); mutex_lock(&ls->ls_requestqueue_mutex); diff --git a/fs/dlm/user.c b/fs/dlm/user.c index c5d27bccc3dc..35129505ddda 100644 --- a/fs/dlm/user.c +++ b/fs/dlm/user.c @@ -25,6 +25,7 @@ #include "user.h" #include "ast.h" #include "config.h" +#include "memory.h" static const char name_prefix[] = "dlm"; static const struct file_operations device_fops; @@ -175,7 +176,7 @@ static int lkb_is_endoflife(int mode, int status) being removed and then remove that lkb from the orphans list and free it */ void dlm_user_add_ast(struct dlm_lkb *lkb, uint32_t flags, int mode, - int status, uint32_t sbflags, uint64_t seq) + int status, uint32_t sbflags) { struct dlm_ls *ls; struct dlm_user_args *ua; @@ -209,16 +210,22 @@ void dlm_user_add_ast(struct dlm_lkb *lkb, uint32_t flags, int mode, spin_lock(&proc->asts_spin); - rv = dlm_add_lkb_callback(lkb, flags, mode, status, sbflags, seq); - if (rv < 0) { + rv = dlm_enqueue_lkb_callback(lkb, flags, mode, status, sbflags); + switch (rv) { + case DLM_ENQUEUE_CALLBACK_FAILURE: spin_unlock(&proc->asts_spin); + WARN_ON_ONCE(1); goto out; - } - - if (list_empty(&lkb->lkb_cb_list)) { + case DLM_ENQUEUE_CALLBACK_NEED_SCHED: kref_get(&lkb->lkb_ref); list_add_tail(&lkb->lkb_cb_list, &proc->asts); wake_up_interruptible(&proc->wait); + break; + case DLM_ENQUEUE_CALLBACK_SUCCESS: + break; + default: + WARN_ON_ONCE(1); + break; } spin_unlock(&proc->asts_spin); @@ -800,8 +807,8 @@ static ssize_t device_read(struct file *file, char __user *buf, size_t count, struct dlm_user_proc *proc = file->private_data; struct dlm_lkb *lkb; DECLARE_WAITQUEUE(wait, current); - struct dlm_callback cb; - int rv, resid, copy_lvb = 0; + struct dlm_callback *cb; + int rv, copy_lvb = 0; int old_mode, new_mode; if (count == sizeof(struct dlm_device_version)) { @@ -857,53 +864,58 @@ static ssize_t device_read(struct file *file, char __user *buf, size_t count, without removing lkb_cb_list; so empty lkb_cb_list is always consistent with empty lkb_callbacks */ - lkb = list_entry(proc->asts.next, struct dlm_lkb, lkb_cb_list); + lkb = list_first_entry(&proc->asts, struct dlm_lkb, lkb_cb_list); /* rem_lkb_callback sets a new lkb_last_cast */ - old_mode = lkb->lkb_last_cast.mode; + old_mode = lkb->lkb_last_cast->mode; - rv = dlm_rem_lkb_callback(lkb->lkb_resource->res_ls, lkb, &cb, &resid); - if (rv < 0) { + rv = dlm_dequeue_lkb_callback(lkb, &cb); + switch (rv) { + case DLM_DEQUEUE_CALLBACK_EMPTY: /* this shouldn't happen; lkb should have been removed from - list when resid was zero */ + * list when last item was dequeued + */ log_print("dlm_rem_lkb_callback empty %x", lkb->lkb_id); list_del_init(&lkb->lkb_cb_list); spin_unlock(&proc->asts_spin); /* removes ref for proc->asts, may cause lkb to be freed */ dlm_put_lkb(lkb); + WARN_ON_ONCE(1); goto try_another; - } - if (!resid) + case DLM_DEQUEUE_CALLBACK_LAST: list_del_init(&lkb->lkb_cb_list); - spin_unlock(&proc->asts_spin); - - if (cb.flags & DLM_CB_SKIP) { - /* removes ref for proc->asts, may cause lkb to be freed */ - if (!resid) - dlm_put_lkb(lkb); - goto try_another; + lkb->lkb_flags &= ~DLM_IFL_CB_PENDING; + break; + case DLM_DEQUEUE_CALLBACK_SUCCESS: + break; + default: + WARN_ON_ONCE(1); + break; } + spin_unlock(&proc->asts_spin); - if (cb.flags & DLM_CB_BAST) { - trace_dlm_bast(lkb->lkb_resource->res_ls, lkb, cb.mode); - } else if (cb.flags & DLM_CB_CAST) { - new_mode = cb.mode; + if (cb->flags & DLM_CB_BAST) { + trace_dlm_bast(lkb->lkb_resource->res_ls, lkb, cb->mode); + } else if (cb->flags & DLM_CB_CAST) { + new_mode = cb->mode; - if (!cb.sb_status && lkb->lkb_lksb->sb_lvbptr && + if (!cb->sb_status && lkb->lkb_lksb->sb_lvbptr && dlm_lvb_operations[old_mode + 1][new_mode + 1]) copy_lvb = 1; - lkb->lkb_lksb->sb_status = cb.sb_status; - lkb->lkb_lksb->sb_flags = cb.sb_flags; + lkb->lkb_lksb->sb_status = cb->sb_status; + lkb->lkb_lksb->sb_flags = cb->sb_flags; trace_dlm_ast(lkb->lkb_resource->res_ls, lkb); } rv = copy_result_to_user(lkb->lkb_ua, test_bit(DLM_PROC_FLAGS_COMPAT, &proc->flags), - cb.flags, cb.mode, copy_lvb, buf, count); + cb->flags, cb->mode, copy_lvb, buf, count); + + kref_put(&cb->ref, dlm_release_callback); /* removes ref for proc->asts, may cause lkb to be freed */ - if (!resid) + if (rv == DLM_DEQUEUE_CALLBACK_LAST) dlm_put_lkb(lkb); return rv; diff --git a/fs/dlm/user.h b/fs/dlm/user.h index 6b9bce6b96e0..33059452d79e 100644 --- a/fs/dlm/user.h +++ b/fs/dlm/user.h @@ -7,7 +7,7 @@ #define __USER_DOT_H__ void dlm_user_add_ast(struct dlm_lkb *lkb, uint32_t flags, int mode, - int status, uint32_t sbflags, uint64_t seq); + int status, uint32_t sbflags); int dlm_user_init(void); void dlm_user_exit(void); int dlm_device_deregister(struct dlm_ls *ls); |