[GFS2] Fix calculation of demote state
[cascardo/linux.git] / fs / gfs2 / glock.c
index 12accb0..6b6ae45 100644 (file)
 #include <linux/module.h>
 #include <linux/rwsem.h>
 #include <asm/uaccess.h>
+#include <linux/seq_file.h>
+#include <linux/debugfs.h>
+#include <linux/module.h>
+#include <linux/kallsyms.h>
 
 #include "gfs2.h"
 #include "incore.h"
@@ -40,20 +44,30 @@ struct gfs2_gl_hash_bucket {
         struct hlist_head hb_list;
 };
 
+struct glock_iter {
+       int hash;                     /* hash bucket index         */
+       struct gfs2_sbd *sdp;         /* incore superblock         */
+       struct gfs2_glock *gl;        /* current glock struct      */
+       struct hlist_head *hb_list;   /* current hash bucket ptr   */
+       struct seq_file *seq;         /* sequence file for debugfs */
+       char string[512];             /* scratch space             */
+};
+
 typedef void (*glock_examiner) (struct gfs2_glock * gl);
 
 static int gfs2_dump_lockstate(struct gfs2_sbd *sdp);
-static int dump_glock(struct gfs2_glock *gl);
-static int dump_inode(struct gfs2_inode *ip);
-static void gfs2_glock_xmote_th(struct gfs2_holder *gh);
+static int dump_glock(struct glock_iter *gi, struct gfs2_glock *gl);
+static void gfs2_glock_xmote_th(struct gfs2_glock *gl, struct gfs2_holder *gh);
 static void gfs2_glock_drop_th(struct gfs2_glock *gl);
 static DECLARE_RWSEM(gfs2_umount_flush_sem);
+static struct dentry *gfs2_root;
 
 #define GFS2_GL_HASH_SHIFT      15
 #define GFS2_GL_HASH_SIZE       (1 << GFS2_GL_HASH_SHIFT)
 #define GFS2_GL_HASH_MASK       (GFS2_GL_HASH_SIZE - 1)
 
 static struct gfs2_gl_hash_bucket gl_hash_table[GFS2_GL_HASH_SIZE];
+static struct dentry *gfs2_root;
 
 /*
  * Despite what you might think, the numbers below are not arbitrary :-)
@@ -202,7 +216,6 @@ int gfs2_glock_put(struct gfs2_glock *gl)
                gfs2_assert(sdp, list_empty(&gl->gl_reclaim));
                gfs2_assert(sdp, list_empty(&gl->gl_holders));
                gfs2_assert(sdp, list_empty(&gl->gl_waiters1));
-               gfs2_assert(sdp, list_empty(&gl->gl_waiters2));
                gfs2_assert(sdp, list_empty(&gl->gl_waiters3));
                glock_free(gl);
                rv = 1;
@@ -303,7 +316,7 @@ int gfs2_glock_get(struct gfs2_sbd *sdp, u64 number,
        atomic_set(&gl->gl_ref, 1);
        gl->gl_state = LM_ST_UNLOCKED;
        gl->gl_hash = hash;
-       gl->gl_owner = NULL;
+       gl->gl_owner_pid = 0;
        gl->gl_ip = 0;
        gl->gl_ops = glops;
        gl->gl_req_gh = NULL;
@@ -367,7 +380,7 @@ void gfs2_holder_init(struct gfs2_glock *gl, unsigned int state, unsigned flags,
        INIT_LIST_HEAD(&gh->gh_list);
        gh->gh_gl = gl;
        gh->gh_ip = (unsigned long)__builtin_return_address(0);
-       gh->gh_owner = current;
+       gh->gh_owner_pid = current->pid;
        gh->gh_state = state;
        gh->gh_flags = flags;
        gh->gh_error = 0;
@@ -389,7 +402,7 @@ void gfs2_holder_reinit(unsigned int state, unsigned flags, struct gfs2_holder *
 {
        gh->gh_state = state;
        gh->gh_flags = flags;
-       gh->gh_iflags &= 1 << HIF_ALLOCED;
+       gh->gh_iflags = 0;
        gh->gh_ip = (unsigned long)__builtin_return_address(0);
 }
 
@@ -406,69 +419,36 @@ void gfs2_holder_uninit(struct gfs2_holder *gh)
        gh->gh_ip = 0;
 }
 
-/**
- * gfs2_holder_get - get a struct gfs2_holder structure
- * @gl: the glock
- * @state: the state we're requesting
- * @flags: the modifier flags
- * @gfp_flags:
- *
- * Figure out how big an impact this function has.  Either:
- * 1) Replace it with a cache of structures hanging off the struct gfs2_sbd
- * 2) Leave it like it is
- *
- * Returns: the holder structure, NULL on ENOMEM
- */
-
-static struct gfs2_holder *gfs2_holder_get(struct gfs2_glock *gl,
-                                          unsigned int state,
-                                          int flags, gfp_t gfp_flags)
+static void gfs2_holder_wake(struct gfs2_holder *gh)
 {
-       struct gfs2_holder *gh;
-
-       gh = kmalloc(sizeof(struct gfs2_holder), gfp_flags);
-       if (!gh)
-               return NULL;
-
-       gfs2_holder_init(gl, state, flags, gh);
-       set_bit(HIF_ALLOCED, &gh->gh_iflags);
-       gh->gh_ip = (unsigned long)__builtin_return_address(0);
-       return gh;
+       clear_bit(HIF_WAIT, &gh->gh_iflags);
+       smp_mb__after_clear_bit();
+       wake_up_bit(&gh->gh_iflags, HIF_WAIT);
 }
 
-/**
- * gfs2_holder_put - get rid of a struct gfs2_holder structure
- * @gh: the holder structure
- *
- */
-
-static void gfs2_holder_put(struct gfs2_holder *gh)
+static int just_schedule(void *word)
 {
-       gfs2_holder_uninit(gh);
-       kfree(gh);
+        schedule();
+        return 0;
 }
 
-static void gfs2_holder_dispose_or_wake(struct gfs2_holder *gh)
+static void wait_on_holder(struct gfs2_holder *gh)
 {
-       if (test_bit(HIF_DEALLOC, &gh->gh_iflags)) {
-               gfs2_holder_put(gh);
-               return;
-       }
-       clear_bit(HIF_WAIT, &gh->gh_iflags);
-       smp_mb();
-       wake_up_bit(&gh->gh_iflags, HIF_WAIT);
+       might_sleep();
+       wait_on_bit(&gh->gh_iflags, HIF_WAIT, just_schedule, TASK_UNINTERRUPTIBLE);
 }
 
-static int holder_wait(void *word)
+static void gfs2_demote_wake(struct gfs2_glock *gl)
 {
-        schedule();
-        return 0;
+        clear_bit(GLF_DEMOTE, &gl->gl_flags);
+        smp_mb__after_clear_bit();
+        wake_up_bit(&gl->gl_flags, GLF_DEMOTE);
 }
 
-static void wait_on_holder(struct gfs2_holder *gh)
+static void wait_on_demote(struct gfs2_glock *gl)
 {
        might_sleep();
-       wait_on_bit(&gh->gh_iflags, HIF_WAIT, holder_wait, TASK_UNINTERRUPTIBLE);
+       wait_on_bit(&gl->gl_flags, GLF_DEMOTE, just_schedule, TASK_UNINTERRUPTIBLE);
 }
 
 /**
@@ -519,7 +499,7 @@ static int rq_promote(struct gfs2_holder *gh)
                                gfs2_reclaim_glock(sdp);
                        }
 
-                       gfs2_glock_xmote_th(gh);
+                       gfs2_glock_xmote_th(gh->gh_gl, gh);
                        spin_lock(&gl->gl_spin);
                }
                return 1;
@@ -542,7 +522,7 @@ static int rq_promote(struct gfs2_holder *gh)
        gh->gh_error = 0;
        set_bit(HIF_HOLDER, &gh->gh_iflags);
 
-       gfs2_holder_dispose_or_wake(gh);
+       gfs2_holder_wake(gh);
 
        return 0;
 }
@@ -554,32 +534,26 @@ static int rq_promote(struct gfs2_holder *gh)
  * Returns: 1 if the queue is blocked
  */
 
-static int rq_demote(struct gfs2_holder *gh)
+static int rq_demote(struct gfs2_glock *gl)
 {
-       struct gfs2_glock *gl = gh->gh_gl;
-
        if (!list_empty(&gl->gl_holders))
                return 1;
 
-       if (gl->gl_state == gh->gh_state || gl->gl_state == LM_ST_UNLOCKED) {
-               list_del_init(&gh->gh_list);
-               gh->gh_error = 0;
+       if (gl->gl_state == gl->gl_demote_state ||
+           gl->gl_state == LM_ST_UNLOCKED) {
+               gfs2_demote_wake(gl);
+               return 0;
+       }
+       set_bit(GLF_LOCK, &gl->gl_flags);
+       if (gl->gl_demote_state == LM_ST_UNLOCKED ||
+           gl->gl_state != LM_ST_EXCLUSIVE) {
                spin_unlock(&gl->gl_spin);
-               gfs2_holder_dispose_or_wake(gh);
-               spin_lock(&gl->gl_spin);
+               gfs2_glock_drop_th(gl);
        } else {
-               gl->gl_req_gh = gh;
-               set_bit(GLF_LOCK, &gl->gl_flags);
                spin_unlock(&gl->gl_spin);
-
-               if (gh->gh_state == LM_ST_UNLOCKED ||
-                   gl->gl_state != LM_ST_EXCLUSIVE)
-                       gfs2_glock_drop_th(gl);
-               else
-                       gfs2_glock_xmote_th(gh);
-
-               spin_lock(&gl->gl_spin);
+               gfs2_glock_xmote_th(gl, NULL);
        }
+       spin_lock(&gl->gl_spin);
 
        return 0;
 }
@@ -607,16 +581,8 @@ static void run_queue(struct gfs2_glock *gl)
                        else
                                gfs2_assert_warn(gl->gl_sbd, 0);
 
-               } else if (!list_empty(&gl->gl_waiters2) &&
-                          !test_bit(GLF_SKIP_WAITERS2, &gl->gl_flags)) {
-                       gh = list_entry(gl->gl_waiters2.next,
-                                       struct gfs2_holder, gh_list);
-
-                       if (test_bit(HIF_DEMOTE, &gh->gh_iflags))
-                               blocked = rq_demote(gh);
-                       else
-                               gfs2_assert_warn(gl->gl_sbd, 0);
-
+               } else if (test_bit(GLF_DEMOTE, &gl->gl_flags)) {
+                       blocked = rq_demote(gl);
                } else if (!list_empty(&gl->gl_waiters3)) {
                        gh = list_entry(gl->gl_waiters3.next,
                                        struct gfs2_holder, gh_list);
@@ -654,7 +620,7 @@ static void gfs2_glmutex_lock(struct gfs2_glock *gl)
        if (test_and_set_bit(GLF_LOCK, &gl->gl_flags)) {
                list_add_tail(&gh.gh_list, &gl->gl_waiters1);
        } else {
-               gl->gl_owner = current;
+               gl->gl_owner_pid = current->pid;
                gl->gl_ip = (unsigned long)__builtin_return_address(0);
                clear_bit(HIF_WAIT, &gh.gh_iflags);
                smp_mb();
@@ -681,7 +647,7 @@ static int gfs2_glmutex_trylock(struct gfs2_glock *gl)
        if (test_and_set_bit(GLF_LOCK, &gl->gl_flags)) {
                acquired = 0;
        } else {
-               gl->gl_owner = current;
+               gl->gl_owner_pid = current->pid;
                gl->gl_ip = (unsigned long)__builtin_return_address(0);
        }
        spin_unlock(&gl->gl_spin);
@@ -699,7 +665,7 @@ static void gfs2_glmutex_unlock(struct gfs2_glock *gl)
 {
        spin_lock(&gl->gl_spin);
        clear_bit(GLF_LOCK, &gl->gl_flags);
-       gl->gl_owner = NULL;
+       gl->gl_owner_pid = 0;
        gl->gl_ip = 0;
        run_queue(gl);
        BUG_ON(!spin_is_locked(&gl->gl_spin));
@@ -707,50 +673,35 @@ static void gfs2_glmutex_unlock(struct gfs2_glock *gl)
 }
 
 /**
- * handle_callback - add a demote request to a lock's queue
+ * handle_callback - process a demote request
  * @gl: the glock
  * @state: the state the caller wants us to change to
  *
- * Note: This may fail sliently if we are out of memory.
+ * There are only two requests that we are going to see in actual
+ * practise: LM_ST_SHARED and LM_ST_UNLOCKED
  */
 
-static void handle_callback(struct gfs2_glock *gl, unsigned int state)
+static void handle_callback(struct gfs2_glock *gl, unsigned int state, int remote)
 {
-       struct gfs2_holder *gh, *new_gh = NULL;
-
-restart:
        spin_lock(&gl->gl_spin);
-
-       list_for_each_entry(gh, &gl->gl_waiters2, gh_list) {
-               if (test_bit(HIF_DEMOTE, &gh->gh_iflags) &&
-                   gl->gl_req_gh != gh) {
-                       if (gh->gh_state != state)
-                               gh->gh_state = LM_ST_UNLOCKED;
-                       goto out;
-               }
-       }
-
-       if (new_gh) {
-               list_add_tail(&new_gh->gh_list, &gl->gl_waiters2);
-               new_gh = NULL;
-       } else {
-               spin_unlock(&gl->gl_spin);
-
-               new_gh = gfs2_holder_get(gl, state, LM_FLAG_TRY, GFP_NOFS);
-               if (!new_gh)
+       if (test_and_set_bit(GLF_DEMOTE, &gl->gl_flags) == 0) {
+               gl->gl_demote_state = state;
+               gl->gl_demote_time = jiffies;
+               if (remote && gl->gl_ops->go_type == LM_TYPE_IOPEN &&
+                   gl->gl_object) {
+                       struct inode *inode = igrab(gl->gl_object);
+                       spin_unlock(&gl->gl_spin);
+                       if (inode) {
+                               d_prune_aliases(inode);
+                               iput(inode);
+                       }
                        return;
-               set_bit(HIF_DEMOTE, &new_gh->gh_iflags);
-               set_bit(HIF_DEALLOC, &new_gh->gh_iflags);
-               set_bit(HIF_WAIT, &new_gh->gh_iflags);
-
-               goto restart;
+               }
+       } else if (gl->gl_demote_state != LM_ST_UNLOCKED &&
+                       gl->gl_demote_state != state) {
+               gl->gl_demote_state = LM_ST_UNLOCKED;
        }
-
-out:
        spin_unlock(&gl->gl_spin);
-
-       if (new_gh)
-               gfs2_holder_put(new_gh);
 }
 
 /**
@@ -810,56 +761,47 @@ static void xmote_bh(struct gfs2_glock *gl, unsigned int ret)
 
        /*  Deal with each possible exit condition  */
 
-       if (!gh)
+       if (!gh) {
                gl->gl_stamp = jiffies;
-       else if (unlikely(test_bit(SDF_SHUTDOWN, &sdp->sd_flags))) {
-               spin_lock(&gl->gl_spin);
-               list_del_init(&gh->gh_list);
-               gh->gh_error = -EIO;
-               spin_unlock(&gl->gl_spin);
-       } else if (test_bit(HIF_DEMOTE, &gh->gh_iflags)) {
-               spin_lock(&gl->gl_spin);
-               list_del_init(&gh->gh_list);
-               if (gl->gl_state == gh->gh_state ||
-                   gl->gl_state == LM_ST_UNLOCKED) {
-                       gh->gh_error = 0;
+               if (ret & LM_OUT_CANCELED) {
+                       op_done = 0;
                } else {
-                       if (gfs2_assert_warn(sdp, gh->gh_flags &
-                                       (LM_FLAG_TRY | LM_FLAG_TRY_1CB)) == -1)
-                               fs_warn(sdp, "ret = 0x%.8X\n", ret);
-                       gh->gh_error = GLR_TRYFAILED;
+                       spin_lock(&gl->gl_spin);
+                       if (gl->gl_state != gl->gl_demote_state) {
+                               gl->gl_req_bh = NULL;
+                               spin_unlock(&gl->gl_spin);
+                               gfs2_glock_drop_th(gl);
+                               gfs2_glock_put(gl);
+                               return;
+                       }
+                       gfs2_demote_wake(gl);
+                       spin_unlock(&gl->gl_spin);
                }
-               spin_unlock(&gl->gl_spin);
-
-               if (ret & LM_OUT_CANCELED)
-                       handle_callback(gl, LM_ST_UNLOCKED);
-
-       } else if (ret & LM_OUT_CANCELED) {
+       } else {
                spin_lock(&gl->gl_spin);
                list_del_init(&gh->gh_list);
+               gh->gh_error = -EIO;
+               if (unlikely(test_bit(SDF_SHUTDOWN, &sdp->sd_flags))) 
+                       goto out;
                gh->gh_error = GLR_CANCELED;
-               spin_unlock(&gl->gl_spin);
-
-       } else if (relaxed_state_ok(gl->gl_state, gh->gh_state, gh->gh_flags)) {
-               spin_lock(&gl->gl_spin);
-               list_move_tail(&gh->gh_list, &gl->gl_holders);
-               gh->gh_error = 0;
-               set_bit(HIF_HOLDER, &gh->gh_iflags);
-               spin_unlock(&gl->gl_spin);
-
-               set_bit(HIF_FIRST, &gh->gh_iflags);
-
-               op_done = 0;
-
-       } else if (gh->gh_flags & (LM_FLAG_TRY | LM_FLAG_TRY_1CB)) {
-               spin_lock(&gl->gl_spin);
-               list_del_init(&gh->gh_list);
+               if (ret & LM_OUT_CANCELED) 
+                       goto out;
+               if (relaxed_state_ok(gl->gl_state, gh->gh_state, gh->gh_flags)) {
+                       list_add_tail(&gh->gh_list, &gl->gl_holders);
+                       gh->gh_error = 0;
+                       set_bit(HIF_HOLDER, &gh->gh_iflags);
+                       set_bit(HIF_FIRST, &gh->gh_iflags);
+                       op_done = 0;
+                       goto out;
+               }
                gh->gh_error = GLR_TRYFAILED;
-               spin_unlock(&gl->gl_spin);
-
-       } else {
+               if (gh->gh_flags & (LM_FLAG_TRY | LM_FLAG_TRY_1CB))
+                       goto out;
+               gh->gh_error = -EINVAL;
                if (gfs2_assert_withdraw(sdp, 0) == -1)
                        fs_err(sdp, "ret = 0x%.8X\n", ret);
+out:
+               spin_unlock(&gl->gl_spin);
        }
 
        if (glops->go_xmote_bh)
@@ -877,7 +819,7 @@ static void xmote_bh(struct gfs2_glock *gl, unsigned int ret)
        gfs2_glock_put(gl);
 
        if (gh)
-               gfs2_holder_dispose_or_wake(gh);
+               gfs2_holder_wake(gh);
 }
 
 /**
@@ -888,12 +830,11 @@ static void xmote_bh(struct gfs2_glock *gl, unsigned int ret)
  *
  */
 
-void gfs2_glock_xmote_th(struct gfs2_holder *gh)
+static void gfs2_glock_xmote_th(struct gfs2_glock *gl, struct gfs2_holder *gh)
 {
-       struct gfs2_glock *gl = gh->gh_gl;
        struct gfs2_sbd *sdp = gl->gl_sbd;
-       int flags = gh->gh_flags;
-       unsigned state = gh->gh_state;
+       int flags = gh ? gh->gh_flags : 0;
+       unsigned state = gh ? gh->gh_state : gl->gl_demote_state;
        const struct gfs2_glock_operations *glops = gl->gl_ops;
        int lck_flags = flags & (LM_FLAG_TRY | LM_FLAG_TRY_1CB |
                                 LM_FLAG_NOEXP | LM_FLAG_ANY |
@@ -943,6 +884,7 @@ static void drop_bh(struct gfs2_glock *gl, unsigned int ret)
        gfs2_assert_warn(sdp, !ret);
 
        state_change(gl, LM_ST_UNLOCKED);
+       gfs2_demote_wake(gl);
 
        if (glops->go_inval)
                glops->go_inval(gl, DIO_METADATA);
@@ -964,7 +906,7 @@ static void drop_bh(struct gfs2_glock *gl, unsigned int ret)
        gfs2_glock_put(gl);
 
        if (gh)
-               gfs2_holder_dispose_or_wake(gh);
+               gfs2_holder_wake(gh);
 }
 
 /**
@@ -1097,18 +1039,32 @@ static int glock_wait_internal(struct gfs2_holder *gh)
 }
 
 static inline struct gfs2_holder *
-find_holder_by_owner(struct list_head *head, struct task_struct *owner)
+find_holder_by_owner(struct list_head *head, pid_t pid)
 {
        struct gfs2_holder *gh;
 
        list_for_each_entry(gh, head, gh_list) {
-               if (gh->gh_owner == owner)
+               if (gh->gh_owner_pid == pid)
                        return gh;
        }
 
        return NULL;
 }
 
+static void print_dbg(struct glock_iter *gi, const char *fmt, ...)
+{
+       va_list args;
+
+       va_start(args, fmt);
+       if (gi) {
+               vsprintf(gi->string, fmt, args);
+               seq_printf(gi->seq, gi->string);
+       }
+       else
+               vprintk(fmt, args);
+       va_end(args);
+}
+
 /**
  * add_to_queue - Add a holder to the wait queue (but look for recursion)
  * @gh: the holder structure to add
@@ -1120,24 +1076,24 @@ static void add_to_queue(struct gfs2_holder *gh)
        struct gfs2_glock *gl = gh->gh_gl;
        struct gfs2_holder *existing;
 
-       BUG_ON(!gh->gh_owner);
+       BUG_ON(!gh->gh_owner_pid);
        if (test_and_set_bit(HIF_WAIT, &gh->gh_iflags))
                BUG();
 
-       existing = find_holder_by_owner(&gl->gl_holders, gh->gh_owner);
+       existing = find_holder_by_owner(&gl->gl_holders, gh->gh_owner_pid);
        if (existing) {
                print_symbol(KERN_WARNING "original: %s\n", existing->gh_ip);
-               printk(KERN_INFO "pid : %d\n", existing->gh_owner->pid);
+               printk(KERN_INFO "pid : %d\n", existing->gh_owner_pid);
                printk(KERN_INFO "lock type : %d lock state : %d\n",
                                existing->gh_gl->gl_name.ln_type, existing->gh_gl->gl_state);
                print_symbol(KERN_WARNING "new: %s\n", gh->gh_ip);
-               printk(KERN_INFO "pid : %d\n", gh->gh_owner->pid);
+               printk(KERN_INFO "pid : %d\n", gh->gh_owner_pid);
                printk(KERN_INFO "lock type : %d lock state : %d\n",
                                gl->gl_name.ln_type, gl->gl_state);
                BUG();
        }
 
-       existing = find_holder_by_owner(&gl->gl_waiters3, gh->gh_owner);
+       existing = find_holder_by_owner(&gl->gl_waiters3, gh->gh_owner_pid);
        if (existing) {
                print_symbol(KERN_WARNING "original: %s\n", existing->gh_ip);
                print_symbol(KERN_WARNING "new: %s\n", gh->gh_ip);
@@ -1254,7 +1210,7 @@ void gfs2_glock_dq(struct gfs2_holder *gh)
        const struct gfs2_glock_operations *glops = gl->gl_ops;
 
        if (gh->gh_flags & GL_NOCACHE)
-               handle_callback(gl, LM_ST_UNLOCKED);
+               handle_callback(gl, LM_ST_UNLOCKED, 0);
 
        gfs2_glmutex_lock(gl);
 
@@ -1267,9 +1223,8 @@ void gfs2_glock_dq(struct gfs2_holder *gh)
                if (glops->go_unlock)
                        glops->go_unlock(gh);
 
-               gl->gl_stamp = jiffies;
-
                spin_lock(&gl->gl_spin);
+               gl->gl_stamp = jiffies;
        }
 
        clear_bit(GLF_LOCK, &gl->gl_flags);
@@ -1277,6 +1232,13 @@ void gfs2_glock_dq(struct gfs2_holder *gh)
        spin_unlock(&gl->gl_spin);
 }
 
+void gfs2_glock_dq_wait(struct gfs2_holder *gh)
+{
+       struct gfs2_glock *gl = gh->gh_gl;
+       gfs2_glock_dq(gh);
+       wait_on_demote(gl);
+}
+
 /**
  * gfs2_glock_dq_uninit - dequeue a holder from a glock and initialize it
  * @gh: the holder structure
@@ -1378,10 +1340,6 @@ static int nq_m_sync(unsigned int num_gh, struct gfs2_holder *ghs,
  * @num_gh: the number of structures
  * @ghs: an array of struct gfs2_holder structures
  *
- * Figure out how big an impact this function has.  Either:
- * 1) Replace this code with code that calls gfs2_glock_prefetch()
- * 2) Forget async stuff and just call nq_m_sync()
- * 3) Leave it like it is
  *
  * Returns: 0 on success (all glocks acquired),
  *          errno on failure (no glocks acquired)
@@ -1389,62 +1347,28 @@ static int nq_m_sync(unsigned int num_gh, struct gfs2_holder *ghs,
 
 int gfs2_glock_nq_m(unsigned int num_gh, struct gfs2_holder *ghs)
 {
-       int *e;
-       unsigned int x;
-       int borked = 0, serious = 0;
+       struct gfs2_holder *tmp[4];
+       struct gfs2_holder **pph = tmp;
        int error = 0;
 
-       if (!num_gh)
+       switch(num_gh) {
+       case 0:
                return 0;
-
-       if (num_gh == 1) {
+       case 1:
                ghs->gh_flags &= ~(LM_FLAG_TRY | GL_ASYNC);
                return gfs2_glock_nq(ghs);
-       }
-
-       e = kcalloc(num_gh, sizeof(struct gfs2_holder *), GFP_KERNEL);
-       if (!e)
-               return -ENOMEM;
-
-       for (x = 0; x < num_gh; x++) {
-               ghs[x].gh_flags |= LM_FLAG_TRY | GL_ASYNC;
-               error = gfs2_glock_nq(&ghs[x]);
-               if (error) {
-                       borked = 1;
-                       serious = error;
-                       num_gh = x;
+       default:
+               if (num_gh <= 4)
                        break;
-               }
+               pph = kmalloc(num_gh * sizeof(struct gfs2_holder *), GFP_NOFS);
+               if (!pph)
+                       return -ENOMEM;
        }
 
-       for (x = 0; x < num_gh; x++) {
-               error = e[x] = glock_wait_internal(&ghs[x]);
-               if (error) {
-                       borked = 1;
-                       if (error != GLR_TRYFAILED && error != GLR_CANCELED)
-                               serious = error;
-               }
-       }
-
-       if (!borked) {
-               kfree(e);
-               return 0;
-       }
+       error = nq_m_sync(num_gh, ghs, pph);
 
-       for (x = 0; x < num_gh; x++)
-               if (!e[x])
-                       gfs2_glock_dq(&ghs[x]);
-
-       if (serious)
-               error = serious;
-       else {
-               for (x = 0; x < num_gh; x++)
-                       gfs2_holder_reinit(ghs[x].gh_state, ghs[x].gh_flags,
-                                         &ghs[x]);
-               error = nq_m_sync(num_gh, ghs, (struct gfs2_holder **)e);
-       }
-
-       kfree(e);
+       if (pph != tmp)
+               kfree(pph);
 
        return error;
 }
@@ -1537,7 +1461,7 @@ static void blocking_cb(struct gfs2_sbd *sdp, struct lm_lockname *name,
        if (!gl)
                return;
 
-       handle_callback(gl, state);
+       handle_callback(gl, state, 1);
 
        spin_lock(&gl->gl_spin);
        run_queue(gl);
@@ -1677,7 +1601,7 @@ void gfs2_reclaim_glock(struct gfs2_sbd *sdp)
        if (gfs2_glmutex_trylock(gl)) {
                if (list_empty(&gl->gl_holders) &&
                    gl->gl_state != LM_ST_UNLOCKED && demote_ok(gl))
-                       handle_callback(gl, LM_ST_UNLOCKED);
+                       handle_callback(gl, LM_ST_UNLOCKED, 0);
                gfs2_glmutex_unlock(gl);
        }
 
@@ -1790,7 +1714,7 @@ static void clear_glock(struct gfs2_glock *gl)
        if (gfs2_glmutex_trylock(gl)) {
                if (list_empty(&gl->gl_holders) &&
                    gl->gl_state != LM_ST_UNLOCKED)
-                       handle_callback(gl, LM_ST_UNLOCKED);
+                       handle_callback(gl, LM_ST_UNLOCKED, 0);
                gfs2_glmutex_unlock(gl);
        }
 }
@@ -1841,6 +1765,15 @@ void gfs2_gl_hash_clear(struct gfs2_sbd *sdp, int wait)
  *  Diagnostic routines to help debug distributed deadlock
  */
 
+static void gfs2_print_symbol(struct glock_iter *gi, const char *fmt,
+                              unsigned long address)
+{
+       char buffer[KSYM_SYMBOL_LEN];
+
+       sprint_symbol(buffer, address);
+       print_dbg(gi, fmt, buffer);
+}
+
 /**
  * dump_holder - print information about a glock holder
  * @str: a string naming the type of holder
@@ -1849,31 +1782,37 @@ void gfs2_gl_hash_clear(struct gfs2_sbd *sdp, int wait)
  * Returns: 0 on success, -ENOBUFS when we run out of space
  */
 
-static int dump_holder(char *str, struct gfs2_holder *gh)
+static int dump_holder(struct glock_iter *gi, char *str,
+                      struct gfs2_holder *gh)
 {
        unsigned int x;
-       int error = -ENOBUFS;
-
-       printk(KERN_INFO "  %s\n", str);
-       printk(KERN_INFO "    owner = %ld\n",
-                  (gh->gh_owner) ? (long)gh->gh_owner->pid : -1);
-       printk(KERN_INFO "    gh_state = %u\n", gh->gh_state);
-       printk(KERN_INFO "    gh_flags =");
+       struct task_struct *gh_owner;
+
+       print_dbg(gi, "  %s\n", str);
+       if (gh->gh_owner_pid) {
+               print_dbg(gi, "    owner = %ld ", (long)gh->gh_owner_pid);
+               gh_owner = find_task_by_pid(gh->gh_owner_pid);
+               if (gh_owner)
+                       print_dbg(gi, "(%s)\n", gh_owner->comm);
+               else
+                       print_dbg(gi, "(ended)\n");
+       } else
+               print_dbg(gi, "    owner = -1\n");
+       print_dbg(gi, "    gh_state = %u\n", gh->gh_state);
+       print_dbg(gi, "    gh_flags =");
        for (x = 0; x < 32; x++)
                if (gh->gh_flags & (1 << x))
-                       printk(" %u", x);
-       printk(" \n");
-       printk(KERN_INFO "    error = %d\n", gh->gh_error);
-       printk(KERN_INFO "    gh_iflags =");
+                       print_dbg(gi, " %u", x);
+       print_dbg(gi, " \n");
+       print_dbg(gi, "    error = %d\n", gh->gh_error);
+       print_dbg(gi, "    gh_iflags =");
        for (x = 0; x < 32; x++)
                if (test_bit(x, &gh->gh_iflags))
-                       printk(" %u", x);
-       printk(" \n");
-       print_symbol(KERN_INFO "    initialized at: %s\n", gh->gh_ip);
-
-       error = 0;
+                       print_dbg(gi, " %u", x);
+       print_dbg(gi, " \n");
+        gfs2_print_symbol(gi, "    initialized at: %s\n", gh->gh_ip);
 
-       return error;
+       return 0;
 }
 
 /**
@@ -1883,25 +1822,21 @@ static int dump_holder(char *str, struct gfs2_holder *gh)
  * Returns: 0 on success, -ENOBUFS when we run out of space
  */
 
-static int dump_inode(struct gfs2_inode *ip)
+static int dump_inode(struct glock_iter *gi, struct gfs2_inode *ip)
 {
        unsigned int x;
-       int error = -ENOBUFS;
 
-       printk(KERN_INFO "  Inode:\n");
-       printk(KERN_INFO "    num = %llu %llu\n",
-                   (unsigned long long)ip->i_num.no_formal_ino,
-                   (unsigned long long)ip->i_num.no_addr);
-       printk(KERN_INFO "    type = %u\n", IF2DT(ip->i_inode.i_mode));
-       printk(KERN_INFO "    i_flags =");
+       print_dbg(gi, "  Inode:\n");
+       print_dbg(gi, "    num = %llu/%llu\n",
+                 (unsigned long long)ip->i_no_formal_ino,
+                 (unsigned long long)ip->i_no_addr);
+       print_dbg(gi, "    type = %u\n", IF2DT(ip->i_inode.i_mode));
+       print_dbg(gi, "    i_flags =");
        for (x = 0; x < 32; x++)
                if (test_bit(x, &ip->i_flags))
-                       printk(" %u", x);
-       printk(" \n");
-
-       error = 0;
-
-       return error;
+                       print_dbg(gi, " %u", x);
+       print_dbg(gi, " \n");
+       return 0;
 }
 
 /**
@@ -1912,74 +1847,86 @@ static int dump_inode(struct gfs2_inode *ip)
  * Returns: 0 on success, -ENOBUFS when we run out of space
  */
 
-static int dump_glock(struct gfs2_glock *gl)
+static int dump_glock(struct glock_iter *gi, struct gfs2_glock *gl)
 {
        struct gfs2_holder *gh;
        unsigned int x;
        int error = -ENOBUFS;
+       struct task_struct *gl_owner;
 
        spin_lock(&gl->gl_spin);
 
-       printk(KERN_INFO "Glock 0x%p (%u, %llu)\n", gl, gl->gl_name.ln_type,
-              (unsigned long long)gl->gl_name.ln_number);
-       printk(KERN_INFO "  gl_flags =");
+       print_dbg(gi, "Glock 0x%p (%u, %llu)\n", gl, gl->gl_name.ln_type,
+                  (unsigned long long)gl->gl_name.ln_number);
+       print_dbg(gi, "  gl_flags =");
        for (x = 0; x < 32; x++) {
                if (test_bit(x, &gl->gl_flags))
-                       printk(" %u", x);
+                       print_dbg(gi, " %u", x);
        }
-       printk(" \n");
-       printk(KERN_INFO "  gl_ref = %d\n", atomic_read(&gl->gl_ref));
-       printk(KERN_INFO "  gl_state = %u\n", gl->gl_state);
-       printk(KERN_INFO "  gl_owner = %s\n", gl->gl_owner->comm);
-       print_symbol(KERN_INFO "  gl_ip = %s\n", gl->gl_ip);
-       printk(KERN_INFO "  req_gh = %s\n", (gl->gl_req_gh) ? "yes" : "no");
-       printk(KERN_INFO "  req_bh = %s\n", (gl->gl_req_bh) ? "yes" : "no");
-       printk(KERN_INFO "  lvb_count = %d\n", atomic_read(&gl->gl_lvb_count));
-       printk(KERN_INFO "  object = %s\n", (gl->gl_object) ? "yes" : "no");
-       printk(KERN_INFO "  le = %s\n",
+       if (!test_bit(GLF_LOCK, &gl->gl_flags))
+               print_dbg(gi, " (unlocked)");
+       print_dbg(gi, " \n");
+       print_dbg(gi, "  gl_ref = %d\n", atomic_read(&gl->gl_ref));
+       print_dbg(gi, "  gl_state = %u\n", gl->gl_state);
+       if (gl->gl_owner_pid) {
+               gl_owner = find_task_by_pid(gl->gl_owner_pid);
+               if (gl_owner)
+                       print_dbg(gi, "  gl_owner = pid %d (%s)\n",
+                                 gl->gl_owner_pid, gl_owner->comm);
+               else
+                       print_dbg(gi, "  gl_owner = %d (ended)\n",
+                                 gl->gl_owner_pid);
+       } else
+               print_dbg(gi, "  gl_owner = -1\n");
+       print_dbg(gi, "  gl_ip = %lu\n", gl->gl_ip);
+       print_dbg(gi, "  req_gh = %s\n", (gl->gl_req_gh) ? "yes" : "no");
+       print_dbg(gi, "  req_bh = %s\n", (gl->gl_req_bh) ? "yes" : "no");
+       print_dbg(gi, "  lvb_count = %d\n", atomic_read(&gl->gl_lvb_count));
+       print_dbg(gi, "  object = %s\n", (gl->gl_object) ? "yes" : "no");
+       print_dbg(gi, "  le = %s\n",
                   (list_empty(&gl->gl_le.le_list)) ? "no" : "yes");
-       printk(KERN_INFO "  reclaim = %s\n",
-                   (list_empty(&gl->gl_reclaim)) ? "no" : "yes");
+       print_dbg(gi, "  reclaim = %s\n",
+                  (list_empty(&gl->gl_reclaim)) ? "no" : "yes");
        if (gl->gl_aspace)
-               printk(KERN_INFO "  aspace = 0x%p nrpages = %lu\n", gl->gl_aspace,
-                      gl->gl_aspace->i_mapping->nrpages);
+               print_dbg(gi, "  aspace = 0x%p nrpages = %lu\n", gl->gl_aspace,
+                          gl->gl_aspace->i_mapping->nrpages);
        else
-               printk(KERN_INFO "  aspace = no\n");
-       printk(KERN_INFO "  ail = %d\n", atomic_read(&gl->gl_ail_count));
+               print_dbg(gi, "  aspace = no\n");
+       print_dbg(gi, "  ail = %d\n", atomic_read(&gl->gl_ail_count));
        if (gl->gl_req_gh) {
-               error = dump_holder("Request", gl->gl_req_gh);
+               error = dump_holder(gi, "Request", gl->gl_req_gh);
                if (error)
                        goto out;
        }
        list_for_each_entry(gh, &gl->gl_holders, gh_list) {
-               error = dump_holder("Holder", gh);
+               error = dump_holder(gi, "Holder", gh);
                if (error)
                        goto out;
        }
        list_for_each_entry(gh, &gl->gl_waiters1, gh_list) {
-               error = dump_holder("Waiter1", gh);
-               if (error)
-                       goto out;
-       }
-       list_for_each_entry(gh, &gl->gl_waiters2, gh_list) {
-               error = dump_holder("Waiter2", gh);
+               error = dump_holder(gi, "Waiter1", gh);
                if (error)
                        goto out;
        }
        list_for_each_entry(gh, &gl->gl_waiters3, gh_list) {
-               error = dump_holder("Waiter3", gh);
+               error = dump_holder(gi, "Waiter3", gh);
                if (error)
                        goto out;
        }
+       if (test_bit(GLF_DEMOTE, &gl->gl_flags)) {
+               print_dbg(gi, "  Demotion req to state %u (%llu uS ago)\n",
+                         gl->gl_demote_state, (unsigned long long)
+                         (jiffies - gl->gl_demote_time)*(1000000/HZ));
+       }
        if (gl->gl_ops == &gfs2_inode_glops && gl->gl_object) {
                if (!test_bit(GLF_LOCK, &gl->gl_flags) &&
-                   list_empty(&gl->gl_holders)) {
-                       error = dump_inode(gl->gl_object);
+                       list_empty(&gl->gl_holders)) {
+                       error = dump_inode(gi, gl->gl_object);
                        if (error)
                                goto out;
                } else {
                        error = -ENOBUFS;
-                       printk(KERN_INFO "  Inode: busy\n");
+                       print_dbg(gi, "  Inode: busy\n");
                }
        }
 
@@ -2014,7 +1961,7 @@ static int gfs2_dump_lockstate(struct gfs2_sbd *sdp)
                        if (gl->gl_sbd != sdp)
                                continue;
 
-                       error = dump_glock(gl);
+                       error = dump_glock(NULL, gl);
                        if (error)
                                break;
                }
@@ -2043,3 +1990,189 @@ int __init gfs2_glock_init(void)
        return 0;
 }
 
+static int gfs2_glock_iter_next(struct glock_iter *gi)
+{
+       read_lock(gl_lock_addr(gi->hash));
+       while (1) {
+               if (!gi->hb_list) {  /* If we don't have a hash bucket yet */
+                       gi->hb_list = &gl_hash_table[gi->hash].hb_list;
+                       if (hlist_empty(gi->hb_list)) {
+                               read_unlock(gl_lock_addr(gi->hash));
+                               gi->hash++;
+                               read_lock(gl_lock_addr(gi->hash));
+                               gi->hb_list = NULL;
+                               if (gi->hash >= GFS2_GL_HASH_SIZE) {
+                                       read_unlock(gl_lock_addr(gi->hash));
+                                       return 1;
+                               }
+                               else
+                                       continue;
+                       }
+                       if (!hlist_empty(gi->hb_list)) {
+                               gi->gl = list_entry(gi->hb_list->first,
+                                                   struct gfs2_glock,
+                                                   gl_list);
+                       }
+               } else {
+                       if (gi->gl->gl_list.next == NULL) {
+                               read_unlock(gl_lock_addr(gi->hash));
+                               gi->hash++;
+                               read_lock(gl_lock_addr(gi->hash));
+                               gi->hb_list = NULL;
+                               continue;
+                       }
+                       gi->gl = list_entry(gi->gl->gl_list.next,
+                                           struct gfs2_glock, gl_list);
+               }
+               if (gi->gl)
+                       break;
+       }
+       read_unlock(gl_lock_addr(gi->hash));
+       return 0;
+}
+
+static void gfs2_glock_iter_free(struct glock_iter *gi)
+{
+       kfree(gi);
+}
+
+static struct glock_iter *gfs2_glock_iter_init(struct gfs2_sbd *sdp)
+{
+       struct glock_iter *gi;
+
+       gi = kmalloc(sizeof (*gi), GFP_KERNEL);
+       if (!gi)
+               return NULL;
+
+       gi->sdp = sdp;
+       gi->hash = 0;
+       gi->gl = NULL;
+       gi->hb_list = NULL;
+       gi->seq = NULL;
+       memset(gi->string, 0, sizeof(gi->string));
+
+       if (gfs2_glock_iter_next(gi)) {
+               gfs2_glock_iter_free(gi);
+               return NULL;
+       }
+
+       return gi;
+}
+
+static void *gfs2_glock_seq_start(struct seq_file *file, loff_t *pos)
+{
+       struct glock_iter *gi;
+       loff_t n = *pos;
+
+       gi = gfs2_glock_iter_init(file->private);
+       if (!gi)
+               return NULL;
+
+       while (n--) {
+               if (gfs2_glock_iter_next(gi)) {
+                       gfs2_glock_iter_free(gi);
+                       return NULL;
+               }
+       }
+
+       return gi;
+}
+
+static void *gfs2_glock_seq_next(struct seq_file *file, void *iter_ptr,
+                                loff_t *pos)
+{
+       struct glock_iter *gi = iter_ptr;
+
+       (*pos)++;
+
+       if (gfs2_glock_iter_next(gi)) {
+               gfs2_glock_iter_free(gi);
+               return NULL;
+       }
+
+       return gi;
+}
+
+static void gfs2_glock_seq_stop(struct seq_file *file, void *iter_ptr)
+{
+       /* nothing for now */
+}
+
+static int gfs2_glock_seq_show(struct seq_file *file, void *iter_ptr)
+{
+       struct glock_iter *gi = iter_ptr;
+
+       gi->seq = file;
+       dump_glock(gi, gi->gl);
+
+       return 0;
+}
+
+static struct seq_operations gfs2_glock_seq_ops = {
+       .start = gfs2_glock_seq_start,
+       .next  = gfs2_glock_seq_next,
+       .stop  = gfs2_glock_seq_stop,
+       .show  = gfs2_glock_seq_show,
+};
+
+static int gfs2_debugfs_open(struct inode *inode, struct file *file)
+{
+       struct seq_file *seq;
+       int ret;
+
+       ret = seq_open(file, &gfs2_glock_seq_ops);
+       if (ret)
+               return ret;
+
+       seq = file->private_data;
+       seq->private = inode->i_private;
+
+       return 0;
+}
+
+static const struct file_operations gfs2_debug_fops = {
+       .owner   = THIS_MODULE,
+       .open    = gfs2_debugfs_open,
+       .read    = seq_read,
+       .llseek  = seq_lseek,
+       .release = seq_release
+};
+
+int gfs2_create_debugfs_file(struct gfs2_sbd *sdp)
+{
+       sdp->debugfs_dir = debugfs_create_dir(sdp->sd_table_name, gfs2_root);
+       if (!sdp->debugfs_dir)
+               return -ENOMEM;
+       sdp->debugfs_dentry_glocks = debugfs_create_file("glocks",
+                                                        S_IFREG | S_IRUGO,
+                                                        sdp->debugfs_dir, sdp,
+                                                        &gfs2_debug_fops);
+       if (!sdp->debugfs_dentry_glocks)
+               return -ENOMEM;
+
+       return 0;
+}
+
+void gfs2_delete_debugfs_file(struct gfs2_sbd *sdp)
+{
+       if (sdp && sdp->debugfs_dir) {
+               if (sdp->debugfs_dentry_glocks) {
+                       debugfs_remove(sdp->debugfs_dentry_glocks);
+                       sdp->debugfs_dentry_glocks = NULL;
+               }
+               debugfs_remove(sdp->debugfs_dir);
+               sdp->debugfs_dir = NULL;
+       }
+}
+
+int gfs2_register_debugfs(void)
+{
+       gfs2_root = debugfs_create_dir("gfs2", NULL);
+       return gfs2_root ? 0 : -ENOMEM;
+}
+
+void gfs2_unregister_debugfs(void)
+{
+       debugfs_remove(gfs2_root);
+       gfs2_root = NULL;
+}