1 files changed, 769 insertions, 219 deletions

diff --git a/kernel/futex.c b/kernel/futex.c
index 600bc9d801f2..b7ce15c67e32 100644
--- a/kernel/futex.c
+++ b/kernel/futex.c
@@ -16,6 +16,9 @@
  *  Copyright (C) 2006 Red Hat, Inc., Ingo Molnar <mingo@redhat.com>
  *  Copyright (C) 2006 Timesys Corp., Thomas Gleixner <tglx@timesys.com>
  *
+ *  PRIVATE futexes by Eric Dumazet
+ *  Copyright (C) 2007 Eric Dumazet <dada1@cosmosbay.com>
+ *
  *  Thanks to Ben LaHaise for yelling "hashed waitqueues" loudly
  *  enough at me, Linus for the original (flawed) idea, Matthew
  *  Kirkwood for proof-of-concept implementation.
@@ -53,6 +56,12 @@
 
 #include "rtmutex_common.h"
 
+#ifdef CONFIG_DEBUG_RT_MUTEXES
+# include "rtmutex-debug.h"
+#else
+# include "rtmutex.h"
+#endif
+
 #define FUTEX_HASHBITS (CONFIG_BASE_SMALL ? 4 : 8)
 
 /*
@@ -81,12 +90,12 @@ struct futex_pi_state {
  * we can wake only the relevant ones (hashed queues may be shared).
  *
  * A futex_q has a woken state, just like tasks have TASK_RUNNING.
- * It is considered woken when list_empty(&q->list) || q->lock_ptr == 0.
+ * It is considered woken when plist_node_empty(&q->list) || q->lock_ptr == 0.
  * The order of wakup is always to make the first condition true, then
  * wake up q->waiters, then make the second condition true.
  */
 struct futex_q {
-        struct list_head list;
+        struct plist_node list;
         wait_queue_head_t waiters;
 
         /* Which hash list lock to use: */
@@ -102,14 +111,20 @@ struct futex_q {
         /* Optional priority inheritance state: */
         struct futex_pi_state *pi_state;
         struct task_struct *task;
+
+        /*
+         * This waiter is used in case of requeue from a
+         * normal futex to a PI-futex
+         */
+        struct rt_mutex_waiter waiter;
 };
 
 /*
  * Split the global futex_lock into every hash list lock.
  */
 struct futex_hash_bucket {
-       spinlock_t              lock;
-       struct list_head       chain;
+        spinlock_t lock;
+        struct plist_head chain;
 };
 
 static struct futex_hash_bucket futex_queues[1<<FUTEX_HASHBITS];
@@ -138,19 +153,26 @@ static inline int match_futex(union futex_key *key1, union futex_key *key2)
                 && key1->both.offset == key2->both.offset);
 }
 
-/*
- * Get parameters which are the keys for a futex.
+/**
+ * get_futex_key - Get parameters which are the keys for a futex.
+ * @uaddr: virtual address of the futex
+ * @shared: NULL for a PROCESS_PRIVATE futex,
+ *      &current->mm->mmap_sem for a PROCESS_SHARED futex
+ * @key: address where result is stored.
+ *
+ * Returns a negative error code or 0
+ * The key words are stored in *key on success.
  *
  * For shared mappings, it's (page->index, vma->vm_file->f_path.dentry->d_inode,
  * offset_within_page).  For private mappings, it's (uaddr, current->mm).
  * We can usually work out the index without swapping in the page.
  *
- * Returns: 0, or negative error code.
- * The key words are stored in *key on success.
- *
- * Should be called with &current->mm->mmap_sem but NOT any spinlocks.
+ * fshared is NULL for PROCESS_PRIVATE futexes
+ * For other futexes, it points to &current->mm->mmap_sem and
+ * caller must have taken the reader lock. but NOT any spinlocks.
  */
-int get_futex_key(u32 __user *uaddr, union futex_key *key)
+int get_futex_key(u32 __user *uaddr, struct rw_semaphore *fshared,
+                  union futex_key *key)
 {
         unsigned long address = (unsigned long)uaddr;
         struct mm_struct *mm = current->mm;
@@ -162,11 +184,25 @@ int get_futex_key(u32 __user *uaddr, union futex_key *key)
          * The futex address must be "naturally" aligned.
          */
         key->both.offset = address % PAGE_SIZE;
-        if (unlikely((key->both.offset % sizeof(u32)) != 0))
+        if (unlikely((address % sizeof(u32)) != 0))
                 return -EINVAL;
         address -= key->both.offset;
 
         /*
+         * PROCESS_PRIVATE futexes are fast.
+         * As the mm cannot disappear under us and the 'key' only needs
+         * virtual address, we dont even have to find the underlying vma.
+         * Note : We do have to check 'uaddr' is a valid user address,
+         *        but access_ok() should be faster than find_vma()
+         */
+        if (!fshared) {
+                if (unlikely(!access_ok(VERIFY_WRITE, uaddr, sizeof(u32))))
+                        return -EFAULT;
+                key->private.mm = mm;
+                key->private.address = address;
+                return 0;
+        }
+        /*
          * The futex is hashed differently depending on whether
          * it's in a shared or private mapping.  So check vma first.
          */
@@ -180,6 +216,9 @@ int get_futex_key(u32 __user *uaddr, union futex_key *key)
         if (unlikely((vma->vm_flags & (VM_IO|VM_READ)) != VM_READ))
                 return (vma->vm_flags & VM_IO) ? -EPERM : -EACCES;
 
+        /* Save the user address in the ley */
+        key->uaddr = uaddr;
+
         /*
          * Private mappings are handled in a simple way.
          *
@@ -190,6 +229,7 @@ int get_futex_key(u32 __user *uaddr, union futex_key *key)
          * mappings of _writable_ handles.
          */
         if (likely(!(vma->vm_flags & VM_MAYSHARE))) {
+                key->both.offset |= FUT_OFF_MMSHARED; /* reference taken on mm */
                 key->private.mm = mm;
                 key->private.address = address;
                 return 0;
@@ -199,7 +239,7 @@ int get_futex_key(u32 __user *uaddr, union futex_key *key)
          * Linear file mappings are also simple.
          */
         key->shared.inode = vma->vm_file->f_path.dentry->d_inode;
-        key->both.offset++; /* Bit 0 of offset indicates inode-based key. */
+        key->both.offset |= FUT_OFF_INODE; /* inode-based key. */
         if (likely(!(vma->vm_flags & VM_NONLINEAR))) {
                 key->shared.pgoff = (((address - vma->vm_start) >> PAGE_SHIFT)
                                      + vma->vm_pgoff);
@@ -227,16 +267,18 @@ EXPORT_SYMBOL_GPL(get_futex_key);
  * Take a reference to the resource addressed by a key.
  * Can be called while holding spinlocks.
  *
- * NOTE: mmap_sem MUST be held between get_futex_key() and calling this
- * function, if it is called at all.  mmap_sem keeps key->shared.inode valid.
  */
 inline void get_futex_key_refs(union futex_key *key)
 {
-        if (key->both.ptr != 0) {
-                if (key->both.offset & 1)
+        if (key->both.ptr == 0)
+                return;
+        switch (key->both.offset & (FUT_OFF_INODE|FUT_OFF_MMSHARED)) {
+                case FUT_OFF_INODE:
                         atomic_inc(&key->shared.inode->i_count);
-                else
+                        break;
+                case FUT_OFF_MMSHARED:
                         atomic_inc(&key->private.mm->mm_count);
+                        break;
         }
 }
 EXPORT_SYMBOL_GPL(get_futex_key_refs);
@@ -247,11 +289,15 @@ EXPORT_SYMBOL_GPL(get_futex_key_refs);
  */
 void drop_futex_key_refs(union futex_key *key)
 {
-        if (key->both.ptr != 0) {
-                if (key->both.offset & 1)
+        if (key->both.ptr == 0)
+                return;
+        switch (key->both.offset & (FUT_OFF_INODE|FUT_OFF_MMSHARED)) {
+                case FUT_OFF_INODE:
                         iput(key->shared.inode);
-                else
+                        break;
+                case FUT_OFF_MMSHARED:
                         mmdrop(key->private.mm);
+                        break;
         }
 }
 EXPORT_SYMBOL_GPL(drop_futex_key_refs);
@@ -268,28 +314,38 @@ static inline int get_futex_value_locked(u32 *dest, u32 __user *from)
 }
 
 /*
- * Fault handling. Called with current->mm->mmap_sem held.
+ * Fault handling.
+ * if fshared is non NULL, current->mm->mmap_sem is already held
  */
-static int futex_handle_fault(unsigned long address, int attempt)
+static int futex_handle_fault(unsigned long address,
+                              struct rw_semaphore *fshared, int attempt)
 {
         struct vm_area_struct * vma;
         struct mm_struct *mm = current->mm;
+        int ret = -EFAULT;
 
-        if (attempt > 2 || !(vma = find_vma(mm, address)) ||
-            vma->vm_start > address || !(vma->vm_flags & VM_WRITE))
-                return -EFAULT;
+        if (attempt > 2)
+                return ret;
 
-        switch (handle_mm_fault(mm, vma, address, 1)) {
-        case VM_FAULT_MINOR:
-                current->min_flt++;
-                break;
-        case VM_FAULT_MAJOR:
-                current->maj_flt++;
-                break;
-        default:
-                return -EFAULT;
+        if (!fshared)
+                down_read(&mm->mmap_sem);
+        vma = find_vma(mm, address);
+        if (vma && address >= vma->vm_start &&
+            (vma->vm_flags & VM_WRITE)) {
+                switch (handle_mm_fault(mm, vma, address, 1)) {
+                case VM_FAULT_MINOR:
+                        ret = 0;
+                        current->min_flt++;
+                        break;
+                case VM_FAULT_MAJOR:
+                        ret = 0;
+                        current->maj_flt++;
+                        break;
+                }
         }
-        return 0;
+        if (!fshared)
+                up_read(&mm->mmap_sem);
+        return ret;
 }
 
 /*
@@ -439,18 +495,19 @@ void exit_pi_state_list(struct task_struct *curr)
 }
 
 static int
-lookup_pi_state(u32 uval, struct futex_hash_bucket *hb, struct futex_q *me)
+lookup_pi_state(u32 uval, struct futex_hash_bucket *hb,
+                union futex_key *key, struct futex_pi_state **ps)
 {
         struct futex_pi_state *pi_state = NULL;
         struct futex_q *this, *next;
-        struct list_head *head;
+        struct plist_head *head;
         struct task_struct *p;
         pid_t pid;
 
         head = &hb->chain;
 
-        list_for_each_entry_safe(this, next, head, list) {
-                if (match_futex(&this->key, &me->key)) {
+        plist_for_each_entry_safe(this, next, head, list) {
+                if (match_futex(&this->key, key)) {
                         /*
                          * Another waiter already exists - bump up
                          * the refcount and return its pi_state:
@@ -465,7 +522,7 @@ lookup_pi_state(u32 uval, struct futex_hash_bucket *hb, struct futex_q *me)
                         WARN_ON(!atomic_read(&pi_state->refcount));
 
                         atomic_inc(&pi_state->refcount);
-                        me->pi_state = pi_state;
+                        *ps = pi_state;
 
                         return 0;
                 }
@@ -492,7 +549,7 @@ lookup_pi_state(u32 uval, struct futex_hash_bucket *hb, struct futex_q *me)
         rt_mutex_init_proxy_locked(&pi_state->pi_mutex, p);
 
         /* Store the key for possible exit cleanups: */
-        pi_state->key = me->key;
+        pi_state->key = *key;
 
         spin_lock_irq(&p->pi_lock);
         WARN_ON(!list_empty(&pi_state->list));
@@ -502,7 +559,7 @@ lookup_pi_state(u32 uval, struct futex_hash_bucket *hb, struct futex_q *me)
 
         put_task_struct(p);
 
-        me->pi_state = pi_state;
+        *ps = pi_state;
 
         return 0;
 }
@@ -513,12 +570,12 @@ lookup_pi_state(u32 uval, struct futex_hash_bucket *hb, struct futex_q *me)
  */
 static void wake_futex(struct futex_q *q)
 {
-        list_del_init(&q->list);
+        plist_del(&q->list, &q->list.plist);
         if (q->filp)
                 send_sigio(&q->filp->f_owner, q->fd, POLL_IN);
         /*
          * The lock in wake_up_all() is a crucial memory barrier after the
-         * list_del_init() and also before assigning to q->lock_ptr.
+         * plist_del() and also before assigning to q->lock_ptr.
          */
         wake_up_all(&q->waiters);
         /*
@@ -562,6 +619,8 @@ static int wake_futex_pi(u32 __user *uaddr, u32 uval, struct futex_q *this)
          */
         if (!(uval & FUTEX_OWNER_DIED)) {
                 newval = FUTEX_WAITERS | new_owner->pid;
+                /* Keep the FUTEX_WAITER_REQUEUED flag if it was set */
+                newval |= (uval & FUTEX_WAITER_REQUEUED);
 
                 pagefault_disable();
                 curval = futex_atomic_cmpxchg_inatomic(uaddr, uval, newval);
@@ -629,17 +688,19 @@ double_lock_hb(struct futex_hash_bucket *hb1, struct futex_hash_bucket *hb2)
  * Wake up all waiters hashed on the physical page that is mapped
  * to this virtual address:
  */
-static int futex_wake(u32 __user *uaddr, int nr_wake)
+static int futex_wake(u32 __user *uaddr, struct rw_semaphore *fshared,
+                      int nr_wake)
 {
         struct futex_hash_bucket *hb;
         struct futex_q *this, *next;
-        struct list_head *head;
+        struct plist_head *head;
         union futex_key key;
         int ret;
 
-        down_read(&current->mm->mmap_sem);
+        if (fshared)
+                down_read(fshared);
 
-        ret = get_futex_key(uaddr, &key);
+        ret = get_futex_key(uaddr, fshared, &key);
         if (unlikely(ret != 0))
                 goto out;
 
@@ -647,7 +708,7 @@ static int futex_wake(u32 __user *uaddr, int nr_wake)
         spin_lock(&hb->lock);
         head = &hb->chain;
 
-        list_for_each_entry_safe(this, next, head, list) {
+        plist_for_each_entry_safe(this, next, head, list) {
                 if (match_futex (&this->key, &key)) {
                         if (this->pi_state) {
                                 ret = -EINVAL;
@@ -661,7 +722,261 @@ static int futex_wake(u32 __user *uaddr, int nr_wake)
 
         spin_unlock(&hb->lock);
 out:
-        up_read(&current->mm->mmap_sem);
+        if (fshared)
+                up_read(fshared);
+        return ret;
+}
+
+/*
+ * Called from futex_requeue_pi.
+ * Set FUTEX_WAITERS and FUTEX_WAITER_REQUEUED flags on the
+ * PI-futex value; search its associated pi_state if an owner exist
+ * or create a new one without owner.
+ */
+static inline int
+lookup_pi_state_for_requeue(u32 __user *uaddr, struct futex_hash_bucket *hb,
+                            union futex_key *key,
+                            struct futex_pi_state **pi_state)
+{
+        u32 curval, uval, newval;
+
+retry:
+        /*
+         * We can't handle a fault cleanly because we can't
+         * release the locks here. Simply return the fault.
+         */
+        if (get_futex_value_locked(&curval, uaddr))
+                return -EFAULT;
+
+        /* set the flags FUTEX_WAITERS and FUTEX_WAITER_REQUEUED */
+        if ((curval & (FUTEX_WAITERS | FUTEX_WAITER_REQUEUED))
+            != (FUTEX_WAITERS | FUTEX_WAITER_REQUEUED)) {
+                /*
+                 * No waiters yet, we prepare the futex to have some waiters.
+                 */
+
+                uval = curval;
+                newval = uval | FUTEX_WAITERS | FUTEX_WAITER_REQUEUED;
+
+                pagefault_disable();
+                curval = futex_atomic_cmpxchg_inatomic(uaddr, uval, newval);
+                pagefault_enable();
+
+                if (unlikely(curval == -EFAULT))
+                        return -EFAULT;
+                if (unlikely(curval != uval))
+                        goto retry;
+        }
+
+        if (!(curval & FUTEX_TID_MASK)
+            || lookup_pi_state(curval, hb, key, pi_state)) {
+                /* the futex has no owner (yet) or the lookup failed:
+                   allocate one pi_state without owner */
+
+                *pi_state = alloc_pi_state();
+
+                /* Already stores the key: */
+                (*pi_state)->key = *key;
+
+                /* init the mutex without owner */
+                __rt_mutex_init(&(*pi_state)->pi_mutex, NULL);
+        }
+
+        return 0;
+}
+
+/*
+ * Keep the first nr_wake waiter from futex1, wake up one,
+ * and requeue the next nr_requeue waiters following hashed on
+ * one physical page to another physical page (PI-futex uaddr2)
+ */
+static int futex_requeue_pi(u32 __user *uaddr1,
+                            struct rw_semaphore *fshared,
+                            u32 __user *uaddr2,
+                            int nr_wake, int nr_requeue, u32 *cmpval)
+{
+        union futex_key key1, key2;
+        struct futex_hash_bucket *hb1, *hb2;
+        struct plist_head *head1;
+        struct futex_q *this, *next;
+        struct futex_pi_state *pi_state2 = NULL;
+        struct rt_mutex_waiter *waiter, *top_waiter = NULL;
+        struct rt_mutex *lock2 = NULL;
+        int ret, drop_count = 0;
+
+        if (refill_pi_state_cache())
+                return -ENOMEM;
+
+retry:
+        /*
+         * First take all the futex related locks:
+         */
+        if (fshared)
+                down_read(fshared);
+
+        ret = get_futex_key(uaddr1, fshared, &key1);
+        if (unlikely(ret != 0))
+                goto out;
+        ret = get_futex_key(uaddr2, fshared, &key2);
+        if (unlikely(ret != 0))
+                goto out;
+
+        hb1 = hash_futex(&key1);
+        hb2 = hash_futex(&key2);
+
+        double_lock_hb(hb1, hb2);
+
+        if (likely(cmpval != NULL)) {
+                u32 curval;
+
+                ret = get_futex_value_locked(&curval, uaddr1);
+
+                if (unlikely(ret)) {
+                        spin_unlock(&hb1->lock);
+                        if (hb1 != hb2)
+                                spin_unlock(&hb2->lock);
+
+                        /*
+                         * If we would have faulted, release mmap_sem, fault
+                         * it in and start all over again.
+                         */
+                        if (fshared)
+                                up_read(fshared);
+
+                        ret = get_user(curval, uaddr1);
+
+                        if (!ret)
+                                goto retry;
+
+                        return ret;
+                }
+                if (curval != *cmpval) {
+                        ret = -EAGAIN;
+                        goto out_unlock;
+                }
+        }
+
+        head1 = &hb1->chain;
+        plist_for_each_entry_safe(this, next, head1, list) {
+                if (!match_futex (&this->key, &key1))
+                        continue;
+                if (++ret <= nr_wake) {
+                        wake_futex(this);
+                } else {
+                        /*
+                         * FIRST: get and set the pi_state
+                         */
+                        if (!pi_state2) {
+                                int s;
+                                /* do this only the first time we requeue someone */
+                                s = lookup_pi_state_for_requeue(uaddr2, hb2,
+                                                                &key2, &pi_state2);
+                                if (s) {
+                                        ret = s;
+                                        goto out_unlock;
+                                }
+
+                                lock2 = &pi_state2->pi_mutex;
+                                spin_lock(&lock2->wait_lock);
+
+                                /* Save the top waiter of the wait_list */
+                                if (rt_mutex_has_waiters(lock2))
+                                        top_waiter = rt_mutex_top_waiter(lock2);
+                        } else
+                                atomic_inc(&pi_state2->refcount);
+
+
+                        this->pi_state = pi_state2;
+
+                        /*
+                         * SECOND: requeue futex_q to the correct hashbucket
+                         */
+
+                        /*
+                         * If key1 and key2 hash to the same bucket, no need to
+                         * requeue.
+                         */
+                        if (likely(head1 != &hb2->chain)) {
+                                plist_del(&this->list, &hb1->chain);
+                                plist_add(&this->list, &hb2->chain);
+                                this->lock_ptr = &hb2->lock;
+#ifdef CONFIG_DEBUG_PI_LIST
+                                this->list.plist.lock = &hb2->lock;
+#endif
+                        }
+                        this->key = key2;
+                        get_futex_key_refs(&key2);
+                        drop_count++;
+
+
+                        /*
+                         * THIRD: queue it to lock2
+                         */
+                        spin_lock_irq(&this->task->pi_lock);
+                        waiter = &this->waiter;
+                        waiter->task = this->task;
+                        waiter->lock = lock2;
+                        plist_node_init(&waiter->list_entry, this->task->prio);
+                        plist_node_init(&waiter->pi_list_entry, this->task->prio);
+                        plist_add(&waiter->list_entry, &lock2->wait_list);
+                        this->task->pi_blocked_on = waiter;
+                        spin_unlock_irq(&this->task->pi_lock);
+
+                        if (ret - nr_wake >= nr_requeue)
+                                break;
+                }
+        }
+
+        /* If we've requeued some tasks and the top_waiter of the rt_mutex
+           has changed, we must adjust the priority of the owner, if any */
+        if (drop_count) {
+                struct task_struct *owner = rt_mutex_owner(lock2);
+                if (owner &&
+                    (top_waiter != (waiter = rt_mutex_top_waiter(lock2)))) {
+                        int chain_walk = 0;
+
+                        spin_lock_irq(&owner->pi_lock);
+                        if (top_waiter)
+                                plist_del(&top_waiter->pi_list_entry, &owner->pi_waiters);
+                        else
+                                /*
+                                 * There was no waiters before the requeue,
+                                 * the flag must be updated
+                                 */
+                                mark_rt_mutex_waiters(lock2);
+
+                        plist_add(&waiter->pi_list_entry, &owner->pi_waiters);
+                        __rt_mutex_adjust_prio(owner);
+                        if (owner->pi_blocked_on) {
+                                chain_walk = 1;
+                                get_task_struct(owner);
+                        }
+
+                        spin_unlock_irq(&owner->pi_lock);
+                        spin_unlock(&lock2->wait_lock);
+
+                        if (chain_walk)
+                                rt_mutex_adjust_prio_chain(owner, 0, lock2, NULL,
+                                                           current);
+                } else {
+                        /* No owner or the top_waiter does not change */
+                        mark_rt_mutex_waiters(lock2);
+                        spin_unlock(&lock2->wait_lock);
+                }
+        }
+
+out_unlock:
+        spin_unlock(&hb1->lock);
+        if (hb1 != hb2)
+                spin_unlock(&hb2->lock);
+
+        /* drop_futex_key_refs() must be called outside the spinlocks. */
+        while (--drop_count >= 0)
+                drop_futex_key_refs(&key1);
+
+out:
+        if (fshared)
+                up_read(fshared);
         return ret;
 }
 
@@ -670,22 +985,24 @@ out:
  * to this virtual address:
  */
 static int
-futex_wake_op(u32 __user *uaddr1, u32 __user *uaddr2,
+futex_wake_op(u32 __user *uaddr1, struct rw_semaphore *fshared,
+              u32 __user *uaddr2,
               int nr_wake, int nr_wake2, int op)
 {
         union futex_key key1, key2;
         struct futex_hash_bucket *hb1, *hb2;
-        struct list_head *head;
+        struct plist_head *head;
         struct futex_q *this, *next;
         int ret, op_ret, attempt = 0;
 
 retryfull:
-        down_read(&current->mm->mmap_sem);
+        if (fshared)
+                down_read(fshared);
 
-        ret = get_futex_key(uaddr1, &key1);
+        ret = get_futex_key(uaddr1, fshared, &key1);
         if (unlikely(ret != 0))
                 goto out;
-        ret = get_futex_key(uaddr2, &key2);
+        ret = get_futex_key(uaddr2, fshared, &key2);
         if (unlikely(ret != 0))
                 goto out;
 
@@ -725,11 +1042,10 @@ retry:
                  * still holding the mmap_sem.
                  */
                 if (attempt++) {
-                        if (futex_handle_fault((unsigned long)uaddr2,
-                                                attempt)) {
-                                ret = -EFAULT;
+                        ret = futex_handle_fault((unsigned long)uaddr2,
+                                                fshared, attempt);
+                        if (ret)
                                 goto out;
-                        }
                         goto retry;
                 }
 
@@ -737,7 +1053,8 @@ retry:
                  * If we would have faulted, release mmap_sem,
                  * fault it in and start all over again.
                  */
-                up_read(&current->mm->mmap_sem);
+                if (fshared)
+                        up_read(fshared);
 
                 ret = get_user(dummy, uaddr2);
                 if (ret)
@@ -748,7 +1065,7 @@ retry:
 
         head = &hb1->chain;
 
-        list_for_each_entry_safe(this, next, head, list) {
+        plist_for_each_entry_safe(this, next, head, list) {
                 if (match_futex (&this->key, &key1)) {
                         wake_futex(this);
                         if (++ret >= nr_wake)
@@ -760,7 +1077,7 @@ retry:
                 head = &hb2->chain;
 
                 op_ret = 0;
-                list_for_each_entry_safe(this, next, head, list) {
+                plist_for_each_entry_safe(this, next, head, list) {
                         if (match_futex (&this->key, &key2)) {
                                 wake_futex(this);
                                 if (++op_ret >= nr_wake2)
@@ -774,7 +1091,8 @@ retry:
         if (hb1 != hb2)
                 spin_unlock(&hb2->lock);
 out:
-        up_read(&current->mm->mmap_sem);
+        if (fshared)
+                up_read(fshared);
         return ret;
 }
 
@@ -782,22 +1100,24 @@ out:
  * Requeue all waiters hashed on one physical page to another
  * physical page.
  */
-static int futex_requeue(u32 __user *uaddr1, u32 __user *uaddr2,
+static int futex_requeue(u32 __user *uaddr1, struct rw_semaphore *fshared,
+                         u32 __user *uaddr2,
                          int nr_wake, int nr_requeue, u32 *cmpval)
 {
         union futex_key key1, key2;
         struct futex_hash_bucket *hb1, *hb2;
-        struct list_head *head1;
+        struct plist_head *head1;
         struct futex_q *this, *next;
         int ret, drop_count = 0;
 
  retry:
-        down_read(&current->mm->mmap_sem);
+        if (fshared)
+                down_read(fshared);
 
-        ret = get_futex_key(uaddr1, &key1);
+        ret = get_futex_key(uaddr1, fshared, &key1);
         if (unlikely(ret != 0))
                 goto out;
-        ret = get_futex_key(uaddr2, &key2);
+        ret = get_futex_key(uaddr2, fshared, &key2);
         if (unlikely(ret != 0))
                 goto out;
 
@@ -820,7 +1140,8 @@ static int futex_requeue(u32 __user *uaddr1, u32 __user *uaddr2,
                          * If we would have faulted, release mmap_sem, fault
                          * it in and start all over again.
                          */
-                        up_read(&current->mm->mmap_sem);
+                        if (fshared)
+                                up_read(fshared);
 
                         ret = get_user(curval, uaddr1);
 
@@ -836,7 +1157,7 @@ static int futex_requeue(u32 __user *uaddr1, u32 __user *uaddr2,
         }
 
         head1 = &hb1->chain;
-        list_for_each_entry_safe(this, next, head1, list) {
+        plist_for_each_entry_safe(this, next, head1, list) {
                 if (!match_futex (&this->key, &key1))
                         continue;
                 if (++ret <= nr_wake) {
@@ -847,9 +1168,13 @@ static int futex_requeue(u32 __user *uaddr1, u32 __user *uaddr2,
                          * requeue.
                          */
                         if (likely(head1 != &hb2->chain)) {
-                                list_move_tail(&this->list, &hb2->chain);
+                                plist_del(&this->list, &hb1->chain);
+                                plist_add(&this->list, &hb2->chain);
                                 this->lock_ptr = &hb2->lock;
-                        }
+#ifdef CONFIG_DEBUG_PI_LIST
+                                this->list.plist.lock = &hb2->lock;
+#endif
+                        }
                         this->key = key2;
                         get_futex_key_refs(&key2);
                         drop_count++;
@@ -869,7 +1194,8 @@ out_unlock:
                 drop_futex_key_refs(&key1);
 
 out:
-        up_read(&current->mm->mmap_sem);
+        if (fshared)
+                up_read(fshared);
         return ret;
 }
 
@@ -894,7 +1220,23 @@ queue_lock(struct futex_q *q, int fd, struct file *filp)
 
 static inline void __queue_me(struct futex_q *q, struct futex_hash_bucket *hb)
 {
-        list_add_tail(&q->list, &hb->chain);
+        int prio;
+
+        /*
+         * The priority used to register this element is
+         * - either the real thread-priority for the real-time threads
+         * (i.e. threads with a priority lower than MAX_RT_PRIO)
+         * - or MAX_RT_PRIO for non-RT threads.
+         * Thus, all RT-threads are woken first in priority order, and
+         * the others are woken last, in FIFO order.
+         */
+        prio = min(current->normal_prio, MAX_RT_PRIO);
+
+        plist_node_init(&q->list, prio);
+#ifdef CONFIG_DEBUG_PI_LIST
+        q->list.plist.lock = &hb->lock;
+#endif
+        plist_add(&q->list, &hb->chain);
         q->task = current;
         spin_unlock(&hb->lock);
 }
@@ -949,8 +1291,8 @@ static int unqueue_me(struct futex_q *q)
                         spin_unlock(lock_ptr);
                         goto retry;
                 }
-                WARN_ON(list_empty(&q->list));
-                list_del(&q->list);
+                WARN_ON(plist_node_empty(&q->list));
+                plist_del(&q->list, &q->list.plist);
 
                 BUG_ON(q->pi_state);
 
@@ -964,39 +1306,104 @@ static int unqueue_me(struct futex_q *q)
 
 /*
  * PI futexes can not be requeued and must remove themself from the
- * hash bucket. The hash bucket lock is held on entry and dropped here.
+ * hash bucket. The hash bucket lock (i.e. lock_ptr) is held on entry
+ * and dropped here.
  */
-static void unqueue_me_pi(struct futex_q *q, struct futex_hash_bucket *hb)
+static void unqueue_me_pi(struct futex_q *q)
 {
-        WARN_ON(list_empty(&q->list));
-        list_del(&q->list);
+        WARN_ON(plist_node_empty(&q->list));
+        plist_del(&q->list, &q->list.plist);
 
         BUG_ON(!q->pi_state);
         free_pi_state(q->pi_state);
         q->pi_state = NULL;
 
-        spin_unlock(&hb->lock);
+        spin_unlock(q->lock_ptr);
 
         drop_futex_key_refs(&q->key);
 }
 
+/*
+ * Fixup the pi_state owner with current.
+ *
+ * The cur->mm semaphore must be  held, it is released at return of this
+ * function.
+ */
+static int fixup_pi_state_owner(u32 __user *uaddr, struct rw_semaphore *fshared,
+                                struct futex_q *q,
+                                struct futex_hash_bucket *hb,
+                                struct task_struct *curr)
+{
+        u32 newtid = curr->pid | FUTEX_WAITERS;
+        struct futex_pi_state *pi_state = q->pi_state;
+        u32 uval, curval, newval;
+        int ret;
+
+        /* Owner died? */
+        if (pi_state->owner != NULL) {
+                spin_lock_irq(&pi_state->owner->pi_lock);
+                WARN_ON(list_empty(&pi_state->list));
+                list_del_init(&pi_state->list);
+                spin_unlock_irq(&pi_state->owner->pi_lock);
+        } else
+                newtid |= FUTEX_OWNER_DIED;
+
+        pi_state->owner = curr;
+
+        spin_lock_irq(&curr->pi_lock);
+        WARN_ON(!list_empty(&pi_state->list));
+        list_add(&pi_state->list, &curr->pi_state_list);
+        spin_unlock_irq(&curr->pi_lock);
+
+        /* Unqueue and drop the lock */
+        unqueue_me_pi(q);
+        if (fshared)
+                up_read(fshared);
+        /*
+         * We own it, so we have to replace the pending owner
+         * TID. This must be atomic as we have preserve the
+         * owner died bit here.
+         */
+        ret = get_user(uval, uaddr);
+        while (!ret) {
+                newval = (uval & FUTEX_OWNER_DIED) | newtid;
+                newval |= (uval & FUTEX_WAITER_REQUEUED);
+                curval = futex_atomic_cmpxchg_inatomic(uaddr,
+                                                       uval, newval);
+                if (curval == -EFAULT)
+                        ret = -EFAULT;
+                if (curval == uval)
+                        break;
+                uval = curval;
+        }
+        return ret;
+}
+
+/*
+ * In case we must use restart_block to restart a futex_wait,
+ * we encode in the 'arg3' shared capability
+ */
+#define ARG3_SHARED  1
+
 static long futex_wait_restart(struct restart_block *restart);
-static int futex_wait_abstime(u32 __user *uaddr, u32 val,
-                        int timed, unsigned long abs_time)
+static int futex_wait(u32 __user *uaddr, struct rw_semaphore *fshared,
+                      u32 val, ktime_t *abs_time)
 {
         struct task_struct *curr = current;
         DECLARE_WAITQUEUE(wait, curr);
         struct futex_hash_bucket *hb;
         struct futex_q q;
-        unsigned long time_left = 0;
         u32 uval;
         int ret;
+        struct hrtimer_sleeper t, *to = NULL;
+        int rem = 0;
 
         q.pi_state = NULL;
  retry:
-        down_read(&curr->mm->mmap_sem);
+        if (fshared)
+                down_read(fshared);
 
-        ret = get_futex_key(uaddr, &q.key);
+        ret = get_futex_key(uaddr, fshared, &q.key);
         if (unlikely(ret != 0))
                 goto out_release_sem;
 
@@ -1019,8 +1426,8 @@ static int futex_wait_abstime(u32 __user *uaddr, u32 val,
          * a wakeup when *uaddr != val on entry to the syscall.  This is
          * rare, but normal.
          *
-         * We hold the mmap semaphore, so the mapping cannot have changed
-         * since we looked it up in get_futex_key.
+         * for shared futexes, we hold the mmap semaphore, so the mapping
+         * cannot have changed since we looked it up in get_futex_key.
          */
         ret = get_futex_value_locked(&uval, uaddr);
 
@@ -1031,7 +1438,8 @@ static int futex_wait_abstime(u32 __user *uaddr, u32 val,
                  * If we would have faulted, release mmap_sem, fault it in and
                  * start all over again.
                  */
-                up_read(&curr->mm->mmap_sem);
+                if (fshared)
+                        up_read(fshared);
 
                 ret = get_user(uval, uaddr);
 
@@ -1043,6 +1451,14 @@ static int futex_wait_abstime(u32 __user *uaddr, u32 val,
         if (uval != val)
                 goto out_unlock_release_sem;
 
+        /*
+         * This rt_mutex_waiter structure is prepared here and will
+         * be used only if this task is requeued from a normal futex to
+         * a PI-futex with futex_requeue_pi.
+         */
+        debug_rt_mutex_init_waiter(&q.waiter);
+        q.waiter.task = NULL;
+
         /* Only actually queue if *uaddr contained val.  */
         __queue_me(&q, hb);
 
@@ -1050,7 +1466,8 @@ static int futex_wait_abstime(u32 __user *uaddr, u32 val,
          * Now the futex is queued and we have checked the data, we
          * don't want to hold mmap_sem while we sleep.
          */
-        up_read(&curr->mm->mmap_sem);
+        if (fshared)
+                up_read(fshared);
 
         /*
          * There might have been scheduling since the queue_me(), as we
@@ -1065,23 +1482,33 @@ static int futex_wait_abstime(u32 __user *uaddr, u32 val,
         __set_current_state(TASK_INTERRUPTIBLE);
         add_wait_queue(&q.waiters, &wait);
         /*
-         * !list_empty() is safe here without any lock.
+         * !plist_node_empty() is safe here without any lock.
          * q.lock_ptr != 0 is not safe, because of ordering against wakeup.
          */
-        time_left = 0;
-        if (likely(!list_empty(&q.list))) {
-                unsigned long rel_time;
-
-                if (timed) {
-                        unsigned long now = jiffies;
-                        if (time_after(now, abs_time))
-                                rel_time = 0;
-                        else
-                                rel_time = abs_time - now;
-                } else
-                        rel_time = MAX_SCHEDULE_TIMEOUT;
+        if (likely(!plist_node_empty(&q.list))) {
+                if (!abs_time)
+                        schedule();
+                else {
+                        to = &t;
+                        hrtimer_init(&t.timer, CLOCK_MONOTONIC, HRTIMER_MODE_ABS);
+                        hrtimer_init_sleeper(&t, current);
+                        t.timer.expires = *abs_time;
 
-                time_left = schedule_timeout(rel_time);
+                        hrtimer_start(&t.timer, t.timer.expires, HRTIMER_MODE_ABS);
+
+                        /*
+                         * the timer could have already expired, in which
+                         * case current would be flagged for rescheduling.
+                         * Don't bother calling schedule.
+                         */
+                        if (likely(t.task))
+                                schedule();
+
+                        hrtimer_cancel(&t.timer);
+
+                        /* Flag if a timeout occured */
+                        rem = (t.task == NULL);
+                }
         }
         __set_current_state(TASK_RUNNING);
 
@@ -1090,17 +1517,80 @@ static int futex_wait_abstime(u32 __user *uaddr, u32 val,
          * we are the only user of it.
          */
 
+        if (q.pi_state) {
+                /*
+                 * We were woken but have been requeued on a PI-futex.
+                 * We have to complete the lock acquisition by taking
+                 * the rtmutex.
+                 */
+
+                struct rt_mutex *lock = &q.pi_state->pi_mutex;
+
+                spin_lock(&lock->wait_lock);
+                if (unlikely(q.waiter.task)) {
+                        remove_waiter(lock, &q.waiter);
+                }
+                spin_unlock(&lock->wait_lock);
+
+                if (rem)
+                        ret = -ETIMEDOUT;
+                else
+                        ret = rt_mutex_timed_lock(lock, to, 1);
+
+                if (fshared)
+                        down_read(fshared);
+                spin_lock(q.lock_ptr);
+
+                /*
+                 * Got the lock. We might not be the anticipated owner if we
+                 * did a lock-steal - fix up the PI-state in that case.
+                 */
+                if (!ret && q.pi_state->owner != curr) {
+                        /*
+                         * We MUST play with the futex we were requeued on,
+                         * NOT the current futex.
+                         * We can retrieve it from the key of the pi_state
+                         */
+                        uaddr = q.pi_state->key.uaddr;
+
+                        /* mmap_sem and hash_bucket lock are unlocked at
+                           return of this function */
+                        ret = fixup_pi_state_owner(uaddr, fshared,
+                                                   &q, hb, curr);
+                } else {
+                        /*
+                         * Catch the rare case, where the lock was released
+                         * when we were on the way back before we locked
+                         * the hash bucket.
+                         */
+                        if (ret && q.pi_state->owner == curr) {
+                                if (rt_mutex_trylock(&q.pi_state->pi_mutex))
+                                        ret = 0;
+                        }
+                        /* Unqueue and drop the lock */
+                        unqueue_me_pi(&q);
+                        if (fshared)
+                                up_read(fshared);
+                }
+
+                debug_rt_mutex_free_waiter(&q.waiter);
+
+                return ret;
+        }
+
+        debug_rt_mutex_free_waiter(&q.waiter);
+
         /* If we were woken (and unqueued), we succeeded, whatever. */
         if (!unqueue_me(&q))
                 return 0;
-        if (time_left == 0)
+        if (rem)
                 return -ETIMEDOUT;
 
         /*
          * We expect signal_pending(current), but another thread may
          * have handled it for us already.
          */
-        if (time_left == MAX_SCHEDULE_TIMEOUT)
+        if (!abs_time)
                 return -ERESTARTSYS;
         else {
                 struct restart_block *restart;
@@ -1108,8 +1598,10 @@ static int futex_wait_abstime(u32 __user *uaddr, u32 val,
                 restart->fn = futex_wait_restart;
                 restart->arg0 = (unsigned long)uaddr;
                 restart->arg1 = (unsigned long)val;
-                restart->arg2 = (unsigned long)timed;
-                restart->arg3 = abs_time;
+                restart->arg2 = (unsigned long)abs_time;
+                restart->arg3 = 0;
+                if (fshared)
+                        restart->arg3 |= ARG3_SHARED;
                 return -ERESTART_RESTARTBLOCK;
         }
 
@@ -1117,65 +1609,111 @@ static int futex_wait_abstime(u32 __user *uaddr, u32 val,
         queue_unlock(&q, hb);
 
  out_release_sem:
-        up_read(&curr->mm->mmap_sem);
+        if (fshared)
+                up_read(fshared);
         return ret;
 }
 
-static int futex_wait(u32 __user *uaddr, u32 val, unsigned long rel_time)
-{
-        int timed = (rel_time != MAX_SCHEDULE_TIMEOUT);
-        return futex_wait_abstime(uaddr, val, timed, jiffies+rel_time);
-}
 
 static long futex_wait_restart(struct restart_block *restart)
 {
         u32 __user *uaddr = (u32 __user *)restart->arg0;
         u32 val = (u32)restart->arg1;
-        int timed = (int)restart->arg2;
-        unsigned long abs_time = restart->arg3;
+        ktime_t *abs_time = (ktime_t *)restart->arg2;
+        struct rw_semaphore *fshared = NULL;
 
         restart->fn = do_no_restart_syscall;
-        return (long)futex_wait_abstime(uaddr, val, timed, abs_time);
+        if (restart->arg3 & ARG3_SHARED)
+                fshared = &current->mm->mmap_sem;
+        return (long)futex_wait(uaddr, fshared, val, abs_time);
 }
 
 
+static void set_pi_futex_owner(struct futex_hash_bucket *hb,
+                               union futex_key *key, struct task_struct *p)
+{
+        struct plist_head *head;
+        struct futex_q *this, *next;
+        struct futex_pi_state *pi_state = NULL;
+        struct rt_mutex *lock;
+
+        /* Search a waiter that should already exists */
+
+        head = &hb->chain;
+
+        plist_for_each_entry_safe(this, next, head, list) {
+                if (match_futex (&this->key, key)) {
+                        pi_state = this->pi_state;
+                        break;
+                }
+        }
+
+        BUG_ON(!pi_state);
+
+        /* set p as pi_state's owner */
+        lock = &pi_state->pi_mutex;
+
+        spin_lock(&lock->wait_lock);
+        spin_lock_irq(&p->pi_lock);
+
+        list_add(&pi_state->list, &p->pi_state_list);
+        pi_state->owner = p;
+
+
+        /* set p as pi_mutex's owner */
+        debug_rt_mutex_proxy_lock(lock, p);
+        WARN_ON(rt_mutex_owner(lock));
+        rt_mutex_set_owner(lock, p, 0);
+        rt_mutex_deadlock_account_lock(lock, p);
+
+        plist_add(&rt_mutex_top_waiter(lock)->pi_list_entry,
+                  &p->pi_waiters);
+        __rt_mutex_adjust_prio(p);
+
+        spin_unlock_irq(&p->pi_lock);
+        spin_unlock(&lock->wait_lock);
+}
+
 /*
  * Userspace tried a 0 -> TID atomic transition of the futex value
  * and failed. The kernel side here does the whole locking operation:
  * if there are waiters then it will block, it does PI, etc. (Due to
  * races the kernel might see a 0 value of the futex too.)
  */
-static int futex_lock_pi(u32 __user *uaddr, int detect, unsigned long sec,
-                         long nsec, int trylock)
+static int futex_lock_pi(u32 __user *uaddr, struct rw_semaphore *fshared,
+                         int detect, ktime_t *time, int trylock)
 {
         struct hrtimer_sleeper timeout, *to = NULL;
         struct task_struct *curr = current;
         struct futex_hash_bucket *hb;
         u32 uval, newval, curval;
         struct futex_q q;
-        int ret, attempt = 0;
+        int ret, lock_held, attempt = 0;
 
         if (refill_pi_state_cache())
                 return -ENOMEM;
 
-        if (sec != MAX_SCHEDULE_TIMEOUT) {
+        if (time) {
                 to = &timeout;
                 hrtimer_init(&to->timer, CLOCK_REALTIME, HRTIMER_MODE_ABS);
                 hrtimer_init_sleeper(to, current);
-                to->timer.expires = ktime_set(sec, nsec);
+                to->timer.expires = *time;
         }
 
         q.pi_state = NULL;
  retry:
-        down_read(&curr->mm->mmap_sem);
+        if (fshared)
+                down_read(fshared);
 
-        ret = get_futex_key(uaddr, &q.key);
+        ret = get_futex_key(uaddr, fshared, &q.key);
         if (unlikely(ret != 0))
                 goto out_release_sem;
 
         hb = queue_lock(&q, -1, NULL);
 
  retry_locked:
+        lock_held = 0;
+
         /*
          * To avoid races, we attempt to take the lock here again
          * (by doing a 0 -> TID atomic cmpxchg), while holding all
@@ -1194,7 +1732,16 @@ static int futex_lock_pi(u32 __user *uaddr, int detect, unsigned long sec,
         if (unlikely((curval & FUTEX_TID_MASK) == current->pid)) {
                 if (!detect && 0)
                         force_sig(SIGKILL, current);
-                ret = -EDEADLK;
+                /*
+                 * Normally, this check is done in user space.
+                 * In case of requeue, the owner may attempt to lock this futex,
+                 * even if the ownership has already been given by the previous
+                 * waker.
+                 * In the usual case, this is a case of deadlock, but not in case
+                 * of REQUEUE_PI.
+                 */
+                if (!(curval & FUTEX_WAITER_REQUEUED))
+                        ret = -EDEADLK;
                 goto out_unlock_release_sem;
         }
 
@@ -1206,7 +1753,18 @@ static int futex_lock_pi(u32 __user *uaddr, int detect, unsigned long sec,
                 goto out_unlock_release_sem;
 
         uval = curval;
-        newval = uval | FUTEX_WAITERS;
+        /*
+         * In case of a requeue, check if there already is an owner
+         * If not, just take the futex.
+         */
+        if ((curval & FUTEX_WAITER_REQUEUED) && !(curval & FUTEX_TID_MASK)) {
+                /* set current as futex owner */
+                newval = curval | current->pid;
+                lock_held = 1;
+        } else
+                /* Set the WAITERS flag, so the owner will know it has someone
+                   to wake at next unlock */
+                newval = curval | FUTEX_WAITERS;
 
         pagefault_disable();
         curval = futex_atomic_cmpxchg_inatomic(uaddr, uval, newval);
@@ -1217,11 +1775,16 @@ static int futex_lock_pi(u32 __user *uaddr, int detect, unsigned long sec,
         if (unlikely(curval != uval))
                 goto retry_locked;
 
+        if (lock_held) {
+                set_pi_futex_owner(hb, &q.key, curr);
+                goto out_unlock_release_sem;
+        }
+
         /*
          * We dont have the lock. Look up the PI state (or create it if
          * we are the first waiter):
          */
-        ret = lookup_pi_state(uval, hb, &q);
+        ret = lookup_pi_state(uval, hb, &q.key, &q.pi_state);
 
         if (unlikely(ret)) {
                 /*
@@ -1263,7 +1826,8 @@ static int futex_lock_pi(u32 __user *uaddr, int detect, unsigned long sec,
          * Now the futex is queued and we have checked the data, we
          * don't want to hold mmap_sem while we sleep.
          */
-        up_read(&curr->mm->mmap_sem);
+        if (fshared)
+                up_read(fshared);
 
         WARN_ON(!q.pi_state);
         /*
@@ -1277,52 +1841,18 @@ static int futex_lock_pi(u32 __user *uaddr, int detect, unsigned long sec,
                 ret = ret ? 0 : -EWOULDBLOCK;
         }
 
-        down_read(&curr->mm->mmap_sem);
+        if (fshared)
+                down_read(fshared);
         spin_lock(q.lock_ptr);
 
         /*
          * Got the lock. We might not be the anticipated owner if we
          * did a lock-steal - fix up the PI-state in that case.
          */
-        if (!ret && q.pi_state->owner != curr) {
-                u32 newtid = current->pid | FUTEX_WAITERS;
-
-                /* Owner died? */
-                if (q.pi_state->owner != NULL) {
-                        spin_lock_irq(&q.pi_state->owner->pi_lock);
-                        WARN_ON(list_empty(&q.pi_state->list));
-                        list_del_init(&q.pi_state->list);
-                        spin_unlock_irq(&q.pi_state->owner->pi_lock);
-                } else
-                        newtid |= FUTEX_OWNER_DIED;
-
-                q.pi_state->owner = current;
-
-                spin_lock_irq(&current->pi_lock);
-                WARN_ON(!list_empty(&q.pi_state->list));
-                list_add(&q.pi_state->list, &current->pi_state_list);
-                spin_unlock_irq(&current->pi_lock);
-
-                /* Unqueue and drop the lock */
-                unqueue_me_pi(&q, hb);
-                up_read(&curr->mm->mmap_sem);
-                /*
-                 * We own it, so we have to replace the pending owner
-                 * TID. This must be atomic as we have preserve the
-                 * owner died bit here.
-                 */
-                ret = get_user(uval, uaddr);
-                while (!ret) {
-                        newval = (uval & FUTEX_OWNER_DIED) | newtid;
-                        curval = futex_atomic_cmpxchg_inatomic(uaddr,
-                                                               uval, newval);
-                        if (curval == -EFAULT)
-                                ret = -EFAULT;
-                        if (curval == uval)
-                                break;
-                        uval = curval;
-                }
-        } else {
+        if (!ret && q.pi_state->owner != curr)
+                /* mmap_sem is unlocked at return of this function */
+                ret = fixup_pi_state_owner(uaddr, fshared, &q, hb, curr);
+        else {
                 /*
                  * Catch the rare case, where the lock was released
                  * when we were on the way back before we locked
@@ -1333,8 +1863,9 @@ static int futex_lock_pi(u32 __user *uaddr, int detect, unsigned long sec,
                                 ret = 0;
                 }
                 /* Unqueue and drop the lock */
-                unqueue_me_pi(&q, hb);
-                up_read(&curr->mm->mmap_sem);
+                unqueue_me_pi(&q);
+                if (fshared)
+                        up_read(fshared);
         }
 
         if (!detect && ret == -EDEADLK && 0)
@@ -1346,7 +1877,8 @@ static int futex_lock_pi(u32 __user *uaddr, int detect, unsigned long sec,
         queue_unlock(&q, hb);
 
  out_release_sem:
-        up_read(&curr->mm->mmap_sem);
+        if (fshared)
+                up_read(fshared);
         return ret;
 
  uaddr_faulted:
@@ -1357,15 +1889,16 @@ static int futex_lock_pi(u32 __user *uaddr, int detect, unsigned long sec,
          * still holding the mmap_sem.
          */
         if (attempt++) {
-                if (futex_handle_fault((unsigned long)uaddr, attempt)) {
-                        ret = -EFAULT;
+                ret = futex_handle_fault((unsigned long)uaddr, fshared,
+                                         attempt);
+                if (ret)
                         goto out_unlock_release_sem;
-                }
                 goto retry_locked;
         }
 
         queue_unlock(&q, hb);
-        up_read(&curr->mm->mmap_sem);
+        if (fshared)
+                up_read(fshared);
 
         ret = get_user(uval, uaddr);
         if (!ret && (uval != -EFAULT))
@@ -1379,12 +1912,12 @@ static int futex_lock_pi(u32 __user *uaddr, int detect, unsigned long sec,
  * This is the in-kernel slowpath: we look up the PI state (if any),
  * and do the rt-mutex unlock.
  */
-static int futex_unlock_pi(u32 __user *uaddr)
+static int futex_unlock_pi(u32 __user *uaddr, struct rw_semaphore *fshared)
 {
         struct futex_hash_bucket *hb;
         struct futex_q *this, *next;
         u32 uval;
-        struct list_head *head;
+        struct plist_head *head;
         union futex_key key;
         int ret, attempt = 0;
 
@@ -1399,9 +1932,10 @@ retry:
         /*
          * First take all the futex related locks:
          */
-        down_read(&current->mm->mmap_sem);
+        if (fshared)
+                down_read(fshared);
 
-        ret = get_futex_key(uaddr, &key);
+        ret = get_futex_key(uaddr, fshared, &key);
         if (unlikely(ret != 0))
                 goto out;
 
@@ -1435,7 +1969,7 @@ retry_locked:
          */
         head = &hb->chain;
 
-        list_for_each_entry_safe(this, next, head, list) {
+        plist_for_each_entry_safe(this, next, head, list) {
                 if (!match_futex (&this->key, &key))
                         continue;
                 ret = wake_futex_pi(uaddr, uval, this);
@@ -1460,7 +1994,8 @@ retry_locked:
 out_unlock:
         spin_unlock(&hb->lock);
 out:
-        up_read(&current->mm->mmap_sem);
+        if (fshared)
+                up_read(fshared);
 
         return ret;
 
@@ -1472,15 +2007,16 @@ pi_faulted:
          * still holding the mmap_sem.
          */
         if (attempt++) {
-                if (futex_handle_fault((unsigned long)uaddr, attempt)) {
-                        ret = -EFAULT;
+                ret = futex_handle_fault((unsigned long)uaddr, fshared,
+                                         attempt);
+                if (ret)
                         goto out_unlock;
-                }
                 goto retry_locked;
         }
 
         spin_unlock(&hb->lock);
-        up_read(&current->mm->mmap_sem);
+        if (fshared)
+                up_read(fshared);
 
         ret = get_user(uval, uaddr);
         if (!ret && (uval != -EFAULT))
@@ -1509,10 +2045,10 @@ static unsigned int futex_poll(struct file *filp,
         poll_wait(filp, &q->waiters, wait);
 
         /*
-         * list_empty() is safe here without any lock.
+         * plist_node_empty() is safe here without any lock.
          * q->lock_ptr != 0 is not safe, because of ordering against wakeup.
          */
-        if (list_empty(&q->list))
+        if (plist_node_empty(&q->list))
                 ret = POLLIN | POLLRDNORM;
 
         return ret;
@@ -1532,6 +2068,7 @@ static int futex_fd(u32 __user *uaddr, int signal)
         struct futex_q *q;
         struct file *filp;
         int ret, err;
+        struct rw_semaphore *fshared;
         static unsigned long printk_interval;
 
         if (printk_timed_ratelimit(&printk_interval, 60 * 60 * 1000)) {
@@ -1573,11 +2110,12 @@ static int futex_fd(u32 __user *uaddr, int signal)
         }
         q->pi_state = NULL;
 
-        down_read(&current->mm->mmap_sem);
-        err = get_futex_key(uaddr, &q->key);
+        fshared = &current->mm->mmap_sem;
+        down_read(fshared);
+        err = get_futex_key(uaddr, fshared, &q->key);
 
         if (unlikely(err != 0)) {
-                up_read(&current->mm->mmap_sem);
+                up_read(fshared);
                 kfree(q);
                 goto error;
         }
@@ -1589,7 +2127,7 @@ static int futex_fd(u32 __user *uaddr, int signal)
         filp->private_data = q;
 
         queue_me(q, ret, filp);
-        up_read(&current->mm->mmap_sem);
+        up_read(fshared);
 
         /* Now we map fd to filp, so userspace can access it */
         fd_install(ret, filp);
@@ -1702,6 +2240,8 @@ retry:
                  * userspace.
                  */
                 mval = (uval & FUTEX_WAITERS) | FUTEX_OWNER_DIED;
+                /* Also keep the FUTEX_WAITER_REQUEUED flag if set */
+                mval |= (uval & FUTEX_WAITER_REQUEUED);
                 nval = futex_atomic_cmpxchg_inatomic(uaddr, uval, mval);
 
                 if (nval == -EFAULT)
@@ -1716,7 +2256,7 @@ retry:
                  */
                 if (!pi) {
                         if (uval & FUTEX_WAITERS)
-                                futex_wake(uaddr, 1);
+                                futex_wake(uaddr, &curr->mm->mmap_sem, 1);
                 }
         }
         return 0;
@@ -1772,7 +2312,8 @@ void exit_robust_list(struct task_struct *curr)
                 return;
 
         if (pending)
-                handle_futex_death((void __user *)pending + futex_offset, curr, pip);
+                handle_futex_death((void __user *)pending + futex_offset,
+                                   curr, pip);
 
         while (entry != &head->list) {
                 /*
@@ -1798,39 +2339,47 @@ void exit_robust_list(struct task_struct *curr)
         }
 }
 
-long do_futex(u32 __user *uaddr, int op, u32 val, unsigned long timeout,
+long do_futex(u32 __user *uaddr, int op, u32 val, ktime_t *timeout,
                 u32 __user *uaddr2, u32 val2, u32 val3)
 {
         int ret;
+        int cmd = op & FUTEX_CMD_MASK;
+        struct rw_semaphore *fshared = NULL;
+
+        if (!(op & FUTEX_PRIVATE_FLAG))
+                fshared = &current->mm->mmap_sem;
 
-        switch (op) {
+        switch (cmd) {
         case FUTEX_WAIT:
-                ret = futex_wait(uaddr, val, timeout);
+                ret = futex_wait(uaddr, fshared, val, timeout);
                 break;
         case FUTEX_WAKE:
-                ret = futex_wake(uaddr, val);
+                ret = futex_wake(uaddr, fshared, val);
                 break;
         case FUTEX_FD:
                 /* non-zero val means F_SETOWN(getpid()) & F_SETSIG(val) */
                 ret = futex_fd(uaddr, val);
                 break;
         case FUTEX_REQUEUE:
-                ret = futex_requeue(uaddr, uaddr2, val, val2, NULL);
+                ret = futex_requeue(uaddr, fshared, uaddr2, val, val2, NULL);
                 break;
         case FUTEX_CMP_REQUEUE:
-                ret = futex_requeue(uaddr, uaddr2, val, val2, &val3);
+                ret = futex_requeue(uaddr, fshared, uaddr2, val, val2, &val3);
                 break;
         case FUTEX_WAKE_OP:
-                ret = futex_wake_op(uaddr, uaddr2, val, val2, val3);
+                ret = futex_wake_op(uaddr, fshared, uaddr2, val, val2, val3);
                 break;
         case FUTEX_LOCK_PI:
-                ret = futex_lock_pi(uaddr, val, timeout, val2, 0);
+                ret = futex_lock_pi(uaddr, fshared, val, timeout, 0);
                 break;
         case FUTEX_UNLOCK_PI:
-                ret = futex_unlock_pi(uaddr);
+                ret = futex_unlock_pi(uaddr, fshared);
                 break;
         case FUTEX_TRYLOCK_PI:
-                ret = futex_lock_pi(uaddr, 0, timeout, val2, 1);
+                ret = futex_lock_pi(uaddr, fshared, 0, timeout, 1);
+                break;
+        case FUTEX_CMP_REQUEUE_PI:
+                ret = futex_requeue_pi(uaddr, fshared, uaddr2, val, val2, &val3);
                 break;
         default:
                 ret = -ENOSYS;
@@ -1843,29 +2392,30 @@ asmlinkage long sys_futex(u32 __user *uaddr, int op, u32 val,
                           struct timespec __user *utime, u32 __user *uaddr2,
                           u32 val3)
 {
-        struct timespec t;
-        unsigned long timeout = MAX_SCHEDULE_TIMEOUT;
+        struct timespec ts;
+        ktime_t t, *tp = NULL;
         u32 val2 = 0;
+        int cmd = op & FUTEX_CMD_MASK;
 
-        if (utime && (op == FUTEX_WAIT || op == FUTEX_LOCK_PI)) {
-                if (copy_from_user(&t, utime, sizeof(t)) != 0)
+        if (utime && (cmd == FUTEX_WAIT || cmd == FUTEX_LOCK_PI)) {
+                if (copy_from_user(&ts, utime, sizeof(ts)) != 0)
                         return -EFAULT;
-                if (!timespec_valid(&t))
+                if (!timespec_valid(&ts))
                         return -EINVAL;
-                if (op == FUTEX_WAIT)
-                        timeout = timespec_to_jiffies(&t) + 1;
-                else {
-                        timeout = t.tv_sec;
-                        val2 = t.tv_nsec;
-                }
+
+                t = timespec_to_ktime(ts);
+                if (cmd == FUTEX_WAIT)
+                        t = ktime_add(ktime_get(), t);
+                tp = &t;
         }
         /*
-         * requeue parameter in 'utime' if op == FUTEX_REQUEUE.
+         * requeue parameter in 'utime' if cmd == FUTEX_REQUEUE.
          */
-        if (op == FUTEX_REQUEUE || op == FUTEX_CMP_REQUEUE)
+        if (cmd == FUTEX_REQUEUE || cmd == FUTEX_CMP_REQUEUE
+            || cmd == FUTEX_CMP_REQUEUE_PI)
                 val2 = (u32) (unsigned long) utime;
 
-        return do_futex(uaddr, op, val, timeout, uaddr2, val2, val3);
+        return do_futex(uaddr, op, val, tp, uaddr2, val2, val3);
 }
 
 static int futexfs_get_sb(struct file_system_type *fs_type,
@@ -1895,7 +2445,7 @@ static int __init init(void)
         }
 
         for (i = 0; i < ARRAY_SIZE(futex_queues); i++) {
-                INIT_LIST_HEAD(&futex_queues[i].chain);
+                plist_head_init(&futex_queues[i].chain, &futex_queues[i].lock);
                 spin_lock_init(&futex_queues[i].lock);
         }
         return 0;