source: libcfa/src/concurrency/ready_queue.cfa @ dca5802

//
// Cforall Version 1.0.0 Copyright (C) 2019 University of Waterloo
//
// The contents of this file are covered under the licence agreement in the
// file "LICENCE" distributed with Cforall.
//
// ready_queue.cfa --
//
// Author           : Thierry Delisle
// Created On       : Mon Nov dd 16:29:18 2019
// Last Modified By :
// Last Modified On :
// Update Count     :
//

#define __cforall_thread__

#include "bits/defs.hfa"
#include "kernel_private.hfa"

#define _GNU_SOURCE
#include "stdlib.hfa"

static const size_t cache_line_size = 64;

// No overriden function, no environment variable, no define
// fall back to a magic number
#ifndef __CFA_MAX_PROCESSORS__
        #define __CFA_MAX_PROCESSORS__ 128
#endif

// returns the maximum number of processors the RWLock support
__attribute__((weak)) unsigned __max_processors() {
        const char * max_cores_s = getenv("CFA_MAX_PROCESSORS");
        if(!max_cores_s) {
                __cfaabi_dbg_print_nolock("No CFA_MAX_PROCESSORS in ENV");
                return __CFA_MAX_PROCESSORS__;
        }

        char * endptr = 0p;
        long int max_cores_l = strtol(max_cores_s, &endptr, 10);
        if(max_cores_l < 1 || max_cores_l > 65535) {
                __cfaabi_dbg_print_nolock("CFA_MAX_PROCESSORS out of range : %ld", max_cores_l);
                return __CFA_MAX_PROCESSORS__;
        }
        if('\0' != *endptr) {
                __cfaabi_dbg_print_nolock("CFA_MAX_PROCESSORS not a decimal number : %s", max_cores_s);
                return __CFA_MAX_PROCESSORS__;
        }

        return max_cores_l;
}

// Picks a random 1 bit in 'mask' according to random number 'rnum'.
static inline unsigned rand_bit(unsigned rnum, __cfa_readyQ_mask_t mask) {
#if defined( __i386 )
        static_assert(sizeof(mask) == 4);
        unsigned bit = mask ? rnum % __builtin_popcount(mask) : 0;
        #if !defined(__BMI2__)
                #error rand_bit not implemented for non __BMI2__ i386
        #else
                uint32_t picked = _pdep_u32(1ul << bit, mask);
                return picked ? __builtin_ctz(picked) : 0;
        #endif
#elif defined( __x86_64 )
        static_assert(sizeof(mask) == 8);
        unsigned bit = mask ? rnum % __builtin_popcountl(mask) : 0;
        #if !defined(__BMI2__)
                uint64_t v = mask;   // Input value to find position with rank r.
                unsigned int r = bit + 1;// Input: bit's desired rank [1-64].
                unsigned int s;      // Output: Resulting position of bit with rank r [1-64]
                uint64_t a, b, c, d; // Intermediate temporaries for bit count.
                unsigned int t;      // Bit count temporary.

                // Do a normal parallel bit count for a 64-bit integer,
                // but store all intermediate steps.
                a =  v - ((v >> 1) & ~0UL/3);
                b = (a & ~0UL/5) + ((a >> 2) & ~0UL/5);
                c = (b + (b >> 4)) & ~0UL/0x11;
                d = (c + (c >> 8)) & ~0UL/0x101;


                t = (d >> 32) + (d >> 48);
                // Now do branchless select!
                s  = 64;
                s -= ((t - r) & 256) >> 3; r -= (t & ((t - r) >> 8));
                t  = (d >> (s - 16)) & 0xff;
                s -= ((t - r) & 256) >> 4; r -= (t & ((t - r) >> 8));
                t  = (c >> (s - 8)) & 0xf;
                s -= ((t - r) & 256) >> 5; r -= (t & ((t - r) >> 8));
                t  = (b >> (s - 4)) & 0x7;
                s -= ((t - r) & 256) >> 6; r -= (t & ((t - r) >> 8));
                t  = (a >> (s - 2)) & 0x3;
                s -= ((t - r) & 256) >> 7; r -= (t & ((t - r) >> 8));
                t  = (v >> (s - 1)) & 0x1;
                s -= ((t - r) & 256) >> 8;
                return s - 1;
        #else
                uint64_t picked = _pdep_u64(1ul << bit, mask);
                return picked ? __builtin_ctzl(picked) : 0;
        #endif
#elif defined( __ARM_ARCH )
        #error rand_bit not implemented for arm
#else
        #error uknown hardware architecture
#endif
}


//-----------------------------------------------------------------------------
// Helpers used by extract
// (_mask_bitsidx() & X) returns a bit index valid for a __cfa_readyQ_mask_t, where X is any integer
static inline __cfa_readyQ_mask_t _mask_bitsidx () { return (8 * sizeof(__cfa_readyQ_mask_t)) - 1; }

// (X >> _mask_shiftidx()) retuns an index into an array of __cfa_readyQ_mask_t
static inline __cfa_readyQ_mask_t _mask_shiftidx() { return (8 * sizeof(__cfa_readyQ_mask_t)) - __builtin_clzl(_mask_bitsidx()); }


// Assuming a large bit mask represented as an array of __cfa_readyQ_mask_t
// Given an index into the large mask, returns the bit index and which __cfa_readyQ_mask_t index in the array
static inline [__cfa_readyQ_mask_t, __cfa_readyQ_mask_t] extract(__cfa_readyQ_mask_t idx) {
        __cfa_readyQ_mask_t word = idx >> _mask_bitsidx();
        __cfa_readyQ_mask_t bit  = idx &  _mask_shiftidx();
        return [bit, word];
}

//=======================================================================
// Cluster wide reader-writer lock
//=======================================================================
void  ?{}(__clusterRWLock_t & this) {
        this.max   = __max_processors();
        this.alloc = 0;
        this.ready = 0;
        this.lock  = false;
        this.data  = alloc(this.max);

        /*paranoid*/ verify( 0 == (((uintptr_t)(this.data    )) % 64) );
        /*paranoid*/ verify( 0 == (((uintptr_t)(this.data + 1)) % 64) );
        /*paranoid*/ verify(__atomic_is_lock_free(sizeof(this.alloc), &this.alloc));
        /*paranoid*/ verify(__atomic_is_lock_free(sizeof(this.ready), &this.ready));

}
void ^?{}(__clusterRWLock_t & this) {
        free(this.data);
}

void ?{}( __processor_id & this, struct processor * proc ) {
        this.handle = proc;
        this.lock   = false;
}

//=======================================================================
// Lock-Free registering/unregistering of threads
unsigned doregister( struct cluster * cltr, struct processor * proc ) with(cltr->ready_lock) {
        // Step - 1 : check if there is already space in the data
        uint_fast32_t s = ready;

        // Check among all the ready
        for(uint_fast32_t i = 0; i < s; i++) {
                processor * null = 0p; // Re-write every loop since compare thrashes it
                if( __atomic_load_n(&data[i].handle, (int)__ATOMIC_RELAXED) == null
                        && __atomic_compare_exchange_n( &data[i].handle, &null, proc, false, __ATOMIC_SEQ_CST, __ATOMIC_SEQ_CST)) {
                        /*paranoid*/ verify(i < ready);
                        /*paranoid*/ verify(__alignof__(data[i]) == cache_line_size);
                        /*paranoid*/ verify((((uintptr_t)&data[i]) % cache_line_size) == 0);
                        return i;
                }
        }

        if(max <= alloc) abort("Trying to create more than %ud processors", cltr->ready_lock.max);

        // Step - 2 : F&A to get a new spot in the array.
        uint_fast32_t n = __atomic_fetch_add(&alloc, 1, __ATOMIC_SEQ_CST);
        if(max <= n) abort("Trying to create more than %ud processors", cltr->ready_lock.max);

        // Step - 3 : Mark space as used and then publish it.
        __processor_id * storage = (__processor_id *)&data[n];
        (*storage){ proc };
        while(true) {
                unsigned copy = n;
                if( __atomic_load_n(&ready, __ATOMIC_RELAXED) == n
                        && __atomic_compare_exchange_n(&ready, &copy, n + 1, true, __ATOMIC_SEQ_CST, __ATOMIC_SEQ_CST))
                        break;
                asm volatile("pause");
        }

        // Return new spot.
        /*paranoid*/ verify(n < ready);
        /*paranoid*/ verify(__alignof__(data[n]) == cache_line_size);
        /*paranoid*/ verify((((uintptr_t)&data[n]) % cache_line_size) == 0);
        return n;
}

void unregister( struct cluster * cltr, struct processor * proc ) with(cltr->ready_lock) {
        unsigned id = proc->id;
        /*paranoid*/ verify(id < ready);
        /*paranoid*/ verify(proc == __atomic_load_n(&data[id].handle, __ATOMIC_RELAXED));
        __atomic_store_n(&data[id].handle, 0p, __ATOMIC_RELEASE);
}

//-----------------------------------------------------------------------
// Writer side : acquire when changing the ready queue, e.g. adding more
//  queues or removing them.
uint_fast32_t ready_mutate_lock( struct cluster & cltr ) with(cltr.ready_lock) {
        // Step 1 : lock global lock
        // It is needed to avoid processors that register mid Critical-Section
        //   to simply lock their own lock and enter.
        __atomic_acquire( &lock );

        // Step 2 : lock per-proc lock
        // Processors that are currently being registered aren't counted
        //   but can't be in read_lock or in the critical section.
        // All other processors are counted
        uint_fast32_t s = ready;
        for(uint_fast32_t i = 0; i < s; i++) {
                __atomic_acquire( &data[i].lock );
        }

        return s;
}

void ready_mutate_unlock( struct cluster & cltr, uint_fast32_t last_s ) with(cltr.ready_lock) {
        // Step 1 : release local locks
        // This must be done while the global lock is held to avoid
        //   threads that where created mid critical section
        //   to race to lock their local locks and have the writer
        //   immidiately unlock them
        // Alternative solution : return s in write_lock and pass it to write_unlock
        for(uint_fast32_t i = 0; i < last_s; i++) {
                verify(data[i].lock);
                __atomic_store_n(&data[i].lock, (bool)false, __ATOMIC_RELEASE);
        }

        // Step 2 : release global lock
        /*paranoid*/ assert(true == lock);
        __atomic_store_n(&lock, (bool)false, __ATOMIC_RELEASE);
}

//=======================================================================
// Intrusive Queue used by ready queue
//=======================================================================
// Get the head pointer (one before the first element) from the anchor
static inline thread_desc * head(const __intrusive_lane_t & this) {
        thread_desc * rhead = (thread_desc *)(
                (uintptr_t)( &this.before ) - offsetof( thread_desc, link )
        );
        /* paranoid */ verify(rhead);
        return rhead;
}

// Get the tail pointer (one after the last element) from the anchor
static inline thread_desc * tail(const __intrusive_lane_t & this) {
        thread_desc * rtail = (thread_desc *)(
                (uintptr_t)( &this.after ) - offsetof( thread_desc, link )
        );
        /* paranoid */ verify(rtail);
        return rtail;
}

// Ctor
void ?{}( __intrusive_lane_t & this ) {
        this.lock = false;
        this.last_id = -1u;
        this.count = 0u;

        this.before.link.prev = 0p;
        this.before.link.next = tail(this);
        this.before.link.ts   = 0;

        this.after .link.prev = head(this);
        this.after .link.next = 0p;
        this.after .link.ts   = 0;

        #if !defined(__CFA_NO_SCHED_STATS__)
                this.stat.diff = 0;
                this.stat.push = 0;
                this.stat.pop  = 0;
        #endif

        // We add a boat-load of assertions here because the anchor code is very fragile
        /* paranoid */ verify(((uintptr_t)( head(this) ) + offsetof( thread_desc, link )) == (uintptr_t)(&this.before));
        /* paranoid */ verify(((uintptr_t)( tail(this) ) + offsetof( thread_desc, link )) == (uintptr_t)(&this.after ));
        /* paranoid */ verify(head(this)->link.prev == 0p );
        /* paranoid */ verify(head(this)->link.next == tail(this) );
        /* paranoid */ verify(tail(this)->link.next == 0p );
        /* paranoid */ verify(tail(this)->link.prev == head(this) );
        /* paranoid */ verify(&head(this)->link.prev == &this.before.link.prev );
        /* paranoid */ verify(&head(this)->link.next == &this.before.link.next );
        /* paranoid */ verify(&tail(this)->link.prev == &this.after .link.prev );
        /* paranoid */ verify(&tail(this)->link.next == &this.after .link.next );
        /* paranoid */ verify(sizeof(__intrusive_lane_t) == 128);
        /* paranoid */ verify(sizeof(this) == 128);
        /* paranoid */ verify(__alignof__(__intrusive_lane_t) == 128);
        /* paranoid */ verify(__alignof__(this) == 128);
        /* paranoid */ verifyf(((intptr_t)(&this) % 128) == 0, "Expected address to be aligned %p %% 128 == %zd", &this, ((intptr_t)(&this) % 128));

        /* paranoid */ verifyf(_mask_shiftidx() == 6 , "%zu", _mask_shiftidx());
        /* paranoid */ verifyf(_mask_bitsidx () == 63, "%zu", _mask_bitsidx());
}

// Dtor is trivial
void ^?{}( __intrusive_lane_t & this ) {
        // Make sure the list is empty
        /* paranoid */ verify(head(this)->link.prev == 0p );
        /* paranoid */ verify(head(this)->link.next == tail(this) );
        /* paranoid */ verify(tail(this)->link.next == 0p );
        /* paranoid */ verify(tail(this)->link.prev == head(this) );
        /* paranoid */ verify(this.count == 0u );
}

// Push a thread onto this lane
// returns true of lane was empty before push, false otherwise
bool push(__intrusive_lane_t & this, thread_desc * node) {
        #if defined(__CFA_WITH_VERIFY__)
                /* paranoid */ verify(this.lock);
                /* paranoid */ verify(node->link.ts != 0);
                /* paranoid */ verify(node->link.next == 0p);
                /* paranoid */ verify(node->link.prev == 0p);

                this.count++;

                if(this.before.link.ts == 0l) {
                        /* paranoid */ verify(tail(this)->link.next == 0p);
                        /* paranoid */ verify(tail(this)->link.prev == head(this));
                        /* paranoid */ verify(head(this)->link.next == tail(this));
                        /* paranoid */ verify(head(this)->link.prev == 0p);
                }
        #endif

        // Get the relevant nodes locally
        thread_desc * tail = tail(this);
        thread_desc * prev = tail->link.prev;

        // Do the push
        node->link.next = tail;
        node->link.prev = prev;
        prev->link.next = node;
        tail->link.prev = node;

        // Update stats
        #if !defined(__CFA_NO_SCHED_STATS__)
                this.stat.diff++;
                this.stat.push++;
        #endif

        verify(node->link.next == tail(this));

        // Check if the queue used to be empty
        if(this.before.link.ts == 0l) {
                this.before.link.ts = node->link.ts;
                /* paranoid */ verify(node->link.prev == head(this));
                return true;
        }
        return false;
}

// Pop a thread from this lane (must be non-empty)
// returns popped
// returns true of lane was empty before push, false otherwise
[thread_desc *, bool] pop(__intrusive_lane_t & this) {
        /* paranoid */ verify(this.lock);
        /* paranoid */ verify(this.before.link.ts != 0ul);

        // Get anchors locally
        thread_desc * head = head(this);
        thread_desc * tail = tail(this);

        // Get the relevant nodes locally
        thread_desc * node = head->link.next;
        thread_desc * next = node->link.next;

        #if defined(__CFA_WITH_VERIFY__)
                this.count--;
                /* paranoid */ verify(node != tail);
                /* paranoid */ verify(node);
        #endif

        // Do the pop
        head->link.next = next;
        next->link.prev = head;
        node->link.[next, prev] = 0p;

        // Update head time stamp
        this.before.link.ts = next->link.ts;

        // Update stats
        #ifndef __CFA_NO_SCHED_STATS__
                this.stat.diff--;
                this.stat.pop ++;
        #endif

        // Check if we emptied list and return accordingly
        if(next == tail) {
                /* paranoid */ verify(this.before.link.ts == 0);
                /* paranoid */ verify(tail(this)->link.next == 0p);
                /* paranoid */ verify(tail(this)->link.prev == head(this));
                /* paranoid */ verify(head(this)->link.next == tail(this));
                /* paranoid */ verify(head(this)->link.prev == 0p);
                return [node, true];
        }
        else {
                /* paranoid */ verify(next->link.ts != 0);
                /* paranoid */ verify(this.before.link.ts != 0);
                return [node, false];
        }
}

// Check whether or not list is empty
static inline bool is_empty(__intrusive_lane_t & this) {
        verify( (this.before.link.ts == 0) == (this.count == 0) );
        return this.before.link.ts == 0;
}

// Return the timestamp
static inline unsigned long long ts(__intrusive_lane_t & this) {
        verify( this.before.link.ts == this.before.link.next->link.ts );
        return this.before.link.ts;
}

//=======================================================================
// Cforall Reqdy Queue used by ready queue
//=======================================================================

// Thread local mirror of ready queue statistics
#if !defined(__CFA_NO_STATISTICS__)
static __attribute__((aligned(128))) thread_local struct {
        struct {
                struct {
                        size_t attempt;
                        size_t success;
                } push;
                struct {
                        size_t maskrds;
                        size_t attempt;
                        size_t success;
                } pop;
        } pick;
        struct {
                size_t value;
                size_t count;
        } used;
} tls = {
        /* pick */{
                /* push */{ 0, 0 },
                /* pop  */{ 0, 0, 0 },
        },
        /* used */{ 0, 0 }
};
#endif

//-----------------------------------------------------------------------

void ?{}(__ready_queue_t & this) with (this) {
        used.count = 0;
        for( i ; __cfa_lane_mask_size ) {
                used.mask[i] = 0;
        }

        lanes.data = alloc(4);
        for( i; 4 ) {
                (lanes.data[i]){};
        }
        lanes.count = 4;

        #if !defined(__CFA_NO_STATISTICS__)
                global_stats.pick.push.attempt = 0;
                global_stats.pick.push.success = 0;
                global_stats.pick.pop .maskrds = 0;
                global_stats.pick.pop .attempt = 0;
                global_stats.pick.pop .success = 0;

                global_stats.used.value = 0;
                global_stats.used.count = 0;
        #endif
}

void ^?{}(__ready_queue_t & this) with (this) {
        verify( 4  == lanes.count );
        verify( 0  == used .count );

        for( i; 4 ) {
                ^(lanes.data[i]){};
        }
        free(lanes.data);


        #if defined(__CFA_WITH_VERIFY__)
                for( i ; __cfa_lane_mask_size ) {
                        assert( 0 == used.mask[i] );
                }
        #endif
}

//-----------------------------------------------------------------------
enum mask_strictness {
        STRICT,
        NOCHECK
};

// Set a given bit in the bit mask array
// strictness determines of the bit had to be cleared before
static inline void mask_set(__cfa_readyQ_mask_t * mask, unsigned index, mask_strictness strict) {
        // Extract the array and bit indexes
        __cfa_readyQ_mask_t word;
        __cfa_readyQ_mask_t bit;
        [bit, word] = extract(index);

        // Conditional check
        verifyf(
                strict == STRICT && // Conditional check if it was expected to be cleared
                ((mask[word] & (1ull << bit)) == 0),
                "Before set %llu:%llu (%u), %llx & %llx", word, bit, index, mask[word], (1ull << bit)
        );

        // Atomically set the bit
        __attribute__((unused)) bool ret = __atomic_bts(&mask[word], bit);

        // Conditional check
        verifyf(
                strict == STRICT && // Conditional check if it was expected to be cleared
                !ret,
                "Bit was not set but bts returned true"
        );

        // Unconditional check
        verifyf(
                (mask[word] & (1ull << bit)) != 0,
                "After set %llu:%llu (%u), %llx & %llx", word, bit, index, mask[word], (1ull << bit)
        );
}

static inline void mask_clear(__cfa_readyQ_mask_t * mask, unsigned index, mask_strictness strict) {
        // Extract the array and bit indexes
        __cfa_readyQ_mask_t word;
        __cfa_readyQ_mask_t bit;
        [bit, word] = extract(index);

        // Conditional check
        verifyf(
                strict == STRICT && // Conditional check if it was expected to be set
                ((mask[word] & (1ull << bit)) != 0),
                "Before clear %llu:%llu (%u), %llx & %llx", word, bit, index, mask[word], (1ull << bit)
        );

        // Atomically clear the bit
        __attribute__((unused)) bool ret = __atomic_btr(&mask[word], bit);

        // Conditional check
        verifyf(
                strict == STRICT && // Conditional check if it was expected to be cleared
                ret,
                "Bit was set but btr returned false"
        );

        // Unconditional check
        verifyf(
                (mask[word] & (1ull << bit)) == 0,
                "After clear %llu:%llu (%u), %llx & %llx", word, bit, index, mask[word], (1ull << bit)
        );
}

//-----------------------------------------------------------------------
__attribute__((hot)) bool push(struct cluster * cltr, struct thread_desc * thrd) with (cltr->ready_queue) {
        // write timestamp
        thrd->link.ts = rdtscl();

        // Try to pick a lane and lock it
        unsigned i;
        do {
                // Pick the index of a lane
                unsigned i = tls_rand() % lanes.count;

                #if !defined(__CFA_NO_STATISTICS__)
                        tls.pick.push.attempt++;
                #endif

                // If we can't lock it retry
        } while( !__atomic_try_acquire( &lanes.data[i].lock ) );

        #if defined(__CFA_WITH_VERIFY__)
                /* paranoid */ verify(lanes.data[i].last_id == -1u);
                /* paranoid */ lanes.data[i].last_id = kernelTLS.this_processor->id;
        #endif

        __attribute__((unused)) size_t num = __atomic_load_n( &used.count, __ATOMIC_RELAXED );
        bool first = false;

        // Actually push it
        bool lane_first = push(lanes.data[i], thrd);

        // If this lane used to be empty we need to do more
        if(lane_first) {
                // Update the global count
                size_t ret = __atomic_fetch_add( &used.count, 1z, __ATOMIC_SEQ_CST);

                // Check if the entire quue used to be empty
                first = (ret == 0);

                // Update the bit mask
                mask_set((__cfa_readyQ_mask_t *)used.mask, i, STRICT);
        }

        #if defined(__CFA_WITH_VERIFY__)
                /* paranoid */ verifyf( used.count <= lanes.count, "Non-empty count (%zu) exceeds actual count (%zu)\n", used.count, lanes.count );
                /* paranoid */ verifyf( lanes.data[i].last_id == kernelTLS.this_processor->id, "Expected last processor to lock queue %u to be %u, was %u\n", i, lanes.data[i].last_id, kernelTLS.this_processor->id );
                /* paranoid */ verifyf( lanes.data[i].lock, "List %u is not locked\n", i );
                /* paranoid */ lanes.data[i].last_id = -1u;
        #endif

        // Unlock and return
        __atomic_unlock( &lanes.data[i].lock );

        // Update statistics
        #if !defined(__CFA_NO_STATISTICS__)
                tls.pick.push.success++;
                tls.used.value += num;
                tls.used.count += 1;
        #endif

        // return whether or not the list was empty before this push
        return first;
}

//-----------------------------------------------------------------------
// Given 2 indexes, pick the list with the oldest push an try to pop from it
static struct thread_desc * try_pop(struct cluster * cltr, unsigned i, unsigned j) with (cltr->ready_queue) {
        #if !defined(__CFA_NO_STATISTICS__)
                tls.pick.pop.attempt++;
        #endif

        // Pick the bet list
        int w = i;
        if( __builtin_expect(!is_empty(lanes.data[j]), true) ) {
                w = (ts(lanes.data[i]) < ts(lanes.data[j])) ? i : j;
        }

        // Get relevant elements locally
        __intrusive_lane_t & lane = lanes.data[w];

        // If list looks empty retry
        if( is_empty(lane) ) return 0p;

        // If we can't get the lock retry
        if( !__atomic_try_acquire(&lane.lock) ) return 0p;

        #if defined(__CFA_WITH_VERIFY__)
                /* paranoid */ verify(lane.last_id == -1u);
                /* paranoid */ lane.last_id = kernelTLS.this_processor->id;
        #endif


        // If list is empty, unlock and retry
        if( is_empty(lane) ) {
                #if defined(__CFA_WITH_VERIFY__)
                        /* paranoid */ verify(lane.last_id == kernelTLS.this_processor->id);
                        /* paranoid */ lane.last_id = -1u;
                #endif

                __atomic_unlock(&lane.lock);
                return 0p;
        }

        // Actually pop the list
        struct thread_desc * thrd;
        bool emptied;
        [thrd, emptied] = pop(lane);

        /* paranoid */ verify(thrd);
        /* paranoid */ verify(lane.last_id == kernelTLS.this_processor->id);
        /* paranoid */ verify(lane.lock);

        // If this was the last element in the lane
        if(emptied) {
                // Update the global count
                __atomic_fetch_sub( &used.count, 1z, __ATOMIC_SEQ_CST);

                // Update the bit mask
                mask_clear((__cfa_readyQ_mask_t *)used.mask, w, STRICT);
        }

        #if defined(__CFA_WITH_VERIFY__)
                /* paranoid */ verify(lane.last_id == kernelTLS.this_processor->id);
                /* paranoid */ lane.last_id = -1u;
        #endif

        // For statistics, check the count before we release the lock
        #if !defined(__CFA_NO_STATISTICS__)
                int num = __atomic_load_n( &used.count, __ATOMIC_RELAXED );
        #endif

        // Unlock and return
        __atomic_unlock(&lane.lock);

        // Update statistics
        #if !defined(__CFA_NO_STATISTICS__)
                tls.pick.pop.success++;
                tls.used.value += num;
                tls.used.count += 1;
        #endif

        // return the popped thread
        return thrd;
}

// Pop from the ready queue from a given cluster
__attribute__((hot)) thread_desc * pop(struct cluster * cltr) with (cltr->ready_queue) {
        /* paranoid */ verify( lanes.count > 0 );

        // As long as the list is not empty, try finding a lane that isn't empty and pop from it
        while( __atomic_load_n( &used.count, __ATOMIC_RELAXED ) != 0) {
                #if !defined(__CFA_READQ_NO_BITMASK__)
                        // If using bit masks
                        #if !defined(__CFA_NO_SCHED_STATS__)
                                tls.pick.pop.maskrds++;
                        #endif

                        // Pick two lists at random
                        unsigned ri = tls_rand();
                        unsigned rj = tls_rand();

                        // Find which __cfa_readyQ_mask_t the two lists belong
                        unsigned num = ((__atomic_load_n( &lanes.count, __ATOMIC_RELAXED ) - 1) >> 6) + 1;
                        unsigned wdxi = (ri >> 6u) % num;
                        unsigned wdxj = (rj >> 6u) % num;

                        // Get the actual __cfa_readyQ_mask_t
                        size_t maski = __atomic_load_n( &used.mask[wdxi], __ATOMIC_RELAXED );
                        size_t maskj = __atomic_load_n( &used.mask[wdxj], __ATOMIC_RELAXED );

                        // If both of these masks are empty, retry
                        if(maski == 0 && maskj == 0) continue;

                        // Pick one of the non-zero bits in the masks and get the bit indexes
                        unsigned bi = rand_bit(ri, maski);
                        unsigned bj = rand_bit(rj, maskj);

                        // some checks
                        /* paranoid */ verifyf(bi < 64, "%zu %u", maski, bi);
                        /* paranoid */ verifyf(bj < 64, "%zu %u", maskj, bj);

                        // get the general list index
                        unsigned i = bi | (wdxi << 6);
                        unsigned j = bj | (wdxj << 6);

                        // some more checks
                        /* paranoid */ verifyf(i < lanes.count, "%u", wdxi << 6);
                        /* paranoid */ verifyf(j < lanes.count, "%u", wdxj << 6);

                        // try popping from the 2 picked lists
                        struct thread_desc * thrd = try_pop(cltr, i, j);
                        if(thrd) return thrd;
                #else
                        // Pick two lists at random
                        int i = tls_rand() % __atomic_load_n( &lanes.count, __ATOMIC_RELAXED );
                        int j = tls_rand() % __atomic_load_n( &lanes.count, __ATOMIC_RELAXED );

                        // try popping from the 2 picked lists
                        struct thread_desc * thrd = try_pop(cltr, i, j);
                        if(thrd) return thrd;
                #endif
        }

        // All lanes where empty return 0p
        return 0p;
}

//-----------------------------------------------------------------------

static void check( __ready_queue_t & q ) with (q) {
        #if defined(__CFA_WITH_VERIFY__)
                {
                        int idx = 0;
                        for( w ; __cfa_lane_mask_size ) {
                                for( b ; 8 * sizeof(__cfa_readyQ_mask_t) ) {
                                        bool is_empty = idx < lanes.count ? (ts(lanes.data[idx]) == 0) : true;
                                        bool should_be_empty = 0 == (used.mask[w] & (1z << b));
                                        assertf(should_be_empty == is_empty, "Inconsistent list %d, mask expect : %d, actual is got %d", idx, should_be_empty, (bool)is_empty);
                                        assert(__cfa_max_lanes > idx);
                                        idx++;
                                }
                        }
                }

                {
                        for( idx ; lanes.count ) {
                                __intrusive_lane_t & sl = lanes.data[idx];
                                assert(!lanes.data[idx].lock);

                                assert(head(sl)->link.prev == 0p );
                                assert(head(sl)->link.next->link.prev == head(sl) );
                                assert(tail(sl)->link.next == 0p );
                                assert(tail(sl)->link.prev->link.next == tail(sl) );

                                if(sl.before.link.ts == 0l) {
                                        assert(tail(sl)->link.next == 0p);
                                        assert(tail(sl)->link.prev == head(sl));
                                        assert(head(sl)->link.next == tail(sl));
                                        assert(head(sl)->link.prev == 0p);
                                }
                        }
                }
        #endif
}

// Call this function of the intrusive list was moved using memcpy
// fixes the list so that the pointers back to anchors aren't left dangling
static inline void fix(__intrusive_lane_t & ll) {
        // if the list is not empty then follow he pointer and fix its reverse
        if(!is_empty(ll)) {
                head(ll)->link.next->link.prev = head(ll);
                tail(ll)->link.prev->link.next = tail(ll);
        }
        // Otherwise just reset the list
        else {
                verify(tail(ll)->link.next == 0p);
                tail(ll)->link.prev = head(ll);
                head(ll)->link.next = tail(ll);
                verify(head(ll)->link.prev == 0p);
        }
}

// Grow the ready queue
void ready_queue_grow  (struct cluster * cltr) {
        // Lock the RWlock so no-one pushes/pops while we are changing the queue
        uint_fast32_t last_size = ready_mutate_lock( *cltr );

        __cfaabi_dbg_print_safe("Kernel : Growing ready queue\n");

        // Make sure that everything is consistent
        /* paranoid */ check( cltr->ready_queue );

        // grow the ready queue
        with( cltr->ready_queue ) {
                size_t ncount = lanes.count;

                // Check that we have some space left
                if(ncount + 4 >= __cfa_max_lanes) abort("Program attempted to create more than maximum number of Ready Queues (%zu)", __cfa_max_lanes);

                // increase count
                ncount += 4;

                // Allocate new array (uses realloc and memcpies the data)
                lanes.data = alloc(lanes.data, ncount);

                // Fix the moved data
                for( idx; (size_t)lanes.count ) {
                        fix(lanes.data[idx]);
                }

                // Construct new data
                for( idx; (size_t)lanes.count ~ ncount) {
                        (lanes.data[idx]){};
                }

                // Update original
                lanes.count = ncount;

                // fields in 'used' don't need to change when growing
        }

        // Make sure that everything is consistent
        /* paranoid */ check( cltr->ready_queue );

        __cfaabi_dbg_print_safe("Kernel : Growing ready queue done\n");

        // Unlock the RWlock
        ready_mutate_unlock( *cltr, last_size );
}

// Shrink the ready queue
void ready_queue_shrink(struct cluster * cltr) {
        // Lock the RWlock so no-one pushes/pops while we are changing the queue
        uint_fast32_t last_size = ready_mutate_lock( *cltr );

        __cfaabi_dbg_print_safe("Kernel : Shrinking ready queue\n");

        // Make sure that everything is consistent
        /* paranoid */ check( cltr->ready_queue );

        with( cltr->ready_queue ) {
                // Make sure that the total thread count stays the same
                #if defined(__CFA_WITH_VERIFY__)
                        size_t nthreads = 0;
                        for( idx; (size_t)lanes.count ) {
                                nthreads += lanes.data[idx].count;
                        }
                #endif

                size_t ocount = lanes.count;
                // Check that we have some space left
                if(ocount < 8) abort("Program attempted to destroy more Ready Queues than were created");

                // reduce the actual count so push doesn't use the old queues
                lanes.count -= 4;
                verify(ocount > lanes.count);

                // for printing count the number of displaced threads
                #if defined(__CFA_DEBUG_PRINT__)
                        __attribute__((unused)) size_t displaced = 0;
                #endif

                // redistribute old data
                for( idx; (size_t)lanes.count ~ ocount) {
                        // Lock is not strictly needed but makes checking invariants much easier
                        bool locked = __atomic_try_acquire(&lanes.data[idx].lock);
                        verify(locked);

                        // As long as we can pop from this lane to push the threads somewhere else in the queue
                        while(!is_empty(lanes.data[idx])) {
                                struct thread_desc * thrd;
                                __attribute__((unused)) bool _;
                                [thrd, _] = pop(lanes.data[idx]);

                                push(cltr, thrd);

                                // for printing count the number of displaced threads
                                #if defined(__CFA_DEBUG_PRINT__)
                                        displaced++;
                                #endif
                        }

                        mask_clear((__cfa_readyQ_mask_t *)used.mask, idx, NOCHECK);

                        // Unlock the lane
                        __atomic_unlock(&lanes.data[idx].lock);

                        // TODO print the queue statistics here

                        ^(lanes.data[idx]){};
                }

                __cfaabi_dbg_print_safe("Kernel : Shrinking ready queue displaced %zu threads\n", displaced);

                // recompute the used.count instead of maintaining it
                used.count = 0;
                for( i ; __cfa_lane_mask_size ) {
                        used.count += __builtin_popcountl(used.mask[i]);
                }

                // Allocate new array (uses realloc and memcpies the data)
                lanes.data = alloc(lanes.data, lanes.count);

                // Fix the moved data
                for( idx; (size_t)lanes.count ) {
                        fix(lanes.data[idx]);
                }

                // Make sure that the total thread count stayed the same
                #if defined(__CFA_WITH_VERIFY__)
                        for( idx; (size_t)lanes.count ) {
                                nthreads -= lanes.data[idx].count;
                        }
                        verifyf(nthreads == 0, "Shrinking changed number of threads");
                #endif
        }

        // Make sure that everything is consistent
        /* paranoid */ check( cltr->ready_queue );

        __cfaabi_dbg_print_safe("Kernel : Shrinking ready queue done\n");

        // Unlock the RWlock
        ready_mutate_unlock( *cltr, last_size );
}

//-----------------------------------------------------------------------

#if !defined(__CFA_NO_STATISTICS__)
void stats_tls_tally(struct cluster * cltr) with (cltr->ready_queue) {
        __atomic_fetch_add( &global_stats.pick.push.attempt, tls.pick.push.attempt, __ATOMIC_SEQ_CST );
        __atomic_fetch_add( &global_stats.pick.push.success, tls.pick.push.success, __ATOMIC_SEQ_CST );
        __atomic_fetch_add( &global_stats.pick.pop .maskrds, tls.pick.pop .maskrds, __ATOMIC_SEQ_CST );
        __atomic_fetch_add( &global_stats.pick.pop .attempt, tls.pick.pop .attempt, __ATOMIC_SEQ_CST );
        __atomic_fetch_add( &global_stats.pick.pop .success, tls.pick.pop .success, __ATOMIC_SEQ_CST );

        __atomic_fetch_add( &global_stats.used.value, tls.used.value, __ATOMIC_SEQ_CST );
        __atomic_fetch_add( &global_stats.used.count, tls.used.count, __ATOMIC_SEQ_CST );
}
#endif