diff --git a/vlib/sync/channel_fill_test.v b/vlib/sync/channel_fill_test.v
index cc68749810..2f02153b23 100644
--- a/vlib/sync/channel_fill_test.v
+++ b/vlib/sync/channel_fill_test.v
@@ -5,7 +5,7 @@ const (
 	queue_fill = 763
 )
 
-fn do_send(ch chan int, fin sync.Semaphore) {
+fn do_send(ch chan int, mut fin sync.Semaphore) {
 	for i in 0 .. queue_fill {
 		ch <- i
 	}
@@ -14,8 +14,8 @@ fn do_send(ch chan int, fin sync.Semaphore) {
 
 fn test_channel_len_cap() {
 	ch := chan int{cap: queue_len}
-	sem := sync.new_semaphore()
-	go do_send(ch, sem)
+	mut sem := sync.new_semaphore()
+	go do_send(ch, mut sem)
 	sem.wait()
 	assert ch.cap == queue_len
 	assert ch.len == queue_fill
diff --git a/vlib/sync/channel_opt_propagate_test.v b/vlib/sync/channel_opt_propagate_test.v
index 4bb4138aee..f9aea58d60 100644
--- a/vlib/sync/channel_opt_propagate_test.v
+++ b/vlib/sync/channel_opt_propagate_test.v
@@ -10,7 +10,7 @@ fn get_val_from_chan(ch chan i64) ?i64 {
 }
 
 // this function gets an array of channels for `i64`
-fn do_rec_calc_send(chs []chan i64, sem sync.Semaphore) {
+fn do_rec_calc_send(chs []chan i64, mut sem sync.Semaphore) {
 	mut msg := ''
 	for {
 		mut s := get_val_from_chan(chs[0]) or {
@@ -26,8 +26,8 @@ fn do_rec_calc_send(chs []chan i64, sem sync.Semaphore) {
 
 fn test_channel_array_mut() {
 	mut chs := [chan i64{}, chan i64{cap: 10}]
-	sem := sync.new_semaphore()
-	go do_rec_calc_send(chs, sem)
+	mut sem := sync.new_semaphore()
+	go do_rec_calc_send(chs, mut sem)
 	mut t := i64(100)
 	for _ in 0 .. num_iterations {
 		chs[0] <- t
diff --git a/vlib/sync/channel_select_3_test.v b/vlib/sync/channel_select_3_test.v
index c78dca5d60..26dda99aca 100644
--- a/vlib/sync/channel_select_3_test.v
+++ b/vlib/sync/channel_select_3_test.v
@@ -9,7 +9,7 @@ fn getint() int {
 	return 8
 }
 
-fn f1(ch1 chan int, ch2 chan St, ch3 chan int, ch4 chan int, ch5 chan int, sem sync.Semaphore) {
+fn f1(ch1 chan int, ch2 chan St, ch3 chan int, ch4 chan int, ch5 chan int, mut sem sync.Semaphore) {
 	mut a := 5
 	select {
 		a = <-ch3 {
@@ -40,7 +40,7 @@ fn f1(ch1 chan int, ch2 chan St, ch3 chan int, ch4 chan int, ch5 chan int, sem s
 	sem.post()
 }
 
-fn f2(ch1 chan St, ch2 chan int, sem sync.Semaphore) {
+fn f2(ch1 chan St, ch2 chan int, mut sem sync.Semaphore) {
 	mut r := 23
 	for i in 0 .. 2 {
 		select {
@@ -66,7 +66,7 @@ fn test_select_blocks() {
 	ch3 := chan int{}
 	ch4 := chan int{}
 	ch5 := chan int{}
-	sem := sync.new_semaphore()
+	mut sem := sync.new_semaphore()
 	mut r := false
 	t := select {
 		b := <-ch1 {
@@ -79,7 +79,7 @@ fn test_select_blocks() {
 	}
 	assert r == true
 	assert t == true
-	go f2(ch2, ch3, sem)
+	go f2(ch2, ch3, mut sem)
 	n := <-ch3
 	assert n == 23
 	ch2 <- St{
@@ -87,7 +87,7 @@ fn test_select_blocks() {
 	}
 	sem.wait()
 	stopwatch := time.new_stopwatch({})
-	go f1(ch1, ch2, ch3, ch4, ch5, sem)
+	go f1(ch1, ch2, ch3, ch4, ch5, mut sem)
 	sem.wait()
 	elapsed_ms := f64(stopwatch.elapsed()) / time.millisecond
 	// https://docs.microsoft.com/en-us/windows-hardware/drivers/kernel/high-resolution-timers
diff --git a/vlib/sync/channels.v b/vlib/sync/channels.v
index a6e4291556..a038fff1ec 100644
--- a/vlib/sync/channels.v
+++ b/vlib/sync/channels.v
@@ -66,7 +66,7 @@ enum BufferElemStat {
 
 struct Subscription {
 mut:
-	sem Semaphore
+	sem &Semaphore
 	prev &&Subscription
 	nxt &Subscription
 }
@@ -77,14 +77,14 @@ enum Direction {
 }
 
 struct Channel {
-	writesem           Semaphore // to wake thread that wanted to write, but buffer was full
-	readsem            Semaphore // to wake thread that wanted to read, but buffer was empty
-	writesem_im        Semaphore
-	readsem_im         Semaphore
 	ringbuf            byteptr // queue for buffered channels
 	statusbuf          byteptr // flags to synchronize write/read in ringbuf
 	objsize            u32
 mut: // atomic
+	writesem           Semaphore // to wake thread that wanted to write, but buffer was full
+	readsem            Semaphore // to wake thread that wanted to read, but buffer was empty
+	writesem_im        Semaphore
+	readsem_im         Semaphore
 	write_adr          C.atomic_uintptr_t // if != NULL the next obj can be written here without wait
 	read_adr           C.atomic_uintptr_t // if != NULL an obj can be read from here without wait
 	adr_read           C.atomic_uintptr_t // used to identify origin of writesem
@@ -113,11 +113,7 @@ fn new_channel_st(n u32, st u32) &Channel {
 	rsem := if n > 0 { u32(0) } else { 1 }
 	rbuf := if n > 0 { malloc(int(n * st)) } else { byteptr(0) }
 	sbuf := if n > 0 { vcalloc(int(n * 2)) } else { byteptr(0) }
-	return &Channel{
-		writesem: new_semaphore_init(wsem)
-		readsem:  new_semaphore_init(rsem)
-		writesem_im: new_semaphore()
-		readsem_im: new_semaphore()
+	mut ch := &Channel{
 		objsize: st
 		cap: n
 		write_free: n
@@ -127,6 +123,11 @@ fn new_channel_st(n u32, st u32) &Channel {
 		write_subscriber: 0
 		read_subscriber: 0
 	}
+	ch.writesem.init(wsem)
+	ch.readsem.init(rsem)
+	ch.writesem_im.init(0)
+	ch.readsem_im.init(0)
+	return ch
 }
 
 pub fn (mut ch Channel) close() {
@@ -528,9 +529,10 @@ pub fn channel_select(mut channels []&Channel, dir []Direction, mut objrefs []vo
 	assert channels.len == dir.len
 	assert dir.len == objrefs.len
 	mut subscr := []Subscription{len: channels.len}
-	sem := new_semaphore()
+	mut sem := Semaphore{}
+	sem.init(0)
 	for i, ch in channels {
-		subscr[i].sem = sem
+		subscr[i].sem = &sem
 		if dir[i] == .push {
 			mut null16 := u16(0)
 			for !C.atomic_compare_exchange_weak_u16(&ch.write_sub_mtx, &null16, u16(1)) {
diff --git a/vlib/sync/sync_default.c.v b/vlib/sync/sync_default.c.v
new file mode 100644
index 0000000000..744fb15e15
--- /dev/null
+++ b/vlib/sync/sync_default.c.v
@@ -0,0 +1,134 @@
+// Copyright (c) 2019-2021 Alexander Medvednikov. All rights reserved.
+// Use of this source code is governed by an MIT license
+// that can be found in the LICENSE file.
+module sync
+
+import time
+
+#flag -lpthread
+#include <semaphore.h>
+
+[trusted]
+fn C.pthread_mutex_init(voidptr, voidptr) int
+fn C.pthread_mutex_lock(voidptr) int
+fn C.pthread_mutex_unlock(voidptr) int
+fn C.pthread_mutex_destroy(voidptr) int
+fn C.pthread_rwlockattr_init(voidptr) int
+fn C.pthread_rwlockattr_setkind_np(voidptr, int) int
+fn C.pthread_rwlockattr_setpshared(voidptr, int) int
+fn C.pthread_rwlock_init(voidptr, voidptr) int
+fn C.pthread_rwlock_rdlock(voidptr) int
+fn C.pthread_rwlock_wrlock(voidptr) int
+fn C.pthread_rwlock_unlock(voidptr) int
+fn C.sem_init(voidptr, int, u32) int
+fn C.sem_post(voidptr) int
+fn C.sem_wait(voidptr) int
+fn C.sem_trywait(voidptr) int
+fn C.sem_timedwait(voidptr, voidptr) int
+fn C.sem_destroy(voidptr) int
+
+// [init_with=new_mutex] // TODO: implement support for this struct attribute, and disallow Mutex{} from outside the sync.new_mutex() function.
+pub struct Mutex {
+	mutex C.pthread_mutex_t
+}
+
+pub struct RwMutex {
+	mutex C.pthread_rwlock_t
+}
+
+struct RwMutexAttr {
+	attr C.pthread_rwlockattr_t
+}
+
+struct Semaphore {
+	sem C.sem_t
+}
+
+pub fn new_mutex() &Mutex {
+	mut m := &Mutex{}
+	m.init()
+	return m
+}
+
+pub fn (mut m Mutex) init() {
+	C.pthread_mutex_init(&m.mutex, C.NULL)
+}
+
+pub fn new_rwmutex() &RwMutex {
+	mut m := &RwMutex{}
+	m.init()
+	return m
+}
+
+pub fn (mut m RwMutex) init() {
+	a := RwMutexAttr{}
+	C.pthread_rwlockattr_init(&a.attr)
+	// Give writer priority over readers
+	C.pthread_rwlockattr_setkind_np(&a.attr, C.PTHREAD_RWLOCK_PREFER_WRITER_NONRECURSIVE_NP)
+	C.pthread_rwlockattr_setpshared(&a.attr, C.PTHREAD_PROCESS_PRIVATE)
+	C.pthread_rwlock_init(&m.mutex, &a.attr)
+}
+
+// m_lock(), for *manual* mutex handling, since `lock` is a keyword
+pub fn (mut m Mutex) m_lock() {
+	C.pthread_mutex_lock(&m.mutex)
+}
+
+pub fn (mut m Mutex) unlock() {
+	C.pthread_mutex_unlock(&m.mutex)
+}
+
+// RwMutex has separate read- and write locks
+pub fn (mut m RwMutex) r_lock() {
+	C.pthread_rwlock_rdlock(&m.mutex)
+}
+
+pub fn (mut m RwMutex) w_lock() {
+	C.pthread_rwlock_wrlock(&m.mutex)
+}
+
+// Windows SRWLocks have different function to unlock
+// So provide two functions here, too, to have a common interface
+pub fn (mut m RwMutex) r_unlock() {
+	C.pthread_rwlock_unlock(&m.mutex)
+}
+
+pub fn (mut m RwMutex) w_unlock() {
+	C.pthread_rwlock_unlock(&m.mutex)
+}
+
+[inline]
+pub fn new_semaphore() &Semaphore {
+	return new_semaphore_init(0)
+}
+
+pub fn new_semaphore_init(n u32) &Semaphore {
+	mut sem := &Semaphore{}
+	sem.init(n)
+	return sem
+}
+
+pub fn (mut sem Semaphore) init(n u32) {
+	C.sem_init(&sem.sem, 0, n)
+}
+
+pub fn (mut sem Semaphore) post() {
+	C.sem_post(&sem.sem)
+}
+
+pub fn (mut sem Semaphore) wait() {
+	C.sem_wait(&sem.sem)
+}
+
+pub fn (mut sem Semaphore) try_wait() bool {
+	return C.sem_trywait(&sem.sem) == 0
+}
+
+pub fn (mut sem Semaphore) timed_wait(timeout time.Duration) bool {
+	t_spec := timeout.timespec()
+	return C.sem_timedwait(&sem.sem, &t_spec) == 0
+}
+
+pub fn (sem Semaphore) destroy() bool {
+	return C.sem_destroy(&sem.sem) == 0
+}
diff --git a/vlib/sync/sync_macos.c.v b/vlib/sync/sync_macos.c.v
new file mode 100644
index 0000000000..c92d389087
--- /dev/null
+++ b/vlib/sync/sync_macos.c.v
@@ -0,0 +1,212 @@
+// Copyright (c) 2019-2021 Alexander Medvednikov. All rights reserved.
+// Use of this source code is governed by an MIT license
+// that can be found in the LICENSE file.
+module sync
+
+import time
+
+#flag -lpthread
+#include <semaphore.h>
+
+[trusted]
+fn C.pthread_mutex_init(voidptr, voidptr) int
+fn C.pthread_mutex_lock(voidptr) int
+fn C.pthread_mutex_unlock(voidptr) int
+fn C.pthread_mutex_destroy(voidptr) int
+fn C.pthread_rwlockattr_init(voidptr) int
+fn C.pthread_rwlockattr_setkind_np(voidptr, int) int
+fn C.pthread_rwlockattr_setpshared(voidptr, int) int
+fn C.pthread_rwlock_init(voidptr, voidptr) int
+fn C.pthread_rwlock_rdlock(voidptr) int
+fn C.pthread_rwlock_wrlock(voidptr) int
+fn C.pthread_rwlock_unlock(voidptr) int
+fn C.pthread_condattr_init(voidptr) int
+fn C.pthread_condattr_setpshared(voidptr, int) int
+fn C.pthread_condattr_destroy(voidptr) int
+fn C.pthread_cond_init(voidptr, voidptr) int
+fn C.pthread_cond_signal(voidptr) int
+fn C.pthread_cond_wait(voidptr, voidptr) int
+fn C.pthread_cond_timedwait(voidptr, voidptr, voidptr) int
+fn C.pthread_cond_destroy(voidptr) int
+
+// [init_with=new_mutex] // TODO: implement support for this struct attribute, and disallow Mutex{} from outside the sync.new_mutex() function.
+pub struct Mutex {
+	mutex C.pthread_mutex_t
+}
+
+pub struct RwMutex {
+	mutex C.pthread_rwlock_t
+}
+
+struct RwMutexAttr {
+	attr C.pthread_rwlockattr_t
+}
+
+struct CondAttr {
+	attr C.pthread_condattr_t
+}
+
+/* MacOSX has no unnamed semaphores and no `timed_wait()` at all
+   so we emulate the behaviour with other devices */
+struct Semaphore {
+	mtx C.pthread_mutex_t
+	cond C.pthread_cond_t
+mut:
+	count u32
+}
+
+pub fn new_mutex() &Mutex {
+	mut m := &Mutex{}
+	m.init()
+	return m
+}
+
+pub fn (mut m Mutex) init() {
+	C.pthread_mutex_init(&m.mutex, C.NULL)
+}
+
+pub fn new_rwmutex() &RwMutex {
+	mut m := &RwMutex{}
+	m.init()
+	return m
+}
+
+pub fn (mut m RwMutex) init() {
+	a := RwMutexAttr{}
+	C.pthread_rwlockattr_init(&a.attr)
+	// Give writer priority over readers
+	C.pthread_rwlockattr_setkind_np(&a.attr, C.PTHREAD_RWLOCK_PREFER_WRITER_NONRECURSIVE_NP)
+	C.pthread_rwlockattr_setpshared(&a.attr, C.PTHREAD_PROCESS_PRIVATE)
+	C.pthread_rwlock_init(&m.mutex, &a.attr)
+}
+
+// m_lock(), for *manual* mutex handling, since `lock` is a keyword
+pub fn (mut m Mutex) m_lock() {
+	C.pthread_mutex_lock(&m.mutex)
+}
+
+pub fn (mut m Mutex) unlock() {
+	C.pthread_mutex_unlock(&m.mutex)
+}
+
+// RwMutex has separate read- and write locks
+pub fn (mut m RwMutex) r_lock() {
+	C.pthread_rwlock_rdlock(&m.mutex)
+}
+
+pub fn (mut m RwMutex) w_lock() {
+	C.pthread_rwlock_wrlock(&m.mutex)
+}
+
+// Windows SRWLocks have different function to unlock
+// So provide two functions here, too, to have a common interface
+pub fn (mut m RwMutex) r_unlock() {
+	C.pthread_rwlock_unlock(&m.mutex)
+}
+
+pub fn (mut m RwMutex) w_unlock() {
+	C.pthread_rwlock_unlock(&m.mutex)
+}
+
+[inline]
+pub fn new_semaphore() &Semaphore {
+	return new_semaphore_init(0)
+}
+
+pub fn new_semaphore_init(n u32) &Semaphore {
+	mut sem := &Semaphore{}
+	sem.init(n)
+	return sem
+}
+
+pub fn (mut sem Semaphore) init(n u32) {
+	C.atomic_store_u32(&sem.count, n)
+	C.pthread_mutex_init(&sem.mtx, C.NULL)
+	attr := CondAttr{}
+	C.pthread_condattr_init(&attr.attr)
+	C.pthread_condattr_setpshared(&attr.attr, C.PTHREAD_PROCESS_PRIVATE)
+	C.pthread_cond_init(&sem.cond, &attr.attr)
+	C.pthread_condattr_destroy(&attr.attr)
+}
+
+pub fn (mut sem Semaphore) post() {
+	mut c := C.atomic_load_u32(&sem.count)
+	for c > 1 {
+		if C.atomic_compare_exchange_weak_u32(&sem.count, &c, c+1) { return }
+	}
+	C.pthread_mutex_lock(&sem.mtx)
+	c = C.atomic_fetch_add_u32(&sem.count, 1)
+	if c == 0 {
+		C.pthread_cond_signal(&sem.cond)
+	}
+	C.pthread_mutex_unlock(&sem.mtx)
+}
+
+pub fn (mut sem Semaphore) wait() {
+	mut c := C.atomic_load_u32(&sem.count)
+	for c > 0 {
+		if C.atomic_compare_exchange_weak_u32(&sem.count, &c, c-1) { return }
+	}
+	C.pthread_mutex_lock(&sem.mtx)
+	c = C.atomic_load_u32(&sem.count)
+	for {
+		if c == 0 {
+			C.pthread_cond_wait(&sem.cond, &sem.mtx)
+			c = C.atomic_load_u32(&sem.count)
+		}
+		for c > 0 {
+			if C.atomic_compare_exchange_weak_u32(&sem.count, &c, c-1) {
+				if c > 1 {
+					C.pthread_cond_signal(&sem.cond)
+				}
+				goto unlock
+			}
+		}
+	}
+unlock:
+	C.pthread_mutex_unlock(&sem.mtx)
+}
+
+pub fn (mut sem Semaphore) try_wait() bool {
+	mut c := C.atomic_load_u32(&sem.count)
+	for c > 0 {
+		if C.atomic_compare_exchange_weak_u32(&sem.count, &c, c-1) { return true }
+	}
+	return false
+}
+
+pub fn (mut sem Semaphore) timed_wait(timeout time.Duration) bool {
+	mut c := C.atomic_load_u32(&sem.count)
+	for c > 0 {
+		if C.atomic_compare_exchange_weak_u32(&sem.count, &c, c-1) { return true }
+	}
+	C.pthread_mutex_lock(&sem.mtx)
+	t_spec := timeout.timespec()
+	mut res := 0
+	c = C.atomic_load_u32(&sem.count)
+	for {
+		if c == 0 {
+			res = C.pthread_cond_timedwait(&sem.cond, &sem.mtx, &t_spec)
+			if res == C.ETIMEDOUT {
+				goto unlock
+			}
+			c = C.atomic_load_u32(&sem.count)
+		}
+		for c > 0 {
+			if C.atomic_compare_exchange_weak_u32(&sem.count, &c, c-1) {
+				if c > 1 {
+					C.pthread_cond_signal(&sem.cond)
+				}
+				goto unlock
+			}
+		}
+	}
+unlock:
+	C.pthread_mutex_unlock(&sem.mtx)
+	return res == 0
+}
+
+pub fn (mut sem Semaphore) destroy() bool {
+	return C.pthread_cond_destroy(&sem.cond) == 0 &&
+		C.pthread_mutex_destroy(&sem.mtx) == 0
+}
diff --git a/vlib/sync/sync_nix.c.v b/vlib/sync/sync_nix.c.v
deleted file mode 100644
index 2972c04d1a..0000000000
--- a/vlib/sync/sync_nix.c.v
+++ /dev/null
@@ -1,153 +0,0 @@
-// Copyright (c) 2019-2021 Alexander Medvednikov. All rights reserved.
-// Use of this source code is governed by an MIT license
-// that can be found in the LICENSE file.
-module sync
-
-import time
-
-#flag -lpthread
-$if macos {
-	#include <dispatch/dispatch.h>
-}
-#include <semaphore.h>
-
-// [init_with=new_mutex] // TODO: implement support for this struct attribute, and disallow Mutex{} from outside the sync.new_mutex() function.
-[ref_only]
-pub struct Mutex {
-	mutex C.pthread_mutex_t
-}
-
-[ref_only]
-pub struct RwMutex {
-	mutex C.pthread_rwlock_t
-}
-
-[ref_only]
-struct RwMutexAttr {
-	attr C.pthread_rwlockattr_t
-}
-
-/* MacOSX has no unnamed semaphores and no `timed_wait()` at all
-   so we emulate the behaviour with other devices */
-// [ref_only]
-// struct MacOSX_Semaphore {
-// 	sem C.dispatch_semaphore_t
-// }
-
-[ref_only]
-struct PosixSemaphore {
-	sem C.sem_t
-}
-
-pub struct Semaphore {
-mut:
-	sem voidptr
-}
-
-pub fn new_mutex() &Mutex {
-	m := &Mutex{}
-	C.pthread_mutex_init(&m.mutex, C.NULL)
-	return m
-}
-
-pub fn new_rwmutex() &RwMutex {
-	m := &RwMutex{}
-	a := &RwMutexAttr{}
-	C.pthread_rwlockattr_init(&a.attr)
-	// Give writer priority over readers
-	C.pthread_rwlockattr_setkind_np(&a.attr, C.PTHREAD_RWLOCK_PREFER_WRITER_NONRECURSIVE_NP)
-	C.pthread_rwlockattr_setpshared(&a.attr, C.PTHREAD_PROCESS_PRIVATE)
-	C.pthread_rwlock_init(&m.mutex, &a.attr)
-	return m
-}
-
-// m_lock(), for *manual* mutex handling, since `lock` is a keyword
-pub fn (mut m Mutex) m_lock() {
-	C.pthread_mutex_lock(&m.mutex)
-}
-
-pub fn (mut m Mutex) unlock() {
-	C.pthread_mutex_unlock(&m.mutex)
-}
-
-// RwMutex has separate read- and write locks
-pub fn (mut m RwMutex) r_lock() {
-	C.pthread_rwlock_rdlock(&m.mutex)
-}
-
-pub fn (mut m RwMutex) w_lock() {
-	C.pthread_rwlock_wrlock(&m.mutex)
-}
-
-// Windows SRWLocks have different function to unlock
-// So provide two functions here, too, to have a common interface
-pub fn (mut m RwMutex) r_unlock() {
-	C.pthread_rwlock_unlock(&m.mutex)
-}
-
-pub fn (mut m RwMutex) w_unlock() {
-	C.pthread_rwlock_unlock(&m.mutex)
-}
-
-[inline]
-pub fn new_semaphore() Semaphore {
-	return new_semaphore_init(0)
-}
-
-pub fn new_semaphore_init(n u32) Semaphore {
-	$if macos {
-		s := Semaphore{
-			sem: C.dispatch_semaphore_create(n)
-		}
-		return s
-	} $else {
-		s := Semaphore{
-			sem: &PosixSemaphore{}
-		}
-		unsafe { C.sem_init(&&PosixSemaphore(s.sem).sem, 0, n) }
-		return s
-	}
-}
-
-pub fn (s Semaphore) post() {
-	$if macos {
-		C.dispatch_semaphore_signal(s.sem)
-	} $else {
-		unsafe { C.sem_post(&&PosixSemaphore(s.sem).sem) }
-	}
-}
-
-pub fn (s Semaphore) wait() {
-	$if macos {
-		C.dispatch_semaphore_wait(s.sem, C.DISPATCH_TIME_FOREVER)
-	} $else {
-		unsafe { C.sem_wait(&&PosixSemaphore(s.sem).sem) }
-	}
-}
-
-pub fn (s Semaphore) try_wait() bool {
-	$if macos {
-		return C.dispatch_semaphore_wait(s.sem, C.DISPATCH_TIME_NOW) == 0
-	} $else {
-		return unsafe { C.sem_trywait(&&PosixSemaphore(s.sem).sem) == 0 }
-	}
-}
-
-pub fn (s Semaphore) timed_wait(timeout time.Duration) bool {
-	$if macos {
-		return C.dispatch_semaphore_wait(s.sem, C.dispatch_time(C.DISPATCH_TIME_NOW, i64(timeout))) == 0
-	} $else {
-		t_spec := timeout.timespec()
-		return unsafe { C.sem_timedwait(&&PosixSemaphore(s.sem).sem, &t_spec) == 0 }
-	}
-}
-
-pub fn (s Semaphore) destroy() bool {
-	$if macos {
-		for s.try_wait() {}
-		C.dispatch_release(s.sem)
-		return true
-	} $else {
-		return unsafe { C.sem_destroy(&&PosixSemaphore(s.sem).sem) == 0 }
-	}
-}
diff --git a/vlib/sync/sync_windows.c.v b/vlib/sync/sync_windows.c.v
index 10da2ffe26..8f9cc31c28 100644
--- a/vlib/sync/sync_windows.c.v
+++ b/vlib/sync/sync_windows.c.v
@@ -5,6 +5,10 @@ module sync
 
 import time
 
+fn C.InitializeConditionVariable(voidptr)
+fn C.WakeConditionVariable(voidptr)
+fn C.SleepConditionVariableSRW(voidptr, voidptr, u32, u32) int
+
 // TODO: The suggestion of using CriticalSection instead of mutex
 // was discussed. Needs consideration.
 
@@ -15,90 +19,50 @@ type SHANDLE = voidptr
 
 //[init_with=new_mutex] // TODO: implement support for this struct attribute, and disallow Mutex{} from outside the sync.new_mutex() function.
 
-[ref_only]
+// `SRWLOCK` is much more performant that `Mutex` on Windows, so use that in both cases since we don't want to share with other processes
 pub struct Mutex {
 mut:
-	mx           MHANDLE    // mutex handle
-	state        MutexState // mutex state
-	cycle_wait   i64        // waiting cycles (implemented only with atomic)
-	cycle_woken  i64        // woken cycles    ^
-	reader_sem   u32        // reader semarphone
-	writer_sem   u32        // writer semaphones
+	mx C.SRWLOCK    // mutex handle
 }
 
-[ref_only]
 pub struct RwMutex {
 mut:
 	mx C.SRWLOCK    // mutex handle
 }
 
-pub struct Semaphore {
+struct Semaphore {
+	mtx C.SRWLOCK
+	cond C.CONDITION_VARIABLE
 mut:
-	sem SHANDLE
-}
-
-enum MutexState {
-	broken
-	waiting
-	released
-	abandoned
-	destroyed
+	count u32
 }
 
 pub fn new_mutex() &Mutex {
-	sm := &Mutex{}
-	unsafe {
-		mut m := sm
-		m.mx = MHANDLE(C.CreateMutex(0, false, 0))
-		if isnil(m.mx) {
-			m.state = .broken // handle broken and mutex state are broken
-			return sm
-		}
-	}
-	return sm
-}
-
-pub fn new_rwmutex() &RwMutex {
-	m := &RwMutex{}
-	C.InitializeSRWLock(&m.mx)
+	mut m := &Mutex{}
+	m.init()
 	return m
 }
 
+pub fn new_rwmutex() &RwMutex {
+	mut m := &RwMutex{}
+	m.init()
+	return m
+}
+
+pub fn (mut m Mutex) init() {
+	C.InitializeSRWLock(&m.mx)
+}
+
+pub fn (mut m RwMutex) init() {
+	C.InitializeSRWLock(&m.mx)
+}
+
 pub fn (mut m Mutex) m_lock() {
-	// if mutex handle not initalized
-	if isnil(m.mx) {
-		m.mx = MHANDLE(C.CreateMutex(0, false, 0))
-		if isnil(m.mx) {
-			m.state = .broken // handle broken and mutex state are broken
-			return
-		}
-	}
-	state := C.WaitForSingleObject(m.mx, C.INFINITE) // infinite wait
-	/* TODO fix match/enum combo
-	m.state = match state {
-		C.WAIT_ABANDONED { .abandoned }
-		C.WAIT_OBJECT_0  { .waiting }
-		else           { .broken }
-	}
-	*/
-	if state == C.WAIT_ABANDONED {
-		m.state = .abandoned
-	// FIXME Use C constant instead
-	} else if state == 0 /* C.WAIT_OBJECT_0 */ {
-		m.state = .waiting
-	} else {
-		m.state = .broken
-	}
+	C.AcquireSRWLockExclusive(&m.mx)
 }
 
 pub fn (mut m Mutex) unlock() {
-	if m.state == .waiting {
-		if C.ReleaseMutex(m.mx) {
-			m.state = .broken
-			return
-		}
-	}
-	m.state = .released
+	C.ReleaseSRWLockExclusive(&m.mx)
 }
 
 // RwMutex has separate read- and write locks
@@ -121,40 +85,103 @@ pub fn (mut m RwMutex) w_unlock() {
 }
 
 pub fn (mut m Mutex) destroy() {
-	if m.state == .waiting {
-		m.unlock() // unlock mutex before destroying
-	}
-	C.CloseHandle(m.mx)  // destroy mutex
-	m.state = .destroyed // setting up reference to invalid state
+	// nothing to do
 }
 
 [inline]
-pub fn new_semaphore() Semaphore {
+pub fn new_semaphore() &Semaphore {
 	return new_semaphore_init(0)
 }
 
-pub fn new_semaphore_init(n u32) Semaphore {
-	return Semaphore{
-		sem: SHANDLE(C.CreateSemaphore(0, n, C.INT32_MAX, 0))
+pub fn new_semaphore_init(n u32) &Semaphore {
+	mut sem := &Semaphore{}
+	sem.init(n)
+	return sem
+}
+
+pub fn (mut sem Semaphore) init(n u32) {
+	C.atomic_store_u32(&sem.count, n)
+	C.InitializeSRWLock(&sem.mtx)
+	C.InitializeConditionVariable(&sem.cond)
+}
+
+pub fn (mut sem Semaphore) post() {
+	mut c := C.atomic_load_u32(&sem.count)
+	for c > 1 {
+		if C.atomic_compare_exchange_weak_u32(&sem.count, &c, c+1) { return }
 	}
+	C.AcquireSRWLockExclusive(&sem.mtx)
+	c = C.atomic_fetch_add_u32(&sem.count, 1)
+	if c == 0 {
+		C.WakeConditionVariable(&sem.cond)
+	}
+	C.ReleaseSRWLockExclusive(&sem.mtx)
 }
 
-pub fn (s Semaphore) post() {
-	C.ReleaseSemaphore(s.sem, 1, 0)
+pub fn (mut sem Semaphore) wait() {
+	mut c := C.atomic_load_u32(&sem.count)
+	for c > 0 {
+		if C.atomic_compare_exchange_weak_u32(&sem.count, &c, c-1) { return }
+	}
+	C.AcquireSRWLockExclusive(&sem.mtx)
+	c = C.atomic_load_u32(&sem.count)
+	for {
+		if c == 0 {
+			C.SleepConditionVariableSRW(&sem.cond, &sem.mtx, C.INFINITE, 0)
+			c = C.atomic_load_u32(&sem.count)
+		}
+		for c > 0 {
+			if C.atomic_compare_exchange_weak_u32(&sem.count, &c, c-1) {
+				if c > 1 {
+					C.WakeConditionVariable(&sem.cond)
+				}
+				goto unlock
+			}
+		}
+	}
+unlock:
+	C.ReleaseSRWLockExclusive(&sem.mtx)
 }
 
-pub fn (s Semaphore) wait() {
-	C.WaitForSingleObject(s.sem, C.INFINITE)
+pub fn (mut sem Semaphore) try_wait() bool {
+	mut c := C.atomic_load_u32(&sem.count)
+	for c > 0 {
+		if C.atomic_compare_exchange_weak_u32(&sem.count, &c, c-1) { return true }
+	}
+	return false
 }
 
-pub fn (s Semaphore) try_wait() bool {
-	return C.WaitForSingleObject(s.sem, 0) == 0
-}
-
-pub fn (s Semaphore) timed_wait(timeout time.Duration) bool {
-	return C.WaitForSingleObject(s.sem, timeout / time.millisecond) == 0
+pub fn (mut sem Semaphore) timed_wait(timeout time.Duration) bool {
+	mut c := C.atomic_load_u32(&sem.count)
+	for c > 0 {
+		if C.atomic_compare_exchange_weak_u32(&sem.count, &c, c-1) { return true }
+	}
+	C.AcquireSRWLockExclusive(&sem.mtx)
+	t_ms := u32(timeout / time.millisecond)
+	mut res := 0
+	c = C.atomic_load_u32(&sem.count)
+	for {
+		if c == 0 {
+			res = C.SleepConditionVariableSRW(&sem.cond, &sem.mtx, t_ms, 0)
+			if res == 0 {
+				goto unlock
+			}
+			c = C.atomic_load_u32(&sem.count)
+		}
+		for c > 0 {
+			if C.atomic_compare_exchange_weak_u32(&sem.count, &c, c-1) {
+				if c > 1 {
+					C.WakeConditionVariable(&sem.cond)
+				}
+				goto unlock
+			}
+		}
+	}
+unlock:
+	C.ReleaseSRWLockExclusive(&sem.mtx)
+	return res != 0
 }
 
 pub fn (s Semaphore) destroy() bool {
-	return C.CloseHandle(s.sem) != 0
+	return true
 }
diff --git a/vlib/sync/waitgroup.v b/vlib/sync/waitgroup.v
index 29bca0b691..58d234d9d8 100644
--- a/vlib/sync/waitgroup.v
+++ b/vlib/sync/waitgroup.v
@@ -14,7 +14,7 @@ module sync
 struct WaitGroup {
 mut:
 	task_count       int // current task count
-	task_count_mutex &Mutex = &Mutex(0) // This mutex protects the task_count count in add()
+	task_count_mutex &Mutex // This mutex protects the task_count count in add()
 	wait_blocker     &Waiter = &Waiter(0) // This blocks the wait() until released by add()
 }
 
diff --git a/vlib/v/checker/tests/chan_ref.out b/vlib/v/checker/tests/chan_ref.out
index d120ce91a8..62637642e8 100644
--- a/vlib/v/checker/tests/chan_ref.out
+++ b/vlib/v/checker/tests/chan_ref.out
@@ -1,5 +1,5 @@
 vlib/v/checker/tests/chan_ref.vv:10:8: error: cannot push non-reference `St` on `chan &St`
-    8 | fn f(ch chan &St, sem sync.Semaphore) {
+    8 | fn f(ch chan &St, mut sem sync.Semaphore) {
     9 |     w := St{}
    10 |     ch <- w
       |           ^
diff --git a/vlib/v/checker/tests/chan_ref.vv b/vlib/v/checker/tests/chan_ref.vv
index f187781747..ad3dcec1a4 100644
--- a/vlib/v/checker/tests/chan_ref.vv
+++ b/vlib/v/checker/tests/chan_ref.vv
@@ -5,7 +5,7 @@ mut:
 	n int
 }
 
-fn f(ch chan &St, sem sync.Semaphore) {
+fn f(ch chan &St, mut sem sync.Semaphore) {
 	w := St{}
 	ch <- w
 	mut x := St{}
@@ -21,8 +21,8 @@ fn f(ch chan &St, sem sync.Semaphore) {
 
 fn main() {
 	c := chan &St{}
-	sem := sync.new_semaphore()
-	go f(c, sem)
+	mut sem := sync.new_semaphore()
+	go f(c, mut sem)
 	y := <-c
 	// this should fail
 	mut z := <-c
diff --git a/vlib/v/gen/array.v b/vlib/v/gen/array.v
index ee379ee9d0..91ebe19c11 100644
--- a/vlib/v/gen/array.v
+++ b/vlib/v/gen/array.v
@@ -17,7 +17,7 @@ fn (mut g Gen) array_init(it ast.ArrayInit) {
 		if g.is_shared {
 			mut shared_typ := it.typ.set_flag(.shared_f)
 			shared_styp = g.typ(shared_typ)
-			g.writeln('($shared_styp*)memdup(&($shared_styp){.val = ')
+			g.writeln('($shared_styp*)__dup_shared_array(&($shared_styp){.val = ')
 		} else {
 			g.write('($styp*)memdup(ADDR($styp, ')
 		}
@@ -95,7 +95,7 @@ fn (mut g Gen) array_init(it ast.ArrayInit) {
 		}
 		if is_amp {
 			if g.is_shared {
-				g.write(', .mtx = sync__new_rwmutex()}, sizeof($shared_styp))')
+				g.write('}, sizeof($shared_styp))')
 			} else {
 				g.write('), sizeof($styp))')
 			}
diff --git a/vlib/v/gen/cgen.v b/vlib/v/gen/cgen.v
index dc1a747a39..7342f8d6a3 100644
--- a/vlib/v/gen/cgen.v
+++ b/vlib/v/gen/cgen.v
@@ -324,6 +324,7 @@ pub fn cgen(files []ast.File, table &table.Table, pref &pref.Preferences) string
 	if g.shared_types.len > 0 {
 		b.writeln('\n// V shared types:')
 		b.write(g.shared_types.str())
+		b.write(c_concurrency_helpers)
 	}
 	if g.channel_definitions.len > 0 {
 		b.writeln('\n// V channel code:')
@@ -572,7 +573,12 @@ fn (mut g Gen) find_or_register_shared(t table.Type, base string) string {
 		return sh_typ
 	}
 	mtx_typ := 'sync__RwMutex'
-	g.shared_types.writeln('struct $sh_typ { $base val; $mtx_typ* mtx; };')
+	g.shared_types.writeln('struct $sh_typ { $base val; $mtx_typ mtx; };')
+	g.shared_types.writeln('static inline voidptr __dup${sh_typ}(voidptr src, int sz) {')
+	g.shared_types.writeln('\t$sh_typ* dest = memdup(src, sz);')
+	g.shared_types.writeln('\tsync__RwMutex_init(&dest->mtx);')
+	g.shared_types.writeln('\treturn dest;')
+	g.shared_types.writeln('}')
 	g.typedefs2.writeln('typedef struct $sh_typ $sh_typ;')
 	// println('registered shared type $sh_typ')
 	g.shareds << t_idx
@@ -2707,7 +2713,7 @@ fn (mut g Gen) expr(node ast.Expr) {
 				if g.is_shared {
 					mut shared_typ := node.typ.set_flag(.shared_f)
 					shared_styp = g.typ(shared_typ)
-					g.writeln('($shared_styp*)memdup(&($shared_styp){.val = ')
+					g.writeln('($shared_styp*)__dup_shared_map(&($shared_styp){.val = ')
 				} else {
 					styp = g.typ(node.typ)
 					g.write('($styp*)memdup(ADDR($styp, ')
@@ -2742,7 +2748,7 @@ fn (mut g Gen) expr(node ast.Expr) {
 				g.write('new_map_2(sizeof($key_typ_str), sizeof($value_typ_str), $hash_fn, $key_eq_fn, $clone_fn, $free_fn)')
 			}
 			if g.is_shared {
-				g.write(', .mtx = sync__new_rwmutex()}')
+				g.write('}')
 				if is_amp {
 					g.write(', sizeof($shared_styp))')
 				}
@@ -2763,7 +2769,7 @@ fn (mut g Gen) expr(node ast.Expr) {
 		}
 		ast.PostfixExpr {
 			if node.auto_locked != '' {
-				g.writeln('sync__RwMutex_w_lock($node.auto_locked->mtx);')
+				g.writeln('sync__RwMutex_w_lock(&$node.auto_locked->mtx);')
 			}
 			g.inside_map_postfix = true
 			g.expr(node.expr)
@@ -2771,7 +2777,7 @@ fn (mut g Gen) expr(node ast.Expr) {
 			g.write(node.op.str())
 			if node.auto_locked != '' {
 				g.writeln(';')
-				g.write('sync__RwMutex_w_unlock($node.auto_locked->mtx)')
+				g.write('sync__RwMutex_w_unlock(&$node.auto_locked->mtx)')
 			}
 		}
 		ast.PrefixExpr {
@@ -3022,7 +3028,7 @@ fn (mut g Gen) infix_expr(node ast.InfixExpr) {
 	// string + string, string == string etc
 	// g.infix_op = node.op
 	if node.auto_locked != '' {
-		g.writeln('sync__RwMutex_w_lock($node.auto_locked->mtx);')
+		g.writeln('sync__RwMutex_w_lock(&$node.auto_locked->mtx);')
 	}
 	left_type := g.unwrap_generic(node.left_type)
 	left_sym := g.table.get_type_symbol(left_type)
@@ -3412,7 +3418,7 @@ fn (mut g Gen) infix_expr(node ast.InfixExpr) {
 	}
 	if node.auto_locked != '' {
 		g.writeln(';')
-		g.write('sync__RwMutex_w_unlock($node.auto_locked->mtx)')
+		g.write('sync__RwMutex_w_unlock(&$node.auto_locked->mtx)')
 	}
 }
 
@@ -3423,7 +3429,7 @@ fn (mut g Gen) lock_expr(node ast.LockExpr) {
 		deref := if id.is_mut { '->' } else { '.' }
 		lock_prefix := if node.is_rlock { `r` } else { `w` }
 		lock_prefixes << lock_prefix // keep for unlock
-		g.writeln('sync__RwMutex_${lock_prefix:c}_lock($name${deref}mtx);')
+		g.writeln('sync__RwMutex_${lock_prefix:c}_lock(&$name${deref}mtx);')
 	}
 	g.stmts(node.stmts)
 	// unlock in reverse order
@@ -3432,7 +3438,7 @@ fn (mut g Gen) lock_expr(node ast.LockExpr) {
 		lock_prefix := lock_prefixes[i]
 		name := id.name
 		deref := if id.is_mut { '->' } else { '.' }
-		g.writeln('sync__RwMutex_${lock_prefix:c}_unlock($name${deref}mtx);')
+		g.writeln('sync__RwMutex_${lock_prefix:c}_unlock(&$name${deref}mtx);')
 	}
 }
 
@@ -4715,7 +4721,7 @@ fn (mut g Gen) struct_init(struct_init ast.StructInit) {
 		if g.is_shared {
 			mut shared_typ := struct_init.typ.set_flag(.shared_f)
 			shared_styp = g.typ(shared_typ)
-			g.writeln('($shared_styp*)memdup(&($shared_styp){.val = ($styp){')
+			g.writeln('($shared_styp*)__dup${shared_styp}(&($shared_styp){.val = ($styp){')
 		} else {
 			g.write('($styp*)memdup(&($styp){')
 		}
@@ -4728,6 +4734,7 @@ fn (mut g Gen) struct_init(struct_init ast.StructInit) {
 		}
 	} else {
 		if g.is_shared {
+			// TODO: non-ref shared should be forbidden
 			g.writeln('{.val = {')
 		} else if is_multiline {
 			g.writeln('($styp){')
@@ -4897,7 +4904,7 @@ fn (mut g Gen) struct_init(struct_init ast.StructInit) {
 	}
 	g.write('}')
 	if g.is_shared {
-		g.write(', .mtx = sync__new_rwmutex()}')
+		g.write('}')
 		if is_amp {
 			g.write(', sizeof($shared_styp))')
 		}
diff --git a/vlib/v/gen/cheaders.v b/vlib/v/gen/cheaders.v
index f7c39d88ec..23b57a27a0 100644
--- a/vlib/v/gen/cheaders.v
+++ b/vlib/v/gen/cheaders.v
@@ -15,6 +15,22 @@ const (
 #ifndef V_CURRENT_COMMIT_HASH
 	#define V_CURRENT_COMMIT_HASH "@@@"
 #endif
+'
+	c_concurrency_helpers         = '
+typedef struct __shared_map __shared_map; 
+struct __shared_map { map val; sync__RwMutex mtx; };
+static inline voidptr __dup_shared_map(voidptr src, int sz) {
+	__shared_map* dest = memdup(src, sz);
+	sync__RwMutex_init(&dest->mtx);
+	return dest;
+}
+typedef struct __shared_array __shared_array;
+struct __shared_array { array val; sync__RwMutex mtx; };
+static inline voidptr __dup_shared_array(voidptr src, int sz) {
+	__shared_array* dest = memdup(src, sz);
+	sync__RwMutex_init(&dest->mtx);
+	return dest;
+}
 '
 	c_common_macros               = '
 #define EMPTY_VARG_INITIALIZATION 0
diff --git a/vlib/v/tests/autolock_array2_test.v b/vlib/v/tests/autolock_array2_test.v
index 75a552b6ab..fa650bb62b 100644
--- a/vlib/v/tests/autolock_array2_test.v
+++ b/vlib/v/tests/autolock_array2_test.v
@@ -4,7 +4,7 @@ const (
 	iterations_per_thread2 = 100000
 )
 
-fn inc_elements(shared foo []int, n int, sem sync.Semaphore) {
+fn inc_elements(shared foo []int, n int, mut sem sync.Semaphore) {
 	for _ in 0 .. iterations_per_thread2 {
 		foo[n]++
 	}
@@ -13,9 +13,9 @@ fn inc_elements(shared foo []int, n int, sem sync.Semaphore) {
 
 fn test_autolocked_array_2() {
 	shared abc := &[0, 0, 0]
-	sem := sync.new_semaphore()
-	go inc_elements(shared abc, 1, sem)
-	go inc_elements(shared abc, 2, sem)
+	mut sem := sync.new_semaphore()
+	go inc_elements(shared abc, 1, mut sem)
+	go inc_elements(shared abc, 2, mut sem)
 	for _ in 0 .. iterations_per_thread2 {
 		unsafe {
 			abc[2]++
diff --git a/vlib/v/tests/semaphore_test.v b/vlib/v/tests/semaphore_test.v
index 5079a996a8..b354746484 100644
--- a/vlib/v/tests/semaphore_test.v
+++ b/vlib/v/tests/semaphore_test.v
@@ -4,7 +4,7 @@ const (
 	signals_per_thread = 100000
 )
 
-fn send_signals(sem sync.Semaphore, sem_end sync.Semaphore) {
+fn send_signals(mut sem sync.Semaphore, mut sem_end sync.Semaphore) {
 	for _ in 0 .. signals_per_thread {
 		sem.post()
 	}
@@ -12,10 +12,10 @@ fn send_signals(sem sync.Semaphore, sem_end sync.Semaphore) {
 }
 
 fn test_semaphores() {
-	sem := sync.new_semaphore()
-	sem_end := sync.new_semaphore()
-	go send_signals(sem, sem_end)
-	go send_signals(sem, sem_end)
+	mut sem := sync.new_semaphore()
+	mut sem_end := sync.new_semaphore()
+	go send_signals(mut sem, mut sem_end)
+	go send_signals(mut sem, mut sem_end)
 	for _ in 0 .. 2 * signals_per_thread {
 		sem.wait()
 	}
diff --git a/vlib/v/tests/semaphore_timed_test.v b/vlib/v/tests/semaphore_timed_test.v
index 62fadf9a48..6b5082f20b 100644
--- a/vlib/v/tests/semaphore_timed_test.v
+++ b/vlib/v/tests/semaphore_timed_test.v
@@ -1,7 +1,7 @@
 import sync
 import time
 
-fn run_forever(shared foo []int, sem sync.Semaphore) {
+fn run_forever(shared foo []int, mut sem sync.Semaphore) {
 	for {
 		foo[0]++
 	}
@@ -10,8 +10,8 @@ fn run_forever(shared foo []int, sem sync.Semaphore) {
 
 fn test_semaphore() {
 	shared abc := &[0]
-	sem := sync.new_semaphore()
-	go run_forever(shared abc, sem)
+	mut sem := sync.new_semaphore()
+	go run_forever(shared abc, mut sem)
 	for _ in 0 .. 1000 {
 		unsafe { abc[0]-- }
 	}
diff --git a/vlib/v/tests/shared_map_test.v b/vlib/v/tests/shared_map_test.v
index 726cbcb051..191b36f33a 100644
--- a/vlib/v/tests/shared_map_test.v
+++ b/vlib/v/tests/shared_map_test.v
@@ -1,6 +1,6 @@
 import sync
 
-fn incr(shared foo map[string]int, key string, sem sync.Semaphore) {
+fn incr(shared foo map[string]int, key string, mut sem sync.Semaphore) {
 	for _ in 0 .. 100000 {
 		lock foo {
 			foo[key] = foo[key] + 1
@@ -14,11 +14,11 @@ fn test_shared_array() {
 	lock foo {
 		foo['q'] = 20
 	}
-	sem := sync.new_semaphore()
-	go incr(shared foo, 'p', sem)
-	go incr(shared foo, 'q', sem)
-	go incr(shared foo, 'p', sem)
-	go incr(shared foo, 'q', sem)
+	mut sem := sync.new_semaphore()
+	go incr(shared foo, 'p', mut sem)
+	go incr(shared foo, 'q', mut sem)
+	go incr(shared foo, 'p', mut sem)
+	go incr(shared foo, 'q', mut sem)
 	for _ in 0 .. 50000 {
 		lock foo {
 			foo['p'] -= 2