Merge branch 'master' of https://github.com/FourteenBrush/Odin

19 files changed, 233 insertions, 73 deletions

diff --git a/core/runtime/core.odin b/core/runtime/core.odin
index 2d176f909..740482493 100644
--- a/core/runtime/core.odin
+++ b/core/runtime/core.odin
@@ -18,6 +18,7 @@
 // This could change at a later date if the all these data structures are
 // implemented within the compiler rather than in this "preload" file
 //
+//+no-instrumentation
 package runtime
 
 import "core:intrinsics"
@@ -306,6 +307,7 @@ Allocator_Mode :: enum byte {
 	Query_Features,
 	Query_Info,
 	Alloc_Non_Zeroed,
+	Resize_Non_Zeroed,
 }
 
 Allocator_Mode_Set :: distinct bit_set[Allocator_Mode]
diff --git a/core/runtime/core_builtin.odin b/core/runtime/core_builtin.odin
index a73a3d712..3f4ebbc74 100644
--- a/core/runtime/core_builtin.odin
+++ b/core/runtime/core_builtin.odin
@@ -109,7 +109,7 @@ remove_range :: proc(array: ^$D/[dynamic]$T, lo, hi: int, loc := #caller_locatio
 
 // `pop` will remove and return the end value of dynamic array `array` and reduces the length of `array` by 1.
 //
-// Note: If the dynamic array as no elements (`len(array) == 0`), this procedure will panic.
+// Note: If the dynamic array has no elements (`len(array) == 0`), this procedure will panic.
 @builtin
 pop :: proc(array: ^$T/[dynamic]$E, loc := #caller_location) -> (res: E) #no_bounds_check {
 	assert(len(array) > 0, loc=loc)
@@ -169,10 +169,16 @@ clear :: proc{clear_dynamic_array, clear_map}
 @builtin
 reserve :: proc{reserve_dynamic_array, reserve_map}
 
-// `resize` will try to resize memory of a passed dynamic array or map to the requested element count (setting the `len`, and possibly `cap`).
+@builtin
+non_zero_reserve :: proc{non_zero_reserve_dynamic_array}
+
+// `resize` will try to resize memory of a passed dynamic array to the requested element count (setting the `len`, and possibly `cap`).
 @builtin
 resize :: proc{resize_dynamic_array}
 
+@builtin
+non_zero_resize :: proc{non_zero_resize_dynamic_array}
+
 // Shrinks the capacity of a dynamic array or map down to the current length, or the given capacity.
 @builtin
 shrink :: proc{shrink_dynamic_array, shrink_map}
@@ -234,6 +240,8 @@ delete :: proc{
 	delete_dynamic_array,
 	delete_slice,
 	delete_map,
+	delete_soa_slice,
+	delete_soa_dynamic_array,
 }
 
 
@@ -346,7 +354,7 @@ make_multi_pointer :: proc($T: typeid/[^]$E, #any_int len: int, allocator := con
 //
 // Similar to `new`, the first argument is a type, not a value. Unlike new, make's return type is the same as the
 // type of its argument, not a pointer to it.
-// Make uses the specified allocator, default is context.allocator, default is context.allocator
+// Make uses the specified allocator, default is context.allocator.
 @builtin
 make :: proc{
 	make_slice,
@@ -404,10 +412,7 @@ delete_key :: proc(m: ^$T/map[$K]$V, key: K) -> (deleted_key: K, deleted_value:
 	return
 }
 
-
-
-@builtin
-append_elem :: proc(array: ^$T/[dynamic]$E, arg: E, loc := #caller_location) -> (n: int, err: Allocator_Error) #optional_allocator_error {
+_append_elem :: #force_inline proc(array: ^$T/[dynamic]$E, arg: E, should_zero: bool, loc := #caller_location) -> (n: int, err: Allocator_Error) #optional_allocator_error {
 	if array == nil {
 		return 0, nil
 	}
@@ -418,7 +423,13 @@ append_elem :: proc(array: ^$T/[dynamic]$E, arg: E, loc := #caller_location) ->
 	} else {
 		if cap(array) < len(array)+1 {
 			cap := 2 * cap(array) + max(8, 1)
-			err = reserve(array, cap, loc) // do not 'or_return' here as it could be a partial success
+
+			// do not 'or_return' here as it could be a partial success
+			if should_zero {
+				err = reserve(array, cap, loc)
+			} else {
+				err = non_zero_reserve(array, cap, loc) 
+			}
 		}
 		if cap(array)-len(array) > 0 {
 			a := (^Raw_Dynamic_Array)(array)
@@ -435,7 +446,16 @@ append_elem :: proc(array: ^$T/[dynamic]$E, arg: E, loc := #caller_location) ->
 }
 
 @builtin
-append_elems :: proc(array: ^$T/[dynamic]$E, args: ..E, loc := #caller_location) -> (n: int, err: Allocator_Error) #optional_allocator_error {
+append_elem :: proc(array: ^$T/[dynamic]$E, arg: E, loc := #caller_location) -> (n: int, err: Allocator_Error) #optional_allocator_error {
+	return _append_elem(array, arg, true, loc=loc)
+}
+
+@builtin
+non_zero_append_elem :: proc(array: ^$T/[dynamic]$E, arg: E, loc := #caller_location) -> (n: int, err: Allocator_Error) #optional_allocator_error {
+	return _append_elem(array, arg, false, loc=loc)
+}
+
+_append_elems :: #force_inline proc(array: ^$T/[dynamic]$E, should_zero: bool, loc := #caller_location, args: ..E) -> (n: int, err: Allocator_Error) #optional_allocator_error {
 	if array == nil {
 		return 0, nil
 	}
@@ -452,7 +472,13 @@ append_elems :: proc(array: ^$T/[dynamic]$E, args: ..E, loc := #caller_location)
 	} else {
 		if cap(array) < len(array)+arg_len {
 			cap := 2 * cap(array) + max(8, arg_len)
-			err = reserve(array, cap, loc)  // do not 'or_return' here as it could be a partial success
+
+			// do not 'or_return' here as it could be a partial success
+			if should_zero {
+				err = reserve(array, cap, loc)
+			} else {
+				err = non_zero_reserve(array, cap, loc)
+			}
 		}
 		arg_len = min(cap(array)-len(array), arg_len)
 		if arg_len > 0 {
@@ -468,11 +494,33 @@ append_elems :: proc(array: ^$T/[dynamic]$E, args: ..E, loc := #caller_location)
 	}
 }
 
+@builtin
+append_elems :: proc(array: ^$T/[dynamic]$E, args: ..E, loc := #caller_location) -> (n: int, err: Allocator_Error) #optional_allocator_error {
+	return _append_elems(array, true, loc, ..args)
+}
+
+@builtin
+non_zero_append_elems :: proc(array: ^$T/[dynamic]$E, args: ..E, loc := #caller_location) -> (n: int, err: Allocator_Error) #optional_allocator_error {
+	return _append_elems(array, false, loc, ..args)
+}
+
 // The append_string built-in procedure appends a string to the end of a [dynamic]u8 like type
+_append_elem_string :: proc(array: ^$T/[dynamic]$E/u8, arg: $A/string, should_zero: bool, loc := #caller_location) -> (n: int, err: Allocator_Error) #optional_allocator_error {
+	args := transmute([]E)arg
+	if should_zero { 
+		return append_elems(array, ..args, loc=loc)
+	} else {
+		return non_zero_append_elems(array, ..args, loc=loc)
+	}
+}
+
 @builtin
 append_elem_string :: proc(array: ^$T/[dynamic]$E/u8, arg: $A/string, loc := #caller_location) -> (n: int, err: Allocator_Error) #optional_allocator_error {
-	args := transmute([]E)arg
-	return append_elems(array, ..args, loc=loc)
+	return _append_elem_string(array, arg, true, loc)
+}
+@builtin
+non_zero_append_elem_string :: proc(array: ^$T/[dynamic]$E/u8, arg: $A/string, loc := #caller_location) -> (n: int, err: Allocator_Error) #optional_allocator_error {
+	return _append_elem_string(array, arg, false, loc)
 }
 
 
@@ -492,6 +540,7 @@ append_string :: proc(array: ^$T/[dynamic]$E/u8, args: ..string, loc := #caller_
 
 // The append built-in procedure appends elements to the end of a dynamic array
 @builtin append :: proc{append_elem, append_elems, append_elem_string}
+@builtin non_zero_append :: proc{non_zero_append_elem, non_zero_append_elems, non_zero_append_elem_string}
 
 
 @builtin
@@ -587,11 +636,14 @@ assign_at_elem :: proc(array: ^$T/[dynamic]$E, index: int, arg: E, loc := #calle
 
 @builtin
 assign_at_elems :: proc(array: ^$T/[dynamic]$E, index: int, args: ..E, loc := #caller_location) -> (ok: bool, err: Allocator_Error) #no_bounds_check #optional_allocator_error {
-	if index+len(args) < len(array) {
+	new_size := index + len(args)
+	if len(args) == 0 {
+		ok = true
+	} else if new_size < len(array) {
 		copy(array[index:], args)
 		ok = true
 	} else {
-		resize(array, index+1+len(args), loc) or_return
+		resize(array, new_size, loc) or_return
 		copy(array[index:], args)
 		ok = true
 	}
@@ -633,8 +685,7 @@ clear_dynamic_array :: proc "contextless" (array: ^$T/[dynamic]$E) {
 // `reserve_dynamic_array` will try to reserve memory of a passed dynamic array or map to the requested element count (setting the `cap`).
 //
 // Note: Prefer the procedure group `reserve`.
-@builtin
-reserve_dynamic_array :: proc(array: ^$T/[dynamic]$E, capacity: int, loc := #caller_location) -> Allocator_Error {
+_reserve_dynamic_array :: #force_inline proc(array: ^$T/[dynamic]$E, capacity: int, should_zero: bool, loc := #caller_location) -> Allocator_Error {
 	if array == nil {
 		return nil
 	}
@@ -653,7 +704,12 @@ reserve_dynamic_array :: proc(array: ^$T/[dynamic]$E, capacity: int, loc := #cal
 	new_size  := capacity * size_of(E)
 	allocator := a.allocator
 
-	new_data := mem_resize(a.data, old_size, new_size, align_of(E), allocator, loc) or_return
+	new_data: []byte
+	if should_zero {
+		new_data = mem_resize(a.data, old_size, new_size, align_of(E), allocator, loc) or_return
+	} else {
+		new_data = non_zero_mem_resize(a.data, old_size, new_size, align_of(E), allocator, loc) or_return
+	}
 	if new_data == nil && new_size > 0 {
 		return .Out_Of_Memory
 	}
@@ -663,11 +719,20 @@ reserve_dynamic_array :: proc(array: ^$T/[dynamic]$E, capacity: int, loc := #cal
 	return nil
 }
 
+@builtin
+reserve_dynamic_array :: proc(array: ^$T/[dynamic]$E, capacity: int, loc := #caller_location) -> Allocator_Error {
+	return _reserve_dynamic_array(array, capacity, true, loc)
+}
+
+@builtin
+non_zero_reserve_dynamic_array :: proc(array: ^$T/[dynamic]$E, capacity: int, loc := #caller_location) -> Allocator_Error {
+	return _reserve_dynamic_array(array, capacity, false, loc)
+}
+
 // `resize_dynamic_array` will try to resize memory of a passed dynamic array or map to the requested element count (setting the `len`, and possibly `cap`).
 //
 // Note: Prefer the procedure group `resize`
-@builtin
-resize_dynamic_array :: proc(array: ^$T/[dynamic]$E, length: int, loc := #caller_location) -> Allocator_Error {
+_resize_dynamic_array :: #force_inline proc(array: ^$T/[dynamic]$E, length: int, should_zero: bool, loc := #caller_location) -> Allocator_Error {
 	if array == nil {
 		return nil
 	}
@@ -687,7 +752,12 @@ resize_dynamic_array :: proc(array: ^$T/[dynamic]$E, length: int, loc := #caller
 	new_size  := length * size_of(E)
 	allocator := a.allocator
 
-	new_data := mem_resize(a.data, old_size, new_size, align_of(E), allocator, loc) or_return
+	new_data : []byte
+	if should_zero {
+		new_data = mem_resize(a.data, old_size, new_size, align_of(E), allocator, loc) or_return
+	} else {
+		new_data = non_zero_mem_resize(a.data, old_size, new_size, align_of(E), allocator, loc) or_return
+	}
 	if new_data == nil && new_size > 0 {
 		return .Out_Of_Memory
 	}
@@ -698,6 +768,16 @@ resize_dynamic_array :: proc(array: ^$T/[dynamic]$E, length: int, loc := #caller
 	return nil
 }
 
+@builtin
+resize_dynamic_array :: proc(array: ^$T/[dynamic]$E, length: int, loc := #caller_location) -> Allocator_Error {
+	return _resize_dynamic_array(array, length, true, loc=loc)
+}
+
+@builtin
+non_zero_resize_dynamic_array :: proc(array: ^$T/[dynamic]$E, length: int, loc := #caller_location) -> Allocator_Error {
+	return _resize_dynamic_array(array, length, false, loc=loc)
+}
+
 /*
 	Shrinks the capacity of a dynamic array down to the current length, or the given capacity.
 
diff --git a/core/runtime/core_builtin_soa.odin b/core/runtime/core_builtin_soa.odin
index f3882a9a8..6313a28f5 100644
--- a/core/runtime/core_builtin_soa.odin
+++ b/core/runtime/core_builtin_soa.odin
@@ -287,7 +287,7 @@ append_soa_elem :: proc(array: ^$T/#soa[dynamic]$E, arg: E, loc := #caller_locat
 	footer := raw_soa_footer(array)
 
 	if size_of(E) > 0 && cap(array)-len(array) > 0 {
-		ti := type_info_of(typeid_of(T))
+		ti := type_info_of(T)
 		ti = type_info_base(ti)
 		si := &ti.variant.(Type_Info_Struct)
 		field_count: uintptr
diff --git a/core/runtime/default_allocators_arena.odin b/core/runtime/default_allocators_arena.odin
index 1c36c4f7c..1fe3c6cfc 100644
--- a/core/runtime/default_allocators_arena.odin
+++ b/core/runtime/default_allocators_arena.odin
@@ -28,11 +28,11 @@ safe_add :: #force_inline proc "contextless" (x, y: uint) -> (uint, bool) {
 }
 
 @(require_results)
-memory_block_alloc :: proc(allocator: Allocator, capacity: uint, loc := #caller_location) -> (block: ^Memory_Block, err: Allocator_Error) {
-	total_size := uint(capacity + size_of(Memory_Block))
-	base_offset    := uintptr(size_of(Memory_Block))
+memory_block_alloc :: proc(allocator: Allocator, capacity: uint, alignment: uint, loc := #caller_location) -> (block: ^Memory_Block, err: Allocator_Error) {
+	total_size  := uint(capacity + max(alignment, size_of(Memory_Block)))
+	base_offset := uintptr(max(alignment, size_of(Memory_Block)))
 
-	min_alignment: int = max(16, align_of(Memory_Block))
+	min_alignment: int = max(16, align_of(Memory_Block), int(alignment))
 	data := mem_alloc(int(total_size), min_alignment, allocator, loc) or_return
 	block = (^Memory_Block)(raw_data(data))
 	end := uintptr(raw_data(data)[len(data):])
@@ -102,20 +102,20 @@ arena_alloc :: proc(arena: ^Arena, size, alignment: uint, loc := #caller_locatio
 	if size == 0 {
 		return
 	}
-
-	if arena.curr_block == nil || (safe_add(arena.curr_block.used, size) or_else 0) > arena.curr_block.capacity {
-		size = align_forward_uint(size, alignment)
+	
+	needed := align_forward_uint(size, alignment)
+	if arena.curr_block == nil || (safe_add(arena.curr_block.used, needed) or_else 0) > arena.curr_block.capacity {
 		if arena.minimum_block_size == 0 {
 			arena.minimum_block_size = DEFAULT_ARENA_GROWING_MINIMUM_BLOCK_SIZE
 		}
 
-		block_size := max(size, arena.minimum_block_size)
+		block_size := max(needed, arena.minimum_block_size)
 
 		if arena.backing_allocator.procedure == nil {
 			arena.backing_allocator = default_allocator()
 		}
 
-		new_block := memory_block_alloc(arena.backing_allocator, block_size, loc) or_return
+		new_block := memory_block_alloc(arena.backing_allocator, block_size, alignment, loc) or_return
 		new_block.prev = arena.curr_block
 		arena.curr_block = new_block
 		arena.total_capacity += new_block.capacity
@@ -134,7 +134,7 @@ arena_init :: proc(arena: ^Arena, size: uint, backing_allocator: Allocator, loc
 	arena^ = {}
 	arena.backing_allocator = backing_allocator
 	arena.minimum_block_size = max(size, 1<<12) // minimum block size of 4 KiB
-	new_block := memory_block_alloc(arena.backing_allocator, arena.minimum_block_size, loc) or_return
+	new_block := memory_block_alloc(arena.backing_allocator, arena.minimum_block_size, 0, loc) or_return
 	arena.curr_block = new_block
 	arena.total_capacity += new_block.capacity
 	return nil
@@ -195,7 +195,7 @@ arena_allocator_proc :: proc(allocator_data: rawptr, mode: Allocator_Mode,
 		err = .Mode_Not_Implemented
 	case .Free_All:
 		arena_free_all(arena, location)
-	case .Resize:
+	case .Resize, .Resize_Non_Zeroed:
 		old_data := ([^]byte)(old_memory)
 
 		switch {
diff --git a/core/runtime/default_allocators_nil.odin b/core/runtime/default_allocators_nil.odin
index a340050eb..c882f5196 100644
--- a/core/runtime/default_allocators_nil.odin
+++ b/core/runtime/default_allocators_nil.odin
@@ -10,7 +10,7 @@ nil_allocator_proc :: proc(allocator_data: rawptr, mode: Allocator_Mode,
 		return nil, .None
 	case .Free_All:
 		return nil, .Mode_Not_Implemented
-	case .Resize:
+	case .Resize, .Resize_Non_Zeroed:
 		if size == 0 {
 			return nil, .None
 		}
@@ -35,7 +35,7 @@ nil_allocator :: proc() -> Allocator {
 when ODIN_OS == .Freestanding {
 	default_allocator_proc :: nil_allocator_proc
 	default_allocator :: nil_allocator
-} 
+}
 
 
 
@@ -55,6 +55,10 @@ panic_allocator_proc :: proc(allocator_data: rawptr, mode: Allocator_Mode,
 		if size > 0 {
 			panic("panic allocator, .Resize called", loc=loc)
 		}
+	case .Resize_Non_Zeroed:
+		if size > 0 {
+			panic("panic allocator, .Alloc_Non_Zeroed called", loc=loc)
+		}
 	case .Free:
 		if old_memory != nil {
 			panic("panic allocator, .Free called", loc=loc)
@@ -78,9 +82,7 @@ panic_allocator_proc :: proc(allocator_data: rawptr, mode: Allocator_Mode,
 
 panic_allocator :: proc() -> Allocator {
 	return Allocator{
-		procedure = nil_allocator_proc,
+		procedure = panic_allocator_proc,
 		data = nil,
 	}
 }
-
-
diff --git a/core/runtime/default_allocators_windows.odin b/core/runtime/default_allocators_windows.odin
index a78a4d04e..1b0f78428 100644
--- a/core/runtime/default_allocators_windows.odin
+++ b/core/runtime/default_allocators_windows.odin
@@ -19,7 +19,7 @@ when ODIN_DEFAULT_TO_NIL_ALLOCATOR {
 		case .Free_All:
 			return nil, .Mode_Not_Implemented
 
-		case .Resize:
+		case .Resize, .Resize_Non_Zeroed:
 			data, err = _windows_default_resize(old_memory, old_size, size, alignment)
 
 		case .Query_Features:
@@ -41,4 +41,4 @@ when ODIN_DEFAULT_TO_NIL_ALLOCATOR {
 			data = nil,
 		}
 	}
-}
\ No newline at end of file
+}
diff --git a/core/runtime/dynamic_map_internal.odin b/core/runtime/dynamic_map_internal.odin
index bdf0979cb..491a7974d 100644
--- a/core/runtime/dynamic_map_internal.odin
+++ b/core/runtime/dynamic_map_internal.odin
@@ -44,7 +44,7 @@ _ :: intrinsics
 MAP_LOAD_FACTOR :: 75
 
 // Minimum log2 capacity.
-MAP_MIN_LOG2_CAPACITY :: 6 // 64 elements
+MAP_MIN_LOG2_CAPACITY :: 3 // 8 elements
 
 // Has to be less than 100% though.
 #assert(MAP_LOAD_FACTOR < 100)
diff --git a/core/runtime/entry_unix.odin b/core/runtime/entry_unix.odin
index 0c718445a..f494a509e 100644
--- a/core/runtime/entry_unix.odin
+++ b/core/runtime/entry_unix.odin
@@ -1,5 +1,6 @@
 //+private
 //+build linux, darwin, freebsd, openbsd
+//+no-instrumentation
 package runtime
 
 import "core:intrinsics"
@@ -26,8 +27,13 @@ when ODIN_BUILD_MODE == .Dynamic {
 		// to retrieve argc and argv from the stack
 		when ODIN_ARCH == .amd64 {
 			@require foreign import entry "entry_unix_no_crt_amd64.asm"
+			SYS_exit :: 60
 		} else when ODIN_ARCH == .i386 {
 			@require foreign import entry "entry_unix_no_crt_i386.asm"
+			SYS_exit :: 1
+		} else when ODIN_OS == .Darwin && ODIN_ARCH == .arm64 {
+			@require foreign import entry "entry_unix_no_crt_darwin_arm64.asm"
+			SYS_exit :: 1
 		}
 		@(link_name="_start_odin", linkage="strong", require)
 		_start_odin :: proc "c" (argc: i32, argv: [^]cstring) -> ! {
@@ -36,11 +42,7 @@ when ODIN_BUILD_MODE == .Dynamic {
 			#force_no_inline _startup_runtime()
 			intrinsics.__entry_point()
 			#force_no_inline _cleanup_runtime()
-			when ODIN_ARCH == .amd64 {
-				intrinsics.syscall(/*SYS_exit = */60)
-			} else when ODIN_ARCH == .i386 {
-				intrinsics.syscall(/*SYS_exit = */1)
-			}
+			intrinsics.syscall(SYS_exit, 0)
 			unreachable()
 		}
 	} else {
diff --git a/core/runtime/entry_unix_no_crt_darwin_arm64.asm b/core/runtime/entry_unix_no_crt_darwin_arm64.asm
new file mode 100644
index 000000000..0f71fbdf8
--- /dev/null
+++ b/core/runtime/entry_unix_no_crt_darwin_arm64.asm
@@ -0,0 +1,20 @@
+	.section __TEXT,__text
+
+	; NOTE(laytan): this should ideally be the -minimum-os-version flag but there is no nice way of preprocessing assembly in Odin.
+	; 10 seems to be the lowest it goes and I don't see it mess with any targeted os version so this seems fine.
+	.build_version macos, 10, 0
+
+	.extern __start_odin
+
+	.global _main
+	.align 2
+_main:
+	mov x5, sp       ; use x5 as the stack pointer
+
+	str x0, [x5]     ; get argc into x0 (kernel passes 32-bit int argc as 64-bits on stack to keep alignment)
+	str x1, [x5, #8] ; get argv into x1
+
+	and sp, x5, #~15 ; force 16-byte alignment of the stack
+	
+	bl __start_odin  ; call into Odin entry point
+	ret              ; should never get here
diff --git a/core/runtime/entry_wasm.odin b/core/runtime/entry_wasm.odin
index 235d5611b..e7f3f156f 100644
--- a/core/runtime/entry_wasm.odin
+++ b/core/runtime/entry_wasm.odin
@@ -1,5 +1,6 @@
 //+private
 //+build wasm32, wasm64p32
+//+no-instrumentation
 package runtime
 
 import "core:intrinsics"
diff --git a/core/runtime/entry_windows.odin b/core/runtime/entry_windows.odin
index a315c1209..b6fbe1dcc 100644
--- a/core/runtime/entry_windows.odin
+++ b/core/runtime/entry_windows.odin
@@ -1,12 +1,13 @@
 //+private
 //+build windows
+//+no-instrumentation
 package runtime
 
 import "core:intrinsics"
 
 when ODIN_BUILD_MODE == .Dynamic {
 	@(link_name="DllMain", linkage="strong", require)
-	DllMain :: proc "stdcall" (hinstDLL: rawptr, fdwReason: u32, lpReserved: rawptr) -> b32 {
+	DllMain :: proc "system" (hinstDLL: rawptr, fdwReason: u32, lpReserved: rawptr) -> b32 {
 		context = default_context()
 
 		// Populate Windows DLL-specific global
@@ -28,7 +29,7 @@ when ODIN_BUILD_MODE == .Dynamic {
 } else when !ODIN_TEST && !ODIN_NO_ENTRY_POINT {
 	when ODIN_ARCH == .i386 || ODIN_NO_CRT {
 		@(link_name="mainCRTStartup", linkage="strong", require)
-		mainCRTStartup :: proc "stdcall" () -> i32 {
+		mainCRTStartup :: proc "system" () -> i32 {
 			context = default_context()
 			#force_no_inline _startup_runtime()
 			intrinsics.__entry_point()
diff --git a/core/runtime/error_checks.odin b/core/runtime/error_checks.odin
index 9d484979a..ea6333c29 100644
--- a/core/runtime/error_checks.odin
+++ b/core/runtime/error_checks.odin
@@ -1,5 +1,6 @@
 package runtime
 
+@(no_instrumentation)
 bounds_trap :: proc "contextless" () -> ! {
 	when ODIN_OS == .Windows {
 		windows_trap_array_bounds()
@@ -8,6 +9,7 @@ bounds_trap :: proc "contextless" () -> ! {
 	}
 }
 
+@(no_instrumentation)
 type_assertion_trap :: proc "contextless" () -> ! {
 	when ODIN_OS == .Windows {
 		windows_trap_type_assertion()
@@ -21,7 +23,7 @@ bounds_check_error :: proc "contextless" (file: string, line, column: i32, index
 	if uint(index) < uint(count) {
 		return
 	}
-	@(cold)
+	@(cold, no_instrumentation)
 	handle_error :: proc "contextless" (file: string, line, column: i32, index, count: int) -> ! {
 		print_caller_location(Source_Code_Location{file, line, column, ""})
 		print_string(" Index ")
@@ -34,6 +36,7 @@ bounds_check_error :: proc "contextless" (file: string, line, column: i32, index
 	handle_error(file, line, column, index, count)
 }
 
+@(no_instrumentation)
 slice_handle_error :: proc "contextless" (file: string, line, column: i32, lo, hi: int, len: int) -> ! {
 	print_caller_location(Source_Code_Location{file, line, column, ""})
 	print_string(" Invalid slice indices ")
@@ -46,6 +49,7 @@ slice_handle_error :: proc "contextless" (file: string, line, column: i32, lo, h
 	bounds_trap()
 }
 
+@(no_instrumentation)
 multi_pointer_slice_handle_error :: proc "contextless" (file: string, line, column: i32, lo, hi: int) -> ! {
 	print_caller_location(Source_Code_Location{file, line, column, ""})
 	print_string(" Invalid slice indices ")
@@ -82,7 +86,7 @@ dynamic_array_expr_error :: proc "contextless" (file: string, line, column: i32,
 	if 0 <= low && low <= high && high <= max {
 		return
 	}
-	@(cold)
+	@(cold, no_instrumentation)
 	handle_error :: proc "contextless" (file: string, line, column: i32, low, high, max: int) -> ! {
 		print_caller_location(Source_Code_Location{file, line, column, ""})
 		print_string(" Invalid dynamic array indices ")
@@ -103,7 +107,7 @@ matrix_bounds_check_error :: proc "contextless" (file: string, line, column: i32
 	   uint(column_index) < uint(column_count) {
 		return
 	}
-	@(cold)
+	@(cold, no_instrumentation)
 	handle_error :: proc "contextless" (file: string, line, column: i32, row_index, column_index, row_count, column_count: int) -> ! {
 		print_caller_location(Source_Code_Location{file, line, column, ""})
 		print_string(" Matrix indices [")
@@ -127,7 +131,7 @@ when ODIN_NO_RTTI {
 		if ok {
 			return
 		}
-		@(cold)
+		@(cold, no_instrumentation)
 		handle_error :: proc "contextless" (file: string, line, column: i32) -> ! {
 			print_caller_location(Source_Code_Location{file, line, column, ""})
 			print_string(" Invalid type assertion\n")
@@ -140,7 +144,7 @@ when ODIN_NO_RTTI {
 		if ok {
 			return
 		}
-		@(cold)
+		@(cold, no_instrumentation)
 		handle_error :: proc "contextless" (file: string, line, column: i32) -> ! {
 			print_caller_location(Source_Code_Location{file, line, column, ""})
 			print_string(" Invalid type assertion\n")
@@ -153,7 +157,7 @@ when ODIN_NO_RTTI {
 		if ok {
 			return
 		}
-		@(cold)
+		@(cold, no_instrumentation)
 		handle_error :: proc "contextless" (file: string, line, column: i32, from, to: typeid) -> ! {
 			print_caller_location(Source_Code_Location{file, line, column, ""})
 			print_string(" Invalid type assertion from ")
@@ -198,7 +202,7 @@ when ODIN_NO_RTTI {
 			return id
 		}
 
-		@(cold)
+		@(cold, no_instrumentation)
 		handle_error :: proc "contextless" (file: string, line, column: i32, from, to: typeid, from_data: rawptr) -> ! {
 
 			actual := variant_type(from, from_data)
@@ -224,7 +228,7 @@ make_slice_error_loc :: #force_inline proc "contextless" (loc := #caller_locatio
 	if 0 <= len {
 		return
 	}
-	@(cold)
+	@(cold, no_instrumentation)
 	handle_error :: proc "contextless" (loc: Source_Code_Location, len: int) -> ! {
 		print_caller_location(loc)
 		print_string(" Invalid slice length for make: ")
@@ -239,7 +243,7 @@ make_dynamic_array_error_loc :: #force_inline proc "contextless" (loc := #caller
 	if 0 <= len && len <= cap {
 		return
 	}
-	@(cold)
+	@(cold, no_instrumentation)
 	handle_error :: proc "contextless" (loc: Source_Code_Location, len, cap: int)  -> ! {
 		print_caller_location(loc)
 		print_string(" Invalid dynamic array parameters for make: ")
@@ -256,7 +260,7 @@ make_map_expr_error_loc :: #force_inline proc "contextless" (loc := #caller_loca
 	if 0 <= cap {
 		return
 	}
-	@(cold)
+	@(cold, no_instrumentation)
 	handle_error :: proc "contextless" (loc: Source_Code_Location, cap: int)  -> ! {
 		print_caller_location(loc)
 		print_string(" Invalid map capacity for make: ")
diff --git a/core/runtime/internal.odin b/core/runtime/internal.odin
index d0e550743..d4c43ed7e 100644
--- a/core/runtime/internal.odin
+++ b/core/runtime/internal.odin
@@ -187,7 +187,7 @@ mem_free_all :: #force_inline proc(allocator := context.allocator, loc := #calle
 	return
 }
 
-mem_resize :: proc(ptr: rawptr, old_size, new_size: int, alignment: int = DEFAULT_ALIGNMENT, allocator := context.allocator, loc := #caller_location) -> (data: []byte, err: Allocator_Error) {
+_mem_resize :: #force_inline proc(ptr: rawptr, old_size, new_size: int, alignment: int = DEFAULT_ALIGNMENT, allocator := context.allocator, should_zero: bool, loc := #caller_location) -> (data: []byte, err: Allocator_Error) {
 	if allocator.procedure == nil {
 		return nil, nil
 	}
@@ -198,15 +198,27 @@ mem_resize :: proc(ptr: rawptr, old_size, new_size: int, alignment: int = DEFAUL
 		}
 		return
 	} else if ptr == nil {
-		return allocator.procedure(allocator.data, .Alloc, new_size, alignment, nil, 0, loc)
+		if should_zero {
+			return allocator.procedure(allocator.data, .Alloc, new_size, alignment, nil, 0, loc)
+		} else {
+			return allocator.procedure(allocator.data, .Alloc_Non_Zeroed, new_size, alignment, nil, 0, loc)
+		}
 	} else if old_size == new_size && uintptr(ptr) % uintptr(alignment) == 0 {
 		data = ([^]byte)(ptr)[:old_size]
 		return
 	}
 
-	data, err = allocator.procedure(allocator.data, .Resize, new_size, alignment, ptr, old_size, loc)
+	if should_zero {
+		data, err = allocator.procedure(allocator.data, .Resize, new_size, alignment, ptr, old_size, loc)
+	} else {
+		data, err = allocator.procedure(allocator.data, .Resize_Non_Zeroed, new_size, alignment, ptr, old_size, loc)
+	}
 	if err == .Mode_Not_Implemented {
-		data, err = allocator.procedure(allocator.data, .Alloc, new_size, alignment, nil, 0, loc)
+		if should_zero {
+			data, err = allocator.procedure(allocator.data, .Alloc, new_size, alignment, nil, 0, loc)
+		} else {
+			data, err = allocator.procedure(allocator.data, .Alloc_Non_Zeroed, new_size, alignment, nil, 0, loc)
+		}
 		if err != nil {
 			return
 		}
@@ -216,6 +228,13 @@ mem_resize :: proc(ptr: rawptr, old_size, new_size: int, alignment: int = DEFAUL
 	return
 }
 
+mem_resize :: proc(ptr: rawptr, old_size, new_size: int, alignment: int = DEFAULT_ALIGNMENT, allocator := context.allocator, loc := #caller_location) -> (data: []byte, err: Allocator_Error) {
+	return _mem_resize(ptr, old_size, new_size, alignment, allocator, true, loc)
+}
+non_zero_mem_resize :: proc(ptr: rawptr, old_size, new_size: int, alignment: int = DEFAULT_ALIGNMENT, allocator := context.allocator, loc := #caller_location) -> (data: []byte, err: Allocator_Error) {
+	return _mem_resize(ptr, old_size, new_size, alignment, allocator, false, loc)
+}
+
 memory_equal :: proc "contextless" (x, y: rawptr, n: int) -> bool {
 	switch {
 	case n == 0: return true
@@ -730,7 +749,7 @@ mul_quaternion64 :: proc "contextless" (q, r: quaternion64) -> quaternion64 {
 	t2 := r0*q2 + r1*q3 + r2*q0 - r3*q1
 	t3 := r0*q3 - r1*q2 + r2*q1 + r3*q0
 
-	return quaternion(t0, t1, t2, t3)
+	return quaternion(w=t0, x=t1, y=t2, z=t3)
 }
 
 mul_quaternion128 :: proc "contextless" (q, r: quaternion128) -> quaternion128 {
@@ -742,7 +761,7 @@ mul_quaternion128 :: proc "contextless" (q, r: quaternion128) -> quaternion128 {
 	t2 := r0*q2 + r1*q3 + r2*q0 - r3*q1
 	t3 := r0*q3 - r1*q2 + r2*q1 + r3*q0
 
-	return quaternion(t0, t1, t2, t3)
+	return quaternion(w=t0, x=t1, y=t2, z=t3)
 }
 
 mul_quaternion256 :: proc "contextless" (q, r: quaternion256) -> quaternion256 {
@@ -754,7 +773,7 @@ mul_quaternion256 :: proc "contextless" (q, r: quaternion256) -> quaternion256 {
 	t2 := r0*q2 + r1*q3 + r2*q0 - r3*q1
 	t3 := r0*q3 - r1*q2 + r2*q1 + r3*q0
 
-	return quaternion(t0, t1, t2, t3)
+	return quaternion(w=t0, x=t1, y=t2, z=t3)
 }
 
 quo_quaternion64 :: proc "contextless" (q, r: quaternion64) -> quaternion64 {
@@ -768,7 +787,7 @@ quo_quaternion64 :: proc "contextless" (q, r: quaternion64) -> quaternion64 {
 	t2 := (r0*q2 - r1*q3 - r2*q0 + r3*q1) * invmag2
 	t3 := (r0*q3 + r1*q2 + r2*q1 - r3*q0) * invmag2
 
-	return quaternion(t0, t1, t2, t3)
+	return quaternion(w=t0, x=t1, y=t2, z=t3)
 }
 
 quo_quaternion128 :: proc "contextless" (q, r: quaternion128) -> quaternion128 {
@@ -782,7 +801,7 @@ quo_quaternion128 :: proc "contextless" (q, r: quaternion128) -> quaternion128 {
 	t2 := (r0*q2 - r1*q3 - r2*q0 + r3*q1) * invmag2
 	t3 := (r0*q3 + r1*q2 + r2*q1 - r3*q0) * invmag2
 
-	return quaternion(t0, t1, t2, t3)
+	return quaternion(w=t0, x=t1, y=t2, z=t3)
 }
 
 quo_quaternion256 :: proc "contextless" (q, r: quaternion256) -> quaternion256 {
@@ -796,7 +815,7 @@ quo_quaternion256 :: proc "contextless" (q, r: quaternion256) -> quaternion256 {
 	t2 := (r0*q2 - r1*q3 - r2*q0 + r3*q1) * invmag2
 	t3 := (r0*q3 + r1*q2 + r2*q1 - r3*q0) * invmag2
 
-	return quaternion(t0, t1, t2, t3)
+	return quaternion(w=t0, x=t1, y=t2, z=t3)
 }
 
 @(link_name="__truncsfhf2", linkage=RUNTIME_LINKAGE, require=RUNTIME_REQUIRE)
diff --git a/core/runtime/os_specific_any.odin b/core/runtime/os_specific_any.odin
index afa106138..6a96655c4 100644
--- a/core/runtime/os_specific_any.odin
+++ b/core/runtime/os_specific_any.odin
@@ -1,4 +1,8 @@
-//+build !freestanding !wasi !windows !js
+//+build !darwin
+//+build !freestanding
+//+build !js
+//+build !wasi
+//+build !windows
 package runtime
 
 import "core:os"
diff --git a/core/runtime/os_specific_darwin.odin b/core/runtime/os_specific_darwin.odin
new file mode 100644
index 000000000..5de9a7d57
--- /dev/null
+++ b/core/runtime/os_specific_darwin.odin
@@ -0,0 +1,12 @@
+//+build darwin
+package runtime
+
+import "core:intrinsics"
+
+_os_write :: proc "contextless" (data: []byte) -> (int, _OS_Errno) {
+	ret := intrinsics.syscall(0x2000004, 1, uintptr(raw_data(data)), uintptr(len(data)))
+	if ret < 0 {
+		return 0, _OS_Errno(-ret)
+	}
+	return int(ret), 0
+}
diff --git a/core/runtime/os_specific_windows.odin b/core/runtime/os_specific_windows.odin
index 9f001fa5a..4a5907466 100644
--- a/core/runtime/os_specific_windows.odin
+++ b/core/runtime/os_specific_windows.odin
@@ -4,7 +4,7 @@ package runtime
 foreign import kernel32 "system:Kernel32.lib"
 
 @(private="file")
-@(default_calling_convention="stdcall")
+@(default_calling_convention="system")
 foreign kernel32 {
 	// NOTE(bill): The types are not using the standard names (e.g. DWORD and LPVOID) to just minimizing the dependency
 
diff --git a/core/runtime/procs.odin b/core/runtime/procs.odin
index 1592c608b..454574c35 100644
--- a/core/runtime/procs.odin
+++ b/core/runtime/procs.odin
@@ -4,7 +4,7 @@ when ODIN_NO_CRT && ODIN_OS == .Windows {
 	foreign import lib "system:NtDll.lib"
 	
 	@(private="file")
-	@(default_calling_convention="stdcall")
+	@(default_calling_convention="system")
 	foreign lib {
 		RtlMoveMemory :: proc(dst, s: rawptr, length: int) ---
 		RtlFillMemory :: proc(dst: rawptr, length: int, fill: i32) ---
@@ -37,7 +37,18 @@ when ODIN_NO_CRT && ODIN_OS == .Windows {
 		}
 		return ptr
 	}
-	
+
+	@(link_name="bzero", linkage="strong", require)
+	bzero :: proc "c" (ptr: rawptr, len: int) -> rawptr {
+		if ptr != nil && len != 0 {
+			p := ([^]byte)(ptr)
+			for i := 0; i < len; i += 1 {
+				p[i] = 0
+			}
+		}
+		return ptr
+	}
+
 	@(link_name="memmove", linkage="strong", require)
 	memmove :: proc "c" (dst, src: rawptr, len: int) -> rawptr {
 		d, s := ([^]byte)(dst), ([^]byte)(src)
diff --git a/core/runtime/procs_windows_amd64.odin b/core/runtime/procs_windows_amd64.odin
index e430357be..ea495f5fa 100644
--- a/core/runtime/procs_windows_amd64.odin
+++ b/core/runtime/procs_windows_amd64.odin
@@ -1,11 +1,12 @@
 //+private
+//+no-instrumentation
 package runtime
 
 foreign import kernel32 "system:Kernel32.lib"
 
 @(private)
 foreign kernel32 {
-	RaiseException :: proc "stdcall" (dwExceptionCode, dwExceptionFlags, nNumberOfArguments: u32, lpArguments: ^uint) -> ! ---
+	RaiseException :: proc "system" (dwExceptionCode, dwExceptionFlags, nNumberOfArguments: u32, lpArguments: ^uint) -> ! ---
 }
 
 windows_trap_array_bounds :: proc "contextless" () -> ! {
diff --git a/core/runtime/procs_windows_i386.odin b/core/runtime/procs_windows_i386.odin
index f810197f1..10422cf07 100644
--- a/core/runtime/procs_windows_i386.odin
+++ b/core/runtime/procs_windows_i386.odin
@@ -1,4 +1,5 @@
 //+private
+//+no-instrumentation
 package runtime
 
 @require foreign import "system:int64.lib"
@@ -12,7 +13,7 @@ windows_trap_array_bounds :: proc "contextless" () -> ! {
 	EXCEPTION_ARRAY_BOUNDS_EXCEEDED :: 0xC000008C
 
 	foreign kernel32 {
-		RaiseException :: proc "stdcall" (dwExceptionCode, dwExceptionFlags, nNumberOfArguments: DWORD, lpArguments: ^ULONG_PTR) -> ! ---
+		RaiseException :: proc "system" (dwExceptionCode, dwExceptionFlags, nNumberOfArguments: DWORD, lpArguments: ^ULONG_PTR) -> ! ---
 	}
 
 	RaiseException(EXCEPTION_ARRAY_BOUNDS_EXCEEDED, 0, 0, nil)