atomics as macros

src/base/base.h

@@ -209,8 +209,6 @@ void __asan_unpoison_memory_region(void const volatile *add, size_t);
 //- Static
 #define PERSIST static
 #define Global static
-/* TODO: Remove this */
-#define internal static
 
 //- Read-only
 #if PlatformIsWindows
@@ -543,29 +541,29 @@ StaticAssert(sizeof(Atomic64Padded) == 64 && alignof(Atomic64Padded) == 64);
 
 #if PlatformIsWindows && ArchIsX64
 //- 8 bit atomic operations
-ForceInline i8 Atomic8Fetch(Atomic8 *x) { return x->_v;  }
-ForceInline i8 Atomic8FetchSet(Atomic8 *x, i8 e) { return (i8)_InterlockedExchange8((volatile char *)&x->_v, e); }
-ForceInline i8 Atomic8FetchTestSet(Atomic8 *x, i8 c, i8 e) { return (i8)_InterlockedCompareExchange8((volatile char *)&x->_v, e, c); }
-ForceInline i8 Atomic8FetchXor(Atomic8 *x, i8 c) { return (i8)_InterlockedXor8((volatile char *)&x->_v, c); }
-ForceInline i8 Atomic8FetchAdd(Atomic8 *x, i8 a) { return (i8)_InterlockedExchangeAdd8((volatile char *)&x->_v, a); }
+#define Atomic8Fetch(x)                 (x)->_v
+#define Atomic8FetchSet(x, e)           (i8)_InterlockedExchange8((volatile char *)&(x)->_v, (e))
+#define Atomic8FetchTestSet(x, c, e)    (i8)_InterlockedCompareExchange8((volatile char *)&(x)->_v, (e), (c))
+#define Atomic8FetchXor(x, c)           (i8)_InterlockedXor8((volatile char *)&(x)->_v, (c))
+#define Atomic8FetchAdd(x, a)           (i8)_InterlockedExchangeAdd8((volatile char *)&(x)->_v, (a))
 //- 16 bit atomic operations
-ForceInline i16 Atomic16Fetch(Atomic16 *x) { return x->_v; }
-ForceInline i16 Atomic16FetchSet(Atomic16 *x, i16 e) { return (i16)_InterlockedExchange16(&x->_v, e); }
-ForceInline i16 Atomic16FetchTestSet(Atomic16 *x, i16 c, i16 e) { return (i16)_InterlockedCompareExchange16(&x->_v, e, c); }
-ForceInline i16 Atomic16FetchTestXor(Atomic16 *x, i16 c) { return (i16)_InterlockedXor16(&x->_v, c); }
-ForceInline i16 Atomic16FetchTestAdd(Atomic16 *x, i16 a) { return (i16)_InterlockedExchangeAdd16(&x->_v, a); }
+#define Atomic16Fetch(x)                (x)->_v
+#define Atomic16FetchSet(x, e)          (i16)_InterlockedExchange16(&(x)->_v, (e))
+#define Atomic16FetchTestSet(x, c, e)   (i16)_InterlockedCompareExchange16(&(x)->_v, (e), (c))
+#define Atomic16FetchTestXor(x, c)      (i16)_InterlockedXor16(&(x)->_v, (c))
+#define Atomic16FetchTestAdd(x, a)      (i16)_InterlockedExchangeAdd16(&(x)->_v, (a))
 //- 32 bit atomic operations
-ForceInline i32 Atomic32Fetch(Atomic32 *x) { return x->_v; }
-ForceInline i32 Atomic32FetchSet(Atomic32 *x, i32 e) { return (i32)_InterlockedExchange((volatile long *)&x->_v, e); }
-ForceInline i32 Atomic32FetchTestSet(Atomic32 *x, i32 c, i32 e) { return (i32)_InterlockedCompareExchange((volatile long *)&x->_v, e, c); }
-ForceInline i32 Atomic32FetchXor(Atomic32 *x, i32 c) { return (i32)_InterlockedXor((volatile long *)&x->_v, c); }
-ForceInline i32 Atomic32FetchAdd(Atomic32 *x, i32 a) { return (i32)_InterlockedExchangeAdd((volatile long *)&x->_v, a); }
+#define Atomic32Fetch(x)                (x)->_v
+#define Atomic32FetchSet(x, e)          (i32)_InterlockedExchange((volatile long *)&(x)->_v, (e))
+#define Atomic32FetchTestSet(x, c, e)   (i32)_InterlockedCompareExchange((volatile long *)&(x)->_v, (e), (c))
+#define Atomic32FetchXor(x, c)          (i32)_InterlockedXor((volatile long *)&(x)->_v, (c))
+#define Atomic32FetchAdd(x, a)          (i32)_InterlockedExchangeAdd((volatile long *)&(x)->_v, (a))
 //- 64 bit atomic operations
-ForceInline i64 Atomic64Fetch(Atomic64 *x) { return x->_v; }
-ForceInline i64 Atomic64FetchSet(Atomic64 *x, i64 e) { return (i64)_InterlockedExchange64(&x->_v, e); }
-ForceInline i64 Atomic64FetchTestSet(Atomic64 *x, i64 c, i64 e) { return (i64)_InterlockedCompareExchange64(&x->_v, e, c); }
-ForceInline i64 Atomic64FetchXor(Atomic64 *x, i64 c) { return (i64)_InterlockedXor64(&x->_v, c); }
-ForceInline i64 Atomic64FetchAdd(Atomic64 *x, i64 a) { return (i64)_InterlockedExchangeAdd64(&x->_v, a); }
+#define Atomic64Fetch(x)                (x)->_v
+#define Atomic64FetchSet(x, e)          (i64)_InterlockedExchange64(&(x)->_v, (e))
+#define Atomic64FetchTestSet(x, c, e)   (i64)_InterlockedCompareExchange64(&(x)->_v, (e), (c))
+#define Atomic64FetchXor(x, c)          (i64)_InterlockedXor64(&(x)->_v, (c))
+#define Atomic64FetchAdd(x, a)          (i64)_InterlockedExchangeAdd64(&(x)->_v, (a))
 #else
 # error Atomics not implemented
 #endif
@@ -705,7 +703,7 @@ Struct(ComputeShader)   { Resource resource; };
 
 #if LanguageIsC
 # if PlatformIsWindows
-#  define FiberId() (*(i16 *)(void *)(volatile u64)__readgsqword(32))
+#  define FiberId() (*(volatile i16 *)__readgsqword(32))
 # else
 #  error FiberId not implemented
 # endif

src/base/base_win32/base_win32_job.c

@@ -837,11 +837,10 @@ ForceInline W32_Fiber *W32_FiberFromId(i16 id)
 }
 
 //- Fiber control flow
-ForceNoInline void W32_FiberResume(W32_Fiber *fiber)
+
+void W32_FiberResume(W32_Fiber *fiber)
 {
-    MemoryBarrier();
     SwitchToFiber(fiber->addr);
-    MemoryBarrier();
 }
 
 void W32_YieldFiber(W32_Fiber *fiber, W32_Fiber *parent_fiber)

src/base/base_win32/base_win32_job.h

@@ -288,7 +288,7 @@ void W32_WakeByTime(u64 time);
 W32_Fiber *W32_AcquireFiber(W32_JobPool *pool);
 void W32_ReleaseFiber(W32_JobPool *pool, W32_Fiber *fiber);
 ForceInline W32_Fiber *W32_FiberFromId(i16 id);
-ForceNoInline void W32_FiberResume(W32_Fiber *fiber);
+void W32_FiberResume(W32_Fiber *fiber);
 void W32_YieldFiber(W32_Fiber *fiber, W32_Fiber *parent_fiber);
 
 ////////////////////////////////

src/meta/meta.c

@@ -884,6 +884,9 @@ void StartupMeta(void)
             PushStringToList(arena, &cp.flags_msvc, Lit("-INCREMENTAL:NO"));
             PushStringToList(arena, &cp.flags_msvc, Lit("-nologo"));
 
+            /* Optimization */
+            PushStringToList(arena, &cp.compiler_only_flags_msvc, Lit("-Od"));
+
             /* Debug info */
             PushStringToList(arena, &cp.flags_msvc, Lit("-DEBUG:FULL"));
             PushStringToList(arena, &cp.compiler_only_flags_msvc, Lit("-Z7"));