diff --git a/osdep/atomics.h b/osdep/atomics.h
index 8431211476..6997ba97b2 100644
--- a/osdep/atomics.h
+++ b/osdep/atomics.h
@@ -30,13 +30,14 @@
 
 // Emulate the parts of C11 stdatomic.h needed by mpv.
 // Still relies on gcc/clang atomic builtins.
+// The t field is a hack to make the non-atomic fallback macro mess work.
 
-typedef struct { volatile unsigned long v;  } atomic_ulong;
-typedef struct { volatile int v;            } atomic_int;
-typedef struct { volatile _Bool v;          } atomic_bool;
-typedef struct { volatile long long v;      } atomic_llong;
-typedef struct { volatile uint_least32_t v; } atomic_uint_least32_t;
-typedef struct { volatile unsigned long long v; } atomic_ullong;
+typedef struct { volatile unsigned long v, t;       } atomic_ulong;
+typedef struct { volatile int v, t;                 } atomic_int;
+typedef struct { volatile _Bool v, t;               } atomic_bool;
+typedef struct { volatile long long v, t;           } atomic_llong;
+typedef struct { volatile uint_least32_t v, t;      } atomic_uint_least32_t;
+typedef struct { volatile unsigned long long v, t;  } atomic_ullong;
 
 #define ATOMIC_VAR_INIT(x) \
     {.v = (x)}
@@ -54,6 +55,10 @@ typedef struct { volatile unsigned long long v; } atomic_ullong;
     __atomic_store_n(&(p)->v, val, __ATOMIC_SEQ_CST)
 #define atomic_fetch_add(a, b) \
     __atomic_fetch_add(&(a)->v, b, __ATOMIC_SEQ_CST)
+#define atomic_fetch_and(a, b) \
+    __atomic_fetch_and(&(a)->v, b, __ATOMIC_SEQ_CST)
+#define atomic_fetch_or(a, b) \
+    __atomic_fetch_or(&(a)->v, b, __ATOMIC_SEQ_CST)
 #define atomic_compare_exchange_strong(a, b, c) \
     __atomic_compare_exchange_n(&(a)->v, b, c, 0, __ATOMIC_SEQ_CST, \
     __ATOMIC_SEQ_CST)
@@ -66,6 +71,10 @@ typedef struct { volatile unsigned long long v; } atomic_ullong;
     (__sync_synchronize(), (p)->v = (val), __sync_synchronize())
 #define atomic_fetch_add(a, b) \
     __sync_fetch_and_add(&(a)->v, b)
+#define atomic_fetch_and(a, b) \
+    __sync_fetch_and_and(&(a)->v, b)
+#define atomic_fetch_or(a, b) \
+    __sync_fetch_and_or(&(a)->v, b)
 // Assumes __sync_val_compare_and_swap is "strong" (using the C11 meaning).
 #define atomic_compare_exchange_strong(p, old, new) \
     ({ __typeof__((p)->v) val_ = __sync_val_compare_and_swap(&(p)->v, *(old), new); \
@@ -79,7 +88,11 @@ typedef struct { volatile unsigned long long v; } atomic_ullong;
 // a serious dependency on working atomics, so this is barely ok.
 #define atomic_load(p) ((p)->v)
 #define atomic_store(p, val) ((p)->v = (val))
-#define atomic_fetch_add(a, b) (((a)->v += (b)) - (b))
+#define atomic_fetch_op_(a, b, op) \
+    ((a)->t = (a)->v, (a)->v = (a)->v op (b), (a)->t)
+#define atomic_fetch_add(a, b) atomic_fetch_op_(a, b, +)
+#define atomic_fetch_and(a, b) atomic_fetch_op_(a, b, &)
+#define atomic_fetch_or(a, b) atomic_fetch_op_(a, b, |)
 #define atomic_compare_exchange_strong(p, old, new) \
     ((p)->v == *(old) ? ((p)->v = (new), 1) : (*(old) = (p)->v, 0))