Overlay draw callback: publish fn+ctx atomically; link it in GL fixtures

Review (#599): fn and ctx were two separate stores read on the render
thread, so an unregister could pair a live fn with a NULL ctx. They are now
one immutable record behind an acquire/release atomic pointer. The GL
fixture runners compile a fixed source list; gpu_gl_renderer.c now calls
psx_host_overlay_draw, so every run_gl_*.py links host_overlay.c.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

runtime/src/host_overlay.c

@@ -1,7 +1,9 @@
 /* host_overlay.c -- see host_overlay.h. */
 #include "host_overlay.h"
 
+#include <stdatomic.h>
 #include <stddef.h>
+#include <stdlib.h>
 
 /* ---- P1: host pause ------------------------------------------------------ */
 #define PSX_HOST_PAUSE_MAX 8
@@ -29,19 +31,35 @@ const char *psx_host_pause_reason(void) {
 void psx_host_pause_set_refuse_probe(PsxHostPauseRefuseFn fn) { s_pause_refuse = fn; }
 
 /* ---- P2: overlay draw callback -------------------------------------------- */
-/* Written on the main thread before the overlay can open, read on the GL
- * thread; a pointer-sized store, and registration happens once at startup. */
-static PsxHostOverlayDrawFn volatile s_draw_fn = NULL;
-static void *volatile s_draw_ctx = NULL;
+/* Registered on the main thread, read on the render / present thread. The
+ * function and its context are published together as one immutable record
+ * behind an atomic pointer (release on store, acquire on load), so a reader
+ * can never pair a live function with a stale or NULL context. Records are
+ * never freed: registration happens a handful of times per process, and a
+ * reader may still hold the previous record when it is replaced. */
+typedef struct {
+    PsxHostOverlayDrawFn fn;
+    void *ctx;
+} PsxHostOverlayDraw;
+
+static _Atomic(PsxHostOverlayDraw *) s_draw = NULL;
 
 void psx_host_overlay_set_draw_cb(PsxHostOverlayDrawFn fn, void *ctx) {
-    s_draw_ctx = ctx;
-    s_draw_fn = fn;
+    PsxHostOverlayDraw *rec = NULL;
+    if (fn) {
+        rec = (PsxHostOverlayDraw *)malloc(sizeof(*rec));
+        if (!rec) return;
+        rec->fn = fn;
+        rec->ctx = ctx;
+    }
+    atomic_store_explicit(&s_draw, rec, memory_order_release);
 }
 
-int psx_host_overlay_has_draw(void) { return s_draw_fn != NULL; }
+int psx_host_overlay_has_draw(void) {
+    return atomic_load_explicit(&s_draw, memory_order_acquire) != NULL;
+}
 
 void psx_host_overlay_draw(int width, int height) {
-    PsxHostOverlayDrawFn fn = s_draw_fn;
-    if (fn && width > 0 && height > 0) fn(width, height, s_draw_ctx);
+    const PsxHostOverlayDraw *rec = atomic_load_explicit(&s_draw, memory_order_acquire);
+    if (rec && rec->fn && width > 0 && height > 0) rec->fn(width, height, rec->ctx);
 }

runtime/tests/run_gl_frame_gen.py

@@ -95,6 +95,9 @@ def main():
     if WINDOWS:
         sections.append("-flto")
     objs = []
+    # The renderer calls the in-game overlay hook (host_overlay.c) at present.
+    if (src / "host_overlay.c").exists():
+        sources.append(("hov", src / "host_overlay.c"))
     for name, path in sources:
         o = dest / (name + ".o")
         r = run([args.cc, "-std=gnu11", "-O1", "-DPSX_SDL3=1", "-DPSX_NO_DEBUG_TOOLS=1",

runtime/tests/run_gl_readback_region.py

@@ -77,6 +77,9 @@ def main():
         sources.append(("pth", framework / "runtime/src/present_thread.c"))
     if (framework / "runtime/src/frame_gen.c").exists():
         sources.append(("fg", framework / "runtime/src/frame_gen.c"))
+    # The renderer calls the in-game overlay hook (host_overlay.c) at present.
+    if (framework / "runtime/src" / "host_overlay.c").exists():
+        sources.append(("hov", framework / "runtime/src" / "host_overlay.c"))
     for name, source in sources:
         if run([compiler / "gcc.exe", "-std=c11", "-O2", "-flto", "-DPSX_SDL3=1",
                 "-DPSX_NO_DEBUG_TOOLS=1", *includes, "-c", source,

runtime/tests/run_gl_render_thread.py

@@ -94,6 +94,9 @@ def main():
     if WINDOWS:
         sections.append("-flto")
     objs = []
+    # The renderer calls the in-game overlay hook (host_overlay.c) at present.
+    if (src / "host_overlay.c").exists():
+        sources.append(("hov", src / "host_overlay.c"))
     for name, path in sources:
         o = dest / (name + ".o")
         r = run([args.cc, "-std=gnu11", "-O1", "-DPSX_SDL3=1", "-DPSX_NO_DEBUG_TOOLS=1",

runtime/tests/run_gl_scale_invariance.py

@@ -154,6 +154,9 @@ def main():
     if passes:
         sections.append("-DPSX_TEST_RENDER_PASSES=1")
     objs = []
+    # The renderer calls the in-game overlay hook (host_overlay.c) at present.
+    if (framework / "runtime/src" / "host_overlay.c").exists():
+        sources.append(("hov", framework / "runtime/src" / "host_overlay.c"))
     for name, src in sources:
         o = dest / (name + ".o")
         r = run([args.cc, "-std=gnu11", "-O1", "-DPSX_SDL3=1", "-DPSX_NO_DEBUG_TOOLS=1",

runtime/tests/run_gl_texture_filter.py

@@ -52,6 +52,9 @@ def main():
     if (fw/'runtime/src/frame_gen.c').exists():
         sources.append(('fg',fw/'runtime/src/frame_gen.c'))
     objects=[]
+    # The renderer calls the in-game overlay hook (host_overlay.c) at present.
+    if (fw/'runtime/src' / "host_overlay.c").exists():
+        sources.append(("hov", fw/'runtime/src' / "host_overlay.c"))
     for name,source in sources:
         obj=out/(name+'.o'); run([args.cc,*flags,'-c',source,'-o',obj]); objects.append(obj)
     if args.hd_pack: