AudioWorklet driver, misc. emscripten fixes #17750
Conversation
|
Just bumped the default audio latency for AudioWorklet back down to 64 ms, though rwebaudio and OpenAL still need the higher latency in emscripten. BTW - for building RetroArch with AudioWorklet, use these extra variables: |
Makefile.emscripten
Outdated
| $(Q)$(LD) -o $@ $(RARCH_OBJ) $(libretro_new) $(LIBS) $(LDFLAGS) | ||
| @if [ "$(HAVE_AUDIOWORKLET)" = 1 ]; then\ | ||
| tr -d '\n' < "$${TARGET%%\.js}.aw.js" | sed -e "s/[\/&]/\\\\&/g" -e "s/'/\\\\\\\\&/g" > _audioworklet.js;\ | ||
| sed -i "s/\"$${TARGET%%\.js}\.aw\.js\"/URL.createObjectURL(new Blob(['$$(cat _audioworklet.js)'],{type:'text\/javascript'}))/" "$@";\ |
There was a problem hiding this comment.
sed and make both have some differences on Mac OS here, so I was able to get it compiling on Mac and Linux with this variation of the $(TARGET) rule (there is a little duplication unfortunately).
I also tweaked the $(libretro_new) rule to only do the mv if the file with the wrong suffix exists. mv -f ... || true would be another valid choice there.
diff --git a/Makefile.emscripten b/Makefile.emscripten
index 83db5df9b0..36aeb6306c 100644
--- a/Makefile.emscripten
+++ b/Makefile.emscripten
@@ -293,18 +293,23 @@ RARCH_OBJ := $(addprefix $(OBJDIR)/,$(OBJ))
all: $(TARGET)
-$(libretro_new) : $(libretro)
+$(libretro_new) : $(wildcard libretro)
mv -f $(libretro) $(libretro_new)
# until emscripten adds something like WASM_WORKERS=2 but for audio worklets, DIY.
+ifeq ($(HAVE_AUDIOWORKLET), 1)
+TARGET_BASE := $(LIBRETRO)_libretro
+$(TARGET): $(RARCH_OBJ) $(libretro_new)
+ @$(if $(Q), $(shell echo echo "LD $@ \<obj\> $(libretro_new) $(LIBS) $(LDFLAGS)"),)
+ $(Q)$(LD) -o $@ $(RARCH_OBJ) $(libretro_new) $(LIBS) $(LDFLAGS)
+ $(Q)tr -d '\n' < "$(TARGET_BASE).aw.js" | sed -e "s/[\/&]/\\\\&/g" -e "s/'/\\\\\\\\&/g" > _audioworklet.js
+ $(Q)sed -i.bak -e "s/\"$(TARGET_BASE)\.aw\.js\"/URL.createObjectURL(new Blob(['$$(cat _audioworklet.js)'],{type:'text\/javascript'}))/" -- "$@"
+ $(Q)rm -f "$(TARGET_BASE).aw.js" _audioworklet.js "$@".bak
+else
$(TARGET): $(RARCH_OBJ) $(libretro_new)
@$(if $(Q), $(shell echo echo "LD $@ \<obj\> $(libretro_new) $(LIBS) $(LDFLAGS)"),)
$(Q)$(LD) -o $@ $(RARCH_OBJ) $(libretro_new) $(LIBS) $(LDFLAGS)
- @if [ "$(HAVE_AUDIOWORKLET)" = 1 ]; then\
- tr -d '\n' < "$${TARGET%%\.js}.aw.js" | sed -e "s/[\/&]/\\\\&/g" -e "s/'/\\\\\\\\&/g" > _audioworklet.js;\
- sed -i "s/\"$${TARGET%%\.js}\.aw\.js\"/URL.createObjectURL(new Blob(['$$(cat _audioworklet.js)'],{type:'text\/javascript'}))/" "$@";\
- rm "$${TARGET%%\.js}.aw.js" _audioworklet.js;\
- fi
+endif
$(OBJDIR)/%.o: %.c
@mkdir -p $(dir $@)
| #elif defined(__wiiu__) | ||
| #include "../audio/drivers/wiiu_audio.c" | ||
| #elif defined(EMSCRIPTEN) | ||
| #elif defined(HAVE_RWEBAUDIO) |
There was a problem hiding this comment.
Do we need to add the audioworklet driver in here somewhere?
There was a problem hiding this comment.
I'm not sure why rwebaudio is in griffin in the first place, maybe emscripten used to build using griffin? It probably doesn't matter, but I fixed this condition just in case.
There was a problem hiding this comment.
Nice, I like these improvements to my sketch for cross platform sed. I’ll review the code changes once I do some testing tomorrow.
|
There are still a couple more (minor and iOS safari specific) issues that I would like to resolve before this is merged. |
|
That should be it! We're thinking the other bug - occasionally freezing - was due to using an out of date emscripten version (I was unable to reproduce it.) |
| emscripten_thread_sleep(msec); | ||
| } | ||
| #elif defined(EMSCRIPTEN) | ||
| #define retro_sleep(msec) (emscripten_thread_sleep(msec)) |
There was a problem hiding this comment.
Shouldn't you also check to make sure this was at least compiled with threads? Or are we assuming asyncify or threads is always true?
There was a problem hiding this comment.
See above condition, which will be preferred if it isn't compiled with threads.
There was a problem hiding this comment.
Oh - emscripten_thread_sleep also works on the main thread, but it busy spins so it just locks the page. It isn't used in that case anyway.
|
Is it alright if I merge this? Or does work still need to be done? |
|
Yep, this is ready! |
4 participants
Changes:
HAVE_EXTRA_WASMFS=1to enable OPFS/FetchFS now;HAVE_WASMFS=1will only enable WasmFS and not the extra FS backends.platform_emscripten.hsince we have several functions used from there nowBTW - regarding #15479 (comment):
This seems to be a firefox windows issue. Even luser.github.io/gamepadtest shows the same behavior.
Let me know what you think! Also, this still needs to be tested on safari.