[IIDX] Crash when beginning play recording

[Radioo]

Game and version

IIDX 30-31

Version of spice2x

2024-01-20

Laptop

No

Attached log.txt file, if available

log.txt

Describe the issue

After implementing necessary network-side logic to enable play recording, when entering a song, the game crashes. I can elaborate what exactly needs to be sent from the server to enable this if needed. I am using an NVIDIA GPU (RTX 2060) which should work for this and I'm also using a 010 DLL.

[LupinThidr]

2024011600 isn't public, the call stack in your log is useless

[LupinThidr]

Is this also LDJ-010? I think the only public LDJ-010 DLL is 2023120600, so you should get a log with that one crashing.

[Radioo]

Apologies, updated the log in issue description. This one is from LDJ-010-2023120600

[sp2xdev]

Can you try disabling spice overlay and see if it helps? I guess you would need a touch screen to card in.

Does it work in btools?

[gar6a9e]

I have done some research on this issue in the past.

• It does not work with spicetools.
  • spice overlay disabled.
• It works with btools, but only in full-screen mode.
  • It does not work in window mode.
• An error occurs when calling NvEncoder::RegisterResource.

I'm not sure if this is entirely correct, but I have compiled it from my old notes. I think there might be some issues with DirectX hooks. I hope this information is helpful.

[sp2xdev]

If it isn't the overlay, I would strongly suspect subscreen swap chain logic (SUB_SWAP_CHAIN) is related, as it runs unconditionally. Might be worth checking if removing logic around it (start with graphics_d3d9_ldj_init_sub_screen) resolves this issue. If not that, perhaps the screenshot logic (which is also always-on).

Also worth checking what status code is returned by NvEncoder::RegisterResource, might give us a hint as to why the game is unhappy.

Is it simple to add enable recordings; e.g., to an asphyxia plugin? Would be super useful for testing purposes.

[Radioo]

I'm afraid it won't be simple to add it to asphyxia as the game requires a separate API server which functions nothing like the base XRPC server (It's just a simple JSON API). But I've just pushed an experimental toggle you can enable on the dev server I develop for. If you'd like to join, feel free to shoot me a PM on discord (my username: services.get).

[sp2xdev]

Okay, we have someone who knows more about iidx who can help out with this & will be contacting you shortly for server access

[sp2xdev]

Resolved in 24-01-27 with the new "disable all d3d9 hook" option. Not the most ideal way to resolve this, but at least brings it to parity with btools. Not fully tested as server support is still being worked on.

[gar6a9e]

Also worth checking what status code is returned by NvEncoder::RegisterResource, might give us a hint as to why the game is unhappy.

According to my old notes, it returned 9.

maybe NV_ENC_ERR_INVALID_CALL https://github.com/NVIDIA/video-sdk-samples/blob/aa3544dcea2fe63122e4feb83bf805ea40e58dbe/Samples/NvCodec/NvEncoder/nvEncodeAPI.h#L472

[llm96]

I was running into another crash despite having hooks disabled on 2023120600.

The game closed shortly before the song started with no logs, but I managed to get something by attaching a debugger.

Didn't happen on IIDX 30 with the same settings, so I guess something changed between the two games.

Switching to a different ASIO driver (FlexASIO to ASIO4ALL) worked for me.

[LupinThidr]

seems to be related to audio format conversion for writing to the video file (guessing from RTTI and log messages), documenting your flexasio and asio4all settings might be useful for future troubleshooting (known bad and good configurations)

[sp2xdev]

User on Radioo's server had similar issues with FlexASIO, but was able to resolve it by changing the configuration.

Bad config:

backend = "Windows WASAPI"
bufferSizeSamples = 386
channels = 2
wasapiExclusiveMode = false

[output]
suggestedLatencySeconds = 0.0

Good (fixed) config:

backend = "Windows WASAPI"
bufferSizeSamples = 128

[output]
device = "Speakers (Removed to protect privacy)"
sampleType = "Int32"
suggestedLatencySeconds = 0.0
wasapiExclusiveMode = false
wasapiAutoConvert = true
channels = 2

Can't tell you for sure what really matters, but if I had to guess, wasapiAutoConvert is needed to allow for resampling. YMMV since some aspects of FlexASIO depends on your actual audio card being output to.

Anyway, the crash seems unrelated to spice.

[llm96]

Thank you for taking a closer look at it. Strangely enough the known good FlexASIO config doesn't work for me either, so I'll just have to chalk it up to my audio hardware. As for my working ASIO4ALL settings, everything is set to default.

I hope documenting it here may help anyone encountering a similar crash in the future. Seems like most common issues with IIDX end up being audio related in one way or another! 😆