Super Mario Galaxy (1/2) Sound - Needs Developer Feedback

[Xtreme2damax]

Well I got an urge to try to help debug the sound issue with Super Mario Galaxy. It may not be considered debugging in the sense of development, but everything and everyone needs to start somewhere. I was testing Super Mario Galaxy 2, at the exact moment the music stopped looping I checked the log in /User/Logs, at exactly the same time the music stopped looping this is what Dolphin listed in the log:

Code:

53:08:946 .\Src\HW\ProcessorInterface.cpp:186 W[PI]: Fifo reset (00000001)
53:08:947 .\Src\HW\ProcessorInterface.cpp:186 W[PI]: Fifo reset (00000000)

The only probable explanation is as was mentioned that the HLE plugin isn't syncing with the core, causing the sound to stop looping on a fifo reset. Since the LLE plugin is in sync with the core, the music continues looping when the fifo is reset.

Perhaps skid or another developer can look into this and see if there is anything that can be done to fix this issue until work on LLE JIT starts progressing.

Then again it could be a coincidence, but it seems odd that this was listed in the logs at the exact time the music stopped looping, which was at 08:53 PM as listed above in the log.

---

Looks like it was just a coincidence, wasn't able to produce the same thing twice. :/

This was the last file loaded according to the logs, and is quite big in comparison to the other files:

38:38:205 .\Src\FileMonitor.cpp:109 N[FileMon]: 9,336 kB AudioRes/Stream/SMG2_galaxy32_strm.ast

Sorry guys, looks like I failed again.

[skid]

Keep looking. Observing the logs as the problem happens is a good idea.

[Xtreme2damax]

I'll keep trying.

Btw, is there any way to observe what the audio is doing in debug mode and compare the results between plugins?

[skid]

Make a debug build via VS2008. Use the Log Window to select Debug level logging. Select DSP Interface and DSP Mails.

The results might not be directly comparable - HLE is too high level to give the detail that LLE provides. But give it a try anyway, you never know.

[Xtreme2damax]

Well even if they aren't the same, with a little debugging I might be able to find out what the game or emulator is doing when the music stops looping. Results could then be compared to the actual hardware, someone else would need to do this as I bricked my Wii a while back and sold it.

[razor1911]

I would be interested in this, so I am willing to test if needed

[Torin]

the second time of debug, the loop worked a bit longer?

[RupeeClock]

Well this is just an observation, but I was testing SMG2 on r5889 just a moment ago, and got a Fifo Overflown Fatal crash.

Code:

08:15:967 .\Src\HW\ProcessorInterface.cpp:186 W[PI]: Fifo reset (00000001)
08:15:967 .\Src\HW\ProcessorInterface.cpp:186 W[PI]: Fifo reset (00000000)
08:15:967 .\Src\MsgHandler.cpp:53 E[*]: Warning: FIFO is overflown by GatherPipe !
CPU thread is too fast, lower the HiWatermark may help.

The music had already stopped playing a while ago, but what's curious is the first two lines are identical to Xtreme2damax's error, but his didn't crash, the music just stops.

Could there be a connection?

[Xtreme2damax]

I still have yet to test with the debugger, will update everyone if I discover anything useful.

[StripTheSoul]

@ RubeeClock: Huh? Are you saying the music stopping could be due to FIFO issues?