custom crashes when exiting the music import screen

[ohrrpgce-bugbot]

[bz#400]

When you exit the music import screen with a music_native build of custom, custom crashes.

It does not crash when exiting the sound effects import screen.

This might be dependant on what types of files you have previewed before exiting the music import screen. I am not sure yet.

From: @bob-the-hamster
Reported version: 20070921 Ubersetzung
Blocker for: bz⁠#255

[ohrrpgce-bugbot]

Comment author: @ziggythehamster

I'm getting something similar (latest SVN) with music_sdl. I added some music, and it was one of the "unsupported" ones (i.e. 16000Hz). Exited the menu, CUSTOM still works. Attempted to go immediately back in, CUSTOM crashes.

[ohrrpgce-bugbot]

Comment author: @ziggythehamster

Okay, it ALWAYS crashes, even if good files are used.

[ohrrpgce-bugbot]

Comment author: @rversteegen

I just got a crash when exiting the import music menu after importing something from the import folder (after a bit of browsing, I think). gfx_fb and music_native, win xp, I was running under gdb:

Program received signal SIGSEGV, Segmentation fault.
0x00470ddd in UNLOADSOUND (SLOT=2) at music_native.bas:947
947 .used = 0
(gdb) info threads
9 thread 2080.0xfc0 0x7c90eb94 in audwrap_dll_iname ()
8 thread 2080.0xb80 0x7c90eb94 in audwrap_dll_iname ()
7 thread 2080.0xd48 0x7c90eb94 in audwrap_dll_iname ()
6 thread 2080.0xc3c 0x7c90eb94 in audwrap_dll_iname ()
5 thread 2080.0x984 0x7c90eb94 in audwrap_dll_iname ()
4 thread 2080.0x99c 0x7c90eb94 in audwrap_dll_iname ()
3 thread 2080.0x654 0x7c90eb94 in audwrap_dll_iname ()
2 thread 2080.0xe2c 0x7c90eb94 in audwrap_dll_iname ()
* 1 thread 2080.0xa40 0x00470ddd in UNLOADSOUND (SLOT=2)
at music_native.bas:947
(gdb) bt
#⁠0 0x00470ddd in UNLOADSOUND (SLOT=2) at music_native.bas:947
#⁠1 0x004705b6 in SOUND_CLOSE () at music_native.bas:767
#⁠2 0x004662fd in CLOSEMUSIC () at allmodex.bas:2177
#⁠3 0x0045bbce in IMPORTSONG () at menus.bas:871
#⁠4 0x0040c122 in main ({ARGC}=1, {ARGV}=0x3d3e38) at custom.bas:252
(gdb) p SOUNDPOOL
$1 = (struct SOUNDEFFECT *) 0x3dffe8
(gdb) p SOUNDPOOL[2]
$2 = {USED = 35651584, EFFECTID = 0, SOUNDID = 590440, PAUSED = 36175880}
(gdb) p SOUNDPOOLSIZE
$3 = 3

I can't reproduce it, though

[ohrrpgce-bugbot]

Comment author: @bob-the-hamster

Kieth, can you reproduce this bug with current music_sdl nightlies?

I am still not sure if this is a non-backend-specific crash, or two separate crashes with separate causes in each backend that just happen to occur at the same place.

[ohrrpgce-bugbot]

Comment author: @ziggythehamster

I need to update the ohrrpgce-custom-wip ebuild for the latest changes (browse.bas, madplay, oggenc). I will test for you probably Sunday and get back to you. The last nightly I used still exhibited this bug.

[ohrrpgce-bugbot]

Comment author: @rversteegen

So, Keith, are you saying that (with music_sdl) it crashes when entering, or exiting, the music screen?

[ohrrpgce-bugbot]

Comment author: @rversteegen

I'm now seeing a crash with music_native 100% of the time.

I open a new musicless RPG, go to import some music, and it crashes instead of showing the browser. Cancel8-Bit.ogg is in my wip folder, and the browser would try to play it.
However, running with gdb (even without compiling with -g) makes the crash disappear. Compiling with -exx reports nothing. Adding too many debugs to track it down makes it disappear too!

Eventually, I found that it's happening in the same place as I reported above.

Actually, before, it was crashing elsewhere:
opening a new RPG, going into the browser and listening to that OGG, and attempting to quit from the music menu to the main menu. It's not doing that now, I think it was doing that when I tried compiling with FB0.17 instead of 0.16.

[ohrrpgce-bugbot]

Comment author: @rversteegen

Opps, it's crashing on loading, not unloading the sound. Specifically, in this block in LoadSound():

with SoundPool[i]
.used = -1
.soundID = s '<-- line causing segfault
.effectID = num
end with

i = 0, SoundPool is a supposedly valid pointer from Reallocate, and .used = -1 does not segfault. How odd.

[ohrrpgce-bugbot]

Comment author: @rversteegen

Of course, SoundPool wasn't allocated correctly. It's hard to notice such things when you can barely focus on the screen in the small hours of the morning.

Now this is just a music_sdl bug.

However, I notice that this didn't fix bug bz⁠#222. We can ask everyone to retest all the music bugs after that one's crushed too.

[ohrrpgce-bugbot]

Comment author: @bob-the-hamster

Marking fixed. If there are still remaining music_sdl problems as described by Keith Gable, they can't share the same cause as this bug, so they either belong in bug bz⁠#372 or in a new bug.