updateCallback3() would call unkOutput2(9) in at least one case. This is
obviously wrong because it a) reads outside _regOffsets[], and b) writes to
invalid Adlib registers.
Now unkOutput2() has the same safeguards as noteOff() already had, making
callbackOutput() and updateCallback3() even more similar.
svn-id: r21335
while still at the very beginning of the "attack" phase. This is the very
lowest point on the attack curve, yet it would continue from the beginning of
the release curve, i.e. its very highest point. This is what caused Kyra to
often play low-frequency notes at the very beginning of a new song. (That, and
a truly bizarre function for initialising the channels.)
The proper fix would be to locate the correct point on the release curve and
continue from there. For now, though, only handle the trivial case.
svn-id: r21302
bug is a bug in our fmopl.cpp. Ken Silverman's Adlib emulator doesn't seem to
exhibit the bug either. (But is apparently less feature complete than MAME's.)
svn-id: r21275
beginning of music" bugs. As an experiment, I tried hooking up ScummVM with the
most recent version of the FMOPL code, and the bug went away.
Unfortunately, I believe we are stuck with the ancient version for license (and
performance?) reasons.
svn-id: r21267
but it seems that it drifts out of sync after a few repeats even with the
original interpreter. It's a feature!
This cleanup clears up some of the more confusing break/continues in the
callbackProcess() function, and gets rid of _continueFlag.
svn-id: r21259
really mean "channel". Fixed that:
OutputState -> Channel
_curTable -> _curChannel
_outputTables[] -> _channels[]
The biggest change was modifying all the "state" and "table" local variables
and function parameters.
Ahd, at no extra charge:
_outputTable[] -> _regOffset[]
svn-id: r21247
generate "leap samples". I doubt this will make any noticeable difference, but
at least now we can probably rule out timer inaccuracies from the list of
possible suspects for the remaining bugs.
Also removed unnecessary mutex locking/unlocking from readBuffer(), spotted by
LordHoto.
svn-id: r21239
timer inaccuracies can cause that kind of problems, or we'd have noticed it in
the other game engines as well", guess what? The other game engines do not
necessary use a timer for their Adlib music. So now Kyra doesn't either.
Fortunately for my dignity, the music is still a bit uneven at times, but the
situation does seem to have improved a bit, and the sound effects sound better
to me now.
svn-id: r21237
update1() -> setupDuration()
updateAndOutput1() -> setupNote()
setInstrument() -> setupInstrument() (just for the symmetry)
updateAndOutput3() -> noteOn()
That made the following two renamings seem natural:
updateCallback10() -> update_playRest()
updateCallback27() -> update_playNote()
I decided to number the secondary effects, even though there's only one:
update_setupSecondaryEffect() -> update_setupSecondaryEffect1()
update_removeSecondaryEffect() -> update_removeSecondaryEffect1()
And finally, I renamed the effects callbacks:
stateCallback1_1() -> primaryEffect1()
stateCallback1_2() -> primaryEffect2()
stateCallback2_1() -> secondaryEffect1()
svn-id: r21232
callbackOutput() and updateCallback3() functions won't change the channel data
pointer unless the new sound has a higher unk2 and the current one.
Since it's set to 0 when the data pointer is nulled, I've changed the priority
to be unsigned.
The updateCallback22() function is now update_setPriority().
svn-id: r21206
level", which I believe is the individual channel volume. (The functions also
return the "scaling level" bits, but they are preserved, not calculated.) I
have renamed these functions calculateOpLevel1() and 2().
The unk25 and unk26 variables have been renamed opLevel1 and opLevel2. These
are called oplvl_1 and oplvl_2 in our MidiDriver_ADLIB class.
The unk26, unk27 and unk28 variables are potentially added to both operator
levels, and have been renamed opExtraLevel1, opExtralevel2 and opExtraLevel3.
The updateCallback25() function has been remamed update_setExtraLevel1().
The updateCallback33() function has been renamed update_setExtraLevel2().
The updateCallback32() function has been renamed update_setExtraLevel3().
Note that these callbacks aren't quite as similar as their names would seem to
indicate: They differ in whether or not the volume is updated and/or how the
parameters are passed.
svn-id: r21199
stuff, it pretty much has to be our "set instrument" function. Also, while I'm
not entirely sure of the implications, "unk23" is the Algorithm bit for the
Feedback / Algorithm register, so I've renamed it "algorithm".
svn-id: r21197
- Adds sound fading support to the adlib player
- Adds support for stopping running tracks
- Changes baseFreq type back to uint8
- corrects octave handling in updateAndOutput1
(Thanks to eriktorbjorn for doing that)
svn-id: r21185
all: it adjusts the frequency even when state.unk16 is zero. Fixed that, and
added some comments to hopefully explain the differences between the cleaned up
code and the original disassembly.
svn-id: r21184
* Rewrote updateCallback41() in terms of updateAndOutput1(), since they were
almost identical to begin with.
* Rewrote updateAndOutput1() for better readability.
* Made unk10 signed, and renamed it baseOctave.
* Made unk14 signed, and renamed it baseNote.
* Made unk15 signed, and renamed it baseFreq.
* Renamed unk17 regAx. It appears to be a cache for that register.
* Renamed unkOutputValue1 regBx. It appears to be a cache for that register.
* Renamed updateCallback8() update_setBaseOctave().
* Renamed updateCallback13() update_setBaseNote().
* Renamed updateCallback20() update_setBaseFreq().
Which still leaves the mystery of unk16 and _unkTables[]. Perhaps *this* is the
pitch bend?
svn-id: r21183
_unkTable[]. Investigating the latter further might help unravelling several of
the "unk" variables. (Right now, I'm suspecting that unk15 is the pitch bend.)
svn-id: r21164
the octave bits.
The unkOuput1() function always turns off the note, so I've renamed it
noteOff(). I've added some comments to unkOuput2() as well, but I'm still not
quite sure what its purpose is. It seems unlikely that it's simply a "note on"
function, given the many things it will clear for the channel. It does end by
turning the note on, though. Strange.
svn-id: r21162
