In NSF export, DPCM notes reset Vxx duty cycle of Pulse 1

[nyanpasu64]

In NSF export, after playing a DPCM note, the Vxx value of Pulse 1 resets to V00 for the next pulse note. The Vxx of Pulse 2 is not corrupted.

Test case: nsf Vxx dpcm bug.zip

The NSF export bug happens in both main, and #262 rebased on main.

[Gumball2415]

there is a stray F volume next to the DPCM note, thus being exported as a volume command. for some reason this is not rendered in the pattern editor.

volume commands affect volume slide channel arrays var_ch_VolSlide and var_ch_VolSlideTarget. they get reset to #$00 and #$80 respectively upon a new volume command.

var_ch_VolSlide and var_ch_VolSlideTarget is reserved with size EFF_CHANS (pulse 1 and 2, triangle, noise in this case).

in processing the stray volume command on the DPCM channel, this causes var_ch_VolSlideTarget+4 to be written with #$80, which is where var_ch_DutyDefault is.

setting var_ch_DutyDefault to this value effectively resets the duty cycle back to 0 (since #$80 & #$03 is being written as pulse 1 duty cycle).

[Gumball2415]

can you test if this build fixes the issue?
https://github.com/Dn-Programming-Core-Management/Dn-FamiTracker/actions/runs/12496933751

[nyanpasu64]

can you test if this build fixes the issue? https://github.com/Dn-Programming-Core-Management/Dn-FamiTracker/actions/runs/12496933751

Bug seems to not occur?