| Re: [hatari-devel] Bug with latest releases on my Eletric night demo |
[ Thread Index |
Date Index
| More lists.tuxfamily.org/hatari-devel Archives
]
Le 26/01/2020 à 18:24, Nicolas Pomarède a écrit :
thanks for the example code you sent me ; meanwhile I looked more
closely and the problem is is fact similar to the fix I made for STE in
commit f9f1a01cea427bd77b7dc47d9dc5a0849f7a0ec4 from 11/1/2020 : the
int signal must be set to low when play stops, but it must also be set
to high when play starts (which is not the case actually).
If not then the "active edge register" will not see the transition
between play/stop and the interrupt will not trigger (previous falcon
code bypassed the AER and always triggered an interrupt when play
stopped, which is not the correct way to emulate the HW for this in the
MFP)
I will send a patch for this soon to fix the issue.
Hi
this is now fixed. There were 2 problems in Hatari :
- in Falcon mode, MFP GPIP7 was wrong since a long time as it should
not be "monitor type" XOR "dma sound state", but only "dma sound state"
(the MONO line is not connected to GPIP7, unlike on ST/STE/TT)
- for Falcon DMA sound, IRQ line was only set to 0 when DMA sound
stopped, but it was not set to 1 when DMA sound was active. Combined
with the change from 9/6/2019, this prevented MFP code to detect the
stop/play transition when DMA sound was in loop mode.
The 'Electric Night' demo is now working again (as with Hatari 2.2.1)
and MFP/DMA sound emulation is now closer to real HW too.
Nicolas