From e5980925652e161f099fce678e7d2df378c82341 Mon Sep 17 00:00:00 2001 From: mamehaze Date: Thu, 19 Mar 2015 10:15:06 +0000 Subject: [PATCH] irq note (nw) --- src/mame/audio/hng64.c | 15 +++++++++++---- 1 file changed, 11 insertions(+), 4 deletions(-) diff --git a/src/mame/audio/hng64.c b/src/mame/audio/hng64.c index dbce6f5c885..2cc54a21ed2 100644 --- a/src/mame/audio/hng64.c +++ b/src/mame/audio/hng64.c @@ -21,11 +21,18 @@ data structures look very similar between all of them IRQ mask register on the internal interrupt controller is set to 0xd8 -so levels 0,1,2,5 are unmasked +so levels 0,1,2,5 are unmasked, vectors get set during the sound CPU init code. -returning random values / triggering random interrupts eventually results in a situation -where the CPU stops writing to the sound related addresses and starts reading / masking the -serial comms register. + level 0/1 irq (fatfurwa) starts at 0xd277 (both the same vector) + serial comms related, maybe to get commands from main CPU if not done with shared ram? + + level 2 irq (fatfurwa) 0xdd20 + simple routine increases counter in RAM, maybe hooked to one / all of the timer irqs + + level 5 irq: (fatfurwa) starts at 0xc1e1 + largest irq, does things with ports 100 / 102 / 104 / 106, 10a (not 108 directly tho) + + no other irqs (or the NMI) are valid. */