Mercurial > repos > blastem
changeset 1997:e08c489d5602
Fix regressions from most recent address/cd latch change. Need to do more research still, but probably good enough for now
| author | Mike Pavone <pavone@retrodev.com> |
|---|---|
| date | Sun, 28 Jun 2020 21:31:09 -0700 |
| parents | e35b00626b3e |
| children | 0740d90812ee |
| files | vdp.c |
| diffstat | 1 files changed, 4 insertions(+), 5 deletions(-) [+] |
line wrap: on
line diff
--- a/vdp.c Thu Jun 18 00:28:53 2020 -0700 +++ b/vdp.c Sun Jun 28 21:31:09 2020 -0700 @@ -3755,6 +3755,10 @@ if ((value & 0xC000) == 0x8000) { //Register write uint8_t reg = (value >> 8) & 0x1F; + // The fact that this is needed seems to pour some cold water on my theory + // about how the address latch actually works. Needs more search to definitively confirm + context->address = (context->address & 0x1C000) | (value & 0x3FFF); + context->cd = (context->cd & 0x3C) | (value >> 14); if (reg < (mode_5 ? VDP_REGS : 0xB)) { //printf("register %d set to %X\n", reg, value & 0xFF); if (reg == REG_MODE_1 && (value & BIT_HVC_LATCH) && !(context->regs[reg] & BIT_HVC_LATCH)) { @@ -3779,11 +3783,6 @@ update_video_params(context); } } - // The fact that this is needed seems to pour some cold water on my theory - // about how the address latch actually works. Needs more search to definitively confirm - clear_pending(context); - context->flags &= ~FLAG_READ_FETCHED; - context->flags2 &= ~FLAG2_READ_PENDING; } else if (mode_5) { context->flags |= FLAG_PENDING; //Should these be taken care of here or after the second write?