Update comments

This commit is contained in:
Lior Halphon 2018-04-07 03:36:47 +03:00
parent 9ce028056a
commit 097b768127
1 changed files with 8 additions and 7 deletions

View File

@ -84,19 +84,19 @@ static void fifo_overlay_object_row(GB_fifo_t *fifo, uint8_t lower, uint8_t uppe
/*
Each line is 456 cycles, approximately:
Each line is 456 cycles. Without scrolling, sprites or a window::
Mode 2 - 80 cycles / OAM Transfer
Mode 3 - 172 cycles / Rendering
Mode 0 - 204 cycles / HBlank
Mode 1 is VBlank
Todo: Mode lengths are not constants, see http://blog.kevtris.org/blogfiles/Nitty%20Gritty%20Gameboy%20VRAM%20Timing.txt
*/
/* Todo: Clean up the glitched line 0 and get rid of these defines */
#define MODE2_LENGTH (80)
#define MODE3_LENGTH (172)
#define MODE0_LENGTH (204)
#define LINE_LENGTH (MODE2_LENGTH + MODE3_LENGTH + MODE0_LENGTH) // = 456
#define LINES (144)
#define WIDTH (160)
@ -238,7 +238,7 @@ static void trigger_oam_interrupt(GB_gameboy_t *gb)
}
}
/* Todo: A proper test ROM of cases where both the PPU and the CPU write to IF in the same M-cycle is needed. */
/* Todo: When the CPU and PPU write to IF at the same T-cycle, the PPU write is ignored. */
void GB_STAT_update(GB_gameboy_t *gb)
{
if (!(gb->io_registers[GB_IO_LCDC] & 0x80)) return;
@ -479,8 +479,8 @@ static void advance_fetcher_state_machine(GB_gameboy_t *gb)
case GB_FETCHER_GET_TILE_DATA_HIGH: {
/* Todo: Verified for DMG (Tested: SGB2), CGB timing is wrong.
Additionally, on the CGB mixing two tiles by changing the tileset bit
mid-fetching causes a glitched mixing of the two, in comparison to the
Additionally, on CGB-D and newer mixing two tiles by changing the tileset
bit mid-fetching causes a glitched mixing of the two, in comparison to the
more logical DMG version. */
uint16_t tile_address = 0;
uint8_t y = gb->is_cgb? gb->fetcher_y : fetcher_y(gb);
@ -736,6 +736,7 @@ void GB_display_run(GB_gameboy_t *gb, uint8_t cycles)
if (object->flags & 0x40) { /* Flip Y */
tile_y ^= height_16? 0xF : 7;
}
/* Todo: I'm not 100% sure an access to OAM can't trigger the OAM bug while we're accessing this */
uint16_t line_address = (height_16? object->tile & 0xFE : object->tile) * 0x10 + tile_y * 2;
if (gb->cgb_mode && (object->flags & 0x8)) { /* Use VRAM bank 2 */