silson
/
gnuboy-for-dfi
2
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

import inofficial 1.0.4pre by joshua

This commit is contained in:
rofl0r 2012-06-23 14:19:47 +02:00
parent 29076d4cef
commit 9859b44888
9 changed files with 615 additions and 558 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

12
README
View File

@ -45,19 +45,20 @@ all of the following and much more:
Wave pattern memory corruption when sound channel 3 is played.
Pad, timer, divide counter, and other basic hardware registers.
CGB double-speed CPU mode.
Sorting sprites by X coordinate in DMG mode.
HALT instruction skipping in DMG mode.
CPU stalls during HDMA and GDMA.
Configurable color filters to provide more authentic LCD look.
Aspects not emulated at this time include:
* Serial IO (link cable).
Undocumented 'extra' ram in OAM space on Gameboy Color.
All Super Gameboy extensions.
* All Super Gameboy extensions.
* GBC, HuC1, and HuC3 IR ports.
* Obscure mappers such as TAMA5.
Sorting sprites by X coordinate in DMG mode.
HALT instruction skipping in DMG mode.
CPU stalls during HDMA and GDMA.
Only the two marked by * are known to affect the playability of
Only the ones marked by * are known to affect the playability of
actual games or demos; the rest are just listed for completeness'
sake.
@ -128,7 +129,6 @@ Here's a brief list of what may appear in gnuboy in the future:
Super Gameboy support.
Serial link over the internet.
Serial link to a real Gameboy with a custom cable.
Configurable color filters to provide more authentic LCD look.
Custom colorization of DMG games on a per-tile basis.
Support for more colorspaces in the hardware scaler.
Recording audio.

2
cpucore.h
View File

@ -7,7 +7,7 @@ const static byte cycles_table[256] =
1, 3, 2, 2, 1, 1, 2, 1, 5, 2, 2, 2, 1, 1, 2, 1,
1, 3, 2, 2, 1, 1, 2, 1, 3, 2, 2, 2, 1, 1, 2, 1,
3, 3, 2, 2, 1, 1, 2, 1, 3, 2, 2, 2, 1, 1, 2, 1,
3, 3, 2, 2, 1, 3, 3, 3, 3, 2, 2, 2, 1, 1, 2, 1,
3, 3, 2, 2, 3, 3, 3, 1, 3, 2, 2, 2, 1, 1, 2, 1,
1, 1, 1, 1, 1, 1, 2, 1, 1, 1, 1, 1, 1, 1, 2, 1,
1, 1, 1, 1, 1, 1, 2, 1, 1, 1, 1, 1, 1, 1, 2, 1,

16
docs/CHANGES
View File

@ -9,6 +9,22 @@ GNUBOY CHANGES FILE -- detailed list of all changes made
For an easy-to-read user-oriented list of major changes in each
release, please refer to the file WHATSNEW.
1.0.4-cvs
fixed a possible out-of-bounds access which could allow a hacked savestate to
crash gnuboy and possibly run arbitrary code on the host machine. Challenge:
try to find a way to exploit it
removed an incorrect restriction on when LY=LYC interrupts can occur, which
broke at least one game and several PD roms. (extra special thanks to beware
for helping us track this down!)
fixed an SDL keymap issue which prevented the keypad '5' key from working when
num-lock was off
replaced the channel 4 PRNG tables with a computer-generated PRNG output table,
as opposed to an approximation based on wavefiles
fixed the HDMA behavior (properly?), several pinball games developed by 'Left
Field Productions' now work correctly (thanks to beware for pointing us in the
right direction)
fixed GDMA so it consumes cpu cycles properly, now the intro to Magical Drop
works correctly
1.0.3
fixed a typo in the SDL keymap file that kept . from working

26
docs/WHATSNEW
View File

@ -8,6 +8,32 @@
CVS 1.0.4
Fixed a possible out-of-bounds access which could allow a hacked savestate to
crash gnuboy and possibly run arbitrary code on the host machine. Challenge:
try to find a way to exploit it
Removed an incorrect restriction on when LY=LYC interrupts can occur (from
0.9.13, possibly to fix montezuma's return or pokemon yellow, though both games
work fine now without it) which broke at least one commercial game and several
PD roms. (extra special thanks to beware for helping us track this down!)
Fixed an SDL keymap issue which prevented the keypad '5' key from working when
num-lock was off
Replaced the channel 4 PRNG tables with a shifter-generated PRNG output table,
as opposed to the old approximations based on wavefiles
Fixed the HDMA timing behavior (properly?), several pinball games developed by
'Left Field Productions' now work correctly (thanks to beware for pointing us
in the right direction)
Fixed the GDMA behavior so it consumes cpu cycles as it should. This fixes the
intro scene animation in 'Magical Drop'
RELEASE 1.0.3
All ANSI C incompatibilities should be fixed. Please report any that

72
hw.c
View File

@ -54,37 +54,12 @@ void hw_dma(byte b)
lcd.oam.mem[i] = readb(a);
}
void hw_hdma_cmd(byte c)
{
int cnt;
addr sa;
int da;
/* Begin or cancel HDMA */
if ((hw.hdma|c) & 0x80)
{
hw.hdma = c;
R_HDMA5 = c & 0x7f;
return;
}
/* Perform GDMA */
sa = ((addr)R_HDMA1 << 8) | (R_HDMA2&0xf0);
da = 0x8000 | ((int)(R_HDMA3&0x1f) << 8) | (R_HDMA4&0xf0);
cnt = ((int)c)+1;
/* FIXME - this should use cpu time! */
/*cpu_timers(102 * cnt);*/
cnt <<= 4;
while (cnt--)
writeb(da++, readb(sa++));
R_HDMA1 = sa >> 8;
R_HDMA2 = sa & 0xF0;
R_HDMA3 = 0x1F & (da >> 8);
R_HDMA4 = da & 0xF0;
R_HDMA5 = 0xFF;
}
/* COMMENT A:
* Beware was pretty sure that this HDMA implementation was incorrect, as when
* he used it in bgb, it broke Pokemon Crystal (J). I tested it with this and
* it seems to work fine, so until I find any problems with it, it's staying.
* (Lord Nightmare)
*/
void hw_hdma()
@ -98,6 +73,7 @@ void hw_hdma()
cnt = 16;
while (cnt--)
writeb(da++, readb(sa++));
cpu_timers(16); /* SEE COMMENT A ABOVE */
R_HDMA1 = sa >> 8;
R_HDMA2 = sa & 0xF0;
R_HDMA3 = 0x1F & (da >> 8);
@ -107,6 +83,40 @@ void hw_hdma()
}
void hw_hdma_cmd(byte c)
{
int cnt;
addr sa;
int da;
/* Begin or cancel HDMA */
if ((hw.hdma|c) & 0x80)
{
hw.hdma = c;
R_HDMA5 = c & 0x7f;
if ((R_STAT&0x03) == 0x00) hw_hdma(); /* SEE COMMENT A ABOVE */
return;
}
/* Perform GDMA */
sa = ((addr)R_HDMA1 << 8) | (R_HDMA2&0xf0);
da = 0x8000 | ((int)(R_HDMA3&0x1f) << 8) | (R_HDMA4&0xf0);
cnt = ((int)c)+1;
/* FIXME - this should use cpu time! */
/*cpu_timers(102 * cnt);*/
cpu_timers((460>>cpu.speed)+cnt*16); /*dalias*/
/*cpu_timers(228 + (16*cnt));*/ /* this should be right according to no$ */
cnt <<= 4;
while (cnt--)
writeb(da++, readb(sa++));
R_HDMA1 = sa >> 8;
R_HDMA2 = sa & 0xF0;
R_HDMA3 = 0x1F & (da >> 8);
R_HDMA4 = da & 0xF0;
R_HDMA5 = 0xFF;
}
/*
* pad_refresh updates the P1 register from the pad states, generating
* the appropriate interrupts (by quickly raising and lowering the

4
lcdc.c
View File

@ -23,10 +23,10 @@
void stat_trigger()
{
static const int condbits[4] = { 0x08, 0x30, 0x20, 0x00 };
static const int condbits[4] = { 0x08, 0x10, 0x20, 0x00 };
int flag = 0;
if ((R_LY < 0x91) && (R_LY == R_LYC))
if (R_LY == R_LYC)
{
R_STAT |= 0x04;
if (R_STAT & 0x40) flag = IF_STAT;

8
mem.c
View File

@ -31,6 +31,9 @@ void mem_updatemap()
int n;
byte **map;
mbc.rombank &= (mbc.romsize - 1);
mbc.rambank &= (mbc.ramsize - 1);
map = mbc.rmap;
map[0x0] = rom.bank[0];
map[0x1] = rom.bank[0];
@ -157,6 +160,9 @@ void ioreg_write(byte r, byte b)
}
R_SC = b; /* & 0x7f; */
break;
case RI_SB:
REG(r) = b;
break;
case RI_DIV:
REG(r) = 0;
break;
@ -428,8 +434,6 @@ void mbc_write(int a, byte b)
}
break;
}
mbc.rombank &= (mbc.romsize - 1);
mbc.rambank &= (mbc.ramsize - 1);
/* printf("%02X\n", mbc.rombank); */
mem_updatemap();
}

1028
noise.h
View File

File diff suppressed because it is too large Load Diff

5
sound.c
View File

@ -134,14 +134,14 @@ void sound_off()
R_NR30 = 0x7F;
R_NR31 = 0xFF;
R_NR32 = 0x9F;
R_NR33 = 0xBF;
R_NR34 = 0xBF;
R_NR41 = 0xFF;
R_NR42 = 0x00;
R_NR43 = 0x00;
R_NR44 = 0xBF;
R_NR50 = 0x77;
R_NR51 = 0xF3;
R_NR52 = 0xF1;
R_NR52 = 0x70;
sound_dirty();
}
@ -153,6 +153,7 @@ void sound_reset()
memcpy(WAVE, hw.cgb ? cgbwave : dmgwave, 16);
memcpy(ram.hi+0x30, WAVE, 16);
sound_off();
R_NR52 = 0xF1;
}