admin/ci-libretro-fceumm
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
Files
f2a5be381d5396cb66b33bca833d4cc1991c5899
ci-libretro-fceumm/src/fds.c

814 lines
20 KiB
C
Raw Normal View History

Initial commit - http://sourceforge.net/p/fceumm/code/160/
2014-03-30 22:15:17 +02:00
/* FCE Ultra - NES/Famicom Emulator
*
* Copyright notice for this file:
* Copyright (C) 2002 Xodnizel
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 2 of the License, or
* (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*/
FDS: Permanently use new disk handling functions FDS was already using the new disk handling code. This PR just removes the old and now unused one.
2022-06-07 18:28:07 +08:00
/* dec-12-19 */
/* 4024(w), 4025(w), 4031(r) by dink(fbneo) */
Initial commit - http://sourceforge.net/p/fceumm/code/160/
2014-03-30 22:15:17 +02:00
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include "fceu-types.h"
#include "x6502.h"
#include "fceu.h"
#include "fds.h"
Allow FDS conversion mappers to use FDS APU
2020-03-05 23:54:34 +08:00
#include "fds_apu.h"
Initial commit - http://sourceforge.net/p/fceumm/code/160/
2014-03-30 22:15:17 +02:00
#include "sound.h"
#include "general.h"
#include "state.h"
#include "file.h"
Rename files endian.c and memory.c
2014-03-30 22:29:30 +02:00
#include "fceu-memory.h"
Initial commit - http://sourceforge.net/p/fceumm/code/160/
2014-03-30 22:15:17 +02:00
#include "cart.h"
#include "md5.h"
Use C-comment style
2017-10-15 03:13:11 +08:00
/* TODO: Add code to put a delay in between the time a disk is inserted
* and the when it can be successfully read/written to. This should
* prevent writes to wrong places OR add code to prevent disk ejects
* when the virtual motor is on(mmm...virtual motor).
*/
Initial commit - http://sourceforge.net/p/fceumm/code/160/
2014-03-30 22:15:17 +02:00
static DECLFR(FDSRead4030);
static DECLFR(FDSRead4031);
static DECLFR(FDSRead4032);
static DECLFR(FDSRead4033);
static DECLFW(FDSWrite);
static void FDSInit(void);
static void FDSClose(void);
static void FP_FASTAPASS(1) FDSFix(int a);
core: stdint typedefs, LE optimizations, frame determinism Three follow-up audit passes on top of the memory-safety / leak / savestate-portability work in 1185db8. ============================================================== Pass 1: replace custom typedefs with C99 stdint types throughout ============================================================== The custom uint8 / uint16 / uint32 / uint64 / int8 / int16 / int32 / int64 typedefs in src/fceu-types.h were just simple aliases for the C99 stdint.h types. Replace them with the standard names directly. - 498 files modified - ~3,400 token replacements (uint8 -> uint8_t, etc) - fceu-types.h slimmed down to just INLINE / GINLINE / FASTAPASS macros and the readfunc / writefunc function-pointer typedefs (those now use uint8_t / uint32_t natively) - Build clean on `make platform=unix` with zero new warnings - Output binary size unchanged - confirming semantic equivalence Mechanical replacement done with a Python script that uses word- boundary regex to avoid false positives (e.g. 'uint32_t' was correctly left alone because '_' is a word character so 'uint32' is not a complete word inside it). ================================================================ Pass 2: prefer memcpy on LE hosts for endian read/write helpers ================================================================ fceu-endian.c's write32le_mem, FCEU_en32lsb, and FCEU_de32lsb performed bytewise composition/decomposition unconditionally. On LE hosts the in-memory representation already matches the desired LE on-disk format, so a single memcpy is equivalent and lets the compiler emit a single load/store rather than four byte ops. - The bytewise path is kept inside #ifdef MSB_FIRST for BE hosts where it implements the actual byte swap - Both forms produce identical results; this is a code-clarity change more than a performance one (the optimizer was already merging the shifts on LE), but it documents the intent and removes a strict-aliasing-flavoured cast through *(uint32_t*)Bufo - Added missing #include <string.h> in fceu-endian.c which was relying on transitive includes for memcpy Other MSB_FIRST sites in the codebase (state.c FlipByteOrder guards, ppu.c sprite-line rendering, boards/unrom512.c flash-write- counter access) were already optimized for LE; they were verified correct rather than changed. ================================================================ Pass 3: frame determinism for replay and netplay ================================================================ Two libc rand() sites in core were replaced with a local xorshift32 PRNG so that NES games which read uninitialised memory or hit hardware "weak bit" emulation produce reproducible behaviour across runs. NES titles routinely read uninitialised RAM (variables not zeroed before use, sprite Y-position set by junk-on-stack), so the RAM contents at power-on subtly affect game behaviour. With libc rand(), those contents depend on whether anyone else seeded rand() in the same process - a different libretro frontend, a different audio backend init order, or any frontend that does srand(time(0)) all break replay / netplay frame-determinism. 1. fceu.c FCEU_MemoryRand. Used to fill RAM (PowerNES) and CHR-RAM (iNES_Init) at power-on when option_ramstate=2 (random init). Replaced with a local xorshift32 PRNG, exposed via a new FCEU_MemoryRand_Reseed(uint32_t) function called once per power-on: - PowerNES seeds from the first 4 bytes of GameInfo->MD5 (set by all loaders before PowerNES runs) so identical ROMs produce identical RAM, different ROMs differ - iNES_Init seeds from iNESCart.PRGCRC32 before the CHR-RAM fill so two builds of the same ROM get the same CHR-RAM - The PRNG state advances across multiple FCEU_MemoryRand calls within one power-on so RAM and CHR-RAM get different content (matching NES hardware reality) 2. boards/rt-01.c UNLRT01Read. The RT-01 board has 'weak bit' protected EPROM regions; reads of 0xCE80-0xCEFF and 0xFE80- 0xFEFF return 0xF2 with the low 3 bits randomised. Replaced libc rand() with a local xorshift32 seeded at power-on, and added the PRNG state to the savestate via AddExState with key "WBKS" so save / load / rewind / netplay rollback all stay deterministic. In addition, two long-double-to-int truncations were changed to double for cross-platform FP determinism: - sound.c SetSoundVariables: soundtsinc - boards/n106.c DoNamcoSound: inc long double has platform-dependent precision (80-bit on x87, 64-bit with -mfpmath=sse, 128-bit on PowerPC), so the truncated integer result varied across these platforms. double is guaranteed 64-bit IEEE-754 portably. After this pass, the core has no time(), clock(), gettimeofday(), clock_gettime(), getpid(), getuid(), getgid(), getenv(), gethostid(), pthread, std::thread, OpenMP, signal handler, or non-deterministic- malloc dependency. Verified with a Python scanner that greps the source for these patterns; runs clean. The PPU / APU / CPU power-on already explicitly memset all state buffers to 0 (deterministic), and ROM/CHR-ROM allocation already memsets to 0xFF before partial fread (deterministic regardless of file truncation). Combined with the memory-safety hardening in 1185db8 (which prevents savestate-loaded indices from going out-of-bounds and producing unpredictable behaviour), the core now offers genuine frame-deterministic replay across runs, builds, and host endian.
2026-05-04 02:46:34 +02:00
static uint8_t FDSRegs[6];
static int32_t IRQLatch, IRQCount;
static uint8_t IRQa;
static uint8_t *FDSROM = NULL;
static uint32_t FDSROMSize = 0;
static uint8_t *FDSRAM = NULL;
static uint32_t FDSRAMSize;
static uint8_t *FDSBIOS = NULL;
static uint32_t FDSBIOSsize;
static uint8_t *CHRRAM = NULL;
static uint32_t CHRRAMSize;
Initial commit - http://sourceforge.net/p/fceumm/code/160/
2014-03-30 22:15:17 +02:00
/* Original disk data backup, to help in creating save states. */
core: stdint typedefs, LE optimizations, frame determinism Three follow-up audit passes on top of the memory-safety / leak / savestate-portability work in 1185db8. ============================================================== Pass 1: replace custom typedefs with C99 stdint types throughout ============================================================== The custom uint8 / uint16 / uint32 / uint64 / int8 / int16 / int32 / int64 typedefs in src/fceu-types.h were just simple aliases for the C99 stdint.h types. Replace them with the standard names directly. - 498 files modified - ~3,400 token replacements (uint8 -> uint8_t, etc) - fceu-types.h slimmed down to just INLINE / GINLINE / FASTAPASS macros and the readfunc / writefunc function-pointer typedefs (those now use uint8_t / uint32_t natively) - Build clean on `make platform=unix` with zero new warnings - Output binary size unchanged - confirming semantic equivalence Mechanical replacement done with a Python script that uses word- boundary regex to avoid false positives (e.g. 'uint32_t' was correctly left alone because '_' is a word character so 'uint32' is not a complete word inside it). ================================================================ Pass 2: prefer memcpy on LE hosts for endian read/write helpers ================================================================ fceu-endian.c's write32le_mem, FCEU_en32lsb, and FCEU_de32lsb performed bytewise composition/decomposition unconditionally. On LE hosts the in-memory representation already matches the desired LE on-disk format, so a single memcpy is equivalent and lets the compiler emit a single load/store rather than four byte ops. - The bytewise path is kept inside #ifdef MSB_FIRST for BE hosts where it implements the actual byte swap - Both forms produce identical results; this is a code-clarity change more than a performance one (the optimizer was already merging the shifts on LE), but it documents the intent and removes a strict-aliasing-flavoured cast through *(uint32_t*)Bufo - Added missing #include <string.h> in fceu-endian.c which was relying on transitive includes for memcpy Other MSB_FIRST sites in the codebase (state.c FlipByteOrder guards, ppu.c sprite-line rendering, boards/unrom512.c flash-write- counter access) were already optimized for LE; they were verified correct rather than changed. ================================================================ Pass 3: frame determinism for replay and netplay ================================================================ Two libc rand() sites in core were replaced with a local xorshift32 PRNG so that NES games which read uninitialised memory or hit hardware "weak bit" emulation produce reproducible behaviour across runs. NES titles routinely read uninitialised RAM (variables not zeroed before use, sprite Y-position set by junk-on-stack), so the RAM contents at power-on subtly affect game behaviour. With libc rand(), those contents depend on whether anyone else seeded rand() in the same process - a different libretro frontend, a different audio backend init order, or any frontend that does srand(time(0)) all break replay / netplay frame-determinism. 1. fceu.c FCEU_MemoryRand. Used to fill RAM (PowerNES) and CHR-RAM (iNES_Init) at power-on when option_ramstate=2 (random init). Replaced with a local xorshift32 PRNG, exposed via a new FCEU_MemoryRand_Reseed(uint32_t) function called once per power-on: - PowerNES seeds from the first 4 bytes of GameInfo->MD5 (set by all loaders before PowerNES runs) so identical ROMs produce identical RAM, different ROMs differ - iNES_Init seeds from iNESCart.PRGCRC32 before the CHR-RAM fill so two builds of the same ROM get the same CHR-RAM - The PRNG state advances across multiple FCEU_MemoryRand calls within one power-on so RAM and CHR-RAM get different content (matching NES hardware reality) 2. boards/rt-01.c UNLRT01Read. The RT-01 board has 'weak bit' protected EPROM regions; reads of 0xCE80-0xCEFF and 0xFE80- 0xFEFF return 0xF2 with the low 3 bits randomised. Replaced libc rand() with a local xorshift32 seeded at power-on, and added the PRNG state to the savestate via AddExState with key "WBKS" so save / load / rewind / netplay rollback all stay deterministic. In addition, two long-double-to-int truncations were changed to double for cross-platform FP determinism: - sound.c SetSoundVariables: soundtsinc - boards/n106.c DoNamcoSound: inc long double has platform-dependent precision (80-bit on x87, 64-bit with -mfpmath=sse, 128-bit on PowerPC), so the truncated integer result varied across these platforms. double is guaranteed 64-bit IEEE-754 portably. After this pass, the core has no time(), clock(), gettimeofday(), clock_gettime(), getpid(), getuid(), getgid(), getenv(), gethostid(), pthread, std::thread, OpenMP, signal handler, or non-deterministic- malloc dependency. Verified with a Python scanner that greps the source for these patterns; runs clean. The PPU / APU / CPU power-on already explicitly memset all state buffers to 0 (deterministic), and ROM/CHR-ROM allocation already memsets to 0xFF before partial fread (deterministic regardless of file truncation). Combined with the memory-safety hardening in 1185db8 (which prevents savestate-loaded indices from going out-of-bounds and producing unpredictable behaviour), the core now offers genuine frame-deterministic replay across runs, builds, and host endian.
2026-05-04 02:46:34 +02:00
static uint8_t *diskdatao[8] = { 0, 0, 0, 0, 0, 0, 0, 0 };
static uint8_t *diskdata[8] = { 0, 0, 0, 0, 0, 0, 0, 0 };
Initial commit - http://sourceforge.net/p/fceumm/code/160/
2014-03-30 22:15:17 +02:00
core: stdint typedefs, LE optimizations, frame determinism Three follow-up audit passes on top of the memory-safety / leak / savestate-portability work in 1185db8. ============================================================== Pass 1: replace custom typedefs with C99 stdint types throughout ============================================================== The custom uint8 / uint16 / uint32 / uint64 / int8 / int16 / int32 / int64 typedefs in src/fceu-types.h were just simple aliases for the C99 stdint.h types. Replace them with the standard names directly. - 498 files modified - ~3,400 token replacements (uint8 -> uint8_t, etc) - fceu-types.h slimmed down to just INLINE / GINLINE / FASTAPASS macros and the readfunc / writefunc function-pointer typedefs (those now use uint8_t / uint32_t natively) - Build clean on `make platform=unix` with zero new warnings - Output binary size unchanged - confirming semantic equivalence Mechanical replacement done with a Python script that uses word- boundary regex to avoid false positives (e.g. 'uint32_t' was correctly left alone because '_' is a word character so 'uint32' is not a complete word inside it). ================================================================ Pass 2: prefer memcpy on LE hosts for endian read/write helpers ================================================================ fceu-endian.c's write32le_mem, FCEU_en32lsb, and FCEU_de32lsb performed bytewise composition/decomposition unconditionally. On LE hosts the in-memory representation already matches the desired LE on-disk format, so a single memcpy is equivalent and lets the compiler emit a single load/store rather than four byte ops. - The bytewise path is kept inside #ifdef MSB_FIRST for BE hosts where it implements the actual byte swap - Both forms produce identical results; this is a code-clarity change more than a performance one (the optimizer was already merging the shifts on LE), but it documents the intent and removes a strict-aliasing-flavoured cast through *(uint32_t*)Bufo - Added missing #include <string.h> in fceu-endian.c which was relying on transitive includes for memcpy Other MSB_FIRST sites in the codebase (state.c FlipByteOrder guards, ppu.c sprite-line rendering, boards/unrom512.c flash-write- counter access) were already optimized for LE; they were verified correct rather than changed. ================================================================ Pass 3: frame determinism for replay and netplay ================================================================ Two libc rand() sites in core were replaced with a local xorshift32 PRNG so that NES games which read uninitialised memory or hit hardware "weak bit" emulation produce reproducible behaviour across runs. NES titles routinely read uninitialised RAM (variables not zeroed before use, sprite Y-position set by junk-on-stack), so the RAM contents at power-on subtly affect game behaviour. With libc rand(), those contents depend on whether anyone else seeded rand() in the same process - a different libretro frontend, a different audio backend init order, or any frontend that does srand(time(0)) all break replay / netplay frame-determinism. 1. fceu.c FCEU_MemoryRand. Used to fill RAM (PowerNES) and CHR-RAM (iNES_Init) at power-on when option_ramstate=2 (random init). Replaced with a local xorshift32 PRNG, exposed via a new FCEU_MemoryRand_Reseed(uint32_t) function called once per power-on: - PowerNES seeds from the first 4 bytes of GameInfo->MD5 (set by all loaders before PowerNES runs) so identical ROMs produce identical RAM, different ROMs differ - iNES_Init seeds from iNESCart.PRGCRC32 before the CHR-RAM fill so two builds of the same ROM get the same CHR-RAM - The PRNG state advances across multiple FCEU_MemoryRand calls within one power-on so RAM and CHR-RAM get different content (matching NES hardware reality) 2. boards/rt-01.c UNLRT01Read. The RT-01 board has 'weak bit' protected EPROM regions; reads of 0xCE80-0xCEFF and 0xFE80- 0xFEFF return 0xF2 with the low 3 bits randomised. Replaced libc rand() with a local xorshift32 seeded at power-on, and added the PRNG state to the savestate via AddExState with key "WBKS" so save / load / rewind / netplay rollback all stay deterministic. In addition, two long-double-to-int truncations were changed to double for cross-platform FP determinism: - sound.c SetSoundVariables: soundtsinc - boards/n106.c DoNamcoSound: inc long double has platform-dependent precision (80-bit on x87, 64-bit with -mfpmath=sse, 128-bit on PowerPC), so the truncated integer result varied across these platforms. double is guaranteed 64-bit IEEE-754 portably. After this pass, the core has no time(), clock(), gettimeofday(), clock_gettime(), getpid(), getuid(), getgid(), getenv(), gethostid(), pthread, std::thread, OpenMP, signal handler, or non-deterministic- malloc dependency. Verified with a Python scanner that greps the source for these patterns; runs clean. The PPU / APU / CPU power-on already explicitly memset all state buffers to 0 (deterministic), and ROM/CHR-ROM allocation already memsets to 0xFF before partial fread (deterministic regardless of file truncation). Combined with the memory-safety hardening in 1185db8 (which prevents savestate-loaded indices from going out-of-bounds and producing unpredictable behaviour), the core now offers genuine frame-deterministic replay across runs, builds, and host endian.
2026-05-04 02:46:34 +02:00
static uint32_t TotalSides;
static uint8_t DiskWritten = 0; /* Set to 1 if disk was written to. */
static uint8_t writeskip;
static int32_t DiskPtr;
static int32_t DiskSeekIRQ;
static uint8_t SelectDisk, InDisk;
Initial commit - http://sourceforge.net/p/fceumm/code/160/
2014-03-30 22:15:17 +02:00
(fds.c) backport new fds disk handling backported from fceux, commit by dinkc64 (fbneo) https://github.com/TASVideos/fceux/commit/0b4be4b639371204642f22def7bd92ea18a4c2f3 Fixes saving issue with some games (Bubble Bobble, Super Lode Runner II, ...)
2020-01-25 08:32:43 +08:00
enum FDS_DiskBlockIDs {
DSK_INIT = 0,
DSK_VOLUME,
DSK_FILECNT,
DSK_FILEHDR,
DSK_FILEDATA
};
core: stdint typedefs, LE optimizations, frame determinism Three follow-up audit passes on top of the memory-safety / leak / savestate-portability work in 1185db8. ============================================================== Pass 1: replace custom typedefs with C99 stdint types throughout ============================================================== The custom uint8 / uint16 / uint32 / uint64 / int8 / int16 / int32 / int64 typedefs in src/fceu-types.h were just simple aliases for the C99 stdint.h types. Replace them with the standard names directly. - 498 files modified - ~3,400 token replacements (uint8 -> uint8_t, etc) - fceu-types.h slimmed down to just INLINE / GINLINE / FASTAPASS macros and the readfunc / writefunc function-pointer typedefs (those now use uint8_t / uint32_t natively) - Build clean on `make platform=unix` with zero new warnings - Output binary size unchanged - confirming semantic equivalence Mechanical replacement done with a Python script that uses word- boundary regex to avoid false positives (e.g. 'uint32_t' was correctly left alone because '_' is a word character so 'uint32' is not a complete word inside it). ================================================================ Pass 2: prefer memcpy on LE hosts for endian read/write helpers ================================================================ fceu-endian.c's write32le_mem, FCEU_en32lsb, and FCEU_de32lsb performed bytewise composition/decomposition unconditionally. On LE hosts the in-memory representation already matches the desired LE on-disk format, so a single memcpy is equivalent and lets the compiler emit a single load/store rather than four byte ops. - The bytewise path is kept inside #ifdef MSB_FIRST for BE hosts where it implements the actual byte swap - Both forms produce identical results; this is a code-clarity change more than a performance one (the optimizer was already merging the shifts on LE), but it documents the intent and removes a strict-aliasing-flavoured cast through *(uint32_t*)Bufo - Added missing #include <string.h> in fceu-endian.c which was relying on transitive includes for memcpy Other MSB_FIRST sites in the codebase (state.c FlipByteOrder guards, ppu.c sprite-line rendering, boards/unrom512.c flash-write- counter access) were already optimized for LE; they were verified correct rather than changed. ================================================================ Pass 3: frame determinism for replay and netplay ================================================================ Two libc rand() sites in core were replaced with a local xorshift32 PRNG so that NES games which read uninitialised memory or hit hardware "weak bit" emulation produce reproducible behaviour across runs. NES titles routinely read uninitialised RAM (variables not zeroed before use, sprite Y-position set by junk-on-stack), so the RAM contents at power-on subtly affect game behaviour. With libc rand(), those contents depend on whether anyone else seeded rand() in the same process - a different libretro frontend, a different audio backend init order, or any frontend that does srand(time(0)) all break replay / netplay frame-determinism. 1. fceu.c FCEU_MemoryRand. Used to fill RAM (PowerNES) and CHR-RAM (iNES_Init) at power-on when option_ramstate=2 (random init). Replaced with a local xorshift32 PRNG, exposed via a new FCEU_MemoryRand_Reseed(uint32_t) function called once per power-on: - PowerNES seeds from the first 4 bytes of GameInfo->MD5 (set by all loaders before PowerNES runs) so identical ROMs produce identical RAM, different ROMs differ - iNES_Init seeds from iNESCart.PRGCRC32 before the CHR-RAM fill so two builds of the same ROM get the same CHR-RAM - The PRNG state advances across multiple FCEU_MemoryRand calls within one power-on so RAM and CHR-RAM get different content (matching NES hardware reality) 2. boards/rt-01.c UNLRT01Read. The RT-01 board has 'weak bit' protected EPROM regions; reads of 0xCE80-0xCEFF and 0xFE80- 0xFEFF return 0xF2 with the low 3 bits randomised. Replaced libc rand() with a local xorshift32 seeded at power-on, and added the PRNG state to the savestate via AddExState with key "WBKS" so save / load / rewind / netplay rollback all stay deterministic. In addition, two long-double-to-int truncations were changed to double for cross-platform FP determinism: - sound.c SetSoundVariables: soundtsinc - boards/n106.c DoNamcoSound: inc long double has platform-dependent precision (80-bit on x87, 64-bit with -mfpmath=sse, 128-bit on PowerPC), so the truncated integer result varied across these platforms. double is guaranteed 64-bit IEEE-754 portably. After this pass, the core has no time(), clock(), gettimeofday(), clock_gettime(), getpid(), getuid(), getgid(), getenv(), gethostid(), pthread, std::thread, OpenMP, signal handler, or non-deterministic- malloc dependency. Verified with a Python scanner that greps the source for these patterns; runs clean. The PPU / APU / CPU power-on already explicitly memset all state buffers to 0 (deterministic), and ROM/CHR-ROM allocation already memsets to 0xFF before partial fread (deterministic regardless of file truncation). Combined with the memory-safety hardening in 1185db8 (which prevents savestate-loaded indices from going out-of-bounds and producing unpredictable behaviour), the core now offers genuine frame-deterministic replay across runs, builds, and host endian.
2026-05-04 02:46:34 +02:00
static uint8_t mapperFDS_control; /* 4025(w) control register */
static uint16_t mapperFDS_filesize; /* size of file being read/written */
static uint8_t mapperFDS_block; /* block-id of current block */
static uint16_t mapperFDS_blockstart; /* start-address of current block */
static uint16_t mapperFDS_blocklen; /* length of current block */
static uint16_t mapperFDS_diskaddr; /* current address relative to blockstart */
static uint8_t mapperFDS_diskaccess; /* disk needs to be accessed at least once before writing */
(fds.c) backport new fds disk handling backported from fceux, commit by dinkc64 (fbneo) https://github.com/TASVideos/fceux/commit/0b4be4b639371204642f22def7bd92ea18a4c2f3 Fixes saving issue with some games (Bubble Bobble, Super Lode Runner II, ...)
2020-01-25 08:32:43 +08:00
#define GET_FDS_DISK() (diskdata[InDisk][mapperFDS_blockstart + mapperFDS_diskaddr])
#define FDS_DISK_INSERTED (InDisk != 255)
Initial commit - http://sourceforge.net/p/fceumm/code/160/
2014-03-30 22:15:17 +02:00
#define DC_INC 1
FDS: Use retro_get_memory_data/size for disk saving
2019-01-04 17:13:09 +08:00
#define BYTES_PER_SIDE 65500
core: stdint typedefs, LE optimizations, frame determinism Three follow-up audit passes on top of the memory-safety / leak / savestate-portability work in 1185db8. ============================================================== Pass 1: replace custom typedefs with C99 stdint types throughout ============================================================== The custom uint8 / uint16 / uint32 / uint64 / int8 / int16 / int32 / int64 typedefs in src/fceu-types.h were just simple aliases for the C99 stdint.h types. Replace them with the standard names directly. - 498 files modified - ~3,400 token replacements (uint8 -> uint8_t, etc) - fceu-types.h slimmed down to just INLINE / GINLINE / FASTAPASS macros and the readfunc / writefunc function-pointer typedefs (those now use uint8_t / uint32_t natively) - Build clean on `make platform=unix` with zero new warnings - Output binary size unchanged - confirming semantic equivalence Mechanical replacement done with a Python script that uses word- boundary regex to avoid false positives (e.g. 'uint32_t' was correctly left alone because '_' is a word character so 'uint32' is not a complete word inside it). ================================================================ Pass 2: prefer memcpy on LE hosts for endian read/write helpers ================================================================ fceu-endian.c's write32le_mem, FCEU_en32lsb, and FCEU_de32lsb performed bytewise composition/decomposition unconditionally. On LE hosts the in-memory representation already matches the desired LE on-disk format, so a single memcpy is equivalent and lets the compiler emit a single load/store rather than four byte ops. - The bytewise path is kept inside #ifdef MSB_FIRST for BE hosts where it implements the actual byte swap - Both forms produce identical results; this is a code-clarity change more than a performance one (the optimizer was already merging the shifts on LE), but it documents the intent and removes a strict-aliasing-flavoured cast through *(uint32_t*)Bufo - Added missing #include <string.h> in fceu-endian.c which was relying on transitive includes for memcpy Other MSB_FIRST sites in the codebase (state.c FlipByteOrder guards, ppu.c sprite-line rendering, boards/unrom512.c flash-write- counter access) were already optimized for LE; they were verified correct rather than changed. ================================================================ Pass 3: frame determinism for replay and netplay ================================================================ Two libc rand() sites in core were replaced with a local xorshift32 PRNG so that NES games which read uninitialised memory or hit hardware "weak bit" emulation produce reproducible behaviour across runs. NES titles routinely read uninitialised RAM (variables not zeroed before use, sprite Y-position set by junk-on-stack), so the RAM contents at power-on subtly affect game behaviour. With libc rand(), those contents depend on whether anyone else seeded rand() in the same process - a different libretro frontend, a different audio backend init order, or any frontend that does srand(time(0)) all break replay / netplay frame-determinism. 1. fceu.c FCEU_MemoryRand. Used to fill RAM (PowerNES) and CHR-RAM (iNES_Init) at power-on when option_ramstate=2 (random init). Replaced with a local xorshift32 PRNG, exposed via a new FCEU_MemoryRand_Reseed(uint32_t) function called once per power-on: - PowerNES seeds from the first 4 bytes of GameInfo->MD5 (set by all loaders before PowerNES runs) so identical ROMs produce identical RAM, different ROMs differ - iNES_Init seeds from iNESCart.PRGCRC32 before the CHR-RAM fill so two builds of the same ROM get the same CHR-RAM - The PRNG state advances across multiple FCEU_MemoryRand calls within one power-on so RAM and CHR-RAM get different content (matching NES hardware reality) 2. boards/rt-01.c UNLRT01Read. The RT-01 board has 'weak bit' protected EPROM regions; reads of 0xCE80-0xCEFF and 0xFE80- 0xFEFF return 0xF2 with the low 3 bits randomised. Replaced libc rand() with a local xorshift32 seeded at power-on, and added the PRNG state to the savestate via AddExState with key "WBKS" so save / load / rewind / netplay rollback all stay deterministic. In addition, two long-double-to-int truncations were changed to double for cross-platform FP determinism: - sound.c SetSoundVariables: soundtsinc - boards/n106.c DoNamcoSound: inc long double has platform-dependent precision (80-bit on x87, 64-bit with -mfpmath=sse, 128-bit on PowerPC), so the truncated integer result varied across these platforms. double is guaranteed 64-bit IEEE-754 portably. After this pass, the core has no time(), clock(), gettimeofday(), clock_gettime(), getpid(), getuid(), getgid(), getenv(), gethostid(), pthread, std::thread, OpenMP, signal handler, or non-deterministic- malloc dependency. Verified with a Python scanner that greps the source for these patterns; runs clean. The PPU / APU / CPU power-on already explicitly memset all state buffers to 0 (deterministic), and ROM/CHR-ROM allocation already memsets to 0xFF before partial fread (deterministic regardless of file truncation). Combined with the memory-safety hardening in 1185db8 (which prevents savestate-loaded indices from going out-of-bounds and producing unpredictable behaviour), the core now offers genuine frame-deterministic replay across runs, builds, and host endian.
2026-05-04 02:46:34 +02:00
uint8_t *FDSROM_ptr(void) {
FDS: Use retro_get_memory_data/size for disk saving
2019-01-04 17:13:09 +08:00
return (FDSROM);
}
core: stdint typedefs, LE optimizations, frame determinism Three follow-up audit passes on top of the memory-safety / leak / savestate-portability work in 1185db8. ============================================================== Pass 1: replace custom typedefs with C99 stdint types throughout ============================================================== The custom uint8 / uint16 / uint32 / uint64 / int8 / int16 / int32 / int64 typedefs in src/fceu-types.h were just simple aliases for the C99 stdint.h types. Replace them with the standard names directly. - 498 files modified - ~3,400 token replacements (uint8 -> uint8_t, etc) - fceu-types.h slimmed down to just INLINE / GINLINE / FASTAPASS macros and the readfunc / writefunc function-pointer typedefs (those now use uint8_t / uint32_t natively) - Build clean on `make platform=unix` with zero new warnings - Output binary size unchanged - confirming semantic equivalence Mechanical replacement done with a Python script that uses word- boundary regex to avoid false positives (e.g. 'uint32_t' was correctly left alone because '_' is a word character so 'uint32' is not a complete word inside it). ================================================================ Pass 2: prefer memcpy on LE hosts for endian read/write helpers ================================================================ fceu-endian.c's write32le_mem, FCEU_en32lsb, and FCEU_de32lsb performed bytewise composition/decomposition unconditionally. On LE hosts the in-memory representation already matches the desired LE on-disk format, so a single memcpy is equivalent and lets the compiler emit a single load/store rather than four byte ops. - The bytewise path is kept inside #ifdef MSB_FIRST for BE hosts where it implements the actual byte swap - Both forms produce identical results; this is a code-clarity change more than a performance one (the optimizer was already merging the shifts on LE), but it documents the intent and removes a strict-aliasing-flavoured cast through *(uint32_t*)Bufo - Added missing #include <string.h> in fceu-endian.c which was relying on transitive includes for memcpy Other MSB_FIRST sites in the codebase (state.c FlipByteOrder guards, ppu.c sprite-line rendering, boards/unrom512.c flash-write- counter access) were already optimized for LE; they were verified correct rather than changed. ================================================================ Pass 3: frame determinism for replay and netplay ================================================================ Two libc rand() sites in core were replaced with a local xorshift32 PRNG so that NES games which read uninitialised memory or hit hardware "weak bit" emulation produce reproducible behaviour across runs. NES titles routinely read uninitialised RAM (variables not zeroed before use, sprite Y-position set by junk-on-stack), so the RAM contents at power-on subtly affect game behaviour. With libc rand(), those contents depend on whether anyone else seeded rand() in the same process - a different libretro frontend, a different audio backend init order, or any frontend that does srand(time(0)) all break replay / netplay frame-determinism. 1. fceu.c FCEU_MemoryRand. Used to fill RAM (PowerNES) and CHR-RAM (iNES_Init) at power-on when option_ramstate=2 (random init). Replaced with a local xorshift32 PRNG, exposed via a new FCEU_MemoryRand_Reseed(uint32_t) function called once per power-on: - PowerNES seeds from the first 4 bytes of GameInfo->MD5 (set by all loaders before PowerNES runs) so identical ROMs produce identical RAM, different ROMs differ - iNES_Init seeds from iNESCart.PRGCRC32 before the CHR-RAM fill so two builds of the same ROM get the same CHR-RAM - The PRNG state advances across multiple FCEU_MemoryRand calls within one power-on so RAM and CHR-RAM get different content (matching NES hardware reality) 2. boards/rt-01.c UNLRT01Read. The RT-01 board has 'weak bit' protected EPROM regions; reads of 0xCE80-0xCEFF and 0xFE80- 0xFEFF return 0xF2 with the low 3 bits randomised. Replaced libc rand() with a local xorshift32 seeded at power-on, and added the PRNG state to the savestate via AddExState with key "WBKS" so save / load / rewind / netplay rollback all stay deterministic. In addition, two long-double-to-int truncations were changed to double for cross-platform FP determinism: - sound.c SetSoundVariables: soundtsinc - boards/n106.c DoNamcoSound: inc long double has platform-dependent precision (80-bit on x87, 64-bit with -mfpmath=sse, 128-bit on PowerPC), so the truncated integer result varied across these platforms. double is guaranteed 64-bit IEEE-754 portably. After this pass, the core has no time(), clock(), gettimeofday(), clock_gettime(), getpid(), getuid(), getgid(), getenv(), gethostid(), pthread, std::thread, OpenMP, signal handler, or non-deterministic- malloc dependency. Verified with a Python scanner that greps the source for these patterns; runs clean. The PPU / APU / CPU power-on already explicitly memset all state buffers to 0 (deterministic), and ROM/CHR-ROM allocation already memsets to 0xFF before partial fread (deterministic regardless of file truncation). Combined with the memory-safety hardening in 1185db8 (which prevents savestate-loaded indices from going out-of-bounds and producing unpredictable behaviour), the core now offers genuine frame-deterministic replay across runs, builds, and host endian.
2026-05-04 02:46:34 +02:00
uint32_t FDSROM_size(void) {
FDS: Use retro_get_memory_data/size for disk saving
2019-01-04 17:13:09 +08:00
return (FDSROMSize);
}
Initial commit - http://sourceforge.net/p/fceumm/code/160/
2014-03-30 22:15:17 +02:00
void FDSGI(int h) {
switch (h) {
case GI_CLOSE: FDSClose(); break;
case GI_POWER: FDSInit(); break;
}
}
static void FDSStateRestore(int version) {
int x;
core: memory-safety, leak, and savestate-portability audit fixes Squashed series of ~50 distinct bugs found during a multi-day security/correctness audit, ranging from ROM-triggerable heap corruption and savestate-triggered OOB read primitives down to obscure cross-platform savestate breakage. Build is clean on `make platform=unix` with zero new warnings. CRITICAL (ROM-triggerable, exploitable on the user's machine) ============================================================= * ines.c iNES 2.0 PRG/CHR exponent overflow leading to undersized allocation followed by heap buffer overflow on fread (pow() with attacker-chosen exponent up to 63). Cap exponent at 30 and compute size in uint32 with explicit cap below 2 GiB before storing in int. * ines.c miscROMSize int wraparound from attacker-controlled PRG/ CHR sizes; previous '& 0x8000000' check missed common underflow cases. Compute in int64 and reject <= 0 or > 128 MiB. * unif.c MAPR chunk OOB heap read on chunks of size < 4. Validate chunk size before allocating board-name buffer. * unif.c chunk size truncation: int conversion + missing cap allowed a 4 GiB chunk size to wrap. Cap at 16 MiB. * unif.c FixRomSize infinite loop on size > 0x80000000. Cap input. * unif.c DINF chunk: months[(m - 1) % 12] is UB for m==0; m comes from the .unf file. Clamp m into [1..12] before subtracting. * unif.c NAME chunk: GameInfo->name = malloc(...) was unchecked. * nsf.c size underflow: FCEU_fgetsize() - 0x80 wraps to ~UINT64_MAX on tiny files, propagating a huge size into uppow2/FCEU_malloc. Validate size > 0x80 before subtracting. * nsf.c NSFMaxBank * 4096 cap tightened from 1<<20 (UB on signed-int overflow) to 1<<19 (fits in int). * general.c uppow2 had signed-int UB (1 << 32) and wrapped to 0 for inputs near uint32 max. Cap at 0x80000000. * cheat.c SubCheats[256] BSS overflow when more than 256 GG/PAR substitution cheats are active. The array is adjacent to MMapPtrs[64] in BSS which is exposed via retro_get_memory_data, making this overflow visible from outside the core. * libretro.c retro_cheat_set strcpy stack overflow with arbitrary- length cheat strings from the frontend. Use strlcpy. CRITICAL (savestate-triggerable on a malicious .fcs file) ========================================================= * fds.c FDS savestate OOB read primitive: InDisk loaded from savestate is used as index into 8-element diskdata[] static pointer array. A value 8..254 dereferences arbitrary memory as a pointer (heap-read primitive). Also bounds-checked SelectDisk, mapperFDS_block, mapperFDS_blockstart, mapperFDS_diskaddr, mapperFDS_blocklen so blockstart+diskaddr stays within the 65500- byte disk buffer. * 11 mappers had savestate-loaded variables masked at write time but not at restore time, used as indices into fixed arrays: - 88.c, KS7037.c, 112.c cmd indexes reg[8] - sachen.c (S8259, S74LS374N) cmd indexes latch[8] - 357.c dipswitch indexes outer_bank[4] - unrom512.c flash_state indexes erase_a/d/b[5] - 368.c preg indexes banks[8] - 69.c sndcmd indexes sreg[14] - mmc2and4.c latch0/latch1 index creg[4] None individually exploitable into RCE - the array writes corrupt adjacent BSS with constrained data flowing in - but each is an out-of-bounds read or write from attacker-controllable input. HIGH (memory-safety, reachable on any load) =========================================== * fceu-memory.c FCEU_malloc deref-of-NULL on allocation failure ('ret = 0; memset(ret, 0, size);'). * file.c multiple FCEUFILE/MakeMemWrap/MakeMemWrapBuffer unchecked allocations and unchecked filestream_tell return. * libretro.c GameInfo NULL-derefs in three entry points (retro_set_controller_port_device, retro_get_memory_data, retro_get_memory_size) reachable on operations called before a successful load. * libretro.c framebuffer leak ~256 KB per failed FCEUI_LoadGame (the libretro frontend doesn't call retro_unload_game on a failed load). * libretro.c 3DS retro_deinit unchecked linearFree. * libretro.c stereo / NTSC filter unchecked malloc returns. * fceu.c FCEUI_LoadGame unchecked GameInfo malloc. * fds.c SubLoad and FDSLoad unchecked FCEU_malloc on diskdatao backup buffers (NULL-deref in subsequent memcpy). * fceu.c AllocGenieRW partial-failure leak: AReadG allocated but BWriteG malloc fails -> 256 KB leak per retry. * input/bworld.c Update(): unbounded strcpy from attacker-supplied data into 20-byte bdata. Replaced with length-bounded copy. * cart.c setprg2r/4r and setchr1r/2r/4r/8r mask-underflow guards. SetupCartPRGMapping/SetupCartCHRMapping compute (size >> N) - 1 which underflows to 0xFFFFFFFF for chips smaller than the unit. setprg8r/16r/32r and setchr8r already had defensive size checks; the smaller units did not. malee.c (2 KB chip) and mapper 218 (2 KB NTARAM-as-CHR) accidentally avoid OOB; the primitive is unsafe for any future board. * video.c FCEU_InitVirtualVideo XBuf/XDBuf partial-failure leak. * video.c, fceu.c FCEU_DispMessage / FCEU_printf / FCEU_PrintError: vsprintf into fixed stack buffers replaced with vsnprintf. * core: validate magic-string read length on FDS/UNIF/NSF load (reject files too short to contain the magic so previous static buffer / stack garbage doesn't spuriously match). LEAKS (every cart load/unload cycle) ==================================== * mmc5.c MMC5 cart-side: WRAM (up to 64K) + MMC5fill (1K) + ExRAM (1K) leaked per cycle. NSFMMC5_Close existed but was only used for NSF code path. * onebus.c Mapper 270/436: 8 KB CHRRAM leaked per cycle. * 330.c, 375.c, 528.c: 8 KB WRAM each, no Close at all. CORRECTNESS (UB / ABI / endianness) =================================== * fceu-endian.c FlipByteOrder loop bound was 'count' instead of 'count/2', making it a no-op for every even count and silently breaking savestate portability for every FCEUSTATE_RLSB-marked field. The '#ifndef GEKKO' workarounds scattered through the codebase (sound.c, vrc7.c, vrc6.c) were symptoms of this root cause. * state.c AddExState bounds check ran AFTER writing the entry, so when SFEXINDEX hit the 63-entry cap the next call would write the entry then immediately overwrite it with the terminator. * state.c FCEUSS_Save_Mem: post-save callback was guarded by 'if (SPreSave)' instead of 'if (SPostSave)'. * Sequence-point UB in counter expressions across 4 board files, e.g. 'x = !x ? a : --x' and 'x = ++x % n'. GCC -Wsequence-point flags all five sites (285.c, 413.c, asic_mmc3.c, jyasic.c). * SFORMAT size mismatches between declared variable type and AddExState size argument: - asic_vrc3.c VRC3_count/VRC3_reload (uint16, saved as 1 byte) - mmc3.c m555_count (uint32, saved as 2 bytes), m555_count_- expired (uint8, saved as 2 bytes - OOB write into adjacent BSS). * unrom512.c UNROM-512 mapper 30 .srm save format: host-endian uint32 flash write counters. Now stored as LE on disk regardless of host (Battle Kid 2, Twin Dragons, Lizard, Sole, etc.). * ~70 multi-byte single-variable SFORMAT/AddExState entries across 36 board files were saved as host-byte-order. After the FlipByteOrder fix, these are now byte-swapped on BE so cross- platform savestates round-trip correctly. * coolgirl.c new ExStateLE() macro added; 22 multi-byte sites. * pic16c5x.c 11 multi-byte AddExState calls (m_PC, m_PREVPC, m_CONFIG, m_WDT, m_prescaler, m_opcode, m_STACK[0/1], m_icount, m_delay_timer, m_rtcc, m_inst_cycles, m_clock2cycle). * onebus.c PowerJoy Supermax submapper detection used non-portable *(uint32*)&info->MD5 cast that read different bytes on LE vs BE. * fds.c clean up redundant FCEUSTATE_RLSB encoding in AddExState calls that also passed type=1 (idempotent OR; readability fix). DOCUMENTATION ============= * input.c UpdateGP's *(uint32*)data cast looks like a typical endian bug but is actually correct (the libretro frontend builds JSReturn with matching host-uint32 shifts). Comment added to prevent future "fixes" from breaking it. Limitations not addressed ========================= * Element-stride-aware byte swapping. The savestate byte-swap mechanism (FlipByteOrder over the entire SFORMAT entry buffer) is structurally wrong for arrays of multi-byte values: it reverses the whole buffer end-to-end instead of byte-swapping each element. Several places that need cross-platform-portable arrays (VRC7 sound state, jyasic chr[8]) work around this by either splitting arrays into per-element SFORMAT entries (n106 PlayIndex, bandai reg) or by skipping save entirely on BE via #ifndef GEKKO. A proper fix would extend the size encoding with an element-stride field. Left for a future change because it would change the savestate format. * Strict-aliasing UB in ppu.c (around 9 sites doing *(uint32*)uint8_buf for fast 4-byte writes via FCEU_dwmemset). Works in practice with all common compilers because the patterns are byte-symmetric, but is formally UB. * FCEU_gmalloc calls exit(1) on OOM. A libretro core should never exit() because that takes down the whole frontend. Used by 100+ call sites; refactoring to return-NULL is out of scope here. Testing ======= * Build: clean on `make platform=unix` with -O2; no new warnings. * FlipByteOrder fix verified by hand-trace and a standalone unit test for counts 2, 4, 8. * uppow2 fix verified by unit test across 13 boundary cases. * SFORMAT size mismatches and missing-RLSB cases identified by Python static-analysis scripts that cross-reference SFORMAT entries against variable declarations. * iNES 2.0 exponent fix verified by hand-tracing what byte 0xFF produces post-fix: exp=30 (capped), mult=7, size=3 GiB nominal, capped to 1 GiB, capped to 2 GiB by uppow2, FCEU_malloc returns NULL on most systems, loader returns 0. No heap overflow for any input byte. * Savestate-loaded array index audit: built a Python scanner that extracts each AddExState/SFORMAT entry and cross-references the variable name against array-index uses in the same file. All flagged sites covered. * A libFuzzer harness and seed corpus generator (fuzz_main.c, gen_seed_corpus.py) accompany this submission for ongoing regression testing.
2026-05-04 02:15:40 +02:00
/* Sanity-check disk indices and block parameters. A malicious
* savestate could otherwise:
* - set InDisk to a value 0..254 (255 means ejected) that's >=
* TotalSides, causing the read path to index past the
* 8-element diskdata[] array and dereference an arbitrary
* pointer-shaped 8 bytes (heap-read primitive);
* - set mapperFDS_block to an out-of-enum value (no current
* case crashes but better to clamp);
* - set mapperFDS_blockstart + mapperFDS_diskaddr beyond 65500
* (BYTES_PER_SIDE), causing GET_FDS_DISK() to OOB-read up
* to 65 KB past the disk buffer. */
if (InDisk != 255 && InDisk >= TotalSides)
InDisk = 255; /* eject */
if (SelectDisk >= TotalSides)
SelectDisk = 0;
if (mapperFDS_block > DSK_FILEDATA) /* enum values 0..4 are valid */
mapperFDS_block = DSK_INIT;
/* BYTES_PER_SIDE is 65500. The read/write sites compute
* (blockstart + diskaddr) as the offset into a 65500-byte buffer
* after gating on diskaddr < blocklen. Clamp each component such
* that any allowed (blockstart + diskaddr) stays within the buffer:
* - blockstart < BYTES_PER_SIDE (so the base is in-range)
* - diskaddr fits in the remaining space, capped by blocklen anyway
* - blocklen capped to BYTES_PER_SIDE - blockstart so the gating
* check 'diskaddr < blocklen' is sufficient. */
if (mapperFDS_blockstart >= BYTES_PER_SIDE)
mapperFDS_blockstart = 0;
if (mapperFDS_blocklen > BYTES_PER_SIDE - mapperFDS_blockstart)
mapperFDS_blocklen = BYTES_PER_SIDE - mapperFDS_blockstart;
if (mapperFDS_diskaddr > mapperFDS_blocklen)
mapperFDS_diskaddr = mapperFDS_blocklen;
Initial commit - http://sourceforge.net/p/fceumm/code/160/
2014-03-30 22:15:17 +02:00
setmirror(((FDSRegs[5] & 8) >> 3) ^ 1);
if (version >= 9810)
for (x = 0; x < TotalSides; x++) {
int b;
for (b = 0; b < 65500; b++)
diskdata[x][b] ^= diskdatao[x][b];
}
}
static void FDSInit(void) {
memset(FDSRegs, 0, sizeof(FDSRegs));
Allow FDS conversion mappers to use FDS APU
2020-03-05 23:54:34 +08:00
writeskip = DiskPtr = DiskSeekIRQ = 0;
Initial commit - http://sourceforge.net/p/fceumm/code/160/
2014-03-30 22:15:17 +02:00
setmirror(1);
Use C-comment style
2017-10-15 03:13:11 +08:00
setprg8(0xE000, 0); /* BIOS */
setprg32r(1, 0x6000, 0); /* 32KB RAM */
setchr8(0); /* 8KB CHR RAM */
Initial commit - http://sourceforge.net/p/fceumm/code/160/
2014-03-30 22:15:17 +02:00
MapIRQHook = FDSFix;
GameStateRestore = FDSStateRestore;
SetReadHandler(0x4030, 0x4030, FDSRead4030);
SetReadHandler(0x4031, 0x4031, FDSRead4031);
SetReadHandler(0x4032, 0x4032, FDSRead4032);
SetReadHandler(0x4033, 0x4033, FDSRead4033);
SetWriteHandler(0x4020, 0x4025, FDSWrite);
SetWriteHandler(0x6000, 0xDFFF, CartBW);
SetReadHandler(0x6000, 0xFFFF, CartBR);
IRQCount = IRQLatch = IRQa = 0;
FDSSoundReset();
InDisk = 0;
SelectDisk = 0;
(fds.c) backport new fds disk handling backported from fceux, commit by dinkc64 (fbneo) https://github.com/TASVideos/fceux/commit/0b4be4b639371204642f22def7bd92ea18a4c2f3 Fixes saving issue with some games (Bubble Bobble, Super Lode Runner II, ...)
2020-01-25 08:32:43 +08:00
mapperFDS_control = 0;
mapperFDS_filesize = 0;
mapperFDS_block = 0;
mapperFDS_blockstart = 0;
mapperFDS_blocklen = 0;
mapperFDS_diskaddr = 0;
mapperFDS_diskaccess = 0;
Initial commit - http://sourceforge.net/p/fceumm/code/160/
2014-03-30 22:15:17 +02:00
}
void FCEU_FDSInsert(int oride) {
if (InDisk == 255) {
Add support for emulated Game Genie add-on
2021-10-21 15:52:40 +01:00
FCEU_DispMessage(RETRO_LOG_INFO, 2000, "Disk %d of %d Side %s Inserted",
1 + (SelectDisk >> 1), (TotalSides + 1) >> 1, (SelectDisk & 1) ? "B" : "A");
Initial commit - http://sourceforge.net/p/fceumm/code/160/
2014-03-30 22:15:17 +02:00
InDisk = SelectDisk;
} else {
Add support for emulated Game Genie add-on
2021-10-21 15:52:40 +01:00
FCEU_DispMessage(RETRO_LOG_INFO, 2000, "Disk %d of %d Side %s Ejected",
1 + (SelectDisk >> 1), (TotalSides + 1) >> 1, (SelectDisk & 1) ? "B" : "A");
Initial commit - http://sourceforge.net/p/fceumm/code/160/
2014-03-30 22:15:17 +02:00
InDisk = 255;
}
}
void FCEU_FDSEject(void) {
InDisk = 255;
}
void FCEU_FDSSelect(void) {
if (InDisk != 255) {
Add support for emulated Game Genie add-on
2021-10-21 15:52:40 +01:00
FCEUD_DispMessage(RETRO_LOG_WARN, 2000, "Eject disk before selecting");
Initial commit - http://sourceforge.net/p/fceumm/code/160/
2014-03-30 22:15:17 +02:00
return;
}
SelectDisk = ((SelectDisk + 1) % TotalSides) & 3;
Add support for emulated Game Genie add-on
2021-10-21 15:52:40 +01:00
FCEU_DispMessage(RETRO_LOG_INFO, 2000, "Disk %d of %d Side %s Selected",
1 + (SelectDisk >> 1), (TotalSides + 1) >> 1, (SelectDisk & 1) ? "B" : "A");
Initial commit - http://sourceforge.net/p/fceumm/code/160/
2014-03-30 22:15:17 +02:00
}
FDS: Fix IRQ timing (#247) - Fixes Druid flickering status window
2018-12-16 08:50:08 +08:00
/* 2018/12/15 - update irq timings */
Initial commit - http://sourceforge.net/p/fceumm/code/160/
2014-03-30 22:15:17 +02:00
static void FP_FASTAPASS(1) FDSFix(int a) {
FDS: Fix IRQ timing (#247) - Fixes Druid flickering status window
2018-12-16 08:50:08 +08:00
if ((IRQa & 2) && (FDSRegs[3] & 0x1)) {
Initial commit - http://sourceforge.net/p/fceumm/code/160/
2014-03-30 22:15:17 +02:00
if (IRQCount <= 0) {
FDS: Fix IRQ timing (#247) - Fixes Druid flickering status window
2018-12-16 08:50:08 +08:00
if (!(IRQa & 1))
IRQa &= ~2; /* does not clear latch, fix Druid */
IRQCount = IRQLatch;
Initial commit - http://sourceforge.net/p/fceumm/code/160/
2014-03-30 22:15:17 +02:00
X6502_IRQBegin(FCEU_IQEXT);
FDS: Fix IRQ timing (#247) - Fixes Druid flickering status window
2018-12-16 08:50:08 +08:00
} else {
IRQCount -= a;
Initial commit - http://sourceforge.net/p/fceumm/code/160/
2014-03-30 22:15:17 +02:00
}
}
FDS: Fix IRQ timing (#247) - Fixes Druid flickering status window
2018-12-16 08:50:08 +08:00
Initial commit - http://sourceforge.net/p/fceumm/code/160/
2014-03-30 22:15:17 +02:00
if (DiskSeekIRQ > 0) {
DiskSeekIRQ -= a;
if (DiskSeekIRQ <= 0) {
if (FDSRegs[5] & 0x80) {
X6502_IRQBegin(FCEU_IQEXT2);
}
}
}
}
static DECLFR(FDSRead4030) {
core: stdint typedefs, LE optimizations, frame determinism Three follow-up audit passes on top of the memory-safety / leak / savestate-portability work in 1185db8. ============================================================== Pass 1: replace custom typedefs with C99 stdint types throughout ============================================================== The custom uint8 / uint16 / uint32 / uint64 / int8 / int16 / int32 / int64 typedefs in src/fceu-types.h were just simple aliases for the C99 stdint.h types. Replace them with the standard names directly. - 498 files modified - ~3,400 token replacements (uint8 -> uint8_t, etc) - fceu-types.h slimmed down to just INLINE / GINLINE / FASTAPASS macros and the readfunc / writefunc function-pointer typedefs (those now use uint8_t / uint32_t natively) - Build clean on `make platform=unix` with zero new warnings - Output binary size unchanged - confirming semantic equivalence Mechanical replacement done with a Python script that uses word- boundary regex to avoid false positives (e.g. 'uint32_t' was correctly left alone because '_' is a word character so 'uint32' is not a complete word inside it). ================================================================ Pass 2: prefer memcpy on LE hosts for endian read/write helpers ================================================================ fceu-endian.c's write32le_mem, FCEU_en32lsb, and FCEU_de32lsb performed bytewise composition/decomposition unconditionally. On LE hosts the in-memory representation already matches the desired LE on-disk format, so a single memcpy is equivalent and lets the compiler emit a single load/store rather than four byte ops. - The bytewise path is kept inside #ifdef MSB_FIRST for BE hosts where it implements the actual byte swap - Both forms produce identical results; this is a code-clarity change more than a performance one (the optimizer was already merging the shifts on LE), but it documents the intent and removes a strict-aliasing-flavoured cast through *(uint32_t*)Bufo - Added missing #include <string.h> in fceu-endian.c which was relying on transitive includes for memcpy Other MSB_FIRST sites in the codebase (state.c FlipByteOrder guards, ppu.c sprite-line rendering, boards/unrom512.c flash-write- counter access) were already optimized for LE; they were verified correct rather than changed. ================================================================ Pass 3: frame determinism for replay and netplay ================================================================ Two libc rand() sites in core were replaced with a local xorshift32 PRNG so that NES games which read uninitialised memory or hit hardware "weak bit" emulation produce reproducible behaviour across runs. NES titles routinely read uninitialised RAM (variables not zeroed before use, sprite Y-position set by junk-on-stack), so the RAM contents at power-on subtly affect game behaviour. With libc rand(), those contents depend on whether anyone else seeded rand() in the same process - a different libretro frontend, a different audio backend init order, or any frontend that does srand(time(0)) all break replay / netplay frame-determinism. 1. fceu.c FCEU_MemoryRand. Used to fill RAM (PowerNES) and CHR-RAM (iNES_Init) at power-on when option_ramstate=2 (random init). Replaced with a local xorshift32 PRNG, exposed via a new FCEU_MemoryRand_Reseed(uint32_t) function called once per power-on: - PowerNES seeds from the first 4 bytes of GameInfo->MD5 (set by all loaders before PowerNES runs) so identical ROMs produce identical RAM, different ROMs differ - iNES_Init seeds from iNESCart.PRGCRC32 before the CHR-RAM fill so two builds of the same ROM get the same CHR-RAM - The PRNG state advances across multiple FCEU_MemoryRand calls within one power-on so RAM and CHR-RAM get different content (matching NES hardware reality) 2. boards/rt-01.c UNLRT01Read. The RT-01 board has 'weak bit' protected EPROM regions; reads of 0xCE80-0xCEFF and 0xFE80- 0xFEFF return 0xF2 with the low 3 bits randomised. Replaced libc rand() with a local xorshift32 seeded at power-on, and added the PRNG state to the savestate via AddExState with key "WBKS" so save / load / rewind / netplay rollback all stay deterministic. In addition, two long-double-to-int truncations were changed to double for cross-platform FP determinism: - sound.c SetSoundVariables: soundtsinc - boards/n106.c DoNamcoSound: inc long double has platform-dependent precision (80-bit on x87, 64-bit with -mfpmath=sse, 128-bit on PowerPC), so the truncated integer result varied across these platforms. double is guaranteed 64-bit IEEE-754 portably. After this pass, the core has no time(), clock(), gettimeofday(), clock_gettime(), getpid(), getuid(), getgid(), getenv(), gethostid(), pthread, std::thread, OpenMP, signal handler, or non-deterministic- malloc dependency. Verified with a Python scanner that greps the source for these patterns; runs clean. The PPU / APU / CPU power-on already explicitly memset all state buffers to 0 (deterministic), and ROM/CHR-ROM allocation already memsets to 0xFF before partial fread (deterministic regardless of file truncation). Combined with the memory-safety hardening in 1185db8 (which prevents savestate-loaded indices from going out-of-bounds and producing unpredictable behaviour), the core now offers genuine frame-deterministic replay across runs, builds, and host endian.
2026-05-04 02:46:34 +02:00
uint8_t ret = 0;
Initial commit - http://sourceforge.net/p/fceumm/code/160/
2014-03-30 22:15:17 +02:00
/* Cheap hack. */
if (X.IRQlow & FCEU_IQEXT) ret |= 1;
if (X.IRQlow & FCEU_IQEXT2) ret |= 2;
#ifdef FCEUDEF_DEBUGGER
if (!fceuindbg)
#endif
{
X6502_IRQEnd(FCEU_IQEXT);
X6502_IRQEnd(FCEU_IQEXT2);
}
return ret;
}
(fds.c) backport new fds disk handling backported from fceux, commit by dinkc64 (fbneo) https://github.com/TASVideos/fceux/commit/0b4be4b639371204642f22def7bd92ea18a4c2f3 Fixes saving issue with some games (Bubble Bobble, Super Lode Runner II, ...)
2020-01-25 08:32:43 +08:00
static DECLFR(FDSRead4031) {
core: stdint typedefs, LE optimizations, frame determinism Three follow-up audit passes on top of the memory-safety / leak / savestate-portability work in 1185db8. ============================================================== Pass 1: replace custom typedefs with C99 stdint types throughout ============================================================== The custom uint8 / uint16 / uint32 / uint64 / int8 / int16 / int32 / int64 typedefs in src/fceu-types.h were just simple aliases for the C99 stdint.h types. Replace them with the standard names directly. - 498 files modified - ~3,400 token replacements (uint8 -> uint8_t, etc) - fceu-types.h slimmed down to just INLINE / GINLINE / FASTAPASS macros and the readfunc / writefunc function-pointer typedefs (those now use uint8_t / uint32_t natively) - Build clean on `make platform=unix` with zero new warnings - Output binary size unchanged - confirming semantic equivalence Mechanical replacement done with a Python script that uses word- boundary regex to avoid false positives (e.g. 'uint32_t' was correctly left alone because '_' is a word character so 'uint32' is not a complete word inside it). ================================================================ Pass 2: prefer memcpy on LE hosts for endian read/write helpers ================================================================ fceu-endian.c's write32le_mem, FCEU_en32lsb, and FCEU_de32lsb performed bytewise composition/decomposition unconditionally. On LE hosts the in-memory representation already matches the desired LE on-disk format, so a single memcpy is equivalent and lets the compiler emit a single load/store rather than four byte ops. - The bytewise path is kept inside #ifdef MSB_FIRST for BE hosts where it implements the actual byte swap - Both forms produce identical results; this is a code-clarity change more than a performance one (the optimizer was already merging the shifts on LE), but it documents the intent and removes a strict-aliasing-flavoured cast through *(uint32_t*)Bufo - Added missing #include <string.h> in fceu-endian.c which was relying on transitive includes for memcpy Other MSB_FIRST sites in the codebase (state.c FlipByteOrder guards, ppu.c sprite-line rendering, boards/unrom512.c flash-write- counter access) were already optimized for LE; they were verified correct rather than changed. ================================================================ Pass 3: frame determinism for replay and netplay ================================================================ Two libc rand() sites in core were replaced with a local xorshift32 PRNG so that NES games which read uninitialised memory or hit hardware "weak bit" emulation produce reproducible behaviour across runs. NES titles routinely read uninitialised RAM (variables not zeroed before use, sprite Y-position set by junk-on-stack), so the RAM contents at power-on subtly affect game behaviour. With libc rand(), those contents depend on whether anyone else seeded rand() in the same process - a different libretro frontend, a different audio backend init order, or any frontend that does srand(time(0)) all break replay / netplay frame-determinism. 1. fceu.c FCEU_MemoryRand. Used to fill RAM (PowerNES) and CHR-RAM (iNES_Init) at power-on when option_ramstate=2 (random init). Replaced with a local xorshift32 PRNG, exposed via a new FCEU_MemoryRand_Reseed(uint32_t) function called once per power-on: - PowerNES seeds from the first 4 bytes of GameInfo->MD5 (set by all loaders before PowerNES runs) so identical ROMs produce identical RAM, different ROMs differ - iNES_Init seeds from iNESCart.PRGCRC32 before the CHR-RAM fill so two builds of the same ROM get the same CHR-RAM - The PRNG state advances across multiple FCEU_MemoryRand calls within one power-on so RAM and CHR-RAM get different content (matching NES hardware reality) 2. boards/rt-01.c UNLRT01Read. The RT-01 board has 'weak bit' protected EPROM regions; reads of 0xCE80-0xCEFF and 0xFE80- 0xFEFF return 0xF2 with the low 3 bits randomised. Replaced libc rand() with a local xorshift32 seeded at power-on, and added the PRNG state to the savestate via AddExState with key "WBKS" so save / load / rewind / netplay rollback all stay deterministic. In addition, two long-double-to-int truncations were changed to double for cross-platform FP determinism: - sound.c SetSoundVariables: soundtsinc - boards/n106.c DoNamcoSound: inc long double has platform-dependent precision (80-bit on x87, 64-bit with -mfpmath=sse, 128-bit on PowerPC), so the truncated integer result varied across these platforms. double is guaranteed 64-bit IEEE-754 portably. After this pass, the core has no time(), clock(), gettimeofday(), clock_gettime(), getpid(), getuid(), getgid(), getenv(), gethostid(), pthread, std::thread, OpenMP, signal handler, or non-deterministic- malloc dependency. Verified with a Python scanner that greps the source for these patterns; runs clean. The PPU / APU / CPU power-on already explicitly memset all state buffers to 0 (deterministic), and ROM/CHR-ROM allocation already memsets to 0xFF before partial fread (deterministic regardless of file truncation). Combined with the memory-safety hardening in 1185db8 (which prevents savestate-loaded indices from going out-of-bounds and producing unpredictable behaviour), the core now offers genuine frame-deterministic replay across runs, builds, and host endian.
2026-05-04 02:46:34 +02:00
uint8_t ret = 0xff;
(fds.c) backport new fds disk handling backported from fceux, commit by dinkc64 (fbneo) https://github.com/TASVideos/fceux/commit/0b4be4b639371204642f22def7bd92ea18a4c2f3 Fixes saving issue with some games (Bubble Bobble, Super Lode Runner II, ...)
2020-01-25 08:32:43 +08:00
if (FDS_DISK_INSERTED && mapperFDS_control & 0x04) {
mapperFDS_diskaccess = 1;
ret = 0;
switch (mapperFDS_block) {
case DSK_FILEHDR:
if (mapperFDS_diskaddr < mapperFDS_blocklen) {
ret = GET_FDS_DISK();
switch (mapperFDS_diskaddr) {
case 13:
mapperFDS_filesize = ret;
break;
case 14:
mapperFDS_filesize |= ret << 8;
break;
}
mapperFDS_diskaddr++;
}
break;
default:
if (mapperFDS_diskaddr < mapperFDS_blocklen) {
ret = GET_FDS_DISK();
mapperFDS_diskaddr++;
}
break;
}
DiskSeekIRQ = 150;
X6502_IRQEnd(FCEU_IQEXT2);
}
return ret;
}
Initial commit - http://sourceforge.net/p/fceumm/code/160/
2014-03-30 22:15:17 +02:00
static DECLFR(FDSRead4032) {
core: stdint typedefs, LE optimizations, frame determinism Three follow-up audit passes on top of the memory-safety / leak / savestate-portability work in 1185db8. ============================================================== Pass 1: replace custom typedefs with C99 stdint types throughout ============================================================== The custom uint8 / uint16 / uint32 / uint64 / int8 / int16 / int32 / int64 typedefs in src/fceu-types.h were just simple aliases for the C99 stdint.h types. Replace them with the standard names directly. - 498 files modified - ~3,400 token replacements (uint8 -> uint8_t, etc) - fceu-types.h slimmed down to just INLINE / GINLINE / FASTAPASS macros and the readfunc / writefunc function-pointer typedefs (those now use uint8_t / uint32_t natively) - Build clean on `make platform=unix` with zero new warnings - Output binary size unchanged - confirming semantic equivalence Mechanical replacement done with a Python script that uses word- boundary regex to avoid false positives (e.g. 'uint32_t' was correctly left alone because '_' is a word character so 'uint32' is not a complete word inside it). ================================================================ Pass 2: prefer memcpy on LE hosts for endian read/write helpers ================================================================ fceu-endian.c's write32le_mem, FCEU_en32lsb, and FCEU_de32lsb performed bytewise composition/decomposition unconditionally. On LE hosts the in-memory representation already matches the desired LE on-disk format, so a single memcpy is equivalent and lets the compiler emit a single load/store rather than four byte ops. - The bytewise path is kept inside #ifdef MSB_FIRST for BE hosts where it implements the actual byte swap - Both forms produce identical results; this is a code-clarity change more than a performance one (the optimizer was already merging the shifts on LE), but it documents the intent and removes a strict-aliasing-flavoured cast through *(uint32_t*)Bufo - Added missing #include <string.h> in fceu-endian.c which was relying on transitive includes for memcpy Other MSB_FIRST sites in the codebase (state.c FlipByteOrder guards, ppu.c sprite-line rendering, boards/unrom512.c flash-write- counter access) were already optimized for LE; they were verified correct rather than changed. ================================================================ Pass 3: frame determinism for replay and netplay ================================================================ Two libc rand() sites in core were replaced with a local xorshift32 PRNG so that NES games which read uninitialised memory or hit hardware "weak bit" emulation produce reproducible behaviour across runs. NES titles routinely read uninitialised RAM (variables not zeroed before use, sprite Y-position set by junk-on-stack), so the RAM contents at power-on subtly affect game behaviour. With libc rand(), those contents depend on whether anyone else seeded rand() in the same process - a different libretro frontend, a different audio backend init order, or any frontend that does srand(time(0)) all break replay / netplay frame-determinism. 1. fceu.c FCEU_MemoryRand. Used to fill RAM (PowerNES) and CHR-RAM (iNES_Init) at power-on when option_ramstate=2 (random init). Replaced with a local xorshift32 PRNG, exposed via a new FCEU_MemoryRand_Reseed(uint32_t) function called once per power-on: - PowerNES seeds from the first 4 bytes of GameInfo->MD5 (set by all loaders before PowerNES runs) so identical ROMs produce identical RAM, different ROMs differ - iNES_Init seeds from iNESCart.PRGCRC32 before the CHR-RAM fill so two builds of the same ROM get the same CHR-RAM - The PRNG state advances across multiple FCEU_MemoryRand calls within one power-on so RAM and CHR-RAM get different content (matching NES hardware reality) 2. boards/rt-01.c UNLRT01Read. The RT-01 board has 'weak bit' protected EPROM regions; reads of 0xCE80-0xCEFF and 0xFE80- 0xFEFF return 0xF2 with the low 3 bits randomised. Replaced libc rand() with a local xorshift32 seeded at power-on, and added the PRNG state to the savestate via AddExState with key "WBKS" so save / load / rewind / netplay rollback all stay deterministic. In addition, two long-double-to-int truncations were changed to double for cross-platform FP determinism: - sound.c SetSoundVariables: soundtsinc - boards/n106.c DoNamcoSound: inc long double has platform-dependent precision (80-bit on x87, 64-bit with -mfpmath=sse, 128-bit on PowerPC), so the truncated integer result varied across these platforms. double is guaranteed 64-bit IEEE-754 portably. After this pass, the core has no time(), clock(), gettimeofday(), clock_gettime(), getpid(), getuid(), getgid(), getenv(), gethostid(), pthread, std::thread, OpenMP, signal handler, or non-deterministic- malloc dependency. Verified with a Python scanner that greps the source for these patterns; runs clean. The PPU / APU / CPU power-on already explicitly memset all state buffers to 0 (deterministic), and ROM/CHR-ROM allocation already memsets to 0xFF before partial fread (deterministic regardless of file truncation). Combined with the memory-safety hardening in 1185db8 (which prevents savestate-loaded indices from going out-of-bounds and producing unpredictable behaviour), the core now offers genuine frame-deterministic replay across runs, builds, and host endian.
2026-05-04 02:46:34 +02:00
uint8_t ret;
Initial commit - http://sourceforge.net/p/fceumm/code/160/
2014-03-30 22:15:17 +02:00
ret = X.DB & ~7;
if (InDisk == 255)
ret |= 5;
if (InDisk == 255 || !(FDSRegs[5] & 1) || (FDSRegs[5] & 2))
ret |= 2;
return ret;
}
static DECLFR(FDSRead4033) {
Use C-comment style
2017-10-15 03:13:11 +08:00
return 0x80; /* battery */
Initial commit - http://sourceforge.net/p/fceumm/code/160/
2014-03-30 22:15:17 +02:00
}
static DECLFW(FDSWrite) {
switch (A) {
case 0x4020:
IRQLatch &= 0xFF00;
IRQLatch |= V;
break;
case 0x4021:
IRQLatch &= 0xFF;
IRQLatch |= V << 8;
break;
case 0x4022:
FDS: Fix IRQ timing (#247) - Fixes Druid flickering status window
2018-12-16 08:50:08 +08:00
if (FDSRegs[3] & 0x1) {
IRQa = (V & 0x3);
if (IRQa & 2) {
IRQCount = IRQLatch;
} else {
X6502_IRQEnd(FCEU_IQEXT);
X6502_IRQEnd(FCEU_IQEXT2);
}
}
Initial commit - http://sourceforge.net/p/fceumm/code/160/
2014-03-30 22:15:17 +02:00
break;
FDS: Fix IRQ timing (#247) - Fixes Druid flickering status window
2018-12-16 08:50:08 +08:00
case 0x4023:
if (!(V & 1)) {
X6502_IRQEnd(FCEU_IQEXT);
X6502_IRQEnd(FCEU_IQEXT2);
}
Fix warnings [-Wimplicit-fallthrough=]
2020-10-21 20:55:08 +08:00
break;
Initial commit - http://sourceforge.net/p/fceumm/code/160/
2014-03-30 22:15:17 +02:00
case 0x4024:
(fds.c) backport new fds disk handling backported from fceux, commit by dinkc64 (fbneo) https://github.com/TASVideos/fceux/commit/0b4be4b639371204642f22def7bd92ea18a4c2f3 Fixes saving issue with some games (Bubble Bobble, Super Lode Runner II, ...)
2020-01-25 08:32:43 +08:00
if (FDS_DISK_INSERTED && ~mapperFDS_control & 0x04) {
if (mapperFDS_diskaccess == 0) {
mapperFDS_diskaccess = 1;
break;
}
switch (mapperFDS_block) {
case DSK_FILEHDR:
if (mapperFDS_diskaddr < mapperFDS_blocklen) {
GET_FDS_DISK() = V;
switch (mapperFDS_diskaddr) {
case 13: mapperFDS_filesize = V; break;
case 14:
mapperFDS_filesize |= V << 8;
break;
}
mapperFDS_diskaddr++;
}
break;
default:
if (mapperFDS_diskaddr < mapperFDS_blocklen) {
GET_FDS_DISK() = V;
mapperFDS_diskaddr++;
}
break;
}
}
Initial commit - http://sourceforge.net/p/fceumm/code/160/
2014-03-30 22:15:17 +02:00
break;
case 0x4025:
(fds.c) backport new fds disk handling backported from fceux, commit by dinkc64 (fbneo) https://github.com/TASVideos/fceux/commit/0b4be4b639371204642f22def7bd92ea18a4c2f3 Fixes saving issue with some games (Bubble Bobble, Super Lode Runner II, ...)
2020-01-25 08:32:43 +08:00
X6502_IRQEnd(FCEU_IQEXT2);
if (FDS_DISK_INSERTED) {
if (V & 0x40 && ~mapperFDS_control & 0x40) {
mapperFDS_diskaccess = 0;
DiskSeekIRQ = 150;
/* blockstart - address of block on disk
* diskaddr - address relative to blockstart
* _block -> _blockID ?
*/
mapperFDS_blockstart += mapperFDS_diskaddr;
mapperFDS_diskaddr = 0;
mapperFDS_block++;
if (mapperFDS_block > DSK_FILEDATA)
mapperFDS_block = DSK_FILEHDR;
switch (mapperFDS_block) {
case DSK_VOLUME:
mapperFDS_blocklen = 0x38;
break;
case DSK_FILECNT:
mapperFDS_blocklen = 0x02;
break;
case DSK_FILEHDR:
mapperFDS_blocklen = 0x10;
break;
case DSK_FILEDATA: /* <blockid><filedata> */
mapperFDS_blocklen = 0x01 + mapperFDS_filesize;
break;
}
}
if (V & 0x02) { /* transfer reset */
mapperFDS_block = DSK_INIT;
mapperFDS_blockstart = 0;
mapperFDS_blocklen = 0;
mapperFDS_diskaddr = 0;
DiskSeekIRQ = 150;
}
if (V & 0x40) { /* turn on motor */
DiskSeekIRQ = 150;
}
}
mapperFDS_control = V;
setmirror(((V >> 3) & 1) ^ 1);
Initial commit - http://sourceforge.net/p/fceumm/code/160/
2014-03-30 22:15:17 +02:00
break;
}
FDSRegs[A & 7] = V;
}
FDS: Print manufacturer label when loading rom
2019-01-04 22:33:26 +08:00
struct codes_t {
core: stdint typedefs, LE optimizations, frame determinism Three follow-up audit passes on top of the memory-safety / leak / savestate-portability work in 1185db8. ============================================================== Pass 1: replace custom typedefs with C99 stdint types throughout ============================================================== The custom uint8 / uint16 / uint32 / uint64 / int8 / int16 / int32 / int64 typedefs in src/fceu-types.h were just simple aliases for the C99 stdint.h types. Replace them with the standard names directly. - 498 files modified - ~3,400 token replacements (uint8 -> uint8_t, etc) - fceu-types.h slimmed down to just INLINE / GINLINE / FASTAPASS macros and the readfunc / writefunc function-pointer typedefs (those now use uint8_t / uint32_t natively) - Build clean on `make platform=unix` with zero new warnings - Output binary size unchanged - confirming semantic equivalence Mechanical replacement done with a Python script that uses word- boundary regex to avoid false positives (e.g. 'uint32_t' was correctly left alone because '_' is a word character so 'uint32' is not a complete word inside it). ================================================================ Pass 2: prefer memcpy on LE hosts for endian read/write helpers ================================================================ fceu-endian.c's write32le_mem, FCEU_en32lsb, and FCEU_de32lsb performed bytewise composition/decomposition unconditionally. On LE hosts the in-memory representation already matches the desired LE on-disk format, so a single memcpy is equivalent and lets the compiler emit a single load/store rather than four byte ops. - The bytewise path is kept inside #ifdef MSB_FIRST for BE hosts where it implements the actual byte swap - Both forms produce identical results; this is a code-clarity change more than a performance one (the optimizer was already merging the shifts on LE), but it documents the intent and removes a strict-aliasing-flavoured cast through *(uint32_t*)Bufo - Added missing #include <string.h> in fceu-endian.c which was relying on transitive includes for memcpy Other MSB_FIRST sites in the codebase (state.c FlipByteOrder guards, ppu.c sprite-line rendering, boards/unrom512.c flash-write- counter access) were already optimized for LE; they were verified correct rather than changed. ================================================================ Pass 3: frame determinism for replay and netplay ================================================================ Two libc rand() sites in core were replaced with a local xorshift32 PRNG so that NES games which read uninitialised memory or hit hardware "weak bit" emulation produce reproducible behaviour across runs. NES titles routinely read uninitialised RAM (variables not zeroed before use, sprite Y-position set by junk-on-stack), so the RAM contents at power-on subtly affect game behaviour. With libc rand(), those contents depend on whether anyone else seeded rand() in the same process - a different libretro frontend, a different audio backend init order, or any frontend that does srand(time(0)) all break replay / netplay frame-determinism. 1. fceu.c FCEU_MemoryRand. Used to fill RAM (PowerNES) and CHR-RAM (iNES_Init) at power-on when option_ramstate=2 (random init). Replaced with a local xorshift32 PRNG, exposed via a new FCEU_MemoryRand_Reseed(uint32_t) function called once per power-on: - PowerNES seeds from the first 4 bytes of GameInfo->MD5 (set by all loaders before PowerNES runs) so identical ROMs produce identical RAM, different ROMs differ - iNES_Init seeds from iNESCart.PRGCRC32 before the CHR-RAM fill so two builds of the same ROM get the same CHR-RAM - The PRNG state advances across multiple FCEU_MemoryRand calls within one power-on so RAM and CHR-RAM get different content (matching NES hardware reality) 2. boards/rt-01.c UNLRT01Read. The RT-01 board has 'weak bit' protected EPROM regions; reads of 0xCE80-0xCEFF and 0xFE80- 0xFEFF return 0xF2 with the low 3 bits randomised. Replaced libc rand() with a local xorshift32 seeded at power-on, and added the PRNG state to the savestate via AddExState with key "WBKS" so save / load / rewind / netplay rollback all stay deterministic. In addition, two long-double-to-int truncations were changed to double for cross-platform FP determinism: - sound.c SetSoundVariables: soundtsinc - boards/n106.c DoNamcoSound: inc long double has platform-dependent precision (80-bit on x87, 64-bit with -mfpmath=sse, 128-bit on PowerPC), so the truncated integer result varied across these platforms. double is guaranteed 64-bit IEEE-754 portably. After this pass, the core has no time(), clock(), gettimeofday(), clock_gettime(), getpid(), getuid(), getgid(), getenv(), gethostid(), pthread, std::thread, OpenMP, signal handler, or non-deterministic- malloc dependency. Verified with a Python scanner that greps the source for these patterns; runs clean. The PPU / APU / CPU power-on already explicitly memset all state buffers to 0 (deterministic), and ROM/CHR-ROM allocation already memsets to 0xFF before partial fread (deterministic regardless of file truncation). Combined with the memory-safety hardening in 1185db8 (which prevents savestate-loaded indices from going out-of-bounds and producing unpredictable behaviour), the core now offers genuine frame-deterministic replay across runs, builds, and host endian.
2026-05-04 02:46:34 +02:00
uint8_t code;
FDS: Print manufacturer label when loading rom
2019-01-04 22:33:26 +08:00
char *name;
};
static const struct codes_t list[] = {
{ 0x01, "Nintendo" },
{ 0x02, "Rocket Games" },
{ 0x08, "Capcom" },
{ 0x09, "Hot B Co." },
{ 0x0A, "Jaleco" },
{ 0x0B, "Coconuts Japan" },
{ 0x0C, "Coconuts Japan/G.X.Media" },
{ 0x13, "Electronic Arts Japan" },
{ 0x18, "Hudson Soft Japan" },
{ 0x19, "S.C.P." },
{ 0x1A, "Yonoman" },
{ 0x20, "Destination Software" },
{ 0x22, "VR 1 Japan" },
{ 0x25, "San-X" },
{ 0x28, "Kemco Japan" },
{ 0x29, "Seta" },
{ 0x36, "Codemasters" },
{ 0x37, "GAGA Communications" },
{ 0x38, "Laguna" },
{ 0x39, "Telstar Fun and Games" },
{ 0x41, "Ubi Soft Entertainment" },
{ 0x42, "Sunsoft" },
{ 0x47, "Spectrum Holobyte" },
{ 0x49, "Irem" },
{ 0x4A, "Gakken" },
{ 0x4D, "Malibu Games" },
{ 0x4F, "Eidos/U.S. Gold" },
{ 0x50, "Absolute Entertainment" },
{ 0x51, "Acclaim" },
{ 0x52, "Activision" },
{ 0x53, "American Sammy Corp." },
{ 0x54, "Take 2 Interactive" },
{ 0x55, "Hi Tech" },
{ 0x56, "LJN LTD." },
{ 0x58, "Mattel" },
{ 0x5A, "Mindscape/Red Orb Ent." },
{ 0x5C, "Taxan" },
{ 0x5D, "Midway" },
{ 0x5F, "American Softworks" },
{ 0x60, "Titus Interactive Studios" },
{ 0x61, "Virgin Interactive" },
{ 0x62, "Maxis" },
{ 0x64, "LucasArts Entertainment" },
{ 0x67, "Ocean" },
{ 0x69, "Electronic Arts" },
{ 0x6E, "Elite Systems Ltd." },
{ 0x6F, "Electro Brain" },
{ 0x70, "Infogrames" },
{ 0x71, "Interplay" },
{ 0x72, "JVC Musical Industries Inc" },
{ 0x73, "Parker Brothers" },
{ 0x75, "SCI" },
{ 0x78, "THQ" },
{ 0x79, "Accolade" },
{ 0x7A, "Triffix Ent. Inc." },
{ 0x7C, "Microprose Software" },
{ 0x7D, "Universal Interactive Studios" },
{ 0x7F, "Kemco" },
{ 0x80, "Misawa" },
{ 0x83, "LOZC" },
{ 0x8B, "Bulletproof Software" },
{ 0x8C, "Vic Tokai Inc." },
{ 0x91, "Chun Soft" },
{ 0x92, "Video System" },
{ 0x93, "BEC" },
{ 0x96, "Yonezawa/S'pal" },
{ 0x97, "Kaneko" },
{ 0x99, "Victor Interactive Software" },
{ 0x9A, "Nichibutsu/Nihon Bussan" },
{ 0x9B, "Tecmo" },
{ 0x9C, "Imagineer" },
{ 0x9F, "Nova" },
{ 0xA0, "Telenet" },
{ 0xA1, "Hori" },
{ 0xA2, "Scorpion Soft " },
{ 0xA4, "Konami" },
{ 0xA6, "Kawada" },
{ 0xA7, "Takara" },
{ 0xA8, "Royal Industries" },
{ 0xA9, "Technos Japan Corp." },
{ 0xAA, "JVC" },
{ 0xAC, "Toei Animation" },
{ 0xAD, "Toho" },
{ 0xAF, "Namco" },
{ 0xB0, "Acclaim Japan" },
{ 0xB1, "ASCII" },
{ 0xB2, "Bandai" },
{ 0xB3, "Soft Pro Inc." },
{ 0xB4, "Enix" },
{ 0xB6, "HAL Laboratory" },
{ 0xB7, "SNK" },
{ 0xB9, "Pony Canyon Hanbai" },
{ 0xBA, "Culture Brain" },
{ 0xBB, "Sunsoft" },
{ 0xBC, "Toshiba EMI" },
{ 0xBD, "Sony Imagesoft" },
{ 0xBF, "Sammy" },
{ 0xC0, "Taito" },
{ 0xC1, "Sunsoft / Ask Co., Ltd." },
{ 0xC2, "Kemco" },
{ 0xC3, "Square Soft" },
{ 0xC4, "Tokuma Shoten " },
{ 0xC5, "Data East" },
{ 0xC6, "Tonkin House" },
{ 0xC7, "East Cube" },
{ 0xC8, "Koei" },
{ 0xCA, "Konami/Palcom/Ultra" },
{ 0xCB, "Vapinc/NTVIC" },
{ 0xCC, "Use Co.,Ltd." },
{ 0xCD, "Meldac" },
{ 0xCE, "FCI/Pony Canyon" },
{ 0xCF, "Angel" },
{ 0xD1, "Sofel" },
/*{ 0xD2, "Quest" },*/
{ 0xD2, "Bothtec, Inc." },
{ 0xD3, "Sigma Enterprises" },
{ 0xD4, "Ask Kodansa" },
{ 0xD6, "Naxat" },
{ 0xD7, "Copya System" },
{ 0xD9, "Banpresto" },
{ 0xDA, "TOMY" },
/*{ 0xDB, "LJN Japan" },*/
{ 0xDB, "Hiro Co., Ltd." },
{ 0xDD, "NCS" },
{ 0xDF, "Altron Corporation" },
{ 0xE2, "Yutaka" },
{ 0xE3, "Varie" },
{ 0xE5, "Epoch" },
{ 0xE7, "Athena" },
{ 0xE8, "Asmik Ace Entertainment Inc." },
{ 0xE9, "Natsume" },
{ 0xEA, "King Records" },
{ 0xEB, "Atlus" },
{ 0xEC, "Epic/Sony Records" },
{ 0xEE, "IGS" },
{ 0xF0, "A Wave" },
{ 0 }
};
core: stdint typedefs, LE optimizations, frame determinism Three follow-up audit passes on top of the memory-safety / leak / savestate-portability work in 1185db8. ============================================================== Pass 1: replace custom typedefs with C99 stdint types throughout ============================================================== The custom uint8 / uint16 / uint32 / uint64 / int8 / int16 / int32 / int64 typedefs in src/fceu-types.h were just simple aliases for the C99 stdint.h types. Replace them with the standard names directly. - 498 files modified - ~3,400 token replacements (uint8 -> uint8_t, etc) - fceu-types.h slimmed down to just INLINE / GINLINE / FASTAPASS macros and the readfunc / writefunc function-pointer typedefs (those now use uint8_t / uint32_t natively) - Build clean on `make platform=unix` with zero new warnings - Output binary size unchanged - confirming semantic equivalence Mechanical replacement done with a Python script that uses word- boundary regex to avoid false positives (e.g. 'uint32_t' was correctly left alone because '_' is a word character so 'uint32' is not a complete word inside it). ================================================================ Pass 2: prefer memcpy on LE hosts for endian read/write helpers ================================================================ fceu-endian.c's write32le_mem, FCEU_en32lsb, and FCEU_de32lsb performed bytewise composition/decomposition unconditionally. On LE hosts the in-memory representation already matches the desired LE on-disk format, so a single memcpy is equivalent and lets the compiler emit a single load/store rather than four byte ops. - The bytewise path is kept inside #ifdef MSB_FIRST for BE hosts where it implements the actual byte swap - Both forms produce identical results; this is a code-clarity change more than a performance one (the optimizer was already merging the shifts on LE), but it documents the intent and removes a strict-aliasing-flavoured cast through *(uint32_t*)Bufo - Added missing #include <string.h> in fceu-endian.c which was relying on transitive includes for memcpy Other MSB_FIRST sites in the codebase (state.c FlipByteOrder guards, ppu.c sprite-line rendering, boards/unrom512.c flash-write- counter access) were already optimized for LE; they were verified correct rather than changed. ================================================================ Pass 3: frame determinism for replay and netplay ================================================================ Two libc rand() sites in core were replaced with a local xorshift32 PRNG so that NES games which read uninitialised memory or hit hardware "weak bit" emulation produce reproducible behaviour across runs. NES titles routinely read uninitialised RAM (variables not zeroed before use, sprite Y-position set by junk-on-stack), so the RAM contents at power-on subtly affect game behaviour. With libc rand(), those contents depend on whether anyone else seeded rand() in the same process - a different libretro frontend, a different audio backend init order, or any frontend that does srand(time(0)) all break replay / netplay frame-determinism. 1. fceu.c FCEU_MemoryRand. Used to fill RAM (PowerNES) and CHR-RAM (iNES_Init) at power-on when option_ramstate=2 (random init). Replaced with a local xorshift32 PRNG, exposed via a new FCEU_MemoryRand_Reseed(uint32_t) function called once per power-on: - PowerNES seeds from the first 4 bytes of GameInfo->MD5 (set by all loaders before PowerNES runs) so identical ROMs produce identical RAM, different ROMs differ - iNES_Init seeds from iNESCart.PRGCRC32 before the CHR-RAM fill so two builds of the same ROM get the same CHR-RAM - The PRNG state advances across multiple FCEU_MemoryRand calls within one power-on so RAM and CHR-RAM get different content (matching NES hardware reality) 2. boards/rt-01.c UNLRT01Read. The RT-01 board has 'weak bit' protected EPROM regions; reads of 0xCE80-0xCEFF and 0xFE80- 0xFEFF return 0xF2 with the low 3 bits randomised. Replaced libc rand() with a local xorshift32 seeded at power-on, and added the PRNG state to the savestate via AddExState with key "WBKS" so save / load / rewind / netplay rollback all stay deterministic. In addition, two long-double-to-int truncations were changed to double for cross-platform FP determinism: - sound.c SetSoundVariables: soundtsinc - boards/n106.c DoNamcoSound: inc long double has platform-dependent precision (80-bit on x87, 64-bit with -mfpmath=sse, 128-bit on PowerPC), so the truncated integer result varied across these platforms. double is guaranteed 64-bit IEEE-754 portably. After this pass, the core has no time(), clock(), gettimeofday(), clock_gettime(), getpid(), getuid(), getgid(), getenv(), gethostid(), pthread, std::thread, OpenMP, signal handler, or non-deterministic- malloc dependency. Verified with a Python scanner that greps the source for these patterns; runs clean. The PPU / APU / CPU power-on already explicitly memset all state buffers to 0 (deterministic), and ROM/CHR-ROM allocation already memsets to 0xFF before partial fread (deterministic regardless of file truncation). Combined with the memory-safety hardening in 1185db8 (which prevents savestate-loaded indices from going out-of-bounds and producing unpredictable behaviour), the core now offers genuine frame-deterministic replay across runs, builds, and host endian.
2026-05-04 02:46:34 +02:00
static const char *getManufacturer(uint8_t code)
FDS: Print manufacturer label when loading rom
2019-01-04 22:33:26 +08:00
{
int x = 0;
char *ret = "unlicensed";
while (list[x].code != 0) {
if (list[x].code == code) {
ret = list[x].name;
break;
}
x++;
}
return ret;
}
Initial commit - http://sourceforge.net/p/fceumm/code/160/
2014-03-30 22:15:17 +02:00
static void FreeFDSMemory(void) {
FDS: Use retro_get_memory_data/size for disk saving
2019-01-04 17:13:09 +08:00
if (FDSROM)
free(FDSROM);
FDSROM = NULL;
if (FDSBIOS)
free(FDSBIOS);
FDSBIOS = NULL;
if (FDSRAM)
free(FDSRAM);
FDSRAM = NULL;
if (CHRRAM)
free(CHRRAM);
CHRRAM = NULL;
Initial commit - http://sourceforge.net/p/fceumm/code/160/
2014-03-30 22:15:17 +02:00
}
static int SubLoad(FCEUFILE *fp) {
struct md5_context md5;
core: stdint typedefs, LE optimizations, frame determinism Three follow-up audit passes on top of the memory-safety / leak / savestate-portability work in 1185db8. ============================================================== Pass 1: replace custom typedefs with C99 stdint types throughout ============================================================== The custom uint8 / uint16 / uint32 / uint64 / int8 / int16 / int32 / int64 typedefs in src/fceu-types.h were just simple aliases for the C99 stdint.h types. Replace them with the standard names directly. - 498 files modified - ~3,400 token replacements (uint8 -> uint8_t, etc) - fceu-types.h slimmed down to just INLINE / GINLINE / FASTAPASS macros and the readfunc / writefunc function-pointer typedefs (those now use uint8_t / uint32_t natively) - Build clean on `make platform=unix` with zero new warnings - Output binary size unchanged - confirming semantic equivalence Mechanical replacement done with a Python script that uses word- boundary regex to avoid false positives (e.g. 'uint32_t' was correctly left alone because '_' is a word character so 'uint32' is not a complete word inside it). ================================================================ Pass 2: prefer memcpy on LE hosts for endian read/write helpers ================================================================ fceu-endian.c's write32le_mem, FCEU_en32lsb, and FCEU_de32lsb performed bytewise composition/decomposition unconditionally. On LE hosts the in-memory representation already matches the desired LE on-disk format, so a single memcpy is equivalent and lets the compiler emit a single load/store rather than four byte ops. - The bytewise path is kept inside #ifdef MSB_FIRST for BE hosts where it implements the actual byte swap - Both forms produce identical results; this is a code-clarity change more than a performance one (the optimizer was already merging the shifts on LE), but it documents the intent and removes a strict-aliasing-flavoured cast through *(uint32_t*)Bufo - Added missing #include <string.h> in fceu-endian.c which was relying on transitive includes for memcpy Other MSB_FIRST sites in the codebase (state.c FlipByteOrder guards, ppu.c sprite-line rendering, boards/unrom512.c flash-write- counter access) were already optimized for LE; they were verified correct rather than changed. ================================================================ Pass 3: frame determinism for replay and netplay ================================================================ Two libc rand() sites in core were replaced with a local xorshift32 PRNG so that NES games which read uninitialised memory or hit hardware "weak bit" emulation produce reproducible behaviour across runs. NES titles routinely read uninitialised RAM (variables not zeroed before use, sprite Y-position set by junk-on-stack), so the RAM contents at power-on subtly affect game behaviour. With libc rand(), those contents depend on whether anyone else seeded rand() in the same process - a different libretro frontend, a different audio backend init order, or any frontend that does srand(time(0)) all break replay / netplay frame-determinism. 1. fceu.c FCEU_MemoryRand. Used to fill RAM (PowerNES) and CHR-RAM (iNES_Init) at power-on when option_ramstate=2 (random init). Replaced with a local xorshift32 PRNG, exposed via a new FCEU_MemoryRand_Reseed(uint32_t) function called once per power-on: - PowerNES seeds from the first 4 bytes of GameInfo->MD5 (set by all loaders before PowerNES runs) so identical ROMs produce identical RAM, different ROMs differ - iNES_Init seeds from iNESCart.PRGCRC32 before the CHR-RAM fill so two builds of the same ROM get the same CHR-RAM - The PRNG state advances across multiple FCEU_MemoryRand calls within one power-on so RAM and CHR-RAM get different content (matching NES hardware reality) 2. boards/rt-01.c UNLRT01Read. The RT-01 board has 'weak bit' protected EPROM regions; reads of 0xCE80-0xCEFF and 0xFE80- 0xFEFF return 0xF2 with the low 3 bits randomised. Replaced libc rand() with a local xorshift32 seeded at power-on, and added the PRNG state to the savestate via AddExState with key "WBKS" so save / load / rewind / netplay rollback all stay deterministic. In addition, two long-double-to-int truncations were changed to double for cross-platform FP determinism: - sound.c SetSoundVariables: soundtsinc - boards/n106.c DoNamcoSound: inc long double has platform-dependent precision (80-bit on x87, 64-bit with -mfpmath=sse, 128-bit on PowerPC), so the truncated integer result varied across these platforms. double is guaranteed 64-bit IEEE-754 portably. After this pass, the core has no time(), clock(), gettimeofday(), clock_gettime(), getpid(), getuid(), getgid(), getenv(), gethostid(), pthread, std::thread, OpenMP, signal handler, or non-deterministic- malloc dependency. Verified with a Python scanner that greps the source for these patterns; runs clean. The PPU / APU / CPU power-on already explicitly memset all state buffers to 0 (deterministic), and ROM/CHR-ROM allocation already memsets to 0xFF before partial fread (deterministic regardless of file truncation). Combined with the memory-safety hardening in 1185db8 (which prevents savestate-loaded indices from going out-of-bounds and producing unpredictable behaviour), the core now offers genuine frame-deterministic replay across runs, builds, and host endian.
2026-05-04 02:46:34 +02:00
uint8_t header[16];
Initial commit - http://sourceforge.net/p/fceumm/code/160/
2014-03-30 22:15:17 +02:00
int x;
core: stdint typedefs, LE optimizations, frame determinism Three follow-up audit passes on top of the memory-safety / leak / savestate-portability work in 1185db8. ============================================================== Pass 1: replace custom typedefs with C99 stdint types throughout ============================================================== The custom uint8 / uint16 / uint32 / uint64 / int8 / int16 / int32 / int64 typedefs in src/fceu-types.h were just simple aliases for the C99 stdint.h types. Replace them with the standard names directly. - 498 files modified - ~3,400 token replacements (uint8 -> uint8_t, etc) - fceu-types.h slimmed down to just INLINE / GINLINE / FASTAPASS macros and the readfunc / writefunc function-pointer typedefs (those now use uint8_t / uint32_t natively) - Build clean on `make platform=unix` with zero new warnings - Output binary size unchanged - confirming semantic equivalence Mechanical replacement done with a Python script that uses word- boundary regex to avoid false positives (e.g. 'uint32_t' was correctly left alone because '_' is a word character so 'uint32' is not a complete word inside it). ================================================================ Pass 2: prefer memcpy on LE hosts for endian read/write helpers ================================================================ fceu-endian.c's write32le_mem, FCEU_en32lsb, and FCEU_de32lsb performed bytewise composition/decomposition unconditionally. On LE hosts the in-memory representation already matches the desired LE on-disk format, so a single memcpy is equivalent and lets the compiler emit a single load/store rather than four byte ops. - The bytewise path is kept inside #ifdef MSB_FIRST for BE hosts where it implements the actual byte swap - Both forms produce identical results; this is a code-clarity change more than a performance one (the optimizer was already merging the shifts on LE), but it documents the intent and removes a strict-aliasing-flavoured cast through *(uint32_t*)Bufo - Added missing #include <string.h> in fceu-endian.c which was relying on transitive includes for memcpy Other MSB_FIRST sites in the codebase (state.c FlipByteOrder guards, ppu.c sprite-line rendering, boards/unrom512.c flash-write- counter access) were already optimized for LE; they were verified correct rather than changed. ================================================================ Pass 3: frame determinism for replay and netplay ================================================================ Two libc rand() sites in core were replaced with a local xorshift32 PRNG so that NES games which read uninitialised memory or hit hardware "weak bit" emulation produce reproducible behaviour across runs. NES titles routinely read uninitialised RAM (variables not zeroed before use, sprite Y-position set by junk-on-stack), so the RAM contents at power-on subtly affect game behaviour. With libc rand(), those contents depend on whether anyone else seeded rand() in the same process - a different libretro frontend, a different audio backend init order, or any frontend that does srand(time(0)) all break replay / netplay frame-determinism. 1. fceu.c FCEU_MemoryRand. Used to fill RAM (PowerNES) and CHR-RAM (iNES_Init) at power-on when option_ramstate=2 (random init). Replaced with a local xorshift32 PRNG, exposed via a new FCEU_MemoryRand_Reseed(uint32_t) function called once per power-on: - PowerNES seeds from the first 4 bytes of GameInfo->MD5 (set by all loaders before PowerNES runs) so identical ROMs produce identical RAM, different ROMs differ - iNES_Init seeds from iNESCart.PRGCRC32 before the CHR-RAM fill so two builds of the same ROM get the same CHR-RAM - The PRNG state advances across multiple FCEU_MemoryRand calls within one power-on so RAM and CHR-RAM get different content (matching NES hardware reality) 2. boards/rt-01.c UNLRT01Read. The RT-01 board has 'weak bit' protected EPROM regions; reads of 0xCE80-0xCEFF and 0xFE80- 0xFEFF return 0xF2 with the low 3 bits randomised. Replaced libc rand() with a local xorshift32 seeded at power-on, and added the PRNG state to the savestate via AddExState with key "WBKS" so save / load / rewind / netplay rollback all stay deterministic. In addition, two long-double-to-int truncations were changed to double for cross-platform FP determinism: - sound.c SetSoundVariables: soundtsinc - boards/n106.c DoNamcoSound: inc long double has platform-dependent precision (80-bit on x87, 64-bit with -mfpmath=sse, 128-bit on PowerPC), so the truncated integer result varied across these platforms. double is guaranteed 64-bit IEEE-754 portably. After this pass, the core has no time(), clock(), gettimeofday(), clock_gettime(), getpid(), getuid(), getgid(), getenv(), gethostid(), pthread, std::thread, OpenMP, signal handler, or non-deterministic- malloc dependency. Verified with a Python scanner that greps the source for these patterns; runs clean. The PPU / APU / CPU power-on already explicitly memset all state buffers to 0 (deterministic), and ROM/CHR-ROM allocation already memsets to 0xFF before partial fread (deterministic regardless of file truncation). Combined with the memory-safety hardening in 1185db8 (which prevents savestate-loaded indices from going out-of-bounds and producing unpredictable behaviour), the core now offers genuine frame-deterministic replay across runs, builds, and host endian.
2026-05-04 02:46:34 +02:00
uint64_t fsize = FCEU_fgetsize(fp);
core: memory-safety, leak, and savestate-portability audit fixes Squashed series of ~50 distinct bugs found during a multi-day security/correctness audit, ranging from ROM-triggerable heap corruption and savestate-triggered OOB read primitives down to obscure cross-platform savestate breakage. Build is clean on `make platform=unix` with zero new warnings. CRITICAL (ROM-triggerable, exploitable on the user's machine) ============================================================= * ines.c iNES 2.0 PRG/CHR exponent overflow leading to undersized allocation followed by heap buffer overflow on fread (pow() with attacker-chosen exponent up to 63). Cap exponent at 30 and compute size in uint32 with explicit cap below 2 GiB before storing in int. * ines.c miscROMSize int wraparound from attacker-controlled PRG/ CHR sizes; previous '& 0x8000000' check missed common underflow cases. Compute in int64 and reject <= 0 or > 128 MiB. * unif.c MAPR chunk OOB heap read on chunks of size < 4. Validate chunk size before allocating board-name buffer. * unif.c chunk size truncation: int conversion + missing cap allowed a 4 GiB chunk size to wrap. Cap at 16 MiB. * unif.c FixRomSize infinite loop on size > 0x80000000. Cap input. * unif.c DINF chunk: months[(m - 1) % 12] is UB for m==0; m comes from the .unf file. Clamp m into [1..12] before subtracting. * unif.c NAME chunk: GameInfo->name = malloc(...) was unchecked. * nsf.c size underflow: FCEU_fgetsize() - 0x80 wraps to ~UINT64_MAX on tiny files, propagating a huge size into uppow2/FCEU_malloc. Validate size > 0x80 before subtracting. * nsf.c NSFMaxBank * 4096 cap tightened from 1<<20 (UB on signed-int overflow) to 1<<19 (fits in int). * general.c uppow2 had signed-int UB (1 << 32) and wrapped to 0 for inputs near uint32 max. Cap at 0x80000000. * cheat.c SubCheats[256] BSS overflow when more than 256 GG/PAR substitution cheats are active. The array is adjacent to MMapPtrs[64] in BSS which is exposed via retro_get_memory_data, making this overflow visible from outside the core. * libretro.c retro_cheat_set strcpy stack overflow with arbitrary- length cheat strings from the frontend. Use strlcpy. CRITICAL (savestate-triggerable on a malicious .fcs file) ========================================================= * fds.c FDS savestate OOB read primitive: InDisk loaded from savestate is used as index into 8-element diskdata[] static pointer array. A value 8..254 dereferences arbitrary memory as a pointer (heap-read primitive). Also bounds-checked SelectDisk, mapperFDS_block, mapperFDS_blockstart, mapperFDS_diskaddr, mapperFDS_blocklen so blockstart+diskaddr stays within the 65500- byte disk buffer. * 11 mappers had savestate-loaded variables masked at write time but not at restore time, used as indices into fixed arrays: - 88.c, KS7037.c, 112.c cmd indexes reg[8] - sachen.c (S8259, S74LS374N) cmd indexes latch[8] - 357.c dipswitch indexes outer_bank[4] - unrom512.c flash_state indexes erase_a/d/b[5] - 368.c preg indexes banks[8] - 69.c sndcmd indexes sreg[14] - mmc2and4.c latch0/latch1 index creg[4] None individually exploitable into RCE - the array writes corrupt adjacent BSS with constrained data flowing in - but each is an out-of-bounds read or write from attacker-controllable input. HIGH (memory-safety, reachable on any load) =========================================== * fceu-memory.c FCEU_malloc deref-of-NULL on allocation failure ('ret = 0; memset(ret, 0, size);'). * file.c multiple FCEUFILE/MakeMemWrap/MakeMemWrapBuffer unchecked allocations and unchecked filestream_tell return. * libretro.c GameInfo NULL-derefs in three entry points (retro_set_controller_port_device, retro_get_memory_data, retro_get_memory_size) reachable on operations called before a successful load. * libretro.c framebuffer leak ~256 KB per failed FCEUI_LoadGame (the libretro frontend doesn't call retro_unload_game on a failed load). * libretro.c 3DS retro_deinit unchecked linearFree. * libretro.c stereo / NTSC filter unchecked malloc returns. * fceu.c FCEUI_LoadGame unchecked GameInfo malloc. * fds.c SubLoad and FDSLoad unchecked FCEU_malloc on diskdatao backup buffers (NULL-deref in subsequent memcpy). * fceu.c AllocGenieRW partial-failure leak: AReadG allocated but BWriteG malloc fails -> 256 KB leak per retry. * input/bworld.c Update(): unbounded strcpy from attacker-supplied data into 20-byte bdata. Replaced with length-bounded copy. * cart.c setprg2r/4r and setchr1r/2r/4r/8r mask-underflow guards. SetupCartPRGMapping/SetupCartCHRMapping compute (size >> N) - 1 which underflows to 0xFFFFFFFF for chips smaller than the unit. setprg8r/16r/32r and setchr8r already had defensive size checks; the smaller units did not. malee.c (2 KB chip) and mapper 218 (2 KB NTARAM-as-CHR) accidentally avoid OOB; the primitive is unsafe for any future board. * video.c FCEU_InitVirtualVideo XBuf/XDBuf partial-failure leak. * video.c, fceu.c FCEU_DispMessage / FCEU_printf / FCEU_PrintError: vsprintf into fixed stack buffers replaced with vsnprintf. * core: validate magic-string read length on FDS/UNIF/NSF load (reject files too short to contain the magic so previous static buffer / stack garbage doesn't spuriously match). LEAKS (every cart load/unload cycle) ==================================== * mmc5.c MMC5 cart-side: WRAM (up to 64K) + MMC5fill (1K) + ExRAM (1K) leaked per cycle. NSFMMC5_Close existed but was only used for NSF code path. * onebus.c Mapper 270/436: 8 KB CHRRAM leaked per cycle. * 330.c, 375.c, 528.c: 8 KB WRAM each, no Close at all. CORRECTNESS (UB / ABI / endianness) =================================== * fceu-endian.c FlipByteOrder loop bound was 'count' instead of 'count/2', making it a no-op for every even count and silently breaking savestate portability for every FCEUSTATE_RLSB-marked field. The '#ifndef GEKKO' workarounds scattered through the codebase (sound.c, vrc7.c, vrc6.c) were symptoms of this root cause. * state.c AddExState bounds check ran AFTER writing the entry, so when SFEXINDEX hit the 63-entry cap the next call would write the entry then immediately overwrite it with the terminator. * state.c FCEUSS_Save_Mem: post-save callback was guarded by 'if (SPreSave)' instead of 'if (SPostSave)'. * Sequence-point UB in counter expressions across 4 board files, e.g. 'x = !x ? a : --x' and 'x = ++x % n'. GCC -Wsequence-point flags all five sites (285.c, 413.c, asic_mmc3.c, jyasic.c). * SFORMAT size mismatches between declared variable type and AddExState size argument: - asic_vrc3.c VRC3_count/VRC3_reload (uint16, saved as 1 byte) - mmc3.c m555_count (uint32, saved as 2 bytes), m555_count_- expired (uint8, saved as 2 bytes - OOB write into adjacent BSS). * unrom512.c UNROM-512 mapper 30 .srm save format: host-endian uint32 flash write counters. Now stored as LE on disk regardless of host (Battle Kid 2, Twin Dragons, Lizard, Sole, etc.). * ~70 multi-byte single-variable SFORMAT/AddExState entries across 36 board files were saved as host-byte-order. After the FlipByteOrder fix, these are now byte-swapped on BE so cross- platform savestates round-trip correctly. * coolgirl.c new ExStateLE() macro added; 22 multi-byte sites. * pic16c5x.c 11 multi-byte AddExState calls (m_PC, m_PREVPC, m_CONFIG, m_WDT, m_prescaler, m_opcode, m_STACK[0/1], m_icount, m_delay_timer, m_rtcc, m_inst_cycles, m_clock2cycle). * onebus.c PowerJoy Supermax submapper detection used non-portable *(uint32*)&info->MD5 cast that read different bytes on LE vs BE. * fds.c clean up redundant FCEUSTATE_RLSB encoding in AddExState calls that also passed type=1 (idempotent OR; readability fix). DOCUMENTATION ============= * input.c UpdateGP's *(uint32*)data cast looks like a typical endian bug but is actually correct (the libretro frontend builds JSReturn with matching host-uint32 shifts). Comment added to prevent future "fixes" from breaking it. Limitations not addressed ========================= * Element-stride-aware byte swapping. The savestate byte-swap mechanism (FlipByteOrder over the entire SFORMAT entry buffer) is structurally wrong for arrays of multi-byte values: it reverses the whole buffer end-to-end instead of byte-swapping each element. Several places that need cross-platform-portable arrays (VRC7 sound state, jyasic chr[8]) work around this by either splitting arrays into per-element SFORMAT entries (n106 PlayIndex, bandai reg) or by skipping save entirely on BE via #ifndef GEKKO. A proper fix would extend the size encoding with an element-stride field. Left for a future change because it would change the savestate format. * Strict-aliasing UB in ppu.c (around 9 sites doing *(uint32*)uint8_buf for fast 4-byte writes via FCEU_dwmemset). Works in practice with all common compilers because the patterns are byte-symmetric, but is formally UB. * FCEU_gmalloc calls exit(1) on OOM. A libretro core should never exit() because that takes down the whole frontend. Used by 100+ call sites; refactoring to return-NULL is out of scope here. Testing ======= * Build: clean on `make platform=unix` with -O2; no new warnings. * FlipByteOrder fix verified by hand-trace and a standalone unit test for counts 2, 4, 8. * uppow2 fix verified by unit test across 13 boundary cases. * SFORMAT size mismatches and missing-RLSB cases identified by Python static-analysis scripts that cross-reference SFORMAT entries against variable declarations. * iNES 2.0 exponent fix verified by hand-tracing what byte 0xFF produces post-fix: exp=30 (capped), mult=7, size=3 GiB nominal, capped to 1 GiB, capped to 2 GiB by uppow2, FCEU_malloc returns NULL on most systems, loader returns 0. No heap overflow for any input byte. * Savestate-loaded array index audit: built a Python scanner that extracts each AddExState/SFORMAT entry and cross-references the variable name against array-index uses in the same file. All flagged sites covered. * A libFuzzer harness and seed corpus generator (fuzz_main.c, gen_seed_corpus.py) accompany this submission for ongoing regression testing.
2026-05-04 02:15:40 +02:00
/* Reject files too short to contain a 16-byte header. Otherwise the
* subsequent FCEU_fread leaves header[] partially uninitialised, and
* the magic-string memcmps and side-count read at header[4] could
* spuriously succeed on stack garbage. */
if (fsize < 16)
return(0);
Initial commit - http://sourceforge.net/p/fceumm/code/160/
2014-03-30 22:15:17 +02:00
FCEU_fread(header, 16, 1, fp);
if (memcmp(header, "FDS\x1a", 4)) {
if (!(memcmp(header + 1, "*NINTENDO-HVC*", 14))) {
long t;
t = FCEU_fgetsize(fp);
if (t < 65500)
t = 65500;
TotalSides = t / 65500;
FCEU_fseek(fp, 0, SEEK_SET);
} else
return(0);
} else
TotalSides = header[4];
if (TotalSides > 8) TotalSides = 8;
if (TotalSides < 1) TotalSides = 1;
FDS: Use retro_get_memory_data/size for disk saving
2019-01-04 17:13:09 +08:00
FDSROMSize = TotalSides * BYTES_PER_SIDE;
core: stdint typedefs, LE optimizations, frame determinism Three follow-up audit passes on top of the memory-safety / leak / savestate-portability work in 1185db8. ============================================================== Pass 1: replace custom typedefs with C99 stdint types throughout ============================================================== The custom uint8 / uint16 / uint32 / uint64 / int8 / int16 / int32 / int64 typedefs in src/fceu-types.h were just simple aliases for the C99 stdint.h types. Replace them with the standard names directly. - 498 files modified - ~3,400 token replacements (uint8 -> uint8_t, etc) - fceu-types.h slimmed down to just INLINE / GINLINE / FASTAPASS macros and the readfunc / writefunc function-pointer typedefs (those now use uint8_t / uint32_t natively) - Build clean on `make platform=unix` with zero new warnings - Output binary size unchanged - confirming semantic equivalence Mechanical replacement done with a Python script that uses word- boundary regex to avoid false positives (e.g. 'uint32_t' was correctly left alone because '_' is a word character so 'uint32' is not a complete word inside it). ================================================================ Pass 2: prefer memcpy on LE hosts for endian read/write helpers ================================================================ fceu-endian.c's write32le_mem, FCEU_en32lsb, and FCEU_de32lsb performed bytewise composition/decomposition unconditionally. On LE hosts the in-memory representation already matches the desired LE on-disk format, so a single memcpy is equivalent and lets the compiler emit a single load/store rather than four byte ops. - The bytewise path is kept inside #ifdef MSB_FIRST for BE hosts where it implements the actual byte swap - Both forms produce identical results; this is a code-clarity change more than a performance one (the optimizer was already merging the shifts on LE), but it documents the intent and removes a strict-aliasing-flavoured cast through *(uint32_t*)Bufo - Added missing #include <string.h> in fceu-endian.c which was relying on transitive includes for memcpy Other MSB_FIRST sites in the codebase (state.c FlipByteOrder guards, ppu.c sprite-line rendering, boards/unrom512.c flash-write- counter access) were already optimized for LE; they were verified correct rather than changed. ================================================================ Pass 3: frame determinism for replay and netplay ================================================================ Two libc rand() sites in core were replaced with a local xorshift32 PRNG so that NES games which read uninitialised memory or hit hardware "weak bit" emulation produce reproducible behaviour across runs. NES titles routinely read uninitialised RAM (variables not zeroed before use, sprite Y-position set by junk-on-stack), so the RAM contents at power-on subtly affect game behaviour. With libc rand(), those contents depend on whether anyone else seeded rand() in the same process - a different libretro frontend, a different audio backend init order, or any frontend that does srand(time(0)) all break replay / netplay frame-determinism. 1. fceu.c FCEU_MemoryRand. Used to fill RAM (PowerNES) and CHR-RAM (iNES_Init) at power-on when option_ramstate=2 (random init). Replaced with a local xorshift32 PRNG, exposed via a new FCEU_MemoryRand_Reseed(uint32_t) function called once per power-on: - PowerNES seeds from the first 4 bytes of GameInfo->MD5 (set by all loaders before PowerNES runs) so identical ROMs produce identical RAM, different ROMs differ - iNES_Init seeds from iNESCart.PRGCRC32 before the CHR-RAM fill so two builds of the same ROM get the same CHR-RAM - The PRNG state advances across multiple FCEU_MemoryRand calls within one power-on so RAM and CHR-RAM get different content (matching NES hardware reality) 2. boards/rt-01.c UNLRT01Read. The RT-01 board has 'weak bit' protected EPROM regions; reads of 0xCE80-0xCEFF and 0xFE80- 0xFEFF return 0xF2 with the low 3 bits randomised. Replaced libc rand() with a local xorshift32 seeded at power-on, and added the PRNG state to the savestate via AddExState with key "WBKS" so save / load / rewind / netplay rollback all stay deterministic. In addition, two long-double-to-int truncations were changed to double for cross-platform FP determinism: - sound.c SetSoundVariables: soundtsinc - boards/n106.c DoNamcoSound: inc long double has platform-dependent precision (80-bit on x87, 64-bit with -mfpmath=sse, 128-bit on PowerPC), so the truncated integer result varied across these platforms. double is guaranteed 64-bit IEEE-754 portably. After this pass, the core has no time(), clock(), gettimeofday(), clock_gettime(), getpid(), getuid(), getgid(), getenv(), gethostid(), pthread, std::thread, OpenMP, signal handler, or non-deterministic- malloc dependency. Verified with a Python scanner that greps the source for these patterns; runs clean. The PPU / APU / CPU power-on already explicitly memset all state buffers to 0 (deterministic), and ROM/CHR-ROM allocation already memsets to 0xFF before partial fread (deterministic regardless of file truncation). Combined with the memory-safety hardening in 1185db8 (which prevents savestate-loaded indices from going out-of-bounds and producing unpredictable behaviour), the core now offers genuine frame-deterministic replay across runs, builds, and host endian.
2026-05-04 02:46:34 +02:00
FDSROM = (uint8_t*)FCEU_malloc(FDSROMSize);
FDS: Use retro_get_memory_data/size for disk saving
2019-01-04 17:13:09 +08:00
if (!FDSROM)
return (0);
for (x = 0; x < TotalSides; x++)
diskdata[x] = &FDSROM[x * BYTES_PER_SIDE];
md5_starts(&md5);
Initial commit - http://sourceforge.net/p/fceumm/code/160/
2014-03-30 22:15:17 +02:00
for (x = 0; x < TotalSides; x++) {
FCEU_fread(diskdata[x], 1, 65500, fp);
md5_update(&md5, diskdata[x], 65500);
}
md5_finish(&md5, GameInfo->MD5);
return(1);
}
static void PreSave(void) {
int x;
for (x = 0; x < TotalSides; x++) {
int b;
for (b = 0; b < 65500; b++)
diskdata[x][b] ^= diskdatao[x][b];
}
}
static void PostSave(void) {
int x;
for (x = 0; x < TotalSides; x++) {
int b;
for (b = 0; b < 65500; b++)
diskdata[x][b] ^= diskdatao[x][b];
}
}
int FDSLoad(const char *name, FCEUFILE *fp) {
FDS: do not use fopen/fclose/fread from bios load functions
2017-05-12 14:59:41 +08:00
FCEUFILE *zp;
Initial commit - http://sourceforge.net/p/fceumm/code/160/
2014-03-30 22:15:17 +02:00
int x;
char *fn = FCEU_MakeFName(FCEUMKF_FDSROM, 0, 0);
Replace direct file access with VFS routines
2021-10-20 16:31:45 +01:00
if (!(zp = FCEU_fopen(fn, NULL, 0))) {
FDS: do not use fopen/fclose/fread from bios load functions
2017-05-12 14:59:41 +08:00
FCEU_PrintError("FDS BIOS ROM image missing!\n");
Add support for emulated Game Genie add-on
2021-10-21 15:52:40 +01:00
FCEUD_DispMessage(RETRO_LOG_ERROR, 3000, "FDS BIOS image (disksys.rom) missing");
Initial commit - http://sourceforge.net/p/fceumm/code/160/
2014-03-30 22:15:17 +02:00
free(fn);
return 0;
}
free(fn);
FDS: Use retro_get_memory_data/size for disk saving
2019-01-04 17:13:09 +08:00
FreeFDSMemory();
Initial commit - http://sourceforge.net/p/fceumm/code/160/
2014-03-30 22:15:17 +02:00
FDS: Use retro_get_memory_data/size for disk saving
2019-01-04 17:13:09 +08:00
ResetCartMapping();
Initial commit - http://sourceforge.net/p/fceumm/code/160/
2014-03-30 22:15:17 +02:00
FDSBIOSsize = 8192;
core: stdint typedefs, LE optimizations, frame determinism Three follow-up audit passes on top of the memory-safety / leak / savestate-portability work in 1185db8. ============================================================== Pass 1: replace custom typedefs with C99 stdint types throughout ============================================================== The custom uint8 / uint16 / uint32 / uint64 / int8 / int16 / int32 / int64 typedefs in src/fceu-types.h were just simple aliases for the C99 stdint.h types. Replace them with the standard names directly. - 498 files modified - ~3,400 token replacements (uint8 -> uint8_t, etc) - fceu-types.h slimmed down to just INLINE / GINLINE / FASTAPASS macros and the readfunc / writefunc function-pointer typedefs (those now use uint8_t / uint32_t natively) - Build clean on `make platform=unix` with zero new warnings - Output binary size unchanged - confirming semantic equivalence Mechanical replacement done with a Python script that uses word- boundary regex to avoid false positives (e.g. 'uint32_t' was correctly left alone because '_' is a word character so 'uint32' is not a complete word inside it). ================================================================ Pass 2: prefer memcpy on LE hosts for endian read/write helpers ================================================================ fceu-endian.c's write32le_mem, FCEU_en32lsb, and FCEU_de32lsb performed bytewise composition/decomposition unconditionally. On LE hosts the in-memory representation already matches the desired LE on-disk format, so a single memcpy is equivalent and lets the compiler emit a single load/store rather than four byte ops. - The bytewise path is kept inside #ifdef MSB_FIRST for BE hosts where it implements the actual byte swap - Both forms produce identical results; this is a code-clarity change more than a performance one (the optimizer was already merging the shifts on LE), but it documents the intent and removes a strict-aliasing-flavoured cast through *(uint32_t*)Bufo - Added missing #include <string.h> in fceu-endian.c which was relying on transitive includes for memcpy Other MSB_FIRST sites in the codebase (state.c FlipByteOrder guards, ppu.c sprite-line rendering, boards/unrom512.c flash-write- counter access) were already optimized for LE; they were verified correct rather than changed. ================================================================ Pass 3: frame determinism for replay and netplay ================================================================ Two libc rand() sites in core were replaced with a local xorshift32 PRNG so that NES games which read uninitialised memory or hit hardware "weak bit" emulation produce reproducible behaviour across runs. NES titles routinely read uninitialised RAM (variables not zeroed before use, sprite Y-position set by junk-on-stack), so the RAM contents at power-on subtly affect game behaviour. With libc rand(), those contents depend on whether anyone else seeded rand() in the same process - a different libretro frontend, a different audio backend init order, or any frontend that does srand(time(0)) all break replay / netplay frame-determinism. 1. fceu.c FCEU_MemoryRand. Used to fill RAM (PowerNES) and CHR-RAM (iNES_Init) at power-on when option_ramstate=2 (random init). Replaced with a local xorshift32 PRNG, exposed via a new FCEU_MemoryRand_Reseed(uint32_t) function called once per power-on: - PowerNES seeds from the first 4 bytes of GameInfo->MD5 (set by all loaders before PowerNES runs) so identical ROMs produce identical RAM, different ROMs differ - iNES_Init seeds from iNESCart.PRGCRC32 before the CHR-RAM fill so two builds of the same ROM get the same CHR-RAM - The PRNG state advances across multiple FCEU_MemoryRand calls within one power-on so RAM and CHR-RAM get different content (matching NES hardware reality) 2. boards/rt-01.c UNLRT01Read. The RT-01 board has 'weak bit' protected EPROM regions; reads of 0xCE80-0xCEFF and 0xFE80- 0xFEFF return 0xF2 with the low 3 bits randomised. Replaced libc rand() with a local xorshift32 seeded at power-on, and added the PRNG state to the savestate via AddExState with key "WBKS" so save / load / rewind / netplay rollback all stay deterministic. In addition, two long-double-to-int truncations were changed to double for cross-platform FP determinism: - sound.c SetSoundVariables: soundtsinc - boards/n106.c DoNamcoSound: inc long double has platform-dependent precision (80-bit on x87, 64-bit with -mfpmath=sse, 128-bit on PowerPC), so the truncated integer result varied across these platforms. double is guaranteed 64-bit IEEE-754 portably. After this pass, the core has no time(), clock(), gettimeofday(), clock_gettime(), getpid(), getuid(), getgid(), getenv(), gethostid(), pthread, std::thread, OpenMP, signal handler, or non-deterministic- malloc dependency. Verified with a Python scanner that greps the source for these patterns; runs clean. The PPU / APU / CPU power-on already explicitly memset all state buffers to 0 (deterministic), and ROM/CHR-ROM allocation already memsets to 0xFF before partial fread (deterministic regardless of file truncation). Combined with the memory-safety hardening in 1185db8 (which prevents savestate-loaded indices from going out-of-bounds and producing unpredictable behaviour), the core now offers genuine frame-deterministic replay across runs, builds, and host endian.
2026-05-04 02:46:34 +02:00
FDSBIOS = (uint8_t*)FCEU_gmalloc(FDSBIOSsize);
Initial commit - http://sourceforge.net/p/fceumm/code/160/
2014-03-30 22:15:17 +02:00
SetupCartPRGMapping(0, FDSBIOS, FDSBIOSsize, 0);
FDS: do not use fopen/fclose/fread from bios load functions
2017-05-12 14:59:41 +08:00
if (FCEU_fread(FDSBIOS, 1, FDSBIOSsize, zp) != FDSBIOSsize) {
Initial commit - http://sourceforge.net/p/fceumm/code/160/
2014-03-30 22:15:17 +02:00
if (FDSBIOS)
free(FDSBIOS);
FDSBIOS = NULL;
FDS: do not use fopen/fclose/fread from bios load functions
2017-05-12 14:59:41 +08:00
FCEU_fclose(zp);
FCEU_PrintError("Error reading FDS BIOS ROM image.\n");
Add support for emulated Game Genie add-on
2021-10-21 15:52:40 +01:00
FCEUD_DispMessage(RETRO_LOG_ERROR, 3000, "Error reading FDS BIOS image (disksys.rom)");
Initial commit - http://sourceforge.net/p/fceumm/code/160/
2014-03-30 22:15:17 +02:00
return 0;
}
FDS: do not use fopen/fclose/fread from bios load functions
2017-05-12 14:59:41 +08:00
FCEU_fclose(zp);
Initial commit - http://sourceforge.net/p/fceumm/code/160/
2014-03-30 22:15:17 +02:00
FCEU_fseek(fp, 0, SEEK_SET);
if (!SubLoad(fp)) {
if (FDSBIOS)
free(FDSBIOS);
FDSBIOS = NULL;
return(0);
}
FDS: Use retro_get_memory_data/size for disk saving
2019-01-04 17:13:09 +08:00
for (x = 0; x < TotalSides; x++) {
core: stdint typedefs, LE optimizations, frame determinism Three follow-up audit passes on top of the memory-safety / leak / savestate-portability work in 1185db8. ============================================================== Pass 1: replace custom typedefs with C99 stdint types throughout ============================================================== The custom uint8 / uint16 / uint32 / uint64 / int8 / int16 / int32 / int64 typedefs in src/fceu-types.h were just simple aliases for the C99 stdint.h types. Replace them with the standard names directly. - 498 files modified - ~3,400 token replacements (uint8 -> uint8_t, etc) - fceu-types.h slimmed down to just INLINE / GINLINE / FASTAPASS macros and the readfunc / writefunc function-pointer typedefs (those now use uint8_t / uint32_t natively) - Build clean on `make platform=unix` with zero new warnings - Output binary size unchanged - confirming semantic equivalence Mechanical replacement done with a Python script that uses word- boundary regex to avoid false positives (e.g. 'uint32_t' was correctly left alone because '_' is a word character so 'uint32' is not a complete word inside it). ================================================================ Pass 2: prefer memcpy on LE hosts for endian read/write helpers ================================================================ fceu-endian.c's write32le_mem, FCEU_en32lsb, and FCEU_de32lsb performed bytewise composition/decomposition unconditionally. On LE hosts the in-memory representation already matches the desired LE on-disk format, so a single memcpy is equivalent and lets the compiler emit a single load/store rather than four byte ops. - The bytewise path is kept inside #ifdef MSB_FIRST for BE hosts where it implements the actual byte swap - Both forms produce identical results; this is a code-clarity change more than a performance one (the optimizer was already merging the shifts on LE), but it documents the intent and removes a strict-aliasing-flavoured cast through *(uint32_t*)Bufo - Added missing #include <string.h> in fceu-endian.c which was relying on transitive includes for memcpy Other MSB_FIRST sites in the codebase (state.c FlipByteOrder guards, ppu.c sprite-line rendering, boards/unrom512.c flash-write- counter access) were already optimized for LE; they were verified correct rather than changed. ================================================================ Pass 3: frame determinism for replay and netplay ================================================================ Two libc rand() sites in core were replaced with a local xorshift32 PRNG so that NES games which read uninitialised memory or hit hardware "weak bit" emulation produce reproducible behaviour across runs. NES titles routinely read uninitialised RAM (variables not zeroed before use, sprite Y-position set by junk-on-stack), so the RAM contents at power-on subtly affect game behaviour. With libc rand(), those contents depend on whether anyone else seeded rand() in the same process - a different libretro frontend, a different audio backend init order, or any frontend that does srand(time(0)) all break replay / netplay frame-determinism. 1. fceu.c FCEU_MemoryRand. Used to fill RAM (PowerNES) and CHR-RAM (iNES_Init) at power-on when option_ramstate=2 (random init). Replaced with a local xorshift32 PRNG, exposed via a new FCEU_MemoryRand_Reseed(uint32_t) function called once per power-on: - PowerNES seeds from the first 4 bytes of GameInfo->MD5 (set by all loaders before PowerNES runs) so identical ROMs produce identical RAM, different ROMs differ - iNES_Init seeds from iNESCart.PRGCRC32 before the CHR-RAM fill so two builds of the same ROM get the same CHR-RAM - The PRNG state advances across multiple FCEU_MemoryRand calls within one power-on so RAM and CHR-RAM get different content (matching NES hardware reality) 2. boards/rt-01.c UNLRT01Read. The RT-01 board has 'weak bit' protected EPROM regions; reads of 0xCE80-0xCEFF and 0xFE80- 0xFEFF return 0xF2 with the low 3 bits randomised. Replaced libc rand() with a local xorshift32 seeded at power-on, and added the PRNG state to the savestate via AddExState with key "WBKS" so save / load / rewind / netplay rollback all stay deterministic. In addition, two long-double-to-int truncations were changed to double for cross-platform FP determinism: - sound.c SetSoundVariables: soundtsinc - boards/n106.c DoNamcoSound: inc long double has platform-dependent precision (80-bit on x87, 64-bit with -mfpmath=sse, 128-bit on PowerPC), so the truncated integer result varied across these platforms. double is guaranteed 64-bit IEEE-754 portably. After this pass, the core has no time(), clock(), gettimeofday(), clock_gettime(), getpid(), getuid(), getgid(), getenv(), gethostid(), pthread, std::thread, OpenMP, signal handler, or non-deterministic- malloc dependency. Verified with a Python scanner that greps the source for these patterns; runs clean. The PPU / APU / CPU power-on already explicitly memset all state buffers to 0 (deterministic), and ROM/CHR-ROM allocation already memsets to 0xFF before partial fread (deterministic regardless of file truncation). Combined with the memory-safety hardening in 1185db8 (which prevents savestate-loaded indices from going out-of-bounds and producing unpredictable behaviour), the core now offers genuine frame-deterministic replay across runs, builds, and host endian.
2026-05-04 02:46:34 +02:00
diskdatao[x] = (uint8_t*)FCEU_malloc(65500);
core: memory-safety, leak, and savestate-portability audit fixes Squashed series of ~50 distinct bugs found during a multi-day security/correctness audit, ranging from ROM-triggerable heap corruption and savestate-triggered OOB read primitives down to obscure cross-platform savestate breakage. Build is clean on `make platform=unix` with zero new warnings. CRITICAL (ROM-triggerable, exploitable on the user's machine) ============================================================= * ines.c iNES 2.0 PRG/CHR exponent overflow leading to undersized allocation followed by heap buffer overflow on fread (pow() with attacker-chosen exponent up to 63). Cap exponent at 30 and compute size in uint32 with explicit cap below 2 GiB before storing in int. * ines.c miscROMSize int wraparound from attacker-controlled PRG/ CHR sizes; previous '& 0x8000000' check missed common underflow cases. Compute in int64 and reject <= 0 or > 128 MiB. * unif.c MAPR chunk OOB heap read on chunks of size < 4. Validate chunk size before allocating board-name buffer. * unif.c chunk size truncation: int conversion + missing cap allowed a 4 GiB chunk size to wrap. Cap at 16 MiB. * unif.c FixRomSize infinite loop on size > 0x80000000. Cap input. * unif.c DINF chunk: months[(m - 1) % 12] is UB for m==0; m comes from the .unf file. Clamp m into [1..12] before subtracting. * unif.c NAME chunk: GameInfo->name = malloc(...) was unchecked. * nsf.c size underflow: FCEU_fgetsize() - 0x80 wraps to ~UINT64_MAX on tiny files, propagating a huge size into uppow2/FCEU_malloc. Validate size > 0x80 before subtracting. * nsf.c NSFMaxBank * 4096 cap tightened from 1<<20 (UB on signed-int overflow) to 1<<19 (fits in int). * general.c uppow2 had signed-int UB (1 << 32) and wrapped to 0 for inputs near uint32 max. Cap at 0x80000000. * cheat.c SubCheats[256] BSS overflow when more than 256 GG/PAR substitution cheats are active. The array is adjacent to MMapPtrs[64] in BSS which is exposed via retro_get_memory_data, making this overflow visible from outside the core. * libretro.c retro_cheat_set strcpy stack overflow with arbitrary- length cheat strings from the frontend. Use strlcpy. CRITICAL (savestate-triggerable on a malicious .fcs file) ========================================================= * fds.c FDS savestate OOB read primitive: InDisk loaded from savestate is used as index into 8-element diskdata[] static pointer array. A value 8..254 dereferences arbitrary memory as a pointer (heap-read primitive). Also bounds-checked SelectDisk, mapperFDS_block, mapperFDS_blockstart, mapperFDS_diskaddr, mapperFDS_blocklen so blockstart+diskaddr stays within the 65500- byte disk buffer. * 11 mappers had savestate-loaded variables masked at write time but not at restore time, used as indices into fixed arrays: - 88.c, KS7037.c, 112.c cmd indexes reg[8] - sachen.c (S8259, S74LS374N) cmd indexes latch[8] - 357.c dipswitch indexes outer_bank[4] - unrom512.c flash_state indexes erase_a/d/b[5] - 368.c preg indexes banks[8] - 69.c sndcmd indexes sreg[14] - mmc2and4.c latch0/latch1 index creg[4] None individually exploitable into RCE - the array writes corrupt adjacent BSS with constrained data flowing in - but each is an out-of-bounds read or write from attacker-controllable input. HIGH (memory-safety, reachable on any load) =========================================== * fceu-memory.c FCEU_malloc deref-of-NULL on allocation failure ('ret = 0; memset(ret, 0, size);'). * file.c multiple FCEUFILE/MakeMemWrap/MakeMemWrapBuffer unchecked allocations and unchecked filestream_tell return. * libretro.c GameInfo NULL-derefs in three entry points (retro_set_controller_port_device, retro_get_memory_data, retro_get_memory_size) reachable on operations called before a successful load. * libretro.c framebuffer leak ~256 KB per failed FCEUI_LoadGame (the libretro frontend doesn't call retro_unload_game on a failed load). * libretro.c 3DS retro_deinit unchecked linearFree. * libretro.c stereo / NTSC filter unchecked malloc returns. * fceu.c FCEUI_LoadGame unchecked GameInfo malloc. * fds.c SubLoad and FDSLoad unchecked FCEU_malloc on diskdatao backup buffers (NULL-deref in subsequent memcpy). * fceu.c AllocGenieRW partial-failure leak: AReadG allocated but BWriteG malloc fails -> 256 KB leak per retry. * input/bworld.c Update(): unbounded strcpy from attacker-supplied data into 20-byte bdata. Replaced with length-bounded copy. * cart.c setprg2r/4r and setchr1r/2r/4r/8r mask-underflow guards. SetupCartPRGMapping/SetupCartCHRMapping compute (size >> N) - 1 which underflows to 0xFFFFFFFF for chips smaller than the unit. setprg8r/16r/32r and setchr8r already had defensive size checks; the smaller units did not. malee.c (2 KB chip) and mapper 218 (2 KB NTARAM-as-CHR) accidentally avoid OOB; the primitive is unsafe for any future board. * video.c FCEU_InitVirtualVideo XBuf/XDBuf partial-failure leak. * video.c, fceu.c FCEU_DispMessage / FCEU_printf / FCEU_PrintError: vsprintf into fixed stack buffers replaced with vsnprintf. * core: validate magic-string read length on FDS/UNIF/NSF load (reject files too short to contain the magic so previous static buffer / stack garbage doesn't spuriously match). LEAKS (every cart load/unload cycle) ==================================== * mmc5.c MMC5 cart-side: WRAM (up to 64K) + MMC5fill (1K) + ExRAM (1K) leaked per cycle. NSFMMC5_Close existed but was only used for NSF code path. * onebus.c Mapper 270/436: 8 KB CHRRAM leaked per cycle. * 330.c, 375.c, 528.c: 8 KB WRAM each, no Close at all. CORRECTNESS (UB / ABI / endianness) =================================== * fceu-endian.c FlipByteOrder loop bound was 'count' instead of 'count/2', making it a no-op for every even count and silently breaking savestate portability for every FCEUSTATE_RLSB-marked field. The '#ifndef GEKKO' workarounds scattered through the codebase (sound.c, vrc7.c, vrc6.c) were symptoms of this root cause. * state.c AddExState bounds check ran AFTER writing the entry, so when SFEXINDEX hit the 63-entry cap the next call would write the entry then immediately overwrite it with the terminator. * state.c FCEUSS_Save_Mem: post-save callback was guarded by 'if (SPreSave)' instead of 'if (SPostSave)'. * Sequence-point UB in counter expressions across 4 board files, e.g. 'x = !x ? a : --x' and 'x = ++x % n'. GCC -Wsequence-point flags all five sites (285.c, 413.c, asic_mmc3.c, jyasic.c). * SFORMAT size mismatches between declared variable type and AddExState size argument: - asic_vrc3.c VRC3_count/VRC3_reload (uint16, saved as 1 byte) - mmc3.c m555_count (uint32, saved as 2 bytes), m555_count_- expired (uint8, saved as 2 bytes - OOB write into adjacent BSS). * unrom512.c UNROM-512 mapper 30 .srm save format: host-endian uint32 flash write counters. Now stored as LE on disk regardless of host (Battle Kid 2, Twin Dragons, Lizard, Sole, etc.). * ~70 multi-byte single-variable SFORMAT/AddExState entries across 36 board files were saved as host-byte-order. After the FlipByteOrder fix, these are now byte-swapped on BE so cross- platform savestates round-trip correctly. * coolgirl.c new ExStateLE() macro added; 22 multi-byte sites. * pic16c5x.c 11 multi-byte AddExState calls (m_PC, m_PREVPC, m_CONFIG, m_WDT, m_prescaler, m_opcode, m_STACK[0/1], m_icount, m_delay_timer, m_rtcc, m_inst_cycles, m_clock2cycle). * onebus.c PowerJoy Supermax submapper detection used non-portable *(uint32*)&info->MD5 cast that read different bytes on LE vs BE. * fds.c clean up redundant FCEUSTATE_RLSB encoding in AddExState calls that also passed type=1 (idempotent OR; readability fix). DOCUMENTATION ============= * input.c UpdateGP's *(uint32*)data cast looks like a typical endian bug but is actually correct (the libretro frontend builds JSReturn with matching host-uint32 shifts). Comment added to prevent future "fixes" from breaking it. Limitations not addressed ========================= * Element-stride-aware byte swapping. The savestate byte-swap mechanism (FlipByteOrder over the entire SFORMAT entry buffer) is structurally wrong for arrays of multi-byte values: it reverses the whole buffer end-to-end instead of byte-swapping each element. Several places that need cross-platform-portable arrays (VRC7 sound state, jyasic chr[8]) work around this by either splitting arrays into per-element SFORMAT entries (n106 PlayIndex, bandai reg) or by skipping save entirely on BE via #ifndef GEKKO. A proper fix would extend the size encoding with an element-stride field. Left for a future change because it would change the savestate format. * Strict-aliasing UB in ppu.c (around 9 sites doing *(uint32*)uint8_buf for fast 4-byte writes via FCEU_dwmemset). Works in practice with all common compilers because the patterns are byte-symmetric, but is formally UB. * FCEU_gmalloc calls exit(1) on OOM. A libretro core should never exit() because that takes down the whole frontend. Used by 100+ call sites; refactoring to return-NULL is out of scope here. Testing ======= * Build: clean on `make platform=unix` with -O2; no new warnings. * FlipByteOrder fix verified by hand-trace and a standalone unit test for counts 2, 4, 8. * uppow2 fix verified by unit test across 13 boundary cases. * SFORMAT size mismatches and missing-RLSB cases identified by Python static-analysis scripts that cross-reference SFORMAT entries against variable declarations. * iNES 2.0 exponent fix verified by hand-tracing what byte 0xFF produces post-fix: exp=30 (capped), mult=7, size=3 GiB nominal, capped to 1 GiB, capped to 2 GiB by uppow2, FCEU_malloc returns NULL on most systems, loader returns 0. No heap overflow for any input byte. * Savestate-loaded array index audit: built a Python scanner that extracts each AddExState/SFORMAT entry and cross-references the variable name against array-index uses in the same file. All flagged sites covered. * A libFuzzer harness and seed corpus generator (fuzz_main.c, gen_seed_corpus.py) accompany this submission for ongoing regression testing.
2026-05-04 02:15:40 +02:00
if (!diskdatao[x]) {
int y;
for (y = 0; y < x; y++) {
free(diskdatao[y]);
diskdatao[y] = NULL;
}
if (FDSBIOS)
free(FDSBIOS);
FDSBIOS = NULL;
free(FDSROM);
FDSROM = NULL;
return(0);
}
FDS: Use retro_get_memory_data/size for disk saving
2019-01-04 17:13:09 +08:00
memcpy(diskdatao[x], diskdata[x], 65500);
}
FDS: Fix pointers are not freed when closing rom
2019-01-06 10:42:12 +08:00
DiskWritten = 1;
FDS: Use retro_get_memory_data/size for disk saving
2019-01-04 17:13:09 +08:00
Initial commit - http://sourceforge.net/p/fceumm/code/160/
2014-03-30 22:15:17 +02:00
GameInfo->type = GIT_FDS;
GameInterface = FDSGI;
SelectDisk = 0;
InDisk = 255;
ResetExState(PreSave, PostSave);
FDSSoundStateAdd();
for (x = 0; x < TotalSides; x++) {
char temp[5];
sprintf(temp, "DDT%d", x);
AddExState(diskdata[x], 65500, 0, temp);
}
FDS: Use retro_get_memory_data/size for disk saving
2019-01-04 17:13:09 +08:00
AddExState(&FDSRegs[0], 1, 0, "REG1");
AddExState(&FDSRegs[1], 1, 0, "REG2");
AddExState(&FDSRegs[2], 1, 0, "REG3");
AddExState(&FDSRegs[3], 1, 0, "REG4");
AddExState(&FDSRegs[4], 1, 0, "REG5");
AddExState(&FDSRegs[5], 1, 0, "REG6");
core: memory-safety, leak, and savestate-portability audit fixes Squashed series of ~50 distinct bugs found during a multi-day security/correctness audit, ranging from ROM-triggerable heap corruption and savestate-triggered OOB read primitives down to obscure cross-platform savestate breakage. Build is clean on `make platform=unix` with zero new warnings. CRITICAL (ROM-triggerable, exploitable on the user's machine) ============================================================= * ines.c iNES 2.0 PRG/CHR exponent overflow leading to undersized allocation followed by heap buffer overflow on fread (pow() with attacker-chosen exponent up to 63). Cap exponent at 30 and compute size in uint32 with explicit cap below 2 GiB before storing in int. * ines.c miscROMSize int wraparound from attacker-controlled PRG/ CHR sizes; previous '& 0x8000000' check missed common underflow cases. Compute in int64 and reject <= 0 or > 128 MiB. * unif.c MAPR chunk OOB heap read on chunks of size < 4. Validate chunk size before allocating board-name buffer. * unif.c chunk size truncation: int conversion + missing cap allowed a 4 GiB chunk size to wrap. Cap at 16 MiB. * unif.c FixRomSize infinite loop on size > 0x80000000. Cap input. * unif.c DINF chunk: months[(m - 1) % 12] is UB for m==0; m comes from the .unf file. Clamp m into [1..12] before subtracting. * unif.c NAME chunk: GameInfo->name = malloc(...) was unchecked. * nsf.c size underflow: FCEU_fgetsize() - 0x80 wraps to ~UINT64_MAX on tiny files, propagating a huge size into uppow2/FCEU_malloc. Validate size > 0x80 before subtracting. * nsf.c NSFMaxBank * 4096 cap tightened from 1<<20 (UB on signed-int overflow) to 1<<19 (fits in int). * general.c uppow2 had signed-int UB (1 << 32) and wrapped to 0 for inputs near uint32 max. Cap at 0x80000000. * cheat.c SubCheats[256] BSS overflow when more than 256 GG/PAR substitution cheats are active. The array is adjacent to MMapPtrs[64] in BSS which is exposed via retro_get_memory_data, making this overflow visible from outside the core. * libretro.c retro_cheat_set strcpy stack overflow with arbitrary- length cheat strings from the frontend. Use strlcpy. CRITICAL (savestate-triggerable on a malicious .fcs file) ========================================================= * fds.c FDS savestate OOB read primitive: InDisk loaded from savestate is used as index into 8-element diskdata[] static pointer array. A value 8..254 dereferences arbitrary memory as a pointer (heap-read primitive). Also bounds-checked SelectDisk, mapperFDS_block, mapperFDS_blockstart, mapperFDS_diskaddr, mapperFDS_blocklen so blockstart+diskaddr stays within the 65500- byte disk buffer. * 11 mappers had savestate-loaded variables masked at write time but not at restore time, used as indices into fixed arrays: - 88.c, KS7037.c, 112.c cmd indexes reg[8] - sachen.c (S8259, S74LS374N) cmd indexes latch[8] - 357.c dipswitch indexes outer_bank[4] - unrom512.c flash_state indexes erase_a/d/b[5] - 368.c preg indexes banks[8] - 69.c sndcmd indexes sreg[14] - mmc2and4.c latch0/latch1 index creg[4] None individually exploitable into RCE - the array writes corrupt adjacent BSS with constrained data flowing in - but each is an out-of-bounds read or write from attacker-controllable input. HIGH (memory-safety, reachable on any load) =========================================== * fceu-memory.c FCEU_malloc deref-of-NULL on allocation failure ('ret = 0; memset(ret, 0, size);'). * file.c multiple FCEUFILE/MakeMemWrap/MakeMemWrapBuffer unchecked allocations and unchecked filestream_tell return. * libretro.c GameInfo NULL-derefs in three entry points (retro_set_controller_port_device, retro_get_memory_data, retro_get_memory_size) reachable on operations called before a successful load. * libretro.c framebuffer leak ~256 KB per failed FCEUI_LoadGame (the libretro frontend doesn't call retro_unload_game on a failed load). * libretro.c 3DS retro_deinit unchecked linearFree. * libretro.c stereo / NTSC filter unchecked malloc returns. * fceu.c FCEUI_LoadGame unchecked GameInfo malloc. * fds.c SubLoad and FDSLoad unchecked FCEU_malloc on diskdatao backup buffers (NULL-deref in subsequent memcpy). * fceu.c AllocGenieRW partial-failure leak: AReadG allocated but BWriteG malloc fails -> 256 KB leak per retry. * input/bworld.c Update(): unbounded strcpy from attacker-supplied data into 20-byte bdata. Replaced with length-bounded copy. * cart.c setprg2r/4r and setchr1r/2r/4r/8r mask-underflow guards. SetupCartPRGMapping/SetupCartCHRMapping compute (size >> N) - 1 which underflows to 0xFFFFFFFF for chips smaller than the unit. setprg8r/16r/32r and setchr8r already had defensive size checks; the smaller units did not. malee.c (2 KB chip) and mapper 218 (2 KB NTARAM-as-CHR) accidentally avoid OOB; the primitive is unsafe for any future board. * video.c FCEU_InitVirtualVideo XBuf/XDBuf partial-failure leak. * video.c, fceu.c FCEU_DispMessage / FCEU_printf / FCEU_PrintError: vsprintf into fixed stack buffers replaced with vsnprintf. * core: validate magic-string read length on FDS/UNIF/NSF load (reject files too short to contain the magic so previous static buffer / stack garbage doesn't spuriously match). LEAKS (every cart load/unload cycle) ==================================== * mmc5.c MMC5 cart-side: WRAM (up to 64K) + MMC5fill (1K) + ExRAM (1K) leaked per cycle. NSFMMC5_Close existed but was only used for NSF code path. * onebus.c Mapper 270/436: 8 KB CHRRAM leaked per cycle. * 330.c, 375.c, 528.c: 8 KB WRAM each, no Close at all. CORRECTNESS (UB / ABI / endianness) =================================== * fceu-endian.c FlipByteOrder loop bound was 'count' instead of 'count/2', making it a no-op for every even count and silently breaking savestate portability for every FCEUSTATE_RLSB-marked field. The '#ifndef GEKKO' workarounds scattered through the codebase (sound.c, vrc7.c, vrc6.c) were symptoms of this root cause. * state.c AddExState bounds check ran AFTER writing the entry, so when SFEXINDEX hit the 63-entry cap the next call would write the entry then immediately overwrite it with the terminator. * state.c FCEUSS_Save_Mem: post-save callback was guarded by 'if (SPreSave)' instead of 'if (SPostSave)'. * Sequence-point UB in counter expressions across 4 board files, e.g. 'x = !x ? a : --x' and 'x = ++x % n'. GCC -Wsequence-point flags all five sites (285.c, 413.c, asic_mmc3.c, jyasic.c). * SFORMAT size mismatches between declared variable type and AddExState size argument: - asic_vrc3.c VRC3_count/VRC3_reload (uint16, saved as 1 byte) - mmc3.c m555_count (uint32, saved as 2 bytes), m555_count_- expired (uint8, saved as 2 bytes - OOB write into adjacent BSS). * unrom512.c UNROM-512 mapper 30 .srm save format: host-endian uint32 flash write counters. Now stored as LE on disk regardless of host (Battle Kid 2, Twin Dragons, Lizard, Sole, etc.). * ~70 multi-byte single-variable SFORMAT/AddExState entries across 36 board files were saved as host-byte-order. After the FlipByteOrder fix, these are now byte-swapped on BE so cross- platform savestates round-trip correctly. * coolgirl.c new ExStateLE() macro added; 22 multi-byte sites. * pic16c5x.c 11 multi-byte AddExState calls (m_PC, m_PREVPC, m_CONFIG, m_WDT, m_prescaler, m_opcode, m_STACK[0/1], m_icount, m_delay_timer, m_rtcc, m_inst_cycles, m_clock2cycle). * onebus.c PowerJoy Supermax submapper detection used non-portable *(uint32*)&info->MD5 cast that read different bytes on LE vs BE. * fds.c clean up redundant FCEUSTATE_RLSB encoding in AddExState calls that also passed type=1 (idempotent OR; readability fix). DOCUMENTATION ============= * input.c UpdateGP's *(uint32*)data cast looks like a typical endian bug but is actually correct (the libretro frontend builds JSReturn with matching host-uint32 shifts). Comment added to prevent future "fixes" from breaking it. Limitations not addressed ========================= * Element-stride-aware byte swapping. The savestate byte-swap mechanism (FlipByteOrder over the entire SFORMAT entry buffer) is structurally wrong for arrays of multi-byte values: it reverses the whole buffer end-to-end instead of byte-swapping each element. Several places that need cross-platform-portable arrays (VRC7 sound state, jyasic chr[8]) work around this by either splitting arrays into per-element SFORMAT entries (n106 PlayIndex, bandai reg) or by skipping save entirely on BE via #ifndef GEKKO. A proper fix would extend the size encoding with an element-stride field. Left for a future change because it would change the savestate format. * Strict-aliasing UB in ppu.c (around 9 sites doing *(uint32*)uint8_buf for fast 4-byte writes via FCEU_dwmemset). Works in practice with all common compilers because the patterns are byte-symmetric, but is formally UB. * FCEU_gmalloc calls exit(1) on OOM. A libretro core should never exit() because that takes down the whole frontend. Used by 100+ call sites; refactoring to return-NULL is out of scope here. Testing ======= * Build: clean on `make platform=unix` with -O2; no new warnings. * FlipByteOrder fix verified by hand-trace and a standalone unit test for counts 2, 4, 8. * uppow2 fix verified by unit test across 13 boundary cases. * SFORMAT size mismatches and missing-RLSB cases identified by Python static-analysis scripts that cross-reference SFORMAT entries against variable declarations. * iNES 2.0 exponent fix verified by hand-tracing what byte 0xFF produces post-fix: exp=30 (capped), mult=7, size=3 GiB nominal, capped to 1 GiB, capped to 2 GiB by uppow2, FCEU_malloc returns NULL on most systems, loader returns 0. No heap overflow for any input byte. * Savestate-loaded array index audit: built a Python scanner that extracts each AddExState/SFORMAT entry and cross-references the variable name against array-index uses in the same file. All flagged sites covered. * A libFuzzer harness and seed corpus generator (fuzz_main.c, gen_seed_corpus.py) accompany this submission for ongoing regression testing.
2026-05-04 02:15:40 +02:00
AddExState(&IRQCount, 4, 1, "IRQC");
AddExState(&IRQLatch, 4, 1, "IQL1");
Initial commit - http://sourceforge.net/p/fceumm/code/160/
2014-03-30 22:15:17 +02:00
AddExState(&IRQa, 1, 0, "IRQA");
AddExState(&writeskip, 1, 0, "WSKI");
core: memory-safety, leak, and savestate-portability audit fixes Squashed series of ~50 distinct bugs found during a multi-day security/correctness audit, ranging from ROM-triggerable heap corruption and savestate-triggered OOB read primitives down to obscure cross-platform savestate breakage. Build is clean on `make platform=unix` with zero new warnings. CRITICAL (ROM-triggerable, exploitable on the user's machine) ============================================================= * ines.c iNES 2.0 PRG/CHR exponent overflow leading to undersized allocation followed by heap buffer overflow on fread (pow() with attacker-chosen exponent up to 63). Cap exponent at 30 and compute size in uint32 with explicit cap below 2 GiB before storing in int. * ines.c miscROMSize int wraparound from attacker-controlled PRG/ CHR sizes; previous '& 0x8000000' check missed common underflow cases. Compute in int64 and reject <= 0 or > 128 MiB. * unif.c MAPR chunk OOB heap read on chunks of size < 4. Validate chunk size before allocating board-name buffer. * unif.c chunk size truncation: int conversion + missing cap allowed a 4 GiB chunk size to wrap. Cap at 16 MiB. * unif.c FixRomSize infinite loop on size > 0x80000000. Cap input. * unif.c DINF chunk: months[(m - 1) % 12] is UB for m==0; m comes from the .unf file. Clamp m into [1..12] before subtracting. * unif.c NAME chunk: GameInfo->name = malloc(...) was unchecked. * nsf.c size underflow: FCEU_fgetsize() - 0x80 wraps to ~UINT64_MAX on tiny files, propagating a huge size into uppow2/FCEU_malloc. Validate size > 0x80 before subtracting. * nsf.c NSFMaxBank * 4096 cap tightened from 1<<20 (UB on signed-int overflow) to 1<<19 (fits in int). * general.c uppow2 had signed-int UB (1 << 32) and wrapped to 0 for inputs near uint32 max. Cap at 0x80000000. * cheat.c SubCheats[256] BSS overflow when more than 256 GG/PAR substitution cheats are active. The array is adjacent to MMapPtrs[64] in BSS which is exposed via retro_get_memory_data, making this overflow visible from outside the core. * libretro.c retro_cheat_set strcpy stack overflow with arbitrary- length cheat strings from the frontend. Use strlcpy. CRITICAL (savestate-triggerable on a malicious .fcs file) ========================================================= * fds.c FDS savestate OOB read primitive: InDisk loaded from savestate is used as index into 8-element diskdata[] static pointer array. A value 8..254 dereferences arbitrary memory as a pointer (heap-read primitive). Also bounds-checked SelectDisk, mapperFDS_block, mapperFDS_blockstart, mapperFDS_diskaddr, mapperFDS_blocklen so blockstart+diskaddr stays within the 65500- byte disk buffer. * 11 mappers had savestate-loaded variables masked at write time but not at restore time, used as indices into fixed arrays: - 88.c, KS7037.c, 112.c cmd indexes reg[8] - sachen.c (S8259, S74LS374N) cmd indexes latch[8] - 357.c dipswitch indexes outer_bank[4] - unrom512.c flash_state indexes erase_a/d/b[5] - 368.c preg indexes banks[8] - 69.c sndcmd indexes sreg[14] - mmc2and4.c latch0/latch1 index creg[4] None individually exploitable into RCE - the array writes corrupt adjacent BSS with constrained data flowing in - but each is an out-of-bounds read or write from attacker-controllable input. HIGH (memory-safety, reachable on any load) =========================================== * fceu-memory.c FCEU_malloc deref-of-NULL on allocation failure ('ret = 0; memset(ret, 0, size);'). * file.c multiple FCEUFILE/MakeMemWrap/MakeMemWrapBuffer unchecked allocations and unchecked filestream_tell return. * libretro.c GameInfo NULL-derefs in three entry points (retro_set_controller_port_device, retro_get_memory_data, retro_get_memory_size) reachable on operations called before a successful load. * libretro.c framebuffer leak ~256 KB per failed FCEUI_LoadGame (the libretro frontend doesn't call retro_unload_game on a failed load). * libretro.c 3DS retro_deinit unchecked linearFree. * libretro.c stereo / NTSC filter unchecked malloc returns. * fceu.c FCEUI_LoadGame unchecked GameInfo malloc. * fds.c SubLoad and FDSLoad unchecked FCEU_malloc on diskdatao backup buffers (NULL-deref in subsequent memcpy). * fceu.c AllocGenieRW partial-failure leak: AReadG allocated but BWriteG malloc fails -> 256 KB leak per retry. * input/bworld.c Update(): unbounded strcpy from attacker-supplied data into 20-byte bdata. Replaced with length-bounded copy. * cart.c setprg2r/4r and setchr1r/2r/4r/8r mask-underflow guards. SetupCartPRGMapping/SetupCartCHRMapping compute (size >> N) - 1 which underflows to 0xFFFFFFFF for chips smaller than the unit. setprg8r/16r/32r and setchr8r already had defensive size checks; the smaller units did not. malee.c (2 KB chip) and mapper 218 (2 KB NTARAM-as-CHR) accidentally avoid OOB; the primitive is unsafe for any future board. * video.c FCEU_InitVirtualVideo XBuf/XDBuf partial-failure leak. * video.c, fceu.c FCEU_DispMessage / FCEU_printf / FCEU_PrintError: vsprintf into fixed stack buffers replaced with vsnprintf. * core: validate magic-string read length on FDS/UNIF/NSF load (reject files too short to contain the magic so previous static buffer / stack garbage doesn't spuriously match). LEAKS (every cart load/unload cycle) ==================================== * mmc5.c MMC5 cart-side: WRAM (up to 64K) + MMC5fill (1K) + ExRAM (1K) leaked per cycle. NSFMMC5_Close existed but was only used for NSF code path. * onebus.c Mapper 270/436: 8 KB CHRRAM leaked per cycle. * 330.c, 375.c, 528.c: 8 KB WRAM each, no Close at all. CORRECTNESS (UB / ABI / endianness) =================================== * fceu-endian.c FlipByteOrder loop bound was 'count' instead of 'count/2', making it a no-op for every even count and silently breaking savestate portability for every FCEUSTATE_RLSB-marked field. The '#ifndef GEKKO' workarounds scattered through the codebase (sound.c, vrc7.c, vrc6.c) were symptoms of this root cause. * state.c AddExState bounds check ran AFTER writing the entry, so when SFEXINDEX hit the 63-entry cap the next call would write the entry then immediately overwrite it with the terminator. * state.c FCEUSS_Save_Mem: post-save callback was guarded by 'if (SPreSave)' instead of 'if (SPostSave)'. * Sequence-point UB in counter expressions across 4 board files, e.g. 'x = !x ? a : --x' and 'x = ++x % n'. GCC -Wsequence-point flags all five sites (285.c, 413.c, asic_mmc3.c, jyasic.c). * SFORMAT size mismatches between declared variable type and AddExState size argument: - asic_vrc3.c VRC3_count/VRC3_reload (uint16, saved as 1 byte) - mmc3.c m555_count (uint32, saved as 2 bytes), m555_count_- expired (uint8, saved as 2 bytes - OOB write into adjacent BSS). * unrom512.c UNROM-512 mapper 30 .srm save format: host-endian uint32 flash write counters. Now stored as LE on disk regardless of host (Battle Kid 2, Twin Dragons, Lizard, Sole, etc.). * ~70 multi-byte single-variable SFORMAT/AddExState entries across 36 board files were saved as host-byte-order. After the FlipByteOrder fix, these are now byte-swapped on BE so cross- platform savestates round-trip correctly. * coolgirl.c new ExStateLE() macro added; 22 multi-byte sites. * pic16c5x.c 11 multi-byte AddExState calls (m_PC, m_PREVPC, m_CONFIG, m_WDT, m_prescaler, m_opcode, m_STACK[0/1], m_icount, m_delay_timer, m_rtcc, m_inst_cycles, m_clock2cycle). * onebus.c PowerJoy Supermax submapper detection used non-portable *(uint32*)&info->MD5 cast that read different bytes on LE vs BE. * fds.c clean up redundant FCEUSTATE_RLSB encoding in AddExState calls that also passed type=1 (idempotent OR; readability fix). DOCUMENTATION ============= * input.c UpdateGP's *(uint32*)data cast looks like a typical endian bug but is actually correct (the libretro frontend builds JSReturn with matching host-uint32 shifts). Comment added to prevent future "fixes" from breaking it. Limitations not addressed ========================= * Element-stride-aware byte swapping. The savestate byte-swap mechanism (FlipByteOrder over the entire SFORMAT entry buffer) is structurally wrong for arrays of multi-byte values: it reverses the whole buffer end-to-end instead of byte-swapping each element. Several places that need cross-platform-portable arrays (VRC7 sound state, jyasic chr[8]) work around this by either splitting arrays into per-element SFORMAT entries (n106 PlayIndex, bandai reg) or by skipping save entirely on BE via #ifndef GEKKO. A proper fix would extend the size encoding with an element-stride field. Left for a future change because it would change the savestate format. * Strict-aliasing UB in ppu.c (around 9 sites doing *(uint32*)uint8_buf for fast 4-byte writes via FCEU_dwmemset). Works in practice with all common compilers because the patterns are byte-symmetric, but is formally UB. * FCEU_gmalloc calls exit(1) on OOM. A libretro core should never exit() because that takes down the whole frontend. Used by 100+ call sites; refactoring to return-NULL is out of scope here. Testing ======= * Build: clean on `make platform=unix` with -O2; no new warnings. * FlipByteOrder fix verified by hand-trace and a standalone unit test for counts 2, 4, 8. * uppow2 fix verified by unit test across 13 boundary cases. * SFORMAT size mismatches and missing-RLSB cases identified by Python static-analysis scripts that cross-reference SFORMAT entries against variable declarations. * iNES 2.0 exponent fix verified by hand-tracing what byte 0xFF produces post-fix: exp=30 (capped), mult=7, size=3 GiB nominal, capped to 1 GiB, capped to 2 GiB by uppow2, FCEU_malloc returns NULL on most systems, loader returns 0. No heap overflow for any input byte. * Savestate-loaded array index audit: built a Python scanner that extracts each AddExState/SFORMAT entry and cross-references the variable name against array-index uses in the same file. All flagged sites covered. * A libFuzzer harness and seed corpus generator (fuzz_main.c, gen_seed_corpus.py) accompany this submission for ongoing regression testing.
2026-05-04 02:15:40 +02:00
AddExState(&DiskPtr, 4, 1, "DPTR");
AddExState(&DiskSeekIRQ, 4, 1, "DSIR");
Initial commit - http://sourceforge.net/p/fceumm/code/160/
2014-03-30 22:15:17 +02:00
AddExState(&SelectDisk, 1, 0, "SELD");
AddExState(&InDisk, 1, 0, "INDI");
AddExState(&DiskWritten, 1, 0, "DSKW");
FDS: Permanently use new disk handling functions FDS was already using the new disk handling code. This PR just removes the old and now unused one.
2022-06-07 18:28:07 +08:00
(fds.c) backport new fds disk handling backported from fceux, commit by dinkc64 (fbneo) https://github.com/TASVideos/fceux/commit/0b4be4b639371204642f22def7bd92ea18a4c2f3 Fixes saving issue with some games (Bubble Bobble, Super Lode Runner II, ...)
2020-01-25 08:32:43 +08:00
AddExState(&mapperFDS_control, 1, 0, "CTRG");
core: memory-safety, leak, and savestate-portability audit fixes Squashed series of ~50 distinct bugs found during a multi-day security/correctness audit, ranging from ROM-triggerable heap corruption and savestate-triggered OOB read primitives down to obscure cross-platform savestate breakage. Build is clean on `make platform=unix` with zero new warnings. CRITICAL (ROM-triggerable, exploitable on the user's machine) ============================================================= * ines.c iNES 2.0 PRG/CHR exponent overflow leading to undersized allocation followed by heap buffer overflow on fread (pow() with attacker-chosen exponent up to 63). Cap exponent at 30 and compute size in uint32 with explicit cap below 2 GiB before storing in int. * ines.c miscROMSize int wraparound from attacker-controlled PRG/ CHR sizes; previous '& 0x8000000' check missed common underflow cases. Compute in int64 and reject <= 0 or > 128 MiB. * unif.c MAPR chunk OOB heap read on chunks of size < 4. Validate chunk size before allocating board-name buffer. * unif.c chunk size truncation: int conversion + missing cap allowed a 4 GiB chunk size to wrap. Cap at 16 MiB. * unif.c FixRomSize infinite loop on size > 0x80000000. Cap input. * unif.c DINF chunk: months[(m - 1) % 12] is UB for m==0; m comes from the .unf file. Clamp m into [1..12] before subtracting. * unif.c NAME chunk: GameInfo->name = malloc(...) was unchecked. * nsf.c size underflow: FCEU_fgetsize() - 0x80 wraps to ~UINT64_MAX on tiny files, propagating a huge size into uppow2/FCEU_malloc. Validate size > 0x80 before subtracting. * nsf.c NSFMaxBank * 4096 cap tightened from 1<<20 (UB on signed-int overflow) to 1<<19 (fits in int). * general.c uppow2 had signed-int UB (1 << 32) and wrapped to 0 for inputs near uint32 max. Cap at 0x80000000. * cheat.c SubCheats[256] BSS overflow when more than 256 GG/PAR substitution cheats are active. The array is adjacent to MMapPtrs[64] in BSS which is exposed via retro_get_memory_data, making this overflow visible from outside the core. * libretro.c retro_cheat_set strcpy stack overflow with arbitrary- length cheat strings from the frontend. Use strlcpy. CRITICAL (savestate-triggerable on a malicious .fcs file) ========================================================= * fds.c FDS savestate OOB read primitive: InDisk loaded from savestate is used as index into 8-element diskdata[] static pointer array. A value 8..254 dereferences arbitrary memory as a pointer (heap-read primitive). Also bounds-checked SelectDisk, mapperFDS_block, mapperFDS_blockstart, mapperFDS_diskaddr, mapperFDS_blocklen so blockstart+diskaddr stays within the 65500- byte disk buffer. * 11 mappers had savestate-loaded variables masked at write time but not at restore time, used as indices into fixed arrays: - 88.c, KS7037.c, 112.c cmd indexes reg[8] - sachen.c (S8259, S74LS374N) cmd indexes latch[8] - 357.c dipswitch indexes outer_bank[4] - unrom512.c flash_state indexes erase_a/d/b[5] - 368.c preg indexes banks[8] - 69.c sndcmd indexes sreg[14] - mmc2and4.c latch0/latch1 index creg[4] None individually exploitable into RCE - the array writes corrupt adjacent BSS with constrained data flowing in - but each is an out-of-bounds read or write from attacker-controllable input. HIGH (memory-safety, reachable on any load) =========================================== * fceu-memory.c FCEU_malloc deref-of-NULL on allocation failure ('ret = 0; memset(ret, 0, size);'). * file.c multiple FCEUFILE/MakeMemWrap/MakeMemWrapBuffer unchecked allocations and unchecked filestream_tell return. * libretro.c GameInfo NULL-derefs in three entry points (retro_set_controller_port_device, retro_get_memory_data, retro_get_memory_size) reachable on operations called before a successful load. * libretro.c framebuffer leak ~256 KB per failed FCEUI_LoadGame (the libretro frontend doesn't call retro_unload_game on a failed load). * libretro.c 3DS retro_deinit unchecked linearFree. * libretro.c stereo / NTSC filter unchecked malloc returns. * fceu.c FCEUI_LoadGame unchecked GameInfo malloc. * fds.c SubLoad and FDSLoad unchecked FCEU_malloc on diskdatao backup buffers (NULL-deref in subsequent memcpy). * fceu.c AllocGenieRW partial-failure leak: AReadG allocated but BWriteG malloc fails -> 256 KB leak per retry. * input/bworld.c Update(): unbounded strcpy from attacker-supplied data into 20-byte bdata. Replaced with length-bounded copy. * cart.c setprg2r/4r and setchr1r/2r/4r/8r mask-underflow guards. SetupCartPRGMapping/SetupCartCHRMapping compute (size >> N) - 1 which underflows to 0xFFFFFFFF for chips smaller than the unit. setprg8r/16r/32r and setchr8r already had defensive size checks; the smaller units did not. malee.c (2 KB chip) and mapper 218 (2 KB NTARAM-as-CHR) accidentally avoid OOB; the primitive is unsafe for any future board. * video.c FCEU_InitVirtualVideo XBuf/XDBuf partial-failure leak. * video.c, fceu.c FCEU_DispMessage / FCEU_printf / FCEU_PrintError: vsprintf into fixed stack buffers replaced with vsnprintf. * core: validate magic-string read length on FDS/UNIF/NSF load (reject files too short to contain the magic so previous static buffer / stack garbage doesn't spuriously match). LEAKS (every cart load/unload cycle) ==================================== * mmc5.c MMC5 cart-side: WRAM (up to 64K) + MMC5fill (1K) + ExRAM (1K) leaked per cycle. NSFMMC5_Close existed but was only used for NSF code path. * onebus.c Mapper 270/436: 8 KB CHRRAM leaked per cycle. * 330.c, 375.c, 528.c: 8 KB WRAM each, no Close at all. CORRECTNESS (UB / ABI / endianness) =================================== * fceu-endian.c FlipByteOrder loop bound was 'count' instead of 'count/2', making it a no-op for every even count and silently breaking savestate portability for every FCEUSTATE_RLSB-marked field. The '#ifndef GEKKO' workarounds scattered through the codebase (sound.c, vrc7.c, vrc6.c) were symptoms of this root cause. * state.c AddExState bounds check ran AFTER writing the entry, so when SFEXINDEX hit the 63-entry cap the next call would write the entry then immediately overwrite it with the terminator. * state.c FCEUSS_Save_Mem: post-save callback was guarded by 'if (SPreSave)' instead of 'if (SPostSave)'. * Sequence-point UB in counter expressions across 4 board files, e.g. 'x = !x ? a : --x' and 'x = ++x % n'. GCC -Wsequence-point flags all five sites (285.c, 413.c, asic_mmc3.c, jyasic.c). * SFORMAT size mismatches between declared variable type and AddExState size argument: - asic_vrc3.c VRC3_count/VRC3_reload (uint16, saved as 1 byte) - mmc3.c m555_count (uint32, saved as 2 bytes), m555_count_- expired (uint8, saved as 2 bytes - OOB write into adjacent BSS). * unrom512.c UNROM-512 mapper 30 .srm save format: host-endian uint32 flash write counters. Now stored as LE on disk regardless of host (Battle Kid 2, Twin Dragons, Lizard, Sole, etc.). * ~70 multi-byte single-variable SFORMAT/AddExState entries across 36 board files were saved as host-byte-order. After the FlipByteOrder fix, these are now byte-swapped on BE so cross- platform savestates round-trip correctly. * coolgirl.c new ExStateLE() macro added; 22 multi-byte sites. * pic16c5x.c 11 multi-byte AddExState calls (m_PC, m_PREVPC, m_CONFIG, m_WDT, m_prescaler, m_opcode, m_STACK[0/1], m_icount, m_delay_timer, m_rtcc, m_inst_cycles, m_clock2cycle). * onebus.c PowerJoy Supermax submapper detection used non-portable *(uint32*)&info->MD5 cast that read different bytes on LE vs BE. * fds.c clean up redundant FCEUSTATE_RLSB encoding in AddExState calls that also passed type=1 (idempotent OR; readability fix). DOCUMENTATION ============= * input.c UpdateGP's *(uint32*)data cast looks like a typical endian bug but is actually correct (the libretro frontend builds JSReturn with matching host-uint32 shifts). Comment added to prevent future "fixes" from breaking it. Limitations not addressed ========================= * Element-stride-aware byte swapping. The savestate byte-swap mechanism (FlipByteOrder over the entire SFORMAT entry buffer) is structurally wrong for arrays of multi-byte values: it reverses the whole buffer end-to-end instead of byte-swapping each element. Several places that need cross-platform-portable arrays (VRC7 sound state, jyasic chr[8]) work around this by either splitting arrays into per-element SFORMAT entries (n106 PlayIndex, bandai reg) or by skipping save entirely on BE via #ifndef GEKKO. A proper fix would extend the size encoding with an element-stride field. Left for a future change because it would change the savestate format. * Strict-aliasing UB in ppu.c (around 9 sites doing *(uint32*)uint8_buf for fast 4-byte writes via FCEU_dwmemset). Works in practice with all common compilers because the patterns are byte-symmetric, but is formally UB. * FCEU_gmalloc calls exit(1) on OOM. A libretro core should never exit() because that takes down the whole frontend. Used by 100+ call sites; refactoring to return-NULL is out of scope here. Testing ======= * Build: clean on `make platform=unix` with -O2; no new warnings. * FlipByteOrder fix verified by hand-trace and a standalone unit test for counts 2, 4, 8. * uppow2 fix verified by unit test across 13 boundary cases. * SFORMAT size mismatches and missing-RLSB cases identified by Python static-analysis scripts that cross-reference SFORMAT entries against variable declarations. * iNES 2.0 exponent fix verified by hand-tracing what byte 0xFF produces post-fix: exp=30 (capped), mult=7, size=3 GiB nominal, capped to 1 GiB, capped to 2 GiB by uppow2, FCEU_malloc returns NULL on most systems, loader returns 0. No heap overflow for any input byte. * Savestate-loaded array index audit: built a Python scanner that extracts each AddExState/SFORMAT entry and cross-references the variable name against array-index uses in the same file. All flagged sites covered. * A libFuzzer harness and seed corpus generator (fuzz_main.c, gen_seed_corpus.py) accompany this submission for ongoing regression testing.
2026-05-04 02:15:40 +02:00
AddExState(&mapperFDS_filesize, 2, 1, "FLSZ");
(fds.c) backport new fds disk handling backported from fceux, commit by dinkc64 (fbneo) https://github.com/TASVideos/fceux/commit/0b4be4b639371204642f22def7bd92ea18a4c2f3 Fixes saving issue with some games (Bubble Bobble, Super Lode Runner II, ...)
2020-01-25 08:32:43 +08:00
AddExState(&mapperFDS_block, 1, 0, "BLCK");
core: memory-safety, leak, and savestate-portability audit fixes Squashed series of ~50 distinct bugs found during a multi-day security/correctness audit, ranging from ROM-triggerable heap corruption and savestate-triggered OOB read primitives down to obscure cross-platform savestate breakage. Build is clean on `make platform=unix` with zero new warnings. CRITICAL (ROM-triggerable, exploitable on the user's machine) ============================================================= * ines.c iNES 2.0 PRG/CHR exponent overflow leading to undersized allocation followed by heap buffer overflow on fread (pow() with attacker-chosen exponent up to 63). Cap exponent at 30 and compute size in uint32 with explicit cap below 2 GiB before storing in int. * ines.c miscROMSize int wraparound from attacker-controlled PRG/ CHR sizes; previous '& 0x8000000' check missed common underflow cases. Compute in int64 and reject <= 0 or > 128 MiB. * unif.c MAPR chunk OOB heap read on chunks of size < 4. Validate chunk size before allocating board-name buffer. * unif.c chunk size truncation: int conversion + missing cap allowed a 4 GiB chunk size to wrap. Cap at 16 MiB. * unif.c FixRomSize infinite loop on size > 0x80000000. Cap input. * unif.c DINF chunk: months[(m - 1) % 12] is UB for m==0; m comes from the .unf file. Clamp m into [1..12] before subtracting. * unif.c NAME chunk: GameInfo->name = malloc(...) was unchecked. * nsf.c size underflow: FCEU_fgetsize() - 0x80 wraps to ~UINT64_MAX on tiny files, propagating a huge size into uppow2/FCEU_malloc. Validate size > 0x80 before subtracting. * nsf.c NSFMaxBank * 4096 cap tightened from 1<<20 (UB on signed-int overflow) to 1<<19 (fits in int). * general.c uppow2 had signed-int UB (1 << 32) and wrapped to 0 for inputs near uint32 max. Cap at 0x80000000. * cheat.c SubCheats[256] BSS overflow when more than 256 GG/PAR substitution cheats are active. The array is adjacent to MMapPtrs[64] in BSS which is exposed via retro_get_memory_data, making this overflow visible from outside the core. * libretro.c retro_cheat_set strcpy stack overflow with arbitrary- length cheat strings from the frontend. Use strlcpy. CRITICAL (savestate-triggerable on a malicious .fcs file) ========================================================= * fds.c FDS savestate OOB read primitive: InDisk loaded from savestate is used as index into 8-element diskdata[] static pointer array. A value 8..254 dereferences arbitrary memory as a pointer (heap-read primitive). Also bounds-checked SelectDisk, mapperFDS_block, mapperFDS_blockstart, mapperFDS_diskaddr, mapperFDS_blocklen so blockstart+diskaddr stays within the 65500- byte disk buffer. * 11 mappers had savestate-loaded variables masked at write time but not at restore time, used as indices into fixed arrays: - 88.c, KS7037.c, 112.c cmd indexes reg[8] - sachen.c (S8259, S74LS374N) cmd indexes latch[8] - 357.c dipswitch indexes outer_bank[4] - unrom512.c flash_state indexes erase_a/d/b[5] - 368.c preg indexes banks[8] - 69.c sndcmd indexes sreg[14] - mmc2and4.c latch0/latch1 index creg[4] None individually exploitable into RCE - the array writes corrupt adjacent BSS with constrained data flowing in - but each is an out-of-bounds read or write from attacker-controllable input. HIGH (memory-safety, reachable on any load) =========================================== * fceu-memory.c FCEU_malloc deref-of-NULL on allocation failure ('ret = 0; memset(ret, 0, size);'). * file.c multiple FCEUFILE/MakeMemWrap/MakeMemWrapBuffer unchecked allocations and unchecked filestream_tell return. * libretro.c GameInfo NULL-derefs in three entry points (retro_set_controller_port_device, retro_get_memory_data, retro_get_memory_size) reachable on operations called before a successful load. * libretro.c framebuffer leak ~256 KB per failed FCEUI_LoadGame (the libretro frontend doesn't call retro_unload_game on a failed load). * libretro.c 3DS retro_deinit unchecked linearFree. * libretro.c stereo / NTSC filter unchecked malloc returns. * fceu.c FCEUI_LoadGame unchecked GameInfo malloc. * fds.c SubLoad and FDSLoad unchecked FCEU_malloc on diskdatao backup buffers (NULL-deref in subsequent memcpy). * fceu.c AllocGenieRW partial-failure leak: AReadG allocated but BWriteG malloc fails -> 256 KB leak per retry. * input/bworld.c Update(): unbounded strcpy from attacker-supplied data into 20-byte bdata. Replaced with length-bounded copy. * cart.c setprg2r/4r and setchr1r/2r/4r/8r mask-underflow guards. SetupCartPRGMapping/SetupCartCHRMapping compute (size >> N) - 1 which underflows to 0xFFFFFFFF for chips smaller than the unit. setprg8r/16r/32r and setchr8r already had defensive size checks; the smaller units did not. malee.c (2 KB chip) and mapper 218 (2 KB NTARAM-as-CHR) accidentally avoid OOB; the primitive is unsafe for any future board. * video.c FCEU_InitVirtualVideo XBuf/XDBuf partial-failure leak. * video.c, fceu.c FCEU_DispMessage / FCEU_printf / FCEU_PrintError: vsprintf into fixed stack buffers replaced with vsnprintf. * core: validate magic-string read length on FDS/UNIF/NSF load (reject files too short to contain the magic so previous static buffer / stack garbage doesn't spuriously match). LEAKS (every cart load/unload cycle) ==================================== * mmc5.c MMC5 cart-side: WRAM (up to 64K) + MMC5fill (1K) + ExRAM (1K) leaked per cycle. NSFMMC5_Close existed but was only used for NSF code path. * onebus.c Mapper 270/436: 8 KB CHRRAM leaked per cycle. * 330.c, 375.c, 528.c: 8 KB WRAM each, no Close at all. CORRECTNESS (UB / ABI / endianness) =================================== * fceu-endian.c FlipByteOrder loop bound was 'count' instead of 'count/2', making it a no-op for every even count and silently breaking savestate portability for every FCEUSTATE_RLSB-marked field. The '#ifndef GEKKO' workarounds scattered through the codebase (sound.c, vrc7.c, vrc6.c) were symptoms of this root cause. * state.c AddExState bounds check ran AFTER writing the entry, so when SFEXINDEX hit the 63-entry cap the next call would write the entry then immediately overwrite it with the terminator. * state.c FCEUSS_Save_Mem: post-save callback was guarded by 'if (SPreSave)' instead of 'if (SPostSave)'. * Sequence-point UB in counter expressions across 4 board files, e.g. 'x = !x ? a : --x' and 'x = ++x % n'. GCC -Wsequence-point flags all five sites (285.c, 413.c, asic_mmc3.c, jyasic.c). * SFORMAT size mismatches between declared variable type and AddExState size argument: - asic_vrc3.c VRC3_count/VRC3_reload (uint16, saved as 1 byte) - mmc3.c m555_count (uint32, saved as 2 bytes), m555_count_- expired (uint8, saved as 2 bytes - OOB write into adjacent BSS). * unrom512.c UNROM-512 mapper 30 .srm save format: host-endian uint32 flash write counters. Now stored as LE on disk regardless of host (Battle Kid 2, Twin Dragons, Lizard, Sole, etc.). * ~70 multi-byte single-variable SFORMAT/AddExState entries across 36 board files were saved as host-byte-order. After the FlipByteOrder fix, these are now byte-swapped on BE so cross- platform savestates round-trip correctly. * coolgirl.c new ExStateLE() macro added; 22 multi-byte sites. * pic16c5x.c 11 multi-byte AddExState calls (m_PC, m_PREVPC, m_CONFIG, m_WDT, m_prescaler, m_opcode, m_STACK[0/1], m_icount, m_delay_timer, m_rtcc, m_inst_cycles, m_clock2cycle). * onebus.c PowerJoy Supermax submapper detection used non-portable *(uint32*)&info->MD5 cast that read different bytes on LE vs BE. * fds.c clean up redundant FCEUSTATE_RLSB encoding in AddExState calls that also passed type=1 (idempotent OR; readability fix). DOCUMENTATION ============= * input.c UpdateGP's *(uint32*)data cast looks like a typical endian bug but is actually correct (the libretro frontend builds JSReturn with matching host-uint32 shifts). Comment added to prevent future "fixes" from breaking it. Limitations not addressed ========================= * Element-stride-aware byte swapping. The savestate byte-swap mechanism (FlipByteOrder over the entire SFORMAT entry buffer) is structurally wrong for arrays of multi-byte values: it reverses the whole buffer end-to-end instead of byte-swapping each element. Several places that need cross-platform-portable arrays (VRC7 sound state, jyasic chr[8]) work around this by either splitting arrays into per-element SFORMAT entries (n106 PlayIndex, bandai reg) or by skipping save entirely on BE via #ifndef GEKKO. A proper fix would extend the size encoding with an element-stride field. Left for a future change because it would change the savestate format. * Strict-aliasing UB in ppu.c (around 9 sites doing *(uint32*)uint8_buf for fast 4-byte writes via FCEU_dwmemset). Works in practice with all common compilers because the patterns are byte-symmetric, but is formally UB. * FCEU_gmalloc calls exit(1) on OOM. A libretro core should never exit() because that takes down the whole frontend. Used by 100+ call sites; refactoring to return-NULL is out of scope here. Testing ======= * Build: clean on `make platform=unix` with -O2; no new warnings. * FlipByteOrder fix verified by hand-trace and a standalone unit test for counts 2, 4, 8. * uppow2 fix verified by unit test across 13 boundary cases. * SFORMAT size mismatches and missing-RLSB cases identified by Python static-analysis scripts that cross-reference SFORMAT entries against variable declarations. * iNES 2.0 exponent fix verified by hand-tracing what byte 0xFF produces post-fix: exp=30 (capped), mult=7, size=3 GiB nominal, capped to 1 GiB, capped to 2 GiB by uppow2, FCEU_malloc returns NULL on most systems, loader returns 0. No heap overflow for any input byte. * Savestate-loaded array index audit: built a Python scanner that extracts each AddExState/SFORMAT entry and cross-references the variable name against array-index uses in the same file. All flagged sites covered. * A libFuzzer harness and seed corpus generator (fuzz_main.c, gen_seed_corpus.py) accompany this submission for ongoing regression testing.
2026-05-04 02:15:40 +02:00
AddExState(&mapperFDS_blockstart, 2, 1, "BLKS");
AddExState(&mapperFDS_blocklen, 2, 1, "BLKL");
AddExState(&mapperFDS_diskaddr, 2, 1, "DADR");
(fds.c) backport new fds disk handling backported from fceux, commit by dinkc64 (fbneo) https://github.com/TASVideos/fceux/commit/0b4be4b639371204642f22def7bd92ea18a4c2f3 Fixes saving issue with some games (Bubble Bobble, Super Lode Runner II, ...)
2020-01-25 08:32:43 +08:00
AddExState(&mapperFDS_diskaccess, 1, 0, "DACC");
Initial commit - http://sourceforge.net/p/fceumm/code/160/
2014-03-30 22:15:17 +02:00
CHRRAMSize = 8192;
core: stdint typedefs, LE optimizations, frame determinism Three follow-up audit passes on top of the memory-safety / leak / savestate-portability work in 1185db8. ============================================================== Pass 1: replace custom typedefs with C99 stdint types throughout ============================================================== The custom uint8 / uint16 / uint32 / uint64 / int8 / int16 / int32 / int64 typedefs in src/fceu-types.h were just simple aliases for the C99 stdint.h types. Replace them with the standard names directly. - 498 files modified - ~3,400 token replacements (uint8 -> uint8_t, etc) - fceu-types.h slimmed down to just INLINE / GINLINE / FASTAPASS macros and the readfunc / writefunc function-pointer typedefs (those now use uint8_t / uint32_t natively) - Build clean on `make platform=unix` with zero new warnings - Output binary size unchanged - confirming semantic equivalence Mechanical replacement done with a Python script that uses word- boundary regex to avoid false positives (e.g. 'uint32_t' was correctly left alone because '_' is a word character so 'uint32' is not a complete word inside it). ================================================================ Pass 2: prefer memcpy on LE hosts for endian read/write helpers ================================================================ fceu-endian.c's write32le_mem, FCEU_en32lsb, and FCEU_de32lsb performed bytewise composition/decomposition unconditionally. On LE hosts the in-memory representation already matches the desired LE on-disk format, so a single memcpy is equivalent and lets the compiler emit a single load/store rather than four byte ops. - The bytewise path is kept inside #ifdef MSB_FIRST for BE hosts where it implements the actual byte swap - Both forms produce identical results; this is a code-clarity change more than a performance one (the optimizer was already merging the shifts on LE), but it documents the intent and removes a strict-aliasing-flavoured cast through *(uint32_t*)Bufo - Added missing #include <string.h> in fceu-endian.c which was relying on transitive includes for memcpy Other MSB_FIRST sites in the codebase (state.c FlipByteOrder guards, ppu.c sprite-line rendering, boards/unrom512.c flash-write- counter access) were already optimized for LE; they were verified correct rather than changed. ================================================================ Pass 3: frame determinism for replay and netplay ================================================================ Two libc rand() sites in core were replaced with a local xorshift32 PRNG so that NES games which read uninitialised memory or hit hardware "weak bit" emulation produce reproducible behaviour across runs. NES titles routinely read uninitialised RAM (variables not zeroed before use, sprite Y-position set by junk-on-stack), so the RAM contents at power-on subtly affect game behaviour. With libc rand(), those contents depend on whether anyone else seeded rand() in the same process - a different libretro frontend, a different audio backend init order, or any frontend that does srand(time(0)) all break replay / netplay frame-determinism. 1. fceu.c FCEU_MemoryRand. Used to fill RAM (PowerNES) and CHR-RAM (iNES_Init) at power-on when option_ramstate=2 (random init). Replaced with a local xorshift32 PRNG, exposed via a new FCEU_MemoryRand_Reseed(uint32_t) function called once per power-on: - PowerNES seeds from the first 4 bytes of GameInfo->MD5 (set by all loaders before PowerNES runs) so identical ROMs produce identical RAM, different ROMs differ - iNES_Init seeds from iNESCart.PRGCRC32 before the CHR-RAM fill so two builds of the same ROM get the same CHR-RAM - The PRNG state advances across multiple FCEU_MemoryRand calls within one power-on so RAM and CHR-RAM get different content (matching NES hardware reality) 2. boards/rt-01.c UNLRT01Read. The RT-01 board has 'weak bit' protected EPROM regions; reads of 0xCE80-0xCEFF and 0xFE80- 0xFEFF return 0xF2 with the low 3 bits randomised. Replaced libc rand() with a local xorshift32 seeded at power-on, and added the PRNG state to the savestate via AddExState with key "WBKS" so save / load / rewind / netplay rollback all stay deterministic. In addition, two long-double-to-int truncations were changed to double for cross-platform FP determinism: - sound.c SetSoundVariables: soundtsinc - boards/n106.c DoNamcoSound: inc long double has platform-dependent precision (80-bit on x87, 64-bit with -mfpmath=sse, 128-bit on PowerPC), so the truncated integer result varied across these platforms. double is guaranteed 64-bit IEEE-754 portably. After this pass, the core has no time(), clock(), gettimeofday(), clock_gettime(), getpid(), getuid(), getgid(), getenv(), gethostid(), pthread, std::thread, OpenMP, signal handler, or non-deterministic- malloc dependency. Verified with a Python scanner that greps the source for these patterns; runs clean. The PPU / APU / CPU power-on already explicitly memset all state buffers to 0 (deterministic), and ROM/CHR-ROM allocation already memsets to 0xFF before partial fread (deterministic regardless of file truncation). Combined with the memory-safety hardening in 1185db8 (which prevents savestate-loaded indices from going out-of-bounds and producing unpredictable behaviour), the core now offers genuine frame-deterministic replay across runs, builds, and host endian.
2026-05-04 02:46:34 +02:00
CHRRAM = (uint8_t*)FCEU_gmalloc(CHRRAMSize);
Initial commit - http://sourceforge.net/p/fceumm/code/160/
2014-03-30 22:15:17 +02:00
SetupCartCHRMapping(0, CHRRAM, CHRRAMSize, 1);
AddExState(CHRRAM, CHRRAMSize, 0, "CHRR");
FDSRAMSize = 32768;
core: stdint typedefs, LE optimizations, frame determinism Three follow-up audit passes on top of the memory-safety / leak / savestate-portability work in 1185db8. ============================================================== Pass 1: replace custom typedefs with C99 stdint types throughout ============================================================== The custom uint8 / uint16 / uint32 / uint64 / int8 / int16 / int32 / int64 typedefs in src/fceu-types.h were just simple aliases for the C99 stdint.h types. Replace them with the standard names directly. - 498 files modified - ~3,400 token replacements (uint8 -> uint8_t, etc) - fceu-types.h slimmed down to just INLINE / GINLINE / FASTAPASS macros and the readfunc / writefunc function-pointer typedefs (those now use uint8_t / uint32_t natively) - Build clean on `make platform=unix` with zero new warnings - Output binary size unchanged - confirming semantic equivalence Mechanical replacement done with a Python script that uses word- boundary regex to avoid false positives (e.g. 'uint32_t' was correctly left alone because '_' is a word character so 'uint32' is not a complete word inside it). ================================================================ Pass 2: prefer memcpy on LE hosts for endian read/write helpers ================================================================ fceu-endian.c's write32le_mem, FCEU_en32lsb, and FCEU_de32lsb performed bytewise composition/decomposition unconditionally. On LE hosts the in-memory representation already matches the desired LE on-disk format, so a single memcpy is equivalent and lets the compiler emit a single load/store rather than four byte ops. - The bytewise path is kept inside #ifdef MSB_FIRST for BE hosts where it implements the actual byte swap - Both forms produce identical results; this is a code-clarity change more than a performance one (the optimizer was already merging the shifts on LE), but it documents the intent and removes a strict-aliasing-flavoured cast through *(uint32_t*)Bufo - Added missing #include <string.h> in fceu-endian.c which was relying on transitive includes for memcpy Other MSB_FIRST sites in the codebase (state.c FlipByteOrder guards, ppu.c sprite-line rendering, boards/unrom512.c flash-write- counter access) were already optimized for LE; they were verified correct rather than changed. ================================================================ Pass 3: frame determinism for replay and netplay ================================================================ Two libc rand() sites in core were replaced with a local xorshift32 PRNG so that NES games which read uninitialised memory or hit hardware "weak bit" emulation produce reproducible behaviour across runs. NES titles routinely read uninitialised RAM (variables not zeroed before use, sprite Y-position set by junk-on-stack), so the RAM contents at power-on subtly affect game behaviour. With libc rand(), those contents depend on whether anyone else seeded rand() in the same process - a different libretro frontend, a different audio backend init order, or any frontend that does srand(time(0)) all break replay / netplay frame-determinism. 1. fceu.c FCEU_MemoryRand. Used to fill RAM (PowerNES) and CHR-RAM (iNES_Init) at power-on when option_ramstate=2 (random init). Replaced with a local xorshift32 PRNG, exposed via a new FCEU_MemoryRand_Reseed(uint32_t) function called once per power-on: - PowerNES seeds from the first 4 bytes of GameInfo->MD5 (set by all loaders before PowerNES runs) so identical ROMs produce identical RAM, different ROMs differ - iNES_Init seeds from iNESCart.PRGCRC32 before the CHR-RAM fill so two builds of the same ROM get the same CHR-RAM - The PRNG state advances across multiple FCEU_MemoryRand calls within one power-on so RAM and CHR-RAM get different content (matching NES hardware reality) 2. boards/rt-01.c UNLRT01Read. The RT-01 board has 'weak bit' protected EPROM regions; reads of 0xCE80-0xCEFF and 0xFE80- 0xFEFF return 0xF2 with the low 3 bits randomised. Replaced libc rand() with a local xorshift32 seeded at power-on, and added the PRNG state to the savestate via AddExState with key "WBKS" so save / load / rewind / netplay rollback all stay deterministic. In addition, two long-double-to-int truncations were changed to double for cross-platform FP determinism: - sound.c SetSoundVariables: soundtsinc - boards/n106.c DoNamcoSound: inc long double has platform-dependent precision (80-bit on x87, 64-bit with -mfpmath=sse, 128-bit on PowerPC), so the truncated integer result varied across these platforms. double is guaranteed 64-bit IEEE-754 portably. After this pass, the core has no time(), clock(), gettimeofday(), clock_gettime(), getpid(), getuid(), getgid(), getenv(), gethostid(), pthread, std::thread, OpenMP, signal handler, or non-deterministic- malloc dependency. Verified with a Python scanner that greps the source for these patterns; runs clean. The PPU / APU / CPU power-on already explicitly memset all state buffers to 0 (deterministic), and ROM/CHR-ROM allocation already memsets to 0xFF before partial fread (deterministic regardless of file truncation). Combined with the memory-safety hardening in 1185db8 (which prevents savestate-loaded indices from going out-of-bounds and producing unpredictable behaviour), the core now offers genuine frame-deterministic replay across runs, builds, and host endian.
2026-05-04 02:46:34 +02:00
FDSRAM = (uint8_t*)FCEU_gmalloc(FDSRAMSize);
Initial commit - http://sourceforge.net/p/fceumm/code/160/
2014-03-30 22:15:17 +02:00
SetupCartPRGMapping(1, FDSRAM, FDSRAMSize, 1);
AddExState(FDSRAM, FDSRAMSize, 0, "FDSR");
SetupCartMirroring(0, 0, 0);
FDS: Print manufacturer label when loading rom
2019-01-04 22:33:26 +08:00
FCEU_printf(" Code : %02x\n", diskdata[0][0xf]);
FCEU_printf(" Manufacturer : %s\n", getManufacturer(diskdata[0][0xf]));
FCEU_printf(" # of Sides : %d\n", TotalSides);
FCEU_printf(" ROM MD5 : 0x%s\n", md5_asciistr(GameInfo->MD5));
Update Region Override
2017-10-07 13:35:00 +08:00
FCEUI_SetVidSystem(0);
Initial commit - http://sourceforge.net/p/fceumm/code/160/
2014-03-30 22:15:17 +02:00
return 1;
}
void FDSClose(void) {
int x;
if (!DiskWritten) return;
for (x = 0; x < TotalSides; x++)
if (diskdatao[x]) {
free(diskdatao[x]);
diskdatao[x] = 0;
}
FreeFDSMemory();
}
Reference in New Issue Copy Permalink