Removed the Enable line from the DISCRETE_ADJUSTMENT modules. They should never be disabled.
Some before/after speed tests:
dkong - from 203% to 204%
firetrk - from 418% to 429%
phoenix - from 253% to 257%
polaris - from 448% to 476%
to whole functions. The other one I'm not so sure about. Commented
code is usually hilighted differently, making it very easy to spot.)
Hi mamedev,
This set of patches has one aim only, to identify chunks of code that
have been disabled by the use of C/C++ comments, and to disable them
instead by using the preprocessor. The C comment approach to
disabling code isn't safe (embedded comments will trip it up), and the
C++ comment approach isn't elegant (you shouldn't need to touch every
line to disable a chunk of code). Using #if...#endif is preferable
always, excepting perhaps if (0) { ... }.
The patch has three parts. The first only handles cases where full
functions were disabled. The second handles cases where parts of
functions were disabled. The third then tries to restore the
whitespace that the use of comments converted from tabs to spaces via
srcclean.exe. It also cleans up the whitespace in a handful of the
files in areas outside of the original two patches.
~aa
Subject: a few cleanups
This patch contains:
- removal of unprintable chars (newlines) in SH-2 disassembler (I
submitted this in the past nd it wasn't included)
- a few unnecessary checks after malloc_ort_die() calls
- changes two romload.c warnings to use GAMENOUN instead
- adds "deprecat.h" in a few src/mame/drivers files (would be
necessary, if the debugger.h one would be removed)
- cleans up the mame.mak by adding all missing defines and grouping
them based on cpu.mak
- renamed video_exit() to winvideo_exit() for consistency in function
names
files. Removed old option 'cheat_file'.
Updated xmlfile.c to specify the line number for each node, for more
accurate error reporting.
Removed old cheat.c/.h and replaced them with code that will read and
write the new XML-based cheat format using the new cheatpath option
(cheats still need to be enabled in order to load at all). Note that
the cheats are not implemented yet, but they are fully parsed. To test
saving, whenever a cheat file is successfully loaded, it is immediately
written back out as 'output.xml' to the cheat directory for validation.
The ROL/ROR/SHL/SHR opcodes provide the carry flags typically generated by
almost all CPUs. The RORC/ROLC opcdes map directly to the rotate through
carry of most CPUs as well.
auto-filled by the device manager before starting the device. This
simplifies the auto-binding process for devices.
Added new selector for device-specific ROM regions. Not yet wired
up in the ROM loader.
From: Atari Ace [mailto:atari_ace@verizon.net]
Subject: [patch] Stricter ADDRESS_MAP checks
Hi mamedev,
The current implementation of address maps is lacking some useful
checks of the initial tokens. In particular, it doesn't validate that
a particular entry doesn't try to define the same handler multiple
times.
The attached patch adds this and some other validations as fatalerrors
in address_map_detokenize, and fixes the affected maps. The errors
generally fall into the following categories.
1. AM_RAM AM_WRITE(...). Should have been AM_RAM_WRITE(...)
2. AM_RAM AM_READWRITE(...). The AM_RAM is a NOP.
3. AM_RAM AM_RAMBANK(...). The AM_RAM is a NOP.
4. AM_ROM AM_ROMBANK(...). The AM_ROM is a NOP.
5. AM_ROM AM_READ(...). The AM_ROM is a NOP.
One peculiar error was in equites.c, where there were two AM_BASE
entries.
~aa
second SH2 for ST-V games.
Fixed buffer overrun in latest SCSP code. Not sure how that ever
worked.
colmns97 still asserts in SCSPDSP, but this happened before the
region change.
integer value, regions are now referred to by a region class and
a region tag. The class specifies the type of region (one of CPU,
gfx, sound, user, disk, prom, pld) while the tag uniquely specifies
the region. This change required updating all the ROM region
definitions in the project to specify the class/tag instead of
region number.
Updated the core memory_region_* functions to accept a class/tag
pair. Added new memory_region_next() function to allow for iteration
over all memory regions of a given class. Added new function
memory_region_class_name() to return the name for a given CPU
memory region class.
Changed the auto-binding behavior of CPU regions. Previously, the
first CPU would auto-bind to REGION_CPU1 (that is, any ROM references
would automatically assume that they lived in the corresponding
region). Now, each CPU automatically binds to the RGNCLASS_CPU region
with the same tag as the CPU itself. This behavior required ensuring
that all previous REGION_CPU* regions were changed to RGNCLASS_CPU
with the same tag as the CPU.
Introduced a new auto-binding mechanism for sound cores. This works
similarly to the CPU binding. Each sound core that requires a memory
region now auto-binds to the RGNCLASS_SOUND with the same tag as the
sound core. In almost all cases, this allowed for the removal of the
explicit region item in the sound configuration, which in turn
allowed for many sound configurations to removed altogether.
Updated the expression engine's memory reference behavior. A recent
update expanded the scope of memory references to allow for referencing
data in non-active CPU spaces, in memory regions, and in EEPROMs.
However, this previous update required an index, which is no longer
appropriate for regions and will become increasingly less appropriate
for CPUs over time. Instead, a new syntax is supported, of the form:
"[tag.][space]size@addr", where 'tag' is an optional tag for the CPU
or memory region you wish to access, followed by a period as a
separator; 'space' is the memory address space or region class you
wish to access (p/d/i for program/data/I/O spaces; o for opcode space;
r for direct RAM; c/u/g/s for CPU/user/gfx/sound regions; e for
EEPROMs); and 'size' is the usual b/w/d/q for byte/word/dword/qword.
Cleaned up ROM definition flags and removed some ugly hacks that had
existed previously. Expanded to support up to 256 BIOSes. Updated
ROM_COPY to support specifying class/tag for the source region.
Updated the address map AM_REGION macro to support specifying a
class/tag for the region.
Updated debugger windows to display the CPU and region tags where
appropriate.
Updated -listxml to output region class and tag for each ROM entry.
- Fixed ADPCM decoder to remove bad optimization and match YMZ280B. DC offset is fixed and output quality is higher
- Improved handling of slot monitor and timers B and C
- Simplified interpolation code
- Rewrote core logic, communications, and interfaces.
- Added three parallel memory moves to the disassembler.
- Initial interrupt logic in place.
Plygonet.c updates.
- All communication hacks have been removed.
- Memory maps have been temporarily reverted while new DSP56k cpu core catches up.
Subject: [patch] Conditional code cleanup resubmit
Hi mamedev,
This is a resubmit of a previous patch. The earlier version would not
compile with 32-bit MSVC, due to the fact that its linker required
external dependencies in dead code to be met before dead code
elimination was done, causing linker errors. The proper fix for this
would be to add the necessary dependencies, so I instead simply left
the conditional code in place in winalloc.c and chd.c.
~aa
Original submission email below:
----
Conditionally compiled code tends to bitrot, so MAME should try to
avoid it as much as possible. I sent a patch six months ago to
eliminate conditional code associated with logging, here's another
patch that does more of this. Some notes:
1. drc_ops.c: I couldn't find a LOG_CODE anywhere, so I used if (0).
2. romload.c: I converted all the users of debugload to use
LOG((...)) instead, following the traditional conditional logging
pattern.
3. windows/sound.c: I eliminated the separate sound log and directed
the few outputs to the error log.
~aa
Subject: [patch] .data removals to fix reset/multisession bugs
Hi mamedev,
One nice artifact of properly constifying data structures in MAME is
that it makes it relatively easy to spot a class of reset/multisession
bugs, namely that almost any object in .data is probably in error.
Unless the value is properly initialized in a reset routine the
initial non-zero value can't be relied upon, so there's no need to
have a non-zero value to begin with.
With that in mind, here's a patch to move more items out of .data by
either applying const, removing the non-zero initializer (if its
overwritten by init/reset) or by adding appropriate initialization
code. In most cases I tried to add initialization code to a reset
routine, but in some cases I chose an init routine, possibly leaving a
reset bug intact.
Some interesting bits:
1. tms9900 core. The use of .data to initialize the irq_level wasn't
correct in some cases as the layout of the structure was core
dependent.
2. bfcobra.c. By introducing a VIDEO_START routine a hack in
VIDEO_UPDATE could be removed.
~aa
Subject: 8257dma
Hi Aaron,
Small changes for 8257.
Fixed terminal count stop for autoload mode and added callback function existance check [Miodrag Milanovic]
Regards,
Miodrag
convert between attotimes and a clock tick at an integral frequency.
Changed the 6532 RIOT device into a proper device. Rewrote the
logic to be simpler and leverage the new attotime functions. Changed
the I/O port setters to specify a mask, and changed the I/O port
callbacks to pass in the previous value. Updated tourtabl and
gameplan drivers to use the new device interface.
Converted audio/starwars.c, audio/exidy.c, and audio/gottlieb.c to
use the new RIOT implementation instead of rolling their own.
Began gottlieb.c cleanup. Converted palette calculations to resistor
weights. Corrected video timing. Reduced the number of separate
machine drivers. Fixed incorrect spriteram sizes. Populated full
memory maps for the main CPU and the rev 1 sound board. More to
come.
- Rearranged decoding behavior.
- Updated code style to MAME standards.
- Tested each single-word opcode with custom ROMs.
(This is a precursor to other major dsp56k changes I have made. I just gotta' get plygonet commanders up and running to do some tests)
- TGP now correctly uses table roms (model1/2 updated accordingly)
- removed FIFO hack on srallyc (game now runs)
- added analog ports reading for model 2A/B/C
- fixed some loading instructions in the TGP. that fixes srallyc automatic transmission bug
Added validity checks to ensure no duplicate sound or CPU tags.
Fixed several duplicate sound tags from the last checkin.
-listxml now outputs the tag names for CPUs and sound chips.
well, the fix is not that but changing them this way:
program_write_dword_32le(i960.r[I960_FP]-16, i960.PC);
program_write_dword_32le(i960.r[I960_FP]-12, i960.AC);
in take_interrupt.
It seems it came from a partial fix in the ret instruction. the manual says that PC is saved at FP-16 and AC at FP-12
it was reversed in mame, so when trying to fix it, i suppose Ernesto forgot to change the push too.
I've tested this fix and apart from having daytona working, pilot kids 2A works too (that I think has been broken for ages).
MDRV_SOUND_ADD. All sound chips must now include a tag.
Laboriously changed all existing drivers to define a
unique tag for each sound chip.
CPUs are next, but will require a more hands-on manual
process to produce reasonable tags.
properly ignore the "break into debugger" keypress and not allow
related characters to filter through. Removed some hacks related to
making that work in the past.
Changed osd_wait_for_debugger() to take a machine parameter and a
"firsttime" parameter, which is set to 1 the first time the function
is called after a break. The Windows debugger uses this to ensure
that the debugger has focus when you break into it.
EEPROM data, and the size is in terms of units, not bytes. Updated all
drivers accordingly.
Changed the ROM loading code to actually alter the region flags based
on the CPU endianness and bus width when creating the region, rather
than fixing them up on the fly. This means that callers to
memory_region_flags() will get the correct results.
Changed the expression engine to use two callbacks for read/write rather
than relying on externally defined functions.
Expanded memory access support in the expression engine. Memory accesses
can now be specified as [space][num]<size>@<address>. 'space' can be
one of the following:
p = program address space of CPU #num (default)
d = data address space of CPU #num
i = I/O address space of CPU #num
o = opcode address space of CPU #num (R/W access to decrypted opcodes)
r = direct RAM space of CPU #num (always allows writes, even for ROM)
e = EEPROM index #num
c = direct REGION_CPU#num access
u = direct REGION_USER#num access
g = direct REGION_GFX#num access
s = direct REGION_SOUND#num access
The 'num' field is optional for p/d/i/o/r, where is defaults to the
current CPU, and for e, where it defaults to EEPROM #0. 'num' is required
for all region-related prefixes. Some examples:
w@curpc = word at 'curpc' in the active CPU's program address space
dd@0 = dword at 0x0 in the active CPU's data address space
r2b@100 = byte at 0x100 from a RAM/ROM region in CPU #2's program space
ew@7f = word from EEPROM address 0x7f
u2q@40 = qword from REGION_USER2, offset 0x40
The 'size' field is always required, and can be b/w/d/q for byte, word,
dword, and qword accesses.
- All games on all drivers should work as before except "colmns97" and "stress" which crash due to sound system trouble.
- All idle skips are still included. They are quite a bit less effective than they were on the interpreter, but they still give a boost.
- Fast RAM bypass is not included yet so this does not represent final performance. That said, it's consistently faster than the interpreter even now. Example: sfiii3 on 0.126 gets 609% on the interpreter and 961% on the DRC.
Major thanks to Aaron for his assistance with several sticky core bugs and other issues encountered writing this.
lurking. If you run into anything odd, please let me know.
Added new module uiinput.c which manages input for the user interface.
The OSD is responsible for pushing mouse events and character events
to this interface in order to support mouse movement and text-based
input (currently only used for the select game menu). Added support
for navigating through the menus using the mouse.
[Nathan Woods, Aaron Giles]
Redesigned the UI menus so that they can maintain a richer state. Now
the menus can be generated once and reused, rather than requiring them
to be regenerated on each frame. All menus also share a comment eventing
system and navigation through them is managed centrally. Rewrote all the
menus to use the new system, apart from the cheat menus, which are now
disabled. Reorganized the video menu to make it easier to understand.
[Aaron Giles]
- Roll back older periodic function
Now standard/old format work in this function
Otherwords, several operations which doesn't exist in older
function and added recently are now disabled
- Separated load_cheat_code() to 3 indipendent functions
And added new cheat options, "Load New/Standard/Old Format Code"
(Default : load all format codes)
NOTE : delete all cheat options in the database before start
or reload default options (Shift + Reload key)
in cheat general menu
- Added format strings structure to accept huge size strings
in loading a database to prevent from breaking other strings
- Changed Link in new format
Label code is now "Link-Label" and sub-label is "Label-Sub-Link"
"Standard-Link" is no longer label
- Added choose_label_index() to manage label selection
- Fixed build_label_index_table() to prevent from crashing
And disabled to build label index table in case of
standard/old format
- Fixed memory free problem in case of standard/old code
- Fixed several reported/found bugs
* marked IDT instructions for R4650 only
* accounted for time taken in throwaway likely branch slots
* fixed jalr to respect the correct link register
* fixed c.eq, c.lt, c.le to fail if unordered
* fixed swxc1, sdxc1 to use the correct source register (fixes many Gauntlet problems)
* fixed CCR31 display in debugger
* fixes assert: src/emu/video.c:664: visarea->min_x < width
* Test whether screen type is vector and ignore min_x < width in this case
* This affected a number of vector games and the only other fix would be to rewrite the vector code
Subject: another Machine -> machine cleanup
This cleans up most of the Machine stuff in src/emu/machine. There is
a bit left to clean up, but it's mostly stuck at some interfaces now.
Subject: [patch] make MSVC_BUILD=1 -j<n> build fix
Hi mamedev,
The following build tweak fixes make MSVC_BUILD=1 -j2. It also fixes
an oversight in the cross-build support.
~aa
Subject: [patch] Fix minor CPU/SOUND core build issues
Hi mamedev,
This small patch fixes a few build problems with the cpu/sound cores
M65CE02, ALPHA8201, TMC0285, TMS5200 and corrects the dependencies for
the mips core.
~aa
macro from the source code. All MAME builds now include
the debugger, and it is enabled/disabled exclusively by
the runtime command-line/ini settings. This is a minor
speed hit for now, but will be further optimized going
forward.
Changed the 'd' suffix in the makefile to apply to DEBUG
builds (versus DEBUGGER builds as it did before).
Changed machine->debug_mode to machine->debug_flags.
These flags now indicate several things, such as whether
debugging is enabled, whether CPU cores should call the
debugger on each instruction, and whether there are live
watchpoints on each address space.
Redesigned a significant portion of debugcpu.c around
the concept of maintaining these flags globally and a
similar, more complete set of flags internally for each
CPU. All previous functionality should work as designed
but should be more robust and faster to work with.
Added new debugger hooks for starting/stopping CPU
execution. This allows the debugger to decide whether
or not a given CPU needs to call the debugger on each
instruction during the coming timeslice.
Added new debugger hook for reporting exceptions.
Proper exception breakpoints are not yet implemented.
Added new module debugger.c which is where global
debugger functions live.
- Merged 3 different search menu functions to search_main_menu()
- Added new cheat options, "Vertical/Horizontal Key Repeat Speed"
NOTE : delete all cheat options in the database before start
or reload default options (Shift + Reload key)
in cheat general menu
- Changed assignment of Entry for Activation Key and Pre-enable
command code. Now there are check the tag in comment field
instead of index number. But it's only for new format command
and old is as before
- Added tag output in save_activation_key() and save_pre_enable()
- Cleaned up save_cheat_options() to output description
- Changed memory read function in watchpoints
No longer "Debugger" watchpoint mis-hit "Cheat" watchpoint
- Fixed several reported/found bugs
- Renamed funcitons/enum based on coding guidlines
All functions are completed in this update
- connected EEPROM (doesn't seem to affect much)
- cleaned up system register access
GTI Club driver:
- altered network IRQ clear to fix several problems
- added Guru readme
- fixed crashes due to missing inputs
- gticlub "works" again
ZR107 driver:
- added Guru readme
- cleaned up system register access
- these games work again with altered network IRQ timing
NWK-TR driver:
- added Guru readme
DRC frontend:
- now passes pointer to previous instruction when describing
PPC frontend:
- attempts to roughly take into account branch and CR logical
folding in timing computations
- Added configurable i8x41/i8x42 subtype support.
- Fixed carry flag handling in ADDC A,#N instruction.
- Fixed carry flag handling in RLC A instruction.
* added save state support to the SHARC CPU core
* added save state support to the PowerPC recompiler
* added save state support to the virtual TLB system
* added save state support to the RF5C400 sound core
* added save state support to konppc module
* added save state support to K056800 host controller
* added save state support to the Konami hornet driver
Fixed poor default CLUT handling in the voodoo driver
Subject: [patch] memory_region madness reloaded
Hi mamedev,
The memory_region and memory_region_length functions are probably the
two most common functions in MAME that don't take a machine parameter
but should given the syntax of the related apis memory_region_type and
memory_region_flags. Clearly they didn't get the parameter because of
the sheer number of changes needed to change the apis. This pair of
patches makes the change, and deals with the consequences.
The second patch then changes the api for memory_region and
memory_region_length, and fixes the fallout. It generally plumbs
through machine parameters where needed, except for the case of sound
apis which I deferred doing so till later. This increased the number
of deprecat.h includes by ~50. Given it is a massive patch, there are
bound to be a few mistakes in it (I had to make ~20% of the changes by
hand), but I exercised care and reviewed the patch several times to
minimize the problems.
Subject: [patch] memory_region madness reloaded
Hi mamedev,
The memory_region and memory_region_length functions are probably the
two most common functions in MAME that don't take a machine parameter
but should given the syntax of the related apis memory_region_type and
memory_region_flags. Clearly they didn't get the parameter because of
the sheer number of changes needed to change the apis. This pair of
patches makes the change, and deals with the consequences.
The first patch makes some changes to help the second patch
along, as well as moves some uses of the memory_region apis within for
loops outside the loops.
* added save state support to the MIPS3 recompiler
* added save state support to CAGE audio system
* added save state support to the voodoo emulator
* added save state support to the smc91c9x emulator
* added save state support to the kinst, seattle, and vegas drivers
* fixed core video handling of save states with dynamic screen resolutions
SMC91C9x:
* converted to proper device
* updated seattle and vegas drivers to allocate devices
* added separate 91C96 device for eventual 2049 use
* cleaned up code
Removed call to discrete_reset() in the initialization code. It is already
called by the sound system during reset, at which point it is safe to read
input ports.
- reactivated back-end validation mechanism
- added back-end validation for ADD/SUB/MUL/DIV/CMP forms
- fixed several errors in dealing with more obscure flag combinations
on a command-line parameter and the configuration. Changed Windows OSD
code to use this instead of its own logic. Changed -snapview to share the
logic as well, enabling 'auto' as a -snapview option.
Enabled by default for snapshots and movie rendering.
Added new option: -snapsize, which lets you specify the target
resolution for snapshots and movies. The existing behavior is still
the default: create snapshots and movies at native pixel
resolutions.
Added new option: -snapview, which lets you specify a particular
view to use for rendering snapshots and movies. The existing
behavior is still the default: use a special internal view and
render each screen to its own snapshot in its own file. When using
this option to specify a view other than 'internal', only a single
snapshot file will be produced regardless of how many screens the
game has.
Improved AVI and MNG recording to properly duplicate/skip frames
as appropriate to keep the correct framerate.
MAMETesters Bugs Fixed
----------------------
00277: [Misc.] groundfx.c, gunbustr.c, superchs.c, taito_f3.c,
undrfire.c, psikyosh.c: Cheat search causes fatal error
Source Changes
--------------
Updated cheat system
- Changed the function of default memory read in case of the search
to prevent from crashing if a game uses cpu_spinutil()
Now basically used do_memory_read() instead of do_cpu_read()
(Try to search in actfancr between this and previous version)
NOTE : It's not fundamental solution but no longer crash
- Cleaned up read_data() and write_data()
- Cleaned up save/load routine and added open_cheat_database()
- Added new menu "Cheat Commands" into the cheat main menu
the following commands are supported in this menu
* reload cheat code (the same as Reload Database key)
* cheats ON/OFF (the same as Toggle Cheat key)
* watchpoints ON/OFF (the same as Shift + Toggle Cheat key)
* save description (save the title of the game)
* save raw code (save raw cheat code with new format)
- Fixed several bugs which has found in testing
- Renamed functions based on coding guidlines
Subject: 8080/8085 fix
Hello Aaron,
While working on drivers for some 8080 machines, I have noticed that there are some errors in flags settings in CPU implementation.
I have done fixes and now sending patch for code . Also there are changes to set right timings for both 8080 and 8085 since not all timings are up to documentation.
Regards,
Miodrag
Subject: [patch] Static qualifiers, header file cleanups, and new
include files for MAME
The first patch adds static qualifiers where appropriate, adds missing
#include statements, source comments and header declarations, as well
as removes dead declarations. The only part that required judgement
was deciding whether audio/galaxian.c declarations should be in
galaxold.h or galaxian.h, it doesn't make sense for them to be
declared in both. This exercise did find a bug, galaxold_init_stars
was declared incorrectly in video/fastfred.c.
Sent: Thursday, June 05, 2008 9:00 PM
To: submit@mamedev.org
Subject: Machine -> machine cleanups
This cleans up the usage of Machine in many of the src/video files.
* fixed adc/sbb so that they don't optimize out ever
* fixed detection of special and/or/xor cases
* fixed GETFLGS opcode so that it doesn't return anything other than requested flags
* changed LZCNT/BSWAP to be more flexible in register selection
C back-end:
* implemented flag variants of SEXT/ROLAND/ROLINS/LZCNT/BSWAP
PPC DRC:
* added more symbols for debugging
* fixed lmw/stmw if rA is one of the loaded/stored registers
* removed unnecessary variables & structure members
* optimized for the XER and CR0 case where XER doesn't need an overflow calculation
*
* changed SPU receive model to a push model; updated drivers accordingly
* added macros for setting the SPU transmit handler and sending bytes
* cleaned up ppc.h
* added detection of MMU enablement in 403GCX case
* fixed bug in protection bounds violation detection for 4XX to get ppd
to the "ppc4xx_spu_rx_data unimplemented" message
* fixed conflicting layout (not super happy with the technique...)
x64 back-end:
* added a "hop" around the prolog of the HANDLE opcode to allow fall-through
x86 back-end:
* redesigned stack management to keep the stack at the same 16-byte alignment
PPC DRC:
* fixed parameter ordering for masked reads/writes
* Added VBLANK IRQ clear
* Added ATAPI IRQ clear
* Tried to add UART IRQ clear but it doesn't yet work
* Most of the games start to show something again with these changes
* Added fast RAM region
PPC DRC:
* Fixed TLB filling for 4XX protection ranges
PC16552d:
* Tried to wire up assert/clear signals for this chip based on guesswork
(anyone want to implement it properly? :)
* reduced address bus width of 4xx series to 31 bits per the
documentation
PPC DRC:
* fixed bug that would jump to incorrect PC after filling the
TLB during a mismatch event
* added explicit address truncation to 31 bits for 4xx series
* added new PPCDRC_ACCURATE_SINGLES option, which removes the
excessive (and very likely unecessary) extra rounding when
performing the "fast" single-precision floating point ops
Konami drivers:
* designated fast RAM accesses for work RAM
* removed unnecessary mirroring
Model 3:
* identified and fixed VBLANK bit in real3d status which was
causing the system to hang at startup
* designated fast RAM accesses for work RAM
53C810:
* changed read/write handles to proper READ8/WRITE8_HANDLER
callbacks so they can eventually be used directly
* changed from tracking "live" registers to tracking "necessary" registers
* genericized register tracking to be more flexible
* added previous instruction pointer to opcode descriptions
PowerPC frontend/DRC:
* cleaned up register tracking implementation
* fixed numerous errors and shortcomings in the tracking
* added support for removing unnecessary XER CA and CR0 computations
* updated UML logging to output new frontend statistics
MIPS3 frontend/DRC:
* tweaked register tracking to match new DRC frontend system
* updated UML logging to output new frontend statistics
* fixed DIVS/DIVU opcode so they properly compute SZ flags when requested;
fixes camera wackiness in nbapbp
PPC DRC:
* added support for 4xx protection registers via the MMU
* added TLBH, TLBU, and DEC to the list of registers in the debugger
* turned off "end of transfer" DMA signals in favor of "transfer count 0"
signals; fixes memory-to-memory DMA in fiveside
* Converted sysreg_r/w to 8-bit handlers
* Added hack to make network IRQs work; brings some games back to life
UML:
* Fixed STORE opcode description to allow immediate source operands
x86/x64 back-ends:
* Added flag support to SEXT, ROLAND, ROLINS, LZCNT opcodes
PPC DRC:
* Rewrote lswi/stswi as subroutines
* Made accesses to tempdata explicitly dword or qword
* Fixed SRR0 when generating a syscall
* Removed no longer necessary TESTs on extsb/w, rlwinm, rlwnm, rlwimi, cntlzw
* Fixed bug where the SO flag was not being computed for compares
* Fixed flag computations for mulhw/mulhwu/mullw
* Fixed subtlety of shifts between 32 and 63 in srw/sraw
* Fixed mffs/mtfsf to use FP registers
* Fixed mtfsfi to use the immediate value properly
* Now marking terminal count bit in DMA status register
* Added simple symbol table to the UML for improved disassembly
* Changed optional disassembler cache parameter to a drcuml object
PPC DRC:
* Added symbols for most common variables
* Fixed bug in handling XER carry flag for subo. forms
* Simplified flag insertion logic for opcodeo. forms
* Added logic to simplify opcodes as much as possible at the UML layer.
* Removed similar logic in the x86 and x64 back-ends.
* Added stricter parameter validation for registers and mapvars
* Removed explicit flag requests from the shorthand opcodes
* Added optimization function to drcuml which is called at block end
* Added logic to compute the necessary flags based on upcoming opcodes
and only select those flags which are required
* Updated the PPC and MIPS3 DRCs to no longer explicitly specify flags
* Wrote new disassembler based on enhanced opcode info structure.
* Moved disassembler into drcuml.c and removed old code.
* Extended disassembler buffer sizes to at least 256 bytes.
* Added more extensive per-opcode information in preparation for UML
optimization step.
* Made validation more thorough using the extended information.
* Disabled back-end validation for now until it can be revisited
using the new tables.
* Changed GETFLGS encoding so that the mask is in parameter 2 instead
of the flags field.
----------------------
- 01845: [Core] all games: cheats are broken
- 01846: [Misc.] -CHEAT trigger causes MAME crash with PSX-type CPU drivers
Source Changes
--------------
Updated cheat system
- Locked search region when malloc fails to allocate gigantic memory
instead of fatal error. Also locked zero malloc error in case of
search speed = all memory in 32-bit CPU. And added search region
range checker to lock a search region with a big risk
- Fixed a bug that old code is always converted to specified
separator in enable/disable menu
- Refined new cheat format.
* Rewritten cheat format again. newest format is
":gamename::type::address::data::extend_data:(description:comment)"
Expanded type field now requires 10 bytes (not 8 bytes) and
it divides internal cpu (2 bytes) and type (8 bytes) parameters
* Removed VWrite, VRWrite. The flag of Value Selection is merged
into expanded type field now.
* Added CBit (Condition Bit). This checks bit in read data then
set or clear bit in case of true.
* Rearranged cheat code viwer based on newest cheat format
- Fixed unworking Load Old Format option. If you want to load old or
older format code, set it (default is OFF)
- Renamed functions/structures based on coding guidlines
All structures and included parameters are completed
- Merged resize_*_no_dispose() into resize_*()
- Removed the hack of search region for SH-2 CPU because required
region is now set as expected without this hack
Subject: [patch] memory_region madness
The memory_region and memory_region_length functions are probably the
two most common functions in MAME that don't take a machine parameter
but probably should given the direction MAME has been going in
removing global variable references. Attached are massive patches to
accomplish this. I wish they could be smaller, but sadly, this is
butchery, not brain surgery.
The first patch makes some simplifications to help the second patch
along. It is a general improvement as well, and hopefully can be
applied even if the second patch is rejected. Specifically:
1. Introduced/updated some include files for files that export
functions whose apis will need to be changed (cps1.h, decocrpt.h,
ms32.h, pgm.h, fd1089.h, konami1.h). In the case of konami.c, I
renamed the file konami1.c and changed the api and callers to only
require one function export.
2. Pulled memory_region*() calls out of for loops and folded the
occasional duplicated call. The compiler can't likely infer that the
results are constant, so this should be a minor performance win as
well.
- rewrote PowerPC implementation as a dynamic recompiler on top
of the universal recompiler engine
- wrote a front-end to analyze PowerPC code paths and register usage
- wrote a common shared module with C implementations of tricky
CPU behaviors
- added separate CPU types for the variants supported, instead of
relying on a hidden model enum
- rewrote the serial port emulation for the 4xx series to be more
accurate and not rely on separate DMA handlers
- rewrote the MMU handling to implement a software TLB that faults
in pages and handles changed bits appropriately
- implemented emulation of the PowerPC 603's software TLB, which
allows the model 3 games to run without a hack to disable the MMU
Updated the PowerPC disassembler to share constants with the rest of
the core, and to more aggressively use simplified mnemonics, especially
for branches. [Aaron Giles]
Universal recompiler:
- fixed frontend to handle opcode widths different from bus width
- added several new opcodes:
* (D)GETFLGS - copies the UML flags to a destination operand
* FDRNDS - rounds a double precision value to single precision
- renamed several opcodes:
* SETC -> CARRY
* XTRACT -> ROLAND
* INSERT -> ROLINS
- consolidated the following opcodes:
* LOAD?U -> LOAD
* LOAD?S -> LOADS
* STORE? -> STORE
* READ?U -> READ
* READ?M -> READM
* WRITE? -> WRITE
* WRITM? -> WRITEM
* SEXT? -> SEXT
* FTOI?? -> FTOINT
* FFRI? -> FFRINT
* FFRF? -> FFRFLT
- removed some opcodes:
* FLAGS - can be done with GETFLGS/LOAD4/ROLINS
* ZEXT - can be achieved with AND
* READ?S - can be achieved with READ/SEXT
- updated C, x86, and x64 back-ends to support these opcode changes
- updated disassembler to support these opcode changes
MIPS3 dynamic recompiler:
- updated to use new/changed opcode forms
- changed context switch so that it only swaps a single pointer
Konami Hornet changes: [Aaron Giles]
- updated to new PowerPC configurations
- updated some memory handlers to be native 8-bit handlers
- cleaned up JVS implementation to work with new serial code
- added fast RAM for the work RAM to give a small speed boost
Konami GTI Club changes: [Aaron Giles]
- updated to new PowerPC configurations
- updated some memory handlers to be native 8-bit handlers
Konami Viper/ZR107 changes: [Aaron Giles]
- updated to new PowerPC configurations
Sega Model 3 changes: [Aaron Giles]
- updated to new PowerPC configurations
- reimplemented/centralized interrupt handling
- these games are broken for the moment
Fixed crasher due to some Konami games using 8 layers in
the K056832 implementation, even though it was only written
for 4. [Aaron Giles]
Added fisttp opcode to i386 disassembler. [Aaron Giles]
- Added Microbus support, clock divisor selection, and CKO mode selection to the COP420
- Changed Thayer's Quest keyboard interface to at least slightly resemble the schematics
- Styling changes
- Added instruction cycle counts.
- Added interrupt support.
- Added HALT support.
- Fixed INT instruction.
- Added incomplete support for DIV and MUL instructions.
- Fixed MOV [#nnnn],BA and MOV [#nnnn],HL instructions.
* added NODE_RELATIVE(NODE, offset) macro to discrete.h
* fix asteroid.c, canyon.c, skydiver.c and tank8.c to use this macro instead of adding directly to the node
fixed Testers bug [01039]
-------------------------------
Updated cheat sytem drastically
* Supported (preliminary) new cheat format
- Added new operations
PDWWrite, RWrite, VRWrite, CWrite, Move, Branch, Loop, Popup
- Added new options
ValueSelectNegative, Return, AddressReadFrom, DataReadFrom
- Added cheat variables
You can store data with Write or Move operations and
read as an address or data from each fields
- Added condition check
for CWrite, Branch and Popup operations
- Added new custom codes
Separator, Layer
- Supported Shared Code [MESS ONLY]
different machine shares common code for a game
* Allowed mixed label-select or user-select with standard code
* Added label selector menu for label-selection code
* Added extend comment displayer for multi-line comments
* Added command menu for cheat list and watchpoint
Easy selectable cheat commands in this menu
* Added debug view menu [DEBUG BUILD ONLY]
It displays internal info to debug cheat system
* Supported unique separator by ui_draw_menu() in several menus
* Added user defined search region
You can set free memory range from cheat database
* Enhanced error check for cheat code to prevent from unexpected working
Error code is locked. If you select this code, the analyser works
instead of the code activater
* Enhanced value input with keyboard-less
Cheat edit key gives new input system in several menus
Left/right keys moves the edit cursor and up/down keys changes a value
* Added cheat command save, reload, reset in option menu
* Supported address shift in searching
Now searchable default region in TMS340*0 cpu games (Trog, Smash T.V. etc)
* Refined cheat menu handling with function pointer called cheat_menu_handler
* Added format strings table to manage in an integrated fashion
* Deleted unused hacks of default search region for Neo Geo and TMS34010
* Refined code based on MAME coding guidelines
- Renamed many functions/structures/constants
But not complete due to too many items
- Converted small but often used functions to INLINE
- Fixed a comparison in case of 0 or NULL
- Added missing function prototype for several functions
names are generated.
Added new astring functions astring_del, astring_replace, and astring_replacec
to help perform simple search/replace substitution.
1. In the MIPS core:
- renamed struct mips3_config -> mips3_config
- updated all drivers to the new names
- removed MIPS3DRC_STRICT_COP0 flag, which is no longer used
- a few minor cleanups
2. In the CPU interface:
- added new 'intention' parameter to the translate callback to
indicate read/write/fetch access, user/supervisor mode, and
a flag for debugging
- updated all call sites to pass an appropriate value
- updated all CPU cores to the new prototype
3. In the UML:
- added new opcode SETC to set the carry flag from a source bit
- added new opcode BSWAP to swap bytes within a value
- updated C, x86, x64 back-ends to support the new opcodes
- updated disassembler to support the new opcodes
4. In the DRC frontend:
- fixed bug in handling edge case with the PC near the 0 or ~0
Changed error reporting during input port detokenization to fill a buffer
rather than fatalerror-ing immediately. Should now properly skip over
any invalid tokens.
Enhanced error detection during input port detokenization to catch
duplicate bits. There are a lot of these!
Updated initialization code to print errors and fatal only if the input
ports were unable to be constructed at all.
Typedef'ed a real eeprom_interface structure.
Added a PORT_CUSTOM for the eeprom bit reading. Updated several
drivers to use that by specifying it in their input ports instead
of having a custom read handler that inserted the bit forcefully.
Added a PORT_CUSTOM for the ticket dispenser callback as well.
Updated the dcheese driver to use it.
Many more drivers can be updated to use these new PORT_CUSTOMs
in the future, eliminating hacky input port read handlers.
mode DIPs on the top of the menu
Fixed behavior of toggle switches so that they don't lose their value
when the UI is up. They also can now be used for multibit DIP switch
settings in which case they toggle through all the options.
New functions input_field_select_next_setting() and
input_field_select_previous_setting() which can be used to iterate
properly through DIP switches. Fixed the behavior for cases where
conditional ports are in play (you could get stuck). Changed uimenu.c
to call these instead of implementing its own.
Changed uimenu.c so that hitting ENTER on a DIP switch resets it to
its default value. This is analagous to how the OSD sliders behave.
- Reworked the color routines switching to resnet system.
- Added a resistor network diagram.
- Switch to pre-defined crystal value.
- Changed the WATCHDOG_TIME_INIT to be based on miliseconds instead of hertz.
- Other minor cleanup/fixes.
- Updated technical notes.
Subject: [patch] Missed Machine->machine changes
Hi mamedev,
A short incremental to my last deprecat.h patches. This one is the
result of a slightly improved fixup script, which finds more cases
where Machine can be replaced with machine. Six more files lose
deprecat.h.
~aa
Subject: [patch] More Machine->machine changes, add machine to irq
callbacks
Hi mamedev,
Here are two more patches to eliminate Machine globals. The first
patch was autogenerated by the attached fixup script. That script has
been updated to catch additional cases which it previously missed
(when Machine is the last parameter to a function or Machine is used
in an assignment). This makes ~50 more files deprecat.h free.
A sizable chunk (~20%) of the remaining uses of the Machine global in
the drivers are due to irq callbacks for sound and machine updates.
Typically such callbacks need to call cpunum_set_input_line, which
requires a machine parameter, so if the callbacks don't pass the
machine parameter, these routines have no choice but to reference the
global variable.
The second patch attempts to address most cases of this by adding the
machine parameter to the callback interfaces. This allows us to
remove #include "deprecat.h" from ~150 files, at the cost of having to
fix up hundreds of callbacks.
In total, these patches reduced the number of files with deprecat.h
from 783 to 575.
~aa
Subject: saturn CPU patch
Dear MAMEDev,
The attached patch corrects several bugs in the emulation of the saturn
CPU. These corrections are needed to make new HP48 drivers (to be
submitted to MESS imminently) work.
AFAIK, MESS is the only place where this CPU is used, so, the patch
should
not cause any regression in MAME.
The patch is against 0125u1.
Best regards,
-
Antoine Mine
* Added new opcode LZCNT which returns the number of leading zeros in
a parameter.
* Added new opcode XTRACT which is a combined rotate/mask (basically
rlwinm from PowerPC)
* Added new opcode INSERT which is a combined rotate/mask/blend
(basically rlwimi from PowerPC). Best. Opcode. Ever.
* Updated all back-ends to support these new opcodes.
* Fixed several bugs relating to shifts/rotates and optimizing out
cases incorrectly.
MIPS3 DRC changes:
* Updated to use INSERT and XTRACT where appropriate
* Cleaned up register usage to enable an additional direct mapping;
this means Linux gets 1 now and Windows gets 3
machine registers if the machine supports it. Currently only x64 on
Windows has enough free registers to do so, though PowerPC will almost
certainly be able to take advantage of this. Gives a minor speedup.
* Fixed front-end so that virtual no-op instructions are still targeted
as branch targets.
* Fixed front-end to mark the beginning of each sequence as needing TLB
validation, since any sequence can be jumped to from anywhere.
* Redid the MIPS3 TLB implementation. Fixed the exception vector and
type handling. Changed the bitfields to directly map from the MIPS TLB
format. Added distinction between TLB fill and TLB valid/modified
exceptions.
* Added separate modes for user, supervisor, and kernel modes. Each mode
does proper verification of addresses now and generates address errors
for invalid accesses.
* Fixed several bugs in the TLB implementation; not everything works
yet but it's a lot closer.
* Made COP0 access checking mandatory in non-kernel modes.
* Fixed several crashes when recompiling virtual no-ops.
* Fixed TLB bug where entries for virtual address 0 were present by
default.
* Fixed bug in the map variable implementation that would sometimes
result in incorrectly recovered values.
This will be expanded in the future.
Added two new opcodes: SAVE and RESTORE to save and restore the entire
virtual machine state for examination/setup.
Added new back-end function get_info() which returns information from
the back-end about how many actual registers will be mapped.
Fixed a bug that mapped the high a low parts of registers to the same
address. This should help the C back-end run better on big-endian
architectures.
