We can't rely on the OS returning files and directories
in a deterministic order, so we should sort them on our own
if we want VolumeDirectory to work for movies and netplay.
This fixes a bug which caused Movie (input recording or playback) or
netplay not to be stopped. DolphinWX previously triggered a STM power
event, and then the STM directly stopped the emulation; the code
which stops Movie/Netplay was completely skipped.
This is fixed by moving it /before/ sending the shutdown event.
This allows removing DSPCore and DSPTables includes from the header file.
Doing allows resolving quite a bit of indirect includes that were present
throughout the DSP source files.
Another plus with this is that changes to the DSPEmitter don't require an
almost total rebuild of all DSP source files. The underlying reason for
most of the files being rebuilt it because DSPMemoryMap is used quite
extensively, however its header includes DSPTables.h. DSPTables.h includes
DSPEmitter.h as it uses the DSPEmitter type in a typedef. So any change to
the emitter would propagate through the DSPMemoryMap header. This will no
longer happen.
This is actually used as the DSP JIT, so this should be with the other JIT
source files.
This commit also makes it so changes to the JIT emitter don't require
recompiling all of the DSP core (i.e. changing the JIT won't require the
interpreter to be rebuilt).
Jit64 inherits from Jitx86Base which inherits from JitBase. JitBase
contains jo and js, which are instances of the JitOptions and JitState
structs. Because of the inheritance, there's no actual need to access the
jit global in order to get to these instances. They're already accessible
via the class hierarchy.
This happened when the geometry shader was disabled, and the uniform
buffer was grown to a larger size. The update would be skipped, leaving
the old buffer to be included in the descriptor set.
OnBootDrive used the "drives" member std::vector for drive paths, but
since PR #4363, this vector is not populated anymore, so we were
accessing it out of bounds.
Actually, drives was not needed in the first place, since we can
get the wxMenu from the event, and from there, get the label directly.
Much of Jit64Util consists of essentials, not utilities. Breaking these
out into their own files also prevents unrelated includes from being
present near other classes.
This also makes it easier to find and change certain components of the
x86-64 JIT, should it be necessary.
These provide the same semantics, however aggregate initialization
doesn't force the structs to be trivially copyable. memset, on the other
hand, does.
JUTWarningConsole_f calls vprintf, but in a way we currently don't
handle (which messes up the printed message). However, it is a standard
debug print function, so we can directly hook it instead of waiting for
the vprintf call.
This is necessary to fix debug output in a few games now that vprintf
is properly detected in more games.
This is more logical as the mic is plugged into an EXI slot so it should be configured via the GameCube config dialog. This also allows to pass the right port number for the new dialog.
This also moves the pipeline and descriptor set layouts used for texture
conversion (texel buffers) to ObjectCache, and shares a binding location
with the SSBO set.
Makes the information panel self-contained.
This was done first, as opposed to isolating the GameConfig panel--the
first panel in the group--as this panel had code all over the place in
ISOProperties, so I figured it'd be best to fix this one up first.
These disc images are encrypted and signed using a different set of keys.
We only care about the master key, so we check the signature issuer. If
it matches the debug issuer, then we'll use the RVT-R key. Otherwise,
the previous set of common and Korean keys are used.
The SDL backend crashes when you close a joystick after SDL_Quit has
been called. Some backends don't need to be shutdown and
re-initialized everytime, we can just ask to enumerate devices again.
It's questionable whether ES_LAUNCH should write *anything* to the
command structure at all, ever, given that it never actually returns
it back through the mailbox. But it *definitely* shouldn't write
anything to it if it has just launched a DOL, because otherwise it might
clobber code/data from the just-loaded application.
When booting "cooked" executables, BATs should already be set up and
enabled. They should only really be disabled when booting NAND contents
in real mode.
Should we ever introduce anything else that has to be done when a command
buffer is executed (e.g. invalidating constants from previous commit), we
don't have to update all the callers.
This could happen because the vertex memory was already committed, if a
uniform buffer allocation failed and caused a command buffer to be
executed, it would be associated with the previous command buffer rather
than the buffer containing the draw that consumed these vertices.
This was happening when a fence wait happened mid-frame. The data written
between the fence being queued and the allocation occuring was incorrectly
assumed to be consumed by the GPU.
Hotkeys
Make a new class that inherits from InputConfigDialog with a specialised constructor. The changes are mainly the top portion and it now uses tabs to categorise the hotkeys.
Redo the GCPad configuration dialog
The layout is similar, but it now allows flexibility to change it more easily.
Redo the GC Keyboard configuration dialog
Same layout.
Redo completely the Wiimote configuration dialog
Separated the controls into 2 tabs to make them less imposing overall.
Redo the Nunchuk configuration dialog
Similar layout, except for 2 control group sizers.
Redo the Classic controller configuration dialog
Same layout.
Redo the Guitar input configuration dialog
Stacked 2 sets of group together.
Redo the Turntable configuration dialog
More stacked groups and the window is much less wide.
Just setting up a switch on the type so that different dialogs can be instantiated. This also makes the extension type an enum because I don't see why not here and finally, it removes ControlGroupSizer. This removal allows to not dynamically generate the UI, but instead, let the specialised constructors do the layout.
Removed the unecessary forced tabbed layout, removed the layout part of the constructor and remade some method in preparation for tabbed styled input dialog such as the new hotkey configuration one. It breaks every inputconfigDialog, but this will get fixed in the next commits.
Also moved to a folder since there will be many more files created in the next commits so it gives better separation.
Making changes to ConfigManager.h has always been a pain, because
it means rebuilding half of Dolphin, since a lot of files depend on
and include this header.
However, it turns out some includes are unnecessary. This commit
removes ConfigManager includes from files which don't contain
SConfig or GPUDeterminismMode or GPU_DETERMINISM (which means the
ConfigManager include is not used).
(I've also had to get rid of some indirect includes.)
Prevents path traversal without needing an absolute path
function, and also improves accuracy (character sequences
like ../ appear to have no special meaning in IOS).
This removes the creation and usage of /sys/replace,
because the new escapes are too complicated to all
be representable in its format and because no other
NAND handling software seems to use /sys/replace.
This reverts commit 141f3bfb3a.
The implementation of getting absolute paths wasn't working
on non-Windows systems, which is a huge problem for IOS HLE.
For hotkeys, changed HotkeyManager to allow to get and make partial groups of hotkeys.
Also preserved the old configuration naming scheme for the ini, this is done to preserve compatibility with the older groups structure.
Add the ability to get GCPad control groups
Used like the HotkeyManager methods, this is used for the new GCPad configuration dialog.
Add the ability to get groups of Keyboard input
Same reasons as the previous ones.
Add ability to get groups of Wiimote input
Add the ability to get extensions group
This needed to pass to 3 classes. Will be used for their respective dialogs.
I know there is already #3521, but it currently needs a rebase and I
needed to add something to IPC_HLE_Device properly, that is, without
putting everything in the header, so this commit cleans up
IPC_HLE_Device first. (And only IPC_HLE_Device: the rest will still
be handled by #3521.)
Also fixes a few indirect includes (removing unused header includes
from IPC_HLE_Device.h broke building)
This is something that was copy-pasted across the IPC_HLE code
(because it's often used). Since all of the duplicated pieces of code
do the same thing as the previous EnqueueReply, except that they also
write to command_address + 0 and + 8 (to write the correct reply type),
this commit changes EnqueueReply to do that instead of having it
duplicated all over IPC HLE.
It was apparently causing heavy slowdowns on game even though it wouldn't spam much, probably caused by the amount of additional check caused by the logs levels changes.
This is mainly for potential Android fifoci usage, and thus is not
exposed anywhere in the UI. To enable, set DumpFramesAsImages under
Settings in GFX.ini.
When the emulated BT device is created, m_HCIEndpoint (which is a
CtrlBuffer)'s m_cmd_address is not initialised to 0. So it ends up
being a random value. This is normally not an issue… but the
emulated Bluetooth code relies on m_cmd_address to know whether the
HCI endpoint is still valid.
This is a problem with ES_Launch, because the bt_emu class is
destructed and re-constructed, and while m_cmd_address is still
uninitialised, the ES_Launch code disconnects all Wii remotes,
which triggers a HCI event and hence the bug.
Instead of resetting two command buffers, now we only have to call
vkResetCommandPool once at the start of a frame.
NV's recommends using one pool per frame/thread. May offer a very small
boost in performance on some systems.
%n writes to a pointer that's provided as a parameter.
We didn't have a custom implementation of this before,
meaning that %n would trigger a write to the host
memory instead of the emulated memory!
The bounds checks in IOCtl were using 0x200 as the size of
m_Registers, which is more than the actual size, 0x200 / 4.
This commit turns m_Registers into an std::array to allow
for a correct and obvious way of getting its size.
anv seems to set this to zero, which is fine according to the spec, but
we were using it as a maximum, which was resulting in a swap chain
without any buffers being created.
This fixes the screenshot stutter, as this needs more than a frame.
So we won't stall on the png writing at all until emulation stops or
a new screenshot is requested.
This increase the performance of good backends a bit, but slows down the bads one a lot.
Let's fix those backends instead of forcing stupid memcpy in the common code.
Makes for a cleaner separation of functionality, as well as removing
multiple includes from the main header file. It also gets a bunch of
structs and enums out of the global namespace.
Coincidentally, this also gets rid of an indirect include cycle that
could have broken compilation of Core.cpp in the future, since it was
relying on IPC network includes to resolve functions in Common/NandPaths.h.
This makes it easier to separate out the individual net classes in a
follow-up. Separating these out would also make it less of a pain to
figure out what's going on, since you wouldn't need to sift through 1000+
lines of code.i
This used an invalid pointer, which was only valid within AddFrame.
This drops a feature which shall dump the last frame as it might was dropped before.
A good implementation however should "overwrite" the last frame if the time matches.
But this needs to delay every frame a bit.
This eliminates public usage of the GetMenuBar() function in CodeWindow.
The benefit of this is it also gets rid of the need to perform direct
access across the config dialog and the main frame. It also gets rid of
the use of the main_frame global.
GetMenuBar() will be removed entirely from CodeWindow in a follow-up that
also removes any related remnants of code made obsolete with its removal.
Gets rid of more menu-related code from CodeWindow and puts it back in
CFrame where it belongs.
This turns the previous menu update function within CodeWindow into one
that simply updates the debugger font for its managed controls. It also
improves how the font is actually updated. Previously, fonts would change,
however this wouldn't actually reflect onto the respective controls until
a refresh or update event occurred. Since codeview, callstack, symbols,
callers, and calls windows are all managed by a wxAuiManager instance,
calling Update() on it after the font has been set will reflect font
changes immediately.
There's not a lot of point in passing these around or storing them
(texture cache/state tracker mainly) as there will only ever be a single
instance of the class.
Also adds downcast helpers such as Vulkan::Renderer::GetInstance().
In the code view, it would never say r1 or r2, but rather sp (stack pointer) and rtoc (register of the table of content) respectively. In the register view, all it says is the register number. This is an inconvenience considering it might not be obvious which register belongs to which of these terms.
Also make r13 named the "sda" for small data area with the same convention as above.
This not only fixes a regression where toggling a breakpoint using the CodeWindow would cause a Center PC, but it also removes several redundant JumpToAddress(PC) calls.
On Linux, the FindFocus method from wx simply doesn't work, it would on some environment report that dolphin has the focus while it doesn't have it. This is why an alternative method has to be used which is to set a focus flag whenever the render frame gets activated.
Some adapters don't have the correct interface class, so they are not
recognised as Bluetooth adapters. It seems that apart from hardcoding
VIDs/PIDs (which is how it's done in the Linux kernel, and which I'm
not very fond of), there is no other way to detect if a device is a
Bluetooth adapter or not.
This change makes Dolphin skip the descriptor check when trying to find
a usable adapter for Bluetooth Passthrough if the use of a specific
adapter was forced; it is assumed that the user knows what they are
doing if they hand-edited their config file.
This allows such adapters to be used.
The usage of "Wii Remote" and "Wiimote" in the interface is inconsistent. "Wiimote" is also not a real word nor is it an official product name. Therefore I have changed instances of "Wiimote" in the UI to instead say "Wii Remote". I also made a couple of minor grammatical changes as well.
This is mostly a resubmission of #4338 but there are some minor other changes as well.
I didn't know that telling that you don't schedule from the CPU thread prevents an assert because it by default assumes you use the CPU thread, but in the case of ClearCacheThreadSafe, it's used from the GUI thread.
Commit 4969415 modified calls to GetInterpolationQualifier, but mistakenly changed the order of some boolean parameters: GetInterpolationQualifier(true, false) was changed to GetInterpolationQualifier(…, false, true).
Should fix#9783.
Currently, `g_controller_interface` is initialized and shut down by each
of `GCKeyboard`, `GCPad`, `Wiimote`, and `HotkeyManager`.
This 1) is weird conceptually, because it necessitates passing a pointer
to the native window to each of those classes, which don't need it, and
2) can cause issues when controller backends are initialized or shutdown
multiple times in succession.
For Wii graceful shutdown to work, the emulated software has to open
the STM event hook and install a hook. Without this, there is no way
to inform them about the shutdown, so trying to do a graceful shutdown
and requiring the use of the shutdown fallback (exiting a second time
to force) is pointless.
Previously Dolphin would only exit if the main window is closed,
and Confirm on Stop is enabled.
This makes Dolphin's behaviour more consistent by always exiting
if the main window is closed or on shutdown signal.
s_dvd_thread_done_working makes the logic more complicated,
and degasus pointed out a race condition that can happen if
the CPU thread calls WaitForIdle right in between the DVD
thread executing done_working.Set() and done_working.Reset()
while there is work left to do. To avoid this, let's just get
rid of s_dvd_thread_done_working. It's a relic from the old
DVDThread design. Thanks to the last few commits, WaitUntilIdle
only gets called rarely (disc change and savestate), so it's
not a problem if WaitUntilIdle ends up being slower.
This is a preparation for adding a queue to DVDThread.
Currently, s_read_request and s_read_result act somewhat like
queues that only can contain one object.
This would previously fail to compile when included in files that do not
include FileUtil.h due to lack of a type declaration.
This moves the constructor and destructor into the cpp file in order to
satisfy the requirements of unique_ptr construction and deletion. That is,
unique_ptr requires a concrete type at the point of construction and
destruction. If the constructor or destructor is left in the header, then
at the point of construction or destruction, IOFile will still be
considered an incomplete type, as unique_ptr's deleter will still only be
able to see the forward declaration, which it can't use.
Unless I'm misreading the code, it doesn't look like this serves any
purpose, and is only polluting the logs.
_Unimplemented_Device_ looked like a device name that was picked to
be used somewhere else in Dolphin, but this doesn't seem to be the case
since 2012 (d95e31a removed the only other usage of this fake device).
Unifies the creation of all the menus into the main frame class.
Now it isn't spread out across the main frame and the code window.
This doesn't alter the placement of the handling functions, as this would
involve unrelated changes, since it would require modifying where
window-related variables are placed. This will be amended in a follow up
changeset.
Keeps related menu items together based on top level menu. This will be
more convenient in the future when debugger menu bar item handling is
moved to CFrame, as it won't be a huge amount of code in one function.
This also makes it easier to locate menu bar code whenever it needs to be
changed.
4319 made Dolphin not read/write directly to the SYSCONF and read
settings from the SYSCONF at boot, and only write Dolphin settings
to the SYSCONF at emulation startup.
However, this also made it a bit confusing, because if settings were
changed, then Dolphin was exited without starting a game in between,
the settings wouldn't actually get persisted. This is fixed by
syncing Dolphin settings with the SYSCONF when Dolphin exits.
Instead of directly reading/storing settings from/to the SYSCONF, we
now store Wii settings to Dolphin's own configuration, and apply them
on boot. This prevents issues with settings not being saved, being
overridden and lost (if the user opens a dialog that writes to the
SYSCONF while a game is running).
This also fixes restoring settings from the config cache after a
graceful shutdown; for some reason, settings were only restored
after a normal shutdown.
Fixes issue 9825 and 9826
Using u8 as indexers is kind of silly, since the rest of the public API
essentially uses int for this sort of thing. Changing these to int also
gets rid of quite a few implicit truncations.
This also allows for getting rid of similar silliness in the netplay API.
There's no official implementation of the Vulkan API,
and Dolphin currently isn't set-up to work with the
single, commercially-available third-party implementation.
This is done to remove confusions among potential debugger users and to also make it more accurately tell what this feature is actually doing. Despite being true that it is using a memcheck (and it certianly checks that memory), the idea being to break on a memory access isn't really obvious especially considering that memchecks are also used in full MMU emulation to handle exceptions. It also doesn't help that memchecks are now supported in every builds.
It also changes the corresponding log because this log would be wanted by the user which means it should be more obvious that it was caused by the "memory breakpoint".
This is an issue because a driver may have to maintain two copies of a
texture if it batches all uploads together at the start of a frame.
In the Vulkan backend, we do something similar to avoid breaking out of a
render pass to copy a texture from the streaming buffer to the destination
image.
This was causing issues in the sms-bubbles fifolog, where an EFB copy to
the same address of a previously-used texture caused the previous texture
to be re-used again for a different image later on in the frame, causing
the original contents to be discarded.
Using cmake and GCC, logs would contain the full file path when logging making logs lines unnecessarily long. This is solved by just removing anything before "/Source/Core/" (where / is whatever your OS uses to separated directory).
This happened when the source texture was an EFB copy, therefore it had
not been populated prior to the draw command buffer being executed, and
the conversion was occurring in the init command list.
Setting a single icon at a single resolution doesn't scale well,
Windows requires a 16x16 icon for the window and a 32x32/48x48 for
the taskbar. Providing all icons produces less pixellated results at
HiDPI.
Changes:
- MemoryWindow was cleaned up and gives more feedback on searches.
Some bugs were fixed as well:
- A complex bug that allowed tearing off tabs and opening multiple
copies of a debug panel which lead to segfaults
- Another segfault related to right-click menus on code/memory views
when those tools were floating in their own window.
Resolved "TODO" for Texture Cache safety, added explanation message.
Resolved "TODO" for default description, no longer uses default text for sizing
Fixed a memory leak in PostProcessingConfigDiag where it was never freeing any
of the objects it allocated in its constructor.
Minor design change to PostProcessingConfigDiag to give padding around elements
consistent with the rest of Dolphin's user interface (5px).
Changed the Cheat Search tab to disable the scan buttons while there is
not a game running and enable when it starts. Also added double-click to
create code to the result list.
Required a partial rewrite of the image loading code because it was working in
unscaled wxBitmaps. Needed to make it produce wxImages and scale them instead.
And related ARCodeAddEdit/PatchAddEdit.
Change ISOFile to use wxImage instead of wxBitmap since bitmaps require
a screen context and banner images have a fixed resolution.
Some OSD messages were displayed in RenderBase.cpp using global variables and some code duplicated
in OnScreeDisplay.cpp.
Now all messages are displayed using functions in the OSD namepace.
* OSDChoice and OSDTime global variables are gone
* All OSD logic is kept at the same place
* All messages are properly aligned
* Clean characters for all OSD messages
Original commit:
commit f0ec61c057
Author: Aestek <thib.gilles@gmail.com>
Date: Sun Aug 7 16:08:41 2016 +0200
Add a context menu entry in main game list to host a netplay game
based on saved settings.
Original commit:
commit 91aaa958e6
Author: Aestek <thib.gilles@gmail.com>
Date: Sun Jul 24 14:51:37 2016 +0200
There was a bug that caused MemoryView to indirectly cause a segfault;
the simplest way to reproduce it is 1) start a game; 2) stop the game;
3) click on the Refresh button and watch Dolphin segfault.
This commit fixes it by only calling PowerPC::HostIsRAMAddress when
emulation is running.
It doesn't make much sense to prefix g_ to static variables and for
some to be completely unprefixed.
Also renames a lot of other variables for the new conventions
This fixes an issue where the Bluetooth info section could be fully
filled up by syncing 5 Wiimotes in passthrough mode then switching to
emulated Bluetooth; emulated Wiimotes were then unable to be used.
The "real" SYSCONF section is now backed up before being replaced with
a blank section that the emulated BT adapter can always fill with 5
Wiimotes without issues.
This backup is restored by the passthrough code, instead of during
the Bluetooth mode switch because this should be done regardless of the
user interface, and even without UI (if the config file is edited
manually).
Homebrew programs seem to rely on getting a reply to the vendor
specific commands, without which Bluetooth initialisation will never
complete.
This vendor-specific command is typically used to patch the Wii's
Bluetooth module, so the replies are only faked when the passed through
adapter is not a Wii Bluetooth module.
This stores the address of paired devices and associated link keys.
It is needed because some adapters forget all stored link keys when
they are reset, which breaks pairings because the Wii relies on the
Bluetooth module to remember them.
It doesn't fix adapters that can't remember any link key
at all and always return 0 for the number of stored/written link keys.
For those adapters, there is no fix.
This also improves the usability of passthrough mode for adapters that
already work, since pairings will now keep working even if the link
keys get cleared by something else (for example by the host Bluetooth
stack).
The ControllerConfigDiag design was getting confusing, so more
significant changes needed to be done.
Firstly, the GC controller and the Wiimote section layouts have been
aligned for consistency.
The Balance Board source chooser is a checkbox.
The "general settings" that affect the SYSCONF have been moved to the
Wii pane in the Config dialog. It makes more sense because those
affect the Wii's settings in the NAND, unlike the other options.
Another reason for moving it is that the Controller Config Dialog was
getting pretty crowded, and the whole section is disabled when
emulation is running, which is wasted space.
The Wiimotes section is now organised by two radio buttons. One is for
the Passthrough Mode, with sync/reset buttons under it; the other is
the emulated Bluetooth mode, which still has the regular Wiimote source
choosers, the Continuous Scanning controls and the Enable Speaker Data
option (which only applies to the emulated BT mode).
Hopefully this should make things a bit clearer and look cleaner.
(This is a monolithic commit because separating UI changes is hard)
Apparently, Nintendo's Bluetooth stack expects the ACL packet buffer
size to be limited to 10.
Reporting anything higher than that could cause memory corruption,
which can result in warning messages in the logs ("event mismatch"),
and more annoyingly, random disconnects.
Thanks to shuffle2 for the fix!
This adds the ability to passthrough a whole Bluetooth adapter and skip
the majority of the Bluetooth emulation code. We use libusb to send HCI
commands, receive HCI events and transfer ACL data directly to the
first adapter that is found or to a specific adapter (if configured to)
This is possible because the Wii's Bluetooth module is actually just
a pretty standard Bluetooth adapter…
…except for two vendor-specific commands, for which replies are faked,
and also for the sync button. This adds a hotkey that works in the
exact same way as the sync button would on a Wii: it triggers an HCI
event, which emulated software interpret as a command to perform
a BT inquiry.
This commit also changes the UI code to expose passthrough mode
and WII_IPC_HLE to be a bit more thread safe (for the device map).
This moves the unordered_map used to store connected Wiimotes IDs to
WiimoteReal, and makes the ID insert/erase logic common so we don't
have to duplicate this code in scanner backends.
The Balance board detection logic is already implemented in a simpler
way in Wiimote::IsBalanceBoard() (since hidapi also needs it).
Therefore, IOWin now only needs to check if a device is a Wiimote.
This is intended to make reconnecting Wiimotes easier with a DolphinBar.
Unfortunately, this change isn't enough as it doesn't always catch
disconnections for Wiimotes connected with a DolphinBar.
But it's better than nothing and eventually a disconnection will be
detected when something tries to write to the Wiimote, instead of never.
There is no other solution as the DolphinBar always exposes 4 HIDs even
when the associated Wiimotes are not connected.
We could try to detect this using the fake input reports sent by the
DolphinBar, but this only works for the first HID (probably because of
a bug in the firmware?), so this method is not an option.
If FindWiimotes() took more time than the UI shutting down, the scanner
would try connecting a Wiimote and sending an event to the UI code
long after it has shut down, which causes a segfault.
This fixes the race by ignoring any found Wiimotes during shutdown.
Normally this would have never happened, but it is possible with hidapi
since Wiimotes can be connected before Dolphin starts.
wxWidgets causes a segfault if Host_ConnectWiimote is called and we try
to create an event from the Wiimote scanner thread while the GUI is
still initialising.
Based on ca0c2efe7a. Credits go to flacs.
However, unlike the original commit, hidapi does not completely replace
the current implementations, so we can still connect Wiimotes with 1+2
(without pairing).
Also, it is only used on Linux and OS X for now. This removes the
advantage of having only one implementation but there is no other
choice: using hidapi on Windows is currently impossible because
hid_write() is implemented in a way that won't work with Wiimotes.
Additionally:
* We now check for the device name in addition to the PID/VID so we can
support the Balance Board and maybe third-party Wiimotes too. This
doesn't achieve anything with the DolphinBar but it does with hidraw.
* Added a check to not connect to the same device more than once.
Should fix a possible reference to deleted framebuffers, as well as fixing
the issues with the render area being correct if the game's source area
changes, or auto-scaling is enabled.
Same thing but allows both GeckoCode and Code to be utilized directly
without predicates for equality/inequality in stardard algorithms
The size check for std::vectors is unnecessary, as this is built into std::vector's operator==
Because of the way this works, randomly overwriting the handler
when loading a savestate will break things because of the
self-modifying nature of the handler.
GeckoCodes require address hooks which don't correspond to any
symbol in the symbol table. The hooks get deleted when repatching
the game because they did not persist across calls to
HLE::PatchFunctions.
Dolphin emulates GeckoCodes by fiddling with the CPU state when a
VI Interrupt occurs. The problem with this is that we don't know
where the PC is so it's non-deterministic and not necessarily
suitable for use with the codehandler.
There are two options: Patch the game like Gecko OS either directly
or using HLE::Patch, or use a trampoline so we can branch from any
PC even if it would otherwise not be valid. The problem with Gecko OS
patches is there are 10 of them and they have to be configured
manually (i.e. Game INIs to would need to have a [Core]GeckoHookType
property).
HLE_Misc::GeckoReturnTrampoline enables the Code Handler to be
entered from anywhere, the trampoline restores all the registers that
had to be secretly saved to the stack.
If the installation fails because codehandler.bin is missing or
unusable then Dolphin will try again every single frame even though
it's highly unlikely a disk file will have changed. Better to just
fail once then only try again when the active code set is changed.
Suppresses generating 60 log messages per second.
Turns out one of the magic numbers was very magic. The gameid is
an ad-hoc comm protocol with HLE_Misc to control the number of times
the ICache is reset.
The code table builder cuts off the end of codes that won't fit
after already writing part of it. That seems quite unlikely to
work the way anyone would find useful since the codes can contain
actual PPC instructions.
The active codes vector cannot safely be used outside the mutex,
move the lock out into RunCodeHandler. s_code_handler_installed was
also racing against SetActiveCodes since it's being written both
inside and outside the lock.
General cleanup. Add s_ prefixes, use constexpr, remove C casts.
Executing PPC code inside an external events callback is a bad idea.
CoreTiming::Advance does not support recursion properly which will
cause timing glitches. The interpreter has a slice length hack that
counters this but without it this would cause a 20000 cycles time
skip. It isn't clear what this was supposed to accomplish that just
changing the current PC would not. Changing the PC works fine.
Deleting instead of overwriting makes the INI cleaner.
Also, in case we change defaults in the future, users will
get the new default when using a new version even if
they have pressed the Reset button in an older version.
