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.
Makes the buffering code a bit more explicit (circular buffer, but
blocks until individual buffers get unqueued by OpenAL), and fixes a
bug in the startup of Super Mario Sunshine:
https://bugs.dolphin-emu.org/issues/9811
When standby mode is enabled, this causes games to ES_Launch the system
menu instead of directly asking IOS (the STM more precisely) to shut
down, which prevents graceful shutdown from working
(it'll appear to hang).
Dolphin never supported WC24 standby mode anyway, so this shouldn't
cause any issues. (This should be reverted if and when WC24 standby is
implemented…)
This adds support for triggering the power event (in the STM), so that
stopping emulation first triggers a shutdown event, which notably gives
emulated software time to save game data (issue 8979) and clean up
SYSCONF (to disconnect Wiimotes and update their state in the SYSCONF).
On the first press, the stop button/hotkey/whatever will trigger a STM
power event. On a second try, we will forcefully stop emulation, just
like how it was working before.
There was as far as I know no reason to put everything in the header.
Separating the declaration from the implementation reduces build
times in case the implementation is updated without changing
any declaration.
This is needed because for some reason the WSI for NV Vulkan drivers
doesn't return VK_ERROR_OUT_OF_DATE_KHR, so there is no other way to know
that a resize has occured apart from polling, which is a poor solution for
X11 (since it is blocking).
Considering there's a public method in the class using it, leaving the
definition in the cpp file can cause a linker error if any method outside
that cpp file calls it for one reason or another.
Single step: Fix an oddity when a breakpoint is hit at the beginning of a block, then after, a single step is performed and finally, hitting play, the breakpoint will be skipped even in the case when it would be hit again. This was done by using the interpreter version of single step. Also, remove some redundant update request.
Step over: fix some GUI lags.
Step out: Add consideration for conditional branching by checking the condition as the interpreter does. Now, every bclr instructions except those that changes the LR (because it would not be the end of the function) will cause the end of the step out and not just blr instructions. Also now stops if a bp is detected and finally, remove redundant GUI updates calls.
This also removes a superfluous draw call on the GUI as the codeView was refreshing twice per event to do so.
It was never used, even when the code tried to make sure it was initialised and passed correctly. This is a supplementary fix for the memCheck dialog as this option will now work correctly.
The old one wasn't very optimal because not only the user would likely want to enter an address instead of a range, but it also made entering just one address confusing (you had to have the same value on both start and end). Also, you should only chose one option between read, write or both, there is no point to not have any.
This is why I made more clear how to add an address and it is the default option using radio buttons and I also made the action flags and the flags to be radio buttons.
It looks like the debug output is also output as SJIS (similar to
OSReport text), so we need to convert it to UTF-8 to prevent it from
all showing up as �.
This doesn't fix all display issues, but fixes all SJIS/UTF-8 related
ones.
This changes GetSymbolFromName to not require the passed name to
completely match with the symbol name. Instead, we now match
against the stripped symbol name (i.e. only the function name).
This fixes a regression introduced by #4160, which prevented
HLE::PatchFunctions() from working properly.
This is the same as PR #3991, but for MainNoGUI.
nogui/headless will shut down cleanly on SIGINT and SIGTERM, just like
it would when closing the render window.
The default signal handler will be restored after a first shutdown
signal so a second signal will exit Dolphin forcefully.
These functions don't actually depend on any state from the class
instance, so they don't really belong in the header, and are just
an implementation detail.
Fixes the issue on macOS where quitting Dolphin from the Dock causes a
crash report (https://bugs.dolphin-emu.org/issues/9794). I'm not
exactly sure why this works, but it feels right and it turns out to fix
the problem.
It didn't really made sense to disable 2 logs levels in releases builds while the level LDEBUG should really be where logs that would impact performance be. Info should be logs that report potentially usefull information and debug should report info that would only be usefull in debug context as they are called very often. To make this work, a lot of info log would have to be made debug log.
It also avoid inaccurate logs level done due to not using debug builds. While searching through the code, I saw a ton of logs that should have been info log, likely done to avoid using a debug build (which shouldn't happen considering the level debug exists anyway).
The whole idea is to have more meaningful logs in release builds while maintaining minimal performance loss from choosing the highest level. This could potentially help to diagnose issues or to know more about what the emulator is actually doing.
The next commit aims to sort the log levels for this purpose.
clang-format really *wants* the two empty lines to be removed;
otherwise, it will always flag MemoryUtil as needing formatting changes
which is an annoyance when it is used as a git filter driver.
clang-format really *wants* the two empty lines to be removed;
otherwise, it will always flag MemoryUtil as needing formatting changes
which is an annoyance when it is used as a git filter driver.
This adds a recenter control binding which allows recentering the
cursor when relative input is enabled.
(EnableSettingControl is renamed to avoid confusions.)
This makes the GUI show the settings that are loaded when the config
gets reloaded, instead of showing potentially outdated settings that
are not applied.
This adds an option to enable relative input for the Wiimote IR
as described in issue 9014.
Enabling it will result in the pointer not going back to the centre
and the inputs will control the direction, not the absolute position.
Also adds a Dead Zone setting which is really needed when relative
input is enabled to prevent the cursor from slowly drifting on
most controllers. (Note: the Deadzone setting has no effect when
relative input is disabled)
Call Advance at the start of each timing slice instead of the end.
Possibly fixed a bug where a slice would randomly branch straight
back to Advance() because the flags were not set to the right
values before branching to the dispatcher entrypoint.
This fix a double break bug when hitting a memcheck and hitting play on the same instruction it broke to earlier. The state is put back to CPU_RUNNING after.
Doing this forces the window to be drawn before reparenting it. It fixes the possibility of creating an empty floating window if the selected tab wasn't corespoinding to the window.
There might be a better way to highlight the options when adding the memory check, but for now, this works and the breakpoint list reports the right settings anyway.
These are needed for the next commit. I had to modify the implementation of the DSP one too, but since it basically isn`t used, I don`t think it matters much. These options only matters when adding one.
It wouldn't impact performance until at least one memcheck is enabled. Because of this, it can be used in release builds without much impact, the only thing that woudl change is the use of HasAny method instead of preprocessor conditionals. Since the perforamnce decrease comes right when the first memcheck is added and restored when the last is removed, it basically is all beneficial and works the same way.
All formatting are individual per registers and they all have one option to go back to their original hexadecimal form.
- GPR: signed integer, unsigned integer, float
- FPR: double
Also happened to come accross an issue where editing the PFR would ignore the higher 32 bits of the new value, this had to be fixed for the format to work.
The min-heap provides no ordering when the key is the same on 2
nodes. Disambiguate identical times by tracking the order items
were added into the queue.
Some games will set q to a different value than 1.0 through
texture matrix manipulations. It seems the console will still
do the division in that case.
This removes a Dolphin-specific patch to the wxWidgets3 code
for the following reasons:
* Calling wxWindowGTK::DoSetSize on a top-level window can end up
calling wxTopLevelWindowGTK::DoMoveWindow, which triggers an assert
because it is not supposed to be called for a top-level wxWindow.
* We should not be patching the wxWidgets code because that means the
toolbars will still be broken if someone builds without using the
WX that is in our Externals.
Instead, we now use a derived class for wxAuiToolBar and override
DoSetSize() to remove the problematic behaviour to get the same effect
(fixing toolbars) but without changing Externals code and without
causing asserts and other issues.
Now that our timings are much more accurate it doesn't look like we
need it anymore. And the instant ARAM DMA mode + scheduling fixes
ctually breaks ATV: Quad Power Racing 2 (causing all sorts of werid
bugs).
Fundamentally, all this does is enforce the invariant that we always
translate effective addresses based on the current BAT registers and
page table before we do anything else with them.
This change can be logically divided into three parts. The first part is
creating a table to represent the current BAT state, and keeping it up to
date (PowerPC::IBATUpdated, PowerPC::DBATUpdated, etc.). This does
nothing by itself, but it's necessary for the other parts.
The second part (mostly in MMU.cpp) is simply removing all the hardcoded
checks for specific untranslated addresses, and consistently translating
addresses using the current BAT configuration. Very straightforward, but a
lot of code changes because we hardcoded assumptions all over the place.
The third part (mostly in Memmap.cpp) is making the fastmem arena reflect
the current BAT configuration. We do this by redoing the mapping (calling
memmap()) based on the BAT table whenever it changes.
One additional minor change is that translation can fail in two ways:
either the segment is a direct store segment, or page table lookup failed.
The difference doesn't usually matter, but the difference affects cache
instructions, like dcbz.
For step over, it was updating twice which actually made the red display on the register view (when a register changes since) malfunction. Since it doesn't seem to be usefull to update before AND after the run, the one before the run was removed.
For step out, well, because there was no chances given for the thread to run as it is single stepping all the time, I only added a call to update after it was done.
Init cannot be called more than once because it registers the
CoreTiming callbacks, that trips the assertions and will cause
anyone with PanicAlerts disabled to crash.
CoreTiming gets restored before ExpansionInterface so CoreTiming
events need to already be registered before the save state loading
begins. This means that the callbacks must be registered
unconditionally instead of on-demand.
Replace adhoc linked list with a priority heap. Performance
characteristics are mostly the same, but is more cache friendly.
[Priority Queues have O(log n) push/pop compared to the linked
list's O(n) push/O(1) pop but the queue is not big enough for
that to matter, so linear is faster over linked. Very slight gains
when framelimit is unlimited (Wind Waker), 1900% -> 1950%]
Some compilers don't have an automatic abs() overload for floats.
Doesn't really matter if they use the integer variant here, but
it's better to be explicit about the fact that we're using floats.
Dolphin no longer lowers itself below the top window when opening.
Dolphin no longer draws garbage lines all over the game list.
Use the correct platform macros so Dolphin can actually find the
translation catalogs on Unix.
Both those in the code window and the ones that appears when the user select edit perpective. The one that isn't fixed by this is in edit perspective mode, when the user drags a panel out and it becomes a floating window.
When n was a multiple of 4, the old implementation would overwrite
the following register with 0.
This was causing Not64 to crash.
Thanks to Extrems for spotting this.
DVDInterface shouldn't need to know anything about
the DTM format's 40-character limitation.
Also replacing "filename" in variable names with "path"
to make it clearer which variables contain the whole path
and which ones only contain the filename.
memory check
Also fix an oddity in the case when the last memory check is deleted,
the jit cache was supposed to be cleared in that case, but it was out of
the for loop that finds the one to delete so it was never run.
Naturally, the same fix for the adding the first memory check was
applied.
EmptyChaos