They weren't sufficient and are made redundant by previous commits; they
also (on master) caused breakage due to Jit64::psq_stXX assuming writes
would be fastmem and not clobber a register under certain conditions.
That really needs to be refactored, but for now, this works.
Change TMemCheck::Action to return whether to break rather than calling
PPCDebugInterface::BreakNow, as this simplified the implementation; then
remove said method, as that was its only caller. One "interface" method
down, many to go...
Without fastmem, the JIT code still does an inline check for RAM
addresses. With watchpoints we have to disable that too. (Hardware
watchpoints would avoid all the slow, but be complicated to implement
and limited in number - I doubt most people debugging games care much if
they run slower.)
With this change and watchpoints enabled, Melee runs at no more than 40%
speed, despite running at full speed without them. Oh well. Better
works slowly than doesn't bloody work.
Incidentally, I'm getting an unrelated crash in
PowerPC::HostIsRAMAddress when shutting down a game. This code sucks.
- Move JitState::memcheck to JitOptions because it's an option.
- Add JitOptions::fastmem; switch JIT code to checking that rather than
bFastmem directly.
- Add JitBase::UpdateMemoryOptions(), which sets both two JIT options
(replacing the duplicate lines in Jit64 and JitIL that set memcheck
from bMMU).
- (!) The ARM JITs both had some lines that checked js.memcheck
despite it being uninitialized in their cases. I've added
UpdateMemoryOptions to both. There is a chance this could make
something slower compared to the old behavior if the uninitialized
value happened to be nonzero... hdkr should check this.
- UpdateMemoryOptions forces jo.fastmem and jo.memcheck off and on,
respectively, if there are any watchpoints set.
- Also call that function from ClearCache.
- Have MemChecks call ClearCache when the {first,last} watchpoint is
{added,removed}.
Enabling jo.memcheck (bah, confusing names) is currently pointless
because hitting a watchpoint does not interrupt the basic block. That
will change in the next commit.
On OS X, this broke Cmd-V to paste in the text boxes. Apparently wx
thinks having mnemonics (which are Alt-* on Windows) be Cmd-* on OS X,
even if this disables standard shortcuts, is a good idea.
Lioncash suggested just getting rid of the accelerators on non-menu
controls, so I'm doing that rather than disabling them only on OS X.
1) Apparently wxString::Format is type safe, and passing a u32 to it
with the format "%lu" crashes with a meaningless assertion failure.
Sure, it's the wrong type, but the error sure doesn't help...
2) "A MenuItem ID of Zero does not work under Mac". Thanks for the
helpful assert message, no thanks for making your construct have random
platform-specific differences for no reason (it's not like menu item IDs
directly correspond to a part of Cocoa's menu API like they do on
Win32).
All the multiplying and dividing by 100 in controller configs is
messy... An attempted solution to the problem was to not multiply
default_value by 100 in ControllerEmu::ControlGroup::LoadConfig,
but that broke other things instead, so I went with this.
This makes the code cleaner and also leads to some user-visible changes:
The wx game properties will no longer let the user
select WAD languages that don't have any names.
The Qt game list will now display names using the languages
set in the configuration instead of always using
English for PAL GC games and Japanese for WADs.
If a WAD doesn't have a name in the user's preferred language,
English is now selected as a fallback before Japanese.
Boot_BS2Emu was trying to read from the inserted disc even when
nothing was inserted, and this happened to not crash (but not
work either) before VolumeHandler was removed. This commit adds
a check that restores the old behavior, so there is no longer a
crash, but the game ID still doesn't get set for WADs. I don't
know if/how it should be set, so this felt like the safest option.
This was causing a race condition where the "absurdly large aux buffer"
panic alert would be triggered in the last bit of fifo processing on the
CPU thread in deterministic mode (i.e. netplay). SyncGPU is supposed to
move the auxiliary queue data to the beginning of the containing buffer
so we don't have to deal with wraparound; if GpuRunningState is false,
however, it just returns, because it's set to false by another thread -
thus it doesn't know whether RunGpuLoop is still executing (in which
case it can't just reset the pointers, because it may still be using the
buffer) or not (in which case the condition variable it normally waits
for to avoid the previous problem will never be signaled). However,
SyncGPU's caller PushFifoAuxBuffer wasn't aware of this, so if the
buffer was filling at just the right time, it'd stay full and that
function would complain that it was about to overflow it. Similar
problem with ReadDataFromFifoOnCPU afaik. Fix this by returning early
from those as well; other callers of SyncGPU should be safe. A
*slightly* cleaner alternative would be giving the CPU thread a way to
tell when RunGpuLoop has actually exited, but whatever, this works.
Core::Shutdown was only called on app exit, yet the emu thread exits
whenever emulation stops; if you launched a new game it would just join
via the destructor when s_emu_thread was set to a new thread.
(Incidentally, the destructor also makes explicitly joining on app exit
rather pointless.)
Because the GUI thread wasn't waiting for the CPU thread to fully shut
down, Core::IsRunning would remain true briefly after CFrame::DoStop
which, given Dolphin's penchant for accessing variables belonging to
other threads, can only mean trouble... In my case, because the previous
commit caused UpdateGUI, which is called at the end of DoStop, to call
PauseAndLock, which checks IsRunning, pressing stop at the right time
would cause strange behavior.
Since its lifetime is managed on the CPU thread, this (or a refactoring)
is absolutely required. One of the functions with a PauseAndLock call
added is CFrame::UpdateGUI; this is fine now, since it's called only
after important events happen, so just make sure not to call it every
frame or something :)
Initially, the dialogs construct in the background when Dolphin initializes. However, it waits until the user actually makes the dialogs visible to decide on whether to create the Wii or GC control layouts.
Therefore, the call to CreateBaseLayout() essentially creates a sizer that isn't actually attached to a main sizer that is set as the sizer for the dialog to use. So upon destruction, these controls would never actually be destroyed if the user didn't open and then close the TAS dialogs.
Incrementing the reference count here isn't necessary, as they construct with a count of 1. Incrementing again results in the attributes not being freed.
The regBuildMemAddress function already clears the address register.
Not only is clearing it again pointless, regBuildMemAddress uses the
bits in IInfo slightly diffrently and the second clear can clear
the wrong registers causing bugs if something else actually needs to
use those registers.
VolumeHandler is basically just a wrapper around a single IVolume object.
This change moves that object to DVDInterface, moves the Read32
function to IVolume, and gets rid of the rest of VolumeHandler.
The instant speed broke some games. SUDTR will now emulate
the transfer from the disc drive buffer to the main memory,
but not the speed of the disc drive itself.
"Dolphin at GitHub" doesn't make much sense IMO. We could also make it say "Source Code" or "Dolphin on GitHub".
I also changed "Dolphin Website" to make it consistent with the Qt version.
With some specific, STUN-hostile routers, the netplay client can get
stuck forever while trying to connect to the stun server. This adds a
5 seconds (much more than should be necessary if it works) timer until
a failure is registered and the attempt stops.
This technically also fixes a memory leak in WatchView.cpp, because the table setting was done such that the grid wouldn't take ownership of the table, which means said table wouldn't be deleted in the grid's destructor.
Previously OSX would crash upon loading. This is because the UI would be polling for hotkey inputs but was doing so before the inputs backends were initialized.
* Changed the SI buffer processing so that transfers do not have to be completed instantly
* Added a second socket at port 49420 (0xc10c) which sends clock information to the GBA slaves
* Handled disconnections from the GBA and GC
* Made the transfers asynchronous
* Blocks the socket before the connection times out
Requires VBA-M SVN 1235 or later.
Prior to this refactor, certain options would cause the game list to refresh when the config modal dialog is closed (such as adding a folder to the path list). This restores that functionality.
Prior to this, ConfigMain.cpp was a large (52KB) cpp file that contained all of the UI setting code.
This breaks up the config code into subclasses of wxPanel, which are then just instantiated to add to the settings wxNoteBook. This keeps all the settings categories separated from one another and also cleans up the code in general.
Otherwise, it would work but any async sending would be delayed by 4ms or
wait until the next packet was received.
Also increase the client timeout to 250ms, since enet_host_service is now
really interrupted.
Add std::unique_ptr<sf::Packet> objects to a queue instead of functions,
makes things easier to read, and avoids headaches while checking the
lifetime of the concerned objects.
Using BindToRegister from the register caches causes the register bound to be marked dirty.
Using the regular R() function from the register caches loads the registers without being dirty.
When implementing new instructions make sure to BindToRegister registers that become dirty.
With my previous changes Dolphin would fail to create the user directory if it didn't exist, and would dump all the configuration options in to the cwdir.
This was a bit more complicated to fix in a clean fashion, so I took to moving around code concerning user directories.
Instead of having GetUserPath serve a dual purpose of both getting and setting our user directories, break out to a new SetUserPath function.
GetUserPath will know only get the configured user path.
SetUserPath will set our user paths and setup the internal user path state.
This ending up being a lot cleaner overall, which is nice. Also less mind bending when attempting to read the code.
So now we won't dump all of our configuration in to the cwdir if ~/.dolphin-emu isn't found.
Fixes issue 8371.
It turns out that the offsets of the data in the return buffer matter: the
offset 0x340 is hardcoded into the code that uses the result.
With that fixed, get rid of the completely wrong 0x80000000 hack.
psq_st performs one store, and psq_ld one load, from the perspective of the
MMU; getting this wrong leads to potentially incorrect behavior (incorrect page
faults, weirdness with the gather pipe, etc.). Fix this, and stop masking
the address when checking for gather pipe writes.
Also a bunch of cleanup.
This came up from the discussion we were having prior about dumping half of a kilobyte of VFP registers to the stack is insanity.
This was due to me basically copying exactly what I did on ARMv7's paired loadstores, where the impact is less since we only use the bottom 64bits of
the VFP registers.
So I decided to think about how to improve upon this since I got called out on my terrible code.
The solution I have come up with is instead of jumping to the common ASM routine and having that check if it needs to take the fastmem or slowmem
routes, just inline the check in to the JIT block and jump to either a fastmem or slowmem handler.
Fairly simple, and this allows us to only flush the registers that are required when doing so. Should give a reasonable increase in performance for
games that use the slowmem path quite a lot.
We are declaring we require ARB_shader_image_load_store in the shader, this isn't an extension on GLES because it is part of the GLSL ES 3.1 spec.
If we are running as GLES then just not put it in the shaders.
Activity Monitor can start activities by using adb to invoke it.
This will allow us to set the user directory and autostart file from adb.
adb shell am start -n org.dolphinemu.dolphinemu/.gamelist.GameListActivity -e AutoStartFile /sdcard/AC.gcz -e UserDir /sdcard/dolphin-emu2/
This allows more automated testing to be done with Dolphin on Android.
Cleared the default mapping of hotkeys as they were conflicting with the key shortcuts.
Removed the Open, Change Disc, Refresh List and Exit hotkeys as they work only when driven from the menu.
- Simplified the locking mechanism when controllers were updated
- Reloaded the config of the controls instead of re-initialising the control plugins
- Fixed controls being unresponsive after the Refresh button was pressed
- Disables the hotkeys while the controller config is open
* Halving the number of hotkeys checked for (there were extra for future use)
* Gets the controller status once before parsing hotkeys
* Checks for the GUI lock once before parsing hotkeys
* Removed some redundant memsets
Clamping a rectangle correctly requires fully clamping all four
coordinates in the general case.
This should fix issue 6923, sort of; at least, it fixes the part where a
rectangle ends up with a nonsensical height after being clamped.
This is unused, and since it had the same value as FL_OUT_D, it was unnecessarily setting the rS register as an output, even on instructions that only have FL_OUT_D set.
This is good for performance when FileMontior is selected
but the log level only is set to notice or warning.
The ability to do this wasn't available until recently:
1ed41672f5
A bit more efficient if we are only pushing two VFP registers.
We can probably be a bit more efficient in the future by mixing paired loadstores in to the other paths as well.
Previously on FPR pushing and popping we would do a single STR/LDR per quad FPR we wanted to push/pop.
In most of our cases when we are pushing and popping VFP registers they will be consecutive registers that will save more efficiently using the NEON
loadstores that can do up to four quad registers.
So this can potentially cutting instructions down to ~1/4th the amount of instructions if the registers are all consecutive.
On the Cortex-A57 this is basically just an icache improvement, but on the Nvidia Denver this may be optimized to be more efficient. Either way it's a
win.
This drops the "feature" to load level 0 from the custom texture
and all other levels from the native one if the size matches.
But in my opinion, when a custom texture only provide one level,
no more should be used at all.
Previously InitControllers() would be called every poll duration, setting this boolean to false which would prevent said sybsystems from deinitializing correctly.
Technically fixes a memory leak (which wouldn't matter because the timer is only created once and destroyed on shutdown).
Also starting and stopping the timer does not cause leaks.
comex