* shader cache: Fix Linux boot issues
This rollback the init logic back to previous state, and replicate the
way PTC handle initialization.
* shader cache: set default state of ready for translation event to false
* Fix cpu unit tests
* shader cache: Fix possible race causing crashes on manifest at startup
This fix a misplace function call ending up causing possibly two write
on the cache.info at the same time.
* shader cache: Make RemoveManifestEntries async too to be sure all operations are perform before starting the game
* shader cache: Fix invalid virtual address clean up
This fix an issue causing the virtual address of texture descriptors to
not be cleaned up when caching and instead cleaning texture format and swizzle.
This should fix duplicate high duplication in the cache for certain
games and possible texture corruption issues.
**THIS WILL INVALIDATE ALL SHADER CACHE LEVELS CONSIDERING THE NATURE OF THE ISSUE**
* shader cache: Address gdk's comment
* infra: Migrate to .NET 5
This migrate projects and CI to .NET 5
* Remove language version restrictions (now on 9.0 by default)
* infra: pin .NET 5 to avoid later issues
* infra: Cleanup csproj files
* infra: update dependencies
* infra: Add temporary workaround for a bug in Vector128.Create
see https://github.com/dotnet/runtime/issues/44704 for more informations
Here come Salieri, my implementation of a disk shader cache!
"I'm sure you know why I named it that."
"It doesn't really mean anything."
This implementation collects shaders at runtime and cache them to be later compiled when starting a game.
This fix a mistake I made during my original reimplementation of SurfaceFlinger by disabling async buffer.
This fix a memory corruption on Super Mario All-Stars 3D (Super Mario Sunshine & Super
Mario Galaxy now go ingame).
Thanks to @gdkchan for tracing the memory corruption.
* hos/gui: Add a check of NCA program index in titleid
This add a check to `ApplicationLoader` for the last 2 digits of the game TitleId who seems to be the NCA program index.
We currently return the last index, instead of the lower one.
Same check is added to ApplicationLibrary in the UI.
I've cleaned up both file too.
* hle: implement partial relaunch logic
TODO: make the emulator auto relauch.
* Handle auto relaunch
* hle: Unify update usage system
* hle: Implement support of multi programs in update system
* Add some documentation
* Address rip's comment
Co-authored-by: Ac_K <Acoustik666@gmail.com>
This add a missing offset on the output buffer of the DataSourceVersion2Command.
This fix music only playing on the left channel on Fairy Tail, Family Mysteries: Poisonous Promises, SEGA AGES Sonic the Hedgehog 2 and probably more.
This update the "No keys" dialog and block starting NSP/XCI/NCA without firmware.
Also propose to the user if they want to install firmware if they start an untrimmed XCI and remove KEYS.md as it was completely outdated.
PS: Also fix a bug with "&" in URL with OpenUrl on Windows.
This project wasn't really used by anyone and isn't worth mantaining.
This commit remove the profiler entirely from Ryujinx and remove the associated CI tasks.
This fix front center and lfe being inverted in input of the
DownMixSurroundToStereoCommand.
This fix:
- Voices being missing on FE3H videos
- Mario Tennis Aces missing backgroun
- Probably more.
Changes:
Implement software surround downmixing (fix#796).
Fix a crash when no audio renderer were created when stopping emulation.
NOTE: This PR also disable support of 5.1 surround on the OpenAL backend as we cannot detect if the hardware directly support it. (the downmixing applied by OpenAL on Windows is terribly slow)
* Amadeus: Final Act
This is my requiem, I present to you Amadeus, a complete reimplementation of the Audio Renderer!
This reimplementation is based on my reversing of every version of the audio system module that I carried for the past 10 months.
This supports every revision (at the time of writing REV1 to REV8 included) and all features proposed by the Audio Renderer on real hardware.
Because this component could be used outside an emulation context, and to avoid possible "inspirations" not crediting the project, I decided to license the Ryujinx.Audio.Renderer project under LGPLv3.
- FE3H voices in videos and chapter intro are not present.
- Games that use two audio renderer **at the same time** are probably going to have issues right now **until we rewrite the audio output interface** (Crash Team Racing is the only known game to use two renderer at the same time).
- Persona 5 Scrambler now goes ingame but audio is garbage. This is caused by the fact that the game engine is syncing audio and video in a really aggressive way. This will disappears the day this game run at full speed.
* Make timing more precise when sleeping on Windows
Improve precision to a 1ms resolution on Windows NT based OS.
This is used to avoid having totally erratic timings and unify all
Windows users to the same resolution.
NOTE: This is only active when emulation is running.
* prepo: Update implementation for 10.x changes
On 10.x, Nintendo changed yet another time the prepo SaveReport &
SaveReportWithUser command ids.
This PR add support for command 10104 & 10105 and update naming of the
old variants to match switchbrew.
* Fix misalign and remove a ToString
* Address jD's comment
* Fix compilation warnings and use new LibHac APIs for executable loading
* Migrate NSO loader to the new reader and fix kip loader
* Fix CS0162 restore
* Remove extra return lines
* Address Moose's comment
* SurfaceFlinger: fix some bugs
This fixes some bugs in the current implementation and make it closer to
the real implementation.
* Fix align of some variables
* am: Implemnet common web applets
This implement parsing of input and output of web applets while making
those close directly.
TODO for the future: Use and hook a web browser.
* Address Ac_K's comments
This should fix OpenGL context creation on OSX (using Cocoa).
NOTE: As macOS doesn't support OpenGL 4.3, the emulator will probably not work but this fix the hardcrash when starting any games.
* nvservice: add a lock to NvHostEvent
* Disable surface flinger release fence and readd infinite timeout
* FenceAction: Add a timeout of 1 seconds as this shouldn't wait forever anyuway
* surfaceflinger: remove leftovers from the release fence
* Don't allow infinite timeout on syncpoint while printing all timeout for better debugging
* nvservices: mitigate abort with heavy load on the GPU processing thread.
This should fix Mario Tennis and LM3 regressions with syncpoints.
NOTE: Mario Tennis seems to have another issue related to the texture
cache that happens randomly when starting a match.
PS: Also add a debug logger for all known ioctl call to facilitate
debugging and add a missing UpdateMin in EventSignal.
* Address LDj3SNuD's comment
* Address gdkchan's comment
This remove Utf8son and JsonPrettyPrinter dependencies.
NOTE: the standard JSON parser doesn't support configurable
indentation, as a result, all the pretty printed JSON are indented with 2
spaces.
This implement the rendering information output informations of
RequestUpdate.
This is needed by some games to keep track of the count of update on the
DSP.
As I was unable to disable the ctrl + f keybinding, this implement
something to make it works with the default GTK code logic.
Also remove unused _treeView static member.
* Add support for dynamic docking/undocking
As SurfaceFlinger is now working more accurately, we can now support
dynamic configuration of docking mode :)
* Simplify a bt the code
* Fix import ordering
* Remove unused argument
This invalidate the GraphicBuffer on the consumer side when
SetPreallocatedBuffer is called on a buffer slot.
This fix rendering issues on games with a dynamic resolution like Yoshi
Crafted World.
* Rewrite SurfaceFlinger
Reimplement accurately SurfaceFlinger (based on my 8.1.0 reversing of it)
TODO: support swap interval properly and reintroduce disabled "game vsync" support.
* Some fixes for SetBufferCount
* uncomment a test from last commit
* SurfaceFlinger: don't free the graphic buffer in SetBufferCount
* SurfaceFlinger: Implement swap interval correctly
* SurfaceFlinger: Reintegrate Game VSync toggle
* SurfaceFlinger: do not push a fence on buffer release on the consumer side
* Revert "SurfaceFlinger: do not push a fence on buffer release on the consumer side"
This reverts commit 586b52b0bfab2d11f361f4b59ab7b7141020bbad.
* Make the game vsync toggle work dynamically again
* Unregister producer's Binder object when closing layer
* Address ripinperi's comments
* Add a timeout on syncpoint wait operation
Syncpoint aren't supposed to be waited on for more than a second.
This effectively workaround issues caused by not having a channel
scheduling in place yet.
PS: Also introduce Android WaitForever warning about fence being not
signaled for 3s
* Fix a print of previous commit
* Address Ac_K's comments
* Address gdkchan's comments
* Address final comments
* Implement GPU syncpoints
This adds support for GPU syncpoints on the GPU backend & nvservices.
Everything that was implemented here is based on my researches,
hardware testing of the GM20B and reversing of nvservices (8.1.0).
Thanks to @fincs for the informations about some behaviours of the pusher
and for the initial informations about syncpoints.
* syncpoint: address gdkchan's comments
* Add some missing logic to handle SubmitGpfifo correctly
* Handle the NV event API correctly
* evnt => hostEvent
* Finish addressing gdkchan's comments
* nvservices: write the output buffer even when an error is returned
* dma pusher: Implemnet prefetch barrier
lso fix when the commands should be prefetch.
* Partially fix prefetch barrier
* Add a missing syncpoint check in QueryEvent of NvHostSyncPt
* Address Ac_K's comments and fix GetSyncpoint for ChannelResourcePolicy == Channel
* fix SyncptWait & SyncptWaitEx cmds logic
* Address ripinperi's comments
* Address gdkchan's comments
* Move user event management to the control channel
* Fix mm implementation, nvdec works again
* Address ripinperi's comments
* Address gdkchan's comments
* Implement nvhost-ctrl close accurately + make nvservices dispose channels when stopping the emulator
* Fix typo in MultiMediaOperationType
* libhac: use ApplicationControlProperty instead of Nacp
Nacp was marked as deprecated in 0.10.0, this PR remove all usage of it
from Ryujinx and use the new struct ApplicationControlProperty.
* Address Moose's comment
* Reduce requirements for running homebrews
This commit change the following behaviours:
- TimeZoneBinary system archive isn't required until guest code call LoadTimeZoneRule.
- Fonts system archives aren't requred until a "pl:u" IPC call is made.
- Custom font support was dropped.
- TimeZoneBinary missing message is now an error and not a warning.
* Address comments
REV8 only added changes on performance buffer and wavebuffer dsp command
generation.
As we don't support any of those, we can just increment the revision
number for now.
* Implement mii:u and mii:e entirely
Co-authored-by: AcK77 <Acoustik666@gmail.com>
This commit implement the mii service accurately.
This is based on Ac_k work but was polished and updated to 7.x.
Please note that the following calls are partially implemented:
- Convert: Used to convert from old console format (Wii/Wii U/3ds)
- Import and Export: this is shouldn't be accesible in production mode.
* Remove some debug leftovers
* Make it possible to load an arbitrary mii database from a Switch
* Address gdk's comments
* Reduce visibility of all the Mii code
* Address Ac_K's comments
* Remove the StructLayout of DatabaseSessionMetadata
* Add a missing line return in DatabaseSessionMetadata
* Misc fixes and style changes
* Fix some issues from last commit
* Fix database server metadata UpdateCounter in MarkDirty (Thanks Moose for the catch)
* MountCounter should only be incremented when no error is reported
* Fix FixDatabase
Co-authored-by: Alex Barney <thealexbarney@gmail.com>
* Implement GetCurrentIpConfigInfo
This is needed by Rocket League.
Also fix GetCurrentIpConfigInfo to not return ipv6 addresses
* Address Ac_K comment
* Some fixes for the new GLRenderer
Changelog:
- Fix transparency of the window on some games on Windows.
- Fix escape key not being able to exit emulation.
- Fix inverted logic in fullscreen event handling.
- Fix a race condition when stoping emulation causing a hang.
- Fix a memory leak of the OpenGL context when stoping emulation (saving ~200MB of RAM when stoping emulation).
- Simplify and document behaviours when exiting the emulator while the
emulation is running.
* Make sure to clear alpha channel when presenting Texture
This fix once and for all the transparency issue on Windows.
* Enforce footer bar size to avoid gl widget to get resized to 1280x724
* Fix full screen inversion in MainWindow and make sure _listStatusBox don't come back when not needed
* Remove previous transparency clear attempt that is useless now
* Remove an extra line return
* Stub the application copyright framebuffer api
As we currently don't support multi layers on vi/nvnflinger, this PR
implement a stub of this API.
* Address Cyuubi's comments
* Add IPC checks and comments for future reversing
Co-authored-by: Ac_K <Acoustik666@gmail.com>
Co-authored-by: Ac_K <Acoustik666@gmail.com>
* Fix a crash when closing the main Ui
Also make sure to dispose the OpenAL context to not leak memory when
unloading the emulation context.
* Improve keys and 'game already running' dialogs
* Make sure to dispose the page table and ThreadContext
Less memory leaks!
* Fix tests
* Address gdk's comments
* Implement IDeliveryCacheProgressService in bcat
This stub IDeliveryCacheProgressService IPC interface as we don't plan
to support cache delivery.
* Address jd's comments
* Address jd's comment correctly
* Address gdk's comments
* Fix inconsistencies with UserId
The account user id isn't an UUID. This PR adds a new UserId type with
the correct value ordering to avoid mismatch with LibHac's Uid. This also fix
an hardcoded value of the UserId.
As the userid has been invalid for quite some time (and to avoid forcing
users to their recreate saves), the userid has been changed to "00000000000000010000000000000000".
Also implement a stub for IApplicationFunctions::GetSaveDataSize. (see
the sources for the reason)
Fix#626
* Address jd's & Ac_k's comments
* Keep the GUI alive when closing a game
Make HLE.Switch init when starting a game and dispose it when closing
the GlScreen.
This also make HLE in charge of disposing the audio and gpu backend.
* Address Ac_k's comments
* Make sure to dispose the Discord module and use GTK quit method
Also update Discord Precense when closing a game.
* Make sure to dispose MainWindow
* Address gdk's comments
* Fix race condition in ContentManager
This fix a race condition happening since #791 when trying to load a
game via command line.
* Address gdk's comments
* Ensure to dispose the FileStream and not the IStorage
* Make HLE disposable safely
This fix the oldest issue with the HLE code: the kernel side
disposability.
Changelog:
- Implement KProcess::UnpauseAndTerminateAllThreadsExcept, KThread::Terminate, KThread::TerminateCurrentProcess, KThread::PrepareForTermiation and the svc post handler accurately.
- Implement svcTerminateProcess and svcExitProcess. (both untested)
- Fix KHandleTable::Destroy not decrementing refcount of all objects stored in the table.
- Spawn a custom KProcess with the maximum priority to terminate every guest KProcess. (terminating kernel emulation safely)
- General system stability improvements to enhance the user's experience.
* Fix a typo in a comment in KProcess.cs
* Address gdk's comments
This add some code to handle usage of poll without any fds.
This is required by Dark Souls Remastered main loop logic as it's
calling it without any fds during initialization.
===
General system stability improvements to enhance the user's experience.
The configuration system was quite fragile and too dependent on everything, this fix#812 .
The changes:
The file configuration is now entirely independent from the internal configuration state.
The file configuration is versioned (current version is 1).
Every configuration elements are now reactive properties that the emulator can register on to handle initialization and configuration changes.
The configuration system is now in Ryujinx.Common to be accessible on every projects.
Discord integration is now independent from the UI and can be reloaded.
The primary controller is now configurable at runtime (NOTE: the UI currently doesn't have any options to configure real controller).
The logger is entirely reloadable.
You can now hotplug your controller when the emulator is running.
The logger now takes name for every LogTarget to make them removable at runtime.
The logger now always add the default "console" target to avoid loosing early init logs.
The configuration system now generates a default file configuration if it's missing or too new.
General system stability improvements to enhance the user's experience
* Fix ILogger type and size decoding
The type and size are custom encoded integer not byte.
This fix issues on games that send messages longer than 127 characters.
* Address gdk's comments
* hle: Improve IRoInterface logic
This commit contains a little rewrite of IRoInterface to fix some issues
that we were facing on some recent games (AC3 Remastered & Final Fantasy
VIII Remastered)
Related issues:
- https://github.com/Ryujinx/Ryujinx-Games-List/issues/196
* Address comments
* Add detail of ZbcSetTableArguments
This is a missing part of the #800 PR that cause an assert to be
triggered in debug mode.
Also, remove Fence in SurfaceFlinger as it's a duplicate of NvFence.
* Fix critical issue in size checking of ioctl
oops
MaryandGitHub
