* Salieri: Add blacklist system and blacklist shaders using bindless
Currently the shader cache doesn't have the right format to support
bindless textures correctly and may cache shaders that it cannot rebuild
after host invalidation.
This PR address the issue by blacklisting shaders using bindless
textures.
THis also support detection of already cached broken shader and handle removal
of those.
* Move to a feature flags design to avoid intrusive changes in the translator
This remove the auto correct behaviour
* Reduce diff on TranslationFlags
* Reduce comma on last entry of TranslationFlags
* Fix inverted logic and remove leftovers
* remove debug edits oops
* Improve linear texture compatibility rules
Fixes an issue where small or width-aligned (rather than byte aligned) textures would fail to create a view of existing data. Creates a copy dependency as size change may be risky.
* Minor cleanup
* Remove Size Change for Copy Depenedencies
The copy to the target (potentially different sized) texture can properly deal with cropping by itself.
* Move StrideAlignment and GobAlignment into Constants
This PR remove the IIpcService.cs interface usage which isn't needed anymore since the interface is only used by IpcService class. We can use it directly.
This PR implement `ApplicationErrorArg` to the Error Applet. It's used by the guest to throw some specific error messages.
The code was done for (and merged) LDN2 build since long time ago and have been tested a bunch of times because of that! In a way to reduce the differences between LDN and master build it's fine to add it to master.
* Minor wording fixes
* Exit dialogue.
* Update MainWindow.cs
* Update DlcWindow.cs
* I found an actual typo for once.
* There we go.
* Fix
* Update MainWindow.cs
* MainWindow is done
* Update ControllerWindow.cs
* Implement friendlier portable mode
* Remove first run dialog
* Disable updates in portable mode for now
* Convert relative custom paths to absolute ones
Also, fix a regression when custom path doesn't exist
This enforce speaker mode to avoid weird audio quality reduction with
output considered as headphones by Windows (It applies a HRTF or crossfeed filter supposed to improve audio quality, might be a bug on their end)
This allows bindless handles to be found for image/texture instructions with predicates, when the assignment of the texture handle is within the same predicate.
This seems to cover the remaining bindless handles that compilers seem to be creating due to optimizations.
Will affect newer UE4 games, and games by NdCube (Super Mario Party, Clubhouse Games)
* Improve Buffer Textures and flush Image Stores
Fixes a number of issues with buffer textures:
- Reworked Buffer Textures to create their buffers in the TextureManager, then bind them with the BufferManager later.
- Fixes an issue where a buffer texture's buffer could be invalidated after it is bound, but before use.
- Fixed width unpacking for large buffer textures. The width is now 32-bit rather than 16.
- Force buffer textures to be rebound whenever any buffer is created, as using the handle id wasn't reliable, and the cost of binding isn't too high.
Fixes vertex explosions and flickering animations in UE4 games.
* Set ImageStore flag... for ImageStore.
* Check the offset and size.
* Improve handling for unmapped GPU resources
- Fixed a memory tracking bug that would set protection on empty PTEs
- When a texture's memory is (partially) unmapped, all pool references are forcibly removed and the texture must be rediscovered to draw with it. This will also force the texture discovery to always compare the texture's range for a match.
- RegionHandles now know if they are unmapped, and automatically unset their dirty flag when unmapped.
- Partial texture sync now loads only the region of texture that has been modified. Unmapped memory tracking handles cause dirty flags for a texture group handle to be ignored.
This greatly improves the emulator's stability for newer UE4 games.
* Address feedback, fix MultiRange slice
Fixed an issue where the size of the multi-range slice would be miscalculated.
* Update Ryujinx.Memory/Range/MultiRange.cs (feedback)
Co-authored-by: Mary <thog@protonmail.com>
Co-authored-by: Mary <thog@protonmail.com>
I happened to notice this when checking Ryu's IPC structures. After double checking with RE it seems the current check is incorrect so use the correct member to fix that.
* Initial implementation (3d tex mips broken)
This works rather well for most games, just need to fix 3d texture mips.
* Cleanup
* Address feedback
* Copy Dependencies and various other fixes
* Fix layer/level offset for copy from view<->view.
* Remove dirty flag from dependency
The dirty flag behaviour is not needed - DeferredCopy is all we need.
* Fix tracking mip slices.
* Propagate granularity (fix astral chain)
* Address Feedback pt 1
* Save slice sizes as part of SizeInfo
* Fix nits
* Fix disposing multiple dependencies causing a crash
This list is obviously modified when removing dependencies, so create a copy of it.
* Fix wrong VA being passed to VirtualMemoryEvent
* Remove unnecessary type in discard
Co-authored-by: gdkchan <gab.dark.100@gmail.com>
Co-authored-by: gdkchan <gab.dark.100@gmail.com>
* Revise SystemInfo
Cleans up and adds a bit more info (logical core count and available mem at launch) to logs.
- Extract CPU name from CPUID when supported.
- Linux: Robust parsing of procfs files
- Windows: Prefer native calls to WMI
- Remove unnecessary virtual specifiers
* Address gdkchan's comments
* Address AcK's comments
* Address formatting nits
* Haydn: Fix race condition in SoundIO Update implementation
This should fix weird crashes happening for some people with SoundIO.
Fix#2062
* haydn: Make SoundIO session lock-free
* ci: Use Github artifacts for PRs
This PR extends the build job by upload artifacts for all PRs.
The version of those builds is set to "1.0.0-<git_short_hash>" and those
builds have the updater explicitly disabled.
With this in place, we will be able to disable ApppVeyor on PRs.
* Add actions to comment build links
As Github doesn't provide links to unlogged users, we use nightly.link for that.
* Address Ac_k's comments
* Fix virtual address overflow near ulong limit
* Fix comments
* Improve overflow checking for large size values
* Add overflow checking to AddressSpaceManager class
* Add overflow protection to read and write functions
* Haydn: Part 1
Based on my reverse of audio 11.0.0.
As always, core implementation under LGPLv3 for the same reasons as for Amadeus.
This place the bases of a more flexible audio system while making audout & audin accurate.
This have the following improvements:
- Complete reimplementation of audout and audin.
- Audin currently only have a dummy backend.
- Dramatically reduce CPU usage by up to 50% in common cases (SoundIO and OpenAL).
- Audio Renderer now can output to 5.1 devices when supported.
- Audio Renderer init its backend on demand instead of keeping two up all the time.
- All backends implementation are now in their own project.
- Ryujinx.Audio.Renderer was renamed Ryujinx.Audio and was refactored because of this.
As a note, games having issues with OpenAL haven't improved and will not
because of OpenAL design (stopping when buffers finish playing causing
possible audio "pops" when buffers are very small).
* Update for latest hexkyz's edits on Switchbrew
* audren: Rollback channel configuration changes
* Address gdkchan's comments
* Fix typo in OpenAL backend driver
* Address last comments
* Fix a nit
* Address gdkchan's comments
* Fix updater breaking user run permissions on *nix
The updater currently sets the permissions to user run only when the yes button is clicked, this changes it to run on Updater.cs after it's done updating as it should.
* Fix nits
* Use multiple threads to download different chunks of an update simultaneously. This reduces time to complete the download significantly.
* Remove dirty-flag check (for test purposes)
* Clean up updater code.
* Include fallback to single-threaded updater if mt fails
* Reduce connection count to 4.
* Improve fallback on error.
Correct issue where data was missing during download due to total build size not being cleanly divisble by the connection count.
Cleaned up unnecessary code.
* Add missing return statements
* Fix alignment
* Alignment
* More alignment
* Rely on content-range request instead of xml/json size property.
* Re-instate dirty checking and version checking to move into review stage.
* Address comments
* Address comments
* Comments
* Comments
* Final...?
* final final
* final final final nit
* Use Array.Copy as requested by rip
* Updated some names for clarity.
* Move addition into for loop (to shorten line width)
* Add missing semicolon -- forgot to stage :9
* Implement VCNT based on AArch64 CNT
Add tests
* Update PTC version
* Address LDj's comments
* Explicit size in encoding
* Tighter tests
* Replace SoftFallback with IR helper
Co-authored-by: LDj3SNuD <35856442+LDj3SNuD@users.noreply.github.com>
* Reduce one BitwiseAnd from IR fallback
Based on popcount64b from https://en.wikipedia.org/wiki/Hamming_weight#Efficient_implementation
* Rename parameter and add assert
Co-authored-by: LDj3SNuD <35856442+LDj3SNuD@users.noreply.github.com>
Co-authored-by: LDj3SNuD <35856442+LDj3SNuD@users.noreply.github.com>
* PPTC & Pool Enhancements.
* Avoid buffer allocations in CodeGenContext.GetCode(). Avoid stream allocations in PTC.PtcInfo.
Refactoring/nits.
* Use XXHash128, for Ptc.Load & Ptc.Save, x10 faster than Md5.
* Why not a nice Span.
* Added a simple PtcFormatter library for deserialization/serialization, which does not require reflection, in use at PtcJumpTable and PtcProfiler; improves maintainability and simplicity/readability of affected code.
* Nits.
* Revert #1987.
* Revert "Revert #1987."
This reverts commit 998be765cf7f7da5ff0c1c08de704c9012b0f49c.
* This should fix issue #1374 in Linux
Changes:
- Bind buttons by detecting the transition from down to up.
- Bind axis by detecting movement from value higher than 50% to a value lower than 50%.
Caveats:
- I have tested only with DS3 in Linux (Fedora 32).
- ZL and ZR detection works by accident. This code doesn't take negative axis into account.
The reason it works is because axis are managed in absolute value. So when pressing ZL/ZR
axis value goes from -1 to 1 (or 1 to 0 and back to 1) and this hits the axis detector.
- Likely I have broken all the other controllers xD (testing needed).
* Assign keyboardPressed
* Make a more robust detection of pressed buttons when using a controller
* Add interface to bind buttons from Joystick and Keyboard
* Fix style issues after code review by @AcK77 (Thanks!)
* Move new classes to Ryujinx.Ui.Input namespace
* Use explicit types instead of var
* Update Ryujinx/Ui/Input/JoystickButtonAssigner.cs
Co-authored-by: Mary <thog@protonmail.com>
* Update Ryujinx/Ui/Input/JoystickButtonAssigner.cs
Co-authored-by: Mary <thog@protonmail.com>
* Update Ryujinx/Ui/Input/JoystickButtonAssigner.cs
Co-authored-by: Mary <thog@protonmail.com>
* Update Ryujinx/Ui/Input/JoystickButtonAssigner.cs
Co-authored-by: Ac_K <Acoustik666@gmail.com>
* Add a new empty line before
* Up
Co-authored-by: Jose Padilla <jose@prensalink.com>
Co-authored-by: Mary <thog@protonmail.com>
Co-authored-by: Ac_K <Acoustik666@gmail.com>
* Make Windows DPI aware to display properly on high-resolution screens.
* remove empty line
* Don't use app manifest, set process dpi aware programatically.
Store variables in Program.cs for use instead of re-creating them per class/ method.
* Fix for linux/osx
* Add braces
* Re-use manifest. It appears to be required on linux.
* Undo previous commit -- it appears linux was simply never affected.
* Addressed AcK's comments
* Remove unused usings
* Address comments by AcK #2
* Re-order
* Move FromHwnd call to ForceDpiAware class. Wrap in Try-Catch to prevent crashes on systems that don't support it.
* Additional code cleanup
* Remove "global::" reference.
This PR fix an issue introduced on last IPC rewrite PRs where some returned buffer size have to be explicit now.
`GetClientId` without an explicit buffer size return some garbage to the guest and then `nifm` service crashes because of a wrong ClientId. It's fixed now.
Horizon Chase Turbo and Doom regression are fixed now. (Probably some other games too)
