The GPU class has been extended in this for Vulkan initialization, this is done to the point of initializing the instance alongside loading in `VK_LAYER_KHRONOS_validation` which is also now packed into all Debug APKs for Skyline. In addition, `VK_EXT_debug_report` is also initialized and it's output is piped directly into the Logger.
A minor change regarding the type of the `Fps` and `Frametime` globals was changed to `skyline::i32`s which is a more suitable type due to those having a smaller chance of overflowing while being signed as Java doesn't have unsigned integral types.
bugs
An engine is effectively a HW block in the GPU, the main one is the
Maxwell 3D which is used for 3D graphics. Engines can be bound to
individual subchannels and then methods within them can be called
through pushbuffers.
The engine side of the GPFIO is also included, it currently does nothing
but will need to be extended in the future with semaphores.
This commit does some minor renaming/reordering in IPC and adds support for strings to IPC Push/Pop. It also fixes a tiny regression with the frametime display.
This commit mainly finishes up refactor by fixing everything brought up in the CR + Improving NCE somewhat and actually killing the child processes properly now.
We earlier moved to LGPLv3.0 or Later. This was a mistake as what we wanted was being able to link to proprietary libraries but LGPL is the opposite and it allows linking proprietary libraries to libskyline instead. After further consideration, we've moved to MPL-2.0, it allows linking to proprietary libraries and is a standardized license as compared to adding an exception to GPL.
This commit adds performance statistics to the emulator that can be toggled in preferences. The layout of `EmulationActivity` was also changed from `ConstraintLayout` to `RelativeLayout` due to poor performance of the former.
This commit adds mutexes to the logger so they produce a valid log file rather than breaking due to a race condition. It also introduced `util::MakeMagic` so the magic functions are far more clear. A small refactor of IPC was also done which cleared up some of the for loops.
This commit addresses the incorrect hierarchy of the GPU and refactors them at the same time. Now, the hierarchy much closely matches HOS. This commit also introduces a texture classes, albeit they're not complete and only partially implemented.
This commit fixed the issues outlined in the CR (Mainly correlated to formatting), moves to a sorted vector from a sorted list for the memory map in addition to using binary search for sorting through rather than iteratively and fixes item duplication in the game list when directory is changed in Settings.
This commit mainly fixes GroupMutex and clock rescaling. In addition, clock rescaling is no longer performed if the CNTFRQ_EL0 of the host device is same as that of the Switch (19.2MHz) which is fairly common on higher end devices.
This commit makes the kernel completely thread-safe and fixes an issue that caused libNX games to not work due to an error with KSharedMemory. In addition, implement GroupMutex to allow the kernel threads to run in parallel but still allow them to not overlap with the JNI thread.
This commit makes the kernel thread safe in some instances (but not fully) and showcases the significantly improved performance over the ptrace method used prior. In addition, scanning for ROMs is now done asynchronously.
This commit is a huge step in the direction of better performance, as we move from ptrace to junction branching and have kernel call overhead similar to that of a native kernel call! In addition, this sets the base for the kernel to go fully multi-threaded. However, the kernel is currently not thread-safe and therefore this commit currently causes a crash.
This commit is the start of moving towards a lockless and faster kernel which can run multiple independent threads with fast userspace synchronization.
This commit fixes JNI race conditions by usage of a mutex, fixes a bug in release builds due to ProGuard member obfuscation and fix searching by fixing the HeaderAdapter filter.
This commit adapts the C++ backend to the Kotlin frontend by moving to usage of file descriptors and, provides an interface to access frontend code via JNI which is used to check the state of the activity and catch events such as surface destruction. In addition, this commit fixes some minor linting errors and changes the CMake version to 3.10.2+.
This commit changes how IPC is interacted with in two ways:
* Simplify all buffers into just InputBuffer and OutputBuffer
* Use pushing/popping for data payload
The deswizzling implementation currently writes linearly and reads non-linearly, this is non optimal as the MMU cannot read ahead. This flips that and reads linearly while it writes non-linearly. This is based on: 324a3624ac/nx/source/display/framebuffer.c (L189).