* Pushbuffer data is now stored in a member buffer to avoid reallocating
it for each pushbuffer which hampered performance before.
* Don't prefetch pushbuffers as it puts unnecessary load on the guest
thread that is better suited for the GPFIFO thread.
* Clean up some misc code to avoid pointless casts of a 4 byte object
and handle GPFIFO control opcodes.
NvHostEvents were renamed to SyncpointEvents which is a much clearer
name that more accurately describes them. Locking is needed as IOCTLs
can be called asynchronously and so event registration and signalling
can race.
The following scheduler bugs were fixed:
* It was assumed that all non-cooperative `Rotate` calls were from a preemptive yield and changed the state of `KThread::isPreempted` incorrectly which could lead to UB, an example of a scenario with it would be:
* * Preemptive thread A gets a signal to yield from cooperative thread B due to it being ready to schedule and higher priority
* * A complies with this request but there's an assumption that the signal was actually from it's preemption timer therefore it doesn't reset it (As it isn't required if the timer was responsible for the signal)
* * A receives the actual preemption signal a while later, causing UB as the signal handler is invoked twice
* `Scheduler::UpdatePriority`
* * A check for `currentIt == core->queue.begin()` existed which caused an incorrect early return
* * The preemption timer was armed correctly when a priority transition from cooperative priority -> preemption priority occurred but not disarmed when a transition from preemption priority -> cooperative priority occurred
* * The timer was unnecessarily disarmed in the case of updating the priority of a non-running thread, this isn't as much a bug as it is just pointless
* Priority inheritance in `KProcess::MutexLock` is fundamentally broken as it performs UB with `waitThread` being accessed prior to being assigned
* When a thread sets its own priority using `SvcSetThreadCoreMask` and its current core is no longer in the affinity mask, it wouldn't actually move to the new thread until the next time the thread is load balanced
This addresses all CR comments including more codebase-wide changes arising from certain review comments like proper usage of its/it's and consistent contraction of it is into it's.
An overhaul was made to the presentation and formatting of `KThread.h` and `LoadBalance` works has been superseded by `GetOptimalCoreForThread` which can be used alongside `InsertThread` or `MigrateToCore`. It makes the API far more atomic and neater. This was a major point of contention for the design prior, it's simplified some code and potentially improved performance.
The case of a thread not being in the core queue during a non-cooperative core affinity change would break things as the thread was non-conditionally removed and inserted, this has been fixed by adding a check to see if the thread exists in the core's queue prior to migration. In addition, `yieldWithCoreMigration` was broken by the previous commit as the fallthrough was intentional and removing it cause core migration without a yield which led to breakage in certain circumstances. The mutex locking logic was also improved in `ConditionalVariableWait` to use atomics in a more effective manner with less atomic operations being performed overall.
The code region's size was previously set at the same value as it is for 36-bit ASes, this value is inadequate for certain larger games and needed to be expanded. We've chosen 4GiB as the new value which should easily encompass all Switch games.
The SVCs improvements are as follows:
* Make SVC logs more concise for:
* * `SleepThread`
* * `ClearEvent`
* * `CloseHandle`
* * `ResetSignal`
* * `WaitSynchronization` (Special case for single handle)
* * `ArbitrateLock`
* * `ArbitrateUnlock`
* * `WaitProcessWideKeyAtomic`
* * `SignalProcessWideKey`
* Fix unintentional fallthrough into `yieldWithoutCoreMigration` from `yieldWithCoreMigration` in `SleepThread`
* Return `result::InvalidState` when an unsignalled handle is reset in `ResetSignal`
* Return `Result{}` (Success) in `CancelSynchronization`
* Do not return `result::InvalidCurrentMemory` in `ArbitrateLock` as it's not a failure condition
* Make `count` in `WaitProcessWideKeyAtomic` a `i32` from a `u32`, zero and all negative values result in waking all waiters
The entirety of the address arbiter is implemented in this commit, all three arbitration types: `WaitIfLessThan`, `DecrementAndWaitIfLessThan` and `WaitIfEqual`, and all three signal types: `Signal`, `SignalAndIncrementIfEqual` and `SignalAndModifyBasedOnWaitingThreadCountIfEqual` have been implemented.
This allows any application which uses levent (Light Events) to function which includes titles such as ARMS.
We did not support migration of threads which were running in a non-cooperative manner, this was partially due to the dependence on per-core conditional variables rather than per-thread which made this harder to do programmatically. This has been fixed by moving to per-thread cvars and therefore the limitation can be removed, this feature is used by Unity games.
SvcClearEvent previously set the `signalled` flag directly rather than
calling `ResetSignal`, which skipped the locking necessary to make it
globally visible. Switch it to use `ResetSignal` to fix this.
We've moved to using RS and GS from ASCII as delimiters rather than
'\n' and '|', this allows more robust parsing and increases the
readability of the log files
This prevents a race where two threads could read at the same time and
end up using the wrong IV leading to garbage data being read. This
caused crashes in several games including Celeste.
This was causing a significant amount of sched thrashing and pinning a
core to 100% as games constantly updated audren, now change it to only
signal on buffer release.
This caused the menus in Sonic Mania to be nonfunctional, futhermore,
default init is not ran for the input structs so the default max
definition in CommonHeader never actually applied.
CircularQueue was looping around too early resulting in the wrong
pushbuffers being used. The debug logging is useful for interpreting the
GPU method call logs.
Exefs loading was changed to check if an NSO exists before trying to
read it, preventing exceptions that get annoying while debugging.
* 'Fix' memory accounting to not measure reserved regions
* Fix some copy bugs introduced by switch to span
* Correct remap the behaviour of Modify so it actually works
An exceptional signal handler allows us to convert an OS signal into a C++ exception, this allows us to alleviate a lot of crashes that would otherwise occur from signals being thrown during execution of games and be able to handle them gracefully.
* Fix alignment handling in NvHostAsGpu::AllocSpace
* Implement Ioctl{2,3} ioctls
These were added in HOS 3.0.0 in order to ease handling ioctl buffers.
* Introduce support for GPU address space remapping
* Fix nvdrv and am service bugs
Syncpoints are supposed to be allocated from ID 1, they were allocated
at 0 before. The ioctl functions were also missing from the service map
* Fix friend:u service name
* Stub NVGPU_IOCTL_CHANNEL_SET_TIMESLICE
* Stub IManagerForApplication::CheckAvailability
* Add OsFileSystem Directory support and add a size field to directory entries
The size field will be needed by the incoming HOS IDirectory support.
* Implement support for IDirectory
This is used by applications to list the contents of a directory.
* Address feedback
This patch reduces the burden of adding services significantly, rather
than having to create an enum entry and add strings in the constructor
it will all be determined at runtime through RTTI. A macro is also used
in the service creation case to reduce clutter.
* Fix NvHostCtrl:EventSignal event ID parsing
* Divide the audout buffer length by the sample size
* Correct audout channel quantity handling
* A few bugfixes for audio tracks
* * Correctly lock in CheckReleasedBuffers and only call the callback once
* * Check if the identifier queue is empty before accessing it's iterator
* Refactor audio to better fit the codestyle
* Explictly specify reference when using GetReference
* Fix CheckReleasedBuffers
This commit significantly increases the accuracy of the prior HID code due to testing on the Switch. It is now fully accurate in all supported scenarios, them being assignment mode, orientation, color writes and system properties. In addition, review comments were addressed and fixed in the PR.
This fixes a Joy-Con Pair bug which caused a crash when a partner device was set to none while being set as a partner. In addition, the following HID service functions were implemented:
* GetSupportedNpadStyleSet
* ActivateNpadWithRevision
* GetNpadJoyHoldType
* AcquireNpadStyleSetUpdateEventHandle
This commit adds support to the C++ end of things for controller configuration. It isn't targeting being 1:1 to HOS for controller assignment but is rather based on intuition of how things should be.
This commit adds in the UI for Controller Configuration to Settings, in addition to introducing the storage and loading of aforementioned configurations to a file that can be saved/loaded at runtime. This commit also fixes updating of individual fields in Settings when changed from an external activity.
This commit focuses on making the UI completely usable using a controller so that a user won't have to switch between their device's touch screen and a controller constantly.
This commit refactors the C++ end of Input so it'll be in line with the rest of the codebase and be ready for the extension with multiple players and controller configuration.
This commit contains the Kotlin side of the initial Input implementation, this is based on the work done in the `hid` branch in `bylaws/skyline`.
Co-authored-by: ◱ PixelyIon <pixelyion@protonmail.com>
This commit contains the C++ side of the initial Input implementation, this is based on the work done in the `hid` branch in `bylaws/skyline`.
Co-authored-by: ◱ PixelyIon <pixelyion@protonmail.com>
interpreter.
The Maxwell 3D engine handles all 3D rendering, currently only non
rendering related methods are implemented. Macros are small pieces of
code that run on the GPU and allow methods to be quickly called for
things like instanced drawing.
These are used to allow the CPU to synchronise with the GPU as it
reaches specific points in its command stream.
Also fixes an nvmap bug where a struct was incorrect.
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.
* Rework VFS to support creating and writing files and introduce OsFileSystem
OsFileSystem abstracts a directory on the device using the filesystem API.
This also introduces GetEntryType and changes FileExists to use it.
* Implement the Horizon FileSystem APIs using our VFS framework
Horizon provides access to files through its IFileSystem class, we can
closely map this to our vfs::FileSystem class.
* Add support for creating application savedata
This implements basic savedata creation using the OsFileSystem API. The
data is stored in Skyline's private directory is stored in the same
format as yuzu.
* Make sure icons have a 1:1 ratio
* Use recyclerview padding to increase grid edge margins
* Fix race condition in searching roms
* Use notify insert for adapter
The GPU has it's own seperate address space to the CPU. It is able to
address 40 bit addresses and accesses the system memory. A sorted vector
has been used to store blocks as insertions are not very frequent.
unmapped regions
svcQueryMemory will return a valid descriptor for anything in the
address space, from 0 to 1 << addrSpaceBits, this was handled
incorrectly before and we were only returning descriptors if the
address was in a mapped region.
If an address in an unmapped region is requested then the extents of the
unmapped region up to the address space end are returned. If the address
requested is outside of the address space then the extents of the
inaccessible address space are returned.
To facilitate this support was added to MemoryManager::Get for
generating the extents of unmapped regions using the chunk list.
As the stack is automatically mapped in the guest by `clone` we do not
need to explicitly map it. This adds a flag to solve the issue.
Also mark the stack as stack rather than reserved.
Not zeroing the sample buffer causes issues when a voice is started but
is playing no samples. The system event handling was also reworked
according to Thog's info.
This fixes two bugs in IPC that were discovered when running Puyo Puyo
Tetris.
The CloneCurrentObject control IPC will now correctly return the handle
of the newly created object through move handles, rather than pushing it
as a result.
The size array of u16s with the sizes of each C buffer is now taken into
account when reading them. Before this change C buffers were entirely
broken.
This implements the base account service and stubs
InitializeApplicationInfoV0 which is used by Puyo Puyo Tetris. Support
for the entirety of account services will be added in the future.
lm is used by applications to print messages to the system log. Log
messages are made up of a header and then several fields containing
metadata or string messages.
In the case of am, IStorage is used to exchange buffers of data such
as application launch parameters or an applets result. It has no
relation to fsp-srv's IStorage.
Fonts are stored in an array of TTF data with an 8 byte header
containing a magic and an XOR'd length. Instead of requiring users to
provide original Nintendo fonts we pack open source replacements.
They are generated with the scripts here
https://github.com/FearlessTobi/yuzu_system_archives. All the fonts are
licenced under the Open Font or Apache 2 License so we can include them
all freely.
An NSP (Nintendo Submission Package) is effectively a PFS0 containing
NCAs, there are also tickets and a CNMT file which contains metadata
about updates. The current implementation is very basic and only
support Control and Program NCAs which is enough for loading games.
Support for updates and dlc will be added at a later date.
Nintendo Content Archives are used to store the assets, executables
and updates of applications. They support holding either a PFS0 or a
RomFS.
An NCA's ExeFS can be loaded by placing each NSO sequentially into
memory, starting with rtld which will link them together.
Currently only decrypted NCAs are supported, encryption and BKTR
handling will be added at a later time.
RomFS is a hierarchial filesystem where each level is made up of a
linked list of files and child directories. It is used in NCAs to store
the applications icon as well as by applications themselves for
accessing assets.
Partition FS encapsulates both the HFS0 found in XCIs and the PFS0 used
for ExeFS images and NSPs, it is purely file based and has no support at
all for directories aside from the root.
Mapping and writing segments into memory is now handled by a common
function that can be shared between all loaders. All they need to do now
is to pack each segment into a common struct.
* Correctly handle -WithContext IPC Requests
They should be treated the same as the non WithContext variants.
* Only send domain data on non-control IPC responses
Control IPC doesn't make use of domains so we shouldn't send extra data
in the response.
* Add the IStorage implementation to CMakeLists
This commit adds support for reading the RomFS data from an NRO and
obtaining an IStorage handle to it through 'OpenDataStorageByCurrentProcess'.
There is currently only support for reading and no support
for enlarging or writing.
Also fixup a few capitalisation issues.
- The backing system provides a flexible way to access a a region of
abstract memory.
- It is currently barebones and only has support for reading data but
this will be expanded as necessary.
The current implementations are:
- OsBacking - A backing that abstracts a linux file descriptor
- RegionBacking - A backing that creates a region from a portion of an
existing one
