The Kepler compute engine is used to run compute jobs encapsulated in to QMDs on the GPU, this commit doesn't implement compute itself but adds the register and QMD structs that will be needed for it in the future.
We wanted views to extend the lifetime of the underlying buffers and at the same time preserve all views until the destruction of the buffer to prevent recreation which might be costly in the future when we need `VkBufferView`s of the buffer but also require a centralized list of all views for recreation of the buffer. It also removes the inconsistency between `BufferView*` being returned in `GetXView` in `GraphicsContext`.
Alised descriptor sets are incorrectly interpreted by the shader compiler causing it to bugger up LLVM function argument types and crash
Co-authored-by: PixelyIon <pixelyion@protonmail.com>
This controls the depth range used by the shader, hades already has support for the necessary patching so we only need to pass the current mode over to it and it'll do the necessary work.
Using `eB5G6R5UnormPack16` (with a swizzle for `R5G6B5Unorm`) removes the need for `VK_IMAGE_CREATE_MUTABLE_FORMAT_BIT` when those formats are aliased which happens in Sonic Mania among other titles.
Adreno GPUs have significant performance penalties from usage of `VK_IMAGE_CREATE_MUTABLE_FORMAT_BIT` which require disabling UBWC and on Turnip, forces linear tiling. As a result, it's been made an optional quirk which doesn't supply the flag in `VkImageCreateInfo` and logs a warning if a view with a different Vulkan format from the original image is created.
We often need to alias the underlying data as multiple Vulkan formats which requires the `eMutableFormat` bit to be set in `VkImageCreateInfo`, without doing this there'll be validation layer errors and potentially GPU bugs.
As we no longer set the layout to general inside the Texture constructor, yet, we need it to be set prior to the image being used as an attachment. We need to transition the layout to `eGeneral` after creation of the texture object.
Any `RecyclerView`s with an app bar in a `CoordinatorLayout` would end up going off-screen due to the layout behavior implementing an offset by using a transform which would not correctly handle focusing on off-screen objects. This has now been fixed by manually adjusting height to be clipped to what is visible on the screen.
We collapse the app bar when the focus is on the app list which only occurs while using a controller, this is required as the app bar will never be collapsed otherwise. It also removes the older code to work around the limitation on `View.FOCUS_DOWN` by collapsing only when the end of the list was reached.
Removes card elevation as it visually conflicts with the scrim, this also makes the scrim a bit darker to emphasize the text and slightly reduces the border radius.
The entire layout is now selectable for grid items rather than just the card, this greatly increases the visibility of the selection when not in touch mode as the contrast of a darken effect on the icon can be minimal depending on how dark the icon already is.
The `InputStream` would not be closed after reading the key file in `KeyReader#import`, it's now wrapped with `use{ }` which handles closing the stream after usage.
Setting the refresh rate via the Display API's`preferredDisplayModeId` is an outdated method to do it on Android 11 and above, we now use `Surface#setFrameRate` alongside it to suggest a refresh rate for the display.
We incorrectly determined an Adreno driver bug to require padding between binding slots but the real issue was not supporting consecutive binding writes for `VK_DESCRIPTOR_TYPE_COMBINED_IMAGE_SAMPLER` and was fixed by the padding slot unintentionally requiring individual writes. The quirk has now been corrected to explicitly specify this as the bug and the solution is more apt.
Any lookups done using `GetAlignedRecursiveRange` incorrectly added intervals in the exclusive interval entry lookups as the condition for adding them was the reverse of what it should've been due to a last minute refactor, it led to graphical glitches and crashes. This has been fixed and the lookups should return the correct results.
On certain devices, accesses to a protected memory region can return `si_code` as non-`SEGV_ACCERR` values, this leads to a crash as we only pass access violations to the trap handler and would lead to not doing so on those devices which would then result in going to the crash handler.
A large amount of Texture/Buffer views would expire before reuse could occur in `Texture::GetView`/`Buffer::GetView`. These can lead to a substantial memory allocation given enough time and they are now deleted during the lookup while iterating on all entries.
It should be noted that there are a lot of duplicate views that don't live long enough to be reused and the ultimate solution here is to make those views live long enough to be reused.
Similar to constant redundant synchronization for textures, there is a lot of redundant synchronization of buffers. Albeit, buffer synchronization is far cheaper than texture synchronization it still has associated costs which have now been reduced by only synchronizing on access.
There was a lot of redundant synchronization of textures to and from host constantly as we were not aware of guest memory access, this has now been averted by tracking any memory accesses to the texture memory using the NCE Memory Trapping API and synchronizing only when required.
An API for trapping accesses to guest memory and performing callbacks based on those accesses alongside managing protection of the memory. This is a fundamental building block for avoiding redundant synchronization of resources from the guest and host.
Note: All accesses are treated as write accesses at the moment, support for picking up read accesses will be implemented later
An interval map is a crucial piece of infrastructure required for memory faulting to track any regions that have an associated callback and their protection. Additionally, efficient page-aligned lookups with semantics optimal for memory faulting are also a requirement and the ability to associate multiple regions with a single callback/protection entry rather than doing so on a per-region basis as we deal with split-mapping resources.
This is a prerequisite to memory trapping as we need to write to the mirror to avoid a race condition with external threads writing to a texture/buffer while we do so ourselves for the sync on a read/write, it also avoids an additional `mprotect` to `-WX`/`RWX` on a read access.
An additional advantage for textures especially is that we now support split-mapping textures due to laying them out in a contiguous mirror and they will not require costly algorithmic changes. Buffers should also benefit from not needing to iterate over every region when they are split into multiple mappings.
`CreateMirror` is limited to creating a mirror of a single contiguous region which does not work when creating a contiguous mirror of multiple non-contiguous regions. To support this functionality, `CreateMirrors` which expects a list of page-aligned regions and maps them into a contiguous mirror.
We want to create arbitrary mirrors in the guest address space and to make this possible, we map the entire address space as a shared memory file. A mirror is mapped by using `mmap` with the offset into the guest address space.
Previously for methods with count > 1 the subchannel and engine would be looked up for each part of the method rather than only doing so at the start. Each call also needed to be looked up to see if it touched a macro or GPFIFO method. Fix this by doing checks outside of the main dispatch loop with templated helper lambdas to avoid needing to repeat lots of code. Maxwell3D is the only subchannel with a fast path for now but more can be added later if needed.
Almost every Maxwell format now directly corresponds to a Vulkan format. This allows formats to be passed through and the swizzle used directly from guest (with some extra swizzle handling for edge cases) thus saving the need to explicitly support each swizzle combination which is adds a lot of code bloat. The format header is additionally reordered with line breaks to separate formats by their bits-per-block.
We always submit pipeline divisor descriptions regardless of binding input rate being vertex rather than instance. This is invalid behavior and has been fixed by only submitting binding descriptors when the input rate is per-instance.
Adreno proprietary drivers suffer from a bug where `VK_DESCRIPTOR_TYPE_COMBINED_IMAGE_SAMPLER` requires 2 descriptor slots rather than one, we add a padding slot to fix this issue. `QuirkManager` was introduced to handle per-vendor/per-device errata and allow enabling this on Adreno proprietary drivers specifically as to not affect the performance of other devices.
Quirk terminology was deemed to be inappropriate for describing the features/extensions of a device. It has been replaced with traits which is far more fitting but quirks will be used as a terminology for errata in devices.
The texture handle offset calculation involved an incorrect shift by descriptor size which was found to be unnecessary and would result in an invalid handle that had the wrong TIC/TSC index and caused broken rendering.
`nodes` and `syncTextures` were cleared after waiting on the `CommandExecutor` fence rather than before, this wasted execution time after the wait for something that could be performed prior to the wait.
We now attempt to enable `VK_KHR_uniform_buffer_standard_layout` when present as lax UBO layout significantly reduces complexity. If a device doesn't support this extension, we still assume that the device supports it implicitly as this has proven to be true across all major mobile GPU vendors regardless of the driver version but enabling this prevents validation layer errors.
We depend on past commands to have completed execution in a renderpass, a subpass dependency on all graphics stages from `VK_SUBPASS_EXTERNAL` to subpass #0 is used to enforce this. Nvidia and Adreno proprietary drivers implicitly do this but Turnip or Mali drivers require this or they execute out of order.
Blocklinear texture decoding was broken for padding blocks and would incorrectly decode them resulting in major texture corruption for any textures with their widths not aligned to 64 bytes. This has now been fixed with neater code which avoids redundant repetition of any code using lambdas and functions where necessary.
Stencil operations are configurable to be the same for both sides or have independent stencil state for both sides. It is controlled via the previously unimplemented `stencilTwoSideEnable`.
Fermi2D supports macros in addition to Maxwell3D, these both share code memory. To support this we rework the macro interpreter to support passing in a target engine and abstract the communications out into an interface that can be implemented by applicable engines.
```
GPFIFO <-> MME <-> Maxwell3D
^ ^---> Fermi2D
X------------> I2M
X------------> MaxwellComputeB
X--Flush-----> MaxwellDMA
```
Shader programs allocate instructions and blocks within an `ObjectPool`, there was a global pool prior that was never reaped aside from on destruction. This led to a leak where the pool would contain resources from shader programs that had been deleted, to avert this the pools are now tied to shader programs.
The size of blocklinear textures did not consider alignment to Block/ROB boundaries before, it is aligned to them now. Incorrect sizes led to textures not being aliased correctly due to different size calculations for GraphicBufferProducer surfaces and Maxwell3D color RTs.
erase invalidated `it` leading to a potential segfault if the GPU was very far behind, bail out early to avoid that since there can only be one occurence at most in the buffer anyway.
Implements the entirety of Maxwell3D Depth/Stencil state for both faces including compare/write masks and reference value. Maxwell3D register `stencilTwoSideEnable` is ignored as its behavior is unknown and could mean the same behavior for both stencils or the back facing stencil being disabled as a result of this it is unimplemented.
We don't respect the host subresource layout in synchronizing linear textures from the guest to the host when mapped to memory directly, this leads to texture corruption and while the real fix would involve respecting the host subresource layout, this has been deferred for later as real world performance advantages/disadvantages associated with this change can be observed more carefully to determine if it's worth it.
Color RTs are disabled by setting their format as `None`, it was removed while transitioning to macros and resulted in a missing format exception. It has been readded as several applications depend on this behavior.
Using `std::vector` for shader bytecode led to a lot of reallocation due to constant resizing, switching over the static vector allows for a single static allocation of the maximum possible guest shader size (1 MiB) to be done for every stage resulting in a 6 MiB preallocation which is unnoticeable given the total memory overhead of running a Switch application.
The `OneMinusSourceAlpha` blending factor was converted to `eOneMinusSrcColor` rather than `eOneMinusSrcAlpha` leading to incorrect blending behavior in certain titles. A similar issue with the order of `MinimumGL`/`MaximumGL` and `SubtractGL`/`ReverseSubtractGL` being the opposite of what it should've been, both of these issues have been fixed.
`NextSubpassNode` didn't increment `subpassIndex` which runs commands with the wrong subpass index resulting in them accessing invalid attachments or other bugs that may arise from using the wrong subpass.
All Maxwell3D state was passed by reference to the draw command lambda, this would break if there was more than one pass or the state was changed in any way before execution. All state has now been serialized by value into the draw command lambda capture, retaining state regardless of mutations of the class state.
Any usage of a resource in a command now requires attaching that resource externally and will not be implicitly attached on usage, this makes attaching of resources consistent and allows for more lax locking requirements on resources as they can be locked while attaching and don't need to be for any commands, it also avoids redundantly attaching a resource in certain cases.
If an object is attached to a `FenceCycle` twice then it would cause `FenceCycleDependency::next` to be overwritten and lead to destruction of dependencies prior to the fence being signaled causing usage of deleted resources. This commit fixes this by tracking what fence cycle a dependency is currently attached to and doesn't reattach if it's already attached to the current fence cycle.
An assumption was hardcoded into `Shader::Profile` regarding devices supporting demotion of shader invocations to helpers. This assumption wasn't backed by enabling the `VK_EXT_shader_demote_to_helper_invocation` extension via a quirk leading to assertions when it was used by the shader compiler, a quirk has now been added for the extension and is supplied to the shader compiler accordingly.
If the controller type was changed from a type with a larger amount of buttons/axes to one with a fewer amount, a crash would occur due to the transition animation retaining those elements as children yet returning `NO_POSITION` from `getChildAdapterPosition` in `DividerItemDecoration` which was an unhandled case and led to an OOB array access.
A bug caused by not passing the index argument to `ControllerActivity` led to all preferences opening the activity that pertained to Controller #1. This was fixed by passing the `index` argument in the activity launch intent.
Fixes texture corruption due to incorrect synchronization, the barrier would not enforce waiting till the texture was entirely rendered causing an incomplete texture to be downloaded which lead to rendering bugs for certain GPUs including ARM's Mali GPUs.
A bug caused an assertion if both `VK_EXT_custom_border_color` and `VK_EXT_vertex_attribute_divisor` due to mistakenly unlinking `PhysicalDeviceVertexAttributeDivisorFeaturesEXT` instead of `PhysicalDeviceCustomBorderColorFeaturesEXT` when `VK_EXT_custom_border_color` isn't supported which would potentially lead to unlinking the same structure twice and cause the assertion.
Implements inline constant buffer updates that are written to the CPU copy of the buffer rather than generating an actual inline buffer write, this works for TIC/TSC index updates but won't work when the buffer is expected to actually be updated inline with regard to sequence rather than just as a buffer upload prior to rendering.
GPU-sided constant buffer updates will be implemented later with optimizations for updating an entire range by handling GPFIFO `Inc`/`NonInc`directly and submitting it as a host inline buffer update.
There should only ever be a single instance of a `ActiveDescriptorSet` that tracks the lifetime of a descriptor set as the destructor is responsible for freeing the descriptor set.
There are cases where a new object inheriting the descriptor set needs to be created in these cases we need to have move semantics and make the destructor of the prior object inert, this allows for moving to the new object without any side effects. If the copy constructor was used in these cases the older object would free the set on its destruction which would lead to the set being invalid on existing instances which is incorrect behavior and would likely lead to driver crashes.
The descriptor sets should now contain a combined image and sampler handle for any sampled textures in the guest shader from the supplied offset into the texture constant buffer.
Note: Games tend to rely on inline constant buffer updates for writing the texture constant buffer and due to it not being implemented, the value will be read as 0 which is incorrect.
We want read semantics inside the constant buffer object via the mappings to avoid a pointless GPU VMM mapping lookup. It is a fairly frequent operation so this is necessary, the ability to write directly will be added in the future as well.
Implements parsing for the Maxwell 3D TIC pool and conversion of a TIC into a `GuestTexture`, support is limited to pitch-linear RGB565/A8R8G8B8 textures at the moment but will be extended as games utilize more formats and layouts. Support for 1D buffers is also omitted at the moment since they need special handling with them effectively being treated as buffers in Vulkan rather than images.
The pitch of the texture should always be supplied in terms of bytes as it denotes alignment on a byte boundary rather than a pixel one, it is also always utilized in terms of bytes rather than pixels so this avoids an unnecessary conversion.
Note: GBP stride unit was assumed to be pixels earlier but is likely bytes which is why there are no changes to the supplied value there, if this is not the case it'll be fixed in the future
Maxwell3D `TextureSamplerControl` (TSC) are fully converted into Vulkan samplers with extension backing for all aspects that require them (border color/reduction mode) and approximations where Vulkan doesn't support certain functionality (sampler address mode) alongside cases where extensions may not be present (border color).
Code involving caching of mappings was copied from `RenderTarget` without much consideration for applicability in buffers, the reason for caching mappings in RTs was that the view may be invalidated by more than the IOVA/Size being changed but this doesn't hold true for buffers generally so invalidation can only be on the view level with the mappings being looked up every time since the invalidation would likely change them.
`std::hash` doesn't have a generic template where it can be utilized for arbitrary trivial objects and implementing this might result in conflicts with other types. To fix this a generic templated hash is now provided as a utility structure, that can be utilized directly in hash-based containers such as `unordered_map`.
Nullability allow for optional semantics where a span may be explicitly invalidated with `nullptr` being used as a sentinel value for it and a boolean operator that allows trivial checking for if the span is valid or not.
Adds support for index buffers including U8 index buffers via the `VK_EXT_index_type_uint8` extension which has been added as an optional quirk but an exception will be thrown if the guest utilizes it but the host doesn't support it.
Add support for parsing and combining `VertexA` and `VertexB` programs into a single vertex pipeline program prior to compilation, atomic reparsing and combining is supported to only reparse the stage that was modified and recombine once at most within a single pipeline compilation.
Atomically invalidate pipeline stages as runtime information that pertains to them changes rather than never recompiling pipelines on runtime information being updated resulting in out of date pipelines or recompiling all pipelines on any runtime information updates.
Shader compilation is now broken into shader program parsing and pipeline shader compilation which will allow for supporting dual vertex shaders and more atomic invalidation depending on runtime state limiting the amount of work that is redone.
Bindings are now properly handled allowing for bound UBOs to be converted to the appropriate host UBO as designated by the shader compiler by creating Vulkan Descriptor Sets that match it.
We need this to make the distinction between a shader and pipeline stage in as shader programs are bound at a different rate than that of pipeline stage resources such as UBO.
An instance of `Shader::Backend::Bindings` must be retained across all stages for correct emission of bindings, which is now done inside `GraphicsContext::GetShaderStages`.
The vertex attribute types supplied prior were just the default which is `Float`, this works for some cases but will entirely break if the attribute type isn't a float. The attribute types are now set correctly.
Only copying a single aspect was supported by `CopyIntoStagingBuffer` earlier due to not supplying a `VkBufferImageCopy` for each aspect separately, this has now been done with Color/Depth/Stencil aspects having their own `VkBufferImageCopy` for the `VkCmdCopyImageToBuffer` command.
The definition of the `TextureView` class was spread across `texture.cpp` and has now been moved to the top of the file above the other half of the definition.
A buffer with 0 as the start/end IOVA should be invalid as there shouldn't be any mappings at 0 in GPU VA, titles such as Puyo Puyo Tetris configure the Vertex Buffer with 0 IOVAs which leads to a segmentation fault without this exception.
The lifetime of a texture and buffer view is now bound by the `FenceCycle` in `CommandExecutor`, this ensures that a `VkImageView` isn't destroyed prior to usage leading to UB.
The lifetime of all textures bound to a RenderPass alongside syncing of textures is already handled by `CommandExecutor` and doesn't need to be redundantly handled by `RenderPassNode`. It's been removed as a result of this.
Adds the depth/stencil RT as an attachment for the draw but with `VkPipelineDepthStencilStateCreateInfo` stubbed out, it'll not function correctly and the contents will not be what the guest expects them to be.
Support for clearing the depth/stencil RT has been added as its own function via either optimized `VkAttachmentLoadOp`-based clears or `vkCmdClearAttachments`. A bit of cleanup has also been done for color RT clears with the lambda for the slow-path purely calling the command rather than creating the parameter structures.
Implements `AddClearDepthStencilSubpass` in `CommandExecutor` which is similar to `ClearColorAttachment` in that it uses `VK_ATTACHMENT_LOAD_OP_CLEAR` for the clear which is far more efficient than using `VK_ATTACHMENT_LOAD_OP_LOAD` then doing the clear.
The stage/access mask for `VkSubpassDependency` were hardcoded to only be valid for color attachments earlier, this has now been fixed by branching based on the format aspect.
Sets `VkImageUsageFlags` correctly rather than hardcoding it for color attachments and adds multiple `VkBufferImageCopy` to `VkCmdCopyBufferToImage` for Color/Depth/Stencil aspects of an image.
Support the Maxwell3D Depth RT for Z-buffering, this just creates an equivalent `RenderTarget` object with no support on the API-user side (IE: `Draw` and `ClearBuffers`).
This prefixes all RT functions that deal with color RTs with `Color` and abstracts out common functions that will be used for both color and depth RTs. All common Maxwell3D structures are also moved out of the `ColorRenderTarget` (`RenderTarget` previously) structure.
To allow for caching of pipelines on the host a `VkPipelineCache` has been added, it is entirely in-memory and is not flushed to the disk which'll be done in the future alongside caching guest shaders to further avoid translation where possible.
Uses all Maxwell3D state converted into Vulkan state to do an equivalent draw on the host GPU, it sets up RT/Vertex Buffer/Vertex Attribute/Shader state and creates a stubbed out `VkPipelineLayout` for the draw. Any descriptor state isn't currently handled and is yet to be implemented, currently there's no Vulkan pipeline cache supplied which will be implemented subsequently.
We require a handle to the current renderpass and the index of the subpass in certain cases, this is now tracked by the `CommandExecutor` and passed in as a parameter to `NextSubpassFunctionNode` and the newly-introduced `SubpassFunctionNode`.
Switch from `SubmitWithCycle` to manually allocating the active command buffer to tag dependencies with the `FenceCycle` that prevents them from being mutated prior to execution. This new paradigm could also allow eager recording of commands with only submission being deferred.
`CommandScheduler` API users can now directly allocate an active command buffer that they need to manage alongside its fence, this can allow for more efficient recording as it doesn't need to be immediately submitted after, it can also allow attaching objects to a `FenceCycle` prior to submission that can be useful for locking resources.
