The-Homebrew-Cloud/skyline
2
0
Fork 0
mirror of https://github.com/skyline-emu/skyline.git synced 2024-11-23 12:49:18 +01:00
Code Issues Packages Projects Releases Wiki Activity

Use weak_ptr for TrapHandler Callbacks

Browse Source 
The lifetime of the `this` pointer in the trap callbacks could be invalid as the lifetime of the underlying `Buffer`/`Texture` object wasn't guaranteed, this commit fixes that by passing a `weak_ptr` of the objects into the callbacks which is locked during the callbacks and ensures that a destroyed object isn't accessed.

Co-authored-by: Billy Laws <blaws05@gmail.com>
This commit is contained in:
PixelyIon 2022-07-25 22:52:45 +05:30
parent 96d8676d5b
commit 9d294b9ccc
No known key found for this signature in database
GPG Key ID: 11BC6C3201BC2C05
6 changed files with 58 additions and 21 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

30
app/src/main/cpp/skyline/gpu/buffer.cpp
View File

@ -15,23 +15,37 @@ namespace skyline::gpu {
alignedMirror = gpu.state.process->memory.CreateMirror(span<u8>{alignedData, alignedSize});
mirror = alignedMirror.subspan(static_cast<size_t>(guest->data() - alignedData), guest->size());
trapHandle = gpu.state.nce->TrapRegions(*guest, true, [this] {
std::scoped_lock lock{*this};
}, [this] {
std::unique_lock lock{*this, std::try_to_lock};
// We can't just capture `this` in the lambda since the lambda could exceed the lifetime of the buffer
std::weak_ptr<Buffer> weakThis{weak_from_this()};
trapHandle = gpu.state.nce->TrapRegions(*guest, true, [weakThis] {
auto buffer{weakThis.lock()};
if (!buffer)
return;
std::scoped_lock lock{*buffer};
}, [weakThis] {
auto buffer{weakThis.lock()};
if (!buffer)
return true;
std::unique_lock lock{*buffer, std::try_to_lock};
if (!lock)
return false;
SynchronizeGuest(true); // We can skip trapping since the caller will do it
return true;
}, [this] {
}, [weakThis] {
auto buffer{weakThis.lock()};
if (!buffer)
return true;
DirtyState expectedState{DirtyState::Clean};
if (dirtyState.compare_exchange_strong(expectedState, DirtyState::CpuDirty, std::memory_order_relaxed) || expectedState == DirtyState::CpuDirty)
if (buffer->dirtyState.compare_exchange_strong(expectedState, DirtyState::CpuDirty, std::memory_order_relaxed) || expectedState == DirtyState::CpuDirty)
return true; // If we can transition the buffer to CPU dirty (from Clean) or if it already is CPU dirty then we can just return, we only need to do the lock and corresponding sync if the buffer is GPU dirty
std::unique_lock lock{*this, std::try_to_lock};
std::unique_lock lock{*buffer, std::try_to_lock};
if (!lock)
return false;
SynchronizeGuest(true, false, true); // We need to assume the buffer is dirty since we don't know what the guest is writing
buffer->SynchronizeGuest(true, false, true); // We need to assume the buffer is dirty since we don't know what the guest is writing
return true;
});
}

6
app/src/main/cpp/skyline/gpu/buffer.h
View File

@ -117,7 +117,7 @@ namespace skyline::gpu {
friend BufferManager;
/**
* @brief Sets up mirror mappings for the guest mappings
* @brief Sets up mirror mappings for the guest mappings, this must be called after construction for the mirror to be valid
*/
void SetupGuestMappings();
@ -138,6 +138,10 @@ namespace skyline::gpu {
return span<u8>(backing);
}
/**
* @brief Creates a buffer object wrapping the guest buffer with a backing that can represent the guest buffer data
* @note The guest mappings will not be setup until SetupGuestMappings() is called
*/
Buffer(GPU &gpu, GuestBuffer guest);
/**

2
app/src/main/cpp/skyline/gpu/buffer_manager.cpp
View File

@ -69,6 +69,7 @@ namespace skyline::gpu {
LockedBuffer newBuffer{std::make_shared<Buffer>(gpu, span<u8>{lowestAddress, highestAddress}), tag}; // If we don't lock the buffer prior to trapping it during synchronization, a race could occur with a guest trap acquiring the lock before we do and mutating the buffer prior to it being ready
newBuffer->SetupGuestMappings();
newBuffer->SynchronizeHost(false); // Overlaps don't necessarily fully cover the buffer so we have to perform a sync here to prevent any gaps
auto copyBuffer{[](auto dstGuest, auto srcGuest, auto dstPtr, auto srcPtr) {
@ -164,6 +165,7 @@ namespace skyline::gpu {
if (overlaps.empty()) {
// If we couldn't find any overlapping buffers, create a new buffer without coalescing
LockedBuffer buffer{std::make_shared<Buffer>(gpu, guestMapping), tag};
buffer->SetupGuestMappings();
buffer->SynchronizeHost();
InsertBuffer(*buffer);
return buffer->GetView(offset, size);

37
app/src/main/cpp/skyline/gpu/texture/texture.cpp
View File

@ -143,25 +143,40 @@ namespace skyline::gpu {
mirror = alignedMirror.subspan(static_cast<size_t>(frontMapping.data() - alignedData), totalSize);
}
trapHandle = gpu.state.nce->TrapRegions(mappings, true, [this] {
std::scoped_lock lock{*this};
}, [this] {
std::unique_lock lock{*this, std::try_to_lock};
// We can't just capture `this` in the lambda since the lambda could exceed the lifetime of the buffer
std::weak_ptr<Texture> weakThis{weak_from_this()};
trapHandle = gpu.state.nce->TrapRegions(mappings, true, [weakThis] {
auto texture{weakThis.lock()};
if (!texture)
return;
std::scoped_lock lock{*texture};
}, [weakThis] {
auto texture{weakThis.lock()};
if (!texture)
return true;
std::unique_lock lock{*texture, std::try_to_lock};
if (!lock)
return false;
SynchronizeGuest(true); // We can skip trapping since the caller will do it
WaitOnFence();
texture->SynchronizeGuest(true); // We can skip trapping since the caller will do it
texture->WaitOnFence();
return true;
}, [this] {
}, [weakThis] {
auto texture{weakThis.lock()};
if (!texture)
return true;
DirtyState expectedState{DirtyState::Clean};
if (dirtyState.compare_exchange_strong(expectedState, DirtyState::CpuDirty, std::memory_order_relaxed) || expectedState == DirtyState::CpuDirty)
if (texture->dirtyState.compare_exchange_strong(expectedState, DirtyState::CpuDirty, std::memory_order_relaxed) || expectedState == DirtyState::CpuDirty)
return true; // If we can transition the texture to CPU dirty (from Clean) or if it already is CPU dirty then we can just return, we only need to do the lock and corresponding sync if the texture is GPU dirty
std::unique_lock lock{*this, std::try_to_lock};
std::unique_lock lock{*texture, std::try_to_lock};
if (!lock)
return false;
SynchronizeGuest(true, true); // We need to assume the texture is dirty since we don't know what the guest is writing
WaitOnFence();
texture->SynchronizeGuest(true, true); // We need to assume the texture is dirty since we don't know what the guest is writing
texture->WaitOnFence();
return true;
});
}

3
app/src/main/cpp/skyline/gpu/texture/texture.h
View File

@ -390,7 +390,7 @@ namespace skyline::gpu {
friend TextureView;
/**
* @brief Sets up mirror mappings for the guest mappings
* @brief Sets up mirror mappings for the guest mappings, this must be called after construction for the mirror to be valid
*/
void SetupGuestMappings();
@ -460,6 +460,7 @@ namespace skyline::gpu {
/**
* @brief Creates a texture object wrapping the guest texture with a backing that can represent the guest texture data
* @note The guest mappings will not be setup until SetupGuestMappings() is called
*/
Texture(GPU &gpu, GuestTexture guest);

1
app/src/main/cpp/skyline/gpu/texture_manager.cpp
View File

@ -87,6 +87,7 @@ namespace skyline::gpu {
// Create a texture as we cannot find one that matches
auto texture{std::make_shared<Texture>(gpu, guestTexture)};
texture->SetupGuestMappings();
texture->TransitionLayout(vk::ImageLayout::eGeneral);
auto it{texture->guest->mappings.begin()};
textures.emplace(mappingEnd, TextureMapping{texture, it, guestMapping});