diff options
author | Daniel Baumann <daniel.baumann@progress-linux.org> | 2024-04-28 14:29:10 +0000 |
---|---|---|
committer | Daniel Baumann <daniel.baumann@progress-linux.org> | 2024-04-28 14:29:10 +0000 |
commit | 2aa4a82499d4becd2284cdb482213d541b8804dd (patch) | |
tree | b80bf8bf13c3766139fbacc530efd0dd9d54394c /gfx/layers/ipc/SharedSurfacesParent.cpp | |
parent | Initial commit. (diff) | |
download | firefox-2aa4a82499d4becd2284cdb482213d541b8804dd.tar.xz firefox-2aa4a82499d4becd2284cdb482213d541b8804dd.zip |
Adding upstream version 86.0.1.upstream/86.0.1upstream
Signed-off-by: Daniel Baumann <daniel.baumann@progress-linux.org>
Diffstat (limited to 'gfx/layers/ipc/SharedSurfacesParent.cpp')
-rw-r--r-- | gfx/layers/ipc/SharedSurfacesParent.cpp | 257 |
1 files changed, 257 insertions, 0 deletions
diff --git a/gfx/layers/ipc/SharedSurfacesParent.cpp b/gfx/layers/ipc/SharedSurfacesParent.cpp new file mode 100644 index 0000000000..844e3d83b3 --- /dev/null +++ b/gfx/layers/ipc/SharedSurfacesParent.cpp @@ -0,0 +1,257 @@ +/* -*- Mode: C++; tab-width: 8; indent-tabs-mode: nil; c-basic-offset: 2 -*- */ +/* vim: set ts=8 sts=2 et sw=2 tw=80: */ +/* This Source Code Form is subject to the terms of the Mozilla Public + * License, v. 2.0. If a copy of the MPL was not distributed with this + * file, You can obtain one at http://mozilla.org/MPL/2.0/. */ + +#include "SharedSurfacesParent.h" +#include "mozilla/DebugOnly.h" +#include "mozilla/gfx/Logging.h" +#include "mozilla/gfx/gfxVars.h" +#include "mozilla/gfx/GPUProcessManager.h" +#include "mozilla/layers/SharedSurfacesMemoryReport.h" +#include "mozilla/layers/SourceSurfaceSharedData.h" +#include "mozilla/layers/CompositorThread.h" +#include "mozilla/webrender/RenderSharedSurfaceTextureHost.h" +#include "mozilla/webrender/RenderSharedSurfaceTextureHostSWGL.h" +#include "mozilla/webrender/RenderThread.h" + +namespace mozilla { +namespace layers { + +using namespace mozilla::gfx; + +StaticMutex SharedSurfacesParent::sMutex; +StaticAutoPtr<SharedSurfacesParent> SharedSurfacesParent::sInstance; + +SharedSurfacesParent::SharedSurfacesParent() = default; + +SharedSurfacesParent::~SharedSurfacesParent() { + for (auto i = mSurfaces.Iter(); !i.Done(); i.Next()) { + // There may be lingering consumers of the surfaces that didn't get shutdown + // yet but since we are here, we know the render thread is finished and we + // can unregister everything. + wr::RenderThread::Get()->UnregisterExternalImageDuringShutdown(i.Key()); + } +} + +/* static */ +void SharedSurfacesParent::Initialize() { + MOZ_ASSERT(NS_IsMainThread()); + StaticMutexAutoLock lock(sMutex); + if (!sInstance) { + sInstance = new SharedSurfacesParent(); + } +} + +/* static */ +void SharedSurfacesParent::Shutdown() { + // The main thread should blocked on waiting for the render thread to + // complete so this should be safe to release off the main thread. + MOZ_ASSERT(wr::RenderThread::IsInRenderThread()); + StaticMutexAutoLock lock(sMutex); + sInstance = nullptr; +} + +/* static */ +already_AddRefed<DataSourceSurface> SharedSurfacesParent::Get( + const wr::ExternalImageId& aId) { + StaticMutexAutoLock lock(sMutex); + if (!sInstance) { + gfxCriticalNote << "SSP:Get " << wr::AsUint64(aId) << " shtd"; + return nullptr; + } + + RefPtr<SourceSurfaceSharedDataWrapper> surface; + sInstance->mSurfaces.Get(wr::AsUint64(aId), getter_AddRefs(surface)); + return surface.forget(); +} + +/* static */ +already_AddRefed<DataSourceSurface> SharedSurfacesParent::Acquire( + const wr::ExternalImageId& aId) { + StaticMutexAutoLock lock(sMutex); + if (!sInstance) { + gfxCriticalNote << "SSP:Acq " << wr::AsUint64(aId) << " shtd"; + return nullptr; + } + + RefPtr<SourceSurfaceSharedDataWrapper> surface; + sInstance->mSurfaces.Get(wr::AsUint64(aId), getter_AddRefs(surface)); + + if (surface) { + DebugOnly<bool> rv = surface->AddConsumer(); + MOZ_ASSERT(!rv); + } + return surface.forget(); +} + +/* static */ +bool SharedSurfacesParent::Release(const wr::ExternalImageId& aId, + bool aForCreator) { + StaticMutexAutoLock lock(sMutex); + if (!sInstance) { + return false; + } + + uint64_t id = wr::AsUint64(aId); + RefPtr<SourceSurfaceSharedDataWrapper> surface; + sInstance->mSurfaces.Get(wr::AsUint64(aId), getter_AddRefs(surface)); + if (!surface) { + return false; + } + + if (surface->RemoveConsumer(aForCreator)) { + wr::RenderThread::Get()->UnregisterExternalImage(id); + sInstance->mSurfaces.Remove(id); + } + + return true; +} + +/* static */ +void SharedSurfacesParent::AddSameProcess(const wr::ExternalImageId& aId, + SourceSurfaceSharedData* aSurface) { + MOZ_ASSERT(XRE_IsParentProcess()); + MOZ_ASSERT(NS_IsMainThread()); + StaticMutexAutoLock lock(sMutex); + if (!sInstance) { + gfxCriticalNote << "SSP:Ads " << wr::AsUint64(aId) << " shtd"; + return; + } + + // If the child bridge detects it is in the combined UI/GPU process, then it + // will insert a wrapper surface holding the shared memory buffer directly. + // This is good because we avoid mapping the same shared memory twice, but + // still allow the original surface to be freed and remove the wrapper from + // the table when it is no longer needed. + RefPtr<SourceSurfaceSharedDataWrapper> surface = + new SourceSurfaceSharedDataWrapper(); + surface->Init(aSurface); + + uint64_t id = wr::AsUint64(aId); + MOZ_ASSERT(!sInstance->mSurfaces.Contains(id)); + + RefPtr<wr::RenderTextureHost> texture; + if (gfx::gfxVars::UseSoftwareWebRender()) { + texture = new wr::RenderSharedSurfaceTextureHostSWGL(surface); + } else { + texture = new wr::RenderSharedSurfaceTextureHost(surface); + } + wr::RenderThread::Get()->RegisterExternalImage(id, texture.forget()); + + surface->AddConsumer(); + sInstance->mSurfaces.Put(id, std::move(surface)); +} + +/* static */ +void SharedSurfacesParent::DestroyProcess(base::ProcessId aPid) { + StaticMutexAutoLock lock(sMutex); + if (!sInstance) { + return; + } + + // Note that the destruction of a parent may not be cheap if it still has a + // lot of surfaces still bound that require unmapping. + for (auto i = sInstance->mSurfaces.Iter(); !i.Done(); i.Next()) { + SourceSurfaceSharedDataWrapper* surface = i.Data(); + if (surface->GetCreatorPid() == aPid && surface->HasCreatorRef() && + surface->RemoveConsumer(/* aForCreator */ true)) { + wr::RenderThread::Get()->UnregisterExternalImage(i.Key()); + i.Remove(); + } + } +} + +/* static */ +void SharedSurfacesParent::Add(const wr::ExternalImageId& aId, + const SurfaceDescriptorShared& aDesc, + base::ProcessId aPid) { + MOZ_ASSERT(CompositorThreadHolder::IsInCompositorThread()); + MOZ_ASSERT(aPid != base::GetCurrentProcId()); + StaticMutexAutoLock lock(sMutex); + if (!sInstance) { + gfxCriticalNote << "SSP:Add " << wr::AsUint64(aId) << " shtd"; + return; + } + + // Note that the surface wrapper maps in the given handle as read only. + RefPtr<SourceSurfaceSharedDataWrapper> surface = + new SourceSurfaceSharedDataWrapper(); + if (NS_WARN_IF(!surface->Init(aDesc.size(), aDesc.stride(), aDesc.format(), + aDesc.handle(), aPid))) { + gfxCriticalNote << "SSP:Add " << wr::AsUint64(aId) << " init"; + return; + } + + uint64_t id = wr::AsUint64(aId); + MOZ_ASSERT(!sInstance->mSurfaces.Contains(id)); + + RefPtr<wr::RenderTextureHost> texture; + if (gfx::gfxVars::UseSoftwareWebRender()) { + texture = new wr::RenderSharedSurfaceTextureHostSWGL(surface); + } else { + texture = new wr::RenderSharedSurfaceTextureHost(surface); + } + wr::RenderThread::Get()->RegisterExternalImage(id, texture.forget()); + + surface->AddConsumer(); + sInstance->mSurfaces.Put(id, std::move(surface)); +} + +/* static */ +void SharedSurfacesParent::Remove(const wr::ExternalImageId& aId) { + DebugOnly<bool> rv = Release(aId, /* aForCreator */ true); + MOZ_ASSERT(rv); +} + +/* static */ +void SharedSurfacesParent::AccumulateMemoryReport( + base::ProcessId aPid, SharedSurfacesMemoryReport& aReport) { + StaticMutexAutoLock lock(sMutex); + if (!sInstance) { + return; + } + + for (auto i = sInstance->mSurfaces.ConstIter(); !i.Done(); i.Next()) { + SourceSurfaceSharedDataWrapper* surface = i.Data(); + if (surface->GetCreatorPid() == aPid) { + aReport.mSurfaces.insert(std::make_pair( + i.Key(), SharedSurfacesMemoryReport::SurfaceEntry{ + aPid, surface->GetSize(), surface->Stride(), + surface->GetConsumers(), surface->HasCreatorRef()})); + } + } +} + +/* static */ +bool SharedSurfacesParent::AccumulateMemoryReport( + SharedSurfacesMemoryReport& aReport) { + if (XRE_IsParentProcess()) { + GPUProcessManager* gpm = GPUProcessManager::Get(); + if (!gpm || gpm->GPUProcessPid() != -1) { + return false; + } + } else if (!XRE_IsGPUProcess()) { + return false; + } + + StaticMutexAutoLock lock(sMutex); + if (!sInstance) { + return true; + } + + for (auto i = sInstance->mSurfaces.ConstIter(); !i.Done(); i.Next()) { + SourceSurfaceSharedDataWrapper* surface = i.Data(); + aReport.mSurfaces.insert(std::make_pair( + i.Key(), + SharedSurfacesMemoryReport::SurfaceEntry{ + surface->GetCreatorPid(), surface->GetSize(), surface->Stride(), + surface->GetConsumers(), surface->HasCreatorRef()})); + } + + return true; +} + +} // namespace layers +} // namespace mozilla |