OpenGL client side glue to compositor header. More...
Go to the source code of this file.
Data Structures | |
| class | client_gl_swapchain |
| Wraps the real compositor swapchain providing a OpenGL based interface. More... | |
| class | client_gl_compositor |
| Wraps the real compositor providing a OpenGL based interface. More... | |
Typedefs | |
| typedef xrt_result_t(* | client_gl_context_begin_locked_func_t) (struct xrt_compositor *xc, enum client_gl_context_reason reason) |
| Fetches the OpenGL context that is current on this thread and makes the OpenGL context given in the graphics binding current instead. More... | |
| typedef void(* | client_gl_context_end_locked_func_t) (struct xrt_compositor *xc, enum client_gl_context_reason reason) |
| Makes the OpenGL context current that was current before client_gl_context_begin_locked_func_t was called. More... | |
| typedef struct xrt_swapchain *(* | client_gl_swapchain_create_func_t) (struct xrt_compositor *xc, const struct xrt_swapchain_create_info *info, struct xrt_swapchain_native *xscn, struct client_gl_swapchain **out_sc) |
| The type of a swapchain create constructor. More... | |
| typedef xrt_result_t(* | client_gl_insert_fence_func_t) (struct xrt_compositor *xc, xrt_graphics_sync_handle_t *out_handle) |
| The type of a fence insertion function. More... | |
Enumerations | |
| enum | client_gl_context_reason { CLIENT_GL_CONTEXT_REASON_SYNCHRONIZE , CLIENT_GL_CONTEXT_REASON_OTHER } |
| What's the reason to make the context current, this is needed currently for EGL where we have to create a shared context in some cases. More... | |
Detailed Description
OpenGL client side glue to compositor header.
Typedef Documentation
◆ client_gl_context_begin_locked_func_t
| typedef xrt_result_t(* client_gl_context_begin_locked_func_t) (struct xrt_compositor *xc, enum client_gl_context_reason reason) |
Fetches the OpenGL context that is current on this thread and makes the OpenGL context given in the graphics binding current instead.
Only one thread at a time can operate on the sections between client_gl_context_begin_locked_func_t and client_gl_context_end_locked_func_t, therefore client_gl_context_end_locked_func_t MUST be called to avoid blocking the next thread calling client_gl_context_begin_locked_func_t.
This function must be called with the context_mutex locked held, that is handled by the helper function client_gl_compositor_context_begin.
If the return value is not XRT_SUCCESS, client_gl_context_end_locked_func_t should not be called.
◆ client_gl_context_end_locked_func_t
| typedef void(* client_gl_context_end_locked_func_t) (struct xrt_compositor *xc, enum client_gl_context_reason reason) |
Makes the OpenGL context current that was current before client_gl_context_begin_locked_func_t was called.
This function must be called with the context_mutex locked held, successful call to client_gl_compositor_context_begin will ensure that. The lock is not released by this function, but client_gl_compositor_context_end does release it.
◆ client_gl_insert_fence_func_t
| typedef xrt_result_t(* client_gl_insert_fence_func_t) (struct xrt_compositor *xc, xrt_graphics_sync_handle_t *out_handle) |
The type of a fence insertion function.
This function is called in xrt_compositor::layer_commit.
The returned graphics sync handle is given to xrt_compositor::layer_commit.
◆ client_gl_swapchain_create_func_t
| typedef struct xrt_swapchain*(* client_gl_swapchain_create_func_t) (struct xrt_compositor *xc, const struct xrt_swapchain_create_info *info, struct xrt_swapchain_native *xscn, struct client_gl_swapchain **out_sc) |
The type of a swapchain create constructor.
Because our swapchain creation varies depending on available extensions and application choices, the swapchain constructor parameter to client_gl_compositor is parameterized.
Note that the "common" swapchain creation function does some setup before invoking this, and some cleanup after.
- Must populate
destroy - Does not need to save/restore texture binding
Enumeration Type Documentation
◆ client_gl_context_reason
What's the reason to make the context current, this is needed currently for EGL where we have to create a shared context in some cases.
But when we want to synchronize (insert a fence or call glFinish) we can not use the shared context and must use the context that the app provided on creation.
