ui: create the GL window surface + first make_current on the main thread - #97
Merged
Conversation
issue #94: on some AMD machines the CLI died during startup with a null read inside atio6axx.dll (the AMD OpenGL ICD) on the main thread, inside a window procedure. The ICD subclasses the GL window when SetPixelFormat is called; its wndproc hook then runs on the window-owning (main) thread. iris was doing SetPixelFormat + the first wglMakeCurrent lazily on the REX3 refresh thread, in GlRenderer::ensure_init - so a startup resize could reach the freshly-installed subclass before wglMakeCurrent had populated the driver's per-HWND state. Ui::new() now builds the Surface and binds the context once (make_current -> make_not_current) on the main thread, before Ui::run starts pumping messages, and hands both the NotCurrentContext and the Surface (initial_surface) to GlRenderer. ensure_init() consumes the surface on the true first frame and only make_current()s it on the refresh thread - the "move a context between threads" handoff. Rendering stays entirely on REX3; only the one-time bind moved. After a stop()/start() cycle (jitcheck checkpoint restore, reset, snapshot load) initial_surface is None and ensure_init makes a fresh surface on the refresh thread as before - safe by then, the pixel format is set and the driver's window state exists. mac/Linux unaffected (macOS window_handle() is already captured on the main thread; Linux/GLX benefits - the drawable calls now happen on the window-creating thread, matching the proprietary-NVIDIA note on not_current_context). Verified on Windows: boots clean, and `reset` (exercises the fresh-surface path) does not crash. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
issue #94: on some AMD machines the CLI died during startup with a null read inside atio6axx.dll (the AMD OpenGL ICD) on the main thread, inside a window procedure. The ICD subclasses the GL window when SetPixelFormat is called; its wndproc hook then runs on the window-owning (main) thread. iris was doing SetPixelFormat + the first wglMakeCurrent lazily on the REX3 refresh thread, in GlRenderer::ensure_init - so a startup resize could reach the freshly-installed subclass before wglMakeCurrent had populated the driver's per-HWND state.
Ui::new() now builds the Surface and binds the context once (make_current -> make_not_current) on the main thread, before Ui::run starts pumping messages, and hands both the NotCurrentContext and the Surface (initial_surface) to GlRenderer. ensure_init() consumes the surface on the true first frame and only make_current()s it on the refresh thread - the "move a context between threads" handoff. Rendering stays entirely on REX3; only the one-time bind moved.
After a stop()/start() cycle (jitcheck checkpoint restore, reset, snapshot load) initial_surface is None and ensure_init makes a fresh surface on the refresh thread as before - safe by then, the pixel format is set and the driver's window state exists.
mac/Linux unaffected (macOS window_handle() is already captured on the main thread; Linux/GLX benefits - the drawable calls now happen on the window-creating thread, matching the proprietary-NVIDIA note on not_current_context). Verified on Windows: boots clean, and
reset(exercises the fresh-surface path) does not crash.