[Fix, Feat, Test] (MG_Util, MG_Backend, MG_Impl, MG_State): correct the log severity ordering and unwind the diagnostics it silenced

This commit is contained in:
Swung0x48
2026-08-13 02:02:49 -04:00
parent 373aa44dd7
commit dd2a62228f
55 changed files with 940 additions and 590 deletions
@@ -594,9 +594,10 @@ namespace MobileGL::MG_Util::BackendLoader {
#endif // !_WIN32
if (!eglLib) {
// MGLOG_F, not MGLOG_E: at the INFO log level every shipping and CI build
// uses, MGLOG_E is compiled out (Log.h orders DEBUG < WARN < ERROR < INFO),
// so this diagnosis was invisible in precisely the builds that needed it.
// MGLOG_F, not MGLOG_E: with no EGL there is no rendering at all, so this is a
// bring-up abort rather than a recoverable error. It was forced to F while the
// Log.h ordering compiled MGLOG_E out of every shipping and CI build; F is still
// the right level on its own merits, so it stays.
MGLOG_F("Failed to open EGL library: none of libEGL.so.1 / libEGL.so could be "
"dlopened; every EGL entry point will be null");
return;
@@ -986,6 +987,12 @@ namespace MobileGL::MG_Util::BackendLoader {
caps.SupportsBaseInstance ? "yes" : "no");
MGLOG_I(" clip distances (EXT_clip_cull_distance): %s", caps.SupportsClipDistance ? "yes" : "no");
// LOAD-BEARING STRING, not just a banner. android-plugin/trace-replay-ci.sh's
// is_angle_surface_lost() greps mobilegl.log for exactly "OpenGL ES capabilities:" to
// decide whether MobileGL got far enough to have a working context: if the probe ran,
// a later surface loss is a real defect rather than an emulator fault worth retrying.
// Demoting this line, renaming it, or moving it before the context is usable silently
// inverts that retry logic. It is init-phase, so MGLOG_I is correct and it stays.
MGLOG_I("OpenGL ES capabilities:");
glesFuncs.glGetIntegerv(GL_UNIFORM_BUFFER_OFFSET_ALIGNMENT, &caps.UniformBufferOffsetAlignment);
MGLOG_I(" GL_UNIFORM_BUFFER_OFFSET_ALIGNMENT: %d", caps.UniformBufferOffsetAlignment);