I am building a zero-copy rendering pipeline on RK3568 (Mali-G52, Qt 5.x with EGL backend).

The idea is:

• Import a dmabuf into an EGLImageKHR
• Bind it to a GL texture
• Render Qt UI (QOpenGLFramebufferObject) directly into this texture

The problem:

• To create an EGLImageKHR, I need a valid EGLDisplay.
• Qt’s QOpenGLContext does not expose the underlying EGLDisplay.
• In Qt5, QOpenGLContext::nativeHandle() always returns QVariant(null).
• Even if I try setNativeHandle(), it only echoes back what I set — Qt still creates its own internal EGL context.

Minimal test code:

QOpenGLContext* ctx = new QOpenGLContext;
ctx->create();

qDebug() << ctx->nativeHandle(); // always null

QVariant nativeHandle = QVariant::fromValue<void*>(eglGetCurrentContext());
ctx->setNativeHandle(nativeHandle);

qDebug() << ctx->nativeHandle(); // echoes back, but not really used

This results in two isolated contexts:

• Qt’s internal context (used by QPainter / QFBO rendering)
• My native EGLContext (used to create EGLImageKHR from dmabuf)

Since these are not shared, Qt cannot render directly into the dmabuf-backed texture. That breaks the zero-copy requirement, because the only fallback is copying with glBlitFramebuffer or PBOs.

My question:
Is there any supported way in Qt5 to:

1. Make QOpenGLContext share resources with an existing native EGLContext, or
2. Retrieve the actual EGLDisplay / EGLContext from Qt so that I can create the EGLImageKHR inside the same context?

Goal: Render Qt FBO content directly into dmabuf (via EGLImage), without extra copies.