Comment 21 for bug 1853266

Revision history for this message
In , ofourdan (ofourdan-redhat-bugs) wrote :

So I think the key to this problem is in comment 14 (thanks Andreas!)

 (WW) glamor: Expect reduced performance.
 (WW) glamor: Failed to allocate 1922x1175 FBO due to GL_OUT_OF_MEMORY.
 (EE) glamor0: GL error: GL_OUT_OF_MEMORY in glTexSubImage

because:

 - All these crashes occur in glamor_get_pixmap_texture() which is a pretty simple function.
 - Only case where this function can segfault is if `pixmap_priv->fbo` in NULL
 - "Failed to allocate ...x... FBO due to GL_OUT_OF_MEMORY" comes from `_glamor_create_tex()` and is called by `glamor_create_fbo()`

So I reckon this issue is caused by glamor failing to allocate a pixmap FBO because GL_OUT_OF_MEMORY and later trying to access it, and crash.

*But* AFAICS, `glamor_get_pixmap_texture()` is used by Xwayland *only* with EGLStream in `xwl_glamor_eglstream_post_damage()`

Are you all using NVidia closed-source driver?