I've submitted the patches corresponding to comment 8 to the upstream. https://lore.kernel.org/netdev/20210714170022.8162-1-tiwai@suse.de/ The main issue wrt napi_disable() is not fixed yet, though.