> I don't think the upstream fix for this is actually complete
this seems to only be an issue in bionic and just requires upstream commit 0dba8a8beb5e2a6f16042421455004c2e7587ed4 to simply ignore the failure.
I'm still concerned about upstream's casual use of stdin/stdout fd in place of sockets, as MuxWrapper() passes the rfile/wfile up to SocketWrapper without changing SocketWrapper's handling of the 'sockets' at all. I suppose that's an issue for upstream to work out, however.
> I don't think the upstream fix for this is actually complete
this seems to only be an issue in bionic and just requires upstream commit 0dba8a8beb5e2a6 f16042421455004 c2e7587ed4 to simply ignore the failure.
I'm still concerned about upstream's casual use of stdin/stdout fd in place of sockets, as MuxWrapper() passes the rfile/wfile up to SocketWrapper without changing SocketWrapper's handling of the 'sockets' at all. I suppose that's an issue for upstream to work out, however.