On Thu, 14 Aug 2025 at 16:39, John Groves <John@xxxxxxxxxx> wrote: > Having a generic approach rather than a '-o' option would be fine with me. > Also happy to entertain other ideas... We could just allow arbitrary options to be set by the server. It might break cases where the server just passes unknown options down into the kernel, which currently are rejected. I don't think this is common practice, but still it sounds a bit risky. Alternatively allow INIT_REPLY to set up misc options, which can only be done explicitly, so no risk there. Thanks, Miklos
