qemu-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [PATCH 07/15] python/aqmp: add send_fd_scm


From: John Snow
Subject: Re: [PATCH 07/15] python/aqmp: add send_fd_scm
Date: Fri, 17 Sep 2021 14:05:54 -0400



On Fri, Sep 17, 2021 at 9:34 AM Hanna Reitz <hreitz@redhat.com> wrote:
On 17.09.21 07:40, John Snow wrote:
> The single space is indeed required to successfully transmit the file
> descriptor to QEMU.

Yeah, socket_scm_helper.c said “Send a blank to notify qemu”.

> Signed-off-by: John Snow <jsnow@redhat.com>
> ---
>   python/qemu/aqmp/qmp_client.py | 17 +++++++++++++++++
>   1 file changed, 17 insertions(+)
>
> diff --git a/python/qemu/aqmp/qmp_client.py b/python/qemu/aqmp/qmp_client.py
> index d2ad7459f9..58f85990bc 100644
> --- a/python/qemu/aqmp/qmp_client.py
> +++ b/python/qemu/aqmp/qmp_client.py
> @@ -9,6 +9,8 @@
>   
>   import asyncio
>   import logging
> +import socket
> +import struct
>   from typing import (
>       Dict,
>       List,
> @@ -624,3 +626,18 @@ async def execute(self, cmd: str,
>           """
>           msg = self.make_execute_msg(cmd, arguments, oob=oob)
>           return await self.execute_msg(msg)
> +
> +    @upper_half
> +    @require(Runstate.RUNNING)
> +    def send_fd_scm(self, fd: int) -> None:
> +        """
> +        Send a file descriptor to the remote via SCM_RIGHTS.
> +        """
> +        assert self._writer is not None
> +        sock = self._writer.transport.get_extra_info('socket')
> +
> +        # Python 3.9+ adds socket.send_fds(...)
> +        sock.sendmsg(
> +            [b' '],
> +            [(socket.SOL_SOCKET, socket.SCM_RIGHTS, struct.pack('@i', fd))]
> +        )

AFAIU the socket can be either TCP or a UNIX socket
(AsyncProtocol._do_accept’s docstring sounds this way), so should we
check that this is a UNIX socket?  (Or is it fine to just run into the
error that I suspect we would get with a TCP socket?)

Hanna


Uhh, hm. I was going to say "Yeah, just let it fail!" but ... upon going to document what error to expect in this case, I am realizing it fails silently. So, uh, that's not ideal.

I'll fix this to make it bark.

--js


reply via email to

[Prev in Thread] Current Thread [Next in Thread]