Mitsuru Kariya <[email protected]> writes:
> The TCP_USER_TIMEOUT socket option (Linux-only) specifies the maximum
> amount of time in milliseconds that transmitted data may remain
> unacknowledged, or buffered data may remain untransmitted, before TCP
> forcibly closes the connection and returns ETIMEDOUT to the
> application. Without it, a peer that becomes silently unreachable can
> take well over an hour to be noticed (the kernel default), which is
> particularly painful for use cases like QEMU live migration.
>
> Expose user-timeout as a new optional InetSocketAddress member, gated
> on a new HAVE_TCP_USER_TIMEOUT meson check, and apply it in
> inet_set_sockopts() alongside the existing keep-alive options. Plumb
> the same option through the legacy QemuOpts-based inet_parse() path so
> that URI-style users (-incoming tcp:, migrate tcp:, HMP
> nbd_server_start, ...) can set it too.
>
> Update the documentation for -netdev stream (addr.user-timeout=N) and
> -incoming tcp: (user-timeout=N, flat key as for the other inet_opts
> entries), and extend test_inet_parse_all_options_good accordingly.
>
> Signed-off-by: Mitsuru Kariya <[email protected]>
> ---
> meson.build | 2 ++
> qapi/sockets.json | 7 +++++++
> qemu-options.hx | 13 +++++++++----
> tests/unit/test-util-sockets.c | 13 +++++++++++++
> util/qemu-sockets.c | 24 ++++++++++++++++++++++++
> 5 files changed, 55 insertions(+), 4 deletions(-)
>
> diff --git a/meson.build b/meson.build
> index eb07491819..c5afe8754d 100644
> --- a/meson.build
> +++ b/meson.build
> @@ -2726,6 +2726,8 @@ config_host_data.set('HAVE_TCP_KEEPINTVL',
> #endif
> int main(void) { return 0; }''',
> name: 'Win32 TCP_KEEPINTVL'))
> +config_host_data.set('HAVE_TCP_USER_TIMEOUT',
> + cc.has_header_symbol('netinet/tcp.h',
> 'TCP_USER_TIMEOUT'))
>
> # has_member
> config_host_data.set('HAVE_SIGEV_NOTIFY_THREAD_ID',
> diff --git a/qapi/sockets.json b/qapi/sockets.json
> index 473be2ac58..543689a2c7 100644
> --- a/qapi/sockets.json
> +++ b/qapi/sockets.json
> @@ -79,6 +79,12 @@
> # defined (this includes Linux, Windows, macOS, FreeBSD, but not
> # OpenBSD). When set to 0, system setting is used. (Since 10.1)
> #
> +# @user-timeout: time in milliseconds that transmitted data may remain
> +# unacknowledged before the connection is closed. Only supported
> +# for TCP sockets on systems where TCP_USER_TIMEOUT socket option
> +# is defined (Linux only). When set to 0, system setting is used.
> +# (Since 11.1)
> +#
Actually, it's "transmitted data may remain unacknowledged, or buffered
data may remain untransmitted". Moreover, the interaction with
@keep-alive is not mentioned. We want to keep the description
reasonably short, though. Have it point to tcp(7) for details?
Perhaps something like
# @user-timeout: time in milliseconds that transmitted data may remain
# unacknowledged, or buffered data may remain untransmitted before
# the connection is closed. See tcp(7) for details. Only
# supported when the host provides socket option TCP_USER_TIMEOUT
# (Linux). 0 means system default. (default 0) (since 11.1)
What do you think?
> # @mptcp: enable multi-path TCP. (Since 6.1)
> #
> # Since: 1.3
> @@ -94,6 +100,7 @@
> '*keep-alive-count': { 'type': 'uint32', 'if': 'HAVE_TCP_KEEPCNT' },
> '*keep-alive-idle': { 'type': 'uint32', 'if': 'HAVE_TCP_KEEPIDLE' },
> '*keep-alive-interval': { 'type': 'uint32', 'if': 'HAVE_TCP_KEEPINTVL' },
> + '*user-timeout': { 'type': 'uint32', 'if': 'HAVE_TCP_USER_TIMEOUT' },
> '*mptcp': { 'type': 'bool', 'if': 'HAVE_IPPROTO_MPTCP' } } }
>
> ##
> diff --git a/qemu-options.hx b/qemu-options.hx
> index d2b816e16f..63587cccd3 100644
> --- a/qemu-options.hx
> +++ b/qemu-options.hx
> @@ -3034,7 +3034,7 @@ DEF("netdev", HAS_ARG, QEMU_OPTION_netdev,
> "-netdev socket,id=str[,fd=h][,udp=host:port][,localaddr=host:port]\n"
> " configure a network backend to connect to another
> network\n"
> " using an UDP tunnel\n"
> - "-netdev
> stream,id=str[,server=on|off],addr.type=inet,addr.host=host,addr.port=port[,addr.to=maxport][,addr.numeric=on|off][,addr.keep-alive=on|off][,addr.keep-alive-count=count][,addr.keep-alive-idle=idle][,addr.keep-alive-interval=interval][,addr.mptcp=on|off][,addr.ipv4=on|off][,addr.ipv6=on|off][,reconnect-ms=milliseconds]\n"
> + "-netdev
> stream,id=str[,server=on|off],addr.type=inet,addr.host=host,addr.port=port[,addr.to=maxport][,addr.numeric=on|off][,addr.keep-alive=on|off][,addr.keep-alive-count=count][,addr.keep-alive-idle=idle][,addr.keep-alive-interval=interval][,addr.user-timeout=timeout][,addr.mptcp=on|off][,addr.ipv4=on|off][,addr.ipv6=on|off][,reconnect-ms=milliseconds]\n"
> "-netdev
> stream,id=str[,server=on|off],addr.type=unix,addr.path=path[,addr.abstract=on|off][,addr.tight=on|off][,reconnect-ms=milliseconds]\n"
> "-netdev
> stream,id=str[,server=on|off],addr.type=fd,addr.str=file-descriptor[,reconnect-ms=milliseconds]\n"
> " configure a network backend to connect to another
> network\n"
> @@ -3640,7 +3640,7 @@ SRST
> -device e1000,netdev=n1,mac=52:54:00:12:34:56 \\
> -netdev
> socket,id=n1,mcast=239.192.168.1:1102,localaddr=1.2.3.4
>
> -``-netdev
> stream,id=str[,server=on|off],addr.type=inet,addr.host=host,addr.port=port[,addr.to=maxport][,addr.numeric=on|off][,addr.keep-alive=on|off][,addr.keep-alive-count=count][,addr.keep-alive-idle=idle][,addr.keep-alive-interval=interval][,addr.mptcp=on|off][,addr.ipv4=on|off][,addr.ipv6=on|off][,reconnect-ms=milliseconds]``
> +``-netdev
> stream,id=str[,server=on|off],addr.type=inet,addr.host=host,addr.port=port[,addr.to=maxport][,addr.numeric=on|off][,addr.keep-alive=on|off][,addr.keep-alive-count=count][,addr.keep-alive-idle=idle][,addr.keep-alive-interval=interval][,addr.user-timeout=timeout][,addr.mptcp=on|off][,addr.ipv4=on|off][,addr.ipv6=on|off][,reconnect-ms=milliseconds]``
> Configure a network backend to connect to another QEMU virtual machine
> or a proxy using a TCP/IP socket.
>
> ``server=on|off``
> @@ -3670,6 +3670,11 @@ SRST
> time in seconds between individual keep-alive packets.
> Set to 0 to use the system default. (default: 0)
>
> + ``addr.user-timeout=timeout``
> + time in milliseconds that transmitted data may remain unacknowledged
> + before the connection is forcibly closed.
> + Set to 0 to use the system default. (default: 0)
> +
If you improve the description in net.json, update this one to match it.
> ``addr.mptcp=on|off``
> enable multipath TCP
>
> @@ -5365,7 +5370,7 @@ SRST
> ERST
>
> DEF("incoming", HAS_ARG, QEMU_OPTION_incoming, \
> - "-incoming
> tcp:[host]:port[,to=maxport][,numeric=on|off][,keep-alive=on|off][,keep-alive-count=count][,keep-alive-idle=idle][,keep-alive-interval=interval][,mptcp=on|off][,ipv4=on|off][,ipv6=on|off]\n"
> \
> + "-incoming
> tcp:[host]:port[,to=maxport][,numeric=on|off][,keep-alive=on|off][,keep-alive-count=count][,keep-alive-idle=idle][,keep-alive-interval=interval][,user-timeout=timeout][,mptcp=on|off][,ipv4=on|off][,ipv6=on|off]\n"
> \
> "-incoming rdma:host:port[,ipv4=on|off][,ipv6=on|off]\n" \
> "-incoming unix:socketpath\n" \
> " prepare for incoming migration, listen on\n" \
> @@ -5387,7 +5392,7 @@ migration channel types. The channel type is specified
> in <channel>,
> or is 'main' for all other forms of -incoming. If multiple -incoming
> options are specified for a channel type, the last one takes precedence.
>
> -``-incoming
> tcp:[host]:port[,to=maxport][,numeric=on|off][,keep-alive=on|off][,keep-alive-count=count][,keep-alive-idle=idle][,keep-alive-interval=interval][,mptcp=on|off][,ipv4=on|off][,ipv6=on|off]``
> +``-incoming
> tcp:[host]:port[,to=maxport][,numeric=on|off][,keep-alive=on|off][,keep-alive-count=count][,keep-alive-idle=idle][,keep-alive-interval=interval][,user-timeout=timeout][,mptcp=on|off][,ipv4=on|off][,ipv6=on|off]``
> \
> ``-incoming rdma:host:port[,ipv4=on|off][,ipv6=on|off]``
> Prepare for incoming migration, listen on a given tcp port.
[...]
