There are several code paths in net_init_socket() depending on how the
socket is created: file descriptor passing, UDP multicast, TCP, or UDP.
Some of these support both listen and connect.
Not all code paths set the socket to non-blocking. This patch addresses
the file descriptor passing and UDP cases which were missing
socket_set_nonblock(fd) calls.
I considered moving socket_set_nonblock(fd) to a central location but it
turns out the code paths are different enough to require non-blocking at
different places.
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Luiz Capitulino <lcapitulino@redhat.com>
The fcntl(fd, F_SETFL, O_NONBLOCK) flag is not specific to sockets.
Rename to qemu_set_nonblock() just like qemu_set_cloexec().
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Luiz Capitulino <lcapitulino@redhat.com>
Socket buffer sizes were hard-coded to 4K for VDE and socket netdevs. Bump this
up to 68K (ala tap netdev) to handle maximum GSO packet size (64k) plus plenty
of room for the ethernet and virtio_net headers.
Originally, ran into this limitation when using -netdev UDP sockets to connect
VM-to-VM, where VM interface is configure with MTU=9000. (Using virtio_net
NIC model). Test is simple: ping -M do -s 8500 <target>. This test will
attempt to ping with unfragmented packet of given size. Without patch, size
is limited to < 4K (minus protocol hdrs). With patch, ping test works with pkt
size up to 9000 (again, minus protocol hdrs).
v2: per Stefan, increase buf size to (4096+65536) as done in tap and apply
to vde and socket netdevs.
v1: increase buf size to 12K just for -netdev UDP sockets
Signed-off-by: Scott Feldman <sfeldma@cumulusnetworks.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reduce -netdev socket latency by disabling the Nagle algorithm on
SOCK_STREAM sockets in net/socket.c. Since we are tunelling Ethernet
over TCP we shouldn't artificially delay outgoing packets, let the guest
decide packet scheduling.
I already get sub-millisecond -netdev socket ping times on localhost, so
there was no measurable difference in my testing. This won't hurt
though and may improve remote socket performance.
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Benoit Canet <benoit@irqsave.net>
Reviewed-by: Daniel P. Berrange <berrange@redhat.com>
Instead of adding missing type casts which are needed by MinGW for the
4th argument, the patch uses qemu_setsockopt which was invented for this
purpose.
Signed-off-by: Stefan Weil <sw@weilnetz.de>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Move public headers to include/net, and leave private headers in net/.
Put the virtio headers in include/net/tap.h, removing the multiple copies
that existed. Leave include/net/tap.h as the interface for NICs, and
net/tap_int.h as the interface for OS-specific parts of the tap backend.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Touching char/char.h basically causes the whole of QEMU to
be rebuilt. Avoid this, it is usually unnecessary.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Fix the problem that can not delete the udp socket.
It's caused by passing "udp" model to net_socket_udp_init,
but we do not have "udp" model in our model list.
Pass the right model "socket" to init function.
https://bugs.launchpad.net/qemu/+bug/1073585?comments=all
Signed-off-by: Lei Li <lilei@linux.vnet.ibm.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
This patch doesn't seem much useful alone, I must admit. However,
it makes sense as part of the upcoming directory reorganization,
where I want to have include/net/tap.h as the net<->hw interface
for tap. Then having both net/tap.h and include/net/tap.h does
not work. "Fixed" by moving all the init functions to a single
header file net/clients.h.
The patch also adopts a uniform style for including net/*.h files
from net/*.c, without the net/ path.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@gmail.com>
Commit 213fd5087e removed a type cast
which is needed for MinGW:
net/socket.c:136: warning:
pointer targets in passing argument 2 of ‘sendto’ differ in signedness
/usr/lib/gcc/amd64-mingw32msvc/4.4.4/../../../../amd64-mingw32msvc/include/winsock2.h:1313: note:
expected ‘const char *’ but argument is of type ‘const uint8_t *’
Add a 'qemu_sendto' macro which provides that type cast where needed
and use the new macro instead of 'sendto'.
Signed-off-by: Stefan Weil <sw@weilnetz.de>
Signed-off-by: Stefan Hajnoczi <stefanha@gmail.com>
Replace spinning send_all() with a proper non-blocking send. When the
socket write buffer limit is reached, we should stop trying to send and
wait for the socket to become writable again.
Non-blocking TCP sockets can return in two different ways when the write
buffer limit is reached:
1. ret = -1 and errno = EAGAIN/EWOULDBLOCK. No data has been written.
2. ret < total_size. Short write, only part of the message was
transmitted.
Handle both cases and keep track of how many bytes have been written in
s->send_index. (This includes the 'length' header before the actual
payload buffer.)
Signed-off-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Implement asynchronous send for UDP (or other SOCK_DGRAM) sockets. If
send fails with EAGAIN we wait for the socket to become writable again.
Signed-off-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
The net/socket.c net client is not truly asynchronous. This patch
borrows the qemu_set_fd_handler2() code from net/tap.c as the basis for
proper asynchronous send/receive.
Only read packets from the socket when the peer is able to receive.
This avoids needless queuing.
Later patches implement asynchronous send.
Signed-off-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
This patch renames+moves the net_handle_fd_param() caller used to
obtain a file descriptor from either qemu_parse_fd() (the normal case)
or from monitor_get_fd() (migration case) into a generically prefixed
monitor_handle_fd_param() to be used by vhost-scsi code.
Also update net/[socket,tap].c consumers to use the new prefix.
Reported-by: Michael S. Tsirkin <mst@redhat.com>
Cc: Michael S. Tsirkin <mst@redhat.com>
Cc: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Anthony Liguori <aliguori@us.ibm.com>
Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
The -net socket,listen option does not work with the newer -netdev
syntax:
http://lists.gnu.org/archive/html/qemu-devel/2011-11/msg01508.html
This patch makes it work now.
For the case where one vlan has multiple listenning sockets,
the patch will also provide the support.
Supported syntax:
1.) -net socket,listen=127.0.0.1:1234,vlan=0
2.) -net socket,listen=127.0.0.1:1234,vlan=0 -net socket,listen=127.0.0.1:1235,vlan=0
3.) -netdev socket,listen=127.0.0.1:1234,id=socket0
Drop the NetSocketListenState struct and add a listen_fd field
to NetSocketState. When a -netdev socket,listen= instance is created
there will be a NetSocketState with fd=-1 and a valid listen_fd. The
net_socket_accept() handler waits for listen_fd to become readable and
then accepts the connection. When this state transition happens, we no
longer monitor listen_fd for incoming connections...until the client
disconnects again.
Signed-off-by: Zhi Yong Wu <wuzhy@linux.vnet.ibm.com>
Signed-off-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Reviewed-by: Laszlo Ersek <lersek@redhat.com>
The vlan feature is no longer part of net core. Rename VLANClientState
to NetClientState because net clients are not explicitly associated with
a vlan at all, instead they have a peer net client to which they are
connected.
This patch is a mechanical search-and-replace except for a few
whitespace fixups where changing VLANClientState to NetClientState
misaligned whitespace.
Signed-off-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Signed-off-by: Zhi Yong Wu <wuzhy@linux.vnet.ibm.com>
Reviewed-by: Laszlo Ersek <lersek@redhat.com>
Since hubs are now used to implement the 'vlan' feature and the vlan
argument is always NULL, remove the argument entirely and update all net
clients that use qemu_new_net_client().
Signed-off-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Signed-off-by: Zhi Yong Wu <wuzhy@linux.vnet.ibm.com>
Reviewed-by: Laszlo Ersek <lersek@redhat.com>
Stop using the special-case vlan code in net.c. Instead use the hub net
client to implement the vlan feature. The next patch will remove vlan
code from net.c completely.
Signed-off-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Signed-off-by: Zhi Yong Wu <wuzhy@linux.vnet.ibm.com>
Reviewed-by: Laszlo Ersek <lersek@redhat.com>
I "reverse engineered" the following permissions between the -socket
sub-options:
fd listen connect mcast udp | localaddr
fd x . . . . | .
listen . x . . . | .
connect . . x . . | .
mcast . . . x . | x
udp . . . . x | x
-------------------------------------------+
localaddr . . . x x x
I transformed the code accordingly. The real fix would be to embed "fd",
"listen", "connect", "mcast" and "udp" in a separate union. However
OptsVisitor's enum parser only supports the type=XXX QemuOpt instance as
union discriminator.
Signed-off-by: Laszlo Ersek <lersek@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
The net_client_init() prototype is kept intact.
Based on "is_netdev", the QemuOpts-rooted QemuOpt-list is parsed as a
Netdev or a NetLegacy. The original meat of net_client_init() is moved to
and simplified in net_client_init1():
Fields not common between -net and -netdev are clearly separated. Getting
the name for the init functions is cleaner: Netdev::id is mandatory, and
all init functions handle a NULL NetLegacy::name. NetLegacy::vlan
explicitly depends on -net (see below).
Verifying the "type=" option for -netdev can be turned into a switch.
Format validation with qemu_opts_validate() can be removed because the
visitor covers it. Relatedly, the "net_client_types" array is reduced to
an array of init functions that can be directly indexed by opts->kind.
(Help text is available in the schema JSON.)
The outermost negation in the condition around qemu_find_vlan() was
flattened, because it expresses the dependent code's requirements more
clearly.
VLAN lookup is avoided if there's no init function to pass the VLAN to.
Whenever the value of type=... is needed, we substitute
NetClientOptionsKind_lookup[kind].
The individual init functions are not converted yet, thus the original
QemuOpts instance is passed transparently.
v1->v2:
- NetLegacy::name is optional. Tracked it through all init functions: they
all handle a NULL name. Updated commit message accordingly.
v2->v3:
- NetLegacy::id is allowed and takes precedence over NetLegacy::name.
Signed-off-by: Laszlo Ersek <lersek@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
The only backend that really uses it is the socket one, which calls
monitor_get_fd(). But it can use 'cur_mon' instead.
Signed-off-by: Luiz Capitulino <lcapitulino@redhat.com>
Reviewed-By: Laszlo Ersek <lersek@redhat.com>
Fix a leak of a file descriptor due to missing closesocket() calls
in error paths in net_socket_listen_init().
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Today net/socket.c has no consistent policy for closing the socket file
descriptor when initialization fails. This means we leak the file
descriptor in some cases or we could also try to close it twice.
Make error paths consistent by taking ownership of the file descriptor
and closing it on error.
Signed-off-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
In order to make later patches sane, expand the tab characters and
conform to QEMU coding style now.
Signed-off-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
Avoid warnings like these by wrapping recv():
CC slirp/ip_icmp.o
/src/qemu/slirp/ip_icmp.c: In function 'icmp_receive':
/src/qemu/slirp/ip_icmp.c:418:5: error: passing argument 2 of 'recv' from incompatible pointer type [-Werror]
/usr/local/lib/gcc/i686-mingw32msvc/4.6.0/../../../../i686-mingw32msvc/include/winsock2.h:547:32: note: expected 'char *' but argument is of type 'struct icmp *'
Remove also casts used to avoid warnings.
Reviewed-by: Anthony Liguori <aliguori@us.ibm.com>
Signed-off-by: Blue Swirl <blauwirbel@gmail.com>
error_report() prepends location, and appends a newline. The message
constructed from the arguments should not contain a newline. Fix the
obvious offenders.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
MSDN includes the following in WSAEALREADY error description for connect()
function: "To preserve backward compatibility, this error is reported as
WSAEINVAL to Winsock applications that link to either Winsock.dll or
Wsock32.dll". So check of this error code was added to allow network
connections through the sockets in Windows.
Signed-off-by: Pavel Dovgalyuk <pavel.dovgaluk@gmail.com>
Signed-off-by: Blue Swirl <blauwirbel@gmail.com>
Avoid this warning like other uses of setsockopt:
/src/qemu/net/socket.c: In function 'net_socket_mcast_create':
/src/qemu/net/socket.c:210: warning: passing argument 4 of 'setsockopt' from incompatible pointer type
Signed-off-by: Blue Swirl <blauwirbel@gmail.com>
Add an option to specify the host IP to send multicast packets from,
when using a multicast socket for networking. The option takes an IP
address and sets the IP_MULTICAST_IF socket option, which causes the
packets to use that IP's interface as an egress.
This is useful if the host machine has several interfaces with several
virtual networks across disparate interfaces.
Signed-off-by: Mike Ryan <mikeryan@ISI.EDU>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
error_report() terminates the message with a newline. Strip it it
from its arguments.
This fixes a few error messages lacking a newline:
net_handle_fd_param()'s "No file descriptor named %s found", and
tap_open()'s "vnet_hdr=1 requested, but no kernel support for
IFF_VNET_HDR available" (all three versions).
There's one place that passes arguments without newlines
intentionally: load_vmstate(). Fix it up.
net_check_clients() prints this when an VLAN has host devices, but no
guest devices. It uses VLANState members nb_guest_devs and
nb_host_devs to keep track of these devices. However, -device does
not update nb_guest_devs, only net_init_nic() does that, for -net nic.
Check the VLAN clients directly, and remove the counters.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
We're leaking file descriptors to child processes. Set FD_CLOEXEC on file
descriptors that don't need to be passed to children to stop this misbehaviour.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
