Live migration with spice works like this today:
(1) client_migrate_info monitor cmd
(2) spice server notifies client, client connects to target host.
(3) qemu waits until spice client connect is finished.
(4) send over vmstate (i.e. main part of live migration).
(5) spice handover to target host.
(3) is implemented by making client_migrate_info a async monitor
command. This is the only async monitor command we have.
The original reason to implement this dance was that qemu did not accept
new tcp connections while the incoming migration was running, so (2) and
(4) could not be done in parallel. That issue was fixed long ago though.
Qemu version 1.3.0 (released Dec 2012) and newer happily accept tcp
connects while the incoming migration runs.
Time to drop step (3). This patch does exactly that, by making the
monitor command synchronous and removing the code needed to handle the
async monitor command in ui/spice-core.c
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
The VNC server websockets decoder will read and buffer data from
websockets clients until it sees the end of the HTTP headers,
as indicated by \r\n\r\n. In theory this allows a malicious to
trick QEMU into consuming an arbitrary amount of RAM. In practice,
because QEMU runs g_strstr_len() across the buffered header data,
it will spend increasingly long burning CPU time searching for
the substring match and less & less time reading data. So while
this does cause arbitrary memory growth, the bigger problem is
that QEMU will be burning 100% of available CPU time.
A novnc websockets client typically sends headers of around
512 bytes in length. As such it is reasonable to place a 4096
byte limit on the amount of data buffered while searching for
the end of HTTP headers.
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
The logic for decoding websocket frames wants to fully
decode the frame header and payload, before allowing the
VNC server to see any of the payload data. There is no
size limit on websocket payloads, so this allows a
malicious network client to consume 2^64 bytes in memory
in QEMU. It can trigger this denial of service before
the VNC server even performs any authentication.
The fix is to decode the header, and then incrementally
decode the payload data as it is needed. With this fix
the websocket decoder will allow at most 4k of data to
be buffered before decoding and processing payload.
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
[ kraxel: fix frequent spurious disconnects, suggested by Peter Maydell ]
@@ -361,7 +361,7 @@ int vncws_decode_frame_payload(Buffer *input,
- *payload_size = input->offset;
+ *payload_size = *payload_remain;
[ kraxel: fix 32bit build ]
@@ -306,7 +306,7 @@ struct VncState
- uint64_t ws_payload_remain;
+ size_t ws_payload_remain;
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
If the x509verify option is requested, the VNC websockets server
was failing to validate that the websockets client provided an
x509 certificate matching the ACL rules.
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
The previous change to the auth scheme handling guarantees we
can never have nested TLS sessions in the VNC websockets server.
Thus we can remove the separate gnutls_session instance.
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
When TLS is required, the primary VNC server considers it to be
mandatory. ie the server admin decides whether or not TLS is used,
and the client has to comply with this decision. The websockets
server, however, treated it as optional, allowing non-TLS clients
to connect to a server which had setup TLS. Thus enabling websockets
lowers the security of the VNC server leaving the admin no way to
enforce use of TLS.
This removes the code that allows non-TLS fallback in the websockets
server, so that if TLS is requested for VNC it is now mandatory for
both the primary VNC server and the websockets VNC server.
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
The way the websockets TLS code was integrated into the VNC server
made it essentially useless. The only time that the websockets TLS
support could be used is if the primary VNC server had its existing
TLS support disabled. ie QEMU had to be launched with:
# qemu -vnc localhost:1,websockets=5902,x509=/path/to/certs
Note the absence of the 'tls' flag. This is already a bug, because
the docs indicate that 'x509' is ignored unless 'tls' is given.
If the primary VNC server had TLS turned on via the 'tls' flag,
then this prevented the websockets TLS support from being used,
because it activates the VeNCrypt auth which would have resulted
in TLS being run over a TLS session. Of course no websockets VNC
client supported VeNCrypt so in practice, since the browser clients
cannot setup a nested TLS session over the main HTTPS connection,
so it would not even get past auth.
This patch causes us to decide our auth scheme separately for the
main VNC server vs the websockets VNC server. We take account of
the fact that if TLS is enabled, then the websockets client will
use https, so setting up VeNCrypt is thus redundant as it would
lead to nested TLS sessions.
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
The vnc_display_open method is quite long and complex, so
move the VNC auth scheme decision logic into a separate
method for clarity.
Also update the comment to better describe what we are
trying to achieve.
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
If the VNC server is built without tls, sasl or websocket support
and the user requests one of these features, they are just silently
ignored. This is bad because it means the VNC server ends up running
in a configuration that is less secure than the user asked for.
It also leads to an tangled mass of preprocessor conditionals when
configuring the VNC server.
This ensures that the tls, sasl & websocket options are always
processed and an error is reported back to the user if any of
them were disabled at build time.
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Handling of VNC audio messages results in printfs to the console.
This is of no use to anyone in production, so should be using the
normal VNC_DEBUG macro instead.
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Error classes are a leftover from the days of "rich" error objects.
New code should always use ERROR_CLASS_GENERIC_ERROR. Commit 1d0d59f
added a use of ERROR_CLASS_DEVICE_NOT_FOUND. Replace it.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
This happens for example when doing ctrl-alt-u and segfaults
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Put them under a #define similar to the VGA model and make them
actually compile. Add a couple too.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
vs->lsock may equal to 0, modify the check condition,
avoid possible vs->lsock leak.
Signed-off-by: Gonglei <arei.gonglei@huawei.com>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
The 'x509verify' parameter is documented as taking a path to the
x509 certificates, ie the same syntax as the 'x509' parameter.
commit 4db14629c3
Author: Gerd Hoffmann <kraxel@redhat.com>
Date: Tue Sep 16 12:33:03 2014 +0200
vnc: switch to QemuOpts, allow multiple servers
caused a regression by turning 'x509verify' into a boolean
parameter instead. This breaks setup from libvirt and is not
consistent with the docs.
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Use inet_listen_opts instead of inet_listen. Allows us to drop some
pointless indirection: Format strings just to parse them again later on.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Tested-by: Gonglei <arei.gonglei@huawei.com>
Reviewed-by: Gonglei <arei.gonglei@huawei.com>
Nobody cares about those strings, they are only used to check whenever
the vnc server / websocket support is enabled or not. Add bools for
this and drop the strings.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Gonglei <arei.gonglei@huawei.com>
Connecting to VNC through websocket crashes in vnc_flush() when trying
to acquire a mutex that hasn't been initialized (vnc_init_state(vs)
hasn't been called at this point).
Signed-off-by: Jorge Acereda Macia <jacereda@gmail.com>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
This way the generated id will be stored in -writeconfig cfg files.
Also we can make vnc_auto_assign_id() local to vnc.c.
Tested-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
This patch adds missing cyrillic character 'numerosign' to the VNC
keysym table, it's needed by Russian keyboard. And I get the keysym from
'<X11/keysymdef.h>', the current keysym table in Qemu was generated from
it.
Signed-off-by: Wang xin <wangxinxin.wang@huawei.com>
Signed-off-by: Gonglei <arei.gonglei@huawei.com>
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
This function is not used anymore, let's remove it.
Signed-off-by: Thomas Huth <thuth@linux.vnet.ibm.com>
Reviewed-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Remove qemu_console_displaystate(), qemu_remove_kbd_event_handler(),
qemu_different_endianness_pixelformat() and cpkey(), since they are
completely unused.
Signed-off-by: Thomas Huth <thuth@linux.vnet.ibm.com>
Reviewed-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Some are called do_COMMAND() (old ones, usually), some hmp_COMMAND(),
and sometimes COMMAND pointlessly differs in spelling.
Normalize to hmp_COMMAND(), where COMMAND is exactly the command name
with '-' replaced by '_'.
Exceptions:
* do_device_add() and client_migrate_info() *not* renamed to
hmp_device_add(), hmp_client_migrate_info(), because they're also
QMP handlers. They still need to be converted to QAPI.
* do_memory_dump(), do_physical_memory_dump(), do_ioport_read(),
do_ioport_write() renamed do hmp_* instead of hmp_x(), hmp_xp(),
hmp_i(), hmp_o(), because those names are too cryptic for my taste.
* do_info_help() renamed to hmp_info_help() instead of hmp_info(),
because it only covers help.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
vnc_display_local_addr will not be called with an invalid display id.
Add assert() to silence coverity warning about a null pointer dereference.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
in case we send and update for a complete scanline increment
the y offset to avoid running to find_next_bit for that lines
twice.
Signed-off-by: Peter Lieven <pl@kamp.de>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Gonglei <arei.gonglei@huawei.com>
Reviewed-by: Thomas Huth <thuth@linux.vnet.ibm.com>
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Add new query vnc qmp command, for the lack of better ideas just name it
"query-vnc-servers". Changes over query-vnc:
* It returns a list of vnc servers, so multiple vnc server instances
are covered.
* Each vnc server returns a list of server sockets. Followup patch
will use that to also report websockets. In case we add support for
multiple server sockets server sockets (to better support ipv4+ipv6
dualstack) we can add them to the list too.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Also track the number of connections in "connecting" and "shared" state
(in addition to the "exclusive" state). Apply a configurable limit to
these connections.
The logic to apply the limit to connections in "shared" state is pretty
simple: When the limit is reached no new connections are allowed.
The logic to apply the limit to connections in "connecting" state (this
is the state you are in *before* successful authentication) is
slightly different: A new connect kicks out the oldest client which is
still in "connecting" state. This avoids a easy DoS by unauthenticated
users by simply opening connections until the limit is reached.
Cc: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
This patch adds a display= parameter to the vnc options. This allows to
bind a vnc server instance to a specific display, allowing to create a
multiseat setup with a vnc server for each seat.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
This patch switches vnc over to QemuOpts, and it (more or less
as side effect) allows multiple vnc server instances.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
In case the display id is "default" (which is the one you get if you
don't explicitly assign one) we keep the old name scheme, without
display, for backward compatibility reasons.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Gonglei <arei.gonglei@huawei.com>
DisplayState isn't used anywhere, drop it. Add the vnc server ID as
parameter instead, so it is possible to specify the server instance.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Gonglei <arei.gonglei@huawei.com>
Replace with a vnc_displays list, so we can have multiple vnc server
instances. Add vnc_server_find function to lookup a display by id.
With no id supplied return the first vnc server, for backward
compatibility reasons.
It is not possible (yet) to actually create multiple vnc server
instances.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Gonglei <arei.gonglei@huawei.com>
Teach qemu to set up a Spice server with a UNIX socket using the
following arguments -spice unix,addr=path.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Markus Armbruster
Daniel P. Berrange
Hervé Poussineau
Peter Maydell
Benjamin Herrenschmidt
Gonglei
Gerd Hoffmann
Jorge Acereda Maciá
Wang Xin
Thomas Huth
Peter Lieven
Marc-André Lureau