Linux now supports Virtual Routing and Forwarding (VRF) as per:
https://www.kernel.org/doc/Documentation/networking/vrf.txt
In order for an application to bind or connect to a socket with an
address in a VRF, they need to first bind the socket to the VRF device:
setsockopt(sd, SOL_SOCKET, SO_BINDTODEVICE, dev, strlen(dev)+1);
Note "dev" is the VRF device, eg. VRF "blue", rather than an interface
enslaved to the VRF.
Add a new socket option, ZMQ_BINDTODEVICE, to bind a socket to a device.
In general, if a socket is bound to a device, eg. an interface, only
packets received from that particular device are processed by the socket.
If device is a VRF device, then subsequent binds/connects to that socket
use addresses in the VRF routing table.
Problem: GSSAPI NAMETYPE options were not documented in man
pages for zmq_getsockopt() and zmq_setsockopt().
Solution: add new options to these manual pages.
These options are confusing and redundant. Their names suggest
they apply to the tcp:// transport, yet they are used for all
stream protocols. The methods zmq::set_tcp_receive_buffer and
zmq::set_tcp_send_buffer don't use these values at all, they use
ZMQ_SNDBUF and ZMQ_RCVBUF.
Solution: merge these new options into ZMQ_SNDBUF and ZMQ_RCVBUF.
This means defaulting these two options to 8192, and removing the
new options. We now have ZMQ_SNDBUF and ZMQ_RCVBUF being used both
for TCP socket control, and for input/output buffering.
Note: the default for SNDBUF and RCVBUF are otherwise 4096.
Solution: add new [set|get]sockopt ZMQ_PRE_ALLOCATED_FD to allow
users to let ZMQ use a pre-allocated file descriptor instead of
allocating a new one. Update [set|get]sockopt documentation and
test accordingly.
The main use case for this feature is a socket-activated systemd
service. For more information about this feature see:
http://0pointer.de/blog/projects/socket-activation.html
VMCI transport allows fast communication between the Host
and a virtual machine, between virtual machines on the same host,
and within a virtual machine (like IPC).
It requires VMware to be installed on the host and Guest Additions
to be installed on a guest.
ZMQ_INVERT_MATCHING reverses the PUB/SUB prefix matching. The subscription
list becomes a rejection list. The PUB socket sends messages to all
connected (X)SUB sockets that do not have any matching subscription.
Whenever the option is used on a PUB/XPUB socket, any connecting SUB
sockets must also set it or they will reject everything the publisher
sends them. XSUB sockets are unaffected because they do not filter out
incoming messages.
Solution: change setsockopts on printable keys to expect 41, nor 40
bytes. Code still accepts 40 bytes for compatibility, and copies the
key to a well-terminated string before using it.
Fixes#1148
- This seems redundant; is there a use case for NOT providing
the IPC credentials to the ZAP authenticator?
- More, why is IPC authentication done via libzmq instead of ZAP?
Is it because we're missing the transport type on the ZAP request?
Another take on LIBZMQ-568 to allow filtering IPC connections, this time
using ZAP. This change is backward compatible. If the
ZMQ_ZAP_IPC_CREDS option is set, the user, group, and process IDs of the
peer process are appended to the address (separated by colons) of a ZAP
request; otherwise, nothing changes. See LIBZMQ-568 and zmq_setsockopt
documentation for more information.
* ZMQ_REQ_STRICT was negative option (default 1) which goes against
the standard, where defaults are zero. I renamed this to
ZMQ_REQ_RELAXED.
* ZMQ_REQ_REQUEST_IDS felt clumsy and describes the technical solution
rather than the problem/requirement. I changed to ZMQ_REQ_CORRELATE
which seems more explicit.
On ZMQ_CURVE_xxxKEY fetches, would return 41 bytes into caller's 40-byte
buffer. Now these fetches only return 41 bytes if the caller explicitly
provides a 41-byte buffer (i.e. the option size is 41).
* This is passed to the ZAP handler in the 'domain' field
* If not set, or empty, then NULL security does not call the ZAP handler
* This resolves the phantom ZAP request syndrome seen with sockets where
security was never intended (e.g. in test cases)
* This means if you install a ZAP handler, it will not get any requests
for new connections until you take some explicit action, which can be
setting a username/password for PLAIN, a key for CURVE, or the domain
for NULL.
The use of binary for CURVE keys is painful; you cannot easily copy
these in e.g. email, or use them directly in source code. There are
various encoding possibilities. Base16 and Base64 are not optimal.
Ascii85 is not safe for source (it generates quotes and escapes).
So, I've designed a new Base85 encoding, Z85, which is safe to use
in code and elsewhere, and I've modified libzmq to use this where
it also uses binary keys (in get/setsockopt).
Very simply, if you use a 32-byte value, it's Base256 (binary),
and if you use a 40-byte value, it's Base85 (Z85).
I've put the Z85 codec into z85_codec.hpp, it's not elegant C++
but it is minimal and it works. Feel free to rewrap as a real class
if this annoys you.
* ZMQ_PLAIN_SERVER, ZMQ_PLAIN_USERNAME, ZMQ_PLAIN_PASSWORD options
* Man page changes to zmq_setsockopt and zmq_getsockopt
* Man pages for ZMQ_NULL, ZMQ_PLAIN, and ZMQ_CURVE
* Test program test_security
contributors and doesn't reflect the real process. I've taken out all named
authors and referred to the contribution policy. Hopefully this will improve
the contributions to the man pages.
This patch, salvaged from a trainwreck accidental merge earlier, adds a
new sockopt, ZMQ_DELAY_ATTACH_ON_CONNECT which prevents a end point
being available to push messages to until it has fully connected, making
connect work more like bind. This also applies to reconnecting sockets,
which may cause message loss of in-queue messages, so it is sensible to
use this in conjunction with a low HWM and potentially an alternative
acknowledgement path.
Notes on most of the individual commits can be found the repository log.
This patch adds a sockopt ZMQ_DELAY_ATTACH_ON_CONNECT, which if set to 1 will attempt to preempt this behavior. It does this by extending the use of the session_base to include in the outbound as well as the inbound pipe, and only associates the pipe with the socket once it receives the connected callback via a process_attach message. This works, and a test has been added to show so, but may introduce unexpected complications. The shutdown logic in this class has become marginally more awkward because of this, requiring the session to serve as the sink for both pipes if shutdown occurs with a still-connecting pipe in place. It is also possible there could be issues around flushing the messages, but as I could not directly think how to create such an issue I have not written any code with regards to that.
The documentation has been updated to reflect the change, but please do check over the code and test and review.
