fix bug in abstract sockets

[walid-git]

Current behavior:

bin/varnishd/varnishd -a @varnish -f /etc/varnish/default.vcl -n /tmp/varnishd`
ss -lnxp | grep varnish
u_str LISTEN 0      1024   @varnish@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@ 2741935            * 0    users:(("cache-main",pid=894543,fd=3),("varnishd",pid=894529,fd=3))  

As you can see, the whole 108 bytes of sun_path are used as the socket name/path.
From man UNIX(7):

The socket's address in this namespace is given by the additional bytes in sun_path that are covered by the specified length of the address structure. (Null bytes in the name have no special significance.) The name has no connection with filesystem pathnames. When the address of an abstract socket is returned, the returned addrlen is greater than sizeof(sa_family_t) (i.e., greater than 2), and the name of the socket is contained in the first (addrlen - sizeof(sa_family_t)) bytes of sun_path.

Which means that socklen is used to determine the length of name and must be set to the length of the struct sockaddr_un up to the last character of name, otherwise the 108 characters of sun_path will be treated as the name of the socket, including NULL bytes.

[nigoroll]

It's interesting how I understand exactly that paragraph from UNIX(7) in the opposite way.

[walid-git]

I agree that it can look confusing, but for me Null bytes in the name have no special significance. means that NULL bytes do not act as string delimiters, which means that the only other way is to use the length argument.

[nigoroll]

So this means we are incompatible with other implementations? Have you actually checked/tested?

[walid-git]

Yes I tried to connect as a client with socat - ABSTRACT-CLIENT:varnish, it is unable to connect without this patch

[nigoroll]

someone(tm) should have done that earlier. Sorry and thank you!

[nigoroll]

Thank you very much!