Please take a look at https://github.com/golang/go/blob/master/CONTRIBUTING.md.
Merged into #7849.

What version of Go are you using (go version)?

go version go1.5beta1 freebsd/amd64

What operating system and processor architecture are you using?

FreeBSD amd64

What did you do?

tar -xzvf go1.5beta1.src.tar.gz
cd go/src
env CC=cc GOROOT_BOOTSTRAP=/usr/local/go bash all.bash

What did you expect to see?

Test pass

What did you see instead?

--- FAIL: TestInterfaces (0.00s)
interface_test.go:74: route ip+net: invalid network interface name
FAIL
FAIL net 13.270s

Thanks. Please let us know your kernel version in next time. The root cause is simply go's routing message parsers still use NET_RT_IFLIST+if_msghdr instead of NET_RT_IFLISTL+if_msghdrl, and seems like the under-development kernel doesn't return correct RTM_IFINFO messages like the following:

# cd $GOROOT/src/syscall
# go test -v -run=Route
    route_bsd_test.go:70: 3 0 ifp [6/unspec/4614 t/n/a/s=2/0/152/0]
    route_bsd_test.go:70: 3 0 ifp [6/unspec/4614 t/n/a/s=2/0/152/0]
    route_bsd_test.go:70: 3 0 ifp [24/unspec/0 t/n/a/s=0/0/152/0]
    route_bsd_test.go:70: 3 2 ifp [6/unspec/4614 t/n/a/s=2/0/152/0]

Ah, I see. Kernel versions is FreeBSD 11.0-CURRENT #0 r284988: Wed Jul 1 03:59:20 UTC 2015 I'll see if I can produce a patch.

I'm not sure whether FreeBSD net subsystem guys want to drop NET_RT_IFLIST in 11-CURRENT. If so, go's routing message parser will use NET_RT_IFLISTL eventually. According to https://www.freebsd.org/releng/index.html, July 2016 is the current scheduled date. So it may happen in go1.7.

For go1.5 The problem is types_freebsd.go strucht if_data8 and its output ztypes_freebsd_amd64.go ifData do not have matching fields to FreeBSD 11, specifically the position of ifi_datalen which is now uint16_t. In short, types_freebsd.go needs to be fixed up to support FreeBSD-11.

Earlier versions of go uses SizeofIfMsghdr (fixed size):

func (any *anyMessage) parseInterfaceMessage(b []byte) *InterfaceMessage {
    p := (*InterfaceMessage)(unsafe.Pointer(any))
    return &InterfaceMessage{Header: p.Header, Data: b[SizeofIfMsghdr:any.Msglen]}
}

where as 1.5 does (referencing p.Header.Data.Datalen)
:

    return &InterfaceMessage{Header: p.Header,
            Data: b[int(unsafe.Offsetof(p.Header.Data))+int(p.Header.Data.Datalen) : any.Msglen]}

types_freebsd.go:

// this needs:
#undef ifi_epoch
// and
#undef ifi_lastchange to work also.

struct if_data8 {
        u_char  ifi_type;
        u_char  ifi_physical;
        u_char  ifi_addrlen;
        u_char  ifi_hdrlen;
        u_char  ifi_link_state;
        u_char  ifi_spare_char1;
        u_char  ifi_spare_char2;
        u_char  ifi_datalen;
        u_long  ifi_mtu;
        ...
};

FreeBSD if.h:

struct if_data {
    /* generic interface information */
    uint8_t ifi_type;       /* ethernet, tokenring, etc */
    uint8_t ifi_physical;       /* e.g., AUI, Thinnet, 10base-T, etc */
    uint8_t ifi_addrlen;        /* media address length */
    uint8_t ifi_hdrlen;     /* media header length */
    uint8_t ifi_link_state;     /* current link state */
    uint8_t ifi_vhid;       /* carp vhid */
    uint16_t    ifi_datalen;    /* length of this data struct */
    uint32_t    ifi_mtu;    /* maximum transmission unit */
        ....
};

CL https://golang.org/cl/14757 mentions this issue.

This was reverted for breaking the build.

This is a duplicate of #7849. Also release-11 branch is not created yet. No need to rush adapting heavily under development kernels because freebee net guys still have a chance to modify abis.

CL https://golang.org/cl/17473 mentions this issue.