Check v6 availablilty mask instead of v4.

go for it deraadt@

IPv6 is no longer on by default. It must be enabled with "inet6 eui64".

Input from sthen@, ok claudio@

Add IPv6 support to umb(4).

ok job@ bluhm@ claudio@

job@ tested with 'telnet -6 towel.blinkenlights.nl' on Fibocom L831-EAU on
IIJ MIO's network (Japan), with 'inet6 autoconf' in /etc/hostname.umb0.

When there is no network around the state timeout fires over and over again.
Change the printf into a log and only under IFF_DEBUG to reduce dmesg spam.
Loudly requested by beck@ OK deraadt@

Add umb_rtrequest() to handle RTM_PROPOSAL events and send back the
DNS proposals in that case. Also clear the DNS proposals when the
interface is deconfigured (e.g. when going down).
OK florian@

Use rtm_proposal() to send out DNS updates. Both when DNS are added and
when the interface is deconfigured (when going down).

Only set the IPv4 address in umb(4) if both MBIM_IPCONF_HAS_ADDRINFO and
MBIM_IPCONF_HAS_GWINFO are available. Configuring umb(4) without gateway
wont work the system needs a destination address for

Only set the IPv4 address in umb(4) if both MBIM_IPCONF_HAS_ADDRINFO and
MBIM_IPCONF_HAS_GWINFO are available. Configuring umb(4) without gateway
wont work the system needs a destination address for the interface.
Problem found by jsg@.
OK jsg@ deraadt@

show more ...

Switch the bpf of umb(4) from DTL_RAW to DTL_LOOP. DTL_RAW is limited to
ip only whereas DTL_LOOP passes the address family of the packet and so
supports more address families.

To make this work umb

Switch the bpf of umb(4) from DTL_RAW to DTL_LOOP. DTL_RAW is limited to
ip only whereas DTL_LOOP passes the address family of the packet and so
supports more address families.

To make this work umb_decap() prepends the AF to the packet and which is
consumed then by umb_input(). Similar umb_output() sets ph_family in the
mbuf header which is used by umb_start().
OK deraadt@ dlg@

show more ...

turn the success paths of FCC registration into debug prints. The
(unlikely) failure path remains noisy.
discussed with claudio

Make umb(4) add a default route when configuring the interface. This route
has a lower priority than wifi or wired LAN and so should only be used when
no other interface is available. With this using

Make umb(4) add a default route when configuring the interface. This route
has a lower priority than wifi or wired LAN and so should only be used when
no other interface is available. With this using umb(4) becomes less painful
Now ifconfig umb0 up will be enough especially if unwind(8) is used to handle
DNS requests.
OK deraadt@ job@ benno@

show more ...

Replace umb_ntop() with sockaddr_ntop() which does almost the same thing.
Also change the storage type for the DNS addresses to struct in_addr since
that is more convinient for userland. This include

Replace umb_ntop() with sockaddr_ntop() which does almost the same thing.
Also change the storage type for the DNS addresses to struct in_addr since
that is more convinient for userland. This includes some minor other cleanup.
OK gerhard@

show more ...

Create IF_WWAN_DEFAULT_PRIORITY which is lower than
IF_WIRELESS_DEFAULT_PRIORITY and use it in umb(4) as default prio.
OK kettenis@, sthen@

Fix typos

ok sthen@

Make some USB ioctls return ENXIO, instead of EIO, if usbd_is_dying().
Brings us one step closer towards making this condition's error code
consistent across all USB drivers.
Patch by Moritz Buhl
ok

Make some USB ioctls return ENXIO, instead of EIO, if usbd_is_dying().
Brings us one step closer towards making this condition's error code
consistent across all USB drivers.
Patch by Moritz Buhl
ok mpi@ bluhm@

show more ...

Prevent a panic in umb(4) when roaming is diabled.

Found by beck@, tested and ok by bluhm@

Reduce the scope of the NET_LOCK() in in_control(). Two functions were
protected: mrt_ioctl() and in_ioctl(). The former has no other callers
and only needs a read lock. The latter will need refactor

Reduce the scope of the NET_LOCK() in in_control(). Two functions were
protected: mrt_ioctl() and in_ioctl(). The former has no other callers
and only needs a read lock. The latter will need refactoring to reduce
the lock's scope further. In a first step, establish a single exit point
and protect most of the function body with the NET_LOCK() while removing
the NET_LOCK() from a handful of callers.

suggested by & ok mpi, ok visa

show more ...

Remove almost unused `flags' argument of suser().

The account flag `ASU' will no longer be set but that makes suser()
mpsafe since it no longer mess with a per-process field.

No objection from mill

Remove almost unused `flags' argument of suser().

The account flag `ASU' will no longer be set but that makes suser()
mpsafe since it no longer mess with a per-process field.

No objection from millert@, ok tedu@, bluhm@

show more ...

Even if a USB transfer fails, call umb_start() so that processing
of the if_snd queue doesn't stall.

ok stsp@

Kill dead ioctl handlers.

SIOCGIF{HARD,}MTU are handled by ifioctl() and not passed down to
drivers.

ok visa@

Remove NET_LOCK()'s argument.

Tested by Hrvoje Popovski, ok bluhm@

Introduce ipv{4,6}_input(), two wrappers around IP queues.

This will help transitionning to an un-KERNEL_LOCK()ed IP
forwarding path.

Disucssed with bluhm@, ok claudio@

Grab the netlock in umb_state_task() and umb_decode_ip_configuration()
when calling in_ioctl(). These ioctls modify the routing table,
which is global. So they need the lock.
Found with Zaur Molotn

Grab the netlock in umb_state_task() and umb_decode_ip_configuration()
when calling in_ioctl(). These ioctls modify the routing table,
which is global. So they need the lock.
Found with Zaur Molotnikov's static lock analyzer.
OK gerhard@ mpi@

show more ...

Byte order of IP addresses was broken on big endian machines.
MBIM already uses network byte order for IP addresses, so just
use them as they are.

ok stsp@, deraadt@

Support packet aggregation for umb(4) on tx.

tested by bluhm@, ststp@ and Bryan Vyhmeister.
ok bluhm@ ststp@

Do not clear IFF_UP, even in the error path.

This flag should only be set by the stack, drivers shouldn't
mess with it.

Discussed with dlg@ and mikeb@, ok gerhard@