Fix 802.1p VLAN priority code points for VLAN_HWTAGGING.
Our in-kernel ether-vtag has a different layout to the vr TXSTAT register.

ok sthen@

cut things that relied on mclgeti for rx ring accounting/restriction over
to using if_rxr.

cut the reporting systat did over to the rxr ioctl.

tested as much as i can on alpha, amd64, and sparc64.
mpi@ has run it on macppc.
ok mpi@

Programme the VT6105M, VT6105 and RhineII-2 to permit frames of up to
1758 bytes (MTU 1740) and raise hardmtu to allow the user to set them.
Tests by Mark Patruck, Michal Markowski, mikeb, beck, dtucker. ok mikeb@

Add TX interrupt mitigation for vr(4) chips, loosely based on FreeBSD,
but with tweaks so it works on 6105Ms like those found in pcengines alix
and soekris 5501s. ok sthen@, help brad chris mikeb dlg jsing

There may be other 610x chips that would benefit from the VR_Q_INTDISABLE
quirk, please send me a dmesg if you can confirm it works.

Add hardware VLAN tag/untag support for VT6105M chips. Help/feedback/ok
brad@ chris@ jsing@ kettenis@ mikeb@ sthen@

Implement transmit DMA segments. The descriptor usage is from FreeBSD,
which contradicts the datasheet (but actually works).

Also, use a segment (instead of a new mbuf and data copy) to zero-pad
small packets, modeled after if_myx.

ok sthen@, gerhard@, henning@

Convert a number of old private copies of code which predates
pci_set_powerstate() to using it instead. Many of these chunks of code had
bugs in them, especially missing delay() calls. Some of them were doing
things our PCI subsystem is now responsible for handling. If you have
any of the affected devices, please keep an eye out for regressions.
ok kettenis

suspend/resume support for vr(4); from brynet@gmail.com

add timeout to fill rx-ring in case of mbuf shortage while in vr_intr();
fixes hanging interfaces similar to in myx(4); ok dlg@, claudio@, henning@

Wake On LAN support for vr(4).
ok deraadt

MCLGETI support for vr(4) interfaces. This will make the newer soekris and
Alix boards suck a little bit less because it will be harder to live lock them
with traffic. Diff mostly adapted from sis(4).
OK kettenis@, dlg@, sthen@

cleanup and simplify promiscuous/multicast setup (setmulti -> iff).
from Brad; ok dlg.

remove local copy of vendor id/device id, there's nothing wrong with
the data from pcidevs. suggested by chris@.

Remove config register gpio definitions for older chips, gpio mii
access disappeared years ago and replace them with config register
definitions for newer chips. From chris@.

- change device matching to use a quirks table.

- avoid the need for unnecessary mbuf copies on modern vr(4) devices
(ancient ones required longword-alignment, but this has not been the case
for several generations now). reduces cpu time, particularly for forwarded
traffic.

from chris@. tested by chris@, myself, Emilio Perea on various devices
including original VIA EPIA, Soekris net5501, pcengines.ch ALIX.

ok mpf. deraadt and dlg agree with committing at this stage.

The TX engine can get stuck if the interface went through a link
state change. Workaround by resetting the chip if necessary.
Also do not try to send packets if our link is down, as this triggers
watchdog timeouts. While there correct the VR_RXSTAT_RX_OK define
(not used yet). Changes inspired from FreeBSDs overhauled vr(4).
Testing, input and OK sthen@

o Use mbufs, for the RX ring, instead of malloc()'ing an MCLBYTES sized buffer.
o On non-strict alignment archs, dont copy the mbuf, every time, unload it, and send
it up the stack and just get a new one for the rx ring. We still do the copy on
strict alignment archs though...
o create a function to handle mbuf allocation for the rx ring, vr_mbuf_alloc(),
use it to allocate the mbufs and shuffle the bus dma setup around.

ideas/code from vic(4) and sis(4);

ok reyk@, brad@, dlg@
tested by many, been in snapshots for a while.

g/c ETHER_CRC_LEN.

ok dlg@

Remove renaming of fields with #define.

Tested by jmc@

ok dlg@

Add VLAN reception support to vr(4). Diff from Brad.

OK claudio@

call vr_setmulti() if adjusting PROMISC or ALLMULTI and do not do a full
re-init if configuring the IP address.

tested by Andrey Matveev

this should fix the VT6103.

- Rewritten TX to use only two pointers to track producer/consumer.
- Bugfix: don't forget to set IFF_OACTIVE when TX list is full.

From FreeBSD

- clear IFF_RUNNING & IFF_OACTIVE in vr_stop() before de-allocating resources.

Thanks to cloder@, millert@ and pedro@ for testing.

kill another forgotten vtophys since bus_dma conversion

Use ETHER_MAX_DIX_LEN/ETHER_MIN_LEN

remove duplication, use ETHER_ALIGN from if_ether.h

only one vtophys left (tx data)... both the rx and tx descriptors are
fully handled by bus_dma now.

rework rx to use bus_dma and sprinkle with a small amount of htole32.

unsmoke drugs in vr_encap() (simplify it quite alot)
prepare for bus_dma of tx data

don't use vtophys on rx lists any more

Sync if_vr code with FreeBSD. This makes the driver much better when it comes
to trying to recover the hardware going nuts under the load.

Reported to work by various people including krw@and I.

Must set BCR0/BCR1 since they can override the RXCFG/TXCFG registers; from freebsd via openbsd pr 3086 (thom at press dot alt-f4 dot org)

branches: 1.5.4;
Bring chip out of suspend mode in case we're booting after a shutdown from
Windows. Needed for the VT6102, but doesn't hurt older chips; from FreeBSD.

add mii support to if_vr (mostly from FreeBSD)
also sync with other FreeBSD changes

sync with freebsd up to switch to mii
also increase the delay for accessing the eeprom (seems it needs a bit
longer to get ether address settled).

branches: 1.2.6;
update to latest freebsd code, fixes stalls during heavy traffic

VIA Rhine driver, ported from FreeBSD; written by wpaul@ctr.columbia.edu.