field and into a separate fast-frames staging pointer in ieee80211_node.
The A-MPDU TX path allows txa_private to be used by drivers. So it will
clash with any attempt to use fast-frames. Now, fast-frames is not really
anything special - it's just a custom ethernet frame type that contains
two MSDUs into one MPDU. So all the NIC has to support doing is transmitting
up to a 4KiB frame with an arbitrary ethertype and bam! Fast-frames.
However, using txa_private means we can /either/ do fast-frames or A-MPDU TX,
so fast frames has been turned off in the Atheros HAL for 11n chipsets.
This is a bit silly - it actually means that 802.11 performance to/from
11abg Atheros chips is actually better than between an 11abg atheros device
and an 11n Atheros device.
So:
* create a new mbuf staging queue for fast frames. It only queues a single
frame in the staging queue (and there's a top-level ic staging queue
used for expiry/tracking) so it's just an mbuf pointer per TID.
* Still use the ampdu TX packet counter to determine whether to do
aggregation or not. It'll double count if we start doing both A-MPDU TX
and fast frames, but that's not all that important right now.
* Initialise the pps tracker so ticks isn't zero. This ensures that
fast-frames actually gets used - without it, the ticks math overflows
and the pps math always sets txa_pps=0. This is the same bug that
plagued A-MPDU TX starting logic.
This actually allows fast-frames transmit to occur between the AR9331
(in 11n HT/20 mode) and AR9170 (if_otus) in 11bg mode.
Now, this is a great big no-op on atheros 11n hardware, so don't worry.
It may mean you start seeing more reliable fast-frames transmission on
11abg hardware which may expose some more amusing bugs.
TODO:
* further testing and debugging of all of this before flipping on
fast-frames in if_ath (for 11n) and if_otus.
in the superg fast-frames code.
This harks back to an earlier commit (r280349) where I found that
initialising the pps code with ticks=0 would cause hilariously bad
hz ticks wraparound failures, leading to never actually aggregating
traffic. This is still true for the superg path and so I have to
do the same thing there.
This is a big no-op; a subsequent commit will flip this on so it
works with the fast-frames transmit path.
Tested:
* AR9170, otus(4) - STA mode, 11bg operation
* AR9331, AP mode
net80211 receive path. This allows drivers (notably USB right now, but
anything/everything!) to optionally defer bulk RX of 802.11 frames until
/outside/ of the driver lock(s), rather than doing:
UNLOCK(sc);
ieee80211_input*()
LOCK(sc);
.. which is really stupid.
The existing API is maintaned - if ieee80211_input() / ieee80211_input_all()
is called then the RSSI/NF values are used. If the MIMO versions are called
with a given rx status pointer then it's used. Else, it'll use whatever
is in the RX mbuf tag.
Some fullmac devices may rely on the stack starting it but not doing it.
Whilst here, remove a duplicate LE_* macro definition, thanks to
Andriy Voskoboinyk <s3erios@gmail.com>.
for re-ordering.
Devices like if_rsu don't pass through action/management frames but do send
firmware commands to inform us of things. One of those notifications is
the RX A-MPDU negotiated parameters.
connectivity interact with the net80211 stack.
Historical background: originally wireless devices created an interface,
just like Ethernet devices do. Name of an interface matched the name of
the driver that created. Later, wlan(4) layer was introduced, and the
wlanX interfaces become the actual interface, leaving original ones as
"a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer
and a driver became a mix of methods that pass a pointer to struct ifnet
as identifier and methods that pass pointer to struct ieee80211com. From
user point of view, the parent interface just hangs on in the ifconfig
list, and user can't do anything useful with it.
Now, the struct ifnet goes away. The struct ieee80211com is the only
KPI between a device driver and net80211. Details:
- The struct ieee80211com is embedded into drivers softc.
- Packets are sent via new ic_transmit method, which is very much like
the previous if_transmit.
- Bringing parent up/down is done via new ic_parent method, which notifies
driver about any changes: number of wlan(4) interfaces, number of them
in promisc or allmulti state.
- Device specific ioctls (if any) are received on new ic_ioctl method.
- Packets/errors accounting are done by the stack. In certain cases, when
driver experiences errors and can not attribute them to any specific
interface, driver updates ic_oerrors or ic_ierrors counters.
Details on interface configuration with new world order:
- A sequence of commands needed to bring up wireless DOESN"T change.
- /etc/rc.conf parameters DON'T change.
- List of devices that can be used to create wlan(4) interfaces is
now provided by net.wlan.devices sysctl.
Most drivers in this change were converted by me, except of wpi(4),
that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing
changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann,
Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in
testing.
Reviewed by: adrian
Sponsored by: Netflix
Sponsored by: Nginx, Inc.
is configured on a channel that isn't valid in the new operating mode.
This isn't strictly true - it should find the first channel that is
available for the given operating mode.
However, I think defaulting to the first channel is fine - it's typically
available for all modes.
If someone would like to correctly implement this feature - try to
find a channel that is valid for the given operating mode and error
out if we can't find one.
This prevents various NICs (eg wpi(4)) from throwing a firmware error.
Tested:
* ath(4), STA/AP mode
* iwn(4), STA/adhoc mode
PR: kern/202502
Submitted by: Andriy Voskoboinyk <s3erios@gmail.com>
* 286410
* 286413
* 286416
The initial commit broke a variety of debug and features that aren't
in the GENERIC kernels but are enabled in other platforms.
with the net80211 stack.
Historical background: originally wireless devices created an interface,
just like Ethernet devices do. Name of an interface matched the name of
the driver that created. Later, wlan(4) layer was introduced, and the
wlanX interfaces become the actual interface, leaving original ones as
"a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer
and a driver became a mix of methods that pass a pointer to struct ifnet
as identifier and methods that pass pointer to struct ieee80211com. From
user point of view, the parent interface just hangs on in the ifconfig
list, and user can't do anything useful with it.
Now, the struct ifnet goes away. The struct ieee80211com is the only
KPI between a device driver and net80211. Details:
- The struct ieee80211com is embedded into drivers softc.
- Packets are sent via new ic_transmit method, which is very much like
the previous if_transmit.
- Bringing parent up/down is done via new ic_parent method, which notifies
driver about any changes: number of wlan(4) interfaces, number of them
in promisc or allmulti state.
- Device specific ioctls (if any) are received on new ic_ioctl method.
- Packets/errors accounting are done by the stack. In certain cases, when
driver experiences errors and can not attribute them to any specific
interface, driver updates ic_oerrors or ic_ierrors counters.
Details on interface configuration with new world order:
- A sequence of commands needed to bring up wireless DOESN"T change.
- /etc/rc.conf parameters DON'T change.
- List of devices that can be used to create wlan(4) interfaces is
now provided by net.wlan.devices sysctl.
Most drivers in this change were converted by me, except of wpi(4),
that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing
changes to at least 8 drivers. Thanks to Olivier Cochard, gjb@, mmoll@,
op@ and lev@, who also participated in testing. Details here:
https://wiki.freebsd.org/projects/ifnet/net80211
Still, drivers: ndis, wtap, mwl, ipw, bwn, wi, upgt, uath were not
tested. Changes to mwl, ipw, bwn, wi, upgt are trivial and chances
of problems are low. The wtap wasn't compilable even before this change.
But the ndis driver is complex, and it is likely to be broken with this
commit. Help with testing and debugging it is appreciated.
Differential Revision: D2655, D2740
Sponsored by: Nginx, Inc.
Sponsored by: Netflix
In order for drivers to provide an alternate set of scan methods,
these have to finally use an indirection table and all of the calls
in ieee80211_scan.c need to use said table.
For all existing drivers - this is basically a glorified, KBI-breaking
functional no-op.
This is also not the final form - too much functionality is currently
hiding in ieee80211_scan_sw.c that should be in ieee80211_scan.c.
That'll be the target of some follow-up commits.
Note:
* You have to recompile your kernel/drivers after this - the net80211 KBI has
changed.
* I'm not yet planning on bumping any versioning - I have a few more things
to shuffle around.
Tested:
* urtwn(4) - STA mode
* Intel 7260 in local repo - overriding the methods and table at
attach time has the desired effect (ie, all the methods are called,
but nothing is ever performed.)
The intel 7260 driver under development requires this - the scans come
in as normal frames but with the frequency provided. The correct method
is to have the driver provide flags (so we can determine if it's 11b
or 11g); this will have to do in the meantime.
Without this, the channel found is 11b, and no ERP (ie "11g") bits
are negotiated with the AP.
This allows the 7260 in 2ghz mode to operate in 11bg, rather than
just 11b.
Tested:
* intel 7260 driver, 11bg channels
with the transmit params.
This allows raw 802.11 frames to be queued in the driver if necessary,
rather than requiring it to be direct-dispatched into the hardware.
Tested:
* ath(4), STA mode
* iwn(4), STA mode
MFC after: 2 weeks
Sponsored by: Norse Corp, Inc.
802.11 mbufs.
The raw transmit path currently doesn't make it easy to queue
these frames:
* there's no node reference stored in the mbuf, like for the normal
path, and
* the bpf supplied raw transmit parameters (rate, rts/cts, etc)
are passed in as an argument, not as an mbuf tag.
In order to support driver queuing of these frames, we need to
be able to put the above into the mbuf before the driver gets it,
so the driver /can/ put it into a queue if needed.
Use an mbuf tag and for now just verbatim copy the bpf parameters
into it. Later on it may grow to include more options but this
will do for now.
Why would you want to queue raw frames? Well, in the case of
iwn(4), we can't send the firmware frames to transmit before
we hear a beacon - the firmware will consider passive channels
as unavailable until it hears a beacon. The firmware "passive"
channel state is cleared upon each RXON command, which is sent to
update association status. So, when we attempt association and
authorisation, the RXON command causes the firmware to clear out
what it's already seen, and so we have to wait for a beacon before
we can transmit.
Before people get overly excited - this alone doesn't "fix" 5GHz
operation - it just makes it (more) possible.
The aim here is to convert all the drivers over to use a raw_xmit()
API that doesn't include the node and params - instead they'd get
those from the mbuf. Then raw_xmit() becomes just a side-channel
version of the normal transmit path for management traffic.
MFC after: 2 weeks
Sponsored by: Norse Corp, Inc.
RTM_IEEE80211_RADIO routing messages, broadcast them on all vap interfaces
instead of sending them on parent.
Reviewed by: adrian
Sponsored by: Netflix
Sponsored by: Nginx, Inc.
DragonflyBSD uses the FreeBSD wireless stack and drivers. Their malloc()
API is named differently, so they don't have userland/kernel symbol
clashes like we do (think libuinet.)
So, to make it easier for them and to port to other BSDs/other operating
systems, start hiding the malloc specific bits behind defines in
ieee80211_freebsd.h.
DragonflyBSD can now put these portability defines in their local
ieee80211_dragonflybsd.h.
This should be a great big no-op for everyone running wifi.
TODO:
* kill M_WAITOK - some platforms just don't want you to use it
* .. and/or handle it returning NULL rather than waiting forever.
* MALLOC_DEFINE() ?
* Migrate the well-known malloc names (eg M_TEMP) to net80211
namespace defines.
Smart NICs with firmware (eg wpi, iwn, the new atheros parts, the intel 7260
series, etc) support doing a lot of things in firmware. This includes but
isn't limited to things like scanning, sending probe requests and receiving
probe responses. However, net80211 doesn't know about any of this - it still
drives the whole scan/probe infrastructure itself.
In order to move towards suppoting smart NICs, the receive path needs to
know about the channel/details for each received packet. In at least
the iwn and 7260 firmware (and I believe wpi, but I haven't tried it yet)
it will do the scanning, power-save and off-channel buffering for you -
all you need to do is handle receiving beacons and probe responses on
channels that aren't what you're currently on. However the whole receive
path is peppered with ic->ic_curchan and manual scan/powersave handling.
The beacon parsing code also checks ic->ic_curchan to determine if the
received beacon is on the correct channel or not.[1]
So:
* add freq/ieee values to ieee80211_rx_stats;
* change ieee80211_parse_beacon() to accept the 'current' channel
as an argument;
* modify the iv_input() and iv_recv_mgmt() methods to include the rx_stats;
* add a new method - ieee80211_lookup_channel_rxstats() - that looks up
a channel based on the contents of ieee80211_rx_stats;
* if it exists, use it in the mgmt path to switch the current channel
(which still defaults to ic->ic_curchan) over to something determined
by rx_stats.
This is enough to kick-start scan offload support in the Intel 7260
driver that Rui/I are working on. It also is a good start for scan
offload support for a handful of existing NICs (wpi, iwn, some USB
parts) and it'll very likely dramatically improve stability/performance
there. It's not the whole thing - notably, we don't need to do powersave,
we should not scan all channels, and we should leave probe request sending
to the firmware and not do it ourselves. But, this allows for continued
development on the above features whilst actually having a somewhat
working NIC.
TODO:
* Finish tidying up how the net80211 input path works.
Right now ieee80211_input / ieee80211_input_all act as the top-level
that everything feeds into; it should change so the MIMO input routines
are those and the legacy routines are phased out.
* The band selection should be done by the driver, not by the net80211
layer.
* ieee80211_lookup_channel_rxstats() only determines 11b or 11g channels
for now - this is enough for scanning, but not 100% true in all cases.
If we ever need to handle off-channel scan support for things like
static-40MHz or static-80MHz, or turbo-G, or half/quarter rates,
then we should extend this.
[1] This is a side effect of frequency-hopping and CCK modes - you
can receive beacons when you think you're on a different channel.
In particular, CCK (which is used by the low 11b rates, eg beacons!)
is decodable from adjacent channels - just at a low SNR.
FH is a side effect of having the hardware/firmware do the frequency
hopping - it may pick up beacons transmitted from other FH networks
that are in a different phase of hopping frequencies.
ieee80211_pwrsave() can fail due to queue overflow, check its return code
and increment oerrors counter when it fails. Also handle more error cases
and update oerrors counter when we don't send mbuf due to some errors.
Return ENETDOWN when parent interface isn't ready. Update obytes and omcasts
counters in corresponding places.
PR: 184626
Differential Revision: https://reviews.freebsd.org/D2621
Reviewed by: adrian
MFC after: 1 week
years for head. However, it is continuously misused as the mpsafe argument
for callout_init(9). Deprecate the flag and clean up callout_init() calls
to make them more consistent.
Differential Revision: https://reviews.freebsd.org/D2613
Reviewed by: jhb
MFC after: 2 weeks
frames to 0
From IEEE Std. 802.11-2012, 8.3.2.1 "Data frame format", p. 415 (513):
"The Sequence Control field for QoS (+)Null frames is ignored by the receiver
upon reception."
At this moment, any <mode>_input() function interprets them as regular QoS data
frames with TID = 0. As a result, stations, that use another TX sequence for
QoS Null frames (e.g. wpi(4), where (QoS) Null frames are generated by the
firmware), may experience significant packet loss with any other NIC in hostap
mode.
Tested:
* wpi(4) (author)
* iwn(4) - Intel 5100, STA mode (me)
PR: kern/200128
Submitted by: Andriy Voskoboinyk <s3erios@gmail.com>
results.
Right now the scan infrastructure assumes the channel is under net80211
control, and that when receiving beacon frames for scanning, the
current channel is indeed what ic_curchan is set to.
But firmware NICs with firmware scan support need more than this -
they can do background scans whilst hiding the off-channel behaviour
from net80211. Ie, net80211 still thinks everything is associated
and on the main channel, but it's getting scan results from all the
background traffic.
However sta_add() pays attention to ic_curchan and discards scan
results that aren't on the right channel. CCK beacon frames can be
decoded from adjacent channels so the receive path and sta_add
discard these as appropriate. This is fine for software scanning
like for ath(4), but not for firmware NICs. So with those, the
whole concept of background firmware scanning won't work without
major hacks (eg, overriding ic_curchan before calling the beacon
input / scan add.)
As part of my scan overhaul, modify sta_add() and the scan_add()
APIs to take an explicit current channel. The normal RX path
will set it to ic_curchan so it's a no-op. However, drivers may
decide to (eventually!) override the scan method to set the
"right" current channel based on what the firmware reports the
scan state is.
So for example, iwn, rsu and other NICs will eventually do this:
* driver issues scan start firmware command;
* firmware sends a "scan start on channel X" notify;
* firmware sends a bunch of beacon RX's as part of
the scan results;
* .. and the driver will replace scan_add() curchan with channel X,
so scan results are correct.
* firmware sends a "scan start on channel Y" notify;
* firmware sends more beacons...
* .. the driver replaces scan_add() curchan with channel Y.
Note:
* Eventually, net80211 should eventually grow the idea of a per-packet
current channel. It's possible in various modes (eg WAVE, P2P, etc)
that individual frames can come in from different channels and that
is under firmware control rather than driver/net80211 control, so
we should support that.
It turns out that ieee80211_start_scan_locked() is only ever called by
the swscan code and it won't likely be required by firmware scanning
implementations.
So, don't bother keeping it in ieee80211_scan.c and it likely won't
become an API call.
Tested:
* Intel 5100, STA mode
* AR5416, STA mode
to us. Instead, advertise what we can do based on what the AP says and what
we're capped at by the VAP settings.
For non-STA modes we still advertise what our VAP settings are.
It may be that I've over-complicated this and instead of capping things
we can just always announce what we're capable of. But this should at least
stop the blatantly wrong handling of A-MPDU parameters.
(I'll happily simplify things if someone can dig up a replacement, better
compliant behaviour.)
PR: kern/176201
However, IBSS merge will be performed only if a driver calls
ieee80211_ibss_merge(); so, this applicable to the ath(4) only.
Also, this should fix bug 167870.
PR: kern/199632
Submitted by: Andriy Voskoboinyk <s3erios@gmail.com>
