mbox series

[v2,00/12] mt76x02: AP support for USB with PS

Message ID 1552991867-5087-1-git-send-email-sgruszka@redhat.com (mailing list archive)
Headers show
Series mt76x02: AP support for USB with PS | expand

Message

Stanislaw Gruszka March 19, 2019, 10:37 a.m. UTC 
On USB we do not have available PSD queue. Using standard AC queues is
unreliable for sending PS buffered frames, stations can go sleep again
before frames are sent. Only reliable way I could find to sent broadcast
PS buffered frames is put them into beacon data registers, then frames
are sent just after proper beacon. HW do the work when we encrypt the 
frames in software.

I tested functionality used arping to iwlwifi station that do
aggressive PS, it works like this:

# arping -c 1 -I wlan0 192.168.9.110 
ARPING 192.168.9.110 from 192.168.9.1 wlan0
Unicast reply from 192.168.9.110 [4C:34:88:5E:76:D5]  324.885ms
Sent 1 probes (1 broadcast(s))

Otherwise arping fails:

# arping -c 1 -I wlan0 192.168.9.110 
ARPING 192.168.9.110 from 192.168.1.148 wlan0
Sent 1 probes (1 broadcast(s))
Received 0 response(s)

Tested on MT7610U and MT7630E (on this device TX hang
in AP mode on 2.4GHz band, but this is not related to this
patch set).

RFC -> v1:
- do no remove set_tim
- add tbtt_start/stop for mt76x2u  
- rebase on top of:
  [PATCH 0/5] introduce mt76_sw_queue data structure
  https://lore.kernel.org/linux-wireless/cover.1551533068.git.lorenzo@kernel.org/

v1 -> v2
- fix true/false bug in patch 3
- change beacon slot size for USB

Stanislaw Gruszka (12):
  mt76x02: introduce mt76x02_beacon.c
  mt76x02: add hrtimer for pre TBTT for USB
  mt76x02: introduce beacon_ops
  mt76x02u: implement beacon_ops
  mt76x02: generalize some mmio beaconing functions
  mt76x02u: add sta_ps
  mt76x02: disable HW encryption for group frames
  mt76x02u: implement pre TBTT work for USB
  mt76x02: make beacon slots bigger for USB
  mt76x02u: add mt76_release_buffered_frames
  mt76: unify set_tim
  mt76x02: enable AP mode for USB

 drivers/net/wireless/mediatek/mt76/Makefile        |   2 +-
 drivers/net/wireless/mediatek/mt76/mac80211.c      |   7 +
 drivers/net/wireless/mediatek/mt76/mt76.h          |   2 +
 drivers/net/wireless/mediatek/mt76/mt7603/main.c   |   8 +-
 drivers/net/wireless/mediatek/mt76/mt76x0/init.c   |   1 -
 drivers/net/wireless/mediatek/mt76/mt76x0/main.c   |   8 +-
 drivers/net/wireless/mediatek/mt76/mt76x0/pci.c    |  11 +-
 drivers/net/wireless/mediatek/mt76/mt76x0/usb.c    |  10 +-
 drivers/net/wireless/mediatek/mt76/mt76x02.h       |  25 +-
 .../net/wireless/mediatek/mt76/mt76x02_beacon.c    | 286 +++++++++++++++++++++
 drivers/net/wireless/mediatek/mt76/mt76x02_mac.c   | 138 ----------
 drivers/net/wireless/mediatek/mt76/mt76x02_mmio.c  | 133 +++-------
 drivers/net/wireless/mediatek/mt76/mt76x02_regs.h  |   5 +-
 drivers/net/wireless/mediatek/mt76/mt76x02_usb.h   |   2 +
 .../net/wireless/mediatek/mt76/mt76x02_usb_core.c  | 157 +++++++++++
 drivers/net/wireless/mediatek/mt76/mt76x02_util.c  |  77 +-----
 .../net/wireless/mediatek/mt76/mt76x2/pci_init.c   |   2 +-
 .../net/wireless/mediatek/mt76/mt76x2/pci_main.c   |   8 +-
 drivers/net/wireless/mediatek/mt76/mt76x2/usb.c    |   1 +
 .../net/wireless/mediatek/mt76/mt76x2/usb_init.c   |   2 +-
 .../net/wireless/mediatek/mt76/mt76x2/usb_main.c   |   6 +
 drivers/net/wireless/mediatek/mt76/usb.c           |   7 +-
 22 files changed, 568 insertions(+), 330 deletions(-)
 create mode 100644 drivers/net/wireless/mediatek/mt76/mt76x02_beacon.c

Comments

Felix Fietkau March 24, 2019, 7:23 a.m. UTC | #1 
On 2019-03-19 11:37, Stanislaw Gruszka wrote:
> On USB we do not have available PSD queue. Using standard AC queues is
> unreliable for sending PS buffered frames, stations can go sleep again
> before frames are sent. Only reliable way I could find to sent broadcast
> PS buffered frames is put them into beacon data registers, then frames
> are sent just after proper beacon. HW do the work when we encrypt the 
> frames in software.
> 
> I tested functionality used arping to iwlwifi station that do
> aggressive PS, it works like this:
> 
> # arping -c 1 -I wlan0 192.168.9.110 
> ARPING 192.168.9.110 from 192.168.9.1 wlan0
> Unicast reply from 192.168.9.110 [4C:34:88:5E:76:D5]  324.885ms
> Sent 1 probes (1 broadcast(s))
> 
> Otherwise arping fails:
> 
> # arping -c 1 -I wlan0 192.168.9.110 
> ARPING 192.168.9.110 from 192.168.1.148 wlan0
> Sent 1 probes (1 broadcast(s))
> Received 0 response(s)
> 
> Tested on MT7610U and MT7630E (on this device TX hang
> in AP mode on 2.4GHz band, but this is not related to this
> patch set).
Applied, thanks.

- Felix