Legacy minstrel is essentially unmaintained and was showing poor performance
Replace it with minstrel_ht and improve rate selection and sampling behavior
Signed-off-by: Felix Fietkau <nbd@nbd.name>
Remove deferred sampling code which does not work well with rate tables +
probing.
Fix tx status handling if the first invalid rate idx is not set to -1
Signed-off-by: Felix Fietkau <nbd@nbd.name>
rt2800 olny gives you survey for current channel.
Survey-based ACS algorithms are failing to perform their job when working
with rt2800.
Make rt2800 save survey for every channel visited and be able to give away
that information.
There is a bug registred https://dev.archive.openwrt.org/ticket/19081 and
this patch solves the issue.
Signed-off-by: Markov Mikhail <markov.mikhail@itmh.ru>
After the status rework, ieee80211_tx_status_ext is leaking un-acknowledged
packets for stations in powersave mode.
To fix this, move the code handling those packets from __ieee80211_tx_status
into ieee80211_tx_status_ext
Reported-by: Tobias Waldvogel <tobias.waldvogel@gmail.com>
Signed-off-by: Felix Fietkau <nbd@nbd.name>
The vendor driver does things differently based on what it finds in the
SoC's CHIP_VER register, which should tell whether this is MT7620N or
MT7620A (PKG) and probably also the revision (VER) and most likely
also something about the silicon implementer (ECO).
Introduce codepaths just like the ones in the vendor driver to handle
the different chips properly.
Some of those paths are most likely dead code and left-overs from FPGA
versions or early prototypes of the chip. It'd thus be great if people
can post their kernel logs, at least the line telling the chip version
and eco, so we know what's actually out there in the wild -- all I
could find is
[ 0.000000] SoC Type: Ralink MT7620A ver:2 eco:6
and
[ 0.000000] SoC Type: Ralink MT7620N ver:2 eco:6
which would make things easier, as then we really just need to know
whether it's MT7620N or MT7620A and not care about FPGA or prototypes
with ver <= 1 and eco <= 2.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
When the nulldata frame was acked, the probe send count needs to be reset,
otherwise it will keep increasing until the connection is considered dead,
even though it fine.
Reported-by: Georgi Valkov <gvalkov@abv.bg>
Signed-off-by: Felix Fietkau <nbd@nbd.name>
Minor cleanup and code reorganization, along with a change to not disable
offload anymore when a tkip or sw crypto key is added
Signed-off-by: Felix Fietkau <nbd@nbd.name>
