It can happen during normal operation and it confuses users more than it
actually helps. workaround trigger counts are still available via
debugfs
Signed-off-by: Felix Fietkau <[email protected]>
Hardware encryption seems to break encrypted unicast mgmt tx.
Unfortunately the hardware TXWI header does not have a bit to indicate
that a frame is software encrypted, so sw-encrypted frames need to use a
different WCID. For that to work, the CCMP PN needs to be generated in
software, which makes things a bit slower, so only do it for keys that
also need to tx management frames.
Signed-off-by: Felix Fietkau <[email protected]>
The rx WCID lookup table and the per-WCID rate table only have 128
entries. All further WCID entries are tx-only.
Signed-off-by: Felix Fietkau <[email protected]>
On some routers, the manufacturer has made a mistake and left the chip
ID part of the EEPROM empty (it is set to 0x0000). One such router is
the ZBT-WE1026-5G (equipped with MT7612), and the consequence is that
the EEPROM check fails and the 5GHz wifi does not work at all. When
setting the chip ID to 0x7612 manually, the 5GHz wifi works fine.
When looking at EEPROM of the WE1026 (and other devices equipped with
mt76x2), I see that the PCI ID is always set to a valid value (7612 or
7662). This PR works around the issue of an empty chip ID by using the
PCI ID, if chip ID is 0. After applying this PR/patch manually to mt76,
the 5GHz on the WE1026 works out of the box. Since the PCI ID is only
used if chip ID is empty, devices with correct EEPROMs are not affected.
Signed-off-by: Kristian Evensen <[email protected]>
Sometimes the code would return a status update to mac80211 with a-mpdu
length of 0, leading to a crash in minstrel.
Make the logic of counting a-mpdu burst length easier to follow and add
an extra check to prevent this issue from happening.
Signed-off-by: Felix Fietkau <[email protected]>
Move device state check in mt76_txq_send_burst() in order to be more
reactive in stopping tx process during hw reset
Signed-off-by: Lorenzo Bianconi <[email protected]>
Fixes rate control for multi-vif. Beacons and buffered multicast packets
still need the adjusted index
Signed-off-by: Felix Fietkau <[email protected]>
Since the TXWI has no room for the MRR reconfigure the MT_WTBL registers
using the sample rate as first tx rate for channel probing avoiding
very low TCP performance in a lossy environment
Signed-off-by: Lorenzo Bianconi <[email protected]>
Signed-off-by: Felix Fietkau <[email protected]>
Allow the hardware to fall back if rates stop working, make the first
rate fail earlier.
Signed-off-by: Felix Fietkau <[email protected]>
Signed-off-by: Lorenzo Bianconi <[email protected]>
Use tx_info reference of last dequeued frame instead of the previous one
to compute 802.11 aggregate burst length
Signed-off-by: Lorenzo Bianconi <[email protected]>
Felix Fietkau