Search Linux Wireless

[BUG] mt7921u: TX path stalls and page_pool errors during VIF concurrency (STA + Monitor)

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



Hello,

I have been debugging a persistent issue with the mt7921u driver on a
6.16.0-rc6 kernel. The driver becomes unstable when a monitor mode
interface is used concurrently with a managed (STA) interface.

Hardware: MediaTek MT7921U (USB ID: 0e8d:7961)

Symptoms:

    When a monitor interface is created while a managed interface is
associated, the driver enters a stalled state.

    dmesg is flooded with page_pool_release_retry() stalled pool
shutdown messages.

    Logs also show timed out waiting for pending tx and wlan0: failed
to remove key (...) from hardware (-19) upon resume from suspend,
indicating a firmware communication failure.

    A deauthentication tool named 'Auth DoS' reports an impossibly
high packet rate (e.g., >300,000 pkts/sec) but has no actual effect on
the target, confirming the TX path is stalled.

Diagnosis:
The core issue seems to be a state management conflict. The TX stall
symptom that occurs in the concurrent (STA + Monitor) scenario is
identical to the symptom that occurs in a standalone monitor mode
scenario if the driver is not patched for raw packet injection.

This suggests that when the STA interface is active, its state
configuration interferes with or overwrites the necessary hardware
settings for the monitor interface's raw packet injection path (e.g.,
WCID configuration, disabling hardware offloads).

Patches Applied:
We have successfully developed several patches that solve preliminary issues:

    A patch to mt7921_set_tx_sar_pwr to correctly set TX power.

    A patch to mt7921_usb_sdio_tx_prepare_skb to handle raw packet
injection in standalone monitor mode.

    A patch to mt7921_add_interface to assign a dedicated omac_idx to
the monitor VIF.

While these patches fix issues in isolation, they are not enough to
prevent the TX stall in the concurrent VIF scenario. The issue appears
to be deeper in the driver's state management logic when multiple BSS
contexts are active.

Could the developers please look into this? This bug severely impacts
the usability of the mt7921u chipset for security research and
advanced networking tasks.

Thank you.




[Index of Archives]     [Linux Host AP]     [ATH6KL]     [Linux Wireless Personal Area Network]     [Linux Bluetooth]     [Wireless Regulations]     [Linux Netdev]     [Kernel Newbies]     [Linux Kernel]     [IDE]     [Git]     [Netfilter]     [Bugtraq]     [Yosemite Hiking]     [MIPS Linux]     [ARM Linux]     [Linux RAID]

  Powered by Linux