CVE-2026-99577 in Linux
Summary
by MITRE • 11/20/2026
In the Linux kernel, the following vulnerability has been resolved:
wifi: rtw88: usb: bound what the driver feeds the after-DTIM queue
Frames routed to the high queue are transmitted right after DTIM beacons only, inside the ATIM window, while they wait in the shared TX page pool. The driver puts no limit on how many it hands over, so as long as one station dozes, any sustained broadcast or multicast traffic outruns the drain and empties the pool: measured on an RTL8822BU AP with a single client in power save and ~40 frames/s of mDNS chatter, the free page count at 0x240 goes from 1803 to 16 in about 100 seconds and stays there for as long as the traffic lasts. From that point every other transmit queues behind the backlog, authentication responses arrive too late for anyone to join, and the reserved page download fails ("error beacon valid"). The AP keeps beaconing and only a reboot recovers.
Feed the high queue through a small budget that refills below the measured drain rate (about 3 frames per DTIM, ~15/s at dtim_period 2 on the default 2 TU ATIM window); whatever exceeds the budget leaves on its access category queue right away. The high queue backlog is now bounded by the burst size under any load, so the page pool cannot run dry, at the price that a dozing station may miss part of a broadcast storm.
If you want to get best quality of vulnerability data, you may have to visit VulDB.
Analysis
by VulDB Data Team • 11/21/2026
The vulnerability identified in the Linux kernel's rtw88 wireless driver represents a critical resource exhaustion flaw within the USB transmission path, specifically affecting how frames are managed during the After-DTIM (Delivery Traffic Indication Message) window. In Wi-Fi networks utilizing power save modes, stations periodically wake up to listen for beacons and DTIMs, which indicate pending buffered traffic at the access point. The rtw88 driver routes certain high-priority or broadcast/multicast frames to a dedicated queue intended for transmission immediately after these beacon intervals. However, the original implementation failed to impose any upper limit on the number of frame descriptors submitted by the driver to this specific hardware queue. This lack of flow control allowed sustained bursts of broadcast traffic, such as mDNS queries common in IoT environments, to rapidly consume all available transmit page buffers allocated for that period.
The operational impact of this flaw is severe and results in a denial-of-service condition for the entire wireless interface rather than just affecting specific clients. As demonstrated by empirical testing on RTL8822BU hardware with a single client in power save mode, continuous broadcast traffic at modest rates can deplete the free page count from over eighteen hundred to nearly zero within approximately one hundred seconds. Once this shared TX page pool is exhausted, all other transmission queues are forced to back up behind the stalled high-priority queue. This congestion prevents critical management frames, such as authentication responses and association requests, from being transmitted in a timely manner. Consequently, new clients cannot join the network because their authentication attempts time out due to delayed responses. Furthermore, the system enters an unstable state where beacon transmission fails with errors indicating invalid beacons, effectively breaking connectivity for all connected devices until the hardware is rebooted or the driver module is reloaded.
From a security and standards perspective, this vulnerability aligns closely with CWE-400: Uncontrolled Resource Consumption, as the application allows external traffic patterns to exhaust internal system resources without restriction. It also relates to CWE-787: Out-of-bounds Write if one considers that exhausting buffer pools can lead to undefined behavior or memory corruption in more complex driver implementations, though here it primarily manifests as a functional denial of service. In terms of the MITRE ATT&CK framework, this flaw could be exploited by an attacker performing a resource exhaustion attack (T1498) against wireless infrastructure. By generating sustained broadcast traffic, such as mDNS or DHCP requests, an adversary can effectively disable network access for legitimate users without needing to compromise credentials or exploit remote code execution flaws. The vulnerability highlights the risks associated with unbounded queueing mechanisms in embedded systems where resources are strictly limited and shared across multiple functional domains.
The resolution involves implementing a strict budget mechanism that limits the number of frames fed into the high-priority after-DTIM queue per beacon interval. This limit is calibrated to be below the measured drain rate, ensuring that the hardware can process the queued items before new ones are added. Specifically, the fix restricts transmission to approximately three frames per DTIM period under default settings. Any traffic exceeding this budget is diverted to standard access category queues rather than being held for immediate post-beacon transmission. While this mitigation effectively bounds the backlog and prevents the page pool from running dry, it introduces a trade-off in network behavior. Dozing stations may miss portions of broadcast storms because not all frames are delivered immediately after the beacon as originally intended by the high-priority queue design. However, this limitation is necessary to preserve overall system stability and ensure that critical management traffic remains responsive, thereby maintaining the availability of the wireless service for all connected clients.