CVE-2026-99801: io_uring/bpf-ops: clear bpf_ctx while submitting SQEs

Published Nov 20, 2026
·
Updated

In the Linux kernel, the following vulnerability has been resolved:

iouring/bpf-ops: clear bpfctx while submitting SQEs

bpfiouringsubmitsqes() calls iosubmitsqes() from inside the loop program, with current->bpfctx set, and requests issued inline run with it set too. The socket code uses hascurrentbpfctx() to recognize calls from bpfsetsockopt() and friends, where the caller holds the socket lock and has already been vetted, and skips both.

So a SOCKETURINGOPSETSOCKOPT from the SQ passes sockoptcapable() without the capability, and runs without locksock(). A task with no capability can set SOMARK, IPTRANSPARENT or TCPREPAIR, and two threads changing IPOPTIONS on one socket, one through the ring, double free in no time. SOCKDESTROY via IORINGOPSENDMSG ends up in tcpabort() without the lock the same way.

Clear the run context around iosubmitsqes(). The SQEs come from userspace and must be issued like those of any other ring. They still run inside the program's RCU tasks trace read section, so a request that sleeps there holds up that grace period, but other request types do that already.

Affected Software

3 affected componentsFixes available
linux/kernel>=7.1.0<7.3.2
7.3.2
debian/linux<7.3.3-1
7.3.3-1
debian/linux
6.1.176-16.1.192-26.12.107-16.12.117-17.3.4-17.3.5-1

Remediation

Recommended actions to resolve this vulnerability, in priority order.

  1. Upgrade

    Upgrade linux/Kernel to a version that resolves this vulnerability.

    Fixed in 7.3.2
  2. Upgrade

    Upgrade debian/linux to a version that resolves this vulnerability.

    Fixed in 7.3.3-1
  3. Upgrade

    Upgrade debian/linux to a version that resolves this vulnerability.

    Fixed in 6.1.176-1Fixed in 6.1.192-2Fixed in 6.12.107-1Fixed in 6.12.117-1Fixed in 7.3.4-1Fixed in 7.3.5-1
  4. Compensating control

    Clear current->bpf_ctx while submitting SQEs in bpf_io_uring_submit_sqes(), around the call to io_submit_sqes().

Frequently Asked Questions

1

Who can exploit this issue?

A task without the relevant capability can exploit it by submitting crafted SQEs through the affected io_uring BPF operations path. The described impact includes setting SO_MARK, IP_TRANSPARENT, or TCP_REPAIR without the required capability.

2

What conditions are required for exploitation?

The attacker needs access to an io_uring setup that can submit SQEs from within the affected loop program path. Exploitation relies on SOCKET_URING_OP_SETSOCKOPT or IORING_OP_SENDMSG requests being issued from that path while current->bpf_ctx is set.

3

What are the potential impacts beyond unauthorized socket options?

Requests can bypass lock_sock() as well as capability checks. Concurrent changes to IP_OPTIONS on the same socket, with one change made through the ring, can cause a double free; SOCK_DESTROY reached through IORING_OP_SENDMSG can call tcp_abort() without the socket lock.

4

What changes with the fix?

The fix clears the BPF run context around io_submit_sqes(), causing SQEs originating from userspace to be handled like requests from any other ring. This prevents the socket code from incorrectly treating those requests as vetted BPF socket-helper calls.

Contact

SecAlerts Pty Ltd.
132 Wickham Terrace
Fortitude Valley,
QLD 4006, Australia
info@secalerts.co
By using SecAlerts services, you agree to our services end-user license agreement. This website is safeguarded by reCAPTCHA and governed by the Google Privacy Policy and Terms of Service. All names, logos, and brands of products are owned by their respective owners, and any usage of these names, logos, and brands for identification purposes only does not imply endorsement. If you possess any content that requires removal, please get in touch with us.
© 2026 SecAlerts Pty Ltd.
ABN: 70 645 966 203, ACN: 645 966 203