{"id":"CVE-2026-99213","published":"2026-10-28T11:17:42.770","lastModified":"2026-10-28T11:17:42.770","description":"In the Linux kernel, the following vulnerability has been resolved:\n\nvsock: treat TCP_CLOSING as once-established\n\nHandle the TCP_ESTABLISHED -> TCP_CLOSING transition on OP_RST, which can\nrace with the connect loop. Immediately break and return 0 on\nESTABLISHED/CLOSING. The return value of connect() should reflect what\nhappened up to the point the connection was established or failed. Events\nthat occur afterwards must not affect it. E.g. even if OP_RW has already\nset sk_err, connect() should still return 0, not ENOBUFS because of it.\nAdapt the inaccurate comment above signal_pending().\n\nResetting a socket that is still present in connected_table can lead to\nmemory corruption. The reporter noted lost transports for in-flight skbs,\nand I have reproduced crashes caused by re-insertion into connected_table.\n\n  list_add double add: new=, prev=, next=.\n  kernel BUG at lib/list_debug.c:35!\n  Oops: invalid opcode: 0000 [#1] SMP KASAN NOPTI\n  Workqueue: vsock-loopback vsock_loopback_work\n  RIP: 0010:__list_add_valid_or_report+0x11f/0x130\n  Call Trace:\n   vsock_insert_connected.cold+0xe/0x13\n   virtio_transport_recv_pkt+0x10e9/0x1460\n   vsock_loopback_work+0x305/0x480\n   process_one_work+0xe4c/0x1560\n   worker_thread+0x4f1/0xd60\n   kthread+0x36e/0x470\n   ret_from_fork+0x47b/0x6b0\n   ret_from_fork_asm+0x1a/0x30\n\nThis fix is supplementary to commit 002541ef650b (\"vsock: Ignore\nsignal/timeout on connect() if already established\"). Details under Link.","cvssScore":null,"cvssSeverity":null,"cvssVector":null,"cwes":[],"vendors":[],"products":[],"references":[{"url":"https://git.kernel.org/stable/c/44d1ca05d6094a5d75c671a0545249186d790b8d","tags":[]},{"url":"https://git.kernel.org/stable/c/4f6ac74fa594a918d41255c941256e1d51592d3b","tags":[]},{"url":"https://git.kernel.org/stable/c/c2c5359ef2c33684fa3f3652fd761a483db8d1e7","tags":[]}],"exploitRefs":[],"hasPoc":false,"ai":null}