{"id":"AZL-109059","summary":"CVE-2026-98455 affecting package kernel 6.6.157.1-1","details":"In the Linux kernel, the following vulnerability has been resolved:\n\nBluetooth: RFCOMM: fix NULL dereference of dlc-\u003esession in RFCOMM_CONNINFO\n\nThe RFCOMM_CONNINFO getsockopt handler accepts a socket that is not\nconnected as long as deferred setup is enabled:\n\n\tif (sk-\u003esk_state != BT_CONNECTED &&\n\t\t\t\t!rfcomm_pi(sk)-\u003edlc-\u003edefer_setup) {\n\t\terr = -ENOTCONN;\n\t\tbreak;\n\t}\n\tl2cap_sk = rfcomm_pi(sk)-\u003edlc-\u003esession-\u003esock-\u003esk;\n\ndlc-\u003edefer_setup is set in rfcomm_sock_init() when rfcomm_connect_ind()\ncreates a child socket for an incoming connection on a listening socket\nthat has BT_DEFER_SETUP enabled. It is never cleared afterwards. The\nsession, however, can go away underneath it.\n\nrfcomm_recv_disc() forces the dlc state before tearing it down:\n\n\td-\u003estate = BT_CLOSED;\n\t__rfcomm_dlc_close(d, err);\n\nThe RFCOMM_DEFER_SETUP early return in __rfcomm_dlc_close() only covers\nBT_CONNECT, BT_CONFIG, BT_OPEN and BT_CONNECT2, so with the state\nalready BT_CLOSED that switch does not match and the function falls\nthrough to rfcomm_dlc_unlink(), which sets d-\u003esession = NULL, while\nd-\u003edefer_setup stays 1.\n\nA getsockopt(SOL_RFCOMM, RFCOMM_CONNINFO) on the accepted socket after\nthat point therefore skips the -ENOTCONN path -- sk-\u003esk_state is\nBT_CLOSED, but dlc-\u003edefer_setup is still set -- and dereferences the\nNULL session. No race is needed: once the DISC has been processed, the\ndereference is unconditional.\n\nReproduced on a KASAN kernel under QEMU with a BR/EDR peer emulated over\n/dev/vhci: the peer brings up an ACL link, opens L2CAP on the RFCOMM\nPSM, starts a session and sends SABM for a channel bound with\nBT_DEFER_SETUP, and sends DISC for that dlci after the socket has been\naccepted. getsockopt(SOL_RFCOMM, RFCOMM_CONNINFO) on the accepted\nsocket then hits:\n\n Oops: general protection fault, probably for non-canonical address\n 0xdffffc0000000002: 0000 [#1] SMP KASAN PTI\n KASAN: null-ptr-deref in range [0x0000000000000010-0x0000000000000017]\n CPU: 1 UID: 0 PID: 150 Comm: init Tainted: G B 7.3.0-rc3-g5dd1818b15d9\n Hardware name: QEMU Standard PC (i440FX + PIIX, 1996)\n RIP: 0010:rfcomm_sock_getsockopt+0x529/0x780\n Call Trace:\n  \u003cTASK\u003e\n  do_sock_getsockopt+0x3ad/0x7d0\n  __sys_getsockopt+0x10e/0x1b0\n  __x64_sys_getsockopt+0xc2/0x160\n  do_syscall_64+0xda/0x4b0\n  entry_SYSCALL_64_after_hwframe+0x77/0x7f\n  \u003c/TASK\u003e\n\n0x10 is the offset of sock in struct rfcomm_session;\nrfcomm_sock_getsockopt_old() is inlined into rfcomm_sock_getsockopt().\n\nCommit 43a556b2fd43 (\"Bluetooth: RFCOMM: take rfcomm_mutex for the\ndeferred setup accept\") fixed the same \"a remote DISC clears the session\nwhile deferred setup is still flagged\" problem in rfcomm_dlc_accept();\nthis is the remaining instance of it, in the getsockopt path.\n\nDeferred setup only leaves a socket usable here once it has reached\nBT_CONNECT2, so restrict the exception to that state and check that a\nsession is actually present before following it.","modified":"2026-10-19T14:16:31.579949133Z","published":"2026-10-18T16:16:55Z","upstream":["CVE-2026-98455"],"references":[{"type":"WEB","url":"https://nvd.nist.gov/vuln/detail/CVE-2026-98455"}],"affected":[{"package":{"name":"kernel","ecosystem":"Azure Linux:3","purl":"pkg:rpm/azure-linux/kernel"},"ranges":[{"type":"ECOSYSTEM","events":[{"introduced":"0"},{"last_affected":"6.6.157.1-1"}]}],"database_specific":{"source":"https://github.com/microsoft/AzureLinuxVulnerabilityData/blob/main/osv/AZL-109059.json"}}],"schema_version":"1.9.0"}