"details": "In the Linux kernel, the following vulnerability has been resolved:\n\nnet: Fix null-ptr-deref by sock_lock_init_class_and_name() and rmmod.\n\nWhen I ran the repro [0] and waited a few seconds, I observed two\nLOCKDEP splats: a warning immediately followed by a null-ptr-deref. [1]\n\nReproduction Steps:\n\n 1) Mount CIFS\n 2) Add an iptables rule to drop incoming FIN packets for CIFS\n 3) Unmount CIFS\n 4) Unload the CIFS module\n 5) Remove the iptables rule\n\nAt step 3), the CIFS module calls sock_release() for the underlying\nTCP socket, and it returns quickly. However, the socket remains in\nFIN_WAIT_1 because incoming FIN packets are dropped.\n\nAt this point, the module's refcnt is 0 while the socket is still\nalive, so the following rmmod command succeeds.\n\n # ss -tan\n State Recv-Q Send-Q Local Address:Port Peer Address:Port\n FIN-WAIT-1 0 477 10.0.2.15:51062 10.0.0.137:445\n\n # lsmod | grep cifs\n cifs 1159168 0\n\nThis highlights a discrepancy between the lifetime of the CIFS module\nand the underlying TCP socket. Even after CIFS calls sock_release()\nand it returns, the TCP socket does not die immediately in order to\nclose the connection gracefully.\n\nWhile this is generally fine, it causes an issue with LOCKDEP because\nCIFS assigns a different lock class to the TCP socket's sk->sk_lock\nusing sock_lock_init_class_and_name().\n\nOnce an incoming packet is processed for the socket or a timer fires,\nsk->sk_lock is acquired.\n\nThen, LOCKDEP checks the lock context in check_wait_context(), where\nhlock_class() is called to retrieve the lock class. However, since\nthe module has already been unloaded, hlock_class() logs a warning\nand returns NULL, triggering the null-ptr-deref.\n\nIf LOCKDEP is enabled, we must ensure that a module calling\nsock_lock_init_class_and_name() (CIFS, NFS, etc) cannot be unloaded\nwhile such a socket is still alive to prevent this issue.\n\nLet's hold the module reference in sock_lock_init_class_and_name()\nand release it when the socket is freed in sk_prot_free().\n\nNote that sock_lock_init() clears sk->sk_owner for svc_create_socket()\nthat calls sock_lock_init_class_and_name() for a listening socket,\nwhich clones a socket by sk_clone_lock() without GFP_ZERO.\n\n[0]:\nCIFS_SERVER=\"10.0.0.137\"\nCIFS_PATH=\"//${CIFS_SERVER}/Users/Administrator/Desktop/CIFS_TEST\"\nDEV=\"enp0s3\"\nCRED=\"/root/WindowsCredential.txt\"\n\nMNT=$(mktemp -d /tmp/XXXXXX)\nmount -t cifs ${CIFS_PATH} ${MNT} -o vers=3.0,credentials=${CRED},cache=none,echo_interval=1\n\niptables -A INPUT -s ${CIFS_SERVER} -j DROP\n\nfor i in $(seq 10);\ndo\n umount ${MNT}\n rmmod cifs\n sleep 1\ndone\n\nrm -r ${MNT}\n\niptables -D INPUT -s ${CIFS_SERVER} -j DROP\n\n[1]:\nDEBUG_LOCKS_WARN_ON(1)\nWARNING: CPU: 10 PID: 0 at kernel/locking/lockdep.c:234 hlock_class (kernel/locking/lockdep.c:234 kernel/locking/lockdep.c:223)\nModules linked in: cifs_arc4 nls_ucs2_utils cifs_md4 [last unloaded: cifs]\nCPU: 10 UID: 0 PID: 0 Comm: swapper/10 Not tainted 6.14.0 #36\nHardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.16.0-0-gd239552ce722-prebuilt.qemu.org 04/01/2014\nRIP: 0010:hlock_class (kernel/locking/lockdep.c:234 kernel/locking/lockdep.c:223)\n...\nCall Trace:\n <IRQ>\n __lock_acquire (kernel/locking/lockdep.c:4853 kernel/locking/lockdep.c:5178)\n lock_acquire (kernel/locking/lockdep.c:469 kernel/locking/lockdep.c:5853 kernel/locking/lockdep.c:5816)\n _raw_spin_lock_nested (kernel/locking/spinlock.c:379)\n tcp_v4_rcv (./include/linux/skbuff.h:1678 ./include/net/tcp.h:2547 net/ipv4/tcp_ipv4.c:2350)\n...\n\nBUG: kernel NULL pointer dereference, address: 00000000000000c4\n PF: supervisor read access in kernel mode\n PF: error_code(0x0000) - not-present page\nPGD 0\nOops: Oops: 0000 [#1] PREEMPT SMP NOPTI\nCPU: 10 UID: 0 PID: 0 Comm: swapper/10 Tainted: G W 6.14.0 #36\nTainted: [W]=WARN\nHardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.16.0-0-gd239552ce722-prebuilt.qemu.org 04/01/2014\nRIP: 0010:__lock_acquire (kernel/\n---truncated---",
![advisory-database[bot]](https://avatars.githubusercontent.com/in/21409?v=4&size=40){kind=avatar}
0 commit comments