nvd-json-data-feeds/CVE-2023/CVE-2023-525xx/CVE-2023-52523.json

{
  "id": "CVE-2023-52523",
  "sourceIdentifier": "416baaa9-dc9f-4396-8d5f-8c081fb06d67",
  "published": "2024-03-02T22:15:48.213",
  "lastModified": "2024-03-04T13:58:23.447",
  "vulnStatus": "Awaiting Analysis",
  "descriptions": [
    {
      "lang": "en",
      "value": "In the Linux kernel, the following vulnerability has been resolved:\n\nbpf, sockmap: Reject sk_msg egress redirects to non-TCP sockets\n\nWith a SOCKMAP/SOCKHASH map and an sk_msg program user can steer messages\nsent from one TCP socket (s1) to actually egress from another TCP\nsocket (s2):\n\ntcp_bpf_sendmsg(s1)\t\t// = sk_prot->sendmsg\n  tcp_bpf_send_verdict(s1)\t// __SK_REDIRECT case\n    tcp_bpf_sendmsg_redir(s2)\n      tcp_bpf_push_locked(s2)\n\ttcp_bpf_push(s2)\n\t  tcp_rate_check_app_limited(s2) // expects tcp_sock\n\t  tcp_sendmsg_locked(s2)\t // ditto\n\nThere is a hard-coded assumption in the call-chain, that the egress\nsocket (s2) is a TCP socket.\n\nHowever in commit 122e6c79efe1 (\"sock_map: Update sock type checks for\nUDP\") we have enabled redirects to non-TCP sockets. This was done for the\nsake of BPF sk_skb programs. There was no indention to support sk_msg\nsend-to-egress use case.\n\nAs a result, attempts to send-to-egress through a non-TCP socket lead to a\ncrash due to invalid downcast from sock to tcp_sock:\n\n BUG: kernel NULL pointer dereference, address: 000000000000002f\n ...\n Call Trace:\n  <TASK>\n  ? show_regs+0x60/0x70\n  ? __die+0x1f/0x70\n  ? page_fault_oops+0x80/0x160\n  ? do_user_addr_fault+0x2d7/0x800\n  ? rcu_is_watching+0x11/0x50\n  ? exc_page_fault+0x70/0x1c0\n  ? asm_exc_page_fault+0x27/0x30\n  ? tcp_tso_segs+0x14/0xa0\n  tcp_write_xmit+0x67/0xce0\n  __tcp_push_pending_frames+0x32/0xf0\n  tcp_push+0x107/0x140\n  tcp_sendmsg_locked+0x99f/0xbb0\n  tcp_bpf_push+0x19d/0x3a0\n  tcp_bpf_sendmsg_redir+0x55/0xd0\n  tcp_bpf_send_verdict+0x407/0x550\n  tcp_bpf_sendmsg+0x1a1/0x390\n  inet_sendmsg+0x6a/0x70\n  sock_sendmsg+0x9d/0xc0\n  ? sockfd_lookup_light+0x12/0x80\n  __sys_sendto+0x10e/0x160\n  ? syscall_enter_from_user_mode+0x20/0x60\n  ? __this_cpu_preempt_check+0x13/0x20\n  ? lockdep_hardirqs_on+0x82/0x110\n  __x64_sys_sendto+0x1f/0x30\n  do_syscall_64+0x38/0x90\n  entry_SYSCALL_64_after_hwframe+0x63/0xcd\n\nReject selecting a non-TCP sockets as redirect target from a BPF sk_msg\nprogram to prevent the crash. When attempted, user will receive an EACCES\nerror from send/sendto/sendmsg() syscall."
    },
    {
      "lang": "es",
      "value": "En el kernel de Linux, se ha resuelto la siguiente vulnerabilidad: bpf, sockmap: rechaza las redirecciones de salida de sk_msg a sockets que no son TCP. Con un mapa SOCKMAP/SOCKHASH y un programa sk_msg, el usuario puede dirigir los mensajes enviados desde un socket TCP (s1) para que realmente salgan desde otro socket TCP (s2): tcp_bpf_sendmsg(s1) // = sk_prot-&gt;sendmsg tcp_bpf_send_verdict(s1) // __SK_REDIRECT caso tcp_bpf_sendmsg_redir(s2) tcp_bpf_push_locked(s2) tcp_bpf_push(s2) tcp_rate_check_ app_limited(s2) // espera tcp_sock tcp_sendmsg_locked(s2 ) // \u00eddem Hay una suposici\u00f3n codificada en la cadena de llamadas de que el socket de salida (s2) es un socket TCP. Sin embargo, en el commit 122e6c79efe1 (\"sock_map: Actualizar comprobaciones de tipo de calcet\u00edn para UDP\") hemos habilitado redirecciones a sockets que no son TCP. Esto se hizo por el bien de los programas BPF sk_skb. No hab\u00eda ninguna sangr\u00eda para admitir el caso de uso de env\u00edo a salida de sk_msg. Como resultado, los intentos de env\u00edo a salida a trav\u00e9s de un socket que no es TCP provocan un bloqueo debido a una conversi\u00f3n no v\u00e1lida de sock a tcp_sock: ERROR: desreferencia del puntero NULL del n\u00facleo, direcci\u00f3n: 0000000000000002f... Seguimiento de llamadas: . mostrar_regs+0x60/0x70? __die+0x1f/0x70 ? page_fault_oops+0x80/0x160? do_user_addr_fault+0x2d7/0x800? rcu_is_watching+0x11/0x50? exc_page_fault+0x70/0x1c0? asm_exc_page_fault+0x27/0x30? tcp_tso_segs+0x14/0xa0 tcp_write_xmit+0x67/0xce0 __tcp_push_pending_frames+0x32/0xf0 tcp_push+0x107/0x140 tcp_sendmsg_locked+0x99f/0xbb0 tcp_bpf_push+0x19d/0x3a0 tcp_bpf_sendmsg_redir+0x55/0xd0 tcp_bpf_send_verdict+0x407/0x550 tcp_bpf_sendmsg+0x1a1/0x390 inet_sendmsg+0x6a/0x70 sock_sendmsg+0x9d/0xc0? sockfd_lookup_light+0x12/0x80 __sys_sendto+0x10e/0x160 ? syscall_enter_from_user_mode+0x20/0x60? __this_cpu_preempt_check+0x13/0x20? lockdep_hardirqs_on+0x82/0x110 __x64_sys_sendto+0x1f/0x30 do_syscall_64+0x38/0x90 Entry_SYSCALL_64_after_hwframe+0x63/0xcd Rechace la selecci\u00f3n de sockets que no sean TCP como destino de redireccionamiento desde un programa BPF sk_msg para evitar el bloqueo. Cuando lo intente, el usuario recibir\u00e1 un error EACCES de la llamada al sistema send/sendto/sendmsg()."
    }
  ],
  "metrics": {},
  "references": [
    {
      "url": "https://git.kernel.org/stable/c/b80e31baa43614e086a9d29dc1151932b1bd7fc5",
      "source": "416baaa9-dc9f-4396-8d5f-8c081fb06d67"
    },
    {
      "url": "https://git.kernel.org/stable/c/b8f97e47b6fb84fcf2f5a22e725eefb6cf5070c2",
      "source": "416baaa9-dc9f-4396-8d5f-8c081fb06d67"
    },
    {
      "url": "https://git.kernel.org/stable/c/bc8b89b6963803a123f64aa9494155a037b3d728",
      "source": "416baaa9-dc9f-4396-8d5f-8c081fb06d67"
    },
    {
      "url": "https://git.kernel.org/stable/c/ded6e448028f0f91b6af35985afca01fa02a9089",
      "source": "416baaa9-dc9f-4396-8d5f-8c081fb06d67"
    }
  ]
}