"value":"In the Linux kernel, the following vulnerability has been resolved:\n\nipvlan: Fix out-of-bound bugs caused by unset skb->mac_header\n\nIf an AF_PACKET socket is used to send packets through ipvlan and the\ndefault xmit function of the AF_PACKET socket is changed from\ndev_queue_xmit() to packet_direct_xmit() via setsockopt() with the option\nname of PACKET_QDISC_BYPASS, the skb->mac_header may not be reset and\nremains as the initial value of 65535, this may trigger slab-out-of-bounds\nbugs as following:\n\n=================================================================\nUG: KASAN: slab-out-of-bounds in ipvlan_xmit_mode_l2+0xdb/0x330 [ipvlan]\nPU: 2 PID: 1768 Comm: raw_send Kdump: loaded Not tainted 6.0.0-rc4+ #6\nardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.14.0-1.fc33\nall Trace:\nprint_address_description.constprop.0+0x1d/0x160\nprint_report.cold+0x4f/0x112\nkasan_report+0xa3/0x130\nipvlan_xmit_mode_l2+0xdb/0x330 [ipvlan]\nipvlan_start_xmit+0x29/0xa0 [ipvlan]\n__dev_direct_xmit+0x2e2/0x380\npacket_direct_xmit+0x22/0x60\npacket_snd+0x7c9/0xc40\nsock_sendmsg+0x9a/0xa0\n__sys_sendto+0x18a/0x230\n__x64_sys_sendto+0x74/0x90\ndo_syscall_64+0x3b/0x90\nentry_SYSCALL_64_after_hwframe+0x63/0xcd\n\nThe root cause is:\n 1. packet_snd() only reset skb->mac_header when sock->type is SOCK_RAW\n and skb->protocol is not specified as in packet_parse_headers()\n\n 2. packet_direct_xmit() doesn't reset skb->mac_header as dev_queue_xmit()\n\nIn this case, skb->mac_header is 65535 when ipvlan_xmit_mode_l2() is\ncalled. So when ipvlan_xmit_mode_l2() gets mac header with eth_hdr() which\nuse \"skb->head + skb->mac_header\", out-of-bound access occurs.\n\nThis patch replaces eth_hdr() with skb_eth_hdr() in ipvlan_xmit_mode_l2()\nand reset mac header in multicast to solve this out-of-bound bug."
"value":"En el kernel de Linux, se ha resuelto la siguiente vulnerabilidad: ipvlan: corrige errores fuera de los l\u00edmites causados por skb->mac_header no configurado si se usa un socket AF_PACKET para enviar paquetes a trav\u00e9s de ipvlan y se cambia la funci\u00f3n xmit predeterminada del socket AF_PACKET desde dev_queue_xmit() a paquete_direct_xmit() a trav\u00e9s de setsockopt() con el nombre de opci\u00f3n PACKET_QDISC_BYPASS, es posible que skb->mac_header no se restablezca y permanezca en el valor inicial de 65535, esto puede desencadenar errores de losa fuera de los l\u00edmites como se muestra a continuaci\u00f3n : =================================================== ================ UG: KASAN: losa fuera de los l\u00edmites en ipvlan_xmit_mode_l2+0xdb/0x330 [ipvlan] PU: 2 PID: 1768 Comm: raw_send Kdump: cargado No contaminado 6.0.0-rc4+ #6 nombre de hardware: PC est\u00e1ndar QEMU (i440FX + PIIX, 1996), BIOS 1.14.0-1.fc33 todo Seguimiento: print_address_description.constprop.0+0x1d/0x160 print_report.cold+0x4f/0x112 kasan_report+ 0xa3/0x130 ipvlan_xmit_mode_l2+0xdb/0x330 [ipvlan] ipvlan_start_xmit+0x29/0xa0 [ipvlan] __dev_direct_xmit+0x2e2/0x380 paquete_direct_xmit+0x22/0x60 paquete_snd+0x7c9/0xc40 0x9a/0xa0 __sys_sendto+0x18a/0x230 __x64_sys_sendto+0x74/0x90 do_syscall_64 +0x3b/0x90 Entry_SYSCALL_64_after_hwframe+0x63/0xcd La causa principal es: 1. paquete_snd() solo restablece skb->mac_header cuando sock->tipo es SOCK_RAW y skb->protocolo no se especifica como en paquete_parse_headers() 2. paquete_direct_xmit() no restablece skb->mac_header como dev_queue_xmit() En este caso, skb->mac_header es 65535 cuando se llama a ipvlan_xmit_mode_l2(). Entonces, cuando ipvlan_xmit_mode_l2() obtiene el encabezado mac con eth_hdr() que usa \"skb->head + skb->mac_header\", se produce un acceso fuera de los l\u00edmites. Este parche reemplaza eth_hdr() con skb_eth_hdr() en ipvlan_xmit_mode_l2() y restablece el encabezado de mac en multidifusi\u00f3n para resolver este error fuera de los l\u00edmites."