CVE-2024-42305

{"id": "CVE-2024-42305", "cveTags": [], "metrics": {}, "published": "2024-08-17T09:15:10.690", "references": [{"url": "https://git.kernel.org/stable/c/19e13b4d7f0303186fcc891aba8d0de7c8fdbda8", "source": "416baaa9-dc9f-4396-8d5f-8c081fb06d67"}, {"url": "https://git.kernel.org/stable/c/42d420517072028fb0eb852c358056b7717ba5aa", "source": "416baaa9-dc9f-4396-8d5f-8c081fb06d67"}, {"url": "https://git.kernel.org/stable/c/50ea741def587a64e08879ce6c6a30131f7111e7", "source": "416baaa9-dc9f-4396-8d5f-8c081fb06d67"}, {"url": "https://git.kernel.org/stable/c/8afe06ed3be7a874b3cd82ef5f8959aca8d6429a", "source": "416baaa9-dc9f-4396-8d5f-8c081fb06d67"}, {"url": "https://git.kernel.org/stable/c/9d241b7a39af192d1bb422714a458982c7cc67a2", "source": "416baaa9-dc9f-4396-8d5f-8c081fb06d67"}, {"url": "https://git.kernel.org/stable/c/abb411ac991810c0bcbe51c2e76d2502bf611b5c", "source": "416baaa9-dc9f-4396-8d5f-8c081fb06d67"}, {"url": "https://git.kernel.org/stable/c/b80575ffa98b5bb3a5d4d392bfe4c2e03e9557db", "source": "416baaa9-dc9f-4396-8d5f-8c081fb06d67"}, {"url": "https://git.kernel.org/stable/c/cdd345321699042ece4a9d2e70754d2397d378c5", "source": "416baaa9-dc9f-4396-8d5f-8c081fb06d67"}], "vulnStatus": "Awaiting Analysis", "descriptions": [{"lang": "en", "value": "In the Linux kernel, the following vulnerability has been resolved:\n\next4: check dot and dotdot of dx_root before making dir indexed\n\nSyzbot reports a issue as follows:\n============================================\nBUG: unable to handle page fault for address: ffffed11022e24fe\nPGD 23ffee067 P4D 23ffee067 PUD 0\nOops: Oops: 0000 [#1] PREEMPT SMP KASAN PTI\nCPU: 0 PID: 5079 Comm: syz-executor306 Not tainted 6.10.0-rc5-g55027e689933 #0\nCall Trace:\n <TASK>\n make_indexed_dir+0xdaf/0x13c0 fs/ext4/namei.c:2341\n ext4_add_entry+0x222a/0x25d0 fs/ext4/namei.c:2451\n ext4_rename fs/ext4/namei.c:3936 [inline]\n ext4_rename2+0x26e5/0x4370 fs/ext4/namei.c:4214\n[...]\n============================================\n\nThe immediate cause of this problem is that there is only one valid dentry\nfor the block to be split during do_split, so split==0 results in out of\nbounds accesses to the map triggering the issue.\n\n do_split\n unsigned split\n dx_make_map\n count = 1\n split = count/2 = 0;\n continued = hash2 == map[split - 1].hash;\n ---> map[4294967295]\n\nThe maximum length of a filename is 255 and the minimum block size is 1024,\nso it is always guaranteed that the number of entries is greater than or\nequal to 2 when do_split() is called.\n\nBut syzbot's crafted image has no dot and dotdot in dir, and the dentry\ndistribution in dirblock is as follows:\n\n bus dentry1 hole dentry2 free\n|xx--|xx-------------|...............|xx-------------|...............|\n0 12 (8+248)=256 268 256 524 (8+256)=264 788 236 1024\n\nSo when renaming dentry1 increases its name_len length by 1, neither hole\nnor free is sufficient to hold the new dentry, and make_indexed_dir() is\ncalled.\n\nIn make_indexed_dir() it is assumed that the first two entries of the\ndirblock must be dot and dotdot, so bus and dentry1 are left in dx_root\nbecause they are treated as dot and dotdot, and only dentry2 is moved\nto the new leaf block. That's why count is equal to 1.\n\nTherefore add the ext4_check_dx_root() helper function to add more sanity\nchecks to dot and dotdot before starting the conversion to avoid the above\nissue."}, {"lang": "es", "value": "En el kernel de Linux, se ha resuelto la siguiente vulnerabilidad: ext4: verifique el punto y el punto de dx_root antes de indexar el directorio Syzbot informa un problema de la siguiente manera: =================== ========================= ERROR: no se puede manejar el error de p\u00e1gina para la direcci\u00f3n: ffffed11022e24fe PGD 23ffee067 P4D 23ffee067 PUD 0 Ups: Ups: 0000 [#1 ] PREEMPT SMP KASAN PTI CPU: 0 PID: 5079 Comm: syz-executor306 No contaminado 6.10.0-rc5-g55027e689933 #0 Seguimiento de llamada: make_indexed_dir+0xdaf/0x13c0 fs/ext4/namei.c:2341 ext4_add_entry+0x222 un /0x25d0 fs/ext4/namei.c:2451 ext4_rename fs/ext4/namei.c:3936 [en l\u00ednea] text4_rename2+0x26e5/0x4370 fs/ext4/namei.c:4214 [...] ======= ====================================== La causa inmediata de este problema es que s\u00f3lo hay una v\u00e1lida dentry para que el bloque se divida durante do_split, por lo que split==0 da como resultado accesos fuera de los l\u00edmites al mapa que desencadenan el problema. do_split divisi\u00f3n sin signo dx_make_map recuento = 1 divisi\u00f3n = recuento/2 = 0; contin\u00faa = hash2 == mapa[dividido - 1].hash; ---&gt; map[4294967295] La longitud m\u00e1xima de un nombre de archivo es 255 y el tama\u00f1o m\u00ednimo de bloque es 1024, por lo que siempre se garantiza que el n\u00famero de entradas sea mayor o igual a 2 cuando se llama a do_split(). Pero la imagen manipulada por syzbot no tiene punto ni puntopunto en el directorio, y la distribuci\u00f3n de dentry en dirblock es la siguiente: bus dentry1 agujero dentry2 gratis |xx--|xx-------------|... ............|xx-------------|...............| 0 12 (8+248)=256 268 256 524 (8+256)=264 788 236 1024 Entonces, cuando se cambia el nombre de dentry1 se aumenta la longitud de name_len en 1, ni el agujero ni el espacio libre son suficientes para contener el nuevo dentry, y make_indexed_dir() es llamado. En make_indexed_dir() se supone que las dos primeras entradas del bloque de directorios deben ser punto y puntopunto, por lo que bus y dentry1 se dejan en dx_root porque se tratan como punto y puntodot, y solo dentry2 se mueve al nuevo bloque de hoja. Es por eso que el recuento es igual a 1. Por lo tanto, agregue la funci\u00f3n auxiliar ext4_check_dx_root() para agregar m\u00e1s controles de cordura a los puntos y puntos antes de comenzar la conversi\u00f3n para evitar el problema anterior."}], "lastModified": "2024-08-19T12:59:59.177", "sourceIdentifier": "416baaa9-dc9f-4396-8d5f-8c081fb06d67"}