CVE-2024-26984

{"id": "CVE-2024-26984", "metrics": {}, "published": "2024-05-01T06:15:15.880", "references": [{"url": "https://git.kernel.org/stable/c/13d76b2f443dc371842916dd8768009ff1594716", "source": "416baaa9-dc9f-4396-8d5f-8c081fb06d67"}, {"url": "https://git.kernel.org/stable/c/1bc4825d4c3ec6abe43cf06c3c39d664d044cbf7", "source": "416baaa9-dc9f-4396-8d5f-8c081fb06d67"}, {"url": "https://git.kernel.org/stable/c/21ca9539f09360fd83654f78f2c361f2f5ddcb52", "source": "416baaa9-dc9f-4396-8d5f-8c081fb06d67"}, {"url": "https://git.kernel.org/stable/c/3ab056814cd8ab84744c9a19ef51360b2271c572", "source": "416baaa9-dc9f-4396-8d5f-8c081fb06d67"}, {"url": "https://git.kernel.org/stable/c/a019b44b1bc6ed224c46fb5f88a8a10dd116e525", "source": "416baaa9-dc9f-4396-8d5f-8c081fb06d67"}, {"url": "https://git.kernel.org/stable/c/ad74d208f213c06d860916ad40f609ade8c13039", "source": "416baaa9-dc9f-4396-8d5f-8c081fb06d67"}, {"url": "https://git.kernel.org/stable/c/bba8ec5e9b16649d85bc9e9086bf7ae5b5716ff9", "source": "416baaa9-dc9f-4396-8d5f-8c081fb06d67"}, {"url": "https://git.kernel.org/stable/c/fff1386cc889d8fb4089d285f883f8cba62d82ce", "source": "416baaa9-dc9f-4396-8d5f-8c081fb06d67"}], "vulnStatus": "Awaiting Analysis", "descriptions": [{"lang": "en", "value": "In the Linux kernel, the following vulnerability has been resolved:\n\nnouveau: fix instmem race condition around ptr stores\n\nRunning a lot of VK CTS in parallel against nouveau, once every\nfew hours you might see something like this crash.\n\nBUG: kernel NULL pointer dereference, address: 0000000000000008\nPGD 8000000114e6e067 P4D 8000000114e6e067 PUD 109046067 PMD 0\nOops: 0000 [#1] PREEMPT SMP PTI\nCPU: 7 PID: 53891 Comm: deqp-vk Not tainted 6.8.0-rc6+ #27\nHardware name: Gigabyte Technology Co., Ltd. Z390 I AORUS PRO WIFI/Z390 I AORUS PRO WIFI-CF, BIOS F8 11/05/2021\nRIP: 0010:gp100_vmm_pgt_mem+0xe3/0x180 [nouveau]\nCode: c7 48 01 c8 49 89 45 58 85 d2 0f 84 95 00 00 00 41 0f b7 46 12 49 8b 7e 08 89 da 42 8d 2c f8 48 8b 47 08 41 83 c7 01 48 89 ee <48> 8b 40 08 ff d0 0f 1f 00 49 8b 7e 08 48 89 d9 48 8d 75 04 48 c1\nRSP: 0000:ffffac20c5857838 EFLAGS: 00010202\nRAX: 0000000000000000 RBX: 00000000004d8001 RCX: 0000000000000001\nRDX: 00000000004d8001 RSI: 00000000000006d8 RDI: ffffa07afe332180\nRBP: 00000000000006d8 R08: ffffac20c5857ad0 R09: 0000000000ffff10\nR10: 0000000000000001 R11: ffffa07af27e2de0 R12: 000000000000001c\nR13: ffffac20c5857ad0 R14: ffffa07a96fe9040 R15: 000000000000001c\nFS: 00007fe395eed7c0(0000) GS:ffffa07e2c980000(0000) knlGS:0000000000000000\nCS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033\nCR2: 0000000000000008 CR3: 000000011febe001 CR4: 00000000003706f0\nDR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000\nDR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400\nCall Trace:\n\n...\n\n ? gp100_vmm_pgt_mem+0xe3/0x180 [nouveau]\n ? gp100_vmm_pgt_mem+0x37/0x180 [nouveau]\n nvkm_vmm_iter+0x351/0xa20 [nouveau]\n ? __pfx_nvkm_vmm_ref_ptes+0x10/0x10 [nouveau]\n ? __pfx_gp100_vmm_pgt_mem+0x10/0x10 [nouveau]\n ? __pfx_gp100_vmm_pgt_mem+0x10/0x10 [nouveau]\n ? __lock_acquire+0x3ed/0x2170\n ? __pfx_gp100_vmm_pgt_mem+0x10/0x10 [nouveau]\n nvkm_vmm_ptes_get_map+0xc2/0x100 [nouveau]\n ? __pfx_nvkm_vmm_ref_ptes+0x10/0x10 [nouveau]\n ? __pfx_gp100_vmm_pgt_mem+0x10/0x10 [nouveau]\n nvkm_vmm_map_locked+0x224/0x3a0 [nouveau]\n\nAdding any sort of useful debug usually makes it go away, so I hand\nwrote the function in a line, and debugged the asm.\n\nEvery so often pt->memory->ptrs is NULL. This ptrs ptr is set in\nthe nv50_instobj_acquire called from nvkm_kmap.\n\nIf Thread A and Thread B both get to nv50_instobj_acquire around\nthe same time, and Thread A hits the refcount_set line, and in\nlockstep thread B succeeds at refcount_inc_not_zero, there is a\nchance the ptrs value won't have been stored since refcount_set\nis unordered. Force a memory barrier here, I picked smp_mb, since\nwe want it on all CPUs and it's write followed by a read.\n\nv2: use paired smp_rmb/smp_wmb."}, {"lang": "es", "value": "En el kernel de Linux, se resolvi\u00f3 la siguiente vulnerabilidad: nouveau: corrige la condici\u00f3n de ejecuci\u00f3n de instmem alrededor de las tiendas ptr Al ejecutar una gran cantidad de VK CTS en paralelo contra nouveau, una vez cada pocas horas es posible que vea algo como este bloqueo. ERROR: desreferencia del puntero NULL del kernel, direcci\u00f3n: 0000000000000008 PGD 8000000114e6e067 P4D 8000000114e6e067 PUD 109046067 PMD 0 Ups: 0000 [#1] PREEMPT SMP PTI CPU: 7 PID: 53891 Comm: deqp-vk No contaminado 6.8.0-rc6+ #27 Nombre del hardware : Gigabyte Technology Co., Ltd. Z390 I AORUS PRO WIFI/Z390 I AORUS PRO WIFI-CF, BIOS F8 05/11/2021 RIP: 0010:gp100_vmm_pgt_mem+0xe3/0x180 [nuevo] C\u00f3digo: c7 48 01 c8 49 89 45 58 85 d2 0f 84 95 00 00 00 41 0f b7 46 12 49 8b 7e 08 89 da 42 8d 2c f8 48 8b 47 08 41 83 c7 01 48 89 ee &lt;48&gt; 8b 40 08 ff d0 0f 1f 0 0 49 8b 7e 08 48 89 d9 48 8d 75 04 48 c1 RSP: 0000:ffffac20c5857838 EFLAGS: 00010202 RAX: 0000000000000000 RBX: 00000000004d8001 RCX: 0000000000000001 RDX: 00000000004d8001 RSI: 00000000000006d8 RDI: ffffa07afe332180 RBP: 00000000000006d8 R08: ffffac20c5857ad0 R09: 0000000000ffff10 R10: 000000001 R11: ffffa07af27e2de0 R12: 000000000000001c R13: ffffac20c5857ad0 R14: ffffa07a96fe9040 R15: 000000000000001c FS: 00007fe395eed7c0(0000) GS:ffffa07e2c980000(0000) nlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 0000000000000008 CR3: 000000011febe001 CR4: 00000000003706f0 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400 Rastreo de llamadas: ... ? gp100_vmm_pgt_mem+0xe3/0x180 [nuevo]? gp100_vmm_pgt_mem+0x37/0x180 [nuevo] nvkm_vmm_iter+0x351/0xa20 [nuevo] ? __pfx_nvkm_vmm_ref_ptes+0x10/0x10 [nuevo] ? __pfx_gp100_vmm_pgt_mem+0x10/0x10 [nuevo] ? __pfx_gp100_vmm_pgt_mem+0x10/0x10 [nuevo] ? __lock_acquire+0x3ed/0x2170? __pfx_gp100_vmm_pgt_mem+0x10/0x10 [nuevo] nvkm_vmm_ptes_get_map+0xc2/0x100 [nuevo] ? __pfx_nvkm_vmm_ref_ptes+0x10/0x10 [nuevo] ? __pfx_gp100_vmm_pgt_mem+0x10/0x10 [nuevo] nvkm_vmm_map_locked+0x224/0x3a0 [nuevo] Agregar cualquier tipo de depuraci\u00f3n \u00fatil generalmente hace que desaparezca, as\u00ed que escrib\u00ed la funci\u00f3n a mano en una l\u00ednea y depur\u00e9 el asm. De vez en cuando pt-&gt;memoria-&gt;ptrs es NULL. Este ptrs ptr se establece en nv50_instobj_acquire llamado desde nvkm_kmap. Si el subproceso A y el subproceso B llegan a nv50_instobj_acquire aproximadamente al mismo tiempo, y el subproceso A llega a la l\u00ednea refcount_set, y en paralelo el subproceso B tiene \u00e9xito en refcount_inc_not_zero, existe la posibilidad de que el valor de ptrs no se haya almacenado ya que refcount_set no est\u00e1 ordenado. Forzar una barrera de memoria aqu\u00ed, eleg\u00ed smp_mb, ya que lo queremos en todas las CPU y es escritura seguida de lectura. v2: use smp_rmb/smp_wmb emparejado."}], "lastModified": "2024-05-03T06:15:11.130", "sourceIdentifier": "416baaa9-dc9f-4396-8d5f-8c081fb06d67"}