nvd-json-data-feeds/CVE-2024/CVE-2024-266xx/CVE-2024-26687.json

{
  "id": "CVE-2024-26687",
  "sourceIdentifier": "416baaa9-dc9f-4396-8d5f-8c081fb06d67",
  "published": "2024-04-03T15:15:52.313",
  "lastModified": "2024-06-25T22:15:20.917",
  "vulnStatus": "Awaiting Analysis",
  "cveTags": [],
  "descriptions": [
    {
      "lang": "en",
      "value": "In the Linux kernel, the following vulnerability has been resolved:\n\nxen/events: close evtchn after mapping cleanup\n\nshutdown_pirq and startup_pirq are not taking the\nirq_mapping_update_lock because they can't due to lock inversion. Both\nare called with the irq_desc->lock being taking. The lock order,\nhowever, is first irq_mapping_update_lock and then irq_desc->lock.\n\nThis opens multiple races:\n- shutdown_pirq can be interrupted by a function that allocates an event\n  channel:\n\n  CPU0                        CPU1\n  shutdown_pirq {\n    xen_evtchn_close(e)\n                              __startup_pirq {\n                                EVTCHNOP_bind_pirq\n                                  -> returns just freed evtchn e\n                                set_evtchn_to_irq(e, irq)\n                              }\n    xen_irq_info_cleanup() {\n      set_evtchn_to_irq(e, -1)\n    }\n  }\n\n  Assume here event channel e refers here to the same event channel\n  number.\n  After this race the evtchn_to_irq mapping for e is invalid (-1).\n\n- __startup_pirq races with __unbind_from_irq in a similar way. Because\n  __startup_pirq doesn't take irq_mapping_update_lock it can grab the\n  evtchn that __unbind_from_irq is currently freeing and cleaning up. In\n  this case even though the event channel is allocated, its mapping can\n  be unset in evtchn_to_irq.\n\nThe fix is to first cleanup the mappings and then close the event\nchannel. In this way, when an event channel gets allocated it's\npotential previous evtchn_to_irq mappings are guaranteed to be unset already.\nThis is also the reverse order of the allocation where first the event\nchannel is allocated and then the mappings are setup.\n\nOn a 5.10 kernel prior to commit 3fcdaf3d7634 (\"xen/events: modify internal\n[un]bind interfaces\"), we hit a BUG like the following during probing of NVMe\ndevices. The issue is that during nvme_setup_io_queues, pci_free_irq\nis called for every device which results in a call to shutdown_pirq.\nWith many nvme devices it's therefore likely to hit this race during\nboot because there will be multiple calls to shutdown_pirq and\nstartup_pirq are running potentially in parallel.\n\n  ------------[ cut here ]------------\n  blkfront: xvda: barrier or flush: disabled; persistent grants: enabled; indirect descriptors: enabled; bounce buffer: enabled\n  kernel BUG at drivers/xen/events/events_base.c:499!\n  invalid opcode: 0000 [#1] SMP PTI\n  CPU: 44 PID: 375 Comm: kworker/u257:23 Not tainted 5.10.201-191.748.amzn2.x86_64 #1\n  Hardware name: Xen HVM domU, BIOS 4.11.amazon 08/24/2006\n  Workqueue: nvme-reset-wq nvme_reset_work\n  RIP: 0010:bind_evtchn_to_cpu+0xdf/0xf0\n  Code: 5d 41 5e c3 cc cc cc cc 44 89 f7 e8 2b 55 ad ff 49 89 c5 48 85 c0 0f 84 64 ff ff ff 4c 8b 68 30 41 83 fe ff 0f 85 60 ff ff ff <0f> 0b 66 66 2e 0f 1f 84 00 00 00 00 00 0f 1f 40 00 0f 1f 44 00 00\n  RSP: 0000:ffffc9000d533b08 EFLAGS: 00010046\n  RAX: 0000000000000000 RBX: 0000000000000000 RCX: 0000000000000006\n  RDX: 0000000000000028 RSI: 00000000ffffffff RDI: 00000000ffffffff\n  RBP: ffff888107419680 R08: 0000000000000000 R09: ffffffff82d72b00\n  R10: 0000000000000000 R11: 0000000000000000 R12: 00000000000001ed\n  R13: 0000000000000000 R14: 00000000ffffffff R15: 0000000000000002\n  FS:  0000000000000000(0000) GS:ffff88bc8b500000(0000) knlGS:0000000000000000\n  CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033\n  CR2: 0000000000000000 CR3: 0000000002610001 CR4: 00000000001706e0\n  DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000\n  DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400\n  Call Trace:\n   ? show_trace_log_lvl+0x1c1/0x2d9\n   ? show_trace_log_lvl+0x1c1/0x2d9\n   ? set_affinity_irq+0xdc/0x1c0\n   ? __die_body.cold+0x8/0xd\n   ? die+0x2b/0x50\n   ? do_trap+0x90/0x110\n   ? bind_evtchn_to_cpu+0xdf/0xf0\n   ? do_error_trap+0x65/0x80\n   ? bind_evtchn_to_cpu+0xdf/0xf0\n   ? exc_invalid_op+0x4e/0x70\n   ? bind_evtchn_to_cpu+0xdf/0xf0\n   ? asm_exc_invalid_op+0x12/0x20\n   ? bind_evtchn_to_cpu+0xdf/0x\n---truncat
    },
    {
      "lang": "es",
      "value": "En el kernel de Linux, se resolvi\u00f3 la siguiente vulnerabilidad: xen/events: cierre evtchn despu\u00e9s de la limpieza del mapeo. Shutdown_pirq y startup_pirq no est\u00e1n tomando irq_mapping_update_lock porque no pueden hacerlo debido a la inversi\u00f3n de bloqueo. Ambos se llaman con el irq_desc-&gt;lock en curso. Sin embargo, el orden de bloqueo es primero irq_mapping_update_lock y luego irq_desc-&gt;lock. Esto abre m\u00faltiples ejecuci\u00f3ns: - Shutdown_pirq puede ser interrumpido por una funci\u00f3n que asigna un canal de evento: CPU0 CPU1 Shutdown_pirq { xen_evtchn_close(e) __startup_pirq { EVTCHNOP_bind_pirq -&gt; devuelve el evtchn e set_evtchn_to_irq(e, irq) reci\u00e9n liberado} xen_irq_info_cleanup() { set_evtchn_to_irq( e, -1) } } Supongamos que aqu\u00ed el canal de eventos e se refiere aqu\u00ed al mismo n\u00famero de canal de eventos. Despu\u00e9s de esta ejecuci\u00f3n, el mapeo evtchn_to_irq para e no es v\u00e1lido (-1). - __startup_pirq compite con __unbind_from_irq de manera similar. Debido a que __startup_pirq no toma irq_mapping_update_lock, puede tomar el evtchn que __unbind_from_irq est\u00e1 liberando y limpiando actualmente. En este caso, aunque el canal de eventos est\u00e9 asignado, su asignaci\u00f3n se puede desarmar en evtchn_to_irq. La soluci\u00f3n es limpiar primero las asignaciones y luego cerrar el canal de eventos. De esta manera, cuando se asigna un canal de eventos, se garantiza que sus posibles asignaciones anteriores de evtchn_to_irq ya no estar\u00e1n configuradas. Este es tambi\u00e9n el orden inverso de la asignaci\u00f3n donde primero se asigna el canal de evento y luego se configuran las asignaciones. En un kernel 5.10 antes de commit 3fcdaf3d7634 (\"xen/events: modificar interfaces internas [un]bind\"), encontramos un ERROR como el siguiente durante el sondeo de dispositivos NVMe. El problema es que durante nvme_setup_io_queues, se llama a pci_free_irq para cada dispositivo, lo que resulta en una llamada a Shutdown_pirq. Por lo tanto, con muchos dispositivos nvme es probable que alcance esta ejecuci\u00f3n durante el arranque porque habr\u00e1 m\u00faltiples llamadas a Shutdown_pirq y startup_pirq que se ejecutan potencialmente en paralelo. ------------[ cortar aqu\u00ed ]------------ blkfront: xvda: barrera o descarga: deshabilitado; subvenciones persistentes: habilitadas; descriptores indirectos: habilitado; b\u00fafer de rebote: \u00a1ERROR del kernel habilitado en drivers/xen/events/events_base.c:499! c\u00f3digo de operaci\u00f3n no v\u00e1lido: 0000 [#1] SMP PTI CPU: 44 PID: 375 Comm: kworker/u257:23 No contaminado 5.10.201-191.748.amzn2.x86_64 #1 Nombre de hardware: Xen HVM domU, BIOS 4.11.amazon 24/08 /2006 Cola de trabajo: nvme-reset-wq nvme_reset_work RIP: 0010:bind_evtchn_to_cpu+0xdf/0xf0 C\u00f3digo: 5d 41 5e c3 cc cc cc cc 44 89 f7 e8 2b 55 ad ff 49 89 c5 48 85 c0 0f 84 64 ff ff ff 4c 8b 68 30 41 83 fe ff 0f 85 60 ff ff ff &lt;0f&gt; 0b 66 66 2e 0f 1f 84 00 00 00 00 00 0f 1f 40 00 0f 1f 44 00 00 RSP: 0000:ffffc9000d533b08 EFLAGS: 00010046 RAX: 0000000000000000 RBX: 0000000000000000 RCX: 0000000000000006 RDX: 0000000000000028 RSI: 00000000ffffffff RDI: 00000000ffffffff RBP: ffff888107419680 R08: 00000000000 00000 R09: ffffffff82d72b00 R10: 00000000000000000 R11: 0000000000000000 R12: 00000000000001ed R13: 00000000000000000 R14: 00000000ffffffff R15 : 0000000000000002 FS: 0000000000000000(0000) GS:ffff88bc8b500000( 0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 00000000000000000 CR3: 0000000002610001 CR4: 000000000017 06e0 DR0: 0000000000000000 DR1: 00000000000000000 DR2: 0000000000000000 DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400 Seguimiento de llamadas: ? show_trace_log_lvl+0x1c1/0x2d9? show_trace_log_lvl+0x1c1/0x2d9? set_affinity_irq+0xdc/0x1c0? __die_body.cold+0x8/0xd ? morir+0x2b/0x50 ? do_trap+0x90/0x110? bind_evtchn_to_cpu+0xdf/0xf0? do_error_trap+0x65/0x80? bind_evtchn_to_cpu+0xdf/0xf0? exc_invalid_op+0x4e/0x70? bind_evtchn_to_cpu+0xdf/0xf0? asm_exc_invalid_op+0x12/0x20
    }
  ],
  "metrics": {},
  "references": [
    {
      "url": "https://git.kernel.org/stable/c/0fc88aeb2e32b76db3fe6a624b8333dbe621b8fd",
      "source": "416baaa9-dc9f-4396-8d5f-8c081fb06d67"
    },
    {
      "url": "https://git.kernel.org/stable/c/20980195ec8d2e41653800c45c8c367fa1b1f2b4",
      "source": "416baaa9-dc9f-4396-8d5f-8c081fb06d67"
    },
    {
      "url": "https://git.kernel.org/stable/c/585a344af6bcac222608a158fc2830ff02712af5",
      "source": "416baaa9-dc9f-4396-8d5f-8c081fb06d67"
    },
    {
      "url": "https://git.kernel.org/stable/c/9470f5b2503cae994098dea9682aee15b313fa44",
      "source": "416baaa9-dc9f-4396-8d5f-8c081fb06d67"
    },
    {
      "url": "https://git.kernel.org/stable/c/9be71aa12afa91dfe457b3fb4a444c42b1ee036b",
      "source": "416baaa9-dc9f-4396-8d5f-8c081fb06d67"
    },
    {
      "url": "https://git.kernel.org/stable/c/ea592baf9e41779fe9a0424c03dd2f324feca3b3",
      "source": "416baaa9-dc9f-4396-8d5f-8c081fb06d67"
    },
    {
      "url": "https://git.kernel.org/stable/c/fa765c4b4aed2d64266b694520ecb025c862c5a9",
      "source": "416baaa9-dc9f-4396-8d5f-8c081fb06d67"
    },
    {
      "url": "https://lists.debian.org/debian-lts-announce/2024/06/msg00017.html",
      "source": "416baaa9-dc9f-4396-8d5f-8c081fb06d67"
    }
  ]
}