nvd-json-data-feeds/CVE-2024/CVE-2024-270xx/CVE-2024-27004.json

{
  "id": "CVE-2024-27004",
  "sourceIdentifier": "416baaa9-dc9f-4396-8d5f-8c081fb06d67",
  "published": "2024-05-01T06:15:18.763",
  "lastModified": "2024-06-25T22:15:27.890",
  "vulnStatus": "Awaiting Analysis",
  "cveTags": [],
  "descriptions": [
    {
      "lang": "en",
      "value": "In the Linux kernel, the following vulnerability has been resolved:\n\nclk: Get runtime PM before walking tree during disable_unused\n\nDoug reported [1] the following hung task:\n\n INFO: task swapper/0:1 blocked for more than 122 seconds.\n       Not tainted 5.15.149-21875-gf795ebc40eb8 #1\n \"echo 0 > /proc/sys/kernel/hung_task_timeout_secs\" disables this message.\n task:swapper/0       state:D stack:    0 pid:    1 ppid:     0 flags:0x00000008\n Call trace:\n  __switch_to+0xf4/0x1f4\n  __schedule+0x418/0xb80\n  schedule+0x5c/0x10c\n  rpm_resume+0xe0/0x52c\n  rpm_resume+0x178/0x52c\n  __pm_runtime_resume+0x58/0x98\n  clk_pm_runtime_get+0x30/0xb0\n  clk_disable_unused_subtree+0x58/0x208\n  clk_disable_unused_subtree+0x38/0x208\n  clk_disable_unused_subtree+0x38/0x208\n  clk_disable_unused_subtree+0x38/0x208\n  clk_disable_unused_subtree+0x38/0x208\n  clk_disable_unused+0x4c/0xe4\n  do_one_initcall+0xcc/0x2d8\n  do_initcall_level+0xa4/0x148\n  do_initcalls+0x5c/0x9c\n  do_basic_setup+0x24/0x30\n  kernel_init_freeable+0xec/0x164\n  kernel_init+0x28/0x120\n  ret_from_fork+0x10/0x20\n INFO: task kworker/u16:0:9 blocked for more than 122 seconds.\n       Not tainted 5.15.149-21875-gf795ebc40eb8 #1\n \"echo 0 > /proc/sys/kernel/hung_task_timeout_secs\" disables this message.\n task:kworker/u16:0   state:D stack:    0 pid:    9 ppid:     2 flags:0x00000008\n Workqueue: events_unbound deferred_probe_work_func\n Call trace:\n  __switch_to+0xf4/0x1f4\n  __schedule+0x418/0xb80\n  schedule+0x5c/0x10c\n  schedule_preempt_disabled+0x2c/0x48\n  __mutex_lock+0x238/0x488\n  __mutex_lock_slowpath+0x1c/0x28\n  mutex_lock+0x50/0x74\n  clk_prepare_lock+0x7c/0x9c\n  clk_core_prepare_lock+0x20/0x44\n  clk_prepare+0x24/0x30\n  clk_bulk_prepare+0x40/0xb0\n  mdss_runtime_resume+0x54/0x1c8\n  pm_generic_runtime_resume+0x30/0x44\n  __genpd_runtime_resume+0x68/0x7c\n  genpd_runtime_resume+0x108/0x1f4\n  __rpm_callback+0x84/0x144\n  rpm_callback+0x30/0x88\n  rpm_resume+0x1f4/0x52c\n  rpm_resume+0x178/0x52c\n  __pm_runtime_resume+0x58/0x98\n  __device_attach+0xe0/0x170\n  device_initial_probe+0x1c/0x28\n  bus_probe_device+0x3c/0x9c\n  device_add+0x644/0x814\n  mipi_dsi_device_register_full+0xe4/0x170\n  devm_mipi_dsi_device_register_full+0x28/0x70\n  ti_sn_bridge_probe+0x1dc/0x2c0\n  auxiliary_bus_probe+0x4c/0x94\n  really_probe+0xcc/0x2c8\n  __driver_probe_device+0xa8/0x130\n  driver_probe_device+0x48/0x110\n  __device_attach_driver+0xa4/0xcc\n  bus_for_each_drv+0x8c/0xd8\n  __device_attach+0xf8/0x170\n  device_initial_probe+0x1c/0x28\n  bus_probe_device+0x3c/0x9c\n  deferred_probe_work_func+0x9c/0xd8\n  process_one_work+0x148/0x518\n  worker_thread+0x138/0x350\n  kthread+0x138/0x1e0\n  ret_from_fork+0x10/0x20\n\nThe first thread is walking the clk tree and calling\nclk_pm_runtime_get() to power on devices required to read the clk\nhardware via struct clk_ops::is_enabled(). This thread holds the clk\nprepare_lock, and is trying to runtime PM resume a device, when it finds\nthat the device is in the process of resuming so the thread schedule()s\naway waiting for the device to finish resuming before continuing. The\nsecond thread is runtime PM resuming the same device, but the runtime\nresume callback is calling clk_prepare(), trying to grab the\nprepare_lock waiting on the first thread.\n\nThis is a classic ABBA deadlock. To properly fix the deadlock, we must\nnever runtime PM resume or suspend a device with the clk prepare_lock\nheld. Actually doing that is near impossible today because the global\nprepare_lock would have to be dropped in the middle of the tree, the\ndevice runtime PM resumed/suspended, and then the prepare_lock grabbed\nagain to ensure consistency of the clk tree topology. If anything\nchanges with the clk tree in the meantime, we've lost and will need to\nstart the operation all over again.\n\nLuckily, most of the time we're simply incrementing or decrementing the\nruntime PM count on an active device, so we don't have the chance to\nschedule away with the prepare_lock held. Let's fix this immediate\nproblem tha
    },
    {
      "lang": "es",
      "value": "En el kernel de Linux, se resolvi\u00f3 la siguiente vulnerabilidad: clk: Obtener PM en tiempo de ejecuci\u00f3n antes de caminar por el \u00e1rbol durante enable_unused Doug inform\u00f3 [1] la siguiente tarea colgada: INFORMACI\u00d3N: intercambio de tareas/0:1 bloqueado durante m\u00e1s de 122 segundos. No contaminado 5.15.149-21875-gf795ebc40eb8 #1 \"echo 0 > /proc/sys/kernel/hung_task_timeout_secs\" desactiva este mensaje. tarea:swapper/0 estado:D pila: 0 pid: 1 ppid: 0 banderas:0x00000008 Rastreo de llamadas: __switch_to+0xf4/0x1f4 __schedule+0x418/0xb80 Schedule+0x5c/0x10c rpm_resume+0xe0/0x52c rpm_resume+0x178/0x52c __pm_run tiempo_resume+ 0x58/0x98 clk_pm_runtime_get+0x30/0xb0 clk_disable_unused_subtree+0x58/0x208 clk_disable_unused_subtree+0x38/0x208 clk_disable_unused_subtree+0x38/0x208 clk_disable_unused_subtree+0x38/0x208 _disable_unused_subtree+0x38/0x208 clk_disable_unused+0x4c/0xe4 do_one_initcall+0xcc/0x2d8 do_initcall_level+0xa4/0x148 do_initcalls+ 0x5c/0x9c do_basic_setup+0x24/0x30 kernel_init_freeable+0xec/0x164 kernel_init+0x28/0x120 ret_from_fork+0x10/0x20 INFORMACI\u00d3N: tarea kworker/u16:0:9 bloqueada durante m\u00e1s de 122 segundos. No contaminado 5.15.149-21875-gf795ebc40eb8 #1 \"echo 0 > /proc/sys/kernel/hung_task_timeout_secs\" desactiva este mensaje. tarea:kworker/u16:0 estado:D pila: 0 pid: 9 ppid: 2 banderas:0x00000008 Cola de trabajo: events_unbound deferred_probe_work_func Rastreo de llamadas: __switch_to+0xf4/0x1f4 __schedule+0x418/0xb80 Schedule+0x5c/0x10c Schedule_preempt_disabled+0x2c /0x48 __mutex_lock +0x238/0x488 __mutex_lock_slowpath+0x1c/0x28 mutex_lock+0x50/0x74 clk_prepare_lock+0x7c/0x9c clk_core_prepare_lock+0x20/0x44 clk_prepare+0x24/0x30 clk_bulk_prepare+0x40/0xb0 curr\u00edculum+0x54/0x1c8 pm_generic_runtime_resume+0x30/0x44 __genpd_runtime_resume+0x68/0x7c genpd_runtime_resume +0x108/0x1f4 __rpm_callback+0x84/0x144 rpm_callback+0x30/0x88 rpm_resume+0x1f4/0x52c rpm_resume+0x178/0x52c __pm_runtime_resume+0x58/0x98 __device_attach+0xe0/0x170 dispositivo_initial_probe+0x 1c/0x28 bus_probe_device+0x3c/0x9c dispositivo_add+0x644/0x814 mipi_dsi_device_register_full +0xe4/0x170 devm_mipi_dsi_device_register_full+0x28/0x70 ti_sn_bridge_probe+0x1dc/0x2c0 auxiliar_bus_probe+0x4c/0x94 very_probe+0xcc/0x2c8 __driver_probe_device+0xa8/0x130 driver_probe_device+0x48/ 0x110 __device_attach_driver+0xa4/0xcc bus_for_each_drv+0x8c/0xd8 __device_attach+0xf8/0x170 dispositivo_inicial_probe +0x1c/0x28 bus_probe_device+0x3c/0x9c deferred_probe_work_func+0x9c/0xd8 Process_one_work+0x148/0x518 Workers_thread+0x138/0x350 kthread+0x138/0x1e0 ret_from_fork+0x10/0x20 El primer hilo est\u00e1 recorriendo el \u00e1rbol clk y llamando clk_pm_runtime_get() para encender dispositivos necesarios para leer el hardware clk a trav\u00e9s de struct clk_ops::is_enabled(). Este hilo contiene clk prepare_lock y est\u00e1 intentando ejecutar PM para reanudar un dispositivo, cuando descubre que el dispositivo est\u00e1 en proceso de reanudaci\u00f3n, por lo que la programaci\u00f3n del hilo est\u00e1 esperando a que el dispositivo termine de reanudarse antes de continuar. El segundo hilo es PM en tiempo de ejecuci\u00f3n que reanuda el mismo dispositivo, pero la devoluci\u00f3n de llamada de reanudaci\u00f3n en tiempo de ejecuci\u00f3n llama a clk_prepare(), intentando capturar el prepare_lock que espera en el primer hilo. Este es un cl\u00e1sico punto muerto de ABBA. Para solucionar correctamente el punto muerto, nunca debemos reanudar el PM en tiempo de ejecuci\u00f3n ni suspender un dispositivo con clk prepare_lock retenido. En realidad, hacer eso es casi imposible hoy en d\u00eda porque el prepare_lock global tendr\u00eda que colocarse en el medio del \u00e1rbol, el tiempo de ejecuci\u00f3n del dispositivo PM se reanudar\u00eda/suspendir\u00eda y luego el prepare_lock se tomar\u00eda nuevamente para garantizar la coherencia de la topolog\u00eda del \u00e1rbol clk. Mientras tanto, si algo cambia con el \u00e1rbol clk, habremos perdido y necesitaremos comenzar la operaci\u00f3n de nuevo. 
    }
  ],
  "metrics": {},
  "references": [
    {
      "url": "https://git.kernel.org/stable/c/115554862294397590088ba02f11f2aba6d5016c",
      "source": "416baaa9-dc9f-4396-8d5f-8c081fb06d67"
    },
    {
      "url": "https://git.kernel.org/stable/c/253ab38d1ee652a596942156978a233970d185ba",
      "source": "416baaa9-dc9f-4396-8d5f-8c081fb06d67"
    },
    {
      "url": "https://git.kernel.org/stable/c/4af115f1a20a3d9093586079206ee37c2ac55123",
      "source": "416baaa9-dc9f-4396-8d5f-8c081fb06d67"
    },
    {
      "url": "https://git.kernel.org/stable/c/60ff482c4205a5aac3b0595ab794cfd62295dab5",
      "source": "416baaa9-dc9f-4396-8d5f-8c081fb06d67"
    },
    {
      "url": "https://git.kernel.org/stable/c/a29ec0465dce0b871003698698ac6fa92c9a5034",
      "source": "416baaa9-dc9f-4396-8d5f-8c081fb06d67"
    },
    {
      "url": "https://git.kernel.org/stable/c/a424e713e0cc33d4b969cfda25b9f46df4d7b5bc",
      "source": "416baaa9-dc9f-4396-8d5f-8c081fb06d67"
    },
    {
      "url": "https://git.kernel.org/stable/c/e581cf5d216289ef292d1a4036d53ce90e122469",
      "source": "416baaa9-dc9f-4396-8d5f-8c081fb06d67"
    },
    {
      "url": "https://lists.debian.org/debian-lts-announce/2024/06/msg00017.html",
      "source": "416baaa9-dc9f-4396-8d5f-8c081fb06d67"
    }
  ]
}