fkie_cve-2024-58057
Vulnerability from fkie_nvd
Published
2025-03-06 16:15
Modified
2025-03-06 16:15
Severity ?
Summary
In the Linux kernel, the following vulnerability has been resolved:
idpf: convert workqueues to unbound
When a workqueue is created with `WQ_UNBOUND`, its work items are
served by special worker-pools, whose host workers are not bound to
any specific CPU. In the default configuration (i.e. when
`queue_delayed_work` and friends do not specify which CPU to run the
work item on), `WQ_UNBOUND` allows the work item to be executed on any
CPU in the same node of the CPU it was enqueued on. While this
solution potentially sacrifices locality, it avoids contention with
other processes that might dominate the CPU time of the processor the
work item was scheduled on.
This is not just a theoretical problem: in a particular scenario
misconfigured process was hogging most of the time from CPU0, leaving
less than 0.5% of its CPU time to the kworker. The IDPF workqueues
that were using the kworker on CPU0 suffered large completion delays
as a result, causing performance degradation, timeouts and eventual
system crash.
* I have also run a manual test to gauge the performance
improvement. The test consists of an antagonist process
(`./stress --cpu 2`) consuming as much of CPU 0 as possible. This
process is run under `taskset 01` to bind it to CPU0, and its
priority is changed with `chrt -pQ 9900 10000 ${pid}` and
`renice -n -20 ${pid}` after start.
Then, the IDPF driver is forced to prefer CPU0 by editing all calls
to `queue_delayed_work`, `mod_delayed_work`, etc... to use CPU 0.
Finally, `ktraces` for the workqueue events are collected.
Without the current patch, the antagonist process can force
arbitrary delays between `workqueue_queue_work` and
`workqueue_execute_start`, that in my tests were as high as
`30ms`. With the current patch applied, the workqueue can be
migrated to another unloaded CPU in the same node, and, keeping
everything else equal, the maximum delay I could see was `6us`.
References
Impacted products
| Vendor | Product | Version |
|---|
{
"cveTags": [],
"descriptions": [
{
"lang": "en",
"value": "In the Linux kernel, the following vulnerability has been resolved:\n\nidpf: convert workqueues to unbound\n\nWhen a workqueue is created with `WQ_UNBOUND`, its work items are\nserved by special worker-pools, whose host workers are not bound to\nany specific CPU. In the default configuration (i.e. when\n`queue_delayed_work` and friends do not specify which CPU to run the\nwork item on), `WQ_UNBOUND` allows the work item to be executed on any\nCPU in the same node of the CPU it was enqueued on. While this\nsolution potentially sacrifices locality, it avoids contention with\nother processes that might dominate the CPU time of the processor the\nwork item was scheduled on.\n\nThis is not just a theoretical problem: in a particular scenario\nmisconfigured process was hogging most of the time from CPU0, leaving\nless than 0.5% of its CPU time to the kworker. The IDPF workqueues\nthat were using the kworker on CPU0 suffered large completion delays\nas a result, causing performance degradation, timeouts and eventual\nsystem crash.\n\n\n* I have also run a manual test to gauge the performance\n improvement. The test consists of an antagonist process\n (`./stress --cpu 2`) consuming as much of CPU 0 as possible. This\n process is run under `taskset 01` to bind it to CPU0, and its\n priority is changed with `chrt -pQ 9900 10000 ${pid}` and\n `renice -n -20 ${pid}` after start.\n\n Then, the IDPF driver is forced to prefer CPU0 by editing all calls\n to `queue_delayed_work`, `mod_delayed_work`, etc... to use CPU 0.\n\n Finally, `ktraces` for the workqueue events are collected.\n\n Without the current patch, the antagonist process can force\n arbitrary delays between `workqueue_queue_work` and\n `workqueue_execute_start`, that in my tests were as high as\n `30ms`. With the current patch applied, the workqueue can be\n migrated to another unloaded CPU in the same node, and, keeping\n everything else equal, the maximum delay I could see was `6us`."
},
{
"lang": "es",
"value": "En el kernel de Linux, se ha resuelto la siguiente vulnerabilidad: idpf: convertir colas de trabajo en no vinculadas Cuando se crea una cola de trabajo con `WQ_UNBOUND`, sus elementos de trabajo son atendidos por grupos de trabajadores especiales, cuyos trabajadores host no est\u00e1n vinculados a ninguna CPU espec\u00edfica. En la configuraci\u00f3n predeterminada (es decir, cuando `queue_delayed_work` y amigos no especifican en qu\u00e9 CPU ejecutar el elemento de trabajo), `WQ_UNBOUND` permite que el elemento de trabajo se ejecute en cualquier CPU en el mismo nodo de la CPU en la que se puso en cola. Si bien esta soluci\u00f3n potencialmente sacrifica la localidad, evita la contenci\u00f3n con otros procesos que podr\u00edan dominar el tiempo de CPU del procesador en el que se program\u00f3 el elemento de trabajo. Este no es solo un problema te\u00f3rico: en un escenario particular, el proceso mal configurado acaparaba la mayor parte del tiempo de la CPU0, dejando menos del 0,5% de su tiempo de CPU al kworker. Las colas de trabajo IDPF que estaban usando el kworker en CPU0 sufrieron grandes retrasos en la finalizaci\u00f3n como resultado, causando degradaci\u00f3n del rendimiento, tiempos de espera y eventualmente falla del sistema. * Tambi\u00e9n he ejecutado una prueba manual para medir la mejora del rendimiento. La prueba consiste en un proceso antagonista (`./stress --cpu 2`) que consume la mayor cantidad posible de CPU 0. Este proceso se ejecuta bajo `taskset 01` para vincularlo a CPU0, y su prioridad se cambia con `chrt -pQ 9900 10000 ${pid}` y `renice -n -20 ${pid}` despu\u00e9s del inicio. Luego, el controlador IDPF se ve obligado a preferir CPU0 editando todas las llamadas a `queue_delayed_work`, `mod_delayed_work`, etc... para usar CPU 0. Finalmente, se recopilan `ktraces` para los eventos de la cola de trabajo. Sin el parche actual, el proceso antagonista puede forzar demoras arbitrarias entre `workqueue_queue_work` y `workqueue_execute_start`, que en mis pruebas fueron tan altas como `30ms`. Con el parche actual aplicado, la cola de trabajo se puede migrar a otra CPU sin carga en el mismo nodo y, manteniendo todo lo dem\u00e1s igual, la demora m\u00e1xima que pude ver fue `6us`."
}
],
"id": "CVE-2024-58057",
"lastModified": "2025-03-06T16:15:51.940",
"metrics": {},
"published": "2025-03-06T16:15:51.940",
"references": [
{
"source": "416baaa9-dc9f-4396-8d5f-8c081fb06d67",
"url": "https://git.kernel.org/stable/c/66bf9b3d9e1658333741f075320dc8e7cd6f8d09"
},
{
"source": "416baaa9-dc9f-4396-8d5f-8c081fb06d67",
"url": "https://git.kernel.org/stable/c/868202ec3854e13de1164e4a3e25521194c5af72"
},
{
"source": "416baaa9-dc9f-4396-8d5f-8c081fb06d67",
"url": "https://git.kernel.org/stable/c/9a5b021cb8186f1854bac2812bd4f396bb1e881c"
}
],
"sourceIdentifier": "416baaa9-dc9f-4396-8d5f-8c081fb06d67",
"vulnStatus": "Awaiting Analysis"
}
Loading…
Loading…
Sightings
| Author | Source | Type | Date |
|---|
Nomenclature
- Seen: The vulnerability was mentioned, discussed, or seen somewhere by the user.
- Confirmed: The vulnerability is confirmed from an analyst perspective.
- Exploited: This vulnerability was exploited and seen by the user reporting the sighting.
- Patched: This vulnerability was successfully patched by the user reporting the sighting.
- Not exploited: This vulnerability was not exploited or seen by the user reporting the sighting.
- Not confirmed: The user expresses doubt about the veracity of the vulnerability.
- Not patched: This vulnerability was not successfully patched by the user reporting the sighting.
Loading…
Loading…