// For flags

CVE-2021-47391

RDMA/cma: Ensure rdma_addr_cancel() happens before issuing more requests

Severity Score

7.8
*CVSS v3

Exploit Likelihood

*EPSS

Affected Versions

*CPE

Public Exploits

0
*Multiple Sources

Exploited in Wild

-
*KEV

Decision

Track
*SSVC
Descriptions

In the Linux kernel, the following vulnerability has been resolved: RDMA/cma: Ensure rdma_addr_cancel() happens before issuing more requests The FSM can run in a circle allowing rdma_resolve_ip() to be called twice
on the same id_priv. While this cannot happen without going through the
work, it violates the invariant that the same address resolution
background request cannot be active twice. CPU 1 CPU 2 rdma_resolve_addr(): RDMA_CM_IDLE -> RDMA_CM_ADDR_QUERY rdma_resolve_ip(addr_handler) #1 process_one_req(): for #1 addr_handler(): RDMA_CM_ADDR_QUERY -> RDMA_CM_ADDR_BOUND mutex_unlock(&id_priv->handler_mutex); [.. handler still running ..] rdma_resolve_addr(): RDMA_CM_ADDR_BOUND -> RDMA_CM_ADDR_QUERY rdma_resolve_ip(addr_handler) !! two requests are now on the req_list rdma_destroy_id(): destroy_id_handler_unlock(): _destroy_id(): cma_cancel_operation(): rdma_addr_cancel() // process_one_req() self removes it spin_lock_bh(&lock); cancel_delayed_work(&req->work); if (!list_empty(&req->list)) == true ! rdma_addr_cancel() returns after process_on_req #1 is done kfree(id_priv) process_one_req(): for #2 addr_handler(): mutex_lock(&id_priv->handler_mutex); !! Use after free on id_priv rdma_addr_cancel() expects there to be one req on the list and only
cancels the first one. The self-removal behavior of the work only happens
after the handler has returned. This yields a situations where the
req_list can have two reqs for the same "handle" but rdma_addr_cancel()
only cancels the first one. The second req remains active beyond rdma_destroy_id() and will
use-after-free id_priv once it inevitably triggers. Fix this by remembering if the id_priv has called rdma_resolve_ip() and
always cancel before calling it again. This ensures the req_list never
gets more than one item in it and doesn't cost anything in the normal flow
that never uses this strange error path.

En el kernel de Linux, se ha resuelto la siguiente vulnerabilidad: RDMA/cma: asegúrese de que rdma_addr_cancel() ocurra antes de emitir más solicitudes. El FSM puede ejecutarse en un círculo permitiendo llamar a rdma_resolve_ip() dos veces en el mismo id_priv. Si bien esto no puede suceder sin realizar el trabajo, viola la invariante de que la misma solicitud en segundo plano de resolución de dirección no puede estar activa dos veces. CPU 1 CPU 2 rdma_resolve_addr(): RDMA_CM_IDLE -> RDMA_CM_ADDR_QUERY rdma_resolve_ip(addr_handler) #1 process_one_req(): para #1 addr_handler(): RDMA_CM_ADDR_QUERY -> RDMA_CM_ADDR_BOUND mutex_unlock(&id_priv->handler_mutex); [.. el controlador sigue ejecutándose...] rdma_resolve_addr(): RDMA_CM_ADDR_BOUND -> RDMA_CM_ADDR_QUERY rdma_resolve_ip(addr_handler)!! ahora hay dos solicitudes en la lista de solicitudes rdma_destroy_id(): destroy_id_handler_unlock(): _destroy_id(): cma_cancel_operation(): rdma_addr_cancel() // Process_one_req() lo elimina automáticamente spin_lock_bh(&lock); cancel_delayed_work(&req->trabajo); if (!list_empty(&req->list)) == verdadero! rdma_addr_cancel() regresa después de que se realiza el proceso_on_req #1 kfree(id_priv) Process_one_req(): para #2 addr_handler(): mutex_lock(&id_priv->handler_mutex); !! El use after free en id_priv rdma_addr_cancel() espera que haya una solicitud en la lista y solo cancela la primera. El comportamiento de autoeliminación del trabajo sólo ocurre después de que el manipulador ha regresado. Esto genera situaciones en las que req_list puede tener dos solicitudes para el mismo "identificador" pero rdma_addr_cancel() solo cancela la primera. El segundo requisito permanece activo más allá de rdma_destroy_id() y usará id_priv después de liberarlo una vez que inevitablemente se active. Solucione este problema recordando si id_priv ha llamado a rdma_resolve_ip() y cancele siempre antes de volver a llamarlo. Esto garantiza que req_list nunca obtenga más de un elemento y no cueste nada en el flujo normal que nunca utiliza esta extraña ruta de error.

In the Linux kernel, the following vulnerability has been resolved: RDMA/cma: Ensure rdma_addr_cancel() happens before issuing more requests The FSM can run in a circle allowing rdma_resolve_ip() to be called twice on the same id_priv. While this cannot happen without going through the work, it violates the invariant that the same address resolution background request cannot be active twice. CPU 1 CPU 2 rdma_resolve_addr(): RDMA_CM_IDLE -> RDMA_CM_ADDR_QUERY rdma_resolve_ip(addr_handler) #1 process_one_req(): for #1 addr_handler(): RDMA_CM_ADDR_QUERY -> RDMA_CM_ADDR_BOUND mutex_unlock(&id_priv->handler_mutex); [.. handler still running ..] rdma_resolve_addr(): RDMA_CM_ADDR_BOUND -> RDMA_CM_ADDR_QUERY rdma_resolve_ip(addr_handler) !! two requests are now on the req_list rdma_destroy_id(): destroy_id_handler_unlock(): _destroy_id(): cma_cancel_operation(): rdma_addr_cancel() // process_one_req() self removes it spin_lock_bh(&lock); cancel_delayed_work(&req->work); if (!list_empty(&req->list)) == true ! rdma_addr_cancel() returns after process_on_req #1 is done kfree(id_priv) process_one_req(): for #2 addr_handler(): mutex_lock(&id_priv->handler_mutex); !! Use after free on id_priv rdma_addr_cancel() expects there to be one req on the list and only cancels the first one. The self-removal behavior of the work only happens after the handler has returned. This yields a situations where the req_list can have two reqs for the same "handle" but rdma_addr_cancel() only cancels the first one. The second req remains active beyond rdma_destroy_id() and will use-after-free id_priv once it inevitably triggers. Fix this by remembering if the id_priv has called rdma_resolve_ip() and always cancel before calling it again. This ensures the req_list never gets more than one item in it and doesn't cost anything in the normal flow that never uses this strange error path.

*Credits: N/A
CVSS Scores
Attack Vector
Local
Attack Complexity
Low
Privileges Required
Low
User Interaction
None
Scope
Unchanged
Confidentiality
High
Integrity
High
Availability
High
Attack Vector
Local
Attack Complexity
Low
Authentication
Single
Confidentiality
Complete
Integrity
Complete
Availability
Complete
* Common Vulnerability Scoring System
SSVC
  • Decision:Track
Exploitation
None
Automatable
No
Tech. Impact
Partial
* Organization's Worst-case Scenario
Timeline
  • 2024-05-21 CVE Reserved
  • 2024-05-21 CVE Published
  • 2024-12-19 CVE Updated
  • 2025-03-18 EPSS Updated
  • ---------- Exploited in Wild
  • ---------- KEV Due Date
  • ---------- First Exploit
CWE
CAPEC
Affected Vendors, Products, and Versions
Vendor Product Version Other Status
Vendor Product Version Other Status <-- --> Vendor Product Version Other Status
Linux
Search vendor "Linux"
Linux Kernel
Search vendor "Linux" for product "Linux Kernel"
>= 2.6.18 < 5.10.188
Search vendor "Linux" for product "Linux Kernel" and version " >= 2.6.18 < 5.10.188"
en
Affected
Linux
Search vendor "Linux"
Linux Kernel
Search vendor "Linux" for product "Linux Kernel"
>= 2.6.18 < 5.14.10
Search vendor "Linux" for product "Linux Kernel" and version " >= 2.6.18 < 5.14.10"
en
Affected
Linux
Search vendor "Linux"
Linux Kernel
Search vendor "Linux" for product "Linux Kernel"
>= 2.6.18 < 5.15
Search vendor "Linux" for product "Linux Kernel" and version " >= 2.6.18 < 5.15"
en
Affected