net/rds: don't let the rdma_cm destroy an id RDS still owns on route failure
Impact
In the Linux kernel, the following vulnerability has been resolved:
Source excerpt · LinuxAffected packages
Affected products
Vendor version evidence. Product records are separate from automatic dependency matching.
Linux · Linux
Other versions: unaffected
| Version / range | Source status |
|---|---|
| 55b7ed0b582f9ff4bf3b9b44969cddda8c6763ea to before a7c7d0ff876377fea7ae876a28d6df8d9349df57 | affected |
| 55b7ed0b582f9ff4bf3b9b44969cddda8c6763ea to before 91ece3efa064f0b3b3c67b9b045d9eeb4a034e43 | affected |
| 55b7ed0b582f9ff4bf3b9b44969cddda8c6763ea to before 98d446b7c013254a8a7b49720557676a3a851fde | affected |
Linux · Linux
Other versions: affected
| Version / range | Source status |
|---|---|
| 2.6.30 | affected |
| 0 to before 2.6.30 | unaffected |
| 6.18.59 through 6.18.* (inclusive) | unaffected |
| 7.2.13 through 7.2.* (inclusive) | unaffected |
| 7.3 through * (inclusive) | unaffected |
Original source description
In the Linux kernel, the following vulnerability has been resolved:
net/rds: don't let the rdma_cm destroy an id RDS still owns on route failure
rds_rdma_cm_event_handler_cmn() hands the return value of rdma_resolve_route() straight back to the rdma_cm on RDMA_CM_EVENT_ADDR_RESOLVED. A synchronous failure there - -ENOMEM from the route work allocation, an SA query that cannot be set up, a RoCE route with no usable device - makes addr_handler() destroy the id the event was delivered on. That id is ic->i_cm_id, and nothing clears the pointer: the connection sits in RDS_CONN_CONNECTING with a freed id until its shutdown calls rdma_disconnect() and rdma_destroy_id() on it.
The ROUTE_RESOLVED case already avoids this: rds_ib_cm_initiate_connect() forces a zero return while ic->i_cm_id == cm_id, as the comment there explains. Do the same here - drop the connection and return 0 - so the id stays RDS's to destroy from the shutdown, and the reconnect gets a fresh one.