{"cve":"CVE-2026-99166","nvd":{"published":"2026-10-28T11:17:37.043","modified":"2026-10-28T11:17:37.043","status":"Received","description":"In the Linux kernel, the following vulnerability has been resolved:\n\nnet: rds: fix uninitialized trans dereference in CM event handler\n\nrds_rdma_cm_event_handler_cmn() assigns trans only when the RDMA\ndevice is an InfiniBand CA (RDMA_NODE_IB_CA).  On any other device\ntype, e.g. an iWARP RNIC such as siw, trans stays uninitialized, but\nthe event switch dereferences it: unconditionally in the\nRDMA_CM_EVENT_CONNECT_REQUEST case via trans->cm_handle_connect(),\nand (with a connection context) in the ROUTE_RESOLVED and\nESTABLISHED cases.\n\nAn RDS listener on an iWARP device therefore crashes the kernel as\nsoon as a connect request arrives: with CONFIG_INIT_STACK_ALL_ZERO\nthe wild load becomes a NULL dereference at offset 0xa0\n(&trans->cm_handle_connect) in the iw_cm_wq workqueue.\n\nGCC masks the bug in default builds by folding the uninitialized\nload into &rds_ib_transport; Clang-built kernels take the real\nuninitialized path and oops.\n\nThe active side can land on a non-IB device too: the id that\nrds_ib_conn_path_connect() creates is not restricted to a node type,\nso rdma_resolve_addr() binds it to whichever device serves the local\naddress, and an iWARP RNIC on the same netdev qualifies.  The crash\nthere is different: resolving a route on an iWARP id never fills in\ncm_id->route.path_rec, and the ROUTE_RESOLVED case writes\npath_rec[0].sl before it calls into the transport.\n\nThe iWARP transport was dropped long ago and IB is the only\ntransport left, so make that explicit: initialize trans to\n&rds_ib_transport at declaration, drop the conditional assignment,\nreject a connect request that arrives on a non-IB device, and drop a\nconnection whose address resolved to one instead of resolving a\nroute on it.  The rejection is limited to RDMA_CM_EVENT_CONNECT_REQUEST\non purpose: a non-zero return from the handler makes rdma_cm destroy\nthe id the event was delivered on, which is right for the request's\nfreshly created id but would free a connection id that RDS still owns\nfor any other event.  On the active side the id stays ic->i_cm_id and\nthe connection's shutdown destroys it, so that case returns 0.\n\n[achender: reject only on RDMA_CM_EVENT_CONNECT_REQUEST instead of\n bailing out for every event on a non-IB device, so that rdma_cm does\n not destroy connection ids RDS still tracks; drop a connection whose\n address resolved to a non-IB device before a route is resolved on it;\n changelog adjusted]","cvss_v4":null,"cvss_v3":null,"cwes":[],"references":["https://git.kernel.org/stable/c/492ba6ac1b1ffb8a06ff52db0825c1af2cdb28c8","https://git.kernel.org/stable/c/5f23e4e3a31dacfd82a8f56be253adc0910ee84f","https://git.kernel.org/stable/c/ca14c289ed9ffed6cc64290d039e923e5720a8bf"],"products":[],"source":"nvd","synced_at":"2026-10-28T11:37:51.394563875+00:00"},"epss":null,"kev":null,"osv":[],"attribution":["This product uses the NVD API but is not endorsed or certified by the NVD.","EPSS scores: FIRST Exploit Prediction Scoring System, https://www.first.org/epss","Known exploited vulnerabilities: CISA Known Exploited Vulnerabilities Catalog, https://www.cisa.gov/known-exploited-vulnerabilities-catalog","Package advisories: OSV (https://osv.dev), from the GitHub Advisory Database, PyPA, Go, RustSec and other ecosystem databases, under CC-BY 4.0 or CC0","CVE records: the CVE Program's CVE List (https://www.cve.org), under its terms of use (https://www.cve.org/Legal/TermsOfUse). CVE is a registered trademark of The MITRE Corporation; this product is not affiliated with the CVE Program."]}