{"id":"CVE-2026-93228","published":"2026-09-24T16:17:18.253","lastModified":"2026-09-25T05:17:00.567","description":"In the Linux kernel, the following vulnerability has been resolved:\n\nsvcrdma: Reject Write/Reply chunks with segcount 0\n\nA peer can send a Write or Reply chunk whose segcount field is zero.\nxdr_check_write_chunk() only rejects segcount > rc_maxpages, so zero\npasses the range check, and xdr_inline_decode(stream, 0) returns the\ncurrent (non-NULL) cursor without advancing. The function returns\ntrue and pcl_alloc_write() then links a struct svc_rdma_chunk with\nch_segcount == 0 onto rc_write_pcl or rc_reply_pcl.\n\nAn earlier patch in this series made pcl_for_each_segment() safe for\nch_segcount == 0, so this no longer drives the memory walk it used\nto. Rejecting the malformed frame at the decode boundary is still\nworthwhile as defense in depth: it keeps degenerate zero-segment\nchunks off the parsed chunk lists entirely, so any future consumer\nthat walks ch_segments directly cannot observe one, and it makes the\nzero-floor easy to backport to trees where the macro change is more\nintrusive. RFC 8166 has no meaning for a Write/Reply chunk that\ndescribes no remote buffer, so no legitimate client is affected.\n\nxdr_check_reply_chunk() funnels Reply chunks through\nxdr_check_write_chunk() and inherits the same rejection.\n\npcl_alloc_write() also links each chunk onto the parsed chunk list\nbefore filling its segment array. If a future change weakens the\nsegcount-0 rejection, an incomplete chunk is visible to consumers\nduring the fill loop. Reorder so that list_add_tail() follows the\nsegment fill loop, ensuring only fully-populated chunks appear on\nthe list.","cvssScore":9.1,"cvssSeverity":"CRITICAL","cvssVector":"CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:N/A:H","cwes":[],"vendors":[],"products":[],"references":[{"url":"https://git.kernel.org/stable/c/45dbdb2637b7fc5f1355b588780d2a7fb0516805","tags":[]},{"url":"https://git.kernel.org/stable/c/9808eb7656666acc7291bae9ab6b987bd16e47e0","tags":[]},{"url":"https://git.kernel.org/stable/c/a798714b58041e88716db3bf8fb03ae13eecd54a","tags":[]},{"url":"https://git.kernel.org/stable/c/fe533ae1bed2ef2b3d54ee7ab410da1874ed3d4f","tags":[]}],"exploitRefs":[],"hasPoc":false,"ai":{"summary":"The flaw allows a peer to send a Write or Reply chunk with a segcount of 0, which is not properly rejected, potentially leading to memory issues or denial of service.","exploitability":"Exploitation requires the ability to send malformed RDMA frames, which is generally difficult without control over the network or kernel.","blast_radius":"If exploited, it could lead to memory corruption or denial of service for the affected system.","remediation":"Upgrade to the Linux kernel version 6.2.11 or later.","detection":"No reliable host or network indicator is derivable from the published description.","tags":["rdma","memory-corruption","dos"],"model":"qwen2.5:7b-instruct","analyzedAt":"2026-09-28T08:52:37.390Z"}}