{"dataType":"CVE_RECORD","dataVersion":"5.2","cveMetadata":{"cveId":"CVE-2025-38523","assignerOrgId":"416baaa9-dc9f-4396-8d5f-8c081fb06d67","state":"PUBLISHED","assignerShortName":"Linux","dateReserved":"2025-04-16T04:51:24.023Z","datePublished":"2025-08-16T11:12:17.254Z","dateUpdated":"2026-08-05T12:02:35.936Z"},"containers":{"cna":{"providerMetadata":{"orgId":"416baaa9-dc9f-4396-8d5f-8c081fb06d67","shortName":"Linux","dateUpdated":"2026-08-05T12:02:35.936Z"},"descriptions":[{"lang":"en","value":"In the Linux kernel, the following vulnerability has been resolved:\n\ncifs: Fix the smbd_response slab to allow usercopy\n\nThe handling of received data in the smbdirect client code involves using\ncopy_to_iter() to copy data from the smbd_reponse struct's packet trailer\nto a folioq buffer provided by netfslib that encapsulates a chunk of\npagecache.\n\nIf, however, CONFIG_HARDENED_USERCOPY=y, this will result in the checks\nthen performed in copy_to_iter() oopsing with something like the following:\n\n CIFS: Attempting to mount //172.31.9.1/test\n CIFS: VFS: RDMA transport established\n usercopy: Kernel memory exposure attempt detected from SLUB object 'smbd_response_0000000091e24ea1' (offset 81, size 63)!\n ------------[ cut here ]------------\n kernel BUG at mm/usercopy.c:102!\n ...\n RIP: 0010:usercopy_abort+0x6c/0x80\n ...\n Call Trace:\n  <TASK>\n  __check_heap_object+0xe3/0x120\n  __check_object_size+0x4dc/0x6d0\n  smbd_recv+0x77f/0xfe0 [cifs]\n  cifs_readv_from_socket+0x276/0x8f0 [cifs]\n  cifs_read_from_socket+0xcd/0x120 [cifs]\n  cifs_demultiplex_thread+0x7e9/0x2d50 [cifs]\n  kthread+0x396/0x830\n  ret_from_fork+0x2b8/0x3b0\n  ret_from_fork_asm+0x1a/0x30\n\nThe problem is that the smbd_response slab's packet field isn't marked as\nbeing permitted for usercopy.\n\nFix this by passing parameters to kmem_slab_create() to indicate that\ncopy_to_iter() is permitted from the packet region of the smbd_response\nslab objects, less the header space."}],"metrics":[{"cvssV3_1":{"version":"3.1","vectorString":"CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H","baseScore":7.5,"baseSeverity":"HIGH"},"scenarios":[{"lang":"en","value":"AV:N - The crash is driven entirely by data received from the remote SMB Direct peer in the CIFS client's transport receive path (`cifs_demultiplex_thread` → `smbd_recv`), and SMB Direct runs over iWARP/RoCEv2/soft-iWARP, which are routable over ordinary IP networks rather than confined to one L2 segment. A malicious or impersonated SMB server, or a MITM on the unauthenticated smbdirect transport headers, reaches the code remotely.\nAC:L - The failure is fully deterministic — the missing slab whitelist makes `__check_heap_object()` reject every single `copy_to_iter()` out of `smbd_response`, so the very first SMB2 response body over the connection panics the client with no race, timing, or memory-layout dependency. `CONFIG_HARDENED_USERCOPY` is a default-on distro hardening option and `CONFIG_CIFS_SMB_DIRECT` is a supported \"say Y\" option, so no rare configuration is needed within the RDMA/HPC/cloud-storage deployments that use this transport.\nPR:N - The attacker is the remote smbdirect peer and holds no privileges or credentials on the victim client; the oops fires while processing the SMB2 NEGOTIATE response, before session setup and authentication complete. Anything answering the client's SMBD connect on port 5445/445 can trigger it.\nUI:N - No human action is part of the attack — RDMA-backed CIFS mounts are established automatically from fstab/autofs/systemd units and the client auto-reconnects, a state an attacker can force simply by dropping the connection. The crash then occurs during ordinary automatic protocol negotiation.\nS:U - The `BUG()` and its consequences are confined to the kernel of the affected client host, with no crossing into another security authority such as a hypervisor, IOMMU, or sandbox boundary.\nC:N - No data is disclosed: the hardening check aborts the transfer rather than permitting an out-of-bounds read, and the source region is legitimate, in-object packet payload the client requested. Removing the whitelist gap loosens the check, so the defect grants no read primitive.\nI:N - Nothing is written or corrupted — the code path ends in a defensive `usercopy_abort()` before any copy occurs, so there is no overwrite, type confusion, or control-flow hijack primitive.\nA:H - `usercopy_abort()` calls `BUG()`, producing a kernel oops in the `cifs_demultiplex_thread` kthread — a full panic under `panic_on_oops` (routinely paired with hardened-usercopy builds) and otherwise a dead demultiplex thread that permanently wedges the mount. The remote peer can trigger this repeatedly on every connection attempt."}]}],"affected":[{"product":"Linux","vendor":"Linux","defaultStatus":"unaffected","repo":"https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git","programFiles":["fs/smb/client/smbdirect.c"],"versions":[{"version":"ee4cdf7ba857a894ad1650d6ab77669cbbfa329e","lessThan":"87dcc7e33fc3dcb8ed32333cec016528b5bb6ce4","status":"affected","versionType":"git"},{"version":"ee4cdf7ba857a894ad1650d6ab77669cbbfa329e","lessThan":"f0dd353d47f7051afa98c6c60c7486831eb1a410","status":"affected","versionType":"git"},{"version":"ee4cdf7ba857a894ad1650d6ab77669cbbfa329e","lessThan":"43e7e284fc77b710d899569360ea46fa3374ae22","status":"affected","versionType":"git"}]},{"product":"Linux","vendor":"Linux","defaultStatus":"affected","repo":"https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git","programFiles":["fs/smb/client/smbdirect.c"],"versions":[{"version":"6.12","status":"affected"},{"version":"0","lessThan":"6.12","status":"unaffected","versionType":"semver"},{"version":"6.12.36","lessThanOrEqual":"6.12.*","status":"unaffected","versionType":"semver"},{"version":"6.15.8","lessThanOrEqual":"6.15.*","status":"unaffected","versionType":"semver"},{"version":"6.16","lessThanOrEqual":"*","status":"unaffected","versionType":"original_commit_for_fix"}]}],"cpeApplicability":[{"nodes":[{"operator":"OR","negate":false,"cpeMatch":[{"vulnerable":true,"criteria":"cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:*","versionStartIncluding":"6.12","versionEndExcluding":"6.12.36"},{"vulnerable":true,"criteria":"cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:*","versionStartIncluding":"6.12","versionEndExcluding":"6.15.8"},{"vulnerable":true,"criteria":"cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:*","versionStartIncluding":"6.12","versionEndExcluding":"6.16"}]}]}],"references":[{"url":"https://git.kernel.org/stable/c/87dcc7e33fc3dcb8ed32333cec016528b5bb6ce4"},{"url":"https://git.kernel.org/stable/c/f0dd353d47f7051afa98c6c60c7486831eb1a410"},{"url":"https://git.kernel.org/stable/c/43e7e284fc77b710d899569360ea46fa3374ae22"}],"title":"cifs: Fix the smbd_response slab to allow usercopy","x_generator":{"engine":"bippy-1.2.0"}}}}