linux.git/fs/smb, branch master

Merge tag 'v7.2-rc4-smb3-server-fixes' of git://git.samba.org/ksmbd

2026-07-25T02:50:05Z

Pull smb server fixes from Steve French: "This contains eight ksmbd fixes covering POSIX ACL handling, SMB signing enforcement, DACL parsing and construction hardening, session lifetime handling, and validation of malformed transform and compressed SMB2 requests: - preserve inherited POSIX ACL mask when creating objects. - enforce the session signing requirement for plaintext SMB requests. - harden DACL/ACE processing against size overflows, incomplete ACE copies, and undersized SIDs. - defer teardown of a previous session until NTLM authentication succeeds. - reject undersized encryption-transform and decompressed SMB2 requests before they can reach normal SMB2 request processing" * tag 'v7.2-rc4-smb3-server-fixes' of git://git.samba.org/ksmbd: ksmbd: reject undersized decompressed SMB2 requests ksmbd: validate minimum PDU size for transform requests ksmbd: defer destroy_previous_session() until after NTLM authentication ksmbd: validate ACE size against SID sub-authorities ksmbd: restore DACL size on check_add_overflow() to avoid malformed ACL ksmbd: bound DACL dedup walk to copied ACEs ksmbd: enforce signing required by the session ksmbd: preserve VFS inherited POSIX ACL mask

ksmbd: reject undersized decompressed SMB2 requests

2026-07-22T14:54:10Z

ksmbd_decompress_request() bounds the decompressed size only against the maximum request size. A compression transform can therefore produce a buffer smaller than an SMB2 PDU and install it as conn->request_buf. The receive path subsequently calls ksmbd_smb_request(), which reads the protocol ID before the normal SMB2 minimum-size check. If the decompressed output is too short, that read can access beyond the request allocation. Require the decompressed output to contain at least a complete minimum SMB2 PDU before allocating and installing the replacement request buffer. Fixes: a08de24c2b85 ("ksmbd: negotiate and decode SMB2 compression") Cc: stable@vger.kernel.org Signed-off-by: Namjae Jeon Signed-off-by: Steve French

ksmbd: validate minimum PDU size for transform requests

2026-07-22T14:54:10Z

The receive path applies the minimum SMB2 PDU size check only when ProtocolId is SMB2_PROTO_NUMBER. A packet carrying SMB2_TRANSFORM_PROTO_NUM bypasses the check even when the negotiated dialect does not provide transform handling. On an SMB 2.1 connection, a short transform packet therefore reaches init_smb2_rsp_hdr(), which interprets the request as a full SMB2 header and reads beyond the request allocation. The copied fields can then be returned to the unauthenticated client. Compression transforms are converted to ordinary SMB2 messages before protocol validation. After that conversion, validate ordinary SMB2 requests against SMB2_MIN_SUPPORTED_PDU_SIZE and require encryption transform requests to contain both a transform header and an SMB2 header. This rejects truncated requests before work allocation. Fixes: 368ba06881c3 ("ksmbd: check the validation of pdu_size in ksmbd_conn_handler_loop") Cc: stable@vger.kernel.org Reported-by: zdi-disclosures@trendmicro.com # ZDI-CAN-31063 Signed-off-by: Namjae Jeon Signed-off-by: Steve French

ksmbd: defer destroy_previous_session() until after NTLM authentication

2026-07-22T14:54:10Z

In ntlm_authenticate(), destroy_previous_session() is called using a user pointer resolved from the client-supplied NTLM blob username field before the NTLMv2 response is validated. An authenticated attacker can set the NTLM blob username to match a victim account and set PreviousSessionId to the victim's session ID; destroy_previous_session() destroys the victim's session while ksmbd_decode_ntlmssp_auth_blob() subsequently rejects the request with -EPERM. Move destroy_previous_session() and the prev_id assignment to after ksmbd_decode_ntlmssp_auth_blob() returns success and use sess->user rather than the pre-authentication lookup result. This matches the ordering already used by krb5_authenticate(), where destroy_previous_session() is called only after ksmbd_krb5_authenticate() returns success. Fixes: e2f34481b24d ("cifsd: add server-side procedures for SMB3") Cc: stable@vger.kernel.org Link: https://lore.kernel.org/linux-cifs/20260702155449.3639773-1-james_montgomery@disroot.org/ Signed-off-by: James Montgomery Acked-by: Namjae Jeon Signed-off-by: Steve French

ksmbd: validate ACE size against SID sub-authorities

2026-07-22T14:54:10Z

set_ntacl_dacl() validates sid.num_subauth before copying an ACE, but does not verify that the declared ACE size contains all sub-authorities described by that field. An undersized ACE can therefore be copied and later make the POSIX ACL deduplication walk inspect data beyond the copied ACE boundary. The existing initial bound check is also too small. It only ensures that the ACE size field is accessible before set_ntacl_dacl() reads sid.num_subauth farther into the input buffer. Require enough input for the fixed SID header before accessing num_subauth, reject ACEs smaller than that header, and skip ACEs whose declared size cannot contain the complete SID. This makes the validation consistent with the other ACE walk paths. Reported-by: LocalHost Signed-off-by: Namjae Jeon Signed-off-by: Steve French

ksmbd: restore DACL size on check_add_overflow() to avoid malformed ACL

2026-07-22T14:54:10Z

check_add_overflow() unconditionally writes the truncated sum into *d even on overflow, per its contract in include/linux/overflow.h. The four check_add_overflow() guards in set_posix_acl_entries_dacl() and set_ntacl_dacl() break out of the ACE-building loops on overflow, but the truncated *size is then consumed downstream at the end of set_ntacl_dacl(): pndacl->size = cpu_to_le16(le16_to_cpu(pndacl->size) + size); This produces an on-wire NT ACL whose pndacl->size under-reports the bytes actually written by the preceding fill_ace_for_sid()/memcpy() calls, yielding a malformed ACL that can trigger out-of-bounds reads when re-parsed by clients or ksmbd itself. Restore *size to its pre-addition value on each overflow branch (via `*size -= ace_sz` / `size -= nt_ace_size`) so that after the break, *size once again holds the cumulative size of the successfully-written ACEs. The committed ACL is then truncated-but-self-consistent rather than malformed. The ksmbd DACL builders are the only check_add_overflow() sites found where an overflow path breaks out of a loop and the destination value is consumed afterward. The other nearby break-style cases either return -EINVAL on overflow (transport_ipc.c) or break without consuming the overflowed destination value afterward (buildid.c). Fixes: 299f962c0b02 ("ksmbd: use check_add_overflow() to prevent u16 DACL size overflow") Assisted-by: atomcode:glm-5.2 Assisted-by: Codex:gpt-5.5 Cc: stable@vger.kernel.org Signed-off-by: Wentao Guan Acked-by: Namjae Jeon Signed-off-by: Steve French

ksmbd: bound DACL dedup walk to copied ACEs

2026-07-22T14:54:10Z

set_ntacl_dacl() can stop copying ACEs before consuming the full input DACL when size accounting overflows. When that happens, num_aces reflects only the ACEs that were actually copied into the output DACL, but set_posix_acl_entries_dacl() still receives nt_num_aces and uses it to walk the existing ACE array during dedup. That makes the dedup walk scan past the copied ACE array and inspect buffer tail that does not contain valid ACEs. Split the two meanings currently carried by the NT ACE count. Pass the number of copied NT ACEs to bound the dedup walk, and preserve the original "input DACL had NT ACEs" state separately for the Everyone/default ACL fallback. This keeps the dedup walk aligned with the ACEs that are actually present in the rebuilt DACL. Signed-off-by: Namjae Jeon Signed-off-by: Steve French

ksmbd: enforce signing required by the session

2026-07-22T14:54:03Z

SMB2_FLAGS_SIGNED is controlled by the incoming request and only indicates that a signature accompanies that request. Do not use it to decide whether a signing-required session must authenticate the request. Reject an unsigned plaintext request before dispatch when the session requires signing. Continue to validate signatures on signed requests, including when signing is optional. Encrypted requests have already been authenticated during decryption. An OPLOCK_BREAK acknowledgment is a session request and is subject to the same signing rule, so do not exclude it from signed-request detection. Reported-by: Charles Vosburgh Tested-by: ChenXiaoSong Reviewed-by: ChenXiaoSong Signed-off-by: Namjae Jeon Signed-off-by: Steve French

ksmbd: preserve VFS inherited POSIX ACL mask

2026-07-22T14:54:01Z

The VFS initializes a child's POSIX ACL from the parent's default ACL and the requested creation mode. Do not mutate the parent ACL or overwrite the child's VFS-computed access and default ACLs afterwards. This preserves restrictive ACL_MASK entries and prevents SMB object creation from widening effective permissions. Reported-by: Charles Vosburgh Signed-off-by: Namjae Jeon Signed-off-by: Steve French

cifs: fix cifsFileInfo leak on kmalloc failure in deferred close drain paths

2026-07-22T01:19:11Z

In cifs_close_deferred_file(), cifs_close_all_deferred_files(), and cifs_close_deferred_file_under_dentry(), when a pending deferred close is cancelled via cancel_delayed_work(), the subsequent kmalloc_obj() to add the file to the local processing list may fail under memory pressure. The loop breaks immediately, but the cancelled work is no longer pending (it would have called _cifsFileInfo_put()), and the cfile is never added to file_head for processing. The cifsFileInfo reference and the open server handle both leak. Fix by saving the cfile that failed allocation in a local variable, breaking as before, and calling _cifsFileInfo_put() on it after releasing the lock. Any files later in the iteration are unaffected since their deferred work is still pending and will fire normally. Fixes: e3fc065682eb ("cifs: Deferred close performance improvements") Signed-off-by: Frank Sorenson Signed-off-by: Steve French