kernel/linux.git/fs/smb, branch v6.18.36

ksmbd: fix use-after-free of a deferred file_lock on double SMB2_CANCEL

2026-06-19T11:44:04+00:00

commit f580d27e8928828693df44ba2db0fffdbe11dfea upstream. A deferred byte-range lock (an SMB2_LOCK that blocks) registers an async work on conn->async_requests via setup_async_work(), with cancel_fn = smb2_remove_blocked_lock and cancel_argv[0] pointing at the struct file_lock. When the request is cancelled, the worker frees the file_lock with locks_free_lock() and takes the cancelled early-exit, which "goto out"s and never reaches release_async_work() -- the only site that unlinks the work from conn->async_requests and clears cancel_fn/cancel_argv. The work therefore stays matchable on async_requests with a live cancel_fn pointing at the freed file_lock, until connection teardown finally runs release_async_work(). smb2_cancel() fires cancel_fn unconditionally with no state guard, so a second SMB2_CANCEL for the same AsyncId, arriving in that window, re-runs smb2_remove_blocked_lock() on the freed file_lock -- a slab use-after-free: BUG: KASAN: slab-use-after-free in __locks_delete_block __locks_delete_block locks_delete_block ksmbd_vfs_posix_lock_unblock smb2_remove_blocked_lock smb2_cancel <- 2nd SMB2_CANCEL fires cancel_fn handle_ksmbd_work Allocated by ...: locks_alloc_lock <- smb2_lock Freed by ...: locks_free_lock <- smb2_lock (cancelled branch) ... cache file_lock_cache of size 192 Reproduced on mainline with KASAN by an authenticated SMB client. Skip a work whose state is already KSMBD_WORK_CANCELLED so its cancel callback cannot be fired a second time. Cc: stable@vger.kernel.org Signed-off-by: Gil Portnoy Acked-by: Namjae Jeon Signed-off-by: Steve French Signed-off-by: Greg Kroah-Hartman

ksmbd: fix NULL-deref of opinfo->conn in oplock/lease break notifiers

2026-06-19T11:43:52+00:00

[ Upstream commit b003086d76968298f22e7cf62239833b5a3a06b1 ] smb2_oplock_break_noti() and smb2_lease_break_noti() read opinfo->conn into a local with neither READ_ONCE() nor a NULL check. Both run from oplock_break() after opinfo_get_list() has dropped ci->m_lock, so a concurrent SMB2 LOGOFF (session_fd_check()) can set op->conn = NULL under ci->m_lock within that window. ksmbd_conn_r_count_inc(conn) then writes through NULL at offset 0xc4 -- a remotely triggerable oops. Guard both reads the way compare_guid_key() already does: read opinfo->conn with READ_ONCE() and return early if it is NULL, before allocating the work struct so nothing leaks. A NULL conn means the client is gone and the break is moot, so return 0; oplock_break() treats that as success and runs the normal teardown. Fixes: c8efcc786146 ("ksmbd: add support for durable handles v1/v2") Assisted-by: Henry (Claude):claude-opus-4 Signed-off-by: Gil Portnoy Acked-by: Namjae Jeon Signed-off-by: Steve French Signed-off-by: Sasha Levin

ksmbd: OOB read regression in smb_check_perm_dacl() ACE-walk loops

2026-06-09T10:28:42+00:00

commit 0e60dafe97eca61721f3db456f97d97a80c6c8ae upstream. Commit d07b26f39246 ("ksmbd: require minimum ACE size in smb_check_perm_dacl()") introduced a transposed bounds check: if (offsetof(struct smb_ace, sid) + aces_size < CIFS_SID_BASE_SIZE) Since offsetof(..sid) is 8 and CIFS_SID_BASE_SIZE is 8, this evaluates to `aces_size < 0`. Because `aces_size` is always non-negative, this check becomes dead code and never breaks the loop. Worse, that commit removed the old 4-byte guard, meaning the loop now reads `ace->size` (offset 2) even when `aces_size` is 0-3 bytes. This re-opens a 2-byte heap out-of-bounds (OOB) read past the pntsd allocation during subsequent SMB2_CREATE operations. Fix this by properly transposing the comparison to require at least 16 bytes (8-byte offset + 8-byte SID base), matching the correct form used in smb_inherit_dacl(). Fixes: d07b26f39246 ("ksmbd: require minimum ACE size in smb_check_perm_dacl()") Cc: stable@vger.kernel.org Signed-off-by: Ali Ganiyev Acked-by: Namjae Jeon Signed-off-by: Steve French Signed-off-by: Greg Kroah-Hartman

smb: client: fix uninitialized variable in smb2_writev_callback

2026-06-09T10:28:35+00:00

commit 9d2491197a00acf8c423512078458c2855102b66 upstream. compiling with W=2 pointed out that "written may be used uninitialized" Fixes: 20d72b00ca81 ("netfs: Fix the request's work item to not require a ref") Cc: stable@vger.kernel.org Reviewed-by: David Howells Signed-off-by: Steve French Signed-off-by: Greg Kroah-Hartman

smb: client: validate the whole DACL before rewriting it in cifsacl

2026-06-09T10:28:32+00:00

[ Upstream commit 0a8cf165566ba55a39fd0f4de172119dd646d39a ] build_sec_desc() and id_mode_to_cifs_acl() derive a DACL pointer from a server-supplied dacloffset and then use the incoming ACL to rebuild the chmod/chown security descriptor. The original fix only checked that the struct smb_acl header fits before reading dacl_ptr->size or dacl_ptr->num_aces. That avoids the immediate header-field OOB read, but the rewrite helpers still walk ACEs based on pdacl->num_aces with no structural validation of the incoming DACL body. A malicious server can return a truncated DACL that still contains a header, claims one or more ACEs, and then drive replace_sids_and_copy_aces() or set_chmod_dacl() past the validated extent while they compare or copy attacker-controlled ACEs. Factor the DACL structural checks into validate_dacl(), extend them to validate each ACE against the DACL bounds, and use the shared validator before the chmod/chown rebuild paths. parse_dacl() reuses the same validator so the read-side parser and write-side rewrite paths agree on what constitutes a well-formed incoming DACL. Fixes: bc3e9dd9d104 ("cifs: Change SIDs in ACEs while transferring file ownership.") Cc: stable@vger.kernel.org Assisted-by: Claude:claude-opus-4-6 Assisted-by: Codex:gpt-5-4 Signed-off-by: Michael Bommarito Signed-off-by: Steve French Signed-off-by: Sasha Levin

ksmbd: fix FSCTL permission bypass by adding a permission check for FSCTL_SET_SPARSE

2026-06-09T10:28:28+00:00

[ Upstream commit cc57232cae23c0df91b4a59d0f519141ce9b5b02 ] FSCTL_SET_SPARSE in fsctl_set_sparse() modifies the file's sparse attribute and saves it through xattr without any permission checks. This exposes two issues: 1) A client on a read-only share can change the sparse attribute on files it opened, even though the share is read-only. Other FSCTL write operations already check test_tree_conn_flag(work->tcon, KSMBD_TREE_CONN_FLAG_WRITABLE), but FSCTL_SET_SPARSE does not. 2) Even on writable shares, clients without FILE_WRITE_DATA or FILE_WRITE_ATTRIBUTES access should not modify the sparse attribute. Similar handle-level checks exist in other functions but are missing here. Add both share-level writable check and per-handle access check. Use goto out on error to avoid leaking file references. Fixes: e2f34481b24d ("cifsd: add server-side procedures for SMB3") Cc: Namjae Jeon Cc: Sergey Senozhatsky Cc: Steve French Signed-off-by: Sean Shen Acked-by: Namjae Jeon Signed-off-by: Steve French Signed-off-by: Sasha Levin

ksmbd: fix durable reconnect error path file lifetime

2026-06-01T15:51:08+00:00

[ Upstream commit 3515503322f4819277091839eed46b695096aca5 ] After a durable reconnect succeeds, ksmbd_reopen_durable_fd() republishes the same ksmbd_file into the session volatile-id table. If smb2_open() then takes a later error path, cleanup first calls ksmbd_fd_put(work, fp) and then unconditionally calls ksmbd_put_durable_fd(dh_info.fp). In this case fp and dh_info.fp are the same object. The first put drops the reconnect lookup reference, but the final durable put can run __ksmbd_close_fd(NULL, fp). Because the final close is not session-aware, it can free the file object without removing the volatile-id entry that was just published into the session table. Use the session-aware put for the final reconnect drop when the reconnect had already succeeded and the error path is cleaning up the republished file. Earlier reconnect failures, before fp is assigned to dh_info.fp, keep using the durable-only put path. Fixes: 1baff47b81f9 ("ksmbd: fix use-after-free in smb2_open during durable reconnect") Signed-off-by: Junyi Liu Acked-by: Namjae Jeon Signed-off-by: Steve French Signed-off-by: Sasha Levin

cifs: Fix busy dentry used after unmounting

2026-06-01T15:50:45+00:00

commit c68337442f03953237a94577beb468ab2662a851 upstream. Since commit 340cea84f691c ("cifs: open files should not hold ref on superblock"), cifs file only holds the dentry ref_cnt, the cifs file close work(cfile->deferred) could be executed after unmounting, which will trigger a warning in generic_shutdown_super: BUG: Dentry 00000000a14a6845{i=c,n=file} still in use (1) [unmount of cifs cifs] The detailed processs is: process A process B kworker fd = open(PATH) vfs_open file->__f_path = *path // dentry->d_lockref.count = 1 cifs_open cifs_new_fileinfo cfile->dentry = dget(dentry) // dentry->d_lockref.count = 2 close(fd) __fput cifs_close queue_delayed_work(deferredclose_wq, cfile->deferred) dput(dentry) // dentry->d_lockref.count = 1 smb2_deferred_work_close _cifsFileInfo_put list_del(&cifs_file->flist) umount cleanup_mnt deactivate_super cifs_kill_sb cifs_close_all_deferred_files_sb cifs_close_all_deferred_files // cannot find cfile, skip _cifsFileInfo_put kill_anon_super generic_shutdown_super shrink_dcache_for_umount umount_check WARN ! // dentry->d_lockref.count = 1 cifsFileInfo_put_final dput(cifs_file->dentry) // dentry->d_lockref.count = 0 Fix it by flushing 'deferredclose_wq' before calling kill_anon_super. Fetch a reproducer in https://bugzilla.kernel.org/show_bug.cgi?id=221548. Fixes: 340cea84f691c ("cifs: open files should not hold ref on superblock") Cc: stable@vger.kernel.org Reviewed-by: Shyam Prasad N Signed-off-by: Zhihao Cheng Signed-off-by: Steve French Signed-off-by: Greg Kroah-Hartman

smb/server: promote S_DEL_ON_CLS to S_DEL_PENDING when close

2026-06-01T15:50:40+00:00

commit 4ec9c8e023c79f613fe4d5ad8cc737112efb2e44 upstream. Reproducer: 1. server: systemctl start ksmbd 2. client: mount -t cifs //${server_ip}/export /mnt 3. client: C program: openat(AT_FDCWD, "/mnt", O_RDWR | O_TMPFILE, 0600) Do not treat `FILE_DELETE_ON_CLOSE_LE` as delete pending while files remain open. This patch fixes xfstests generic/004. Cc: stable@vger.kernel.org Link: https://chenxiaosong.com/en/smb-xfstests-generic-004.html Co-developed-by: Huiwen He Signed-off-by: Huiwen He Signed-off-by: ChenXiaoSong Tested-by: Steve French Acked-by: Namjae Jeon Signed-off-by: Steve French Signed-off-by: Greg Kroah-Hartman

smb: client: use data_len for SMB2 READ encrypted folioq copy

2026-06-01T15:50:40+00:00

commit d4d76c9ee1997cc8c977a63f6c43551c253c1066 upstream. In handle_read_data() the encrypted/folioq branch (buf_len <= data_offset, reached via receive_encrypted_read for transform PDUs > CIFSMaxBufSize + MAX_HEADER_SIZE) copies the READ payload using buffer_len rather than data_len: rdata->result = cifs_copy_folioq_to_iter(buffer, buffer_len, cur_off, &rdata->subreq.io_iter); ... rdata->got_bytes = buffer_len; buffer_len comes from the SMB3 transform header OriginalMessageSize field (OriginalMessageSize - read_rsp_size); it represents the size of the decrypted message after the SMB2 header. data_len comes from the SMB2 READ response DataLength field; it represents the actual READ payload size and may be smaller than buffer_len when the decrypted message contains padding or other trailing bytes after the READ payload. The existing check `data_len > buffer_len - pad_len` only enforces an upper bound, so a server that emits OriginalMessageSize larger than read_rsp_size + pad_len + data_len passes the check and the kernel copies buffer_len bytes per response, ignoring the server-asserted DataLength. Two observable failures with a crafted server (DataLength=4, buffer_len=20000): - the kernel returns 20000 bytes per sub-request to userspace and sets got_bytes = buffer_len, even though the response claimed only 4 bytes of payload; - on a partial netfs sub-request whose iterator is sized to data_len, the over-large copy_folio_to_iter() short-reads, cifs_copy_folioq_to_iter() returns -EIO via the n != len path, and the entire netfs read collapses to -EIO even though the leading sub-requests succeeded. Use data_len for the copy length and for got_bytes so the kernel honours the server-asserted READ payload size. For well-formed servers (where buffer_len == pad_len + data_len) the change is behaviour-equivalent. Cc: stable@vger.kernel.org Signed-off-by: Jeremy Erazo Acked-by: David Howells Signed-off-by: Steve French Signed-off-by: Greg Kroah-Hartman