Never return 1 on failure.
Reviewed-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Namjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
Return zero in always success case.
Reviewed-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Namjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
Set error return value for memcmp() difference.
Reviewed-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Namjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
Remove unneeded type casting.
Reviewed-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Namjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
simplify error handling in ksmbd_auth_ntlm().
Reviewed-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Namjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
Move ret check before the out label.
Reviewed-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Namjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
Just return smbhash() instead of using rc return value.
Reviewed-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Namjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
Move fips_enabled check before the str_to_key().
Reviewed-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Namjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
Add goto fail in neg_token_init_mech_type().
Reviewed-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Namjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
Use memcmp instead of for loop check in oid_eq().
Reviewed-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Namjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
Add goto fail in asn1_oid_decode() to clean-up exception handling code.
Reviewed-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Namjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
Fix build error when CONFIG_OID_REGISTRY is not set:
mips-linux-gnu-ld: fs/cifsd/asn1.o: in function `gssapi_this_mech':
asn1.c:(.text+0xaa0): undefined reference to `sprint_oid'
mips-linux-gnu-ld: fs/cifsd/asn1.o: in function `neg_token_init_mech_type':
asn1.c:(.text+0xbec): undefined reference to `sprint_oid'
Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: Wei Yongjun <weiyongjun1@huawei.com>
Signed-off-by: Namjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
Add support for FSCTL_DUPLICATE_EXTENTS_TO_FILE in smb2 ioctl.
Reviewed-by: Hyunchul Lee <hyc.lee@gmail.com>
Signed-off-by: Namjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
Returning TreeID=0 is valid behaviour according to [MS-SMB2] 2.2.1.2:
TreeId (4 bytes): Uniquely identifies the tree connect for the command.
This MUST be 0 for the SMB2 TREE_CONNECT Request. The TreeId can be
any unsigned 32-bit integer that is received from a previous
SMB2 TREE_CONNECT Response. TreeId SHOULD be set to 0 for the
following commands:
[...]
However, some client implementations reject it as invalid. Windows10
assigns ids starting from 1, and samba4 returns a random uint32_t
which suggests there may be other clients that consider it is
invalid behaviour.
Signed-off-by: Marios Makassikis <mmakassikis@freebox.fr>
Signed-off-by: Namjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
If lock length in smb2 lock request from client is over
flock max length size, lock length is changed to flock max length
and don't return error response.
Signed-off-by: Namjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
fs/cifsd/smb2pdu.c:8098:8-9: WARNING: return of 0/1 in function
'smb2_is_sign_req' with return type bool
Return statements in functions returning bool should use true/false
instead of 1/0.
Generated by: scripts/coccinelle/misc/boolreturn.cocci
Reported-by: kernel test robot <lkp@intel.com>
Signed-off-by: kernel test robot <lkp@intel.com>
Signed-off-by: Namjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
WARNING: Too many leading tabs - consider code refactoring
3066: FILE: fs/cifsd/smb2pdu.c:2733:
+ if (fattr.cf_dacls)
Fix a warning from checkpatch.pl.
Signed-off-by: Namjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
WARNING: Possible unnecessary 'out of memory' message
902: FILE: fs/cifsd/smb2pdu.c:569:
+ if (!work->response_buf) {
+ ksmbd_err("Failed to allocate %zu bytes buffer\n", sz);
Fix a warning from checkpatch.pl.
Signed-off-by: Namjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
Decode gss token of SMB2_SESSSION_SETUP using
lib/asn1_decoder.c
Signed-off-by: Hyunchul Lee <hyc.lee@gmail.com>
Signed-off-by: Namjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
Now that 256 bit encryption can be negotiated, update
names of the nonces to match the updated official protocol
documentation (e.g. AES_GCM_NONCE instead of AES_128GCM_NONCE)
since they apply to both 128 bit and 256 bit encryption.
update smb encryption code to set 32 byte key length and to
set gcm256/ccm256 when requested on mount.
Signed-off-by: Namjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
The user field in the session structure is allocated when the client is
authenticated. If the client explicitly logs off, the user field is freed,
but the session is kept around in case the user reconnects. If the TCP
connection hasn't been closed and the client sends a session setup with
a PreviousSessionId set, destroy_previous_session() will be called to
check if the session needs to be cleaned up.
Signed-off-by: Marios Makassikis <mmakassikis@freebox.fr>
Signed-off-by: Namjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
When processing a SMB2 QUERY_DIRECTORY request,
smb2_populate_readdir_entry() is called first to fill the dot/dotdot
entries. This moves the d_info->wptr pointer but out_buf_len remains
unchanged. As a result, reserve_populate_dentry() may end up writing
past the end of the buffer since the bounds checking is done on
invalid values.
Signed-off-by: Marios Makassikis <mmakassikis@freebox.fr>
Signed-off-by: Namjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
Currently, a SMB2 client does not receive an error message if
ksmbd_session_rpc_open() fails when opening a pipe.
Fix this by responding with STATUS_NO_MEMORY or STATUS_INVALID_PARAMETER
depending on the error that occurred.
Signed-off-by: Marios Makassikis <mmakassikis@freebox.fr>
Signed-off-by: Namjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
a Windows 10 client isn't able to store files from ksmbd servers due
unknown local permission problems (code 0x8007003A) if smb3 encryption
is enabled. Windows 10 is requesting for ATTRIBUTE_SECINFO (mask 0x20)
which is not yet handled by ksmbd, this leads to a invalid response.
For now we just reintroduce the old check to avoid processing of unhandled
flags until ATTRIBUTE_SECINFO is properly handled.
Signed-off-by: Sebastian Gottschall <s.gottschall@dd-wrt.com>
Signed-off-by: Namjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
Inline it in the only place it is used and remove it.
Signed-off-by: Marios Makassikis <mmakassikis@freebox.fr>
Signed-off-by: Namjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
[MS-SMB2] 3.3.5.21.1
If the object store supports security and FileInfoClass is
FileAllocationInformation, FileEndOfFileInformation, or
FileValidDataLengthInformation, and Open.GrantedAccess does not
include FILE_WRITE_DATA, the server MUST fail the request with
STATUS_ACCESS_DENIED.
Signed-off-by: Marios Makassikis <mmakassikis@freebox.fr>
Signed-off-by: Namjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
Remove the dead code of unimplemented durable handle.
Signed-off-by: Namjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
Move nt time functions in netmisc.c to misc.c to remove netmisc.c file.
Signed-off-by: Namjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
smberr.h is a leftover of SMB1. This patch remove unused smberr.h.
Signed-off-by: Namjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
smb2_put_name() is called twice, and both call sites do the IS_ERR() check
before.
Signed-off-by: Marios Makassikis <mmakassikis@freebox.fr>
Signed-off-by: Namjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
__ksmbd_lookup_fd could decrement the reference count of
unclaimed ksmbd_file to 0 but not release this ksmbd_file.
ksmbd_file cannot be unclaimed except ksmbd_close_inode_fds(),
because ksmbd_file is only removed from the m_fp_list list
after the reference count of ksmbd_file becomes 0. And if the
count is 0, __ksmbd_lookup_fd does not use ksmbd_file found
from idr due to atomic_inc_not_zero.
Signed-off-by: Hyunchul Lee <hyc.lee@gmail.com>
Signed-off-by: Namjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
re-implement ksmbd_vfs_kern_path() to change
recursion to iteration.
Signed-off-by: Hyunchul Lee <hyc.lee@gmail.com>
Signed-off-by: Namjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
To remove the file, We have parsed full pathname to divide parent path and
filename. It is a better way to get parent dentry from child dentry that
obtained by lookup with given pathname.
Signed-off-by: Namjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
This patch add the check if parent is stable by unexpected rename.
Signed-off-by: Hyunchul Lee <hyc.lee@gmail.com>
Signed-off-by: Namjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
Matthew pointed out that embedding struct ida into the struct is
better than having a pointer to it.
This patch initialise it statically using DEFINE_IDA() or ida_init()
and remove ksmbd_ida_alloc/free().
Signed-off-by: Namjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
Remove including <linux/version.h> that don't need it.
Signed-off-by: Tian Tao <tiantao6@hisilicon.com>
Signed-off-by: Zhiqi Song <songzhiqi1@huawei.com>
Signed-off-by: Namjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
hulk-robot following build error:
fs/cifsd/transport_rdma.c: In function 'read_write_done':
fs/cifsd/transport_rdma.c:1297:2: error: implicit declaration of
function 'sg_free_table_chained'
[-Werror=implicit-function-declaration]
1297 | sg_free_table_chained(&msg->sgt, SG_CHUNK_SIZE);
The reason is CONFIG_SG_POOL is not enabled in the config, to
avoid such failure, select SG_POOL in Kconfig for SMB_SERVER_SMBDIRECT.
Signed-off-by: Zhang Xiaoxu <zhangxiaoxu5@huawei.com>
Signed-off-by: Namjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
Dan Carpenter pointed out that there there is a possibility of
integer overflow. This patch prevent a integer overflow in wm_alloc().
Signed-off-by: Namjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
fs/cifsd/smb2pdu.c:1177:27-34: WARNING opportunity for kmemdup
Use kmemdup rather than duplicating its implementation
Generated by: scripts/coccinelle/api/memdup.cocci
Reported-by: kernel test robot <lkp@intel.com>
Signed-off-by: kernel test robot <lkp@intel.com>
Signed-off-by: Namjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
kfree should be used to free memory allocated by kmalloc or kzalloc to
avoid any overhead and for maintaining consistency.
Signed-off-by: Muhammad Usama Anjum <musamaanjum@gmail.com>
Signed-off-by: Namjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
Memory is being allocated and if veto_list is zero, the loop breaks
without cleaning up the allocated memory. In this patch, the length
check has been moved before allocation. If loop breaks, the memory isn't
allocated in the first place. Thus the memory is being protected from
leaking.
Reported-by: coverity-bot <keescook+coverity-bot@chromium.org>
Addresses-Coverity-ID: 1503590 ("Resource leaks")
Signed-off-by: Muhammad Usama Anjum <musamaanjum@gmail.com>
Signed-off-by: Namjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
Remove unused function prototype and variables.
Signed-off-by: Gibeom Kim <gibeomii.kim@samsung.com>
Signed-off-by: Namjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
Matthew suggest to change linked list of tree connect list to xarray.
It will be tree connect lookup in O(log(n)) time instead of O(n) time.
Signed-off-by: Namjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
