2022-09-23 7:28 GMT+09:00, Tom Talpey <tom@xxxxxxxxxx>:
> On 9/20/2022 6:20 PM, Namjae Jeon wrote:
>> 2022-09-21 6:05 GMT+09:00, Tom Talpey <tom@xxxxxxxxxx>:
>>> On 9/20/2022 9:20 AM, Namjae Jeon wrote:
>>>> This patch fill missing sids in SMB_FIND_FILE_POSIX_INFO response.
>>>>
>>>> Signed-off-by: Namjae Jeon <linkinjeon@xxxxxxxxxx>
>>>> ---
>>>> fs/ksmbd/smb2pdu.c | 15 ++++++++++++---
>>>> 1 file changed, 12 insertions(+), 3 deletions(-)
>>>>
>>>> diff --git a/fs/ksmbd/smb2pdu.c b/fs/ksmbd/smb2pdu.c
>>>> index 5c797cc09494..9dd6033bc4de 100644
>>>> --- a/fs/ksmbd/smb2pdu.c
>>>> +++ b/fs/ksmbd/smb2pdu.c
>>>> @@ -4717,6 +4717,9 @@ static int find_file_posix_info(struct
>>>> smb2_query_info_rsp *rsp,
>>>> {
>>>> struct smb311_posix_qinfo *file_info;
>>>> struct inode *inode = file_inode(fp->filp);
>>>> + struct user_namespace *user_ns = file_mnt_user_ns(fp->filp);
>>>> + vfsuid_t vfsuid = i_uid_into_vfsuid(user_ns, inode);
>>>> + vfsgid_t vfsgid = i_gid_into_vfsgid(user_ns, inode);
>>>> u64 time;
>>>>
>>>> file_info = (struct smb311_posix_qinfo *)rsp->Buffer;
>>>> @@ -4734,9 +4737,15 @@ static int find_file_posix_info(struct
>>>> smb2_query_info_rsp *rsp,
>>>> file_info->HardLinks = cpu_to_le32(inode->i_nlink);
>>>> file_info->Mode = cpu_to_le32(inode->i_mode & 0777);
>>>> file_info->DeviceId = cpu_to_le32(inode->i_rdev);
>>>> +
>>>> + id_to_sid(from_kuid_munged(&init_user_ns, vfsuid_into_kuid(vfsuid)),
>>>> + SIDUNIX_USER, (struct smb_sid *)&file_info->Sids[0]);
>>>> + id_to_sid(from_kgid_munged(&init_user_ns, vfsgid_into_kgid(vfsgid)),
>>>> + SIDUNIX_GROUP, (struct smb_sid *)&file_info->Sids[16]);
>>>> +
>>>> rsp->OutputBufferLength =
>>>> - cpu_to_le32(sizeof(struct smb311_posix_qinfo));
>>>> - inc_rfc1001_len(rsp_org, sizeof(struct smb311_posix_qinfo));
>>>> + cpu_to_le32(sizeof(struct smb311_posix_qinfo) + 32);
>>>> + inc_rfc1001_len(rsp_org, sizeof(struct smb311_posix_qinfo) + 32);
>>>
>>> These 32's, and the one just below, are really sizeof(sidbuffer), right?
>> Yes.
>>>
>>> Why code it as a raw number?
>> Sids is declared as flexible-array members.
>
> Ugh - worse than that. The smb311_posix_qinfo looks to have
> even more undefined payload:
When I checked the behavior of samba, I figured out that samba does
not return filename and filename_length. Only user/group sids... It
seems that the comment is probably wrong, but there is no
specification documentation and there is no way to check it except
packet analysis.
>
> u8 Sids[];
> /*
> * var sized owner SID
> * var sized group SID
> * le32 filenamelength
> * u8 filename[]
> */
>
> This is pre-existing, nothing your patch should address, but
> does need attention before we attempt to standardize it!!
It can be more updated when standardization is completed, Anyway,
currently I am trying to match the behavior of samba. Please check if
v2 patch is a little better.
Thanks!
>
> MHO anyway.
>
> Tom.
>
>
>>>
>>> Tom.
>>>
>>>> return 0;
>>>> }
>>>>
>>>> @@ -4858,7 +4867,7 @@ static int smb2_get_info_file(struct ksmbd_work
>>>> *work,
>>>> rc = -EOPNOTSUPP;
>>>> } else {
>>>> rc = find_file_posix_info(rsp, fp, work->response_buf);
>>>> - file_infoclass_size = sizeof(struct smb311_posix_qinfo);
>>>> + file_infoclass_size = sizeof(struct smb311_posix_qinfo) + 32;
>>>> }
>>>> break;
>>>> default:
>>>
>>
>
