1.. SPDX-License-Identifier: GPL-2.0 2 3================================= 4The Linux NTFS filesystem driver 5================================= 6 7 8.. Table of contents 9 10 - Overview 11 - Utilities support 12 - Linux file attributes 13 - Supported mount options 14 15 16Overview 17======== 18 19NTFS is a Linux kernel filesystem driver that provides full read and write 20support for NTFS volumes. It is designed for high performance, modern 21kernel infrastructure (iomap, folio), and stable long-term maintenance. 22 23 24Utilities support 25================= 26 27The NTFS utilities project, called ntfsprogs-plus, provides mkfs.ntfs, 28fsck.ntfs, and other related tools (e.g., ntfsinfo, ntfsclone, etc.) for 29creating, checking, and managing NTFS volumes. These utilities can be used 30for filesystem testing with xfstests as well as for recovering corrupted 31NTFS devices. 32 33The project is available at: 34 35 https://github.com/ntfsprogs-plus/ntfsprogs-plus 36 37 38Linux file attributes 39===================== 40 41The driver supports lsattr(1) and chattr(1). Immutable, append-only, nodump 42and noatime are stored in a private ``$LINUXFLAGS`` extended attribute within the 43standard NTFS ``$EA``/``$EA_INFORMATION`` attributes. Its value is a 4432-bit little-endian bitmask: bit 0 is immutable, bit 1 is append-only, 45bit 2 is nodump, and bit 3 is noatime. Unknown bits are preserved when updating 46the flags; the entry is removed when the entire value becomes zero. These settings 47survive inode eviction, unmount and reboot. Direct writes to this EA 48through setxattr(2) or removexattr(2) are rejected; use chattr(1) instead. 49 50The noatime attribute (``chattr +A``) suppresses automatic access-time 51updates for the inode, including when mounted with ``strictatime``. It 52applies to directories as well as files, but is not inherited by newly 53created children. Use ``chattr -A`` to restore the mount's access-time policy. 54 55System metadata files and files protected by the ``sys_immutable`` mount 56option cannot have their immutable protection cleared with chattr(1). 57Protection derived solely from the mount option is not stored in the EA. 58Case folding is reported according to the mount options and cannot be 59configured per file. Compression and encryption are reported but cannot 60be changed through chattr(1). 61 62These Linux flags are enforced by Linux. Storing them in an NTFS EA does 63not make Windows enforce immutable, append-only, nodump or noatime semantics. 64 65 66Supported mount options 67======================= 68 69The NTFS driver supports the following mount options: 70 71======================= ==================================================== 72iocharset=name Character set to use for converting between 73 the encoding is used for user visible filename and 74 16 bit Unicode characters. 75 76nls=name Deprecated option. Still supported but please use 77 iocharset=name in the future. 78 79uid= 80gid= 81umask= Provide default owner, group, and access mode mask. 82 These options work as documented in mount(8). By 83 default, the files/directories are owned by root 84 and he/she has read and write permissions, as well 85 as browse permission for directories. No one else 86 has any access permissions. I.e. the mode on all 87 files is by default rw------- and 88 for directories rwx------, a consequence of 89 the default fmask=0177 and dmask=0077. 90 Using a umask of zero will grant all permissions to 91 everyone, i.e. all files and directories will have 92 mode rwxrwxrwx. 93 94fmask= 95dmask= Instead of specifying umask which applies both to 96 files and directories, fmask applies only to files 97 and dmask only to directories. 98 99showmeta=<BOOL> 100show_sys_files=<BOOL> If show_sys_files is specified, show the system 101 files in directory listings. Otherwise the default 102 behaviour is to hide the system files. 103 Note that even when show_sys_files is specified, 104 "$MFT" will not be visible due to bugs/mis-features 105 in glibc. Further, note that irrespective of 106 show_sys_files, all files are accessible by name, 107 i.e. you can always do "ls -l \$UpCase" for example 108 to specifically show the system file containing 109 the Unicode upcase table. 110 111case_sensitive=<BOOL> If case_sensitive is specified, treat all filenames 112 as case sensitive and create file names in 113 the POSIX namespace (default behavior). Note, 114 the Linux NTFS driver will never create short 115 filenames and will remove them on rename/delete of 116 the corresponding long file name. Note that files 117 remain accessible via their short file name, if it 118 exists. 119 120nocase=<BOOL> If nocase is specified, treat filenames 121 case-insensitively. 122 123disable_sparse=<BOOL> If disable_sparse is specified, creation of sparse 124 regions, i.e. holes, inside files is disabled for 125 the volume (for the duration of this mount only). 126 By default, creation of sparse regions is enabled, 127 which is consistent with the behaviour of 128 traditional Unix filesystems. 129 130errors=opt Specify NTFS behavior on critical errors: panic, 131 remount the partition in read-only mode or 132 continue without doing anything (default behavior). 133 134mft_zone_multiplier= Set the MFT zone multiplier for the volume (this 135 setting is not persistent across mounts and can be 136 changed from mount to mount but cannot be changed 137 on remount). Values of 1 to 4 are allowed, 1 being 138 the default. The MFT zone multiplier determines 139 how much space is reserved for the MFT on the 140 volume. If all other space is used up, then the 141 MFT zone will be shrunk dynamically, so this has no 142 impact on the amount of free space. However, it 143 can have an impact on performance by affecting 144 fragmentation of the MFT. In general use the 145 default. If you have a lot of small files then use 146 a higher value. The values have the following 147 meaning: 148 149 ===== ================================= 150 Value MFT zone size (% of volume size) 151 ===== ================================= 152 1 12.5% 153 2 25% 154 3 37.5% 155 4 50% 156 ===== ================================= 157 158 Note this option is irrelevant for read-only mount. 159 160preallocated_size= Set preallocated size to optimize runlist merge 161 overhead with small chunck size.(64KB size by 162 default) 163 164acl=<BOOL> Enable POSIX ACL support. When specified, POSIX 165 ACLs stored in extended attributes are enforced. 166 Default is off. Requires kernel config 167 NTFS_FS_POSIX_ACL enabled. 168 169sys_immutable=<BOOL> Make NTFS system files (e.g. $MFT, $LogFile, 170 $Bitmap, $UpCase, etc.) immutable to user initiated 171 modifications for extra safety. Default is off. 172 173nohidden=<BOOL> Hide files and directories marked with the Windows 174 "hidden" attribute. By default hidden items are 175 shown. 176 177hide_dot_files=<BOOL> Hide names beginning with a dot ("."). By default 178 dot files are shown. When enabled, files and 179 directories created with a leading '.' will be 180 hidden from directory listings. 181 182windows_names=<BOOL> Refuse creation/rename of files with characters or 183 reserved device names disallowed on Windows (e.g. 184 CON, NUL, AUX, COM1, LPT1, etc.). Default is off. 185discard=<BOOL> Issue block device discard for clusters freed on 186 file deletion/truncation to inform underlying 187 storage. 188 189native_symlink=raw|rel Configure how absolute symbolic links and mount 190 points (junctions) are handled. Under "raw" 191 (default), the absolute target path is returned 192 as-is without translation. Under "rel", it is 193 rewritten as a relative path anchored at 194 the volume root. 195 196symlink=wsl|native Configure how symbolic links are created. Under 197 "wsl" (default), WSL (Windows Subsystem for 198 Linux) compatible symlinks are created. Under 199 "native", Windows native symbolic links are 200 created. 201 202streams_interface= Select the named-stream pathname interface: 203 "none" (default) treats ':' as an ordinary filename 204 character, while "windows" interprets the final 205 component of a path such as "file:stream" as a named 206 $DATA stream. In "windows" mode, filenames containing 207 ':' are hidden from directory listings and cannot be 208 addressed through the pathname interface. Only the 209 file:stream form is recognized. Typed forms such 210 as file::$DATA and file:stream:$DATA are not 211 supported. The ioctl interface is available in both 212 modes. The base file or directory must already exist 213 before a pathname stream can be created. Creating 214 file:stream does not create file. Removing a stream 215 while it is open fails with -EBUSY. Removing the base 216 file while a stream is open detaches the stream from 217 the namespace; NTFS has no orphan-stream list, so a 218 crash can leave that stream's data unreachable on 219 disk. 220======================= ==================================================== 221 222Named-stream ioctls 223=================== 224 225The named-stream ioctl interface is available regardless of 226streams_interface. Open the base regular file or directory and use 227NTFS_IOC_STREAM_READ and NTFS_IOC_STREAM_WRITE with a 228struct ntfs_stream request. The request contains the stream name, 229byte offset, and transfer length. Write requests create the stream when it 230does not exist. Writes are not atomic. Error can leave partial data 231changes, an extended stream size, or a newly created stream. bytes_returned 232is valid only on success. Failed write does not report a completed byte 233count. Use NTFS_IOC_STREAM_REMOVE to remove a stream and NTFS_IOC_LIST_STREAMS 234to enumerate streams. Stream names are encoded with 235the mounted NLS by default. Set NTFS_STREAM_FL_UTF16 to pass or receive 236raw UTF-16LE names. These ioctls transfer stream data directly and do not 237return a separate stream file descriptor. 238