diff options
Diffstat (limited to 'Documentation/filesystems')
19 files changed, 1330 insertions, 1986 deletions
diff --git a/Documentation/filesystems/cifs/AUTHORS b/Documentation/filesystems/cifs/AUTHORS deleted file mode 100644 index 75865da2ce14..000000000000 --- a/Documentation/filesystems/cifs/AUTHORS +++ /dev/null @@ -1,63 +0,0 @@ -Original Author -=============== -Steve French (sfrench@samba.org) - -The author wishes to express his appreciation and thanks to: -Andrew Tridgell (Samba team) for his early suggestions about smb/cifs VFS -improvements. Thanks to IBM for allowing me time and test resources to pursue -this project, to Jim McDonough from IBM (and the Samba Team) for his help, to -the IBM Linux JFS team for explaining many esoteric Linux filesystem features. -Jeremy Allison of the Samba team has done invaluable work in adding the server -side of the original CIFS Unix extensions and reviewing and implementing -portions of the newer CIFS POSIX extensions into the Samba 3 file server. Thank -Dave Boutcher of IBM Rochester (author of the OS/400 smb/cifs filesystem client) -for proving years ago that very good smb/cifs clients could be done on Unix-like -operating systems. Volker Lendecke, Andrew Tridgell, Urban Widmark, John -Newbigin and others for their work on the Linux smbfs module. Thanks to -the other members of the Storage Network Industry Association CIFS Technical -Workgroup for their work specifying this highly complex protocol and finally -thanks to the Samba team for their technical advice and encouragement. - -Patch Contributors ------------------- -Zwane Mwaikambo -Andi Kleen -Amrut Joshi -Shobhit Dayal -Sergey Vlasov -Richard Hughes -Yury Umanets -Mark Hamzy (for some of the early cifs IPv6 work) -Domen Puncer -Jesper Juhl (in particular for lots of whitespace/formatting cleanup) -Vince Negri and Dave Stahl (for finding an important caching bug) -Adrian Bunk (kcalloc cleanups) -Miklos Szeredi -Kazeon team for various fixes especially for 2.4 version. -Asser Ferno (Change Notify support) -Shaggy (Dave Kleikamp) for innumerable small fs suggestions and some good cleanup -Gunter Kukkukk (testing and suggestions for support of old servers) -Igor Mammedov (DFS support) -Jeff Layton (many, many fixes, as well as great work on the cifs Kerberos code) -Scott Lovenberg -Pavel Shilovsky (for great work adding SMB2 support, and various SMB3 features) -Aurelien Aptel (for DFS SMB3 work and some key bug fixes) -Ronnie Sahlberg (for SMB3 xattr work, bug fixes, and lots of great work on compounding) -Shirish Pargaonkar (for many ACL patches over the years) -Sachin Prabhu (many bug fixes, including for reconnect, copy offload and security) -Paulo Alcantara -Long Li (some great work on RDMA, SMB Direct) - - -Test case and Bug Report contributors -------------------------------------- -Thanks to those in the community who have submitted detailed bug reports -and debug of problems they have found: Jochen Dolze, David Blaine, -Rene Scharfe, Martin Josefsson, Alexander Wild, Anthony Liguori, -Lars Muller, Urban Widmark, Massimiliano Ferrero, Howard Owen, -Olaf Kirch, Kieron Briggs, Nick Millington and others. Also special -mention to the Stanford Checker (SWAT) which pointed out many minor -bugs in error paths. Valuable suggestions also have come from Al Viro -and Dave Miller. - -And thanks to the IBM LTC and Power test teams and SuSE and Citrix and RedHat testers for finding multiple bugs during excellent stress test runs. diff --git a/Documentation/filesystems/cifs/CHANGES b/Documentation/filesystems/cifs/CHANGES deleted file mode 100644 index 1df7f4910eb2..000000000000 --- a/Documentation/filesystems/cifs/CHANGES +++ /dev/null @@ -1,4 +0,0 @@ -See https://wiki.samba.org/index.php/LinuxCIFSKernel for summary -information (that may be easier to read than parsing the output of -"git log fs/cifs") about fixes/improvements to CIFS/SMB2/SMB3 support (changes -to cifs.ko module) by kernel version (and cifs internal module version). diff --git a/Documentation/filesystems/cifs/README b/Documentation/filesystems/cifs/README deleted file mode 100644 index 4a804619cff2..000000000000 --- a/Documentation/filesystems/cifs/README +++ /dev/null @@ -1,743 +0,0 @@ -This module supports the SMB3 family of advanced network protocols (as well -as older dialects, originally called "CIFS" or SMB1). - -The CIFS VFS module for Linux supports many advanced network filesystem -features such as hierarchical DFS like namespace, hardlinks, locking and more. -It was designed to comply with the SNIA CIFS Technical Reference (which -supersedes the 1992 X/Open SMB Standard) as well as to perform best practice -practical interoperability with Windows 2000, Windows XP, Samba and equivalent -servers. This code was developed in participation with the Protocol Freedom -Information Foundation. CIFS and now SMB3 has now become a defacto -standard for interoperating between Macs and Windows and major NAS appliances. - -Please see - MS-SMB2 (for detailed SMB2/SMB3/SMB3.1.1 protocol specification) - http://protocolfreedom.org/ and - http://samba.org/samba/PFIF/ -for more details. - - -For questions or bug reports please contact: - smfrench@gmail.com - -See the project page at: https://wiki.samba.org/index.php/LinuxCIFS_utils - -Build instructions: -================== -For Linux: -1) Download the kernel (e.g. from http://www.kernel.org) -and change directory into the top of the kernel directory tree -(e.g. /usr/src/linux-2.5.73) -2) make menuconfig (or make xconfig) -3) select cifs from within the network filesystem choices -4) save and exit -5) make - - -Installation instructions: -========================= -If you have built the CIFS vfs as module (successfully) simply -type "make modules_install" (or if you prefer, manually copy the file to -the modules directory e.g. /lib/modules/2.4.10-4GB/kernel/fs/cifs/cifs.ko). - -If you have built the CIFS vfs into the kernel itself, follow the instructions -for your distribution on how to install a new kernel (usually you -would simply type "make install"). - -If you do not have the utility mount.cifs (in the Samba 4.x source tree and on -the CIFS VFS web site) copy it to the same directory in which mount helpers -reside (usually /sbin). Although the helper software is not -required, mount.cifs is recommended. Most distros include a "cifs-utils" -package that includes this utility so it is recommended to install this. - -Note that running the Winbind pam/nss module (logon service) on all of your -Linux clients is useful in mapping Uids and Gids consistently across the -domain to the proper network user. The mount.cifs mount helper can be -found at cifs-utils.git on git.samba.org - -If cifs is built as a module, then the size and number of network buffers -and maximum number of simultaneous requests to one server can be configured. -Changing these from their defaults is not recommended. By executing modinfo - modinfo kernel/fs/cifs/cifs.ko -on kernel/fs/cifs/cifs.ko the list of configuration changes that can be made -at module initialization time (by running insmod cifs.ko) can be seen. - -Recommendations -=============== -To improve security the SMB2.1 dialect or later (usually will get SMB3) is now -the new default. To use old dialects (e.g. to mount Windows XP) use "vers=1.0" -on mount (or vers=2.0 for Windows Vista). Note that the CIFS (vers=1.0) is -much older and less secure than the default dialect SMB3 which includes -many advanced security features such as downgrade attack detection -and encrypted shares and stronger signing and authentication algorithms. -There are additional mount options that may be helpful for SMB3 to get -improved POSIX behavior (NB: can use vers=3.0 to force only SMB3, never 2.1): - "mfsymlinks" and "cifsacl" and "idsfromsid" - -Allowing User Mounts -==================== -To permit users to mount and unmount over directories they own is possible -with the cifs vfs. A way to enable such mounting is to mark the mount.cifs -utility as suid (e.g. "chmod +s /sbin/mount.cifs). To enable users to -umount shares they mount requires -1) mount.cifs version 1.4 or later -2) an entry for the share in /etc/fstab indicating that a user may -unmount it e.g. -//server/usersharename /mnt/username cifs user 0 0 - -Note that when the mount.cifs utility is run suid (allowing user mounts), -in order to reduce risks, the "nosuid" mount flag is passed in on mount to -disallow execution of an suid program mounted on the remote target. -When mount is executed as root, nosuid is not passed in by default, -and execution of suid programs on the remote target would be enabled -by default. This can be changed, as with nfs and other filesystems, -by simply specifying "nosuid" among the mount options. For user mounts -though to be able to pass the suid flag to mount requires rebuilding -mount.cifs with the following flag: CIFS_ALLOW_USR_SUID - -There is a corresponding manual page for cifs mounting in the Samba 3.0 and -later source tree in docs/manpages/mount.cifs.8 - -Allowing User Unmounts -====================== -To permit users to ummount directories that they have user mounted (see above), -the utility umount.cifs may be used. It may be invoked directly, or if -umount.cifs is placed in /sbin, umount can invoke the cifs umount helper -(at least for most versions of the umount utility) for umount of cifs -mounts, unless umount is invoked with -i (which will avoid invoking a umount -helper). As with mount.cifs, to enable user unmounts umount.cifs must be marked -as suid (e.g. "chmod +s /sbin/umount.cifs") or equivalent (some distributions -allow adding entries to a file to the /etc/permissions file to achieve the -equivalent suid effect). For this utility to succeed the target path -must be a cifs mount, and the uid of the current user must match the uid -of the user who mounted the resource. - -Also note that the customary way of allowing user mounts and unmounts is -(instead of using mount.cifs and unmount.cifs as suid) to add a line -to the file /etc/fstab for each //server/share you wish to mount, but -this can become unwieldy when potential mount targets include many -or unpredictable UNC names. - -Samba Considerations -==================== -Most current servers support SMB2.1 and SMB3 which are more secure, -but there are useful protocol extensions for the older less secure CIFS -dialect, so to get the maximum benefit if mounting using the older dialect -(CIFS/SMB1), we recommend using a server that supports the SNIA CIFS -Unix Extensions standard (e.g. almost any version of Samba ie version -2.2.5 or later) but the CIFS vfs works fine with a wide variety of CIFS servers. -Note that uid, gid and file permissions will display default values if you do -not have a server that supports the Unix extensions for CIFS (such as Samba -2.2.5 or later). To enable the Unix CIFS Extensions in the Samba server, add -the line: - - unix extensions = yes - -to your smb.conf file on the server. Note that the following smb.conf settings -are also useful (on the Samba server) when the majority of clients are Unix or -Linux: - - case sensitive = yes - delete readonly = yes - ea support = yes - -Note that server ea support is required for supporting xattrs from the Linux -cifs client, and that EA support is present in later versions of Samba (e.g. -3.0.6 and later (also EA support works in all versions of Windows, at least to -shares on NTFS filesystems). Extended Attribute (xattr) support is an optional -feature of most Linux filesystems which may require enabling via -make menuconfig. Client support for extended attributes (user xattr) can be -disabled on a per-mount basis by specifying "nouser_xattr" on mount. - -The CIFS client can get and set POSIX ACLs (getfacl, setfacl) to Samba servers -version 3.10 and later. Setting POSIX ACLs requires enabling both XATTR and -then POSIX support in the CIFS configuration options when building the cifs -module. POSIX ACL support can be disabled on a per mount basic by specifying -"noacl" on mount. - -Some administrators may want to change Samba's smb.conf "map archive" and -"create mask" parameters from the default. Unless the create mask is changed -newly created files can end up with an unnecessarily restrictive default mode, -which may not be what you want, although if the CIFS Unix extensions are -enabled on the server and client, subsequent setattr calls (e.g. chmod) can -fix the mode. Note that creating special devices (mknod) remotely -may require specifying a mkdev function to Samba if you are not using -Samba 3.0.6 or later. For more information on these see the manual pages -("man smb.conf") on the Samba server system. Note that the cifs vfs, -unlike the smbfs vfs, does not read the smb.conf on the client system -(the few optional settings are passed in on mount via -o parameters instead). -Note that Samba 2.2.7 or later includes a fix that allows the CIFS VFS to delete -open files (required for strict POSIX compliance). Windows Servers already -supported this feature. Samba server does not allow symlinks that refer to files -outside of the share, so in Samba versions prior to 3.0.6, most symlinks to -files with absolute paths (ie beginning with slash) such as: - ln -s /mnt/foo bar -would be forbidden. Samba 3.0.6 server or later includes the ability to create -such symlinks safely by converting unsafe symlinks (ie symlinks to server -files that are outside of the share) to a samba specific format on the server -that is ignored by local server applications and non-cifs clients and that will -not be traversed by the Samba server). This is opaque to the Linux client -application using the cifs vfs. Absolute symlinks will work to Samba 3.0.5 or -later, but only for remote clients using the CIFS Unix extensions, and will -be invisbile to Windows clients and typically will not affect local -applications running on the same server as Samba. - -Use instructions: -================ -Once the CIFS VFS support is built into the kernel or installed as a module -(cifs.ko), you can use mount syntax like the following to access Samba or -Mac or Windows servers: - - mount -t cifs //9.53.216.11/e$ /mnt -o username=myname,password=mypassword - -Before -o the option -v may be specified to make the mount.cifs -mount helper display the mount steps more verbosely. -After -o the following commonly used cifs vfs specific options -are supported: - - username=<username> - password=<password> - domain=<domain name> - -Other cifs mount options are described below. Use of TCP names (in addition to -ip addresses) is available if the mount helper (mount.cifs) is installed. If -you do not trust the server to which are mounted, or if you do not have -cifs signing enabled (and the physical network is insecure), consider use -of the standard mount options "noexec" and "nosuid" to reduce the risk of -running an altered binary on your local system (downloaded from a hostile server -or altered by a hostile router). - -Although mounting using format corresponding to the CIFS URL specification is -not possible in mount.cifs yet, it is possible to use an alternate format -for the server and sharename (which is somewhat similar to NFS style mount -syntax) instead of the more widely used UNC format (i.e. \\server\share): - mount -t cifs tcp_name_of_server:share_name /mnt -o user=myname,pass=mypasswd - -When using the mount helper mount.cifs, passwords may be specified via alternate -mechanisms, instead of specifying it after -o using the normal "pass=" syntax -on the command line: -1) By including it in a credential file. Specify credentials=filename as one -of the mount options. Credential files contain two lines - username=someuser - password=your_password -2) By specifying the password in the PASSWD environment variable (similarly -the user name can be taken from the USER environment variable). -3) By specifying the password in a file by name via PASSWD_FILE -4) By specifying the password in a file by file descriptor via PASSWD_FD - -If no password is provided, mount.cifs will prompt for password entry - -Restrictions -============ -Servers must support either "pure-TCP" (port 445 TCP/IP CIFS connections) or RFC -1001/1002 support for "Netbios-Over-TCP/IP." This is not likely to be a -problem as most servers support this. - -Valid filenames differ between Windows and Linux. Windows typically restricts -filenames which contain certain reserved characters (e.g.the character : -which is used to delimit the beginning of a stream name by Windows), while -Linux allows a slightly wider set of valid characters in filenames. Windows -servers can remap such characters when an explicit mapping is specified in -the Server's registry. Samba starting with version 3.10 will allow such -filenames (ie those which contain valid Linux characters, which normally -would be forbidden for Windows/CIFS semantics) as long as the server is -configured for Unix Extensions (and the client has not disabled -/proc/fs/cifs/LinuxExtensionsEnabled). In addition the mount option -"mapposix" can be used on CIFS (vers=1.0) to force the mapping of -illegal Windows/NTFS/SMB characters to a remap range (this mount parm -is the default for SMB3). This remap ("mapposix") range is also -compatible with Mac (and "Services for Mac" on some older Windows). - -CIFS VFS Mount Options -====================== -A partial list of the supported mount options follows: - username The user name to use when trying to establish - the CIFS session. - password The user password. If the mount helper is - installed, the user will be prompted for password - if not supplied. - ip The ip address of the target server - unc The target server Universal Network Name (export) to - mount. - domain Set the SMB/CIFS workgroup name prepended to the - username during CIFS session establishment - forceuid Set the default uid for inodes to the uid - passed in on mount. For mounts to servers - which do support the CIFS Unix extensions, such as a - properly configured Samba server, the server provides - the uid, gid and mode so this parameter should not be - specified unless the server and clients uid and gid - numbering differ. If the server and client are in the - same domain (e.g. running winbind or nss_ldap) and - the server supports the Unix Extensions then the uid - and gid can be retrieved from the server (and uid - and gid would not have to be specified on the mount. - For servers which do not support the CIFS Unix - extensions, the default uid (and gid) returned on lookup - of existing files will be the uid (gid) of the person - who executed the mount (root, except when mount.cifs - is configured setuid for user mounts) unless the "uid=" - (gid) mount option is specified. Also note that permission - checks (authorization checks) on accesses to a file occur - at the server, but there are cases in which an administrator - may want to restrict at the client as well. For those - servers which do not report a uid/gid owner - (such as Windows), permissions can also be checked at the - client, and a crude form of client side permission checking - can be enabled by specifying file_mode and dir_mode on - the client. (default) - forcegid (similar to above but for the groupid instead of uid) (default) - noforceuid Fill in file owner information (uid) by requesting it from - the server if possible. With this option, the value given in - the uid= option (on mount) will only be used if the server - can not support returning uids on inodes. - noforcegid (similar to above but for the group owner, gid, instead of uid) - uid Set the default uid for inodes, and indicate to the - cifs kernel driver which local user mounted. If the server - supports the unix extensions the default uid is - not used to fill in the owner fields of inodes (files) - unless the "forceuid" parameter is specified. - gid Set the default gid for inodes (similar to above). - file_mode If CIFS Unix extensions are not supported by the server - this overrides the default mode for file inodes. - fsc Enable local disk caching using FS-Cache (off by default). This - option could be useful to improve performance on a slow link, - heavily loaded server and/or network where reading from the - disk is faster than reading from the server (over the network). - This could also impact scalability positively as the - number of calls to the server are reduced. However, local - caching is not suitable for all workloads for e.g. read-once - type workloads. So, you need to consider carefully your - workload/scenario before using this option. Currently, local - disk caching is functional for CIFS files opened as read-only. - dir_mode If CIFS Unix extensions are not supported by the server - this overrides the default mode for directory inodes. - port attempt to contact the server on this tcp port, before - trying the usual ports (port 445, then 139). - iocharset Codepage used to convert local path names to and from - Unicode. Unicode is used by default for network path - names if the server supports it. If iocharset is - not specified then the nls_default specified - during the local client kernel build will be used. - If server does not support Unicode, this parameter is - unused. - rsize default read size (usually 16K). The client currently - can not use rsize larger than CIFSMaxBufSize. CIFSMaxBufSize - defaults to 16K and may be changed (from 8K to the maximum - kmalloc size allowed by your kernel) at module install time - for cifs.ko. Setting CIFSMaxBufSize to a very large value - will cause cifs to use more memory and may reduce performance - in some cases. To use rsize greater than 127K (the original - cifs protocol maximum) also requires that the server support - a new Unix Capability flag (for very large read) which some - newer servers (e.g. Samba 3.0.26 or later) do. rsize can be - set from a minimum of 2048 to a maximum of 130048 (127K or - CIFSMaxBufSize, whichever is smaller) - wsize default write size (default 57344) - maximum wsize currently allowed by CIFS is 57344 (fourteen - 4096 byte pages) - actimeo=n attribute cache timeout in seconds (default 1 second). - After this timeout, the cifs client requests fresh attribute - information from the server. This option allows to tune the - attribute cache timeout to suit the workload needs. Shorter - timeouts mean better the cache coherency, but increased number - of calls to the server. Longer timeouts mean reduced number - of calls to the server at the expense of less stricter cache - coherency checks (i.e. incorrect attribute cache for a short - period of time). - rw mount the network share read-write (note that the - server may still consider the share read-only) - ro mount network share read-only - version used to distinguish different versions of the - mount helper utility (not typically needed) - sep if first mount option (after the -o), overrides - the comma as the separator between the mount - parms. e.g. - -o user=myname,password=mypassword,domain=mydom - could be passed instead with period as the separator by - -o sep=.user=myname.password=mypassword.domain=mydom - this might be useful when comma is contained within username - or password or domain. This option is less important - when the cifs mount helper cifs.mount (version 1.1 or later) - is used. - nosuid Do not allow remote executables with the suid bit - program to be executed. This is only meaningful for mounts - to servers such as Samba which support the CIFS Unix Extensions. - If you do not trust the servers in your network (your mount - targets) it is recommended that you specify this option for - greater security. - exec Permit execution of binaries on the mount. - noexec Do not permit execution of binaries on the mount. - dev Recognize block devices on the remote mount. - nodev Do not recognize devices on the remote mount. - suid Allow remote files on this mountpoint with suid enabled to - be executed (default for mounts when executed as root, - nosuid is default for user mounts). - credentials Although ignored by the cifs kernel component, it is used by - the mount helper, mount.cifs. When mount.cifs is installed it - opens and reads the credential file specified in order - to obtain the userid and password arguments which are passed to - the cifs vfs. - guest Although ignored by the kernel component, the mount.cifs - mount helper will not prompt the user for a password - if guest is specified on the mount options. If no - password is specified a null password will be used. - perm Client does permission checks (vfs_permission check of uid - and gid of the file against the mode and desired operation), - Note that this is in addition to the normal ACL check on the - target machine done by the server software. - Client permission checking is enabled by default. - noperm Client does not do permission checks. This can expose - files on this mount to access by other users on the local - client system. It is typically only needed when the server - supports the CIFS Unix Extensions but the UIDs/GIDs on the - client and server system do not match closely enough to allow - access by the user doing the mount, but it may be useful with - non CIFS Unix Extension mounts for cases in which the default - mode is specified on the mount but is not to be enforced on the - client (e.g. perhaps when MultiUserMount is enabled) - Note that this does not affect the normal ACL check on the - target machine done by the server software (of the server - ACL against the user name provided at mount time). - serverino Use server's inode numbers instead of generating automatically - incrementing inode numbers on the client. Although this will - make it easier to spot hardlinked files (as they will have - the same inode numbers) and inode numbers may be persistent, - note that the server does not guarantee that the inode numbers - are unique if multiple server side mounts are exported under a - single share (since inode numbers on the servers might not - be unique if multiple filesystems are mounted under the same - shared higher level directory). Note that some older - (e.g. pre-Windows 2000) do not support returning UniqueIDs - or the CIFS Unix Extensions equivalent and for those - this mount option will have no effect. Exporting cifs mounts - under nfsd requires this mount option on the cifs mount. - This is now the default if server supports the - required network operation. - noserverino Client generates inode numbers (rather than using the actual one - from the server). These inode numbers will vary after - unmount or reboot which can confuse some applications, - but not all server filesystems support unique inode - numbers. - setuids If the CIFS Unix extensions are negotiated with the server - the client will attempt to set the effective uid and gid of - the local process on newly created files, directories, and - devices (create, mkdir, mknod). If the CIFS Unix Extensions - are not negotiated, for newly created files and directories - instead of using the default uid and gid specified on - the mount, cache the new file's uid and gid locally which means - that the uid for the file can change when the inode is - reloaded (or the user remounts the share). - nosetuids The client will not attempt to set the uid and gid on - on newly created files, directories, and devices (create, - mkdir, mknod) which will result in the server setting the - uid and gid to the default (usually the server uid of the - user who mounted the share). Letting the server (rather than - the client) set the uid and gid is the default. If the CIFS - Unix Extensions are not negotiated then the uid and gid for - new files will appear to be the uid (gid) of the mounter or the - uid (gid) parameter specified on the mount. - netbiosname When mounting to servers via port 139, specifies the RFC1001 - source name to use to represent the client netbios machine - name when doing the RFC1001 netbios session initialize. - direct Do not do inode data caching on files opened on this mount. - This precludes mmapping files on this mount. In some cases - with fast networks and little or no caching benefits on the - client (e.g. when the application is doing large sequential - reads bigger than page size without rereading the same data) - this can provide better performance than the default - behavior which caches reads (readahead) and writes - (writebehind) through the local Linux client pagecache - if oplock (caching token) is granted and held. Note that - direct allows write operations larger than page size - to be sent to the server. - strictcache Use for switching on strict cache mode. In this mode the - client read from the cache all the time it has Oplock Level II, - otherwise - read from the server. All written data are stored - in the cache, but if the client doesn't have Exclusive Oplock, - it writes the data to the server. - rwpidforward Forward pid of a process who opened a file to any read or write - operation on that file. This prevent applications like WINE - from failing on read and write if we use mandatory brlock style. - acl Allow setfacl and getfacl to manage posix ACLs if server - supports them. (default) - noacl Do not allow setfacl and getfacl calls on this mount - user_xattr Allow getting and setting user xattrs (those attributes whose - name begins with "user." or "os2.") as OS/2 EAs (extended - attributes) to the server. This allows support of the - setfattr and getfattr utilities. (default) - nouser_xattr Do not allow getfattr/setfattr to get/set/list xattrs - mapchars Translate six of the seven reserved characters (not backslash) - *?<>|: - to the remap range (above 0xF000), which also - allows the CIFS client to recognize files created with - such characters by Windows's POSIX emulation. This can - also be useful when mounting to most versions of Samba - (which also forbids creating and opening files - whose names contain any of these seven characters). - This has no effect if the server does not support - Unicode on the wire. - nomapchars Do not translate any of these seven characters (default). - nocase Request case insensitive path name matching (case - sensitive is the default if the server supports it). - (mount option "ignorecase" is identical to "nocase") - posixpaths If CIFS Unix extensions are supported, attempt to - negotiate posix path name support which allows certain - characters forbidden in typical CIFS filenames, without - requiring remapping. (default) - noposixpaths If CIFS Unix extensions are supported, do not request - posix path name support (this may cause servers to - reject creatingfile with certain reserved characters). - nounix Disable the CIFS Unix Extensions for this mount (tree - connection). This is rarely needed, but it may be useful - in order to turn off multiple settings all at once (ie - posix acls, posix locks, posix paths, symlink support - and retrieving uids/gids/mode from the server) or to - work around a bug in server which implement the Unix - Extensions. - nobrl Do not send byte range lock requests to the server. - This is necessary for certain applications that break - with cifs style mandatory byte range locks (and most - cifs servers do not yet support requesting advisory - byte range locks). - forcemandatorylock Even if the server supports posix (advisory) byte range - locking, send only mandatory lock requests. For some - (presumably rare) applications, originally coded for - DOS/Windows, which require Windows style mandatory byte range - locking, they may be able to take advantage of this option, - forcing the cifs client to only send mandatory locks - even if the cifs server would support posix advisory locks. - "forcemand" is accepted as a shorter form of this mount - option. - nostrictsync If this mount option is set, when an application does an - fsync call then the cifs client does not send an SMB Flush - to the server (to force the server to write all dirty data - for this file immediately to disk), although cifs still sends - all dirty (cached) file data to the server and waits for the - server to respond to the write. Since SMB Flush can be - very slow, and some servers may be reliable enough (to risk - delaying slightly flushing the data to disk on the server), - turning on this option may be useful to improve performance for - applications that fsync too much, at a small risk of server - crash. If this mount option is not set, by default cifs will - send an SMB flush request (and wait for a response) on every - fsync call. - nodfs Disable DFS (global name space support) even if the - server claims to support it. This can help work around - a problem with parsing of DFS paths with Samba server - versions 3.0.24 and 3.0.25. - remount remount the share (often used to change from ro to rw mounts - or vice versa) - cifsacl Report mode bits (e.g. on stat) based on the Windows ACL for - the file. (EXPERIMENTAL) - servern Specify the server 's netbios name (RFC1001 name) to use - when attempting to setup a session to the server. - This is needed for mounting to some older servers (such - as OS/2 or Windows 98 and Windows ME) since they do not - support a default server name. A server name can be up - to 15 characters long and is usually uppercased. - sfu When the CIFS Unix Extensions are not negotiated, attempt to - create device files and fifos in a format compatible with - Services for Unix (SFU). In addition retrieve bits 10-12 - of the mode via the SETFILEBITS extended attribute (as - SFU does). In the future the bottom 9 bits of the - mode also will be emulated using queries of the security - descriptor (ACL). - mfsymlinks Enable support for Minshall+French symlinks - (see http://wiki.samba.org/index.php/UNIX_Extensions#Minshall.2BFrench_symlinks) - This option is ignored when specified together with the - 'sfu' option. Minshall+French symlinks are used even if - the server supports the CIFS Unix Extensions. - sign Must use packet signing (helps avoid unwanted data modification - by intermediate systems in the route). Note that signing - does not work with lanman or plaintext authentication. - seal Must seal (encrypt) all data on this mounted share before - sending on the network. Requires support for Unix Extensions. - Note that this differs from the sign mount option in that it - causes encryption of data sent over this mounted share but other - shares mounted to the same server are unaffected. - locallease This option is rarely needed. Fcntl F_SETLEASE is - used by some applications such as Samba and NFSv4 server to - check to see whether a file is cacheable. CIFS has no way - to explicitly request a lease, but can check whether a file - is cacheable (oplocked). Unfortunately, even if a file - is not oplocked, it could still be cacheable (ie cifs client - could grant fcntl leases if no other local processes are using - the file) for cases for example such as when the server does not - support oplocks and the user is sure that the only updates to - the file will be from this client. Specifying this mount option - will allow the cifs client to check for leases (only) locally - for files which are not oplocked instead of denying leases - in that case. (EXPERIMENTAL) - sec Security mode. Allowed values are: - none attempt to connection as a null user (no name) - krb5 Use Kerberos version 5 authentication - krb5i Use Kerberos authentication and packet signing - ntlm Use NTLM password hashing (default) - ntlmi Use NTLM password hashing with signing (if - /proc/fs/cifs/PacketSigningEnabled on or if - server requires signing also can be the default) - ntlmv2 Use NTLMv2 password hashing - ntlmv2i Use NTLMv2 password hashing with packet signing - lanman (if configured in kernel config) use older - lanman hash -hard Retry file operations if server is not responding -soft Limit retries to unresponsive servers (usually only - one retry) before returning an error. (default) - -The mount.cifs mount helper also accepts a few mount options before -o -including: - - -S take password from stdin (equivalent to setting the environment - variable "PASSWD_FD=0" - -V print mount.cifs version - -? display simple usage information - -With most 2.6 kernel versions of modutils, the version of the cifs kernel -module can be displayed via modinfo. - -Misc /proc/fs/cifs Flags and Debug Info -======================================= -Informational pseudo-files: -DebugData Displays information about active CIFS sessions and - shares, features enabled as well as the cifs.ko - version. -Stats Lists summary resource usage information as well as per - share statistics. - -Configuration pseudo-files: -SecurityFlags Flags which control security negotiation and - also packet signing. Authentication (may/must) - flags (e.g. for NTLM and/or NTLMv2) may be combined with - the signing flags. Specifying two different password - hashing mechanisms (as "must use") on the other hand - does not make much sense. Default flags are - 0x07007 - (NTLM, NTLMv2 and packet signing allowed). The maximum - allowable flags if you want to allow mounts to servers - using weaker password hashes is 0x37037 (lanman, - plaintext, ntlm, ntlmv2, signing allowed). Some - SecurityFlags require the corresponding menuconfig - options to be enabled (lanman and plaintext require - CONFIG_CIFS_WEAK_PW_HASH for example). Enabling - plaintext authentication currently requires also - enabling lanman authentication in the security flags - because the cifs module only supports sending - laintext passwords using the older lanman dialect - form of the session setup SMB. (e.g. for authentication - using plain text passwords, set the SecurityFlags - to 0x30030): - - may use packet signing 0x00001 - must use packet signing 0x01001 - may use NTLM (most common password hash) 0x00002 - must use NTLM 0x02002 - may use NTLMv2 0x00004 - must use NTLMv2 0x04004 - may use Kerberos security 0x00008 - must use Kerberos 0x08008 - may use lanman (weak) password hash 0x00010 - must use lanman password hash 0x10010 - may use plaintext passwords 0x00020 - must use plaintext passwords 0x20020 - (reserved for future packet encryption) 0x00040 - -cifsFYI If set to non-zero value, additional debug information - will be logged to the system error log. This field - contains three flags controlling different classes of - debugging entries. The maximum value it can be set - to is 7 which enables all debugging points (default 0). - Some debugging statements are not compiled into the - cifs kernel unless CONFIG_CIFS_DEBUG2 is enabled in the - kernel configuration. cifsFYI may be set to one or - nore of the following flags (7 sets them all): - - log cifs informational messages 0x01 - log return codes from cifs entry points 0x02 - log slow responses (ie which take longer than 1 second) - CONFIG_CIFS_STATS2 must be enabled in .config 0x04 - - -traceSMB If set to one, debug information is logged to the - system error log with the start of smb requests - and responses (default 0) -LookupCacheEnable If set to one, inode information is kept cached - for one second improving performance of lookups - (default 1) -LinuxExtensionsEnabled If set to one then the client will attempt to - use the CIFS "UNIX" extensions which are optional - protocol enhancements that allow CIFS servers - to return accurate UID/GID information as well - as support symbolic links. If you use servers - such as Samba that support the CIFS Unix - extensions but do not want to use symbolic link - support and want to map the uid and gid fields - to values supplied at mount (rather than the - actual values, then set this to zero. (default 1) - -These experimental features and tracing can be enabled by changing flags in -/proc/fs/cifs (after the cifs module has been installed or built into the -kernel, e.g. insmod cifs). To enable a feature set it to 1 e.g. to enable -tracing to the kernel message log type: - - echo 7 > /proc/fs/cifs/cifsFYI - -cifsFYI functions as a bit mask. Setting it to 1 enables additional kernel -logging of various informational messages. 2 enables logging of non-zero -SMB return codes while 4 enables logging of requests that take longer -than one second to complete (except for byte range lock requests). -Setting it to 4 requires CONFIG_CIFS_STATS2 to be set in kernel configuration -(.config). Setting it to seven enables all three. Finally, tracing -the start of smb requests and responses can be enabled via: - - echo 1 > /proc/fs/cifs/traceSMB - -Per share (per client mount) statistics are available in /proc/fs/cifs/Stats. -Additional information is available if CONFIG_CIFS_STATS2 is enabled in the -kernel configuration (.config). The statistics returned include counters which -represent the number of attempted and failed (ie non-zero return code from the -server) SMB3 (or cifs) requests grouped by request type (read, write, close etc.). -Also recorded is the total bytes read and bytes written to the server for -that share. Note that due to client caching effects this can be less than the -number of bytes read and written by the application running on the client. -Statistics can be reset to zero by "echo 0 > /proc/fs/cifs/Stats" which may be -useful if comparing performance of two different scenarios. - -Also note that "cat /proc/fs/cifs/DebugData" will display information about -the active sessions and the shares that are mounted. - -Enabling Kerberos (extended security) works but requires version 1.2 or later -of the helper program cifs.upcall to be present and to be configured in the -/etc/request-key.conf file. The cifs.upcall helper program is from the Samba -project(http://www.samba.org). NTLM and NTLMv2 and LANMAN support do not -require this helper. Note that NTLMv2 security (which does not require the -cifs.upcall helper program), instead of using Kerberos, is sufficient for -some use cases. - -DFS support allows transparent redirection to shares in an MS-DFS name space. -In addition, DFS support for target shares which are specified as UNC -names which begin with host names (rather than IP addresses) requires -a user space helper (such as cifs.upcall) to be present in order to -translate host names to ip address, and the user space helper must also -be configured in the file /etc/request-key.conf. Samba, Windows servers and -many NAS appliances support DFS as a way of constructing a global name -space to ease network configuration and improve reliability. - -To use cifs Kerberos and DFS support, the Linux keyutils package should be -installed and something like the following lines should be added to the -/etc/request-key.conf file: - -create cifs.spnego * * /usr/local/sbin/cifs.upcall %k -create dns_resolver * * /usr/local/sbin/cifs.upcall %k - -CIFS kernel module parameters -============================= -These module parameters can be specified or modified either during the time of -module loading or during the runtime by using the interface - /proc/module/cifs/parameters/<param> - -i.e. echo "value" > /sys/module/cifs/parameters/<param> - -1. enable_oplocks - Enable or disable oplocks. Oplocks are enabled by default. - [Y/y/1]. To disable use any of [N/n/0]. - diff --git a/Documentation/filesystems/cifs/TODO b/Documentation/filesystems/cifs/TODO deleted file mode 100644 index 9267f3fb131f..000000000000 --- a/Documentation/filesystems/cifs/TODO +++ /dev/null @@ -1,119 +0,0 @@ -Version 2.14 December 21, 2018 - -A Partial List of Missing Features -================================== - -Contributions are welcome. There are plenty of opportunities -for visible, important contributions to this module. Here -is a partial list of the known problems and missing features: - -a) SMB3 (and SMB3.1.1) missing optional features: - - multichannel (started), integration with RDMA - - directory leases (improved metadata caching), started (root dir only) - - T10 copy offload ie "ODX" (copy chunk, and "Duplicate Extents" ioctl - currently the only two server side copy mechanisms supported) - -b) improved sparse file support - -c) Directory entry caching relies on a 1 second timer, rather than -using Directory Leases, currently only the root file handle is cached longer - -d) quota support (needs minor kernel change since quota calls -to make it to network filesystems or deviceless filesystems) - -e) Additional use cases where we use "compoounding" (e.g. open/query/close -and open/setinfo/close) to reduce the number of roundtrips, and also -open to reduce redundant opens (using deferred close and reference counts more). - -f) Finish inotify support so kde and gnome file list windows -will autorefresh (partially complete by Asser). Needs minor kernel -vfs change to support removing D_NOTIFY on a file. - -g) Add GUI tool to configure /proc/fs/cifs settings and for display of -the CIFS statistics (started) - -h) implement support for security and trusted categories of xattrs -(requires minor protocol extension) to enable better support for SELINUX - -i) Add support for tree connect contexts (see MS-SMB2) a new SMB3.1.1 protocol - feature (may be especially useful for virtualization). - -j) Create UID mapping facility so server UIDs can be mapped on a per -mount or a per server basis to client UIDs or nobody if no mapping -exists. Also better integration with winbind for resolving SID owners - -k) Add tools to take advantage of more smb3 specific ioctls and features -(passthrough ioctl/fsctl for sending various SMB3 fsctls to the server -is in progress, and a passthrough query_info call is already implemented -in cifs.ko to allow smb3 info levels queries to be sent from userspace) - -l) encrypted file support - -m) improved stats gathering tools (perhaps integration with nfsometer?) -to extend and make easier to use what is currently in /proc/fs/cifs/Stats - -n) allow setting more NTFS/SMB3 file attributes remotely (currently limited to compressed -file attribute via chflags) and improve user space tools for managing and -viewing them. - -o) mount helper GUI (to simplify the various configuration options on mount) - -p) Add support for witness protocol (perhaps ioctl to cifs.ko from user space - tool listening on witness protocol RPC) to allow for notification of share - move, server failover, and server adapter changes. And also improve other - failover scenarios, e.g. when client knows multiple DFS entries point to - different servers, and the server we are connected to has gone down. - -q) Allow mount.cifs to be more verbose in reporting errors with dialect -or unsupported feature errors. - -r) updating cifs documentation, and user guide. - -s) Addressing bugs found by running a broader set of xfstests in standard -file system xfstest suite. - -t) split cifs and smb3 support into separate modules so legacy (and less -secure) CIFS dialect can be disabled in environments that don't need it -and simplify the code. - -v) POSIX Extensions for SMB3.1.1 (started, create and mkdir support added -so far). - -w) Add support for additional strong encryption types, and additional spnego -authentication mechanisms (see MS-SMB2) - -KNOWN BUGS -==================================== -See http://bugzilla.samba.org - search on product "CifsVFS" for -current bug list. Also check http://bugzilla.kernel.org (Product = File System, Component = CIFS) - -1) existing symbolic links (Windows reparse points) are recognized but -can not be created remotely. They are implemented for Samba and those that -support the CIFS Unix extensions, although earlier versions of Samba -overly restrict the pathnames. -2) follow_link and readdir code does not follow dfs junctions -but recognizes them - -Misc testing to do -================== -1) check out max path names and max path name components against various server -types. Try nested symlinks (8 deep). Return max path name in stat -f information - -2) Improve xfstest's cifs/smb3 enablement and adapt xfstests where needed to test -cifs/smb3 better - -3) Additional performance testing and optimization using iozone and similar - -there are some easy changes that can be done to parallelize sequential writes, -and when signing is disabled to request larger read sizes (larger than -negotiated size) and send larger write sizes to modern servers. - -4) More exhaustively test against less common servers - -5) Continue to extend the smb3 "buildbot" which does automated xfstesting -against Windows, Samba and Azure currently - to add additional tests and -to allow the buildbot to execute the tests faster. The URL for the -buildbot is: http://smb3-test-rhel-75.southcentralus.cloudapp.azure.com - -6) Address various coverity warnings (most are not bugs per-se, but -the more warnings are addressed, the easier it is to spot real -problems that static analyzers will point out in the future). diff --git a/Documentation/filesystems/cifs/cifs.txt b/Documentation/filesystems/cifs/cifs.txt deleted file mode 100644 index 1be3d21c286e..000000000000 --- a/Documentation/filesystems/cifs/cifs.txt +++ /dev/null @@ -1,45 +0,0 @@ - This is the client VFS module for the SMB3 NAS protocol as well - as for older dialects such as the Common Internet File System (CIFS) - protocol which was the successor to the Server Message Block - (SMB) protocol, the native file sharing mechanism for most early - PC operating systems. New and improved versions of CIFS are now - called SMB2 and SMB3. Use of SMB3 (and later, including SMB3.1.1) - is strongly preferred over using older dialects like CIFS due to - security reaasons. All modern dialects, including the most recent, - SMB3.1.1 are supported by the CIFS VFS module. The SMB3 protocol - is implemented and supported by all major file servers - such as all modern versions of Windows (including Windows 2016 - Server), as well as by Samba (which provides excellent - CIFS/SMB2/SMB3 server support and tools for Linux and many other - operating systems). Apple systems also support SMB3 well, as - do most Network Attached Storage vendors, so this network - filesystem client can mount to a wide variety of systems. - It also supports mounting to the cloud (for example - Microsoft Azure), including the necessary security features. - - The intent of this module is to provide the most advanced network - file system function for SMB3 compliant servers, including advanced - security features, excellent parallelized high performance i/o, better - POSIX compliance, secure per-user session establishment, encryption, - high performance safe distributed caching (leases/oplocks), optional packet - signing, large files, Unicode support and other internationalization - improvements. Since both Samba server and this filesystem client support - the CIFS Unix extensions (and in the future SMB3 POSIX extensions), - the combination can provide a reasonable alternative to other network and - cluster file systems for fileserving in some Linux to Linux environments, - not just in Linux to Windows (or Linux to Mac) environments. - - This filesystem has a mount utility (mount.cifs) and various user space - tools (including smbinfo and setcifsacl) that can be obtained from - - https://git.samba.org/?p=cifs-utils.git - or - git://git.samba.org/cifs-utils.git - - mount.cifs should be installed in the directory with the other mount helpers. - - For more information on the module see the project wiki page at - - https://wiki.samba.org/index.php/LinuxCIFS - and - https://wiki.samba.org/index.php/LinuxCIFS_utils diff --git a/Documentation/filesystems/cifs/winucase_convert.pl b/Documentation/filesystems/cifs/winucase_convert.pl deleted file mode 100755 index 322a9c833f23..000000000000 --- a/Documentation/filesystems/cifs/winucase_convert.pl +++ /dev/null @@ -1,62 +0,0 @@ -#!/usr/bin/perl -w -# -# winucase_convert.pl -- convert "Windows 8 Upper Case Mapping Table.txt" to -# a two-level set of C arrays. -# -# Copyright 2013: Jeff Layton <jlayton@redhat.com> -# -# This program is free software: you can redistribute it and/or modify -# it under the terms of the GNU General Public License as published by -# the Free Software Foundation, either version 3 of the License, or -# (at your option) any later version. -# -# This program is distributed in the hope that it will be useful, -# but WITHOUT ANY WARRANTY; without even the implied warranty of -# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the -# GNU General Public License for more details. -# -# You should have received a copy of the GNU General Public License -# along with this program. If not, see <http://www.gnu.org/licenses/>. -# - -while(<>) { - next if (!/^0x(..)(..)\t0x(....)\t/); - $firstchar = hex($1); - $secondchar = hex($2); - $uppercase = hex($3); - - $top[$firstchar][$secondchar] = $uppercase; -} - -for ($i = 0; $i < 256; $i++) { - next if (!$top[$i]); - - printf("static const wchar_t t2_%2.2x[256] = {", $i); - for ($j = 0; $j < 256; $j++) { - if (($j % 8) == 0) { - print "\n\t"; - } else { - print " "; - } - printf("0x%4.4x,", $top[$i][$j] ? $top[$i][$j] : 0); - } - print "\n};\n\n"; -} - -printf("static const wchar_t *const toplevel[256] = {", $i); -for ($i = 0; $i < 256; $i++) { - if (($i % 8) == 0) { - print "\n\t"; - } elsif ($top[$i]) { - print " "; - } else { - print " "; - } - - if ($top[$i]) { - printf("t2_%2.2x,", $i); - } else { - print "NULL,"; - } -} -print "\n};\n\n"; diff --git a/Documentation/filesystems/coda.txt b/Documentation/filesystems/coda.txt index 545262c167c3..1711ad48e38a 100644 --- a/Documentation/filesystems/coda.txt +++ b/Documentation/filesystems/coda.txt @@ -421,14 +421,14 @@ kernel support. The CodaCred structure defines a variety of user and group ids as - they are set for the calling process. The vuid_t and guid_t are 32 bit + they are set for the calling process. The vuid_t and vgid_t are 32 bit unsigned integers. It also defines group membership in an array. On Unix the CodaCred has proven sufficient to implement good security semantics for Coda but the structure may have to undergo modification for the Windows environment when these mature. struct CodaCred { - vuid_t cr_uid, cr_euid, cr_suid, cr_fsuid; /* Real, effective, set, fs uid*/ + vuid_t cr_uid, cr_euid, cr_suid, cr_fsuid; /* Real, effective, set, fs uid */ vgid_t cr_gid, cr_egid, cr_sgid, cr_fsgid; /* same for groups */ vgid_t cr_groups[NGROUPS]; /* Group membership for caller */ }; diff --git a/Documentation/filesystems/conf.py b/Documentation/filesystems/conf.py deleted file mode 100644 index ea44172af5c4..000000000000 --- a/Documentation/filesystems/conf.py +++ /dev/null @@ -1,10 +0,0 @@ -# -*- coding: utf-8; mode: python -*- - -project = "Linux Filesystems API" - -tags.add("subproject") - -latex_documents = [ - ('index', 'filesystems.tex', project, - 'The kernel development community', 'manual'), -] diff --git a/Documentation/filesystems/directory-locking b/Documentation/filesystems/directory-locking.rst index 4e32cb961e5b..de12016ee419 100644 --- a/Documentation/filesystems/directory-locking +++ b/Documentation/filesystems/directory-locking.rst @@ -1,12 +1,17 @@ - Locking scheme used for directory operations is based on two +================= +Directory Locking +================= + + +Locking scheme used for directory operations is based on two kinds of locks - per-inode (->i_rwsem) and per-filesystem (->s_vfs_rename_mutex). - When taking the i_rwsem on multiple non-directory objects, we +When taking the i_rwsem on multiple non-directory objects, we always acquire the locks in order by increasing address. We'll call that "inode pointer" order in the following. - For our purposes all operations fall in 5 classes: +For our purposes all operations fall in 5 classes: 1) read access. Locking rules: caller locks directory we are accessing. The lock is taken shared. @@ -27,25 +32,29 @@ NB: we might get away with locking the the source (and target in exchange case) shared. 5) link creation. Locking rules: + * lock parent * check that source is not a directory * lock source * call the method. + All locks are exclusive. 6) cross-directory rename. The trickiest in the whole bunch. Locking rules: + * lock the filesystem * lock parents in "ancestors first" order. * find source and target. * if old parent is equal to or is a descendent of target - fail with -ENOTEMPTY + fail with -ENOTEMPTY * if new parent is equal to or is a descendent of source - fail with -ELOOP + fail with -ELOOP * If it's an exchange, lock both the source and the target. * If the target exists, lock it. If the source is a non-directory, lock it. If we need to lock both, do so in inode pointer order. * call the method. + All ->i_rwsem are taken exclusive. Again, we might get away with locking the the source (and target in exchange case) shared. @@ -54,10 +63,11 @@ read, modified or removed by method will be locked by caller. If no directory is its own ancestor, the scheme above is deadlock-free. + Proof: First of all, at any moment we have a partial ordering of the -objects - A < B iff A is an ancestor of B. + objects - A < B iff A is an ancestor of B. That ordering can change. However, the following is true: @@ -77,32 +87,32 @@ objects - A < B iff A is an ancestor of B. non-directory object, except renames, which take locks on source and target in inode pointer order in the case they are not directories.) - Now consider the minimal deadlock. Each process is blocked on +Now consider the minimal deadlock. Each process is blocked on attempt to acquire some lock and already holds at least one lock. Let's consider the set of contended locks. First of all, filesystem lock is not contended, since any process blocked on it is not holding any locks. Thus all processes are blocked on ->i_rwsem. - By (3), any process holding a non-directory lock can only be +By (3), any process holding a non-directory lock can only be waiting on another non-directory lock with a larger address. Therefore the process holding the "largest" such lock can always make progress, and non-directory objects are not included in the set of contended locks. - Thus link creation can't be a part of deadlock - it can't be +Thus link creation can't be a part of deadlock - it can't be blocked on source and it means that it doesn't hold any locks. - Any contended object is either held by cross-directory rename or +Any contended object is either held by cross-directory rename or has a child that is also contended. Indeed, suppose that it is held by operation other than cross-directory rename. Then the lock this operation is blocked on belongs to child of that object due to (1). - It means that one of the operations is cross-directory rename. +It means that one of the operations is cross-directory rename. Otherwise the set of contended objects would be infinite - each of them would have a contended child and we had assumed that no object is its own descendent. Moreover, there is exactly one cross-directory rename (see above). - Consider the object blocking the cross-directory rename. One +Consider the object blocking the cross-directory rename. One of its descendents is locked by cross-directory rename (otherwise we would again have an infinite set of contended objects). But that means that cross-directory rename is taking locks out of order. Due @@ -112,7 +122,7 @@ try to acquire lock on descendent before the lock on ancestor. Contradiction. I.e. deadlock is impossible. Q.E.D. - These operations are guaranteed to avoid loop creation. Indeed, +These operations are guaranteed to avoid loop creation. Indeed, the only operation that could introduce loops is cross-directory rename. Since the only new (parent, child) pair added by rename() is (new parent, source), such loop would have to contain these objects and the rest of it @@ -123,13 +133,13 @@ new parent had been equal to or a descendent of source since the moment when we had acquired filesystem lock and rename() would fail with -ELOOP in that case. - While this locking scheme works for arbitrary DAGs, it relies on +While this locking scheme works for arbitrary DAGs, it relies on ability to check that directory is a descendent of another object. Current implementation assumes that directory graph is a tree. This assumption is also preserved by all operations (cross-directory rename on a tree that would not introduce a cycle will leave it a tree and link() fails for directories). - Notice that "directory" in the above == "anything that might have +Notice that "directory" in the above == "anything that might have children", so if we are going to introduce hybrid objects we will need either to make sure that link(2) doesn't work for them or to make changes in is_subdir() that would make it work even in presence of such beasts. diff --git a/Documentation/filesystems/erofs.txt b/Documentation/filesystems/erofs.txt new file mode 100644 index 000000000000..b0c085326e2e --- /dev/null +++ b/Documentation/filesystems/erofs.txt @@ -0,0 +1,210 @@ +Overview +======== + +EROFS file-system stands for Enhanced Read-Only File System. Different +from other read-only file systems, it aims to be designed for flexibility, +scalability, but be kept simple and high performance. + +It is designed as a better filesystem solution for the following scenarios: + - read-only storage media or + + - part of a fully trusted read-only solution, which means it needs to be + immutable and bit-for-bit identical to the official golden image for + their releases due to security and other considerations and + + - hope to save some extra storage space with guaranteed end-to-end performance + by using reduced metadata and transparent file compression, especially + for those embedded devices with limited memory (ex, smartphone); + +Here is the main features of EROFS: + - Little endian on-disk design; + + - Currently 4KB block size (nobh) and therefore maximum 16TB address space; + + - Metadata & data could be mixed by design; + + - 2 inode versions for different requirements: + v1 v2 + Inode metadata size: 32 bytes 64 bytes + Max file size: 4 GB 16 EB (also limited by max. vol size) + Max uids/gids: 65536 4294967296 + File creation time: no yes (64 + 32-bit timestamp) + Max hardlinks: 65536 4294967296 + Metadata reserved: 4 bytes 14 bytes + + - Support extended attributes (xattrs) as an option; + + - Support xattr inline and tail-end data inline for all files; + + - Support POSIX.1e ACLs by using xattrs; + + - Support transparent file compression as an option: + LZ4 algorithm with 4 KB fixed-output compression for high performance; + +The following git tree provides the file system user-space tools under +development (ex, formatting tool mkfs.erofs): +>> git://git.kernel.org/pub/scm/linux/kernel/git/xiang/erofs-utils.git + +Bugs and patches are welcome, please kindly help us and send to the following +linux-erofs mailing list: +>> linux-erofs mailing list <linux-erofs@lists.ozlabs.org> + +Mount options +============= + +(no)user_xattr Setup Extended User Attributes. Note: xattr is enabled + by default if CONFIG_EROFS_FS_XATTR is selected. +(no)acl Setup POSIX Access Control List. Note: acl is enabled + by default if CONFIG_EROFS_FS_POSIX_ACL is selected. +cache_strategy=%s Select a strategy for cached decompression from now on: + disabled: In-place I/O decompression only; + readahead: Cache the last incomplete compressed physical + cluster for further reading. It still does + in-place I/O decompression for the rest + compressed physical clusters; + readaround: Cache the both ends of incomplete compressed + physical clusters for further reading. + It still does in-place I/O decompression + for the rest compressed physical clusters. + +On-disk details +=============== + +Summary +------- +Different from other read-only file systems, an EROFS volume is designed +to be as simple as possible: + + |-> aligned with the block size + ____________________________________________________________ + | |SB| | ... | Metadata | ... | Data | Metadata | ... | Data | + |_|__|_|_____|__________|_____|______|__________|_____|______| + 0 +1K + +All data areas should be aligned with the block size, but metadata areas +may not. All metadatas can be now observed in two different spaces (views): + 1. Inode metadata space + Each valid inode should be aligned with an inode slot, which is a fixed + value (32 bytes) and designed to be kept in line with v1 inode size. + + Each inode can be directly found with the following formula: + inode offset = meta_blkaddr * block_size + 32 * nid + + |-> aligned with 8B + |-> followed closely + + meta_blkaddr blocks |-> another slot + _____________________________________________________________________ + | ... | inode | xattrs | extents | data inline | ... | inode ... + |________|_______|(optional)|(optional)|__(optional)_|_____|__________ + |-> aligned with the inode slot size + . . + . . + . . + . . + . . + . . + .____________________________________________________|-> aligned with 4B + | xattr_ibody_header | shared xattrs | inline xattrs | + |____________________|_______________|_______________| + |-> 12 bytes <-|->x * 4 bytes<-| . + . . . + . . . + . . . + ._______________________________.______________________. + | id | id | id | id | ... | id | ent | ... | ent| ... | + |____|____|____|____|______|____|_____|_____|____|_____| + |-> aligned with 4B + |-> aligned with 4B + + Inode could be 32 or 64 bytes, which can be distinguished from a common + field which all inode versions have -- i_advise: + + __________________ __________________ + | i_advise | | i_advise | + |__________________| |__________________| + | ... | | ... | + | | | | + |__________________| 32 bytes | | + | | + |__________________| 64 bytes + + Xattrs, extents, data inline are followed by the corresponding inode with + proper alignes, and they could be optional for different data mappings, + _currently_ there are totally 3 valid data mappings supported: + + 1) flat file data without data inline (no extent); + 2) fixed-output size data compression (must have extents); + 3) flat file data with tail-end data inline (no extent); + + The size of the optional xattrs is indicated by i_xattr_count in inode + header. Large xattrs or xattrs shared by many different files can be + stored in shared xattrs metadata rather than inlined right after inode. + + 2. Shared xattrs metadata space + Shared xattrs space is similar to the above inode space, started with + a specific block indicated by xattr_blkaddr, organized one by one with + proper align. + + Each share xattr can also be directly found by the following formula: + xattr offset = xattr_blkaddr * block_size + 4 * xattr_id + + |-> aligned by 4 bytes + + xattr_blkaddr blocks |-> aligned with 4 bytes + _________________________________________________________________________ + | ... | xattr_entry | xattr data | ... | xattr_entry | xattr data ... + |________|_____________|_____________|_____|______________|_______________ + +Directories +----------- +All directories are now organized in a compact on-disk format. Note that +each directory block is divided into index and name areas in order to support +random file lookup, and all directory entries are _strictly_ recorded in +alphabetical order in order to support improved prefix binary search +algorithm (could refer to the related source code). + + ___________________________ + / | + / ______________|________________ + / / | nameoff1 | nameoffN-1 + ____________.______________._______________v________________v__________ +| dirent | dirent | ... | dirent | filename | filename | ... | filename | +|___.0___|____1___|_____|___N-1__|____0_____|____1_____|_____|___N-1____| + \ ^ + \ | * could have + \ | trailing '\0' + \________________________| nameoff0 + + Directory block + +Note that apart from the offset of the first filename, nameoff0 also indicates +the total number of directory entries in this block since it is no need to +introduce another on-disk field at all. + +Compression +----------- +Currently, EROFS supports 4KB fixed-output clustersize transparent file +compression, as illustrated below: + + |---- Variant-Length Extent ----|-------- VLE --------|----- VLE ----- + clusterofs clusterofs clusterofs + | | | logical data +_________v_______________________________v_____________________v_______________ +... | . | | . | | . | ... +____|____.________|_____________|________.____|_____________|__.__________|____ + |-> cluster <-|-> cluster <-|-> cluster <-|-> cluster <-|-> cluster <-| + size size size size size + . . . . + . . . . + . . . . + _______._____________._____________._____________._____________________ + ... | | | | ... physical data + _______|_____________|_____________|_____________|_____________________ + |-> cluster <-|-> cluster <-|-> cluster <-| + size size size + +Currently each on-disk physical cluster can contain 4KB (un)compressed data +at most. For each logical cluster, there is a corresponding on-disk index to +describe its cluster type, physical cluster address, etc. + +See "struct z_erofs_vle_decompressed_index" in erofs_fs.h for more details. + diff --git a/Documentation/filesystems/index.rst b/Documentation/filesystems/index.rst index 2de2fe2ab078..96653ebefd7e 100644 --- a/Documentation/filesystems/index.rst +++ b/Documentation/filesystems/index.rst @@ -20,6 +20,10 @@ algorithms work. path-lookup api-summary splice + locking + directory-locking + + porting Filesystem support layers ========================= diff --git a/Documentation/filesystems/jfs.txt b/Documentation/filesystems/jfs.txt deleted file mode 100644 index 41fd757997b3..000000000000 --- a/Documentation/filesystems/jfs.txt +++ /dev/null @@ -1,52 +0,0 @@ -IBM's Journaled File System (JFS) for Linux - -JFS Homepage: http://jfs.sourceforge.net/ - -The following mount options are supported: -(*) == default - -iocharset=name Character set to use for converting from Unicode to - ASCII. The default is to do no conversion. Use - iocharset=utf8 for UTF-8 translations. This requires - CONFIG_NLS_UTF8 to be set in the kernel .config file. - iocharset=none specifies the default behavior explicitly. - -resize=value Resize the volume to <value> blocks. JFS only supports - growing a volume, not shrinking it. This option is only - valid during a remount, when the volume is mounted - read-write. The resize keyword with no value will grow - the volume to the full size of the partition. - -nointegrity Do not write to the journal. The primary use of this option - is to allow for higher performance when restoring a volume - from backup media. The integrity of the volume is not - guaranteed if the system abnormally abends. - -integrity(*) Commit metadata changes to the journal. Use this option to - remount a volume where the nointegrity option was - previously specified in order to restore normal behavior. - -errors=continue Keep going on a filesystem error. -errors=remount-ro(*) Remount the filesystem read-only on an error. -errors=panic Panic and halt the machine if an error occurs. - -uid=value Override on-disk uid with specified value -gid=value Override on-disk gid with specified value -umask=value Override on-disk umask with specified octal value. For - directories, the execute bit will be set if the corresponding - read bit is set. - -discard=minlen This enables/disables the use of discard/TRIM commands. -discard The discard/TRIM commands are sent to the underlying -nodiscard(*) block device when blocks are freed. This is useful for SSD - devices and sparse/thinly-provisioned LUNs. The FITRIM ioctl - command is also available together with the nodiscard option. - The value of minlen specifies the minimum blockcount, when - a TRIM command to the block device is considered useful. - When no value is given to the discard option, it defaults to - 64 blocks, which means 256KiB in JFS. - The minlen value of discard overrides the minlen value given - on an FITRIM ioctl(). - -The JFS mailing list can be subscribed to by using the link labeled -"Mail list Subscribe" at our web page http://jfs.sourceforge.net/ diff --git a/Documentation/filesystems/Locking b/Documentation/filesystems/locking.rst index 204dd3ea36bb..fc3a0704553c 100644 --- a/Documentation/filesystems/Locking +++ b/Documentation/filesystems/locking.rst @@ -1,14 +1,22 @@ - The text below describes the locking rules for VFS-related methods. +======= +Locking +======= + +The text below describes the locking rules for VFS-related methods. It is (believed to be) up-to-date. *Please*, if you change anything in prototypes or locking protocols - update this file. And update the relevant instances in the tree, don't leave that to maintainers of filesystems/devices/ etc. At the very least, put the list of dubious cases in the end of this file. Don't turn it into log - maintainers of out-of-the-tree code are supposed to be able to use diff(1). - Thing currently missing here: socket operations. Alexey? ---------------------------- dentry_operations -------------------------- -prototypes: +Thing currently missing here: socket operations. Alexey? + +dentry_operations +================= + +prototypes:: + int (*d_revalidate)(struct dentry *, unsigned int); int (*d_weak_revalidate)(struct dentry *, unsigned int); int (*d_hash)(const struct dentry *, struct qstr *); @@ -24,23 +32,30 @@ prototypes: struct dentry *(*d_real)(struct dentry *, const struct inode *); locking rules: - rename_lock ->d_lock may block rcu-walk -d_revalidate: no no yes (ref-walk) maybe -d_weak_revalidate:no no yes no -d_hash no no no maybe -d_compare: yes no no maybe -d_delete: no yes no no -d_init: no no yes no -d_release: no no yes no -d_prune: no yes no no -d_iput: no no yes no -d_dname: no no no no -d_automount: no no yes no -d_manage: no no yes (ref-walk) maybe -d_real no no yes no - ---------------------------- inode_operations --------------------------- -prototypes: + +================== =========== ======== ============== ======== +ops rename_lock ->d_lock may block rcu-walk +================== =========== ======== ============== ======== +d_revalidate: no no yes (ref-walk) maybe +d_weak_revalidate: no no yes no +d_hash no no no maybe +d_compare: yes no no maybe +d_delete: no yes no no +d_init: no no yes no +d_release: no no yes no +d_prune: no yes no no +d_iput: no no yes no +d_dname: no no no no +d_automount: no no yes no +d_manage: no no yes (ref-walk) maybe +d_real no no yes no +================== =========== ======== ============== ======== + +inode_operations +================ + +prototypes:: + int (*create) (struct inode *,struct dentry *,umode_t, bool); struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int); int (*link) (struct dentry *,struct inode *,struct dentry *); @@ -68,7 +83,10 @@ prototypes: locking rules: all may block - i_rwsem(inode) + +============ ============================================= +ops i_rwsem(inode) +============ ============================================= lookup: shared create: exclusive link: exclusive (both) @@ -89,17 +107,21 @@ fiemap: no update_time: no atomic_open: exclusive tmpfile: no +============ ============================================= Additionally, ->rmdir(), ->unlink() and ->rename() have ->i_rwsem exclusive on victim. cross-directory ->rename() has (per-superblock) ->s_vfs_rename_sem. -See Documentation/filesystems/directory-locking for more detailed discussion +See Documentation/filesystems/directory-locking.rst for more detailed discussion of the locking scheme for directory operations. ------------------------ xattr_handler operations ----------------------- -prototypes: +xattr_handler operations +======================== + +prototypes:: + bool (*list)(struct dentry *dentry); int (*get)(const struct xattr_handler *handler, struct dentry *dentry, struct inode *inode, const char *name, void *buffer, @@ -110,13 +132,20 @@ prototypes: locking rules: all may block - i_rwsem(inode) + +===== ============== +ops i_rwsem(inode) +===== ============== list: no get: no set: exclusive +===== ============== + +super_operations +================ + +prototypes:: ---------------------------- super_operations --------------------------- -prototypes: struct inode *(*alloc_inode)(struct super_block *sb); void (*free_inode)(struct inode *); void (*destroy_inode)(struct inode *); @@ -138,7 +167,10 @@ prototypes: locking rules: All may block [not true, see below] - s_umount + +====================== ============ ======================== +ops s_umount note +====================== ============ ======================== alloc_inode: free_inode: called from RCU callback destroy_inode: @@ -157,6 +189,7 @@ show_options: no (namespace_sem) quota_read: no (see below) quota_write: no (see below) bdev_try_to_free_page: no (see below) +====================== ============ ======================== ->statfs() has s_umount (shared) when called by ustat(2) (native or compat), but that's an accident of bad API; s_umount is used to pin @@ -164,31 +197,44 @@ the superblock down when we only have dev_t given us by userland to identify the superblock. Everything else (statfs(), fstatfs(), etc.) doesn't hold it when calling ->statfs() - superblock is pinned down by resolving the pathname passed to syscall. + ->quota_read() and ->quota_write() functions are both guaranteed to be the only ones operating on the quota file by the quota code (via dqio_sem) (unless an admin really wants to screw up something and writes to quota files with quotas on). For other details about locking see also dquot_operations section. + ->bdev_try_to_free_page is called from the ->releasepage handler of the block device inode. See there for more details. ---------------------------- file_system_type --------------------------- -prototypes: +file_system_type +================ + +prototypes:: + struct dentry *(*mount) (struct file_system_type *, int, const char *, void *); void (*kill_sb) (struct super_block *); + locking rules: - may block + +======= ========= +ops may block +======= ========= mount yes kill_sb yes +======= ========= ->mount() returns ERR_PTR or the root dentry; its superblock should be locked on return. + ->kill_sb() takes a write-locked superblock, does all shutdown work on it, unlocks and drops the reference. ---------------------------- address_space_operations -------------------------- -prototypes: +address_space_operations +======================== +prototypes:: + int (*writepage)(struct page *page, struct writeback_control *wbc); int (*readpage)(struct file *, struct page *); int (*writepages)(struct address_space *, struct writeback_control *); @@ -218,14 +264,16 @@ prototypes: locking rules: All except set_page_dirty and freepage may block - PageLocked(page) i_rwsem +====================== ======================== ========= +ops PageLocked(page) i_rwsem +====================== ======================== ========= writepage: yes, unlocks (see below) readpage: yes, unlocks writepages: set_page_dirty no readpages: -write_begin: locks the page exclusive -write_end: yes, unlocks exclusive +write_begin: locks the page exclusive +write_end: yes, unlocks exclusive bmap: invalidatepage: yes releasepage: yes @@ -239,17 +287,18 @@ is_partially_uptodate: yes error_remove_page: yes swap_activate: no swap_deactivate: no +====================== ======================== ========= - ->write_begin(), ->write_end() and ->readpage() may be called from +->write_begin(), ->write_end() and ->readpage() may be called from the request handler (/dev/loop). - ->readpage() unlocks the page, either synchronously or via I/O +->readpage() unlocks the page, either synchronously or via I/O completion. - ->readpages() populates the pagecache with the passed pages and starts +->readpages() populates the pagecache with the passed pages and starts I/O against them. They come unlocked upon I/O completion. - ->writepage() is used for two purposes: for "memory cleansing" and for +->writepage() is used for two purposes: for "memory cleansing" and for "sync". These are quite different operations and the behaviour may differ depending upon the mode. @@ -297,70 +346,81 @@ will leave the page itself marked clean but it will be tagged as dirty in the radix tree. This incoherency can lead to all sorts of hard-to-debug problems in the filesystem like having dirty inodes at umount and losing written data. - ->writepages() is used for periodic writeback and for syscall-initiated +->writepages() is used for periodic writeback and for syscall-initiated sync operations. The address_space should start I/O against at least -*nr_to_write pages. *nr_to_write must be decremented for each page which is -written. The address_space implementation may write more (or less) pages -than *nr_to_write asks for, but it should try to be reasonably close. If -nr_to_write is NULL, all dirty pages must be written. +``*nr_to_write`` pages. ``*nr_to_write`` must be decremented for each page +which is written. The address_space implementation may write more (or less) +pages than ``*nr_to_write`` asks for, but it should try to be reasonably close. +If nr_to_write is NULL, all dirty pages must be written. writepages should _only_ write pages which are present on mapping->io_pages. - ->set_page_dirty() is called from various places in the kernel +->set_page_dirty() is called from various places in the kernel when the target page is marked as needing writeback. It may be called under spinlock (it cannot block) and is sometimes called with the page not locked. - ->bmap() is currently used by legacy ioctl() (FIBMAP) provided by some +->bmap() is currently used by legacy ioctl() (FIBMAP) provided by some filesystems and by the swapper. The latter will eventually go away. Please, keep it that way and don't breed new callers. - ->invalidatepage() is called when the filesystem must attempt to drop +->invalidatepage() is called when the filesystem must attempt to drop some or all of the buffers from the page when it is being truncated. It returns zero on success. If ->invalidatepage is zero, the kernel uses block_invalidatepage() instead. - ->releasepage() is called when the kernel is about to try to drop the +->releasepage() is called when the kernel is about to try to drop the buffers from the page in preparation for freeing it. It returns zero to indicate that the buffers are (or may be) freeable. If ->releasepage is zero, the kernel assumes that the fs has no private interest in the buffers. - ->freepage() is called when the kernel is done dropping the page +->freepage() is called when the kernel is done dropping the page from the page cache. - ->launder_page() may be called prior to releasing a page if +->launder_page() may be called prior to releasing a page if it is still found to be dirty. It returns zero if the page was successfully cleaned, or an error value if not. Note that in order to prevent the page getting mapped back in and redirtied, it needs to be kept locked across the entire operation. - ->swap_activate will be called with a non-zero argument on +->swap_activate will be called with a non-zero argument on files backing (non block device backed) swapfiles. A return value of zero indicates success, in which case this file can be used for backing swapspace. The swapspace operations will be proxied to the address space operations. - ->swap_deactivate() will be called in the sys_swapoff() +->swap_deactivate() will be called in the sys_swapoff() path after ->swap_activate() returned success. ------------------------ file_lock_operations ------------------------------ -prototypes: +file_lock_operations +==================== + +prototypes:: + void (*fl_copy_lock)(struct file_lock *, struct file_lock *); void (*fl_release_private)(struct file_lock *); locking rules: - inode->i_lock may block + +=================== ============= ========= +ops inode->i_lock may block +=================== ============= ========= fl_copy_lock: yes no -fl_release_private: maybe maybe[1] +fl_release_private: maybe maybe[1]_ +=================== ============= ========= + +.. [1]: + ->fl_release_private for flock or POSIX locks is currently allowed + to block. Leases however can still be freed while the i_lock is held and + so fl_release_private called on a lease should not block. -[1]: ->fl_release_private for flock or POSIX locks is currently allowed -to block. Leases however can still be freed while the i_lock is held and -so fl_release_private called on a lease should not block. +lock_manager_operations +======================= + +prototypes:: ------------------------ lock_manager_operations --------------------------- -prototypes: void (*lm_notify)(struct file_lock *); /* unblock callback */ int (*lm_grant)(struct file_lock *, struct file_lock *, int); void (*lm_break)(struct file_lock *); /* break_lease callback */ @@ -368,24 +428,33 @@ prototypes: locking rules: - inode->i_lock blocked_lock_lock may block +========== ============= ================= ========= +ops inode->i_lock blocked_lock_lock may block +========== ============= ================= ========= lm_notify: yes yes no lm_grant: no no no lm_break: yes no no lm_change yes no no +========== ============= ================= ========= + +buffer_head +=========== + +prototypes:: ---------------------------- buffer_head ----------------------------------- -prototypes: void (*b_end_io)(struct buffer_head *bh, int uptodate); locking rules: - called from interrupts. In other words, extreme care is needed here. + +called from interrupts. In other words, extreme care is needed here. bh is locked, but that's all warranties we have here. Currently only RAID1, highmem, fs/buffer.c, and fs/ntfs/aops.c are providing these. Block devices call this method upon the IO completion. ---------------------------- block_device_operations ----------------------- -prototypes: +block_device_operations +======================= +prototypes:: + int (*open) (struct block_device *, fmode_t); int (*release) (struct gendisk *, fmode_t); int (*ioctl) (struct block_device *, fmode_t, unsigned, unsigned long); @@ -399,7 +468,10 @@ prototypes: void (*swap_slot_free_notify) (struct block_device *, unsigned long); locking rules: - bd_mutex + +======================= =================== +ops bd_mutex +======================= =================== open: yes release: yes ioctl: no @@ -410,6 +482,7 @@ unlock_native_capacity: no revalidate_disk: no getgeo: no swap_slot_free_notify: no (see below) +======================= =================== media_changed, unlock_native_capacity and revalidate_disk are called only from check_disk_change(). @@ -418,8 +491,11 @@ swap_slot_free_notify is called with swap_lock and sometimes the page lock held. ---------------------------- file_operations ------------------------------- -prototypes: +file_operations +=============== + +prototypes:: + loff_t (*llseek) (struct file *, loff_t, int); ssize_t (*read) (struct file *, char __user *, size_t, loff_t *); ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *); @@ -455,7 +531,6 @@ prototypes: size_t, unsigned int); int (*setlease)(struct file *, long, struct file_lock **, void **); long (*fallocate)(struct file *, int, loff_t, loff_t); -}; locking rules: All may block. @@ -490,8 +565,11 @@ in sys_read() and friends. the lease within the individual filesystem to record the result of the operation ---------------------------- dquot_operations ------------------------------- -prototypes: +dquot_operations +================ + +prototypes:: + int (*write_dquot) (struct dquot *); int (*acquire_dquot) (struct dquot *); int (*release_dquot) (struct dquot *); @@ -503,20 +581,26 @@ a proper locking wrt the filesystem and call the generic quota operations. What filesystem should expect from the generic quota functions: - FS recursion Held locks when called +============== ============ ========================= +ops FS recursion Held locks when called +============== ============ ========================= write_dquot: yes dqonoff_sem or dqptr_sem acquire_dquot: yes dqonoff_sem or dqptr_sem release_dquot: yes dqonoff_sem or dqptr_sem mark_dirty: no - write_info: yes dqonoff_sem +============== ============ ========================= FS recursion means calling ->quota_read() and ->quota_write() from superblock operations. More details about quota locking can be found in fs/dquot.c. ---------------------------- vm_operations_struct ----------------------------- -prototypes: +vm_operations_struct +==================== + +prototypes:: + void (*open)(struct vm_area_struct*); void (*close)(struct vm_area_struct*); vm_fault_t (*fault)(struct vm_area_struct*, struct vm_fault *); @@ -525,7 +609,10 @@ prototypes: int (*access)(struct vm_area_struct *, unsigned long, void*, int, int); locking rules: - mmap_sem PageLocked(page) + +============= ======== =========================== +ops mmap_sem PageLocked(page) +============= ======== =========================== open: yes close: yes fault: yes can return with page locked @@ -533,8 +620,9 @@ map_pages: yes page_mkwrite: yes can return with page locked pfn_mkwrite: yes access: yes +============= ======== =========================== - ->fault() is called when a previously not present pte is about +->fault() is called when a previously not present pte is about to be faulted in. The filesystem must find and return the page associated with the passed in "pgoff" in the vm_fault structure. If it is possible that the page may be truncated and/or invalidated, then the filesystem must lock @@ -542,7 +630,7 @@ the page, then ensure it is not already truncated (the page lock will block subsequent truncate), and then return with VM_FAULT_LOCKED, and the page locked. The VM will unlock the page. - ->map_pages() is called when VM asks to map easy accessible pages. +->map_pages() is called when VM asks to map easy accessible pages. Filesystem should find and map pages associated with offsets from "start_pgoff" till "end_pgoff". ->map_pages() is called with page table locked and must not block. If it's not possible to reach a page without blocking, @@ -551,25 +639,26 @@ page table entry. Pointer to entry associated with the page is passed in "pte" field in vm_fault structure. Pointers to entries for other offsets should be calculated relative to "pte". - ->page_mkwrite() is called when a previously read-only pte is +->page_mkwrite() is called when a previously read-only pte is about to become writeable. The filesystem again must ensure that there are no truncate/invalidate races, and then return with the page locked. If the page has been truncated, the filesystem should not look up a new page like the ->fault() handler, but simply return with VM_FAULT_NOPAGE, which will cause the VM to retry the fault. - ->pfn_mkwrite() is the same as page_mkwrite but when the pte is +->pfn_mkwrite() is the same as page_mkwrite but when the pte is VM_PFNMAP or VM_MIXEDMAP with a page-less entry. Expected return is VM_FAULT_NOPAGE. Or one of the VM_FAULT_ERROR types. The default behavior after this call is to make the pte read-write, unless pfn_mkwrite returns an error. - ->access() is called when get_user_pages() fails in +->access() is called when get_user_pages() fails in access_process_vm(), typically used to debug a process through /proc/pid/mem or ptrace. This function is needed only for VM_IO | VM_PFNMAP VMAs. -================================================================================ +-------------------------------------------------------------------------------- + Dubious stuff (if you break something or notice that it is broken and do not fix it yourself diff --git a/Documentation/filesystems/nfs/Exporting b/Documentation/filesystems/nfs/exporting.rst index 63889149f532..33d588a01ace 100644 --- a/Documentation/filesystems/nfs/Exporting +++ b/Documentation/filesystems/nfs/exporting.rst @@ -1,3 +1,4 @@ +:orphan: Making Filesystems Exportable ============================= @@ -42,9 +43,9 @@ filehandle fragment, there is no automatic creation of a path prefix for the object. This leads to two related but distinct features of the dcache that are not needed for normal filesystem access. -1/ The dcache must sometimes contain objects that are not part of the +1. The dcache must sometimes contain objects that are not part of the proper prefix. i.e that are not connected to the root. -2/ The dcache must be prepared for a newly found (via ->lookup) directory +2. The dcache must be prepared for a newly found (via ->lookup) directory to already have a (non-connected) dentry, and must be able to move that dentry into place (based on the parent and name in the ->lookup). This is particularly needed for directories as @@ -52,7 +53,7 @@ the dcache that are not needed for normal filesystem access. To implement these features, the dcache has: -a/ A dentry flag DCACHE_DISCONNECTED which is set on +a. A dentry flag DCACHE_DISCONNECTED which is set on any dentry that might not be part of the proper prefix. This is set when anonymous dentries are created, and cleared when a dentry is noticed to be a child of a dentry which is in the proper @@ -71,48 +72,52 @@ a/ A dentry flag DCACHE_DISCONNECTED which is set on dentries. That guarantees that we won't need to hunt them down upon umount. -b/ A primitive for creation of secondary roots - d_obtain_root(inode). +b. A primitive for creation of secondary roots - d_obtain_root(inode). Those do _not_ bear DCACHE_DISCONNECTED. They are placed on the per-superblock list (->s_roots), so they can be located at umount time for eviction purposes. -c/ Helper routines to allocate anonymous dentries, and to help attach +c. Helper routines to allocate anonymous dentries, and to help attach loose directory dentries at lookup time. They are: + d_obtain_alias(inode) will return a dentry for the given inode. If the inode already has a dentry, one of those is returned. + If it doesn't, a new anonymous (IS_ROOT and - DCACHE_DISCONNECTED) dentry is allocated and attached. + DCACHE_DISCONNECTED) dentry is allocated and attached. + In the case of a directory, care is taken that only one dentry can ever be attached. + d_splice_alias(inode, dentry) will introduce a new dentry into the tree; either the passed-in dentry or a preexisting alias for the given inode (such as an anonymous one created by d_obtain_alias), if appropriate. It returns NULL when the passed-in dentry is used, following the calling convention of ->lookup. - + Filesystem Issues ----------------- For a filesystem to be exportable it must: - - 1/ provide the filehandle fragment routines described below. - 2/ make sure that d_splice_alias is used rather than d_add + + 1. provide the filehandle fragment routines described below. + 2. make sure that d_splice_alias is used rather than d_add when ->lookup finds an inode for a given parent and name. - If inode is NULL, d_splice_alias(inode, dentry) is equivalent to + If inode is NULL, d_splice_alias(inode, dentry) is equivalent to:: d_add(dentry, inode), NULL Similarly, d_splice_alias(ERR_PTR(err), dentry) = ERR_PTR(err) - Typically the ->lookup routine will simply end with a: + Typically the ->lookup routine will simply end with a:: return d_splice_alias(inode, dentry); } - A file system implementation declares that instances of the filesystem +A file system implementation declares that instances of the filesystem are exportable by setting the s_export_op field in the struct super_block. This field must point to a "struct export_operations" struct which has the following members: diff --git a/Documentation/filesystems/porting b/Documentation/filesystems/porting deleted file mode 100644 index 6b7a41cfcaed..000000000000 --- a/Documentation/filesystems/porting +++ /dev/null @@ -1,686 +0,0 @@ -Changes since 2.5.0: - ---- -[recommended] - -New helpers: sb_bread(), sb_getblk(), sb_find_get_block(), set_bh(), - sb_set_blocksize() and sb_min_blocksize(). - -Use them. - -(sb_find_get_block() replaces 2.4's get_hash_table()) - ---- -[recommended] - -New methods: ->alloc_inode() and ->destroy_inode(). - -Remove inode->u.foo_inode_i -Declare - struct foo_inode_info { - /* fs-private stuff */ - struct inode vfs_inode; - }; - static inline struct foo_inode_info *FOO_I(struct inode *inode) - { - return list_entry(inode, struct foo_inode_info, vfs_inode); - } - -Use FOO_I(inode) instead of &inode->u.foo_inode_i; - -Add foo_alloc_inode() and foo_destroy_inode() - the former should allocate -foo_inode_info and return the address of ->vfs_inode, the latter should free -FOO_I(inode) (see in-tree filesystems for examples). - -Make them ->alloc_inode and ->destroy_inode in your super_operations. - -Keep in mind that now you need explicit initialization of private data -typically between calling iget_locked() and unlocking the inode. - -At some point that will become mandatory. - ---- -[mandatory] - -Change of file_system_type method (->read_super to ->get_sb) - -->read_super() is no more. Ditto for DECLARE_FSTYPE and DECLARE_FSTYPE_DEV. - -Turn your foo_read_super() into a function that would return 0 in case of -success and negative number in case of error (-EINVAL unless you have more -informative error value to report). Call it foo_fill_super(). Now declare - -int foo_get_sb(struct file_system_type *fs_type, - int flags, const char *dev_name, void *data, struct vfsmount *mnt) -{ - return get_sb_bdev(fs_type, flags, dev_name, data, foo_fill_super, - mnt); -} - -(or similar with s/bdev/nodev/ or s/bdev/single/, depending on the kind of -filesystem). - -Replace DECLARE_FSTYPE... with explicit initializer and have ->get_sb set as -foo_get_sb. - ---- -[mandatory] - -Locking change: ->s_vfs_rename_sem is taken only by cross-directory renames. -Most likely there is no need to change anything, but if you relied on -global exclusion between renames for some internal purpose - you need to -change your internal locking. Otherwise exclusion warranties remain the -same (i.e. parents and victim are locked, etc.). - ---- -[informational] - -Now we have the exclusion between ->lookup() and directory removal (by -->rmdir() and ->rename()). If you used to need that exclusion and do -it by internal locking (most of filesystems couldn't care less) - you -can relax your locking. - ---- -[mandatory] - -->lookup(), ->truncate(), ->create(), ->unlink(), ->mknod(), ->mkdir(), -->rmdir(), ->link(), ->lseek(), ->symlink(), ->rename() -and ->readdir() are called without BKL now. Grab it on entry, drop upon return -- that will guarantee the same locking you used to have. If your method or its -parts do not need BKL - better yet, now you can shift lock_kernel() and -unlock_kernel() so that they would protect exactly what needs to be -protected. - ---- -[mandatory] - -BKL is also moved from around sb operations. BKL should have been shifted into -individual fs sb_op functions. If you don't need it, remove it. - ---- -[informational] - -check for ->link() target not being a directory is done by callers. Feel -free to drop it... - ---- -[informational] - -->link() callers hold ->i_mutex on the object we are linking to. Some of your -problems might be over... - ---- -[mandatory] - -new file_system_type method - kill_sb(superblock). If you are converting -an existing filesystem, set it according to ->fs_flags: - FS_REQUIRES_DEV - kill_block_super - FS_LITTER - kill_litter_super - neither - kill_anon_super -FS_LITTER is gone - just remove it from fs_flags. - ---- -[mandatory] - - FS_SINGLE is gone (actually, that had happened back when ->get_sb() -went in - and hadn't been documented ;-/). Just remove it from fs_flags -(and see ->get_sb() entry for other actions). - ---- -[mandatory] - -->setattr() is called without BKL now. Caller _always_ holds ->i_mutex, so -watch for ->i_mutex-grabbing code that might be used by your ->setattr(). -Callers of notify_change() need ->i_mutex now. - ---- -[recommended] - -New super_block field "struct export_operations *s_export_op" for -explicit support for exporting, e.g. via NFS. The structure is fully -documented at its declaration in include/linux/fs.h, and in -Documentation/filesystems/nfs/Exporting. - -Briefly it allows for the definition of decode_fh and encode_fh operations -to encode and decode filehandles, and allows the filesystem to use -a standard helper function for decode_fh, and provide file-system specific -support for this helper, particularly get_parent. - -It is planned that this will be required for exporting once the code -settles down a bit. - -[mandatory] - -s_export_op is now required for exporting a filesystem. -isofs, ext2, ext3, resierfs, fat -can be used as examples of very different filesystems. - ---- -[mandatory] - -iget4() and the read_inode2 callback have been superseded by iget5_locked() -which has the following prototype, - - struct inode *iget5_locked(struct super_block *sb, unsigned long ino, - int (*test)(struct inode *, void *), - int (*set)(struct inode *, void *), - void *data); - -'test' is an additional function that can be used when the inode -number is not sufficient to identify the actual file object. 'set' -should be a non-blocking function that initializes those parts of a -newly created inode to allow the test function to succeed. 'data' is -passed as an opaque value to both test and set functions. - -When the inode has been created by iget5_locked(), it will be returned with the -I_NEW flag set and will still be locked. The filesystem then needs to finalize -the initialization. Once the inode is initialized it must be unlocked by -calling unlock_new_inode(). - -The filesystem is responsible for setting (and possibly testing) i_ino -when appropriate. There is also a simpler iget_locked function that -just takes the superblock and inode number as arguments and does the -test and set for you. - -e.g. - inode = iget_locked(sb, ino); - if (inode->i_state & I_NEW) { - err = read_inode_from_disk(inode); - if (err < 0) { - iget_failed(inode); - return err; - } - unlock_new_inode(inode); - } - -Note that if the process of setting up a new inode fails, then iget_failed() -should be called on the inode to render it dead, and an appropriate error -should be passed back to the caller. - ---- -[recommended] - -->getattr() finally getting used. See instances in nfs, minix, etc. - ---- -[mandatory] - -->revalidate() is gone. If your filesystem had it - provide ->getattr() -and let it call whatever you had as ->revlidate() + (for symlinks that -had ->revalidate()) add calls in ->follow_link()/->readlink(). - ---- -[mandatory] - -->d_parent changes are not protected by BKL anymore. Read access is safe -if at least one of the following is true: - * filesystem has no cross-directory rename() - * we know that parent had been locked (e.g. we are looking at -->d_parent of ->lookup() argument). - * we are called from ->rename(). - * the child's ->d_lock is held -Audit your code and add locking if needed. Notice that any place that is -not protected by the conditions above is risky even in the old tree - you -had been relying on BKL and that's prone to screwups. Old tree had quite -a few holes of that kind - unprotected access to ->d_parent leading to -anything from oops to silent memory corruption. - ---- -[mandatory] - - FS_NOMOUNT is gone. If you use it - just set SB_NOUSER in flags -(see rootfs for one kind of solution and bdev/socket/pipe for another). - ---- -[recommended] - - Use bdev_read_only(bdev) instead of is_read_only(kdev). The latter -is still alive, but only because of the mess in drivers/s390/block/dasd.c. -As soon as it gets fixed is_read_only() will die. - ---- -[mandatory] - -->permission() is called without BKL now. Grab it on entry, drop upon -return - that will guarantee the same locking you used to have. If -your method or its parts do not need BKL - better yet, now you can -shift lock_kernel() and unlock_kernel() so that they would protect -exactly what needs to be protected. - ---- -[mandatory] - -->statfs() is now called without BKL held. BKL should have been -shifted into individual fs sb_op functions where it's not clear that -it's safe to remove it. If you don't need it, remove it. - ---- -[mandatory] - - is_read_only() is gone; use bdev_read_only() instead. - ---- -[mandatory] - - destroy_buffers() is gone; use invalidate_bdev(). - ---- -[mandatory] - - fsync_dev() is gone; use fsync_bdev(). NOTE: lvm breakage is -deliberate; as soon as struct block_device * is propagated in a reasonable -way by that code fixing will become trivial; until then nothing can be -done. - -[mandatory] - - block truncatation on error exit from ->write_begin, and ->direct_IO -moved from generic methods (block_write_begin, cont_write_begin, -nobh_write_begin, blockdev_direct_IO*) to callers. Take a look at -ext2_write_failed and callers for an example. - -[mandatory] - - ->truncate is gone. The whole truncate sequence needs to be -implemented in ->setattr, which is now mandatory for filesystems -implementing on-disk size changes. Start with a copy of the old inode_setattr -and vmtruncate, and the reorder the vmtruncate + foofs_vmtruncate sequence to -be in order of zeroing blocks using block_truncate_page or similar helpers, -size update and on finally on-disk truncation which should not fail. -setattr_prepare (which used to be inode_change_ok) now includes the size checks -for ATTR_SIZE and must be called in the beginning of ->setattr unconditionally. - -[mandatory] - - ->clear_inode() and ->delete_inode() are gone; ->evict_inode() should -be used instead. It gets called whenever the inode is evicted, whether it has -remaining links or not. Caller does *not* evict the pagecache or inode-associated -metadata buffers; the method has to use truncate_inode_pages_final() to get rid -of those. Caller makes sure async writeback cannot be running for the inode while -(or after) ->evict_inode() is called. - - ->drop_inode() returns int now; it's called on final iput() with -inode->i_lock held and it returns true if filesystems wants the inode to be -dropped. As before, generic_drop_inode() is still the default and it's been -updated appropriately. generic_delete_inode() is also alive and it consists -simply of return 1. Note that all actual eviction work is done by caller after -->drop_inode() returns. - - As before, clear_inode() must be called exactly once on each call of -->evict_inode() (as it used to be for each call of ->delete_inode()). Unlike -before, if you are using inode-associated metadata buffers (i.e. -mark_buffer_dirty_inode()), it's your responsibility to call -invalidate_inode_buffers() before clear_inode(). - - NOTE: checking i_nlink in the beginning of ->write_inode() and bailing out -if it's zero is not *and* *never* *had* *been* enough. Final unlink() and iput() -may happen while the inode is in the middle of ->write_inode(); e.g. if you blindly -free the on-disk inode, you may end up doing that while ->write_inode() is writing -to it. - ---- -[mandatory] - - .d_delete() now only advises the dcache as to whether or not to cache -unreferenced dentries, and is now only called when the dentry refcount goes to -0. Even on 0 refcount transition, it must be able to tolerate being called 0, -1, or more times (eg. constant, idempotent). - ---- -[mandatory] - - .d_compare() calling convention and locking rules are significantly -changed. Read updated documentation in Documentation/filesystems/vfs.rst (and -look at examples of other filesystems) for guidance. - ---- -[mandatory] - - .d_hash() calling convention and locking rules are significantly -changed. Read updated documentation in Documentation/filesystems/vfs.rst (and -look at examples of other filesystems) for guidance. - ---- -[mandatory] - dcache_lock is gone, replaced by fine grained locks. See fs/dcache.c -for details of what locks to replace dcache_lock with in order to protect -particular things. Most of the time, a filesystem only needs ->d_lock, which -protects *all* the dcache state of a given dentry. - --- -[mandatory] - - Filesystems must RCU-free their inodes, if they can have been accessed -via rcu-walk path walk (basically, if the file can have had a path name in the -vfs namespace). - - Even though i_dentry and i_rcu share storage in a union, we will -initialize the former in inode_init_always(), so just leave it alone in -the callback. It used to be necessary to clean it there, but not anymore -(starting at 3.2). - --- -[recommended] - vfs now tries to do path walking in "rcu-walk mode", which avoids -atomic operations and scalability hazards on dentries and inodes (see -Documentation/filesystems/path-lookup.txt). d_hash and d_compare changes -(above) are examples of the changes required to support this. For more complex -filesystem callbacks, the vfs drops out of rcu-walk mode before the fs call, so -no changes are required to the filesystem. However, this is costly and loses -the benefits of rcu-walk mode. We will begin to add filesystem callbacks that -are rcu-walk aware, shown below. Filesystems should take advantage of this -where possible. - --- -[mandatory] - d_revalidate is a callback that is made on every path element (if -the filesystem provides it), which requires dropping out of rcu-walk mode. This -may now be called in rcu-walk mode (nd->flags & LOOKUP_RCU). -ECHILD should be -returned if the filesystem cannot handle rcu-walk. See -Documentation/filesystems/vfs.rst for more details. - - permission is an inode permission check that is called on many or all -directory inodes on the way down a path walk (to check for exec permission). It -must now be rcu-walk aware (mask & MAY_NOT_BLOCK). See -Documentation/filesystems/vfs.rst for more details. - --- -[mandatory] - In ->fallocate() you must check the mode option passed in. If your -filesystem does not support hole punching (deallocating space in the middle of a -file) you must return -EOPNOTSUPP if FALLOC_FL_PUNCH_HOLE is set in mode. -Currently you can only have FALLOC_FL_PUNCH_HOLE with FALLOC_FL_KEEP_SIZE set, -so the i_size should not change when hole punching, even when puching the end of -a file off. - --- -[mandatory] - ->get_sb() is gone. Switch to use of ->mount(). Typically it's just -a matter of switching from calling get_sb_... to mount_... and changing the -function type. If you were doing it manually, just switch from setting ->mnt_root -to some pointer to returning that pointer. On errors return ERR_PTR(...). - --- -[mandatory] - ->permission() and generic_permission()have lost flags -argument; instead of passing IPERM_FLAG_RCU we add MAY_NOT_BLOCK into mask. - generic_permission() has also lost the check_acl argument; ACL checking -has been taken to VFS and filesystems need to provide a non-NULL ->i_op->get_acl -to read an ACL from disk. - --- -[mandatory] - If you implement your own ->llseek() you must handle SEEK_HOLE and -SEEK_DATA. You can hanle this by returning -EINVAL, but it would be nicer to -support it in some way. The generic handler assumes that the entire file is -data and there is a virtual hole at the end of the file. So if the provided -offset is less than i_size and SEEK_DATA is specified, return the same offset. -If the above is true for the offset and you are given SEEK_HOLE, return the end -of the file. If the offset is i_size or greater return -ENXIO in either case. - -[mandatory] - If you have your own ->fsync() you must make sure to call -filemap_write_and_wait_range() so that all dirty pages are synced out properly. -You must also keep in mind that ->fsync() is not called with i_mutex held -anymore, so if you require i_mutex locking you must make sure to take it and -release it yourself. - --- -[mandatory] - d_alloc_root() is gone, along with a lot of bugs caused by code -misusing it. Replacement: d_make_root(inode). On success d_make_root(inode) -allocates and returns a new dentry instantiated with the passed in inode. -On failure NULL is returned and the passed in inode is dropped so the reference -to inode is consumed in all cases and failure handling need not do any cleanup -for the inode. If d_make_root(inode) is passed a NULL inode it returns NULL -and also requires no further error handling. Typical usage is: - - inode = foofs_new_inode(....); - s->s_root = d_make_root(inode); - if (!s->s_root) - /* Nothing needed for the inode cleanup */ - return -ENOMEM; - ... - --- -[mandatory] - The witch is dead! Well, 2/3 of it, anyway. ->d_revalidate() and -->lookup() do *not* take struct nameidata anymore; just the flags. --- -[mandatory] - ->create() doesn't take struct nameidata *; unlike the previous -two, it gets "is it an O_EXCL or equivalent?" boolean argument. Note that -local filesystems can ignore tha argument - they are guaranteed that the -object doesn't exist. It's remote/distributed ones that might care... --- -[mandatory] - FS_REVAL_DOT is gone; if you used to have it, add ->d_weak_revalidate() -in your dentry operations instead. --- -[mandatory] - vfs_readdir() is gone; switch to iterate_dir() instead --- -[mandatory] - ->readdir() is gone now; switch to ->iterate() -[mandatory] - vfs_follow_link has been removed. Filesystems must use nd_set_link - from ->follow_link for normal symlinks, or nd_jump_link for magic - /proc/<pid> style links. --- -[mandatory] - iget5_locked()/ilookup5()/ilookup5_nowait() test() callback used to be - called with both ->i_lock and inode_hash_lock held; the former is *not* - taken anymore, so verify that your callbacks do not rely on it (none - of the in-tree instances did). inode_hash_lock is still held, - of course, so they are still serialized wrt removal from inode hash, - as well as wrt set() callback of iget5_locked(). --- -[mandatory] - d_materialise_unique() is gone; d_splice_alias() does everything you - need now. Remember that they have opposite orders of arguments ;-/ --- -[mandatory] - f_dentry is gone; use f_path.dentry, or, better yet, see if you can avoid - it entirely. --- -[mandatory] - never call ->read() and ->write() directly; use __vfs_{read,write} or - wrappers; instead of checking for ->write or ->read being NULL, look for - FMODE_CAN_{WRITE,READ} in file->f_mode. --- -[mandatory] - do _not_ use new_sync_{read,write} for ->read/->write; leave it NULL - instead. --- -[mandatory] - ->aio_read/->aio_write are gone. Use ->read_iter/->write_iter. ---- -[recommended] - for embedded ("fast") symlinks just set inode->i_link to wherever the - symlink body is and use simple_follow_link() as ->follow_link(). --- -[mandatory] - calling conventions for ->follow_link() have changed. Instead of returning - cookie and using nd_set_link() to store the body to traverse, we return - the body to traverse and store the cookie using explicit void ** argument. - nameidata isn't passed at all - nd_jump_link() doesn't need it and - nd_[gs]et_link() is gone. --- -[mandatory] - calling conventions for ->put_link() have changed. It gets inode instead of - dentry, it does not get nameidata at all and it gets called only when cookie - is non-NULL. Note that link body isn't available anymore, so if you need it, - store it as cookie. --- -[mandatory] - any symlink that might use page_follow_link_light/page_put_link() must - have inode_nohighmem(inode) called before anything might start playing with - its pagecache. No highmem pages should end up in the pagecache of such - symlinks. That includes any preseeding that might be done during symlink - creation. __page_symlink() will honour the mapping gfp flags, so once - you've done inode_nohighmem() it's safe to use, but if you allocate and - insert the page manually, make sure to use the right gfp flags. --- -[mandatory] - ->follow_link() is replaced with ->get_link(); same API, except that - * ->get_link() gets inode as a separate argument - * ->get_link() may be called in RCU mode - in that case NULL - dentry is passed --- -[mandatory] - ->get_link() gets struct delayed_call *done now, and should do - set_delayed_call() where it used to set *cookie. - ->put_link() is gone - just give the destructor to set_delayed_call() - in ->get_link(). --- -[mandatory] - ->getxattr() and xattr_handler.get() get dentry and inode passed separately. - dentry might be yet to be attached to inode, so do _not_ use its ->d_inode - in the instances. Rationale: !@#!@# security_d_instantiate() needs to be - called before we attach dentry to inode. --- -[mandatory] - symlinks are no longer the only inodes that do *not* have i_bdev/i_cdev/ - i_pipe/i_link union zeroed out at inode eviction. As the result, you can't - assume that non-NULL value in ->i_nlink at ->destroy_inode() implies that - it's a symlink. Checking ->i_mode is really needed now. In-tree we had - to fix shmem_destroy_callback() that used to take that kind of shortcut; - watch out, since that shortcut is no longer valid. --- -[mandatory] - ->i_mutex is replaced with ->i_rwsem now. inode_lock() et.al. work as - they used to - they just take it exclusive. However, ->lookup() may be - called with parent locked shared. Its instances must not - * use d_instantiate) and d_rehash() separately - use d_add() or - d_splice_alias() instead. - * use d_rehash() alone - call d_add(new_dentry, NULL) instead. - * in the unlikely case when (read-only) access to filesystem - data structures needs exclusion for some reason, arrange it - yourself. None of the in-tree filesystems needed that. - * rely on ->d_parent and ->d_name not changing after dentry has - been fed to d_add() or d_splice_alias(). Again, none of the - in-tree instances relied upon that. - We are guaranteed that lookups of the same name in the same directory - will not happen in parallel ("same" in the sense of your ->d_compare()). - Lookups on different names in the same directory can and do happen in - parallel now. --- -[recommended] - ->iterate_shared() is added; it's a parallel variant of ->iterate(). - Exclusion on struct file level is still provided (as well as that - between it and lseek on the same struct file), but if your directory - has been opened several times, you can get these called in parallel. - Exclusion between that method and all directory-modifying ones is - still provided, of course. - - Often enough ->iterate() can serve as ->iterate_shared() without any - changes - it is a read-only operation, after all. If you have any - per-inode or per-dentry in-core data structures modified by ->iterate(), - you might need something to serialize the access to them. If you - do dcache pre-seeding, you'll need to switch to d_alloc_parallel() for - that; look for in-tree examples. - - Old method is only used if the new one is absent; eventually it will - be removed. Switch while you still can; the old one won't stay. --- -[mandatory] - ->atomic_open() calls without O_CREAT may happen in parallel. --- -[mandatory] - ->setxattr() and xattr_handler.set() get dentry and inode passed separately. - dentry might be yet to be attached to inode, so do _not_ use its ->d_inode - in the instances. Rationale: !@#!@# security_d_instantiate() needs to be - called before we attach dentry to inode and !@#!@##!@$!$#!@#$!@$!@$ smack - ->d_instantiate() uses not just ->getxattr() but ->setxattr() as well. --- -[mandatory] - ->d_compare() doesn't get parent as a separate argument anymore. If you - used it for finding the struct super_block involved, dentry->d_sb will - work just as well; if it's something more complicated, use dentry->d_parent. - Just be careful not to assume that fetching it more than once will yield - the same value - in RCU mode it could change under you. --- -[mandatory] - ->rename() has an added flags argument. Any flags not handled by the - filesystem should result in EINVAL being returned. --- -[recommended] - ->readlink is optional for symlinks. Don't set, unless filesystem needs - to fake something for readlink(2). --- -[mandatory] - ->getattr() is now passed a struct path rather than a vfsmount and - dentry separately, and it now has request_mask and query_flags arguments - to specify the fields and sync type requested by statx. Filesystems not - supporting any statx-specific features may ignore the new arguments. --- -[mandatory] - ->atomic_open() calling conventions have changed. Gone is int *opened, - along with FILE_OPENED/FILE_CREATED. In place of those we have - FMODE_OPENED/FMODE_CREATED, set in file->f_mode. Additionally, return - value for 'called finish_no_open(), open it yourself' case has become - 0, not 1. Since finish_no_open() itself is returning 0 now, that part - does not need any changes in ->atomic_open() instances. --- -[mandatory] - alloc_file() has become static now; two wrappers are to be used instead. - alloc_file_pseudo(inode, vfsmount, name, flags, ops) is for the cases - when dentry needs to be created; that's the majority of old alloc_file() - users. Calling conventions: on success a reference to new struct file - is returned and callers reference to inode is subsumed by that. On - failure, ERR_PTR() is returned and no caller's references are affected, - so the caller needs to drop the inode reference it held. - alloc_file_clone(file, flags, ops) does not affect any caller's references. - On success you get a new struct file sharing the mount/dentry with the - original, on failure - ERR_PTR(). --- -[mandatory] - ->clone_file_range() and ->dedupe_file_range have been replaced with - ->remap_file_range(). See Documentation/filesystems/vfs.rst for more - information. --- -[recommended] - ->lookup() instances doing an equivalent of - if (IS_ERR(inode)) - return ERR_CAST(inode); - return d_splice_alias(inode, dentry); - don't need to bother with the check - d_splice_alias() will do the - right thing when given ERR_PTR(...) as inode. Moreover, passing NULL - inode to d_splice_alias() will also do the right thing (equivalent of - d_add(dentry, NULL); return NULL;), so that kind of special cases - also doesn't need a separate treatment. --- -[strongly recommended] - take the RCU-delayed parts of ->destroy_inode() into a new method - - ->free_inode(). If ->destroy_inode() becomes empty - all the better, - just get rid of it. Synchronous work (e.g. the stuff that can't - be done from an RCU callback, or any WARN_ON() where we want the - stack trace) *might* be movable to ->evict_inode(); however, - that goes only for the things that are not needed to balance something - done by ->alloc_inode(). IOW, if it's cleaning up the stuff that - might have accumulated over the life of in-core inode, ->evict_inode() - might be a fit. - - Rules for inode destruction: - * if ->destroy_inode() is non-NULL, it gets called - * if ->free_inode() is non-NULL, it gets scheduled by call_rcu() - * combination of NULL ->destroy_inode and NULL ->free_inode is - treated as NULL/free_inode_nonrcu, to preserve the compatibility. - - Note that the callback (be it via ->free_inode() or explicit call_rcu() - in ->destroy_inode()) is *NOT* ordered wrt superblock destruction; - as the matter of fact, the superblock and all associated structures - might be already gone. The filesystem driver is guaranteed to be still - there, but that's it. Freeing memory in the callback is fine; doing - more than that is possible, but requires a lot of care and is best - avoided. --- -[mandatory] - DCACHE_RCUACCESS is gone; having an RCU delay on dentry freeing is the - default. DCACHE_NORCU opts out, and only d_alloc_pseudo() has any - business doing so. --- -[mandatory] - d_alloc_pseudo() is internal-only; uses outside of alloc_file_pseudo() are - very suspect (and won't work in modules). Such uses are very likely to - be misspelled d_alloc_anon(). diff --git a/Documentation/filesystems/porting.rst b/Documentation/filesystems/porting.rst new file mode 100644 index 000000000000..f18506083ced --- /dev/null +++ b/Documentation/filesystems/porting.rst @@ -0,0 +1,852 @@ +==================== +Changes since 2.5.0: +==================== + +--- + +**recommended** + +New helpers: sb_bread(), sb_getblk(), sb_find_get_block(), set_bh(), +sb_set_blocksize() and sb_min_blocksize(). + +Use them. + +(sb_find_get_block() replaces 2.4's get_hash_table()) + +--- + +**recommended** + +New methods: ->alloc_inode() and ->destroy_inode(). + +Remove inode->u.foo_inode_i + +Declare:: + + struct foo_inode_info { + /* fs-private stuff */ + struct inode vfs_inode; + }; + static inline struct foo_inode_info *FOO_I(struct inode *inode) + { + return list_entry(inode, struct foo_inode_info, vfs_inode); + } + +Use FOO_I(inode) instead of &inode->u.foo_inode_i; + +Add foo_alloc_inode() and foo_destroy_inode() - the former should allocate +foo_inode_info and return the address of ->vfs_inode, the latter should free +FOO_I(inode) (see in-tree filesystems for examples). + +Make them ->alloc_inode and ->destroy_inode in your super_operations. + +Keep in mind that now you need explicit initialization of private data +typically between calling iget_locked() and unlocking the inode. + +At some point that will become mandatory. + +--- + +**mandatory** + +Change of file_system_type method (->read_super to ->get_sb) + +->read_super() is no more. Ditto for DECLARE_FSTYPE and DECLARE_FSTYPE_DEV. + +Turn your foo_read_super() into a function that would return 0 in case of +success and negative number in case of error (-EINVAL unless you have more +informative error value to report). Call it foo_fill_super(). Now declare:: + + int foo_get_sb(struct file_system_type *fs_type, + int flags, const char *dev_name, void *data, struct vfsmount *mnt) + { + return get_sb_bdev(fs_type, flags, dev_name, data, foo_fill_super, + mnt); + } + +(or similar with s/bdev/nodev/ or s/bdev/single/, depending on the kind of +filesystem). + +Replace DECLARE_FSTYPE... with explicit initializer and have ->get_sb set as +foo_get_sb. + +--- + +**mandatory** + +Locking change: ->s_vfs_rename_sem is taken only by cross-directory renames. +Most likely there is no need to change anything, but if you relied on +global exclusion between renames for some internal purpose - you need to +change your internal locking. Otherwise exclusion warranties remain the +same (i.e. parents and victim are locked, etc.). + +--- + +**informational** + +Now we have the exclusion between ->lookup() and directory removal (by +->rmdir() and ->rename()). If you used to need that exclusion and do +it by internal locking (most of filesystems couldn't care less) - you +can relax your locking. + +--- + +**mandatory** + +->lookup(), ->truncate(), ->create(), ->unlink(), ->mknod(), ->mkdir(), +->rmdir(), ->link(), ->lseek(), ->symlink(), ->rename() +and ->readdir() are called without BKL now. Grab it on entry, drop upon return +- that will guarantee the same locking you used to have. If your method or its +parts do not need BKL - better yet, now you can shift lock_kernel() and +unlock_kernel() so that they would protect exactly what needs to be +protected. + +--- + +**mandatory** + +BKL is also moved from around sb operations. BKL should have been shifted into +individual fs sb_op functions. If you don't need it, remove it. + +--- + +**informational** + +check for ->link() target not being a directory is done by callers. Feel +free to drop it... + +--- + +**informational** + +->link() callers hold ->i_mutex on the object we are linking to. Some of your +problems might be over... + +--- + +**mandatory** + +new file_system_type method - kill_sb(superblock). If you are converting +an existing filesystem, set it according to ->fs_flags:: + + FS_REQUIRES_DEV - kill_block_super + FS_LITTER - kill_litter_super + neither - kill_anon_super + +FS_LITTER is gone - just remove it from fs_flags. + +--- + +**mandatory** + +FS_SINGLE is gone (actually, that had happened back when ->get_sb() +went in - and hadn't been documented ;-/). Just remove it from fs_flags +(and see ->get_sb() entry for other actions). + +--- + +**mandatory** + +->setattr() is called without BKL now. Caller _always_ holds ->i_mutex, so +watch for ->i_mutex-grabbing code that might be used by your ->setattr(). +Callers of notify_change() need ->i_mutex now. + +--- + +**recommended** + +New super_block field ``struct export_operations *s_export_op`` for +explicit support for exporting, e.g. via NFS. The structure is fully +documented at its declaration in include/linux/fs.h, and in +Documentation/filesystems/nfs/exporting.rst. + +Briefly it allows for the definition of decode_fh and encode_fh operations +to encode and decode filehandles, and allows the filesystem to use +a standard helper function for decode_fh, and provide file-system specific +support for this helper, particularly get_parent. + +It is planned that this will be required for exporting once the code +settles down a bit. + +**mandatory** + +s_export_op is now required for exporting a filesystem. +isofs, ext2, ext3, resierfs, fat +can be used as examples of very different filesystems. + +--- + +**mandatory** + +iget4() and the read_inode2 callback have been superseded by iget5_locked() +which has the following prototype:: + + struct inode *iget5_locked(struct super_block *sb, unsigned long ino, + int (*test)(struct inode *, void *), + int (*set)(struct inode *, void *), + void *data); + +'test' is an additional function that can be used when the inode +number is not sufficient to identify the actual file object. 'set' +should be a non-blocking function that initializes those parts of a +newly created inode to allow the test function to succeed. 'data' is +passed as an opaque value to both test and set functions. + +When the inode has been created by iget5_locked(), it will be returned with the +I_NEW flag set and will still be locked. The filesystem then needs to finalize +the initialization. Once the inode is initialized it must be unlocked by +calling unlock_new_inode(). + +The filesystem is responsible for setting (and possibly testing) i_ino +when appropriate. There is also a simpler iget_locked function that +just takes the superblock and inode number as arguments and does the +test and set for you. + +e.g.:: + + inode = iget_locked(sb, ino); + if (inode->i_state & I_NEW) { + err = read_inode_from_disk(inode); + if (err < 0) { + iget_failed(inode); + return err; + } + unlock_new_inode(inode); + } + +Note that if the process of setting up a new inode fails, then iget_failed() +should be called on the inode to render it dead, and an appropriate error +should be passed back to the caller. + +--- + +**recommended** + +->getattr() finally getting used. See instances in nfs, minix, etc. + +--- + +**mandatory** + +->revalidate() is gone. If your filesystem had it - provide ->getattr() +and let it call whatever you had as ->revlidate() + (for symlinks that +had ->revalidate()) add calls in ->follow_link()/->readlink(). + +--- + +**mandatory** + +->d_parent changes are not protected by BKL anymore. Read access is safe +if at least one of the following is true: + + * filesystem has no cross-directory rename() + * we know that parent had been locked (e.g. we are looking at + ->d_parent of ->lookup() argument). + * we are called from ->rename(). + * the child's ->d_lock is held + +Audit your code and add locking if needed. Notice that any place that is +not protected by the conditions above is risky even in the old tree - you +had been relying on BKL and that's prone to screwups. Old tree had quite +a few holes of that kind - unprotected access to ->d_parent leading to +anything from oops to silent memory corruption. + +--- + +**mandatory** + +FS_NOMOUNT is gone. If you use it - just set SB_NOUSER in flags +(see rootfs for one kind of solution and bdev/socket/pipe for another). + +--- + +**recommended** + +Use bdev_read_only(bdev) instead of is_read_only(kdev). The latter +is still alive, but only because of the mess in drivers/s390/block/dasd.c. +As soon as it gets fixed is_read_only() will die. + +--- + +**mandatory** + +->permission() is called without BKL now. Grab it on entry, drop upon +return - that will guarantee the same locking you used to have. If +your method or its parts do not need BKL - better yet, now you can +shift lock_kernel() and unlock_kernel() so that they would protect +exactly what needs to be protected. + +--- + +**mandatory** + +->statfs() is now called without BKL held. BKL should have been +shifted into individual fs sb_op functions where it's not clear that +it's safe to remove it. If you don't need it, remove it. + +--- + +**mandatory** + +is_read_only() is gone; use bdev_read_only() instead. + +--- + +**mandatory** + +destroy_buffers() is gone; use invalidate_bdev(). + +--- + +**mandatory** + +fsync_dev() is gone; use fsync_bdev(). NOTE: lvm breakage is +deliberate; as soon as struct block_device * is propagated in a reasonable +way by that code fixing will become trivial; until then nothing can be +done. + +**mandatory** + +block truncatation on error exit from ->write_begin, and ->direct_IO +moved from generic methods (block_write_begin, cont_write_begin, +nobh_write_begin, blockdev_direct_IO*) to callers. Take a look at +ext2_write_failed and callers for an example. + +**mandatory** + +->truncate is gone. The whole truncate sequence needs to be +implemented in ->setattr, which is now mandatory for filesystems +implementing on-disk size changes. Start with a copy of the old inode_setattr +and vmtruncate, and the reorder the vmtruncate + foofs_vmtruncate sequence to +be in order of zeroing blocks using block_truncate_page or similar helpers, +size update and on finally on-disk truncation which should not fail. +setattr_prepare (which used to be inode_change_ok) now includes the size checks +for ATTR_SIZE and must be called in the beginning of ->setattr unconditionally. + +**mandatory** + +->clear_inode() and ->delete_inode() are gone; ->evict_inode() should +be used instead. It gets called whenever the inode is evicted, whether it has +remaining links or not. Caller does *not* evict the pagecache or inode-associated +metadata buffers; the method has to use truncate_inode_pages_final() to get rid +of those. Caller makes sure async writeback cannot be running for the inode while +(or after) ->evict_inode() is called. + +->drop_inode() returns int now; it's called on final iput() with +inode->i_lock held and it returns true if filesystems wants the inode to be +dropped. As before, generic_drop_inode() is still the default and it's been +updated appropriately. generic_delete_inode() is also alive and it consists +simply of return 1. Note that all actual eviction work is done by caller after +->drop_inode() returns. + +As before, clear_inode() must be called exactly once on each call of +->evict_inode() (as it used to be for each call of ->delete_inode()). Unlike +before, if you are using inode-associated metadata buffers (i.e. +mark_buffer_dirty_inode()), it's your responsibility to call +invalidate_inode_buffers() before clear_inode(). + +NOTE: checking i_nlink in the beginning of ->write_inode() and bailing out +if it's zero is not *and* *never* *had* *been* enough. Final unlink() and iput() +may happen while the inode is in the middle of ->write_inode(); e.g. if you blindly +free the on-disk inode, you may end up doing that while ->write_inode() is writing +to it. + +--- + +**mandatory** + +.d_delete() now only advises the dcache as to whether or not to cache +unreferenced dentries, and is now only called when the dentry refcount goes to +0. Even on 0 refcount transition, it must be able to tolerate being called 0, +1, or more times (eg. constant, idempotent). + +--- + +**mandatory** + +.d_compare() calling convention and locking rules are significantly +changed. Read updated documentation in Documentation/filesystems/vfs.rst (and +look at examples of other filesystems) for guidance. + +--- + +**mandatory** + +.d_hash() calling convention and locking rules are significantly +changed. Read updated documentation in Documentation/filesystems/vfs.rst (and +look at examples of other filesystems) for guidance. + +--- + +**mandatory** + +dcache_lock is gone, replaced by fine grained locks. See fs/dcache.c +for details of what locks to replace dcache_lock with in order to protect +particular things. Most of the time, a filesystem only needs ->d_lock, which +protects *all* the dcache state of a given dentry. + +--- + +**mandatory** + +Filesystems must RCU-free their inodes, if they can have been accessed +via rcu-walk path walk (basically, if the file can have had a path name in the +vfs namespace). + +Even though i_dentry and i_rcu share storage in a union, we will +initialize the former in inode_init_always(), so just leave it alone in +the callback. It used to be necessary to clean it there, but not anymore +(starting at 3.2). + +--- + +**recommended** + +vfs now tries to do path walking in "rcu-walk mode", which avoids +atomic operations and scalability hazards on dentries and inodes (see +Documentation/filesystems/path-lookup.txt). d_hash and d_compare changes +(above) are examples of the changes required to support this. For more complex +filesystem callbacks, the vfs drops out of rcu-walk mode before the fs call, so +no changes are required to the filesystem. However, this is costly and loses +the benefits of rcu-walk mode. We will begin to add filesystem callbacks that +are rcu-walk aware, shown below. Filesystems should take advantage of this +where possible. + +--- + +**mandatory** + +d_revalidate is a callback that is made on every path element (if +the filesystem provides it), which requires dropping out of rcu-walk mode. This +may now be called in rcu-walk mode (nd->flags & LOOKUP_RCU). -ECHILD should be +returned if the filesystem cannot handle rcu-walk. See +Documentation/filesystems/vfs.rst for more details. + +permission is an inode permission check that is called on many or all +directory inodes on the way down a path walk (to check for exec permission). It +must now be rcu-walk aware (mask & MAY_NOT_BLOCK). See +Documentation/filesystems/vfs.rst for more details. + +--- + +**mandatory** + +In ->fallocate() you must check the mode option passed in. If your +filesystem does not support hole punching (deallocating space in the middle of a +file) you must return -EOPNOTSUPP if FALLOC_FL_PUNCH_HOLE is set in mode. +Currently you can only have FALLOC_FL_PUNCH_HOLE with FALLOC_FL_KEEP_SIZE set, +so the i_size should not change when hole punching, even when puching the end of +a file off. + +--- + +**mandatory** + +->get_sb() is gone. Switch to use of ->mount(). Typically it's just +a matter of switching from calling ``get_sb_``... to ``mount_``... and changing +the function type. If you were doing it manually, just switch from setting +->mnt_root to some pointer to returning that pointer. On errors return +ERR_PTR(...). + +--- + +**mandatory** + +->permission() and generic_permission()have lost flags +argument; instead of passing IPERM_FLAG_RCU we add MAY_NOT_BLOCK into mask. + +generic_permission() has also lost the check_acl argument; ACL checking +has been taken to VFS and filesystems need to provide a non-NULL ->i_op->get_acl +to read an ACL from disk. + +--- + +**mandatory** + +If you implement your own ->llseek() you must handle SEEK_HOLE and +SEEK_DATA. You can hanle this by returning -EINVAL, but it would be nicer to +support it in some way. The generic handler assumes that the entire file is +data and there is a virtual hole at the end of the file. So if the provided +offset is less than i_size and SEEK_DATA is specified, return the same offset. +If the above is true for the offset and you are given SEEK_HOLE, return the end +of the file. If the offset is i_size or greater return -ENXIO in either case. + +**mandatory** + +If you have your own ->fsync() you must make sure to call +filemap_write_and_wait_range() so that all dirty pages are synced out properly. +You must also keep in mind that ->fsync() is not called with i_mutex held +anymore, so if you require i_mutex locking you must make sure to take it and +release it yourself. + +--- + +**mandatory** + +d_alloc_root() is gone, along with a lot of bugs caused by code +misusing it. Replacement: d_make_root(inode). On success d_make_root(inode) +allocates and returns a new dentry instantiated with the passed in inode. +On failure NULL is returned and the passed in inode is dropped so the reference +to inode is consumed in all cases and failure handling need not do any cleanup +for the inode. If d_make_root(inode) is passed a NULL inode it returns NULL +and also requires no further error handling. Typical usage is:: + + inode = foofs_new_inode(....); + s->s_root = d_make_root(inode); + if (!s->s_root) + /* Nothing needed for the inode cleanup */ + return -ENOMEM; + ... + +--- + +**mandatory** + +The witch is dead! Well, 2/3 of it, anyway. ->d_revalidate() and +->lookup() do *not* take struct nameidata anymore; just the flags. + +--- + +**mandatory** + +->create() doesn't take ``struct nameidata *``; unlike the previous +two, it gets "is it an O_EXCL or equivalent?" boolean argument. Note that +local filesystems can ignore tha argument - they are guaranteed that the +object doesn't exist. It's remote/distributed ones that might care... + +--- + +**mandatory** + +FS_REVAL_DOT is gone; if you used to have it, add ->d_weak_revalidate() +in your dentry operations instead. + +--- + +**mandatory** + +vfs_readdir() is gone; switch to iterate_dir() instead + +--- + +**mandatory** + +->readdir() is gone now; switch to ->iterate() + +**mandatory** + +vfs_follow_link has been removed. Filesystems must use nd_set_link +from ->follow_link for normal symlinks, or nd_jump_link for magic +/proc/<pid> style links. + +--- + +**mandatory** + +iget5_locked()/ilookup5()/ilookup5_nowait() test() callback used to be +called with both ->i_lock and inode_hash_lock held; the former is *not* +taken anymore, so verify that your callbacks do not rely on it (none +of the in-tree instances did). inode_hash_lock is still held, +of course, so they are still serialized wrt removal from inode hash, +as well as wrt set() callback of iget5_locked(). + +--- + +**mandatory** + +d_materialise_unique() is gone; d_splice_alias() does everything you +need now. Remember that they have opposite orders of arguments ;-/ + +--- + +**mandatory** + +f_dentry is gone; use f_path.dentry, or, better yet, see if you can avoid +it entirely. + +--- + +**mandatory** + +never call ->read() and ->write() directly; use __vfs_{read,write} or +wrappers; instead of checking for ->write or ->read being NULL, look for +FMODE_CAN_{WRITE,READ} in file->f_mode. + +--- + +**mandatory** + +do _not_ use new_sync_{read,write} for ->read/->write; leave it NULL +instead. + +--- + +**mandatory** + ->aio_read/->aio_write are gone. Use ->read_iter/->write_iter. + +--- + +**recommended** + +for embedded ("fast") symlinks just set inode->i_link to wherever the +symlink body is and use simple_follow_link() as ->follow_link(). + +--- + +**mandatory** + +calling conventions for ->follow_link() have changed. Instead of returning +cookie and using nd_set_link() to store the body to traverse, we return +the body to traverse and store the cookie using explicit void ** argument. +nameidata isn't passed at all - nd_jump_link() doesn't need it and +nd_[gs]et_link() is gone. + +--- + +**mandatory** + +calling conventions for ->put_link() have changed. It gets inode instead of +dentry, it does not get nameidata at all and it gets called only when cookie +is non-NULL. Note that link body isn't available anymore, so if you need it, +store it as cookie. + +--- + +**mandatory** + +any symlink that might use page_follow_link_light/page_put_link() must +have inode_nohighmem(inode) called before anything might start playing with +its pagecache. No highmem pages should end up in the pagecache of such +symlinks. That includes any preseeding that might be done during symlink +creation. __page_symlink() will honour the mapping gfp flags, so once +you've done inode_nohighmem() it's safe to use, but if you allocate and +insert the page manually, make sure to use the right gfp flags. + +--- + +**mandatory** + +->follow_link() is replaced with ->get_link(); same API, except that + + * ->get_link() gets inode as a separate argument + * ->get_link() may be called in RCU mode - in that case NULL + dentry is passed + +--- + +**mandatory** + +->get_link() gets struct delayed_call ``*done`` now, and should do +set_delayed_call() where it used to set ``*cookie``. + +->put_link() is gone - just give the destructor to set_delayed_call() +in ->get_link(). + +--- + +**mandatory** + +->getxattr() and xattr_handler.get() get dentry and inode passed separately. +dentry might be yet to be attached to inode, so do _not_ use its ->d_inode +in the instances. Rationale: !@#!@# security_d_instantiate() needs to be +called before we attach dentry to inode. + +--- + +**mandatory** + +symlinks are no longer the only inodes that do *not* have i_bdev/i_cdev/ +i_pipe/i_link union zeroed out at inode eviction. As the result, you can't +assume that non-NULL value in ->i_nlink at ->destroy_inode() implies that +it's a symlink. Checking ->i_mode is really needed now. In-tree we had +to fix shmem_destroy_callback() that used to take that kind of shortcut; +watch out, since that shortcut is no longer valid. + +--- + +**mandatory** + +->i_mutex is replaced with ->i_rwsem now. inode_lock() et.al. work as +they used to - they just take it exclusive. However, ->lookup() may be +called with parent locked shared. Its instances must not + + * use d_instantiate) and d_rehash() separately - use d_add() or + d_splice_alias() instead. + * use d_rehash() alone - call d_add(new_dentry, NULL) instead. + * in the unlikely case when (read-only) access to filesystem + data structures needs exclusion for some reason, arrange it + yourself. None of the in-tree filesystems needed that. + * rely on ->d_parent and ->d_name not changing after dentry has + been fed to d_add() or d_splice_alias(). Again, none of the + in-tree instances relied upon that. + +We are guaranteed that lookups of the same name in the same directory +will not happen in parallel ("same" in the sense of your ->d_compare()). +Lookups on different names in the same directory can and do happen in +parallel now. + +--- + +**recommended** + +->iterate_shared() is added; it's a parallel variant of ->iterate(). +Exclusion on struct file level is still provided (as well as that +between it and lseek on the same struct file), but if your directory +has been opened several times, you can get these called in parallel. +Exclusion between that method and all directory-modifying ones is +still provided, of course. + +Often enough ->iterate() can serve as ->iterate_shared() without any +changes - it is a read-only operation, after all. If you have any +per-inode or per-dentry in-core data structures modified by ->iterate(), +you might need something to serialize the access to them. If you +do dcache pre-seeding, you'll need to switch to d_alloc_parallel() for +that; look for in-tree examples. + +Old method is only used if the new one is absent; eventually it will +be removed. Switch while you still can; the old one won't stay. + +--- + +**mandatory** + +->atomic_open() calls without O_CREAT may happen in parallel. + +--- + +**mandatory** + +->setxattr() and xattr_handler.set() get dentry and inode passed separately. +dentry might be yet to be attached to inode, so do _not_ use its ->d_inode +in the instances. Rationale: !@#!@# security_d_instantiate() needs to be +called before we attach dentry to inode and !@#!@##!@$!$#!@#$!@$!@$ smack +->d_instantiate() uses not just ->getxattr() but ->setxattr() as well. + +--- + +**mandatory** + +->d_compare() doesn't get parent as a separate argument anymore. If you +used it for finding the struct super_block involved, dentry->d_sb will +work just as well; if it's something more complicated, use dentry->d_parent. +Just be careful not to assume that fetching it more than once will yield +the same value - in RCU mode it could change under you. + +--- + +**mandatory** + +->rename() has an added flags argument. Any flags not handled by the +filesystem should result in EINVAL being returned. + +--- + + +**recommended** + +->readlink is optional for symlinks. Don't set, unless filesystem needs +to fake something for readlink(2). + +--- + +**mandatory** + +->getattr() is now passed a struct path rather than a vfsmount and +dentry separately, and it now has request_mask and query_flags arguments +to specify the fields and sync type requested by statx. Filesystems not +supporting any statx-specific features may ignore the new arguments. + +--- + +**mandatory** + +->atomic_open() calling conventions have changed. Gone is ``int *opened``, +along with FILE_OPENED/FILE_CREATED. In place of those we have +FMODE_OPENED/FMODE_CREATED, set in file->f_mode. Additionally, return +value for 'called finish_no_open(), open it yourself' case has become +0, not 1. Since finish_no_open() itself is returning 0 now, that part +does not need any changes in ->atomic_open() instances. + +--- + +**mandatory** + +alloc_file() has become static now; two wrappers are to be used instead. +alloc_file_pseudo(inode, vfsmount, name, flags, ops) is for the cases +when dentry needs to be created; that's the majority of old alloc_file() +users. Calling conventions: on success a reference to new struct file +is returned and callers reference to inode is subsumed by that. On +failure, ERR_PTR() is returned and no caller's references are affected, +so the caller needs to drop the inode reference it held. +alloc_file_clone(file, flags, ops) does not affect any caller's references. +On success you get a new struct file sharing the mount/dentry with the +original, on failure - ERR_PTR(). + +--- + +**mandatory** + +->clone_file_range() and ->dedupe_file_range have been replaced with +->remap_file_range(). See Documentation/filesystems/vfs.rst for more +information. + +--- + +**recommended** + +->lookup() instances doing an equivalent of:: + + if (IS_ERR(inode)) + return ERR_CAST(inode); + return d_splice_alias(inode, dentry); + +don't need to bother with the check - d_splice_alias() will do the +right thing when given ERR_PTR(...) as inode. Moreover, passing NULL +inode to d_splice_alias() will also do the right thing (equivalent of +d_add(dentry, NULL); return NULL;), so that kind of special cases +also doesn't need a separate treatment. + +--- + +**strongly recommended** + +take the RCU-delayed parts of ->destroy_inode() into a new method - +->free_inode(). If ->destroy_inode() becomes empty - all the better, +just get rid of it. Synchronous work (e.g. the stuff that can't +be done from an RCU callback, or any WARN_ON() where we want the +stack trace) *might* be movable to ->evict_inode(); however, +that goes only for the things that are not needed to balance something +done by ->alloc_inode(). IOW, if it's cleaning up the stuff that +might have accumulated over the life of in-core inode, ->evict_inode() +might be a fit. + +Rules for inode destruction: + + * if ->destroy_inode() is non-NULL, it gets called + * if ->free_inode() is non-NULL, it gets scheduled by call_rcu() + * combination of NULL ->destroy_inode and NULL ->free_inode is + treated as NULL/free_inode_nonrcu, to preserve the compatibility. + +Note that the callback (be it via ->free_inode() or explicit call_rcu() +in ->destroy_inode()) is *NOT* ordered wrt superblock destruction; +as the matter of fact, the superblock and all associated structures +might be already gone. The filesystem driver is guaranteed to be still +there, but that's it. Freeing memory in the callback is fine; doing +more than that is possible, but requires a lot of care and is best +avoided. + +--- + +**mandatory** + +DCACHE_RCUACCESS is gone; having an RCU delay on dentry freeing is the +default. DCACHE_NORCU opts out, and only d_alloc_pseudo() has any +business doing so. + +--- + +**mandatory** + +d_alloc_pseudo() is internal-only; uses outside of alloc_file_pseudo() are +very suspect (and won't work in modules). Such uses are very likely to +be misspelled d_alloc_anon(). diff --git a/Documentation/filesystems/ubifs-authentication.md b/Documentation/filesystems/ubifs-authentication.rst index 23e698167141..6a9584f6ff46 100644 --- a/Documentation/filesystems/ubifs-authentication.md +++ b/Documentation/filesystems/ubifs-authentication.rst @@ -1,8 +1,11 @@ -% UBIFS Authentication -% sigma star gmbh -% 2018 +:orphan: -# Introduction +.. UBIFS Authentication +.. sigma star gmbh +.. 2018 + +Introduction +============ UBIFS utilizes the fscrypt framework to provide confidentiality for file contents and file names. This prevents attacks where an attacker is able to @@ -33,7 +36,8 @@ existing features like key derivation can be utilized. It should however also be possible to use UBIFS authentication without using encryption. -## MTD, UBI & UBIFS +MTD, UBI & UBIFS +---------------- On Linux, the MTD (Memory Technology Devices) subsystem provides a uniform interface to access raw flash devices. One of the more prominent subsystems that @@ -47,7 +51,7 @@ UBIFS is a filesystem for raw flash which operates on top of UBI. Thus, wear leveling and some flash specifics are left to UBI, while UBIFS focuses on scalability, performance and recoverability. - +:: +------------+ +*******+ +-----------+ +-----+ | | * UBIFS * | UBI-BLOCK | | ... | @@ -84,7 +88,8 @@ persisted onto the flash directly. More details on UBIFS can also be found in [UBIFS-WP]. -### UBIFS Index & Tree Node Cache +UBIFS Index & Tree Node Cache +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Basic on-flash UBIFS entities are called *nodes*. UBIFS knows different types of nodes. Eg. data nodes (`struct ubifs_data_node`) which store chunks of file @@ -118,17 +123,18 @@ on-flash filesystem structures like the index. On every commit, the TNC nodes marked as dirty are written to the flash to update the persisted index. -### Journal +Journal +~~~~~~~ To avoid wearing out the flash, the index is only persisted (*commited*) when -certain conditions are met (eg. `fsync(2)`). The journal is used to record +certain conditions are met (eg. ``fsync(2)``). The journal is used to record any changes (in form of inode nodes, data nodes etc.) between commits of the index. During mount, the journal is read from the flash and replayed onto the TNC (which will be created on-demand from the on-flash index). UBIFS reserves a bunch of LEBs just for the journal called *log area*. The amount of log area LEBs is configured on filesystem creation (using -`mkfs.ubifs`) and stored in the superblock node. The log area contains only +``mkfs.ubifs``) and stored in the superblock node. The log area contains only two types of nodes: *reference nodes* and *commit start nodes*. A commit start node is written whenever an index commit is performed. Reference nodes are written on every journal update. Each reference node points to the position of @@ -152,6 +158,7 @@ done for the last referenced LEB of the journal. Only this can become corrupt because of a power cut. If the recovery fails, UBIFS will not mount. An error for every other LEB will directly cause UBIFS to fail the mount operation. +:: | ---- LOG AREA ---- | ---------- MAIN AREA ------------ | @@ -172,10 +179,11 @@ for every other LEB will directly cause UBIFS to fail the mount operation. containing their buds -### LEB Property Tree/Table +LEB Property Tree/Table +~~~~~~~~~~~~~~~~~~~~~~~ The LEB property tree is used to store per-LEB information. This includes the -LEB type and amount of free and *dirty* (old, obsolete content) space [1] on +LEB type and amount of free and *dirty* (old, obsolete content) space [1]_ on the LEB. The type is important, because UBIFS never mixes index nodes with data nodes on a single LEB and thus each LEB has a specific purpose. This again is useful for free space calculations. See [UBIFS-WP] for more details. @@ -185,19 +193,21 @@ index. Due to its smaller size it is always written as one chunk on every commit. Thus, saving the LPT is an atomic operation. -[1] Since LEBs can only be appended and never overwritten, there is a -difference between free space ie. the remaining space left on the LEB to be -written to without erasing it and previously written content that is obsolete -but can't be overwritten without erasing the full LEB. +.. [1] Since LEBs can only be appended and never overwritten, there is a + difference between free space ie. the remaining space left on the LEB to be + written to without erasing it and previously written content that is obsolete + but can't be overwritten without erasing the full LEB. -# UBIFS Authentication +UBIFS Authentication +==================== This chapter introduces UBIFS authentication which enables UBIFS to verify the authenticity and integrity of metadata and file contents stored on flash. -## Threat Model +Threat Model +------------ UBIFS authentication enables detection of offline data modification. While it does not prevent it, it enables (trusted) code to check the integrity and @@ -224,7 +234,8 @@ Additional measures like secure boot and trusted boot have to be taken to ensure that only trusted code is executed on a device. -## Authentication +Authentication +-------------- To be able to fully trust data read from flash, all UBIFS data structures stored on flash are authenticated. That is: @@ -236,7 +247,8 @@ stored on flash are authenticated. That is: - The LPT which stores UBI LEB metadata which UBIFS uses for free space accounting -### Index Authentication +Index Authentication +~~~~~~~~~~~~~~~~~~~~ Through UBIFS' concept of a wandering tree, it already takes care of only updating and persisting changed parts from leaf node up to the root node @@ -260,6 +272,7 @@ include a hash. All other types of nodes will remain unchanged. This reduces the storage overhead which is precious for users of UBIFS (ie. embedded devices). +:: +---------------+ | Master Node | @@ -303,7 +316,8 @@ hashes to index nodes does not change this since each hash will be persisted atomically together with its respective node. -### Journal Authentication +Journal Authentication +~~~~~~~~~~~~~~~~~~~~~~ The journal is authenticated too. Since the journal is continuously written it is necessary to also add authentication information frequently to the @@ -316,7 +330,7 @@ of the hash chain. That way a journal can be authenticated up to the last authentication node. The tail of the journal which may not have a authentication node cannot be authenticated and is skipped during journal replay. -We get this picture for journal authentication: +We get this picture for journal authentication:: ,,,,,,,, ,......,........................................... @@ -352,7 +366,8 @@ the superblock struct. The superblock node is stored in LEB 0 and is only modified on feature flag or similar changes, but never on file changes. -### LPT Authentication +LPT Authentication +~~~~~~~~~~~~~~~~~~ The location of the LPT root node on the flash is stored in the UBIFS master node. Since the LPT is written and read atomically on every commit, there is @@ -363,7 +378,8 @@ be verified by verifying the authenticity of the master node and comparing the LTP hash stored there with the hash computed from the read on-flash LPT. -## Key Management +Key Management +-------------- For simplicity, UBIFS authentication uses a single key to compute the HMACs of superblock, master, commit start and reference nodes. This key has to be @@ -399,7 +415,8 @@ approach is similar to the approach proposed for fscrypt encryption policy v2 [FSCRYPT-POLICY2]. -# Future Extensions +Future Extensions +================= In certain cases where a vendor wants to provide an authenticated filesystem image to customers, it should be possible to do so without sharing the secret @@ -411,7 +428,8 @@ to the way the IMA/EVM subsystem deals with such situations. The HMAC key will then have to be provided beforehand in the normal way. -# References +References +========== [CRYPTSETUP2] http://www.saout.de/pipermail/dm-crypt/2017-November/005745.html diff --git a/Documentation/filesystems/ufs.txt b/Documentation/filesystems/ufs.txt deleted file mode 100644 index 7a602adeca2b..000000000000 --- a/Documentation/filesystems/ufs.txt +++ /dev/null @@ -1,60 +0,0 @@ -USING UFS -========= - -mount -t ufs -o ufstype=type_of_ufs device dir - - -UFS OPTIONS -=========== - -ufstype=type_of_ufs - UFS is a file system widely used in different operating systems. - The problem are differences among implementations. Features of - some implementations are undocumented, so its hard to recognize - type of ufs automatically. That's why user must specify type of - ufs manually by mount option ufstype. Possible values are: - - old old format of ufs - default value, supported as read-only - - 44bsd used in FreeBSD, NetBSD, OpenBSD - supported as read-write - - ufs2 used in FreeBSD 5.x - supported as read-write - - 5xbsd synonym for ufs2 - - sun used in SunOS (Solaris) - supported as read-write - - sunx86 used in SunOS for Intel (Solarisx86) - supported as read-write - - hp used in HP-UX - supported as read-only - - nextstep - used in NextStep - supported as read-only - - nextstep-cd - used for NextStep CDROMs (block_size == 2048) - supported as read-only - - openstep - used in OpenStep - supported as read-only - - -POSSIBLE PROBLEMS -================= - -See next section, if you have any. - - -BUG REPORTS -=========== - -Any ufs bug report you can send to daniel.pirkl@email.cz or -to dushistov@mail.ru (do not send partition tables bug reports). diff --git a/Documentation/filesystems/vfs.rst b/Documentation/filesystems/vfs.rst index 0f85ab21c2ca..7d4d09dd5e6d 100644 --- a/Documentation/filesystems/vfs.rst +++ b/Documentation/filesystems/vfs.rst @@ -20,7 +20,7 @@ kernel which allows different filesystem implementations to coexist. VFS system calls open(2), stat(2), read(2), write(2), chmod(2) and so on are called from a process context. Filesystem locking is described in -the document Documentation/filesystems/Locking. +the document Documentation/filesystems/locking.rst. Directory Entry Cache (dcache) |