GitLab

Copy-up the ancestors of the target.
Signed-off-by: J. R. Okajima <hooanon05g@gmail.com>

Because the copy-up operation is big and has many parameters and
functions, consolidate them in this commit.
Signed-off-by: J. R. Okajima <hooanon05g@gmail.com>

The functions for
- create the copy-up target file
- copy filedata
- copy metadata

In copying filedata, I had tried splice_direct() instead of repeating
read/write. Surprisingly, I could not see a big difference. So let's
keep this approach for a while. Someday SEEK_DATA/SEEK_HOLE become more
popular, it may help optimizing this read/write.
Signed-off-by: J. R. Okajima <hooanon05g@gmail.com>

The internal file read/write for copy-up in kernelspace.
Signed-off-by: J. R. Okajima <hooanon05g@gmail.com>

Basically copy-up is done by these steps using au_pin (in another
commit).
- lock the target parent mutex
- lookup a negative dentry with a whiteout-ed temporary unique name
- create it
- unlock the target parent mutex
- copy filedata
- copy metadata (inode attributes)
- lock the target parent mutex
- rename the temporary name to the target name
- unlock the target parent mutex

This commit contains step2 mainly.
I hope someday aufs uses O_TMPFILE for this.
Signed-off-by: J. R. Okajima <hooanon05g@gmail.com>

Copy the inode attributes between branches.
See also the document in this commit.
Signed-off-by: J. R. Okajima <hooanon05g@gmail.com>

To create/delete/rename files including copy-up, aufs acquires several
locks on the branch fs internally. These lock/unlock operations are
consolidated into struct au_pin in this commit.
au_pin handles
- LOCKDEP class
- re-validate/verify
- suspend/resume HNOTIFY

See also lookup.txt in later commit.
Signed-off-by: J. R. Okajima <hooanon05g@gmail.com>

Some filesystems are natively readonly. And aufs can make a few
optimization for them. This new attribute tells aufs.
Signed-off-by: J. R. Okajima <hooanon05g@gmail.com>

The functions to create/delete the opaque directory marker (called
'diropq') on the added writable branch.
For details, refer to previous commit.
Signed-off-by: J. R. Okajima <hooanon05g@gmail.com>

Actually prepare the whiteout bases on the adding writable branch.
For details, refer to previous commit.
Signed-off-by: J. R. Okajima <hooanon05g@gmail.com>

The writable branch prepares a few files and dirs for whiteouts.
For branch filesystems which doesn't support link(2), there is "nolwh"
attribute. On the branch which is specified this attribute, aufs never
try link(2) for whitout and always creat(2) it.
Signed-off-by: J. R. Okajima <hooanon05g@gmail.com>

While the whiteout on the writable branch have its effect
unconditionally (in latter commit), the one on the readonly branch can
have its effect only when this attribute is specified explicitly.
For the branch attributes, refer to the manual in aufs-util.git.
Signed-off-by: J. R. Okajima <hooanon05g@gmail.com>

In freeing aufs iinfo objects, it acquires the internal rw_sem (see
another commit in detail). Since iinfo can be freed anytime, a deadlock
may happen due to the rw_sem. To prevent this problem, this commit
introduces a flag per task.
This is another (very) ugly approach which I don't like.
Signed-off-by: J. R. Okajima <hooanon05g@gmail.com>

Aufs pseudo-link (plink) represents a virtual hardlink across the
branches. To implement the plink maintenance mode, aufs uses procfs.
See also the document in this commit.

There is an external user-space utility called 'auplink' in
aufs-util.git, which has these features.
- 'list' shows the pseudo-linked inode numbers and filenames.
- 'cpup' copies-up all pseudo-link to the writable branch.
- 'flush' calls 'cpup', and then 'mount -o remount,clean_plink=inum'
Signed-off-by: J. R. Okajima <hooanon05g@gmail.com>

When user accesses aufs via other than fs related systemcalls, aufs
needs to identify which superblock is the target.  Here is the trick.
It is just a list of aufs superblocks.  Such way will be procfs and
MagicSysRq key.  For MagicSysRq support, see the later commit.
This is a dirty approach which I don't like, but I just don't have
another idea.
Signed-off-by: J. R. Okajima <hooanon05g@gmail.com>

The marker to represent that the directory is opaque (stop digging
down the branch stack) is implemented as a special whiteout.

See also the document in previous commit.
Signed-off-by: J. R. Okajima <hooanon05g@gmail.com>

The whiteout represents a logical deletion.
Although the document in this commit mentioned about rmdir(2) and
rename(2) for dir, this commit doesn't contain such functions. They will
be added in later commits.
Signed-off-by: J. R. Okajima <hooanon05g@gmail.com>

As mentioned earlier, sometimes the size of XINO file is a problem.
Aufs has a feature to truncate it asynchronously using workqueue. But it
may not be so effective in some cases, and you may want to stop
discontiguous distribution of the inode numbers on branch fs.
See also the log in another commit.
Signed-off-by: J. R. Okajima <hooanon05g@gmail.com>

The branch path can be much longer and it is not suitable to print via
/proc/mounts as a part of mount options. Aufs can show it either
separately via sysfs or /proc/mounts (as a part of mount options).
This approach affects the lifetime of aufs objects and sbinfo contains
kobject (in another commit). Theoretically user can disable
CONFIG_SYSFS, but the lifetime management is always necessary. So
supporting sysfs is split into two files, sysaufs.c and sysfs.c.
sysaufs.c is always compiled, but sysfs.c is compiled only when
CONFIG_SYSFS is enabled.
Signed-off-by: J. R. Okajima <hooanon05g@gmail.com>

XINO and XIB files are read and written frequently after unlinked, and
it means that the remote filesystems are not suitable for them.
Additionally aufs shows their metadata via debugfs (in later commit).
To make it easier to do this, aufs expects branch filesystems to
maintain their i_size and i_blocks. And it means some filesystem are not
suitable for XINO.
Signed-off-by: J. R. Okajima <hooanon05g@gmail.com>

XINO and XIB files are to maintain the inode numbers in aufs
(cf. struct.txt and aufs manual in aufs-util.git).

XINO file contains just a sequence of the inode numbers, and their
offset in the file is real_inum x sizeof(inum).  So the size is limited
by s_maxbytes of the filesystem where XINO file is located.  In order to
support the larger inum, aufs stores XINO files as an internal array.

Sometimes the size of XINO file can be a problem, ie. too big,
particularly when XINO files are located on tmpfs. In this case, another
separate patch tmpfs-ino.patch in aufs4-standalone.git is recommended
(as well as vfs-ino.patch). The patch makes tmpfs to maintain inode
number within itself and suppress its discontiguous distribution.

See also the document in next commit.
Signed-off-by: J. R. Okajima <hooanon05g@gmail.com>

Aufs uses the workqueue both synchronously and asynchronously.
For sync-use-case, aufs uses its own specific wkq since doesn't want to
be disturbed by other tasks on the system. For async-use-case, aufs uses
the system global workqueue.
Aufs has to prevent itself to being unmounted during the async-task is
queued.

See also the document in this commit.
Signed-off-by: J. R. Okajima <hooanon05g@gmail.com>

Print various info about aufs branch and superblock. This feature is
enabled when CONFIG_AUFS_DEBUG and the module parameter 'debug' are set.
Signed-off-by: J. R. Okajima <hooanon05g@gmail.com>

For details, see previous commit.
Signed-off-by: J. R. Okajima <hooanon05g@gmail.com>

The branch object is managed by the sbinfo object as an element of its
internal array. The iinfo and dinfo objects contain the branch id, and
it will be used to implement the correct order in branch management
(add/del).

See also the documents in this commit.
Signed-off-by: J. R. Okajima <hooanon05g@gmail.com>

The generation of iinfo and dinfo inherit sbinfo's.
Also iinfo generation tracks the branch inode's generation to test the
matching after the branch management.
Signed-off-by: J. R. Okajima <hooanon05g@gmail.com>

The generation of aufs objects will be updated by the branch
management (add/del branches, by later commit), and aufs will refresh
the objects based upon the generation.
In refreshing dinfos, aufs will find all ancestors of the given dentry
and store the pointers in dynamically allocated memory. I don't think it
beautiful, but I don't have any other idea.
Signed-off-by: J. R. Okajima <hooanon05g@gmail.com>

The structure is very similar to inode and dentry infos (in previous
commits), but the internal array is for 'struct au_branch' instead of
'superblock.' Additionally the lifetime of 'struct au_sbinfo' is managed
by kobject since it will be connected to sysfs by later commit.
Signed-off-by: J. R. Okajima <hooanon05g@gmail.com>

The structure is very similar to aufs inode info (in previous commit).
Signed-off-by: J. R. Okajima <hooanon05g@gmail.com>

Print various info about aufs inode. This feature is enabled when
CONFIG_AUFS_DEBUG and the module parameter 'debug' are set.
Signed-off-by: J. R. Okajima <hooanon05g@gmail.com>

This parameter is available only when CONFIG_AUFS_DEBUG is enabled, and
its default value is 0. Setting 1 will enable the various verifications
and debug outputs.
Signed-off-by: J. R. Okajima <hooanon05g@gmail.com>

This is a very old debugging routine for rw_semaphore I was using
privately and less meaningful to other people. It was (probably) before
LOCKDEP feature was introduced, but now it is based upon LOCKDEP. This
is compiled when CONFIG_AUFS_DEBUG is enabled.
Signed-off-by: J. R. Okajima <hooanon05g@gmail.com>

See the documents in this commit.
Signed-off-by: J. R. Okajima <hooanon05g@gmail.com>

Very basic, simple and stupid wrappers for kmalloc family.
It tries kfree_ruc() as possible, with hoping a better performance.
Signed-off-by: J. R. Okajima <hooanon05g@gmail.com>

A header file for both of kernelspace and userspace.

For the new file fs/aufs/Kconfig, the maximum number of branches is
customizable, and it determines the type (size) of 'aufs_bindex_t.' The
type is always 'signed.' If we made it 'unsigned,' then more branches
would be available. But generally I think 127 (default) is enough and it
won't be a big issue.

For those who wants more than 127 branches, other values are
available. But we should care the size of the internal pointer arrays,
and it is good for the performance to keep it in a page at most.
AUFS_BRANCH_MAX_511 is mainly for 64bit systems which limits the
internal array size less than 4k (511 x 8bytes < 4k). Similarly for
32bit systems, AUFS_BRANCH_MAX_1023 (1023 x 4 bytes < 4k).

See also the documents in this commit.
Signed-off-by: J. R. Okajima <hooanon05g@gmail.com>