kernel.git/fs/afs/inode.c, branch linux-4.16.y

afs: Support the AFS dynamic root

2018-02-06T14:43:37Z

Support the AFS dynamic root which is a pseudo-volume that doesn't connect to any server resource, but rather is just a root directory that dynamically creates mountpoint directories where the name of such a directory is the name of the cell. Such a mount can be created thus: mount -t afs none /afs -o dyn Dynamic root superblocks aren't shared except by bind mounts and propagation. Cell root volumes can then be mounted by referring to them by name, e.g.: ls /afs/grand.central.org/ ls /afs/.grand.central.org/ The kernel will upcall to consult the DNS if the address wasn't supplied directly. Signed-off-by: David Howells

Merge tag 'iversion-v4.16-1' of git://git.kernel.org/pub/scm/linux/kernel/git/jlayton/linux

2018-01-29T21:33:53Z

Pull inode->i_version rework from Jeff Layton: "This pile of patches is a rework of the inode->i_version field. We have traditionally incremented that field on every inode data or metadata change. Typically this increment needs to be logged on disk even when nothing else has changed, which is rather expensive. It turns out though that none of the consumers of that field actually require this behavior. The only real requirement for all of them is that it be different iff the inode has changed since the last time the field was checked. Given that, we can optimize away most of the i_version increments and avoid dirtying inode metadata when the only change is to the i_version and no one is querying it. Queries of the i_version field are rather rare, so we can help write performance under many common workloads. This patch series converts existing accesses of the i_version field to a new API, and then converts all of the in-kernel filesystems to use it. The last patch in the series then converts the backend implementation to a scheme that optimizes away a large portion of the metadata updates when no one is looking at it. In my own testing this series significantly helps performance with small I/O sizes. I also got this email for Christmas this year from the kernel test robot (a 244% r/w bandwidth improvement with XFS over DAX, with 4k writes): https://lkml.org/lkml/2017/12/25/8 A few of the earlier patches in this pile are also flowing to you via other trees (mm, integrity, and nfsd trees in particular)". * tag 'iversion-v4.16-1' of git://git.kernel.org/pub/scm/linux/kernel/git/jlayton/linux: (22 commits) fs: handle inode->i_version more efficiently btrfs: only dirty the inode in btrfs_update_time if something was changed xfs: avoid setting XFS_ILOG_CORE if i_version doesn't need incrementing fs: only set S_VERSION when updating times if necessary IMA: switch IMA over to new i_version API xfs: convert to new i_version API ufs: use new i_version API ocfs2: convert to new i_version API nfsd: convert to new i_version API nfs: convert to new i_version API ext4: convert to new i_version API ext2: convert to new i_version API exofs: switch to new i_version API btrfs: convert to new i_version API afs: convert to new i_version API affs: convert to new i_version API fat: convert to new i_version API fs: don't take the i_lock in inode_inc_iversion fs: new API for handling inode->i_version ntfs: remove i_version handling ...

afs: convert to new i_version API

2018-01-29T11:42:20Z

For AFS, it's generally treated as an opaque value, so we use the *_raw variants of the API here. Note that AFS has quite a different definition for this counter. AFS only increments it on changes to the data to the data in regular files and contents of the directories. Inode metadata changes do not result in a version increment. We'll need to reconcile that somehow if we ever want to present this to userspace via statx. Signed-off-by: Jeff Layton

afs: Fix unlink

2018-01-02T10:02:19Z

Repeating creation and deletion of a file on an afs mount will run the box out of memory, e.g.: dd if=/dev/zero of=/afs/scratch/m0 bs=$((1024*1024)) count=512 rm /afs/scratch/m0 The problem seems to be that it's not properly decrementing the nlink count so that the inode can be scrapped. Note that this doesn't fix local creation followed by remote deletion. That's harder to handle and will require a separate patch as we're not told that the file has been deleted - only that the directory has changed. Reported-by: Marc Dionne Signed-off-by: David Howells

afs: Get rid of the afs_writeback record

2017-11-13T15:38:20Z

Get rid of the afs_writeback record that kAFS is using to match keys with writes made by that key. Instead, keep a list of keys that have a file open for writing and/or sync'ing and iterate through those. Signed-off-by: David Howells

afs: Introduce a file-private data record

2017-11-13T15:38:20Z

Introduce a file-private data record for kAFS and put the key into it rather than storing the key in file->private_data. Signed-off-by: David Howells

afs: Overhaul volume and server record caching and fileserver rotation

2017-11-13T15:38:19Z

The current code assumes that volumes and servers are per-cell and are never shared, but this is not enforced, and, indeed, public cells do exist that are aliases of each other. Further, an organisation can, say, set up a public cell and a private cell with overlapping, but not identical, sets of servers. The difference is purely in the database attached to the VL servers. The current code will malfunction if it sees a server in two cells as it assumes global address -> server record mappings and that each server is in just one cell. Further, each server may have multiple addresses - and may have addresses of different families (IPv4 and IPv6, say). To this end, the following structural changes are made: (1) Server record management is overhauled: (a) Server records are made independent of cell. The namespace keeps track of them, volume records have lists of them and each vnode has a server on which its callback interest currently resides. (b) The cell record no longer keeps a list of servers known to be in that cell. (c) The server records are now kept in a flat list because there's no single address to sort on. (d) Server records are now keyed by their UUID within the namespace. (e) The addresses for a server are obtained with the VL.GetAddrsU rather than with VL.GetEntryByName, using the server's UUID as a parameter. (f) Cached server records are garbage collected after a period of non-use and are counted out of existence before purging is allowed to complete. This protects the work functions against rmmod. (g) The servers list is now in /proc/fs/afs/servers. (2) Volume record management is overhauled: (a) An RCU-replaceable server list is introduced. This tracks both servers and their coresponding callback interests. (b) The superblock is now keyed on cell record and numeric volume ID. (c) The volume record is now tied to the superblock which mounts it, and is activated when mounted and deactivated when unmounted. This makes it easier to handle the cache cookie without causing a double-use in fscache. (d) The volume record is loaded from the VLDB using VL.GetEntryByNameU to get the server UUID list. (e) The volume name is updated if it is seen to have changed when the volume is updated (the update is keyed on the volume ID). (3) The vlocation record is got rid of and VLDB records are no longer cached. Sufficient information is stored in the volume record, though an update to a volume record is now no longer shared between related volumes (volumes come in bundles of three: R/W, R/O and backup). and the following procedural changes are made: (1) The fileserver cursor introduced previously is now fleshed out and used to iterate over fileservers and their addresses. (2) Volume status is checked during iteration, and the server list is replaced if a change is detected. (3) Server status is checked during iteration, and the address list is replaced if a change is detected. (4) The abort code is saved into the address list cursor and -ECONNABORTED returned in afs_make_call() if a remote abort happened rather than translating the abort into an error message. This allows actions to be taken depending on the abort code more easily. (a) If a VMOVED abort is seen then this is handled by rechecking the volume and restarting the iteration. (b) If a VBUSY, VRESTARTING or VSALVAGING abort is seen then this is handled by sleeping for a short period and retrying and/or trying other servers that might serve that volume. A message is also displayed once until the condition has cleared. (c) If a VOFFLINE abort is seen, then this is handled as VBUSY for the moment. (d) If a VNOVOL abort is seen, the volume is rechecked in the VLDB to see if it has been deleted; if not, the fileserver is probably indicating that the volume couldn't be attached and needs salvaging. (e) If statfs() sees one of these aborts, it does not sleep, but rather returns an error, so as not to block the umount program. (5) The fileserver iteration functions in vnode.c are now merged into their callers and more heavily macroised around the cursor. vnode.c is removed. (6) Operations on a particular vnode are serialised on that vnode because the server will lock that vnode whilst it operates on it, so a second op sent will just have to wait. (7) Fileservers are probed with FS.GetCapabilities before being used. This is where service upgrade will be done. (8) A callback interest on a fileserver is set up before an FS operation is performed and passed through to afs_make_call() so that it can be set on the vnode if the operation returns a callback. The callback interest is passed through to afs_iget() also so that it can be set there too. In general, record updating is done on an as-needed basis when we try to access servers, volumes or vnodes rather than offloading it to work items and special threads. Notes: (1) Pre AFS-3.4 servers are no longer supported, though this can be added back if necessary (AFS-3.4 was released in 1998). (2) VBUSY is retried forever for the moment at intervals of 1s. (3) /proc/fs/afs//servers no longer exists. Signed-off-by: David Howells

afs: Overhaul permit caching

2017-11-13T15:38:18Z

Overhaul permit caching in AFS by making it per-vnode and sharing permit lists where possible. When most of the fileserver operations are called, they return a status structure indicating the (revised) details of the vnode or vnodes involved in the operation. This includes the access mark derived from the ACL (named CallerAccess in the protocol definition file). This is cacheable and if the ACL changes, the server will tell us that it is breaking the callback promise, at which point we can discard the currently cached permits. With this patch, the afs_permits structure has, at the end, an array of { key, CallerAccess } elements, sorted by key pointer. This is then cached in a hash table so that it can be shared between vnodes with the same access permits. Permit lists can only be shared if they contain the exact same set of key->CallerAccess mappings. Note that that table is global rather than being per-net_ns. If the keys in a permit list cross net_ns boundaries, there is no problem sharing the cached permits, since the permits are just integer masks. Since permit lists pin keys, the permit cache also makes it easier for a future patch to find all occurrences of a key and remove them by means of setting the afs_permits::invalidated flag and then clearing the appropriate key pointer. In such an event, memory barriers will need adding. Lastly, the permit caching is skipped if the server has sent either a vnode-specific or an entire-server callback since the start of the operation. Signed-off-by: David Howells

afs: Overhaul the callback handling

2017-11-13T15:38:18Z

Overhaul the AFS callback handling by the following means: (1) Don't give up callback promises on vnodes that we are no longer using, rather let them just expire on the server or let the server break them. This is actually more efficient for the server as the callback lookup is expensive if there are lots of extant callbacks. (2) Only give up the callback promises we have from a server when the server record is destroyed. Then we can just give up *all* the callback promises on it in one go. (3) Servers can end up being shared between cells if cells are aliased, so don't add all the vnodes being backed by a particular server into a big FID-indexed tree on that server as there may be duplicates. Instead have each volume instance (~= superblock) register an interest in a server as it starts to make use of it and use this to allow the processor for callbacks from the server to find the superblock and thence the inode corresponding to the FID being broken by means of ilookup_nowait(). (4) Rather than iterating over the entire callback list when a mass-break comes in from the server, maintain a counter of mass-breaks in afs_server (cb_seq) and make afs_validate() check it against the copy in afs_vnode. It would be nice not to have to take a read_lock whilst doing this, but that's tricky without using RCU. (5) Save a ref on the fileserver we're using for a call in the afs_call struct so that we can access its cb_s_break during call decoding. (6) Write-lock around callback and status storage in a vnode and read-lock around getattr so that we don't see the status mid-update. This has the following consequences: (1) Data invalidation isn't seen until someone calls afs_validate() on a vnode. Unfortunately, we need to use a key to query the server, but getting one from a background thread is tricky without caching loads of keys all over the place. (2) Mass invalidation isn't seen until someone calls afs_validate(). (3) Callback breaking is going to hit the inode_hash_lock quite a bit. Could this be replaced with rcu_read_lock() since inodes are destroyed under RCU conditions. Signed-off-by: David Howells

afs: Push the net ns pointer to more places

2017-11-13T15:38:17Z

Push the network namespace pointer to more places in AFS, including the afs_server structure (which doesn't hold a ref on the netns). In particular, afs_put_cell() now takes requires a net ns parameter so that it can safely alter the netns after decrementing the cell usage count - the cell will be deallocated by a background thread after being cached for a period, which means that it's not safe to access it after reducing its usage count. Signed-off-by: David Howells