linux/fs/afs/callback.c
David Howells c435ee3455 afs: Overhaul the callback handling
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 <dhowells@redhat.com>
2017-11-13 15:38:18 +00:00

245 lines
5.6 KiB
C

/*
* Copyright (c) 2002, 2007 Red Hat, Inc. All rights reserved.
*
* This software may be freely redistributed under the terms of the
* GNU General Public License.
*
* You should have received a copy of the GNU General Public License
* along with this program; if not, write to the Free Software
* Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
*
* Authors: David Woodhouse <dwmw2@infradead.org>
* David Howells <dhowells@redhat.com>
*
*/
#include <linux/kernel.h>
#include <linux/module.h>
#include <linux/init.h>
#include <linux/circ_buf.h>
#include <linux/sched.h>
#include "internal.h"
/*
* Set up an interest-in-callbacks record for a volume on a server and
* register it with the server.
* - Called with volume->server_sem held.
*/
int afs_register_server_cb_interest(struct afs_vnode *vnode,
struct afs_cb_interest **ppcbi,
struct afs_server *server)
{
struct afs_cb_interest *cbi = *ppcbi, *vcbi, *new, *x;
again:
vcbi = vnode->cb_interest;
if (vcbi) {
if (vcbi == cbi)
return 0;
if (cbi && vcbi->server == cbi->server) {
write_seqlock(&vnode->cb_lock);
vnode->cb_interest = afs_get_cb_interest(cbi);
write_sequnlock(&vnode->cb_lock);
afs_put_cb_interest(afs_v2net(vnode), cbi);
return 0;
}
if (!cbi && vcbi->server == server) {
afs_get_cb_interest(vcbi);
x = cmpxchg(ppcbi, cbi, vcbi);
if (x != cbi) {
cbi = x;
afs_put_cb_interest(afs_v2net(vnode), vcbi);
goto again;
}
return 0;
}
}
if (!cbi) {
new = kzalloc(sizeof(struct afs_cb_interest), GFP_KERNEL);
if (!new)
return -ENOMEM;
refcount_set(&new->usage, 1);
new->sb = vnode->vfs_inode.i_sb;
new->vid = vnode->volume->vid;
new->server = afs_get_server(server);
INIT_LIST_HEAD(&new->cb_link);
write_lock(&server->cb_break_lock);
list_add_tail(&new->cb_link, &server->cb_interests);
write_unlock(&server->cb_break_lock);
x = cmpxchg(ppcbi, cbi, new);
if (x == cbi) {
cbi = new;
} else {
cbi = x;
afs_put_cb_interest(afs_v2net(vnode), new);
}
}
ASSERT(cbi);
/* Change the server the vnode is using. This entails scrubbing any
* interest the vnode had in the previous server it was using.
*/
write_seqlock(&vnode->cb_lock);
vnode->cb_interest = afs_get_cb_interest(cbi);
vnode->cb_s_break = cbi->server->cb_s_break;
clear_bit(AFS_VNODE_CB_PROMISED, &vnode->flags);
write_sequnlock(&vnode->cb_lock);
return 0;
}
/*
* Set a vnode's interest on a server.
*/
void afs_set_cb_interest(struct afs_vnode *vnode, struct afs_cb_interest *cbi)
{
struct afs_cb_interest *old_cbi = NULL;
if (vnode->cb_interest == cbi)
return;
write_seqlock(&vnode->cb_lock);
if (vnode->cb_interest != cbi) {
afs_get_cb_interest(cbi);
old_cbi = vnode->cb_interest;
vnode->cb_interest = cbi;
}
write_sequnlock(&vnode->cb_lock);
afs_put_cb_interest(afs_v2net(vnode), cbi);
}
/*
* Remove an interest on a server.
*/
void afs_put_cb_interest(struct afs_net *net, struct afs_cb_interest *cbi)
{
if (cbi && refcount_dec_and_test(&cbi->usage)) {
if (!list_empty(&cbi->cb_link)) {
write_lock(&cbi->server->cb_break_lock);
list_del_init(&cbi->cb_link);
write_unlock(&cbi->server->cb_break_lock);
afs_put_server(net, cbi->server);
}
kfree(cbi);
}
}
/*
* allow the fileserver to request callback state (re-)initialisation
*/
void afs_init_callback_state(struct afs_server *server)
{
if (!test_and_clear_bit(AFS_SERVER_NEW, &server->flags))
server->cb_s_break++;
}
/*
* actually break a callback
*/
void afs_break_callback(struct afs_vnode *vnode)
{
_enter("");
write_seqlock(&vnode->cb_lock);
if (test_and_clear_bit(AFS_VNODE_CB_PROMISED, &vnode->flags)) {
vnode->cb_break++;
afs_clear_permits(vnode);
spin_lock(&vnode->lock);
_debug("break callback");
if (list_empty(&vnode->granted_locks) &&
!list_empty(&vnode->pending_locks))
afs_lock_may_be_available(vnode);
spin_unlock(&vnode->lock);
}
write_sequnlock(&vnode->cb_lock);
}
/*
* allow the fileserver to explicitly break one callback
* - happens when
* - the backing file is changed
* - a lock is released
*/
static void afs_break_one_callback(struct afs_server *server,
struct afs_fid *fid)
{
struct afs_cb_interest *cbi;
struct afs_iget_data data;
struct afs_vnode *vnode;
struct inode *inode;
read_lock(&server->cb_break_lock);
/* Step through all interested superblocks. There may be more than one
* because of cell aliasing.
*/
list_for_each_entry(cbi, &server->cb_interests, cb_link) {
if (cbi->vid != fid->vid)
continue;
data.volume = NULL;
data.fid = *fid;
inode = ilookup5_nowait(cbi->sb, fid->vnode, afs_iget5_test, &data);
if (inode) {
vnode = AFS_FS_I(inode);
afs_break_callback(vnode);
iput(inode);
}
}
read_unlock(&server->cb_break_lock);
}
/*
* allow the fileserver to break callback promises
*/
void afs_break_callbacks(struct afs_server *server, size_t count,
struct afs_callback callbacks[])
{
_enter("%p,%zu,", server, count);
ASSERT(server != NULL);
ASSERTCMP(count, <=, AFSCBMAX);
for (; count > 0; callbacks++, count--) {
_debug("- Fid { vl=%08x n=%u u=%u } CB { v=%u x=%u t=%u }",
callbacks->fid.vid,
callbacks->fid.vnode,
callbacks->fid.unique,
callbacks->version,
callbacks->expiry,
callbacks->type
);
afs_break_one_callback(server, &callbacks->fid);
}
_leave("");
return;
}
/*
* Clear the callback interests in a server list.
*/
void afs_clear_callback_interests(struct afs_net *net, struct afs_volume *volume)
{
int i;
for (i = 0; i < ARRAY_SIZE(volume->cb_interests); i++) {
afs_put_cb_interest(net, volume->cb_interests[i]);
volume->cb_interests[i] = NULL;
}
}