summaryrefslogtreecommitdiff
path: root/fs/afs/volume.c
diff options
context:
space:
mode:
authorDavid Howells <dhowells@redhat.com>2017-11-02 15:27:49 +0000
committerDavid Howells <dhowells@redhat.com>2017-11-13 15:38:18 +0000
commitc435ee34551e1f5a02a253ca8e235287efd2727c (patch)
treefaafdec41e83f4cbe15638d2145b1b339d13757d /fs/afs/volume.c
parentd0676a16781d0972969dff8b3f3f819599cc4b07 (diff)
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>
Diffstat (limited to 'fs/afs/volume.c')
-rw-r--r--fs/afs/volume.c43
1 files changed, 27 insertions, 16 deletions
diff --git a/fs/afs/volume.c b/fs/afs/volume.c
index 52f0dc40732b..4f6fd10094c6 100644
--- a/fs/afs/volume.c
+++ b/fs/afs/volume.c
@@ -153,8 +153,10 @@ error:
error_discard:
up_write(&params->cell->vl_sem);
- for (loop = volume->nservers - 1; loop >= 0; loop--)
+ for (loop = volume->nservers - 1; loop >= 0; loop--) {
+ afs_put_cb_interest(params->net, volume->cb_interests[loop]);
afs_put_server(params->net, volume->servers[loop]);
+ }
kfree(volume);
goto error;
@@ -197,8 +199,10 @@ void afs_put_volume(struct afs_cell *cell, struct afs_volume *volume)
#endif
afs_put_vlocation(cell->net, vlocation);
- for (loop = volume->nservers - 1; loop >= 0; loop--)
+ for (loop = volume->nservers - 1; loop >= 0; loop--) {
+ afs_put_cb_interest(cell->net, volume->cb_interests[loop]);
afs_put_server(cell->net, volume->servers[loop]);
+ }
kfree(volume);
@@ -218,10 +222,10 @@ struct afs_server *afs_volume_pick_fileserver(struct afs_vnode *vnode)
_enter("%s", volume->vlocation->vldb.name);
/* stick with the server we're already using if we can */
- if (vnode->server && vnode->server->fs_state == 0) {
- afs_get_server(vnode->server);
- _leave(" = %p [current]", vnode->server);
- return vnode->server;
+ if (vnode->cb_interest && vnode->cb_interest->server->fs_state == 0) {
+ afs_get_server(vnode->cb_interest->server);
+ _leave(" = %p [current]", vnode->cb_interest->server);
+ return vnode->cb_interest->server;
}
down_read(&volume->server_sem);
@@ -244,13 +248,8 @@ struct afs_server *afs_volume_pick_fileserver(struct afs_vnode *vnode)
_debug("consider %d [%d]", loop, state);
switch (state) {
- /* found an apparently healthy server */
case 0:
- afs_get_server(server);
- up_read(&volume->server_sem);
- _leave(" = %p (picked %pIS)",
- server, &server->addr.transport);
- return server;
+ goto picked_server;
case -ENETUNREACH:
if (ret == 0)
@@ -284,9 +283,25 @@ struct afs_server *afs_volume_pick_fileserver(struct afs_vnode *vnode)
/* no available servers
* - TODO: handle the no active servers case better
*/
+error:
up_read(&volume->server_sem);
_leave(" = %d", ret);
return ERR_PTR(ret);
+
+picked_server:
+ /* Found an apparently healthy server. We need to register an interest
+ * in receiving callbacks before we talk to it.
+ */
+ ret = afs_register_server_cb_interest(vnode,
+ &volume->cb_interests[loop], server);
+ if (ret < 0)
+ goto error;
+
+ afs_get_server(server);
+ up_read(&volume->server_sem);
+ _leave(" = %p (picked %pIS)",
+ server, &server->addr.transport);
+ return server;
}
/*
@@ -309,14 +324,12 @@ int afs_volume_release_fileserver(struct afs_vnode *vnode,
switch (result) {
/* success */
case 0:
- server->fs_act_jif = jiffies;
server->fs_state = 0;
_leave("");
return 1;
/* the fileserver denied all knowledge of the volume */
case -ENOMEDIUM:
- server->fs_act_jif = jiffies;
down_write(&volume->server_sem);
/* firstly, find where the server is in the active list (if it
@@ -365,7 +378,6 @@ int afs_volume_release_fileserver(struct afs_vnode *vnode,
*/
spin_lock(&server->fs_lock);
if (!server->fs_state) {
- server->fs_dead_jif = jiffies + HZ * 10;
server->fs_state = result;
printk("kAFS: SERVER DEAD state=%d\n", result);
}
@@ -374,7 +386,6 @@ int afs_volume_release_fileserver(struct afs_vnode *vnode,
/* miscellaneous error */
default:
- server->fs_act_jif = jiffies;
case -ENOMEM:
case -ENONET:
/* tell the caller to accept the result */