dect
/
linux-2.6
Archived
13
0
Fork 0
Code Releases Activity

[AFS]: Eliminate cmpxchg() usage in vlocation code. 

cmpxchg() is not available on every processor so can't
be used in generic code.

Replace with spinlock protection on the ->state changes,
wakeups, and wait loops.

Add what appears to be a missing wakeup on transition
to AFS_VL_VALID state in afs_vlocation_updater().

Signed-off-by: David S. Miller <davem@davemloft.net>
This commit is contained in:
David S. Miller 2007-04-26 20:39:14 -07:00
parent 68c708fd5e
commit 39bf094930
2 changed files with 16 additions and 6 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
fs/afs/internal.h
View File

@ -219,7 +219,7 @@ struct afs_vlocation {
struct afs_volume *vols[3]; /* volume access record pointer (index by type) */
wait_queue_head_t waitq; /* status change waitqueue */
time_t update_at; /* time at which record should be updated */
rwlock_t lock; /* access lock */
spinlock_t lock; /* access lock */
afs_vlocation_state_t state; /* volume location state */
unsigned short upd_rej_cnt; /* ENOMEDIUM count during update */
unsigned short upd_busy_cnt; /* EBUSY count during update */

20
fs/afs/vlocation.c
View File

@ -178,7 +178,7 @@ static struct afs_vlocation *afs_vlocation_alloc(struct afs_cell *cell,
INIT_LIST_HEAD(&vl->grave);
INIT_LIST_HEAD(&vl->update);
init_waitqueue_head(&vl->waitq);
rwlock_init(&vl->lock);
spin_lock_init(&vl->lock);
memcpy(vl->vldb.name, name, namesz);
}
@ -414,8 +414,10 @@ fill_in_record:
ret = afs_vlocation_fill_in_record(vl, key);
if (ret < 0)
goto error_abandon;
spin_lock(&vl->lock);
vl->state = AFS_VL_VALID;
wake_up(&vl->waitq);
spin_unlock(&vl->lock);
/* schedule for regular updates */
afs_vlocation_queue_for_updates(vl);
@ -434,22 +436,23 @@ found_in_memory:
up_write(&cell->vl_sem);
/* see if it was an abandoned record that we might try filling in */
spin_lock(&vl->lock);
while (vl->state != AFS_VL_VALID) {
afs_vlocation_state_t state = vl->state;
_debug("invalid [state %d]", state);
if ((state == AFS_VL_NEW || state == AFS_VL_NO_VOLUME)) {
if (cmpxchg(&vl->state, state, AFS_VL_CREATING) ==
state)
goto fill_in_record;
continue;
vl->state = AFS_VL_CREATING;
spin_unlock(&vl->lock);
goto fill_in_record;
}
/* must now wait for creation or update by someone else to
* complete */
_debug("wait");
spin_unlock(&vl->lock);
ret = wait_event_interruptible(
vl->waitq,
vl->state == AFS_VL_NEW ||
@ -457,15 +460,19 @@ found_in_memory:
vl->state == AFS_VL_NO_VOLUME);
if (ret < 0)
goto error;
spin_lock(&vl->lock);
}
spin_unlock(&vl->lock);
success:
_leave(" = %p",vl);
return vl;
error_abandon:
spin_lock(&vl->lock);
vl->state = AFS_VL_NEW;
wake_up(&vl->waitq);
spin_unlock(&vl->lock);
error:
ASSERT(vl != NULL);
afs_put_vlocation(vl);
@ -663,10 +670,12 @@ static void afs_vlocation_updater(struct work_struct *work)
vl->upd_busy_cnt = 0;
ret = afs_vlocation_update_record(vl, NULL, &vldb);
spin_lock(&vl->lock);
switch (ret) {
case 0:
afs_vlocation_apply_update(vl, &vldb);
vl->state = AFS_VL_VALID;
wake_up(&vl->waitq);
break;
case -ENOMEDIUM:
vl->state = AFS_VL_VOLUME_DELETED;
@ -675,6 +684,7 @@ static void afs_vlocation_updater(struct work_struct *work)
vl->state = AFS_VL_UNCERTAIN;
break;
}
spin_unlock(&vl->lock);
/* and then reschedule */
_debug("reschedule");