NFSv4.1: Ensure smooth handover of slots from one task to the next waiting
Currently, we see a lot of bouncing for the value of highest_used_slotid
due to the fact that slots are getting freed, instead of getting instantly
transmitted to the next waiting task.
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
diff --git a/fs/nfs/nfs4state.c b/fs/nfs/nfs4state.c
index 7d73df5..78e90a8 100644
--- a/fs/nfs/nfs4state.c
+++ b/fs/nfs/nfs4state.c
@@ -255,17 +255,13 @@
{
struct nfs4_session *ses = clp->cl_session;
struct nfs4_slot_table *tbl;
- unsigned int i;
if (ses == NULL)
return;
tbl = &ses->fc_slot_table;
if (test_and_clear_bit(NFS4_SESSION_DRAINING, &ses->session_state)) {
spin_lock(&tbl->slot_tbl_lock);
- for (i = 0; i <= tbl->max_slotid; i++) {
- if (rpc_wake_up_next(&tbl->slot_tbl_waitq) == NULL)
- break;
- }
+ nfs41_wake_slot_table(tbl);
spin_unlock(&tbl->slot_tbl_lock);
}
}