rcu: Avoid spurious RCU CPU stall warnings
authorPaul E. McKenney <paul.mckenney@linaro.org>
Mon, 13 Aug 2012 18:17:06 +0000 (11:17 -0700)
committerPaul E. McKenney <paulmck@linux.vnet.ibm.com>
Sun, 23 Sep 2012 14:42:51 +0000 (07:42 -0700)
If a given CPU avoids the idle loop but also avoids starting a new
RCU grace period for a full minute, RCU can issue spurious RCU CPU
stall warnings.  This commit fixes this issue by adding a check for
ongoing grace period to avoid these spurious stall warnings.

Reported-by: Becky Bruce <bgillbruce@gmail.com>
Signed-off-by: Paul E. McKenney <paul.mckenney@linaro.org>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Reviewed-by: Josh Triplett <josh@joshtriplett.org>
kernel/rcutree.c

index 2cf8eb3e2d435828afb1e149606aa8277efee947..98f275296c6d80d7cb4cad25acf6fb76f8408e59 100644 (file)
@@ -819,7 +819,8 @@ static void check_cpu_stall(struct rcu_state *rsp, struct rcu_data *rdp)
        j = ACCESS_ONCE(jiffies);
        js = ACCESS_ONCE(rsp->jiffies_stall);
        rnp = rdp->mynode;
-       if ((ACCESS_ONCE(rnp->qsmask) & rdp->grpmask) && ULONG_CMP_GE(j, js)) {
+       if (rcu_gp_in_progress(rsp) &&
+           (ACCESS_ONCE(rnp->qsmask) & rdp->grpmask) && ULONG_CMP_GE(j, js)) {
 
                /* We haven't checked in, so go dump stack. */
                print_cpu_stall(rsp);