linux-kernel - [tip:core/rcu] rcu: Adjust force_quiescent

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <tip-f96e9232e04856c781d4f71923a46dd3f7b429fa@git.kernel.org>
Date:	Wed, 13 Jan 2010 10:24:45 GMT
From:	"tip-bot for Paul E. McKenney" <paulmck@...ux.vnet.ibm.com>
To:	linux-tip-commits@...r.kernel.org
Cc:	linux-kernel@...r.kernel.org, paulmck@...ux.vnet.ibm.com,
	hpa@...or.com, mingo@...hat.com, tglx@...utronix.de, mingo@...e.hu
Subject: [tip:core/rcu] rcu: Adjust force_quiescent_state() locking, step 1

Commit-ID:  f96e9232e04856c781d4f71923a46dd3f7b429fa
Gitweb:     http://git.kernel.org/tip/f96e9232e04856c781d4f71923a46dd3f7b429fa
Author:     Paul E. McKenney <paulmck@...ux.vnet.ibm.com>
AuthorDate: Mon, 4 Jan 2010 15:09:00 -0800
Committer:  Ingo Molnar <mingo@...e.hu>
CommitDate: Wed, 13 Jan 2010 09:06:01 +0100

rcu: Adjust force_quiescent_state() locking, step 1

This causes rnp->lock to be held on entry to
force_quiescent_state()'s switch statement.  This is a first
step towards prohibiting starting grace periods while
force_quiescent_state() is executing, which will reduce the
number and complexity of races that force_quiescent_state() is
involved in.

Signed-off-by: Paul E. McKenney <paulmck@...ux.vnet.ibm.com>
Cc: laijs@...fujitsu.com
Cc: dipankar@...ibm.com
Cc: mathieu.desnoyers@...ymtl.ca
Cc: josh@...htriplett.org
Cc: dvhltc@...ibm.com
Cc: niv@...ibm.com
Cc: peterz@...radead.org
Cc: rostedt@...dmis.org
Cc: Valdis.Kletnieks@...edu
Cc: dhowells@...hat.com
LKML-Reference: <12626465501455-git-send-email->
Signed-off-by: Ingo Molnar <mingo@...e.hu>
---
 kernel/rcutree.c |   27 ++++++++++++++++++---------
 1 files changed, 18 insertions(+), 9 deletions(-)

diff --git a/kernel/rcutree.c b/kernel/rcutree.c
index 53ae959..eae331d 100644
--- a/kernel/rcutree.c
+++ b/kernel/rcutree.c
@@ -1204,7 +1204,7 @@ static void force_quiescent_state(struct rcu_state *rsp, int relaxed)
 	}
 	if (relaxed &&
 	    (long)(rsp->jiffies_force_qs - jiffies) >= 0)
-		goto unlock_ret; /* no emergency and done recently. */
+		goto unlock_fqs_ret; /* no emergency and done recently. */
 	rsp->n_force_qs++;
 	spin_lock(&rnp->lock);
 	lastcomp = rsp->gpnum - 1;
@@ -1213,31 +1213,32 @@ static void force_quiescent_state(struct rcu_state *rsp, int relaxed)
 	if(!rcu_gp_in_progress(rsp)) {
 		rsp->n_force_qs_ngp++;
 		spin_unlock(&rnp->lock);
-		goto unlock_ret;  /* no GP in progress, time updated. */
+		goto unlock_fqs_ret;  /* no GP in progress, time updated. */
 	}
-	spin_unlock(&rnp->lock);
 	switch (signaled) {
 	case RCU_GP_IDLE:
 	case RCU_GP_INIT:
 
+		spin_unlock(&rnp->lock);
 		break; /* grace period idle or initializing, ignore. */
 
 	case RCU_SAVE_DYNTICK:
 
+		spin_unlock(&rnp->lock);
 		if (RCU_SIGNAL_INIT != RCU_SAVE_DYNTICK)
 			break; /* So gcc recognizes the dead code. */
 
 		/* Record dyntick-idle state. */
 		if (rcu_process_dyntick(rsp, lastcomp,
 					dyntick_save_progress_counter))
-			goto unlock_ret;
+			goto unlock_fqs_ret;
+		spin_lock(&rnp->lock);
 		/* fall into next case. */
 
 	case RCU_SAVE_COMPLETED:
 
 		/* Update state, record completion counter. */
 		forcenow = 0;
-		spin_lock(&rnp->lock);
 		if (lastcomp + 1 == rsp->gpnum &&
 		    lastcomp == rsp->completed &&
 		    rsp->signaled == signaled) {
@@ -1245,23 +1246,31 @@ static void force_quiescent_state(struct rcu_state *rsp, int relaxed)
 			rsp->completed_fqs = lastcomp;
 			forcenow = signaled == RCU_SAVE_COMPLETED;
 		}
-		spin_unlock(&rnp->lock);
-		if (!forcenow)
+		if (!forcenow) {
+			spin_unlock(&rnp->lock);
 			break;
+		}
 		/* fall into next case. */
 
 	case RCU_FORCE_QS:
 
 		/* Check dyntick-idle state, send IPI to laggarts. */
+		spin_unlock(&rnp->lock);
 		if (rcu_process_dyntick(rsp, rsp->completed_fqs,
 					rcu_implicit_dynticks_qs))
-			goto unlock_ret;
+			goto unlock_fqs_ret;
 
 		/* Leave state in case more forcing is required. */
 
 		break;
+
+	default:
+
+		spin_unlock(&rnp->lock);
+		WARN_ON_ONCE(1);
+		break;
 	}
-unlock_ret:
+unlock_fqs_ret:
 	spin_unlock_irqrestore(&rsp->fqslock, flags);
 }
 
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/