From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1752595AbbF3Vsv (ORCPT ); Tue, 30 Jun 2015 17:48:51 -0400 Received: from e36.co.us.ibm.com ([32.97.110.154]:40180 "EHLO e36.co.us.ibm.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753151AbbF3Vsh (ORCPT ); Tue, 30 Jun 2015 17:48:37 -0400 X-Helo: d03dlp02.boulder.ibm.com X-MailFrom: paulmck@linux.vnet.ibm.com X-RcptTo: linux-kernel@vger.kernel.org From: "Paul E. McKenney" To: linux-kernel@vger.kernel.org Cc: mingo@kernel.org, laijs@cn.fujitsu.com, dipankar@in.ibm.com, akpm@linux-foundation.org, mathieu.desnoyers@efficios.com, josh@joshtriplett.org, tglx@linutronix.de, peterz@infradead.org, rostedt@goodmis.org, dhowells@redhat.com, edumazet@google.com, dvhart@linux.intel.com, fweisbec@gmail.com, oleg@redhat.com, bobby.prani@gmail.com, "Paul E. McKenney" Subject: [PATCH RFC tip/core/rcu 3/5] rcutorture: Ensure that normal GPs advance without expedited GPs Date: Tue, 30 Jun 2015 14:48:28 -0700 Message-Id: <1435700910-9104-3-git-send-email-paulmck@linux.vnet.ibm.com> X-Mailer: git-send-email 1.8.1.5 In-Reply-To: <1435700910-9104-1-git-send-email-paulmck@linux.vnet.ibm.com> References: <20150630214805.GA7795@linux.vnet.ibm.com> <1435700910-9104-1-git-send-email-paulmck@linux.vnet.ibm.com> X-TM-AS-MML: disable X-Content-Scanned: Fidelis XPS MAILER x-cbid: 15063021-0021-0000-0000-00000E0338A1 Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org From: "Paul E. McKenney" Currently, rcutorture runs between 25% and 50% of grace periods as expedited grace periods. This means that on large configurations, there are expected to be a number of expedited grace periods executing concurrently with each normal grace period. Now that normal grace periods can be assisted by expedited grace periods, rcutorture would be unable to detect a bug that prevented normal grace periods from completing if there was no expedited grace period helping it. This commit therefore alternates 30-second phases that run expedited grace periods and 30-second phases that do not, so that this sort of bug will provoke an RCU CPU stall warning. Signed-off-by: Paul E. McKenney --- kernel/rcu/rcutorture.c | 22 +++++++++++++++++++++- 1 file changed, 21 insertions(+), 1 deletion(-) diff --git a/kernel/rcu/rcutorture.c b/kernel/rcu/rcutorture.c index 59aa76b4460e..c62c1c34d73c 100644 --- a/kernel/rcu/rcutorture.c +++ b/kernel/rcu/rcutorture.c @@ -54,6 +54,11 @@ MODULE_LICENSE("GPL"); MODULE_AUTHOR("Paul E. McKenney and Josh Triplett "); +#ifdef CONFIG_RCU_CPU_STALL_TIMEOUT +#define NOEXP_SECS_DEFAULT (CONFIG_RCU_CPU_STALL_TIMEOUT + 10) +#else /* #ifdef CONFIG_RCU_CPU_STALL_TIMEOUT */ +#define NOEXP_SECS_DEFAULT 30 +#endif /* #else #ifdef CONFIG_RCU_CPU_STALL_TIMEOUT */ torture_param(int, cbflood_inter_holdoff, HZ, "Holdoff between floods (jiffies)"); @@ -75,6 +80,8 @@ torture_param(int, irqreader, 1, "Allow RCU readers from irq handlers"); torture_param(int, n_barrier_cbs, 0, "# of callbacks/kthreads for barrier testing"); torture_param(int, nfakewriters, 4, "Number of RCU fake writer threads"); +torture_param(int, noexp_secs, NOEXP_SECS_DEFAULT, + "Time to suppress/enable expedited GPs (s), 0 to disable"); torture_param(int, nreaders, -1, "Number of RCU reader threads"); torture_param(int, object_debug, 0, "Enable debug-object double call_rcu() testing"); @@ -192,6 +199,8 @@ static u64 notrace rcu_trace_clock_local(void) } #endif /* #else #ifdef CONFIG_RCU_TRACE */ +static unsigned long rcutorture_noexp; + static unsigned long boost_starttime; /* jiffies of next boost test start. */ static DEFINE_MUTEX(boost_mutex); /* protect setting boost_starttime */ /* and boost task create/destroy. */ @@ -894,6 +903,8 @@ rcu_torture_writer(void *arg) bool gp_cond1 = gp_cond, gp_exp1 = gp_exp, gp_normal1 = gp_normal; bool gp_sync1 = gp_sync; int i; + int noexp_jiffies = noexp_secs * HZ; + unsigned long noexp_jiffies_next = jiffies - 1; struct rcu_torture *rp; struct rcu_torture *old_rp; static DEFINE_TORTURE_RANDOM(rand); @@ -942,6 +953,10 @@ rcu_torture_writer(void *arg) do { rcu_torture_writer_state = RTWS_FIXED_DELAY; schedule_timeout_uninterruptible(1); + if (time_after(jiffies, noexp_jiffies_next)) { + rcutorture_noexp = jiffies + noexp_jiffies; + noexp_jiffies_next = rcutorture_noexp + noexp_jiffies; + } rp = rcu_torture_alloc(); if (rp == NULL) continue; @@ -966,6 +981,9 @@ rcu_torture_writer(void *arg) cur_ops->deferred_free(old_rp); break; case RTWS_EXP_SYNC: + if (time_before(jiffies, rcutorture_noexp) && + cur_ops->sync && gp_sync1) + goto noexp; rcu_torture_writer_state = RTWS_EXP_SYNC; cur_ops->exp_sync(); rcu_torture_pipe_update(old_rp); @@ -982,6 +1000,7 @@ rcu_torture_writer(void *arg) rcu_torture_pipe_update(old_rp); break; case RTWS_SYNC: +noexp: rcu_torture_writer_state = RTWS_SYNC; cur_ops->sync(); rcu_torture_pipe_update(old_rp); @@ -1036,7 +1055,8 @@ rcu_torture_fakewriter(void *arg) torture_random(&rand) % (nfakewriters * 8) == 0) { cur_ops->cb_barrier(); } else if (gp_normal == gp_exp) { - if (torture_random(&rand) & 0x80) + if (time_before(jiffies, rcutorture_noexp) || + torture_random(&rand) & 0x80) cur_ops->sync(); else cur_ops->exp_sync(); -- 1.8.1.5