From: Chris Metcalf <cmetcalf@ezchip.com>
To: Gilad Ben Yossef <giladb@ezchip.com>, Steven Rostedt <rostedt@goodmis.org>,
        Ingo Molnar <mingo@kernel.org>, Peter Zijlstra <peterz@infradead.org>,
        Andrew Morton <akpm@linux-foundation.org>,
        Rik van Riel <riel@redhat.com>, Tejun Heo <tj@kernel.org>,
        Frederic Weisbecker <fweisbec@gmail.com>,
        Thomas Gleixner <tglx@linutronix.de>,
        "Paul E. McKenney" <paulmck@linux.vnet.ibm.com>,
        Christoph Lameter <cl@linux.com>,
        Viresh Kumar <viresh.kumar@linaro.org>, <linux-kernel@vger.kernel.org>
CC: Chris Metcalf <cmetcalf@ezchip.com>
Subject: [PATCH v4 5/5] nohz: cpu_isolated: allow tick to be fully disabled
Date: Mon, 13 Jul 2015 15:58:01 -0400
Message-ID: <1436817481-8732-6-git-send-email-cmetcalf@ezchip.com>
In-Reply-To: <1436817481-8732-1-git-send-email-cmetcalf@ezchip.com>
References: <1436817481-8732-1-git-send-email-cmetcalf@ezchip.com>
MIME-Version: 1.0
Content-Type: text/plain
AM3PR02MB0421: X-MS-Exchange-Organization-RulesExecuted
SpamDiagnosticOutput: 1:23
SpamDiagnosticMetadata: NSPM
X-MS-Exchange-CrossTenant-OriginalArrivalTime: 13 Jul 2015 19:58:30.4648
 (UTC)
X-MS-Exchange-CrossTenant-Id: 0fc16e0a-3cd3-4092-8b2f-0a42cff122c3
X-MS-Exchange-CrossTenant-OriginalAttributedTenantConnectingIp: TenantId=0fc16e0a-3cd3-4092-8b2f-0a42cff122c3;Ip=[12.216.194.146];Helo=[ld-1.internal.tilera.com]
X-MS-Exchange-CrossTenant-FromEntityHeader: HybridOnPrem
X-MS-Exchange-Transport-CrossTenantHeadersStamped: AM3PR02MB0421
Sender: linux-kernel-owner@vger.kernel.org
List-ID: <linux-kernel.vger.kernel.org>
X-Mailing-List: linux-kernel@vger.kernel.org
Content-Length: 2687
Lines: 56

While the current fallback to 1-second tick is still helpful for
maintaining completely correct kernel semantics, processes using
prctl(PR_SET_CPU_ISOLATED) semantics place a higher priority on running
completely tickless, so don't bound the time_delta for such processes.
In addition, due to the way such processes quiesce by waiting for
the timer tick to stop prior to returning to userspace, without this
commit it won't be possible to use the cpu_isolated mode at all.

Removing the 1-second cap was previously discussed (see link below)
and Thomas Gleixner observed that vruntime, load balancing data, load
accounting, and other things might be impacted.  Frederic Weisbecker
similarly observed that allowing the tick to be indefinitely deferred just
meant that no one would ever fix the underlying bugs.  However it's at
least true that the mode proposed in this patch can only be enabled on an
isolcpus core by a process requesting cpu_isolated mode, which may limit
how important it is to maintain scheduler data correctly, for example.

Paul McKenney observed that if provide a mode where the 1Hz fallback timer
is removed, this will provide an environment where new code that relies
on that tick will get punished, and we won't forgive such assumptions
silently, so it may also be worth it from that perspective.

Finally, it's worth observing that the tile architecture has been using
similar code for its Zero-Overhead Linux for many years (starting in
2008) and customers are very enthusiastic about the resulting bare-metal
performance on cores that are available to run full Linux semantics
on demand (crash, logging, shutdown, etc).  So this semantics is very
useful if we can convince ourselves that doing this is safe.

Link: https://lkml.kernel.org/r/alpine.DEB.2.11.1410311058500.32582@gentwo.org
Signed-off-by: Chris Metcalf <cmetcalf@ezchip.com>
---
 kernel/time/tick-sched.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/kernel/time/tick-sched.c b/kernel/time/tick-sched.c
index c5eca9c99fad..8187b4b4c91c 100644
--- a/kernel/time/tick-sched.c
+++ b/kernel/time/tick-sched.c
@@ -754,7 +754,7 @@ static ktime_t tick_nohz_stop_sched_tick(struct tick_sched *ts,
 
 #ifdef CONFIG_NO_HZ_FULL
 	/* Limit the tick delta to the maximum scheduler deferment */
-	if (!ts->inidle)
+	if (!ts->inidle && !tick_nohz_is_cpu_isolated())
 		delta = min(delta, scheduler_tick_max_deferment());
 #endif
 
-- 
2.1.2

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/