linux-kernel - Re: [PATCH 06/12 UPDATED] irq: implement irq_schedule

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: <4C1F687A.5030907@kernel.org>
Date:	Mon, 21 Jun 2010 15:26:18 +0200
From:	Tejun Heo <tj@...nel.org>
To:	Jonathan Corbet <corbet@....net>
CC:	mingo@...e.hu, tglx@...utronix.de, bphilips@...e.de,
	yinghai@...nel.org, akpm@...ux-foundation.org,
	torvalds@...ux-foundation.org, linux-kernel@...r.kernel.org,
	jeff@...zik.org
Subject: Re: [PATCH 06/12 UPDATED] irq: implement irq_schedule_poll()

Implement and use irq_schedule_poll() to schedule desc->poll_timer
instead of calling mod_timer directly.  irq_schedule_poll() is called
with desc->lock held and schedules the timer iff necessary - ie. if
the timer is offline or scheduled to expire later than requested.
This will be used to share desc->poll_timer.

* Jiri and Jonathan pointed out that IRQ_POLL_SLACK of HZ / 1000 would
  yield 0 slack on the popular 250HZ config.  Adjusted to HZ / 250.

Signed-off-by: Tejun Heo <tj@...nel.org>
Cc: Jiri Slaby <jirislaby@...il.com>
Cc: Jonathan Corbet <corbet@....net>
---
Minor update.  IRQ_POLL_SLACK bumped up to HZ / 250.

Thanks.

 kernel/irq/spurious.c |   47 +++++++++++++++++++++++++++++++++++++++++------
 1 file changed, 41 insertions(+), 6 deletions(-)

Index: work/kernel/irq/spurious.c
===================================================================
--- work.orig/kernel/irq/spurious.c
+++ work/kernel/irq/spurious.c
@@ -23,6 +23,8 @@ enum {

 	/* IRQ polling common parameters */
 	IRQ_POLL_INTV			= HZ / 100,	/* from the good ol' 100HZ tick */
+
+	IRQ_POLL_SLACK			= HZ / 250,	/* 1 tick slack w/ the popular 250HZ config */
 };

 int noirqdebug __read_mostly;
@@ -43,6 +45,38 @@ static void print_irq_handlers(struct ir
 	}
 }

+static unsigned long irq_poll_slack(unsigned long intv)
+{
+	return IRQ_POLL_SLACK;
+}
+
+/**
+ * irq_schedule_poll - schedule IRQ poll
+ * @desc: IRQ desc to schedule poll for
+ * @intv: poll interval
+ *
+ * Schedules @desc->poll_timer.  If the timer is already scheduled,
+ * it's modified iff jiffies + @intv + slack is before the timer's
+ * expires.  poll_timers aren't taken offline behind this function's
+ * back and the users of this function are guaranteed that poll_irq()
+ * will be called at or before jiffies + @intv + slack.
+ *
+ * CONTEXT:
+ * desc->lock
+ */
+static void irq_schedule_poll(struct irq_desc *desc, unsigned long intv)
+{
+	unsigned long expires = jiffies + intv;
+	int slack = irq_poll_slack(intv);
+
+	if (timer_pending(&desc->poll_timer) &&
+	    time_before_eq(desc->poll_timer.expires, expires + slack))
+		return;
+
+	set_timer_slack(&desc->poll_timer, slack);
+	mod_timer(&desc->poll_timer, expires);
+}
+
 /*
  * Recovery handler for misrouted interrupts.
  */
@@ -207,7 +241,9 @@ void __note_interrupt(unsigned int irq,
 		desc->depth++;
 		desc->chip->disable(irq);

-		mod_timer(&desc->poll_timer, jiffies + IRQ_POLL_INTV);
+		raw_spin_lock(&desc->lock);
+		irq_schedule_poll(desc, IRQ_POLL_INTV);
+		raw_spin_unlock(&desc->lock);
 	}
 	desc->irqs_unhandled = 0;
 }
@@ -221,9 +257,8 @@ void poll_irq(unsigned long arg)

 	raw_spin_lock_irq(&desc->lock);
 	try_one_irq(desc->irq, desc);
+	irq_schedule_poll(desc, IRQ_POLL_INTV);
 	raw_spin_unlock_irq(&desc->lock);
-
-	mod_timer(&desc->poll_timer, jiffies + IRQ_POLL_INTV);
 }

 void irq_poll_action_added(struct irq_desc *desc, struct irqaction *action)
@@ -238,10 +273,10 @@ void irq_poll_action_added(struct irq_de
 		__enable_irq(desc, desc->irq, false);
 	}

-	raw_spin_unlock_irqrestore(&desc->lock, flags);
-
 	if ((action->flags & IRQF_SHARED) && irqfixup >= IRQFIXUP_POLL)
-		mod_timer(&desc->poll_timer, jiffies + IRQ_POLL_INTV);
+		irq_schedule_poll(desc, IRQ_POLL_INTV);
+
+	raw_spin_unlock_irqrestore(&desc->lock, flags);
 }

 void irq_poll_action_removed(struct irq_desc *desc, struct irqaction *action)
--
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/