Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1757459Ab0F3Xxq (ORCPT ); Wed, 30 Jun 2010 19:53:46 -0400 Received: from ogre.sisk.pl ([217.79.144.158]:37897 "EHLO ogre.sisk.pl" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752551Ab0F3Xxo (ORCPT ); Wed, 30 Jun 2010 19:53:44 -0400 From: "Rafael J. Wysocki" To: Alan Stern Subject: Re: [update] Re: [PATCH] PM: Make it possible to avoid wakeup events from being lost Date: Thu, 1 Jul 2010 01:52:03 +0200 User-Agent: KMail/1.13.3 (Linux/2.6.35-rc3-rjw+; KDE/4.4.3; x86_64; ; ) Cc: linux-pm@lists.linux-foundation.org, Linux Kernel Mailing List , Neil Brown , Matthew Garrett , mark gross <640e9920@gmail.com>, Arve =?iso-8859-1?q?Hj=F8nnev=E5g?= , Dmitry Torokhov , Florian Mickler , linux-pci@vger.kernel.org, Jesse Barnes References: In-Reply-To: MIME-Version: 1.0 Content-Type: Text/Plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Message-Id: <201007010152.03869.rjw@sisk.pl> Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 4575 Lines: 144 On Wednesday, June 30, 2010, Alan Stern wrote: > On Wed, 30 Jun 2010, Rafael J. Wysocki wrote: > > > > Since there's no longer any way to cancel a call to pm_wakeup_event() > > > or close the "no suspend" period early, there is no need to use > > > dynamically-allocated delayed_work structures. You can make do with a > > > single static timer; always keep it set to expire at the latest time > > > passed to pm_wakeup_event(). > > > > The decremenations of events_in_progress wouldn't be balanced with > > incrementations this way. Or do you have any clever way of dealing with > > that in mind? > > Keep track of the current expiration time in a static variable called > wakeup_timeout, and use 0 to indicate there is no timeout. I invented a slightly different version in the meantime, which is appended as a patch on top of the original one (I don't want to modify the original patch, since it's been reviewed already by several people and went to my linux-next branch). > In pm_wakeup_event() (everything protected by the spinlock): > > unsigned long new_timeout = jiffies + msecs_to_jiffies(msecs); > if (new_timeout == 0) > new_timeout = 1; > > ++event_count; > if (!wakeup_timeout || time_after(new_timeout, wakeup_timeout)) { > if (!wakeup_timeout) > ++events_in_progress; > wakeup_timeout = new_timeout; > mod_timer(&wakeup_timer, wakeup_timeout); > } > > In the timer routine: > > if (wakeup_timeout && time_before_eq(wakeup_timeout, jiffies)) { > --events_in_progres; > wakeup_timeout = 0; > } > --- drivers/base/power/wakeup.c | 47 +++++++++++++++++++++++++++++++++----------- 1 file changed, 36 insertions(+), 11 deletions(-) Index: linux-2.6/drivers/base/power/wakeup.c =================================================================== --- linux-2.6.orig/drivers/base/power/wakeup.c +++ linux-2.6/drivers/base/power/wakeup.c @@ -9,6 +9,7 @@ #include #include #include +#include #include #include #include @@ -28,6 +29,12 @@ static unsigned long events_in_progress; static DEFINE_SPINLOCK(events_lock); +static void pm_wakeup_timer_fn(unsigned long data); + +static DEFINE_TIMER(events_timer, pm_wakeup_timer_fn, 0, 0); +static unsigned long events_timer_expires; +static unsigned long delayed_count; + /* * The functions below use the observation that each wakeup event starts a * period in which the system should not be suspended. The moment this period @@ -103,17 +110,23 @@ void pm_relax(void) } /** - * pm_wakeup_work_fn - Deferred closing of a wakeup event. + * pm_wakeup_timer_fn - Deferred closing of a wakeup event. * * Execute pm_relax() for a wakeup event detected in the past and free the * work item object used for queuing up the work. */ -static void pm_wakeup_work_fn(struct work_struct *work) +static void pm_wakeup_timer_fn(unsigned long data) { - struct delayed_work *dwork = to_delayed_work(work); + unsigned long flags; - pm_relax(); - kfree(dwork); + spin_lock_irqsave(&events_lock, flags); + if (events_timer_expires && time_after(jiffies, events_timer_expires)) { + events_in_progress -= delayed_count; + event_count += delayed_count; + delayed_count = 0; + events_timer_expires = 0; + } + spin_unlock_irqrestore(&events_lock, flags); } /** @@ -132,19 +145,31 @@ static void pm_wakeup_work_fn(struct wor void pm_wakeup_event(struct device *dev, unsigned int msec) { unsigned long flags; - struct delayed_work *dwork; - - dwork = msec ? kzalloc(sizeof(*dwork), GFP_ATOMIC) : NULL; spin_lock_irqsave(&events_lock, flags); if (dev) dev->power.wakeup_count++; - if (dwork) { - INIT_DELAYED_WORK(dwork, pm_wakeup_work_fn); - schedule_delayed_work(dwork, msecs_to_jiffies(msec)); + if (msec) { + ktime_t kt; + struct timespec ts; + unsigned long expires; + + kt = ktime_get(); + kt = ktime_add_ns(kt, msec * NSEC_PER_MSEC); + ts = ktime_to_timespec(kt); + expires = timespec_to_jiffies(&ts); + if (!expires) + expires = 1; + + if (!events_timer_expires + || time_after(expires, events_timer_expires)) { + mod_timer(&events_timer, expires); + events_timer_expires = expires; + } events_in_progress++; + delayed_count++; } else { event_count++; } -- 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/