From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1161058AbWBYS6A (ORCPT ); Sat, 25 Feb 2006 13:58:00 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1161059AbWBYS6A (ORCPT ); Sat, 25 Feb 2006 13:58:00 -0500 Received: from ylpvm43-ext.prodigy.net ([207.115.57.74]:60587 "EHLO ylpvm43.prodigy.net") by vger.kernel.org with ESMTP id S1161058AbWBYS6A (ORCPT ); Sat, 25 Feb 2006 13:58:00 -0500 X-ORBL: [67.117.73.34] Date: Sat, 25 Feb 2006 10:57:31 -0800 From: Tony Lindgren To: Thomas Gleixner Cc: akpm@osdl.org, heiko.carstens@de.ibm.com, johnstul@us.ibm.com, rmk@arm.linux.org.uk, schwidefsky@de.ibm.com, LKML Subject: Re: + fix-next_timer_interrupt-for-hrtimer.patch added to -mm tree Message-ID: <20060225185731.GA4294@atomide.com> References: <200602250219.k1P2JLqY018864@shell0.pdx.osdl.net> <1140884243.5237.104.camel@localhost.localdomain> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <1140884243.5237.104.camel@localhost.localdomain> User-Agent: Mutt/1.5.11 Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org * Thomas Gleixner [060225 08:16]: > > +#ifdef CONFIG_NO_IDLE_HZ > > +/** > > + * hrtimer_get_next - get next hrtimer to expire > > + * > > + * @bases: ktimer base array > > + */ > > +static inline struct hrtimer *hrtimer_get_next(struct hrtimer_base *bases) > > +{ > > + unsigned long flags; > > + struct hrtimer *timer = NULL; > > + int i; > > + > > + for (i = 0; i < MAX_HRTIMER_BASES; i++) { > > + struct hrtimer_base *base; > > + struct hrtimer *cur; > > + > > + base = &bases[i]; > > + spin_lock_irqsave(&base->lock, flags); > > + cur = rb_entry(base->first, struct hrtimer, node); > > + spin_unlock_irqrestore(&base->lock, flags); > > + > > + if (cur == NULL) > > + continue; > > + > > + if (timer == NULL || cur->expires.tv64 < timer->expires.tv64) > > + timer = cur; > > + } > > + > > + return timer; > > +} > > This is racy on SMP. nanosleep hrtimers are on the stack and can go away > due to a signal. posix timers can be removed and destroyed on another > CPU. This should be fixed. But just as a note, we can tolerate some removed timer values values as it would be just an extra timer interrupt. > Also the expires fields of CLOCK_MONOTONIC and CLOCK_REALTIME based > tiemrs can not really be compared. Expiry value is absolute time of the > respective base clock. OK. How can we get the first event in nanoseconds easily? > > +/** > > + * ktime_to_jiffies - converts ktime to jiffies > > + * > > + * @event: ktime event to be converted to jiffies > > + * > > + * Caller must take care of xtime locking. > > + */ > > +static inline unsigned long ktime_to_jiffies(const ktime_t event) > > +{ > > + ktime_t now, delta; > > + > > + now = timespec_to_ktime(xtime); > > + delta = ktime_sub(event, now); > > + > > + return jiffies + (((delta.tv64 * NSEC_CONVERSION) >> > > + (NSEC_JIFFIE_SC - SEC_JIFFIE_SC)) >> SEC_JIFFIE_SC); > > +} > > Only CLOCK_REALTIME based timers are based on xtime. For CLOCK_MONOTONIC > based timers this calculation is off by wall_to_monotonic. Needs to be fixed then. > > +/** > > + * hrtimer_next_jiffie - get next hrtimer event in jiffies > > + * > > + * Called from next_timer_interrupt() to get the next hrtimer event. > > + * Eventually we should change next_timer_interrupt() to return > > + * results in nanoseconds instead of jiffies. Caller must host xtime_lock. > > S390 does not hold xtime lock when calling next_timer_interrupt ! Good point. Needs to be fixed. > I look for a sane solution. We really should fix this for 2.6.16. Then maybe after 2.6.16 we can convert next_timer_interrupt() to return nanosecond instead of jiffies. And then we can get rid of jiffies again in the hrtimer code :) Tony