Received-SPF: pass (google.com: domain of linux-kernel+bounces-195287-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:45d1:ec00::1 as permitted sender) client-ip=2604:1380:45d1:ec00::1;
Date: Thu, 30 May 2024 12:10:44 +0100
From: Qais Yousef <qyousef@layalina.io>
To: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Ingo Molnar <mingo@kernel.org>, Peter Zijlstra <peterz@infradead.org>,
	Juri Lelli <juri.lelli@redhat.com>,
	Steven Rostedt <rostedt@goodmis.org>,
	Vincent Guittot <vincent.guittot@linaro.org>,
	Daniel Bristot de Oliveira <bristot@redhat.com>,
	Thomas Gleixner <tglx@linutronix.de>,
	Alexander Viro <viro@zeniv.linux.org.uk>,
	Christian Brauner <brauner@kernel.org>,
	Andrew Morton <akpm@linux-foundation.org>,
	Jens Axboe <axboe@kernel.dk>, linux-kernel@vger.kernel.org,
	linux-fsdevel@vger.kernel.org, linux-trace-kernel@vger.kernel.org,
	linux-mm@kvack.org, Phil Auld <pauld@redhat.com>
Subject: Re: [PATCH v2] sched/rt: Clean up usage of rt_task()
Message-ID: <20240530111044.d4jegeiueizvdjrg@airbuntu>
References: <20240515220536.823145-1-qyousef@layalina.io>
 <20240521110035.KRIwllGe@linutronix.de>
 <20240527172650.kieptfl3zhyljkzx@airbuntu>
 <20240529082912.gPDpgVy3@linutronix.de>
 <20240529103409.3iiemroaavv5lh2p@airbuntu>
 <20240529105528.9QBTCqCr@linutronix.de>
Precedence: bulk
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
In-Reply-To: <20240529105528.9QBTCqCr@linutronix.de>

On 05/29/24 12:55, Sebastian Andrzej Siewior wrote:
> On 2024-05-29 11:34:09 [+0100], Qais Yousef wrote:
> > > behaviour. But then it is insistent which matters only in the RT case.
> > > Puh. Any sched folks regarding policy?
> > 
> > I am not sure I understood you here. Could you rephrase please?
> 
> Right now a SCHED_OTHER task boosted to a realtime priority gets
> slack=0. In the !RT scenario everything is fine.
> For RT the slack=0 also happens but the init of the timer looks at the
> policy instead at the possible boosted priority and uses a different
> clock attribute. This can lead to a delayed wake up (so avoiding the
> slack does not solve the problem).
> 
> This is not consistent because IMHO the clock setup & slack should be
> handled equally. So I am asking the sched folks for a policy and I am
> leaning towards looking at task-policy in this case instead of prio
> because you shouldn't do anything that can delay.

Can't we do that based on is_soft/is_hard flag in hrtimer struct when we apply
the slack in hrtimer_set_expires_range_ns() instead?

(not compile tested even)

diff --git a/include/linux/hrtimer.h b/include/linux/hrtimer.h
index aa1e65ccb615..e001f20bbea9 100644
--- a/include/linux/hrtimer.h
+++ b/include/linux/hrtimer.h
@@ -102,12 +102,16 @@ static inline void hrtimer_set_expires(struct hrtimer *timer, ktime_t time)
 
 static inline void hrtimer_set_expires_range(struct hrtimer *timer, ktime_t time, ktime_t delta)
 {
+       if (timer->is_soft || timer->is_hard)
+               delta = 0;
        timer->_softexpires = time;
        timer->node.expires = ktime_add_safe(time, delta);
 }
 
 static inline void hrtimer_set_expires_range_ns(struct hrtimer *timer, ktime_t time, u64 delta)
 {
+       if (timer->is_soft || timer->is_hard)
+               delta = 0;
        timer->_softexpires = time;
        timer->node.expires = ktime_add_safe(time, ns_to_ktime(delta));
 }