linux-kernel - Re: [PATCH 8/9] psi: pressure stall information for CPU, memory, and IO

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: <20180907102146.GI24142@hirez.programming.kicks-ass.net>
Date:   Fri, 7 Sep 2018 12:21:46 +0200
From:   Peter Zijlstra <peterz@...radead.org>
To:     Johannes Weiner <hannes@...xchg.org>
Cc:     Ingo Molnar <mingo@...hat.com>,
        Andrew Morton <akpm@...ux-foundation.org>,
        Linus Torvalds <torvalds@...ux-foundation.org>,
        Tejun Heo <tj@...nel.org>,
        Suren Baghdasaryan <surenb@...gle.com>,
        Daniel Drake <drake@...lessm.com>,
        Vinayak Menon <vinmenon@...eaurora.org>,
        Christopher Lameter <cl@...ux.com>,
        Peter Enderborg <peter.enderborg@...y.com>,
        Shakeel Butt <shakeelb@...gle.com>,
        Mike Galbraith <efault@....de>, linux-mm@...ck.org,
        cgroups@...r.kernel.org, linux-kernel@...r.kernel.org,
        kernel-team@...com
Subject: Re: [PATCH 8/9] psi: pressure stall information for CPU, memory, and
 IO

On Fri, Sep 07, 2018 at 12:16:34PM +0200, Peter Zijlstra wrote:
> This does the whole seqcount thing 6x, which is a bit of a waste.
> 
> struct snapshot {
> 	u32 times[NR_PSI_STATES];
> };
> 
> static inline struct snapshot get_times_snapshot(struct psi_group *pg, int cpu)
> {
> 	struct pci_group_cpu *pgc = per_cpu_ptr(pg->pcpu, cpu);
> 	struct snapshot s;
> 	unsigned int seq;
> 	u32 delta;
> 	int i;
> 
> 	do {
> 		seq = read_seqcount_begin(&pgc->seq);
> 
> 		delta = cpu_clock(cpu) - pgc->state_start;
> 		for (i = 0; i < NR_PSI_STATES; i++) {
> 			s.times[i] = gpc->times[i];
> 			if (test_state(pgc->tasks, i))
> 				s.times[i] += delta;
> 		}
> 
> 	} while (read_seqcount_retry(&pgc->seq, seq);

Sorry, I forgot the whole times_prev thing:

	for (i = 0; i < NR_PSI_STATES; i++) {
		tmp = s.times[i];
		s.times[i] -= pgc->times_prev[i];
		pgc->times_prev[i] = tmp;
	}

> 	return s;
> }
> 
> 
> 	for_each_possible_cpu(cpu) {
> 		struct snapshot s = get_times_snapshot(pg, cpu);
> 
> 		nonidle = nsecs_to_jiffies(s.times[PSI_NONIDLE]);
> 		nonidle_total += nonidle;
> 
> 		for (i = 0; i < PSI_NONIDLE; i++)
> 			deltas[s] += (u64)s.times[i] * nonidle;
> 
> 		/* ... */
> 
> 	}
> 
> 
> It's a bit cumbersome, but that's because of C.