From: Gabriele Monaco <gmonaco@redhat.com>
To: Tobias Schaffner <tobias.schaffner@siemens.com>,
Steven Rostedt <rostedt@goodmis.org>
Cc: Jonathan Corbet <corbet@lwn.net>, Shuah Khan <shuah@kernel.org>,
Randy Dunlap <rdunlap@infradead.org>,
Masami Hiramatsu <mhiramat@kernel.org>,
Mathieu Desnoyers <mathieu.desnoyers@efficios.com>,
linux-trace-kernel@vger.kernel.org, linux-doc@vger.kernel.org,
linux-kselftest@vger.kernel.org, linux-kernel@vger.kernel.org,
rpm@xenomai.org, jan.kiszka@siemens.com
Subject: Re: [PATCH v3 2/4] rv: add per-edge dwell-time statistics
Date: Tue, 17 Nov 2026 16:17:29 +0100 [thread overview]
Message-ID: <eb08b0075a686d0f1907160c4e4880ed779e02d8.camel@redhat.com> (raw)
In-Reply-To: <20261020095512.863754-3-tobias.schaffner@siemens.com>
On Tue, 2026-10-20 at 11:55 +0200, Tobias Schaffner wrote:
> Add CONFIG_RV_EDGE_STAT, an optional facility that records how long a
> monitor's automaton dwells in a state before leaving it and accounts that
> dwell to the edge it leaves by.
>
> Keep the state and its entry timestamp in one word so a transition commits
> both with a single cmpxchg and nested transitions cannot charge a dwell to
> the wrong edge. This requires a native 64-bit cmpxchg.
>
> The counters are per-CPU and lock-free, so a monitor's hot path can update
> them without disabling interrupts and without perturbing the latency being
> measured. Only the owning CPU writes them, so a reader snapshots them with
> local64_read() with no IPI and no locking on the accounting path.
>
> Storage is a static per-CPU array of STATE_MAX * EVENT_MAX edges per
> monitor, exposed through a per-monitor "stats" tracefs file. Report whether
> the model defines each edge, so a count on an edge that should never be
> taken is visible. Nothing uses the facility yet.
>
> Signed-off-by: Tobias Schaffner <tobias.schaffner@siemens.com>
I took too long to finally review this. It looks good, just a few minor
comments.
> ---
> include/linux/rv.h | 7 ++
> include/rv/edge_stat.h | 211 ++++++++++++++++++++++++++++++++++++++++
> kernel/trace/rv/Kconfig | 12 +++
> 3 files changed, 230 insertions(+)
> create mode 100644 include/rv/edge_stat.h
>
> diff --git a/include/linux/rv.h b/include/linux/rv.h
> index 541ba404926a..1dc7c45f476c 100644
> --- a/include/linux/rv.h
> +++ b/include/linux/rv.h
> @@ -21,6 +21,13 @@
> #include <linux/list.h>
> #include <linux/types.h>
>
> +/* Edge statistics pack the state and its entry timestamp into one word. */
> +#ifdef CONFIG_RV_EDGE_STAT
> +#define da_state_t unsigned long
> +#else
> +#define da_state_t unsigned int
> +#endif
> +
I presume here you are using a define instead of a typedef just to avoid
checkpatch and friends to complain. Right?
I'm not sure this is any better to be fair. Typedef is what you want here and I
would argue it's the right tool for the job.
This exact use is mentioned as an exception (b) to the no typedefs rule [1].
[1] - https://www.kernel.org/doc/html/latest/process/coding-style.html#typedefs
> /*
> * Deterministic automaton per-object variables.
> */
> diff --git a/include/rv/edge_stat.h b/include/rv/edge_stat.h
> new file mode 100644
> index 000000000000..8b99e8520e1e
> --- /dev/null
> +++ b/include/rv/edge_stat.h
> @@ -0,0 +1,211 @@
> +/* SPDX-License-Identifier: GPL-2.0 */
> +/*
> + * Per-edge dwell-time statistics for RV monitors.
> + *
> + * Copyright (C) 2026 Siemens AG
> + * Author: Tobias Schaffner <tobias.schaffner@siemens.com>
> + */
> +#ifndef _RV_EDGE_STAT_H
> +#define _RV_EDGE_STAT_H
> +
> +#include <linux/rv.h>
> +
> +#ifdef CONFIG_RV_EDGE_STAT
> +
> +#include <linux/args.h>
> +#include <linux/atomic.h>
> +#include <linux/bitfield.h>
> +#include <linux/bug.h>
> +#include <linux/compiler.h>
> +#include <linux/percpu.h>
> +#include <linux/sched/clock.h>
> +#include <linux/seq_file.h>
> +#include <linux/types.h>
> +#include <asm/local64.h>
> +#include <rv/automata.h>
> +#include "rv.h"
Including this apparently opened a can of worms and BTF infos no longer build.
I'll send a fix separately, it isn't really related to your change.
> +
> +/*
> + * Per-CPU counters kept in local64_t so accounting is safe against interrupt
> + * and NMI nesting on the owning CPU without disabling interrupts. Only the
> + * owning CPU writes.
> + */
> +struct rv_edge_stat {
> + local64_t count;
> + local64_t sum_ns;
> + local64_t max_ns;
> +};
> +
> +/*
> + * State and entry timestamp share one word so a transition commits both with
> + * one cmpxchg. Comparing the timestamp also detects nested transitions that
> + * return to the same state.
> + */
> +#define RV_STATE_BITS 8
> +#define RV_STATE_MASK GENMASK(RV_STATE_BITS - 1, 0)
> +#define RV_TS_MASK GENMASK(BITS_PER_LONG - 1, RV_STATE_BITS)
> +
> +#define da_state_of(w) FIELD_GET(RV_STATE_MASK, w)
> +#define da_ts_of(w) FIELD_GET(RV_TS_MASK, w)
> +#define da_state_pack(s, ts) (FIELD_PREP(RV_TS_MASK, ts) | \
> + FIELD_PREP(RV_STATE_MASK, s))
> +
> +static __always_inline
> +void rv_edge_stat_account(struct rv_edge_stat *s, u64 dwell_ns)
> +{
> + s64 max;
> + int i;
> +
> + local64_inc(&s->count);
> + local64_add(dwell_ns, &s->sum_ns);
> +
> + /* Keep the largest dwell; bound retries if a nested update races us.
> */
> + max = local64_read(&s->max_ns);
> + for (i = 0; dwell_ns > (u64)max; i++) {
> + if (i == MAX_DA_RETRY_RACING_EVENTS) {
> + WARN_ONCE(1, "rv: edge-stat max update exceeded %d
> retries\n",
> + MAX_DA_RETRY_RACING_EVENTS);
> + break;
> + }
> + if (local64_try_cmpxchg(&s->max_ns, &max, dwell_ns))
> + break;
> + }
> +}
> +
> +static_assert(STATE_MAX <= (1U << RV_STATE_BITS),
> + "automaton has more states than the packed state word can
> hold");
> +
> +#if RV_MON_TYPE == RV_MON_PER_CPU
> +
> +/*
> + * One edge per (state, event) pair. Static storage avoids allocation
> lifetime
> + * races. Only the owning CPU writes, so readers can snapshot counters with
> + * local64_read().
> + */
> +#define RV_THIS_NR_EDGES (STATE_MAX * EVENT_MAX)
> +
> +#define DA_EDGE(state, event) ((state) * EVENT_MAX + (event))
> +#define DA_EDGE_STATE(edge) ((edge) / EVENT_MAX)
> +#define DA_EDGE_EVENT(edge) ((edge) % EVENT_MAX)
> +
> +#define DA_MON_EDGES CONCATENATE(da_mon_edges_, MONITOR_NAME)
> +
> +static DEFINE_PER_CPU(struct rv_edge_stat[RV_THIS_NR_EDGES], DA_MON_EDGES);
> +static struct dentry *rv_this_stats_file;
> +
> +/* Called before tracepoints are registered, so memset cannot race an update.
> */
> +static void rv_edge_stats_reset(void)
> +{
> + int cpu;
> +
> + for_each_possible_cpu(cpu)
> + memset(per_cpu_ptr(DA_MON_EDGES, cpu), 0,
> sizeof(DA_MON_EDGES));
> +}
> +
> +static int rv_edge_stats_show(struct seq_file *seq, void *v)
> +{
> + unsigned int e;
> + int cpu;
> +
> + seq_puts(seq, "# cpu edge label valid count sum_ns max_ns\n");
> + /* Include offline CPUs and keep output stable across CPU hotplug. */
> + for_each_possible_cpu(cpu) {
> + struct rv_edge_stat *edges = per_cpu_ptr(DA_MON_EDGES, cpu);
> +
> + for (e = 0; e < RV_THIS_NR_EDGES; e++) {
> + enum states state = DA_EDGE_STATE(e);
> + enum events event = DA_EDGE_EVENT(e);
> +
> + /* Edges the model does not define must stay at zero.
> */
> + seq_printf(seq, "%d %u %s:%s %d %llu %llu %llu\n",
> + cpu, e,
> + model_get_state_name(state),
> + model_get_event_name(event),
> + model_get_next_state(state, event) !=
> INVALID_STATE,
> + (u64)local64_read(&edges[e].count),
> + (u64)local64_read(&edges[e].sum_ns),
> + (u64)local64_read(&edges[e].max_ns));
> + }
> + }
> + return 0;
> +}
> +DEFINE_SHOW_ATTRIBUTE(rv_edge_stats);
> +
> +static int rv_edge_stats_create(struct dentry *parent)
> +{
> + rv_edge_stats_reset();
> + rv_this_stats_file = rv_create_file("stats", RV_MODE_READ, parent,
> + NULL, &rv_edge_stats_fops);
> + return rv_this_stats_file ? 0 : -ENOMEM;
> +}
> +
> +static void rv_edge_stats_remove(void)
> +{
> + rv_remove(rv_this_stats_file);
> + rv_this_stats_file = NULL;
> +}
> +
> +/* A zero timestamp marks resets, which can run on a different CPU. */
> +#define da_state_entered(s) da_state_pack((s), local_clock())
> +
> +static __always_inline void
> +rv_edge_account(da_state_t old, da_state_t new, enum states curr, enum events
> ev)
> +{
> + struct rv_edge_stat *edges;
> + u64 prev = da_ts_of(old);
> +
> + if (unlikely(!prev))
> + return;
> +
> + edges = this_cpu_ptr(DA_MON_EDGES);
> + rv_edge_stat_account(&edges[DA_EDGE(curr, ev)],
> + (da_ts_of(new) - prev) & FIELD_MAX(RV_TS_MASK));
> +}
> +
> +#else /* per-cpu accounting off for this monitor type */
> +
> +#define da_state_entered(s) ((da_state_t)(s))
> +
> +static __always_inline void
> +rv_edge_account(da_state_t old, da_state_t new, enum states curr, enum events
> ev) { }
> +
> +#endif /* RV_MON_TYPE == RV_MON_PER_CPU */
> +
> +static __always_inline bool
> +da_state_try_commit(da_state_t *word, da_state_t *old, enum states next,
> + enum events event)
> +{
> + da_state_t prev = *old;
> + da_state_t new = da_state_entered(next);
> +
> + if (!try_cmpxchg(word, old, new))
> + return false;
> +
> + rv_edge_account(prev, new, da_state_of(prev), event);
> + return true;
> +}
> +
> +#else /* !CONFIG_RV_EDGE_STAT */
> +
> +/*
> + * Feature off: only the hooks the DA/HA layer calls are provided, as trivial
> + * pass-throughs. da_state_t is a plain state word with no packed timestamp.
> + */
> +#define da_state_of(w) (w)
> +#define da_state_entered(s) (s)
> +
> +static __always_inline bool
> +da_state_try_commit(da_state_t *word, da_state_t *old, enum states next,
> + enum events event)
> +{
> + return try_cmpxchg(word, old, next);
> +}
> +
> +#if RV_MON_TYPE == RV_MON_PER_CPU
> +static inline int rv_edge_stats_create(struct dentry *parent) { return 0; }
> +static inline void rv_edge_stats_remove(void) { }
> +#endif
> +
> +#endif /* CONFIG_RV_EDGE_STAT */
> +
> +#endif /* _RV_EDGE_STAT_H */
> diff --git a/kernel/trace/rv/Kconfig b/kernel/trace/rv/Kconfig
> index efa930f94ea4..54f7c50a74a8 100644
> --- a/kernel/trace/rv/Kconfig
> +++ b/kernel/trace/rv/Kconfig
> @@ -59,6 +59,18 @@ config RV_PER_TASK_MONITORS
> This option configures the maximum number of per-task RV monitors
> that can run
> simultaneously.
>
> +config RV_EDGE_STAT
> + bool "Per-edge dwell-time statistics"
> + depends on RV && 64BIT && DA_MON_EVENTS_IMPLICIT
> + help
> + Record per-edge dwell-time statistics for per-cpu deterministic and
> + hybrid automaton monitors and expose them through a per-monitor
> + "stats" tracefs file. This times each monitored automaton
> transition
> + with local_clock(), so leave it off if you do not need the
> + statistics.
> +
> + If unsure, say N.
> +
Mostly personal preference, but I would add new configs at the end of the file
(e.g. after reactors and KUnit).
Thanks,
Gabriele
> source "kernel/trace/rv/monitors/wip/Kconfig"
> source "kernel/trace/rv/monitors/wwnr/Kconfig"
>
next prev parent reply other threads:[~2026-11-17 15:17 UTC|newest]
Thread overview: 7+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-10-20 9:55 [PATCH v3 0/4] rv: per-edge dwell-time statistics for per-cpu monitors Tobias Schaffner
2026-10-20 9:55 ` [PATCH v3 1/4] rv: guard rv.h against double inclusion Tobias Schaffner
2026-10-20 9:55 ` [PATCH v3 2/4] rv: add per-edge dwell-time statistics Tobias Schaffner
2026-11-17 15:17 ` Gabriele Monaco [this message]
2026-10-20 9:55 ` [PATCH v3 3/4] rv: collect per-edge dwell time for per-cpu DA/HA monitors Tobias Schaffner
2026-10-20 10:06 ` sashiko-bot
2026-10-20 9:55 ` [PATCH v3 4/4] selftests/verification: test per-edge dwell-time statistics Tobias Schaffner
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=eb08b0075a686d0f1907160c4e4880ed779e02d8.camel@redhat.com \
--to=gmonaco@redhat.com \
--cc=corbet@lwn.net \
--cc=jan.kiszka@siemens.com \
--cc=linux-doc@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-kselftest@vger.kernel.org \
--cc=linux-trace-kernel@vger.kernel.org \
--cc=mathieu.desnoyers@efficios.com \
--cc=mhiramat@kernel.org \
--cc=rdunlap@infradead.org \
--cc=rostedt@goodmis.org \
--cc=rpm@xenomai.org \
--cc=shuah@kernel.org \
--cc=tobias.schaffner@siemens.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox