mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: Leo Yan <leo.yan@arm.com>
To: Peter Zijlstra <peterz@infradead.org>,
	Ingo Molnar <mingo@redhat.com>,
	 Arnaldo Carvalho de Melo <acme@kernel.org>,
	 Namhyung Kim <namhyung@kernel.org>,
	Ian Rogers <irogers@google.com>,
	 Adrian Hunter <adrian.hunter@intel.com>,
	 Suzuki K Poulose <suzuki.poulose@arm.com>,
	 James Clark <james.clark@linaro.org>,
	Mike Leach <mike.leach@arm.com>,
	 Anshuman Khandual <anshuman.khandual@arm.com>,
	 Yeoreum Yun <yeoreum.yun@arm.com>,
	Jonathan Corbet <corbet@lwn.net>,  Suyash Mahar <smahar@meta.com>,
	Amir Ayupov <aaupov@fb.com>
Cc: Ingo Molnar <mingo@kernel.org>,
	linux-perf-users@vger.kernel.org,  linux-kernel@vger.kernel.org,
	coresight@lists.linaro.org,
	 linux-arm-kernel@lists.infradead.org, linux-doc@vger.kernel.org,
	 Leo Yan <leo.yan@arm.com>
Subject: [PATCH 24/25] perf test: Add CoreSight AUX sample decoding test
Date: Tue, 15 Sep 2026 16:49:10 +0100	[thread overview]
Message-ID: <20260915-arm_cs_support_aux_sample-v1-24-35cf6f297d20@arm.com> (raw)
In-Reply-To: <20260915-arm_cs_support_aux_sample-v1-0-35cf6f297d20@arm.com>

Add a shell test that records the brstack workload with 8 KiB AUX
samples attached to cycle events. Pin the workload to a CPU with a TRBE
sink and use per-thread recording with trace timestamps and context IDs
disabled, exercising decoding with the owning sample's context.

Check branch stacks and callchains with --itrace=L4, L64, G3 and G3L64.
Require at least one sample to reach both requested depths, with all
callchain frames belonging to the workload, and reject samples exceeding
either limit. Use a callchain depth of three to match the nested calls
in brstack.

Compare with AUX decoding enabled and disabled to verify that adding
history preserves the original PMU samples.

Assisted-by: Codex:gpt-6
Signed-off-by: Leo Yan <leo.yan@arm.com>
---
 tools/perf/tests/shell/coresight/aux_sample.sh | 189 +++++++++++++++++++++++++
 1 file changed, 189 insertions(+)

diff --git a/tools/perf/tests/shell/coresight/aux_sample.sh b/tools/perf/tests/shell/coresight/aux_sample.sh
new file mode 100755
index 0000000000000000000000000000000000000000..887ad6a085eacdfd192db3863fbb61dae1aa19b5
--- /dev/null
+++ b/tools/perf/tests/shell/coresight/aux_sample.sh
@@ -0,0 +1,189 @@
+#!/bin/bash -e
+# SPDX-License-Identifier: GPL-2.0
+# CoreSight AUX samples with callchains and branch stacks (exclusive)
+
+export LC_ALL=C
+
+skip()
+{
+	echo "[Skip] $1"
+	exit 2
+}
+
+perf list pmu | grep -q 'cs_etm//' || skip "cs_etm is not available"
+perf check feature -q libopencsd || skip "perf was built without OpenCSD"
+[ "$(id -u)" = 0 ] || skip "No root permission"
+command -v taskset >/dev/null || skip "taskset is not available"
+
+# AUX sample now supports per CPU sink (e.g., TRBE). Select a usable CPU
+# and name its sink explicitly to avoid a shared sink.
+sink=
+for dev in /sys/bus/coresight/devices/trbe[0-9]*; do
+	[ -d "$dev" ] || continue
+	name=${dev##*/}
+	cpu=${name#trbe}
+	[ -e "/sys/bus/event_source/devices/cs_etm/cpu$cpu" ] || continue
+	if taskset -c "$cpu" true 2>/dev/null; then
+		sink=$name
+		break
+	fi
+done
+[ -n "$sink" ] || skip "No usable TRBE CPU is available"
+
+tmpdir=$(mktemp -d /tmp/perf-cs-aux-sample.XXXXXX)
+
+cleanup()
+{
+	rm -rf "$tmpdir"
+}
+
+trap cleanup EXIT
+trap 'exit 1' TERM INT
+
+fail()
+{
+	echo "$1" >&2
+	cat "$tmpdir/stderr" >&2
+	exit 1
+}
+
+cf="$tmpdir/ctl"
+af="$tmpdir/ack"
+mkfifo "$cf" "$af"
+
+# Disable trace timestamps and context IDs to use the owning sample's context.
+# Use a per-thread mmap so the explicit sink is only used on the pinned CPU.
+if ! taskset -c "$cpu" perf record -o "$tmpdir/data" --aux-sample=8192 \
+	-e "{cs_etm/@$sink,timestamp=0,contextid=0/u,cycles/period=100003/u}" \
+	--per-thread -D -1 --control fifo:"$cf","$af" -- \
+	perf test --record-ctl fifo:"$cf","$af" -w brstack 1000000 \
+	>/dev/null 2>"$tmpdir/stderr"; then
+	# Kernels without snapshot_aux reject the sampling event at open time.
+	if grep -Eq 'sys_perf_event_open.*event \(.*cycles.*\): Invalid argument' \
+		"$tmpdir/stderr"; then
+		skip "Kernel does not accept CoreSight AUX sampling"
+	fi
+	fail "Failed to record CoreSight AUX samples"
+fi
+
+perf evlist -v -i "$tmpdir/data" >"$tmpdir/evlist" 2>"$tmpdir/stderr" ||
+	fail "Failed to read AUX sample attributes"
+grep -Eq 'sample_type:.*\|AUX(,|\|).*aux_sample_size: 8192(,|$)' \
+	"$tmpdir/evlist" || fail "Missing AUX sample attributes"
+
+check_history()
+{
+	local options=$1
+	local max_branches=$2
+	local max_callchains=$3
+	local output="$tmpdir/script-$options"
+
+	perf script -i "$tmpdir/data" --itrace="$options" \
+		-F comm,pid,tid,cpu,event,ip,sym,brstack >"$output" \
+		2>"$tmpdir/stderr" || fail "Failed to decode AUX samples with $options"
+
+	# Some windows may contain incomplete trace. Require at least one cycle
+	# sample to reach both requested depths, and reject any that exceed them.
+	if ! awk -v max_branches="$max_branches" \
+		-v max_callchains="$max_callchains" '
+		function check_sample() {
+			if (in_sample == 0) {
+				return
+			}
+
+			if (branch_entries > max_branches || callchain_frames > max_callchains) {
+				depth_exceeded = 1
+			}
+
+			# Both depths must be reached in the same sample, with all
+			# callchain frames belonging to the brstack workload.
+			if (branch_entries == max_branches &&
+			    callchain_frames == max_callchains &&
+			    workload_frames == max_callchains) {
+				found_sample = 1
+			}
+
+			branch_entries = 0
+			callchain_frames = 0
+			workload_frames = 0
+			in_sample = 0
+		}
+		{
+			# A blank line marks the end of a sample.
+			if (NF == 0) {
+				check_sample()
+				next
+			}
+
+			# Sample header: comm pid/tid [cpu] event ...
+			# The event may include a PMU prefix and configuration terms.
+			if ($4 ~ /(^|\/)cycles([,\/:]|$)/) {
+				check_sample()
+				in_sample = 1
+			}
+			if (in_sample == 0) {
+				next
+			}
+
+			# Each branch entry is a field in 0xFROM/0xTO/... form.
+			for (i = 1; i <= NF; i++) {
+				if ($i ~ /^0x[[:xdigit:]]+\/0x[[:xdigit:]]+\//) {
+					branch_entries++
+				}
+			}
+
+			# Callchain rows start with a hexadecimal IP, then a symbol.
+			if ($1 ~ /^[[:xdigit:]]+$/) {
+				callchain_frames++
+				if ($2 ~ /^brstack(_|$)/) {
+					workload_frames++
+				}
+			}
+		}
+		END {
+			# Account for the last sample even without a trailing blank line.
+			check_sample()
+
+			if (depth_exceeded != 0) {
+				print "AUX sample history exceeds the requested depth" > "/dev/stderr"
+				exit 1
+			}
+			if (found_sample == 0) {
+				printf "No sample has %d branches and %d workload callchain frames\n", \
+					max_branches, max_callchains > "/dev/stderr"
+				exit 1
+			}
+			exit 0
+		}
+	' "$output"; then
+		head -n 80 "$output" >&2
+		fail "Missing AUX sample history or incorrect depth with $options"
+	fi
+}
+
+check_sample_identity()
+{
+	local fields=comm,pid,tid,cpu,event,ip
+
+	# Hiding history must leave exactly the original PMU samples, including
+	# their PID/TID, CPU and sampled IP, with no synthesized events. Per-thread
+	# recording does not require sample timestamps.
+	perf script -i "$tmpdir/data" --no-itrace -F "$fields" \
+		>"$tmpdir/original" 2>"$tmpdir/stderr" || fail "Failed to read cycle samples"
+	perf script -i "$tmpdir/data" --itrace=L4 -F "$fields" \
+		>"$tmpdir/decoded" 2>"$tmpdir/stderr" || fail "Failed to decode cycle samples"
+	diff -u "$tmpdir/original" "$tmpdir/decoded" ||
+		fail "AUX decoding changed the original cycle samples"
+}
+
+check_history L4 4 0
+check_history L64 64 0
+
+# brstack -> brstack_bench -> brstack_foo -> brstack_bar provides three
+# nested calls whose caller frames can be reconstructed from an AUX window.
+check_history G3 0 3
+check_history G3L64 64 3
+
+check_sample_identity
+
+echo "CoreSight AUX sample decoding: PASS"

-- 
2.34.1


  parent reply	other threads:[~2026-09-15 15:50 UTC|newest]

Thread overview: 26+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-09-15 15:48 [PATCH 00/25] CoreSight / perf: Support AUX sampling Leo Yan
2026-09-15 15:48 ` [PATCH 01/25] perf/core: Reject non-positive AUX snapshot sizes Leo Yan
2026-09-15 15:48 ` [PATCH 02/25] perf/core: Prevent AUX sampling from racing with pause/resume Leo Yan
2026-09-15 15:48 ` [PATCH 03/25] perf/core: Export output copy helpers for CoreSight Leo Yan
2026-09-15 15:48 ` [PATCH 04/25] coresight: perf: End AUX output when an event is throttled Leo Yan
2026-09-15 15:48 ` [PATCH 05/25] coresight: perf: Extract AUX buffer update helper Leo Yan
2026-09-15 15:48 ` [PATCH 06/25] coresight: perf: Simplify flow for CPUs without a path Leo Yan
2026-09-15 15:48 ` [PATCH 07/25] coresight: perf: Validate the live context through its path Leo Yan
2026-09-15 15:48 ` [PATCH 08/25] coresight: perf: Validate context before resuming trace Leo Yan
2026-09-15 15:48 ` [PATCH 09/25] coresight: perf: Improve hardware state transitions Leo Yan
2026-09-15 15:48 ` [PATCH 10/25] coresight: trbe: Pad snapshot buffers Leo Yan
2026-09-15 15:48 ` [PATCH 11/25] coresight: trbe: Use circular buffer mode for snapshots Leo Yan
2026-09-15 15:48 ` [PATCH 12/25] coresight: trbe: Track per-CPU sink interrupt handling Leo Yan
2026-09-15 15:48 ` [PATCH 13/25] coresight: trbe: Handle an already disabled sink Leo Yan
2026-09-15 15:49 ` [PATCH 14/25] coresight: perf: Look up the trace path in etm_event_pause() Leo Yan
2026-09-15 15:49 ` [PATCH 15/25] coresight: perf: Serialize AUX pause and resume with event stops Leo Yan
2026-09-15 15:49 ` [PATCH 16/25] coresight: perf: Support AUX sampling with per-CPU sinks Leo Yan
2026-09-15 15:49 ` [PATCH 17/25] perf cs-etm: Allow history collection for samples with zero IP Leo Yan
2026-09-15 15:49 ` [PATCH 18/25] perf cs-etm: Unify timeless buffer decoding Leo Yan
2026-09-15 15:49 ` [PATCH 19/25] perf cs-etm: Propagate errors from trace queue flushing Leo Yan
2026-09-15 15:49 ` [PATCH 20/25] perf cs-etm: Drain pending packets before finishing trace blocks Leo Yan
2026-09-15 15:49 ` [PATCH 21/25] perf cs-etm: Complete packet draining with end of trace Leo Yan
2026-09-15 15:49 ` [PATCH 22/25] perf cs-etm: Centralize sample history collection Leo Yan
2026-09-15 15:49 ` [PATCH 23/25] perf cs-etm: Decode AUX samples into callchains and branch stacks Leo Yan
2026-09-15 15:49 ` Leo Yan [this message]
2026-09-15 15:49 ` [PATCH 25/25] Documentation: coresight: Document AUX sample decoding Leo Yan

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=20260915-arm_cs_support_aux_sample-v1-24-35cf6f297d20@arm.com \
    --to=leo.yan@arm.com \
    --cc=aaupov@fb.com \
    --cc=acme@kernel.org \
    --cc=adrian.hunter@intel.com \
    --cc=anshuman.khandual@arm.com \
    --cc=corbet@lwn.net \
    --cc=coresight@lists.linaro.org \
    --cc=irogers@google.com \
    --cc=james.clark@linaro.org \
    --cc=linux-arm-kernel@lists.infradead.org \
    --cc=linux-doc@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-perf-users@vger.kernel.org \
    --cc=mike.leach@arm.com \
    --cc=mingo@kernel.org \
    --cc=mingo@redhat.com \
    --cc=namhyung@kernel.org \
    --cc=peterz@infradead.org \
    --cc=smahar@meta.com \
    --cc=suzuki.poulose@arm.com \
    --cc=yeoreum.yun@arm.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

all inboxes | Powered by JetHome®