| 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
| ||
|
Message-Id: <20230428073809.1803624-3-irogers@google.com>
Date: Fri, 28 Apr 2023 00:37:28 -0700
From: Ian Rogers <irogers@...gle.com>
To: Arnaldo Carvalho de Melo <acme@...nel.org>,
Kan Liang <kan.liang@...ux.intel.com>,
Ahmad Yasin <ahmad.yasin@...el.com>,
Peter Zijlstra <peterz@...radead.org>,
Ingo Molnar <mingo@...hat.com>,
Stephane Eranian <eranian@...gle.com>,
Andi Kleen <ak@...ux.intel.com>,
Perry Taylor <perry.taylor@...el.com>,
Samantha Alt <samantha.alt@...el.com>,
Caleb Biggers <caleb.biggers@...el.com>,
Weilin Wang <weilin.wang@...el.com>,
Edward Baker <edward.baker@...el.com>,
Mark Rutland <mark.rutland@....com>,
Alexander Shishkin <alexander.shishkin@...ux.intel.com>,
Jiri Olsa <jolsa@...nel.org>,
Namhyung Kim <namhyung@...nel.org>,
Adrian Hunter <adrian.hunter@...el.com>,
Florian Fischer <florian.fischer@...q.space>,
Rob Herring <robh@...nel.org>,
Zhengjun Xing <zhengjun.xing@...ux.intel.com>,
John Garry <john.g.garry@...cle.com>,
Kajol Jain <kjain@...ux.ibm.com>,
Sumanth Korikkar <sumanthk@...ux.ibm.com>,
Thomas Richter <tmricht@...ux.ibm.com>,
Tiezhu Yang <yangtiezhu@...ngson.cn>,
Ravi Bangoria <ravi.bangoria@....com>,
Leo Yan <leo.yan@...aro.org>,
Yang Jihong <yangjihong1@...wei.com>,
James Clark <james.clark@....com>,
Suzuki Poulouse <suzuki.poulose@....com>,
Kang Minchul <tegongkang@...il.com>,
Athira Rajeev <atrajeev@...ux.vnet.ibm.com>,
linux-perf-users@...r.kernel.org, linux-kernel@...r.kernel.org
Cc: Ian Rogers <irogers@...gle.com>
Subject: [PATCH v2 02/43] perf stat: Introduce skippable evsels
Perf stat with no arguments will use default events and metrics. These
events may fail to open even with kernel and hypervisor disabled. When
these fail then the permissions error appears even though they were
implicitly selected. This is particularly a problem with the automatic
selection of the TopdownL1 metric group on certain architectures like
Skylake:
'''
$ perf stat true
Error:
Access to performance monitoring and observability operations is limited.
Consider adjusting /proc/sys/kernel/perf_event_paranoid setting to open
access to performance monitoring and observability operations for processes
without CAP_PERFMON, CAP_SYS_PTRACE or CAP_SYS_ADMIN Linux capability.
More information can be found at 'Perf events and tool security' document:
https://www.kernel.org/doc/html/latest/admin-guide/perf-security.html
perf_event_paranoid setting is 2:
-1: Allow use of (almost) all events by all users
Ignore mlock limit after perf_event_mlock_kb without CAP_IPC_LOCK
>= 0: Disallow raw and ftrace function tracepoint access
>= 1: Disallow CPU event access
>= 2: Disallow kernel profiling
To make the adjusted perf_event_paranoid setting permanent preserve it
in /etc/sysctl.conf (e.g. kernel.perf_event_paranoid = <setting>)
'''
This patch adds skippable evsels that when they fail to open won't
fail and won't appear in output. The TopdownL1 events, from the metric
group, are marked as skippable. This turns the failure above to:
'''
$ perf stat true
Performance counter stats for 'true':
1.26 msec task-clock:u # 0.389 CPUs utilized
0 context-switches:u # 0.000 /sec
0 cpu-migrations:u # 0.000 /sec
48 page-faults:u # 38.068 K/sec
264,719 cycles:u # 0.210 GHz (78.34%)
122,746 instructions:u # 0.46 insn per cycle
28,219 branches:u # 22.380 M/sec
2,526 branch-misses:u # 8.95% of all branches
4,712 CPU_CLK_UNHALTED.REF_XCLK:u # 3.737 M/sec
350,040 IDQ_UOPS_NOT_DELIVERED.CORE:u # 277.610 M/sec
<not counted> CPU_CLK_UNHALTED.ONE_THREAD_ACTIVE:u (0.00%)
<not counted> CPU_CLK_UNHALTED.THREAD:u (0.00%)
<not counted> UOPS_RETIRED.RETIRE_SLOTS:u (0.00%)
<not counted> UOPS_ISSUED.ANY:u (0.00%)
0.003238907 seconds time elapsed
0.000000000 seconds user
0.003412000 seconds sys
'''
Some events aren't counted leading to <not counted>/0 counts which
then lead to divides by zero in the metric and no display - running a
longer running program corrects that and is shown next. The event
INT_MISC.RECOVERY_CYCLES_ANY:u is skipped as it can't be opened with
paranoia 2 on Skylake.
'''
$ perf stat perf bench internals synthesize
Computing performance of single threaded perf event synthesis by
synthesizing events on the perf process itself:
Average synthesis took: 49.840 usec (+- 0.084 usec)
Average num. events: 3.000 (+- 0.000)
Average time per event 16.613 usec
Average data synthesis took: 50.131 usec (+- 0.087 usec)
Average num. events: 11.000 (+- 0.000)
Average time per event 4.557 usec
Performance counter stats for '/tmp/perf/perf bench internals synthesize':
1,232.99 msec task-clock:u # 0.993 CPUs utilized
0 context-switches:u # 0.000 /sec
0 cpu-migrations:u # 0.000 /sec
162 page-faults:u # 131.388 /sec
786,004,400 cycles:u # 0.637 GHz (50.05%)
1,640,825,583 instructions:u # 2.09 insn per cycle (60.01%)
301,257,667 branches:u # 244.330 M/sec (60.02%)
1,808,506 branch-misses:u # 0.60% of all branches (59.99%)
5,321,529 CPU_CLK_UNHALTED.REF_XCLK:u # 4.316 M/sec
# 16.5 % tma_frontend_bound
# 55.0 % tma_retiring
# 22.5 % tma_backend_bound
# 6.0 % tma_bad_speculation (59.99%)
523,301,560 IDQ_UOPS_NOT_DELIVERED.CORE:u # 424.416 M/sec (60.05%)
5,373,764 CPU_CLK_UNHALTED.ONE_THREAD_ACTIVE:u # 4.358 M/sec (39.98%)
789,470,378 CPU_CLK_UNHALTED.THREAD:u # 640.288 M/sec (49.93%)
1,744,823,827 UOPS_RETIRED.RETIRE_SLOTS:u # 1.415 G/sec (50.06%)
1,934,825,111 UOPS_ISSUED.ANY:u # 1.569 G/sec (49.91%)
1.242166915 seconds time elapsed
0.274824000 seconds user
0.959892000 seconds sys
'''
And this likewise works if paranoia allows or running as root.
Signed-off-by: Ian Rogers <irogers@...gle.com>
---
tools/perf/builtin-stat.c | 38 ++++++++++++++++++++++++++--------
tools/perf/util/evsel.c | 15 ++++++++++++--
tools/perf/util/evsel.h | 1 +
tools/perf/util/stat-display.c | 11 +++++++++-
4 files changed, 53 insertions(+), 12 deletions(-)
diff --git a/tools/perf/builtin-stat.c b/tools/perf/builtin-stat.c
index be9677aa642f..ffb47b166098 100644
--- a/tools/perf/builtin-stat.c
+++ b/tools/perf/builtin-stat.c
@@ -667,6 +667,13 @@ static enum counter_recovery stat_handle_error(struct evsel *counter)
evsel_list->core.threads->err_thread = -1;
return COUNTER_RETRY;
}
+ } else if (counter->skippable) {
+ if (verbose > 0)
+ ui__warning("skipping event %s that kernel failed to open .\n",
+ evsel__name(counter));
+ counter->supported = false;
+ counter->errored = true;
+ return COUNTER_SKIP;
}
evsel__open_strerror(counter, &target, errno, msg, sizeof(msg));
@@ -1890,15 +1897,28 @@ static int add_default_attributes(void)
* caused by exposing latent bugs. This is fixed properly in:
* https://lore.kernel.org/lkml/bff481ba-e60a-763f-0aa0-3ee53302c480@linux.intel.com/
*/
- if (metricgroup__has_metric("TopdownL1") && !perf_pmu__has_hybrid() &&
- metricgroup__parse_groups(evsel_list, "TopdownL1",
- /*metric_no_group=*/false,
- /*metric_no_merge=*/false,
- /*metric_no_threshold=*/true,
- stat_config.user_requested_cpu_list,
- stat_config.system_wide,
- &stat_config.metric_events) < 0)
- return -1;
+ if (metricgroup__has_metric("TopdownL1") && !perf_pmu__has_hybrid()) {
+ struct evlist *metric_evlist = evlist__new();
+ struct evsel *metric_evsel;
+
+ if (!metric_evlist)
+ return -1;
+
+ if (metricgroup__parse_groups(metric_evlist, "TopdownL1",
+ /*metric_no_group=*/false,
+ /*metric_no_merge=*/false,
+ /*metric_no_threshold=*/true,
+ stat_config.user_requested_cpu_list,
+ stat_config.system_wide,
+ &stat_config.metric_events) < 0)
+ return -1;
+
+ evlist__for_each_entry(metric_evlist, metric_evsel) {
+ metric_evsel->skippable = true;
+ }
+ evlist__splice_list_tail(evsel_list, &metric_evlist->core.entries);
+ evlist__delete(metric_evlist);
+ }
/* Platform specific attrs */
if (evlist__add_default_attrs(evsel_list, default_null_attrs) < 0)
diff --git a/tools/perf/util/evsel.c b/tools/perf/util/evsel.c
index 356c07f03be6..1cd04b5998d2 100644
--- a/tools/perf/util/evsel.c
+++ b/tools/perf/util/evsel.c
@@ -290,6 +290,7 @@ void evsel__init(struct evsel *evsel,
evsel->per_pkg_mask = NULL;
evsel->collect_stat = false;
evsel->pmu_name = NULL;
+ evsel->skippable = false;
}
struct evsel *evsel__new_idx(struct perf_event_attr *attr, int idx)
@@ -1725,9 +1726,13 @@ static int get_group_fd(struct evsel *evsel, int cpu_map_idx, int thread)
return -1;
fd = FD(leader, cpu_map_idx, thread);
- BUG_ON(fd == -1);
+ BUG_ON(fd == -1 && !leader->skippable);
- return fd;
+ /*
+ * When the leader has been skipped, return -2 to distinguish from no
+ * group leader case.
+ */
+ return fd == -1 ? -2 : fd;
}
static void evsel__remove_fd(struct evsel *pos, int nr_cpus, int nr_threads, int thread_idx)
@@ -2109,6 +2114,12 @@ static int evsel__open_cpu(struct evsel *evsel, struct perf_cpu_map *cpus,
group_fd = get_group_fd(evsel, idx, thread);
+ if (group_fd == -2) {
+ pr_debug("broken group leader for %s\n", evsel->name);
+ err = -EINVAL;
+ goto out_close;
+ }
+
test_attr__ready();
/* Debug message used by test scripts */
diff --git a/tools/perf/util/evsel.h b/tools/perf/util/evsel.h
index 35805dcdb1b9..bf8f01af1c0b 100644
--- a/tools/perf/util/evsel.h
+++ b/tools/perf/util/evsel.h
@@ -95,6 +95,7 @@ struct evsel {
bool weak_group;
bool bpf_counter;
bool use_config_name;
+ bool skippable;
int bpf_fd;
struct bpf_object *bpf_obj;
struct list_head config_terms;
diff --git a/tools/perf/util/stat-display.c b/tools/perf/util/stat-display.c
index e6035ecbeee8..df6337f2480b 100644
--- a/tools/perf/util/stat-display.c
+++ b/tools/perf/util/stat-display.c
@@ -7,6 +7,7 @@
#include <perf/cpumap.h>
#include "color.h"
#include "counts.h"
+#include "debug.h"
#include "evlist.h"
#include "evsel.h"
#include "stat.h"
@@ -791,6 +792,7 @@ static void uniquify_counter(struct perf_stat_config *config, struct evsel *coun
* should_skip_zero_count() - Check if the event should print 0 values.
* @config: The perf stat configuration (including aggregation mode).
* @counter: The evsel with its associated cpumap.
+ * @ena: The enabled time for the counter.
* @id: The aggregation id that is being queried.
*
* Due to mismatch between the event cpumap or thread-map and the
@@ -805,11 +807,18 @@ static void uniquify_counter(struct perf_stat_config *config, struct evsel *coun
*/
static bool should_skip_zero_counter(struct perf_stat_config *config,
struct evsel *counter,
+ u64 ena,
const struct aggr_cpu_id *id)
{
struct perf_cpu cpu;
int idx;
+ /* Skip counters that were speculatively/default enabled rather than requested. */
+ if (ena == 0 && counter->skippable) {
+ pr_debug("Skipping counter '%s'\n", evsel__name(counter));
+ return true;
+ }
+
/*
* Skip value 0 when enabling --per-thread globally,
* otherwise it will have too many 0 output.
@@ -859,7 +868,7 @@ static void print_counter_aggrdata(struct perf_stat_config *config,
ena = aggr->counts.ena;
run = aggr->counts.run;
- if (val == 0 && should_skip_zero_counter(config, counter, &id))
+ if (val == 0 && should_skip_zero_counter(config, counter, ena, &id))
return;
if (!metric_only) {
--
2.40.1.495.gc816e09b53d-goog
Powered by blists - more mailing lists