perf inject: Add --vm-time-correlation option

Intel PT timestamps are affected by virtualization. Add a new option
that will allow the Intel PT decoder to correlate the timestamps and
translate the virtual machine timestamps to host timestamps.

The advantages of making this a separate step, rather than a part of
normal decoding are that it is simpler to implement, and it needs to
be done only once.

This patch adds only the option. Later patches add Intel PT support.
Signed-off-by: Adrian Hunter <adrian.hunter@intel.com>
Reviewed-by: Andi Kleen <ak@linux.intel.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Link: https://lore.kernel.org/r/20210430070309.17624-6-adrian.hunter@intel.comSigned-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

tools/perf/Documentation/perf-inject.txt

@@ -68,6 +68,16 @@ include::itrace.txt[]
 --force::
 	Don't complain, do it.
 
+--vm-time-correlation[=OPTIONS]::
+	Some architectures may capture AUX area data which contains timestamps
+	affected by virtualization. This option will update those timestamps
+	in place, to correlate with host timestamps. The in-place update means
+	that an output file is not specified, and instead the input file is
+	modified.  The options are architecture specific, except that they may
+	start with "dry-run" which will cause the file to be processed but
+	without updating it. Currently this option is supported only by
+	Intel PT, refer linkperf:perf-intel-pt[1]
+
 SEE ALSO
 --------
 linkperf:perf-record[1], linkperf:perf-report[1], linkperf:perf-archive[1],

tools/perf/builtin-inject.c

@@ -31,6 +31,7 @@
 #include <uapi/linux/mman.h> /* To get things like MAP_HUGETLB even on older libc headers */
 
 #include <linux/list.h>
+#include <linux/string.h>
 #include <errno.h>
 #include <signal.h>
 
@@ -698,6 +699,36 @@ static void strip_init(struct perf_inject *inject)
 		evsel->handler = drop_sample;
 }
 
+static int parse_vm_time_correlation(const struct option *opt, const char *str, int unset)
+{
+	struct perf_inject *inject = opt->value;
+	const char *args;
+	char *dry_run;
+
+	if (unset)
+		return 0;
+
+	inject->itrace_synth_opts.set = true;
+	inject->itrace_synth_opts.vm_time_correlation = true;
+	inject->in_place_update = true;
+
+	if (!str)
+		return 0;
+
+	dry_run = skip_spaces(str);
+	if (!strncmp(dry_run, "dry-run", strlen("dry-run"))) {
+		inject->itrace_synth_opts.vm_tm_corr_dry_run = true;
+		inject->in_place_update_dry_run = true;
+		args = dry_run + strlen("dry-run");
+	} else {
+		args = str;
+	}
+
+	inject->itrace_synth_opts.vm_tm_corr_args = strdup(args);
+
+	return inject->itrace_synth_opts.vm_tm_corr_args ? 0 : -ENOMEM;
+}
+
 static int __cmd_inject(struct perf_inject *inject)
 {
 	int ret = -EINVAL;
@@ -739,6 +770,15 @@ static int __cmd_inject(struct perf_inject *inject)
 			else if (!strncmp(name, "sched:sched_stat_", 17))
 				evsel->handler = perf_inject__sched_stat;
 		}
+	} else if (inject->itrace_synth_opts.vm_time_correlation) {
+		session->itrace_synth_opts = &inject->itrace_synth_opts;
+		memset(&inject->tool, 0, sizeof(inject->tool));
+		inject->tool.id_index	    = perf_event__process_id_index;
+		inject->tool.auxtrace_info  = perf_event__process_auxtrace_info;
+		inject->tool.auxtrace	    = perf_event__process_auxtrace;
+		inject->tool.auxtrace_error = perf_event__process_auxtrace_error;
+		inject->tool.ordered_events = true;
+		inject->tool.ordering_requires_timestamps = true;
 	} else if (inject->itrace_synth_opts.set) {
 		session->itrace_synth_opts = &inject->itrace_synth_opts;
 		inject->itrace_synth_opts.inject = true;
@@ -880,6 +920,9 @@ int cmd_inject(int argc, const char **argv)
 				    itrace_parse_synth_opts),
 		OPT_BOOLEAN(0, "strip", &inject.strip,
 			    "strip non-synthesized events (use with --itrace)"),
+		OPT_CALLBACK_OPTARG(0, "vm-time-correlation", &inject, NULL, "opts",
+				    "correlate time between VM guests and the host",
+				    parse_vm_time_correlation),
 		OPT_END()
 	};
 	const char * const inject_usage[] = {
@@ -968,5 +1011,6 @@ int cmd_inject(int argc, const char **argv)
 out_delete:
 	zstd_fini(&(inject.session->zstd_data));
 	perf_session__delete(inject.session);
+	free(inject.itrace_synth_opts.vm_tm_corr_args);
 	return ret;
 }

tools/perf/util/auxtrace.h

@@ -90,6 +90,9 @@ enum itrace_period_type {
  * @remote_access: whether to synthesize remote access events
  * @mem: whether to synthesize memory events
  * @timeless_decoding: prefer "timeless" decoding i.e. ignore timestamps
+ * @vm_time_correlation: perform VM Time Correlation
+ * @vm_tm_corr_dry_run: VM Time Correlation dry-run
+ * @vm_tm_corr_args:  VM Time Correlation implementation-specific arguments
  * @callchain_sz: maximum callchain size
  * @last_branch_sz: branch context size
  * @period: 'instructions' events period
@@ -130,6 +133,9 @@ struct itrace_synth_opts {
 	bool			remote_access;
 	bool			mem;
 	bool			timeless_decoding;
+	bool			vm_time_correlation;
+	bool			vm_tm_corr_dry_run;
+	char			*vm_tm_corr_args;
 	unsigned int		callchain_sz;
 	unsigned int		last_branch_sz;
 	unsigned long long	period;