Merge tag 'perf-core-for-mingo' of git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux into perf/core

Pull perf/core improvements and fixes from Arnaldo Carvalho de Melo:

 * Don't install scripting files files when perl/python support is disabled.

 * Support ! in -e expressions in 'perf trace', to filter a list of syscalls.

 * Add --verbose and -o/--output options to 'perf trace'.

 * Introduce better formatting of syscall arguments in 'perf trace',
   including so far beautifiers for mmap, madvise, syscall return
   values.

 * Fixup jobserver setup in libtraceevent makefile.

 * Debug improvements from Adrian Hunter.

 * Try to increase the file descriptor limits on EMFILE, from Andi Kleen.

 * Remove unused force option in 'perf kvm', from David Ahern.

 * Make 'perf trace' command line arguments consistent with 'perf record',
   from David Ahern.

 * Fix correlation of samples coming after PERF_RECORD_EXIT event, from
   David Ahern.

Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: Ingo Molnar <mingo@kernel.org>

1 files changed, 93 insertions, 0 deletions

diff --git a/tools/perf/util/evsel.c b/tools/perf/util/evsel.c
index 960394e..47cbe1e 100644
--- a/tools/perf/util/evsel.c
+++ b/tools/perf/util/evsel.c
@@ -13,6 +13,7 @@
 #include <traceevent/event-parse.h>
 #include <linux/hw_breakpoint.h>
 #include <linux/perf_event.h>
+#include <sys/resource.h>
 #include "asm/bug.h"
 #include "evsel.h"
 #include "evlist.h"
@@ -21,6 +22,7 @@
 #include "thread_map.h"
 #include "target.h"
 #include "perf_regs.h"
+#include "debug.h"
 
 static struct {
 	bool sample_id_all;
@@ -861,12 +863,72 @@ static int get_group_fd(struct perf_evsel *evsel, int cpu, int thread)
 	return fd;
 }
 
+#define __PRINT_ATTR(fmt, cast, field)  \
+	fprintf(fp, "  %-19s "fmt"\n", #field, cast attr->field)
+
+#define PRINT_ATTR_U32(field)  __PRINT_ATTR("%u" , , field)
+#define PRINT_ATTR_X32(field)  __PRINT_ATTR("%#x", , field)
+#define PRINT_ATTR_U64(field)  __PRINT_ATTR("%" PRIu64, (uint64_t), field)
+#define PRINT_ATTR_X64(field)  __PRINT_ATTR("%#"PRIx64, (uint64_t), field)
+
+#define PRINT_ATTR2N(name1, field1, name2, field2)	\
+	fprintf(fp, "  %-19s %u    %-19s %u\n",		\
+	name1, attr->field1, name2, attr->field2)
+
+#define PRINT_ATTR2(field1, field2) \
+	PRINT_ATTR2N(#field1, field1, #field2, field2)
+
+static size_t perf_event_attr__fprintf(struct perf_event_attr *attr, FILE *fp)
+{
+	size_t ret = 0;
+
+	ret += fprintf(fp, "%.60s\n", graph_dotted_line);
+	ret += fprintf(fp, "perf_event_attr:\n");
+
+	ret += PRINT_ATTR_U32(type);
+	ret += PRINT_ATTR_U32(size);
+	ret += PRINT_ATTR_X64(config);
+	ret += PRINT_ATTR_U64(sample_period);
+	ret += PRINT_ATTR_U64(sample_freq);
+	ret += PRINT_ATTR_X64(sample_type);
+	ret += PRINT_ATTR_X64(read_format);
+
+	ret += PRINT_ATTR2(disabled, inherit);
+	ret += PRINT_ATTR2(pinned, exclusive);
+	ret += PRINT_ATTR2(exclude_user, exclude_kernel);
+	ret += PRINT_ATTR2(exclude_hv, exclude_idle);
+	ret += PRINT_ATTR2(mmap, comm);
+	ret += PRINT_ATTR2(freq, inherit_stat);
+	ret += PRINT_ATTR2(enable_on_exec, task);
+	ret += PRINT_ATTR2(watermark, precise_ip);
+	ret += PRINT_ATTR2(mmap_data, sample_id_all);
+	ret += PRINT_ATTR2(exclude_host, exclude_guest);
+	ret += PRINT_ATTR2N("excl.callchain_kern", exclude_callchain_kernel,
+			    "excl.callchain_user", exclude_callchain_user);
+
+	ret += PRINT_ATTR_U32(wakeup_events);
+	ret += PRINT_ATTR_U32(wakeup_watermark);
+	ret += PRINT_ATTR_X32(bp_type);
+	ret += PRINT_ATTR_X64(bp_addr);
+	ret += PRINT_ATTR_X64(config1);
+	ret += PRINT_ATTR_U64(bp_len);
+	ret += PRINT_ATTR_X64(config2);
+	ret += PRINT_ATTR_X64(branch_sample_type);
+	ret += PRINT_ATTR_X64(sample_regs_user);
+	ret += PRINT_ATTR_U32(sample_stack_user);
+
+	ret += fprintf(fp, "%.60s\n", graph_dotted_line);
+
+	return ret;
+}
+
 static int __perf_evsel__open(struct perf_evsel *evsel, struct cpu_map *cpus,
 			      struct thread_map *threads)
 {
 	int cpu, thread;
 	unsigned long flags = 0;
 	int pid = -1, err;
+	enum { NO_CHANGE, SET_TO_MAX, INCREASED_MAX } set_rlimit = NO_CHANGE;
 
 	if (evsel->fd == NULL &&
 	    perf_evsel__alloc_fd(evsel, cpus->nr, threads->nr) < 0)
@@ -884,6 +946,9 @@ retry_sample_id:
 	if (perf_missing_features.sample_id_all)
 		evsel->attr.sample_id_all = 0;
 
+	if (verbose >= 2)
+		perf_event_attr__fprintf(&evsel->attr, stderr);
+
 	for (cpu = 0; cpu < cpus->nr; cpu++) {
 
 		for (thread = 0; thread < threads->nr; thread++) {
@@ -893,6 +958,9 @@ retry_sample_id:
 				pid = threads->map[thread];
 
 			group_fd = get_group_fd(evsel, cpu, thread);
+retry_open:
+			pr_debug2("perf_event_open: pid %d  cpu %d  group_fd %d  flags %#lx\n",
+				  pid, cpus->map[cpu], group_fd, flags);
 
 			FD(evsel, cpu, thread) = sys_perf_event_open(&evsel->attr,
 								     pid,
@@ -902,12 +970,37 @@ retry_sample_id:
 				err = -errno;
 				goto try_fallback;
 			}
+			set_rlimit = NO_CHANGE;
 		}
 	}
 
 	return 0;
 
 try_fallback:
+	/*
+	 * perf stat needs between 5 and 22 fds per CPU. When we run out
+	 * of them try to increase the limits.
+	 */
+	if (err == -EMFILE && set_rlimit < INCREASED_MAX) {
+		struct rlimit l;
+		int old_errno = errno;
+
+		if (getrlimit(RLIMIT_NOFILE, &l) == 0) {
+			if (set_rlimit == NO_CHANGE)
+				l.rlim_cur = l.rlim_max;
+			else {
+				l.rlim_cur = l.rlim_max + 1000;
+				l.rlim_max = l.rlim_cur;
+			}
+			if (setrlimit(RLIMIT_NOFILE, &l) == 0) {
+				set_rlimit++;
+				errno = old_errno;
+				goto retry_open;
+			}
+		}
+		errno = old_errno;
+	}
+
 	if (err != -EINVAL || cpu > 0 || thread > 0)
 		goto out_close;