linux-kernel - Re: [PATCH v5 4/4] perf tools: add JVMTI agent library

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

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <20150327140117.GF23123@twins.programming.kicks-ass.net>
Date:	Fri, 27 Mar 2015 15:01:17 +0100
From:	Peter Zijlstra <peterz@...radead.org>
To:	Stephane Eranian <eranian@...gle.com>
Cc:	linux-kernel@...r.kernel.org, acme@...hat.com, mingo@...e.hu,
	ak@...ux.intel.com, jolsa@...hat.com, namhyung@...nel.org,
	cel@...ibm.com, sukadev@...ux.vnet.ibm.com, sonnyrao@...omium.org,
	johnmccutchan@...gle.com, dsahern@...il.com,
	adrian.hunter@...el.com, pawel.moll@....com
Subject: Re: [PATCH v5 4/4] perf tools: add JVMTI agent library

On Thu, Mar 26, 2015 at 08:17:09PM +0100, Stephane Eranian wrote:

> This version of the JVMTI agent is using Pawel Moll's CLOCK_MONOTONIC
> perf clock patch posted here:
> 	https://lkml.org/lkml/2015/1/21/711


Seeing how we just merged its replacement:

  http://git.kernel.org/tip/34f439278cef7b1177f8ce24f9fc81dfc6221d3b

how about something like this to make it work?

Completely not been near a compiler at all. I'll go do the regular perf
userspace part now I suppose -- I had somewhat forgotten about that.

---
--- a/tools/perf/jvmti/jvmti_agent.c
+++ b/tools/perf/jvmti/jvmti_agent.c
@@ -36,6 +36,8 @@
 #include <syscall.h> /* for gettid() */
 #include <err.h>
 
+#include "perf-sys.h"
+
 #include "jvmti_agent.h"
 #include "../util/jitdump.h"
 
@@ -92,37 +94,44 @@ static int get_e_machine(struct jitheade
 	return ret;
 }
 
-#define NSEC_PER_SEC	1000000000
-static int perf_clk_id;
-static int
-perf_get_clock_id(void)
-{
-	FILE *fp;
-	int ret;
-
-	fp = fopen("/proc/sys/kernel/perf_sample_time_clk_id", "r");
-	if (!fp)
-		return -1;
-	ret = fscanf(fp, "%d",  &perf_clk_id);
+static clockid_t jit_clockid = CLOCK_MONOTONIC_RAW;
 
-	fclose(fp);
-
-	return ret == 1 ? 0 : -1;
-}
+#define NSEC_PER_SEC	1000000000
 
 static int
 perf_open_timestamp(void)
 {
+	struct perf_event_attr perf_attr = {
+		.type = PERF_TYPE_SOFTWARE,
+		.config = PERF_COUNT_SW_DUMMY,
+		.exclude_kernel = 1,
+		.use_clockid = 1,
+	};
 	struct timespec ts;
+	char *clk;
 	int ret;
 
-	if (perf_get_clock_id())
-		return -1;
+	clk = getenv("JITCLOCKID");
+	if (clk) {
+		ret = sscanf(clk, "%d", &jit_clockid);
+		if (ret != 1)
+			return -1;
+	}
+
 	/*
-	 * check if clock_id is actually available
+	 * check if jit_clockid is actually available
 	 */
-	ret = clock_gettime(perf_clk_id, &ts);
-	return ret ? -1 : 0;
+	ret = clock_gettime(jit_clockid, &ts);
+	if (ret)
+		return ret;
+
+	perf_attr.clockid = jit_clockid;
+	ret = sys_perf_event_open(&perf_attr, 0, -1, -1, 0);
+	if (ret < 0)
+		return ret;
+
+	close(ret);
+	return 0;
 }
 
 static inline uint64_t
@@ -137,7 +146,7 @@ perf_get_timestamp(void)
 	struct timespec ts;
 	int ret;
 
-	ret = clock_gettime(perf_clk_id, &ts);
+	ret = clock_gettime(jit_clockid, &ts);
 	if (ret)
 		return 0;
 
@@ -242,7 +251,7 @@ void *jvmti_open(void)
 	FILE *fp;
 
 	if (perf_open_timestamp())
-		warnx("jvmti: kernel does not support %d clock id", perf_clk_id);
+		warnx("jvmti: kernel does not support %d clock id", jit_clockid);
 
 	memset(&header, 0, sizeof(header));
 
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/