linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Arnaldo Carvalho de Melo <acme@kernel.org>
To: Ingo Molnar <mingo@kernel.org>
Cc: Clark Williams <williams@redhat.com>,
	linux-kernel@vger.kernel.org, linux-perf-users@vger.kernel.org,
	Leo Yan <leo.yan@linaro.org>,
	Alexander Shishkin <alexander.shishkin@linux.intel.com>,
	Jiri Olsa <jolsa@redhat.com>, Mike Leach <mike.leach@linaro.org>,
	Namhyung Kim <namhyung@kernel.org>,
	Robert Walker <robert.walker@arm.com>,
	coresight@lists.linaro.org, linux-arm-kernel@lists.infradead.org,
	Arnaldo Carvalho de Melo <acme@redhat.com>
Subject: [PATCH 27/63] perf cs-etm: Avoid stale branch samples when flush packet
Date: Tue, 18 Dec 2018 19:06:57 -0300	[thread overview]
Message-ID: <20181218220733.15839-28-acme@kernel.org> (raw)
In-Reply-To: <20181218220733.15839-1-acme@kernel.org>

From: Leo Yan <leo.yan@linaro.org>

At the end of trace buffer handling, function cs_etm__flush() is invoked
to flush any remaining branch stack entries.  As a side effect, it also
generates branch sample, because the 'etmq->packet' doesn't contains any
new coming packet but point to one stale packet after packets swapping,
so it wrongly makes synthesize branch samples with stale packet info.

We could review below detailed flow which causes issue:

  Packet1: start_addr=0xffff000008b1fbf0 end_addr=0xffff000008b1fbfc
  Packet2: start_addr=0xffff000008b1fb5c end_addr=0xffff000008b1fb6c

  step 1: cs_etm__sample():
	sample: ip=(0xffff000008b1fbfc-4) addr=0xffff000008b1fb5c

  step 2: flush packet in cs_etm__run_decoder():
	cs_etm__run_decoder()
	  `-> err = cs_etm__flush(etmq, false);
	sample: ip=(0xffff000008b1fb6c-4) addr=0xffff000008b1fbf0

Packet1 and packet2 are two continuous packets, when packet2 is the new
coming packet, cs_etm__sample() generates branch sample for these two
packets and use [packet1::end_addr - 4 => packet2::start_addr] as branch
jump flow, thus we can see the first generated branch sample in step 1.
At the end of cs_etm__sample() it swaps packets so 'etm->prev_packet'=
packet2 and 'etm->packet'=packet1, so far it's okay for branch sample.

If packet2 is the last one packet in trace buffer, even there have no
any new coming packet, cs_etm__run_decoder() invokes cs_etm__flush() to
flush branch stack entries as expected, but it also generates branch
samples by taking 'etm->packet' as a new coming packet, thus the branch
jump flow is as [packet2::end_addr - 4 =>  packet1::start_addr]; this
is the second sample which is generated in step 2.  So actually the
second sample is a stale sample and we should not generate it.

This patch introduces a new function cs_etm__end_block(), at the end of
trace block this function is invoked to only flush branch stack entries
and thus can avoid to generate branch sample for stale packet.

Signed-off-by: Leo Yan <leo.yan@linaro.org>
Reviewed-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Mike Leach <mike.leach@linaro.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Robert Walker <robert.walker@arm.com>
Cc: coresight@lists.linaro.org
Cc: linux-arm-kernel@lists.infradead.org
Link: http://lkml.kernel.org/r/1544513908-16805-3-git-send-email-leo.yan@linaro.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
---
 tools/perf/util/cs-etm.c | 35 ++++++++++++++++++++++++++++++++++-
 1 file changed, 34 insertions(+), 1 deletion(-)

diff --git a/tools/perf/util/cs-etm.c b/tools/perf/util/cs-etm.c
index 789707bc4e9e..ffc4fe5c0b7e 100644
--- a/tools/perf/util/cs-etm.c
+++ b/tools/perf/util/cs-etm.c
@@ -1055,6 +1055,39 @@ static int cs_etm__flush(struct cs_etm_queue *etmq)
 	return err;
 }
 
+static int cs_etm__end_block(struct cs_etm_queue *etmq)
+{
+	int err;
+
+	/*
+	 * It has no new packet coming and 'etmq->packet' contains the stale
+	 * packet which was set at the previous time with packets swapping;
+	 * so skip to generate branch sample to avoid stale packet.
+	 *
+	 * For this case only flush branch stack and generate a last branch
+	 * event for the branches left in the circular buffer at the end of
+	 * the trace.
+	 */
+	if (etmq->etm->synth_opts.last_branch &&
+	    etmq->prev_packet->sample_type == CS_ETM_RANGE) {
+		/*
+		 * Use the address of the end of the last reported execution
+		 * range.
+		 */
+		u64 addr = cs_etm__last_executed_instr(etmq->prev_packet);
+
+		err = cs_etm__synth_instruction_sample(
+			etmq, addr,
+			etmq->period_instructions);
+		if (err)
+			return err;
+
+		etmq->period_instructions = 0;
+	}
+
+	return 0;
+}
+
 static int cs_etm__run_decoder(struct cs_etm_queue *etmq)
 {
 	struct cs_etm_auxtrace *etm = etmq->etm;
@@ -1137,7 +1170,7 @@ static int cs_etm__run_decoder(struct cs_etm_queue *etmq)
 
 		if (err == 0)
 			/* Flush any remaining branch stack entries */
-			err = cs_etm__flush(etmq);
+			err = cs_etm__end_block(etmq);
 	}
 
 	return err;
-- 
2.19.2


  parent reply	other threads:[~2018-12-18 22:10 UTC|newest]

Thread overview: 69+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2018-12-18 22:06 [GIT PULL 00/63] perf/core improvements and fixes Arnaldo Carvalho de Melo
2018-12-18 22:06 ` [PATCH 01/63] perf dso: Export data_file_size() method there are no symbols Arnaldo Carvalho de Melo
2018-12-18 22:06 ` [PATCH 02/63] perf auxtrace: Alter addr_filter__entire_dso() to work if " Arnaldo Carvalho de Melo
2018-12-18 22:06 ` [PATCH 03/63] perf tests: Use shebangs in the shell scripts Arnaldo Carvalho de Melo
2018-12-18 22:06 ` [PATCH 04/63] perf stat: Avoid segfaults caused by negated options Arnaldo Carvalho de Melo
2018-12-18 22:06 ` [PATCH 05/63] tools lib traceevent: Fix processing of dereferenced args in bprintk events Arnaldo Carvalho de Melo
2018-12-18 22:06 ` [PATCH 06/63] perf trace: Rename delivery functions to ease making ordered_events selectable Arnaldo Carvalho de Melo
2018-12-18 22:06 ` [PATCH 07/63] perf trace: Allow selecting use the use of the ordered_events code Arnaldo Carvalho de Melo
2018-12-18 22:06 ` [PATCH 08/63] perf trace beauty: Beautify renameat2's fd arg wrt AT_FDCWD Arnaldo Carvalho de Melo
2018-12-18 22:06 ` [PATCH 09/63] perf beauty: Add a string table generator for renameat2's flags constants Arnaldo Carvalho de Melo
2018-12-18 22:06 ` [PATCH 10/63] perf beauty: Wire up the renameat flags table generator to the Makefile Arnaldo Carvalho de Melo
2018-12-18 22:06 ` [PATCH 11/63] perf trace: Beautify renameat2's flags argument Arnaldo Carvalho de Melo
2018-12-18 22:06 ` [PATCH 12/63] perf trace beauty: renameat's newdirfd may also be AT_FDCWD Arnaldo Carvalho de Melo
2018-12-18 22:06 ` [PATCH 13/63] tools lib subcmd: Don't add the kernel sources to the include path Arnaldo Carvalho de Melo
2018-12-18 22:06 ` [PATCH 14/63] perf tools: Add missing sigqueue() prototype for systems lacking it Arnaldo Carvalho de Melo
2018-12-18 22:06 ` [PATCH 15/63] perf header: Fix up argument to ctime() Arnaldo Carvalho de Melo
2018-12-18 22:06 ` [PATCH 16/63] perf tools: Add missing open_memstream() prototype for systems lacking it Arnaldo Carvalho de Melo
2018-12-18 22:06 ` [PATCH 17/63] perf tools: Cast off_t to s64 to avoid warning on bionic libc Arnaldo Carvalho de Melo
2018-12-18 22:06 ` [PATCH 18/63] tools lib traceevent: Use LDFLAGS in the build commands Arnaldo Carvalho de Melo
2018-12-18 22:06 ` [PATCH 19/63] perf tools: Link libperf-jvmti.so with LDFLAGS variable Arnaldo Carvalho de Melo
2018-12-18 22:06 ` [PATCH 20/63] perf trace: Rename set_ev_qualifier_filter to clarify its a tracepoint filter Arnaldo Carvalho de Melo
2018-12-18 22:06 ` [PATCH 21/63] perf trace: Avoid using raw_syscalls in duplicity with eBPF augmentation Arnaldo Carvalho de Melo
2018-12-18 22:06 ` [PATCH 22/63] perf trace: Implement syscall filtering in augmented_syscalls Arnaldo Carvalho de Melo
2018-12-18 22:06 ` [PATCH 23/63] perf bpf: Move perf_event_output() from stdio.h to bpf.h Arnaldo Carvalho de Melo
2018-12-18 22:06 ` [PATCH 24/63] perf augmented_syscalls: Switch to using a struct for the syscalls map values Arnaldo Carvalho de Melo
2018-12-18 22:06 ` [PATCH 25/63] perf trace: Switch to using a struct for the aumented_raw_syscalls " Arnaldo Carvalho de Melo
2018-12-18 22:06 ` [PATCH 26/63] perf cs-etm: Correct packets swapping in cs_etm__flush() Arnaldo Carvalho de Melo
2018-12-18 22:06 ` Arnaldo Carvalho de Melo [this message]
2018-12-18 22:06 ` [PATCH 28/63] perf cs-etm: Remove unused 'trace_on' in cs_etm_decoder Arnaldo Carvalho de Melo
2018-12-18 22:06 ` [PATCH 29/63] perf cs-etm: Refactor enumeration cs_etm_sample_type Arnaldo Carvalho de Melo
2018-12-18 22:07 ` [PATCH 30/63] perf cs-etm: Rename CS_ETM_TRACE_ON to CS_ETM_DISCONTINUITY Arnaldo Carvalho de Melo
2018-12-18 22:07 ` [PATCH 31/63] perf cs-etm: Treat NO_SYNC element as trace discontinuity Arnaldo Carvalho de Melo
2018-12-18 22:07 ` [PATCH 32/63] perf cs-etm: Treat EO_TRACE " Arnaldo Carvalho de Melo
2018-12-18 22:07 ` [PATCH 33/63] perf cs-etm: Generate branch sample for exception packet Arnaldo Carvalho de Melo
2018-12-18 22:07 ` [PATCH 34/63] perf augmented_raw_syscalls: Do not include stdio.h Arnaldo Carvalho de Melo
2018-12-18 22:07 ` [PATCH 35/63] perf trace: Allow specifying a set of events to add in perfconfig Arnaldo Carvalho de Melo
2018-12-19  8:40   ` Namhyung Kim
2018-12-19 12:50     ` Arnaldo Carvalho de Melo
2018-12-18 22:07 ` [PATCH 36/63] perf trace: Allow configuring if zeroed syscall args should be printed Arnaldo Carvalho de Melo
2018-12-18 22:07 ` [PATCH 37/63] perf trace: Allow configuring if the syscall duration " Arnaldo Carvalho de Melo
2018-12-18 22:07 ` [PATCH 38/63] perf config: Show the configuration when no arguments are provided Arnaldo Carvalho de Melo
2018-12-18 22:07 ` [PATCH 39/63] perf trace: Allow configuring default for perf_event_attr.inherit Arnaldo Carvalho de Melo
2018-12-18 22:07 ` [PATCH 40/63] perf trace: Allow configuring if the syscall start timestamp should be printed Arnaldo Carvalho de Melo
2018-12-18 22:07 ` [PATCH 41/63] perf trace: Allow suppressing the syscall argument names Arnaldo Carvalho de Melo
2018-12-18 22:07 ` [PATCH 42/63] perf trace: Make the alignment of the syscall args be configurable Arnaldo Carvalho de Melo
2018-12-18 22:07 ` [PATCH 43/63] perf trace: Enclose strings with double quotes Arnaldo Carvalho de Melo
2018-12-18 22:07 ` [PATCH 44/63] perf trace: Add a prefix member to the strarray class Arnaldo Carvalho de Melo
2018-12-18 22:07 ` [PATCH 45/63] perf trace: Allow asking for not suppressing common string prefixes Arnaldo Carvalho de Melo
2018-12-18 22:07 ` [PATCH 46/63] perf trace beauty: Print O_RDONLY when (flags & O_ACCMODE) == 0 Arnaldo Carvalho de Melo
2018-12-18 22:07 ` [PATCH 47/63] perf trace: Add alignment spaces after the closing parens Arnaldo Carvalho de Melo
2018-12-18 22:07 ` [PATCH 48/63] perf augmented_raw_syscalls: Copy 'access' arg as well Arnaldo Carvalho de Melo
2018-12-18 22:07 ` [PATCH 49/63] perf trace: Enclose the errno strings with () Arnaldo Carvalho de Melo
2018-12-18 22:07 ` [PATCH 50/63] perf trace: Show NULL when syscall pointer args are 0 Arnaldo Carvalho de Melo
2018-12-18 22:07 ` [PATCH 51/63] tools include arch: Grab a copy of x86's prctl.h Arnaldo Carvalho de Melo
2018-12-18 22:07 ` [PATCH 52/63] perf beauty: Add a string table generator for x86's 'arch_prctl' codes Arnaldo Carvalho de Melo
2018-12-18 22:07 ` [PATCH 53/63] perf beauty: Wire up the x86_arch prctl code table generator Arnaldo Carvalho de Melo
2018-12-18 22:07 ` [PATCH 54/63] perf trace: Move strarrays to beauty.h for further reuse Arnaldo Carvalho de Melo
2018-12-18 22:07 ` [PATCH 55/63] perf trace: When showing string prefixes show prefix + ??? for unknown entries Arnaldo Carvalho de Melo
2018-12-18 22:07 ` [PATCH 56/63] perf trace beauty: Beautify arch_prctl()'s arguments Arnaldo Carvalho de Melo
2018-12-18 22:07 ` [PATCH 57/63] perf beauty mmap: Print PROT_READ before PROT_EXEC to match strace output Arnaldo Carvalho de Melo
2018-12-19  9:15   ` Namhyung Kim
2018-12-19 13:10     ` Arnaldo Carvalho de Melo
2018-12-18 22:07 ` [PATCH 58/63] perf beauty mmap: Print mmap's 'offset' arg in hexadecimal Arnaldo Carvalho de Melo
2018-12-18 22:07 ` [PATCH 59/63] tools headers uapi: Grab a copy of fadvise.h Arnaldo Carvalho de Melo
2018-12-18 22:07 ` [PATCH 60/63] perf beauty: Add generator for fadvise64's 'advice' arg constants Arnaldo Carvalho de Melo
2018-12-18 22:07 ` [PATCH 61/63] perf trace: Wire up the fadvise 'advice' table generator Arnaldo Carvalho de Melo
2018-12-18 22:07 ` [PATCH 62/63] perf symbols: Relax checks on perf-PID.map ownership Arnaldo Carvalho de Melo
2018-12-18 22:07 ` [PATCH 63/63] tools uapi asm: Update asm-generic/unistd.h copy Arnaldo Carvalho de Melo
2018-12-20 17:53 ` [GIT PULL 00/63] perf/core improvements and fixes Ingo Molnar

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=20181218220733.15839-28-acme@kernel.org \
    --to=acme@kernel.org \
    --cc=acme@redhat.com \
    --cc=alexander.shishkin@linux.intel.com \
    --cc=coresight@lists.linaro.org \
    --cc=jolsa@redhat.com \
    --cc=leo.yan@linaro.org \
    --cc=linux-arm-kernel@lists.infradead.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-perf-users@vger.kernel.org \
    --cc=mike.leach@linaro.org \
    --cc=mingo@kernel.org \
    --cc=namhyung@kernel.org \
    --cc=robert.walker@arm.com \
    --cc=williams@redhat.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;
as well as URLs for NNTP newsgroup(s).