All of lore.kernel.org
 help / color / mirror / Atom feed
From: Arnaldo Carvalho de Melo <acme@kernel.org>
To: Ingo Molnar <mingo@kernel.org>
Cc: linux-kernel@vger.kernel.org,
	Alexander Yarygin <yarygin@linux.vnet.ibm.com>,
	Christian Borntraeger <borntraeger@de.ibm.com>,
	Cornelia Huck <cornelia.huck@de.ibm.com>,
	David Ahern <dsahern@gmail.com>, Jiri Olsa <jolsa@redhat.com>,
	Paul Mackerras <paulus@samba.org>,
	Peter Zijlstra <a.p.zijlstra@chello.nl>,
	Arnaldo Carvalho de Melo <acme@redhat.com>
Subject: [PATCH 04/33] perf kvm: Add stat support on s390
Date: Thu, 17 Jul 2014 13:12:06 -0300	[thread overview]
Message-ID: <1405613555-23380-5-git-send-email-acme@kernel.org> (raw)
In-Reply-To: <1405613555-23380-1-git-send-email-acme@kernel.org>

From: Alexander Yarygin <yarygin@linux.vnet.ibm.com>

On s390, the vmexit event has a tree-like structure: between
exit_event_begin and exit_event_end several other events may happen and
with each of them refining the previous ones.

This patch adds a decoder for such events to the generic code and also
the files <asm/kvm_perf.h> and kvm-stat.c for s390.

Commands 'perf kvm stat record', 'report' and 'live' are supported.

Reviewed-by: David Ahern <dsahern@gmail.com>
Signed-off-by: Alexander Yarygin <yarygin@linux.vnet.ibm.com>
Acked-by: Christian Borntraeger <borntraeger@de.ibm.com>
Cc: Christian Borntraeger <borntraeger@de.ibm.com>
Cc: Cornelia Huck <cornelia.huck@de.ibm.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/1404397747-20939-5-git-send-email-yarygin@linux.vnet.ibm.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
---
 arch/s390/include/uapi/asm/Kbuild     |   1 +
 arch/s390/include/uapi/asm/kvm_perf.h |  25 ++++++++
 tools/perf/Documentation/perf-kvm.txt |  16 +++---
 tools/perf/MANIFEST                   |   2 +
 tools/perf/arch/s390/Makefile         |   2 +
 tools/perf/arch/s390/util/kvm-stat.c  | 105 ++++++++++++++++++++++++++++++++++
 tools/perf/builtin-kvm.c              |  52 +++++++++++++++--
 tools/perf/util/kvm-stat.h            |   9 +++
 8 files changed, 201 insertions(+), 11 deletions(-)
 create mode 100644 arch/s390/include/uapi/asm/kvm_perf.h
 create mode 100644 tools/perf/arch/s390/util/kvm-stat.c

diff --git a/arch/s390/include/uapi/asm/Kbuild b/arch/s390/include/uapi/asm/Kbuild
index 6a9a9eb645f5..0e2b54db82bc 100644
--- a/arch/s390/include/uapi/asm/Kbuild
+++ b/arch/s390/include/uapi/asm/Kbuild
@@ -16,6 +16,7 @@ header-y += ioctls.h
 header-y += ipcbuf.h
 header-y += kvm.h
 header-y += kvm_para.h
+header-y += kvm_perf.h
 header-y += kvm_virtio.h
 header-y += mman.h
 header-y += monwriter.h
diff --git a/arch/s390/include/uapi/asm/kvm_perf.h b/arch/s390/include/uapi/asm/kvm_perf.h
new file mode 100644
index 000000000000..397282727e21
--- /dev/null
+++ b/arch/s390/include/uapi/asm/kvm_perf.h
@@ -0,0 +1,25 @@
+/*
+ * Definitions for perf-kvm on s390
+ *
+ * Copyright 2014 IBM Corp.
+ * Author(s): Alexander Yarygin <yarygin@linux.vnet.ibm.com>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License (version 2 only)
+ * as published by the Free Software Foundation.
+ */
+
+#ifndef __LINUX_KVM_PERF_S390_H
+#define __LINUX_KVM_PERF_S390_H
+
+#include <asm/sie.h>
+
+#define DECODE_STR_LEN 40
+
+#define VCPU_ID "id"
+
+#define KVM_ENTRY_TRACE "kvm:kvm_s390_sie_enter"
+#define KVM_EXIT_TRACE "kvm:kvm_s390_sie_exit"
+#define KVM_EXIT_REASON "icptcode"
+
+#endif
diff --git a/tools/perf/Documentation/perf-kvm.txt b/tools/perf/Documentation/perf-kvm.txt
index 52276a6d2b75..6e689dc89a2f 100644
--- a/tools/perf/Documentation/perf-kvm.txt
+++ b/tools/perf/Documentation/perf-kvm.txt
@@ -51,9 +51,9 @@ There are a couple of variants of perf kvm:
   'perf kvm stat <command>' to run a command and gather performance counter
   statistics.
   Especially, perf 'kvm stat record/report' generates a statistical analysis
-  of KVM events. Currently, vmexit, mmio and ioport events are supported.
-  'perf kvm stat record <command>' records kvm events and the events between
-  start and end <command>.
+  of KVM events. Currently, vmexit, mmio (x86 only) and ioport (x86 only)
+  events are supported. 'perf kvm stat record <command>' records kvm events
+  and the events between start and end <command>.
   And this command produces a file which contains tracing results of kvm
   events.
 
@@ -103,8 +103,8 @@ STAT REPORT OPTIONS
        analyze events which occures on this vcpu. (default: all vcpus)
 
 --event=<value>::
-       event to be analyzed. Possible values: vmexit, mmio, ioport.
-       (default: vmexit)
+       event to be analyzed. Possible values: vmexit, mmio (x86 only),
+       ioport (x86 only). (default: vmexit)
 -k::
 --key=<value>::
        Sorting key. Possible values: sample (default, sort by samples
@@ -138,7 +138,8 @@ STAT LIVE OPTIONS
 
 
 --event=<value>::
-       event to be analyzed. Possible values: vmexit, mmio, ioport.
+       event to be analyzed. Possible values: vmexit,
+       mmio (x86 only), ioport (x86 only).
        (default: vmexit)
 
 -k::
@@ -147,7 +148,8 @@ STAT LIVE OPTIONS
        number), time (sort by average time).
 
 --duration=<value>::
-       Show events other than HLT that take longer than duration usecs.
+       Show events other than HLT (x86 only) or Wait state (s390 only)
+       that take longer than duration usecs.
 
 SEE ALSO
 --------
diff --git a/tools/perf/MANIFEST b/tools/perf/MANIFEST
index 02b485d619cd..344c4d3d0a4a 100644
--- a/tools/perf/MANIFEST
+++ b/tools/perf/MANIFEST
@@ -38,3 +38,5 @@ arch/x86/include/uapi/asm/svm.h
 arch/x86/include/uapi/asm/vmx.h
 arch/x86/include/uapi/asm/kvm.h
 arch/x86/include/uapi/asm/kvm_perf.h
+arch/s390/include/uapi/asm/sie.h
+arch/s390/include/uapi/asm/kvm_perf.h
diff --git a/tools/perf/arch/s390/Makefile b/tools/perf/arch/s390/Makefile
index 744e629797be..798ac7379c5f 100644
--- a/tools/perf/arch/s390/Makefile
+++ b/tools/perf/arch/s390/Makefile
@@ -3,3 +3,5 @@ PERF_HAVE_DWARF_REGS := 1
 LIB_OBJS += $(OUTPUT)arch/$(ARCH)/util/dwarf-regs.o
 endif
 LIB_OBJS += $(OUTPUT)arch/$(ARCH)/util/header.o
+HAVE_KVM_STAT_SUPPORT := 1
+LIB_OBJS += $(OUTPUT)arch/$(ARCH)/util/kvm-stat.o
diff --git a/tools/perf/arch/s390/util/kvm-stat.c b/tools/perf/arch/s390/util/kvm-stat.c
new file mode 100644
index 000000000000..a5dbc07ec9dc
--- /dev/null
+++ b/tools/perf/arch/s390/util/kvm-stat.c
@@ -0,0 +1,105 @@
+/*
+ * Arch specific functions for perf kvm stat.
+ *
+ * Copyright 2014 IBM Corp.
+ * Author(s): Alexander Yarygin <yarygin@linux.vnet.ibm.com>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License (version 2 only)
+ * as published by the Free Software Foundation.
+ */
+
+#include "../../util/kvm-stat.h"
+#include <asm/kvm_perf.h>
+
+define_exit_reasons_table(sie_exit_reasons, sie_intercept_code);
+define_exit_reasons_table(sie_icpt_insn_codes, icpt_insn_codes);
+define_exit_reasons_table(sie_sigp_order_codes, sigp_order_codes);
+define_exit_reasons_table(sie_diagnose_codes, diagnose_codes);
+define_exit_reasons_table(sie_icpt_prog_codes, icpt_prog_codes);
+
+static void event_icpt_insn_get_key(struct perf_evsel *evsel,
+				    struct perf_sample *sample,
+				    struct event_key *key)
+{
+	unsigned long insn;
+
+	insn = perf_evsel__intval(evsel, sample, "instruction");
+	key->key = icpt_insn_decoder(insn);
+	key->exit_reasons = sie_icpt_insn_codes;
+}
+
+static void event_sigp_get_key(struct perf_evsel *evsel,
+			       struct perf_sample *sample,
+			       struct event_key *key)
+{
+	key->key = perf_evsel__intval(evsel, sample, "order_code");
+	key->exit_reasons = sie_sigp_order_codes;
+}
+
+static void event_diag_get_key(struct perf_evsel *evsel,
+			       struct perf_sample *sample,
+			       struct event_key *key)
+{
+	key->key = perf_evsel__intval(evsel, sample, "code");
+	key->exit_reasons = sie_diagnose_codes;
+}
+
+static void event_icpt_prog_get_key(struct perf_evsel *evsel,
+				    struct perf_sample *sample,
+				    struct event_key *key)
+{
+	key->key = perf_evsel__intval(evsel, sample, "code");
+	key->exit_reasons = sie_icpt_prog_codes;
+}
+
+static struct child_event_ops child_events[] = {
+	{ .name = "kvm:kvm_s390_intercept_instruction",
+	  .get_key = event_icpt_insn_get_key },
+	{ .name = "kvm:kvm_s390_handle_sigp",
+	  .get_key = event_sigp_get_key },
+	{ .name = "kvm:kvm_s390_handle_diag",
+	  .get_key = event_diag_get_key },
+	{ .name = "kvm:kvm_s390_intercept_prog",
+	  .get_key = event_icpt_prog_get_key },
+	{ NULL, NULL },
+};
+
+static struct kvm_events_ops exit_events = {
+	.is_begin_event = exit_event_begin,
+	.is_end_event = exit_event_end,
+	.child_ops = child_events,
+	.decode_key = exit_event_decode_key,
+	.name = "VM-EXIT"
+};
+
+const char * const kvm_events_tp[] = {
+	"kvm:kvm_s390_sie_enter",
+	"kvm:kvm_s390_sie_exit",
+	"kvm:kvm_s390_intercept_instruction",
+	"kvm:kvm_s390_handle_sigp",
+	"kvm:kvm_s390_handle_diag",
+	"kvm:kvm_s390_intercept_prog",
+	NULL,
+};
+
+struct kvm_reg_events_ops kvm_reg_events_ops[] = {
+	{ .name = "vmexit", .ops = &exit_events },
+	{ NULL, NULL },
+};
+
+const char * const kvm_skip_events[] = {
+	"Wait state",
+	NULL,
+};
+
+int cpu_isa_init(struct perf_kvm_stat *kvm, const char *cpuid)
+{
+	if (strstr(cpuid, "IBM/S390")) {
+		kvm->exit_reasons = sie_exit_reasons;
+		kvm->exit_reasons_isa = "SIE";
+	} else
+		return -ENOTSUP;
+
+	return 0;
+}
diff --git a/tools/perf/builtin-kvm.c b/tools/perf/builtin-kvm.c
index fc2d63d3e791..43367eb00510 100644
--- a/tools/perf/builtin-kvm.c
+++ b/tools/perf/builtin-kvm.c
@@ -88,7 +88,7 @@ void exit_event_decode_key(struct perf_kvm_stat *kvm,
 			   struct event_key *key,
 			   char *decode)
 {
-	const char *exit_reason = get_exit_reason(kvm, kvm->exit_reasons,
+	const char *exit_reason = get_exit_reason(kvm, key->exit_reasons,
 						  key->key);
 
 	scnprintf(decode, DECODE_STR_LEN, "%s", exit_reason);
@@ -261,6 +261,43 @@ static bool update_kvm_event(struct kvm_event *event, int vcpu_id,
 	return true;
 }
 
+static bool is_child_event(struct perf_kvm_stat *kvm,
+			   struct perf_evsel *evsel,
+			   struct perf_sample *sample,
+			   struct event_key *key)
+{
+	struct child_event_ops *child_ops;
+
+	child_ops = kvm->events_ops->child_ops;
+
+	if (!child_ops)
+		return false;
+
+	for (; child_ops->name; child_ops++) {
+		if (!strcmp(evsel->name, child_ops->name)) {
+			child_ops->get_key(evsel, sample, key);
+			return true;
+		}
+	}
+
+	return false;
+}
+
+static bool handle_child_event(struct perf_kvm_stat *kvm,
+			       struct vcpu_event_record *vcpu_record,
+			       struct event_key *key,
+			       struct perf_sample *sample __maybe_unused)
+{
+	struct kvm_event *event = NULL;
+
+	if (key->key != INVALID_KEY)
+		event = find_create_kvm_event(kvm, key);
+
+	vcpu_record->last_event = event;
+
+	return true;
+}
+
 static bool skip_event(const char *event)
 {
 	const char * const *skip_events;
@@ -361,7 +398,8 @@ static bool handle_kvm_event(struct perf_kvm_stat *kvm,
 			     struct perf_sample *sample)
 {
 	struct vcpu_event_record *vcpu_record;
-	struct event_key key = {.key = INVALID_KEY};
+	struct event_key key = { .key = INVALID_KEY,
+				 .exit_reasons = kvm->exit_reasons };
 
 	vcpu_record = per_vcpu_record(thread, evsel, sample);
 	if (!vcpu_record)
@@ -375,6 +413,9 @@ static bool handle_kvm_event(struct perf_kvm_stat *kvm,
 	if (kvm->events_ops->is_begin_event(evsel, sample, &key))
 		return handle_begin_event(kvm, vcpu_record, &key, sample->time);
 
+	if (is_child_event(kvm, evsel, sample, &key))
+		return handle_child_event(kvm, vcpu_record, &key, sample);
+
 	if (kvm->events_ops->is_end_event(evsel, sample, &key))
 		return handle_end_event(kvm, vcpu_record, &key, sample);
 
@@ -1143,7 +1184,8 @@ kvm_events_report(struct perf_kvm_stat *kvm, int argc, const char **argv)
 {
 	const struct option kvm_events_report_options[] = {
 		OPT_STRING(0, "event", &kvm->report_event, "report event",
-			    "event for reporting: vmexit, mmio, ioport"),
+			   "event for reporting: vmexit, "
+			   "mmio (x86 only), ioport (x86 only)"),
 		OPT_INTEGER(0, "vcpu", &kvm->trace_vcpu,
 			    "vcpu id to report"),
 		OPT_STRING('k', "key", &kvm->sort_key, "sort-key",
@@ -1249,7 +1291,9 @@ static int kvm_events_live(struct perf_kvm_stat *kvm,
 			"key for sorting: sample(sort by samples number)"
 			" time (sort by avg time)"),
 		OPT_U64(0, "duration", &kvm->duration,
-		    "show events other than HALT that take longer than duration usecs"),
+			"show events other than"
+			" HLT (x86 only) or Wait state (s390 only)"
+			" that take longer than duration usecs"),
 		OPT_END()
 	};
 	const char * const live_usage[] = {
diff --git a/tools/perf/util/kvm-stat.h b/tools/perf/util/kvm-stat.h
index ba937caa28ac..0b5a8cd2ee79 100644
--- a/tools/perf/util/kvm-stat.h
+++ b/tools/perf/util/kvm-stat.h
@@ -12,6 +12,7 @@ struct event_key {
 	#define INVALID_KEY     (~0ULL)
 	u64 key;
 	int info;
+	struct exit_reasons_table *exit_reasons;
 };
 
 struct kvm_event_stats {
@@ -41,12 +42,20 @@ struct kvm_event_key {
 
 struct perf_kvm_stat;
 
+struct child_event_ops {
+	void (*get_key)(struct perf_evsel *evsel,
+			struct perf_sample *sample,
+			struct event_key *key);
+	const char *name;
+};
+
 struct kvm_events_ops {
 	bool (*is_begin_event)(struct perf_evsel *evsel,
 			       struct perf_sample *sample,
 			       struct event_key *key);
 	bool (*is_end_event)(struct perf_evsel *evsel,
 			     struct perf_sample *sample, struct event_key *key);
+	struct child_event_ops *child_ops;
 	void (*decode_key)(struct perf_kvm_stat *kvm, struct event_key *key,
 			   char *decode);
 	const char *name;
-- 
1.9.3


  parent reply	other threads:[~2014-07-17 16:23 UTC|newest]

Thread overview: 35+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2014-07-17 16:12 [GIT PULL 00/33] perf/core improvements and fixes Arnaldo Carvalho de Melo
2014-07-17 16:12 ` [PATCH 01/33] perf kvm: Use defines of kvm events Arnaldo Carvalho de Melo
2014-07-17 16:12 ` [PATCH 02/33] perf kvm: Move arch specific code into arch/ Arnaldo Carvalho de Melo
2014-07-17 16:12 ` [PATCH 03/33] perf kvm: Add skip_event() for --duration option Arnaldo Carvalho de Melo
2014-07-17 16:12 ` Arnaldo Carvalho de Melo [this message]
2014-07-17 16:12 ` [PATCH 05/33] perf script: Add missing calls to Py_DECREF for return values Arnaldo Carvalho de Melo
2014-07-17 16:12 ` [PATCH 06/33] perf script: Add callchain to generic and tracepoint events Arnaldo Carvalho de Melo
2014-07-17 16:12 ` [PATCH 07/33] perf script: Provide additional sample information on generic events Arnaldo Carvalho de Melo
2014-07-17 16:12 ` [PATCH 08/33] perf machine: Fix the value used for unknown pids Arnaldo Carvalho de Melo
2014-07-17 16:12 ` [PATCH 09/33] perf script: Display PERF_RECORD_MISC_COMM_EXEC flag Arnaldo Carvalho de Melo
2014-07-17 16:12 ` [PATCH 10/33] perf record: Select comm_exec flag if supported Arnaldo Carvalho de Melo
2014-07-17 16:12 ` [PATCH 11/33] perf tools: Fix missing kernel map load Arnaldo Carvalho de Melo
2014-07-17 16:12 ` [PATCH 12/33] perf symbols: Fix missing GNU IFUNC symbols Arnaldo Carvalho de Melo
2014-07-17 16:12 ` [PATCH 13/33] perf inject: Fix build id injection Arnaldo Carvalho de Melo
2014-07-17 16:12 ` [PATCH 14/33] perf callchain: Fix appending a callchain from a previous sample Arnaldo Carvalho de Melo
2014-07-17 16:12 ` [PATCH 15/33] perf buildid-cache: Apply force option to copying kcore Arnaldo Carvalho de Melo
2014-07-17 16:12 ` [PATCH 16/33] perf symbols: Record whether a dso is 64-bit Arnaldo Carvalho de Melo
2014-07-17 16:12 ` [PATCH 17/33] perf symbols: Do not attempt to read data from kallsyms Arnaldo Carvalho de Melo
2014-07-17 16:12 ` [PATCH 18/33] perf symbols: Add ability to iterate over a dso's symbols Arnaldo Carvalho de Melo
2014-07-17 16:12 ` [PATCH 19/33] perf session: Flag if the event stream is entirely in memory Arnaldo Carvalho de Melo
2014-07-17 16:12 ` [PATCH 20/33] perf evlist: Pass mmap parameters in a struct Arnaldo Carvalho de Melo
2014-07-17 16:12 ` [PATCH 21/33] perf tools: Add feature test for __sync_val_compare_and_swap Arnaldo Carvalho de Melo
2014-07-17 16:12 ` [PATCH 22/33] perf tools: Add option macro OPT_CALLBACK_OPTARG Arnaldo Carvalho de Melo
2014-07-17 16:12 ` [PATCH 23/33] perf evsel: Add 'no_aux_samples' option Arnaldo Carvalho de Melo
2014-07-17 16:12 ` [PATCH 24/33] perf evsel: Add 'immediate' option Arnaldo Carvalho de Melo
2014-07-17 16:12 ` [PATCH 25/33] perf machine: Fix map groups of threads with unknown pids Arnaldo Carvalho de Melo
2014-07-17 16:12 ` [PATCH 26/33] perf thread: Allow deletion of a thread with no map groups Arnaldo Carvalho de Melo
2014-07-17 16:12 ` [PATCH 27/33] perf machine: Fix leak of 'struct thread' on error path Arnaldo Carvalho de Melo
2014-07-17 16:12 ` [PATCH 28/33] perf tools: Remove verbose from functions prototypes Arnaldo Carvalho de Melo
2014-07-17 16:12 ` [PATCH 29/33] perf tools: Move pr_* debug macros into debug object Arnaldo Carvalho de Melo
2014-07-17 16:12 ` [PATCH 30/33] perf tools: Factor eprintf to allow different debug variables Arnaldo Carvalho de Melo
2014-07-17 16:12 ` [PATCH 31/33] perf tools: Add --debug optionto set debug variable Arnaldo Carvalho de Melo
2014-07-17 16:12 ` [PATCH 32/33] perf tools: Remove needless getopt.h includes Arnaldo Carvalho de Melo
2014-07-17 16:12 ` [PATCH 33/33] perf tools: Allow TSC conversion on any arch Arnaldo Carvalho de Melo
2014-07-18  4:17 ` [GIT PULL 00/33] 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=1405613555-23380-5-git-send-email-acme@kernel.org \
    --to=acme@kernel.org \
    --cc=a.p.zijlstra@chello.nl \
    --cc=acme@redhat.com \
    --cc=borntraeger@de.ibm.com \
    --cc=cornelia.huck@de.ibm.com \
    --cc=dsahern@gmail.com \
    --cc=jolsa@redhat.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mingo@kernel.org \
    --cc=paulus@samba.org \
    --cc=yarygin@linux.vnet.ibm.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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.