Message-Id: <1509732006-5917-3-git-send-email-megha.dey@linux.intel.com>
Date:   Fri,  3 Nov 2017 11:00:05 -0700
From:   Megha Dey <megha.dey@...ux.intel.com>
To:     x86@...nel.org, linux-kernel@...r.kernel.org,
        linux-doc@...r.kernel.org
Cc:     tglx@...utronix.de, mingo@...hat.com, hpa@...or.com,
        andriy.shevchenko@...ux.intel.com, kstewart@...uxfoundation.org,
        yu-cheng.yu@...el.com, len.brown@...el.com,
        gregkh@...uxfoundation.org, peterz@...radead.org, acme@...nel.org,
        alexander.shishkin@...ux.intel.com, jolsa@...hat.com,
        namhyung@...nel.org, vikas.shivappa@...ux.intel.com,
        pombredanne@...b.com, me@...ehuey.com, bp@...e.de,
        grzegorz.andrejczuk@...el.com, tony.luck@...el.com, corbet@....net,
        ravi.v.shankar@...el.com, megha.dey@...el.com,
        Megha Dey <megha.dey@...ux.intel.com>
Subject: [PATCH V0 2/3] perf/x86/intel/bm.c: Add Intel Branch Monitoring support

Currently, the cannonlake family of Intel processors support the
branch monitoring feature. Intel's Branch monitoring feature is trying
to utilize heuristics to detect the occurrence of an ROP (Return
Oriented Programming) attack.

A perf-based kernel driver has been used to monitor the occurrence of
one of the 6 branch monitoring events. There are 2 counters that each
can select between one of these events for evaluation over a specified
instruction window size (0 to 1023). For each counter, a threshold value
(0 to 127) can be configured to set a point at which ROP detection event
action is taken (determined by user-space). The entire system can monitor
a maximum of 2 events(either from the same or different tasks) at any
given time.

Apart from window_size(global) and threshold(per-counter), various sysfs
entries are provided for the user to configure: guest_disable, lbr_freeze,
window_cnt_sel, cnt_and_mode (all global) and mispred_evt_cnt(per-counter).

Everytime a task is scheduled out, we save current window and count
associated with the event being monitored. When the task is scheduled
next, we start counting from previous count associated with this event.
Thus, a full context switch in this case is not necessary.

To monitor a user space application for ROP related events, perf command
line can be used as follows:

perf stat -e <name of event> <application to be monitored>

eg. For the following test program (test.c) and threshold = 100
(echo 100 > /sys/devices/intel_bm/threshold)

void func(void)
{
        return;
}

void main(void)
{
        int i;

        for (i = 0; i < 128; i++) {
                func();
        }

        return;
}

perf stat -e intel_bm/rets/ ./test

 Performance counter stats for './test':

                 1      intel_bm/rets/

       0.104705937 seconds time elapsed

perf returns the number of branch monitoring interrupts occurred during
the execution of the user-space application.

Signed-off-by: Megha Dey <megha.dey@...ux.intel.com>
Signed-off-by: Yu-Cheng Yu <yu-cheng.yu@...el.com>
---
 arch/x86/events/Kconfig          |  10 +
 arch/x86/events/intel/Makefile   |   2 +
 arch/x86/events/intel/bm.c       | 635 +++++++++++++++++++++++++++++++++++++++
 arch/x86/include/asm/msr-index.h |   5 +
 include/linux/perf_event.h       |   8 +-
 kernel/events/core.c             |   8 +
 6 files changed, 667 insertions(+), 1 deletion(-)
 create mode 100644 arch/x86/events/intel/bm.c

diff --git a/arch/x86/events/Kconfig b/arch/x86/events/Kconfig
index 9a7a144..40903ca 100644
--- a/arch/x86/events/Kconfig
+++ b/arch/x86/events/Kconfig
@@ -9,6 +9,16 @@ config PERF_EVENTS_INTEL_UNCORE
 	Include support for Intel uncore performance events. These are
 	available on NehalemEX and more modern processors.
 
+config PERF_EVENTS_INTEL_BM
+	bool "Intel Branch Monitoring support"
+	depends on PERF_EVENTS && CPU_SUP_INTEL && PCI
+	---help---
+	  Include support for Intel Branch monitoring. This feature utilizes
+	  heuristics for detecting ROP(Return oriented programming) like
+	  attacks. These heuristics are based off certain performance
+	  monitoring statistics, measured dynamically over a short
+	  configurable window period.
+
 config PERF_EVENTS_INTEL_RAPL
 	tristate "Intel rapl performance events"
 	depends on PERF_EVENTS && CPU_SUP_INTEL && PCI
diff --git a/arch/x86/events/intel/Makefile b/arch/x86/events/intel/Makefile
index 3468b0c..14235ec 100644
--- a/arch/x86/events/intel/Makefile
+++ b/arch/x86/events/intel/Makefile
@@ -2,6 +2,8 @@
 obj-$(CONFIG_CPU_SUP_INTEL)		+= core.o bts.o
 obj-$(CONFIG_CPU_SUP_INTEL)		+= ds.o knc.o
 obj-$(CONFIG_CPU_SUP_INTEL)		+= lbr.o p4.o p6.o pt.o
+obj-$(CONFIG_PERF_EVENTS_INTEL_BM)	+= intel-bm-perf.o
+intel-bm-perf-objs			:= bm.o
 obj-$(CONFIG_PERF_EVENTS_INTEL_RAPL)	+= intel-rapl-perf.o
 intel-rapl-perf-objs			:= rapl.o
 obj-$(CONFIG_PERF_EVENTS_INTEL_UNCORE)	+= intel-uncore.o
diff --git a/arch/x86/events/intel/bm.c b/arch/x86/events/intel/bm.c
new file mode 100644
index 0000000..79f4923
--- /dev/null
+++ b/arch/x86/events/intel/bm.c
@@ -0,0 +1,635 @@
+/*
+ * Support for Intel branch monitoring counters
+ *
+ * Intel branch monitoring MSRs are specified in the Intel® 64 and IA-32
+ * Software Developer’s Manual Volume 4 section 2.16.2 (October 2017)
+ *
+ * Copyright (c) 2017, Intel Corporation.
+ *
+ * Contact Information:
+ * Megha Dey <megha.dey@...ux.intel.com>
+ * Yu-Cheng Yu <yu-cheng.yu@...el.com>
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
+ * more details.
+ */
+
+#include <linux/module.h>
+#include <linux/perf_event.h>
+#include <linux/slab.h>
+#include <linux/poll.h>
+#include <linux/err.h>
+#include <asm/apic.h>
+#include <asm/cpu_device_id.h>
+#include <asm/intel-family.h>
+#include <asm/nmi.h>
+
+#include "../perf_event.h"
+
+/* Branch Monitoring default and mask values */
+#define BM_MAX_WINDOW_SIZE		0x3ff
+#define BM_MAX_THRESHOLD		0x7f
+#define BM_MAX_EVENTS			6
+#define BM_WINDOW_SIZE_SHIFT		8
+#define BM_THRESHOLD_SHIFT		8
+#define BM_EVENT_TYPE_SHIFT		1
+#define BM_GUEST_DISABLE_SHIFT		3
+#define BM_LBR_FREEZE_SHIFT		2
+#define BM_WINDOW_CNT_SEL_SHIFT		24
+#define BM_CNT_AND_MODE_SHIFT		26
+#define BM_MISPRED_EVT_CNT_SHIFT	15
+#define BM_ENABLE			0x3
+
+static unsigned int bm_window_size = BM_MAX_WINDOW_SIZE;
+static unsigned int bm_guest_disable;
+static unsigned int bm_lbr_freeze;
+static unsigned int bm_window_cnt_sel;
+static unsigned int bm_cnt_and_mode;
+
+static unsigned int bm_threshold = BM_MAX_THRESHOLD;
+static unsigned int bm_mispred_evt_cnt;
+
+/* Branch monitoring counter owners */
+static struct perf_event *bm_counter_owner[2];
+
+static struct pmu intel_bm_pmu;
+
+static int bm_num_counters;
+
+DEFINE_PER_CPU(int, bm_unmask_apic);
+
+union bm_detect_status {
+	struct {
+		__u8 event: 1;
+		__u8 lbrs_valid: 1;
+		__u8 reserved0: 6;
+		__u8 ctrl_hit: 4;
+		__u8 reserved1: 4;
+		__u16 count_window: 10;
+		__u8 reserved2: 6;
+		__u8 count[4];
+	} __packed;
+	uint64_t raw;
+};
+
+static int intel_bm_event_nmi_handler(unsigned int cmd, struct pt_regs *regs)
+{
+	struct perf_event *event;
+	union bm_detect_status stat;
+	struct perf_sample_data data;
+	int i;
+	unsigned long x;
+
+	rdmsrl(BR_DETECT_STATUS_MSR, stat.raw);
+
+	if (stat.event) {
+		wrmsrl(BR_DETECT_STATUS_MSR, 0);
+		apic_write(APIC_LVTPC, APIC_DM_NMI);
+		/*
+		 * Issue wake-up to corrresponding polling event
+		 */
+		x = stat.ctrl_hit;
+		for_each_set_bit(i, &x, bm_num_counters) {
+			event = bm_counter_owner[i];
+			perf_sample_data_init(&data, 0, event->hw.last_period);
+			perf_event_overflow(event, &data, regs);
+			local64_inc(&event->count);
+			atomic_set(&event->hw.bm_poll, POLLIN);
+			event->pending_wakeup = 1;
+			irq_work_queue(&event->pending);
+		}
+		return NMI_HANDLED;
+	}
+	return NMI_DONE;
+}
+
+static void intel_bm_event_start(struct perf_event *event, int mode)
+{
+	wrmsrl(BR_DETECT_COUNTER_CONFIG_BASE + event->id,
+	       (event->hw.bm_counter_conf | 1));
+}
+
+/*
+ * Unmask the NMI bit of the local APIC the first time task is scheduled
+ * on a particular CPU.
+ */
+static void intel_bm_unmask_nmi(void)
+{
+	this_cpu_write(bm_unmask_apic, 0);
+
+	if (!(this_cpu_read(bm_unmask_apic))) {
+		apic_write(APIC_LVTPC, APIC_DM_NMI);
+		this_cpu_inc(bm_unmask_apic);
+	}
+}
+
+static int intel_bm_event_add(struct perf_event *event, int mode)
+{
+	union bm_detect_status cur_stat, prev_stat;
+
+	prev_stat.raw = local64_read(&event->hw.prev_count);
+
+	/*
+	 * Start counting from previous count associated with this event
+	 */
+	rdmsrl(BR_DETECT_STATUS_MSR, cur_stat.raw);
+
+	cur_stat.count[event->id] = prev_stat.count[event->id];
+	cur_stat.count_window = prev_stat.count_window;
+	wrmsrl(BR_DETECT_STATUS_MSR, cur_stat.raw);
+
+	wrmsrl(BR_DETECT_CONTROL_MSR, event->hw.bm_ctrl);
+
+	intel_bm_unmask_nmi();
+
+	intel_bm_event_start(event, mode);
+
+	return 0;
+}
+
+static void intel_bm_event_update(struct perf_event *event)
+{
+	union bm_detect_status cur_stat;
+
+	rdmsrl(BR_DETECT_STATUS_MSR, cur_stat.raw);
+	local64_set(&event->hw.prev_count, (uint64_t)cur_stat.raw);
+}
+
+static void intel_bm_event_stop(struct perf_event *event, int mode)
+{
+	wrmsrl(BR_DETECT_COUNTER_CONFIG_BASE + event->id,
+	       (event->hw.bm_counter_conf & ~1));
+
+	intel_bm_event_update(event);
+}
+
+static void intel_bm_event_del(struct perf_event *event, int flags)
+{
+	intel_bm_event_stop(event, flags);
+}
+
+static void intel_bm_event_read(struct perf_event *event)
+{
+}
+
+static void intel_bm_event_destroy(struct perf_event *event)
+{
+	bm_counter_owner[event->id] = NULL;
+}
+
+static int intel_bm_event_init(struct perf_event *event)
+{
+	u64 cfg;
+	int counter_to_use = -1, i;
+
+	local64_set(&event->hw.prev_count, 0);
+
+	if (perf_paranoid_cpu() && !capable(CAP_SYS_ADMIN))
+		return -EACCES;
+
+	/*
+	 * Type is assigned by kernel, see /sys/devices/intel_bm/type
+	 */
+	if (event->attr.type != intel_bm_pmu.type)
+		return -ENOENT;
+
+	/*
+	 * Only per tasks events are supported. It does not make sense to
+	 * monitor all tasks for an ROP attack. This could generate a lot
+	 * of false positives.
+	 */
+	if (event->hw.target == NULL)
+		return -EINVAL;
+
+	event->event_caps |= PERF_EV_CAP_BM;
+	/*
+	 * cfg contains one of the 6 possible Branch Monitoring events
+	 */
+	cfg = event->attr.config;
+	if (cfg < 0 || cfg > (BM_MAX_EVENTS - 1))
+		return -EINVAL;
+
+	if (event->attr.sample_period) /* no sampling */
+		return -EINVAL;
+
+	/*
+	 * Find a hardware counter for the target task
+	 */
+	for (i = 0; i < bm_num_counters; i++) {
+		if ((bm_counter_owner[i] == NULL) ||
+			(bm_counter_owner[i]->state == PERF_EVENT_STATE_DEAD)) {
+			counter_to_use = i;
+			bm_counter_owner[i] = event;
+			break;
+		}
+	}
+
+	if (counter_to_use == -1)
+		return -EBUSY;
+
+	event->hw.bm_ctrl = (bm_window_size << BM_WINDOW_SIZE_SHIFT) |
+			    (bm_guest_disable << BM_GUEST_DISABLE_SHIFT) |
+			    (bm_lbr_freeze << BM_LBR_FREEZE_SHIFT) |
+			    (bm_window_cnt_sel << BM_WINDOW_CNT_SEL_SHIFT) |
+			    (bm_cnt_and_mode << BM_CNT_AND_MODE_SHIFT) |
+								BM_ENABLE;
+	event->hw.bm_counter_conf = (bm_threshold << BM_THRESHOLD_SHIFT) |
+			(bm_mispred_evt_cnt << BM_MISPRED_EVT_CNT_SHIFT) |
+					(cfg << BM_EVENT_TYPE_SHIFT);
+
+	wrmsrl(BR_DETECT_COUNTER_CONFIG_BASE + counter_to_use,
+						event->hw.bm_counter_conf);
+	wrmsrl(BR_DETECT_STATUS_MSR, 0);
+	event->id = counter_to_use;
+	local64_set(&event->count, 0);
+
+	event->destroy = intel_bm_event_destroy;
+
+	return 0;
+}
+
+EVENT_ATTR_STR(rets, rets, "event=0x0");
+EVENT_ATTR_STR(call-ret, call_ret, "event=0x01");
+EVENT_ATTR_STR(ret-misp, ret_misp, "event=0x02");
+EVENT_ATTR_STR(branch-misp, branch_mispredict, "event=0x03");
+EVENT_ATTR_STR(indirect-branch-misp, indirect_branch_mispredict, "event=0x04");
+EVENT_ATTR_STR(far-branch, far_branch, "event=0x05");
+
+static struct attribute *intel_bm_events_attr[] = {
+	EVENT_PTR(rets),
+	EVENT_PTR(call_ret),
+	EVENT_PTR(ret_misp),
+	EVENT_PTR(branch_mispredict),
+	EVENT_PTR(indirect_branch_mispredict),
+	EVENT_PTR(far_branch),
+	NULL,
+};
+
+static struct attribute_group intel_bm_events_group = {
+	.name = "events",
+	.attrs = intel_bm_events_attr,
+};
+
+PMU_FORMAT_ATTR(event, "config:0-7");
+static struct attribute *intel_bm_formats_attr[] = {
+	&format_attr_event.attr,
+	NULL,
+};
+
+static struct attribute_group intel_bm_format_group = {
+	.name = "format",
+	.attrs = intel_bm_formats_attr,
+};
+
+/*
+ * User can configure the BM MSRs using the corresponding sysfs entries
+ */
+
+static ssize_t
+threshold_show(struct device *dev, struct device_attribute *attr,
+			char *buf)
+{
+	ssize_t rv;
+
+	rv = sprintf(buf, "%d\n", bm_threshold);
+
+	return rv;
+}
+
+static ssize_t
+threshold_store(struct device *dev,
+			   struct device_attribute *attr,
+			   const char *buf, size_t count)
+{
+	unsigned int threshold;
+	int err;
+
+	err = kstrtouint(buf, 0, &threshold);
+	if (err)
+		return err;
+
+	if (threshold > BM_MAX_THRESHOLD) {
+		pr_err("invalid threshold value\n");
+		return -EINVAL;
+	}
+
+	bm_threshold = threshold;
+
+	return count;
+}
+
+static DEVICE_ATTR_RW(threshold);
+
+static ssize_t
+window_size_show(struct device *dev, struct device_attribute *attr,
+			char *buf)
+{
+	ssize_t rv;
+
+	rv = sprintf(buf, "%d\n", bm_window_size);
+
+	return rv;
+}
+
+static ssize_t
+window_size_store(struct device *dev,
+			    struct device_attribute *attr,
+			    const char *buf, size_t count)
+{
+	unsigned int window_size;
+	int err;
+
+	err = kstrtouint(buf, 0, &window_size);
+	if (err)
+		return err;
+
+	if (window_size > BM_MAX_WINDOW_SIZE) {
+		pr_err("illegal window size\n");
+		return -EINVAL;
+	}
+
+	bm_window_size = window_size;
+
+	return count;
+}
+
+static DEVICE_ATTR_RW(window_size);
+
+static ssize_t
+lbr_freeze_show(struct device *dev, struct device_attribute *attr,
+			char *buf)
+{
+	ssize_t rv;
+
+	rv = sprintf(buf, "%d\n", bm_lbr_freeze);
+
+	return rv;
+}
+
+static ssize_t
+lbr_freeze_store(struct device *dev,
+				struct device_attribute *attr,
+				const char *buf, size_t count)
+{
+	unsigned int lbr_freeze;
+	int err;
+
+	err = kstrtouint(buf, 0, &lbr_freeze);
+	if (err)
+		return err;
+
+	if (lbr_freeze > 1) {
+		pr_err("lbr freeze can only be 0 or 1\n");
+		return -EINVAL;
+	}
+
+	bm_lbr_freeze = lbr_freeze;
+
+	return count;
+}
+
+static DEVICE_ATTR_RW(lbr_freeze);
+
+static ssize_t
+guest_disable_show(struct device *dev, struct device_attribute *attr,
+			char *buf)
+{
+	ssize_t rv;
+
+	rv = sprintf(buf, "%d\n", bm_guest_disable);
+
+	return rv;
+}
+
+static ssize_t
+guest_disable_store(struct device *dev,
+			struct device_attribute *attr,
+			const char *buf, size_t count)
+{
+	unsigned int guest_disable;
+	int err;
+
+	err = kstrtouint(buf, 0, &guest_disable);
+	if (err)
+		return err;
+
+	if (guest_disable > 1) {
+		pr_err("guest disable can only be 0 or 1\n");
+		return -EINVAL;
+	}
+
+	bm_guest_disable = guest_disable;
+
+	return count;
+}
+
+static DEVICE_ATTR_RW(guest_disable);
+
+static ssize_t
+window_cnt_sel_show(struct device *dev, struct device_attribute *attr,
+			char *buf)
+{
+	ssize_t rv;
+
+	rv = sprintf(buf, "%d\n", bm_window_cnt_sel);
+
+	return rv;
+}
+
+static ssize_t
+window_cnt_sel_store(struct device *dev,
+				struct device_attribute *attr,
+				const char *buf, size_t count)
+{
+	unsigned int window_cnt_sel;
+	int err;
+
+	err = kstrtouint(buf, 0, &window_cnt_sel);
+	if (err)
+		return err;
+
+	if (window_cnt_sel > 3) {
+		pr_err("invalid window_cnt_sel value\n");
+		return -EINVAL;
+	}
+
+	bm_window_cnt_sel = window_cnt_sel;
+
+	return count;
+}
+
+static DEVICE_ATTR_RW(window_cnt_sel);
+
+static ssize_t
+cnt_and_mode_show(struct device *dev, struct device_attribute *attr,
+			char *buf)
+{
+	ssize_t rv;
+
+	rv = sprintf(buf, "%d\n", bm_cnt_and_mode);
+
+	return rv;
+}
+
+static ssize_t
+cnt_and_mode_store(struct device *dev,
+				struct device_attribute *attr,
+				const char *buf, size_t count)
+{
+	unsigned int cnt_and_mode;
+	int err;
+
+	err = kstrtouint(buf, 0, &cnt_and_mode);
+	if (err)
+		return err;
+
+	if (cnt_and_mode > 1) {
+		pr_err("invalid cnt_and_mode value\n");
+		return -EINVAL;
+	}
+
+	bm_cnt_and_mode = cnt_and_mode;
+
+	return count;
+}
+
+static DEVICE_ATTR_RW(cnt_and_mode);
+
+static ssize_t
+mispred_evt_cnt_show(struct device *dev, struct device_attribute *attr,
+				char *buf)
+{
+	ssize_t rv;
+
+	rv = sprintf(buf, "%d\n", bm_mispred_evt_cnt);
+
+	return rv;
+}
+
+static ssize_t
+mispred_evt_cnt_store(struct device *dev,
+				struct device_attribute *attr,
+				const char *buf, size_t count)
+{
+	unsigned int mispred_evt_cnt;
+	int err;
+
+	err = kstrtouint(buf, 0, &mispred_evt_cnt);
+	if (err)
+		return err;
+
+	if (mispred_evt_cnt > 1) {
+		pr_err("invalid mispred_evt_cnt value\n");
+		return -EINVAL;
+	}
+
+	bm_mispred_evt_cnt = mispred_evt_cnt;
+
+	return count;
+}
+
+static DEVICE_ATTR_RW(mispred_evt_cnt);
+
+static ssize_t
+num_counters_show(struct device *dev, struct device_attribute *attr,
+			char *buf)
+{
+	ssize_t rv;
+
+	rv = sprintf(buf, "%d\n", bm_num_counters);
+
+	return rv;
+}
+
+static DEVICE_ATTR_RO(num_counters);
+
+static struct attribute *intel_bm_attrs[] = {
+	&dev_attr_window_size.attr,
+	&dev_attr_threshold.attr,
+	&dev_attr_lbr_freeze.attr,
+	&dev_attr_guest_disable.attr,
+	&dev_attr_window_cnt_sel.attr,
+	&dev_attr_cnt_and_mode.attr,
+	&dev_attr_mispred_evt_cnt.attr,
+	&dev_attr_num_counters.attr,
+	NULL,
+};
+
+static const struct attribute_group intel_bm_group = {
+	.attrs = intel_bm_attrs,
+};
+
+static const struct attribute_group *intel_bm_attr_groups[] = {
+	&intel_bm_events_group,
+	&intel_bm_format_group,
+	&intel_bm_group,
+	NULL,
+};
+
+static struct pmu intel_bm_pmu = {
+	.task_ctx_nr     = perf_sw_context,
+	.attr_groups     = intel_bm_attr_groups,
+	.event_init      = intel_bm_event_init,
+	.add             = intel_bm_event_add,
+	.del             = intel_bm_event_del,
+	.start           = intel_bm_event_start,
+	.stop            = intel_bm_event_stop,
+	.read            = intel_bm_event_read,
+};
+
+#define X86_BM_MODEL_MATCH(model)       \
+	{ X86_VENDOR_INTEL, 6, model, X86_FEATURE_ANY }
+
+static const struct x86_cpu_id bm_cpu_match[] __initconst = {
+	X86_BM_MODEL_MATCH(INTEL_FAM6_CANNONLAKE_MOBILE),
+	{},
+};
+
+MODULE_DEVICE_TABLE(x86cpu, bm_cpu_match);
+
+static __init int intel_bm_init(void)
+{
+	int ret, err;
+
+	/*
+	 * Only CNL supports branch monitoring
+	 */
+	if (!(x86_match_cpu(bm_cpu_match)))
+		return -ENODEV;
+
+	bm_num_counters = 2;
+
+	err = register_nmi_handler(NMI_LOCAL, intel_bm_event_nmi_handler,
+								0, "BM");
+
+	if (err)
+		goto fail_nmi;
+
+	ret =  perf_pmu_register(&intel_bm_pmu, "intel_bm", -1);
+	if (ret) {
+		pr_err("Intel BM perf registration failed: %d\n", ret);
+		return ret;
+	}
+
+	return 0;
+
+fail_nmi:
+	unregister_nmi_handler(NMI_LOCAL, "BM");
+	return err;
+}
+module_init(intel_bm_init);
+
+static void __exit intel_bm_exit(void)
+{
+	perf_pmu_unregister(&intel_bm_pmu);
+	unregister_nmi_handler(NMI_LOCAL, "BM");
+}
+module_exit(intel_bm_exit);
+
+MODULE_LICENSE("GPL");
diff --git a/arch/x86/include/asm/msr-index.h b/arch/x86/include/asm/msr-index.h
index ab02261..f72de49 100644
--- a/arch/x86/include/asm/msr-index.h
+++ b/arch/x86/include/asm/msr-index.h
@@ -294,6 +294,11 @@
 /* Alternative perfctr range with full access. */
 #define MSR_IA32_PMC0			0x000004c1
 
+/* Intel Branch Monitoring MSRs */
+#define BR_DETECT_CONTROL_MSR		0x00000350
+#define BR_DETECT_STATUS_MSR		0x00000351
+#define BR_DETECT_COUNTER_CONFIG_BASE	0x00000354
+
 /* AMD64 MSRs. Not complete. See the architecture manual for a more
    complete list. */
 
diff --git a/include/linux/perf_event.h b/include/linux/perf_event.h
index 8e22f24..6d929f3 100644
--- a/include/linux/perf_event.h
+++ b/include/linux/perf_event.h
@@ -168,6 +168,12 @@ struct hw_perf_event {
 	 */
 	struct task_struct		*target;
 
+	struct { /* intel_bm */
+			u64 bm_ctrl;
+			u64 bm_counter_conf;
+			atomic_t bm_poll;
+	};
+
 	/*
 	 * PMU would store hardware filter configuration
 	 * here.
@@ -191,7 +197,6 @@ struct hw_perf_event {
 	 * local64_cmpxchg() such that pmu::read() can be called nested.
 	 */
 	local64_t			prev_count;
-
 	/*
 	 * The period to start the next sample with.
 	 */
@@ -512,6 +517,7 @@ typedef void (*perf_overflow_handler_t)(struct perf_event *,
  */
 #define PERF_EV_CAP_SOFTWARE		BIT(0)
 #define PERF_EV_CAP_READ_ACTIVE_PKG	BIT(1)
+#define PERF_EV_CAP_BM			BIT(2)
 
 #define SWEVENT_HLIST_BITS		8
 #define SWEVENT_HLIST_SIZE		(1 << SWEVENT_HLIST_BITS)
diff --git a/kernel/events/core.c b/kernel/events/core.c
index 9d93db8..52b3512 100644
--- a/kernel/events/core.c
+++ b/kernel/events/core.c
@@ -4596,6 +4596,11 @@ static unsigned int perf_poll(struct file *file, poll_table *wait)
 
 	poll_wait(file, &event->waitq, wait);
 
+	if (event->event_caps & PERF_EV_CAP_BM) {
+		events = atomic_xchg(&event->hw.bm_poll, 0);
+		return events;
+	}
+
 	if (is_event_hup(event))
 		return events;
 
@@ -5498,6 +5503,9 @@ void perf_event_wakeup(struct perf_event *event)
 {
 	ring_buffer_wakeup(event);
 
+	if (event->event_caps & PERF_EV_CAP_BM)
+		wake_up_all(&event->waitq);
+
 	if (event->pending_kill) {
 		kill_fasync(perf_event_fasync(event), SIGIO, event->pending_kill);
 		event->pending_kill = 0;
-- 
1.9.1

Powered by blists - more mailing lists