diff mbox

[igt,v4,01/13] igt/perf: add i915 perf stream tests for Haswell

Message ID 20161114205122.10742-2-robert@sixbynine.org (mailing list archive)
State New, archived
Headers show

Commit Message

Robert Bragg Nov. 14, 2016, 8:51 p.m. UTC
Signed-off-by: Robert Bragg <robert@sixbynine.org>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
---
 tests/Makefile.sources |    1 +
 tests/perf.c           | 2373 ++++++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 2374 insertions(+)
 create mode 100644 tests/perf.c
diff mbox

Patch

diff --git a/tests/Makefile.sources b/tests/Makefile.sources
index 65e0792..a818963 100644
--- a/tests/Makefile.sources
+++ b/tests/Makefile.sources
@@ -214,6 +214,7 @@  TESTS_progs = \
 	kms_pwrite_crc \
 	kms_sink_crc_basic \
 	prime_udl \
+	perf \
 	$(NULL)
 
 # IMPORTANT: The ZZ_ tests need to be run last!
diff --git a/tests/perf.c b/tests/perf.c
new file mode 100644
index 0000000..1aafca9
--- /dev/null
+++ b/tests/perf.c
@@ -0,0 +1,2373 @@ 
+/*
+ * Copyright © 2016 Intel Corporation
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ *
+ */
+
+#include <stdlib.h>
+#include <stdio.h>
+#include <string.h>
+#include <fcntl.h>
+#include <inttypes.h>
+#include <errno.h>
+#include <sys/stat.h>
+#include <sys/time.h>
+#include <sys/times.h>
+#include <sys/types.h>
+#include <dirent.h>
+#include <time.h>
+#include <poll.h>
+#include <math.h>
+
+#include "igt.h"
+#include "drm.h"
+
+IGT_TEST_DESCRIPTION("Test the i915 perf metrics streaming interface");
+
+#define GEN6_MI_REPORT_PERF_COUNT ((0x28 << 23) | (3 - 2))
+
+#define GFX_OP_PIPE_CONTROL     ((3 << 29) | (3 << 27) | (2 << 24))
+#define PIPE_CONTROL_CS_STALL	   (1 << 20)
+#define PIPE_CONTROL_GLOBAL_SNAPSHOT_COUNT_RESET	(1 << 19)
+#define PIPE_CONTROL_TLB_INVALIDATE     (1 << 18)
+#define PIPE_CONTROL_SYNC_GFDT	  (1 << 17)
+#define PIPE_CONTROL_MEDIA_STATE_CLEAR  (1 << 16)
+#define PIPE_CONTROL_NO_WRITE	   (0 << 14)
+#define PIPE_CONTROL_WRITE_IMMEDIATE    (1 << 14)
+#define PIPE_CONTROL_WRITE_DEPTH_COUNT  (2 << 14)
+#define PIPE_CONTROL_WRITE_TIMESTAMP    (3 << 14)
+#define PIPE_CONTROL_DEPTH_STALL	(1 << 13)
+#define PIPE_CONTROL_RENDER_TARGET_FLUSH (1 << 12)
+#define PIPE_CONTROL_INSTRUCTION_INVALIDATE (1 << 11)
+#define PIPE_CONTROL_TEXTURE_CACHE_INVALIDATE   (1 << 10) /* GM45+ only */
+#define PIPE_CONTROL_ISP_DIS	    (1 << 9)
+#define PIPE_CONTROL_INTERRUPT_ENABLE   (1 << 8)
+#define PIPE_CONTROL_FLUSH_ENABLE       (1 << 7) /* Gen7+ only */
+/* GT */
+#define PIPE_CONTROL_DATA_CACHE_INVALIDATE      (1 << 5)
+#define PIPE_CONTROL_VF_CACHE_INVALIDATE	(1 << 4)
+#define PIPE_CONTROL_CONST_CACHE_INVALIDATE     (1 << 3)
+#define PIPE_CONTROL_STATE_CACHE_INVALIDATE     (1 << 2)
+#define PIPE_CONTROL_STALL_AT_SCOREBOARD	(1 << 1)
+#define PIPE_CONTROL_DEPTH_CACHE_FLUSH	  (1 << 0)
+#define PIPE_CONTROL_PPGTT_WRITE	(0 << 2)
+#define PIPE_CONTROL_GLOBAL_GTT_WRITE   (1 << 2)
+
+#define NSEC_PER_SEC 1000000000ull
+
+/* Temporarily copy i915-perf uapi here to avoid a dependency on libdrm's
+ * i915_drm.h copy being updated with the i915-perf interface before this
+ * test can land in i-g-t.
+ *
+ * TODO: remove this once the interface lands in libdrm
+ */
+#ifndef DRM_I915_PERF_OPEN
+#define DRM_I915_PERF_OPEN		0x36
+#define DRM_IOCTL_I915_PERF_OPEN	DRM_IOW(DRM_COMMAND_BASE + DRM_I915_PERF_OPEN, struct drm_i915_perf_open_param)
+
+enum drm_i915_oa_format {
+       I915_OA_FORMAT_A13 = 1,
+       I915_OA_FORMAT_A29,
+       I915_OA_FORMAT_A13_B8_C8,
+       I915_OA_FORMAT_B4_C8,
+       I915_OA_FORMAT_A45_B8_C8,
+       I915_OA_FORMAT_B4_C8_A16,
+       I915_OA_FORMAT_C4_B8,
+
+       I915_OA_FORMAT_MAX /* non-ABI */
+};
+
+enum drm_i915_perf_property_id {
+       DRM_I915_PERF_PROP_CTX_HANDLE = 1,
+       DRM_I915_PERF_PROP_SAMPLE_OA,
+       DRM_I915_PERF_PROP_OA_METRICS_SET,
+       DRM_I915_PERF_PROP_OA_FORMAT,
+       DRM_I915_PERF_PROP_OA_EXPONENT,
+
+       DRM_I915_PERF_PROP_MAX /* non-ABI */
+};
+
+struct drm_i915_perf_open_param {
+       __u32 flags;
+#define I915_PERF_FLAG_FD_CLOEXEC	(1<<0)
+#define I915_PERF_FLAG_FD_NONBLOCK	(1<<1)
+#define I915_PERF_FLAG_DISABLED		(1<<2)
+
+       __u32 num_properties;
+       __u64 properties_ptr;
+};
+
+#define I915_PERF_IOCTL_ENABLE _IO('i', 0x0)
+#define I915_PERF_IOCTL_DISABLE	_IO('i', 0x1)
+
+struct drm_i915_perf_record_header {
+       __u32 type;
+       __u16 pad;
+       __u16 size;
+};
+
+enum drm_i915_perf_record_type {
+       DRM_I915_PERF_RECORD_SAMPLE = 1,
+       DRM_I915_PERF_RECORD_OA_REPORT_LOST = 2,
+       DRM_I915_PERF_RECORD_OA_BUFFER_LOST = 3,
+
+       DRM_I915_PERF_RECORD_MAX /* non-ABI */
+};
+#endif /* !DRM_I915_PERF_OPEN */
+
+static struct {
+	const char *name;
+	size_t size;
+	int a_off; /* bytes */
+	int n_a;
+	int first_a;
+	int b_off;
+	int n_b;
+	int c_off;
+	int n_c;
+} oa_formats[I915_OA_FORMAT_MAX] = {
+	[I915_OA_FORMAT_A13] = {
+		"A13", .size = 64,
+		.a_off = 12, .n_a = 13 },
+	[I915_OA_FORMAT_A29] = {
+		"A29", .size = 128,
+		.a_off = 12, .n_a = 29 },
+	[I915_OA_FORMAT_A13_B8_C8] = {
+		"A13_B8_C8", .size = 128,
+		.a_off = 12, .n_a = 13,
+		.b_off = 64, .n_b = 8,
+		.c_off = 96, .n_c = 8 },
+	[I915_OA_FORMAT_A45_B8_C8] = {
+		"A45_B8_C8", .size = 256,
+		.a_off = 12,  .n_a = 45,
+		.b_off = 192, .n_b = 8,
+		.c_off = 224, .n_c = 8 },
+	[I915_OA_FORMAT_B4_C8] = {
+		"B4_C8", .size = 64,
+		.b_off = 16, .n_b = 4,
+		.c_off = 32, .n_c = 8 },
+	[I915_OA_FORMAT_B4_C8_A16] = {
+		"B4_C8_A16", .size = 128,
+		.b_off = 16, .n_b = 4,
+		.c_off = 32, .n_c = 8,
+		.a_off = 60, .n_a = 16, .first_a = 29 },
+	[I915_OA_FORMAT_C4_B8] = {
+		"C4_B8", .size = 64,
+		.c_off = 16, .n_c = 4,
+		.b_off = 28, .n_b = 8 },
+};
+
+static bool hsw_undefined_a_counters[45] = {
+	[4] = true,
+	[6] = true,
+	[9] = true,
+	[11] = true,
+	[14] = true,
+	[16] = true,
+	[19] = true,
+	[21] = true,
+	[24] = true,
+	[26] = true,
+	[29] = true,
+	[31] = true,
+	[34] = true,
+	[43] = true,
+	[44] = true,
+};
+
+static int drm_fd = -1;
+static uint32_t devid;
+static int device = -1;
+
+static uint64_t hsw_render_basic_id = UINT64_MAX;
+static uint64_t gt_min_freq_mhz_saved = 0;
+static uint64_t gt_max_freq_mhz_saved = 0;
+static uint64_t gt_min_freq_mhz = 0;
+static uint64_t gt_max_freq_mhz = 0;
+
+static uint64_t timestamp_frequency = 12500000;
+
+static igt_render_copyfunc_t render_copy = NULL;
+
+static int
+__perf_open(int fd, struct drm_i915_perf_open_param *param)
+{
+	int ret = igt_ioctl(fd, DRM_IOCTL_I915_PERF_OPEN, param);
+
+	igt_assert(ret >= 0);
+	errno = 0;
+
+	return ret;
+}
+
+static int
+lookup_format(int i915_perf_fmt_id)
+{
+	igt_assert(i915_perf_fmt_id < I915_OA_FORMAT_MAX);
+	igt_assert(oa_formats[i915_perf_fmt_id].name);
+
+	return i915_perf_fmt_id;
+}
+
+static bool
+try_read_u64_file(const char *file, uint64_t *val)
+{
+	char buf[32];
+	int fd, n;
+
+	fd = open(file, O_RDONLY);
+	if (fd < 0)
+		return false;
+
+	while ((n = read(fd, buf, sizeof(buf) - 1)) < 0 && errno == EINTR)
+		;
+	igt_assert(n >= 0);
+
+	close(fd);
+
+	buf[n] = '\0';
+	*val = strtoull(buf, NULL, 0);
+
+	return true;
+}
+
+static uint64_t
+read_u64_file(const char *file)
+{
+	uint64_t val;
+
+	igt_assert_eq(try_read_u64_file(file, &val), true);
+
+	return val;
+}
+
+static void
+write_u64_file(const char *file, uint64_t val)
+{
+	char buf[32];
+	int fd, len, ret;
+
+	fd = open(file, O_WRONLY);
+	igt_assert(fd >= 0);
+
+	len = snprintf(buf, sizeof(buf), "%"PRIu64, val);
+	igt_assert(len > 0);
+
+	while ((ret = write(fd, buf, len)) < 0 && errno == EINTR)
+		;
+	igt_assert_eq(ret, len);
+
+	close(fd);
+}
+
+static uint64_t
+sysfs_read(const char *file)
+{
+	char buf[512];
+
+	snprintf(buf, sizeof(buf), "/sys/class/drm/card%d/%s", device, file);
+
+	return read_u64_file(buf);
+}
+
+static void
+sysfs_write(const char *file, uint64_t val)
+{
+	char buf[512];
+
+	snprintf(buf, sizeof(buf), "/sys/class/drm/card%d/%s", device, file);
+
+	write_u64_file(buf, val);
+}
+
+static char *
+read_debugfs_record(const char *file, const char *key)
+{
+	FILE *fp;
+	char *line = NULL;
+	size_t line_buf_size = 0;
+	int len = 0;
+	int key_len = strlen(key);
+	char *value = NULL;
+
+	fp = igt_debugfs_fopen(file, "r");
+	igt_require(fp);
+
+	while ((len = getline(&line, &line_buf_size, fp)) > 0) {
+
+		if (line[len - 1] == '\n')
+			line[len - 1] = '\0';
+
+		if (strncmp(key, line, key_len) == 0 &&
+		    line[key_len] == ':' &&
+		    line[key_len + 1] == ' ')
+		{
+			value = strdup(line + key_len + 2);
+			goto done;
+		}
+	}
+
+	igt_assert(!"reached");
+done:
+	free(line);
+	if (fp)
+		fclose(fp);
+	return value;
+}
+
+static uint64_t
+read_debugfs_u64_record(const char *file, const char *key)
+{
+	char *str_val = read_debugfs_record(file, key);
+	uint64_t val;
+
+	igt_require(str_val);
+
+	val = strtoull(str_val, NULL, 0);
+	free(str_val);
+
+	return val;
+}
+
+static bool
+lookup_hsw_render_basic_id(void)
+{
+	char buf[256];
+
+	igt_assert_neq(device, -1);
+
+	snprintf(buf, sizeof(buf),
+		 "/sys/class/drm/card%d/metrics/403d8832-1a27-4aa6-a64e-f5389ce7b212/id",
+		 device);
+
+	return try_read_u64_file(buf, &hsw_render_basic_id);
+}
+
+static void
+gt_frequency_range_save(void)
+{
+	gt_min_freq_mhz_saved = sysfs_read("gt_min_freq_mhz");
+	gt_max_freq_mhz_saved = sysfs_read("gt_max_freq_mhz");
+
+	gt_min_freq_mhz = gt_min_freq_mhz_saved;
+	gt_max_freq_mhz = gt_max_freq_mhz_saved;
+}
+
+static void
+gt_frequency_pin(int gt_freq_mhz)
+{
+	igt_debug("requesting pinned GT freq = %dmhz\n", gt_freq_mhz);
+
+	if (gt_freq_mhz > gt_max_freq_mhz) {
+		sysfs_write("gt_max_freq_mhz", gt_freq_mhz);
+		sysfs_write("gt_min_freq_mhz", gt_freq_mhz);
+	} else {
+		sysfs_write("gt_min_freq_mhz", gt_freq_mhz);
+		sysfs_write("gt_max_freq_mhz", gt_freq_mhz);
+	}
+	gt_min_freq_mhz = gt_freq_mhz;
+	gt_max_freq_mhz = gt_freq_mhz;
+}
+
+static void
+gt_frequency_range_restore(void)
+{
+	igt_debug("restoring GT frequency range: min = %dmhz, max =%dmhz, current: min=%dmhz, max=%dmhz\n",
+		  (int)gt_min_freq_mhz_saved,
+		  (int)gt_max_freq_mhz_saved,
+		  (int)gt_min_freq_mhz,
+		  (int)gt_max_freq_mhz);
+
+	/* Assume current min/max are the same */
+	if (gt_min_freq_mhz_saved > gt_max_freq_mhz) {
+		sysfs_write("gt_max_freq_mhz", gt_max_freq_mhz_saved);
+		sysfs_write("gt_min_freq_mhz", gt_min_freq_mhz_saved);
+	} else {
+		sysfs_write("gt_min_freq_mhz", gt_min_freq_mhz_saved);
+		sysfs_write("gt_max_freq_mhz", gt_max_freq_mhz_saved);
+	}
+
+	gt_min_freq_mhz = gt_min_freq_mhz_saved;
+	gt_max_freq_mhz = gt_max_freq_mhz_saved;
+}
+
+static uint64_t
+timebase_scale(uint32_t u32_delta)
+{
+	return ((uint64_t)u32_delta * NSEC_PER_SEC) / timestamp_frequency;
+}
+
+/* CAP_SYS_ADMIN is required to open system wide metrics, unless the system
+ * control parameter dev.i915.perf_stream_paranoid == 0 */
+static void
+test_system_wide_paranoid(void)
+{
+	igt_fork(child, 1) {
+		uint64_t properties[] = {
+			/* Include OA reports in samples */
+			DRM_I915_PERF_PROP_SAMPLE_OA, true,
+
+			/* OA unit configuration */
+			DRM_I915_PERF_PROP_OA_METRICS_SET, hsw_render_basic_id,
+			DRM_I915_PERF_PROP_OA_FORMAT, I915_OA_FORMAT_A45_B8_C8,
+			DRM_I915_PERF_PROP_OA_EXPONENT, 13, /* 1 millisecond */
+		};
+		struct drm_i915_perf_open_param param = {
+			.flags = I915_PERF_FLAG_FD_CLOEXEC |
+				I915_PERF_FLAG_FD_NONBLOCK,
+			.num_properties = sizeof(properties) / 16,
+			.properties_ptr = (uint64_t)properties,
+		};
+
+		write_u64_file("/proc/sys/dev/i915/perf_stream_paranoid", 1);
+
+		igt_drop_root();
+
+		do_ioctl_err(drm_fd, DRM_IOCTL_I915_PERF_OPEN, &param, EACCES);
+	}
+
+	igt_waitchildren();
+
+	igt_fork(child, 1) {
+		uint64_t properties[] = {
+			/* Include OA reports in samples */
+			DRM_I915_PERF_PROP_SAMPLE_OA, true,
+
+			/* OA unit configuration */
+			DRM_I915_PERF_PROP_OA_METRICS_SET, hsw_render_basic_id,
+			DRM_I915_PERF_PROP_OA_FORMAT, I915_OA_FORMAT_A45_B8_C8,
+			DRM_I915_PERF_PROP_OA_EXPONENT, 13, /* 1 millisecond */
+		};
+		struct drm_i915_perf_open_param param = {
+			.flags = I915_PERF_FLAG_FD_CLOEXEC |
+				I915_PERF_FLAG_FD_NONBLOCK,
+			.num_properties = sizeof(properties) / 16,
+			.properties_ptr = (uint64_t)properties,
+		};
+		int stream_fd;
+
+		write_u64_file("/proc/sys/dev/i915/perf_stream_paranoid", 0);
+
+		igt_drop_root();
+
+		stream_fd = __perf_open(drm_fd, &param);
+		close(stream_fd);
+	}
+
+	igt_waitchildren();
+
+	/* leave in paranoid state */
+	write_u64_file("/proc/sys/dev/i915/perf_stream_paranoid", 1);
+}
+
+static void
+test_invalid_open_flags(void)
+{
+	uint64_t properties[] = {
+		/* Include OA reports in samples */
+		DRM_I915_PERF_PROP_SAMPLE_OA, true,
+
+		/* OA unit configuration */
+		DRM_I915_PERF_PROP_OA_METRICS_SET, hsw_render_basic_id,
+		DRM_I915_PERF_PROP_OA_FORMAT, I915_OA_FORMAT_A45_B8_C8,
+		DRM_I915_PERF_PROP_OA_EXPONENT, 13, /* 1 millisecond */
+	};
+	struct drm_i915_perf_open_param param = {
+		.flags = ~0, /* Undefined flag bits set! */
+		.num_properties = sizeof(properties) / 16,
+		.properties_ptr = (uint64_t)properties,
+	};
+
+	do_ioctl_err(drm_fd, DRM_IOCTL_I915_PERF_OPEN, &param, EINVAL);
+}
+
+static void
+test_invalid_oa_metric_set_id(void)
+{
+	uint64_t properties[] = {
+		/* Include OA reports in samples */
+		DRM_I915_PERF_PROP_SAMPLE_OA, true,
+
+		/* OA unit configuration */
+		DRM_I915_PERF_PROP_OA_FORMAT, I915_OA_FORMAT_A45_B8_C8,
+		DRM_I915_PERF_PROP_OA_EXPONENT, 13, /* 1 millisecond */
+		DRM_I915_PERF_PROP_OA_METRICS_SET, UINT64_MAX,
+	};
+	struct drm_i915_perf_open_param param = {
+		.flags = I915_PERF_FLAG_FD_CLOEXEC |
+			I915_PERF_FLAG_FD_NONBLOCK,
+		.num_properties = sizeof(properties) / 16,
+		.properties_ptr = (uint64_t)properties,
+	};
+	int stream_fd;
+
+	do_ioctl_err(drm_fd, DRM_IOCTL_I915_PERF_OPEN, &param, EINVAL);
+
+	properties[ARRAY_SIZE(properties) - 1] = 0; /* ID 0 is also be reserved as invalid */
+	do_ioctl_err(drm_fd, DRM_IOCTL_I915_PERF_OPEN, &param, EINVAL);
+
+	/* Check that we aren't just seeing false positives... */
+	properties[ARRAY_SIZE(properties) - 1] = hsw_render_basic_id;
+	stream_fd = __perf_open(drm_fd, &param);
+	close(stream_fd);
+
+	/* There's no valid default OA metric set ID... */
+	param.num_properties--;
+	do_ioctl_err(drm_fd, DRM_IOCTL_I915_PERF_OPEN, &param, EINVAL);
+}
+
+static void
+test_invalid_oa_format_id(void)
+{
+	uint64_t properties[] = {
+		/* Include OA reports in samples */
+		DRM_I915_PERF_PROP_SAMPLE_OA, true,
+
+		/* OA unit configuration */
+		DRM_I915_PERF_PROP_OA_METRICS_SET, hsw_render_basic_id,
+		DRM_I915_PERF_PROP_OA_EXPONENT, 13, /* 1 millisecond */
+		DRM_I915_PERF_PROP_OA_FORMAT, UINT64_MAX,
+	};
+	struct drm_i915_perf_open_param param = {
+		.flags = I915_PERF_FLAG_FD_CLOEXEC |
+			I915_PERF_FLAG_FD_NONBLOCK,
+		.num_properties = sizeof(properties) / 16,
+		.properties_ptr = (uint64_t)properties,
+	};
+	int stream_fd;
+
+	do_ioctl_err(drm_fd, DRM_IOCTL_I915_PERF_OPEN, &param, EINVAL);
+
+	properties[ARRAY_SIZE(properties) - 1] = 0; /* ID 0 is also be reserved as invalid */
+	do_ioctl_err(drm_fd, DRM_IOCTL_I915_PERF_OPEN, &param, EINVAL);
+
+	/* Check that we aren't just seeing false positives... */
+	properties[ARRAY_SIZE(properties) - 1] = I915_OA_FORMAT_A45_B8_C8;
+	stream_fd = __perf_open(drm_fd, &param);
+	close(stream_fd);
+
+	/* There's no valid default OA format... */
+	param.num_properties--;
+	do_ioctl_err(drm_fd, DRM_IOCTL_I915_PERF_OPEN, &param, EINVAL);
+}
+
+static void
+test_missing_sample_flags(void)
+{
+	uint64_t properties[] = {
+		/* No _PROP_SAMPLE_xyz flags */
+
+		/* OA unit configuration */
+		DRM_I915_PERF_PROP_OA_METRICS_SET, hsw_render_basic_id,
+		DRM_I915_PERF_PROP_OA_EXPONENT, 13, /* 1 millisecond */
+		DRM_I915_PERF_PROP_OA_FORMAT, I915_OA_FORMAT_A45_B8_C8,
+	};
+	struct drm_i915_perf_open_param param = {
+		.flags = I915_PERF_FLAG_FD_CLOEXEC,
+		.num_properties = sizeof(properties) / 16,
+		.properties_ptr = (uint64_t)properties,
+	};
+
+	do_ioctl_err(drm_fd, DRM_IOCTL_I915_PERF_OPEN, &param, EINVAL);
+}
+
+static void
+read_2_oa_reports(int stream_fd,
+		  int format_id,
+		  int exponent,
+		  uint32_t *oa_report0,
+		  uint32_t *oa_report1,
+		  bool timer_only)
+{
+	size_t format_size = oa_formats[format_id].size;
+	size_t sample_size = (sizeof(struct drm_i915_perf_record_header) +
+			      format_size);
+	const struct drm_i915_perf_record_header *header;
+	uint32_t exponent_mask = (1 << (exponent + 1)) - 1;
+
+	/* Note: we allocate a large buffer so that each read() iteration
+	 * should scrape *all* pending records.
+	 *
+	 * The largest buffer the OA unit supports is 16MB and the smallest
+	 * OA report format is 64bytes allowing up to 262144 reports to
+	 * be buffered.
+	 *
+	 * Being sure we are fetching all buffered reports allows us to
+	 * potentially throw away / skip all reports whenever we see
+	 * a _REPORT_LOST notification as a way of being sure are
+	 * measurements aren't skewed by a lost report.
+	 *
+	 * Note: that is is useful for some tests but also not something
+	 * applications would be expected to resort to. Lost reports are
+	 * somewhat unpredictable but typically don't pose a problem - except
+	 * to indicate that the OA unit may be over taxed if lots of reports
+	 * are being lost.
+	 */
+	int buf_size = 262144 * (64 + sizeof(struct drm_i915_perf_record_header));
+	uint8_t *buf = malloc(buf_size);
+	int n = 0;
+
+	for (int i = 0; i < 1000; i++) {
+		ssize_t len;
+
+		while ((len = read(stream_fd, buf, buf_size)) < 0 &&
+		       errno == EINTR)
+			;
+
+		igt_assert(len > 0);
+
+		for (size_t offset = 0; offset < len; offset += header->size) {
+			const uint32_t *report;
+
+			header = (void *)(buf + offset);
+
+			igt_assert_eq(header->pad, 0); /* Reserved */
+
+			/* Currently the only test that should ever expect to
+			 * see a _BUFFER_LOST error is the buffer_fill test,
+			 * otherwise something bad has probably happened...
+			 */
+			igt_assert_neq(header->type, DRM_I915_PERF_RECORD_OA_BUFFER_LOST);
+
+			/* At high sampling frequencies the OA HW might not be
+			 * able to cope with all write requests and will notify
+			 * us that a report was lost. We restart our read of
+			 * two sequential reports due to the timeline blip this
+			 * implies
+			 */
+			if (header->type == DRM_I915_PERF_RECORD_OA_REPORT_LOST) {
+				igt_debug("read restart: OA trigger collision / report lost\n");
+				n = 0;
+
+				/* XXX: break, because we don't know where
+				 * within the series of already read reports
+				 * there could be a blip from the lost report.
+				 */
+				break;
+			}
+
+			/* Currently the only other record type expected is a
+			 * _SAMPLE. Notably this test will need updating if
+			 * i915-perf is extended in the future with additional
+			 * record types.
+			 */
+			igt_assert_eq(header->type, DRM_I915_PERF_RECORD_SAMPLE);
+
+			igt_assert_eq(header->size, sample_size);
+
+			report = (const void *)(header + 1);
+
+			igt_debug("read report: reason = %x, timestamp = %x, exponent mask=%x\n",
+				  report[0], report[1], exponent_mask);
+
+			/* Don't expect zero for timestamps */
+			igt_assert_neq(report[1], 0);
+
+			if (timer_only) {
+				/* For Haswell we don't have a documented
+				 * report reason field (though empirically
+				 * report[0] bit 10 does seem to correlate with
+				 * a timer trigger reason) so we instead infer
+				 * which reports are timer triggered by
+				 * checking if the least significant bits are
+				 * zero and the exponent bit is set.
+				 */
+				if ((report[1] & exponent_mask) != (1 << exponent)) {
+					igt_debug("skipping non timer report reason=%x\n",
+						  report[0]);
+
+					/* Also assert our hypothesis about the
+					 * reason bit...
+					 */
+					igt_assert_eq(report[0] & (1 << 10), 0);
+					continue;
+				}
+			}
+
+			if (n++ == 0)
+				memcpy(oa_report0, report, format_size);
+			else {
+				memcpy(oa_report1, report, format_size);
+				free(buf);
+				return;
+			}
+		}
+	}
+
+	free(buf);
+
+	igt_assert(!"reached");
+}
+
+static void
+open_and_read_2_oa_reports(int format_id,
+			   int exponent,
+			   uint32_t *oa_report0,
+			   uint32_t *oa_report1,
+			   bool timer_only)
+{
+	uint64_t properties[] = {
+		/* Include OA reports in samples */
+		DRM_I915_PERF_PROP_SAMPLE_OA, true,
+
+		/* OA unit configuration */
+		DRM_I915_PERF_PROP_OA_METRICS_SET, hsw_render_basic_id,
+		DRM_I915_PERF_PROP_OA_FORMAT, format_id,
+		DRM_I915_PERF_PROP_OA_EXPONENT, exponent,
+
+	};
+	struct drm_i915_perf_open_param param = {
+		.flags = I915_PERF_FLAG_FD_CLOEXEC,
+		.num_properties = sizeof(properties) / 16,
+		.properties_ptr = (uint64_t)properties,
+	};
+	int stream_fd = __perf_open(drm_fd, &param);
+
+	read_2_oa_reports(stream_fd, format_id, exponent,
+			  oa_report0, oa_report1, timer_only);
+
+	close(stream_fd);
+}
+
+static void
+print_reports(uint32_t *oa_report0, uint32_t *oa_report1, int fmt)
+{
+	uint32_t *a0, *b0, *c0;
+	uint32_t *a1, *b1, *c1;
+
+	/* Not ideal naming here with a0 or a1
+	 * differentiating report0 or 1 not A counter 0 or 1....
+	 */
+	a0 = (uint32_t *)(((uint8_t *)oa_report0) + oa_formats[fmt].a_off);
+	b0 = (uint32_t *)(((uint8_t *)oa_report0) + oa_formats[fmt].b_off);
+	c0 = (uint32_t *)(((uint8_t *)oa_report0) + oa_formats[fmt].c_off);
+
+	a1 = (uint32_t *)(((uint8_t *)oa_report1) + oa_formats[fmt].a_off);
+	b1 = (uint32_t *)(((uint8_t *)oa_report1) + oa_formats[fmt].b_off);
+	c1 = (uint32_t *)(((uint8_t *)oa_report1) + oa_formats[fmt].c_off);
+
+	igt_debug("TIMESTAMP: 1st = %"PRIu32", 2nd = %"PRIu32", delta = %"PRIu32"\n",
+		  oa_report0[1], oa_report1[1], oa_report1[1] - oa_report0[1]);
+
+	if (oa_formats[fmt].n_c) {
+		igt_debug("CLOCK: 1st = %"PRIu32", 2nd = %"PRIu32", delta = %"PRIu32"\n",
+			  c0[2], c1[2], c1[2] - c0[2]);
+	} else
+		igt_debug("CLOCK = N/A\n");
+
+	for (int j = oa_formats[fmt].first_a;
+	     j < oa_formats[fmt].n_a;
+	     j++)
+	{
+		uint32_t delta = a1[j] - a0[j];
+
+		if (hsw_undefined_a_counters[j])
+			continue;
+
+		igt_debug("A%d: 1st = %"PRIu32", 2nd = %"PRIu32", delta = %"PRIu32"\n",
+			  j, a0[j], a1[j], delta);
+	}
+
+	for (int j = 0; j < oa_formats[fmt].n_b; j++) {
+		uint32_t delta = b1[j] - b0[j];
+		igt_debug("B%d: 1st = %"PRIu32", 2nd = %"PRIu32", delta = %"PRIu32"\n",
+			  j, b0[j], b1[j], delta);
+	}
+
+	for (int j = 0; j < oa_formats[fmt].n_c; j++) {
+		uint32_t delta = c1[j] - c0[j];
+		igt_debug("C%d: 1st = %"PRIu32", 2nd = %"PRIu32", delta = %"PRIu32"\n",
+			  j, c0[j], c1[j], delta);
+	}
+}
+
+static void
+test_oa_formats(void)
+{
+	int oa_exponent = 13;
+
+	for (int i = 0; i < ARRAY_SIZE(oa_formats); i++) {
+		uint32_t oa_report0[64];
+		uint32_t oa_report1[64];
+		uint32_t *a0, *b0, *c0;
+		uint32_t *a1, *b1, *c1;
+		uint32_t time_delta;
+		uint32_t clock_delta;
+		uint32_t max_delta;
+
+		if (!oa_formats[i].name) /* sparse, indexed by ID */
+			continue;
+
+		igt_debug("Checking OA format %s\n", oa_formats[i].name);
+
+		open_and_read_2_oa_reports(i,
+					   oa_exponent,
+					   oa_report0,
+					   oa_report1,
+					   false); /* timer reports only */
+
+		print_reports(oa_report0, oa_report1, i);
+
+		a0 = (uint32_t *)(((uint8_t *)oa_report0) + oa_formats[i].a_off);
+		b0 = (uint32_t *)(((uint8_t *)oa_report0) + oa_formats[i].b_off);
+		c0 = (uint32_t *)(((uint8_t *)oa_report0) + oa_formats[i].c_off);
+
+		a1 = (uint32_t *)(((uint8_t *)oa_report1) + oa_formats[i].a_off);
+		b1 = (uint32_t *)(((uint8_t *)oa_report1) + oa_formats[i].b_off);
+		c1 = (uint32_t *)(((uint8_t *)oa_report1) + oa_formats[i].c_off);
+
+		time_delta = timebase_scale(oa_report1[1] - oa_report0[1]);
+		igt_assert_neq(time_delta, 0);
+
+		/* C2 corresponds to a clock counter for the Haswell render
+		 * basic metric set but it's not included in all of the
+		 * formats.
+		 */
+		if (oa_formats[i].n_c) {
+			uint64_t freq;
+
+			/* The first report might have a clock count of zero
+			 * but we wouldn't expect that in the second report...
+			 */
+			igt_assert_neq(c1[2], 0);
+
+			clock_delta = c1[2] - c0[2];
+			igt_assert_neq(clock_delta, 0);
+
+			freq = ((uint64_t)clock_delta * 1000) / time_delta;
+			igt_debug("freq = %"PRIu64"\n", freq);
+
+			igt_assert(freq <= gt_max_freq_mhz);
+		} else {
+			/* Assume running at max freq for sake of
+			 * below sanity check on counters... */
+			clock_delta = (gt_max_freq_mhz *
+				       (uint64_t)time_delta) / 1000;
+		}
+
+		igt_debug("clock delta = %"PRIu32"\n", clock_delta);
+
+		/* The maximum rate for any HSW counter =
+		 *   clock_delta * 40 EUs
+		 *
+		 * Sanity check that no counters exceed this delta.
+		 */
+		max_delta = clock_delta * 40;
+
+		for (int j = oa_formats[i].first_a;
+		     j < oa_formats[i].n_a;
+		     j++)
+		{
+			uint32_t delta = a1[j] - a0[j];
+
+			if (hsw_undefined_a_counters[j])
+				continue;
+
+			igt_debug("A%d: delta = %"PRIu32"\n", j, delta);
+			igt_assert(delta <= max_delta);
+		}
+
+		for (int j = 0; j < oa_formats[i].n_b; j++) {
+			uint32_t delta = b1[j] - b0[j];
+			igt_debug("B%d: delta = %"PRIu32"\n", j, delta);
+			igt_assert(delta <= max_delta);
+		}
+
+		for (int j = 0; j < oa_formats[i].n_c; j++) {
+			uint32_t delta = c1[j] - c0[j];
+			igt_debug("C%d: delta = %"PRIu32"\n", j, delta);
+			igt_assert(delta <= max_delta);
+		}
+	}
+}
+
+static void
+test_oa_exponents(int gt_freq_mhz)
+{
+	/* This test tries to use the sysfs interface for pinning the GT
+	 * frequency so we have another point of reference for comparing with
+	 * the clock frequency as derived from OA reports.
+	 *
+	 * This test has been finicky to stabilise while the
+	 * gt_min/max_freq_mhz files in sysfs don't seem to be a reliable
+	 * mechanism for fixing the gpu frequency.
+	 *
+	 * Since these unit tests are focused on the OA unit not the ability to
+	 * pin the frequency via sysfs we make the test account for pinning not
+	 * being reliable and read back the current frequency for each
+	 * iteration of this test to take this into account.
+	 */
+	gt_frequency_pin(gt_freq_mhz);
+
+	igt_debug("Testing OA timer exponents with requested GT frequency = %dmhz\n",
+		  gt_freq_mhz);
+
+	/* It's asking a lot to sample with a 160 nanosecond period and the
+	 * test can fail due to buffer overflows if it wasn't possible to
+	 * keep up, so we don't start from an exponent of zero...
+	 */
+	for (int i = 2; i < 20; i++) {
+		uint32_t expected_timestamp_delta;
+		uint32_t timestamp_delta;
+		uint32_t oa_report0[64];
+		uint32_t oa_report1[64];
+		uint32_t *c0, *c1;
+		uint32_t time_delta;
+		uint32_t clock_delta;
+		uint32_t freq;
+		int n_tested = 0;
+		int n_freq_matches = 0;
+
+		/* The exponent is effectively selecting a bit in the timestamp
+		 * to trigger reports on and so in practice we expect the raw
+		 * timestamp deltas for periodic reports to exactly match the
+		 * value of next bit.
+		 */
+		expected_timestamp_delta = 2 << i;
+
+		for (int j = 0; n_tested < 10 && j < 100; j++) {
+			int gt_freq_mhz_0, gt_freq_mhz_1;
+			int c_off;
+
+			gt_freq_mhz_0 = sysfs_read("gt_act_freq_mhz");
+
+			igt_debug("ITER %d: testing OA exponent %d with sysfs GT freq = %dmhz\n",
+				  j, i, gt_freq_mhz_0);
+
+			open_and_read_2_oa_reports(I915_OA_FORMAT_A45_B8_C8,
+						   i, /* exponent */
+						   oa_report0,
+						   oa_report1,
+						   true); /* timer triggered
+							     reports only */
+
+			gt_freq_mhz_1 = sysfs_read("gt_act_freq_mhz");
+
+			/* If it looks like the frequency has changed according
+			 * to sysfs then skip looking at this pair of reports
+			 */
+			if (gt_freq_mhz_0 != gt_freq_mhz_1) {
+				igt_debug("skipping OA reports pair due to GT frequency change according to sysfs\n");
+				continue;
+			}
+
+			timestamp_delta = oa_report1[1] - oa_report0[1];
+			igt_assert_neq(timestamp_delta, 0);
+
+			if (timestamp_delta != expected_timestamp_delta) {
+				igt_debug("timestamp0 = %u/0x%x\n",
+					  oa_report0[1], oa_report0[1]);
+				igt_debug("timestamp1 = %u/0x%x\n",
+					  oa_report1[1], oa_report1[1]);
+			}
+
+			igt_assert_eq(timestamp_delta, expected_timestamp_delta);
+
+			/* NB: for the render basic metric set opened above by
+			 * open_and_read_2_oa_reports(), the C2 counter is
+			 * configured as the gpu clock counter...
+			 */
+			c_off = oa_formats[I915_OA_FORMAT_A45_B8_C8].c_off;
+			igt_assert(c_off);
+			c0 = (uint32_t *)(((uint8_t *)oa_report0) + c_off);
+			c1 = (uint32_t *)(((uint8_t *)oa_report1) + c_off);
+			clock_delta = c1[2] - c0[2];
+
+			time_delta = timebase_scale(timestamp_delta);
+
+			freq = ((uint64_t)clock_delta * 1000) / time_delta;
+			igt_debug("ITER %d: time delta = %"PRIu32"(ns) clock delta = %"PRIu32" freq = %"PRIu32"(mhz)\n",
+				  j, time_delta, clock_delta, freq);
+
+			if (freq == gt_freq_mhz_1)
+				n_freq_matches++;
+
+			n_tested++;
+		}
+
+		if (n_tested < 10)
+			igt_debug("sysfs frequency pinning too unstable for cross-referencing with OA derived frequency");
+		igt_assert_eq(n_tested, 10);
+
+		igt_debug("number of iterations with expected clock frequency = %d\n",
+			  n_freq_matches);
+
+		/* Don't assert the calculated frequency for extremely short
+		 * durations.
+		 *
+		 * Allow some mismatches since can't be can't be sure about
+		 * frequency changes between sysfs reads.
+		 */
+		if (i > 3)
+			igt_assert(n_freq_matches >= 7);
+	}
+
+	gt_frequency_range_restore();
+}
+
+/* The OA exponent selects a timestamp counter bit to trigger reports on.
+ *
+ * With a 64bit timestamp and least significant bit approx == 80ns then the MSB
+ * equates to > 40 thousand years and isn't exposed via the i915 perf interface.
+ *
+ * The max exponent exposed is expected to be 31, which is still a fairly
+ * ridiculous period (>5min) but is the maximum exponent where it's still
+ * possible to use periodic sampling as a means for tracking the overflow of
+ * 32bit OA report timestamps.
+ */
+static void
+test_invalid_oa_exponent(void)
+{
+	uint64_t properties[] = {
+		/* Include OA reports in samples */
+		DRM_I915_PERF_PROP_SAMPLE_OA, true,
+
+		/* OA unit configuration */
+		DRM_I915_PERF_PROP_OA_METRICS_SET, hsw_render_basic_id,
+		DRM_I915_PERF_PROP_OA_FORMAT, I915_OA_FORMAT_A45_B8_C8,
+		DRM_I915_PERF_PROP_OA_EXPONENT, 31, /* maximum exponent expected
+						       to be accepted */
+	};
+	struct drm_i915_perf_open_param param = {
+		.flags = I915_PERF_FLAG_FD_CLOEXEC,
+		.num_properties = sizeof(properties) / 16,
+		.properties_ptr = (uint64_t)properties,
+	};
+	int stream_fd = __perf_open(drm_fd, &param);
+
+	close(stream_fd);
+
+	for (int i = 32; i < 65; i++) {
+		properties[7] = i;
+		do_ioctl_err(drm_fd, DRM_IOCTL_I915_PERF_OPEN, &param, EINVAL);
+	}
+}
+
+/* Return the largest OA exponent that will still result in a sampling
+ * frequency higher than the given frequency.
+ */
+static int
+max_oa_exponent_for_higher_freq(uint64_t freq)
+{
+	/* NB: timebase_scale() takes a uint32_t and an exponent of 30
+	 * would already represent a period of ~3 minutes so there's
+	 * really no need to consider higher exponents.
+	 */
+	for (int i = 0; i < 30; i++) {
+		uint64_t oa_period = timebase_scale(2 << i);
+		uint32_t oa_freq = NSEC_PER_SEC / oa_period;
+
+		if (oa_freq <= freq)
+			return max(0, i - 1);
+	}
+
+	igt_assert(!"reached");
+	return -1;
+}
+
+/* The lowest periodic sampling exponent equates to a period of 160 nanoseconds
+ * or a frequency of 6.25MHz which is only possible to request as root by
+ * default. By default the maximum OA sampling rate is 100KHz
+ */
+static void
+test_low_oa_exponent_permissions(void)
+{
+	int max_freq = read_u64_file("/proc/sys/dev/i915/oa_max_sample_rate");
+	int bad_exponent = max_oa_exponent_for_higher_freq(max_freq);
+	int ok_exponent = bad_exponent + 1;
+	uint64_t properties[] = {
+		/* Include OA reports in samples */
+		DRM_I915_PERF_PROP_SAMPLE_OA, true,
+
+		/* OA unit configuration */
+		DRM_I915_PERF_PROP_OA_METRICS_SET, hsw_render_basic_id,
+		DRM_I915_PERF_PROP_OA_FORMAT, I915_OA_FORMAT_A45_B8_C8,
+		DRM_I915_PERF_PROP_OA_EXPONENT, bad_exponent,
+	};
+	struct drm_i915_perf_open_param param = {
+		.flags = I915_PERF_FLAG_FD_CLOEXEC,
+		.num_properties = sizeof(properties) / 16,
+		.properties_ptr = (uint64_t)properties,
+	};
+	uint64_t oa_period, oa_freq;
+
+	igt_assert_eq(max_freq, 100000);
+
+	/* Avoid EACCES errors opening a stream without CAP_SYS_ADMIN */
+	write_u64_file("/proc/sys/dev/i915/perf_stream_paranoid", 0);
+
+	igt_fork(child, 1) {
+		igt_drop_root();
+
+		do_ioctl_err(drm_fd, DRM_IOCTL_I915_PERF_OPEN, &param, EACCES);
+	}
+
+	igt_waitchildren();
+
+	properties[7] = ok_exponent;
+
+	igt_fork(child, 1) {
+		int stream_fd;
+
+		igt_drop_root();
+
+		stream_fd = __perf_open(drm_fd, &param);
+		close(stream_fd);
+	}
+
+	igt_waitchildren();
+
+	oa_period = timebase_scale(2 << ok_exponent);
+	oa_freq = NSEC_PER_SEC / oa_period;
+	write_u64_file("/proc/sys/dev/i915/oa_max_sample_rate", oa_freq - 100);
+
+	igt_fork(child, 1) {
+		igt_drop_root();
+
+		do_ioctl_err(drm_fd, DRM_IOCTL_I915_PERF_OPEN, &param, EACCES);
+	}
+
+	igt_waitchildren();
+
+	/* restore the defaults */
+	write_u64_file("/proc/sys/dev/i915/oa_max_sample_rate", 100000);
+	write_u64_file("/proc/sys/dev/i915/perf_stream_paranoid", 1);
+}
+
+static void
+test_per_context_mode_unprivileged(void)
+{
+	uint64_t properties[] = {
+		/* Single context sampling */
+		DRM_I915_PERF_PROP_CTX_HANDLE, UINT64_MAX, /* updated below */
+
+		/* Include OA reports in samples */
+		DRM_I915_PERF_PROP_SAMPLE_OA, true,
+
+		/* OA unit configuration */
+		DRM_I915_PERF_PROP_OA_METRICS_SET, hsw_render_basic_id,
+		DRM_I915_PERF_PROP_OA_FORMAT, I915_OA_FORMAT_A45_B8_C8,
+		DRM_I915_PERF_PROP_OA_EXPONENT, 13, /* 1 millisecond */
+	};
+	struct drm_i915_perf_open_param param = {
+		.flags = I915_PERF_FLAG_FD_CLOEXEC,
+		.num_properties = sizeof(properties) / 16,
+		.properties_ptr = (uint64_t)properties,
+	};
+
+	/* should be default, but just to be sure... */
+	write_u64_file("/proc/sys/dev/i915/perf_stream_paranoid", 1);
+
+	igt_fork(child, 1) {
+		drm_intel_context *context;
+		drm_intel_bufmgr *bufmgr;
+		int stream_fd;
+
+		igt_drop_root();
+
+		bufmgr = drm_intel_bufmgr_gem_init(drm_fd, 4096);
+		context = drm_intel_gem_context_create(bufmgr);
+
+		igt_assert(context);
+
+		properties[1] = drm_intel_gem_context_get_context_id(context);
+
+		stream_fd = __perf_open(drm_fd, &param);
+		close(stream_fd);
+
+		drm_intel_gem_context_destroy(context);
+		drm_intel_bufmgr_destroy(bufmgr);
+	}
+
+	igt_waitchildren();
+}
+
+static int64_t
+get_time(void)
+{
+	struct timespec ts;
+
+	clock_gettime(CLOCK_MONOTONIC, &ts);
+
+	return ts.tv_sec * 1000000000 + ts.tv_nsec;
+}
+
+/* Note: The interface doesn't currently provide strict guarantees or control
+ * over the upper bound for how long it might take for a POLLIN event after
+ * some OA report is written by the OA unit.
+ *
+ * The plan is to add a property later that gives some control over the maximum
+ * latency, but for now we expect it is tuned for a fairly low latency
+ * suitable for applications wanting to provide live feedback for captured
+ * metrics.
+ *
+ * At the time of writing this test the driver was using a fixed 200Hz hrtimer
+ * regardless of the OA sampling exponent.
+ *
+ * There is no lower bound since a stream configured for periodic sampling may
+ * still contain other automatically triggered reports.
+ *
+ * What we try and check for here is that blocking reads don't return EAGAIN
+ * and that we aren't spending any significant time burning the cpu in
+ * kernelspace.
+ */
+static void
+test_blocking(void)
+{
+	/* 40 milliseconds
+	 *
+	 * Having a period somewhat > sysconf(_SC_CLK_TCK) helps to stop
+	 * scheduling (liable to kick in when we make blocking poll()s/reads)
+	 * from interfering with the test.
+	 */
+	int oa_exponent = 18;
+	uint64_t properties[] = {
+		/* Include OA reports in samples */
+		DRM_I915_PERF_PROP_SAMPLE_OA, true,
+
+		/* OA unit configuration */
+		DRM_I915_PERF_PROP_OA_METRICS_SET, hsw_render_basic_id,
+		DRM_I915_PERF_PROP_OA_FORMAT, I915_OA_FORMAT_A45_B8_C8,
+		DRM_I915_PERF_PROP_OA_EXPONENT, oa_exponent,
+	};
+	struct drm_i915_perf_open_param param = {
+		.flags = I915_PERF_FLAG_FD_CLOEXEC,
+		.num_properties = sizeof(properties) / 16,
+		.properties_ptr = (uint64_t)properties,
+	};
+	int stream_fd = __perf_open(drm_fd, &param);
+	uint8_t buf[1024 * 1024];
+	struct tms start_times;
+	struct tms end_times;
+	int64_t user_ns, kernel_ns;
+	int64_t tick_ns = 1000000000 / sysconf(_SC_CLK_TCK);
+	int64_t start;
+	int n = 0;
+
+	times(&start_times);
+
+	/* Loop for 600ms performing blocking reads while the HW is sampling at
+	 * ~25Hz, with the expectation that we spend most of our time blocked
+	 * in the kernel, and shouldn't be burning cpu cycles in the kernel in
+	 * association with this process (verified by looking at stime before
+	 * and after loop).
+	 */
+	for (start = get_time(); (get_time() - start) < 600000000; /* nop */) {
+		int ret;
+
+		while ((ret = read(stream_fd, buf, sizeof(buf))) < 0 &&
+		       errno == EINTR)
+			;
+
+		igt_assert(ret > 0);
+
+		n++;
+	}
+
+	times(&end_times);
+
+	/* Using nanosecond units is fairly silly here, given the tick in-
+	 * precision - ah well, it's consistent with the get_time() units.
+	 */
+	user_ns = (end_times.tms_utime - start_times.tms_utime) * tick_ns;
+	kernel_ns = (end_times.tms_stime - start_times.tms_stime) * tick_ns;
+
+	igt_debug("%d blocking reads in 500 milliseconds, with 1KHz OA sampling\n", n);
+	igt_debug("time in userspace = %"PRIu64"ns (start utime = %d, end = %d, ns ticks per sec = %d)\n",
+		  user_ns, (int)start_times.tms_utime, (int)end_times.tms_utime, (int)tick_ns);
+	igt_debug("time in kernelspace = %"PRIu64"ns (start stime = %d, end = %d, ns ticks per sec = %d)\n",
+		  kernel_ns, (int)start_times.tms_stime, (int)end_times.tms_stime, (int)tick_ns);
+
+	/* With completely broken blocking (but also not returning an error) we
+	 * could end up with an open loop, hopefully recognisable with > 15
+	 * (600/40)iterations.
+	 */
+	igt_assert(n <= 15);
+
+	/* It's a bit tricky to put a lower limit here, but we expect a
+	 * relatively low latency for seeing reports, while we don't currently
+	 * give any control over this in the api.
+	 *
+	 * Limited to a 5 millisecond latency and 45ms (worst case)
+	 * per-iteration that could give 13.3 iterations. Rounding gives a tiny
+	 * bit more latency slack (6ms)...
+	 */
+	igt_assert(n > 13);
+
+	/* A bit tricky to put a number on this, but we don't expect the kernel
+	 * to use any significant cpu while waiting and given the in precision
+	 * of stime (multiple of CLK_TCK) we expect this to round to zero.
+	 */
+	igt_assert_eq(kernel_ns, 0);
+
+	close(stream_fd);
+}
+
+static void
+test_polling(void)
+{
+	/* 40 milliseconds
+	 *
+	 * Having a period somewhat > sysconf(_SC_CLK_TCK) helps to stop
+	 * scheduling (liable to kick in when we make blocking poll()s/reads)
+	 * from interfering with the test.
+	 */
+	int oa_exponent = 18;
+	uint64_t properties[] = {
+		/* Include OA reports in samples */
+		DRM_I915_PERF_PROP_SAMPLE_OA, true,
+
+		/* OA unit configuration */
+		DRM_I915_PERF_PROP_OA_METRICS_SET, hsw_render_basic_id,
+		DRM_I915_PERF_PROP_OA_FORMAT, I915_OA_FORMAT_A45_B8_C8,
+		DRM_I915_PERF_PROP_OA_EXPONENT, oa_exponent,
+	};
+	struct drm_i915_perf_open_param param = {
+		.flags = I915_PERF_FLAG_FD_CLOEXEC |
+			I915_PERF_FLAG_FD_NONBLOCK,
+		.num_properties = sizeof(properties) / 16,
+		.properties_ptr = (uint64_t)properties,
+	};
+	int stream_fd = __perf_open(drm_fd, &param);
+	uint8_t buf[1024 * 1024];
+	struct tms start_times;
+	struct tms end_times;
+	int64_t user_ns, kernel_ns;
+	int64_t tick_ns = 1000000000 / sysconf(_SC_CLK_TCK);
+	int64_t start;
+	int n = 0;
+
+	times(&start_times);
+
+	/* Loop for 600ms performing blocking polls while the HW is sampling at
+	 * ~25Hz, with the expectation that we spend most of our time blocked
+	 * in the kernel, and shouldn't be burning cpu cycles in the kernel in
+	 * association with this process (verified by looking at stime before
+	 * and after loop).
+	 */
+	for (start = get_time(); (get_time() - start) < 600000000; /* nop */) {
+		struct pollfd pollfd = { .fd = stream_fd, .events = POLLIN };
+		int ret;
+
+		while ((ret = poll(&pollfd, 1, -1)) < 0 &&
+		       errno == EINTR)
+			;
+		igt_assert_eq(ret, 1);
+		igt_assert(pollfd.revents & POLLIN);
+
+		while ((ret = read(stream_fd, buf, sizeof(buf))) < 0 &&
+		       errno == EINTR)
+			;
+
+		/* Don't expect to see EAGAIN if we've had a POLLIN event
+		 *
+		 * XXX: actually this is technically overly strict since we do
+		 * knowingly allow false positive POLLIN events. At least in
+		 * the future when supporting context filtering of metrics for
+		 * Gen8+ handled in the kernel then POLLIN events may be
+		 * delivered when we know there are pending reports to process
+		 * but before we've done any filtering to know for certain that
+		 * any reports are destined to be copied to userspace.
+		 *
+		 * Still, for now it's a reasonable sanity check.
+		 */
+		if (ret < 0)
+			igt_debug("Unexpected error when reading after poll = %d\n", errno);
+		igt_assert_neq(ret, -1);
+
+		/* At this point, after consuming pending reports (and hoping
+		 * the scheduler hasn't stopped us for too long we now
+		 * expect EAGAIN on read.
+		 */
+		while ((ret = read(stream_fd, buf, sizeof(buf))) < 0 &&
+		       errno == EINTR)
+			;
+		igt_assert_eq(ret, -1);
+		igt_assert_eq(errno, EAGAIN);
+
+		n++;
+	}
+
+	times(&end_times);
+
+	/* Using nanosecond units is fairly silly here, given the tick in-
+	 * precision - ah well, it's consistent with the get_time() units.
+	 */
+	user_ns = (end_times.tms_utime - start_times.tms_utime) * tick_ns;
+	kernel_ns = (end_times.tms_stime - start_times.tms_stime) * tick_ns;
+
+	igt_debug("%d blocking poll()s in 600 milliseconds, with 25Hz OA sampling\n", n);
+	igt_debug("time in userspace = %"PRIu64"ns (start utime = %d, end = %d, ns ticks per sec = %d)\n",
+		  user_ns, (int)start_times.tms_utime, (int)end_times.tms_utime, (int)tick_ns);
+	igt_debug("time in kernelspace = %"PRIu64"ns (start stime = %d, end = %d, ns ticks per sec = %d)\n",
+		  kernel_ns, (int)start_times.tms_stime, (int)end_times.tms_stime, (int)tick_ns);
+
+	/* With completely broken blocking while polling (but still somehow
+	 * reporting a POLLIN event) we could end up with an open loop,
+	 * hopefully recognisable with > 15 (600/40)iterations.
+	 */
+	igt_assert(n <= 15);
+
+	/* It's a bit tricky to put a lower limit here, but we expect a
+	 * relatively low latency for seeing reports, while we don't currently
+	 * give any control over this in the api.
+	 *
+	 * Limited to a 5 millisecond latency and 45ms (worst case)
+	 * per-iteration that could give 13.3 iterations. Rounding gives a tiny
+	 * bit more latency slack (6ms)...
+	 */
+	igt_assert(n > 13);
+
+	/* A bit tricky to put a number on this, but we don't expect the kernel
+	 * to use any significant cpu while waiting and given the in precision
+	 * of stime (multiple of CLK_TCK) we expect this to round to zero.
+	 */
+	igt_assert_eq(kernel_ns, 0);
+
+	close(stream_fd);
+}
+
+static void
+test_buffer_fill(void)
+{
+	int oa_exponent = 5; /* 5 micro seconds */
+	uint64_t properties[] = {
+		/* Include OA reports in samples */
+		DRM_I915_PERF_PROP_SAMPLE_OA, true,
+
+		/* OA unit configuration */
+		DRM_I915_PERF_PROP_OA_METRICS_SET, hsw_render_basic_id,
+		DRM_I915_PERF_PROP_OA_FORMAT, I915_OA_FORMAT_A45_B8_C8,
+		DRM_I915_PERF_PROP_OA_EXPONENT, oa_exponent,
+	};
+	struct drm_i915_perf_open_param param = {
+		.flags = I915_PERF_FLAG_FD_CLOEXEC,
+		.num_properties = sizeof(properties) / 16,
+		.properties_ptr = (uint64_t)properties,
+	};
+	int stream_fd = __perf_open(drm_fd, &param);
+	int buf_size = 65536 * (256 + sizeof(struct drm_i915_perf_record_header));
+	uint8_t *buf = malloc(buf_size);
+
+
+	for (int i = 0; i < 5; i++) {
+		struct drm_i915_perf_record_header *header;
+		bool overflow_seen;
+		int offset = 0;
+		int len;
+
+		/* It should take ~330 milliseconds to fill a 16MB OA buffer with a
+		 * 5 microsecond sampling period and 256 byte reports. */
+		nanosleep(&(struct timespec){ .tv_sec = 0, .tv_nsec = 500000000 }, NULL);
+
+		while ((len = read(stream_fd, buf, buf_size)) == -1 && errno == EINTR)
+			;
+
+		igt_assert_neq(len, -1);
+
+		overflow_seen = false;
+		for (offset = 0; offset < len; offset += header->size) {
+			header = (void *)(buf + offset);
+
+			if (header->type == DRM_I915_PERF_RECORD_OA_BUFFER_LOST)
+				overflow_seen = true;
+		}
+
+		igt_assert_eq(overflow_seen, true);
+
+		nanosleep(&(struct timespec){ .tv_sec = 0, .tv_nsec = 1000000 }, NULL);
+
+		while ((len = read(stream_fd, buf, buf_size)) == -1 && errno == EINTR)
+			;
+
+		igt_assert_neq(len, -1);
+
+		/* expect ~ 200 records in 1 millisecond */
+		igt_assert(len > 256 * 150);
+
+		overflow_seen = false;
+		for (offset = 0; offset < len; offset += header->size) {
+			header = (void *)(buf + offset);
+
+			if (header->type == DRM_I915_PERF_RECORD_OA_BUFFER_LOST)
+				overflow_seen = true;
+		}
+
+		igt_assert_eq(overflow_seen, false);
+	}
+
+	free(buf);
+
+	close(stream_fd);
+}
+
+static void
+test_enable_disable(void)
+{
+	int oa_exponent = 5; /* 5 micro seconds */
+	uint64_t properties[] = {
+		/* Include OA reports in samples */
+		DRM_I915_PERF_PROP_SAMPLE_OA, true,
+
+		/* OA unit configuration */
+		DRM_I915_PERF_PROP_OA_METRICS_SET, hsw_render_basic_id,
+		DRM_I915_PERF_PROP_OA_FORMAT, I915_OA_FORMAT_A45_B8_C8,
+		DRM_I915_PERF_PROP_OA_EXPONENT, oa_exponent,
+	};
+	struct drm_i915_perf_open_param param = {
+		.flags = I915_PERF_FLAG_FD_CLOEXEC |
+			 I915_PERF_FLAG_DISABLED, /* Verify we start disabled */
+		.num_properties = sizeof(properties) / 16,
+		.properties_ptr = (uint64_t)properties,
+	};
+	int stream_fd = __perf_open(drm_fd, &param);
+	int buf_size = 65536 * (256 + sizeof(struct drm_i915_perf_record_header));
+	uint8_t *buf = malloc(buf_size);
+
+
+	for (int i = 0; i < 5; i++) {
+		int len;
+
+		/* If the stream were enabled then it would take ~330
+		 * milliseconds to fill a 16MB OA buffer with a 5 microsecond
+		 * sampling period and 256 byte reports.
+		 *
+		 * Giving enough time for an overflow might help catch whether
+		 * the OA unit has been enabled even if the driver might at
+		 * least avoid copying reports while disabled.
+		 */
+		nanosleep(&(struct timespec){ .tv_sec = 0, .tv_nsec = 500000000 }, NULL);
+
+		while ((len = read(stream_fd, buf, buf_size)) == -1 && errno == EINTR)
+			;
+
+		igt_assert_eq(len, -1);
+		igt_assert_eq(errno, EIO);
+
+		do_ioctl(stream_fd, I915_PERF_IOCTL_ENABLE, 0);
+
+		nanosleep(&(struct timespec){ .tv_sec = 0, .tv_nsec = 1000000 }, NULL);
+
+		while ((len = read(stream_fd, buf, buf_size)) == -1 && errno == EINTR)
+			;
+
+		igt_assert_neq(len, -1);
+
+		/* expect ~ 200 records in 1 millisecond */
+		igt_assert(len > 256 * 150 && len < 256 * 2000);
+
+		do_ioctl(stream_fd, I915_PERF_IOCTL_DISABLE, 0);
+
+		/* It's considered an error to read a stream while it's disabled
+		 * since it would block indefinitely...
+		 */
+		len = read(stream_fd, buf, buf_size);
+
+		igt_assert_eq(len, -1);
+		igt_assert_eq(errno, EIO);
+	}
+
+	free(buf);
+
+	close(stream_fd);
+}
+
+static void
+test_short_reads(void)
+{
+	int oa_exponent = 5; /* 5 micro seconds */
+	uint64_t properties[] = {
+		/* Include OA reports in samples */
+		DRM_I915_PERF_PROP_SAMPLE_OA, true,
+
+		/* OA unit configuration */
+		DRM_I915_PERF_PROP_OA_METRICS_SET, hsw_render_basic_id,
+		DRM_I915_PERF_PROP_OA_FORMAT, I915_OA_FORMAT_A45_B8_C8,
+		DRM_I915_PERF_PROP_OA_EXPONENT, oa_exponent,
+	};
+	struct drm_i915_perf_open_param param = {
+		.flags = I915_PERF_FLAG_FD_CLOEXEC,
+		.num_properties = sizeof(properties) / 16,
+		.properties_ptr = (uint64_t)properties,
+	};
+	size_t record_size = 256 + sizeof(struct drm_i915_perf_record_header);
+	size_t page_size = sysconf(_SC_PAGE_SIZE);
+	int zero_fd = open("/dev/zero", O_RDWR|O_CLOEXEC);
+	uint8_t *pages = mmap(NULL, page_size * 2,
+			      PROT_READ|PROT_WRITE, MAP_PRIVATE, zero_fd, 0);
+	struct drm_i915_perf_record_header *header;
+	int stream_fd;
+	int ret;
+
+	igt_assert_neq(zero_fd, -1);
+	close(zero_fd);
+	zero_fd = -1;
+
+	igt_assert(pages);
+
+	ret = mprotect(pages + page_size, page_size, PROT_NONE);
+	igt_assert_eq(ret, 0);
+
+	stream_fd = __perf_open(drm_fd, &param);
+
+	nanosleep(&(struct timespec){ .tv_sec = 0, .tv_nsec = 5000000 }, NULL);
+
+	/* At this point there should be lots of pending reports to read */
+
+	/* A read that can return at least one record should result in a short
+	 * read not an EFAULT if the buffer is smaller than the requested read
+	 * size...
+	 *
+	 * Expect to see a sample record here, but at least skip over any
+	 * _RECORD_LOST notifications.
+	 */
+	do {
+		header = (void *)(pages + page_size - record_size);
+		ret = read(stream_fd,
+			   header,
+			   page_size);
+		igt_assert(ret > 0);
+	} while (header->type == DRM_I915_PERF_RECORD_OA_REPORT_LOST);
+
+	igt_assert_eq(ret, record_size);
+
+	/* A read that can't return a single record because it would result
+	 * in a fault on buffer overrun should result in an EFAULT error...
+	 */
+	ret = read(stream_fd, pages + page_size - 16, page_size);
+	igt_assert_eq(ret, -1);
+	igt_assert_eq(errno, EFAULT);
+
+	/* A read that can't return a single record because the buffer is too
+	 * small should result in an ENOSPC error..
+	 *
+	 * Again, skip over _RECORD_LOST records (smaller than record_size/2)
+	 */
+	do {
+		header = (void *)(pages + page_size - record_size / 2);
+		ret = read(stream_fd,
+			   header,
+			   record_size / 2);
+	} while (ret > 0 && header->type == DRM_I915_PERF_RECORD_OA_REPORT_LOST);
+
+	igt_assert_eq(ret, -1);
+	igt_assert_eq(errno, ENOSPC);
+
+	close(stream_fd);
+
+	munmap(pages, page_size * 2);
+}
+
+static void
+test_non_sampling_read_error(void)
+{
+	uint64_t properties[] = {
+		/* XXX: even without periodic sampling we have to
+		 * specify at least one sample layout property...
+		 */
+		DRM_I915_PERF_PROP_SAMPLE_OA, true,
+
+		/* OA unit configuration */
+		DRM_I915_PERF_PROP_OA_METRICS_SET, hsw_render_basic_id,
+		DRM_I915_PERF_PROP_OA_FORMAT, I915_OA_FORMAT_A45_B8_C8,
+
+		/* XXX: no sampling exponent */
+	};
+	struct drm_i915_perf_open_param param = {
+		.flags = I915_PERF_FLAG_FD_CLOEXEC,
+		.num_properties = sizeof(properties) / 16,
+		.properties_ptr = (uint64_t)properties,
+	};
+	int stream_fd = __perf_open(drm_fd, &param);
+	uint8_t buf[1024];
+
+	int ret = read(stream_fd, buf, sizeof(buf));
+	igt_assert_eq(ret, -1);
+	igt_assert_eq(errno, EIO);
+
+	close(stream_fd);
+}
+
+/* Check that attempts to read from a stream while it is disable will return
+ * EIO instead of blocking indefinitely.
+ */
+static void
+test_disabled_read_error(void)
+{
+	int oa_exponent = 5; /* 5 micro seconds */
+	uint64_t properties[] = {
+		/* XXX: even without periodic sampling we have to
+		 * specify at least one sample layout property...
+		 */
+		DRM_I915_PERF_PROP_SAMPLE_OA, true,
+
+		/* OA unit configuration */
+		DRM_I915_PERF_PROP_OA_METRICS_SET, hsw_render_basic_id,
+		DRM_I915_PERF_PROP_OA_FORMAT, I915_OA_FORMAT_A45_B8_C8,
+		DRM_I915_PERF_PROP_OA_EXPONENT, oa_exponent,
+	};
+	struct drm_i915_perf_open_param param = {
+		.flags = I915_PERF_FLAG_FD_CLOEXEC |
+			 I915_PERF_FLAG_DISABLED, /* XXX: open disabled */
+		.num_properties = sizeof(properties) / 16,
+		.properties_ptr = (uint64_t)properties,
+	};
+	int stream_fd = __perf_open(drm_fd, &param);
+	uint32_t oa_report0[64];
+	uint32_t oa_report1[64];
+	uint32_t buf[128] = { 0 };
+	int ret;
+
+
+	ret = read(stream_fd, buf, sizeof(buf));
+	igt_assert_eq(ret, -1);
+	igt_assert_eq(errno, EIO);
+
+	close(stream_fd);
+
+
+	param.flags &= ~I915_PERF_FLAG_DISABLED;
+	stream_fd = __perf_open(drm_fd, &param);
+
+	read_2_oa_reports(stream_fd,
+			  I915_OA_FORMAT_A45_B8_C8,
+			  oa_exponent,
+			  oa_report0,
+			  oa_report1,
+			  false); /* not just timer reports */
+
+	do_ioctl(stream_fd, I915_PERF_IOCTL_DISABLE, 0);
+
+	ret = read(stream_fd, buf, sizeof(buf));
+	igt_assert_eq(ret, -1);
+	igt_assert_eq(errno, EIO);
+
+	do_ioctl(stream_fd, I915_PERF_IOCTL_ENABLE, 0);
+
+	read_2_oa_reports(stream_fd,
+			  I915_OA_FORMAT_A45_B8_C8,
+			  oa_exponent,
+			  oa_report0,
+			  oa_report1,
+			  false); /* not just timer reports */
+
+	close(stream_fd);
+}
+
+static void
+test_mi_rpc(void)
+{
+	uint64_t properties[] = {
+		/* Note: we have to specify at least one sample property even
+		 * though we aren't interested in samples in this case.
+		 */
+		DRM_I915_PERF_PROP_SAMPLE_OA, true,
+
+		/* OA unit configuration */
+		DRM_I915_PERF_PROP_OA_METRICS_SET, hsw_render_basic_id,
+		DRM_I915_PERF_PROP_OA_FORMAT, I915_OA_FORMAT_A45_B8_C8,
+
+		/* Note: no OA exponent specified in this case */
+	};
+	struct drm_i915_perf_open_param param = {
+		.flags = I915_PERF_FLAG_FD_CLOEXEC,
+		.num_properties = sizeof(properties) / 16,
+		.properties_ptr = (uint64_t)properties,
+	};
+	int stream_fd = __perf_open(drm_fd, &param);
+	drm_intel_bufmgr *bufmgr = drm_intel_bufmgr_gem_init(drm_fd, 4096);
+	drm_intel_context *context;
+	struct intel_batchbuffer *batch;
+	drm_intel_bo *bo;
+	uint32_t *report32;
+	int ret;
+
+	drm_intel_bufmgr_gem_enable_reuse(bufmgr);
+
+	context = drm_intel_gem_context_create(bufmgr);
+	igt_assert(context);
+
+	batch = intel_batchbuffer_alloc(bufmgr, devid);
+
+	bo = drm_intel_bo_alloc(bufmgr, "mi_rpc dest bo", 4096, 64);
+
+	ret = drm_intel_bo_map(bo, true);
+	igt_assert_eq(ret, 0);
+
+	memset(bo->virtual, 0x80, 4096);
+	drm_intel_bo_unmap(bo);
+
+	BEGIN_BATCH(3, 1);
+	OUT_BATCH(GEN6_MI_REPORT_PERF_COUNT);
+	OUT_RELOC(bo, I915_GEM_DOMAIN_INSTRUCTION, I915_GEM_DOMAIN_INSTRUCTION,
+		  0); /* offset in bytes */
+	OUT_BATCH(0xdeadbeef); /* report ID */
+	ADVANCE_BATCH();
+
+	intel_batchbuffer_flush_with_context(batch, context);
+
+	ret = drm_intel_bo_map(bo, false /* write enable */);
+	igt_assert_eq(ret, 0);
+
+	report32 = bo->virtual;
+	igt_assert_eq(report32[0], 0xdeadbeef); /* report ID */
+	igt_assert_neq(report32[1], 0); /* timestamp */
+
+	igt_assert_neq(report32[63], 0x80808080); /* end of report */
+	igt_assert_eq(report32[64], 0x80808080); /* after 256 byte report */
+
+	drm_intel_bo_unmap(bo);
+	drm_intel_bo_unreference(bo);
+	intel_batchbuffer_free(batch);
+	drm_intel_gem_context_destroy(context);
+	drm_intel_bufmgr_destroy(bufmgr);
+	close(stream_fd);
+}
+
+static void
+scratch_buf_init(drm_intel_bufmgr *bufmgr,
+		 struct igt_buf *buf,
+		 int width, int height,
+		 uint32_t color)
+{
+	size_t stride = width * 4;
+	size_t size = stride * height;
+	drm_intel_bo *bo = drm_intel_bo_alloc(bufmgr, "", size, 4096);
+	int ret;
+
+	ret = drm_intel_bo_map(bo, true /* writable */);
+	igt_assert_eq(ret, 0);
+
+	for (int i = 0; i < width * height; i++)
+		((uint32_t *)bo->virtual)[i] = color;
+
+	drm_intel_bo_unmap(bo);
+
+	buf->bo = bo;
+	buf->stride = stride;
+	buf->tiling = I915_TILING_NONE;
+	buf->size = size;
+}
+
+static void
+emit_stall_timestamp_and_rpc(struct intel_batchbuffer *batch,
+			     drm_intel_bo *dst,
+			     int timestamp_offset,
+			     int report_dst_offset,
+			     uint32_t report_id)
+{
+	uint32_t pipe_ctl_flags = (PIPE_CONTROL_CS_STALL |
+				   PIPE_CONTROL_RENDER_TARGET_FLUSH |
+				   PIPE_CONTROL_WRITE_TIMESTAMP);
+
+	BEGIN_BATCH(5, 1);
+	OUT_BATCH(GFX_OP_PIPE_CONTROL | (5 - 2));
+	OUT_BATCH(pipe_ctl_flags);
+	OUT_RELOC(dst, I915_GEM_DOMAIN_INSTRUCTION, I915_GEM_DOMAIN_INSTRUCTION,
+		  timestamp_offset);
+	OUT_BATCH(0); /* imm lower */
+	OUT_BATCH(0); /* imm upper */
+	ADVANCE_BATCH();
+
+	BEGIN_BATCH(3, 1);
+	OUT_BATCH(GEN6_MI_REPORT_PERF_COUNT);
+	OUT_RELOC(dst, I915_GEM_DOMAIN_INSTRUCTION, I915_GEM_DOMAIN_INSTRUCTION,
+		  report_dst_offset);
+	OUT_BATCH(report_id);
+	ADVANCE_BATCH();
+}
+
+/* Tests the INTEL_performance_query use case where an unprivileged process
+ * should be able to configure the OA unit for per-context metrics (for a
+ * context associated with that process' drm file descriptor) and the counters
+ * should only relate to that specific context.
+ */
+static void
+test_per_ctx_mi_rpc(void)
+{
+	uint64_t properties[] = {
+		DRM_I915_PERF_PROP_CTX_HANDLE, UINT64_MAX, /* updated below */
+
+		/* Note: we have to specify at least one sample property even
+		 * though we aren't interested in samples in this case
+		 */
+		DRM_I915_PERF_PROP_SAMPLE_OA, true,
+
+		/* OA unit configuration */
+		DRM_I915_PERF_PROP_OA_METRICS_SET, hsw_render_basic_id,
+		DRM_I915_PERF_PROP_OA_FORMAT, I915_OA_FORMAT_A45_B8_C8,
+
+		/* Note: no OA exponent specified in this case */
+	};
+	struct drm_i915_perf_open_param param = {
+		.flags = I915_PERF_FLAG_FD_CLOEXEC,
+		.num_properties = sizeof(properties) / 16,
+		.properties_ptr = (uint64_t)properties,
+	};
+
+	/* should be default, but just to be sure... */
+	write_u64_file("/proc/sys/dev/i915/perf_stream_paranoid", 1);
+
+	igt_fork(child, 1) {
+		drm_intel_bufmgr *bufmgr;
+		drm_intel_context *context0, *context1;
+		int stream_fd;
+		struct intel_batchbuffer *batch;
+		struct igt_buf src, dst;
+		drm_intel_bo *bo;
+		uint32_t *report0_32, *report1_32;
+		uint64_t timestamp0_64, timestamp1_64;
+		uint32_t delta_ts64, delta_oa32;
+		uint64_t delta_ts64_ns, delta_oa32_ns;
+		uint32_t delta_delta;
+		int n_samples_written;
+		int width = 800;
+		int height = 600;
+		int ret;
+
+		igt_drop_root();
+
+		bufmgr = drm_intel_bufmgr_gem_init(drm_fd, 4096);
+		drm_intel_bufmgr_gem_enable_reuse(bufmgr);
+
+		scratch_buf_init(bufmgr, &src, width, height, 0xff0000ff);
+		scratch_buf_init(bufmgr, &dst, width, height, 0x00ff00ff);
+
+		batch = intel_batchbuffer_alloc(bufmgr, devid);
+
+		context0 = drm_intel_gem_context_create(bufmgr);
+		igt_assert(context0);
+
+		context1 = drm_intel_gem_context_create(bufmgr);
+		igt_assert(context1);
+
+		igt_debug("submitting warm up render_copy\n");
+
+		/* Submit some early, unmeasured, work to the context we want
+		 * to measure to try and catch issues with i915-perf
+		 * initializing the HW context ID for filtering.
+		 *
+		 * We do this because i915-perf single context filtering had
+		 * previously only relied on a hook into context pinning to
+		 * initialize the HW context ID, instead of also trying to
+		 * determine the HW ID while opening the stream, in case it
+		 * has already been pinned.
+		 *
+		 * This wasn't noticed by the previous unit test because we
+		 * were opening the stream while the context hadn't been
+		 * touched or pinned yet and so it worked out correctly to wait
+		 * for the pinning hook.
+		 *
+		 * Now a buggy version of i915-perf will fail to measure
+		 * anything for context0 once this initial render_copy() ends
+		 * up pinning the context since there won't ever be a pinning
+		 * hook callback.
+		 */
+		render_copy(batch,
+			    context0,
+			    &src, 0, 0, width, height,
+			    &dst, 0, 0);
+
+		properties[1] = drm_intel_gem_context_get_context_id(context0);
+
+		igt_debug("opening i915-perf stream\n");
+		stream_fd = __perf_open(drm_fd, &param);
+
+		bo = drm_intel_bo_alloc(bufmgr, "mi_rpc dest bo", 4096, 64);
+
+		ret = drm_intel_bo_map(bo, true /* write enable */);
+		igt_assert_eq(ret, 0);
+
+		memset(bo->virtual, 0x80, 4096);
+		drm_intel_bo_unmap(bo);
+
+		emit_stall_timestamp_and_rpc(batch,
+					     bo,
+					     512 /* timestamp offset */,
+					     0, /* report dst offset */
+					     0xdeadbeef); /* report id */
+
+		/* Explicitly flush here (even though the render_copy() call
+		 * will itself flush before/after the copy) to clarify that
+		 * that the PIPE_CONTROL + MI_RPC commands will be in a
+		 * separate batch from the copy.
+		 */
+		intel_batchbuffer_flush_with_context(batch, context0);
+
+		render_copy(batch,
+			    context0,
+			    &src, 0, 0, width, height,
+			    &dst, 0, 0);
+
+		/* Another redundant flush to clarify batch bo is free to reuse */
+		intel_batchbuffer_flush_with_context(batch, context0);
+
+		/* submit two copies on the other context to avoid a false
+		 * positive in case the driver somehow ended up filtering for
+		 * context1
+		 */
+		render_copy(batch,
+			    context1,
+			    &src, 0, 0, width, height,
+			    &dst, 0, 0);
+
+		render_copy(batch,
+			    context1,
+			    &src, 0, 0, width, height,
+			    &dst, 0, 0);
+
+		/* And another */
+		intel_batchbuffer_flush_with_context(batch, context1);
+
+		emit_stall_timestamp_and_rpc(batch,
+					     bo,
+					     520 /* timestamp offset */,
+					     256, /* report dst offset */
+					     0xbeefbeef); /* report id */
+
+		intel_batchbuffer_flush_with_context(batch, context0);
+
+		ret = drm_intel_bo_map(bo, false /* write enable */);
+		igt_assert_eq(ret, 0);
+
+		report0_32 = bo->virtual;
+		igt_assert_eq(report0_32[0], 0xdeadbeef); /* report ID */
+		igt_assert_neq(report0_32[1], 0); /* timestamp */
+
+		report1_32 = report0_32 + 64;
+		igt_assert_eq(report1_32[0], 0xbeefbeef); /* report ID */
+		igt_assert_neq(report1_32[1], 0); /* timestamp */
+
+		print_reports(report0_32, report1_32,
+			      lookup_format(I915_OA_FORMAT_A45_B8_C8));
+
+		/* A40 == N samples written to all render targets */
+		n_samples_written = report1_32[43] - report0_32[43];
+		igt_debug("n samples written = %d\n", n_samples_written);
+		igt_assert_eq(n_samples_written, width * height);
+
+		igt_debug("timestamp32 0 = %u\n", report0_32[1]);
+		igt_debug("timestamp32 1 = %u\n", report1_32[1]);
+
+		timestamp0_64 = *(uint64_t *)(((uint8_t *)bo->virtual) + 512);
+		timestamp1_64 = *(uint64_t *)(((uint8_t *)bo->virtual) + 520);
+
+		igt_debug("timestamp64 0 = %"PRIu64"\n", timestamp0_64);
+		igt_debug("timestamp64 1 = %"PRIu64"\n", timestamp1_64);
+
+		delta_ts64 = timestamp1_64 - timestamp0_64;
+		delta_oa32 = report1_32[1] - report0_32[1];
+
+		/* sanity check that we can pass the delta to timebase_scale */
+		igt_assert(delta_ts64 < UINT32_MAX);
+		delta_oa32_ns = timebase_scale(delta_oa32);
+		delta_ts64_ns = timebase_scale(delta_ts64);
+
+		igt_debug("ts32 delta = %u, = %uns\n",
+			  delta_oa32, (unsigned)delta_oa32_ns);
+		igt_debug("ts64 delta = %u, = %uns\n",
+			  delta_ts64, (unsigned)delta_ts64_ns);
+
+		/* The delta as calculated via the PIPE_CONTROL timestamp or
+		 * the OA report timestamps should be almost identical but
+		 * allow a 320 nanoseconds margin.
+		 */
+		delta_delta = delta_ts64_ns > delta_oa32_ns ?
+			(delta_ts64_ns - delta_oa32_ns) :
+			(delta_oa32_ns - delta_ts64_ns);
+		igt_assert(delta_delta <= 320);
+
+		drm_intel_bo_unreference(src.bo);
+		drm_intel_bo_unreference(dst.bo);
+
+		drm_intel_bo_unmap(bo);
+		drm_intel_bo_unreference(bo);
+		intel_batchbuffer_free(batch);
+		drm_intel_gem_context_destroy(context0);
+		drm_intel_gem_context_destroy(context1);
+		drm_intel_bufmgr_destroy(bufmgr);
+		close(stream_fd);
+	}
+
+	igt_waitchildren();
+}
+
+static void
+test_rc6_disable(void)
+{
+	int oa_exponent = 13; /* 1 millisecond */
+	uint64_t properties[] = {
+		/* Include OA reports in samples */
+		DRM_I915_PERF_PROP_SAMPLE_OA, true,
+
+		/* OA unit configuration */
+		DRM_I915_PERF_PROP_OA_METRICS_SET, hsw_render_basic_id,
+		DRM_I915_PERF_PROP_OA_FORMAT, I915_OA_FORMAT_A45_B8_C8,
+		DRM_I915_PERF_PROP_OA_EXPONENT, oa_exponent,
+	};
+	struct drm_i915_perf_open_param param = {
+		.flags = I915_PERF_FLAG_FD_CLOEXEC,
+		.num_properties = sizeof(properties) / 16,
+		.properties_ptr = (uint64_t)properties,
+	};
+	int stream_fd = __perf_open(drm_fd, &param);
+	uint64_t n_events_start = read_debugfs_u64_record("i915_drpc_info",
+							  "RC6 residency since boot");
+	uint64_t n_events_end;
+
+	nanosleep(&(struct timespec){ .tv_sec = 0, .tv_nsec = 500000000 }, NULL);
+
+	n_events_end = read_debugfs_u64_record("i915_drpc_info",
+					       "RC6 residency since boot");
+
+	igt_assert_eq(n_events_end - n_events_start, 0);
+
+	close(stream_fd);
+
+	n_events_start = read_debugfs_u64_record("i915_drpc_info",
+						 "RC6 residency since boot");
+
+	nanosleep(&(struct timespec){ .tv_sec = 0, .tv_nsec = 500000000 }, NULL);
+
+	n_events_end = read_debugfs_u64_record("i915_drpc_info",
+					       "RC6 residency since boot");
+
+	igt_assert_neq(n_events_end - n_events_start, 0);
+}
+
+static unsigned
+read_i915_module_ref(void)
+{
+	FILE *fp = fopen("/proc/modules", "r");
+	char *line = NULL;
+	size_t line_buf_size = 0;
+	int len = 0;
+	unsigned ref_count;
+
+	igt_assert(fp);
+
+	while ((len = getline(&line, &line_buf_size, fp)) > 0) {
+		if (strncmp(line, "i915 ", 5) == 0) {
+			unsigned long mem;
+			int ret = sscanf(line + 5, "%lu %u", &mem, &ref_count);
+			igt_assert(ret == 2);
+			goto done;
+		}
+	}
+
+	igt_assert(!"reached");
+
+done:
+	free(line);
+	fclose(fp);
+	return ref_count;
+}
+
+/* check that an open i915 perf stream holds a reference on the drm i915 module
+ * including in the corner case where the original drm fd has been closed.
+ */
+static void
+test_i915_ref_count(void)
+{
+	int oa_exponent = 13; /* 1 millisecond */
+	uint64_t properties[] = {
+		/* Include OA reports in samples */
+		DRM_I915_PERF_PROP_SAMPLE_OA, true,
+
+		/* OA unit configuration */
+		DRM_I915_PERF_PROP_OA_METRICS_SET, 0 /* updated below */,
+		DRM_I915_PERF_PROP_OA_FORMAT, I915_OA_FORMAT_A45_B8_C8,
+		DRM_I915_PERF_PROP_OA_EXPONENT, oa_exponent,
+	};
+	struct drm_i915_perf_open_param param = {
+		.flags = I915_PERF_FLAG_FD_CLOEXEC,
+		.num_properties = sizeof(properties) / 16,
+		.properties_ptr = (uint64_t)properties,
+	};
+	unsigned baseline, ref_count0, ref_count1;
+	int stream_fd;
+	uint32_t oa_report0[64];
+	uint32_t oa_report1[64];
+
+	/* This should be the first test before the first fixture so no drm_fd
+	 * should have been opened so far...
+	 */
+	igt_assert_eq(drm_fd, -1);
+
+	baseline = read_i915_module_ref();
+	igt_debug("baseline ref count (drm fd closed) = %u\n", baseline);
+
+	drm_fd = __drm_open_driver(DRIVER_INTEL);
+	devid = intel_get_drm_devid(drm_fd);
+	device = drm_get_card();
+
+	igt_require(IS_HASWELL(devid));
+	igt_require(lookup_hsw_render_basic_id());
+	properties[3] = hsw_render_basic_id;
+
+	ref_count0 = read_i915_module_ref();
+	igt_debug("initial ref count with drm_fd open = %u\n", ref_count0);
+	igt_assert(ref_count0 > baseline);
+
+	stream_fd = __perf_open(drm_fd, &param);
+	ref_count1 = read_i915_module_ref();
+	igt_debug("ref count after opening i915 perf stream = %u\n", ref_count1);
+	igt_assert(ref_count1 > ref_count0);
+
+	close(drm_fd);
+	drm_fd = -1;
+	ref_count0 = read_i915_module_ref();
+	igt_debug("ref count after closing drm fd = %u\n", ref_count0);
+
+	igt_assert(ref_count0 > baseline);
+
+	read_2_oa_reports(stream_fd,
+			  I915_OA_FORMAT_A45_B8_C8,
+			  oa_exponent,
+			  oa_report0,
+			  oa_report1,
+			  false); /* not just timer reports */
+
+	close(stream_fd);
+	ref_count0 = read_i915_module_ref();
+	igt_debug("ref count after closing i915 perf stream fd = %u\n", ref_count0);
+	igt_assert_eq(ref_count0, baseline);
+}
+
+static void
+test_sysctl_defaults(void)
+{
+	int paranoid = read_u64_file("/proc/sys/dev/i915/perf_stream_paranoid");
+	int max_freq = read_u64_file("/proc/sys/dev/i915/oa_max_sample_rate");
+
+	igt_assert_eq(paranoid, 1);
+	igt_assert_eq(max_freq, 100000);
+}
+
+igt_main
+{
+	igt_skip_on_simulation();
+
+	igt_fixture {
+		struct stat sb;
+
+		igt_require(stat("/proc/sys/dev/i915/perf_stream_paranoid", &sb)
+			    == 0);
+		igt_require(stat("/proc/sys/dev/i915/oa_max_sample_rate", &sb)
+			    == 0);
+	}
+
+	igt_subtest("i915-ref-count")
+		test_i915_ref_count();
+
+	igt_subtest("sysctl-defaults")
+		test_sysctl_defaults();
+
+	igt_fixture {
+		/* We expect that the ref count test before these fixtures
+		 * should have closed drm_fd...
+		 */
+		igt_assert_eq(drm_fd, -1);
+		drm_fd = drm_open_driver_render(DRIVER_INTEL);
+		devid = intel_get_drm_devid(drm_fd);
+		device = drm_get_card();
+
+		igt_require(IS_HASWELL(devid));
+		igt_require(lookup_hsw_render_basic_id());
+
+		gt_frequency_range_save();
+
+		write_u64_file("/proc/sys/dev/i915/perf_stream_paranoid", 1);
+		write_u64_file("/proc/sys/dev/i915/oa_max_sample_rate", 100000);
+
+		render_copy = igt_get_render_copyfunc(devid);
+		igt_require_f(render_copy, "no render-copy function\n");
+	}
+
+	igt_subtest("non-system-wide-paranoid")
+		test_system_wide_paranoid();
+
+	igt_subtest("invalid-open-flags")
+		test_invalid_open_flags();
+
+	igt_subtest("invalid-oa-metric-set-id")
+		test_invalid_oa_metric_set_id();
+
+	igt_subtest("invalid-oa-format-id")
+		test_invalid_oa_format_id();
+
+	igt_subtest("missing-sample-flags")
+		test_missing_sample_flags();
+
+	igt_subtest("oa-formats")
+		test_oa_formats();
+
+	igt_subtest("invalid-oa-exponent")
+		test_invalid_oa_exponent();
+	igt_subtest("low-oa-exponent-permissions")
+		test_low_oa_exponent_permissions();
+	igt_subtest("oa-exponents") {
+		test_oa_exponents(450);
+		test_oa_exponents(550);
+	}
+
+	igt_subtest("per-context-mode-unprivileged")
+		test_per_context_mode_unprivileged();
+
+	igt_subtest("buffer-fill")
+		test_buffer_fill();
+
+	igt_subtest("disabled-read-error")
+		test_disabled_read_error();
+	igt_subtest("non-sampling-read-error")
+		test_non_sampling_read_error();
+
+	igt_subtest("enable-disable")
+		test_enable_disable();
+
+	igt_subtest("blocking")
+		test_blocking();
+
+	igt_subtest("polling")
+		test_polling();
+
+	igt_subtest("short-reads")
+		test_short_reads();
+
+	igt_subtest("mi-rpc")
+		test_mi_rpc();
+
+	igt_subtest("mi-rpc-per-ctx")
+		test_per_ctx_mi_rpc();
+
+	igt_subtest("rc6-disable")
+		test_rc6_disable();
+
+	igt_fixture {
+		/* leave sysctl options in their default state... */
+		write_u64_file("/proc/sys/dev/i915/oa_max_sample_rate", 100000);
+		write_u64_file("/proc/sys/dev/i915/perf_stream_paranoid", 1);
+
+		gt_frequency_range_restore();
+
+		close(drm_fd);
+	}
+}