summaryrefslogtreecommitdiff
path: root/drivers/gpu/drm/i915/selftests/i915_gem_timeline.c
blob: 4795877abe5611876c7a3fe0f4401cd10b0e4bfe (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
/*
 * Copyright © 2017 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 "../i915_selftest.h"
#include "i915_random.h"

#include "mock_gem_device.h"
#include "mock_timeline.h"

struct __igt_sync {
	const char *name;
	u32 seqno;
	bool expected;
	bool set;
};

static int __igt_sync(struct intel_timeline *tl,
		      u64 ctx,
		      const struct __igt_sync *p,
		      const char *name)
{
	int ret;

	if (__intel_timeline_sync_is_later(tl, ctx, p->seqno) != p->expected) {
		pr_err("%s: %s(ctx=%llu, seqno=%u) expected passed %s but failed\n",
		       name, p->name, ctx, p->seqno, yesno(p->expected));
		return -EINVAL;
	}

	if (p->set) {
		ret = __intel_timeline_sync_set(tl, ctx, p->seqno);
		if (ret)
			return ret;
	}

	return 0;
}

static int igt_sync(void *arg)
{
	const struct __igt_sync pass[] = {
		{ "unset", 0, false, false },
		{ "new", 0, false, true },
		{ "0a", 0, true, true },
		{ "1a", 1, false, true },
		{ "1b", 1, true, true },
		{ "0b", 0, true, false },
		{ "2a", 2, false, true },
		{ "4", 4, false, true },
		{ "INT_MAX", INT_MAX, false, true },
		{ "INT_MAX-1", INT_MAX-1, true, false },
		{ "INT_MAX+1", (u32)INT_MAX+1, false, true },
		{ "INT_MAX", INT_MAX, true, false },
		{ "UINT_MAX", UINT_MAX, false, true },
		{ "wrap", 0, false, true },
		{ "unwrap", UINT_MAX, true, false },
		{},
	}, *p;
	struct intel_timeline *tl;
	int order, offset;
	int ret;

	tl = mock_timeline(0);
	if (!tl)
		return -ENOMEM;

	for (p = pass; p->name; p++) {
		for (order = 1; order < 64; order++) {
			for (offset = -1; offset <= (order > 1); offset++) {
				u64 ctx = BIT_ULL(order) + offset;

				ret = __igt_sync(tl, ctx, p, "1");
				if (ret)
					goto out;
			}
		}
	}
	mock_timeline_destroy(tl);

	tl = mock_timeline(0);
	if (!tl)
		return -ENOMEM;

	for (order = 1; order < 64; order++) {
		for (offset = -1; offset <= (order > 1); offset++) {
			u64 ctx = BIT_ULL(order) + offset;

			for (p = pass; p->name; p++) {
				ret = __igt_sync(tl, ctx, p, "2");
				if (ret)
					goto out;
			}
		}
	}

out:
	mock_timeline_destroy(tl);
	return ret;
}

static unsigned int random_engine(struct rnd_state *rnd)
{
	return i915_prandom_u32_max_state(I915_NUM_ENGINES, rnd);
}

static int bench_sync(void *arg)
{
	struct rnd_state prng;
	struct intel_timeline *tl;
	unsigned long end_time, count;
	u64 prng32_1M;
	ktime_t kt;
	int order, last_order;

	tl = mock_timeline(0);
	if (!tl)
		return -ENOMEM;

	/* Lookups from cache are very fast and so the random number generation
	 * and the loop itself becomes a significant factor in the per-iteration
	 * timings. We try to compensate the results by measuring the overhead
	 * of the prng and subtract it from the reported results.
	 */
	prandom_seed_state(&prng, i915_selftest.random_seed);
	count = 0;
	kt = ktime_get();
	end_time = jiffies + HZ/10;
	do {
		u32 x;

		/* Make sure the compiler doesn't optimise away the prng call */
		WRITE_ONCE(x, prandom_u32_state(&prng));

		count++;
	} while (!time_after(jiffies, end_time));
	kt = ktime_sub(ktime_get(), kt);
	pr_debug("%s: %lu random evaluations, %lluns/prng\n",
		 __func__, count, (long long)div64_ul(ktime_to_ns(kt), count));
	prng32_1M = div64_ul(ktime_to_ns(kt) << 20, count);

	/* Benchmark (only) setting random context ids */
	prandom_seed_state(&prng, i915_selftest.random_seed);
	count = 0;
	kt = ktime_get();
	end_time = jiffies + HZ/10;
	do {
		u64 id = i915_prandom_u64_state(&prng);

		__intel_timeline_sync_set(tl, id, 0);
		count++;
	} while (!time_after(jiffies, end_time));
	kt = ktime_sub(ktime_get(), kt);
	kt = ktime_sub_ns(kt, (count * prng32_1M * 2) >> 20);
	pr_info("%s: %lu random insertions, %lluns/insert\n",
		__func__, count, (long long)div64_ul(ktime_to_ns(kt), count));

	/* Benchmark looking up the exact same context ids as we just set */
	prandom_seed_state(&prng, i915_selftest.random_seed);
	end_time = count;
	kt = ktime_get();
	while (end_time--) {
		u64 id = i915_prandom_u64_state(&prng);

		if (!__intel_timeline_sync_is_later(tl, id, 0)) {
			mock_timeline_destroy(tl);
			pr_err("Lookup of %llu failed\n", id);
			return -EINVAL;
		}
	}
	kt = ktime_sub(ktime_get(), kt);
	kt = ktime_sub_ns(kt, (count * prng32_1M * 2) >> 20);
	pr_info("%s: %lu random lookups, %lluns/lookup\n",
		__func__, count, (long long)div64_ul(ktime_to_ns(kt), count));

	mock_timeline_destroy(tl);
	cond_resched();

	tl = mock_timeline(0);
	if (!tl)
		return -ENOMEM;

	/* Benchmark setting the first N (in order) contexts */
	count = 0;
	kt = ktime_get();
	end_time = jiffies + HZ/10;
	do {
		__intel_timeline_sync_set(tl, count++, 0);
	} while (!time_after(jiffies, end_time));
	kt = ktime_sub(ktime_get(), kt);
	pr_info("%s: %lu in-order insertions, %lluns/insert\n",
		__func__, count, (long long)div64_ul(ktime_to_ns(kt), count));

	/* Benchmark looking up the exact same context ids as we just set */
	end_time = count;
	kt = ktime_get();
	while (end_time--) {
		if (!__intel_timeline_sync_is_later(tl, end_time, 0)) {
			pr_err("Lookup of %lu failed\n", end_time);
			mock_timeline_destroy(tl);
			return -EINVAL;
		}
	}
	kt = ktime_sub(ktime_get(), kt);
	pr_info("%s: %lu in-order lookups, %lluns/lookup\n",
		__func__, count, (long long)div64_ul(ktime_to_ns(kt), count));

	mock_timeline_destroy(tl);
	cond_resched();

	tl = mock_timeline(0);
	if (!tl)
		return -ENOMEM;

	/* Benchmark searching for a random context id and maybe changing it */
	prandom_seed_state(&prng, i915_selftest.random_seed);
	count = 0;
	kt = ktime_get();
	end_time = jiffies + HZ/10;
	do {
		u32 id = random_engine(&prng);
		u32 seqno = prandom_u32_state(&prng);

		if (!__intel_timeline_sync_is_later(tl, id, seqno))
			__intel_timeline_sync_set(tl, id, seqno);

		count++;
	} while (!time_after(jiffies, end_time));
	kt = ktime_sub(ktime_get(), kt);
	kt = ktime_sub_ns(kt, (count * prng32_1M * 2) >> 20);
	pr_info("%s: %lu repeated insert/lookups, %lluns/op\n",
		__func__, count, (long long)div64_ul(ktime_to_ns(kt), count));
	mock_timeline_destroy(tl);
	cond_resched();

	/* Benchmark searching for a known context id and changing the seqno */
	for (last_order = 1, order = 1; order < 32;
	     ({ int tmp = last_order; last_order = order; order += tmp; })) {
		unsigned int mask = BIT(order) - 1;

		tl = mock_timeline(0);
		if (!tl)
			return -ENOMEM;

		count = 0;
		kt = ktime_get();
		end_time = jiffies + HZ/10;
		do {
			/* Without assuming too many details of the underlying
			 * implementation, try to identify its phase-changes
			 * (if any)!
			 */
			u64 id = (u64)(count & mask) << order;

			__intel_timeline_sync_is_later(tl, id, 0);
			__intel_timeline_sync_set(tl, id, 0);

			count++;
		} while (!time_after(jiffies, end_time));
		kt = ktime_sub(ktime_get(), kt);
		pr_info("%s: %lu cyclic/%d insert/lookups, %lluns/op\n",
			__func__, count, order,
			(long long)div64_ul(ktime_to_ns(kt), count));
		mock_timeline_destroy(tl);
		cond_resched();
	}

	return 0;
}

int i915_gem_timeline_mock_selftests(void)
{
	static const struct i915_subtest tests[] = {
		SUBTEST(igt_sync),
		SUBTEST(bench_sync),
	};

	return i915_subtests(tests, NULL);
}