xen-selfballoon.c 17.7 KB
Newer Older
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
/******************************************************************************
 * Xen selfballoon driver (and optional frontswap self-shrinking driver)
 *
 * Copyright (c) 2009-2011, Dan Magenheimer, Oracle Corp.
 *
 * This code complements the cleancache and frontswap patchsets to optimize
 * support for Xen Transcendent Memory ("tmem").  The policy it implements
 * is rudimentary and will likely improve over time, but it does work well
 * enough today.
 *
 * Two functionalities are implemented here which both use "control theory"
 * (feedback) to optimize memory utilization. In a virtualized environment
 * such as Xen, RAM is often a scarce resource and we would like to ensure
 * that each of a possibly large number of virtual machines is using RAM
 * efficiently, i.e. using as little as possible when under light load
 * and obtaining as much as possible when memory demands are high.
 * Since RAM needs vary highly dynamically and sometimes dramatically,
 * "hysteresis" is used, that is, memory target is determined not just
 * on current data but also on past data stored in the system.
 *
 * "Selfballooning" creates memory pressure by managing the Xen balloon
 * driver to decrease and increase available kernel memory, driven
 * largely by the target value of "Committed_AS" (see /proc/meminfo).
 * Since Committed_AS does not account for clean mapped pages (i.e. pages
 * in RAM that are identical to pages on disk), selfballooning has the
 * affect of pushing less frequently used clean pagecache pages out of
 * kernel RAM and, presumably using cleancache, into Xen tmem where
 * Xen can more efficiently optimize RAM utilization for such pages.
 *
 * When kernel memory demand unexpectedly increases faster than Xen, via
 * the selfballoon driver, is able to (or chooses to) provide usable RAM,
 * the kernel may invoke swapping.  In most cases, frontswap is able
 * to absorb this swapping into Xen tmem.  However, due to the fact
 * that the kernel swap subsystem assumes swapping occurs to a disk,
 * swapped pages may sit on the disk for a very long time; even if
 * the kernel knows the page will never be used again.  This is because
 * the disk space costs very little and can be overwritten when
 * necessary.  When such stale pages are in frontswap, however, they
 * are taking up valuable real estate.  "Frontswap selfshrinking" works
 * to resolve this:  When frontswap activity is otherwise stable
 * and the guest kernel is not under memory pressure, the "frontswap
 * selfshrinking" accounts for this by providing pressure to remove some
 * pages from frontswap and return them to kernel memory.
 *
 * For both "selfballooning" and "frontswap-selfshrinking", a worker
 * thread is used and sysfs tunables are provided to adjust the frequency
 * and rate of adjustments to achieve the goal, as well as to disable one
 * or both functions independently.
 *
 * While some argue that this functionality can and should be implemented
 * in userspace, it has been observed that bad things happen (e.g. OOMs).
 *
 * System configuration note: Selfballooning should not be enabled on
 * systems without a sufficiently large swap device configured; for best
 * results, it is recommended that total swap be increased by the size
56 57 58 59 60 61
 * of the guest memory. Note, that selfballooning should be disabled by default
 * if frontswap is not configured.  Similarly selfballooning should be enabled
 * by default if frontswap is configured and can be disabled with the
 * "tmem.selfballooning=0" kernel boot option.  Finally, when frontswap is
 * configured, frontswap-selfshrinking can be disabled  with the
 * "tmem.selfshrink=0" kernel boot option.
62 63 64 65 66
 *
 * Selfballooning is disallowed in domain0 and force-disabled.
 *
 */

J
Joe Perches 已提交
67 68
#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt

69
#include <linux/kernel.h>
70 71
#include <linux/bootmem.h>
#include <linux/swap.h>
72 73
#include <linux/mm.h>
#include <linux/mman.h>
74
#include <linux/workqueue.h>
75
#include <linux/device.h>
76 77
#include <xen/balloon.h>
#include <xen/tmem.h>
78
#include <xen/xen.h>
79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96

/* Enable/disable with sysfs. */
static int xen_selfballooning_enabled __read_mostly;

/*
 * Controls rate at which memory target (this iteration) approaches
 * ultimate goal when memory need is increasing (up-hysteresis) or
 * decreasing (down-hysteresis). Higher values of hysteresis cause
 * slower increases/decreases. The default values for the various
 * parameters were deemed reasonable by experimentation, may be
 * workload-dependent, and can all be adjusted via sysfs.
 */
static unsigned int selfballoon_downhysteresis __read_mostly = 8;
static unsigned int selfballoon_uphysteresis __read_mostly = 1;

/* In HZ, controls frequency of worker invocation. */
static unsigned int selfballoon_interval __read_mostly = 5;

97 98 99 100 101 102 103 104 105
/*
 * Minimum usable RAM in MB for selfballooning target for balloon.
 * If non-zero, it is added to totalreserve_pages and self-ballooning
 * will not balloon below the sum.  If zero, a piecewise linear function
 * is calculated as a minimum and added to totalreserve_pages.  Note that
 * setting this value indiscriminately may cause OOMs and crashes.
 */
static unsigned int selfballoon_min_usable_mb;

106 107 108 109 110 111
/*
 * Amount of RAM in MB to add to the target number of pages.
 * Can be used to reserve some more room for caches and the like.
 */
static unsigned int selfballoon_reserved_mb;

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
static void selfballoon_process(struct work_struct *work);
static DECLARE_DELAYED_WORK(selfballoon_worker, selfballoon_process);

#ifdef CONFIG_FRONTSWAP
#include <linux/frontswap.h>

/* Enable/disable with sysfs. */
static bool frontswap_selfshrinking __read_mostly;

/*
 * The default values for the following parameters were deemed reasonable
 * by experimentation, may be workload-dependent, and can all be
 * adjusted via sysfs.
 */

/* Control rate for frontswap shrinking. Higher hysteresis is slower. */
static unsigned int frontswap_hysteresis __read_mostly = 20;

/*
 * Number of selfballoon worker invocations to wait before observing that
 * frontswap selfshrinking should commence. Note that selfshrinking does
 * not use a separate worker thread.
 */
static unsigned int frontswap_inertia __read_mostly = 3;

/* Countdown to next invocation of frontswap_shrink() */
static unsigned long frontswap_inertia_counter;

/*
 * Invoked by the selfballoon worker thread, uses current number of pages
 * in frontswap (frontswap_curr_pages()), previous status, and control
 * values (hysteresis and inertia) to determine if frontswap should be
 * shrunk and what the new frontswap size should be.  Note that
 * frontswap_shrink is essentially a partial swapoff that immediately
 * transfers pages from the "swap device" (frontswap) back into kernel
 * RAM; despite the name, frontswap "shrinking" is very different from
 * the "shrinker" interface used by the kernel MM subsystem to reclaim
 * memory.
 */
static void frontswap_selfshrink(void)
{
	static unsigned long cur_frontswap_pages;
154 155
	unsigned long last_frontswap_pages;
	unsigned long tgt_frontswap_pages;
156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171

	last_frontswap_pages = cur_frontswap_pages;
	cur_frontswap_pages = frontswap_curr_pages();
	if (!cur_frontswap_pages ||
			(cur_frontswap_pages > last_frontswap_pages)) {
		frontswap_inertia_counter = frontswap_inertia;
		return;
	}
	if (frontswap_inertia_counter && --frontswap_inertia_counter)
		return;
	if (cur_frontswap_pages <= frontswap_hysteresis)
		tgt_frontswap_pages = 0;
	else
		tgt_frontswap_pages = cur_frontswap_pages -
			(cur_frontswap_pages / frontswap_hysteresis);
	frontswap_shrink(tgt_frontswap_pages);
172
	frontswap_inertia_counter = frontswap_inertia;
173 174 175 176
}

#endif /* CONFIG_FRONTSWAP */

177
#define MB2PAGES(mb)	((mb) << (20 - PAGE_SHIFT))
178
#define PAGES2MB(pages) ((pages) >> (20 - PAGE_SHIFT))
179

180 181 182 183 184 185
/*
 * Use current balloon size, the goal (vm_committed_as), and hysteresis
 * parameters to set a new target balloon size
 */
static void selfballoon_process(struct work_struct *work)
{
186 187
	unsigned long cur_pages, goal_pages, tgt_pages, floor_pages;
	unsigned long useful_pages;
188 189 190
	bool reset_timer = false;

	if (xen_selfballooning_enabled) {
191
		cur_pages = totalram_pages;
192
		tgt_pages = cur_pages; /* default is no change */
193
		goal_pages = vm_memory_committed() +
194 195
				totalreserve_pages +
				MB2PAGES(selfballoon_reserved_mb);
196 197
#ifdef CONFIG_FRONTSWAP
		/* allow space for frontswap pages to be repatriated */
198
		if (frontswap_selfshrinking)
199 200 201 202 203 204 205 206 207 208 209
			goal_pages += frontswap_curr_pages();
#endif
		if (cur_pages > goal_pages)
			tgt_pages = cur_pages -
				((cur_pages - goal_pages) /
				  selfballoon_downhysteresis);
		else if (cur_pages < goal_pages)
			tgt_pages = cur_pages +
				((goal_pages - cur_pages) /
				  selfballoon_uphysteresis);
		/* else if cur_pages == goal_pages, no change */
210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229
		useful_pages = max_pfn - totalreserve_pages;
		if (selfballoon_min_usable_mb != 0)
			floor_pages = totalreserve_pages +
					MB2PAGES(selfballoon_min_usable_mb);
		/* piecewise linear function ending in ~3% slope */
		else if (useful_pages < MB2PAGES(16))
			floor_pages = max_pfn; /* not worth ballooning */
		else if (useful_pages < MB2PAGES(64))
			floor_pages = totalreserve_pages + MB2PAGES(16) +
					((useful_pages - MB2PAGES(16)) >> 1);
		else if (useful_pages < MB2PAGES(512))
			floor_pages = totalreserve_pages + MB2PAGES(40) +
					((useful_pages - MB2PAGES(40)) >> 3);
		else /* useful_pages >= MB2PAGES(512) */
			floor_pages = totalreserve_pages + MB2PAGES(99) +
					((useful_pages - MB2PAGES(99)) >> 5);
		if (tgt_pages < floor_pages)
			tgt_pages = floor_pages;
		balloon_set_new_target(tgt_pages +
			balloon_stats.current_pages - totalram_pages);
230 231 232
		reset_timer = true;
	}
#ifdef CONFIG_FRONTSWAP
233
	if (frontswap_selfshrinking) {
234 235 236 237 238 239 240 241 242 243 244 245 246 247
		frontswap_selfshrink();
		reset_timer = true;
	}
#endif
	if (reset_timer)
		schedule_delayed_work(&selfballoon_worker,
			selfballoon_interval * HZ);
}

#ifdef CONFIG_SYSFS

#include <linux/capability.h>

#define SELFBALLOON_SHOW(name, format, args...)				\
248 249 250
	static ssize_t show_##name(struct device *dev,	\
					  struct device_attribute *attr, \
					  char *buf) \
251 252 253 254 255 256
	{ \
		return sprintf(buf, format, ##args); \
	}

SELFBALLOON_SHOW(selfballooning, "%d\n", xen_selfballooning_enabled);

257 258
static ssize_t store_selfballooning(struct device *dev,
			    struct device_attribute *attr,
259 260 261 262 263 264 265 266 267 268
			    const char *buf,
			    size_t count)
{
	bool was_enabled = xen_selfballooning_enabled;
	unsigned long tmp;
	int err;

	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;

269 270 271 272
	err = kstrtoul(buf, 10, &tmp);
	if (err)
		return err;
	if ((tmp != 0) && (tmp != 1))
273 274 275 276 277 278 279 280 281 282
		return -EINVAL;

	xen_selfballooning_enabled = !!tmp;
	if (!was_enabled && xen_selfballooning_enabled)
		schedule_delayed_work(&selfballoon_worker,
			selfballoon_interval * HZ);

	return count;
}

283
static DEVICE_ATTR(selfballooning, S_IRUGO | S_IWUSR,
284 285 286 287
		   show_selfballooning, store_selfballooning);

SELFBALLOON_SHOW(selfballoon_interval, "%d\n", selfballoon_interval);

288 289
static ssize_t store_selfballoon_interval(struct device *dev,
					  struct device_attribute *attr,
290 291 292 293 294 295 296 297
					  const char *buf,
					  size_t count)
{
	unsigned long val;
	int err;

	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;
298 299 300 301
	err = kstrtoul(buf, 10, &val);
	if (err)
		return err;
	if (val == 0)
302 303 304 305 306
		return -EINVAL;
	selfballoon_interval = val;
	return count;
}

307
static DEVICE_ATTR(selfballoon_interval, S_IRUGO | S_IWUSR,
308 309 310 311
		   show_selfballoon_interval, store_selfballoon_interval);

SELFBALLOON_SHOW(selfballoon_downhys, "%d\n", selfballoon_downhysteresis);

312 313
static ssize_t store_selfballoon_downhys(struct device *dev,
					 struct device_attribute *attr,
314 315 316 317 318 319 320 321
					 const char *buf,
					 size_t count)
{
	unsigned long val;
	int err;

	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;
322 323 324 325
	err = kstrtoul(buf, 10, &val);
	if (err)
		return err;
	if (val == 0)
326 327 328 329 330
		return -EINVAL;
	selfballoon_downhysteresis = val;
	return count;
}

331
static DEVICE_ATTR(selfballoon_downhysteresis, S_IRUGO | S_IWUSR,
332 333 334 335 336
		   show_selfballoon_downhys, store_selfballoon_downhys);


SELFBALLOON_SHOW(selfballoon_uphys, "%d\n", selfballoon_uphysteresis);

337 338
static ssize_t store_selfballoon_uphys(struct device *dev,
				       struct device_attribute *attr,
339 340 341 342 343 344 345 346
				       const char *buf,
				       size_t count)
{
	unsigned long val;
	int err;

	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;
347 348 349 350
	err = kstrtoul(buf, 10, &val);
	if (err)
		return err;
	if (val == 0)
351 352 353 354 355
		return -EINVAL;
	selfballoon_uphysteresis = val;
	return count;
}

356
static DEVICE_ATTR(selfballoon_uphysteresis, S_IRUGO | S_IWUSR,
357 358
		   show_selfballoon_uphys, store_selfballoon_uphys);

359 360 361
SELFBALLOON_SHOW(selfballoon_min_usable_mb, "%d\n",
				selfballoon_min_usable_mb);

362 363
static ssize_t store_selfballoon_min_usable_mb(struct device *dev,
					       struct device_attribute *attr,
364 365 366 367 368 369 370 371
					       const char *buf,
					       size_t count)
{
	unsigned long val;
	int err;

	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;
372 373 374 375
	err = kstrtoul(buf, 10, &val);
	if (err)
		return err;
	if (val == 0)
376 377 378 379 380
		return -EINVAL;
	selfballoon_min_usable_mb = val;
	return count;
}

381
static DEVICE_ATTR(selfballoon_min_usable_mb, S_IRUGO | S_IWUSR,
382 383 384
		   show_selfballoon_min_usable_mb,
		   store_selfballoon_min_usable_mb);

385 386 387 388 389 390 391 392 393 394 395 396 397
SELFBALLOON_SHOW(selfballoon_reserved_mb, "%d\n",
				selfballoon_reserved_mb);

static ssize_t store_selfballoon_reserved_mb(struct device *dev,
					     struct device_attribute *attr,
					     const char *buf,
					     size_t count)
{
	unsigned long val;
	int err;

	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;
398 399 400 401
	err = kstrtoul(buf, 10, &val);
	if (err)
		return err;
	if (val == 0)
402 403 404 405 406 407 408 409 410
		return -EINVAL;
	selfballoon_reserved_mb = val;
	return count;
}

static DEVICE_ATTR(selfballoon_reserved_mb, S_IRUGO | S_IWUSR,
		   show_selfballoon_reserved_mb,
		   store_selfballoon_reserved_mb);

411

412 413 414
#ifdef CONFIG_FRONTSWAP
SELFBALLOON_SHOW(frontswap_selfshrinking, "%d\n", frontswap_selfshrinking);

415 416
static ssize_t store_frontswap_selfshrinking(struct device *dev,
					     struct device_attribute *attr,
417 418 419 420 421 422 423 424 425
					     const char *buf,
					     size_t count)
{
	bool was_enabled = frontswap_selfshrinking;
	unsigned long tmp;
	int err;

	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;
426 427 428 429
	err = kstrtoul(buf, 10, &tmp);
	if (err)
		return err;
	if ((tmp != 0) && (tmp != 1))
430 431 432 433 434 435 436 437 438 439
		return -EINVAL;
	frontswap_selfshrinking = !!tmp;
	if (!was_enabled && !xen_selfballooning_enabled &&
	     frontswap_selfshrinking)
		schedule_delayed_work(&selfballoon_worker,
			selfballoon_interval * HZ);

	return count;
}

440
static DEVICE_ATTR(frontswap_selfshrinking, S_IRUGO | S_IWUSR,
441 442 443 444
		   show_frontswap_selfshrinking, store_frontswap_selfshrinking);

SELFBALLOON_SHOW(frontswap_inertia, "%d\n", frontswap_inertia);

445 446
static ssize_t store_frontswap_inertia(struct device *dev,
				       struct device_attribute *attr,
447 448 449 450 451 452 453 454
				       const char *buf,
				       size_t count)
{
	unsigned long val;
	int err;

	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;
455 456 457 458
	err = kstrtoul(buf, 10, &val);
	if (err)
		return err;
	if (val == 0)
459 460 461 462 463 464
		return -EINVAL;
	frontswap_inertia = val;
	frontswap_inertia_counter = val;
	return count;
}

465
static DEVICE_ATTR(frontswap_inertia, S_IRUGO | S_IWUSR,
466 467 468 469
		   show_frontswap_inertia, store_frontswap_inertia);

SELFBALLOON_SHOW(frontswap_hysteresis, "%d\n", frontswap_hysteresis);

470 471
static ssize_t store_frontswap_hysteresis(struct device *dev,
					  struct device_attribute *attr,
472 473 474 475 476 477 478 479
					  const char *buf,
					  size_t count)
{
	unsigned long val;
	int err;

	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;
480 481 482 483
	err = kstrtoul(buf, 10, &val);
	if (err)
		return err;
	if (val == 0)
484 485 486 487 488
		return -EINVAL;
	frontswap_hysteresis = val;
	return count;
}

489
static DEVICE_ATTR(frontswap_hysteresis, S_IRUGO | S_IWUSR,
490 491 492 493 494
		   show_frontswap_hysteresis, store_frontswap_hysteresis);

#endif /* CONFIG_FRONTSWAP */

static struct attribute *selfballoon_attrs[] = {
495 496 497 498 499
	&dev_attr_selfballooning.attr,
	&dev_attr_selfballoon_interval.attr,
	&dev_attr_selfballoon_downhysteresis.attr,
	&dev_attr_selfballoon_uphysteresis.attr,
	&dev_attr_selfballoon_min_usable_mb.attr,
500
	&dev_attr_selfballoon_reserved_mb.attr,
501
#ifdef CONFIG_FRONTSWAP
502 503 504
	&dev_attr_frontswap_selfshrinking.attr,
	&dev_attr_frontswap_hysteresis.attr,
	&dev_attr_frontswap_inertia.attr,
505 506 507 508
#endif
	NULL
};

509
static const struct attribute_group selfballoon_group = {
510 511 512 513 514
	.name = "selfballoon",
	.attrs = selfballoon_attrs
};
#endif

515
int register_xen_selfballooning(struct device *dev)
516 517 518 519
{
	int error = -1;

#ifdef CONFIG_SYSFS
520
	error = sysfs_create_group(&dev->kobj, &selfballoon_group);
521 522 523 524 525
#endif
	return error;
}
EXPORT_SYMBOL(register_xen_selfballooning);

526
int xen_selfballoon_init(bool use_selfballooning, bool use_frontswap_selfshrink)
527 528
{
	bool enable = false;
529
	unsigned long reserve_pages;
530 531 532 533 534

	if (!xen_domain())
		return -ENODEV;

	if (xen_initial_domain()) {
J
Joe Perches 已提交
535
		pr_info("Xen selfballooning driver disabled for domain0\n");
536 537 538 539 540
		return -ENODEV;
	}

	xen_selfballooning_enabled = tmem_enabled && use_selfballooning;
	if (xen_selfballooning_enabled) {
J
Joe Perches 已提交
541
		pr_info("Initializing Xen selfballooning driver\n");
542 543 544 545 546
		enable = true;
	}
#ifdef CONFIG_FRONTSWAP
	frontswap_selfshrinking = tmem_enabled && use_frontswap_selfshrink;
	if (frontswap_selfshrinking) {
J
Joe Perches 已提交
547
		pr_info("Initializing frontswap selfshrinking driver\n");
548 549 550 551 552 553
		enable = true;
	}
#endif
	if (!enable)
		return -ENODEV;

554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573
	/*
	 * Give selfballoon_reserved_mb a default value(10% of total ram pages)
	 * to make selfballoon not so aggressive.
	 *
	 * There are mainly two reasons:
	 * 1) The original goal_page didn't consider some pages used by kernel
	 *    space, like slab pages and memory used by device drivers.
	 *
	 * 2) The balloon driver may not give back memory to guest OS fast
	 *    enough when the workload suddenly aquries a lot of physical memory.
	 *
	 * In both cases, the guest OS will suffer from memory pressure and
	 * OOM killer may be triggered.
	 * By reserving extra 10% of total ram pages, we can keep the system
	 * much more reliably and response faster in some cases.
	 */
	if (!selfballoon_reserved_mb) {
		reserve_pages = totalram_pages / 10;
		selfballoon_reserved_mb = PAGES2MB(reserve_pages);
	}
574 575 576 577
	schedule_delayed_work(&selfballoon_worker, selfballoon_interval * HZ);

	return 0;
}
578
EXPORT_SYMBOL(xen_selfballoon_init);