Commit bc2c0303226ec716854d3c208c7f84fe7aa35cd7

Authored by Ian Campbell
Committed by Jeremy Fitzhardinge
1 parent 3d65c9488c

xen: try harder to balloon up under memory pressure.

Currently if the balloon driver is unable to increase the guest's
reservation it assumes the failure was due to reaching its full
allocation, gives up on the ballooning operation and records the limit
it reached as the "hard limit". The driver will not try again until
the target is set again (even to the same value).

However it is possible that ballooning has in fact failed due to
memory pressure in the host and therefore it is desirable to keep
attempting to reach the target in case memory becomes available. The
most likely scenario is that some guests are ballooning down while
others are ballooning up and therefore there is temporary memory
pressure while things stabilise. You would not expect a well behaved
toolstack to ask a domain to balloon to more than its allocation nor
would you expect it to deliberately over-commit memory by setting
balloon targets which exceed the total host memory.

This patch drops the concept of a hard limit and causes the balloon
driver to retry increasing the reservation on a timer in the same
manner as when decreasing the reservation.

Also if we partially succeed in increasing the reservation
(i.e. receive less pages than we asked for) then we may as well keep
those pages rather than returning them to Xen.

Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Cc: Stable Kernel <stable@kernel.org>

Showing 1 changed file with 5 additions and 26 deletions Side-by-side Diff

drivers/xen/balloon.c
... ... @@ -66,8 +66,6 @@
66 66 /* We aim for 'current allocation' == 'target allocation'. */
67 67 unsigned long current_pages;
68 68 unsigned long target_pages;
69   - /* We may hit the hard limit in Xen. If we do then we remember it. */
70   - unsigned long hard_limit;
71 69 /*
72 70 * Drivers may alter the memory reservation independently, but they
73 71 * must inform the balloon driver so we avoid hitting the hard limit.
... ... @@ -185,7 +183,7 @@
185 183  
186 184 static unsigned long current_target(void)
187 185 {
188   - unsigned long target = min(balloon_stats.target_pages, balloon_stats.hard_limit);
  186 + unsigned long target = balloon_stats.target_pages;
189 187  
190 188 target = min(target,
191 189 balloon_stats.current_pages +
192 190  
193 191  
... ... @@ -221,23 +219,10 @@
221 219 set_xen_guest_handle(reservation.extent_start, frame_list);
222 220 reservation.nr_extents = nr_pages;
223 221 rc = HYPERVISOR_memory_op(XENMEM_populate_physmap, &reservation);
224   - if (rc < nr_pages) {
225   - if (rc > 0) {
226   - int ret;
227   -
228   - /* We hit the Xen hard limit: reprobe. */
229   - reservation.nr_extents = rc;
230   - ret = HYPERVISOR_memory_op(XENMEM_decrease_reservation,
231   - &reservation);
232   - BUG_ON(ret != rc);
233   - }
234   - if (rc >= 0)
235   - balloon_stats.hard_limit = (balloon_stats.current_pages + rc -
236   - balloon_stats.driver_pages);
  222 + if (rc < 0)
237 223 goto out;
238   - }
239 224  
240   - for (i = 0; i < nr_pages; i++) {
  225 + for (i = 0; i < rc; i++) {
241 226 page = balloon_retrieve();
242 227 BUG_ON(page == NULL);
243 228  
244 229  
... ... @@ -263,12 +248,12 @@
263 248 __free_page(page);
264 249 }
265 250  
266   - balloon_stats.current_pages += nr_pages;
  251 + balloon_stats.current_pages += rc;
267 252  
268 253 out:
269 254 spin_unlock_irqrestore(&balloon_lock, flags);
270 255  
271   - return 0;
  256 + return rc < 0 ? rc : rc != nr_pages;
272 257 }
273 258  
274 259 static int decrease_reservation(unsigned long nr_pages)
... ... @@ -369,7 +354,6 @@
369 354 static void balloon_set_new_target(unsigned long target)
370 355 {
371 356 /* No need for lock. Not read-modify-write updates. */
372   - balloon_stats.hard_limit = ~0UL;
373 357 balloon_stats.target_pages = target;
374 358 schedule_work(&balloon_worker);
375 359 }
... ... @@ -428,7 +412,6 @@
428 412 balloon_stats.balloon_low = 0;
429 413 balloon_stats.balloon_high = 0;
430 414 balloon_stats.driver_pages = 0UL;
431   - balloon_stats.hard_limit = ~0UL;
432 415  
433 416 init_timer(&balloon_timer);
434 417 balloon_timer.data = 0;
... ... @@ -473,9 +456,6 @@
473 456 BALLOON_SHOW(current_kb, "%lu\n", PAGES2KB(balloon_stats.current_pages));
474 457 BALLOON_SHOW(low_kb, "%lu\n", PAGES2KB(balloon_stats.balloon_low));
475 458 BALLOON_SHOW(high_kb, "%lu\n", PAGES2KB(balloon_stats.balloon_high));
476   -BALLOON_SHOW(hard_limit_kb,
477   - (balloon_stats.hard_limit!=~0UL) ? "%lu\n" : "???\n",
478   - (balloon_stats.hard_limit!=~0UL) ? PAGES2KB(balloon_stats.hard_limit) : 0);
479 459 BALLOON_SHOW(driver_kb, "%lu\n", PAGES2KB(balloon_stats.driver_pages));
480 460  
481 461 static ssize_t show_target_kb(struct sys_device *dev, struct sysdev_attribute *attr,
... ... @@ -545,7 +525,6 @@
545 525 &attr_current_kb.attr,
546 526 &attr_low_kb.attr,
547 527 &attr_high_kb.attr,
548   - &attr_hard_limit_kb.attr,
549 528 &attr_driver_kb.attr,
550 529 NULL
551 530 };