]> Pileus Git - ~andy/linux/blob - drivers/xen/balloon.c
37d06ea624aa953d40448bcd2a2d4943baf79fa9
[~andy/linux] / drivers / xen / balloon.c
1 /******************************************************************************
2  * Xen balloon driver - enables returning/claiming memory to/from Xen.
3  *
4  * Copyright (c) 2003, B Dragovic
5  * Copyright (c) 2003-2004, M Williamson, K Fraser
6  * Copyright (c) 2005 Dan M. Smith, IBM Corporation
7  * Copyright (c) 2010 Daniel Kiper
8  *
9  * Memory hotplug support was written by Daniel Kiper. Work on
10  * it was sponsored by Google under Google Summer of Code 2010
11  * program. Jeremy Fitzhardinge from Citrix was the mentor for
12  * this project.
13  *
14  * This program is free software; you can redistribute it and/or
15  * modify it under the terms of the GNU General Public License version 2
16  * as published by the Free Software Foundation; or, when distributed
17  * separately from the Linux kernel or incorporated into other
18  * software packages, subject to the following license:
19  *
20  * Permission is hereby granted, free of charge, to any person obtaining a copy
21  * of this source file (the "Software"), to deal in the Software without
22  * restriction, including without limitation the rights to use, copy, modify,
23  * merge, publish, distribute, sublicense, and/or sell copies of the Software,
24  * and to permit persons to whom the Software is furnished to do so, subject to
25  * the following conditions:
26  *
27  * The above copyright notice and this permission notice shall be included in
28  * all copies or substantial portions of the Software.
29  *
30  * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
31  * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
32  * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
33  * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
34  * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
35  * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
36  * IN THE SOFTWARE.
37  */
38
39 #define pr_fmt(fmt) "xen:" KBUILD_MODNAME ": " fmt
40
41 #include <linux/cpu.h>
42 #include <linux/kernel.h>
43 #include <linux/sched.h>
44 #include <linux/errno.h>
45 #include <linux/module.h>
46 #include <linux/mm.h>
47 #include <linux/bootmem.h>
48 #include <linux/pagemap.h>
49 #include <linux/highmem.h>
50 #include <linux/mutex.h>
51 #include <linux/list.h>
52 #include <linux/gfp.h>
53 #include <linux/notifier.h>
54 #include <linux/memory.h>
55 #include <linux/memory_hotplug.h>
56 #include <linux/percpu-defs.h>
57
58 #include <asm/page.h>
59 #include <asm/pgalloc.h>
60 #include <asm/pgtable.h>
61 #include <asm/tlb.h>
62
63 #include <asm/xen/hypervisor.h>
64 #include <asm/xen/hypercall.h>
65
66 #include <xen/xen.h>
67 #include <xen/interface/xen.h>
68 #include <xen/interface/memory.h>
69 #include <xen/balloon.h>
70 #include <xen/features.h>
71 #include <xen/page.h>
72
73 /*
74  * balloon_process() state:
75  *
76  * BP_DONE: done or nothing to do,
77  * BP_EAGAIN: error, go to sleep,
78  * BP_ECANCELED: error, balloon operation canceled.
79  */
80
81 enum bp_state {
82         BP_DONE,
83         BP_EAGAIN,
84         BP_ECANCELED
85 };
86
87
88 static DEFINE_MUTEX(balloon_mutex);
89
90 struct balloon_stats balloon_stats;
91 EXPORT_SYMBOL_GPL(balloon_stats);
92
93 /* We increase/decrease in batches which fit in a page */
94 static xen_pfn_t frame_list[PAGE_SIZE / sizeof(unsigned long)];
95 static DEFINE_PER_CPU(struct page *, balloon_scratch_page);
96
97
98 /* List of ballooned pages, threaded through the mem_map array. */
99 static LIST_HEAD(ballooned_pages);
100
101 /* Main work function, always executed in process context. */
102 static void balloon_process(struct work_struct *work);
103 static DECLARE_DELAYED_WORK(balloon_worker, balloon_process);
104
105 /* When ballooning out (allocating memory to return to Xen) we don't really
106    want the kernel to try too hard since that can trigger the oom killer. */
107 #define GFP_BALLOON \
108         (GFP_HIGHUSER | __GFP_NOWARN | __GFP_NORETRY | __GFP_NOMEMALLOC)
109
110 static void scrub_page(struct page *page)
111 {
112 #ifdef CONFIG_XEN_SCRUB_PAGES
113         clear_highpage(page);
114 #endif
115 }
116
117 /* balloon_append: add the given page to the balloon. */
118 static void __balloon_append(struct page *page)
119 {
120         /* Lowmem is re-populated first, so highmem pages go at list tail. */
121         if (PageHighMem(page)) {
122                 list_add_tail(&page->lru, &ballooned_pages);
123                 balloon_stats.balloon_high++;
124         } else {
125                 list_add(&page->lru, &ballooned_pages);
126                 balloon_stats.balloon_low++;
127         }
128 }
129
130 static void balloon_append(struct page *page)
131 {
132         __balloon_append(page);
133         adjust_managed_page_count(page, -1);
134 }
135
136 /* balloon_retrieve: rescue a page from the balloon, if it is not empty. */
137 static struct page *balloon_retrieve(bool prefer_highmem)
138 {
139         struct page *page;
140
141         if (list_empty(&ballooned_pages))
142                 return NULL;
143
144         if (prefer_highmem)
145                 page = list_entry(ballooned_pages.prev, struct page, lru);
146         else
147                 page = list_entry(ballooned_pages.next, struct page, lru);
148         list_del(&page->lru);
149
150         if (PageHighMem(page))
151                 balloon_stats.balloon_high--;
152         else
153                 balloon_stats.balloon_low--;
154
155         adjust_managed_page_count(page, 1);
156
157         return page;
158 }
159
160 static struct page *balloon_next_page(struct page *page)
161 {
162         struct list_head *next = page->lru.next;
163         if (next == &ballooned_pages)
164                 return NULL;
165         return list_entry(next, struct page, lru);
166 }
167
168 static enum bp_state update_schedule(enum bp_state state)
169 {
170         if (state == BP_DONE) {
171                 balloon_stats.schedule_delay = 1;
172                 balloon_stats.retry_count = 1;
173                 return BP_DONE;
174         }
175
176         ++balloon_stats.retry_count;
177
178         if (balloon_stats.max_retry_count != RETRY_UNLIMITED &&
179                         balloon_stats.retry_count > balloon_stats.max_retry_count) {
180                 balloon_stats.schedule_delay = 1;
181                 balloon_stats.retry_count = 1;
182                 return BP_ECANCELED;
183         }
184
185         balloon_stats.schedule_delay <<= 1;
186
187         if (balloon_stats.schedule_delay > balloon_stats.max_schedule_delay)
188                 balloon_stats.schedule_delay = balloon_stats.max_schedule_delay;
189
190         return BP_EAGAIN;
191 }
192
193 #ifdef CONFIG_XEN_BALLOON_MEMORY_HOTPLUG
194 static long current_credit(void)
195 {
196         return balloon_stats.target_pages - balloon_stats.current_pages -
197                 balloon_stats.hotplug_pages;
198 }
199
200 static bool balloon_is_inflated(void)
201 {
202         if (balloon_stats.balloon_low || balloon_stats.balloon_high ||
203                         balloon_stats.balloon_hotplug)
204                 return true;
205         else
206                 return false;
207 }
208
209 /*
210  * reserve_additional_memory() adds memory region of size >= credit above
211  * max_pfn. New region is section aligned and size is modified to be multiple
212  * of section size. Those features allow optimal use of address space and
213  * establish proper alignment when this function is called first time after
214  * boot (last section not fully populated at boot time contains unused memory
215  * pages with PG_reserved bit not set; online_pages_range() does not allow page
216  * onlining in whole range if first onlined page does not have PG_reserved
217  * bit set). Real size of added memory is established at page onlining stage.
218  */
219
220 static enum bp_state reserve_additional_memory(long credit)
221 {
222         int nid, rc;
223         u64 hotplug_start_paddr;
224         unsigned long balloon_hotplug = credit;
225
226         hotplug_start_paddr = PFN_PHYS(SECTION_ALIGN_UP(max_pfn));
227         balloon_hotplug = round_up(balloon_hotplug, PAGES_PER_SECTION);
228         nid = memory_add_physaddr_to_nid(hotplug_start_paddr);
229
230         rc = add_memory(nid, hotplug_start_paddr, balloon_hotplug << PAGE_SHIFT);
231
232         if (rc) {
233                 pr_info("%s: add_memory() failed: %i\n", __func__, rc);
234                 return BP_EAGAIN;
235         }
236
237         balloon_hotplug -= credit;
238
239         balloon_stats.hotplug_pages += credit;
240         balloon_stats.balloon_hotplug = balloon_hotplug;
241
242         return BP_DONE;
243 }
244
245 static void xen_online_page(struct page *page)
246 {
247         __online_page_set_limits(page);
248
249         mutex_lock(&balloon_mutex);
250
251         __balloon_append(page);
252
253         if (balloon_stats.hotplug_pages)
254                 --balloon_stats.hotplug_pages;
255         else
256                 --balloon_stats.balloon_hotplug;
257
258         mutex_unlock(&balloon_mutex);
259 }
260
261 static int xen_memory_notifier(struct notifier_block *nb, unsigned long val, void *v)
262 {
263         if (val == MEM_ONLINE)
264                 schedule_delayed_work(&balloon_worker, 0);
265
266         return NOTIFY_OK;
267 }
268
269 static struct notifier_block xen_memory_nb = {
270         .notifier_call = xen_memory_notifier,
271         .priority = 0
272 };
273 #else
274 static long current_credit(void)
275 {
276         unsigned long target = balloon_stats.target_pages;
277
278         target = min(target,
279                      balloon_stats.current_pages +
280                      balloon_stats.balloon_low +
281                      balloon_stats.balloon_high);
282
283         return target - balloon_stats.current_pages;
284 }
285
286 static bool balloon_is_inflated(void)
287 {
288         if (balloon_stats.balloon_low || balloon_stats.balloon_high)
289                 return true;
290         else
291                 return false;
292 }
293
294 static enum bp_state reserve_additional_memory(long credit)
295 {
296         balloon_stats.target_pages = balloon_stats.current_pages;
297         return BP_DONE;
298 }
299 #endif /* CONFIG_XEN_BALLOON_MEMORY_HOTPLUG */
300
301 static enum bp_state increase_reservation(unsigned long nr_pages)
302 {
303         int rc;
304         unsigned long  pfn, i;
305         struct page   *page;
306         struct xen_memory_reservation reservation = {
307                 .address_bits = 0,
308                 .extent_order = 0,
309                 .domid        = DOMID_SELF
310         };
311
312 #ifdef CONFIG_XEN_BALLOON_MEMORY_HOTPLUG
313         if (!balloon_stats.balloon_low && !balloon_stats.balloon_high) {
314                 nr_pages = min(nr_pages, balloon_stats.balloon_hotplug);
315                 balloon_stats.hotplug_pages += nr_pages;
316                 balloon_stats.balloon_hotplug -= nr_pages;
317                 return BP_DONE;
318         }
319 #endif
320
321         if (nr_pages > ARRAY_SIZE(frame_list))
322                 nr_pages = ARRAY_SIZE(frame_list);
323
324         page = list_first_entry_or_null(&ballooned_pages, struct page, lru);
325         for (i = 0; i < nr_pages; i++) {
326                 if (!page) {
327                         nr_pages = i;
328                         break;
329                 }
330                 frame_list[i] = page_to_pfn(page);
331                 page = balloon_next_page(page);
332         }
333
334         set_xen_guest_handle(reservation.extent_start, frame_list);
335         reservation.nr_extents = nr_pages;
336         rc = HYPERVISOR_memory_op(XENMEM_populate_physmap, &reservation);
337         if (rc <= 0)
338                 return BP_EAGAIN;
339
340         for (i = 0; i < rc; i++) {
341                 page = balloon_retrieve(false);
342                 BUG_ON(page == NULL);
343
344                 pfn = page_to_pfn(page);
345
346 #ifdef CONFIG_XEN_HAVE_PVMMU
347                 if (!xen_feature(XENFEAT_auto_translated_physmap)) {
348                         set_phys_to_machine(pfn, frame_list[i]);
349
350                         /* Link back into the page tables if not highmem. */
351                         if (!PageHighMem(page)) {
352                                 int ret;
353                                 ret = HYPERVISOR_update_va_mapping(
354                                                 (unsigned long)__va(pfn << PAGE_SHIFT),
355                                                 mfn_pte(frame_list[i], PAGE_KERNEL),
356                                                 0);
357                                 BUG_ON(ret);
358                         }
359                 }
360 #endif
361
362                 /* Relinquish the page back to the allocator. */
363                 __free_reserved_page(page);
364         }
365
366         balloon_stats.current_pages += rc;
367
368         return BP_DONE;
369 }
370
371 static enum bp_state decrease_reservation(unsigned long nr_pages, gfp_t gfp)
372 {
373         enum bp_state state = BP_DONE;
374         unsigned long  pfn, i;
375         struct page   *page;
376         int ret;
377         struct xen_memory_reservation reservation = {
378                 .address_bits = 0,
379                 .extent_order = 0,
380                 .domid        = DOMID_SELF
381         };
382
383 #ifdef CONFIG_XEN_BALLOON_MEMORY_HOTPLUG
384         if (balloon_stats.hotplug_pages) {
385                 nr_pages = min(nr_pages, balloon_stats.hotplug_pages);
386                 balloon_stats.hotplug_pages -= nr_pages;
387                 balloon_stats.balloon_hotplug += nr_pages;
388                 return BP_DONE;
389         }
390 #endif
391
392         if (nr_pages > ARRAY_SIZE(frame_list))
393                 nr_pages = ARRAY_SIZE(frame_list);
394
395         for (i = 0; i < nr_pages; i++) {
396                 page = alloc_page(gfp);
397                 if (page == NULL) {
398                         nr_pages = i;
399                         state = BP_EAGAIN;
400                         break;
401                 }
402
403                 pfn = page_to_pfn(page);
404                 frame_list[i] = pfn_to_mfn(pfn);
405
406                 scrub_page(page);
407
408 #ifdef CONFIG_XEN_HAVE_PVMMU
409                 /*
410                  * Ballooned out frames are effectively replaced with
411                  * a scratch frame.  Ensure direct mappings and the
412                  * p2m are consistent.
413                  */
414                 if (!xen_feature(XENFEAT_auto_translated_physmap)) {
415                         unsigned long p;
416                         struct page   *scratch_page = get_balloon_scratch_page();
417
418                         if (!PageHighMem(page)) {
419                                 ret = HYPERVISOR_update_va_mapping(
420                                                 (unsigned long)__va(pfn << PAGE_SHIFT),
421                                                 pfn_pte(page_to_pfn(scratch_page),
422                                                         PAGE_KERNEL_RO), 0);
423                                 BUG_ON(ret);
424                         }
425                         p = page_to_pfn(scratch_page);
426                         __set_phys_to_machine(pfn, pfn_to_mfn(p));
427
428                         put_balloon_scratch_page();
429                 }
430 #endif
431
432                 balloon_append(pfn_to_page(pfn));
433         }
434
435         /* Ensure that ballooned highmem pages don't have kmaps. */
436         kmap_flush_unused();
437         flush_tlb_all();
438
439         set_xen_guest_handle(reservation.extent_start, frame_list);
440         reservation.nr_extents   = nr_pages;
441         ret = HYPERVISOR_memory_op(XENMEM_decrease_reservation, &reservation);
442         BUG_ON(ret != nr_pages);
443
444         balloon_stats.current_pages -= nr_pages;
445
446         return state;
447 }
448
449 /*
450  * We avoid multiple worker processes conflicting via the balloon mutex.
451  * We may of course race updates of the target counts (which are protected
452  * by the balloon lock), or with changes to the Xen hard limit, but we will
453  * recover from these in time.
454  */
455 static void balloon_process(struct work_struct *work)
456 {
457         enum bp_state state = BP_DONE;
458         long credit;
459
460         mutex_lock(&balloon_mutex);
461
462         do {
463                 credit = current_credit();
464
465                 if (credit > 0) {
466                         if (balloon_is_inflated())
467                                 state = increase_reservation(credit);
468                         else
469                                 state = reserve_additional_memory(credit);
470                 }
471
472                 if (credit < 0)
473                         state = decrease_reservation(-credit, GFP_BALLOON);
474
475                 state = update_schedule(state);
476
477 #ifndef CONFIG_PREEMPT
478                 if (need_resched())
479                         schedule();
480 #endif
481         } while (credit && state == BP_DONE);
482
483         /* Schedule more work if there is some still to be done. */
484         if (state == BP_EAGAIN)
485                 schedule_delayed_work(&balloon_worker, balloon_stats.schedule_delay * HZ);
486
487         mutex_unlock(&balloon_mutex);
488 }
489
490 struct page *get_balloon_scratch_page(void)
491 {
492         struct page *ret = get_cpu_var(balloon_scratch_page);
493         BUG_ON(ret == NULL);
494         return ret;
495 }
496
497 void put_balloon_scratch_page(void)
498 {
499         put_cpu_var(balloon_scratch_page);
500 }
501
502 /* Resets the Xen limit, sets new target, and kicks off processing. */
503 void balloon_set_new_target(unsigned long target)
504 {
505         /* No need for lock. Not read-modify-write updates. */
506         balloon_stats.target_pages = target;
507         schedule_delayed_work(&balloon_worker, 0);
508 }
509 EXPORT_SYMBOL_GPL(balloon_set_new_target);
510
511 /**
512  * alloc_xenballooned_pages - get pages that have been ballooned out
513  * @nr_pages: Number of pages to get
514  * @pages: pages returned
515  * @highmem: allow highmem pages
516  * @return 0 on success, error otherwise
517  */
518 int alloc_xenballooned_pages(int nr_pages, struct page **pages, bool highmem)
519 {
520         int pgno = 0;
521         struct page *page;
522         mutex_lock(&balloon_mutex);
523         while (pgno < nr_pages) {
524                 page = balloon_retrieve(highmem);
525                 if (page && (highmem || !PageHighMem(page))) {
526                         pages[pgno++] = page;
527                 } else {
528                         enum bp_state st;
529                         if (page)
530                                 balloon_append(page);
531                         st = decrease_reservation(nr_pages - pgno,
532                                         highmem ? GFP_HIGHUSER : GFP_USER);
533                         if (st != BP_DONE)
534                                 goto out_undo;
535                 }
536         }
537         mutex_unlock(&balloon_mutex);
538         return 0;
539  out_undo:
540         while (pgno)
541                 balloon_append(pages[--pgno]);
542         /* Free the memory back to the kernel soon */
543         schedule_delayed_work(&balloon_worker, 0);
544         mutex_unlock(&balloon_mutex);
545         return -ENOMEM;
546 }
547 EXPORT_SYMBOL(alloc_xenballooned_pages);
548
549 /**
550  * free_xenballooned_pages - return pages retrieved with get_ballooned_pages
551  * @nr_pages: Number of pages
552  * @pages: pages to return
553  */
554 void free_xenballooned_pages(int nr_pages, struct page **pages)
555 {
556         int i;
557
558         mutex_lock(&balloon_mutex);
559
560         for (i = 0; i < nr_pages; i++) {
561                 if (pages[i])
562                         balloon_append(pages[i]);
563         }
564
565         /* The balloon may be too large now. Shrink it if needed. */
566         if (current_credit())
567                 schedule_delayed_work(&balloon_worker, 0);
568
569         mutex_unlock(&balloon_mutex);
570 }
571 EXPORT_SYMBOL(free_xenballooned_pages);
572
573 static void __init balloon_add_region(unsigned long start_pfn,
574                                       unsigned long pages)
575 {
576         unsigned long pfn, extra_pfn_end;
577         struct page *page;
578
579         /*
580          * If the amount of usable memory has been limited (e.g., with
581          * the 'mem' command line parameter), don't add pages beyond
582          * this limit.
583          */
584         extra_pfn_end = min(max_pfn, start_pfn + pages);
585
586         for (pfn = start_pfn; pfn < extra_pfn_end; pfn++) {
587                 page = pfn_to_page(pfn);
588                 /* totalram_pages and totalhigh_pages do not
589                    include the boot-time balloon extension, so
590                    don't subtract from it. */
591                 __balloon_append(page);
592         }
593 }
594
595 static int balloon_cpu_notify(struct notifier_block *self,
596                                     unsigned long action, void *hcpu)
597 {
598         int cpu = (long)hcpu;
599         switch (action) {
600         case CPU_UP_PREPARE:
601                 if (per_cpu(balloon_scratch_page, cpu) != NULL)
602                         break;
603                 per_cpu(balloon_scratch_page, cpu) = alloc_page(GFP_KERNEL);
604                 if (per_cpu(balloon_scratch_page, cpu) == NULL) {
605                         pr_warn("Failed to allocate balloon_scratch_page for cpu %d\n", cpu);
606                         return NOTIFY_BAD;
607                 }
608                 break;
609         default:
610                 break;
611         }
612         return NOTIFY_OK;
613 }
614
615 static struct notifier_block balloon_cpu_notifier = {
616         .notifier_call  = balloon_cpu_notify,
617 };
618
619 static int __init balloon_init(void)
620 {
621         int i, cpu;
622
623         if (!xen_domain())
624                 return -ENODEV;
625
626         if (!xen_feature(XENFEAT_auto_translated_physmap)) {
627                 for_each_online_cpu(cpu)
628                 {
629                         per_cpu(balloon_scratch_page, cpu) = alloc_page(GFP_KERNEL);
630                         if (per_cpu(balloon_scratch_page, cpu) == NULL) {
631                                 pr_warn("Failed to allocate balloon_scratch_page for cpu %d\n", cpu);
632                                 return -ENOMEM;
633                         }
634                 }
635                 register_cpu_notifier(&balloon_cpu_notifier);
636         }
637
638         pr_info("Initialising balloon driver\n");
639
640         balloon_stats.current_pages = xen_pv_domain()
641                 ? min(xen_start_info->nr_pages - xen_released_pages, max_pfn)
642                 : get_num_physpages();
643         balloon_stats.target_pages  = balloon_stats.current_pages;
644         balloon_stats.balloon_low   = 0;
645         balloon_stats.balloon_high  = 0;
646
647         balloon_stats.schedule_delay = 1;
648         balloon_stats.max_schedule_delay = 32;
649         balloon_stats.retry_count = 1;
650         balloon_stats.max_retry_count = RETRY_UNLIMITED;
651
652 #ifdef CONFIG_XEN_BALLOON_MEMORY_HOTPLUG
653         balloon_stats.hotplug_pages = 0;
654         balloon_stats.balloon_hotplug = 0;
655
656         set_online_page_callback(&xen_online_page);
657         register_memory_notifier(&xen_memory_nb);
658 #endif
659
660         /*
661          * Initialize the balloon with pages from the extra memory
662          * regions (see arch/x86/xen/setup.c).
663          */
664         for (i = 0; i < XEN_EXTRA_MEM_MAX_REGIONS; i++)
665                 if (xen_extra_mem[i].size)
666                         balloon_add_region(PFN_UP(xen_extra_mem[i].start),
667                                            PFN_DOWN(xen_extra_mem[i].size));
668
669         return 0;
670 }
671
672 subsys_initcall(balloon_init);
673
674 static int __init balloon_clear(void)
675 {
676         int cpu;
677
678         for_each_possible_cpu(cpu)
679                 per_cpu(balloon_scratch_page, cpu) = NULL;
680
681         return 0;
682 }
683 early_initcall(balloon_clear);
684
685 MODULE_LICENSE("GPL");