1 /******************************************************************************
2 * Xen balloon driver - enables returning/claiming memory to/from Xen.
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
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
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:
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:
27 * The above copyright notice and this permission notice shall be included in
28 * all copies or substantial portions of the Software.
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
39 #define pr_fmt(fmt) "xen:" KBUILD_MODNAME ": " fmt
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>
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 #include <linux/slab.h>
58 #include <linux/sysctl.h>
61 #include <asm/pgalloc.h>
62 #include <asm/pgtable.h>
65 #include <asm/xen/hypervisor.h>
66 #include <asm/xen/hypercall.h>
69 #include <xen/interface/xen.h>
70 #include <xen/interface/memory.h>
71 #include <xen/balloon.h>
72 #include <xen/features.h>
75 static int xen_hotplug_unpopulated;
77 #ifdef CONFIG_XEN_BALLOON_MEMORY_HOTPLUG
82 static struct ctl_table balloon_table[] = {
84 .procname = "hotplug_unpopulated",
85 .data = &xen_hotplug_unpopulated,
86 .maxlen = sizeof(int),
88 .proc_handler = proc_dointvec_minmax,
95 static struct ctl_table balloon_root[] = {
97 .procname = "balloon",
99 .child = balloon_table,
104 static struct ctl_table xen_root[] = {
108 .child = balloon_root,
116 * balloon_process() state:
118 * BP_DONE: done or nothing to do,
119 * BP_WAIT: wait to be rescheduled,
120 * BP_EAGAIN: error, go to sleep,
121 * BP_ECANCELED: error, balloon operation canceled.
132 static DEFINE_MUTEX(balloon_mutex);
134 struct balloon_stats balloon_stats;
135 EXPORT_SYMBOL_GPL(balloon_stats);
137 /* We increase/decrease in batches which fit in a page */
138 static xen_pfn_t frame_list[PAGE_SIZE / sizeof(unsigned long)];
141 /* List of ballooned pages, threaded through the mem_map array. */
142 static LIST_HEAD(ballooned_pages);
143 static DECLARE_WAIT_QUEUE_HEAD(balloon_wq);
145 /* Main work function, always executed in process context. */
146 static void balloon_process(struct work_struct *work);
147 static DECLARE_DELAYED_WORK(balloon_worker, balloon_process);
149 /* When ballooning out (allocating memory to return to Xen) we don't really
150 want the kernel to try too hard since that can trigger the oom killer. */
151 #define GFP_BALLOON \
152 (GFP_HIGHUSER | __GFP_NOWARN | __GFP_NORETRY | __GFP_NOMEMALLOC)
154 static void scrub_page(struct page *page)
156 #ifdef CONFIG_XEN_SCRUB_PAGES
157 clear_highpage(page);
161 /* balloon_append: add the given page to the balloon. */
162 static void __balloon_append(struct page *page)
164 /* Lowmem is re-populated first, so highmem pages go at list tail. */
165 if (PageHighMem(page)) {
166 list_add_tail(&page->lru, &ballooned_pages);
167 balloon_stats.balloon_high++;
169 list_add(&page->lru, &ballooned_pages);
170 balloon_stats.balloon_low++;
172 wake_up(&balloon_wq);
175 static void balloon_append(struct page *page)
177 __balloon_append(page);
178 adjust_managed_page_count(page, -1);
181 /* balloon_retrieve: rescue a page from the balloon, if it is not empty. */
182 static struct page *balloon_retrieve(bool require_lowmem)
186 if (list_empty(&ballooned_pages))
189 page = list_entry(ballooned_pages.next, struct page, lru);
190 if (require_lowmem && PageHighMem(page))
192 list_del(&page->lru);
194 if (PageHighMem(page))
195 balloon_stats.balloon_high--;
197 balloon_stats.balloon_low--;
199 adjust_managed_page_count(page, 1);
204 static struct page *balloon_next_page(struct page *page)
206 struct list_head *next = page->lru.next;
207 if (next == &ballooned_pages)
209 return list_entry(next, struct page, lru);
212 static enum bp_state update_schedule(enum bp_state state)
214 if (state == BP_WAIT)
217 if (state == BP_ECANCELED)
220 if (state == BP_DONE) {
221 balloon_stats.schedule_delay = 1;
222 balloon_stats.retry_count = 1;
226 ++balloon_stats.retry_count;
228 if (balloon_stats.max_retry_count != RETRY_UNLIMITED &&
229 balloon_stats.retry_count > balloon_stats.max_retry_count) {
230 balloon_stats.schedule_delay = 1;
231 balloon_stats.retry_count = 1;
235 balloon_stats.schedule_delay <<= 1;
237 if (balloon_stats.schedule_delay > balloon_stats.max_schedule_delay)
238 balloon_stats.schedule_delay = balloon_stats.max_schedule_delay;
243 #ifdef CONFIG_XEN_BALLOON_MEMORY_HOTPLUG
244 static struct resource *additional_memory_resource(phys_addr_t size)
246 struct resource *res;
249 res = kzalloc(sizeof(*res), GFP_KERNEL);
253 res->name = "System RAM";
254 res->flags = IORESOURCE_MEM | IORESOURCE_BUSY;
256 ret = allocate_resource(&iomem_resource, res,
258 PAGES_PER_SECTION * PAGE_SIZE, NULL, NULL);
260 pr_err("Cannot allocate new System RAM resource\n");
268 static void release_memory_resource(struct resource *resource)
274 * No need to reset region to identity mapped since we now
275 * know that no I/O can be in this region
277 release_resource(resource);
281 static enum bp_state reserve_additional_memory(void)
284 struct resource *resource;
286 unsigned long balloon_hotplug;
288 credit = balloon_stats.target_pages + balloon_stats.target_unpopulated
289 - balloon_stats.total_pages;
292 * Already hotplugged enough pages? Wait for them to be
298 balloon_hotplug = round_up(credit, PAGES_PER_SECTION);
300 resource = additional_memory_resource(balloon_hotplug * PAGE_SIZE);
304 nid = memory_add_physaddr_to_nid(resource->start);
306 #ifdef CONFIG_XEN_HAVE_PVMMU
308 * add_memory() will build page tables for the new memory so
309 * the p2m must contain invalid entries so the correct
310 * non-present PTEs will be written.
312 * If a failure occurs, the original (identity) p2m entries
313 * are not restored since this region is now known not to
314 * conflict with any devices.
316 if (!xen_feature(XENFEAT_auto_translated_physmap)) {
317 unsigned long pfn, i;
319 pfn = PFN_DOWN(resource->start);
320 for (i = 0; i < balloon_hotplug; i++) {
321 if (!set_phys_to_machine(pfn + i, INVALID_P2M_ENTRY)) {
322 pr_warn("set_phys_to_machine() failed, no memory added\n");
329 rc = add_memory_resource(nid, resource);
331 pr_warn("Cannot add additional memory (%i)\n", rc);
335 balloon_stats.total_pages += balloon_hotplug;
339 release_memory_resource(resource);
343 static void xen_online_page(struct page *page)
345 __online_page_set_limits(page);
347 mutex_lock(&balloon_mutex);
349 __balloon_append(page);
351 mutex_unlock(&balloon_mutex);
354 static int xen_memory_notifier(struct notifier_block *nb, unsigned long val, void *v)
356 if (val == MEM_ONLINE)
357 schedule_delayed_work(&balloon_worker, 0);
362 static struct notifier_block xen_memory_nb = {
363 .notifier_call = xen_memory_notifier,
367 static enum bp_state reserve_additional_memory(void)
369 balloon_stats.target_pages = balloon_stats.current_pages;
372 #endif /* CONFIG_XEN_BALLOON_MEMORY_HOTPLUG */
374 static long current_credit(void)
376 return balloon_stats.target_pages - balloon_stats.current_pages;
379 static bool balloon_is_inflated(void)
381 return balloon_stats.balloon_low || balloon_stats.balloon_high;
384 static enum bp_state increase_reservation(unsigned long nr_pages)
387 unsigned long pfn, i;
389 struct xen_memory_reservation reservation = {
395 if (nr_pages > ARRAY_SIZE(frame_list))
396 nr_pages = ARRAY_SIZE(frame_list);
398 page = list_first_entry_or_null(&ballooned_pages, struct page, lru);
399 for (i = 0; i < nr_pages; i++) {
404 frame_list[i] = page_to_pfn(page);
405 page = balloon_next_page(page);
408 set_xen_guest_handle(reservation.extent_start, frame_list);
409 reservation.nr_extents = nr_pages;
410 rc = HYPERVISOR_memory_op(XENMEM_populate_physmap, &reservation);
414 for (i = 0; i < rc; i++) {
415 page = balloon_retrieve(false);
416 BUG_ON(page == NULL);
418 pfn = page_to_pfn(page);
420 #ifdef CONFIG_XEN_HAVE_PVMMU
421 if (!xen_feature(XENFEAT_auto_translated_physmap)) {
422 set_phys_to_machine(pfn, frame_list[i]);
424 /* Link back into the page tables if not highmem. */
425 if (!PageHighMem(page)) {
427 ret = HYPERVISOR_update_va_mapping(
428 (unsigned long)__va(pfn << PAGE_SHIFT),
429 mfn_pte(frame_list[i], PAGE_KERNEL),
436 /* Relinquish the page back to the allocator. */
437 __free_reserved_page(page);
440 balloon_stats.current_pages += rc;
445 static enum bp_state decrease_reservation(unsigned long nr_pages, gfp_t gfp)
447 enum bp_state state = BP_DONE;
448 unsigned long pfn, i;
451 struct xen_memory_reservation reservation = {
457 if (nr_pages > ARRAY_SIZE(frame_list))
458 nr_pages = ARRAY_SIZE(frame_list);
460 for (i = 0; i < nr_pages; i++) {
461 page = alloc_page(gfp);
469 frame_list[i] = page_to_pfn(page);
473 * Ensure that ballooned highmem pages don't have kmaps.
475 * Do this before changing the p2m as kmap_flush_unused()
476 * reads PTEs to obtain pages (and hence needs the original
481 /* Update direct mapping, invalidate P2M, and add to balloon. */
482 for (i = 0; i < nr_pages; i++) {
484 frame_list[i] = pfn_to_gfn(pfn);
485 page = pfn_to_page(pfn);
487 #ifdef CONFIG_XEN_HAVE_PVMMU
488 if (!xen_feature(XENFEAT_auto_translated_physmap)) {
489 if (!PageHighMem(page)) {
490 ret = HYPERVISOR_update_va_mapping(
491 (unsigned long)__va(pfn << PAGE_SHIFT),
495 __set_phys_to_machine(pfn, INVALID_P2M_ENTRY);
499 balloon_append(page);
504 set_xen_guest_handle(reservation.extent_start, frame_list);
505 reservation.nr_extents = nr_pages;
506 ret = HYPERVISOR_memory_op(XENMEM_decrease_reservation, &reservation);
507 BUG_ON(ret != nr_pages);
509 balloon_stats.current_pages -= nr_pages;
515 * As this is a work item it is guaranteed to run as a single instance only.
516 * We may of course race updates of the target counts (which are protected
517 * by the balloon lock), or with changes to the Xen hard limit, but we will
518 * recover from these in time.
520 static void balloon_process(struct work_struct *work)
522 enum bp_state state = BP_DONE;
527 mutex_lock(&balloon_mutex);
529 credit = current_credit();
532 if (balloon_is_inflated())
533 state = increase_reservation(credit);
535 state = reserve_additional_memory();
539 state = decrease_reservation(-credit, GFP_BALLOON);
541 state = update_schedule(state);
543 mutex_unlock(&balloon_mutex);
547 } while (credit && state == BP_DONE);
549 /* Schedule more work if there is some still to be done. */
550 if (state == BP_EAGAIN)
551 schedule_delayed_work(&balloon_worker, balloon_stats.schedule_delay * HZ);
554 /* Resets the Xen limit, sets new target, and kicks off processing. */
555 void balloon_set_new_target(unsigned long target)
557 /* No need for lock. Not read-modify-write updates. */
558 balloon_stats.target_pages = target;
559 schedule_delayed_work(&balloon_worker, 0);
561 EXPORT_SYMBOL_GPL(balloon_set_new_target);
563 static int add_ballooned_pages(int nr_pages)
567 if (xen_hotplug_unpopulated) {
568 st = reserve_additional_memory();
569 if (st != BP_ECANCELED) {
570 mutex_unlock(&balloon_mutex);
571 wait_event(balloon_wq,
572 !list_empty(&ballooned_pages));
573 mutex_lock(&balloon_mutex);
578 st = decrease_reservation(nr_pages, GFP_USER);
586 * alloc_xenballooned_pages - get pages that have been ballooned out
587 * @nr_pages: Number of pages to get
588 * @pages: pages returned
589 * @return 0 on success, error otherwise
591 int alloc_xenballooned_pages(int nr_pages, struct page **pages)
597 mutex_lock(&balloon_mutex);
599 balloon_stats.target_unpopulated += nr_pages;
601 while (pgno < nr_pages) {
602 page = balloon_retrieve(true);
604 pages[pgno++] = page;
606 ret = add_ballooned_pages(nr_pages - pgno);
611 mutex_unlock(&balloon_mutex);
614 mutex_unlock(&balloon_mutex);
615 free_xenballooned_pages(pgno, pages);
618 EXPORT_SYMBOL(alloc_xenballooned_pages);
621 * free_xenballooned_pages - return pages retrieved with get_ballooned_pages
622 * @nr_pages: Number of pages
623 * @pages: pages to return
625 void free_xenballooned_pages(int nr_pages, struct page **pages)
629 mutex_lock(&balloon_mutex);
631 for (i = 0; i < nr_pages; i++) {
633 balloon_append(pages[i]);
636 balloon_stats.target_unpopulated -= nr_pages;
638 /* The balloon may be too large now. Shrink it if needed. */
639 if (current_credit())
640 schedule_delayed_work(&balloon_worker, 0);
642 mutex_unlock(&balloon_mutex);
644 EXPORT_SYMBOL(free_xenballooned_pages);
646 static void __init balloon_add_region(unsigned long start_pfn,
649 unsigned long pfn, extra_pfn_end;
653 * If the amount of usable memory has been limited (e.g., with
654 * the 'mem' command line parameter), don't add pages beyond
657 extra_pfn_end = min(max_pfn, start_pfn + pages);
659 for (pfn = start_pfn; pfn < extra_pfn_end; pfn++) {
660 page = pfn_to_page(pfn);
661 /* totalram_pages and totalhigh_pages do not
662 include the boot-time balloon extension, so
663 don't subtract from it. */
664 __balloon_append(page);
667 balloon_stats.total_pages += extra_pfn_end - start_pfn;
670 static int __init balloon_init(void)
677 pr_info("Initialising balloon driver\n");
679 balloon_stats.current_pages = xen_pv_domain()
680 ? min(xen_start_info->nr_pages - xen_released_pages, max_pfn)
681 : get_num_physpages();
682 balloon_stats.target_pages = balloon_stats.current_pages;
683 balloon_stats.balloon_low = 0;
684 balloon_stats.balloon_high = 0;
685 balloon_stats.total_pages = balloon_stats.current_pages;
687 balloon_stats.schedule_delay = 1;
688 balloon_stats.max_schedule_delay = 32;
689 balloon_stats.retry_count = 1;
690 balloon_stats.max_retry_count = RETRY_UNLIMITED;
692 #ifdef CONFIG_XEN_BALLOON_MEMORY_HOTPLUG
693 set_online_page_callback(&xen_online_page);
694 register_memory_notifier(&xen_memory_nb);
695 register_sysctl_table(xen_root);
699 * Initialize the balloon with pages from the extra memory
700 * regions (see arch/x86/xen/setup.c).
702 for (i = 0; i < XEN_EXTRA_MEM_MAX_REGIONS; i++)
703 if (xen_extra_mem[i].n_pfns)
704 balloon_add_region(xen_extra_mem[i].start_pfn,
705 xen_extra_mem[i].n_pfns);
710 subsys_initcall(balloon_init);
712 MODULE_LICENSE("GPL");