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/cred.h>
  45 #include <linux/errno.h>
  46 #include <linux/freezer.h>
  47 #include <linux/kthread.h>
  48 #include <linux/mm.h>
  49 #include <linux/bootmem.h>
  50 #include <linux/pagemap.h>
  51 #include <linux/highmem.h>
  52 #include <linux/mutex.h>
  53 #include <linux/list.h>
  54 #include <linux/gfp.h>
  55 #include <linux/notifier.h>
  56 #include <linux/memory.h>
  57 #include <linux/memory_hotplug.h>
  58 #include <linux/percpu-defs.h>
  59 #include <linux/slab.h>
  60 #include <linux/sysctl.h>
  61 #include <linux/moduleparam.h>
  62
  63 #include <asm/page.h>
  64 #include <asm/pgalloc.h>
  65 #include <asm/pgtable.h>
  66 #include <asm/tlb.h>
  67
  68 #include <asm/xen/hypervisor.h>
  69 #include <asm/xen/hypercall.h>
  70
  71 #include <xen/xen.h>
  72 #include <xen/interface/xen.h>
  73 #include <xen/interface/memory.h>
  74 #include <xen/balloon.h>
  75 #include <xen/features.h>
  76 #include <xen/page.h>
  77 #include <xen/mem-reservation.h>
  78
  79 #undef MODULE_PARAM_PREFIX
  80 #define MODULE_PARAM_PREFIX "xen."
  81
  82 static uint __read_mostly balloon_boot_timeout = 180;
  83 module_param(balloon_boot_timeout, uint, 0444);
  84
  85 static int xen_hotplug_unpopulated;
  86
  87 #ifdef CONFIG_XEN_BALLOON_MEMORY_HOTPLUG
  88
  89 static int zero;
  90 static int one = 1;
  91
  92 static struct ctl_table balloon_table[] = {
  93         {
  94                 .procname       = "hotplug_unpopulated",
  95                 .data           = &xen_hotplug_unpopulated,
  96                 .maxlen         = sizeof(int),
  97                 .mode           = 0644,
  98                 .proc_handler   = proc_dointvec_minmax,
  99                 .extra1         = &zero,
 100                 .extra2         = &one,
 101         },
 102         { }
 103 };
 104
 105 static struct ctl_table balloon_root[] = {
 106         {
 107                 .procname       = "balloon",
 108                 .mode           = 0555,
 109                 .child          = balloon_table,
 110         },
 111         { }
 112 };
 113
 114 static struct ctl_table xen_root[] = {
 115         {
 116                 .procname       = "xen",
 117                 .mode           = 0555,
 118                 .child          = balloon_root,
 119         },
 120         { }
 121 };
 122
 123 #endif
 124
 125 /*
 126  * Use one extent per PAGE_SIZE to avoid to break down the page into
 127  * multiple frame.
 128  */
 129 #define EXTENT_ORDER (fls(XEN_PFN_PER_PAGE) - 1)
 130
 131 /*
 132  * balloon_thread() state:
 133  *
 134  * BP_DONE: done or nothing to do,
 135  * BP_WAIT: wait to be rescheduled,
 136  * BP_EAGAIN: error, go to sleep,
 137  * BP_ECANCELED: error, balloon operation canceled.
 138  */
 139
 140 static enum bp_state {
 141         BP_DONE,
 142         BP_WAIT,
 143         BP_EAGAIN,
 144         BP_ECANCELED
 145 } balloon_state = BP_DONE;
 146
 147 /* Main waiting point for xen-balloon thread. */
 148 static DECLARE_WAIT_QUEUE_HEAD(balloon_thread_wq);
 149
 150 static DEFINE_MUTEX(balloon_mutex);
 151
 152 struct balloon_stats balloon_stats;
 153 EXPORT_SYMBOL_GPL(balloon_stats);
 154
 155 /* We increase/decrease in batches which fit in a page */
 156 static xen_pfn_t frame_list[PAGE_SIZE / sizeof(xen_pfn_t)];
 157
 158
 159 /* List of ballooned pages, threaded through the mem_map array. */
 160 static LIST_HEAD(ballooned_pages);
 161 static DECLARE_WAIT_QUEUE_HEAD(balloon_wq);
 162
 163 /* When ballooning out (allocating memory to return to Xen) we don't really
 164    want the kernel to try too hard since that can trigger the oom killer. */
 165 #define GFP_BALLOON \
 166         (GFP_HIGHUSER | __GFP_NOWARN | __GFP_NORETRY | __GFP_NOMEMALLOC)
 167
 168 /* balloon_append: add the given page to the balloon. */
 169 static void __balloon_append(struct page *page)
 170 {
 171         /* Lowmem is re-populated first, so highmem pages go at list tail. */
 172         if (PageHighMem(page)) {
 173                 list_add_tail(&page->lru, &ballooned_pages);
 174                 balloon_stats.balloon_high++;
 175         } else {
 176                 list_add(&page->lru, &ballooned_pages);
 177                 balloon_stats.balloon_low++;
 178         }
 179         wake_up(&balloon_wq);
 180 }
 181
 182 static void balloon_append(struct page *page)
 183 {
 184         __balloon_append(page);
 185 }
 186
 187 /* balloon_retrieve: rescue a page from the balloon, if it is not empty. */
 188 static struct page *balloon_retrieve(bool require_lowmem)
 189 {
 190         struct page *page;
 191
 192         if (list_empty(&ballooned_pages))
 193                 return NULL;
 194
 195         page = list_entry(ballooned_pages.next, struct page, lru);
 196         if (require_lowmem && PageHighMem(page))
 197                 return NULL;
 198         list_del(&page->lru);
 199
 200         if (PageHighMem(page))
 201                 balloon_stats.balloon_high--;
 202         else
 203                 balloon_stats.balloon_low--;
 204
 205         return page;
 206 }
 207
 208 static struct page *balloon_next_page(struct page *page)
 209 {
 210         struct list_head *next = page->lru.next;
 211         if (next == &ballooned_pages)
 212                 return NULL;
 213         return list_entry(next, struct page, lru);
 214 }
 215
 216 static void update_schedule(void)
 217 {
 218         if (balloon_state == BP_WAIT || balloon_state == BP_ECANCELED)
 219                 return;
 220
 221         if (balloon_state == BP_DONE) {
 222                 balloon_stats.schedule_delay = 1;
 223                 balloon_stats.retry_count = 1;
 224                 return;
 225         }
 226
 227         ++balloon_stats.retry_count;
 228
 229         if (balloon_stats.max_retry_count != RETRY_UNLIMITED &&
 230                         balloon_stats.retry_count > balloon_stats.max_retry_count) {
 231                 balloon_stats.schedule_delay = 1;
 232                 balloon_stats.retry_count = 1;
 233                 balloon_state = BP_ECANCELED;
 234                 return;
 235         }
 236
 237         balloon_stats.schedule_delay <<= 1;
 238
 239         if (balloon_stats.schedule_delay > balloon_stats.max_schedule_delay)
 240                 balloon_stats.schedule_delay = balloon_stats.max_schedule_delay;
 241
 242         balloon_state = BP_EAGAIN;
 243 }
 244
 245 #ifdef CONFIG_XEN_BALLOON_MEMORY_HOTPLUG
 246 static void release_memory_resource(struct resource *resource)
 247 {
 248         if (!resource)
 249                 return;
 250
 251         /*
 252          * No need to reset region to identity mapped since we now
 253          * know that no I/O can be in this region
 254          */
 255         release_resource(resource);
 256         kfree(resource);
 257 }
 258
 259 static struct resource *additional_memory_resource(phys_addr_t size)
 260 {
 261         struct resource *res;
 262         int ret;
 263
 264         res = kzalloc(sizeof(*res), GFP_KERNEL);
 265         if (!res)
 266                 return NULL;
 267
 268         res->name = "System RAM";
 269         res->flags = IORESOURCE_SYSTEM_RAM | IORESOURCE_BUSY;
 270
 271         ret = allocate_resource(&iomem_resource, res,
 272                                 size, 0, -1,
 273                                 PAGES_PER_SECTION * PAGE_SIZE, NULL, NULL);
 274         if (ret < 0) {
 275                 pr_err("Cannot allocate new System RAM resource\n");
 276                 kfree(res);
 277                 return NULL;
 278         }
 279
 280 #ifdef CONFIG_SPARSEMEM
 281         {
 282                 unsigned long limit = 1UL << (MAX_PHYSMEM_BITS - PAGE_SHIFT);
 283                 unsigned long pfn = res->start >> PAGE_SHIFT;
 284
 285                 if (pfn > limit) {
 286                         pr_err("New System RAM resource outside addressable RAM (%lu > %lu)\n",
 287                                pfn, limit);
 288                         release_memory_resource(res);
 289                         return NULL;
 290                 }
 291         }
 292 #endif
 293
 294         return res;
 295 }
 296
 297 static enum bp_state reserve_additional_memory(void)
 298 {
 299         long credit;
 300         struct resource *resource;
 301         int nid, rc;
 302         unsigned long balloon_hotplug;
 303
 304         credit = balloon_stats.target_pages + balloon_stats.target_unpopulated
 305                 - balloon_stats.total_pages;
 306
 307         /*
 308          * Already hotplugged enough pages?  Wait for them to be
 309          * onlined.
 310          */
 311         if (credit <= 0)
 312                 return BP_WAIT;
 313
 314         balloon_hotplug = round_up(credit, PAGES_PER_SECTION);
 315
 316         resource = additional_memory_resource(balloon_hotplug * PAGE_SIZE);
 317         if (!resource)
 318                 goto err;
 319
 320         nid = memory_add_physaddr_to_nid(resource->start);
 321
 322 #ifdef CONFIG_XEN_HAVE_PVMMU
 323         /*
 324          * We don't support PV MMU when Linux and Xen is using
 325          * different page granularity.
 326          */
 327         BUILD_BUG_ON(XEN_PAGE_SIZE != PAGE_SIZE);
 328
 329         /*
 330          * add_memory() will build page tables for the new memory so
 331          * the p2m must contain invalid entries so the correct
 332          * non-present PTEs will be written.
 333          *
 334          * If a failure occurs, the original (identity) p2m entries
 335          * are not restored since this region is now known not to
 336          * conflict with any devices.
 337          */
 338         if (!xen_feature(XENFEAT_auto_translated_physmap)) {
 339                 unsigned long pfn, i;
 340
 341                 pfn = PFN_DOWN(resource->start);
 342                 for (i = 0; i < balloon_hotplug; i++) {
 343                         if (!set_phys_to_machine(pfn + i, INVALID_P2M_ENTRY)) {
 344                                 pr_warn("set_phys_to_machine() failed, no memory added\n");
 345                                 goto err;
 346                         }
 347                 }
 348         }
 349 #endif
 350
 351         /*
 352          * add_memory_resource() will call online_pages() which in its turn
 353          * will call xen_online_page() callback causing deadlock if we don't
 354          * release balloon_mutex here. Unlocking here is safe because the
 355          * callers drop the mutex before trying again.
 356          */
 357         mutex_unlock(&balloon_mutex);
 358         /* add_memory_resource() requires the device_hotplug lock */
 359         lock_device_hotplug();
 360         rc = add_memory_resource(nid, resource, memhp_auto_online);
 361         unlock_device_hotplug();
 362         mutex_lock(&balloon_mutex);
 363
 364         if (rc) {
 365                 pr_warn("Cannot add additional memory (%i)\n", rc);
 366                 goto err;
 367         }
 368
 369         balloon_stats.total_pages += balloon_hotplug;
 370
 371         return BP_WAIT;
 372   err:
 373         release_memory_resource(resource);
 374         return BP_ECANCELED;
 375 }
 376
 377 static void xen_online_page(struct page *page)
 378 {
 379         __online_page_set_limits(page);
 380
 381         mutex_lock(&balloon_mutex);
 382
 383         __balloon_append(page);
 384
 385         mutex_unlock(&balloon_mutex);
 386 }
 387
 388 static int xen_memory_notifier(struct notifier_block *nb, unsigned long val, void *v)
 389 {
 390         if (val == MEM_ONLINE)
 391                 wake_up(&balloon_thread_wq);
 392
 393         return NOTIFY_OK;
 394 }
 395
 396 static struct notifier_block xen_memory_nb = {
 397         .notifier_call = xen_memory_notifier,
 398         .priority = 0
 399 };
 400 #else
 401 static enum bp_state reserve_additional_memory(void)
 402 {
 403         balloon_stats.target_pages = balloon_stats.current_pages +
 404                                      balloon_stats.target_unpopulated;
 405         return BP_ECANCELED;
 406 }
 407 #endif /* CONFIG_XEN_BALLOON_MEMORY_HOTPLUG */
 408
 409 static long current_credit(void)
 410 {
 411         return balloon_stats.target_pages - balloon_stats.current_pages;
 412 }
 413
 414 static bool balloon_is_inflated(void)
 415 {
 416         return balloon_stats.balloon_low || balloon_stats.balloon_high;
 417 }
 418
 419 static enum bp_state increase_reservation(unsigned long nr_pages)
 420 {
 421         int rc;
 422         unsigned long i;
 423         struct page   *page;
 424
 425         if (nr_pages > ARRAY_SIZE(frame_list))
 426                 nr_pages = ARRAY_SIZE(frame_list);
 427
 428         page = list_first_entry_or_null(&ballooned_pages, struct page, lru);
 429         for (i = 0; i < nr_pages; i++) {
 430                 if (!page) {
 431                         nr_pages = i;
 432                         break;
 433                 }
 434
 435                 frame_list[i] = page_to_xen_pfn(page);
 436                 page = balloon_next_page(page);
 437         }
 438
 439         rc = xenmem_reservation_increase(nr_pages, frame_list);
 440         if (rc <= 0)
 441                 return BP_EAGAIN;
 442
 443         for (i = 0; i < rc; i++) {
 444                 page = balloon_retrieve(false);
 445                 BUG_ON(page == NULL);
 446
 447                 xenmem_reservation_va_mapping_update(1, &page, &frame_list[i]);
 448
 449                 /* Relinquish the page back to the allocator. */
 450                 free_reserved_page(page);
 451         }
 452
 453         balloon_stats.current_pages += rc;
 454
 455         return BP_DONE;
 456 }
 457
 458 static enum bp_state decrease_reservation(unsigned long nr_pages, gfp_t gfp)
 459 {
 460         enum bp_state state = BP_DONE;
 461         unsigned long i;
 462         struct page *page, *tmp;
 463         int ret;
 464         LIST_HEAD(pages);
 465
 466         if (nr_pages > ARRAY_SIZE(frame_list))
 467                 nr_pages = ARRAY_SIZE(frame_list);
 468
 469         for (i = 0; i < nr_pages; i++) {
 470                 page = alloc_page(gfp);
 471                 if (page == NULL) {
 472                         nr_pages = i;
 473                         state = BP_EAGAIN;
 474                         break;
 475                 }
 476                 adjust_managed_page_count(page, -1);
 477                 xenmem_reservation_scrub_page(page);
 478                 list_add(&page->lru, &pages);
 479         }
 480
 481         /*
 482          * Ensure that ballooned highmem pages don't have kmaps.
 483          *
 484          * Do this before changing the p2m as kmap_flush_unused()
 485          * reads PTEs to obtain pages (and hence needs the original
 486          * p2m entry).
 487          */
 488         kmap_flush_unused();
 489
 490         /*
 491          * Setup the frame, update direct mapping, invalidate P2M,
 492          * and add to balloon.
 493          */
 494         i = 0;
 495         list_for_each_entry_safe(page, tmp, &pages, lru) {
 496                 frame_list[i++] = xen_page_to_gfn(page);
 497
 498                 xenmem_reservation_va_mapping_reset(1, &page);
 499
 500                 list_del(&page->lru);
 501
 502                 balloon_append(page);
 503         }
 504
 505         flush_tlb_all();
 506
 507         ret = xenmem_reservation_decrease(nr_pages, frame_list);
 508         BUG_ON(ret != nr_pages);
 509
 510         balloon_stats.current_pages -= nr_pages;
 511
 512         return state;
 513 }
 514
 515 /*
 516  * Stop waiting if either state is BP_DONE and ballooning action is
 517  * needed, or if the credit has changed while state is not BP_DONE.
 518  */
 519 static bool balloon_thread_cond(long credit)
 520 {
 521         if (balloon_state == BP_DONE)
 522                 credit = 0;
 523
 524         return current_credit() != credit || kthread_should_stop();
 525 }
 526
 527 /*
 528  * As this is a kthread it is guaranteed to run as a single instance only.
 529  * We may of course race updates of the target counts (which are protected
 530  * by the balloon lock), or with changes to the Xen hard limit, but we will
 531  * recover from these in time.
 532  */
 533 static int balloon_thread(void *unused)
 534 {
 535         long credit;
 536         unsigned long timeout;
 537
 538         set_freezable();
 539         for (;;) {
 540                 switch (balloon_state) {
 541                 case BP_DONE:
 542                 case BP_ECANCELED:
 543                         timeout = 3600 * HZ;
 544                         break;
 545                 case BP_EAGAIN:
 546                         timeout = balloon_stats.schedule_delay * HZ;
 547                         break;
 548                 case BP_WAIT:
 549                         timeout = HZ;
 550                         break;
 551                 }
 552
 553                 credit = current_credit();
 554
 555                 wait_event_freezable_timeout(balloon_thread_wq,
 556                         balloon_thread_cond(credit), timeout);
 557
 558                 if (kthread_should_stop())
 559                         return 0;
 560
 561                 mutex_lock(&balloon_mutex);
 562
 563                 credit = current_credit();
 564
 565                 if (credit > 0) {
 566                         if (balloon_is_inflated())
 567                                 balloon_state = increase_reservation(credit);
 568                         else
 569                                 balloon_state = reserve_additional_memory();
 570                 }
 571
 572                 if (credit < 0) {
 573                         long n_pages;
 574
 575                         n_pages = min(-credit, si_mem_available());
 576                         balloon_state = decrease_reservation(n_pages,
 577                                                              GFP_BALLOON);
 578                         if (balloon_state == BP_DONE && n_pages != -credit &&
 579                             n_pages < totalreserve_pages)
 580                                 balloon_state = BP_EAGAIN;
 581                 }
 582
 583                 update_schedule();
 584
 585                 mutex_unlock(&balloon_mutex);
 586
 587                 cond_resched();
 588         }
 589 }
 590
 591 /* Resets the Xen limit, sets new target, and kicks off processing. */
 592 void balloon_set_new_target(unsigned long target)
 593 {
 594         /* No need for lock. Not read-modify-write updates. */
 595         balloon_stats.target_pages = target;
 596         wake_up(&balloon_thread_wq);
 597 }
 598 EXPORT_SYMBOL_GPL(balloon_set_new_target);
 599
 600 static int add_ballooned_pages(int nr_pages)
 601 {
 602         enum bp_state st;
 603
 604         if (xen_hotplug_unpopulated) {
 605                 st = reserve_additional_memory();
 606                 if (st != BP_ECANCELED) {
 607                         int rc;
 608
 609                         mutex_unlock(&balloon_mutex);
 610                         rc = wait_event_interruptible(balloon_wq,
 611                                    !list_empty(&ballooned_pages));
 612                         mutex_lock(&balloon_mutex);
 613                         return rc ? -ENOMEM : 0;
 614                 }
 615         }
 616
 617         if (si_mem_available() < nr_pages)
 618                 return -ENOMEM;
 619
 620         st = decrease_reservation(nr_pages, GFP_USER);
 621         if (st != BP_DONE)
 622                 return -ENOMEM;
 623
 624         return 0;
 625 }
 626
 627 /**
 628  * alloc_xenballooned_pages - get pages that have been ballooned out
 629  * @nr_pages: Number of pages to get
 630  * @pages: pages returned
 631  * @return 0 on success, error otherwise
 632  */
 633 int alloc_xenballooned_pages(int nr_pages, struct page **pages)
 634 {
 635         int pgno = 0;
 636         struct page *page;
 637         int ret;
 638
 639         mutex_lock(&balloon_mutex);
 640
 641         balloon_stats.target_unpopulated += nr_pages;
 642
 643         while (pgno < nr_pages) {
 644                 page = balloon_retrieve(true);
 645                 if (page) {
 646                         pages[pgno++] = page;
 647 #ifdef CONFIG_XEN_HAVE_PVMMU
 648                         /*
 649                          * We don't support PV MMU when Linux and Xen is using
 650                          * different page granularity.
 651                          */
 652                         BUILD_BUG_ON(XEN_PAGE_SIZE != PAGE_SIZE);
 653
 654                         if (!xen_feature(XENFEAT_auto_translated_physmap)) {
 655                                 ret = xen_alloc_p2m_entry(page_to_pfn(page));
 656                                 if (ret < 0)
 657                                         goto out_undo;
 658                         }
 659 #endif
 660                 } else {
 661                         ret = add_ballooned_pages(nr_pages - pgno);
 662                         if (ret < 0)
 663                                 goto out_undo;
 664                 }
 665         }
 666         mutex_unlock(&balloon_mutex);
 667         return 0;
 668  out_undo:
 669         mutex_unlock(&balloon_mutex);
 670         free_xenballooned_pages(pgno, pages);
 671         /*
 672          * NB: free_xenballooned_pages will only subtract pgno pages, but since
 673          * target_unpopulated is incremented with nr_pages at the start we need
 674          * to remove the remaining ones also, or accounting will be screwed.
 675          */
 676         balloon_stats.target_unpopulated -= nr_pages - pgno;
 677         return ret;
 678 }
 679 EXPORT_SYMBOL(alloc_xenballooned_pages);
 680
 681 /**
 682  * free_xenballooned_pages - return pages retrieved with get_ballooned_pages
 683  * @nr_pages: Number of pages
 684  * @pages: pages to return
 685  */
 686 void free_xenballooned_pages(int nr_pages, struct page **pages)
 687 {
 688         int i;
 689
 690         mutex_lock(&balloon_mutex);
 691
 692         for (i = 0; i < nr_pages; i++) {
 693                 if (pages[i])
 694                         balloon_append(pages[i]);
 695         }
 696
 697         balloon_stats.target_unpopulated -= nr_pages;
 698
 699         /* The balloon may be too large now. Shrink it if needed. */
 700         if (current_credit())
 701                 wake_up(&balloon_thread_wq);
 702
 703         mutex_unlock(&balloon_mutex);
 704 }
 705 EXPORT_SYMBOL(free_xenballooned_pages);
 706
 707 #ifdef CONFIG_XEN_PV
 708 static void __init balloon_add_region(unsigned long start_pfn,
 709                                       unsigned long pages)
 710 {
 711         unsigned long pfn, extra_pfn_end;
 712         struct page *page;
 713
 714         /*
 715          * If the amount of usable memory has been limited (e.g., with
 716          * the 'mem' command line parameter), don't add pages beyond
 717          * this limit.
 718          */
 719         extra_pfn_end = min(max_pfn, start_pfn + pages);
 720
 721         for (pfn = start_pfn; pfn < extra_pfn_end; pfn++) {
 722                 page = pfn_to_page(pfn);
 723                 /* totalram_pages and totalhigh_pages do not
 724                    include the boot-time balloon extension, so
 725                    don't subtract from it. */
 726                 __balloon_append(page);
 727         }
 728
 729         balloon_stats.total_pages += extra_pfn_end - start_pfn;
 730 }
 731 #endif
 732
 733 static int __init balloon_init(void)
 734 {
 735         struct task_struct *task;
 736
 737         if (!xen_domain())
 738                 return -ENODEV;
 739
 740         pr_info("Initialising balloon driver\n");
 741
 742 #ifdef CONFIG_XEN_PV
 743         balloon_stats.current_pages = xen_pv_domain()
 744                 ? min(xen_start_info->nr_pages - xen_released_pages, max_pfn)
 745                 : get_num_physpages();
 746 #else
 747         balloon_stats.current_pages = get_num_physpages();
 748 #endif
 749         balloon_stats.target_pages  = balloon_stats.current_pages;
 750         balloon_stats.balloon_low   = 0;
 751         balloon_stats.balloon_high  = 0;
 752         balloon_stats.total_pages   = balloon_stats.current_pages;
 753
 754         balloon_stats.schedule_delay = 1;
 755         balloon_stats.max_schedule_delay = 32;
 756         balloon_stats.retry_count = 1;
 757         balloon_stats.max_retry_count = 4;
 758
 759 #ifdef CONFIG_XEN_BALLOON_MEMORY_HOTPLUG
 760         set_online_page_callback(&xen_online_page);
 761         register_memory_notifier(&xen_memory_nb);
 762         register_sysctl_table(xen_root);
 763 #endif
 764
 765 #ifdef CONFIG_XEN_PV
 766         {
 767                 int i;
 768
 769                 /*
 770                  * Initialize the balloon with pages from the extra memory
 771                  * regions (see arch/x86/xen/setup.c).
 772                  */
 773                 for (i = 0; i < XEN_EXTRA_MEM_MAX_REGIONS; i++)
 774                         if (xen_extra_mem[i].n_pfns)
 775                                 balloon_add_region(xen_extra_mem[i].start_pfn,
 776                                                    xen_extra_mem[i].n_pfns);
 777         }
 778 #endif
 779
 780         task = kthread_run(balloon_thread, NULL, "xen-balloon");
 781         if (IS_ERR(task)) {
 782                 pr_err("xen-balloon thread could not be started, ballooning will not work!\n");
 783                 return PTR_ERR(task);
 784         }
 785
 786         /* Init the xen-balloon driver. */
 787         xen_balloon_init();
 788
 789         return 0;
 790 }
 791 subsys_initcall(balloon_init);
 792
 793 static int __init balloon_wait_finish(void)
 794 {
 795         long credit, last_credit = 0;
 796         unsigned long last_changed = 0;
 797
 798         if (!xen_domain())
 799                 return -ENODEV;
 800
 801         /* PV guests don't need to wait. */
 802         if (xen_pv_domain() || !current_credit())
 803                 return 0;
 804
 805         pr_notice("Waiting for initial ballooning down having finished.\n");
 806
 807         while ((credit = current_credit()) < 0) {
 808                 if (credit != last_credit) {
 809                         last_changed = jiffies;
 810                         last_credit = credit;
 811                 }
 812                 if (balloon_state == BP_ECANCELED) {
 813                         pr_warn_once("Initial ballooning failed, %ld pages need to be freed.\n",
 814                                      -credit);
 815                         if (jiffies - last_changed >= HZ * balloon_boot_timeout)
 816                                 panic("Initial ballooning failed!\n");
 817                 }
 818
 819                 schedule_timeout_interruptible(HZ / 10);
 820         }
 821
 822         pr_notice("Initial ballooning down finished.\n");
 823
 824         return 0;
 825 }
 826 late_initcall_sync(balloon_wait_finish);