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  *
   8  * This program is free software; you can redistribute it and/or
   9  * modify it under the terms of the GNU General Public License version 2
  10  * as published by the Free Software Foundation; or, when distributed
  11  * separately from the Linux kernel or incorporated into other
  12  * software packages, subject to the following license:
  13  *
  14  * Permission is hereby granted, free of charge, to any person obtaining a copy
  15  * of this source file (the "Software"), to deal in the Software without
  16  * restriction, including without limitation the rights to use, copy, modify,
  17  * merge, publish, distribute, sublicense, and/or sell copies of the Software,
  18  * and to permit persons to whom the Software is furnished to do so, subject to
  19  * the following conditions:
  20  *
  21  * The above copyright notice and this permission notice shall be included in
  22  * all copies or substantial portions of the Software.
  23  *
  24  * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
  25  * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
  26  * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
  27  * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
  28  * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
  29  * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
  30  * IN THE SOFTWARE.
  31  */
  32
  33 #include <linux/kernel.h>
  34 #include <linux/sched.h>
  35 #include <linux/errno.h>
  36 #include <linux/mm.h>
  37 #include <linux/bootmem.h>
  38 #include <linux/pagemap.h>
  39 #include <linux/highmem.h>
  40 #include <linux/mutex.h>
  41 #include <linux/list.h>
  42 #include <linux/gfp.h>
  43
  44 #include <asm/page.h>
  45 #include <asm/pgalloc.h>
  46 #include <asm/pgtable.h>
  47 #include <asm/tlb.h>
  48 #include <asm/e820.h>
  49
  50 #include <asm/xen/hypervisor.h>
  51 #include <asm/xen/hypercall.h>
  52
  53 #include <xen/xen.h>
  54 #include <xen/interface/xen.h>
  55 #include <xen/interface/memory.h>
  56 #include <xen/balloon.h>
  57 #include <xen/features.h>
  58 #include <xen/page.h>
  59
  60 /*
  61  * balloon_process() state:
  62  *
  63  * BP_DONE: done or nothing to do,
  64  * BP_EAGAIN: error, go to sleep,
  65  * BP_ECANCELED: error, balloon operation canceled.
  66  */
  67
  68 enum bp_state {
  69         BP_DONE,
  70         BP_EAGAIN,
  71         BP_ECANCELED
  72 };
  73
  74
  75 static DEFINE_MUTEX(balloon_mutex);
  76
  77 struct balloon_stats balloon_stats;
  78 EXPORT_SYMBOL_GPL(balloon_stats);
  79
  80 /* We increase/decrease in batches which fit in a page */
  81 static unsigned long frame_list[PAGE_SIZE / sizeof(unsigned long)];
  82
  83 #ifdef CONFIG_HIGHMEM
  84 #define inc_totalhigh_pages() (totalhigh_pages++)
  85 #define dec_totalhigh_pages() (totalhigh_pages--)
  86 #else
  87 #define inc_totalhigh_pages() do {} while(0)
  88 #define dec_totalhigh_pages() do {} while(0)
  89 #endif
  90
  91 /* List of ballooned pages, threaded through the mem_map array. */
  92 static LIST_HEAD(ballooned_pages);
  93
  94 /* Main work function, always executed in process context. */
  95 static void balloon_process(struct work_struct *work);
  96 static DECLARE_DELAYED_WORK(balloon_worker, balloon_process);
  97
  98 /* When ballooning out (allocating memory to return to Xen) we don't really
  99    want the kernel to try too hard since that can trigger the oom killer. */
 100 #define GFP_BALLOON \
 101         (GFP_HIGHUSER | __GFP_NOWARN | __GFP_NORETRY | __GFP_NOMEMALLOC)
 102
 103 static void scrub_page(struct page *page)
 104 {
 105 #ifdef CONFIG_XEN_SCRUB_PAGES
 106         clear_highpage(page);
 107 #endif
 108 }
 109
 110 /* balloon_append: add the given page to the balloon. */
 111 static void __balloon_append(struct page *page)
 112 {
 113         /* Lowmem is re-populated first, so highmem pages go at list tail. */
 114         if (PageHighMem(page)) {
 115                 list_add_tail(&page->lru, &ballooned_pages);
 116                 balloon_stats.balloon_high++;
 117         } else {
 118                 list_add(&page->lru, &ballooned_pages);
 119                 balloon_stats.balloon_low++;
 120         }
 121 }
 122
 123 static void balloon_append(struct page *page)
 124 {
 125         __balloon_append(page);
 126         if (PageHighMem(page))
 127                 dec_totalhigh_pages();
 128         totalram_pages--;
 129 }
 130
 131 /* balloon_retrieve: rescue a page from the balloon, if it is not empty. */
 132 static struct page *balloon_retrieve(bool prefer_highmem)
 133 {
 134         struct page *page;
 135
 136         if (list_empty(&ballooned_pages))
 137                 return NULL;
 138
 139         if (prefer_highmem)
 140                 page = list_entry(ballooned_pages.prev, struct page, lru);
 141         else
 142                 page = list_entry(ballooned_pages.next, struct page, lru);
 143         list_del(&page->lru);
 144
 145         if (PageHighMem(page)) {
 146                 balloon_stats.balloon_high--;
 147                 inc_totalhigh_pages();
 148         }
 149         else
 150                 balloon_stats.balloon_low--;
 151
 152         totalram_pages++;
 153
 154         return page;
 155 }
 156
 157 static struct page *balloon_first_page(void)
 158 {
 159         if (list_empty(&ballooned_pages))
 160                 return NULL;
 161         return list_entry(ballooned_pages.next, struct page, lru);
 162 }
 163
 164 static struct page *balloon_next_page(struct page *page)
 165 {
 166         struct list_head *next = page->lru.next;
 167         if (next == &ballooned_pages)
 168                 return NULL;
 169         return list_entry(next, struct page, lru);
 170 }
 171
 172 static enum bp_state update_schedule(enum bp_state state)
 173 {
 174         if (state == BP_DONE) {
 175                 balloon_stats.schedule_delay = 1;
 176                 balloon_stats.retry_count = 1;
 177                 return BP_DONE;
 178         }
 179
 180         ++balloon_stats.retry_count;
 181
 182         if (balloon_stats.max_retry_count != RETRY_UNLIMITED &&
 183                         balloon_stats.retry_count > balloon_stats.max_retry_count) {
 184                 balloon_stats.schedule_delay = 1;
 185                 balloon_stats.retry_count = 1;
 186                 return BP_ECANCELED;
 187         }
 188
 189         balloon_stats.schedule_delay <<= 1;
 190
 191         if (balloon_stats.schedule_delay > balloon_stats.max_schedule_delay)
 192                 balloon_stats.schedule_delay = balloon_stats.max_schedule_delay;
 193
 194         return BP_EAGAIN;
 195 }
 196
 197 static long current_credit(void)
 198 {
 199         unsigned long target = balloon_stats.target_pages;
 200
 201         target = min(target,
 202                      balloon_stats.current_pages +
 203                      balloon_stats.balloon_low +
 204                      balloon_stats.balloon_high);
 205
 206         return target - balloon_stats.current_pages;
 207 }
 208
 209 static enum bp_state increase_reservation(unsigned long nr_pages)
 210 {
 211         int rc;
 212         unsigned long  pfn, i;
 213         struct page   *page;
 214         struct xen_memory_reservation reservation = {
 215                 .address_bits = 0,
 216                 .extent_order = 0,
 217                 .domid        = DOMID_SELF
 218         };
 219
 220         if (nr_pages > ARRAY_SIZE(frame_list))
 221                 nr_pages = ARRAY_SIZE(frame_list);
 222
 223         page = balloon_first_page();
 224         for (i = 0; i < nr_pages; i++) {
 225                 if (!page) {
 226                         nr_pages = i;
 227                         break;
 228                 }
 229                 frame_list[i] = page_to_pfn(page);
 230                 page = balloon_next_page(page);
 231         }
 232
 233         set_xen_guest_handle(reservation.extent_start, frame_list);
 234         reservation.nr_extents = nr_pages;
 235         rc = HYPERVISOR_memory_op(XENMEM_populate_physmap, &reservation);
 236         if (rc <= 0)
 237                 return BP_EAGAIN;
 238
 239         for (i = 0; i < rc; i++) {
 240                 page = balloon_retrieve(false);
 241                 BUG_ON(page == NULL);
 242
 243                 pfn = page_to_pfn(page);
 244                 BUG_ON(!xen_feature(XENFEAT_auto_translated_physmap) &&
 245                        phys_to_machine_mapping_valid(pfn));
 246
 247                 set_phys_to_machine(pfn, frame_list[i]);
 248
 249                 /* Link back into the page tables if not highmem. */
 250                 if (xen_pv_domain() && !PageHighMem(page)) {
 251                         int ret;
 252                         ret = HYPERVISOR_update_va_mapping(
 253                                 (unsigned long)__va(pfn << PAGE_SHIFT),
 254                                 mfn_pte(frame_list[i], PAGE_KERNEL),
 255                                 0);
 256                         BUG_ON(ret);
 257                 }
 258
 259                 /* Relinquish the page back to the allocator. */
 260                 ClearPageReserved(page);
 261                 init_page_count(page);
 262                 __free_page(page);
 263         }
 264
 265         balloon_stats.current_pages += rc;
 266
 267         return BP_DONE;
 268 }
 269
 270 static enum bp_state decrease_reservation(unsigned long nr_pages, gfp_t gfp)
 271 {
 272         enum bp_state state = BP_DONE;
 273         unsigned long  pfn, i;
 274         struct page   *page;
 275         int ret;
 276         struct xen_memory_reservation reservation = {
 277                 .address_bits = 0,
 278                 .extent_order = 0,
 279                 .domid        = DOMID_SELF
 280         };
 281
 282         if (nr_pages > ARRAY_SIZE(frame_list))
 283                 nr_pages = ARRAY_SIZE(frame_list);
 284
 285         for (i = 0; i < nr_pages; i++) {
 286                 if ((page = alloc_page(gfp)) == NULL) {
 287                         nr_pages = i;
 288                         state = BP_EAGAIN;
 289                         break;
 290                 }
 291
 292                 pfn = page_to_pfn(page);
 293                 frame_list[i] = pfn_to_mfn(pfn);
 294
 295                 scrub_page(page);
 296
 297                 if (xen_pv_domain() && !PageHighMem(page)) {
 298                         ret = HYPERVISOR_update_va_mapping(
 299                                 (unsigned long)__va(pfn << PAGE_SHIFT),
 300                                 __pte_ma(0), 0);
 301                         BUG_ON(ret);
 302                 }
 303
 304         }
 305
 306         /* Ensure that ballooned highmem pages don't have kmaps. */
 307         kmap_flush_unused();
 308         flush_tlb_all();
 309
 310         /* No more mappings: invalidate P2M and add to balloon. */
 311         for (i = 0; i < nr_pages; i++) {
 312                 pfn = mfn_to_pfn(frame_list[i]);
 313                 __set_phys_to_machine(pfn, INVALID_P2M_ENTRY);
 314                 balloon_append(pfn_to_page(pfn));
 315         }
 316
 317         set_xen_guest_handle(reservation.extent_start, frame_list);
 318         reservation.nr_extents   = nr_pages;
 319         ret = HYPERVISOR_memory_op(XENMEM_decrease_reservation, &reservation);
 320         BUG_ON(ret != nr_pages);
 321
 322         balloon_stats.current_pages -= nr_pages;
 323
 324         return state;
 325 }
 326
 327 /*
 328  * We avoid multiple worker processes conflicting via the balloon mutex.
 329  * We may of course race updates of the target counts (which are protected
 330  * by the balloon lock), or with changes to the Xen hard limit, but we will
 331  * recover from these in time.
 332  */
 333 static void balloon_process(struct work_struct *work)
 334 {
 335         enum bp_state state = BP_DONE;
 336         long credit;
 337
 338         mutex_lock(&balloon_mutex);
 339
 340         do {
 341                 credit = current_credit();
 342
 343                 if (credit > 0)
 344                         state = increase_reservation(credit);
 345
 346                 if (credit < 0)
 347                         state = decrease_reservation(-credit, GFP_BALLOON);
 348
 349                 state = update_schedule(state);
 350
 351 #ifndef CONFIG_PREEMPT
 352                 if (need_resched())
 353                         schedule();
 354 #endif
 355         } while (credit && state == BP_DONE);
 356
 357         /* Schedule more work if there is some still to be done. */
 358         if (state == BP_EAGAIN)
 359                 schedule_delayed_work(&balloon_worker, balloon_stats.schedule_delay * HZ);
 360
 361         mutex_unlock(&balloon_mutex);
 362 }
 363
 364 /* Resets the Xen limit, sets new target, and kicks off processing. */
 365 void balloon_set_new_target(unsigned long target)
 366 {
 367         /* No need for lock. Not read-modify-write updates. */
 368         balloon_stats.target_pages = target;
 369         schedule_delayed_work(&balloon_worker, 0);
 370 }
 371 EXPORT_SYMBOL_GPL(balloon_set_new_target);
 372
 373 /**
 374  * alloc_xenballooned_pages - get pages that have been ballooned out
 375  * @nr_pages: Number of pages to get
 376  * @pages: pages returned
 377  * @return 0 on success, error otherwise
 378  */
 379 int alloc_xenballooned_pages(int nr_pages, struct page** pages)
 380 {
 381         int pgno = 0;
 382         struct page* page;
 383         mutex_lock(&balloon_mutex);
 384         while (pgno < nr_pages) {
 385                 page = balloon_retrieve(true);
 386                 if (page) {
 387                         pages[pgno++] = page;
 388                 } else {
 389                         enum bp_state st;
 390                         st = decrease_reservation(nr_pages - pgno, GFP_HIGHUSER);
 391                         if (st != BP_DONE)
 392                                 goto out_undo;
 393                 }
 394         }
 395         mutex_unlock(&balloon_mutex);
 396         return 0;
 397  out_undo:
 398         while (pgno)
 399                 balloon_append(pages[--pgno]);
 400         /* Free the memory back to the kernel soon */
 401         schedule_delayed_work(&balloon_worker, 0);
 402         mutex_unlock(&balloon_mutex);
 403         return -ENOMEM;
 404 }
 405 EXPORT_SYMBOL(alloc_xenballooned_pages);
 406
 407 /**
 408  * free_xenballooned_pages - return pages retrieved with get_ballooned_pages
 409  * @nr_pages: Number of pages
 410  * @pages: pages to return
 411  */
 412 void free_xenballooned_pages(int nr_pages, struct page** pages)
 413 {
 414         int i;
 415
 416         mutex_lock(&balloon_mutex);
 417
 418         for (i = 0; i < nr_pages; i++) {
 419                 if (pages[i])
 420                         balloon_append(pages[i]);
 421         }
 422
 423         /* The balloon may be too large now. Shrink it if needed. */
 424         if (current_credit())
 425                 schedule_delayed_work(&balloon_worker, 0);
 426
 427         mutex_unlock(&balloon_mutex);
 428 }
 429 EXPORT_SYMBOL(free_xenballooned_pages);
 430
 431 static int __init balloon_init(void)
 432 {
 433         unsigned long pfn, extra_pfn_end;
 434         struct page *page;
 435
 436         if (!xen_domain())
 437                 return -ENODEV;
 438
 439         pr_info("xen/balloon: Initialising balloon driver.\n");
 440
 441         balloon_stats.current_pages = xen_pv_domain() ? min(xen_start_info->nr_pages, max_pfn) : max_pfn;
 442         balloon_stats.target_pages  = balloon_stats.current_pages;
 443         balloon_stats.balloon_low   = 0;
 444         balloon_stats.balloon_high  = 0;
 445
 446         balloon_stats.schedule_delay = 1;
 447         balloon_stats.max_schedule_delay = 32;
 448         balloon_stats.retry_count = 1;
 449         balloon_stats.max_retry_count = RETRY_UNLIMITED;
 450
 451         /*
 452          * Initialise the balloon with excess memory space.  We need
 453          * to make sure we don't add memory which doesn't exist or
 454          * logically exist.  The E820 map can be trimmed to be smaller
 455          * than the amount of physical memory due to the mem= command
 456          * line parameter.  And if this is a 32-bit non-HIGHMEM kernel
 457          * on a system with memory which requires highmem to access,
 458          * don't try to use it.
 459          */
 460         extra_pfn_end = min(min(max_pfn, e820_end_of_ram_pfn()),
 461                             (unsigned long)PFN_DOWN(xen_extra_mem_start + xen_extra_mem_size));
 462         for (pfn = PFN_UP(xen_extra_mem_start);
 463              pfn < extra_pfn_end;
 464              pfn++) {
 465                 page = pfn_to_page(pfn);
 466                 /* totalram_pages and totalhigh_pages do not include the boot-time
 467                    balloon extension, so don't subtract from it. */
 468                 __balloon_append(page);
 469         }
 470
 471         return 0;
 472 }
 473
 474 subsys_initcall(balloon_init);
 475
 476 MODULE_LICENSE("GPL");