]> git.karo-electronics.de Git - karo-tx-linux.git/blob - drivers/xen/balloon.c
Merge branch 'for-linus' of git://www.jni.nu/cris
[karo-tx-linux.git] / drivers / xen / balloon.c
1 /******************************************************************************
2  * balloon.c
3  *
4  * Xen balloon driver - enables returning/claiming memory to/from Xen.
5  *
6  * Copyright (c) 2003, B Dragovic
7  * Copyright (c) 2003-2004, M Williamson, K Fraser
8  * Copyright (c) 2005 Dan M. Smith, IBM Corporation
9  *
10  * This program is free software; you can redistribute it and/or
11  * modify it under the terms of the GNU General Public License version 2
12  * as published by the Free Software Foundation; or, when distributed
13  * separately from the Linux kernel or incorporated into other
14  * software packages, subject to the following license:
15  *
16  * Permission is hereby granted, free of charge, to any person obtaining a copy
17  * of this source file (the "Software"), to deal in the Software without
18  * restriction, including without limitation the rights to use, copy, modify,
19  * merge, publish, distribute, sublicense, and/or sell copies of the Software,
20  * and to permit persons to whom the Software is furnished to do so, subject to
21  * the following conditions:
22  *
23  * The above copyright notice and this permission notice shall be included in
24  * all copies or substantial portions of the Software.
25  *
26  * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
27  * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
28  * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
29  * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
30  * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
31  * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
32  * IN THE SOFTWARE.
33  */
34
35 #include <linux/kernel.h>
36 #include <linux/module.h>
37 #include <linux/sched.h>
38 #include <linux/errno.h>
39 #include <linux/mm.h>
40 #include <linux/bootmem.h>
41 #include <linux/pagemap.h>
42 #include <linux/highmem.h>
43 #include <linux/mutex.h>
44 #include <linux/highmem.h>
45 #include <linux/list.h>
46 #include <linux/sysdev.h>
47
48 #include <asm/xen/hypervisor.h>
49 #include <asm/page.h>
50 #include <asm/pgalloc.h>
51 #include <asm/pgtable.h>
52 #include <asm/uaccess.h>
53 #include <asm/tlb.h>
54
55 #include <xen/interface/memory.h>
56 #include <xen/xenbus.h>
57 #include <xen/features.h>
58 #include <xen/page.h>
59
60 #define PAGES2KB(_p) ((_p)<<(PAGE_SHIFT-10))
61
62 #define BALLOON_CLASS_NAME "xen_memory"
63
64 struct balloon_stats {
65         /* We aim for 'current allocation' == 'target allocation'. */
66         unsigned long current_pages;
67         unsigned long target_pages;
68         /* We may hit the hard limit in Xen. If we do then we remember it. */
69         unsigned long hard_limit;
70         /*
71          * Drivers may alter the memory reservation independently, but they
72          * must inform the balloon driver so we avoid hitting the hard limit.
73          */
74         unsigned long driver_pages;
75         /* Number of pages in high- and low-memory balloons. */
76         unsigned long balloon_low;
77         unsigned long balloon_high;
78 };
79
80 static DEFINE_MUTEX(balloon_mutex);
81
82 static struct sys_device balloon_sysdev;
83
84 static int register_balloon(struct sys_device *sysdev);
85
86 /*
87  * Protects atomic reservation decrease/increase against concurrent increases.
88  * Also protects non-atomic updates of current_pages and driver_pages, and
89  * balloon lists.
90  */
91 static DEFINE_SPINLOCK(balloon_lock);
92
93 static struct balloon_stats balloon_stats;
94
95 /* We increase/decrease in batches which fit in a page */
96 static unsigned long frame_list[PAGE_SIZE / sizeof(unsigned long)];
97
98 /* VM /proc information for memory */
99 extern unsigned long totalram_pages;
100
101 #ifdef CONFIG_HIGHMEM
102 extern unsigned long totalhigh_pages;
103 #define inc_totalhigh_pages() (totalhigh_pages++)
104 #define dec_totalhigh_pages() (totalhigh_pages--)
105 #else
106 #define inc_totalhigh_pages() do {} while(0)
107 #define dec_totalhigh_pages() do {} while(0)
108 #endif
109
110 /* List of ballooned pages, threaded through the mem_map array. */
111 static LIST_HEAD(ballooned_pages);
112
113 /* Main work function, always executed in process context. */
114 static void balloon_process(struct work_struct *work);
115 static DECLARE_WORK(balloon_worker, balloon_process);
116 static struct timer_list balloon_timer;
117
118 /* When ballooning out (allocating memory to return to Xen) we don't really
119    want the kernel to try too hard since that can trigger the oom killer. */
120 #define GFP_BALLOON \
121         (GFP_HIGHUSER | __GFP_NOWARN | __GFP_NORETRY | __GFP_NOMEMALLOC)
122
123 static void scrub_page(struct page *page)
124 {
125 #ifdef CONFIG_XEN_SCRUB_PAGES
126         if (PageHighMem(page)) {
127                 void *v = kmap(page);
128                 clear_page(v);
129                 kunmap(v);
130         } else {
131                 void *v = page_address(page);
132                 clear_page(v);
133         }
134 #endif
135 }
136
137 /* balloon_append: add the given page to the balloon. */
138 static void balloon_append(struct page *page)
139 {
140         /* Lowmem is re-populated first, so highmem pages go at list tail. */
141         if (PageHighMem(page)) {
142                 list_add_tail(&page->lru, &ballooned_pages);
143                 balloon_stats.balloon_high++;
144                 dec_totalhigh_pages();
145         } else {
146                 list_add(&page->lru, &ballooned_pages);
147                 balloon_stats.balloon_low++;
148         }
149 }
150
151 /* balloon_retrieve: rescue a page from the balloon, if it is not empty. */
152 static struct page *balloon_retrieve(void)
153 {
154         struct page *page;
155
156         if (list_empty(&ballooned_pages))
157                 return NULL;
158
159         page = list_entry(ballooned_pages.next, struct page, lru);
160         list_del(&page->lru);
161
162         if (PageHighMem(page)) {
163                 balloon_stats.balloon_high--;
164                 inc_totalhigh_pages();
165         }
166         else
167                 balloon_stats.balloon_low--;
168
169         return page;
170 }
171
172 static struct page *balloon_first_page(void)
173 {
174         if (list_empty(&ballooned_pages))
175                 return NULL;
176         return list_entry(ballooned_pages.next, struct page, lru);
177 }
178
179 static struct page *balloon_next_page(struct page *page)
180 {
181         struct list_head *next = page->lru.next;
182         if (next == &ballooned_pages)
183                 return NULL;
184         return list_entry(next, struct page, lru);
185 }
186
187 static void balloon_alarm(unsigned long unused)
188 {
189         schedule_work(&balloon_worker);
190 }
191
192 static unsigned long current_target(void)
193 {
194         unsigned long target = min(balloon_stats.target_pages, balloon_stats.hard_limit);
195
196         target = min(target,
197                      balloon_stats.current_pages +
198                      balloon_stats.balloon_low +
199                      balloon_stats.balloon_high);
200
201         return target;
202 }
203
204 static int increase_reservation(unsigned long nr_pages)
205 {
206         unsigned long  pfn, i, flags;
207         struct page   *page;
208         long           rc;
209         struct xen_memory_reservation reservation = {
210                 .address_bits = 0,
211                 .extent_order = 0,
212                 .domid        = DOMID_SELF
213         };
214
215         if (nr_pages > ARRAY_SIZE(frame_list))
216                 nr_pages = ARRAY_SIZE(frame_list);
217
218         spin_lock_irqsave(&balloon_lock, flags);
219
220         page = balloon_first_page();
221         for (i = 0; i < nr_pages; i++) {
222                 BUG_ON(page == NULL);
223                 frame_list[i] = page_to_pfn(page);;
224                 page = balloon_next_page(page);
225         }
226
227         set_xen_guest_handle(reservation.extent_start, frame_list);
228         reservation.nr_extents = nr_pages;
229         rc = HYPERVISOR_memory_op(XENMEM_populate_physmap, &reservation);
230         if (rc < nr_pages) {
231                 if (rc > 0) {
232                         int ret;
233
234                         /* We hit the Xen hard limit: reprobe. */
235                         reservation.nr_extents = rc;
236                         ret = HYPERVISOR_memory_op(XENMEM_decrease_reservation,
237                                                    &reservation);
238                         BUG_ON(ret != rc);
239                 }
240                 if (rc >= 0)
241                         balloon_stats.hard_limit = (balloon_stats.current_pages + rc -
242                                                     balloon_stats.driver_pages);
243                 goto out;
244         }
245
246         for (i = 0; i < nr_pages; i++) {
247                 page = balloon_retrieve();
248                 BUG_ON(page == NULL);
249
250                 pfn = page_to_pfn(page);
251                 BUG_ON(!xen_feature(XENFEAT_auto_translated_physmap) &&
252                        phys_to_machine_mapping_valid(pfn));
253
254                 set_phys_to_machine(pfn, frame_list[i]);
255
256                 /* Link back into the page tables if not highmem. */
257                 if (pfn < max_low_pfn) {
258                         int ret;
259                         ret = HYPERVISOR_update_va_mapping(
260                                 (unsigned long)__va(pfn << PAGE_SHIFT),
261                                 mfn_pte(frame_list[i], PAGE_KERNEL),
262                                 0);
263                         BUG_ON(ret);
264                 }
265
266                 /* Relinquish the page back to the allocator. */
267                 ClearPageReserved(page);
268                 init_page_count(page);
269                 __free_page(page);
270         }
271
272         balloon_stats.current_pages += nr_pages;
273         totalram_pages = balloon_stats.current_pages;
274
275  out:
276         spin_unlock_irqrestore(&balloon_lock, flags);
277
278         return 0;
279 }
280
281 static int decrease_reservation(unsigned long nr_pages)
282 {
283         unsigned long  pfn, i, flags;
284         struct page   *page;
285         int            need_sleep = 0;
286         int ret;
287         struct xen_memory_reservation reservation = {
288                 .address_bits = 0,
289                 .extent_order = 0,
290                 .domid        = DOMID_SELF
291         };
292
293         if (nr_pages > ARRAY_SIZE(frame_list))
294                 nr_pages = ARRAY_SIZE(frame_list);
295
296         for (i = 0; i < nr_pages; i++) {
297                 if ((page = alloc_page(GFP_BALLOON)) == NULL) {
298                         nr_pages = i;
299                         need_sleep = 1;
300                         break;
301                 }
302
303                 pfn = page_to_pfn(page);
304                 frame_list[i] = pfn_to_mfn(pfn);
305
306                 scrub_page(page);
307         }
308
309         /* Ensure that ballooned highmem pages don't have kmaps. */
310         kmap_flush_unused();
311         flush_tlb_all();
312
313         spin_lock_irqsave(&balloon_lock, flags);
314
315         /* No more mappings: invalidate P2M and add to balloon. */
316         for (i = 0; i < nr_pages; i++) {
317                 pfn = mfn_to_pfn(frame_list[i]);
318                 set_phys_to_machine(pfn, INVALID_P2M_ENTRY);
319                 balloon_append(pfn_to_page(pfn));
320         }
321
322         set_xen_guest_handle(reservation.extent_start, frame_list);
323         reservation.nr_extents   = nr_pages;
324         ret = HYPERVISOR_memory_op(XENMEM_decrease_reservation, &reservation);
325         BUG_ON(ret != nr_pages);
326
327         balloon_stats.current_pages -= nr_pages;
328         totalram_pages = balloon_stats.current_pages;
329
330         spin_unlock_irqrestore(&balloon_lock, flags);
331
332         return need_sleep;
333 }
334
335 /*
336  * We avoid multiple worker processes conflicting via the balloon mutex.
337  * We may of course race updates of the target counts (which are protected
338  * by the balloon lock), or with changes to the Xen hard limit, but we will
339  * recover from these in time.
340  */
341 static void balloon_process(struct work_struct *work)
342 {
343         int need_sleep = 0;
344         long credit;
345
346         mutex_lock(&balloon_mutex);
347
348         do {
349                 credit = current_target() - balloon_stats.current_pages;
350                 if (credit > 0)
351                         need_sleep = (increase_reservation(credit) != 0);
352                 if (credit < 0)
353                         need_sleep = (decrease_reservation(-credit) != 0);
354
355 #ifndef CONFIG_PREEMPT
356                 if (need_resched())
357                         schedule();
358 #endif
359         } while ((credit != 0) && !need_sleep);
360
361         /* Schedule more work if there is some still to be done. */
362         if (current_target() != balloon_stats.current_pages)
363                 mod_timer(&balloon_timer, jiffies + HZ);
364
365         mutex_unlock(&balloon_mutex);
366 }
367
368 /* Resets the Xen limit, sets new target, and kicks off processing. */
369 static void balloon_set_new_target(unsigned long target)
370 {
371         /* No need for lock. Not read-modify-write updates. */
372         balloon_stats.hard_limit   = ~0UL;
373         balloon_stats.target_pages = target;
374         schedule_work(&balloon_worker);
375 }
376
377 static struct xenbus_watch target_watch =
378 {
379         .node = "memory/target"
380 };
381
382 /* React to a change in the target key */
383 static void watch_target(struct xenbus_watch *watch,
384                          const char **vec, unsigned int len)
385 {
386         unsigned long long new_target;
387         int err;
388
389         err = xenbus_scanf(XBT_NIL, "memory", "target", "%llu", &new_target);
390         if (err != 1) {
391                 /* This is ok (for domain0 at least) - so just return */
392                 return;
393         }
394
395         /* The given memory/target value is in KiB, so it needs converting to
396          * pages. PAGE_SHIFT converts bytes to pages, hence PAGE_SHIFT - 10.
397          */
398         balloon_set_new_target(new_target >> (PAGE_SHIFT - 10));
399 }
400
401 static int balloon_init_watcher(struct notifier_block *notifier,
402                                 unsigned long event,
403                                 void *data)
404 {
405         int err;
406
407         err = register_xenbus_watch(&target_watch);
408         if (err)
409                 printk(KERN_ERR "Failed to set balloon watcher\n");
410
411         return NOTIFY_DONE;
412 }
413
414 static struct notifier_block xenstore_notifier;
415
416 static int __init balloon_init(void)
417 {
418         unsigned long pfn;
419         struct page *page;
420
421         if (!xen_pv_domain())
422                 return -ENODEV;
423
424         pr_info("xen_balloon: Initialising balloon driver.\n");
425
426         balloon_stats.current_pages = min(xen_start_info->nr_pages, max_pfn);
427         totalram_pages   = balloon_stats.current_pages;
428         balloon_stats.target_pages  = balloon_stats.current_pages;
429         balloon_stats.balloon_low   = 0;
430         balloon_stats.balloon_high  = 0;
431         balloon_stats.driver_pages  = 0UL;
432         balloon_stats.hard_limit    = ~0UL;
433
434         init_timer(&balloon_timer);
435         balloon_timer.data = 0;
436         balloon_timer.function = balloon_alarm;
437
438         register_balloon(&balloon_sysdev);
439
440         /* Initialise the balloon with excess memory space. */
441         for (pfn = xen_start_info->nr_pages; pfn < max_pfn; pfn++) {
442                 page = pfn_to_page(pfn);
443                 if (!PageReserved(page))
444                         balloon_append(page);
445         }
446
447         target_watch.callback = watch_target;
448         xenstore_notifier.notifier_call = balloon_init_watcher;
449
450         register_xenstore_notifier(&xenstore_notifier);
451
452         return 0;
453 }
454
455 subsys_initcall(balloon_init);
456
457 static void balloon_exit(void)
458 {
459     /* XXX - release balloon here */
460     return;
461 }
462
463 module_exit(balloon_exit);
464
465 #define BALLOON_SHOW(name, format, args...)                             \
466         static ssize_t show_##name(struct sys_device *dev,              \
467                                    struct sysdev_attribute *attr,       \
468                                    char *buf)                           \
469         {                                                               \
470                 return sprintf(buf, format, ##args);                    \
471         }                                                               \
472         static SYSDEV_ATTR(name, S_IRUGO, show_##name, NULL)
473
474 BALLOON_SHOW(current_kb, "%lu\n", PAGES2KB(balloon_stats.current_pages));
475 BALLOON_SHOW(low_kb, "%lu\n", PAGES2KB(balloon_stats.balloon_low));
476 BALLOON_SHOW(high_kb, "%lu\n", PAGES2KB(balloon_stats.balloon_high));
477 BALLOON_SHOW(hard_limit_kb,
478              (balloon_stats.hard_limit!=~0UL) ? "%lu\n" : "???\n",
479              (balloon_stats.hard_limit!=~0UL) ? PAGES2KB(balloon_stats.hard_limit) : 0);
480 BALLOON_SHOW(driver_kb, "%lu\n", PAGES2KB(balloon_stats.driver_pages));
481
482 static ssize_t show_target_kb(struct sys_device *dev, struct sysdev_attribute *attr,
483                               char *buf)
484 {
485         return sprintf(buf, "%lu\n", PAGES2KB(balloon_stats.target_pages));
486 }
487
488 static ssize_t store_target_kb(struct sys_device *dev,
489                                struct sysdev_attribute *attr,
490                                const char *buf,
491                                size_t count)
492 {
493         char *endchar;
494         unsigned long long target_bytes;
495
496         if (!capable(CAP_SYS_ADMIN))
497                 return -EPERM;
498
499         target_bytes = memparse(buf, &endchar);
500
501         balloon_set_new_target(target_bytes >> PAGE_SHIFT);
502
503         return count;
504 }
505
506 static SYSDEV_ATTR(target_kb, S_IRUGO | S_IWUSR,
507                    show_target_kb, store_target_kb);
508
509 static struct sysdev_attribute *balloon_attrs[] = {
510         &attr_target_kb,
511 };
512
513 static struct attribute *balloon_info_attrs[] = {
514         &attr_current_kb.attr,
515         &attr_low_kb.attr,
516         &attr_high_kb.attr,
517         &attr_hard_limit_kb.attr,
518         &attr_driver_kb.attr,
519         NULL
520 };
521
522 static struct attribute_group balloon_info_group = {
523         .name = "info",
524         .attrs = balloon_info_attrs,
525 };
526
527 static struct sysdev_class balloon_sysdev_class = {
528         .name = BALLOON_CLASS_NAME,
529 };
530
531 static int register_balloon(struct sys_device *sysdev)
532 {
533         int i, error;
534
535         error = sysdev_class_register(&balloon_sysdev_class);
536         if (error)
537                 return error;
538
539         sysdev->id = 0;
540         sysdev->cls = &balloon_sysdev_class;
541
542         error = sysdev_register(sysdev);
543         if (error) {
544                 sysdev_class_unregister(&balloon_sysdev_class);
545                 return error;
546         }
547
548         for (i = 0; i < ARRAY_SIZE(balloon_attrs); i++) {
549                 error = sysdev_create_file(sysdev, balloon_attrs[i]);
550                 if (error)
551                         goto fail;
552         }
553
554         error = sysfs_create_group(&sysdev->kobj, &balloon_info_group);
555         if (error)
556                 goto fail;
557
558         return 0;
559
560  fail:
561         while (--i >= 0)
562                 sysdev_remove_file(sysdev, balloon_attrs[i]);
563         sysdev_unregister(sysdev);
564         sysdev_class_unregister(&balloon_sysdev_class);
565         return error;
566 }
567
568 MODULE_LICENSE("GPL");