Blame view
mm/page_isolation.c
8.36 KB
a5d76b54a memory unplug: pa... |
1 2 3 |
/* * linux/mm/page_isolation.c */ |
a5d76b54a memory unplug: pa... |
4 5 6 |
#include <linux/mm.h> #include <linux/page-isolation.h> #include <linux/pageblock-flags.h> |
ee6f509c3 mm: factor out me... |
7 |
#include <linux/memory.h> |
c8721bbbd mm: memory-hotplu... |
8 |
#include <linux/hugetlb.h> |
83358ece2 mm/page_owner: in... |
9 |
#include <linux/page_owner.h> |
a5d76b54a memory unplug: pa... |
10 |
#include "internal.h" |
0f0848e51 mm/page_isolation... |
11 12 |
#define CREATE_TRACE_POINTS #include <trace/events/page_isolation.h> |
c5b4e1b02 mm, page_isolatio... |
13 14 |
static int set_migratetype_isolate(struct page *page, bool skip_hwpoisoned_pages) |
ee6f509c3 mm: factor out me... |
15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 |
{ struct zone *zone; unsigned long flags, pfn; struct memory_isolate_notify arg; int notifier_ret; int ret = -EBUSY; zone = page_zone(page); spin_lock_irqsave(&zone->lock, flags); pfn = page_to_pfn(page); arg.start_pfn = pfn; arg.nr_pages = pageblock_nr_pages; arg.pages_found = 0; /* * It may be possible to isolate a pageblock even if the * migratetype is not MIGRATE_MOVABLE. The memory isolation * notifier chain is used by balloon drivers to return the * number of pages in a range that are held by the balloon * driver to shrink memory. If all the pages are accounted for * by balloons, are free, or on the LRU, isolation can continue. * Later, for example, when memory hotplug notifier runs, these * pages reported as "can be isolated" should be isolated(freed) * by the balloon driver through the memory notifier chain. */ notifier_ret = memory_isolate_notify(MEM_ISOLATE_COUNT, &arg); notifier_ret = notifier_to_errno(notifier_ret); if (notifier_ret) goto out; /* * FIXME: Now, memory hotplug doesn't call shrink_slab() by itself. * We just check MOVABLE pages. */ |
b023f4681 memory-hotplug: s... |
50 51 |
if (!has_unmovable_pages(zone, page, arg.pages_found, skip_hwpoisoned_pages)) |
ee6f509c3 mm: factor out me... |
52 53 54 |
ret = 0; /* |
ac34dcd26 mm/page_isolation... |
55 |
* immobile means "not-on-lru" pages. If immobile is larger than |
ee6f509c3 mm: factor out me... |
56 57 58 59 60 |
* removable-by-driver pages reported by notifier, we'll fail. */ out: if (!ret) { |
2139cbe62 cma: fix counting... |
61 |
unsigned long nr_pages; |
d1ce749a0 cma: count free C... |
62 |
int migratetype = get_pageblock_migratetype(page); |
2139cbe62 cma: fix counting... |
63 |
|
a458431e1 mm: fix zone_wate... |
64 |
set_pageblock_migratetype(page, MIGRATE_ISOLATE); |
ad53f92eb mm/page_alloc: fi... |
65 |
zone->nr_isolate_pageblock++; |
2139cbe62 cma: fix counting... |
66 |
nr_pages = move_freepages_block(zone, page, MIGRATE_ISOLATE); |
d1ce749a0 cma: count free C... |
67 |
__mod_zone_freepage_state(zone, -nr_pages, migratetype); |
ee6f509c3 mm: factor out me... |
68 69 70 71 |
} spin_unlock_irqrestore(&zone->lock, flags); if (!ret) |
ec25af84b mm, page_isolatio... |
72 |
drain_all_pages(zone); |
ee6f509c3 mm: factor out me... |
73 74 |
return ret; } |
c5b4e1b02 mm, page_isolatio... |
75 |
static void unset_migratetype_isolate(struct page *page, unsigned migratetype) |
ee6f509c3 mm: factor out me... |
76 77 |
{ struct zone *zone; |
2139cbe62 cma: fix counting... |
78 |
unsigned long flags, nr_pages; |
e3a2713c3 mm/page_isolation... |
79 |
bool isolated_page = false; |
3c605096d mm/page_alloc: re... |
80 81 82 |
unsigned int order; unsigned long page_idx, buddy_idx; struct page *buddy; |
2139cbe62 cma: fix counting... |
83 |
|
ee6f509c3 mm: factor out me... |
84 85 86 87 |
zone = page_zone(page); spin_lock_irqsave(&zone->lock, flags); if (get_pageblock_migratetype(page) != MIGRATE_ISOLATE) goto out; |
3c605096d mm/page_alloc: re... |
88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 |
/* * Because freepage with more than pageblock_order on isolated * pageblock is restricted to merge due to freepage counting problem, * it is possible that there is free buddy page. * move_freepages_block() doesn't care of merge so we need other * approach in order to merge them. Isolation and free will make * these pages to be merged. */ if (PageBuddy(page)) { order = page_order(page); if (order >= pageblock_order) { page_idx = page_to_pfn(page) & ((1 << MAX_ORDER) - 1); buddy_idx = __find_buddy_index(page_idx, order); buddy = page + (buddy_idx - page_idx); |
1ae7013df CMA: page_isolati... |
103 104 |
if (pfn_valid_within(page_to_pfn(buddy)) && !is_migrate_isolate_page(buddy)) { |
3c605096d mm/page_alloc: re... |
105 |
__isolate_free_page(page, order); |
e3a2713c3 mm/page_isolation... |
106 |
isolated_page = true; |
3c605096d mm/page_alloc: re... |
107 108 109 110 111 112 113 114 115 116 117 118 119 |
} } } /* * If we isolate freepage with more than pageblock_order, there * should be no freepage in the range, so we could avoid costly * pageblock scanning for freepage moving. */ if (!isolated_page) { nr_pages = move_freepages_block(zone, page, migratetype); __mod_zone_freepage_state(zone, nr_pages, migratetype); } |
a458431e1 mm: fix zone_wate... |
120 |
set_pageblock_migratetype(page, migratetype); |
ad53f92eb mm/page_alloc: fi... |
121 |
zone->nr_isolate_pageblock--; |
ee6f509c3 mm: factor out me... |
122 123 |
out: spin_unlock_irqrestore(&zone->lock, flags); |
83358ece2 mm/page_owner: in... |
124 |
if (isolated_page) { |
46f24fd85 mm/page_alloc: in... |
125 |
post_alloc_hook(page, order, __GFP_MOVABLE); |
e3a2713c3 mm/page_isolation... |
126 |
__free_pages(page, order); |
83358ece2 mm/page_owner: in... |
127 |
} |
ee6f509c3 mm: factor out me... |
128 |
} |
a5d76b54a memory unplug: pa... |
129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 |
static inline struct page * __first_valid_page(unsigned long pfn, unsigned long nr_pages) { int i; for (i = 0; i < nr_pages; i++) if (pfn_valid_within(pfn + i)) break; if (unlikely(i == nr_pages)) return NULL; return pfn_to_page(pfn + i); } /* * start_isolate_page_range() -- make page-allocation-type of range of pages * to be MIGRATE_ISOLATE. * @start_pfn: The lower PFN of the range to be isolated. * @end_pfn: The upper PFN of the range to be isolated. |
0815f3d81 mm: page_isolatio... |
146 |
* @migratetype: migrate type to set in error recovery. |
a5d76b54a memory unplug: pa... |
147 148 149 150 151 152 153 154 |
* * Making page-allocation-type to be MIGRATE_ISOLATE means free pages in * the range will never be allocated. Any free pages and pages freed in the * future will not be allocated again. * * start_pfn/end_pfn must be aligned to pageblock_order. * Returns 0 on success and -EBUSY if any part of range cannot be isolated. */ |
0815f3d81 mm: page_isolatio... |
155 |
int start_isolate_page_range(unsigned long start_pfn, unsigned long end_pfn, |
b023f4681 memory-hotplug: s... |
156 |
unsigned migratetype, bool skip_hwpoisoned_pages) |
a5d76b54a memory unplug: pa... |
157 158 159 160 |
{ unsigned long pfn; unsigned long undo_pfn; struct page *page; |
fec174d66 mm/page_isolation... |
161 162 |
BUG_ON(!IS_ALIGNED(start_pfn, pageblock_nr_pages)); BUG_ON(!IS_ALIGNED(end_pfn, pageblock_nr_pages)); |
a5d76b54a memory unplug: pa... |
163 164 165 166 167 |
for (pfn = start_pfn; pfn < end_pfn; pfn += pageblock_nr_pages) { page = __first_valid_page(pfn, pageblock_nr_pages); |
b023f4681 memory-hotplug: s... |
168 169 |
if (page && set_migratetype_isolate(page, skip_hwpoisoned_pages)) { |
a5d76b54a memory unplug: pa... |
170 171 172 173 174 175 176 |
undo_pfn = pfn; goto undo; } } return 0; undo: for (pfn = start_pfn; |
dbc0e4cef memory hotremove:... |
177 |
pfn < undo_pfn; |
a5d76b54a memory unplug: pa... |
178 |
pfn += pageblock_nr_pages) |
0815f3d81 mm: page_isolatio... |
179 |
unset_migratetype_isolate(pfn_to_page(pfn), migratetype); |
a5d76b54a memory unplug: pa... |
180 181 182 183 184 185 186 |
return -EBUSY; } /* * Make isolated pages available again. */ |
0815f3d81 mm: page_isolatio... |
187 188 |
int undo_isolate_page_range(unsigned long start_pfn, unsigned long end_pfn, unsigned migratetype) |
a5d76b54a memory unplug: pa... |
189 190 191 |
{ unsigned long pfn; struct page *page; |
6f8d2b8a2 mm/page_isolation... |
192 193 194 |
BUG_ON(!IS_ALIGNED(start_pfn, pageblock_nr_pages)); BUG_ON(!IS_ALIGNED(end_pfn, pageblock_nr_pages)); |
a5d76b54a memory unplug: pa... |
195 196 197 198 |
for (pfn = start_pfn; pfn < end_pfn; pfn += pageblock_nr_pages) { page = __first_valid_page(pfn, pageblock_nr_pages); |
dbc0e4cef memory hotremove:... |
199 |
if (!page || get_pageblock_migratetype(page) != MIGRATE_ISOLATE) |
a5d76b54a memory unplug: pa... |
200 |
continue; |
0815f3d81 mm: page_isolatio... |
201 |
unset_migratetype_isolate(page, migratetype); |
a5d76b54a memory unplug: pa... |
202 203 204 205 206 207 208 209 |
} return 0; } /* * Test all pages in the range is free(means isolated) or not. * all pages in [start_pfn...end_pfn) must be in the same zone. * zone->lock must be held before call this. * |
ec3b68825 mm/page_isolation... |
210 |
* Returns the last tested pfn. |
a5d76b54a memory unplug: pa... |
211 |
*/ |
fea85cff1 mm/page_isolation... |
212 |
static unsigned long |
b023f4681 memory-hotplug: s... |
213 214 |
__test_page_isolated_in_pageblock(unsigned long pfn, unsigned long end_pfn, bool skip_hwpoisoned_pages) |
a5d76b54a memory unplug: pa... |
215 216 217 218 219 220 221 222 223 |
{ struct page *page; while (pfn < end_pfn) { if (!pfn_valid_within(pfn)) { pfn++; continue; } page = pfn_to_page(pfn); |
aa016d145 mm, page_isolatio... |
224 |
if (PageBuddy(page)) |
435b405c0 memory-hotplug: f... |
225 |
/* |
aa016d145 mm, page_isolatio... |
226 227 228 |
* If the page is on a free list, it has to be on * the correct MIGRATE_ISOLATE freelist. There is no * simple way to verify that as VM_BUG_ON(), though. |
435b405c0 memory-hotplug: f... |
229 |
*/ |
a5d76b54a memory unplug: pa... |
230 |
pfn += 1 << page_order(page); |
aa016d145 mm, page_isolatio... |
231 232 |
else if (skip_hwpoisoned_pages && PageHWPoison(page)) /* A HWPoisoned page cannot be also PageBuddy */ |
b023f4681 memory-hotplug: s... |
233 |
pfn++; |
a5d76b54a memory unplug: pa... |
234 235 236 |
else break; } |
fea85cff1 mm/page_isolation... |
237 238 |
return pfn; |
a5d76b54a memory unplug: pa... |
239 |
} |
b9eb63191 mm/memory_hotplug... |
240 |
/* Caller should ensure that requested range is in a single zone */ |
b023f4681 memory-hotplug: s... |
241 242 |
int test_pages_isolated(unsigned long start_pfn, unsigned long end_pfn, bool skip_hwpoisoned_pages) |
a5d76b54a memory unplug: pa... |
243 |
{ |
6c1b7f680 memory hotplug: m... |
244 |
unsigned long pfn, flags; |
a5d76b54a memory unplug: pa... |
245 |
struct page *page; |
6c1b7f680 memory hotplug: m... |
246 |
struct zone *zone; |
a5d76b54a memory unplug: pa... |
247 |
|
a5d76b54a memory unplug: pa... |
248 |
/* |
85dbe7060 page_isolation: F... |
249 250 251 |
* Note: pageblock_nr_pages != MAX_ORDER. Then, chunks of free pages * are not aligned to pageblock_nr_pages. * Then we just check migratetype first. |
a5d76b54a memory unplug: pa... |
252 253 254 |
*/ for (pfn = start_pfn; pfn < end_pfn; pfn += pageblock_nr_pages) { page = __first_valid_page(pfn, pageblock_nr_pages); |
dbc0e4cef memory hotremove:... |
255 |
if (page && get_pageblock_migratetype(page) != MIGRATE_ISOLATE) |
a5d76b54a memory unplug: pa... |
256 257 |
break; } |
a70dcb969 memory hotplug: f... |
258 259 |
page = __first_valid_page(start_pfn, end_pfn - start_pfn); if ((pfn < end_pfn) || !page) |
a5d76b54a memory unplug: pa... |
260 |
return -EBUSY; |
85dbe7060 page_isolation: F... |
261 |
/* Check all pages are free or marked as ISOLATED */ |
a70dcb969 memory hotplug: f... |
262 |
zone = page_zone(page); |
6c1b7f680 memory hotplug: m... |
263 |
spin_lock_irqsave(&zone->lock, flags); |
fea85cff1 mm/page_isolation... |
264 |
pfn = __test_page_isolated_in_pageblock(start_pfn, end_pfn, |
b023f4681 memory-hotplug: s... |
265 |
skip_hwpoisoned_pages); |
6c1b7f680 memory hotplug: m... |
266 |
spin_unlock_irqrestore(&zone->lock, flags); |
fea85cff1 mm/page_isolation... |
267 |
|
0f0848e51 mm/page_isolation... |
268 |
trace_test_pages_isolated(start_pfn, end_pfn, pfn); |
fea85cff1 mm/page_isolation... |
269 |
return pfn < end_pfn ? -EBUSY : 0; |
a5d76b54a memory unplug: pa... |
270 |
} |
723a0644a mm/page_alloc: re... |
271 272 273 274 275 |
struct page *alloc_migrate_target(struct page *page, unsigned long private, int **resultp) { gfp_t gfp_mask = GFP_USER | __GFP_MOVABLE; |
c8721bbbd mm: memory-hotplu... |
276 277 278 279 280 |
/* * TODO: allocate a destination hugepage from a nearest neighbor node, * accordance with memory policy of the user process if possible. For * now as a simple work-around, we use the next node for destination. */ |
0edaf86cf include/linux/nod... |
281 |
if (PageHuge(page)) |
c8721bbbd mm: memory-hotplu... |
282 |
return alloc_huge_page_node(page_hstate(compound_head(page)), |
0edaf86cf include/linux/nod... |
283 284 |
next_node_in(page_to_nid(page), node_online_map)); |
c8721bbbd mm: memory-hotplu... |
285 |
|
723a0644a mm/page_alloc: re... |
286 287 288 289 290 |
if (PageHighMem(page)) gfp_mask |= __GFP_HIGHMEM; return alloc_page(gfp_mask); } |