Blame view
mm/sparse.c
26.4 KB
b24413180
|
1 |
// SPDX-License-Identifier: GPL-2.0 |
d41dee369
|
2 3 4 |
/* * sparse memory mappings. */ |
d41dee369
|
5 |
#include <linux/mm.h> |
5a0e3ad6a
|
6 |
#include <linux/slab.h> |
d41dee369
|
7 |
#include <linux/mmzone.h> |
97ad1087e
|
8 |
#include <linux/memblock.h> |
3b32123d7
|
9 |
#include <linux/compiler.h> |
0b0acbec1
|
10 |
#include <linux/highmem.h> |
b95f1b31b
|
11 |
#include <linux/export.h> |
28ae55c98
|
12 |
#include <linux/spinlock.h> |
0b0acbec1
|
13 |
#include <linux/vmalloc.h> |
9f82883c6
|
14 15 |
#include <linux/swap.h> #include <linux/swapops.h> |
3b32123d7
|
16 |
|
0c0a4a517
|
17 |
#include "internal.h" |
d41dee369
|
18 19 20 21 22 23 24 |
#include <asm/dma.h> /* * Permanent SPARSEMEM data: * * 1) mem_section - memory sections, mem_map's for valid memory */ |
3e347261a
|
25 |
#ifdef CONFIG_SPARSEMEM_EXTREME |
83e3c4872
|
26 |
struct mem_section **mem_section; |
3e347261a
|
27 28 |
#else struct mem_section mem_section[NR_SECTION_ROOTS][SECTIONS_PER_ROOT] |
22fc6eccb
|
29 |
____cacheline_internodealigned_in_smp; |
3e347261a
|
30 31 |
#endif EXPORT_SYMBOL(mem_section); |
89689ae7f
|
32 33 34 35 36 37 38 39 40 41 42 |
#ifdef NODE_NOT_IN_PAGE_FLAGS /* * If we did not store the node number in the page then we have to * do a lookup in the section_to_node_table in order to find which * node the page belongs to. */ #if MAX_NUMNODES <= 256 static u8 section_to_node_table[NR_MEM_SECTIONS] __cacheline_aligned; #else static u16 section_to_node_table[NR_MEM_SECTIONS] __cacheline_aligned; #endif |
33dd4e0ec
|
43 |
int page_to_nid(const struct page *page) |
89689ae7f
|
44 45 46 47 |
{ return section_to_node_table[page_to_section(page)]; } EXPORT_SYMBOL(page_to_nid); |
85770ffe4
|
48 49 50 51 52 53 54 55 56 |
static void set_section_nid(unsigned long section_nr, int nid) { section_to_node_table[section_nr] = nid; } #else /* !NODE_NOT_IN_PAGE_FLAGS */ static inline void set_section_nid(unsigned long section_nr, int nid) { } |
89689ae7f
|
57 |
#endif |
3e347261a
|
58 |
#ifdef CONFIG_SPARSEMEM_EXTREME |
bd721ea73
|
59 |
static noinline struct mem_section __ref *sparse_index_alloc(int nid) |
28ae55c98
|
60 61 62 63 |
{ struct mem_section *section = NULL; unsigned long array_size = SECTIONS_PER_ROOT * sizeof(struct mem_section); |
8a7f97b90
|
64 |
if (slab_is_available()) { |
b95046b04
|
65 |
section = kzalloc_node(array_size, GFP_KERNEL, nid); |
8a7f97b90
|
66 |
} else { |
7e1c4e279
|
67 68 |
section = memblock_alloc_node(array_size, SMP_CACHE_BYTES, nid); |
8a7f97b90
|
69 70 71 72 73 |
if (!section) panic("%s: Failed to allocate %lu bytes nid=%d ", __func__, array_size, nid); } |
28ae55c98
|
74 75 |
return section; |
3e347261a
|
76 |
} |
802f192e4
|
77 |
|
a3142c8e1
|
78 |
static int __meminit sparse_index_init(unsigned long section_nr, int nid) |
802f192e4
|
79 |
{ |
28ae55c98
|
80 81 |
unsigned long root = SECTION_NR_TO_ROOT(section_nr); struct mem_section *section; |
802f192e4
|
82 |
|
ba72b4c8c
|
83 84 85 86 87 88 89 |
/* * An existing section is possible in the sub-section hotplug * case. First hot-add instantiates, follow-on hot-add reuses * the existing section. * * The mem_hotplug_lock resolves the apparent race below. */ |
802f192e4
|
90 |
if (mem_section[root]) |
ba72b4c8c
|
91 |
return 0; |
3e347261a
|
92 |
|
28ae55c98
|
93 |
section = sparse_index_alloc(nid); |
af0cd5a7c
|
94 95 |
if (!section) return -ENOMEM; |
28ae55c98
|
96 97 |
mem_section[root] = section; |
c1c951833
|
98 |
|
9d1936cf8
|
99 |
return 0; |
28ae55c98
|
100 101 102 103 104 |
} #else /* !SPARSEMEM_EXTREME */ static inline int sparse_index_init(unsigned long section_nr, int nid) { return 0; |
802f192e4
|
105 |
} |
28ae55c98
|
106 |
#endif |
91fd8b95d
|
107 |
#ifdef CONFIG_SPARSEMEM_EXTREME |
2491f0a2c
|
108 |
unsigned long __section_nr(struct mem_section *ms) |
4ca644d97
|
109 110 |
{ unsigned long root_nr; |
83e3c4872
|
111 |
struct mem_section *root = NULL; |
4ca644d97
|
112 |
|
12783b002
|
113 114 |
for (root_nr = 0; root_nr < NR_SECTION_ROOTS; root_nr++) { root = __nr_to_section(root_nr * SECTIONS_PER_ROOT); |
4ca644d97
|
115 116 117 118 119 120 |
if (!root) continue; if ((ms >= root) && (ms < (root + SECTIONS_PER_ROOT))) break; } |
83e3c4872
|
121 |
VM_BUG_ON(!root); |
db36a4611
|
122 |
|
4ca644d97
|
123 124 |
return (root_nr * SECTIONS_PER_ROOT) + (ms - root); } |
91fd8b95d
|
125 |
#else |
2491f0a2c
|
126 |
unsigned long __section_nr(struct mem_section *ms) |
91fd8b95d
|
127 |
{ |
2491f0a2c
|
128 |
return (unsigned long)(ms - mem_section[0]); |
91fd8b95d
|
129 130 |
} #endif |
4ca644d97
|
131 |
|
30c253e6d
|
132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 |
/* * During early boot, before section_mem_map is used for an actual * mem_map, we use section_mem_map to store the section's NUMA * node. This keeps us from having to use another data structure. The * node information is cleared just before we store the real mem_map. */ static inline unsigned long sparse_encode_early_nid(int nid) { return (nid << SECTION_NID_SHIFT); } static inline int sparse_early_nid(struct mem_section *section) { return (section->section_mem_map >> SECTION_NID_SHIFT); } |
2dbb51c49
|
147 148 149 |
/* Validate the physical addressing limitations of the model */ void __meminit mminit_validate_memmodel_limits(unsigned long *start_pfn, unsigned long *end_pfn) |
d41dee369
|
150 |
{ |
2dbb51c49
|
151 |
unsigned long max_sparsemem_pfn = 1UL << (MAX_PHYSMEM_BITS-PAGE_SHIFT); |
d41dee369
|
152 |
|
bead9a3ab
|
153 154 155 156 |
/* * Sanity checks - do not allow an architecture to pass * in larger pfns than the maximum scope of sparsemem: */ |
2dbb51c49
|
157 158 159 160 161 162 163 164 |
if (*start_pfn > max_sparsemem_pfn) { mminit_dprintk(MMINIT_WARNING, "pfnvalidation", "Start of range %lu -> %lu exceeds SPARSEMEM max %lu ", *start_pfn, *end_pfn, max_sparsemem_pfn); WARN_ON_ONCE(1); *start_pfn = max_sparsemem_pfn; *end_pfn = max_sparsemem_pfn; |
ef161a986
|
165 |
} else if (*end_pfn > max_sparsemem_pfn) { |
2dbb51c49
|
166 167 168 169 170 171 172 173 |
mminit_dprintk(MMINIT_WARNING, "pfnvalidation", "End of range %lu -> %lu exceeds SPARSEMEM max %lu ", *start_pfn, *end_pfn, max_sparsemem_pfn); WARN_ON_ONCE(1); *end_pfn = max_sparsemem_pfn; } } |
c4e1be9ec
|
174 175 176 177 178 179 180 181 182 |
/* * There are a number of times that we loop over NR_MEM_SECTIONS, * looking for section_present() on each. But, when we have very * large physical address spaces, NR_MEM_SECTIONS can also be * very large which makes the loops quite long. * * Keeping track of this gives us an easy way to break out of * those loops early. */ |
2491f0a2c
|
183 |
unsigned long __highest_present_section_nr; |
c4e1be9ec
|
184 185 |
static void section_mark_present(struct mem_section *ms) { |
2491f0a2c
|
186 |
unsigned long section_nr = __section_nr(ms); |
c4e1be9ec
|
187 188 189 190 191 192 |
if (section_nr > __highest_present_section_nr) __highest_present_section_nr = section_nr; ms->section_mem_map |= SECTION_MARKED_PRESENT; } |
c4e1be9ec
|
193 194 |
#define for_each_present_section_nr(start, section_nr) \ for (section_nr = next_present_section_nr(start-1); \ |
d778015ac
|
195 |
((section_nr != -1) && \ |
c4e1be9ec
|
196 197 |
(section_nr <= __highest_present_section_nr)); \ section_nr = next_present_section_nr(section_nr)) |
85c77f791
|
198 199 200 201 |
static inline unsigned long first_present_section_nr(void) { return next_present_section_nr(-1); } |
0a9f9f623
|
202 |
#ifdef CONFIG_SPARSEMEM_VMEMMAP |
758b8db4a
|
203 |
static void subsection_mask_set(unsigned long *map, unsigned long pfn, |
f46edbd1b
|
204 205 206 207 208 209 210 211 212 213 214 |
unsigned long nr_pages) { int idx = subsection_map_index(pfn); int end = subsection_map_index(pfn + nr_pages - 1); bitmap_set(map, idx, end - idx + 1); } void __init subsection_map_init(unsigned long pfn, unsigned long nr_pages) { int end_sec = pfn_to_section_nr(pfn + nr_pages - 1); |
9a8450304
|
215 |
unsigned long nr, start_sec = pfn_to_section_nr(pfn); |
f46edbd1b
|
216 217 218 |
if (!nr_pages) return; |
9a8450304
|
219 |
for (nr = start_sec; nr <= end_sec; nr++) { |
f46edbd1b
|
220 221 222 223 224 |
struct mem_section *ms; unsigned long pfns; pfns = min(nr_pages, PAGES_PER_SECTION - (pfn & ~PAGE_SECTION_MASK)); |
9a8450304
|
225 |
ms = __nr_to_section(nr); |
f46edbd1b
|
226 |
subsection_mask_set(ms->usage->subsection_map, pfn, pfns); |
9a8450304
|
227 228 |
pr_debug("%s: sec: %lu pfns: %lu set(%d, %d) ", __func__, nr, |
f46edbd1b
|
229 230 231 232 233 234 235 |
pfns, subsection_map_index(pfn), subsection_map_index(pfn + pfns - 1)); pfn += pfns; nr_pages -= pfns; } } |
0a9f9f623
|
236 237 238 239 240 |
#else void __init subsection_map_init(unsigned long pfn, unsigned long nr_pages) { } #endif |
f46edbd1b
|
241 |
|
2dbb51c49
|
242 |
/* Record a memory area against a node. */ |
c89ab04fe
|
243 |
static void __init memory_present(int nid, unsigned long start, unsigned long end) |
2dbb51c49
|
244 245 |
{ unsigned long pfn; |
bead9a3ab
|
246 |
|
629a359bd
|
247 248 249 |
#ifdef CONFIG_SPARSEMEM_EXTREME if (unlikely(!mem_section)) { unsigned long size, align; |
d09cfbbfa
|
250 |
size = sizeof(struct mem_section*) * NR_SECTION_ROOTS; |
629a359bd
|
251 |
align = 1 << (INTERNODE_CACHE_SHIFT); |
eb31d559f
|
252 |
mem_section = memblock_alloc(size, align); |
8a7f97b90
|
253 254 255 256 |
if (!mem_section) panic("%s: Failed to allocate %lu bytes align=0x%lx ", __func__, size, align); |
629a359bd
|
257 258 |
} #endif |
d41dee369
|
259 |
start &= PAGE_SECTION_MASK; |
2dbb51c49
|
260 |
mminit_validate_memmodel_limits(&start, &end); |
d41dee369
|
261 262 |
for (pfn = start; pfn < end; pfn += PAGES_PER_SECTION) { unsigned long section = pfn_to_section_nr(pfn); |
802f192e4
|
263 264 265 |
struct mem_section *ms; sparse_index_init(section, nid); |
85770ffe4
|
266 |
set_section_nid(section, nid); |
802f192e4
|
267 268 |
ms = __nr_to_section(section); |
c4e1be9ec
|
269 |
if (!ms->section_mem_map) { |
2d070eab2
|
270 271 |
ms->section_mem_map = sparse_encode_early_nid(nid) | SECTION_IS_ONLINE; |
c4e1be9ec
|
272 273 |
section_mark_present(ms); } |
d41dee369
|
274 275 276 277 |
} } /* |
c89ab04fe
|
278 279 280 |
* Mark all memblocks as present using memory_present(). * This is a convenience function that is useful to mark all of the systems * memory as present during initialization. |
9def36e0f
|
281 |
*/ |
c89ab04fe
|
282 |
static void __init memblocks_present(void) |
9def36e0f
|
283 |
{ |
c9118e6c3
|
284 285 |
unsigned long start, end; int i, nid; |
9def36e0f
|
286 |
|
c9118e6c3
|
287 288 |
for_each_mem_pfn_range(i, MAX_NUMNODES, &start, &end, &nid) memory_present(nid, start, end); |
9def36e0f
|
289 290 291 |
} /* |
29751f699
|
292 293 294 295 296 297 |
* Subtle, we encode the real pfn into the mem_map such that * the identity pfn - section_mem_map will return the actual * physical page frame number. */ static unsigned long sparse_encode_mem_map(struct page *mem_map, unsigned long pnum) { |
def9b71ee
|
298 299 300 301 302 |
unsigned long coded_mem_map = (unsigned long)(mem_map - (section_nr_to_pfn(pnum))); BUILD_BUG_ON(SECTION_MAP_LAST_BIT > (1UL<<PFN_SECTION_SHIFT)); BUG_ON(coded_mem_map & ~SECTION_MAP_MASK); return coded_mem_map; |
29751f699
|
303 |
} |
3a0aaefe4
|
304 |
#ifdef CONFIG_MEMORY_HOTPLUG |
29751f699
|
305 |
/* |
ea01ea937
|
306 |
* Decode mem_map from the coded memmap |
29751f699
|
307 |
*/ |
29751f699
|
308 309 |
struct page *sparse_decode_mem_map(unsigned long coded_mem_map, unsigned long pnum) { |
ea01ea937
|
310 311 |
/* mask off the extra low bits of information */ coded_mem_map &= SECTION_MAP_MASK; |
29751f699
|
312 313 |
return ((struct page *)coded_mem_map) + section_nr_to_pfn(pnum); } |
3a0aaefe4
|
314 |
#endif /* CONFIG_MEMORY_HOTPLUG */ |
29751f699
|
315 |
|
4e40987f1
|
316 |
static void __meminit sparse_init_one_section(struct mem_section *ms, |
5c0e30664
|
317 |
unsigned long pnum, struct page *mem_map, |
326e1b8f8
|
318 |
struct mem_section_usage *usage, unsigned long flags) |
29751f699
|
319 |
{ |
30c253e6d
|
320 |
ms->section_mem_map &= ~SECTION_MAP_MASK; |
326e1b8f8
|
321 322 |
ms->section_mem_map |= sparse_encode_mem_map(mem_map, pnum) | SECTION_HAS_MEM_MAP | flags; |
f1eca35a0
|
323 |
ms->usage = usage; |
29751f699
|
324 |
} |
f1eca35a0
|
325 |
static unsigned long usemap_size(void) |
5c0e30664
|
326 |
{ |
60a7a88db
|
327 |
return BITS_TO_LONGS(SECTION_BLOCKFLAGS_BITS) * sizeof(unsigned long); |
5c0e30664
|
328 |
} |
f1eca35a0
|
329 |
size_t mem_section_usage_size(void) |
5c0e30664
|
330 |
{ |
f1eca35a0
|
331 |
return sizeof(struct mem_section_usage) + usemap_size(); |
5c0e30664
|
332 |
} |
5c0e30664
|
333 |
|
48c906823
|
334 |
#ifdef CONFIG_MEMORY_HOTREMOVE |
f1eca35a0
|
335 |
static struct mem_section_usage * __init |
a4322e1ba
|
336 |
sparse_early_usemaps_alloc_pgdat_section(struct pglist_data *pgdat, |
238305bb4
|
337 |
unsigned long size) |
48c906823
|
338 |
{ |
f1eca35a0
|
339 |
struct mem_section_usage *usage; |
99ab7b194
|
340 |
unsigned long goal, limit; |
99ab7b194
|
341 |
int nid; |
48c906823
|
342 343 344 |
/* * A page may contain usemaps for other sections preventing the * page being freed and making a section unremovable while |
c800bcd5f
|
345 |
* other sections referencing the usemap remain active. Similarly, |
48c906823
|
346 347 348 349 350 351 |
* a pgdat can prevent a section being removed. If section A * contains a pgdat and section B contains the usemap, both * sections become inter-dependent. This allocates usemaps * from the same section as the pgdat where possible to avoid * this problem. */ |
07b4e2bc9
|
352 |
goal = __pa(pgdat) & (PAGE_SECTION_MASK << PAGE_SHIFT); |
99ab7b194
|
353 354 355 |
limit = goal + (1UL << PA_SECTION_SHIFT); nid = early_pfn_to_nid(goal >> PAGE_SHIFT); again: |
f1eca35a0
|
356 357 |
usage = memblock_alloc_try_nid(size, SMP_CACHE_BYTES, goal, limit, nid); if (!usage && limit) { |
99ab7b194
|
358 359 360 |
limit = 0; goto again; } |
f1eca35a0
|
361 |
return usage; |
48c906823
|
362 |
} |
f1eca35a0
|
363 364 |
static void __init check_usemap_section_nr(int nid, struct mem_section_usage *usage) |
48c906823
|
365 366 |
{ unsigned long usemap_snr, pgdat_snr; |
83e3c4872
|
367 368 |
static unsigned long old_usemap_snr; static unsigned long old_pgdat_snr; |
48c906823
|
369 370 |
struct pglist_data *pgdat = NODE_DATA(nid); int usemap_nid; |
83e3c4872
|
371 372 373 374 375 |
/* First call */ if (!old_usemap_snr) { old_usemap_snr = NR_MEM_SECTIONS; old_pgdat_snr = NR_MEM_SECTIONS; } |
f1eca35a0
|
376 |
usemap_snr = pfn_to_section_nr(__pa(usage) >> PAGE_SHIFT); |
48c906823
|
377 378 379 380 381 382 383 384 385 386 387 388 389 |
pgdat_snr = pfn_to_section_nr(__pa(pgdat) >> PAGE_SHIFT); if (usemap_snr == pgdat_snr) return; if (old_usemap_snr == usemap_snr && old_pgdat_snr == pgdat_snr) /* skip redundant message */ return; old_usemap_snr = usemap_snr; old_pgdat_snr = pgdat_snr; usemap_nid = sparse_early_nid(__nr_to_section(usemap_snr)); if (usemap_nid != nid) { |
1170532bb
|
390 391 392 |
pr_info("node %d must be removed before remove section %ld ", nid, usemap_snr); |
48c906823
|
393 394 395 396 397 398 399 400 |
return; } /* * There is a circular dependency. * Some platforms allow un-removable section because they will just * gather other removable sections for dynamic partitioning. * Just notify un-removable section's number here. */ |
1170532bb
|
401 402 403 |
pr_info("Section %ld and %ld (node %d) have a circular dependency on usemap and pgdat allocations ", usemap_snr, pgdat_snr, nid); |
48c906823
|
404 405 |
} #else |
f1eca35a0
|
406 |
static struct mem_section_usage * __init |
a4322e1ba
|
407 |
sparse_early_usemaps_alloc_pgdat_section(struct pglist_data *pgdat, |
238305bb4
|
408 |
unsigned long size) |
48c906823
|
409 |
{ |
26fb3dae0
|
410 |
return memblock_alloc_node(size, SMP_CACHE_BYTES, pgdat->node_id); |
48c906823
|
411 |
} |
f1eca35a0
|
412 413 |
static void __init check_usemap_section_nr(int nid, struct mem_section_usage *usage) |
48c906823
|
414 415 416 |
{ } #endif /* CONFIG_MEMORY_HOTREMOVE */ |
35fd1eb1e
|
417 |
#ifdef CONFIG_SPARSEMEM_VMEMMAP |
afda57bc1
|
418 |
static unsigned long __init section_map_size(void) |
35fd1eb1e
|
419 420 421 422 423 |
{ return ALIGN(sizeof(struct page) * PAGES_PER_SECTION, PMD_SIZE); } #else |
afda57bc1
|
424 |
static unsigned long __init section_map_size(void) |
e131c06b1
|
425 426 427 |
{ return PAGE_ALIGN(sizeof(struct page) * PAGES_PER_SECTION); } |
e9c0a3f05
|
428 429 |
struct page __init *__populate_section_memmap(unsigned long pfn, unsigned long nr_pages, int nid, struct vmem_altmap *altmap) |
29751f699
|
430 |
{ |
e131c06b1
|
431 432 |
unsigned long size = section_map_size(); struct page *map = sparse_buffer_alloc(size); |
8a7f97b90
|
433 |
phys_addr_t addr = __pa(MAX_DMA_ADDRESS); |
e131c06b1
|
434 435 436 |
if (map) return map; |
29751f699
|
437 |
|
09dbcf422
|
438 |
map = memblock_alloc_try_nid_raw(size, size, addr, |
97ad1087e
|
439 |
MEMBLOCK_ALLOC_ACCESSIBLE, nid); |
8a7f97b90
|
440 441 442 443 |
if (!map) panic("%s: Failed to allocate %lu bytes align=0x%lx nid=%d from=%pa ", __func__, size, PAGE_SIZE, nid, &addr); |
8f6aac419
|
444 445 446 |
return map; } #endif /* !CONFIG_SPARSEMEM_VMEMMAP */ |
35fd1eb1e
|
447 448 |
static void *sparsemap_buf __meminitdata; static void *sparsemap_buf_end __meminitdata; |
ae8318940
|
449 450 451 452 453 |
static inline void __meminit sparse_buffer_free(unsigned long size) { WARN_ON(!sparsemap_buf || size == 0); memblock_free_early(__pa(sparsemap_buf), size); } |
afda57bc1
|
454 |
static void __init sparse_buffer_init(unsigned long size, int nid) |
35fd1eb1e
|
455 |
{ |
8a7f97b90
|
456 |
phys_addr_t addr = __pa(MAX_DMA_ADDRESS); |
35fd1eb1e
|
457 |
WARN_ON(sparsemap_buf); /* forgot to call sparse_buffer_fini()? */ |
09dbcf422
|
458 459 460 461 462 |
/* * Pre-allocated buffer is mainly used by __populate_section_memmap * and we want it to be properly aligned to the section size - this is * especially the case for VMEMMAP which maps memmap to PMDs */ |
0ac398b17
|
463 |
sparsemap_buf = memblock_alloc_exact_nid_raw(size, section_map_size(), |
09dbcf422
|
464 |
addr, MEMBLOCK_ALLOC_ACCESSIBLE, nid); |
35fd1eb1e
|
465 466 |
sparsemap_buf_end = sparsemap_buf + size; } |
afda57bc1
|
467 |
static void __init sparse_buffer_fini(void) |
35fd1eb1e
|
468 469 470 471 |
{ unsigned long size = sparsemap_buf_end - sparsemap_buf; if (sparsemap_buf && size > 0) |
ae8318940
|
472 |
sparse_buffer_free(size); |
35fd1eb1e
|
473 474 475 476 477 478 479 480 |
sparsemap_buf = NULL; } void * __meminit sparse_buffer_alloc(unsigned long size) { void *ptr = NULL; if (sparsemap_buf) { |
db57e98d8
|
481 |
ptr = (void *) roundup((unsigned long)sparsemap_buf, size); |
35fd1eb1e
|
482 483 |
if (ptr + size > sparsemap_buf_end) ptr = NULL; |
ae8318940
|
484 485 486 487 |
else { /* Free redundant aligned space */ if ((unsigned long)(ptr - sparsemap_buf) > 0) sparse_buffer_free((unsigned long)(ptr - sparsemap_buf)); |
35fd1eb1e
|
488 |
sparsemap_buf = ptr + size; |
ae8318940
|
489 |
} |
35fd1eb1e
|
490 491 492 |
} return ptr; } |
3b32123d7
|
493 |
void __weak __meminit vmemmap_populate_print_last(void) |
c2b91e2ee
|
494 495 |
{ } |
a4322e1ba
|
496 |
|
85c77f791
|
497 498 499 500 501 502 503 504 |
/* * Initialize sparse on a specific node. The node spans [pnum_begin, pnum_end) * And number of present sections in this node is map_count. */ static void __init sparse_init_nid(int nid, unsigned long pnum_begin, unsigned long pnum_end, unsigned long map_count) { |
f1eca35a0
|
505 506 |
struct mem_section_usage *usage; unsigned long pnum; |
85c77f791
|
507 |
struct page *map; |
f1eca35a0
|
508 509 510 |
usage = sparse_early_usemaps_alloc_pgdat_section(NODE_DATA(nid), mem_section_usage_size() * map_count); if (!usage) { |
85c77f791
|
511 512 513 514 515 |
pr_err("%s: node[%d] usemap allocation failed", __func__, nid); goto failed; } sparse_buffer_init(map_count * section_map_size(), nid); for_each_present_section_nr(pnum_begin, pnum) { |
e9c0a3f05
|
516 |
unsigned long pfn = section_nr_to_pfn(pnum); |
85c77f791
|
517 518 |
if (pnum >= pnum_end) break; |
e9c0a3f05
|
519 520 |
map = __populate_section_memmap(pfn, PAGES_PER_SECTION, nid, NULL); |
85c77f791
|
521 522 523 524 525 526 |
if (!map) { pr_err("%s: node[%d] memory map backing failed. Some memory will not be available.", __func__, nid); pnum_begin = pnum; goto failed; } |
f1eca35a0
|
527 |
check_usemap_section_nr(nid, usage); |
326e1b8f8
|
528 529 |
sparse_init_one_section(__nr_to_section(pnum), pnum, map, usage, SECTION_IS_EARLY); |
f1eca35a0
|
530 |
usage = (void *) usage + mem_section_usage_size(); |
85c77f791
|
531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 |
} sparse_buffer_fini(); return; failed: /* We failed to allocate, mark all the following pnums as not present */ for_each_present_section_nr(pnum_begin, pnum) { struct mem_section *ms; if (pnum >= pnum_end) break; ms = __nr_to_section(pnum); ms->section_mem_map = 0; } } /* * Allocate the accumulated non-linear sections, allocate a mem_map * for each and record the physical to section mapping. */ |
2a3cb8bae
|
550 |
void __init sparse_init(void) |
85c77f791
|
551 |
{ |
c89ab04fe
|
552 553 554 555 556 557 558 |
unsigned long pnum_end, pnum_begin, map_count = 1; int nid_begin; memblocks_present(); pnum_begin = first_present_section_nr(); nid_begin = sparse_early_nid(__nr_to_section(pnum_begin)); |
85c77f791
|
559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 |
/* Setup pageblock_order for HUGETLB_PAGE_SIZE_VARIABLE */ set_pageblock_order(); for_each_present_section_nr(pnum_begin + 1, pnum_end) { int nid = sparse_early_nid(__nr_to_section(pnum_end)); if (nid == nid_begin) { map_count++; continue; } /* Init node with sections in range [pnum_begin, pnum_end) */ sparse_init_nid(nid_begin, pnum_begin, pnum_end, map_count); nid_begin = nid; pnum_begin = pnum_end; map_count = 1; } /* cover the last node */ sparse_init_nid(nid_begin, pnum_begin, pnum_end, map_count); vmemmap_populate_print_last(); } |
193faea92
|
580 |
#ifdef CONFIG_MEMORY_HOTPLUG |
2d070eab2
|
581 582 583 584 585 586 587 |
/* Mark all memory sections within the pfn range as online */ void online_mem_sections(unsigned long start_pfn, unsigned long end_pfn) { unsigned long pfn; for (pfn = start_pfn; pfn < end_pfn; pfn += PAGES_PER_SECTION) { |
b4ccec41a
|
588 |
unsigned long section_nr = pfn_to_section_nr(pfn); |
2d070eab2
|
589 590 591 592 593 594 595 596 597 598 599 600 |
struct mem_section *ms; /* onlining code should never touch invalid ranges */ if (WARN_ON(!valid_section_nr(section_nr))) continue; ms = __nr_to_section(section_nr); ms->section_mem_map |= SECTION_IS_ONLINE; } } #ifdef CONFIG_MEMORY_HOTREMOVE |
9b7ea46a8
|
601 |
/* Mark all memory sections within the pfn range as offline */ |
2d070eab2
|
602 603 604 605 606 |
void offline_mem_sections(unsigned long start_pfn, unsigned long end_pfn) { unsigned long pfn; for (pfn = start_pfn; pfn < end_pfn; pfn += PAGES_PER_SECTION) { |
27227c733
|
607 |
unsigned long section_nr = pfn_to_section_nr(pfn); |
2d070eab2
|
608 609 610 611 612 613 614 615 616 617 618 619 620 621 |
struct mem_section *ms; /* * TODO this needs some double checking. Offlining code makes * sure to check pfn_valid but those checks might be just bogus */ if (WARN_ON(!valid_section_nr(section_nr))) continue; ms = __nr_to_section(section_nr); ms->section_mem_map &= ~SECTION_IS_ONLINE; } } #endif |
98f3cfc1d
|
622 |
#ifdef CONFIG_SPARSEMEM_VMEMMAP |
030eab4f9
|
623 |
static struct page * __meminit populate_section_memmap(unsigned long pfn, |
e9c0a3f05
|
624 |
unsigned long nr_pages, int nid, struct vmem_altmap *altmap) |
98f3cfc1d
|
625 |
{ |
e9c0a3f05
|
626 |
return __populate_section_memmap(pfn, nr_pages, nid, altmap); |
98f3cfc1d
|
627 |
} |
e9c0a3f05
|
628 629 |
static void depopulate_section_memmap(unsigned long pfn, unsigned long nr_pages, |
24b6d4164
|
630 |
struct vmem_altmap *altmap) |
98f3cfc1d
|
631 |
{ |
e9c0a3f05
|
632 633 |
unsigned long start = (unsigned long) pfn_to_page(pfn); unsigned long end = start + nr_pages * sizeof(struct page); |
0aad818b2
|
634 |
|
24b6d4164
|
635 |
vmemmap_free(start, end, altmap); |
98f3cfc1d
|
636 |
} |
81556b025
|
637 |
static void free_map_bootmem(struct page *memmap) |
0c0a4a517
|
638 |
{ |
0aad818b2
|
639 |
unsigned long start = (unsigned long)memmap; |
81556b025
|
640 |
unsigned long end = (unsigned long)(memmap + PAGES_PER_SECTION); |
0aad818b2
|
641 |
|
24b6d4164
|
642 |
vmemmap_free(start, end, NULL); |
0c0a4a517
|
643 |
} |
6ecb0fc61
|
644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 |
static int clear_subsection_map(unsigned long pfn, unsigned long nr_pages) { DECLARE_BITMAP(map, SUBSECTIONS_PER_SECTION) = { 0 }; DECLARE_BITMAP(tmp, SUBSECTIONS_PER_SECTION) = { 0 }; struct mem_section *ms = __pfn_to_section(pfn); unsigned long *subsection_map = ms->usage ? &ms->usage->subsection_map[0] : NULL; subsection_mask_set(map, pfn, nr_pages); if (subsection_map) bitmap_and(tmp, map, subsection_map, SUBSECTIONS_PER_SECTION); if (WARN(!subsection_map || !bitmap_equal(tmp, map, SUBSECTIONS_PER_SECTION), "section already deactivated (%#lx + %ld) ", pfn, nr_pages)) return -EINVAL; bitmap_xor(subsection_map, map, subsection_map, SUBSECTIONS_PER_SECTION); return 0; } static bool is_subsection_map_empty(struct mem_section *ms) { return bitmap_empty(&ms->usage->subsection_map[0], SUBSECTIONS_PER_SECTION); } static int fill_subsection_map(unsigned long pfn, unsigned long nr_pages) { struct mem_section *ms = __pfn_to_section(pfn); DECLARE_BITMAP(map, SUBSECTIONS_PER_SECTION) = { 0 }; unsigned long *subsection_map; int rc = 0; subsection_mask_set(map, pfn, nr_pages); subsection_map = &ms->usage->subsection_map[0]; if (bitmap_empty(map, SUBSECTIONS_PER_SECTION)) rc = -EINVAL; else if (bitmap_intersects(map, subsection_map, SUBSECTIONS_PER_SECTION)) rc = -EEXIST; else bitmap_or(subsection_map, map, subsection_map, SUBSECTIONS_PER_SECTION); return rc; } |
98f3cfc1d
|
694 |
#else |
030eab4f9
|
695 |
struct page * __meminit populate_section_memmap(unsigned long pfn, |
e9c0a3f05
|
696 |
unsigned long nr_pages, int nid, struct vmem_altmap *altmap) |
0b0acbec1
|
697 |
{ |
4027149ab
|
698 699 |
return kvmalloc_node(array_size(sizeof(struct page), PAGES_PER_SECTION), GFP_KERNEL, nid); |
0b0acbec1
|
700 |
} |
e9c0a3f05
|
701 |
static void depopulate_section_memmap(unsigned long pfn, unsigned long nr_pages, |
7b73d978a
|
702 |
struct vmem_altmap *altmap) |
98f3cfc1d
|
703 |
{ |
3af776f60
|
704 |
kvfree(pfn_to_page(pfn)); |
0b0acbec1
|
705 |
} |
0c0a4a517
|
706 |
|
81556b025
|
707 |
static void free_map_bootmem(struct page *memmap) |
0c0a4a517
|
708 709 |
{ unsigned long maps_section_nr, removing_section_nr, i; |
81556b025
|
710 |
unsigned long magic, nr_pages; |
ae64ffcac
|
711 |
struct page *page = virt_to_page(memmap); |
0c0a4a517
|
712 |
|
81556b025
|
713 714 |
nr_pages = PAGE_ALIGN(PAGES_PER_SECTION * sizeof(struct page)) >> PAGE_SHIFT; |
0c0a4a517
|
715 |
for (i = 0; i < nr_pages; i++, page++) { |
ddffe98d1
|
716 |
magic = (unsigned long) page->freelist; |
0c0a4a517
|
717 718 719 720 |
BUG_ON(magic == NODE_INFO); maps_section_nr = pfn_to_section_nr(page_to_pfn(page)); |
857e522a0
|
721 |
removing_section_nr = page_private(page); |
0c0a4a517
|
722 723 724 725 726 727 728 729 730 731 732 733 734 |
/* * When this function is called, the removing section is * logical offlined state. This means all pages are isolated * from page allocator. If removing section's memmap is placed * on the same section, it must not be freed. * If it is freed, page allocator may allocate it which will * be removed physically soon. */ if (maps_section_nr != removing_section_nr) put_page_bootmem(page); } } |
0b0acbec1
|
735 |
|
37bc15020
|
736 |
static int clear_subsection_map(unsigned long pfn, unsigned long nr_pages) |
ba72b4c8c
|
737 |
{ |
37bc15020
|
738 739 740 741 742 |
return 0; } static bool is_subsection_map_empty(struct mem_section *ms) { |
6ecb0fc61
|
743 |
return true; |
0a9f9f623
|
744 |
} |
6ecb0fc61
|
745 |
static int fill_subsection_map(unsigned long pfn, unsigned long nr_pages) |
0a9f9f623
|
746 |
{ |
6ecb0fc61
|
747 |
return 0; |
0a9f9f623
|
748 |
} |
6ecb0fc61
|
749 |
#endif /* CONFIG_SPARSEMEM_VMEMMAP */ |
37bc15020
|
750 |
|
95a5a34df
|
751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 |
/* * To deactivate a memory region, there are 3 cases to handle across * two configurations (SPARSEMEM_VMEMMAP={y,n}): * * 1. deactivation of a partial hot-added section (only possible in * the SPARSEMEM_VMEMMAP=y case). * a) section was present at memory init. * b) section was hot-added post memory init. * 2. deactivation of a complete hot-added section. * 3. deactivation of a complete section from memory init. * * For 1, when subsection_map does not empty we will not be freeing the * usage map, but still need to free the vmemmap range. * * For 2 and 3, the SPARSEMEM_VMEMMAP={y,n} cases are unified */ |
37bc15020
|
767 768 769 770 771 772 773 774 775 776 |
static void section_deactivate(unsigned long pfn, unsigned long nr_pages, struct vmem_altmap *altmap) { struct mem_section *ms = __pfn_to_section(pfn); bool section_is_early = early_section(ms); struct page *memmap = NULL; bool empty; if (clear_subsection_map(pfn, nr_pages)) return; |
95a5a34df
|
777 |
|
37bc15020
|
778 |
empty = is_subsection_map_empty(ms); |
d41e2f3bd
|
779 |
if (empty) { |
ba72b4c8c
|
780 |
unsigned long section_nr = pfn_to_section_nr(pfn); |
8068df3b6
|
781 782 783 784 785 786 787 788 |
/* * When removing an early section, the usage map is kept (as the * usage maps of other sections fall into the same page). It * will be re-used when re-adding the section - which is then no * longer an early section. If the usage map is PageReserved, it * was allocated during boot. */ if (!PageReserved(virt_to_page(ms->usage))) { |
ba72b4c8c
|
789 790 791 792 |
kfree(ms->usage); ms->usage = NULL; } memmap = sparse_decode_mem_map(ms->section_mem_map, section_nr); |
b943f045a
|
793 794 795 796 797 798 |
/* * Mark the section invalid so that valid_section() * return false. This prevents code from dereferencing * ms->usage array. */ ms->section_mem_map &= ~SECTION_HAS_MEM_MAP; |
ba72b4c8c
|
799 |
} |
ef69bc9f6
|
800 801 802 803 804 |
/* * The memmap of early sections is always fully populated. See * section_activate() and pfn_valid() . */ if (!section_is_early) |
ba72b4c8c
|
805 |
depopulate_section_memmap(pfn, nr_pages, altmap); |
ef69bc9f6
|
806 807 |
else if (memmap) free_map_bootmem(memmap); |
d41e2f3bd
|
808 809 810 |
if (empty) ms->section_mem_map = (unsigned long)NULL; |
ba72b4c8c
|
811 |
} |
5d87255ca
|
812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 |
static struct page * __meminit section_activate(int nid, unsigned long pfn, unsigned long nr_pages, struct vmem_altmap *altmap) { struct mem_section *ms = __pfn_to_section(pfn); struct mem_section_usage *usage = NULL; struct page *memmap; int rc = 0; if (!ms->usage) { usage = kzalloc(mem_section_usage_size(), GFP_KERNEL); if (!usage) return ERR_PTR(-ENOMEM); ms->usage = usage; } rc = fill_subsection_map(pfn, nr_pages); |
ba72b4c8c
|
828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 |
if (rc) { if (usage) ms->usage = NULL; kfree(usage); return ERR_PTR(rc); } /* * The early init code does not consider partially populated * initial sections, it simply assumes that memory will never be * referenced. If we hot-add memory into such a section then we * do not need to populate the memmap and can simply reuse what * is already there. */ if (nr_pages < PAGES_PER_SECTION && early_section(ms)) return pfn_to_page(pfn); memmap = populate_section_memmap(pfn, nr_pages, nid, altmap); if (!memmap) { section_deactivate(pfn, nr_pages, altmap); return ERR_PTR(-ENOMEM); } return memmap; } |
7567cfc5d
|
853 |
/** |
ba72b4c8c
|
854 |
* sparse_add_section - add a memory section, or populate an existing one |
7567cfc5d
|
855 856 |
* @nid: The node to add section on * @start_pfn: start pfn of the memory range |
ba72b4c8c
|
857 |
* @nr_pages: number of pfns to add in the section |
7567cfc5d
|
858 859 860 861 |
* @altmap: device page map * * This is only intended for hotplug. * |
95a5a34df
|
862 863 864 865 |
* Note that only VMEMMAP supports sub-section aligned hotplug, * the proper alignment and size are gated by check_pfn_span(). * * |
7567cfc5d
|
866 867 868 869 |
* Return: * * 0 - On success. * * -EEXIST - Section has been present. * * -ENOMEM - Out of memory. |
29751f699
|
870 |
*/ |
7ea621604
|
871 872 |
int __meminit sparse_add_section(int nid, unsigned long start_pfn, unsigned long nr_pages, struct vmem_altmap *altmap) |
29751f699
|
873 |
{ |
0b0acbec1
|
874 |
unsigned long section_nr = pfn_to_section_nr(start_pfn); |
0b0acbec1
|
875 876 |
struct mem_section *ms; struct page *memmap; |
0b0acbec1
|
877 |
int ret; |
29751f699
|
878 |
|
4e0d2e7ef
|
879 |
ret = sparse_index_init(section_nr, nid); |
ba72b4c8c
|
880 |
if (ret < 0) |
bbd068259
|
881 |
return ret; |
0b0acbec1
|
882 |
|
ba72b4c8c
|
883 884 885 |
memmap = section_activate(nid, start_pfn, nr_pages, altmap); if (IS_ERR(memmap)) return PTR_ERR(memmap); |
5c0e30664
|
886 |
|
d0dc12e86
|
887 888 889 890 |
/* * Poison uninitialized struct pages in order to catch invalid flags * combinations. */ |
18e19f195
|
891 |
page_init_poison(memmap, sizeof(struct page) * nr_pages); |
3ac19f8ef
|
892 |
|
c1cbc3eeb
|
893 |
ms = __nr_to_section(section_nr); |
26f26beda
|
894 |
set_section_nid(section_nr, nid); |
c4e1be9ec
|
895 |
section_mark_present(ms); |
0b0acbec1
|
896 |
|
ba72b4c8c
|
897 898 |
/* Align memmap to section boundary in the subsection case */ if (section_nr_to_pfn(section_nr) != start_pfn) |
4627d76dc
|
899 |
memmap = pfn_to_page(section_nr_to_pfn(section_nr)); |
ba72b4c8c
|
900 901 902 |
sparse_init_one_section(ms, section_nr, memmap, ms->usage, 0); return 0; |
29751f699
|
903 |
} |
ea01ea937
|
904 |
|
95a4774d0
|
905 906 907 908 |
#ifdef CONFIG_MEMORY_FAILURE static void clear_hwpoisoned_pages(struct page *memmap, int nr_pages) { int i; |
5eb570a8d
|
909 910 911 912 913 914 915 916 |
/* * A further optimization is to have per section refcounted * num_poisoned_pages. But that would need more space per memmap, so * for now just do a quick global check to speed up this routine in the * absence of bad pages. */ if (atomic_long_read(&num_poisoned_pages) == 0) return; |
4b94ffdc4
|
917 |
for (i = 0; i < nr_pages; i++) { |
95a4774d0
|
918 |
if (PageHWPoison(&memmap[i])) { |
9f82883c6
|
919 |
num_poisoned_pages_dec(); |
95a4774d0
|
920 921 922 923 924 925 926 927 928 |
ClearPageHWPoison(&memmap[i]); } } } #else static inline void clear_hwpoisoned_pages(struct page *memmap, int nr_pages) { } #endif |
ba72b4c8c
|
929 |
void sparse_remove_section(struct mem_section *ms, unsigned long pfn, |
7ea621604
|
930 931 |
unsigned long nr_pages, unsigned long map_offset, struct vmem_altmap *altmap) |
ea01ea937
|
932 |
{ |
ba72b4c8c
|
933 934 935 |
clear_hwpoisoned_pages(pfn_to_page(pfn) + map_offset, nr_pages - map_offset); section_deactivate(pfn, nr_pages, altmap); |
ea01ea937
|
936 |
} |
4edd7ceff
|
937 |
#endif /* CONFIG_MEMORY_HOTPLUG */ |