1 /*- 2 * Copyright (c) 2002-2005, 2009, 2013 Jeffrey Roberson <jeff@FreeBSD.org> 3 * Copyright (c) 2004, 2005 Bosko Milekic <bmilekic@FreeBSD.org> 4 * All rights reserved. 5 * 6 * Redistribution and use in source and binary forms, with or without 7 * modification, are permitted provided that the following conditions 8 * are met: 9 * 1. Redistributions of source code must retain the above copyright 10 * notice unmodified, this list of conditions, and the following 11 * disclaimer. 12 * 2. Redistributions in binary form must reproduce the above copyright 13 * notice, this list of conditions and the following disclaimer in the 14 * documentation and/or other materials provided with the distribution. 15 * 16 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR 17 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES 18 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. 19 * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, 20 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT 21 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, 22 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY 23 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT 24 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF 25 * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 26 * 27 * $FreeBSD$ 28 * 29 */ 30 31 /* 32 * This file includes definitions, structures, prototypes, and inlines that 33 * should not be used outside of the actual implementation of UMA. 34 */ 35 36 /* 37 * Here's a quick description of the relationship between the objects: 38 * 39 * Kegs contain lists of slabs which are stored in either the full bin, empty 40 * bin, or partially allocated bin, to reduce fragmentation. They also contain 41 * the user supplied value for size, which is adjusted for alignment purposes 42 * and rsize is the result of that. The Keg also stores information for 43 * managing a hash of page addresses that maps pages to uma_slab_t structures 44 * for pages that don't have embedded uma_slab_t's. 45 * 46 * The uma_slab_t may be embedded in a UMA_SLAB_SIZE chunk of memory or it may 47 * be allocated off the page from a special slab zone. The free list within a 48 * slab is managed with a bitmask. For item sizes that would yield more than 49 * 10% memory waste we potentially allocate a separate uma_slab_t if this will 50 * improve the number of items per slab that will fit. 51 * 52 * The only really gross cases, with regards to memory waste, are for those 53 * items that are just over half the page size. You can get nearly 50% waste, 54 * so you fall back to the memory footprint of the power of two allocator. I 55 * have looked at memory allocation sizes on many of the machines available to 56 * me, and there does not seem to be an abundance of allocations at this range 57 * so at this time it may not make sense to optimize for it. This can, of 58 * course, be solved with dynamic slab sizes. 59 * 60 * Kegs may serve multiple Zones but by far most of the time they only serve 61 * one. When a Zone is created, a Keg is allocated and setup for it. While 62 * the backing Keg stores slabs, the Zone caches Buckets of items allocated 63 * from the slabs. Each Zone is equipped with an init/fini and ctor/dtor 64 * pair, as well as with its own set of small per-CPU caches, layered above 65 * the Zone's general Bucket cache. 66 * 67 * The PCPU caches are protected by critical sections, and may be accessed 68 * safely only from their associated CPU, while the Zones backed by the same 69 * Keg all share a common Keg lock (to coalesce contention on the backing 70 * slabs). The backing Keg typically only serves one Zone but in the case of 71 * multiple Zones, one of the Zones is considered the Master Zone and all 72 * Zone-related stats from the Keg are done in the Master Zone. For an 73 * example of a Multi-Zone setup, refer to the Mbuf allocation code. 74 */ 75 76 /* 77 * This is the representation for normal (Non OFFPAGE slab) 78 * 79 * i == item 80 * s == slab pointer 81 * 82 * <---------------- Page (UMA_SLAB_SIZE) ------------------> 83 * ___________________________________________________________ 84 * | _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ ___________ | 85 * ||i||i||i||i||i||i||i||i||i||i||i||i||i||i||i| |slab header|| 86 * ||_||_||_||_||_||_||_||_||_||_||_||_||_||_||_| |___________|| 87 * |___________________________________________________________| 88 * 89 * 90 * This is an OFFPAGE slab. These can be larger than UMA_SLAB_SIZE. 91 * 92 * ___________________________________________________________ 93 * | _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ | 94 * ||i||i||i||i||i||i||i||i||i||i||i||i||i||i||i||i||i||i||i| | 95 * ||_||_||_||_||_||_||_||_||_||_||_||_||_||_||_||_||_||_||_| | 96 * |___________________________________________________________| 97 * ___________ ^ 98 * |slab header| | 99 * |___________|---* 100 * 101 */ 102 103 #ifndef VM_UMA_INT_H 104 #define VM_UMA_INT_H 105 106 #define UMA_SLAB_SIZE PAGE_SIZE /* How big are our slabs? */ 107 #define UMA_SLAB_MASK (PAGE_SIZE - 1) /* Mask to get back to the page */ 108 #define UMA_SLAB_SHIFT PAGE_SHIFT /* Number of bits PAGE_MASK */ 109 110 #define UMA_BOOT_PAGES 64 /* Pages allocated for startup */ 111 112 /* Max waste percentage before going to off page slab management */ 113 #define UMA_MAX_WASTE 10 114 115 /* 116 * I doubt there will be many cases where this is exceeded. This is the initial 117 * size of the hash table for uma_slabs that are managed off page. This hash 118 * does expand by powers of two. Currently it doesn't get smaller. 119 */ 120 #define UMA_HASH_SIZE_INIT 32 121 122 /* 123 * I should investigate other hashing algorithms. This should yield a low 124 * number of collisions if the pages are relatively contiguous. 125 */ 126 127 #define UMA_HASH(h, s) ((((uintptr_t)s) >> UMA_SLAB_SHIFT) & (h)->uh_hashmask) 128 129 #define UMA_HASH_INSERT(h, s, mem) \ 130 SLIST_INSERT_HEAD(&(h)->uh_slab_hash[UMA_HASH((h), \ 131 (mem))], (s), us_hlink) 132 #define UMA_HASH_REMOVE(h, s, mem) \ 133 SLIST_REMOVE(&(h)->uh_slab_hash[UMA_HASH((h), \ 134 (mem))], (s), uma_slab, us_hlink) 135 136 /* Hash table for freed address -> slab translation */ 137 138 SLIST_HEAD(slabhead, uma_slab); 139 140 struct uma_hash { 141 struct slabhead *uh_slab_hash; /* Hash table for slabs */ 142 int uh_hashsize; /* Current size of the hash table */ 143 int uh_hashmask; /* Mask used during hashing */ 144 }; 145 146 /* 147 * align field or structure to cache line 148 */ 149 #if defined(__amd64__) 150 #define UMA_ALIGN __aligned(CACHE_LINE_SIZE) 151 #else 152 #define UMA_ALIGN 153 #endif 154 155 /* 156 * Structures for per cpu queues. 157 */ 158 159 struct uma_bucket { 160 LIST_ENTRY(uma_bucket) ub_link; /* Link into the zone */ 161 int16_t ub_cnt; /* Count of free items. */ 162 int16_t ub_entries; /* Max items. */ 163 void *ub_bucket[]; /* actual allocation storage */ 164 }; 165 166 typedef struct uma_bucket * uma_bucket_t; 167 168 struct uma_cache { 169 uma_bucket_t uc_freebucket; /* Bucket we're freeing to */ 170 uma_bucket_t uc_allocbucket; /* Bucket to allocate from */ 171 uint64_t uc_allocs; /* Count of allocations */ 172 uint64_t uc_frees; /* Count of frees */ 173 } UMA_ALIGN; 174 175 typedef struct uma_cache * uma_cache_t; 176 177 /* 178 * Keg management structure 179 * 180 * TODO: Optimize for cache line size 181 * 182 */ 183 struct uma_keg { 184 struct mtx_padalign uk_lock; /* Lock for the keg */ 185 struct uma_hash uk_hash; 186 187 LIST_HEAD(,uma_zone) uk_zones; /* Keg's zones */ 188 LIST_HEAD(,uma_slab) uk_part_slab; /* partially allocated slabs */ 189 LIST_HEAD(,uma_slab) uk_free_slab; /* empty slab list */ 190 LIST_HEAD(,uma_slab) uk_full_slab; /* full slabs */ 191 192 uint32_t uk_align; /* Alignment mask */ 193 uint32_t uk_pages; /* Total page count */ 194 uint32_t uk_free; /* Count of items free in slabs */ 195 uint32_t uk_reserve; /* Number of reserved items. */ 196 uint32_t uk_size; /* Requested size of each item */ 197 uint32_t uk_rsize; /* Real size of each item */ 198 uint32_t uk_maxpages; /* Maximum number of pages to alloc */ 199 200 uma_init uk_init; /* Keg's init routine */ 201 uma_fini uk_fini; /* Keg's fini routine */ 202 uma_alloc uk_allocf; /* Allocation function */ 203 uma_free uk_freef; /* Free routine */ 204 205 u_long uk_offset; /* Next free offset from base KVA */ 206 vm_offset_t uk_kva; /* Zone base KVA */ 207 uma_zone_t uk_slabzone; /* Slab zone backing us, if OFFPAGE */ 208 209 uint16_t uk_slabsize; /* Slab size for this keg */ 210 uint16_t uk_pgoff; /* Offset to uma_slab struct */ 211 uint16_t uk_ppera; /* pages per allocation from backend */ 212 uint16_t uk_ipers; /* Items per slab */ 213 uint32_t uk_flags; /* Internal flags */ 214 215 /* Least used fields go to the last cache line. */ 216 const char *uk_name; /* Name of creating zone. */ 217 LIST_ENTRY(uma_keg) uk_link; /* List of all kegs */ 218 }; 219 typedef struct uma_keg * uma_keg_t; 220 221 /* 222 * Free bits per-slab. 223 */ 224 #define SLAB_SETSIZE (PAGE_SIZE / UMA_SMALLEST_UNIT) 225 BITSET_DEFINE(slabbits, SLAB_SETSIZE); 226 227 /* 228 * The slab structure manages a single contiguous allocation from backing 229 * store and subdivides it into individually allocatable items. 230 */ 231 struct uma_slab { 232 uma_keg_t us_keg; /* Keg we live in */ 233 union { 234 LIST_ENTRY(uma_slab) _us_link; /* slabs in zone */ 235 unsigned long _us_size; /* Size of allocation */ 236 } us_type; 237 SLIST_ENTRY(uma_slab) us_hlink; /* Link for hash table */ 238 uint8_t *us_data; /* First item */ 239 struct slabbits us_free; /* Free bitmask. */ 240 #ifdef INVARIANTS 241 struct slabbits us_debugfree; /* Debug bitmask. */ 242 #endif 243 uint16_t us_freecount; /* How many are free? */ 244 uint8_t us_flags; /* Page flags see uma.h */ 245 uint8_t us_pad; /* Pad to 32bits, unused. */ 246 }; 247 248 #define us_link us_type._us_link 249 #define us_size us_type._us_size 250 251 /* 252 * The slab structure for UMA_ZONE_REFCNT zones for whose items we 253 * maintain reference counters in the slab for. 254 */ 255 struct uma_slab_refcnt { 256 struct uma_slab us_head; /* slab header data */ 257 uint32_t us_refcnt[0]; /* Actually larger. */ 258 }; 259 260 typedef struct uma_slab * uma_slab_t; 261 typedef struct uma_slab_refcnt * uma_slabrefcnt_t; 262 typedef uma_slab_t (*uma_slaballoc)(uma_zone_t, uma_keg_t, int); 263 264 struct uma_klink { 265 LIST_ENTRY(uma_klink) kl_link; 266 uma_keg_t kl_keg; 267 }; 268 typedef struct uma_klink *uma_klink_t; 269 270 /* 271 * Zone management structure 272 * 273 * TODO: Optimize for cache line size 274 * 275 */ 276 struct uma_zone { 277 struct mtx_padalign uz_lock; /* Lock for the zone */ 278 struct mtx_padalign *uz_lockptr; 279 const char *uz_name; /* Text name of the zone */ 280 281 LIST_ENTRY(uma_zone) uz_link; /* List of all zones in keg */ 282 LIST_HEAD(,uma_bucket) uz_buckets; /* full buckets */ 283 284 LIST_HEAD(,uma_klink) uz_kegs; /* List of kegs. */ 285 struct uma_klink uz_klink; /* klink for first keg. */ 286 287 uma_slaballoc uz_slab; /* Allocate a slab from the backend. */ 288 uma_ctor uz_ctor; /* Constructor for each allocation */ 289 uma_dtor uz_dtor; /* Destructor */ 290 uma_init uz_init; /* Initializer for each item */ 291 uma_fini uz_fini; /* Finalizer for each item. */ 292 uma_import uz_import; /* Import new memory to cache. */ 293 uma_release uz_release; /* Release memory from cache. */ 294 void *uz_arg; /* Import/release argument. */ 295 296 uint32_t uz_flags; /* Flags inherited from kegs */ 297 uint32_t uz_size; /* Size inherited from kegs */ 298 299 volatile u_long uz_allocs UMA_ALIGN; /* Total number of allocations */ 300 volatile u_long uz_fails; /* Total number of alloc failures */ 301 volatile u_long uz_frees; /* Total number of frees */ 302 uint64_t uz_sleeps; /* Total number of alloc sleeps */ 303 uint16_t uz_count; /* Highest amount of items in bucket */ 304 305 /* The next three fields are used to print a rate-limited warnings. */ 306 const char *uz_warning; /* Warning to print on failure */ 307 struct timeval uz_ratecheck; /* Warnings rate-limiting */ 308 309 /* 310 * This HAS to be the last item because we adjust the zone size 311 * based on NCPU and then allocate the space for the zones. 312 */ 313 struct uma_cache uz_cpu[1]; /* Per cpu caches */ 314 }; 315 316 /* 317 * These flags must not overlap with the UMA_ZONE flags specified in uma.h. 318 */ 319 #define UMA_ZFLAG_MULTI 0x04000000 /* Multiple kegs in the zone. */ 320 #define UMA_ZFLAG_DRAINING 0x08000000 /* Running zone_drain. */ 321 #define UMA_ZFLAG_BUCKET 0x10000000 /* Bucket zone. */ 322 #define UMA_ZFLAG_INTERNAL 0x20000000 /* No offpage no PCPU. */ 323 #define UMA_ZFLAG_FULL 0x40000000 /* Reached uz_maxpages */ 324 #define UMA_ZFLAG_CACHEONLY 0x80000000 /* Don't ask VM for buckets. */ 325 326 #define UMA_ZFLAG_INHERIT \ 327 (UMA_ZFLAG_INTERNAL | UMA_ZFLAG_CACHEONLY | UMA_ZFLAG_BUCKET) 328 329 static inline uma_keg_t 330 zone_first_keg(uma_zone_t zone) 331 { 332 uma_klink_t klink; 333 334 klink = LIST_FIRST(&zone->uz_kegs); 335 return (klink != NULL) ? klink->kl_keg : NULL; 336 } 337 338 #undef UMA_ALIGN 339 340 #ifdef _KERNEL 341 /* Internal prototypes */ 342 static __inline uma_slab_t hash_sfind(struct uma_hash *hash, uint8_t *data); 343 void *uma_large_malloc(int size, int wait); 344 void uma_large_free(uma_slab_t slab); 345 346 /* Lock Macros */ 347 348 #define KEG_LOCK_INIT(k, lc) \ 349 do { \ 350 if ((lc)) \ 351 mtx_init(&(k)->uk_lock, (k)->uk_name, \ 352 (k)->uk_name, MTX_DEF | MTX_DUPOK); \ 353 else \ 354 mtx_init(&(k)->uk_lock, (k)->uk_name, \ 355 "UMA zone", MTX_DEF | MTX_DUPOK); \ 356 } while (0) 357 358 #define KEG_LOCK_FINI(k) mtx_destroy(&(k)->uk_lock) 359 #define KEG_LOCK(k) mtx_lock(&(k)->uk_lock) 360 #define KEG_UNLOCK(k) mtx_unlock(&(k)->uk_lock) 361 362 #define ZONE_LOCK_INIT(z, lc) \ 363 do { \ 364 if ((lc)) \ 365 mtx_init(&(z)->uz_lock, (z)->uz_name, \ 366 (z)->uz_name, MTX_DEF | MTX_DUPOK); \ 367 else \ 368 mtx_init(&(z)->uz_lock, (z)->uz_name, \ 369 "UMA zone", MTX_DEF | MTX_DUPOK); \ 370 } while (0) 371 372 #define ZONE_LOCK(z) mtx_lock((z)->uz_lockptr) 373 #define ZONE_TRYLOCK(z) mtx_trylock((z)->uz_lockptr) 374 #define ZONE_UNLOCK(z) mtx_unlock((z)->uz_lockptr) 375 #define ZONE_LOCK_FINI(z) mtx_destroy(&(z)->uz_lock) 376 377 /* 378 * Find a slab within a hash table. This is used for OFFPAGE zones to lookup 379 * the slab structure. 380 * 381 * Arguments: 382 * hash The hash table to search. 383 * data The base page of the item. 384 * 385 * Returns: 386 * A pointer to a slab if successful, else NULL. 387 */ 388 static __inline uma_slab_t 389 hash_sfind(struct uma_hash *hash, uint8_t *data) 390 { 391 uma_slab_t slab; 392 int hval; 393 394 hval = UMA_HASH(hash, data); 395 396 SLIST_FOREACH(slab, &hash->uh_slab_hash[hval], us_hlink) { 397 if ((uint8_t *)slab->us_data == data) 398 return (slab); 399 } 400 return (NULL); 401 } 402 403 static __inline uma_slab_t 404 vtoslab(vm_offset_t va) 405 { 406 vm_page_t p; 407 uma_slab_t slab; 408 409 p = PHYS_TO_VM_PAGE(pmap_kextract(va)); 410 slab = (uma_slab_t )p->plinks.s.pv; 411 412 if (p->flags & PG_SLAB) 413 return (slab); 414 else 415 return (NULL); 416 } 417 418 static __inline void 419 vsetslab(vm_offset_t va, uma_slab_t slab) 420 { 421 vm_page_t p; 422 423 p = PHYS_TO_VM_PAGE(pmap_kextract(va)); 424 p->plinks.s.pv = slab; 425 p->flags |= PG_SLAB; 426 } 427 428 /* 429 * The following two functions may be defined by architecture specific code 430 * if they can provide more effecient allocation functions. This is useful 431 * for using direct mapped addresses. 432 */ 433 void *uma_small_alloc(uma_zone_t zone, int bytes, uint8_t *pflag, int wait); 434 void uma_small_free(void *mem, int size, uint8_t flags); 435 #endif /* _KERNEL */ 436 437 #endif /* VM_UMA_INT_H */ 438