aboutsummaryrefslogtreecommitdiffstats
path: root/mm
diff options
context:
space:
mode:
authorMatt Mackall <mpm@selenic.com>2006-01-08 04:01:45 -0500
committerLinus Torvalds <torvalds@g5.osdl.org>2006-01-08 23:13:41 -0500
commit10cef6029502915bdb3cf0821d425cf9dc30c817 (patch)
tree2c9dfef95d58b64dcf4cdf3c32b18164928b438e /mm
parent30992c97ae9d01b17374fbfab76a869fb4bba500 (diff)
[PATCH] slob: introduce the SLOB allocator
configurable replacement for slab allocator This adds a CONFIG_SLAB option under CONFIG_EMBEDDED. When CONFIG_SLAB is disabled, the kernel falls back to using the 'SLOB' allocator. SLOB is a traditional K&R/UNIX allocator with a SLAB emulation layer, similar to the original Linux kmalloc allocator that SLAB replaced. It's signicantly smaller code and is more memory efficient. But like all similar allocators, it scales poorly and suffers from fragmentation more than SLAB, so it's only appropriate for small systems. It's been tested extensively in the Linux-tiny tree. I've also stress-tested it with make -j 8 compiles on a 3G SMP+PREEMPT box (not recommended). Here's a comparison for otherwise identical builds, showing SLOB saving nearly half a megabyte of RAM: $ size vmlinux* text data bss dec hex filename 3336372 529360 190812 4056544 3de5e0 vmlinux-slab 3323208 527948 190684 4041840 3dac70 vmlinux-slob $ size mm/{slab,slob}.o text data bss dec hex filename 13221 752 48 14021 36c5 mm/slab.o 1896 52 8 1956 7a4 mm/slob.o /proc/meminfo: SLAB SLOB delta MemTotal: 27964 kB 27980 kB +16 kB MemFree: 24596 kB 25092 kB +496 kB Buffers: 36 kB 36 kB 0 kB Cached: 1188 kB 1188 kB 0 kB SwapCached: 0 kB 0 kB 0 kB Active: 608 kB 600 kB -8 kB Inactive: 808 kB 812 kB +4 kB HighTotal: 0 kB 0 kB 0 kB HighFree: 0 kB 0 kB 0 kB LowTotal: 27964 kB 27980 kB +16 kB LowFree: 24596 kB 25092 kB +496 kB SwapTotal: 0 kB 0 kB 0 kB SwapFree: 0 kB 0 kB 0 kB Dirty: 4 kB 12 kB +8 kB Writeback: 0 kB 0 kB 0 kB Mapped: 560 kB 556 kB -4 kB Slab: 1756 kB 0 kB -1756 kB CommitLimit: 13980 kB 13988 kB +8 kB Committed_AS: 4208 kB 4208 kB 0 kB PageTables: 28 kB 28 kB 0 kB VmallocTotal: 1007312 kB 1007312 kB 0 kB VmallocUsed: 48 kB 48 kB 0 kB VmallocChunk: 1007264 kB 1007264 kB 0 kB (this work has been sponsored in part by CELF) From: Ingo Molnar <mingo@elte.hu> Fix 32-bitness bugs in mm/slob.c. Signed-off-by: Matt Mackall <mpm@selenic.com> Signed-off-by: Ingo Molnar <mingo@elte.hu> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Diffstat (limited to 'mm')
-rw-r--r--mm/Makefile4
-rw-r--r--mm/slob.c385
2 files changed, 388 insertions, 1 deletions
diff --git a/mm/Makefile b/mm/Makefile
index 74c85ddc9176..9aa03fa1dcc3 100644
--- a/mm/Makefile
+++ b/mm/Makefile
@@ -9,7 +9,7 @@ mmu-$(CONFIG_MMU) := fremap.o highmem.o madvise.o memory.o mincore.o \
9 9
10obj-y := bootmem.o filemap.o mempool.o oom_kill.o fadvise.o \ 10obj-y := bootmem.o filemap.o mempool.o oom_kill.o fadvise.o \
11 page_alloc.o page-writeback.o pdflush.o \ 11 page_alloc.o page-writeback.o pdflush.o \
12 readahead.o slab.o swap.o truncate.o vmscan.o \ 12 readahead.o swap.o truncate.o vmscan.o \
13 prio_tree.o util.o $(mmu-y) 13 prio_tree.o util.o $(mmu-y)
14 14
15obj-$(CONFIG_SWAP) += page_io.o swap_state.o swapfile.o thrash.o 15obj-$(CONFIG_SWAP) += page_io.o swap_state.o swapfile.o thrash.o
@@ -18,5 +18,7 @@ obj-$(CONFIG_NUMA) += mempolicy.o
18obj-$(CONFIG_SPARSEMEM) += sparse.o 18obj-$(CONFIG_SPARSEMEM) += sparse.o
19obj-$(CONFIG_SHMEM) += shmem.o 19obj-$(CONFIG_SHMEM) += shmem.o
20obj-$(CONFIG_TINY_SHMEM) += tiny-shmem.o 20obj-$(CONFIG_TINY_SHMEM) += tiny-shmem.o
21obj-$(CONFIG_SLOB) += slob.o
22obj-$(CONFIG_SLAB) += slab.o
21obj-$(CONFIG_MEMORY_HOTPLUG) += memory_hotplug.o 23obj-$(CONFIG_MEMORY_HOTPLUG) += memory_hotplug.o
22obj-$(CONFIG_FS_XIP) += filemap_xip.o 24obj-$(CONFIG_FS_XIP) += filemap_xip.o
diff --git a/mm/slob.c b/mm/slob.c
new file mode 100644
index 000000000000..1c240c4b71d9
--- /dev/null
+++ b/mm/slob.c
@@ -0,0 +1,385 @@
1/*
2 * SLOB Allocator: Simple List Of Blocks
3 *
4 * Matt Mackall <mpm@selenic.com> 12/30/03
5 *
6 * How SLOB works:
7 *
8 * The core of SLOB is a traditional K&R style heap allocator, with
9 * support for returning aligned objects. The granularity of this
10 * allocator is 8 bytes on x86, though it's perhaps possible to reduce
11 * this to 4 if it's deemed worth the effort. The slob heap is a
12 * singly-linked list of pages from __get_free_page, grown on demand
13 * and allocation from the heap is currently first-fit.
14 *
15 * Above this is an implementation of kmalloc/kfree. Blocks returned
16 * from kmalloc are 8-byte aligned and prepended with a 8-byte header.
17 * If kmalloc is asked for objects of PAGE_SIZE or larger, it calls
18 * __get_free_pages directly so that it can return page-aligned blocks
19 * and keeps a linked list of such pages and their orders. These
20 * objects are detected in kfree() by their page alignment.
21 *
22 * SLAB is emulated on top of SLOB by simply calling constructors and
23 * destructors for every SLAB allocation. Objects are returned with
24 * the 8-byte alignment unless the SLAB_MUST_HWCACHE_ALIGN flag is
25 * set, in which case the low-level allocator will fragment blocks to
26 * create the proper alignment. Again, objects of page-size or greater
27 * are allocated by calling __get_free_pages. As SLAB objects know
28 * their size, no separate size bookkeeping is necessary and there is
29 * essentially no allocation space overhead.
30 */
31
32#include <linux/config.h>
33#include <linux/slab.h>
34#include <linux/mm.h>
35#include <linux/cache.h>
36#include <linux/init.h>
37#include <linux/module.h>
38#include <linux/timer.h>
39
40struct slob_block {
41 int units;
42 struct slob_block *next;
43};
44typedef struct slob_block slob_t;
45
46#define SLOB_UNIT sizeof(slob_t)
47#define SLOB_UNITS(size) (((size) + SLOB_UNIT - 1)/SLOB_UNIT)
48#define SLOB_ALIGN L1_CACHE_BYTES
49
50struct bigblock {
51 int order;
52 void *pages;
53 struct bigblock *next;
54};
55typedef struct bigblock bigblock_t;
56
57static slob_t arena = { .next = &arena, .units = 1 };
58static slob_t *slobfree = &arena;
59static bigblock_t *bigblocks;
60static DEFINE_SPINLOCK(slob_lock);
61static DEFINE_SPINLOCK(block_lock);
62
63static void slob_free(void *b, int size);
64
65static void *slob_alloc(size_t size, gfp_t gfp, int align)
66{
67 slob_t *prev, *cur, *aligned = 0;
68 int delta = 0, units = SLOB_UNITS(size);
69 unsigned long flags;
70
71 spin_lock_irqsave(&slob_lock, flags);
72 prev = slobfree;
73 for (cur = prev->next; ; prev = cur, cur = cur->next) {
74 if (align) {
75 aligned = (slob_t *)ALIGN((unsigned long)cur, align);
76 delta = aligned - cur;
77 }
78 if (cur->units >= units + delta) { /* room enough? */
79 if (delta) { /* need to fragment head to align? */
80 aligned->units = cur->units - delta;
81 aligned->next = cur->next;
82 cur->next = aligned;
83 cur->units = delta;
84 prev = cur;
85 cur = aligned;
86 }
87
88 if (cur->units == units) /* exact fit? */
89 prev->next = cur->next; /* unlink */
90 else { /* fragment */
91 prev->next = cur + units;
92 prev->next->units = cur->units - units;
93 prev->next->next = cur->next;
94 cur->units = units;
95 }
96
97 slobfree = prev;
98 spin_unlock_irqrestore(&slob_lock, flags);
99 return cur;
100 }
101 if (cur == slobfree) {
102 spin_unlock_irqrestore(&slob_lock, flags);
103
104 if (size == PAGE_SIZE) /* trying to shrink arena? */
105 return 0;
106
107 cur = (slob_t *)__get_free_page(gfp);
108 if (!cur)
109 return 0;
110
111 slob_free(cur, PAGE_SIZE);
112 spin_lock_irqsave(&slob_lock, flags);
113 cur = slobfree;
114 }
115 }
116}
117
118static void slob_free(void *block, int size)
119{
120 slob_t *cur, *b = (slob_t *)block;
121 unsigned long flags;
122
123 if (!block)
124 return;
125
126 if (size)
127 b->units = SLOB_UNITS(size);
128
129 /* Find reinsertion point */
130 spin_lock_irqsave(&slob_lock, flags);
131 for (cur = slobfree; !(b > cur && b < cur->next); cur = cur->next)
132 if (cur >= cur->next && (b > cur || b < cur->next))
133 break;
134
135 if (b + b->units == cur->next) {
136 b->units += cur->next->units;
137 b->next = cur->next->next;
138 } else
139 b->next = cur->next;
140
141 if (cur + cur->units == b) {
142 cur->units += b->units;
143 cur->next = b->next;
144 } else
145 cur->next = b;
146
147 slobfree = cur;
148
149 spin_unlock_irqrestore(&slob_lock, flags);
150}
151
152static int FASTCALL(find_order(int size));
153static int fastcall find_order(int size)
154{
155 int order = 0;
156 for ( ; size > 4096 ; size >>=1)
157 order++;
158 return order;
159}
160
161void *kmalloc(size_t size, gfp_t gfp)
162{
163 slob_t *m;
164 bigblock_t *bb;
165 unsigned long flags;
166
167 if (size < PAGE_SIZE - SLOB_UNIT) {
168 m = slob_alloc(size + SLOB_UNIT, gfp, 0);
169 return m ? (void *)(m + 1) : 0;
170 }
171
172 bb = slob_alloc(sizeof(bigblock_t), gfp, 0);
173 if (!bb)
174 return 0;
175
176 bb->order = find_order(size);
177 bb->pages = (void *)__get_free_pages(gfp, bb->order);
178
179 if (bb->pages) {
180 spin_lock_irqsave(&block_lock, flags);
181 bb->next = bigblocks;
182 bigblocks = bb;
183 spin_unlock_irqrestore(&block_lock, flags);
184 return bb->pages;
185 }
186
187 slob_free(bb, sizeof(bigblock_t));
188 return 0;
189}
190
191EXPORT_SYMBOL(kmalloc);
192
193void kfree(const void *block)
194{
195 bigblock_t *bb, **last = &bigblocks;
196 unsigned long flags;
197
198 if (!block)
199 return;
200
201 if (!((unsigned long)block & (PAGE_SIZE-1))) {
202 /* might be on the big block list */
203 spin_lock_irqsave(&block_lock, flags);
204 for (bb = bigblocks; bb; last = &bb->next, bb = bb->next) {
205 if (bb->pages == block) {
206 *last = bb->next;
207 spin_unlock_irqrestore(&block_lock, flags);
208 free_pages((unsigned long)block, bb->order);
209 slob_free(bb, sizeof(bigblock_t));
210 return;
211 }
212 }
213 spin_unlock_irqrestore(&block_lock, flags);
214 }
215
216 slob_free((slob_t *)block - 1, 0);
217 return;
218}
219
220EXPORT_SYMBOL(kfree);
221
222unsigned int ksize(const void *block)
223{
224 bigblock_t *bb;
225 unsigned long flags;
226
227 if (!block)
228 return 0;
229
230 if (!((unsigned long)block & (PAGE_SIZE-1))) {
231 spin_lock_irqsave(&block_lock, flags);
232 for (bb = bigblocks; bb; bb = bb->next)
233 if (bb->pages == block) {
234 spin_unlock_irqrestore(&slob_lock, flags);
235 return PAGE_SIZE << bb->order;
236 }
237 spin_unlock_irqrestore(&block_lock, flags);
238 }
239
240 return ((slob_t *)block - 1)->units * SLOB_UNIT;
241}
242
243struct kmem_cache {
244 unsigned int size, align;
245 const char *name;
246 void (*ctor)(void *, struct kmem_cache *, unsigned long);
247 void (*dtor)(void *, struct kmem_cache *, unsigned long);
248};
249
250struct kmem_cache *kmem_cache_create(const char *name, size_t size,
251 size_t align, unsigned long flags,
252 void (*ctor)(void*, struct kmem_cache *, unsigned long),
253 void (*dtor)(void*, struct kmem_cache *, unsigned long))
254{
255 struct kmem_cache *c;
256
257 c = slob_alloc(sizeof(struct kmem_cache), flags, 0);
258
259 if (c) {
260 c->name = name;
261 c->size = size;
262 c->ctor = ctor;
263 c->dtor = dtor;
264 /* ignore alignment unless it's forced */
265 c->align = (flags & SLAB_MUST_HWCACHE_ALIGN) ? SLOB_ALIGN : 0;
266 if (c->align < align)
267 c->align = align;
268 }
269
270 return c;
271}
272EXPORT_SYMBOL(kmem_cache_create);
273
274int kmem_cache_destroy(struct kmem_cache *c)
275{
276 slob_free(c, sizeof(struct kmem_cache));
277 return 0;
278}
279EXPORT_SYMBOL(kmem_cache_destroy);
280
281void *kmem_cache_alloc(struct kmem_cache *c, gfp_t flags)
282{
283 void *b;
284
285 if (c->size < PAGE_SIZE)
286 b = slob_alloc(c->size, flags, c->align);
287 else
288 b = (void *)__get_free_pages(flags, find_order(c->size));
289
290 if (c->ctor)
291 c->ctor(b, c, SLAB_CTOR_CONSTRUCTOR);
292
293 return b;
294}
295EXPORT_SYMBOL(kmem_cache_alloc);
296
297void kmem_cache_free(struct kmem_cache *c, void *b)
298{
299 if (c->dtor)
300 c->dtor(b, c, 0);
301
302 if (c->size < PAGE_SIZE)
303 slob_free(b, c->size);
304 else
305 free_pages((unsigned long)b, find_order(c->size));
306}
307EXPORT_SYMBOL(kmem_cache_free);
308
309unsigned int kmem_cache_size(struct kmem_cache *c)
310{
311 return c->size;
312}
313EXPORT_SYMBOL(kmem_cache_size);
314
315const char *kmem_cache_name(struct kmem_cache *c)
316{
317 return c->name;
318}
319EXPORT_SYMBOL(kmem_cache_name);
320
321static struct timer_list slob_timer = TIMER_INITIALIZER(
322 (void (*)(unsigned long))kmem_cache_init, 0, 0);
323
324void kmem_cache_init(void)
325{
326 void *p = slob_alloc(PAGE_SIZE, 0, PAGE_SIZE-1);
327
328 if (p)
329 free_page((unsigned long)p);
330
331 mod_timer(&slob_timer, jiffies + HZ);
332}
333
334atomic_t slab_reclaim_pages = ATOMIC_INIT(0);
335EXPORT_SYMBOL(slab_reclaim_pages);
336
337#ifdef CONFIG_SMP
338
339void *__alloc_percpu(size_t size, size_t align)
340{
341 int i;
342 struct percpu_data *pdata = kmalloc(sizeof (*pdata), GFP_KERNEL);
343
344 if (!pdata)
345 return NULL;
346
347 for (i = 0; i < NR_CPUS; i++) {
348 if (!cpu_possible(i))
349 continue;
350 pdata->ptrs[i] = kmalloc(size, GFP_KERNEL);
351 if (!pdata->ptrs[i])
352 goto unwind_oom;
353 memset(pdata->ptrs[i], 0, size);
354 }
355
356 /* Catch derefs w/o wrappers */
357 return (void *) (~(unsigned long) pdata);
358
359unwind_oom:
360 while (--i >= 0) {
361 if (!cpu_possible(i))
362 continue;
363 kfree(pdata->ptrs[i]);
364 }
365 kfree(pdata);
366 return NULL;
367}
368EXPORT_SYMBOL(__alloc_percpu);
369
370void
371free_percpu(const void *objp)
372{
373 int i;
374 struct percpu_data *p = (struct percpu_data *) (~(unsigned long) objp);
375
376 for (i = 0; i < NR_CPUS; i++) {
377 if (!cpu_possible(i))
378 continue;
379 kfree(p->ptrs[i]);
380 }
381 kfree(p);
382}
383EXPORT_SYMBOL(free_percpu);
384
385#endif