From 4df7b3e0370ab6161ea2f258f51dd7c43bef2bda Mon Sep 17 00:00:00 2001 From: Roel Kluin Date: Wed, 15 Jul 2009 20:29:07 +0200 Subject: Dynamic debug: fix typo: -/-> The member was intended, not the local variable. Signed-off-by: Roel Kluin Cc: Jason Baron Cc: Greg Banks Signed-off-by: Andrew Morton Signed-off-by: Greg Kroah-Hartman --- lib/dynamic_debug.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'lib') diff --git a/lib/dynamic_debug.c b/lib/dynamic_debug.c index 833139ce1e22..e22c148e4b7f 100644 --- a/lib/dynamic_debug.c +++ b/lib/dynamic_debug.c @@ -164,7 +164,7 @@ static void ddebug_change(const struct ddebug_query *query, if (!newflags) dt->num_enabled--; - else if (!dp-flags) + else if (!dp->flags) dt->num_enabled++; dp->flags = newflags; if (newflags) { -- cgit v1.2.3 From 3fc7b4b220c7e830a5b3ce0ea5f85a635e0c50f0 Mon Sep 17 00:00:00 2001 From: Roland Dreier Date: Wed, 29 Jul 2009 15:04:02 -0700 Subject: lib: export generic atomic64_t functions The generic atomic64_t implementation in lib/ did not export the functions it defined, which means that modules that use atomic64_t would not link on platforms (such as 32-bit powerpc). For example, trying to build a kernel with CONFIG_NET_RDS on such a platform would fail with: ERROR: "atomic64_read" [net/rds/rds.ko] undefined! ERROR: "atomic64_set" [net/rds/rds.ko] undefined! Fix this by exporting the atomic64_t functions to modules. (I export the entire API even if it's not all currently used by in-tree modules to avoid having to continue fixing this in dribs and drabs) Signed-off-by: Roland Dreier Acked-by: Paul Mackerras Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- lib/atomic64.c | 11 +++++++++++ 1 file changed, 11 insertions(+) (limited to 'lib') diff --git a/lib/atomic64.c b/lib/atomic64.c index c5e725562416..8bee16ec7524 100644 --- a/lib/atomic64.c +++ b/lib/atomic64.c @@ -13,6 +13,7 @@ #include #include #include +#include #include /* @@ -52,6 +53,7 @@ long long atomic64_read(const atomic64_t *v) spin_unlock_irqrestore(lock, flags); return val; } +EXPORT_SYMBOL(atomic64_read); void atomic64_set(atomic64_t *v, long long i) { @@ -62,6 +64,7 @@ void atomic64_set(atomic64_t *v, long long i) v->counter = i; spin_unlock_irqrestore(lock, flags); } +EXPORT_SYMBOL(atomic64_set); void atomic64_add(long long a, atomic64_t *v) { @@ -72,6 +75,7 @@ void atomic64_add(long long a, atomic64_t *v) v->counter += a; spin_unlock_irqrestore(lock, flags); } +EXPORT_SYMBOL(atomic64_add); long long atomic64_add_return(long long a, atomic64_t *v) { @@ -84,6 +88,7 @@ long long atomic64_add_return(long long a, atomic64_t *v) spin_unlock_irqrestore(lock, flags); return val; } +EXPORT_SYMBOL(atomic64_add_return); void atomic64_sub(long long a, atomic64_t *v) { @@ -94,6 +99,7 @@ void atomic64_sub(long long a, atomic64_t *v) v->counter -= a; spin_unlock_irqrestore(lock, flags); } +EXPORT_SYMBOL(atomic64_sub); long long atomic64_sub_return(long long a, atomic64_t *v) { @@ -106,6 +112,7 @@ long long atomic64_sub_return(long long a, atomic64_t *v) spin_unlock_irqrestore(lock, flags); return val; } +EXPORT_SYMBOL(atomic64_sub_return); long long atomic64_dec_if_positive(atomic64_t *v) { @@ -120,6 +127,7 @@ long long atomic64_dec_if_positive(atomic64_t *v) spin_unlock_irqrestore(lock, flags); return val; } +EXPORT_SYMBOL(atomic64_dec_if_positive); long long atomic64_cmpxchg(atomic64_t *v, long long o, long long n) { @@ -134,6 +142,7 @@ long long atomic64_cmpxchg(atomic64_t *v, long long o, long long n) spin_unlock_irqrestore(lock, flags); return val; } +EXPORT_SYMBOL(atomic64_cmpxchg); long long atomic64_xchg(atomic64_t *v, long long new) { @@ -147,6 +156,7 @@ long long atomic64_xchg(atomic64_t *v, long long new) spin_unlock_irqrestore(lock, flags); return val; } +EXPORT_SYMBOL(atomic64_xchg); int atomic64_add_unless(atomic64_t *v, long long a, long long u) { @@ -162,6 +172,7 @@ int atomic64_add_unless(atomic64_t *v, long long a, long long u) spin_unlock_irqrestore(lock, flags); return ret; } +EXPORT_SYMBOL(atomic64_add_unless); static int init_atomic64_lock(void) { -- cgit v1.2.3 From 534acc057b5a08ec33fa57cdd2f5a09ef124e7f2 Mon Sep 17 00:00:00 2001 From: Dave Hansen Date: Wed, 29 Jul 2009 15:04:18 -0700 Subject: lib: flexible array implementation Once a structure goes over PAGE_SIZE*2, we see occasional allocation failures. Some people have chosen to switch over to things like vmalloc() that will let them keep array-like access to such a large structures. But, vmalloc() has plenty of downsides. Here's an alternative. I think it's what Andrew was suggesting here: http://lkml.org/lkml/2009/7/2/518 I call it a flexible array. It does all of its work in PAGE_SIZE bits, so never does an order>0 allocation. The base level has PAGE_SIZE-2*sizeof(int) bytes of storage for pointers to the second level. So, with a 32-bit arch, you get about 4MB (4183112 bytes) of total storage when the objects pack nicely into a page. It is half that on 64-bit because the pointers are twice the size. There's a table detailing this in the code. There are kerneldocs for the functions, but here's an overview: flex_array_alloc() - dynamically allocate a base structure flex_array_free() - free the array and all of the second-level pages flex_array_free_parts() - free the second-level pages, but not the base (for static bases) flex_array_put() - copy into the array at the given index flex_array_get() - copy out of the array at the given index flex_array_prealloc() - preallocate the second-level pages between the given indexes to guarantee no allocs will occur at put() time. We could also potentially just pass the "element_size" into each of the API functions instead of storing it internally. That would get us one more base pointer on 32-bit. I've been testing this by running it in userspace. The header and patch that I've been using are here, as well as the little script I'm using to generate the size table which goes in the kerneldocs. http://sr71.net/~dave/linux/flexarray/ [akpm@linux-foundation.org: coding-style fixes] Signed-off-by: Dave Hansen Reviewed-by: KAMEZAWA Hiroyuki Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- include/linux/flex_array.h | 47 ++++++++ lib/Makefile | 2 +- lib/flex_array.c | 269 +++++++++++++++++++++++++++++++++++++++++++++ 3 files changed, 317 insertions(+), 1 deletion(-) create mode 100644 include/linux/flex_array.h create mode 100644 lib/flex_array.c (limited to 'lib') diff --git a/include/linux/flex_array.h b/include/linux/flex_array.h new file mode 100644 index 000000000000..23c1ec79a31b --- /dev/null +++ b/include/linux/flex_array.h @@ -0,0 +1,47 @@ +#ifndef _FLEX_ARRAY_H +#define _FLEX_ARRAY_H + +#include +#include + +#define FLEX_ARRAY_PART_SIZE PAGE_SIZE +#define FLEX_ARRAY_BASE_SIZE PAGE_SIZE + +struct flex_array_part; + +/* + * This is meant to replace cases where an array-like + * structure has gotten too big to fit into kmalloc() + * and the developer is getting tempted to use + * vmalloc(). + */ + +struct flex_array { + union { + struct { + int element_size; + int total_nr_elements; + struct flex_array_part *parts[0]; + }; + /* + * This little trick makes sure that + * sizeof(flex_array) == PAGE_SIZE + */ + char padding[FLEX_ARRAY_BASE_SIZE]; + }; +}; + +#define FLEX_ARRAY_INIT(size, total) { { {\ + .element_size = (size), \ + .total_nr_elements = (total), \ +} } } + +struct flex_array *flex_array_alloc(int element_size, int total, gfp_t flags); +int flex_array_prealloc(struct flex_array *fa, int start, int end, gfp_t flags); +void flex_array_free(struct flex_array *fa); +void flex_array_free_parts(struct flex_array *fa); +int flex_array_put(struct flex_array *fa, int element_nr, void *src, + gfp_t flags); +void *flex_array_get(struct flex_array *fa, int element_nr); + +#endif /* _FLEX_ARRAY_H */ diff --git a/lib/Makefile b/lib/Makefile index b6d1857bbf08..2e78277eff9d 100644 --- a/lib/Makefile +++ b/lib/Makefile @@ -12,7 +12,7 @@ lib-y := ctype.o string.o vsprintf.o cmdline.o \ idr.o int_sqrt.o extable.o prio_tree.o \ sha1.o irq_regs.o reciprocal_div.o argv_split.o \ proportions.o prio_heap.o ratelimit.o show_mem.o \ - is_single_threaded.o plist.o decompress.o + is_single_threaded.o plist.o decompress.o flex_array.o lib-$(CONFIG_MMU) += ioremap.o lib-$(CONFIG_SMP) += cpumask.o diff --git a/lib/flex_array.c b/lib/flex_array.c new file mode 100644 index 000000000000..0e7894ce8882 --- /dev/null +++ b/lib/flex_array.c @@ -0,0 +1,269 @@ +/* + * Flexible array managed in PAGE_SIZE parts + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. + * + * Copyright IBM Corporation, 2009 + * + * Author: Dave Hansen + */ + +#include +#include +#include + +struct flex_array_part { + char elements[FLEX_ARRAY_PART_SIZE]; +}; + +static inline int __elements_per_part(int element_size) +{ + return FLEX_ARRAY_PART_SIZE / element_size; +} + +static inline int bytes_left_in_base(void) +{ + int element_offset = offsetof(struct flex_array, parts); + int bytes_left = FLEX_ARRAY_BASE_SIZE - element_offset; + return bytes_left; +} + +static inline int nr_base_part_ptrs(void) +{ + return bytes_left_in_base() / sizeof(struct flex_array_part *); +} + +/* + * If a user requests an allocation which is small + * enough, we may simply use the space in the + * flex_array->parts[] array to store the user + * data. + */ +static inline int elements_fit_in_base(struct flex_array *fa) +{ + int data_size = fa->element_size * fa->total_nr_elements; + if (data_size <= bytes_left_in_base()) + return 1; + return 0; +} + +/** + * flex_array_alloc - allocate a new flexible array + * @element_size: the size of individual elements in the array + * @total: total number of elements that this should hold + * + * Note: all locking must be provided by the caller. + * + * @total is used to size internal structures. If the user ever + * accesses any array indexes >=@total, it will produce errors. + * + * The maximum number of elements is defined as: the number of + * elements that can be stored in a page times the number of + * page pointers that we can fit in the base structure or (using + * integer math): + * + * (PAGE_SIZE/element_size) * (PAGE_SIZE-8)/sizeof(void *) + * + * Here's a table showing example capacities. Note that the maximum + * index that the get/put() functions is just nr_objects-1. This + * basically means that you get 4MB of storage on 32-bit and 2MB on + * 64-bit. + * + * + * Element size | Objects | Objects | + * PAGE_SIZE=4k | 32-bit | 64-bit | + * ---------------------------------| + * 1 bytes | 4186112 | 2093056 | + * 2 bytes | 2093056 | 1046528 | + * 3 bytes | 1395030 | 697515 | + * 4 bytes | 1046528 | 523264 | + * 32 bytes | 130816 | 65408 | + * 33 bytes | 126728 | 63364 | + * 2048 bytes | 2044 | 1022 | + * 2049 bytes | 1022 | 511 | + * void * | 1046528 | 261632 | + * + * Since 64-bit pointers are twice the size, we lose half the + * capacity in the base structure. Also note that no effort is made + * to efficiently pack objects across page boundaries. + */ +struct flex_array *flex_array_alloc(int element_size, int total, gfp_t flags) +{ + struct flex_array *ret; + int max_size = nr_base_part_ptrs() * __elements_per_part(element_size); + + /* max_size will end up 0 if element_size > PAGE_SIZE */ + if (total > max_size) + return NULL; + ret = kzalloc(sizeof(struct flex_array), flags); + if (!ret) + return NULL; + ret->element_size = element_size; + ret->total_nr_elements = total; + return ret; +} + +static int fa_element_to_part_nr(struct flex_array *fa, int element_nr) +{ + return element_nr / __elements_per_part(fa->element_size); +} + +/** + * flex_array_free_parts - just free the second-level pages + * @src: address of data to copy into the array + * @element_nr: index of the position in which to insert + * the new element. + * + * This is to be used in cases where the base 'struct flex_array' + * has been statically allocated and should not be free. + */ +void flex_array_free_parts(struct flex_array *fa) +{ + int part_nr; + int max_part = nr_base_part_ptrs(); + + if (elements_fit_in_base(fa)) + return; + for (part_nr = 0; part_nr < max_part; part_nr++) + kfree(fa->parts[part_nr]); +} + +void flex_array_free(struct flex_array *fa) +{ + flex_array_free_parts(fa); + kfree(fa); +} + +static int fa_index_inside_part(struct flex_array *fa, int element_nr) +{ + return element_nr % __elements_per_part(fa->element_size); +} + +static int index_inside_part(struct flex_array *fa, int element_nr) +{ + int part_offset = fa_index_inside_part(fa, element_nr); + return part_offset * fa->element_size; +} + +static struct flex_array_part * +__fa_get_part(struct flex_array *fa, int part_nr, gfp_t flags) +{ + struct flex_array_part *part = fa->parts[part_nr]; + if (!part) { + /* + * This leaves the part pages uninitialized + * and with potentially random data, just + * as if the user had kmalloc()'d the whole. + * __GFP_ZERO can be used to zero it. + */ + part = kmalloc(FLEX_ARRAY_PART_SIZE, flags); + if (!part) + return NULL; + fa->parts[part_nr] = part; + } + return part; +} + +/** + * flex_array_put - copy data into the array at @element_nr + * @src: address of data to copy into the array + * @element_nr: index of the position in which to insert + * the new element. + * + * Note that this *copies* the contents of @src into + * the array. If you are trying to store an array of + * pointers, make sure to pass in &ptr instead of ptr. + * + * Locking must be provided by the caller. + */ +int flex_array_put(struct flex_array *fa, int element_nr, void *src, gfp_t flags) +{ + int part_nr = fa_element_to_part_nr(fa, element_nr); + struct flex_array_part *part; + void *dst; + + if (element_nr >= fa->total_nr_elements) + return -ENOSPC; + if (elements_fit_in_base(fa)) + part = (struct flex_array_part *)&fa->parts[0]; + else + part = __fa_get_part(fa, part_nr, flags); + if (!part) + return -ENOMEM; + dst = &part->elements[index_inside_part(fa, element_nr)]; + memcpy(dst, src, fa->element_size); + return 0; +} + +/** + * flex_array_prealloc - guarantee that array space exists + * @start: index of first array element for which space is allocated + * @end: index of last (inclusive) element for which space is allocated + * + * This will guarantee that no future calls to flex_array_put() + * will allocate memory. It can be used if you are expecting to + * be holding a lock or in some atomic context while writing + * data into the array. + * + * Locking must be provided by the caller. + */ +int flex_array_prealloc(struct flex_array *fa, int start, int end, gfp_t flags) +{ + int start_part; + int end_part; + int part_nr; + struct flex_array_part *part; + + if (start >= fa->total_nr_elements || end >= fa->total_nr_elements) + return -ENOSPC; + if (elements_fit_in_base(fa)) + return 0; + start_part = fa_element_to_part_nr(fa, start); + end_part = fa_element_to_part_nr(fa, end); + for (part_nr = start_part; part_nr <= end_part; part_nr++) { + part = __fa_get_part(fa, part_nr, flags); + if (!part) + return -ENOMEM; + } + return 0; +} + +/** + * flex_array_get - pull data back out of the array + * @element_nr: index of the element to fetch from the array + * + * Returns a pointer to the data at index @element_nr. Note + * that this is a copy of the data that was passed in. If you + * are using this to store pointers, you'll get back &ptr. + * + * Locking must be provided by the caller. + */ +void *flex_array_get(struct flex_array *fa, int element_nr) +{ + int part_nr = fa_element_to_part_nr(fa, element_nr); + struct flex_array_part *part; + int index; + + if (element_nr >= fa->total_nr_elements) + return NULL; + if (!fa->parts[part_nr]) + return NULL; + if (elements_fit_in_base(fa)) + part = (struct flex_array_part *)&fa->parts[0]; + else + part = fa->parts[part_nr]; + index = index_inside_part(fa, element_nr); + return &part->elements[index_inside_part(fa, element_nr)]; +} -- cgit v1.2.3 From 6de7e356faf54aa75de5b624bbce28a5b776dfa8 Mon Sep 17 00:00:00 2001 From: Sebastian Andrzej Siewior Date: Thu, 18 Jun 2009 10:19:12 +0200 Subject: lib/scatterlist: add a flags to signalize mapping direction sg_miter_start() is currently unaware of the direction of the copy process (to or from the scatter list). It is important to know the direction because the page has to be flushed in case the data written is seen on a different mapping in user land on cache incoherent architectures. Signed-off-by: Sebastian Andrzej Siewior Acked-by: FUJITA Tomonori Acked-by: Tejun Heo Signed-off-by: Pierre Ossman --- include/linux/scatterlist.h | 2 ++ lib/scatterlist.c | 16 ++++++++++++---- 2 files changed, 14 insertions(+), 4 deletions(-) (limited to 'lib') diff --git a/include/linux/scatterlist.h b/include/linux/scatterlist.h index e5996984ddd0..9aaf5bfdad1a 100644 --- a/include/linux/scatterlist.h +++ b/include/linux/scatterlist.h @@ -242,6 +242,8 @@ size_t sg_copy_to_buffer(struct scatterlist *sgl, unsigned int nents, */ #define SG_MITER_ATOMIC (1 << 0) /* use kmap_atomic */ +#define SG_MITER_TO_SG (1 << 1) /* flush back to phys on unmap */ +#define SG_MITER_FROM_SG (1 << 2) /* nop */ struct sg_mapping_iter { /* the following three fields can be accessed directly */ diff --git a/lib/scatterlist.c b/lib/scatterlist.c index a295e404e908..0d475d8167bf 100644 --- a/lib/scatterlist.c +++ b/lib/scatterlist.c @@ -314,6 +314,7 @@ void sg_miter_start(struct sg_mapping_iter *miter, struct scatterlist *sgl, miter->__sg = sgl; miter->__nents = nents; miter->__offset = 0; + WARN_ON(!(flags & (SG_MITER_TO_SG | SG_MITER_FROM_SG))); miter->__flags = flags; } EXPORT_SYMBOL(sg_miter_start); @@ -394,6 +395,9 @@ void sg_miter_stop(struct sg_mapping_iter *miter) if (miter->addr) { miter->__offset += miter->consumed; + if (miter->__flags & SG_MITER_TO_SG) + flush_kernel_dcache_page(miter->page); + if (miter->__flags & SG_MITER_ATOMIC) { WARN_ON(!irqs_disabled()); kunmap_atomic(miter->addr, KM_BIO_SRC_IRQ); @@ -426,8 +430,14 @@ static size_t sg_copy_buffer(struct scatterlist *sgl, unsigned int nents, unsigned int offset = 0; struct sg_mapping_iter miter; unsigned long flags; + unsigned int sg_flags = SG_MITER_ATOMIC; + + if (to_buffer) + sg_flags |= SG_MITER_FROM_SG; + else + sg_flags |= SG_MITER_TO_SG; - sg_miter_start(&miter, sgl, nents, SG_MITER_ATOMIC); + sg_miter_start(&miter, sgl, nents, sg_flags); local_irq_save(flags); @@ -438,10 +448,8 @@ static size_t sg_copy_buffer(struct scatterlist *sgl, unsigned int nents, if (to_buffer) memcpy(buf + offset, miter.addr, len); - else { + else memcpy(miter.addr, buf + offset, len); - flush_kernel_dcache_page(miter.page); - } offset += len; } -- cgit v1.2.3 From 07868201070d87484bd00610a4921e879be78746 Mon Sep 17 00:00:00 2001 From: Jonathan Corbet Date: Tue, 4 Aug 2009 13:35:17 -0600 Subject: flex_array: remove unneeded index calculation flex_array_get() calculates an index value, then drops it on the floor; simply remove it. Signed-off-by: Jonathan Corbet Acked-by: Dave Hansen Signed-off-by: Linus Torvalds --- lib/flex_array.c | 2 -- 1 file changed, 2 deletions(-) (limited to 'lib') diff --git a/lib/flex_array.c b/lib/flex_array.c index 0e7894ce8882..08f1636d296a 100644 --- a/lib/flex_array.c +++ b/lib/flex_array.c @@ -254,7 +254,6 @@ void *flex_array_get(struct flex_array *fa, int element_nr) { int part_nr = fa_element_to_part_nr(fa, element_nr); struct flex_array_part *part; - int index; if (element_nr >= fa->total_nr_elements) return NULL; @@ -264,6 +263,5 @@ void *flex_array_get(struct flex_array *fa, int element_nr) part = (struct flex_array_part *)&fa->parts[0]; else part = fa->parts[part_nr]; - index = index_inside_part(fa, element_nr); return &part->elements[index_inside_part(fa, element_nr)]; } -- cgit v1.2.3