* - define CONFIG_HAVE_DYNAMIC_PER_CPU_AREA
*
* - define __addr_to_pcpu_ptr() and __pcpu_ptr_to_addr() to translate
- * regular address to percpu pointer and back
+ * regular address to percpu pointer and back if they need to be
+ * different from the default
*
* - use pcpu_setup_first_chunk() during percpu area initialization to
* setup the first chunk containing the kernel static percpu area
#include <linux/pfn.h>
#include <linux/rbtree.h>
#include <linux/slab.h>
+#include <linux/spinlock.h>
#include <linux/vmalloc.h>
+#include <linux/workqueue.h>
#include <asm/cacheflush.h>
+#include <asm/sections.h>
#include <asm/tlbflush.h>
#define PCPU_SLOT_BASE_SHIFT 5 /* 1-31 shares the same slot */
#define PCPU_DFL_MAP_ALLOC 16 /* start a map with 16 ents */
+/* default addr <-> pcpu_ptr mapping, override in asm/percpu.h if necessary */
+#ifndef __addr_to_pcpu_ptr
+#define __addr_to_pcpu_ptr(addr) \
+ (void *)((unsigned long)(addr) - (unsigned long)pcpu_base_addr \
+ + (unsigned long)__per_cpu_start)
+#endif
+#ifndef __pcpu_ptr_to_addr
+#define __pcpu_ptr_to_addr(ptr) \
+ (void *)((unsigned long)(ptr) + (unsigned long)pcpu_base_addr \
+ - (unsigned long)__per_cpu_start)
+#endif
+
struct pcpu_chunk {
struct list_head list; /* linked to pcpu_slot lists */
struct rb_node rb_node; /* key is chunk->vm->addr */
static int pcpu_reserved_chunk_limit;
/*
- * One mutex to rule them all.
- *
- * The following mutex is grabbed in the outermost public alloc/free
- * interface functions and released only when the operation is
- * complete. As such, every function in this file other than the
- * outermost functions are called under pcpu_mutex.
- *
- * It can easily be switched to use spinlock such that only the area
- * allocation and page population commit are protected with it doing
- * actual [de]allocation without holding any lock. However, given
- * what this allocator does, I think it's better to let them run
- * sequentially.
+ * Synchronization rules.
+ *
+ * There are two locks - pcpu_alloc_mutex and pcpu_lock. The former
+ * protects allocation/reclaim paths, chunks and chunk->page arrays.
+ * The latter is a spinlock and protects the index data structures -
+ * chunk slots, rbtree, chunks and area maps in chunks.
+ *
+ * During allocation, pcpu_alloc_mutex is kept locked all the time and
+ * pcpu_lock is grabbed and released as necessary. All actual memory
+ * allocations are done using GFP_KERNEL with pcpu_lock released.
+ *
+ * Free path accesses and alters only the index data structures, so it
+ * can be safely called from atomic context. When memory needs to be
+ * returned to the system, free path schedules reclaim_work which
+ * grabs both pcpu_alloc_mutex and pcpu_lock, unlinks chunks to be
+ * reclaimed, release both locks and frees the chunks. Note that it's
+ * necessary to grab both locks to remove a chunk from circulation as
+ * allocation path might be referencing the chunk with only
+ * pcpu_alloc_mutex locked.
*/
-static DEFINE_MUTEX(pcpu_mutex);
+static DEFINE_MUTEX(pcpu_alloc_mutex); /* protects whole alloc and reclaim */
+static DEFINE_SPINLOCK(pcpu_lock); /* protects index data structures */
static struct list_head *pcpu_slot __read_mostly; /* chunk list slots */
static struct rb_root pcpu_addr_root = RB_ROOT; /* chunks by address */
+/* reclaim work to release fully free chunks, scheduled from free path */
+static void pcpu_reclaim(struct work_struct *work);
+static DECLARE_WORK(pcpu_reclaim_work, pcpu_reclaim);
+
static int __pcpu_size_to_slot(int size)
{
int highbit = fls(size); /* size is in bytes */
* kzalloc() is used; otherwise, vmalloc() is used. The returned
* memory is always zeroed.
*
+ * CONTEXT:
+ * Does GFP_KERNEL allocation.
+ *
* RETURNS:
* Pointer to the allocated area on success, NULL on failure.
*/
* New slot according to the changed state is determined and @chunk is
* moved to the slot. Note that the reserved chunk is never put on
* chunk slots.
+ *
+ * CONTEXT:
+ * pcpu_lock.
*/
static void pcpu_chunk_relocate(struct pcpu_chunk *chunk, int oslot)
{
* searchs for the chunk with the highest start address which isn't
* beyond @addr.
*
+ * CONTEXT:
+ * pcpu_lock.
+ *
* RETURNS:
* The address of the found chunk.
*/
* @new: chunk to insert
*
* Insert @new into address rb tree.
+ *
+ * CONTEXT:
+ * pcpu_lock.
*/
static void pcpu_chunk_addr_insert(struct pcpu_chunk *new)
{
* A single allocation can split an area into three areas, so this
* function makes sure that @chunk->map has at least two extra slots.
*
+ * CONTEXT:
+ * pcpu_alloc_mutex, pcpu_lock. pcpu_lock is released and reacquired
+ * if area map is extended.
+ *
* RETURNS:
* 0 if noop, 1 if successfully extended, -errno on failure.
*/
if (chunk->map_alloc >= chunk->map_used + 2)
return 0;
+ spin_unlock_irq(&pcpu_lock);
+
new_alloc = PCPU_DFL_MAP_ALLOC;
while (new_alloc < chunk->map_used + 2)
new_alloc *= 2;
new = pcpu_mem_alloc(new_alloc * sizeof(new[0]));
- if (!new)
+ if (!new) {
+ spin_lock_irq(&pcpu_lock);
return -ENOMEM;
+ }
+
+ /*
+ * Acquire pcpu_lock and switch to new area map. Only free
+ * could have happened inbetween, so map_used couldn't have
+ * grown.
+ */
+ spin_lock_irq(&pcpu_lock);
+ BUG_ON(new_alloc < chunk->map_used + 2);
size = chunk->map_alloc * sizeof(chunk->map[0]);
memcpy(new, chunk->map, size);
* is inserted after the target block.
*
* @chunk->map must have enough free slots to accomodate the split.
+ *
+ * CONTEXT:
+ * pcpu_lock.
*/
static void pcpu_split_block(struct pcpu_chunk *chunk, int i,
int head, int tail)
*
* @chunk->map must have at least two free slots.
*
+ * CONTEXT:
+ * pcpu_lock.
+ *
* RETURNS:
* Allocated offset in @chunk on success, -1 if no matching area is
* found.
* Free area starting from @freeme to @chunk. Note that this function
* only modifies the allocation map. It doesn't depopulate or unmap
* the area.
+ *
+ * CONTEXT:
+ * pcpu_lock.
*/
static void pcpu_free_area(struct pcpu_chunk *chunk, int freeme)
{
* For each cpu, depopulate and unmap pages [@page_start,@page_end)
* from @chunk. If @flush is true, vcache is flushed before unmapping
* and tlb after.
+ *
+ * CONTEXT:
+ * pcpu_alloc_mutex.
*/
static void pcpu_depopulate_chunk(struct pcpu_chunk *chunk, int off, int size,
bool flush)
*
* For each cpu, populate and map pages [@page_start,@page_end) into
* @chunk. The area is cleared on return.
+ *
+ * CONTEXT:
+ * pcpu_alloc_mutex, does GFP_KERNEL allocation.
*/
static int pcpu_populate_chunk(struct pcpu_chunk *chunk, int off, int size)
{
* @align: alignment of area (max PAGE_SIZE)
* @reserved: allocate from the reserved chunk if available
*
- * Allocate percpu area of @size bytes aligned at @align. Might
- * sleep. Might trigger writeouts.
+ * Allocate percpu area of @size bytes aligned at @align.
+ *
+ * CONTEXT:
+ * Does GFP_KERNEL allocation.
*
* RETURNS:
* Percpu pointer to the allocated area on success, NULL on failure.
*/
static void *pcpu_alloc(size_t size, size_t align, bool reserved)
{
- void *ptr = NULL;
struct pcpu_chunk *chunk;
int slot, off;
return NULL;
}
- mutex_lock(&pcpu_mutex);
+ mutex_lock(&pcpu_alloc_mutex);
+ spin_lock_irq(&pcpu_lock);
/* serve reserved allocations from the reserved chunk if available */
if (reserved && pcpu_reserved_chunk) {
chunk = pcpu_reserved_chunk;
if (size > chunk->contig_hint ||
pcpu_extend_area_map(chunk) < 0)
- goto out_unlock;
+ goto fail_unlock;
off = pcpu_alloc_area(chunk, size, align);
if (off >= 0)
goto area_found;
- goto out_unlock;
+ goto fail_unlock;
}
+restart:
/* search through normal chunks */
for (slot = pcpu_size_to_slot(size); slot < pcpu_nr_slots; slot++) {
list_for_each_entry(chunk, &pcpu_slot[slot], list) {
if (size > chunk->contig_hint)
continue;
- if (pcpu_extend_area_map(chunk) < 0)
- goto out_unlock;
+
+ switch (pcpu_extend_area_map(chunk)) {
+ case 0:
+ break;
+ case 1:
+ goto restart; /* pcpu_lock dropped, restart */
+ default:
+ goto fail_unlock;
+ }
+
off = pcpu_alloc_area(chunk, size, align);
if (off >= 0)
goto area_found;
}
/* hmmm... no space left, create a new chunk */
+ spin_unlock_irq(&pcpu_lock);
+
chunk = alloc_pcpu_chunk();
if (!chunk)
- goto out_unlock;
+ goto fail_unlock_mutex;
+
+ spin_lock_irq(&pcpu_lock);
pcpu_chunk_relocate(chunk, -1);
pcpu_chunk_addr_insert(chunk);
-
- off = pcpu_alloc_area(chunk, size, align);
- if (off < 0)
- goto out_unlock;
+ goto restart;
area_found:
+ spin_unlock_irq(&pcpu_lock);
+
/* populate, map and clear the area */
if (pcpu_populate_chunk(chunk, off, size)) {
+ spin_lock_irq(&pcpu_lock);
pcpu_free_area(chunk, off);
- goto out_unlock;
+ goto fail_unlock;
}
- ptr = __addr_to_pcpu_ptr(chunk->vm->addr + off);
-out_unlock:
- mutex_unlock(&pcpu_mutex);
- return ptr;
+ mutex_unlock(&pcpu_alloc_mutex);
+
+ return __addr_to_pcpu_ptr(chunk->vm->addr + off);
+
+fail_unlock:
+ spin_unlock_irq(&pcpu_lock);
+fail_unlock_mutex:
+ mutex_unlock(&pcpu_alloc_mutex);
+ return NULL;
}
/**
* Allocate percpu area of @size bytes aligned at @align. Might
* sleep. Might trigger writeouts.
*
+ * CONTEXT:
+ * Does GFP_KERNEL allocation.
+ *
* RETURNS:
* Percpu pointer to the allocated area on success, NULL on failure.
*/
* percpu area if arch has set it up; otherwise, allocation is served
* from the same dynamic area. Might sleep. Might trigger writeouts.
*
+ * CONTEXT:
+ * Does GFP_KERNEL allocation.
+ *
* RETURNS:
* Percpu pointer to the allocated area on success, NULL on failure.
*/
return pcpu_alloc(size, align, true);
}
-static void pcpu_kill_chunk(struct pcpu_chunk *chunk)
+/**
+ * pcpu_reclaim - reclaim fully free chunks, workqueue function
+ * @work: unused
+ *
+ * Reclaim all fully free chunks except for the first one.
+ *
+ * CONTEXT:
+ * workqueue context.
+ */
+static void pcpu_reclaim(struct work_struct *work)
{
- WARN_ON(chunk->immutable);
- pcpu_depopulate_chunk(chunk, 0, pcpu_unit_size, false);
- list_del(&chunk->list);
- rb_erase(&chunk->rb_node, &pcpu_addr_root);
- free_pcpu_chunk(chunk);
+ LIST_HEAD(todo);
+ struct list_head *head = &pcpu_slot[pcpu_nr_slots - 1];
+ struct pcpu_chunk *chunk, *next;
+
+ mutex_lock(&pcpu_alloc_mutex);
+ spin_lock_irq(&pcpu_lock);
+
+ list_for_each_entry_safe(chunk, next, head, list) {
+ WARN_ON(chunk->immutable);
+
+ /* spare the first one */
+ if (chunk == list_first_entry(head, struct pcpu_chunk, list))
+ continue;
+
+ rb_erase(&chunk->rb_node, &pcpu_addr_root);
+ list_move(&chunk->list, &todo);
+ }
+
+ spin_unlock_irq(&pcpu_lock);
+ mutex_unlock(&pcpu_alloc_mutex);
+
+ list_for_each_entry_safe(chunk, next, &todo, list) {
+ pcpu_depopulate_chunk(chunk, 0, pcpu_unit_size, false);
+ free_pcpu_chunk(chunk);
+ }
}
/**
* free_percpu - free percpu area
* @ptr: pointer to area to free
*
- * Free percpu area @ptr. Might sleep.
+ * Free percpu area @ptr.
+ *
+ * CONTEXT:
+ * Can be called from atomic context.
*/
void free_percpu(void *ptr)
{
void *addr = __pcpu_ptr_to_addr(ptr);
struct pcpu_chunk *chunk;
+ unsigned long flags;
int off;
if (!ptr)
return;
- mutex_lock(&pcpu_mutex);
+ spin_lock_irqsave(&pcpu_lock, flags);
chunk = pcpu_chunk_addr_search(addr);
off = addr - chunk->vm->addr;
pcpu_free_area(chunk, off);
- /* the chunk became fully free, kill one if there are other free ones */
+ /* if there are more than one fully free chunks, wake up grim reaper */
if (chunk->free_size == pcpu_unit_size) {
struct pcpu_chunk *pos;
- list_for_each_entry(pos,
- &pcpu_slot[pcpu_chunk_slot(chunk)], list)
+ list_for_each_entry(pos, &pcpu_slot[pcpu_nr_slots - 1], list)
if (pos != chunk) {
- pcpu_kill_chunk(pos);
+ schedule_work(&pcpu_reclaim_work);
break;
}
}
- mutex_unlock(&pcpu_mutex);
+ spin_unlock_irqrestore(&pcpu_lock, flags);
}
EXPORT_SYMBOL_GPL(free_percpu);
* @get_page_fn: callback to fetch page pointer
* @static_size: the size of static percpu area in bytes
* @reserved_size: the size of reserved percpu area in bytes
- * @unit_size: unit size in bytes, must be multiple of PAGE_SIZE, -1 for auto
* @dyn_size: free size for dynamic allocation in bytes, -1 for auto
+ * @unit_size: unit size in bytes, must be multiple of PAGE_SIZE, -1 for auto
* @base_addr: mapped address, NULL for auto
* @populate_pte_fn: callback to allocate pagetable, NULL if unnecessary
*
* limited offset range for symbol relocations to guarantee module
* percpu symbols fall inside the relocatable range.
*
+ * @dyn_size, if non-negative, determines the number of bytes
+ * available for dynamic allocation in the first chunk. Specifying
+ * non-negative value makes percpu leave alone the area beyond
+ * @static_size + @reserved_size + @dyn_size.
+ *
* @unit_size, if non-negative, specifies unit size and must be
* aligned to PAGE_SIZE and equal to or larger than @static_size +
- * @reserved_size + @dyn_size.
- *
- * @dyn_size, if non-negative, limits the number of bytes available
- * for dynamic allocation in the first chunk. Specifying non-negative
- * value make percpu leave alone the area beyond @static_size +
- * @reserved_size + @dyn_size.
+ * @reserved_size + if non-negative, @dyn_size.
*
* Non-null @base_addr means that the caller already allocated virtual
* region for the first chunk and mapped it. percpu must not mess
*/
size_t __init pcpu_setup_first_chunk(pcpu_get_page_fn_t get_page_fn,
size_t static_size, size_t reserved_size,
- ssize_t unit_size, ssize_t dyn_size,
+ ssize_t dyn_size, ssize_t unit_size,
void *base_addr,
pcpu_populate_pte_fn_t populate_pte_fn)
{
static struct vm_struct first_vm;
static int smap[2], dmap[2];
+ size_t size_sum = static_size + reserved_size +
+ (dyn_size >= 0 ? dyn_size : 0);
struct pcpu_chunk *schunk, *dchunk = NULL;
unsigned int cpu;
int nr_pages;
ARRAY_SIZE(dmap) >= PCPU_DFL_MAP_ALLOC);
BUG_ON(!static_size);
if (unit_size >= 0) {
- BUG_ON(unit_size < static_size + reserved_size +
- (dyn_size >= 0 ? dyn_size : 0));
+ BUG_ON(unit_size < size_sum);
BUG_ON(unit_size & ~PAGE_MASK);
- } else {
- BUG_ON(dyn_size >= 0);
+ BUG_ON(unit_size < PCPU_MIN_UNIT_SIZE);
+ } else
BUG_ON(base_addr);
- }
BUG_ON(base_addr && populate_pte_fn);
if (unit_size >= 0)
pcpu_unit_pages = unit_size >> PAGE_SHIFT;
else
pcpu_unit_pages = max_t(int, PCPU_MIN_UNIT_SIZE >> PAGE_SHIFT,
- PFN_UP(static_size + reserved_size));
+ PFN_UP(size_sum));
pcpu_unit_size = pcpu_unit_pages << PAGE_SHIFT;
pcpu_chunk_size = num_possible_cpus() * pcpu_unit_size;
pcpu_base_addr = (void *)pcpu_chunk_addr(schunk, 0, 0);
return pcpu_unit_size;
}
+
+/*
+ * Embedding first chunk setup helper.
+ */
+static void *pcpue_ptr __initdata;
+static size_t pcpue_size __initdata;
+static size_t pcpue_unit_size __initdata;
+
+static struct page * __init pcpue_get_page(unsigned int cpu, int pageno)
+{
+ size_t off = (size_t)pageno << PAGE_SHIFT;
+
+ if (off >= pcpue_size)
+ return NULL;
+
+ return virt_to_page(pcpue_ptr + cpu * pcpue_unit_size + off);
+}
+
+/**
+ * pcpu_embed_first_chunk - embed the first percpu chunk into bootmem
+ * @static_size: the size of static percpu area in bytes
+ * @reserved_size: the size of reserved percpu area in bytes
+ * @dyn_size: free size for dynamic allocation in bytes, -1 for auto
+ * @unit_size: unit size in bytes, must be multiple of PAGE_SIZE, -1 for auto
+ *
+ * This is a helper to ease setting up embedded first percpu chunk and
+ * can be called where pcpu_setup_first_chunk() is expected.
+ *
+ * If this function is used to setup the first chunk, it is allocated
+ * as a contiguous area using bootmem allocator and used as-is without
+ * being mapped into vmalloc area. This enables the first chunk to
+ * piggy back on the linear physical mapping which often uses larger
+ * page size.
+ *
+ * When @dyn_size is positive, dynamic area might be larger than
+ * specified to fill page alignment. Also, when @dyn_size is auto,
+ * @dyn_size does not fill the whole first chunk but only what's
+ * necessary for page alignment after static and reserved areas.
+ *
+ * If the needed size is smaller than the minimum or specified unit
+ * size, the leftover is returned to the bootmem allocator.
+ *
+ * RETURNS:
+ * The determined pcpu_unit_size which can be used to initialize
+ * percpu access on success, -errno on failure.
+ */
+ssize_t __init pcpu_embed_first_chunk(size_t static_size, size_t reserved_size,
+ ssize_t dyn_size, ssize_t unit_size)
+{
+ unsigned int cpu;
+
+ /* determine parameters and allocate */
+ pcpue_size = PFN_ALIGN(static_size + reserved_size +
+ (dyn_size >= 0 ? dyn_size : 0));
+ if (dyn_size != 0)
+ dyn_size = pcpue_size - static_size - reserved_size;
+
+ if (unit_size >= 0) {
+ BUG_ON(unit_size < pcpue_size);
+ pcpue_unit_size = unit_size;
+ } else
+ pcpue_unit_size = max_t(size_t, pcpue_size, PCPU_MIN_UNIT_SIZE);
+
+ pcpue_ptr = __alloc_bootmem_nopanic(
+ num_possible_cpus() * pcpue_unit_size,
+ PAGE_SIZE, __pa(MAX_DMA_ADDRESS));
+ if (!pcpue_ptr)
+ return -ENOMEM;
+
+ /* return the leftover and copy */
+ for_each_possible_cpu(cpu) {
+ void *ptr = pcpue_ptr + cpu * pcpue_unit_size;
+
+ free_bootmem(__pa(ptr + pcpue_size),
+ pcpue_unit_size - pcpue_size);
+ memcpy(ptr, __per_cpu_load, static_size);
+ }
+
+ /* we're ready, commit */
+ pr_info("PERCPU: Embedded %zu pages at %p, static data %zu bytes\n",
+ pcpue_size >> PAGE_SHIFT, pcpue_ptr, static_size);
+
+ return pcpu_setup_first_chunk(pcpue_get_page, static_size,
+ reserved_size, dyn_size,
+ pcpue_unit_size, pcpue_ptr, NULL);
+}