#include "ptlrpc_internal.h"
-static int mult = 20 - PAGE_SHIFT;
+#define PPOOL_MIN_CHUNK_BITS 16 /* 2^16 bytes = 64KiB */
+#define PPOOL_MAX_CHUNK_BITS PTLRPC_MAX_BRW_BITS
+#define POOLS_COUNT (PPOOL_MAX_CHUNK_BITS - PPOOL_MIN_CHUNK_BITS + 1)
+#define PPOOL_SIZE_TO_INDEX(bits) ((bits) - PPOOL_MIN_CHUNK_BITS + 1)
+#define POOL_BITS(pool) ((pool) + PPOOL_MIN_CHUNK_BITS - 1)
+#define ELEMENT_SIZE(pool) (1 << (PPOOL_MIN_CHUNK_BITS + (pool) - 1))
+#define mult (20 - PAGE_SHIFT)
static int enc_pool_max_memory_mb;
module_param(enc_pool_max_memory_mb, int, 0644);
MODULE_PARM_DESC(enc_pool_max_memory_mb,
/*
* pointers to pools, may be vmalloc'd
*/
- struct page ***epp_pools;
-} page_pools;
+ void ***epp_pools;
+ /*
+ * memory shrinker
+ */
+ struct shrinker pool_shrinker;
+ struct mutex add_pages_mutex;
+} **page_pools;
/*
- * /proc/fs/lustre/sptlrpc/encrypt_page_pools
+ * /sys/kernel/debug/lustre/sptlrpc/encrypt_page_pools
*/
-int sptlrpc_proc_enc_pool_seq_show(struct seq_file *m, void *v)
+int encrypt_page_pools_seq_show(struct seq_file *m, void *v)
{
- spin_lock(&page_pools.epp_lock);
-
+ spin_lock(&page_pools[PAGES_POOL]->epp_lock);
seq_printf(m, "physical pages: %lu\n"
- "pages per pool: %lu\n"
- "max pages: %lu\n"
- "max pools: %u\n"
- "total pages: %lu\n"
- "total free: %lu\n"
- "idle index: %lu/100\n"
- "last shrink: %llds\n"
- "last access: %llds\n"
- "max pages reached: %lu\n"
- "grows: %u\n"
- "grows failure: %u\n"
- "shrinks: %u\n"
- "cache access: %lu\n"
- "cache missing: %lu\n"
- "low free mark: %lu\n"
- "max waitqueue depth: %u\n"
- "max wait time ms: %lld\n"
- "out of mem: %lu\n",
- cfs_totalram_pages(), PAGES_PER_POOL,
- page_pools.epp_max_pages,
- page_pools.epp_max_pools,
- page_pools.epp_total_pages,
- page_pools.epp_free_pages,
- page_pools.epp_idle_idx,
- ktime_get_seconds() - page_pools.epp_last_shrink,
- ktime_get_seconds() - page_pools.epp_last_access,
- page_pools.epp_st_max_pages,
- page_pools.epp_st_grows,
- page_pools.epp_st_grow_fails,
- page_pools.epp_st_shrinks,
- page_pools.epp_st_access,
- page_pools.epp_st_missings,
- page_pools.epp_st_lowfree,
- page_pools.epp_st_max_wqlen,
- ktime_to_ms(page_pools.epp_st_max_wait),
- page_pools.epp_st_outofmem);
-
- spin_unlock(&page_pools.epp_lock);
+ "pages per pool: %lu\n"
+ "max pages: %lu\n"
+ "max pools: %u\n"
+ "total pages: %lu\n"
+ "total free: %lu\n"
+ "idle index: %lu/100\n"
+ "last shrink: %llds\n"
+ "last access: %llds\n"
+ "max pages reached: %lu\n"
+ "grows: %u\n"
+ "grows failure: %u\n"
+ "shrinks: %u\n"
+ "cache access: %lu\n"
+ "cache missing: %lu\n"
+ "low free mark: %lu\n"
+ "max waitqueue depth: %u\n"
+ "max wait time ms: %lld\n"
+ "out of mem: %lu\n",
+ cfs_totalram_pages(), PAGES_PER_POOL,
+ page_pools[PAGES_POOL]->epp_max_pages,
+ page_pools[PAGES_POOL]->epp_max_pools,
+ page_pools[PAGES_POOL]->epp_total_pages,
+ page_pools[PAGES_POOL]->epp_free_pages,
+ page_pools[PAGES_POOL]->epp_idle_idx,
+ ktime_get_seconds() - page_pools[PAGES_POOL]->epp_last_shrink,
+ ktime_get_seconds() - page_pools[PAGES_POOL]->epp_last_access,
+ page_pools[PAGES_POOL]->epp_st_max_pages,
+ page_pools[PAGES_POOL]->epp_st_grows,
+ page_pools[PAGES_POOL]->epp_st_grow_fails,
+ page_pools[PAGES_POOL]->epp_st_shrinks,
+ page_pools[PAGES_POOL]->epp_st_access,
+ page_pools[PAGES_POOL]->epp_st_missings,
+ page_pools[PAGES_POOL]->epp_st_lowfree,
+ page_pools[PAGES_POOL]->epp_st_max_wqlen,
+ ktime_to_ms(page_pools[PAGES_POOL]->epp_st_max_wait),
+ page_pools[PAGES_POOL]->epp_st_outofmem);
+ spin_unlock(&page_pools[PAGES_POOL]->epp_lock);
+
+ return 0;
+}
+
+/*
+ * /sys/kernel/debug/lustre/sptlrpc/page_pools
+ */
+int page_pools_seq_show(struct seq_file *m, void *v)
+{
+ int pool_index;
+ struct ptlrpc_enc_page_pool *pool;
+
+ seq_printf(m, "physical_pages: %lu\n"
+ "pages per pool: %lu\n\n"
+ "pools:\n",
+ cfs_totalram_pages(), PAGES_PER_POOL);
+
+ for (pool_index = 0; pool_index < POOLS_COUNT; pool_index++) {
+ pool = page_pools[pool_index];
+ if (!pool->epp_st_access)
+ continue;
+ spin_lock(&pool->epp_lock);
+ seq_printf(m, " pool_%luk:\n"
+ " max_pages: %lu\n"
+ " max_pools: %u\n"
+ " total_pages: %lu\n"
+ " total_free: %lu\n"
+ " idle_index: %lu/100\n"
+ " last_shrink: %llds\n"
+ " last_access: %llds\n"
+ " max_pages_reached: %lu\n"
+ " grows: %u\n"
+ " grows_failure: %u\n"
+ " shrinks: %u\n"
+ " cache_access: %lu\n"
+ " cache_missing: %lu\n"
+ " low_free_mark: %lu\n"
+ " max_waitqueue_depth: %u\n"
+ " max_wait_time_ms: %lld\n"
+ " out_of_mem: %lu\n",
+ (pool_index ? ELEMENT_SIZE(pool_index - 10) :
+ PAGE_SIZE >> 10),
+ pool->epp_max_pages,
+ pool->epp_max_pools,
+ pool->epp_total_pages,
+ pool->epp_free_pages,
+ pool->epp_idle_idx,
+ ktime_get_seconds() - pool->epp_last_shrink,
+ ktime_get_seconds() - pool->epp_last_access,
+ pool->epp_st_max_pages,
+ pool->epp_st_grows,
+ pool->epp_st_grow_fails,
+ pool->epp_st_shrinks,
+ pool->epp_st_access,
+ pool->epp_st_missings,
+ pool->epp_st_lowfree,
+ pool->epp_st_max_wqlen,
+ ktime_to_ms(pool->epp_st_max_wait),
+ pool->epp_st_outofmem);
+
+ spin_unlock(&pool->epp_lock);
+ }
return 0;
}
-static void enc_pools_release_free_pages(long npages)
+static void enc_pools_release_free_pages(long npages, unsigned int pool_idx)
{
int p_idx, g_idx;
int p_idx_max1, p_idx_max2;
+ struct ptlrpc_enc_page_pool *pool = page_pools[pool_idx];
LASSERT(npages > 0);
- LASSERT(npages <= page_pools.epp_free_pages);
- LASSERT(page_pools.epp_free_pages <= page_pools.epp_total_pages);
+ LASSERT(npages <= pool->epp_free_pages);
+ LASSERT(pool->epp_free_pages <= pool->epp_total_pages);
/* max pool index before the release */
- p_idx_max2 = (page_pools.epp_total_pages - 1) / PAGES_PER_POOL;
+ p_idx_max2 = (pool->epp_total_pages - 1) / PAGES_PER_POOL;
- page_pools.epp_free_pages -= npages;
- page_pools.epp_total_pages -= npages;
+ pool->epp_free_pages -= npages;
+ pool->epp_total_pages -= npages;
/* max pool index after the release */
- p_idx_max1 = page_pools.epp_total_pages == 0 ? -1 :
- ((page_pools.epp_total_pages - 1) / PAGES_PER_POOL);
+ p_idx_max1 = pool->epp_total_pages == 0 ? -1 :
+ ((pool->epp_total_pages - 1) / PAGES_PER_POOL);
- p_idx = page_pools.epp_free_pages / PAGES_PER_POOL;
- g_idx = page_pools.epp_free_pages % PAGES_PER_POOL;
- LASSERT(page_pools.epp_pools[p_idx]);
+ p_idx = pool->epp_free_pages / PAGES_PER_POOL;
+ g_idx = pool->epp_free_pages % PAGES_PER_POOL;
+ LASSERT(pool->epp_pools[p_idx]);
while (npages--) {
- LASSERT(page_pools.epp_pools[p_idx]);
- LASSERT(page_pools.epp_pools[p_idx][g_idx] != NULL);
+ LASSERT(pool->epp_pools[p_idx]);
+ LASSERT(pool->epp_pools[p_idx][g_idx] != NULL);
- __free_page(page_pools.epp_pools[p_idx][g_idx]);
- page_pools.epp_pools[p_idx][g_idx] = NULL;
+ if (pool_idx == 0)
+ __free_page(pool->epp_pools[p_idx][g_idx]);
+ else
+ OBD_FREE_LARGE(pool->epp_pools[p_idx][g_idx],
+ ELEMENT_SIZE(pool_idx));
+ pool->epp_pools[p_idx][g_idx] = NULL;
if (++g_idx == PAGES_PER_POOL) {
p_idx++;
/* free unused pools */
while (p_idx_max1 < p_idx_max2) {
- LASSERT(page_pools.epp_pools[p_idx_max2]);
- OBD_FREE(page_pools.epp_pools[p_idx_max2], PAGE_SIZE);
- page_pools.epp_pools[p_idx_max2] = NULL;
+ LASSERT(pool->epp_pools[p_idx_max2]);
+ OBD_FREE(pool->epp_pools[p_idx_max2], PAGE_SIZE);
+ pool->epp_pools[p_idx_max2] = NULL;
p_idx_max2--;
}
}
+#define SEEKS_TO_INDEX(s) (((s)->seeks >> 8) & 0xff)
+#define INDEX_TO_SEEKS(i) (DEFAULT_SEEKS | (i << 8))
/*
* we try to keep at least PTLRPC_MAX_BRW_PAGES pages in the pool.
*/
static unsigned long enc_pools_shrink_count(struct shrinker *s,
struct shrink_control *sc)
{
+ unsigned int pool_index = SEEKS_TO_INDEX(s);
+ struct ptlrpc_enc_page_pool *pool = page_pools[pool_index];
/*
- * if no pool access for a long time, we consider it's fully idle.
- * a little race here is fine.
+ * if no pool access for a long time, we consider it's fully
+ * idle. A little race here is fine.
*/
- if (unlikely(ktime_get_seconds() - page_pools.epp_last_access >
+ if (unlikely(ktime_get_seconds() - pool->epp_last_access >
CACHE_QUIESCENT_PERIOD)) {
- spin_lock(&page_pools.epp_lock);
- page_pools.epp_idle_idx = IDLE_IDX_MAX;
- spin_unlock(&page_pools.epp_lock);
+ spin_lock(&pool->epp_lock);
+ pool->epp_idle_idx = IDLE_IDX_MAX;
+ spin_unlock(&pool->epp_lock);
}
- LASSERT(page_pools.epp_idle_idx <= IDLE_IDX_MAX);
- return (page_pools.epp_free_pages <= PTLRPC_MAX_BRW_PAGES) ? 0 :
- (page_pools.epp_free_pages - PTLRPC_MAX_BRW_PAGES) *
- (IDLE_IDX_MAX - page_pools.epp_idle_idx) / IDLE_IDX_MAX;
+ LASSERT(pool->epp_idle_idx <= IDLE_IDX_MAX);
+
+ return (pool->epp_free_pages <= PTLRPC_MAX_BRW_PAGES) ? 0 :
+ (pool->epp_free_pages - PTLRPC_MAX_BRW_PAGES) *
+ (IDLE_IDX_MAX - pool->epp_idle_idx) / IDLE_IDX_MAX;
}
/*
static unsigned long enc_pools_shrink_scan(struct shrinker *s,
struct shrink_control *sc)
{
- spin_lock(&page_pools.epp_lock);
- if (page_pools.epp_free_pages <= PTLRPC_MAX_BRW_PAGES)
+ /* Get pool number passed as part of pools_shrinker_seeks value */
+ unsigned int pool_index = SEEKS_TO_INDEX(s);
+ struct ptlrpc_enc_page_pool *pool = page_pools[pool_index];
+
+ spin_lock(&pool->epp_lock);
+ if (pool->epp_free_pages <= PTLRPC_MAX_BRW_PAGES)
sc->nr_to_scan = 0;
else
sc->nr_to_scan = min_t(unsigned long, sc->nr_to_scan,
- page_pools.epp_free_pages - PTLRPC_MAX_BRW_PAGES);
+ pool->epp_free_pages - PTLRPC_MAX_BRW_PAGES);
if (sc->nr_to_scan > 0) {
- enc_pools_release_free_pages(sc->nr_to_scan);
+ enc_pools_release_free_pages(sc->nr_to_scan, pool_index);
CDEBUG(D_SEC, "released %ld pages, %ld left\n",
- (long)sc->nr_to_scan, page_pools.epp_free_pages);
+ (long)sc->nr_to_scan, pool->epp_free_pages);
- page_pools.epp_st_shrinks++;
- page_pools.epp_last_shrink = ktime_get_seconds();
+ pool->epp_st_shrinks++;
+ pool->epp_last_shrink = ktime_get_seconds();
}
- spin_unlock(&page_pools.epp_lock);
+ spin_unlock(&pool->epp_lock);
/*
* if no pool access for a long time, we consider it's fully idle.
* a little race here is fine.
*/
- if (unlikely(ktime_get_seconds() - page_pools.epp_last_access >
+ if (unlikely(ktime_get_seconds() - pool->epp_last_access >
CACHE_QUIESCENT_PERIOD)) {
- spin_lock(&page_pools.epp_lock);
- page_pools.epp_idle_idx = IDLE_IDX_MAX;
- spin_unlock(&page_pools.epp_lock);
+ spin_lock(&pool->epp_lock);
+ pool->epp_idle_idx = IDLE_IDX_MAX;
+ spin_unlock(&pool->epp_lock);
}
- LASSERT(page_pools.epp_idle_idx <= IDLE_IDX_MAX);
+ LASSERT(pool->epp_idle_idx <= IDLE_IDX_MAX);
+
return sc->nr_to_scan;
}
-#ifdef HAVE_SHRINKER_COUNT
-static struct shrinker pools_shrinker = {
- .count_objects = enc_pools_shrink_count,
- .scan_objects = enc_pools_shrink_scan,
- .seeks = DEFAULT_SEEKS,
-};
-#else
+#ifndef HAVE_SHRINKER_COUNT
/*
* could be called frequently for query (@nr_to_scan == 0).
* we try to keep at least PTLRPC_MAX_BRW_PAGES pages in the pool.
return enc_pools_shrink_count(shrinker, sc);
}
-
-static struct shrinker pools_shrinker = {
- .shrink = enc_pools_shrink,
- .seeks = DEFAULT_SEEKS,
-};
#endif /* HAVE_SHRINKER_COUNT */
static inline
/*
* return how many pages cleaned up.
*/
-static unsigned long enc_pools_cleanup(struct page ***pools, int npools)
+static unsigned long enc_pools_cleanup(void ***pools, int npools, int pool_idx)
{
unsigned long cleaned = 0;
int i, j;
if (pools[i]) {
for (j = 0; j < PAGES_PER_POOL; j++) {
if (pools[i][j]) {
- __free_page(pools[i][j]);
+ if (pool_idx == 0) {
+ __free_page(pools[i][j]);
+ } else {
+ OBD_FREE_LARGE(pools[i][j],
+ ELEMENT_SIZE(pool_idx));
+ }
cleaned++;
}
}
* we have options to avoid most memory copy with some tricks. but we choose
* the simplest way to avoid complexity. It's not frequently called.
*/
-static void enc_pools_insert(struct page ***pools, int npools, int npages)
+static void enc_pools_insert(void ***pools, int npools, int npages,
+ unsigned int pool_idx)
{
int freeslot;
int op_idx, np_idx, og_idx, ng_idx;
int cur_npools, end_npools;
+ struct ptlrpc_enc_page_pool *page_pool = page_pools[pool_idx];
LASSERT(npages > 0);
- LASSERT(page_pools.epp_total_pages+npages <= page_pools.epp_max_pages);
+ LASSERT(page_pool->epp_total_pages+npages <= page_pool->epp_max_pages);
LASSERT(npages_to_npools(npages) == npools);
- LASSERT(page_pools.epp_growing);
+ LASSERT(page_pool->epp_growing);
- spin_lock(&page_pools.epp_lock);
+ spin_lock(&page_pool->epp_lock);
/*
* (1) fill all the free slots of current pools.
* free slots are those left by rent pages, and the extra ones with
* index >= total_pages, locate at the tail of last pool.
*/
- freeslot = page_pools.epp_total_pages % PAGES_PER_POOL;
+ freeslot = page_pool->epp_total_pages % PAGES_PER_POOL;
if (freeslot != 0)
freeslot = PAGES_PER_POOL - freeslot;
- freeslot += page_pools.epp_total_pages - page_pools.epp_free_pages;
+ freeslot += page_pool->epp_total_pages - page_pool->epp_free_pages;
- op_idx = page_pools.epp_free_pages / PAGES_PER_POOL;
- og_idx = page_pools.epp_free_pages % PAGES_PER_POOL;
+ op_idx = page_pool->epp_free_pages / PAGES_PER_POOL;
+ og_idx = page_pool->epp_free_pages % PAGES_PER_POOL;
np_idx = npools - 1;
ng_idx = (npages - 1) % PAGES_PER_POOL;
while (freeslot) {
- LASSERT(page_pools.epp_pools[op_idx][og_idx] == NULL);
+ LASSERT(page_pool->epp_pools[op_idx][og_idx] == NULL);
LASSERT(pools[np_idx][ng_idx] != NULL);
- page_pools.epp_pools[op_idx][og_idx] = pools[np_idx][ng_idx];
+ page_pool->epp_pools[op_idx][og_idx] = pools[np_idx][ng_idx];
pools[np_idx][ng_idx] = NULL;
freeslot--;
/*
* (2) add pools if needed.
*/
- cur_npools = (page_pools.epp_total_pages + PAGES_PER_POOL - 1) /
+ cur_npools = (page_pool->epp_total_pages + PAGES_PER_POOL - 1) /
PAGES_PER_POOL;
- end_npools = (page_pools.epp_total_pages + npages +
+ end_npools = (page_pool->epp_total_pages + npages +
PAGES_PER_POOL - 1) / PAGES_PER_POOL;
- LASSERT(end_npools <= page_pools.epp_max_pools);
+ LASSERT(end_npools <= page_pool->epp_max_pools);
np_idx = 0;
while (cur_npools < end_npools) {
- LASSERT(page_pools.epp_pools[cur_npools] == NULL);
+ LASSERT(page_pool->epp_pools[cur_npools] == NULL);
LASSERT(np_idx < npools);
LASSERT(pools[np_idx] != NULL);
- page_pools.epp_pools[cur_npools++] = pools[np_idx];
+ page_pool->epp_pools[cur_npools++] = pools[np_idx];
pools[np_idx++] = NULL;
}
- page_pools.epp_total_pages += npages;
- page_pools.epp_free_pages += npages;
- page_pools.epp_st_lowfree = page_pools.epp_free_pages;
+ /*
+ * (3) free useless source pools
+ */
+ while (np_idx < npools) {
+ LASSERT(pools[np_idx] != NULL);
+ CDEBUG(D_SEC, "Free useless pool buffer: %i, %p\n", np_idx,
+ pools[np_idx]);
+ OBD_FREE(pools[np_idx], PAGE_SIZE);
+ pools[np_idx++] = NULL;
+ }
- if (page_pools.epp_total_pages > page_pools.epp_st_max_pages)
- page_pools.epp_st_max_pages = page_pools.epp_total_pages;
+ page_pool->epp_total_pages += npages;
+ page_pool->epp_free_pages += npages;
+ page_pool->epp_st_lowfree = page_pool->epp_free_pages;
+
+ if (page_pool->epp_total_pages > page_pool->epp_st_max_pages)
+ page_pool->epp_st_max_pages = page_pool->epp_total_pages;
CDEBUG(D_SEC, "add %d pages to total %lu\n", npages,
- page_pools.epp_total_pages);
+ page_pool->epp_total_pages);
- spin_unlock(&page_pools.epp_lock);
+ spin_unlock(&page_pool->epp_lock);
}
-static int enc_pools_add_pages(int npages)
+#define POOL_INIT_SIZE (PTLRPC_MAX_BRW_SIZE / 4)
+static int enc_pools_add_pages(int npages, int pool_index)
{
- static DEFINE_MUTEX(add_pages_mutex);
- struct page ***pools;
+ void ***pools;
int npools, alloced = 0;
int i, j, rc = -ENOMEM;
+ struct ptlrpc_enc_page_pool *page_pool = page_pools[pool_index];
- if (npages < PTLRPC_MAX_BRW_PAGES)
- npages = PTLRPC_MAX_BRW_PAGES;
+ if (pool_index == 0) {
+ if (npages < POOL_INIT_SIZE >> PAGE_SHIFT)
+ npages = POOL_INIT_SIZE >> PAGE_SHIFT;
+ } else {
+ if (npages < POOL_INIT_SIZE / ELEMENT_SIZE(pool_index))
+ npages = POOL_INIT_SIZE / ELEMENT_SIZE(pool_index);
+ }
- mutex_lock(&add_pages_mutex);
+ mutex_lock(&page_pool->add_pages_mutex);
- if (npages + page_pools.epp_total_pages > page_pools.epp_max_pages)
- npages = page_pools.epp_max_pages - page_pools.epp_total_pages;
+ if (npages + page_pool->epp_total_pages > page_pool->epp_max_pages)
+ npages = page_pool->epp_max_pages - page_pool->epp_total_pages;
LASSERT(npages > 0);
- page_pools.epp_st_grows++;
+ page_pool->epp_st_grows++;
npools = npages_to_npools(npages);
OBD_ALLOC_PTR_ARRAY(pools, npools);
goto out_pools;
for (j = 0; j < PAGES_PER_POOL && alloced < npages; j++) {
- pools[i][j] = alloc_page(GFP_NOFS |
- __GFP_HIGHMEM);
+ if (pool_index == 0)
+ pools[i][j] = alloc_page(GFP_NOFS |
+ __GFP_HIGHMEM);
+ else {
+ OBD_ALLOC_LARGE(pools[i][j],
+ ELEMENT_SIZE(pool_index));
+ }
if (pools[i][j] == NULL)
goto out_pools;
}
LASSERT(alloced == npages);
- enc_pools_insert(pools, npools, npages);
+ enc_pools_insert(pools, npools, npages, pool_index);
CDEBUG(D_SEC, "added %d pages into pools\n", npages);
+ OBD_FREE_PTR_ARRAY(pools, npools);
rc = 0;
out_pools:
- enc_pools_cleanup(pools, npools);
- OBD_FREE_PTR_ARRAY(pools, npools);
+ if (rc) {
+ enc_pools_cleanup(pools, npools, pool_index);
+ }
out:
if (rc) {
- page_pools.epp_st_grow_fails++;
+ page_pool->epp_st_grow_fails++;
CERROR("Failed to allocate %d enc pages\n", npages);
}
- mutex_unlock(&add_pages_mutex);
+ mutex_unlock(&page_pool->add_pages_mutex);
return rc;
}
-static inline void enc_pools_wakeup(void)
+static inline void enc_pools_wakeup(unsigned int pool)
{
- assert_spin_locked(&page_pools.epp_lock);
+ assert_spin_locked(&page_pools[pool]->epp_lock);
/* waitqueue_active */
- if (unlikely(waitqueue_active(&page_pools.epp_waitq)))
- wake_up(&page_pools.epp_waitq);
+ if (unlikely(waitqueue_active(&page_pools[pool]->epp_waitq)))
+ wake_up_all(&page_pools[pool]->epp_waitq);
}
-static int enc_pools_should_grow(int page_needed, time64_t now)
+static int enc_pools_should_grow(int page_needed, time64_t now,
+ unsigned int pool_index)
{
/*
* don't grow if someone else is growing the pools right now,
* or the pools has reached its full capacity
*/
- if (page_pools.epp_growing ||
- page_pools.epp_total_pages == page_pools.epp_max_pages)
+ if (page_pools[pool_index]->epp_growing ||
+ page_pools[pool_index]->epp_total_pages ==
+ page_pools[pool_index]->epp_max_pages)
return 0;
/* if total pages is not enough, we need to grow */
- if (page_pools.epp_total_pages < page_needed)
+ if (page_pools[pool_index]->epp_total_pages < page_needed)
return 1;
-
/*
* we wanted to return 0 here if there was a shrink just
* happened a moment ago, but this may cause deadlock if both
/*
* Export the number of free pages in the pool
*/
-int get_free_pages_in_pool(void)
+int sptlrpc_enc_pool_get_free_pages(unsigned int pool)
+{
+ return page_pools[pool]->epp_free_pages;
+}
+EXPORT_SYMBOL(sptlrpc_enc_pool_get_free_pages);
+
+/*
+ * Let outside world know if enc_pool full capacity is reached
+ */
+int __pool_is_at_full_capacity(unsigned int pool)
{
- return page_pools.epp_free_pages;
+ return (page_pools[pool]->epp_total_pages ==
+ page_pools[pool]->epp_max_pages);
}
-EXPORT_SYMBOL(get_free_pages_in_pool);
/*
* Let outside world know if enc_pool full capacity is reached
*/
int pool_is_at_full_capacity(void)
{
- return (page_pools.epp_total_pages == page_pools.epp_max_pages);
+ return __pool_is_at_full_capacity(PAGES_POOL);
}
EXPORT_SYMBOL(pool_is_at_full_capacity);
-static inline struct page **page_from_bulkdesc(void *array, int index)
+static inline void **page_from_bulkdesc(void *array, int index)
{
struct ptlrpc_bulk_desc *desc = (struct ptlrpc_bulk_desc *)array;
- return &desc->bd_enc_vec[index].bv_page;
+ return (void **)&desc->bd_enc_vec[index].bv_page;
}
-static inline struct page **page_from_pagearray(void *array, int index)
+static inline void **page_from_pagearray(void *array, int index)
{
struct page **pa = (struct page **)array;
- return &pa[index];
+ return (void **)&pa[index];
+}
+
+static inline void **page_from_bufarray(void *array, int index)
+{
+ return (void **)array;
}
/*
* we allocate the requested pages atomically.
*/
static inline int __sptlrpc_enc_pool_get_pages(void *array, unsigned int count,
- struct page **(*page_from)(void *, int))
+ unsigned int pool,
+ void **(*page_from)(void *, int))
{
+ struct ptlrpc_enc_page_pool *page_pool = page_pools[pool];
wait_queue_entry_t waitlink;
unsigned long this_idle = -1;
u64 tick_ns = 0;
int p_idx, g_idx;
int i, rc = 0;
- if (!array || count <= 0 || count > page_pools.epp_max_pages)
+ if (pool)
+ count = 1;
+
+ if (!array || count <= 0 || count > page_pool->epp_max_pages)
return -EINVAL;
- spin_lock(&page_pools.epp_lock);
+ spin_lock(&page_pool->epp_lock);
- page_pools.epp_st_access++;
+ page_pool->epp_st_access++;
again:
- if (unlikely(page_pools.epp_free_pages < count)) {
+ if (unlikely(page_pool->epp_free_pages < count)) {
if (tick_ns == 0)
tick_ns = ktime_get_ns();
now = ktime_get_real_seconds();
- page_pools.epp_st_missings++;
- page_pools.epp_pages_short += count;
+ page_pool->epp_st_missings++;
+ page_pool->epp_pages_short += count;
- if (enc_pools_should_grow(count, now)) {
- page_pools.epp_growing = 1;
+ if (enc_pools_should_grow(count, now, pool)) {
+ page_pool->epp_growing = 1;
- spin_unlock(&page_pools.epp_lock);
- enc_pools_add_pages(page_pools.epp_pages_short / 2);
- spin_lock(&page_pools.epp_lock);
+ spin_unlock(&page_pool->epp_lock);
+ CDEBUG(D_SEC, "epp_pages_short: %lu\n", page_pool->epp_pages_short);
+ enc_pools_add_pages(8, pool);
+ spin_lock(&page_pool->epp_lock);
- page_pools.epp_growing = 0;
+ page_pool->epp_growing = 0;
- enc_pools_wakeup();
+ enc_pools_wakeup(pool);
} else {
- if (page_pools.epp_growing) {
- if (++page_pools.epp_waitqlen >
- page_pools.epp_st_max_wqlen)
- page_pools.epp_st_max_wqlen =
- page_pools.epp_waitqlen;
+ if (page_pool->epp_growing) {
+ if (++page_pool->epp_waitqlen >
+ page_pool->epp_st_max_wqlen)
+ page_pool->epp_st_max_wqlen =
+ page_pool->epp_waitqlen;
set_current_state(TASK_UNINTERRUPTIBLE);
init_wait(&waitlink);
- add_wait_queue(&page_pools.epp_waitq,
+ add_wait_queue(&page_pool->epp_waitq,
&waitlink);
- spin_unlock(&page_pools.epp_lock);
+ spin_unlock(&page_pool->epp_lock);
schedule();
- remove_wait_queue(&page_pools.epp_waitq,
+ remove_wait_queue(&page_pool->epp_waitq,
&waitlink);
- spin_lock(&page_pools.epp_lock);
- page_pools.epp_waitqlen--;
+ spin_lock(&page_pool->epp_lock);
+ page_pool->epp_waitqlen--;
} else {
/*
* ptlrpcd thread should not sleep in that case,
* Instead, return -ENOMEM so that upper layers
* will put request back in queue.
*/
- page_pools.epp_st_outofmem++;
+ page_pool->epp_st_outofmem++;
GOTO(out_unlock, rc = -ENOMEM);
}
}
- if (page_pools.epp_pages_short < count)
+ if (page_pool->epp_pages_short < count)
GOTO(out_unlock, rc = -EPROTO);
- page_pools.epp_pages_short -= count;
+ page_pool->epp_pages_short -= count;
this_idle = 0;
goto again;
if (unlikely(tick_ns)) {
ktime_t tick = ktime_sub_ns(ktime_get(), tick_ns);
- if (ktime_after(tick, page_pools.epp_st_max_wait))
- page_pools.epp_st_max_wait = tick;
+ if (ktime_after(tick, page_pool->epp_st_max_wait))
+ page_pool->epp_st_max_wait = tick;
}
/* proceed with rest of allocation */
- page_pools.epp_free_pages -= count;
+ page_pool->epp_free_pages -= count;
- p_idx = page_pools.epp_free_pages / PAGES_PER_POOL;
- g_idx = page_pools.epp_free_pages % PAGES_PER_POOL;
+ p_idx = page_pool->epp_free_pages / PAGES_PER_POOL;
+ g_idx = page_pool->epp_free_pages % PAGES_PER_POOL;
for (i = 0; i < count; i++) {
- struct page **pagep = page_from(array, i);
+ void **pagep = page_from(array, i);
- if (page_pools.epp_pools[p_idx][g_idx] == NULL)
+ if (page_pool->epp_pools[p_idx][g_idx] == NULL)
GOTO(out_unlock, rc = -EPROTO);
- *pagep = page_pools.epp_pools[p_idx][g_idx];
- page_pools.epp_pools[p_idx][g_idx] = NULL;
+ *pagep = page_pool->epp_pools[p_idx][g_idx];
+ page_pool->epp_pools[p_idx][g_idx] = NULL;
if (++g_idx == PAGES_PER_POOL) {
p_idx++;
}
}
- if (page_pools.epp_free_pages < page_pools.epp_st_lowfree)
- page_pools.epp_st_lowfree = page_pools.epp_free_pages;
+ if (page_pool->epp_free_pages < page_pool->epp_st_lowfree)
+ page_pool->epp_st_lowfree =
+ page_pool->epp_free_pages;
/*
* new idle index = (old * weight + new) / (weight + 1)
*/
if (this_idle == -1) {
- this_idle = page_pools.epp_free_pages * IDLE_IDX_MAX /
- page_pools.epp_total_pages;
+ this_idle = page_pool->epp_free_pages * IDLE_IDX_MAX /
+ page_pool->epp_total_pages;
}
- page_pools.epp_idle_idx = (page_pools.epp_idle_idx * IDLE_IDX_WEIGHT +
- this_idle) /
- (IDLE_IDX_WEIGHT + 1);
+ page_pool->epp_idle_idx = (page_pool->epp_idle_idx *
+ IDLE_IDX_WEIGHT + this_idle) /
+ (IDLE_IDX_WEIGHT + 1);
- page_pools.epp_last_access = ktime_get_seconds();
+ page_pool->epp_last_access = ktime_get_seconds();
out_unlock:
- spin_unlock(&page_pools.epp_lock);
+ spin_unlock(&page_pool->epp_lock);
return rc;
}
int rc;
LASSERT(desc->bd_iov_count > 0);
- LASSERT(desc->bd_iov_count <= page_pools.epp_max_pages);
+ LASSERT(desc->bd_iov_count <= page_pools[PAGES_POOL]->epp_max_pages);
/* resent bulk, enc iov might have been allocated previously */
if (desc->bd_enc_vec != NULL)
return -ENOMEM;
rc = __sptlrpc_enc_pool_get_pages((void *)desc, desc->bd_iov_count,
- page_from_bulkdesc);
+ PAGES_POOL, page_from_bulkdesc);
if (rc) {
OBD_FREE_LARGE(desc->bd_enc_vec,
desc->bd_iov_count *
int sptlrpc_enc_pool_get_pages_array(struct page **pa, unsigned int count)
{
- return __sptlrpc_enc_pool_get_pages((void *)pa, count,
+ return __sptlrpc_enc_pool_get_pages((void *)pa, count, PAGES_POOL,
page_from_pagearray);
}
EXPORT_SYMBOL(sptlrpc_enc_pool_get_pages_array);
+int sptlrpc_enc_pool_get_buf(void **buf, unsigned int size_bits)
+{
+ return __sptlrpc_enc_pool_get_pages((void *)buf, 0,
+ PPOOL_SIZE_TO_INDEX(size_bits),
+ page_from_bufarray);
+}
+EXPORT_SYMBOL(sptlrpc_enc_pool_get_buf);
+
static int __sptlrpc_enc_pool_put_pages(void *array, unsigned int count,
- struct page **(*page_from)(void *, int))
+ unsigned int pool,
+ void **(*page_from)(void *, int))
{
int p_idx, g_idx;
int i, rc = 0;
+ struct ptlrpc_enc_page_pool *page_pool;
- if (!array || count <= 0)
+ LASSERTF(pool < POOLS_COUNT, "count %u, pool %u\n", count, pool);
+ if (!array || pool >= POOLS_COUNT) {
+ CERROR("Faled to put %u pages, from pull %u\n", count, pool);
return -EINVAL;
+ }
- spin_lock(&page_pools.epp_lock);
+ page_pool = page_pools[pool];
+ LASSERTF(page_pool != NULL, "count %u, pool %u\n", count, pool);
- p_idx = page_pools.epp_free_pages / PAGES_PER_POOL;
- g_idx = page_pools.epp_free_pages % PAGES_PER_POOL;
+ spin_lock(&page_pool->epp_lock);
- if (page_pools.epp_free_pages + count > page_pools.epp_total_pages)
+ p_idx = page_pool->epp_free_pages / PAGES_PER_POOL;
+ g_idx = page_pool->epp_free_pages % PAGES_PER_POOL;
+
+ if (page_pool->epp_free_pages + count > page_pool->epp_total_pages)
GOTO(out_unlock, rc = -EPROTO);
- if (!page_pools.epp_pools[p_idx])
+ if (!page_pool->epp_pools[p_idx])
GOTO(out_unlock, rc = -EPROTO);
for (i = 0; i < count; i++) {
- struct page **pagep = page_from(array, i);
+ void **pagep = page_from(array, i);
if (!*pagep ||
- page_pools.epp_pools[p_idx][g_idx] != NULL)
+ page_pool->epp_pools[p_idx][g_idx] != NULL)
GOTO(out_unlock, rc = -EPROTO);
- page_pools.epp_pools[p_idx][g_idx] = *pagep;
+ page_pool->epp_pools[p_idx][g_idx] = *pagep;
if (++g_idx == PAGES_PER_POOL) {
p_idx++;
g_idx = 0;
}
}
- page_pools.epp_free_pages += count;
- enc_pools_wakeup();
+ page_pool->epp_free_pages += count;
+ enc_pools_wakeup(pool);
out_unlock:
- spin_unlock(&page_pools.epp_lock);
+ spin_unlock(&page_pool->epp_lock);
return rc;
}
return;
rc = __sptlrpc_enc_pool_put_pages((void *)desc, desc->bd_iov_count,
- page_from_bulkdesc);
+ PAGES_POOL, page_from_bulkdesc);
if (rc)
CDEBUG(D_SEC, "error putting pages in enc pool: %d\n", rc);
{
int rc;
- rc = __sptlrpc_enc_pool_put_pages((void *)pa, count,
+ rc = __sptlrpc_enc_pool_put_pages((void *)pa, count, PAGES_POOL,
page_from_pagearray);
if (rc)
}
EXPORT_SYMBOL(sptlrpc_enc_pool_put_pages_array);
+void sptlrpc_enc_pool_put_buf(void *buf, unsigned int size_bits)
+{
+ int rc;
+
+ rc = __sptlrpc_enc_pool_put_pages(buf, 1,
+ PPOOL_SIZE_TO_INDEX(size_bits),
+ page_from_bufarray);
+ if (rc)
+ CDEBUG(D_SEC, "error putting pages in enc pool: %d\n", rc);
+}
+EXPORT_SYMBOL(sptlrpc_enc_pool_put_buf);
+
+
/*
* we don't do much stuff for add_user/del_user anymore, except adding some
* initial pages in add_user() if current pools are empty, rest would be
{
int need_grow = 0;
- spin_lock(&page_pools.epp_lock);
- if (page_pools.epp_growing == 0 && page_pools.epp_total_pages == 0) {
- page_pools.epp_growing = 1;
+ spin_lock(&page_pools[PAGES_POOL]->epp_lock);
+ if (page_pools[PAGES_POOL]->epp_growing == 0 &&
+ page_pools[PAGES_POOL]->epp_total_pages == 0) {
+ page_pools[PAGES_POOL]->epp_growing = 1;
need_grow = 1;
}
- spin_unlock(&page_pools.epp_lock);
+ spin_unlock(&page_pools[PAGES_POOL]->epp_lock);
+
if (need_grow) {
enc_pools_add_pages(PTLRPC_MAX_BRW_PAGES +
- PTLRPC_MAX_BRW_PAGES);
+ PTLRPC_MAX_BRW_PAGES, 0);
- spin_lock(&page_pools.epp_lock);
- page_pools.epp_growing = 0;
- enc_pools_wakeup();
- spin_unlock(&page_pools.epp_lock);
+ spin_lock(&page_pools[PAGES_POOL]->epp_lock);
+ page_pools[PAGES_POOL]->epp_growing = 0;
+ enc_pools_wakeup(PAGES_POOL);
+ spin_unlock(&page_pools[PAGES_POOL]->epp_lock);
}
return 0;
}
EXPORT_SYMBOL(sptlrpc_enc_pool_add_user);
-static inline void enc_pools_alloc(void)
+static inline void enc_pools_alloc(struct ptlrpc_enc_page_pool *pool)
{
- LASSERT(page_pools.epp_max_pools);
- OBD_ALLOC_LARGE(page_pools.epp_pools,
- page_pools.epp_max_pools *
- sizeof(*page_pools.epp_pools));
+ LASSERT(pool->epp_max_pools);
+ OBD_ALLOC_LARGE(pool->epp_pools,
+ pool->epp_max_pools *
+ sizeof(*pool->epp_pools));
}
-static inline void enc_pools_free(void)
+static inline void enc_pools_free(unsigned int i)
{
- LASSERT(page_pools.epp_max_pools);
- LASSERT(page_pools.epp_pools);
+ LASSERT(page_pools[i]->epp_max_pools);
+ LASSERT(page_pools[i]->epp_pools);
- OBD_FREE_LARGE(page_pools.epp_pools,
- page_pools.epp_max_pools *
- sizeof(*page_pools.epp_pools));
+ OBD_FREE_LARGE(page_pools[i]->epp_pools,
+ page_pools[i]->epp_max_pools *
+ sizeof(*page_pools[i]->epp_pools));
}
int sptlrpc_enc_pool_init(void)
{
- int rc;
-
- page_pools.epp_max_pages = cfs_totalram_pages() / 8;
- if (enc_pool_max_memory_mb > 0 &&
- enc_pool_max_memory_mb <= (cfs_totalram_pages() >> mult))
- page_pools.epp_max_pages = enc_pool_max_memory_mb << mult;
-
- page_pools.epp_max_pools = npages_to_npools(page_pools.epp_max_pages);
-
- init_waitqueue_head(&page_pools.epp_waitq);
- page_pools.epp_waitqlen = 0;
- page_pools.epp_pages_short = 0;
-
- page_pools.epp_growing = 0;
-
- page_pools.epp_idle_idx = 0;
- page_pools.epp_last_shrink = ktime_get_seconds();
- page_pools.epp_last_access = ktime_get_seconds();
-
- spin_lock_init(&page_pools.epp_lock);
- page_pools.epp_total_pages = 0;
- page_pools.epp_free_pages = 0;
+ int pool_index = 0, to_revert;
+ int rc = 0;
+ struct ptlrpc_enc_page_pool *pool;
+
+ ENTRY;
+ OBD_ALLOC(page_pools, POOLS_COUNT * sizeof(*page_pools));
+ if (page_pools == NULL)
+ RETURN(-ENOMEM);
+ for (pool_index = 0; pool_index < POOLS_COUNT; pool_index++) {
+ OBD_ALLOC(page_pools[pool_index], sizeof(**page_pools));
+ if (page_pools[pool_index] == NULL)
+ GOTO(fail, rc = -ENOMEM);
+
+ pool = page_pools[pool_index];
+ pool->epp_max_pages =
+ cfs_totalram_pages() / POOLS_COUNT;
+ if (enc_pool_max_memory_mb > 0 &&
+ enc_pool_max_memory_mb <= (cfs_totalram_pages() >> mult))
+ pool->epp_max_pages =
+ enc_pool_max_memory_mb << mult;
+
+ pool->epp_max_pools =
+ npages_to_npools(pool->epp_max_pages);
+
+ init_waitqueue_head(&pool->epp_waitq);
+ pool->epp_last_shrink = ktime_get_seconds();
+ pool->epp_last_access = ktime_get_seconds();
+
+ spin_lock_init(&pool->epp_lock);
+ pool->epp_st_max_wait = ktime_set(0, 0);
+
+ enc_pools_alloc(pool);
+ CDEBUG(D_SEC, "Allocated pool %i\n", pool_index);
+ if (pool->epp_pools == NULL)
+ GOTO(fail, rc = -ENOMEM);
+ /* Pass pool number as part of pools_shrinker_seeks value */
+#ifdef HAVE_SHRINKER_COUNT
+ pool->pool_shrinker.count_objects = enc_pools_shrink_count;
+ pool->pool_shrinker.scan_objects = enc_pools_shrink_scan;
+#else
+ pool->pool_shrinker.shrink = enc_pools_shrink;
+#endif
+ pool->pool_shrinker.seeks = INDEX_TO_SEEKS(pool_index);
- page_pools.epp_st_max_pages = 0;
- page_pools.epp_st_grows = 0;
- page_pools.epp_st_grow_fails = 0;
- page_pools.epp_st_shrinks = 0;
- page_pools.epp_st_access = 0;
- page_pools.epp_st_missings = 0;
- page_pools.epp_st_lowfree = 0;
- page_pools.epp_st_max_wqlen = 0;
- page_pools.epp_st_max_wait = ktime_set(0, 0);
- page_pools.epp_st_outofmem = 0;
+ rc = register_shrinker(&pool->pool_shrinker);
+ if (rc)
+ GOTO(fail, rc);
- enc_pools_alloc();
- if (page_pools.epp_pools == NULL)
- return -ENOMEM;
+ mutex_init(&pool->add_pages_mutex);
+ }
- rc = register_shrinker(&pools_shrinker);
- if (rc)
- enc_pools_free();
+ RETURN(0);
+fail:
+ to_revert = pool_index;
+ for (pool_index = 0; pool_index <= to_revert; pool_index++) {
+ pool = page_pools[pool_index];
+ if (pool) {
+ if (pool->epp_pools)
+ enc_pools_free(pool_index);
+ OBD_FREE(pool, sizeof(**page_pools));
+ }
+ }
+ OBD_FREE(page_pools, POOLS_COUNT * sizeof(*page_pools));
- return rc;
+ RETURN(rc);
}
void sptlrpc_enc_pool_fini(void)
{
unsigned long cleaned, npools;
+ int pool_index;
+ struct ptlrpc_enc_page_pool *pool;
+
+ for (pool_index = 0; pool_index < POOLS_COUNT; pool_index++) {
+ pool = page_pools[pool_index];
+ unregister_shrinker(&pool->pool_shrinker);
+ LASSERT(pool->epp_pools);
+ LASSERT(pool->epp_total_pages == pool->epp_free_pages);
+
+ npools = npages_to_npools(pool->epp_total_pages);
+ cleaned = enc_pools_cleanup(pool->epp_pools,
+ npools, pool_index);
+ LASSERT(cleaned == pool->epp_total_pages);
+
+ enc_pools_free(pool_index);
+
+ if (pool->epp_st_access > 0) {
+ CDEBUG(D_SEC,
+ "max pages %lu, grows %u, grow fails %u, shrinks %u, access %lu, missing %lu, max qlen %u, max wait ms %lld, out of mem %lu\n",
+ pool->epp_st_max_pages,
+ pool->epp_st_grows,
+ pool->epp_st_grow_fails,
+ pool->epp_st_shrinks,
+ pool->epp_st_access,
+ pool->epp_st_missings,
+ pool->epp_st_max_wqlen,
+ ktime_to_ms(pool->epp_st_max_wait),
+ pool->epp_st_outofmem);
+ }
- LASSERT(page_pools.epp_pools);
- LASSERT(page_pools.epp_total_pages == page_pools.epp_free_pages);
-
- unregister_shrinker(&pools_shrinker);
-
- npools = npages_to_npools(page_pools.epp_total_pages);
- cleaned = enc_pools_cleanup(page_pools.epp_pools, npools);
- LASSERT(cleaned == page_pools.epp_total_pages);
-
- enc_pools_free();
-
- if (page_pools.epp_st_access > 0) {
- CDEBUG(D_SEC,
- "max pages %lu, grows %u, grow fails %u, shrinks %u, access %lu, missing %lu, max qlen %u, max wait ms %lld, out of mem %lu\n",
- page_pools.epp_st_max_pages, page_pools.epp_st_grows,
- page_pools.epp_st_grow_fails,
- page_pools.epp_st_shrinks, page_pools.epp_st_access,
- page_pools.epp_st_missings, page_pools.epp_st_max_wqlen,
- ktime_to_ms(page_pools.epp_st_max_wait),
- page_pools.epp_st_outofmem);
+ OBD_FREE(pool, sizeof(**page_pools));
}
-}
+ OBD_FREE(page_pools, POOLS_COUNT * sizeof(*page_pools));
+}
static int cfs_hash_alg_id[] = {
[BULK_HASH_ALG_NULL] = CFS_HASH_ALG_NULL,
git://git.whamcloud.com / fs / lustre-release.git / commitdiff