*/
/*
* This file is part of Lustre, http://www.lustre.org/
- * Lustre is a trademark of Sun Microsystems, Inc.
*
* lustre/llite/rw.c
*
unsigned long pages,
unsigned long pages_min)
{
- struct ll_ra_info *ra = &sbi->ll_ra_info;
- long ret;
+ struct ll_ra_info *ra = &sbi->ll_ra_info;
+ long ret;
+
ENTRY;
- /* If read-ahead pages left are less than 1M, do not do read-ahead,
- * otherwise it will form small read RPC(< 1M), which hurt server
- * performance a lot. */
+ WARN_ON_ONCE(pages_min > pages);
+ /**
+ * Don't try readahead aggresively if we are limited
+ * LRU pages, otherwise, it could cause deadlock.
+ */
+ pages = min(sbi->ll_cache->ccc_lru_max >> 2, pages);
+ /**
+ * if this happen, we reserve more pages than needed,
+ * this will make us leak @ra_cur_pages, because
+ * ll_ra_count_put() acutally freed @pages.
+ */
+ if (unlikely(pages_min > pages))
+ pages_min = pages;
+
+ /*
+ * If read-ahead pages left are less than 1M, do not do read-ahead,
+ * otherwise it will form small read RPC(< 1M), which hurt server
+ * performance a lot.
+ */
ret = min(ra->ra_max_pages - atomic_read(&ra->ra_cur_pages),
pages);
- if (ret < 0 || ret < min_t(long, PTLRPC_MAX_BRW_PAGES, pages))
- GOTO(out, ret = 0);
+ if (ret < 0 || ret < min_t(long, PTLRPC_MAX_BRW_PAGES, pages))
+ GOTO(out, ret = 0);
if (atomic_add_return(ret, &ra->ra_cur_pages) > ra->ra_max_pages) {
atomic_sub(ret, &ra->ra_cur_pages);
cl_page_assume(env, io, page);
vpg = cl2vvp_page(cl_object_page_slice(clob, page));
if (!vpg->vpg_defer_uptodate && !PageUptodate(vmpage)) {
- vpg->vpg_defer_uptodate = 1;
- vpg->vpg_ra_used = 0;
- cl_page_list_add(queue, page);
+ if (hint == MAYNEED) {
+ vpg->vpg_defer_uptodate = 1;
+ vpg->vpg_ra_used = 0;
+ }
+ cl_page_list_add(queue, page, true);
} else {
/* skip completed pages */
cl_page_unassume(env, io, page);
static pgoff_t ras_align(struct ll_readahead_state *ras, pgoff_t index)
{
- return index - (index % ras->ras_rpc_pages);
+ unsigned opt_size = min(ras->ras_window_pages, ras->ras_rpc_pages);
+
+ if (opt_size == 0)
+ opt_size = 1;
+ return index - (index % opt_size);
}
/* Check whether the index is in the defined ra-window */
static unsigned long
ll_read_ahead_pages(const struct lu_env *env, struct cl_io *io,
struct cl_page_list *queue, struct ll_readahead_state *ras,
- struct ra_io_arg *ria, pgoff_t *ra_end)
+ struct ra_io_arg *ria, pgoff_t *ra_end, pgoff_t skip_index)
{
struct cl_read_ahead ra = { 0 };
/* busy page count is per stride */
for (page_idx = ria->ria_start_idx;
page_idx <= ria->ria_end_idx && ria->ria_reserved > 0;
page_idx++) {
+ if (skip_index && page_idx == skip_index)
+ continue;
if (ras_inside_ra_window(page_idx, ria)) {
if (ra.cra_end_idx == 0 || ra.cra_end_idx < page_idx) {
pgoff_t end_idx;
if (ras->ras_rpc_pages != ra.cra_rpc_pages &&
ra.cra_rpc_pages > 0)
ras->ras_rpc_pages = ra.cra_rpc_pages;
- /* trim it to align with optimal RPC size */
- end_idx = ras_align(ras, ria->ria_end_idx + 1);
- if (end_idx > 0 && !ria->ria_eof)
- ria->ria_end_idx = end_idx - 1;
+ if (!skip_index) {
+ /* trim it to align with optimal RPC size */
+ end_idx = ras_align(ras, ria->ria_end_idx + 1);
+ if (end_idx > 0 && !ria->ria_eof)
+ ria->ria_end_idx = end_idx - 1;
+ }
if (ria->ria_end_idx < ria->ria_end_idx_min)
ria->ria_end_idx = ria->ria_end_idx_min;
}
memcpy(ll_i2info(inode)->lli_jobid, work->lrw_jobid,
sizeof(work->lrw_jobid));
- vvp_env_io(env)->vui_io_subtype = IO_NORMAL;
vvp_env_io(env)->vui_fd = fd;
io->ci_state = CIS_LOCKED;
io->ci_async_readahead = true;
cl_2queue_init(queue);
rc = ll_read_ahead_pages(env, io, &queue->c2_qin, ras, ria,
- &ra_end_idx);
+ &ra_end_idx, 0);
if (ria->ria_reserved != 0)
ll_ra_count_put(ll_i2sbi(inode), ria->ria_reserved);
if (queue->c2_qin.pl_nr > 0) {
static int ll_readahead(const struct lu_env *env, struct cl_io *io,
struct cl_page_list *queue,
struct ll_readahead_state *ras, bool hit,
- struct file *file)
+ struct file *file, pgoff_t skip_index)
{
struct vvp_io *vio = vvp_env_io(env);
struct ll_thread_info *lti = ll_env_info(env);
struct cl_object *clob;
int ret = 0;
__u64 kms;
+ struct ll_sb_info *sbi;
+ struct ll_ra_info *ra;
+
ENTRY;
+ ENTRY;
+
clob = io->ci_obj;
inode = vvp_object_inode(clob);
+ sbi = ll_i2sbi(inode);
+ ra = &sbi->ll_ra_info;
+
+ /**
+ * In case we have a limited max_cached_mb, readahead
+ * should be stopped if it have run out of all LRU slots.
+ */
+ if (atomic_read(&ra->ra_cur_pages) >= sbi->ll_cache->ccc_lru_max) {
+ ll_ra_stats_inc(inode, RA_STAT_MAX_IN_FLIGHT);
+ RETURN(0);
+ }
memset(ria, 0, sizeof(*ria));
ret = ll_readahead_file_kms(env, io, &kms);
if (ras->ras_window_pages > 0)
end_idx = ras->ras_window_start_idx + ras->ras_window_pages - 1;
+ if (skip_index)
+ end_idx = start_idx + ras->ras_window_pages - 1;
+
/* Enlarge the RA window to encompass the full read */
if (vio->vui_ra_valid &&
end_idx < vio->vui_ra_start_idx + vio->vui_ra_pages - 1)
vio->vui_ra_start_idx + vio->vui_ra_pages - 1;
pages_min = vio->vui_ra_start_idx + vio->vui_ra_pages -
ria->ria_start_idx;
+ /**
+ * For performance reason, exceeding @ra_max_pages
+ * are allowed, but this should be limited with RPC
+ * size in case a large block size read issued. Trim
+ * to RPC boundary.
+ */
+ pages_min = min(pages_min, ras->ras_rpc_pages -
+ (ria->ria_start_idx % ras->ras_rpc_pages));
}
+ /* don't over reserved for mmap range read */
+ if (skip_index)
+ pages_min = 0;
+ if (pages_min > pages)
+ pages = pages_min;
ria->ria_reserved = ll_ra_count_get(ll_i2sbi(inode), ria, pages,
pages_min);
if (ria->ria_reserved < pages)
atomic_read(&ll_i2sbi(inode)->ll_ra_info.ra_cur_pages),
ll_i2sbi(inode)->ll_ra_info.ra_max_pages);
- ret = ll_read_ahead_pages(env, io, queue, ras, ria, &ra_end_idx);
-
+ ret = ll_read_ahead_pages(env, io, queue, ras, ria, &ra_end_idx,
+ skip_index);
if (ria->ria_reserved != 0)
ll_ra_count_put(ll_i2sbi(inode), ria->ria_reserved);
ras_reset(ras, 0);
ras->ras_last_read_end_bytes = 0;
ras->ras_requests = 0;
+ ras->ras_range_min_start_idx = 0;
+ ras->ras_range_max_end_idx = 0;
+ ras->ras_range_requests = 0;
+ ras->ras_last_range_pages = 0;
}
/*
8UL << PAGE_SHIFT, 8UL << PAGE_SHIFT);
}
+static inline bool is_loose_mmap_read(struct ll_sb_info *sbi,
+ struct ll_readahead_state *ras,
+ unsigned long pos)
+{
+ unsigned long range_pages = sbi->ll_ra_info.ra_range_pages;
+
+ return pos_in_window(pos, ras->ras_last_read_end_bytes,
+ range_pages << PAGE_SHIFT,
+ range_pages << PAGE_SHIFT);
+}
+
+/**
+ * We have observed slow mmap read performances for some
+ * applications. The problem is if access pattern is neither
+ * sequential nor stride, but could be still adjacent in a
+ * small range and then seek a random position.
+ *
+ * So the pattern could be something like this:
+ *
+ * [1M data] [hole] [0.5M data] [hole] [0.7M data] [1M data]
+ *
+ *
+ * Every time an application reads mmap data, it may not only
+ * read a single 4KB page, but aslo a cluster of nearby pages in
+ * a range(e.g. 1MB) of the first page after a cache miss.
+ *
+ * The readahead engine is modified to track the range size of
+ * a cluster of mmap reads, so that after a seek and/or cache miss,
+ * the range size is used to efficiently prefetch multiple pages
+ * in a single RPC rather than many small RPCs.
+ */
+static void ras_detect_cluster_range(struct ll_readahead_state *ras,
+ struct ll_sb_info *sbi,
+ unsigned long pos, unsigned long count)
+{
+ pgoff_t last_pages, pages;
+ pgoff_t end_idx = (pos + count - 1) >> PAGE_SHIFT;
+
+ last_pages = ras->ras_range_max_end_idx -
+ ras->ras_range_min_start_idx + 1;
+ /* First time come here */
+ if (!ras->ras_range_max_end_idx)
+ goto out;
+
+ /* Random or Stride read */
+ if (!is_loose_mmap_read(sbi, ras, pos))
+ goto out;
+
+ ras->ras_range_requests++;
+ if (ras->ras_range_max_end_idx < end_idx)
+ ras->ras_range_max_end_idx = end_idx;
+
+ if (ras->ras_range_min_start_idx > (pos >> PAGE_SHIFT))
+ ras->ras_range_min_start_idx = pos >> PAGE_SHIFT;
+
+ /* Out of range, consider it as random or stride */
+ pages = ras->ras_range_max_end_idx -
+ ras->ras_range_min_start_idx + 1;
+ if (pages <= sbi->ll_ra_info.ra_range_pages)
+ return;
+out:
+ ras->ras_last_range_pages = last_pages;
+ ras->ras_range_requests = 0;
+ ras->ras_range_min_start_idx = pos >> PAGE_SHIFT;
+ ras->ras_range_max_end_idx = end_idx;
+}
+
static void ras_detect_read_pattern(struct ll_readahead_state *ras,
struct ll_sb_info *sbi,
loff_t pos, size_t count, bool mmap)
ras->ras_consecutive_bytes += count;
if (mmap) {
pgoff_t idx = ras->ras_consecutive_bytes >> PAGE_SHIFT;
+ unsigned long ra_range_pages =
+ max_t(unsigned long, RA_MIN_MMAP_RANGE_PAGES,
+ sbi->ll_ra_info.ra_range_pages);
- if ((idx >= 4 && (idx & 3UL) == 0) || stride_detect)
+ if ((idx >= ra_range_pages &&
+ idx % ra_range_pages == 0) || stride_detect)
ras->ras_need_increase_window = true;
} else if ((ras->ras_consecutive_requests > 1 || stride_detect)) {
ras->ras_need_increase_window = true;
*/
static void ras_update(struct ll_sb_info *sbi, struct inode *inode,
struct ll_readahead_state *ras, pgoff_t index,
- enum ras_update_flags flags)
+ enum ras_update_flags flags, struct cl_io *io)
{
struct ll_ra_info *ra = &sbi->ll_ra_info;
bool hit = flags & LL_RAS_HIT;
if (ras->ras_no_miss_check)
GOTO(out_unlock, 0);
- if (flags & LL_RAS_MMAP)
+ if (io && io->ci_rand_read)
+ GOTO(out_unlock, 0);
+
+ if (io && io->ci_seq_read) {
+ if (!hit) {
+ /* to avoid many small read RPC here */
+ ras->ras_window_pages = sbi->ll_ra_info.ra_range_pages;
+ ll_ra_stats_inc_sbi(sbi, RA_STAT_MMAP_RANGE_READ);
+ }
+ goto skip;
+ }
+
+ if (flags & LL_RAS_MMAP) {
+ unsigned long ra_pages;
+
+ ras_detect_cluster_range(ras, sbi, index << PAGE_SHIFT,
+ PAGE_SIZE);
ras_detect_read_pattern(ras, sbi, (loff_t)index << PAGE_SHIFT,
PAGE_SIZE, true);
+ /* we did not detect anything but we could prefetch */
+ if (!ras->ras_need_increase_window &&
+ ras->ras_window_pages <= sbi->ll_ra_info.ra_range_pages &&
+ ras->ras_range_requests >= 2) {
+ if (!hit) {
+ ra_pages = max_t(unsigned long,
+ RA_MIN_MMAP_RANGE_PAGES,
+ ras->ras_last_range_pages);
+ if (index < ra_pages / 2)
+ index = 0;
+ else
+ index -= ra_pages / 2;
+ ras->ras_window_pages = ra_pages;
+ ll_ra_stats_inc_sbi(sbi,
+ RA_STAT_MMAP_RANGE_READ);
+ } else {
+ ras->ras_window_pages = 0;
+ }
+ goto skip;
+ }
+ }
+
if (!hit && ras->ras_window_pages &&
index < ras->ras_next_readahead_idx &&
pos_in_window(index, ras->ras_window_start_idx, 0,
GOTO(out_unlock, 0);
}
}
+
+skip:
ras_set_start(ras, index);
if (stride_io_mode(ras)) {
{
struct inode *inode = vvp_object_inode(page->cp_obj);
struct ll_sb_info *sbi = ll_i2sbi(inode);
- struct ll_file_data *fd = file->private_data;
- struct ll_readahead_state *ras = &fd->fd_ras;
+ struct ll_file_data *fd = NULL;
+ struct ll_readahead_state *ras = NULL;
struct cl_2queue *queue = &io->ci_queue;
struct cl_sync_io *anchor = NULL;
struct vvp_page *vpg;
pgoff_t io_end_index;
ENTRY;
+ if (file) {
+ fd = file->private_data;
+ ras = &fd->fd_ras;
+ }
+
vpg = cl2vvp_page(cl_object_page_slice(page->cp_obj, page));
uptodate = vpg->vpg_defer_uptodate;
- if (ll_readahead_enabled(sbi) && !vpg->vpg_ra_updated) {
+ if (ll_readahead_enabled(sbi) && !vpg->vpg_ra_updated && ras) {
struct vvp_io *vio = vvp_env_io(env);
enum ras_update_flags flags = 0;
flags |= LL_RAS_HIT;
if (!vio->vui_ra_valid)
flags |= LL_RAS_MMAP;
- ras_update(sbi, inode, ras, vvp_index(vpg), flags);
+ ras_update(sbi, inode, ras, vvp_index(vpg), flags, io);
}
cl_2queue_init(queue);
cl_sync_io_init(anchor, 1);
page->cp_sync_io = anchor;
- cl_2queue_add(queue, page);
+ cl_2queue_add(queue, page, true);
}
io_start_index = cl_index(io->ci_obj, io->u.ci_rw.crw_pos);
io_end_index = cl_index(io->ci_obj, io->u.ci_rw.crw_pos +
io->u.ci_rw.crw_count - 1);
- if (ll_readahead_enabled(sbi)) {
+ if (ll_readahead_enabled(sbi) && ras && !io->ci_rand_read) {
+ pgoff_t skip_index = 0;
+
+ if (ras->ras_next_readahead_idx < vvp_index(vpg))
+ skip_index = vvp_index(vpg);
rc2 = ll_readahead(env, io, &queue->c2_qin, ras,
- uptodate, file);
+ uptodate, file, skip_index);
CDEBUG(D_READA, DFID " %d pages read ahead at %lu\n",
PFID(ll_inode2fid(inode)), rc2, vvp_index(vpg));
} else if (vvp_index(vpg) == io_start_index &&
pgoff_t start_idx = ras_align(ras, ras->ras_next_readahead_idx);
pgoff_t end_idx = start_idx + pages - 1;
+ /**
+ * In case we have a limited max_cached_mb, readahead
+ * should be stopped if it have run out of all LRU slots.
+ */
+ if (atomic_read(&ra->ra_cur_pages) >= sbi->ll_cache->ccc_lru_max) {
+ ll_ra_stats_inc(inode, RA_STAT_MAX_IN_FLIGHT);
+ return 0;
+ }
+
throttle = min(ra->ra_async_pages_per_file_threshold,
ra->ra_max_pages_per_file);
/*
unsigned long fast_read_pages =
max(RA_REMAIN_WINDOW_MIN, ras->ras_rpc_pages);
loff_t skip_pages;
+ loff_t stride_bytes = ras->ras_stride_bytes;
- if (stride_io_mode(ras)) {
+ if (stride_io_mode(ras) && stride_bytes) {
skip_pages = (ras->ras_stride_length +
- ras->ras_stride_bytes - 1) / ras->ras_stride_bytes;
+ ras->ras_stride_bytes - 1) / stride_bytes;
skip_pages *= fast_read_pages;
} else {
skip_pages = fast_read_pages;
/* For fast read, it updates read ahead state only
* if the page is hit in cache because non cache page
* case will be handled by slow read later. */
- ras_update(sbi, inode, ras, vvp_index(vpg), flags);
+ ras_update(sbi, inode, ras, vvp_index(vpg), flags, io);
/* avoid duplicate ras_update() call */
vpg->vpg_ra_updated = 1;
*/
if (file->f_flags & O_DIRECT &&
lcc && lcc->lcc_type == LCC_RW &&
- !io->ci_ignore_lockless) {
+ !io->ci_dio_lock) {
unlock_page(vmpage);
- io->ci_ignore_lockless = 1;
+ io->ci_dio_lock = 1;
io->ci_need_restart = 1;
RETURN(-ENOLCK);
}