/*
* This is the predictive prefetch entry point. It associates dnode access
* specified with blkid and nblks arguments with prefetch stream, predicts
* further accesses based on that stats and initiates speculative prefetch.
* fetch_data argument specifies whether actual data blocks should be fetched:
* FALSE -- prefetch only indirect blocks for predicted data blocks;
* TRUE -- prefetch predicted data blocks plus following indirect blocks.
*/
void
dmu_zfetch(zfetch_t *zf, uint64_t blkid, uint64_t nblks, boolean_t fetch_data)
{
zstream_t *zs;
int64_t pf_start, ipf_start, ipf_istart, ipf_iend;
int64_t pf_ahead_blks, max_blks;
int epbs, max_dist_blks, pf_nblks, ipf_nblks;
uint64_t end_of_access_blkid = blkid + nblks;
if (zfs_prefetch_disable)
return;
/*
* As a fast path for small (single-block) files, ignore access
* to the first block.
*/
if (blkid == 0)
return;
rw_enter(&zf->zf_rwlock, RW_READER);
for (zs = list_head(&zf->zf_stream); zs != NULL;
zs = list_next(&zf->zf_stream, zs)) {
if (blkid == zs->zs_blkid) {
mutex_enter(&zs->zs_lock);
/*
* zs_blkid could have changed before we
* acquired zs_lock; re-check them here.
*/
if (blkid != zs->zs_blkid) {
mutex_exit(&zs->zs_lock);
continue;
}
break;
}
}
if (zs == NULL) {
/*
* This access is not part of any existing stream. Create
* a new stream for it.
*/
ZFETCHSTAT_BUMP(zfetchstat_misses);
if (rw_tryupgrade(&zf->zf_rwlock))
dmu_zfetch_stream_create(zf, end_of_access_blkid);
rw_exit(&zf->zf_rwlock);
return;
}
/*
* This access was to a block that we issued a prefetch for on
* behalf of this stream. Issue further prefetches for this stream.
*
* Normally, we start prefetching where we stopped
* prefetching last (zs_pf_blkid). But when we get our first
* hit on this stream, zs_pf_blkid == zs_blkid, we don't
* want to prefetch the block we just accessed. In this case,
* start just after the block we just accessed.
*/
pf_start = MAX(zs->zs_pf_blkid, end_of_access_blkid);
/*
* Double our amount of prefetched data, but don't let the
* prefetch get further ahead than zfetch_max_distance.
*/
if (fetch_data) {
max_dist_blks =
zfetch_max_distance >> zf->zf_dnode->dn_datablkshift;
/*
* Previously, we were (zs_pf_blkid - blkid) ahead. We
* want to now be double that, so read that amount again,
* plus the amount we are catching up by (i.e. the amount
* read just now).
*/
pf_ahead_blks = zs->zs_pf_blkid - blkid + nblks;
max_blks = max_dist_blks - (pf_start - end_of_access_blkid);
pf_nblks = MIN(pf_ahead_blks, max_blks);
} else {
/*
* This is the predictive prefetch entry point. It associates dnode access
* specified with blkid and nblks arguments with prefetch stream, predicts
* further accesses based on that stats and initiates speculative prefetch.
* fetch_data argument specifies whether actual data blocks should be fetched:
* FALSE -- prefetch only indirect blocks for predicted data blocks;
* TRUE -- prefetch predicted data blocks plus following indirect blocks.
*/
void
dmu_zfetch(zfetch_t *zf, uint64_t blkid, uint64_t nblks, boolean_t fetch_data)
{
zstream_t *zs;
int64_t pf_start, ipf_start, ipf_istart, ipf_iend;
int64_t pf_ahead_blks, max_blks;
int epbs, max_dist_blks, pf_nblks, ipf_nblks;
uint64_t end_of_access_blkid = blkid + nblks;
spa_t *spa = zf->zf_dnode->dn_objset->os_spa;
if (zfs_prefetch_disable)
return;
/*
* If we haven't yet loaded the indirect vdevs' mappings, we
* can only read from blocks that we carefully ensure are on
* concrete vdevs (or previously-loaded indirect vdevs). So we
* can't allow the predictive prefetcher to attempt reads of other
* blocks (e.g. of the MOS's dnode obejct).
*/
if (!spa_indirect_vdevs_loaded(spa))
return;
/*
* As a fast path for small (single-block) files, ignore access
* to the first block.
*/
if (blkid == 0)
return;
rw_enter(&zf->zf_rwlock, RW_READER);
/*
* Find matching prefetch stream. Depending on whether the accesses
* are block-aligned, first block of the new access may either follow
* the last block of the previous access, or be equal to it.
*/
for (zs = list_head(&zf->zf_stream); zs != NULL;
zs = list_next(&zf->zf_stream, zs)) {
if (blkid == zs->zs_blkid || blkid + 1 == zs->zs_blkid) {
mutex_enter(&zs->zs_lock);
/*
* zs_blkid could have changed before we
* acquired zs_lock; re-check them here.
*/
if (blkid == zs->zs_blkid) {
break;
} else if (blkid + 1 == zs->zs_blkid) {
blkid++;
nblks--;
if (nblks == 0) {
/* Already prefetched this before. */
mutex_exit(&zs->zs_lock);
rw_exit(&zf->zf_rwlock);
return;
}
break;
}
mutex_exit(&zs->zs_lock);
}
}
if (zs == NULL) {
/*
* This access is not part of any existing stream. Create
* a new stream for it.
*/
ZFETCHSTAT_BUMP(zfetchstat_misses);
if (rw_tryupgrade(&zf->zf_rwlock))
dmu_zfetch_stream_create(zf, end_of_access_blkid);
rw_exit(&zf->zf_rwlock);
return;
}
/*
* This access was to a block that we issued a prefetch for on
* behalf of this stream. Issue further prefetches for this stream.
*
* Normally, we start prefetching where we stopped
* prefetching last (zs_pf_blkid). But when we get our first
* hit on this stream, zs_pf_blkid == zs_blkid, we don't
* want to prefetch the block we just accessed. In this case,
* start just after the block we just accessed.
*/
pf_start = MAX(zs->zs_pf_blkid, end_of_access_blkid);
/*
* Double our amount of prefetched data, but don't let the
* prefetch get further ahead than zfetch_max_distance.
*/
if (fetch_data) {
max_dist_blks =
zfetch_max_distance >> zf->zf_dnode->dn_datablkshift;
/*
* Previously, we were (zs_pf_blkid - blkid) ahead. We
* want to now be double that, so read that amount again,
* plus the amount we are catching up by (i.e. the amount
* read just now).
*/
pf_ahead_blks = zs->zs_pf_blkid - blkid + nblks;
max_blks = max_dist_blks - (pf_start - end_of_access_blkid);
pf_nblks = MIN(pf_ahead_blks, max_blks);
} else {
