X-Git-Url: https://git.camperquake.de/gitweb.cgi?a=blobdiff_plain;f=module%2Fzfs%2Fdbuf.c;h=faa6cc345b8f53b02bbc2e646efa3eb8040fcacd;hb=refs%2Fheads%2Frertzinger%2Ffeature-zpool-get--p;hp=42d82bbfb8f474b62da89ee3342eb03670e578d4;hpb=4047414a6a340dff5fd52d48b163991bc9760ad2;p=zfs.git diff --git a/module/zfs/dbuf.c b/module/zfs/dbuf.c index 42d82bb..faa6cc3 100644 --- a/module/zfs/dbuf.c +++ b/module/zfs/dbuf.c @@ -21,6 +21,8 @@ /* * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. * Copyright 2011 Nexenta Systems, Inc. All rights reserved. + * Copyright (c) 2012 by Delphix. All rights reserved. + * Copyright (c) 2013 by Saso Kiselkov. All rights reserved. */ #include @@ -260,7 +262,7 @@ dbuf_is_metadata(dmu_buf_impl_t *db) boolean_t is_metadata; DB_DNODE_ENTER(db); - is_metadata = dmu_ot[DB_DNODE(db)->dn_type].ot_metadata; + is_metadata = DMU_OT_IS_METADATA(DB_DNODE(db)->dn_type); DB_DNODE_EXIT(db); return (is_metadata); @@ -298,7 +300,7 @@ retry: #if defined(_KERNEL) && defined(HAVE_SPL) /* Large allocations which do not require contiguous pages * should be using vmem_alloc() in the linux kernel */ - h->hash_table = vmem_zalloc(hsize * sizeof (void *), KM_SLEEP); + h->hash_table = vmem_zalloc(hsize * sizeof (void *), KM_PUSHPAGE); #else h->hash_table = kmem_zalloc(hsize * sizeof (void *), KM_NOSLEEP); #endif @@ -372,7 +374,7 @@ dbuf_verify(dmu_buf_impl_t *db) } else if (db->db_blkid == DMU_SPILL_BLKID) { ASSERT(dn != NULL); ASSERT3U(db->db.db_size, >=, dn->dn_bonuslen); - ASSERT3U(db->db.db_offset, ==, 0); + ASSERT0(db->db.db_offset); } else { ASSERT3U(db->db.db_offset, ==, db->db_blkid * db->db.db_size); } @@ -558,7 +560,6 @@ dbuf_read_impl(dmu_buf_impl_t *db, zio_t *zio, uint32_t *flags) spa_t *spa; zbookmark_t zb; uint32_t aflags = ARC_NOWAIT; - arc_buf_t *pbuf; DB_DNODE_ENTER(db); dn = DB_DNODE(db); @@ -614,20 +615,16 @@ dbuf_read_impl(dmu_buf_impl_t *db, zio_t *zio, uint32_t *flags) if (DBUF_IS_L2CACHEABLE(db)) aflags |= ARC_L2CACHE; + if (DBUF_IS_L2COMPRESSIBLE(db)) + aflags |= ARC_L2COMPRESS; SET_BOOKMARK(&zb, db->db_objset->os_dsl_dataset ? db->db_objset->os_dsl_dataset->ds_object : DMU_META_OBJSET, db->db.db_object, db->db_level, db->db_blkid); dbuf_add_ref(db, NULL); - /* ZIO_FLAG_CANFAIL callers have to check the parent zio's error */ - if (db->db_parent) - pbuf = db->db_parent->db_buf; - else - pbuf = db->db_objset->os_phys_buf; - - (void) dsl_read(zio, spa, db->db_blkptr, pbuf, + (void) arc_read(zio, spa, db->db_blkptr, dbuf_read_done, db, ZIO_PRIORITY_SYNC_READ, (*flags & DB_RF_CANFAIL) ? ZIO_FLAG_CANFAIL : ZIO_FLAG_MUSTSUCCEED, &aflags, &zb); @@ -1025,7 +1022,6 @@ void dbuf_release_bp(dmu_buf_impl_t *db) { objset_t *os; - zbookmark_t zb; DB_GET_OBJSET(&os, db); ASSERT(dsl_pool_sync_context(dmu_objset_pool(os))); @@ -1033,13 +1029,7 @@ dbuf_release_bp(dmu_buf_impl_t *db) list_link_active(&os->os_dsl_dataset->ds_synced_link)); ASSERT(db->db_parent == NULL || arc_released(db->db_parent->db_buf)); - zb.zb_objset = os->os_dsl_dataset ? - os->os_dsl_dataset->ds_object : 0; - zb.zb_object = db->db.db_object; - zb.zb_level = db->db_level; - zb.zb_blkid = db->db_blkid; - (void) arc_release_bp(db->db_buf, db, - db->db_blkptr, os->os_spa, &zb); + (void) arc_release(db->db_buf, db); } dbuf_dirty_record_t * @@ -1719,7 +1709,7 @@ dbuf_create(dnode_t *dn, uint8_t level, uint64_t blkid, ASSERT(RW_LOCK_HELD(&dn->dn_struct_rwlock)); ASSERT(dn->dn_type != DMU_OT_NONE); - db = kmem_cache_alloc(dbuf_cache, KM_SLEEP); + db = kmem_cache_alloc(dbuf_cache, KM_PUSHPAGE); db->db_objset = os; db->db.db_object = dn->dn_object; @@ -1885,7 +1875,6 @@ dbuf_prefetch(dnode_t *dn, uint64_t blkid) if (bp && !BP_IS_HOLE(bp)) { int priority = dn->dn_type == DMU_OT_DDT_ZAP ? ZIO_PRIORITY_DDT_PREFETCH : ZIO_PRIORITY_ASYNC_READ; - arc_buf_t *pbuf; dsl_dataset_t *ds = dn->dn_objset->os_dsl_dataset; uint32_t aflags = ARC_NOWAIT | ARC_PREFETCH; zbookmark_t zb; @@ -1893,13 +1882,8 @@ dbuf_prefetch(dnode_t *dn, uint64_t blkid) SET_BOOKMARK(&zb, ds ? ds->ds_object : DMU_META_OBJSET, dn->dn_object, 0, blkid); - if (db) - pbuf = db->db_buf; - else - pbuf = dn->dn_objset->os_phys_buf; - - (void) dsl_read(NULL, dn->dn_objset->os_spa, - bp, pbuf, NULL, NULL, priority, + (void) arc_read(NULL, dn->dn_objset->os_spa, + bp, NULL, NULL, priority, ZIO_FLAG_CANFAIL | ZIO_FLAG_SPECULATIVE, &aflags, &zb); } @@ -2019,7 +2003,7 @@ dbuf_hold_impl(dnode_t *dn, uint8_t level, uint64_t blkid, int fail_sparse, int error; dh = kmem_zalloc(sizeof(struct dbuf_hold_impl_data) * - DBUF_HOLD_IMPL_MAX_DEPTH, KM_SLEEP); + DBUF_HOLD_IMPL_MAX_DEPTH, KM_PUSHPAGE); __dbuf_hold_impl_init(dh, dn, level, blkid, fail_sparse, tag, dbp, 0); error = __dbuf_hold_impl(dh); @@ -2188,7 +2172,24 @@ dbuf_rele_and_unlock(dmu_buf_impl_t *db, void *tag) dbuf_evict(db); } else { VERIFY(arc_buf_remove_ref(db->db_buf, db) == 0); - if (!DBUF_IS_CACHEABLE(db)) + + /* + * A dbuf will be eligible for eviction if either the + * 'primarycache' property is set or a duplicate + * copy of this buffer is already cached in the arc. + * + * In the case of the 'primarycache' a buffer + * is considered for eviction if it matches the + * criteria set in the property. + * + * To decide if our buffer is considered a + * duplicate, we must call into the arc to determine + * if multiple buffers are referencing the same + * block on-disk. If so, then we simply evict + * ourselves. + */ + if (!DBUF_IS_CACHEABLE(db) || + arc_buf_eviction_needed(db->db_buf)) dbuf_clear(db); else mutex_exit(&db->db_mtx); @@ -2415,7 +2416,7 @@ dbuf_sync_leaf(dbuf_dirty_record_t *dr, dmu_tx_t *tx) dbuf_dirty_record_t **drp; ASSERT(*datap != NULL); - ASSERT3U(db->db_level, ==, 0); + ASSERT0(db->db_level); ASSERT3U(dn->dn_phys->dn_bonuslen, <=, DN_MAX_BONUSLEN); bcopy(*datap, DN_BONUS(dn->dn_phys), dn->dn_phys->dn_bonuslen); DB_DNODE_EXIT(db); @@ -2618,7 +2619,7 @@ dbuf_write_done(zio_t *zio, arc_buf_t *buf, void *vdb) uint64_t txg = zio->io_txg; dbuf_dirty_record_t **drp, *dr; - ASSERT3U(zio->io_error, ==, 0); + ASSERT0(zio->io_error); ASSERT(db->db_blkptr == bp); if (zio->io_flags & ZIO_FLAG_IO_REWRITE) { @@ -2824,9 +2825,10 @@ dbuf_write(dbuf_dirty_record_t *dr, arc_buf_t *data, dmu_tx_t *tx) } else { ASSERT(arc_released(data)); dr->dr_zio = arc_write(zio, os->os_spa, txg, - db->db_blkptr, data, DBUF_IS_L2CACHEABLE(db), &zp, - dbuf_write_ready, dbuf_write_done, db, - ZIO_PRIORITY_ASYNC_WRITE, ZIO_FLAG_MUSTSUCCEED, &zb); + db->db_blkptr, data, DBUF_IS_L2CACHEABLE(db), + DBUF_IS_L2COMPRESSIBLE(db), &zp, dbuf_write_ready, + dbuf_write_done, db, ZIO_PRIORITY_ASYNC_WRITE, + ZIO_FLAG_MUSTSUCCEED, &zb); } }