X-Git-Url: https://git.camperquake.de/gitweb.cgi?a=blobdiff_plain;f=module%2Fzfs%2Fdmu_traverse.c;h=1c39723714379b2596c233f9b1ec6f0b192df70f;hb=294f68063b49c06d3118d51016811063e69cf97a;hp=980c1aa2f8222f0acb959e2bb134ef2e5f587c79;hpb=9ae529ec5dbdc828ff8326beae58062971d74b2e;p=zfs.git diff --git a/module/zfs/dmu_traverse.c b/module/zfs/dmu_traverse.c index 980c1aa..1c39723 100644 --- a/module/zfs/dmu_traverse.c +++ b/module/zfs/dmu_traverse.c @@ -62,7 +62,9 @@ typedef struct traverse_data { } traverse_data_t; static int traverse_dnode(traverse_data_t *td, const dnode_phys_t *dnp, - arc_buf_t *buf, uint64_t objset, uint64_t object); + uint64_t objset, uint64_t object); +static void prefetch_dnode_metadata(traverse_data_t *td, const dnode_phys_t *, + uint64_t objset, uint64_t object); static int traverse_zil_block(zilog_t *zilog, blkptr_t *bp, void *arg, uint64_t claim_txg) @@ -79,7 +81,7 @@ traverse_zil_block(zilog_t *zilog, blkptr_t *bp, void *arg, uint64_t claim_txg) SET_BOOKMARK(&zb, td->td_objset, ZB_ZIL_OBJECT, ZB_ZIL_LEVEL, bp->blk_cksum.zc_word[ZIL_ZC_SEQ]); - (void) td->td_func(td->td_spa, zilog, bp, NULL, &zb, NULL, td->td_arg); + (void) td->td_func(td->td_spa, zilog, bp, &zb, NULL, td->td_arg); return (0); } @@ -103,7 +105,7 @@ traverse_zil_record(zilog_t *zilog, lr_t *lrc, void *arg, uint64_t claim_txg) SET_BOOKMARK(&zb, td->td_objset, lr->lr_foid, ZB_ZIL_LEVEL, lr->lr_offset / BP_GET_LSIZE(bp)); - (void) td->td_func(td->td_spa, zilog, bp, NULL, &zb, NULL, + (void) td->td_func(td->td_spa, zilog, bp, &zb, NULL, td->td_arg); } return (0); @@ -174,13 +176,37 @@ static void traverse_pause(traverse_data_t *td, const zbookmark_t *zb) { ASSERT(td->td_resume != NULL); - ASSERT3U(zb->zb_level, ==, 0); + ASSERT0(zb->zb_level); bcopy(zb, td->td_resume, sizeof (*td->td_resume)); } +static void +traverse_prefetch_metadata(traverse_data_t *td, + const blkptr_t *bp, const zbookmark_t *zb) +{ + uint32_t flags = ARC_NOWAIT | ARC_PREFETCH; + + if (!(td->td_flags & TRAVERSE_PREFETCH_METADATA)) + return; + /* + * If we are in the process of resuming, don't prefetch, because + * some children will not be needed (and in fact may have already + * been freed). + */ + if (td->td_resume != NULL && !ZB_IS_ZERO(td->td_resume)) + return; + if (BP_IS_HOLE(bp) || bp->blk_birth <= td->td_min_txg) + return; + if (BP_GET_LEVEL(bp) == 0 && BP_GET_TYPE(bp) != DMU_OT_DNODE) + return; + + (void) arc_read(NULL, td->td_spa, bp, NULL, NULL, + ZIO_PRIORITY_ASYNC_READ, ZIO_FLAG_CANFAIL, &flags, zb); +} + static int traverse_visitbp(traverse_data_t *td, const dnode_phys_t *dnp, - arc_buf_t *pbuf, blkptr_t *bp, const zbookmark_t *zb) + const blkptr_t *bp, const zbookmark_t *zb) { zbookmark_t czb; int err = 0, lasterr = 0; @@ -201,8 +227,7 @@ traverse_visitbp(traverse_data_t *td, const dnode_phys_t *dnp, } if (BP_IS_HOLE(bp)) { - err = td->td_func(td->td_spa, NULL, NULL, pbuf, zb, dnp, - td->td_arg); + err = td->td_func(td->td_spa, NULL, NULL, zb, dnp, td->td_arg); return (err); } @@ -222,7 +247,7 @@ traverse_visitbp(traverse_data_t *td, const dnode_phys_t *dnp, } if (td->td_flags & TRAVERSE_PRE) { - err = td->td_func(td->td_spa, NULL, bp, pbuf, zb, dnp, + err = td->td_func(td->td_spa, NULL, bp, zb, dnp, td->td_arg); if (err == TRAVERSE_VISIT_NO_CHILDREN) return (0); @@ -238,19 +263,25 @@ traverse_visitbp(traverse_data_t *td, const dnode_phys_t *dnp, blkptr_t *cbp; int epb = BP_GET_LSIZE(bp) >> SPA_BLKPTRSHIFT; - err = dsl_read(NULL, td->td_spa, bp, pbuf, - arc_getbuf_func, &buf, + err = arc_read(NULL, td->td_spa, bp, arc_getbuf_func, &buf, ZIO_PRIORITY_ASYNC_READ, ZIO_FLAG_CANFAIL, &flags, zb); if (err) return (err); + cbp = buf->b_data; + + for (i = 0; i < epb; i++) { + SET_BOOKMARK(&czb, zb->zb_objset, zb->zb_object, + zb->zb_level - 1, + zb->zb_blkid * epb + i); + traverse_prefetch_metadata(td, &cbp[i], &czb); + } /* recursively visitbp() blocks below this */ - cbp = buf->b_data; - for (i = 0; i < epb; i++, cbp++) { + for (i = 0; i < epb; i++) { SET_BOOKMARK(&czb, zb->zb_objset, zb->zb_object, zb->zb_level - 1, zb->zb_blkid * epb + i); - err = traverse_visitbp(td, dnp, buf, cbp, &czb); + err = traverse_visitbp(td, dnp, &cbp[i], &czb); if (err) { if (!hard) break; @@ -262,16 +293,20 @@ traverse_visitbp(traverse_data_t *td, const dnode_phys_t *dnp, int i; int epb = BP_GET_LSIZE(bp) >> DNODE_SHIFT; - err = dsl_read(NULL, td->td_spa, bp, pbuf, - arc_getbuf_func, &buf, + err = arc_read(NULL, td->td_spa, bp, arc_getbuf_func, &buf, ZIO_PRIORITY_ASYNC_READ, ZIO_FLAG_CANFAIL, &flags, zb); if (err) return (err); + dnp = buf->b_data; + + for (i = 0; i < epb; i++) { + prefetch_dnode_metadata(td, &dnp[i], zb->zb_objset, + zb->zb_blkid * epb + i); + } /* recursively visitbp() blocks below this */ - dnp = buf->b_data; - for (i = 0; i < epb; i++, dnp++) { - err = traverse_dnode(td, dnp, buf, zb->zb_objset, + for (i = 0; i < epb; i++) { + err = traverse_dnode(td, &dnp[i], zb->zb_objset, zb->zb_blkid * epb + i); if (err) { if (!hard) @@ -284,15 +319,23 @@ traverse_visitbp(traverse_data_t *td, const dnode_phys_t *dnp, objset_phys_t *osp; dnode_phys_t *dnp; - err = dsl_read_nolock(NULL, td->td_spa, bp, - arc_getbuf_func, &buf, + err = arc_read(NULL, td->td_spa, bp, arc_getbuf_func, &buf, ZIO_PRIORITY_ASYNC_READ, ZIO_FLAG_CANFAIL, &flags, zb); if (err) return (err); osp = buf->b_data; dnp = &osp->os_meta_dnode; - err = traverse_dnode(td, dnp, buf, zb->zb_objset, + prefetch_dnode_metadata(td, dnp, zb->zb_objset, + DMU_META_DNODE_OBJECT); + if (arc_buf_size(buf) >= sizeof (objset_phys_t)) { + prefetch_dnode_metadata(td, &osp->os_userused_dnode, + zb->zb_objset, DMU_USERUSED_OBJECT); + prefetch_dnode_metadata(td, &osp->os_groupused_dnode, + zb->zb_objset, DMU_USERUSED_OBJECT); + } + + err = traverse_dnode(td, dnp, zb->zb_objset, DMU_META_DNODE_OBJECT); if (err && hard) { lasterr = err; @@ -300,7 +343,7 @@ traverse_visitbp(traverse_data_t *td, const dnode_phys_t *dnp, } if (err == 0 && arc_buf_size(buf) >= sizeof (objset_phys_t)) { dnp = &osp->os_userused_dnode; - err = traverse_dnode(td, dnp, buf, zb->zb_objset, + err = traverse_dnode(td, dnp, zb->zb_objset, DMU_USERUSED_OBJECT); } if (err && hard) { @@ -309,7 +352,7 @@ traverse_visitbp(traverse_data_t *td, const dnode_phys_t *dnp, } if (err == 0 && arc_buf_size(buf) >= sizeof (objset_phys_t)) { dnp = &osp->os_groupused_dnode; - err = traverse_dnode(td, dnp, buf, zb->zb_objset, + err = traverse_dnode(td, dnp, zb->zb_objset, DMU_GROUPUSED_OBJECT); } } @@ -319,8 +362,7 @@ traverse_visitbp(traverse_data_t *td, const dnode_phys_t *dnp, post: if (err == 0 && lasterr == 0 && (td->td_flags & TRAVERSE_POST)) { - err = td->td_func(td->td_spa, NULL, bp, pbuf, zb, dnp, - td->td_arg); + err = td->td_func(td->td_spa, NULL, bp, zb, dnp, td->td_arg); if (err == ERESTART) pause = B_TRUE; } @@ -334,9 +376,27 @@ post: return (err != 0 ? err : lasterr); } +static void +prefetch_dnode_metadata(traverse_data_t *td, const dnode_phys_t *dnp, + uint64_t objset, uint64_t object) +{ + int j; + zbookmark_t czb; + + for (j = 0; j < dnp->dn_nblkptr; j++) { + SET_BOOKMARK(&czb, objset, object, dnp->dn_nlevels - 1, j); + traverse_prefetch_metadata(td, &dnp->dn_blkptr[j], &czb); + } + + if (dnp->dn_flags & DNODE_FLAG_SPILL_BLKPTR) { + SET_BOOKMARK(&czb, objset, object, 0, DMU_SPILL_BLKID); + traverse_prefetch_metadata(td, &dnp->dn_spill, &czb); + } +} + static int traverse_dnode(traverse_data_t *td, const dnode_phys_t *dnp, - arc_buf_t *buf, uint64_t objset, uint64_t object) + uint64_t objset, uint64_t object) { int j, err = 0, lasterr = 0; zbookmark_t czb; @@ -344,8 +404,7 @@ traverse_dnode(traverse_data_t *td, const dnode_phys_t *dnp, for (j = 0; j < dnp->dn_nblkptr; j++) { SET_BOOKMARK(&czb, objset, object, dnp->dn_nlevels - 1, j); - err = traverse_visitbp(td, dnp, buf, - (blkptr_t *)&dnp->dn_blkptr[j], &czb); + err = traverse_visitbp(td, dnp, &dnp->dn_blkptr[j], &czb); if (err) { if (!hard) break; @@ -354,10 +413,8 @@ traverse_dnode(traverse_data_t *td, const dnode_phys_t *dnp, } if (dnp->dn_flags & DNODE_FLAG_SPILL_BLKPTR) { - SET_BOOKMARK(&czb, objset, - object, 0, DMU_SPILL_BLKID); - err = traverse_visitbp(td, dnp, buf, - (blkptr_t *)&dnp->dn_spill, &czb); + SET_BOOKMARK(&czb, objset, object, 0, DMU_SPILL_BLKID); + err = traverse_visitbp(td, dnp, &dnp->dn_spill, &czb); if (err) { if (!hard) return (err); @@ -370,8 +427,7 @@ traverse_dnode(traverse_data_t *td, const dnode_phys_t *dnp, /* ARGSUSED */ static int traverse_prefetcher(spa_t *spa, zilog_t *zilog, const blkptr_t *bp, - arc_buf_t *pbuf, const zbookmark_t *zb, const dnode_phys_t *dnp, - void *arg) + const zbookmark_t *zb, const dnode_phys_t *dnp, void *arg) { prefetch_data_t *pfd = arg; uint32_t aflags = ARC_NOWAIT | ARC_PREFETCH; @@ -392,10 +448,8 @@ traverse_prefetcher(spa_t *spa, zilog_t *zilog, const blkptr_t *bp, cv_broadcast(&pfd->pd_cv); mutex_exit(&pfd->pd_mtx); - (void) dsl_read(NULL, spa, bp, pbuf, NULL, NULL, - ZIO_PRIORITY_ASYNC_READ, - ZIO_FLAG_CANFAIL | ZIO_FLAG_SPECULATIVE, - &aflags, zb); + (void) arc_read(NULL, spa, bp, NULL, NULL, ZIO_PRIORITY_ASYNC_READ, + ZIO_FLAG_CANFAIL | ZIO_FLAG_SPECULATIVE, &aflags, zb); return (0); } @@ -413,7 +467,7 @@ traverse_prefetch_thread(void *arg) SET_BOOKMARK(&czb, td.td_objset, ZB_ROOT_OBJECT, ZB_ROOT_LEVEL, ZB_ROOT_BLKID); - (void) traverse_visitbp(&td, NULL, NULL, td.td_rootbp, &czb); + (void) traverse_visitbp(&td, NULL, td.td_rootbp, &czb); mutex_enter(&td_main->td_pfd->pd_mtx); td_main->td_pfd->pd_exited = B_TRUE; @@ -438,6 +492,12 @@ traverse_impl(spa_t *spa, dsl_dataset_t *ds, uint64_t objset, blkptr_t *rootbp, ASSERT(ds == NULL || objset == ds->ds_object); ASSERT(!(flags & TRAVERSE_PRE) || !(flags & TRAVERSE_POST)); + /* + * The data prefetching mechanism (the prefetch thread) is incompatible + * with resuming from a bookmark. + */ + ASSERT(resume == NULL || !(flags & TRAVERSE_PREFETCH_DATA)); + td = kmem_alloc(sizeof(traverse_data_t), KM_PUSHPAGE); pd = kmem_zalloc(sizeof(prefetch_data_t), KM_PUSHPAGE); czb = kmem_alloc(sizeof(zbookmark_t), KM_PUSHPAGE); @@ -468,14 +528,14 @@ traverse_impl(spa_t *spa, dsl_dataset_t *ds, uint64_t objset, blkptr_t *rootbp, traverse_zil(td, &os->os_zil_header); } - if (!(flags & TRAVERSE_PREFETCH) || + if (!(flags & TRAVERSE_PREFETCH_DATA) || 0 == taskq_dispatch(system_taskq, traverse_prefetch_thread, td, TQ_NOQUEUE)) pd->pd_exited = B_TRUE; SET_BOOKMARK(czb, td->td_objset, ZB_ROOT_OBJECT, ZB_ROOT_LEVEL, ZB_ROOT_BLKID); - err = traverse_visitbp(td, NULL, NULL, rootbp, czb); + err = traverse_visitbp(td, NULL, rootbp, czb); mutex_enter(&pd->pd_mtx); pd->pd_cancel = B_TRUE;