/*- * See the file LICENSE for redistribution information. * * Copyright (c) 1996, 1997, 1998, 1999 * Sleepycat Software. All rights reserved. */ #include "db_config.h" #ifndef lint static const char sccsid[] = "@(#)db_dup.c 11.11 (Sleepycat) 11/3/99"; #endif /* not lint */ #ifndef NO_SYSTEM_INCLUDES #include #include #endif #include "db_int.h" #include "db_page.h" #include "db_shash.h" #include "btree.h" #include "hash.h" #include "lock.h" #include "db_am.h" static int CDB___db_addpage __P((DBC *, PAGE **, db_indx_t *)); static int CDB___db_dsplit __P((DBC *, PAGE **, db_indx_t *, u_int32_t)); /* * CDB___db_dput -- * Put a duplicate item onto a duplicate page at the given index. * * PUBLIC: int CDB___db_dput __P((DBC *, DBT *, PAGE **, db_indx_t *)); */ int CDB___db_dput(dbc, dbt, pp, indxp) DBC *dbc; DBT *dbt; PAGE **pp; db_indx_t *indxp; { BOVERFLOW bo; DBT *data_dbtp, hdr_dbt, *hdr_dbtp; PAGE *pagep; db_indx_t size, isize; db_pgno_t pgno; int ret; /* * We need some access method independent threshold for when we put * a duplicate item onto an overflow page. */ if (dbt->size > 0.25 * dbc->dbp->pgsize) { if ((ret = CDB___db_poff(dbc, dbt, &pgno)) != 0) return (ret); UMRW(bo.unused1); B_TSET(bo.type, B_OVERFLOW, 0); UMRW(bo.unused2); bo.tlen = dbt->size; bo.pgno = pgno; hdr_dbt.data = &bo; hdr_dbt.size = isize = BOVERFLOW_SIZE; hdr_dbtp = &hdr_dbt; size = BOVERFLOW_PSIZE; data_dbtp = NULL; } else { size = BKEYDATA_PSIZE(dbt->size); isize = BKEYDATA_SIZE(dbt->size); hdr_dbtp = NULL; data_dbtp = dbt; } pagep = *pp; if (size > P_FREESPACE(pagep)) { if (*indxp == NUM_ENT(*pp) && NEXT_PGNO(*pp) == PGNO_INVALID) ret = CDB___db_addpage(dbc, pp, indxp); else ret = CDB___db_dsplit(dbc, pp, indxp, isize); if (ret != 0) /* * XXX * Pages not returned to free list. */ return (ret); pagep = *pp; } /* * Now, pagep references the page on which to insert and indx is the * the location to insert. */ if ((ret = CDB___db_pitem(dbc, pagep, (u_int32_t)*indxp, isize, hdr_dbtp, data_dbtp)) != 0) return (ret); (void)CDB_memp_fset(dbc->dbp->mpf, pagep, DB_MPOOL_DIRTY); return (0); } /* * CDB___db_drem -- * Remove a duplicate at the given index on the given page. * * PUBLIC: int CDB___db_drem __P((DBC *, PAGE **, u_int32_t)); */ int CDB___db_drem(dbc, pp, indx) DBC *dbc; PAGE **pp; u_int32_t indx; { PAGE *pagep; int ret; pagep = *pp; /* Check if we are freeing a big item. */ if (B_TYPE(GET_BKEYDATA(pagep, indx)->type) == B_OVERFLOW) { if ((ret = CDB___db_doff(dbc, GET_BOVERFLOW(pagep, indx)->pgno)) != 0) return (ret); ret = CDB___db_ditem(dbc, pagep, indx, BOVERFLOW_SIZE); } else ret = CDB___db_ditem(dbc, pagep, indx, BKEYDATA_SIZE(GET_BKEYDATA(pagep, indx)->len)); if (ret != 0) return (ret); if (NUM_ENT(pagep) == 0) { /* * If the page is emptied, then the page is freed and the pp * parameter is set to reference the next, locked page in the * duplicate chain, if one exists. If there was no such page, * then it is set to NULL. * * !!! * CDB___db_relink will set the dirty bit for us. */ if ((ret = CDB___db_relink(dbc, DB_REM_PAGE, pagep, pp, 0)) != 0) return (ret); if ((ret = CDB___db_free(dbc, pagep)) != 0) return (ret); } else (void)CDB_memp_fset(dbc->dbp->mpf, pagep, DB_MPOOL_DIRTY); return (0); } /* * CDB___db_dend -- * Find the last page in a set of offpage duplicates. * * PUBLIC: int CDB___db_dend __P((DBC *, db_pgno_t, PAGE **)); */ int CDB___db_dend(dbc, pgno, pp) DBC *dbc; db_pgno_t pgno; PAGE **pp; { DB *dbp; PAGE *h; int ret; dbp = dbc->dbp; /* * This implements DB_KEYLAST. The last page is returned in pp; pgno * should be the page number of the first page of the duplicate chain. * * *pp may be non-NULL -- if given a valid page use it. */ if (*pp != NULL) goto started; for (;;) { if ((ret = CDB_memp_fget(dbp->mpf, &pgno, 0, pp)) != 0) { (void)CDB___db_pgerr(dbp, pgno); return (ret); } started: h = *pp; if ((pgno = NEXT_PGNO(h)) == PGNO_INVALID) break; if ((ret = CDB_memp_fput(dbp->mpf, h, 0)) != 0) return (ret); } return (0); } /* * CDB___db_dsplit -- * Split a page of duplicates, calculating the split point based * on an element of size "size" being added at "*indxp". * On entry hp contains a pointer to the page-pointer of the original * page. On exit, it returns a pointer to the page containing "*indxp" * and "indxp" has been modified to reflect the index on the new page * where the element should be added. The function returns with * the page on which the insert should happen, not yet put. */ static int CDB___db_dsplit(dbc, hp, indxp, size) DBC *dbc; PAGE **hp; db_indx_t *indxp; u_int32_t size; { PAGE *h, *np, *tp; BKEYDATA *bk; DBT page_dbt; DB *dbp; size_t pgsize; db_indx_t halfbytes, i, indx, lastsum, nindex, oindex, s, sum; int did_indx, ret, t_ret; h = *hp; indx = *indxp; ret = 0; dbp = dbc->dbp; pgsize = dbp->pgsize; /* Create a temporary page to do compaction onto. */ if ((ret = CDB___os_malloc(pgsize, NULL, &tp)) != 0) return (ret); /* Create new page for the split. */ if ((ret = CDB___db_new(dbc, P_DUPLICATE, &np)) != 0) { CDB___os_free(tp, pgsize); return (ret); } P_INIT(np, pgsize, PGNO(np), PGNO(h), NEXT_PGNO(h), 0, P_DUPLICATE); P_INIT(tp, pgsize, PGNO(h), PREV_PGNO(h), PGNO(np), 0, P_DUPLICATE); /* Figure out the split point */ halfbytes = (pgsize - HOFFSET(h)) / 2; did_indx = 0; for (sum = 0, lastsum = 0, i = 0; i < NUM_ENT(h); i++) { if (i == indx) { sum += size; did_indx = 1; if (lastsum < halfbytes && sum >= halfbytes) { /* We've crossed the halfway point. */ if ((db_indx_t)(halfbytes - lastsum) < (db_indx_t)(sum - halfbytes)) { *hp = np; *indxp = 0; } else *indxp = i; break; } *indxp = i; lastsum = sum; } if (B_TYPE(GET_BKEYDATA(h, i)->type) == B_KEYDATA) sum += BKEYDATA_SIZE(GET_BKEYDATA(h, i)->len); else sum += BOVERFLOW_SIZE; if (lastsum < halfbytes && sum >= halfbytes) { /* We've crossed the halfway point. */ if ((db_indx_t)(sum - halfbytes) < (db_indx_t)(halfbytes - lastsum)) i++; break; } } /* * Check if we have set the return values of the index pointer and * page pointer. */ if (!did_indx) { *hp = np; *indxp = indx - i; } if (DB_LOGGING(dbc)) { page_dbt.size = dbp->pgsize; page_dbt.data = h; if ((ret = CDB___db_split_log(dbp->dbenv, dbc->txn, &LSN(h), 0, DB_SPLITOLD, dbp->log_fileid, PGNO(h), &page_dbt, &LSN(h))) != 0) { CDB___os_free(tp, pgsize); return (ret); } LSN(tp) = LSN(h); } /* * If it's a btree, adjust the cursors. * * i is the index of the first element to move onto the new page. */ if (dbp->type == DB_BTREE) CDB___bam_ca_split(dbp, PGNO(h), PGNO(h), PGNO(np), i, 0); if (dbp->type == DB_HASH) CDB___ham_ca_split(dbp, PGNO(h), PGNO(h), PGNO(np), i, 0); for (nindex = 0, oindex = i; oindex < NUM_ENT(h); oindex++) { bk = GET_BKEYDATA(h, oindex); if (B_TYPE(bk->type) == B_KEYDATA) s = BKEYDATA_SIZE(bk->len); else s = BOVERFLOW_SIZE; np->inp[nindex++] = HOFFSET(np) -= s; memcpy((u_int8_t *)np + HOFFSET(np), bk, s); NUM_ENT(np)++; } /* * Now do data compaction by copying the remaining stuff onto the * temporary page and then copying it back to the real page. */ for (nindex = 0, oindex = 0; oindex < i; oindex++) { bk = GET_BKEYDATA(h, oindex); if (B_TYPE(bk->type) == B_KEYDATA) s = BKEYDATA_SIZE(bk->len); else s = BOVERFLOW_SIZE; tp->inp[nindex++] = HOFFSET(tp) -= s; memcpy((u_int8_t *)tp + HOFFSET(tp), bk, s); NUM_ENT(tp)++; } /* * This page (the temporary) should be only half full, so we do two * memcpy's, one for the top of the page and one for the bottom of * the page. This way we avoid copying the middle which should be * about half a page. */ memcpy(h, tp, LOFFSET(tp)); memcpy((u_int8_t *)h + HOFFSET(tp), (u_int8_t *)tp + HOFFSET(tp), pgsize - HOFFSET(tp)); CDB___os_free(tp, pgsize); if (DB_LOGGING(dbc)) { /* * XXX * If either of these fails, are we leaving pages pinned? * Yes, but it seems like this happens in error case. */ page_dbt.size = pgsize; page_dbt.data = h; if ((ret = CDB___db_split_log(dbp->dbenv, dbc->txn, &LSN(h), 0, DB_SPLITNEW, dbp->log_fileid, PGNO(h), &page_dbt, &LSN(h))) != 0) return (ret); page_dbt.size = pgsize; page_dbt.data = np; if ((ret = CDB___db_split_log(dbp->dbenv, dbc->txn, &LSN(np), 0, DB_SPLITNEW, dbp->log_fileid, PGNO(np), &page_dbt, &LSN(np))) != 0) return (ret); } /* * Finally, if there was a next page after the page being * split, fix its prev pointer. */ if (np->next_pgno != PGNO_INVALID) ret = CDB___db_relink(dbc, DB_ADD_PAGE, np, NULL, 1); /* * Figure out if the location we're interested in is on the new * page, and if so, reset the callers' pointer. Push the other * page back to the store. */ if (*hp == h) t_ret = CDB_memp_fput(dbp->mpf, np, DB_MPOOL_DIRTY); else t_ret = CDB_memp_fput(dbp->mpf, h, DB_MPOOL_DIRTY); return (ret != 0 ? ret : t_ret); } /* * CDB___db_ditem -- * Remove an item from a page. * * PUBLIC: int CDB___db_ditem __P((DBC *, PAGE *, u_int32_t, u_int32_t)); */ int CDB___db_ditem(dbc, pagep, indx, nbytes) DBC *dbc; PAGE *pagep; u_int32_t indx, nbytes; { DB *dbp; DBT ldbt; db_indx_t cnt, offset; int ret; u_int8_t *from; dbp = dbc->dbp; if (DB_LOGGING(dbc)) { ldbt.data = P_ENTRY(pagep, indx); ldbt.size = nbytes; if ((ret = CDB___db_addrem_log(dbp->dbenv, dbc->txn, &LSN(pagep), 0, DB_REM_DUP, dbp->log_fileid, PGNO(pagep), (u_int32_t)indx, nbytes, &ldbt, NULL, &LSN(pagep))) != 0) return (ret); } /* * If there's only a single item on the page, we don't have to * work hard. */ if (NUM_ENT(pagep) == 1) { NUM_ENT(pagep) = 0; HOFFSET(pagep) = dbp->pgsize; return (0); } /* * Pack the remaining key/data items at the end of the page. Use * memmove(3), the regions may overlap. */ from = (u_int8_t *)pagep + HOFFSET(pagep); memmove(from + nbytes, from, pagep->inp[indx] - HOFFSET(pagep)); HOFFSET(pagep) += nbytes; /* Adjust the indices' offsets. */ offset = pagep->inp[indx]; for (cnt = 0; cnt < NUM_ENT(pagep); ++cnt) if (pagep->inp[cnt] < offset) pagep->inp[cnt] += nbytes; /* Shift the indices down. */ --NUM_ENT(pagep); if (indx != NUM_ENT(pagep)) memmove(&pagep->inp[indx], &pagep->inp[indx + 1], sizeof(db_indx_t) * (NUM_ENT(pagep) - indx)); return (0); } /* * CDB___db_pitem -- * Put an item on a page. * * PUBLIC: int CDB___db_pitem * PUBLIC: __P((DBC *, PAGE *, u_int32_t, u_int32_t, DBT *, DBT *)); */ int CDB___db_pitem(dbc, pagep, indx, nbytes, hdr, data) DBC *dbc; PAGE *pagep; u_int32_t indx; u_int32_t nbytes; DBT *hdr, *data; { DB *dbp; BKEYDATA bk; DBT thdr; int ret; u_int8_t *p; /* * Put a single item onto a page. The logic figuring out where to * insert and whether it fits is handled in the caller. All we do * here is manage the page shuffling. We cheat a little bit in that * we don't want to copy the dbt on a normal put twice. If hdr is * NULL, we create a BKEYDATA structure on the page, otherwise, just * copy the caller's information onto the page. * * This routine is also used to put entries onto the page where the * entry is pre-built, e.g., during recovery. In this case, the hdr * will point to the entry, and the data argument will be NULL. * * !!! * There's a tremendous potential for off-by-one errors here, since * the passed in header sizes must be adjusted for the structure's * placeholder for the trailing variable-length data field. */ dbp = dbc->dbp; if (DB_LOGGING(dbc)) if ((ret = CDB___db_addrem_log(dbp->dbenv, dbc->txn, &LSN(pagep), 0, DB_ADD_DUP, dbp->log_fileid, PGNO(pagep), (u_int32_t)indx, nbytes, hdr, data, &LSN(pagep))) != 0) return (ret); if (hdr == NULL) { B_TSET(bk.type, B_KEYDATA, 0); bk.len = data == NULL ? 0 : data->size; thdr.data = &bk; thdr.size = SSZA(BKEYDATA, data); hdr = &thdr; } /* Adjust the index table, then put the item on the page. */ if (indx != NUM_ENT(pagep)) memmove(&pagep->inp[indx + 1], &pagep->inp[indx], sizeof(db_indx_t) * (NUM_ENT(pagep) - indx)); HOFFSET(pagep) -= nbytes; pagep->inp[indx] = HOFFSET(pagep); ++NUM_ENT(pagep); p = P_ENTRY(pagep, indx); memcpy(p, hdr->data, hdr->size); if (data != NULL) memcpy(p + hdr->size, data->data, data->size); return (0); } /* * CDB___db_relink -- * Relink around a deleted page. * * PUBLIC: int CDB___db_relink __P((DBC *, u_int32_t, PAGE *, PAGE **, int)); */ int CDB___db_relink(dbc, add_rem, pagep, new_next, needlock) DBC *dbc; u_int32_t add_rem; PAGE *pagep, **new_next; int needlock; { DB *dbp; PAGE *np, *pp; DB_LOCK npl, ppl; DB_LSN *nlsnp, *plsnp; int ret; ret = 0; np = pp = NULL; npl.off = ppl.off = LOCK_INVALID; nlsnp = plsnp = NULL; dbp = dbc->dbp; /* * Retrieve and lock the one/two pages. For a remove, we may need * two pages (the before and after). For an add, we only need one * because, the split took care of the prev. */ if (pagep->next_pgno != PGNO_INVALID) { if (needlock && (ret = CDB___db_lget(dbc, 0, pagep->next_pgno, DB_LOCK_WRITE, 0, &npl)) != 0) goto err; if ((ret = CDB_memp_fget(dbp->mpf, &pagep->next_pgno, 0, &np)) != 0) { (void)CDB___db_pgerr(dbp, pagep->next_pgno); goto err; } nlsnp = &np->lsn; } if (add_rem == DB_REM_PAGE && pagep->prev_pgno != PGNO_INVALID) { if (needlock && (ret = CDB___db_lget(dbc, 0, pagep->prev_pgno, DB_LOCK_WRITE, 0, &ppl)) != 0) goto err; if ((ret = CDB_memp_fget(dbp->mpf, &pagep->prev_pgno, 0, &pp)) != 0) { (void)CDB___db_pgerr(dbp, pagep->next_pgno); goto err; } plsnp = &pp->lsn; } /* Log the change. */ if (DB_LOGGING(dbc)) { if ((ret = CDB___db_relink_log(dbp->dbenv, dbc->txn, &pagep->lsn, 0, add_rem, dbp->log_fileid, pagep->pgno, &pagep->lsn, pagep->prev_pgno, plsnp, pagep->next_pgno, nlsnp)) != 0) goto err; if (np != NULL) np->lsn = pagep->lsn; if (pp != NULL) pp->lsn = pagep->lsn; } /* * Modify and release the two pages. * * !!! * The parameter new_next gets set to the page following the page we * are removing. If there is no following page, then new_next gets * set to NULL. */ if (np != NULL) { if (add_rem == DB_ADD_PAGE) np->prev_pgno = pagep->pgno; else np->prev_pgno = pagep->prev_pgno; if (new_next == NULL) ret = CDB_memp_fput(dbp->mpf, np, DB_MPOOL_DIRTY); else { *new_next = np; ret = CDB_memp_fset(dbp->mpf, np, DB_MPOOL_DIRTY); } if (ret != 0) goto err; if (needlock) (void)__TLPUT(dbc, npl); } else if (new_next != NULL) *new_next = NULL; if (pp != NULL) { pp->next_pgno = pagep->next_pgno; if ((ret = CDB_memp_fput(dbp->mpf, pp, DB_MPOOL_DIRTY)) != 0) goto err; if (needlock) (void)__TLPUT(dbc, ppl); } return (0); err: if (np != NULL) (void)CDB_memp_fput(dbp->mpf, np, 0); if (needlock && npl.off != LOCK_INVALID) (void)__TLPUT(dbc, npl); if (pp != NULL) (void)CDB_memp_fput(dbp->mpf, pp, 0); if (needlock && ppl.off != LOCK_INVALID) (void)__TLPUT(dbc, ppl); return (ret); } /* * CDB___db_ddup -- * Delete an offpage chain of duplicates. * * PUBLIC: int CDB___db_ddup __P((DBC *, db_pgno_t)); */ int CDB___db_ddup(dbc, pgno) DBC *dbc; db_pgno_t pgno; { DB *dbp; PAGE *pagep; DBT tmp_dbt; int ret; dbp = dbc->dbp; do { if ((ret = CDB_memp_fget(dbp->mpf, &pgno, 0, &pagep)) != 0) { (void)CDB___db_pgerr(dbp, pgno); return (ret); } if (DB_LOGGING(dbc)) { tmp_dbt.data = pagep; tmp_dbt.size = dbp->pgsize; if ((ret = CDB___db_split_log(dbp->dbenv, dbc->txn, &LSN(pagep), 0, DB_SPLITOLD, dbp->log_fileid, PGNO(pagep), &tmp_dbt, &LSN(pagep))) != 0) return (ret); } pgno = pagep->next_pgno; if ((ret = CDB___db_free(dbc, pagep)) != 0) return (ret); } while (pgno != PGNO_INVALID); return (0); } /* * CDB___db_addpage -- * Create a new page and link it onto the next_pgno field of the * current page. */ static int CDB___db_addpage(dbc, hp, indxp) DBC *dbc; PAGE **hp; db_indx_t *indxp; { DB *dbp; PAGE *newpage; int ret; dbp = dbc->dbp; if ((ret = CDB___db_new(dbc, P_DUPLICATE, &newpage)) != 0) return (ret); if (DB_LOGGING(dbc)) { if ((ret = CDB___db_addpage_log(dbp->dbenv, dbc->txn, &LSN(*hp), 0, dbp->log_fileid, PGNO(*hp), &LSN(*hp), PGNO(newpage), &LSN(newpage))) != 0) { return (ret); } LSN(newpage) = LSN(*hp); } PREV_PGNO(newpage) = PGNO(*hp); NEXT_PGNO(*hp) = PGNO(newpage); if ((ret = CDB_memp_fput(dbp->mpf, *hp, DB_MPOOL_DIRTY)) != 0) return (ret); *hp = newpage; *indxp = 0; return (0); } /* * CDB___db_dsearch -- * Search a set of duplicates for the proper position for a new duplicate. * * + pgno is the page number of the page on which to begin searching. * Since we can continue duplicate searches, it might not be the first * page. * * + If we are continuing a search, then *pp may be non-NULL in which * case we do not have to retrieve the page. * * + If we are continuing a search, then *indxp contains the first * on pgno of where we should begin the search. * * NOTE: if there is no comparison function, then continuing is * meaningless, and *pp should always be NULL and *indxp will be * ignored. * * 3 return values:: * * + pp is the returned page pointer of where this element should go. * + indxp is the returned index on that page * + cmpp is the returned final comparison result. * * PUBLIC: int CDB___db_dsearch __P((DBC *, * PUBLIC: int, DBT *, db_pgno_t, db_indx_t *, PAGE **, int *)); */ int CDB___db_dsearch(dbc, is_insert, dbt, pgno, indxp, pp, cmpp) DBC *dbc; int is_insert, *cmpp; DBT *dbt; db_pgno_t pgno; db_indx_t *indxp; PAGE **pp; { DB *dbp; PAGE *h; db_indx_t base, indx, lim; db_pgno_t save_pgno; int ret, firstpg; dbp = dbc->dbp; if (dbp->dup_compare == NULL) { /* * We may have been given a valid page, but we may not be * able to use it. The problem is that the application is * doing a join and we're trying to continue the search, * but since the items aren't sorted, we can't. Discard * the page if it's not the one we're going to start with * anyway. */ if (*pp != NULL && (*pp)->pgno != pgno) { if ((ret = CDB_memp_fput(dbp->mpf, *pp, 0)) != 0) return (ret); *pp = NULL; } /* * If no duplicate function is specified, just go to the end * of the duplicate set. */ if (is_insert) { if ((ret = CDB___db_dend(dbc, pgno, pp)) != 0) return (ret); *indxp = NUM_ENT(*pp); return (0); } /* * We are looking for a specific duplicate, so do a linear * search. */ firstpg = 1; /* Is this the first page we're searching? */ if (*pp != NULL) goto nocmp_started; for (;;) { if ((ret = CDB_memp_fget(dbp->mpf, &pgno, 0, pp)) != 0) goto pg_err; nocmp_started: h = *pp; base = (F_ISSET(dbc, DBC_CONTINUE) && (firstpg == 1)) ? *indxp : 0; for (*indxp = base; *indxp < NUM_ENT(h); ++*indxp) { if ((*cmpp = CDB___bam_cmp(dbp, dbt, h, *indxp, CDB___bam_defcmp)) != 0) continue; /* * The duplicate may have already been deleted, * if it's a btree page, in which case we skip * it. */ if (dbp->type == DB_BTREE && B_DISSET(GET_BKEYDATA(h, *indxp)->type)) continue; return (0); } if ((pgno = h->next_pgno) == PGNO_INVALID) break; /* * Moving on to another page; make sure we * search from the beginning rather than *indxp. */ firstpg = 0; if ((ret = CDB_memp_fput(dbp->mpf, h, 0)) != 0) return (ret); } *cmpp = 1; /* We didn't succeed... */ return (0); } /* * We have a comparison routine, i.e., the duplicates are sorted. * Walk through the chain of duplicates, checking the last entry * on each page to decide if it's the page we want to search. * * *pp may be non-NULL -- if we were given a valid page (e.g., are * in mid-search), then use the provided page. */ if (*pp != NULL) goto cmp_started; for (;;) { if ((ret = CDB_memp_fget(dbp->mpf, &pgno, 0, pp)) != 0) goto pg_err; cmp_started: h = *pp; if ((pgno = h->next_pgno) == PGNO_INVALID || CDB___bam_cmp(dbp, dbt, h, h->entries - 1, dbp->dup_compare) <= 0) break; /* * Even when continuing a search, make sure we don't skip * entries on a new page */ *indxp = 0; if ((ret = CDB_memp_fput(dbp->mpf, h, 0)) != 0) return (ret); } /* Next, do a binary search on the page. */ base = F_ISSET(dbc, DBC_CONTINUE) ? *indxp : 0; for (lim = NUM_ENT(h) - base; lim != 0; lim >>= 1) { indx = base + (lim >> 1); if ((*cmpp = CDB___bam_cmp(dbp, dbt, h, indx, dbp->dup_compare)) == 0) { /* * If DBC_CONTINUE is set, we want to make sure * we return the first duplicate duplicate * with index larger than *indxp. Search * backwards until we're sure we have such. * * XXX: For the various uses of CDB___db_dsearch * outside of joins, what's the correct behavior * in the presence of duplicate duplicates? Someone * should ponder this, and perhaps remove the * following check and/or add different ones. */ if (F_ISSET(dbc, DBC_CONTINUE)) { if (indx != 0) { for (indx--; indx >= *indxp; indx--) { if (CDB___bam_cmp(dbp, dbt, h, indx, dbp->dup_compare) != 0) break; /* * Check for exit; if indx * == 0 here, we know we * want to return 0 without * decrementing into an * overflow condition (indx * is unsigned), so jump out. */ if (indx == 0) goto postinc; } /* * We've decremented one too far; * either indx is now less than * *indxp, or it's pointing at a dup * that doesn't match. */ indx++; } } postinc: *indxp = indx; if (dbp->type != DB_BTREE || !B_DISSET(GET_BKEYDATA(h, *indxp)->type)) return (0); goto check_delete; } if (*cmpp > 0) { base = indx + 1; lim--; } } /* * Base references the smallest index larger than the supplied DBT's * data item, potentially both 0 and NUM_ENT. */ *indxp = base; return (0); check_delete: /* * The duplicate may have already been deleted, if it's a btree page, * in which case we wander around, hoping to find an entry that hasn't * been deleted. First, wander in a forwardly direction. */ save_pgno = (*pp)->pgno; for (++*indxp;;) { for (; *indxp < NUM_ENT(h); ++*indxp) { if ((*cmpp = CDB___bam_cmp(dbp, dbt, h, *indxp, dbp->dup_compare)) != 0) goto check_delete_rev; if (!B_DISSET(GET_BKEYDATA(h, *indxp)->type)) return (0); } if ((pgno = h->next_pgno) == PGNO_INVALID) break; if ((ret = CDB_memp_fput(dbp->mpf, h, 0)) != 0) return (ret); if ((ret = CDB_memp_fget(dbp->mpf, &pgno, 0, pp)) != 0) goto pg_err; h = *pp; *indxp = 0; } check_delete_rev: /* Go back to where we started, and wander in a backwardly direction. */ if (h->pgno != save_pgno) { if ((ret = CDB_memp_fput(dbp->mpf, h, 0)) != 0) return (ret); if ((ret = CDB_memp_fget(dbp->mpf, &save_pgno, 0, pp)) != 0) goto pg_err; h = *pp; } for (;;) { while (*indxp > 0) { --*indxp; if ((*cmpp = CDB___bam_cmp(dbp, dbt, h, *indxp, dbp->dup_compare)) != 0) goto check_delete_fail; if (!B_DISSET(GET_BKEYDATA(h, *indxp)->type)) return (0); } if ((pgno = h->prev_pgno) == PGNO_INVALID) break; if ((ret = CDB_memp_fput(dbp->mpf, h, 0)) != 0) return (ret); if ((ret = CDB_memp_fget(dbp->mpf, &pgno, 0, pp)) != 0) goto pg_err; h = *pp; *indxp = NUM_ENT(h); } check_delete_fail: *cmpp = 1; /* We didn't succeed... */ return (0); pg_err: CDB___db_pgerr(dbp, pgno); return (ret); }