[Date Prev][Date Next] [Chronological] [Thread] [Top]

Re: (ITS#7594) incomplete MDB subdb cursor cleanup



h.b.furuseth@usit.uio.no wrote:
> Howard Chu writes:
>> h.b.furuseth@usit.uio.no wrote:
>>> If a cursor is at a clean subDB page, and its current item is deleted
>>> by another cursor, then MDB_GET_CURRENT returns the deleted item.
>>
>> Fixed now in mdb.master. The subcursor on the clean page will be invalidated
>> when the page is touched.
>
> MDB_NEXT/MDB_PREV crash on assert(mc->mc_flags & C_INITIALIZED)
> since commit e31c7d3b31d8d5073195e31b283e8ccb46bd13cf, with the
> patch below to the test program.

OK, I've reverted that commit.

> mdb.master output with patched test program:
>
> With clean page:
> Name:	Key  -> Data  (mv_data[mv_len]):
> m2:	a[2] -> x[2]
> mc/del:	a[2] -> [0]
> a.out: mdb.c:4386: mdb_cursor_next: Assertion `mc->mc_flags & 0x01' failed.

No longer asserts. Also, [0] was the wrong result here. Also fixed.

> Branch mdb/its7594 output:
>
> With clean page:
> Name:	Key  -> Data  (mv_data[mv_len]):
> m2:	a[2] -> x[2]
> mc/del:	a[2] -> [0]
> its7594a.c:48: mdb_cursor_get(m2, &key, &data, MDB_NEXT): MDB_BAD_CURSOR: Cursor is unusable
>
> That assert is OK since it's the test program asserting success.
>
> --- its7594.c
> +++ its7594a.c
> @@ -28,4 +28,6 @@
>       E(mdb_cursor_put(mc, STR2VAL("a"), STR2VAL("x"), 0));
>       E(mdb_cursor_put(mc, STR2VAL("a"), STR2VAL("y"), 0));
> +    E(mdb_cursor_put(mc, STR2VAL("b"), STR2VAL("z"), 0));
> +    E(mdb_cursor_put(mc, STR2VAL("c"), STR2VAL("w"), 0));
>
>       if (argc < 2) {
> @@ -42,8 +44,8 @@
>
>       E(mdb_cursor_del(mc, 0));
> -    E(mdb_cursor_get(m2, &key, &data, MDB_GET_CURRENT));
> -    SHOW("m2/del"); /* Should output the same as... */
> -    E(mdb_cursor_get(mc, &key, &data, MDB_GET_CURRENT));
> -    SHOW("mc/del"); /* ...this: "a[2] -> [0]". */
> +    E(mdb_cursor_get(mc, &key, &data, MDB_NEXT));
> +    SHOW("mc/del"); /* Outputs "a[2] -> [0]"... */
> +    E(mdb_cursor_get(m2, &key, &data, MDB_NEXT));
> +    SHOW("m2/del"); /* ...and this should output the same. */
>
>       mdb_txn_abort(txn);
>


-- 
   -- Howard Chu
   CTO, Symas Corp.           http://www.symas.com
   Director, Highland Sun     http://highlandsun.com/hyc/
   Chief Architect, OpenLDAP  http://www.openldap.org/project/