Re: Bug in huge simplehash - Mailing list pgsql-hackers

Good day, hackers.

Our client caught process stuck in tuplehash_grow. There was a query
like
`select ts, count(*) from really_huge_partitioned_table group by ts`,
and
planner decided to use hash aggregation.

Backtrace shows that oldsize were 2147483648 at the moment. While
newsize
were optimized, looks like it were SH_MAX_SIZE.

#0  0x0000000000603d0c in tuplehash_grow (tb=0x7f18c3c284c8,
newsize=<optimized out>) at ../../../src/include/lib/simplehash.h:457
         hash = 2147483654
         startelem = 1
         curelem = 1
         oldentry = 0x7f00c299e0d8
         oldsize = 2147483648
         olddata = 0x7f00c299e048
         newdata = 0x32e0448
         i = 6
         copyelem = 6

EXPLAIN shows that there are 2604186278 rows in all partitions, but
planner
thinks there will be only 200 unique rows after group by. Looks like we
was
mistaken.

  Finalize GroupAggregate  (cost=154211885.42..154211936.09 rows=200
width=16)
     Group Key: really_huge_partitioned_table.ts
     ->  Gather Merge  (cost=154211885.42..154211932.09 rows=400
width=16)
           Workers Planned: 2
           ->  Sort  (cost=154210885.39..154210885.89 rows=200 width=16)
                 Sort Key: really_huge_partitioned_table.ts
                 ->  Partial HashAggregate 
(cost=154210875.75..154210877.75 rows=200 width=16)
                       Group Key: really_huge_partitioned_table.ts
                       ->  Append  (cost=0.43..141189944.36
rows=2604186278 width=8)
                             ->  Parallel Index Only Scan using
really_huge_partitioned_table_001_idx2 on
really_huge_partitioned_table_001  (cost=0.43..108117.92 rows=2236977
width=8)
                             ->  Parallel Index Only Scan using
really_huge_partitioned_table_002_idx2 on
really_huge_partitioned_table_002  (cost=0.43..114928.19 rows=2377989
width=8)
                             .... and more than 400 partitions more

After some investigation I found bug that is present in simplehash from
its
beginning:

- sizemask is set only in SH_COMPUTE_PARAMETERS . And it is set in this
way:

         /* now set size */
         tb->size = size;

         if (tb->size == SH_MAX_SIZE)
                 tb->sizemask = 0;
         else
                 tb->sizemask = tb->size - 1;

   that means, when we are resizing to SH_MAX_SIZE, sizemask becomes
zero.

- then sizemask is used to SH_INITIAL_BUCKET and SH_NEXT to compute
initial and
   next position:

   SH_INITIAL_BUCKET(SH_TYPE * tb, uint32 hash)
     return hash & tb->sizemask;
   SH_NEXT(SH_TYPE * tb, uint32 curelem, uint32 startelem)
     curelem = (curelem + 1) & tb->sizemask;

- and then SH_GROW stuck in element placing loop:

   startelem = SH_INITIAL_BUCKET(tb, hash);
   curelem = startelem;
   while (true)
     curelem = SH_NEXT(tb, curelem, startelem);

There is Assert(curelem != startelem) in SH_NEXT, but since no one test
it
with 2 billion elements, it were not triggered. And Assert is not
compiled
in production code.

Attached patch fixes it with removing condition and type casting:

         /* now set size */
         tb->size = size;
         tb->sizemask = (uint32)(size - 1);


OOOOOOPS

While writting this letter, I looke at newdata in the frame of
tuplehash_grow:

     newdata = 0x32e0448

It is bellow 4GB border. Allocator does not allocate many-gigabytes
chunks
(and we certainly need 96GB in this case) in sub 4GB address space.
Because
mmap doesn't do this.

I went to check SH_GROW and.... It is `SH_GROW(SH_TYPE *tb, uint32
newsize)`
:-(((
Therefore when `tb->size == SH_MAX_SIZE/2` and we call `SH_GROW(tb,
tb->size * 2)`,
then SH_GROW(tb, 0) is called due to truncation.
And SH_COMPUTE_PARAMETERS is also accepts `uint32 newsize`.

Ahh... ok, patch is updated to fix this as well.