Horton

Re: Parallel bitmap heap scan

Robert Haas <robertmhaas@gmail.com>

View thread View on postgr.es
From: Robert Haas <robertmhaas@gmail.com>
To: Dilip Kumar <dilipbalaut@gmail.com>
Cc: Thomas Munro <thomas.munro@enterprisedb.com>, Haribabu Kommi <kommi.haribabu@gmail.com>, Rafia Sabih <rafia.sabih@enterprisedb.com>, tushar <tushar.ahuja@enterprisedb.com>, Amit Khandekar <amitdkhan.pg@gmail.com>, Amit Kapila <amit.kapila16@gmail.com>, Andres Freund <andres@anarazel.de>, pgsql-hackers <pgsql-hackers@postgresql.org>
Date: 2017-03-07T16:11:19Z
Lists: pgsql-hackers 
(On Tue, Feb 28, 2017 at 10:48 AM, Dilip Kumar <dilipbalaut@gmail.com> wrote:
> 0001- same as previous with some changes for freeing the shared memory stuff.

+    if (--ptbase->refcount == 0)
+        dsa_free(dsa, istate->pagetable);
+
+    if (istate->spages)
+    {
+        ptpages = dsa_get_address(dsa, istate->spages);
+        if (--ptpages->refcount == 0)
+            dsa_free(dsa, istate->spages);
+    }
+    if (istate->schunks)
+    {
+        ptchunks = dsa_get_address(dsa, istate->schunks);
+        if (--ptchunks->refcount == 0)
+            dsa_free(dsa, istate->schunks);
+    }

This doesn't involve any locking, which I think will happen to work
with the current usage pattern but doesn't seem very robust in
general.  I think you either need the refcounts to be protected by a
spinlock, or maybe better, use pg_atomic_uint32 for them.  You want
something like if (pg_atomic_sub_fetch_u32(&refcount, 1) == 0) {
dsa_free(...) }

Otherwise, there's no guarantee it will get freed exactly once.

-- 
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company

Commits

  1. Pass DSA_ALLOC_HUGE when allocating a shared TIDBitmap.

  2. Support parallel bitmap heap scans.

  3. tidbitmap: Support shared iteration.

  4. Shut down Gather's children before shutting down Gather itself.

  5. Revise the way the element allocator for a simplehash is specified.

  6. Refactor bitmap heap scan estimation of heap pages fetched.

  7. Use more efficient hashtable for tidbitmap.c to speed up bitmap scans.