Horton

RE: Is this a problem in GenericXLogFinish()?

Hayato Kuroda (Fujitsu) <kuroda.hayato@fujitsu.com>

View thread View on postgr.es
From: "Hayato Kuroda (Fujitsu)" <kuroda.hayato@fujitsu.com>
To: 'Amit Kapila' <amit.kapila16@gmail.com>, Michael Paquier <michael@paquier.xyz>
Cc: Robert Haas <robertmhaas@gmail.com>, Jeff Davis <pgsql@j-davis.com>, Alexander Lakhin <exclusion@gmail.com>, Heikki Linnakangas <hlinnaka@iki.fi>, "pgsql-hackers@postgresql.org" <pgsql-hackers@postgresql.org>, "Hayato Kuroda (Fujitsu)" <kuroda.hayato@fujitsu.com>
Date: 2023-11-06T06:46:16Z
Lists: pgsql-hackers

Attachments

Dear hackers,

> Dear Amit, Michael,
> 
> Thanks for making the patch!
> 
> > Why register wbuf at all if there are no tuples to add and it is not
> > the same as bucketbuf? Also, I think this won't be correct if prevbuf
> > and wrtbuf are the same and also we have no tuples to add to wbuf. I
> > have attached a naive and crude way to achieve it. This needs more
> > work both in terms of trying to find a better way to change the code
> > or ensure this won't break any existing case. I have just run the
> > existing tests. Such a fix certainly required more testing.
> 
> I'm verifying the idea (currently it seems OK), but at least there is a coding error -
> wbuf_flags should be uint8 here. PSA the fixed patch.

Here is a new patch which is bit refactored. It did:

* If-conditions in _hash_freeovflpage() were swapped.
* Based on above, an Assert(xlrec.ntups == 0) was added.
* A condition in hash_xlog_squeeze_page() was followed the change as well
* comments were adjusted

Next we should add some test codes. I will continue considering but please post anything
If you have idea.

Best Regards,
Hayato Kuroda
FUJITSU LIMITED

Commits

  1. Fix inconsistency with replay of hash squeeze record for clean buffers

  2. Set LSN for wbuf in _hash_freeovflpage() iff wbuf is modified.

  3. Fix an uninitialized access in hash_xlog_squeeze_page().

  4. Use REGBUF_NO_CHANGE at one more place in the hash index.

  5. Assert that buffers are marked dirty before XLogRegisterBuffer().

  6. Fix bug in GenericXLogFinish().

  7. Improve hash index bucket split behavior.