Horton

Re: longfin and tamandua aren't too happy but I'm not sure why

Dilip Kumar <dilipbalaut@gmail.com>

View thread View on postgr.es
From: Dilip Kumar <dilipbalaut@gmail.com>
To: Tom Lane <tgl@sss.pgh.pa.us>
Cc: Robert Haas <robertmhaas@gmail.com>, pgsql-hackers@lists.postgresql.org, Justin Pryzby <pryzby@telsasoft.com>
Date: 2022-09-28T11:52:28Z
Lists: pgsql-hackers 
On Wed, Sep 28, 2022 at 11:57 AM Tom Lane <tgl@sss.pgh.pa.us> wrote:
>
> Dilip Kumar <dilipbalaut@gmail.com> writes:
> > Btw, I think the reason for the bus error on wrasse is the same as
> > what is creating failure on longfin[1], I mean this unaligned access
> > is causing Bus error during startup, IMHO.
>
> Maybe, but there's not a lot of evidence for that.  wrasse got
> through the test_decoding check where longfin, tamandua, kestrel,
> and now skink are failing.  It's evidently not the same issue
> that the 32-bit animals are choking on, either.  Looks like yet
> a third bug to me.

I think the reason is that "longfin" is configured with the
-fsanitize=alignment option so it will report the failure for any
unaligned access.  Whereas "wrasse" actually generates the "Bus error"
due to architecture.  So the difference is that with
-fsanitize=alignment, it will always complain for any unaligned access
but all unaligned access will not end up in the "Bus error", and I
think that could be the reason "wrasse" is not failing in the test
decoding.

Yeah but anyway this is just a theory behind why failing at different
places but we still do not have evidence/call stack to prove that.

-- 
Regards,
Dilip Kumar
EnterpriseDB: http://www.enterprisedb.com

Commits

  1. meson: Split 'main' suite into 'regress' and 'isolation'

  2. ci: Add 32bit build and test

  3. Fix InitializeRelfilenumberMap for 05d4cbf9b6ba708858984b01ca0fc56d59d4ec7c

  4. Fix alignment problems with SharedInvalSmgrMsg.

  5. In BufTagGetForkNum, cast to the correct type.

  6. Increase width of RelFileNumbers from 32 bits to 56 bits.

  7. Harden pg_filenode_relation test against concurrent DROP TABLE.