Horton

Re: BUG #16476: pgp_sym_encrypt_bytea with compress-level=6 : Wrong key or corrupt data

Michael Paquier <michael@paquier.xyz>

View thread View on postgr.es
From: Michael Paquier <michael@paquier.xyz>
To: Kyotaro Horiguchi <horikyota.ntt@gmail.com>
Cc: tgl@sss.pgh.pa.us, jeff.janes@gmail.com, frank.gagnepain@intm.fr, pgsql-bugs@lists.postgresql.org, Mark Wong <mark@2ndquadrant.com>
Date: 2020-07-22T09:37:49Z
Lists: pgsql-bugs 
On Wed, Jul 22, 2020 at 05:45:54PM +0900, Kyotaro Horiguchi wrote:
> Might be help if the test prints the orignal and the decrypted bytes?

And turning the buildfarm completely red?  No, thanks.  Those random
strings are also too long to be analyzed easily.  The issue is not
directly related to s390 though, shelduck, that runs on SLES 12 SP3 is
reporting green, as well as sheartail.  I am wondering what
barbastella, barbthroat and pipistrelles that run SLES 15 are going to
say though.

We have a couple of options to move on, but a failure of this test
means that the input string does not match with the output once
compressed and then decompressed:
1) Revert the patch, look more into the problem in those
environments.  I don't have an access to that though.
2) Remove the test.
3) Switch the buildfarm members to not use libz, and file a bug to
SLES to find out what's wrong.  For example, could it be possible that
a flavor of zlib-ng with zlib compatibility gets used?  This has some
code specific to s390.  I am adding Mark in CC, perhaps he has an
idea.

For now I am waiting more to see what other animals have to say.
--
Michael

Commits

  1. Fix corner case with 16kB-long decompression in pgcrypto, take 2

  2. Fix ancient violation of zlib's API spec.

  3. Revert "Fix corner case with PGP decompression in pgcrypto"

  4. Fix corner case with PGP decompression in pgcrypto