Assert that we don't insert nulls into attnotnull catalog columns.
authorTom Lane
Tue, 21 Jul 2020 16:38:08 +0000 (12:38 -0400)
committerTom Lane
Tue, 21 Jul 2020 16:38:08 +0000 (12:38 -0400)
commitbca409e5b160f81ccd980bef2aeb32f8b731b0fd
treed79bf2fafd5294d000ac602b63df5071b851f09c
parente5372b48b94f17dcce5d8f3b26d55b0f182e4306
Assert that we don't insert nulls into attnotnull catalog columns.

The executor checks for this error, and so does the bootstrap catalog
loader, but we never checked for it in retail catalog manipulations.
The folly of that has now been exposed, so let's add assertions
checking it.  Checking in CatalogTupleInsert[WithInfo] and
CatalogTupleUpdate[WithInfo] should be enough to cover this.

Back-patch to v10; the aforesaid functions didn't exist before that,
and it didn't seem worth adapting the patch to the oldest branches.
But given the risk of JIT crashes, I think we certainly need this
as far back as v11.

Pre-v13, we have to explicitly exclude pg_subscription.subslotname
and pg_subscription_rel.srsublsn from the checks, since they are
mismarked.  (Even if we change our mind about applying BKI_FORCE_NULL
in the branch tips, it doesn't seem wise to have assertions that
would fire in existing databases.)

Discussion: https://postgr.es/m/298837.1595196283@sss.pgh.pa.us
doc/src/sgml/bki.sgml
src/backend/catalog/indexing.c