Horton

Re: HOT chain validation in verify_heapam()

Himanshu Upadhyaya <upadhyaya.himanshu@gmail.com>

View thread View on postgr.es
From: Himanshu Upadhyaya <upadhyaya.himanshu@gmail.com>
To: Robert Haas <robertmhaas@gmail.com>
Cc: Andres Freund <andres@anarazel.de>, Peter Geoghegan <pg@bowt.ie>, PostgreSQL Hackers <pgsql-hackers@postgresql.org>, Andrew Dunstan <andrew@dunslane.net>, Mark Dilger <mark.dilger@enterprisedb.com>
Date: 2022-09-19T08:28:03Z
Lists: pgsql-hackers

Attachments

On Wed, Sep 7, 2022 at 2:49 AM Robert Haas <robertmhaas@gmail.com> wrote:

>
> But here's one random idea: add a successor[] array and an lp_valid[]
> array. In the first loop, set lp_valid[offset] = true if it passes the
> check_lp() checks, and set successor[A] = B if A redirects to B or has
> a CTID link to B, without matching xmin/xmax. Then, in a second loop,
> iterate over the successor[] array. If successor[A] = B && lp_valid[A]
> && lp_valid[B], then check whether A.xmax = B.xmin; if so, then
> complain if predecessor[B] is already set, else set predecessor[B] =
> A. Then, in the third loop, iterate over the predecessor array just as
> you're doing now. Then it's clear that we do the lp_valid checks
> exactly once for every offset that might need them, and in order. And
> it's also clear that the predecessor-based checks can never happen
> unless the lp_valid checks passed for both of the offsets involved.
>
>
ok, I have introduced a new approach to first construct a successor array
and then loop over the successor array to construct a predecessor array.

-- 
Regards,
Himanshu Upadhyaya
EnterpriseDB: http://www.enterprisedb.com

Commits

  1. amcheck: Generalize one of the recently-added update chain checks.

  2. amcheck: Tighten up validation of redirect line pointers.

  3. amcheck: Fix verify_heapam for tuples where xmin or xmax is 0.

  4. amcheck: Fix a few bugs in new update chain validation.

  5. Fix new test case to work on (some?) big-endian architectures.

  6. Don't test HEAP_XMAX_INVALID when freezing xmax.