Re: [sqlsmith] Crash in mcv_get_match_bitmap - Mailing list pgsql-hackers

From Tomas Vondra
Subject Re: [sqlsmith] Crash in mcv_get_match_bitmap
Date
Msg-id 20190710215904.ncaqlswrylliisv3@development
Whole thread Raw
In response to Re: [sqlsmith] Crash in mcv_get_match_bitmap  (Tom Lane <tgl@sss.pgh.pa.us>)
Responses Re: [sqlsmith] Crash in mcv_get_match_bitmap  (Tom Lane <tgl@sss.pgh.pa.us>)
List pgsql-hackers
On Wed, Jul 10, 2019 at 05:45:24PM -0400, Tom Lane wrote:
>Tomas Vondra <tomas.vondra@2ndquadrant.com> writes:
>> Yeah, that's a bug. Will fix (not sure how yet).
>
>You could do worse than replace this:
>
>        ok = (NumRelids(clause) == 1) &&
>            (is_pseudo_constant_clause(lsecond(expr->args)) ||
>             (varonleft = false,
>              is_pseudo_constant_clause(linitial(expr->args))));
>
>with something like
>
>    if (IsA(linitial(expr->args), Var) &&
>        IsA(lsecond(expr->args), Const))
>       ok = true, varonleft = true;
>    else if (IsA(linitial(expr->args), Const) &&
>             IsA(lsecond(expr->args), Var))
>       ok = true, varonleft = false;
>
>Or possibly get rid of varonleft as such, and merge extraction of the
>"var" and "cst" variables into this test.
>

OK, thanks for the suggestion.

I probably also need to look at the "is compatible" test in
extended_stats.c which also looks at the clauses. It may not crash as it
does not attempt to extract the const values etc. but it likely needs to
be in sync with this part.

>BTW, I bet passing a unary-argument OpExpr also makes this code
>unhappy.

Whooops :-(


regards

-- 
Tomas Vondra                  http://www.2ndQuadrant.com
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services 



pgsql-hackers by date:

Previous
From: Tom Lane
Date:
Subject: Re: [sqlsmith] Crash in mcv_get_match_bitmap
Next
From: Tomas Vondra
Date:
Subject: Re: [Proposal] Table-level Transparent Data Encryption (TDE) and KeyManagement Service (KMS)