Setting pd_lower in GIN metapage

From: Amit Langote <Langote_Amit_f8(at)lab(dot)ntt(dot)co(dot)jp>
To: Pg Hackers <pgsql-hackers(at)postgresql(dot)org>
Subject: Setting pd_lower in GIN metapage
Date: 2017-06-19 06:07:39
Message-ID: 0d273805-0e9e-ec1a-cb84-d4da400b8f85@lab.ntt.co.jp
Views: Raw Message | Whole Thread | Download mbox
Thread:
Lists: pgsql-hackers

What are some arguments against setting pd_lower in the GIN metapage as
follows?

GinMetaPageData *metad = GinPageGetMeta(page);

((PageHeader) page)->pd_lower =
((char *) metad + sizeof(GinMetaPageData)) - (char *) page;

I saw that _bt_initmetapage() does it, so was wondering why doesn't GIN.

How about porting such a change to the back-branches if we do this at all?
I couldn't find any discussion in the archives about this. I read in
comments that server versions older than 9.4 didn't set pd_lower correctly
in any of GIN index pages, so relying on pd_lower value of GIN pages is
unreliable in general.

The reason I'm asking is that a certain backup tool relies on pd_lower
values of data pages (disk blocks in relation files that are known to have
a valid PageHeaderData) to be correct to discard the portion of every page
that supposedly does not contain any useful information. The assumption
doesn't hold in the case of GIN metapage, so any GIN indexes contain
corrupted metapage after recovery (metadata overwritten with zeros).

Thanks,
Amit

Responses

Browse pgsql-hackers by date

  From Date Subject
Next Message Ashutosh Sharma 2017-06-19 06:12:25 Re: BUG: pg_dump generates corrupted gzip file in Windows
Previous Message Amit Khandekar 2017-06-19 05:33:01 Re: UPDATE of partition key