Re: amcheck: add index-all-keys-match verification for B-Tree - Mailing list pgsql-hackers

From Jakub Wartak
Subject Re: amcheck: add index-all-keys-match verification for B-Tree
Date
Msg-id CAKZiRmwAg8UWkoZzYq8XNe+oacJiicgJyED0GZNWk1HGY4zwRQ@mail.gmail.com
Whole thread
In response to Re: amcheck: add index-all-keys-match verification for B-Tree  (Andrey Borodin <x4mmm@yandex-team.ru>)
List pgsql-hackers
On Mon, Jul 27, 2026 at 4:12 PM Andrey Borodin <x4mmm@yandex-team.ru> wrote:
>
> Jakub, thanks for the next review round! I'll provide thoughtful answers later.
> [v6]

Hi Andrey,

I've spotted in the commitfest that the cfbot complained [1], so I've took a
quick look. It appears that recent preparation work for the index prefetch
(shifting the responsibilty of the index scans to the tableam alone), commit
ddce1da5b1b1 removed the IndexFetchTableData and table_index_fetch_{begin,
tuple,end}. To my understanding this new stuff - that is:
* table_fetch_tid() -> heapam_fetch_tid()
* table_tuple_fetch_row_version() -> heapam_fetch_row_version()
still follows LP_REDIRECT/HOT, so it should be OKay-ish, but please take a
deeper view if possible.

Attached v6fixup is patch that fixes the compilation issue, and passes
CI tests for
me (I didn't want to modify the v6 patchset to make review easier).

-J.

[1] - https://github.com/postgres/postgres-cfbot/actions/runs/36987206460/job/110774874058

Attachment

pgsql-hackers by date:

Previous
From: Osama Abdul Qader
Date:
Subject: Re: Improving scalability of Parallel Bitmap Heap/Index Scan
Next
From: Bertrand Drouvot
Date:
Subject: Re: WAL segment file descriptor leak on read errors can PANIC the server