Re: a problem about XLogReader callback system

From: Alvaro Herrera <alvherre(at)2ndquadrant(dot)com>
To: "Wang, Shenhao" <wangsh(dot)fnst(at)cn(dot)fujitsu(dot)com>
Cc: "pgsql-hackers(at)lists(dot)postgresql(dot)org" <pgsql-hackers(at)lists(dot)postgresql(dot)org>
Subject: Re: a problem about XLogReader callback system
Date: 2020-09-28 13:56:10
Message-ID: 20200928135610.GA1123@alvherre.pgsql
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

Hello,

On 2020-Sep-28, Wang, Shenhao wrote:

> I have a problem about XLogReader callback system
>
> In xlog.c, function StartupXLOG
>
> xlogreader =
> XLogReaderAllocate(wal_segment_size, NULL,
> XL_ROUTINE(.page_read = &XLogPageRead,
> .segment_open = NULL,
> .segment_close = wal_segment_close),
> &private);
>
> XLogPageReader uses readFile to store the fd, and I can't find any
> location to set the value of seg.ws_file.

Hmm, AFAICS (admittedly a quick look) you're right: because XLogPageRead
does not touch ws_file, wal_segment_close would always find that it's
unset (-1) so XLogReaderFree would have nothing to do.

The whole XLogReader abstraction remains pretty leaky, I'm afraid :-(
Ideally we would plug those holes. Perhaps getting rid of readFile
completely would be a solution, but I haven't analyzed it.

> Is it necessary to set a segment_close callback?

For this case it seems useless but not actually harmful. Have you come
across some scenario in which it causes a problem?

--
Álvaro Herrera https://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services

In response to

Browse pgsql-hackers by date

  From Date Subject
Next Message Tom Lane 2020-09-28 14:02:31 Re: The return value of SPI_connect
Previous Message Vladimir Sitnikov 2020-09-28 13:46:43 Re: BLOB / CLOB support in PostgreSQL