Re: Reduce the instruction overhead of OpenSSL calls

From: Daniel Gustafsson <daniel(at)yesql(dot)se>
To: "ryanewang(王蕾)" <ryanewang(at)tencent(dot)com>
Cc: PostgreSQL Hackers <pgsql-hackers(at)postgresql(dot)org>
Subject: Re: Reduce the instruction overhead of OpenSSL calls
Date: 2025-03-04 09:36:08
Message-ID: 751A4F25-79A6-425B-8323-C626BD300875@yesql.se
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

> On 27 Feb 2025, at 14:16, ryanewang(王蕾) <ryanewang(at)tencent(dot)com> wrote:
>
> Hi,
> From the openssl documentation, when the value of ret is greater than 0, the SSL_get_error() function returns SSL_ERROR_NONE.
> So, it seems that when the return value of SSL_read() or SSL_write() function is greater than 0, we don't need to make an error judgment.
> The attached patch attempts to reduce unnecessary error judgments.

Calling SSL_get_error() is needed to clear the error queue for subsequent IO
operations, so I don't skipping it would be wise. I imagine that any overhead
is dwarfed by the time spent in network IO anyways.

--
Daniel Gustafsson

In response to

Browse pgsql-hackers by date

  From Date Subject
Next Message Josef Šimánek 2025-03-04 09:40:04 Re: tests for pg_stat_progress_copy.tuples_skipped
Previous Message Anthonin Bonnefoy 2025-03-04 09:31:46 Re: Add Pipelining support in psql