Re: Use pg_pwritev_with_retry() instead of write() in dir_open_for_write() to avoid partial writes?

From: Andres Freund <andres(at)anarazel(dot)de>
To: Bharath Rupireddy <bharath(dot)rupireddyforpostgres(at)gmail(dot)com>
Cc: Michael Paquier <michael(at)paquier(dot)xyz>, Alvaro Herrera <alvherre(at)alvh(dot)no-ip(dot)org>, Nathan Bossart <nathandbossart(at)gmail(dot)com>, Thomas Munro <thomas(dot)munro(at)gmail(dot)com>, PostgreSQL Hackers <pgsql-hackers(at)lists(dot)postgresql(dot)org>
Subject: Re: Use pg_pwritev_with_retry() instead of write() in dir_open_for_write() to avoid partial writes?
Date: 2023-02-11 22:44:24
Message-ID: 20230211224424.r25uw4rsv6taukxk@awork3.anarazel.de
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

Hi,

On 2022-11-01 08:32:48 +0530, Bharath Rupireddy wrote:
> +/*
> + * pg_pwrite_zeros
> + *
> + * Writes zeros to a given file. Input parameters are "fd" (file descriptor of
> + * the file), "size" (size of the file in bytes).
> + *
> + * On failure, a negative value is returned and errno is set appropriately so
> + * that the caller can use it accordingly.
> + */
> +ssize_t
> +pg_pwrite_zeros(int fd, size_t size)
> +{
> + PGAlignedBlock zbuffer;
> + size_t zbuffer_sz;
> + struct iovec iov[PG_IOV_MAX];
> + int blocks;
> + size_t remaining_size = 0;
> + int i;
> + ssize_t written;
> + ssize_t total_written = 0;
> +
> + zbuffer_sz = sizeof(zbuffer.data);
> +
> + /* Zero-fill the buffer. */
> + memset(zbuffer.data, 0, zbuffer_sz);

I previously commented on this - why are we memseting a buffer on every call
to this? That's not at all free.

Something like
static const PGAlignedBlock zerobuf = {0};
would do the trick. You do need to cast the const away, to assign to
iov_base, but that's not too ugly.

Greetings,

Andres Freund

In response to

Responses

Browse pgsql-hackers by date

  From Date Subject
Next Message Tom Lane 2023-02-11 23:54:00 Re: Importing pg_bsd_indent into our source tree
Previous Message Andres Freund 2023-02-11 22:33:40 Re: refactoring relation extension and BufferAlloc(), faster COPY