Horton

Re: Flushing large data immediately in pqcomm

Jelte Fennema-Nio <postgres@jeltef.nl>

View thread View on postgr.es
From: Jelte Fennema-Nio <postgres@jeltef.nl>
To: Andres Freund <andres@anarazel.de>
Cc: David Rowley <dgrowleyml@gmail.com>, Melih Mutlu <m.melihmutlu@gmail.com>, PostgreSQL Hackers <pgsql-hackers@lists.postgresql.org>, Heikki Linnakangas <hlinnaka@iki.fi>, Robert Haas <robertmhaas@gmail.com>, Thomas Munro <thomas.munro@gmail.com>, Ranier Vilela <ranier.vf@gmail.com>
Date: 2024-04-07T10:04:48Z
Lists: pgsql-hackers

Attachments

On Sun, 7 Apr 2024 at 03:39, Andres Freund <andres@anarazel.de> wrote:
> Changing the global vars to size_t seems mildly bogus to me. All it's
> achieving is to use slightly more memory. It also just seems unrelated to the
> change.

I took a closer look at this. I agree that changing PqSendBufferSize
to size_t is unnecessary: given the locations that it is used I see no
risk of overflow anywhere. Changing the type of PqSendPointer and
PqSendStart is needed though, because (as described by Heiki and David
upthread) the argument type of internal_flush_buffer is size_t*. So if
you actually pass int* there, and the sizes are not the same then you
will start writing out of bounds. And because internal_flush_buffer is
introduced in this patch, it is related to this change.

This is what David just committed too.

However, the "required" var actually should be of size_t to avoid
overflow if len is larger than int even without this change. So
attached is a tiny patch that does that.

Commits

  1. Avoid needless large memcpys in libpq socket writing