From: | Greg Stark <stark(at)mit(dot)edu> |
---|---|
To: | Robert Haas <robertmhaas(at)gmail(dot)com> |
Cc: | "pgsql-hackers(at)postgresql(dot)org" <pgsql-hackers(at)postgresql(dot)org> |
Subject: | Re: sb_alloc: a new memory allocator for PostgreSQL |
Date: | 2014-05-06 13:25:47 |
Message-ID: | CAM-w4HPBgoLcaT-xqEXzVg9ONVSn1AB_Z7qbSxN_iN1JccmJJw@mail.gmail.com |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-hackers |
On Tue, May 6, 2014 at 2:04 PM, Robert Haas <robertmhaas(at)gmail(dot)com> wrote:
> I also didn't find anything that looked like our
> memory context paradigm, and in particular the ability to cheaply
> reset a context, in any other allocator.
You probably knew this but just in case the term for this strategy is
called a "ripcord allocator". I believe GCC had one, not sure if it
still uses it. I doubt any existing ones are especially helpful though
compared to looking at regular allocators and seeing how to integrate
them.
I assume you're also aware of lock-free data structures and the like.
A memory allocator seems like something that needs to be super aware
of causing concurrency bottlenecks since it has no idea where it'll be
used.
--
greg
From | Date | Subject | |
---|---|---|---|
Next Message | Heikki Linnakangas | 2014-05-06 13:31:13 | Re: sb_alloc: a new memory allocator for PostgreSQL |
Previous Message | Noah Misch | 2014-05-06 13:22:20 | Re: [PATCH] `pg_dump -Fd` doesn't check write return status... |