Re: [BUGS] BUG #2401: spinlocks not available on amd64 - Mailing list pgsql-patches

From Bruce Momjian
Subject Re: [BUGS] BUG #2401: spinlocks not available on amd64
Date
Msg-id 200604300016.k3U0GGu13388@candle.pha.pa.us
Whole thread Raw
In response to Re: [BUGS] BUG #2401: spinlocks not available on amd64  (Tom Lane <tgl@sss.pgh.pa.us>)
Responses Re: [BUGS] BUG #2401: spinlocks not available on amd64  (Tom Lane <tgl@sss.pgh.pa.us>)
Re: [BUGS] BUG #2401: spinlocks not available on amd64  (Robert Lor <Robert.Lor@Sun.COM>)
List pgsql-patches
Tom Lane wrote:
> Theo Schlossnagle <jesus@omniti.com> writes:
> > I'd remind everyone that the spinlock stuff is entirely optional at
> > build time.
>
> Not really.  The performance hit for not having hardware spinlocks is
> so severe that it's not considered a reasonable fallback.
>
> > I also think it immensely useful to replace all of the tas subsystem
> > with cas so that one could reliabily lock these atomics with the process
> > id of the locker.
>
> I cannot, ever once in my years working on Postgres, remember having
> wanted such a thing.  I am strongly against mucking with the spinlock
> code for mere aesthetics --- it's too fragile and hard to test,
> especially on platforms you don't have ready access to.
>
> In short, it ain't broken and we don't need to fix it.

Agreed.  Should the new Solaris ASM code be modified?

--
  Bruce Momjian   http://candle.pha.pa.us
  EnterpriseDB    http://www.enterprisedb.com

  + If your life is a hard drive, Christ can be your backup. +

pgsql-patches by date:

Previous
From: Tom Lane
Date:
Subject: Re: Patch for BUG #2073: Can't drop sequence when created
Next
From: Tom Lane
Date:
Subject: Re: [BUGS] BUG #2401: spinlocks not available on amd64