Subject: Re: [vserver] kernel stacksize
From: Herbert Poetzl <herbert@13thfloor.at>
Date: Fri, 12 Oct 2007 16:18:26 +0200

On Fri, Oct 12, 2007 at 08:54:43AM -0400, Chuck wrote:
> On Friday 12 October 2007, Herbert Poetzl wrote:
> > On Wed, Oct 10, 2007 at 09:59:45PM -0400, Chuck wrote:
> > > On Wednesday 10 October 2007, Herbert Poetzl wrote:
> > > > On Wed, Oct 10, 2007 at 05:59:22PM -0400, Chuck wrote:
> > > > > 
> > > > > does vserver performance benefit from 4k stack sizes instead of 8? 
> > > > 
> > > > assumed that the page size of your architecture is
> > > > 4k and not 8k, then this will probably give you
> > > > a (small) performance improvement and of course
> > > > better memory usage
> > > > 
> > > 
> > > i don't know about the page size of the arch.. its amd64 ..
> > > opteron processors.. if its going to be only a small benefit i
> > > might best leave it at the default which is 8k.
> > 
> > page size for x86_64 is usually the same as for x86
> > so the assumption that it has a 4k page size is most
> > likely correct 
> 
> so then if it has a 4k page size, if for no other reason than to match
> it, it would be sensible to set the kernel stacksize to 4k.

yes, but note that in the past, the size was 8k
(even for x86) and certain combinations of 'heavy'
stack users (modules) might not work with the 4k
stack (should not be the case for in kernel modules)

best,
Herbert

> > best,
> > Herbert
> > 
> > > > HTC,
> > > > Herbert
> > > > 
> > > > > -- 
> > > > > 
> > > > > Chuck
> > > > > 
> > > > > 
> > > > 
> > > 
> > > 
> > > 
> > > -- 
> > > 
> > > Chuck
> > > 
> > 
> 
> 
> 
> -- 
> 
> Chuck
> 
> "...and the hordes of M$*ft users descended upon me in their anger,
> and asked 'Why do you not get the viruses or the BlueScreensOfDeath
> or insecure system troubles and slowness or pay through the nose 
> for an OS as *we* do?!!', and I answered...'I use Linux'. "
> The Book of John, chapter 1, page 1, and end of book
>