Re: [PATCH] General purpose zeroed page slab

From: Denis Vlasenko <vda_at_port.imtp.ilyichevsk.odessa.ua>
Date: 2004-10-15 05:08:49
On Thursday 14 October 2004 21:49, Dave Jones wrote:
> On Thu, Oct 14, 2004 at 06:36:37PM +0100, Matthew Wilcox wrote:
>  > On Thu, Oct 14, 2004 at 01:30:21PM -0400, Brian Gerst wrote:
>  > > This doesn't work as you expect it does.  The constructor is only called 
>  > > when a new slab is created, for each new object on the slab.  It is 
>  > > _not_ run again when an object is freed.  So if a page is freed then 
>  > > immediately reallocated it will contain garbage.
>  > 
>  > The user is responsible for zeroing the page before handing it back to
>  > the slab allocator.

BTW, zeroing with non-temporal stores may be a huge win here.
It is 300% faster on Athlon.

> That sounds like an accident waiting to happen.
> How about a CONFIG_DEBUG option to check its zeroed on free ?
--
vda

-
To unsubscribe from this list: send the line "unsubscribe linux-ia64" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Received on Thu Oct 14 15:19:56 2004

This archive was generated by hypermail 2.1.8 : 2005-08-02 09:20:31 EST