Skip site navigation (1)Skip section navigation (2)
Date:      Sat, 8 Feb 2003 23:57:06 -0800
From:      Alfred Perlstein <bright@mu.org>
To:        Peter Wemm <peter@wemm.org>
Cc:        Jake Burkholder <jake@locore.ca>, Jeff Roberson <jroberson@chesapeake.net>, arch@FreeBSD.ORG
Subject:   Re: New kernel allocation API
Message-ID:  <20030209075706.GO88781@elvis.mu.org>
In-Reply-To: <20030209074604.EC6962A8B4@canning.wemm.org>
References:  <20030207234049.A32864@locore.ca> <20030209074604.EC6962A8B4@canning.wemm.org>

next in thread | previous in thread | raw e-mail | index | archive | help
* Peter Wemm <peter@wemm.org> [030208 23:46] wrote:
> 
> And there is hardware support for direct mapping on alpha as well, so it
> completely bypasses the TLB entirely for significant benefits.  For x86-64
> there'll likely be up to 2^39 of direct mapped space as well, but using 2MB
> pages (use an entire level 4 directory slot for direct mapping to avoid the
> APTD evilness).  We just dont have the space to do this on plain x86 :-(.

It seems to me that someone should run some tests to get numbers
to determine if the tlb optimization helps more than the kmem-array
optimization.  However the problem with that is that much more of
the kernel should be converted to use zones/slabs before counting
on the data gathered.  I'm trusting you guys that avoiding the tlb
shootdowns help quite a bit, but then again the cost of each free(9)
seems quite high.

I may see if I can convert some paths in the kernel to use zones. :)

-- 
-Alfred Perlstein [alfred@freebsd.org]
'Instead of asking why a piece of software is using "1970s technology,"
 start asking why software is ignoring 30 years of accumulated wisdom.'

To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-arch" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20030209075706.GO88781>