WARNING - OLD ARCHIVES

This is an archived copy of the Xen.org mailing list, which we have preserved to ensure that existing links to archives are not broken. The live archive, which contains the latest emails, can be found at http://lists.xen.org/
   
 
 
Xen 
 
Home Products Support Community News
 
   
 

xen-devel

[Xen-devel] Re: [patch] pae: tlbflush linear page table updates

To: Andi Kleen <ak@xxxxxxx>
Subject: [Xen-devel] Re: [patch] pae: tlbflush linear page table updates
From: Keir Fraser <Keir.Fraser@xxxxxxxxxxxx>
Date: Fri, 12 Aug 2005 11:18:12 +0100
Cc: Gerd Knorr <kraxel@xxxxxxx>, xen-devel@xxxxxxxxxxxxxxxxxxx, "Tian, Kevin" <kevin.tian@xxxxxxxxx>
Delivery-date: Fri, 12 Aug 2005 10:12:05 +0000
Envelope-to: www-data@xxxxxxxxxxxxxxxxxxx
In-reply-to: <20050812095648.GG8974@xxxxxxxxxxxxx>
List-help: <mailto:xen-devel-request@lists.xensource.com?subject=help>
List-id: Xen developer discussion <xen-devel.lists.xensource.com>
List-post: <mailto:xen-devel@lists.xensource.com>
List-subscribe: <http://lists.xensource.com/cgi-bin/mailman/listinfo/xen-devel>, <mailto:xen-devel-request@lists.xensource.com?subject=subscribe>
List-unsubscribe: <http://lists.xensource.com/cgi-bin/mailman/listinfo/xen-devel>, <mailto:xen-devel-request@lists.xensource.com?subject=unsubscribe>
References: <571ACEFD467F7749BC50E0A98C17CDD807959968@pdsmsx403> <20050812095648.GG8974@xxxxxxxxxxxxx>
Sender: xen-devel-bounces@xxxxxxxxxxxxxxxxxxx

On 12 Aug 2005, at 10:56, Andi Kleen wrote:

"early clobber" is good to prevent output alias as input, if that output
may be clobbered before input is used. However there seems no point to
simply alias among inputs. Or else, the only way I can see is that
compiler insert some extra lines in the middle of inline asm... Any
benefit for this likelihood?

There can be cases e.g. when the inputs are dependent and you use
general enough constraints. e.g. one input can be %reg and the other
offset(%reg). In this case they will essentially alias.

That would be a nasty one to track down. :-)

It seems the general policy in Linux also is not to bother with '&' on outputs that have a forced alias in the input list. I'd be happy to see a cleanup patch that adds more '&'s to Xen though -- in cases where we are relying on the forced alias then it's unlikely to change the generated assembly code.

 -- Keir


_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel

<Prev in Thread] Current Thread [Next in Thread>