From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id ; Thu, 22 Nov 2001 10:26:41 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id ; Thu, 22 Nov 2001 10:26:31 -0500 Received: from ns.suse.de ([213.95.15.193]:7692 "HELO Cantor.suse.de") by vger.kernel.org with SMTP id ; Thu, 22 Nov 2001 10:26:18 -0500 To: Terje Eggestad Cc: linux-kernel@vger.kernel.org Subject: Re: [Q] was the SYSENTER/SYSCALL fast system calls completed or discared in the end?? In-Reply-To: <1006184327.19902.2.camel@pc-16.office.scali.no.suse.lists.linux.kernel> <20011121225402.A175@elf.ucw.cz.suse.lists.linux.kernel> <1006440069.22598.4.camel@pc-16.office.scali.no.suse.lists.linux.kernel> From: Andi Kleen Date: 22 Nov 2001 16:26:15 +0100 In-Reply-To: Terje Eggestad's message of "22 Nov 2001 15:46:31 +0100" Message-ID: X-Mailer: Gnus v5.7/Emacs 20.7 Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Terje Eggestad writes: > ons, 2001-11-21 kl. 22:54 skrev Pavel Machek: > > > > On Mon 19-11-01 16:38:45, Terje Eggestad wrote: > > > subject says it all.... > > > > > > I remember there was a discussion and a patch floating around that > > > implemented SYSCALL/SYSRET, just want to know what happened to it.... > > > > discarded > > Because there was no perf benefit or because the patch was in poor > quality? The x86-64 port uses SYSCALL/SYSRET as the native call method for 64bit programs. Before x86-64 it turned out that SYSCALL is not really usable because of some builtin limitations (which are fixed in x86-64). This is only a problem for the K6, K7 should have support for SYSENTER/SYSEXIT (the Intel equivalent), which seems to be better designed. There is a patch to use SYSENTER/SYSEXIT. -Andi