From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S932667AbXAWDLB (ORCPT ); Mon, 22 Jan 2007 22:11:01 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S932091AbXAWDLB (ORCPT ); Mon, 22 Jan 2007 22:11:01 -0500 Received: from rhun.apana.org.au ([64.62.148.172]:2047 "EHLO arnor.apana.org.au" rhost-flags-OK-OK-OK-FAIL) by vger.kernel.org with ESMTP id S932667AbXAWDLA (ORCPT ); Mon, 22 Jan 2007 22:11:00 -0500 From: Herbert Xu To: avi@qumranet.com (Avi Kivity), torvalds@osdl.org Subject: Re: [PATCH 4/5] KVM: Fix asm constraints with CONFIG_FRAME_POINTER=n Cc: kvm-devel@lists.sourceforge.net, linux-kernel@vger.kernel.org, akpm@osdl.org, mingo@elte.hu Organization: Core In-Reply-To: <20070111100630.55461250595@il.qumranet.com> X-Newsgroups: apana.lists.os.linux.kernel User-Agent: tin/1.7.4-20040225 ("Benbecula") (UNIX) (Linux/2.6.17-rc4 (i686)) Message-Id: Date: Tue, 23 Jan 2007 14:10:00 +1100 Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Avi Kivity wrote: > A "g" constraint may place a local variable in an %rsp-relative memory operand. > but if your assembly changes %rsp, the operand points to the wrong location. > > An "r" constraint fixes that. > > Thanks to Ingo Molnar for neatly bisecting the problem. > > Signed-off-by: Avi Kivity > > Index: linux-2.6/drivers/kvm/vmx.c > =================================================================== > --- linux-2.6.orig/drivers/kvm/vmx.c > +++ linux-2.6/drivers/kvm/vmx.c > @@ -1825,7 +1825,7 @@ again: > #endif > "setbe %0 \n\t" > "popf \n\t" > - : "=g" (fail) > + : "=r" (fail) > : "r"(vcpu->launched), "d"((unsigned long)HOST_RSP), > "c"(vcpu), > [rax]"i"(offsetof(struct kvm_vcpu, regs[VCPU_REGS_RAX])), We need the following fix for 2.6.20. [KVM] vmx: Fix register constraint in launch code Both "=r" and "=g" breaks my build on i386: $ make CC [M] drivers/kvm/vmx.o {standard input}: Assembler messages: {standard input}:3318: Error: bad register name `%sil' make[1]: *** [drivers/kvm/vmx.o] Error 1 make: *** [_module_drivers/kvm] Error 2 The reason is that setbe requires an 8-bit register but "=r" does not constrain the target register to be one that has an 8-bit version on i386. According to http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10153 the correct constraint is "=q". Signed-off-by: Herbert Xu Cheers, -- Visit Openswan at http://www.openswan.org/ Email: Herbert Xu ~{PmV>HI~} Home Page: http://gondor.apana.org.au/~herbert/ PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt -- diff --git a/drivers/kvm/vmx.c b/drivers/kvm/vmx.c index ce219e3..0aa2659 100644 --- a/drivers/kvm/vmx.c +++ b/drivers/kvm/vmx.c @@ -1824,7 +1824,7 @@ again: #endif "setbe %0 \n\t" "popf \n\t" - : "=g" (fail) + : "=q" (fail) : "r"(vcpu->launched), "d"((unsigned long)HOST_RSP), "c"(vcpu), [rax]"i"(offsetof(struct kvm_vcpu, regs[VCPU_REGS_RAX])),