From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1752562AbbKMXPu (ORCPT ); Fri, 13 Nov 2015 18:15:50 -0500 Received: from userp1040.oracle.com ([156.151.31.81]:25168 "EHLO userp1040.oracle.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752174AbbKMXPt (ORCPT ); Fri, 13 Nov 2015 18:15:49 -0500 From: Boris Ostrovsky To: konrad.wilk@oracle.com, david.vrabel@citrix.com Cc: xen-devel@lists.xen.org, linux-kernel@vger.kernel.org, luto@kernel.org, boris.ostrovsky@oracle.com Subject: [PATCH] xen/x86: Adjust stack pointer in xen_sysexit Date: Fri, 13 Nov 2015 18:18:26 -0500 Message-Id: <1447456706-24347-1-git-send-email-boris.ostrovsky@oracle.com> X-Mailer: git-send-email 1.7.1 X-Source-IP: userv0022.oracle.com [156.151.31.74] Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org After 32-bit syscall rewrite, and specifically after commit 5f310f739b4c ("x86/entry/32: Re-implement SYSENTER using the new C path"), the stack frame that is passed to xen_sysexit is no longer a "standard" one (i.e. it's not pt_regs). We need to adjust it so that subsequent xen_iret can use it. Signed-off-by: Boris Ostrovsky --- Alternatively, we could return 0 from do_fast_syscall_32() if paravirt_enabled() is true since Xen PV guests will end up using xen_iret one way or the other. And then we won't need xen_sysexit at all. arch/x86/xen/xen-asm_32.S | 23 ++++++++++++++++------- 1 files changed, 16 insertions(+), 7 deletions(-) diff --git a/arch/x86/xen/xen-asm_32.S b/arch/x86/xen/xen-asm_32.S index fd92a64..c70ec37 100644 --- a/arch/x86/xen/xen-asm_32.S +++ b/arch/x86/xen/xen-asm_32.S @@ -36,15 +36,24 @@ check_events: /* * We can't use sysexit directly, because we're not running in ring0. - * But we can easily fake it up using iret. Assuming xen_sysexit is - * jumped to with a standard stack frame, we can just strip it back to - * a standard iret frame and use iret. + * But we can easily fake it up using iret. + * We came here from the opportunistic SYSEXIT path in entry_SYSENTER_32 + * which left the stack looking like this: + * $__USER_DS + * %ecx + * eflags + * $__USER_CS + * %eip + * %eax + * %gs + * %fs + * %es + * %ds <-- %esp + * + * so we need to adjust it to look like a standard iret frame */ ENTRY(xen_sysexit) - movl PT_EAX(%esp), %eax /* Shouldn't be necessary? */ - orl $X86_EFLAGS_IF, PT_EFLAGS(%esp) - lea PT_EIP(%esp), %esp - + add $5*4, %esp jmp xen_iret ENDPROC(xen_sysexit) -- 1.7.1