Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org Received: from vger.kernel.org (vger.kernel.org [23.128.96.18]) by smtp.lore.kernel.org (Postfix) with ESMTP id 332BBC54EAA for ; Thu, 26 Jan 2023 20:09:06 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S230334AbjAZUJD (ORCPT ); Thu, 26 Jan 2023 15:09:03 -0500 Received: from lindbergh.monkeyblade.net ([23.128.96.19]:33034 "EHLO lindbergh.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S232747AbjAZUIu (ORCPT ); Thu, 26 Jan 2023 15:08:50 -0500 Received: from gnuweeb.org (gnuweeb.org [51.81.211.47]) by lindbergh.monkeyblade.net (Postfix) with ESMTPS id 10A747166B; Thu, 26 Jan 2023 12:08:43 -0800 (PST) Received: from biznet-home.integral.gnuweeb.org (unknown [182.253.88.152]) by gnuweeb.org (Postfix) with ESMTPSA id 5FF2082F37; Thu, 26 Jan 2023 20:08:37 +0000 (UTC) X-GW-Data: lPqxHiMPbJw1wb7CM9QUryAGzr0yq5atzVDdxTR0iA== DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=gnuweeb.org; s=default; t=1674763722; bh=2uK4PjvwK03HI9Z+rE8QjRWxVNSJBr+QdWaWqL0x9cI=; h=Date:From:To:Cc:Subject:References:In-Reply-To:From; b=I9aZGjdfHpodIG9EaE0S7ixL3e/iR4Hr+sYIC1sUq4qzwhhjDmj0pG32tpGRNfoOv Wg8yaXBchbgTmPx6/ur8qr6g5HsOBFEca/beD3gGIVscG035d6rFZDWldJFWMPdU3x ypP+R45NYS+7QGDOp2zzXuLC3W1vm6UxCrwRDdRErvuA1U62fGgN/DUXVd/NHg3gvv wO4dqMVGVMlMAZ/40uVuPGmm1RTgj3WC7OPPdheclkiUpHypXD5eV7lN+T3UBVskFQ M3vfF4ZfWiyMWKmegqPs6oZMfRSVIf7OYiLW5JoO1H0+oTk0ozQBSDNdzl1hrigdlE Z7+ZiAHD7mEMw== Date: Fri, 27 Jan 2023 03:08:33 +0700 From: Ammar Faizi To: "H. Peter Anvin" Cc: Dave Hansen , Dave Hansen , "Kirill A. Shutemov" , Thomas Gleixner , Andrew Cooper , Peter Zijlstra , Brian Gerst , Borislav Petkov , Shuah Khan , Ingo Molnar , Andy Lutomirski , x86 Mailing List , Linux Kselftest Mailing List , Linux Kernel Mailing List Subject: Re: [RFC PATCH v1 1/2] selftests/x86: sysret_rip: Handle syscall in a FRED system Message-ID: References: <25b96960-a07e-a952-5c23-786b55054126@zytor.com> <509443c8-e0fd-935f-63d8-7264f5dd3c05@zytor.com> <20230124002625.581323-1-ammarfaizi2@gnuweeb.org> <20230124002625.581323-2-ammarfaizi2@gnuweeb.org> <8f5c24df-514d-5d89-f58f-ec8c3eb1e049@zytor.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <8f5c24df-514d-5d89-f58f-ec8c3eb1e049@zytor.com> Precedence: bulk List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Mon, Jan 23, 2023 at 05:40:23PM -0800, H. Peter Anvin wrote: > So as per Andrew's comment, add: > > register void * rsp asm("%rsp"); > > ... > > "+r" (rsp) /* clobber the redzone */ > > ... as the right way to avoid redzone problems. I played with this more. I found something wrong with this. This doesn't work for me. The compiler still uses red zone despite I use "+r" (rsp). What did I do wrong? ----- ammarfaizi2@integral2:/tmp$ gcc -fno-stack-protector -O2 -Wall -Wextra test.c -o test ammarfaizi2@integral2:/tmp$ objdump --no-show-raw-insn -d test | grep "a_leaf_func_with_red_zone>:" -A8 0000000000001180 : 1180: endbr64 1184: mov $0x1,%eax 1189: mov %rax,-0x8(%rsp) ## BUG!!! 118e: pushf 118f: pop %rax 1190: mov -0x8(%rsp),%rax ## BUG!!! 1195: ret ammarfaizi2@integral2:/tmp$ clang -O2 -Wall -Wextra test.c -o test ammarfaizi2@integral2:/tmp$ objdump --no-show-raw-insn -d test | grep "a_leaf_func_with_red_zone>:" -A6 0000000000001140 : 1140: mov $0x1,%eax 1145: mov %rax,-0x8(%rsp) ## BUG!!! 114a: pushf 114b: pop %rax 114c: mov -0x8(%rsp),%rax ## BUG!!! 1151: ret ----- ammarfaizi2@integral2:~$ gcc --version gcc (Ubuntu 11.3.0-1ubuntu1~22.04) 11.3.0 Copyright (C) 2021 Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. ammarfaizi2@integral2:~$ clang --version Ubuntu clang version 16.0.0 (++20230124031324+d63e492562f2-1~exp1~20230124151444.705) Target: x86_64-pc-linux-gnu Thread model: posix InstalledDir: /usr/bin ----- test.c: #include static inline void clobber_redzone(void) { register void *rsp __asm__("%rsp"); unsigned long rflags; __asm__ volatile ("pushf; popq %1" : "+r" (rsp), "=r" (rflags)); } static inline void set_red_zone(long *mem, long val) { __asm__ volatile ("movq %[val], %[mem]" : [mem] "=m" (*mem) : [val] "r" (val)); } static inline long get_red_zone(long *mem) { long ret; __asm__ volatile ("movq %[in], %[out]" : [out] "=r" (ret) : [in] "m" (*mem)); return ret; } __attribute__((__noinline__)) long a_leaf_func_with_red_zone(void) { long x; set_red_zone(&x, 1); clobber_redzone(); /* The correct retval is 1 */ return get_red_zone(&x); } int main(void) { printf("ret = %ld\n", a_leaf_func_with_red_zone()); return 0; } -- Ammar Faizi