From patchwork Tue Feb 22 14:01:47 2022 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Stefan Hajnoczi X-Patchwork-Id: 1596007 Return-Path: X-Original-To: incoming@patchwork.ozlabs.org Delivered-To: patchwork-incoming@bilbo.ozlabs.org Authentication-Results: bilbo.ozlabs.org; dkim=fail reason="signature verification failed" (1024-bit key; unprotected) header.d=redhat.com header.i=@redhat.com header.a=rsa-sha256 header.s=mimecast20190719 header.b=a/fJb/Qh; dkim-atps=neutral Authentication-Results: ozlabs.org; spf=pass (sender SPF authorized) smtp.mailfrom=nongnu.org (client-ip=209.51.188.17; helo=lists.gnu.org; envelope-from=qemu-devel-bounces+incoming=patchwork.ozlabs.org@nongnu.org; receiver=) Received: from lists.gnu.org (lists.gnu.org [209.51.188.17]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by bilbo.ozlabs.org (Postfix) with ESMTPS id 4K31Lb46PRz9sGD for ; Wed, 23 Feb 2022 01:11:34 +1100 (AEDT) Received: from localhost ([::1]:36976 helo=lists1p.gnu.org) by lists.gnu.org with esmtp (Exim 4.90_1) (envelope-from ) id 1nMVt4-0005zr-R4 for incoming@patchwork.ozlabs.org; Tue, 22 Feb 2022 09:11:30 -0500 Received: from eggs.gnu.org ([209.51.188.92]:60742) by lists.gnu.org with esmtps (TLS1.2:ECDHE_RSA_AES_256_GCM_SHA384:256) (Exim 4.90_1) (envelope-from ) id 1nMVkp-0003OO-K8 for qemu-devel@nongnu.org; Tue, 22 Feb 2022 09:03:01 -0500 Received: from us-smtp-delivery-124.mimecast.com ([170.10.133.124]:22549) by eggs.gnu.org with esmtps (TLS1.2:ECDHE_RSA_AES_256_GCM_SHA384:256) (Exim 4.90_1) (envelope-from ) id 1nMVkj-00064Q-FO for qemu-devel@nongnu.org; Tue, 22 Feb 2022 09:02:59 -0500 DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=redhat.com; s=mimecast20190719; t=1645538562; h=from:from:reply-to:subject:subject:date:date:message-id:message-id: to:to:cc:cc:mime-version:mime-version:content-type:content-type: content-transfer-encoding:content-transfer-encoding: in-reply-to:in-reply-to:references:references; bh=AF5Sp/89TquTLE2+tiea8vAXBykqv2OvxpkhGk+W5+g=; b=a/fJb/QhYjLaFCrMtMSHAjz84EiZNBbrEXf/E2bbKrHNIZJW8lCP7rRPbE2HdZYtuFdfgi lxSZChWl3htSL2cP490/XUHjHUF+0rOnl2U0Kp6ivgMNayu/+trr0ccfIJwgQZLBc2SxIs XJwTaH4t8UEn3Ls/9YXiYNgdlznZZ6k= Received: from mimecast-mx01.redhat.com (mimecast-mx01.redhat.com [209.132.183.4]) by relay.mimecast.com with ESMTP with STARTTLS (version=TLSv1.2, cipher=TLS_ECDHE_RSA_WITH_AES_256_GCM_SHA384) id us-mta-620-kPPTYBxMN_ydcNP3ZEhQaA-1; Tue, 22 Feb 2022 09:02:39 -0500 X-MC-Unique: kPPTYBxMN_ydcNP3ZEhQaA-1 Received: from smtp.corp.redhat.com (int-mx05.intmail.prod.int.phx2.redhat.com [10.5.11.15]) (using TLSv1.2 with cipher AECDH-AES256-SHA (256/256 bits)) (No client certificate requested) by mimecast-mx01.redhat.com (Postfix) with ESMTPS id 411A26C504; Tue, 22 Feb 2022 14:02:38 +0000 (UTC) Received: from localhost (unknown [10.39.195.202]) by smtp.corp.redhat.com (Postfix) with ESMTP id C93657C3FD; Tue, 22 Feb 2022 14:02:37 +0000 (UTC) From: Stefan Hajnoczi To: qemu-devel@nongnu.org Subject: [PATCH v5 1/4] tls: add macros for coroutine-safe TLS variables Date: Tue, 22 Feb 2022 14:01:47 +0000 Message-Id: <20220222140150.27240-2-stefanha@redhat.com> In-Reply-To: <20220222140150.27240-1-stefanha@redhat.com> References: <20220222140150.27240-1-stefanha@redhat.com> MIME-Version: 1.0 X-Scanned-By: MIMEDefang 2.79 on 10.5.11.15 Authentication-Results: relay.mimecast.com; auth=pass smtp.auth=CUSA124A263 smtp.mailfrom=stefanha@redhat.com X-Mimecast-Spam-Score: 0 X-Mimecast-Originator: redhat.com Received-SPF: pass client-ip=170.10.133.124; envelope-from=stefanha@redhat.com; helo=us-smtp-delivery-124.mimecast.com X-Spam_score_int: -28 X-Spam_score: -2.9 X-Spam_bar: -- X-Spam_report: (-2.9 / 5.0 requ) BAYES_00=-1.9, DKIMWL_WL_HIGH=-0.082, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, DKIM_VALID_EF=-0.1, RCVD_IN_DNSWL_LOW=-0.7, RCVD_IN_MSPIKE_H5=0.001, RCVD_IN_MSPIKE_WL=0.001, SPF_HELO_NONE=0.001, SPF_PASS=-0.001, T_SCC_BODY_TEXT_LINE=-0.01 autolearn=ham autolearn_force=no X-Spam_action: no action X-BeenThere: qemu-devel@nongnu.org X-Mailman-Version: 2.1.29 Precedence: list List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Cc: Fam Zheng , Peter Maydell , Florian Weimer , qemu-block@nongnu.org, Richard Henderson , Stefan Hajnoczi , Paolo Bonzini , Kevin Wolf , Serge Guelton Errors-To: qemu-devel-bounces+incoming=patchwork.ozlabs.org@nongnu.org Sender: "Qemu-devel" Compiler optimizations can cache TLS values across coroutine yield points, resulting in stale values from the previous thread when a coroutine is re-entered by a new thread. Serge Guelton developed an __attribute__((noinline)) wrapper and tested it with clang and gcc. I formatted his idea according to QEMU's coding style and wrote documentation. The compiler can still optimize based on analyzing noinline code, so an asm volatile barrier with an output constraint is required to prevent unwanted optimizations. Buglink: https://bugzilla.redhat.com/show_bug.cgi?id=1952483 Suggested-by: Serge Guelton Signed-off-by: Stefan Hajnoczi --- include/qemu/coroutine-tls.h | 165 +++++++++++++++++++++++++++++++++++ 1 file changed, 165 insertions(+) create mode 100644 include/qemu/coroutine-tls.h diff --git a/include/qemu/coroutine-tls.h b/include/qemu/coroutine-tls.h new file mode 100644 index 0000000000..1558a826aa --- /dev/null +++ b/include/qemu/coroutine-tls.h @@ -0,0 +1,165 @@ +/* + * QEMU Thread Local Storage for coroutines + * + * Copyright Red Hat + * + * SPDX-License-Identifier: LGPL-2.1-or-later + * + * This work is licensed under the terms of the GNU LGPL, version 2.1 or later. + * See the COPYING.LIB file in the top-level directory. + * + * It is forbidden to access Thread Local Storage in coroutines because + * compiler optimizations may cause values to be cached across coroutine + * re-entry. Coroutines can run in more than one thread through the course of + * their life, leading bugs when stale TLS values from the wrong thread are + * used as a result of compiler optimization. + * + * An example is: + * + * ..code-block:: c + * :caption: A coroutine that may see the wrong TLS value + * + * static __thread AioContext *current_aio_context; + * ... + * static void coroutine_fn foo(void) + * { + * aio_notify(current_aio_context); + * qemu_coroutine_yield(); + * aio_notify(current_aio_context); // <-- may be stale after yielding! + * } + * + * This header provides macros for safely defining variables in Thread Local + * Storage: + * + * ..code-block:: c + * :caption: A coroutine that safely uses TLS + * + * QEMU_DEFINE_STATIC_CO_TLS(AioContext *, current_aio_context) + * ... + * static void coroutine_fn foo(void) + * { + * aio_notify(get_current_aio_context()); + * qemu_coroutine_yield(); + * aio_notify(get_current_aio_context()); // <-- safe + * } + */ + +#ifndef QEMU_COROUTINE_TLS_H +#define QEMU_COROUTINE_TLS_H + +/* + * To stop the compiler from caching TLS values we define accessor functions + * with __attribute__((noinline)) plus asm volatile("") to prevent + * optimizations that override noinline. + * + * The compiler can still analyze noinline code and make optimizations based on + * that knowledge, so an inline asm output operand is used to prevent + * optimizations that make assumptions about the address of the TLS variable. + * + * This is fragile and ultimately needs to be solved by a mechanism that is + * guaranteed to work by the compiler (e.g. stackless coroutines), but for now + * we use this approach to prevent issues. + */ + +/** + * QEMU_DECLARE_CO_TLS: + * @type: the variable's C type + * @var: the variable name + * + * Declare an extern variable in Thread Local Storage from a header file: + * + * .. code-block:: c + * :caption: Declaring an extern variable in Thread Local Storage + * + * QEMU_DECLARE_CO_TLS(int, my_count) + * ... + * int c = get_my_count(); + * set_my_count(c + 1); + * *get_ptr_my_count() = 0; + * + * This is a coroutine-safe replacement for the __thread keyword and is + * equivalent to the following code: + * + * .. code-block:: c + * :caption: Declaring a TLS variable using __thread + * + * extern __thread int my_count; + * ... + * int c = my_count; + * my_count = c + 1; + * *(&my_count) = 0; + */ +#define QEMU_DECLARE_CO_TLS(type, var) \ + __attribute__((noinline)) type get_##var(void); \ + __attribute__((noinline)) void set_##var(type v); \ + __attribute__((noinline)) type *get_ptr_##var(void); + +/** + * QEMU_DEFINE_CO_TLS: + * @type: the variable's C type + * @var: the variable name + * + * Define a variable in Thread Local Storage that was previously declared from + * a header file with QEMU_DECLARE_CO_TLS(): + * + * .. code-block:: c + * :caption: Defining a variable in Thread Local Storage + * + * QEMU_DEFINE_CO_TLS(int, my_count) + * + * This is a coroutine-safe replacement for the __thread keyword and is + * equivalent to the following code: + * + * .. code-block:: c + * :caption: Defining a TLS variable using __thread + * + * __thread int my_count; + */ +#define QEMU_DEFINE_CO_TLS(type, var) \ + static __thread type co_tls_##var; \ + type get_##var(void) { asm volatile(""); return co_tls_##var; } \ + void set_##var(type v) { asm volatile(""); co_tls_##var = v; } \ + type *get_ptr_##var(void) \ + { type *ptr = &co_tls_##var; asm volatile("" : "+rm" (ptr)); return ptr; } + +/** + * QEMU_DEFINE_STATIC_CO_TLS: + * @type: the variable's C type + * @var: the variable name + * + * Define a static variable in Thread Local Storage: + * + * .. code-block:: c + * :caption: Defining a static variable in Thread Local Storage + * + * QEMU_DEFINE_STATIC_CO_TLS(int, my_count) + * ... + * int c = get_my_count(); + * set_my_count(c + 1); + * *get_ptr_my_count() = 0; + * + * This is a coroutine-safe replacement for the __thread keyword and is + * equivalent to the following code: + * + * .. code-block:: c + * :caption: Defining a static TLS variable using __thread + * + * static __thread int my_count; + * ... + * int c = my_count; + * my_count = c + 1; + * *(&my_count) = 0; + */ +#define QEMU_DEFINE_STATIC_CO_TLS(type, var) \ + static __thread type co_tls_##var; \ + static __attribute__((noinline, unused)) \ + type get_##var(void) \ + { asm volatile(""); return co_tls_##var; } \ + static __attribute__((noinline, unused)) \ + void set_##var(type v) \ + { asm volatile(""); co_tls_##var = v; } \ + static __attribute__((noinline, unused)) \ + type *get_ptr_##var(void) \ + { type *ptr = &co_tls_##var; asm volatile("" : "+rm" (ptr)); return ptr; } + +#endif /* QEMU_COROUTINE_TLS_H */