From patchwork Mon Jul 2 17:55:43 2018 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Chris von Recklinghausen X-Patchwork-Id: 10502185 Return-Path: Received: from mail.wl.linuxfoundation.org (pdx-wl-mail.web.codeaurora.org [172.30.200.125]) by pdx-korg-patchwork.web.codeaurora.org (Postfix) with ESMTP id 8FD8C60325 for ; Mon, 2 Jul 2018 17:56:05 +0000 (UTC) Received: from mail.wl.linuxfoundation.org (localhost [127.0.0.1]) by mail.wl.linuxfoundation.org (Postfix) with ESMTP id 8261F286B4 for ; Mon, 2 Jul 2018 17:56:05 +0000 (UTC) Received: by mail.wl.linuxfoundation.org (Postfix, from userid 486) id 768DD28E19; Mon, 2 Jul 2018 17:56:05 +0000 (UTC) X-Spam-Checker-Version: SpamAssassin 3.3.1 (2010-03-16) on pdx-wl-mail.web.codeaurora.org X-Spam-Level: X-Spam-Status: No, score=-5.2 required=2.0 tests=BAYES_00, MAILING_LIST_MULTI, RCVD_IN_DNSWL_MED autolearn=ham version=3.3.1 Received: from mother.openwall.net (mother.openwall.net [195.42.179.200]) by mail.wl.linuxfoundation.org (Postfix) with SMTP id DB319286B4 for ; Mon, 2 Jul 2018 17:56:02 +0000 (UTC) Received: (qmail 32141 invoked by uid 550); 2 Jul 2018 17:56:00 -0000 Mailing-List: contact kernel-hardening-help@lists.openwall.com; run by ezmlm Precedence: bulk List-Post: List-Help: List-Unsubscribe: List-Subscribe: List-ID: Delivered-To: mailing list kernel-hardening@lists.openwall.com Received: (qmail 32100 invoked from network); 2 Jul 2018 17:55:59 -0000 From: Chris von Recklinghausen To: keescook@chromium.org, labbott@redhat.com, pabeni@redhat.com, linux-kernel@vger.kernel.org, linux-mm@kvack.org, kernel-hardening@lists.openwall.com Subject: [PATCH v4] add param that allows bootline control of hardened usercopy Date: Mon, 2 Jul 2018 13:55:43 -0400 Message-Id: <1530554143-17218-1-git-send-email-crecklin@redhat.com> X-Scanned-By: MIMEDefang 2.78 on 10.11.54.3 X-Greylist: Sender IP whitelisted, not delayed by milter-greylist-4.5.16 (mx1.redhat.com [10.11.55.6]); Mon, 02 Jul 2018 17:55:47 +0000 (UTC) X-Greylist: inspected by milter-greylist-4.5.16 (mx1.redhat.com [10.11.55.6]); Mon, 02 Jul 2018 17:55:47 +0000 (UTC) for IP:'10.11.54.3' DOMAIN:'int-mx03.intmail.prod.int.rdu2.redhat.com' HELO:'smtp.corp.redhat.com' FROM:'crecklin@redhat.com' RCPT:'' X-Virus-Scanned: ClamAV using ClamSMTP From: root Enabling HARDENED_USERCOPY causes measurable regressions in networking performance, up to 8% under UDP flood. I'm running an a small packet UDP flood using pktgen vs. a host b2b connected. On the receiver side the UDP packets are processed by a simple user space process that just reads and drops them: https://github.com/netoptimizer/network-testing/blob/master/src/udp_sink.c Not very useful from a functional PoV, but it helps to pin-point bottlenecks in the networking stack. When running a kernel with CONFIG_HARDENED_USERCOPY=y, I see a 5-8% regression in the receive tput, compared to the same kernel without this option enabled. With CONFIG_HARDENED_USERCOPY=y, perf shows ~6% of CPU time spent cumulatively in __check_object_size (~4%) and __virt_addr_valid (~2%). The call-chain is: __GI___libc_recvfrom entry_SYSCALL_64_after_hwframe do_syscall_64 __x64_sys_recvfrom __sys_recvfrom inet_recvmsg udp_recvmsg __check_object_size udp_recvmsg() actually calls copy_to_iter() (inlined) and the latters calls check_copy_size() (again, inlined). A generic distro may want to enable HARDENED_USERCOPY in their default kernel config, but at the same time, such distro may want to be able to avoid the performance penalties in with the default configuration and disable the stricter check on a per-boot basis. This change adds a boot parameter that conditionally disables HARDENED_USERCOPY at boot time. Atomics needed by jump label code are not available on platforms with CONFIG_BROKEN_ON_SMP set, so don't enable this functionality on those platforms. v3->v4: fix a couple of nits in commit comments declaration of bypass_usercopy_checks moved inside mm/usercopy.c and made static add blurb to commit comments about not enabling this functionality on platforms with CONFIG_BROKEN_ON_SMP set. v2->v3: add benchmark details to commit comments Don't add new item to Documentation/admin-guide/kernel-parameters.rst rename boot param to "hardened_usercopy=" update description in Documentation/admin-guide/kernel-parameters.txt static_branch_likely -> static_branch_unlikely add __ro_after_init versions of DEFINE_STATIC_KEY_FALSE, DEFINE_STATIC_KEY_TRUE disable_huc_atboot -> enable_checks (strtobool "on" == true) v1->v2: remove CONFIG_HUC_DEFAULT_OFF default is now enabled, boot param disables move check to __check_object_size so as to not break optimization of __builtin_constant_p() include linux/atomic.h before linux/jump_label.h Signed-off-by: Chris von Recklinghausen --- Documentation/admin-guide/kernel-parameters.txt | 11 ++++++++ include/linux/jump_label.h | 6 +++++ mm/usercopy.c | 35 +++++++++++++++++++++++++ 3 files changed, 52 insertions(+) diff --git a/Documentation/admin-guide/kernel-parameters.txt b/Documentation/admin-guide/kernel-parameters.txt index efc7aa7..560d4dc 100644 --- a/Documentation/admin-guide/kernel-parameters.txt +++ b/Documentation/admin-guide/kernel-parameters.txt @@ -816,6 +816,17 @@ disable= [IPV6] See Documentation/networking/ipv6.txt. + hardened_usercopy= + [KNL] Under CONFIG_HARDENED_USERCOPY, whether + hardening is enabled for this boot. Hardened + usercopy checking is used to protect the kernel + from reading or writing beyond known memory + allocation boundaries as a proactive defense + against bounds-checking flaws in the kernel's + copy_to_user()/copy_from_user() interface. + on Perform hardened usercopy checks (default). + off Disable hardened usercopy checks. + disable_radix [PPC] Disable RADIX MMU mode on POWER9 diff --git a/include/linux/jump_label.h b/include/linux/jump_label.h index b46b541..1a0b6f1 100644 --- a/include/linux/jump_label.h +++ b/include/linux/jump_label.h @@ -299,12 +299,18 @@ struct static_key_false { #define DEFINE_STATIC_KEY_TRUE(name) \ struct static_key_true name = STATIC_KEY_TRUE_INIT +#define DEFINE_STATIC_KEY_TRUE_RO(name) \ + struct static_key_true name __ro_after_init = STATIC_KEY_TRUE_INIT + #define DECLARE_STATIC_KEY_TRUE(name) \ extern struct static_key_true name #define DEFINE_STATIC_KEY_FALSE(name) \ struct static_key_false name = STATIC_KEY_FALSE_INIT +#define DEFINE_STATIC_KEY_FALSE_RO(name) \ + struct static_key_false name __ro_after_init = STATIC_KEY_FALSE_INIT + #define DECLARE_STATIC_KEY_FALSE(name) \ extern struct static_key_false name diff --git a/mm/usercopy.c b/mm/usercopy.c index e9e9325..a5f8be5 100644 --- a/mm/usercopy.c +++ b/mm/usercopy.c @@ -20,6 +20,10 @@ #include #include #include +#ifndef CONFIG_BROKEN_ON_SMP +#include +#include +#endif #include /* @@ -240,6 +244,10 @@ static inline void check_heap_object(const void *ptr, unsigned long n, } } +#ifndef CONFIG_BROKEN_ON_SMP +static DEFINE_STATIC_KEY_FALSE_RO(bypass_usercopy_checks); +#endif + /* * Validates that the given object is: * - not bogus address @@ -248,6 +256,11 @@ static inline void check_heap_object(const void *ptr, unsigned long n, */ void __check_object_size(const void *ptr, unsigned long n, bool to_user) { +#ifndef CONFIG_BROKEN_ON_SMP + if (static_branch_unlikely(&bypass_usercopy_checks)) + return; +#endif + /* Skip all tests if size is zero. */ if (!n) return; @@ -279,3 +292,25 @@ void __check_object_size(const void *ptr, unsigned long n, bool to_user) check_kernel_text_object((const unsigned long)ptr, n, to_user); } EXPORT_SYMBOL(__check_object_size); + +#ifndef CONFIG_BROKEN_ON_SMP +EXPORT_SYMBOL(bypass_usercopy_checks); + +static bool enable_checks __initdata = true; + +static int __init parse_hardened_usercopy(char *str) +{ + return strtobool(str, &enable_checks); +} + +__setup("hardened_usercopy=", parse_hardened_usercopy); + +static int __init set_hardened_usercopy(void) +{ + if (enable_checks == false) + static_branch_enable(&bypass_usercopy_checks); + return 1; +} + +late_initcall(set_hardened_usercopy); +#endif