LKML Archive on lore.kernel.org help / color / mirror / Atom feed
From: Fenghua Yu <fenghua.yu@intel.com> To: "Thomas Gleixner" <tglx@linutronix.de>, "Ingo Molnar" <mingo@elte.hu>, "H. Peter Anvin" <hpa@linux.intel.com>, "Ashok Raj" <ashok.raj@intel.com>, "Ravi V Shankar" <ravi.v.shankar@intel.com>, "Tony Luck" <tony.luck@intel.com>, "Dave Hansen" <dave.hansen@intel.com>, "Rafael Wysocki" <rafael.j.wysocki@intel.com>, "Arjan van de Ven" <arjan@infradead.org>, "Alan Cox" <alan@linux.intel.com> Cc: "x86" <x86@kernel.org>, "linux-kernel" <linux-kernel@vger.kernel.org>, Fenghua Yu <fenghua.yu@intel.com> Subject: [PATCH 15/15] x86/split_lock: Add split lock user space test in selftest Date: Mon, 14 May 2018 11:52:25 -0700 [thread overview] Message-ID: <1526323945-211107-16-git-send-email-fenghua.yu@intel.com> (raw) In-Reply-To: <1526323945-211107-1-git-send-email-fenghua.yu@intel.com> The test generates split locked access from user space. If #AC exception is enabled for split lock, a SIGBUS is delivered to the test. Signed-off-by: Fenghua Yu <fenghua.yu@intel.com> --- tools/testing/selftests/x86/Makefile | 3 +- tools/testing/selftests/x86/split_lock_user_test.c | 187 +++++++++++++++++++++ 2 files changed, 189 insertions(+), 1 deletion(-) create mode 100644 tools/testing/selftests/x86/split_lock_user_test.c diff --git a/tools/testing/selftests/x86/Makefile b/tools/testing/selftests/x86/Makefile index d744991c0f4f..19d65729b100 100644 --- a/tools/testing/selftests/x86/Makefile +++ b/tools/testing/selftests/x86/Makefile @@ -11,7 +11,8 @@ CAN_BUILD_X86_64 := $(shell ./check_cc.sh $(CC) trivial_64bit_program.c) TARGETS_C_BOTHBITS := single_step_syscall sysret_ss_attrs syscall_nt test_mremap_vdso \ check_initial_reg_state sigreturn iopl mpx-mini-test ioperm \ - protection_keys test_vdso test_vsyscall + protection_keys test_vdso test_vsyscall \ + split_lock_user_test TARGETS_C_32BIT_ONLY := entry_from_vm86 syscall_arg_fault test_syscall_vdso unwind_vdso \ test_FCMOV test_FCOMI test_FISTTP \ vdso_restorer diff --git a/tools/testing/selftests/x86/split_lock_user_test.c b/tools/testing/selftests/x86/split_lock_user_test.c new file mode 100644 index 000000000000..e6f3367dfb6b --- /dev/null +++ b/tools/testing/selftests/x86/split_lock_user_test.c @@ -0,0 +1,187 @@ +// SPDX-License-Identifier: GPL-2.0 +/* + * Copyright (C) 2018 Intel Corporation + * Author: Fenghua Yu <fenghua.yu@intel.com> + * + * Pre-request: + * Kernel is built with CONFIG_SPLIT_LOCK_AC=y. + * Split lock is enabled. If not, enable it by: + * #echo 1 >/sys/kernel/split_lock/enable + * + * Usage: + * Run the test alone and it should show: + * TEST PASS: locked instruction is re-executed. + * TEST PASS: Caught SIGBUS/#AC due to split locked access + * + * Or launch the test from perf and watch "split_lock_user" event count. + * #/perf stat -e exceptions:split_lock* /root/split_lock_user_test_64 + * TEST PASS: locked instruction is re-executed. + * TEST PASS: Caught SIGBUS/#AC due to split locked access + * + * Performance counter stats for '/root/split_lock_user_test_64': + * + * 1 exceptions:split_lock_user + * 0 exceptions:split_lock_kernel + * + * 1.000893458 seconds time elapsed + */ +#include <sys/types.h> +#include <sys/stat.h> +#include <fcntl.h> +#include <stdlib.h> +#include <signal.h> +#include <stdio.h> +#include <unistd.h> +#include <string.h> + +int split_lock_exception; + +void catch_sigbus(int sig) +{ + split_lock_exception = 1; + printf("TEST PASS: Caught SIGBUS/#AC due to split locked access\n"); + + exit(-1); +} + +int split_lock_ac_enabled(void) +{ + int fd, enable, ret; + char buf[16]; + + fd = open("/sys/kernel/split_lock/enable", O_RDONLY); + if (fd < 0) + return 0; + + if (read(fd, buf, sizeof(int)) < 0) { + ret = 0; + goto out; + } + + enable = atoi(buf); + if (enable == 1) + ret = 1; + else + ret = 0; + +out: + close(fd); + + return ret; +} + +int setup_user_mode(char *user_mode_reaction) +{ + ssize_t count; + int fd; + + if (strcmp(user_mode_reaction, "sigbus") && + strcmp(user_mode_reaction, "re-execute")) + return -1; + + fd = open("/sys/kernel/split_lock/user_mode", O_RDWR); + if (fd < 0) + return -1; + + count = write(fd, user_mode_reaction, strlen(user_mode_reaction)); + if (count != strlen(user_mode_reaction)) + return -1; + + close(fd); + + return 0; +} + +void do_split_locked_inst(int a, int b, int *iptr) +{ + /* + * Since eax is equal to *iptr, the instruction loads value in b + * (i.e. 11) into iptr. If the instruction is executed correctly, + * the content of *iptr is changed * to 11 from previous value 1. + * + * Accessing iptr cross two cache lines will trigger #AC in hardware + * and kernel either delivers SIGBUS to this process or re-execute + * the instruction depending on /sys/kernel/split_lock/user_mode + * setting. + */ + asm volatile ("movl %1, %%eax\n\t" + "movl %1, %0\n\t" + "lock\n cmpxchgl %2, %0\n\t" + : "=m" (*iptr) + : "r"(a), "r"(b) + : "%eax"); +} + +void test_re_execute(int a, int b, int *iptr) +{ + setup_user_mode("re-execute"); + + /* The locked instruction triggers #AC and then it's re-executed. */ + do_split_locked_inst(a, b, iptr); + + if (*iptr == b) { + printf("TEST PASS: locked instruction is re-executed.\n"); + } else { + printf("TEST FAIL: No #AC exception is caught and "); + printf("instruction is not executed correctly.\n"); + } +} + +void test_sigbus(int a, int b, int *iptr) +{ + setup_user_mode("sigbus"); + + /* + * The locked instruction triggers #AC and kernel delivers SIGBUS + * to this process. + */ + do_split_locked_inst(a, b, iptr); +} + +int main(int argc, char **argv) +{ + int *iptr, a = 10, b = 11; + char *cptr; + + if (!split_lock_ac_enabled()) { + printf("#AC exception for split lock is NOT enabled!!\n"); + printf("Please make sure split lock feature is supported,\n"); + printf("CONFIG_SPLIT_LOCK_AC is turned on,\n"); + printf("and /sys/kernel/split_lock/enable is 1 before test!\n"); + + return 0; + } + + signal(SIGBUS, catch_sigbus); + + /* + * Enable Alignment Checking on x86_64. + * This will generate alignment check on not only split lock but also + * on any misalignment. + * Turn on this for reference only. + */ + /* __asm__("pushf\norl $0x40000,(%rsp)\npopf"); */ + + /* aligned_alloc() provides 64-byte aligned memory */ + cptr = (char *)aligned_alloc(64, 128); + + /* Increment the pointer by 61, making it misaligned */ + iptr = (int *)(cptr + 61); + + /* Initial value in iptr is 1. */ + *iptr = 1; + + test_re_execute(a, b, iptr); + /* + * The split lock is disabled after the last locked instruction is + * re-executed. + * + * Wait for the split lock is re-enabled again before next test. + */ + sleep(1); + test_sigbus(a, b, iptr); + + free(cptr); + + return 0; +} -- 2.5.0
next prev parent reply other threads:[~2018-05-14 18:52 UTC|newest] Thread overview: 31+ messages / expand[flat|nested] mbox.gz Atom feed top 2018-05-14 18:52 [PATCH 0/15] x86/split_lock: Enable #AC exception for split locked accesses Fenghua Yu 2018-05-14 18:52 ` [PATCH 01/15] x86/split_lock: Add CONFIG and enumerate #AC exception for split locked access feature Fenghua Yu 2018-05-15 15:36 ` Dave Hansen 2018-05-15 15:41 ` Fenghua Yu 2018-05-15 15:54 ` Dave Hansen 2018-05-14 18:52 ` [PATCH 02/15] x86/split_lock: Set up #AC exception for split locked accesses Fenghua Yu 2018-05-14 18:52 ` [PATCH 03/15] x86/split_lock: Handle #AC exception for split lock in kernel mode Fenghua Yu 2018-05-15 15:51 ` Dave Hansen 2018-05-15 16:35 ` Luck, Tony 2018-05-15 17:21 ` Fenghua Yu 2018-05-16 16:44 ` Dave Hansen 2018-05-16 21:35 ` Fenghua Yu 2018-05-14 18:52 ` [PATCH 04/15] x86/split_lock: Use non locked bit set instruction in set_cpu_cap Fenghua Yu 2018-05-14 18:52 ` [PATCH 05/15] x86/split_lock: Use non atomic set and clear bit instructions to clear cpufeature Fenghua Yu 2018-05-14 18:52 ` [PATCH 06/15] x86/split_lock: Save #AC setting for split lock in BIOS in boot time and restore the setting in reboot Fenghua Yu 2018-05-14 18:52 ` [PATCH 07/15] x86/split_lock: Handle suspend/hibernate and resume Fenghua Yu 2018-05-14 21:42 ` Rafael J. Wysocki 2018-05-14 18:52 ` [PATCH 08/15] x86/split_lock: Set split lock during EFI runtime service Fenghua Yu 2018-05-14 18:52 ` [PATCH 09/15] x86/split_lock: Explicitly enable or disable #AC for split locked accesses Fenghua Yu 2018-05-15 16:15 ` Dave Hansen 2018-05-15 17:29 ` Fenghua Yu 2018-05-16 16:37 ` Dave Hansen 2018-05-14 18:52 ` [PATCH 10/15] x86/split_lock: Add a sysfs interface to allow user to enable or disable split lock during run time Fenghua Yu 2018-05-14 18:52 ` [PATCH 11/15] x86/split_lock: Add sysfs interface to control user mode behavior Fenghua Yu 2018-05-14 18:52 ` [PATCH 12/15] x86/split_lock: Add sysfs interface to show and control BIOS split lock setting Fenghua Yu 2018-05-14 18:52 ` [PATCH 13/15] x86/split_lock: Trace #AC exception for split lock Fenghua Yu 2018-05-14 18:52 ` [PATCH 14/15] x86/split_lock: Add CONFIG and testing sysfs interface Fenghua Yu 2018-05-14 18:52 ` Fenghua Yu [this message] 2018-05-15 15:10 ` [PATCH 0/15] x86/split_lock: Enable #AC exception for split locked accesses Dave Hansen 2018-05-15 16:26 ` Alan Cox 2018-05-15 16:30 ` Dave Hansen
Reply instructions: You may reply publicly to this message via plain-text email using any one of the following methods: * Save the following mbox file, import it into your mail client, and reply-to-all from there: mbox Avoid top-posting and favor interleaved quoting: https://en.wikipedia.org/wiki/Posting_style#Interleaved_style * Reply using the --to, --cc, and --in-reply-to switches of git-send-email(1): git send-email \ --in-reply-to=1526323945-211107-16-git-send-email-fenghua.yu@intel.com \ --to=fenghua.yu@intel.com \ --cc=alan@linux.intel.com \ --cc=arjan@infradead.org \ --cc=ashok.raj@intel.com \ --cc=dave.hansen@intel.com \ --cc=hpa@linux.intel.com \ --cc=linux-kernel@vger.kernel.org \ --cc=mingo@elte.hu \ --cc=rafael.j.wysocki@intel.com \ --cc=ravi.v.shankar@intel.com \ --cc=tglx@linutronix.de \ --cc=tony.luck@intel.com \ --cc=x86@kernel.org \ /path/to/YOUR_REPLY https://kernel.org/pub/software/scm/git/docs/git-send-email.html * If your mail client supports setting the In-Reply-To header via mailto: links, try the mailto: linkBe sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions for how to clone and mirror all data and code used for this inbox; as well as URLs for NNTP newsgroup(s).