From patchwork Wed Jul 27 17:11:30 2022 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Daniel Bristot de Oliveira X-Patchwork-Id: 12930659 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 38F25C04A68 for ; Wed, 27 Jul 2022 18:09:39 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S243270AbiG0SJi (ORCPT ); Wed, 27 Jul 2022 14:09:38 -0400 Received: from lindbergh.monkeyblade.net ([23.128.96.19]:51820 "EHLO lindbergh.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S243345AbiG0SJH (ORCPT ); Wed, 27 Jul 2022 14:09:07 -0400 Received: from ams.source.kernel.org (ams.source.kernel.org [IPv6:2604:1380:4601:e00::1]) by lindbergh.monkeyblade.net (Postfix) with ESMTPS id 73912CE50D; Wed, 27 Jul 2022 10:12:19 -0700 (PDT) Received: from smtp.kernel.org (relay.kernel.org [52.25.139.140]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by ams.source.kernel.org (Postfix) with ESMTPS id DC9BBB8200D; Wed, 27 Jul 2022 17:12:17 +0000 (UTC) Received: by smtp.kernel.org (Postfix) with ESMTPSA id 22001C433D6; Wed, 27 Jul 2022 17:12:10 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=kernel.org; s=k20201202; t=1658941936; bh=qv2SZkmPVZ8CTohftcki0CGonzCuKvmTJUkw5fAZ9M8=; h=From:To:Cc:Subject:Date:In-Reply-To:References:From; b=SF4VuHUdh4zY1UeNQUzZMABQitrUDzyZm9BcO+PhNzw3qsiANuywu0QhkzW3dIoVk U5NzWtdNmTVqJ7b/F1y3daDvgeJUly++7bgdH21LstPqksQSzJy+L0ZjedDaZKwlVS QdCHG2uWsPT4B5tsHtneedmPaGwG+ayk/mu5vuvXlaLEYXzQWKD8PJ4vdpGnMfDdXt vSfXTbdcLGFszUTM4D50Fg9JTZ5a7AkQeoFIGNzZ/yvEsH8TMtffrdUC3b5ggMlkKx Dc1W420EdRTb0QB5d3Dlp15X8OnfxaZaO4UD5eRoT71eQPX2usHHxwWUW0TEkorMvf qO60cK6jfrjCA== From: Daniel Bristot de Oliveira To: Steven Rostedt Cc: Daniel Bristot de Oliveira , Wim Van Sebroeck , Guenter Roeck , Jonathan Corbet , Ingo Molnar , Thomas Gleixner , Peter Zijlstra , Will Deacon , Catalin Marinas , Marco Elver , Dmitry Vyukov , "Paul E. McKenney" , Shuah Khan , Gabriele Paoloni , Juri Lelli , Clark Williams , Tao Zhou , Randy Dunlap , linux-doc@vger.kernel.org, linux-kernel@vger.kernel.org, linux-trace-devel@vger.kernel.org Subject: [PATCH V8 02/16] rv: Add runtime reactors interface Date: Wed, 27 Jul 2022 19:11:30 +0200 Message-Id: <8be3e66a8f0ab42feb3de4e6a630775b68df2d58.1658940828.git.bristot@kernel.org> X-Mailer: git-send-email 2.35.1 In-Reply-To: References: MIME-Version: 1.0 Precedence: bulk List-ID: X-Mailing-List: linux-trace-devel@vger.kernel.org A runtime monitor can cause a reaction to the detection of an exception on the model's execution. By default, the monitors have tracing reactions, printing the monitor output via tracepoints. But other reactions can be added (on-demand) via this interface. The user interface resembles the kernel tracing interface and presents these files: "available_reactors" - Reading shows the available reactors, one per line. For example: # cat available_reactors nop panic printk "reacting_on" - It is an on/off general switch for reactors, disabling all reactions. "monitors/MONITOR/reactors" - List available reactors, with the select reaction for the given MONITOR inside []. The default one is the nop (no operation) reactor. - Writing the name of a reactor enables it to the given MONITOR. For example: # cat monitors/wip/reactors [nop] panic printk # echo panic > monitors/wip/reactors # cat monitors/wip/reactors nop [panic] printk Cc: Wim Van Sebroeck Cc: Guenter Roeck Cc: Jonathan Corbet Cc: Steven Rostedt Cc: Ingo Molnar Cc: Thomas Gleixner Cc: Peter Zijlstra Cc: Will Deacon Cc: Catalin Marinas Cc: Marco Elver Cc: Dmitry Vyukov Cc: "Paul E. McKenney" Cc: Shuah Khan Cc: Gabriele Paoloni Cc: Juri Lelli Cc: Clark Williams Cc: Tao Zhou Cc: Randy Dunlap Cc: linux-doc@vger.kernel.org Cc: linux-kernel@vger.kernel.org Cc: linux-trace-devel@vger.kernel.org Signed-off-by: Daniel Bristot de Oliveira --- include/linux/rv.h | 17 ++ kernel/trace/rv/Kconfig | 11 + kernel/trace/rv/Makefile | 1 + kernel/trace/rv/rv.c | 8 + kernel/trace/rv/rv.h | 35 +++ kernel/trace/rv/rv_reactors.c | 508 ++++++++++++++++++++++++++++++++++ 6 files changed, 580 insertions(+) create mode 100644 kernel/trace/rv/rv_reactors.c diff --git a/include/linux/rv.h b/include/linux/rv.h index d8fa9e8be94a..dcce56987cf7 100644 --- a/include/linux/rv.h +++ b/include/linux/rv.h @@ -24,6 +24,14 @@ union rv_task_monitor { }; +#ifdef CONFIG_RV_REACTORS +struct rv_reactor { + const char *name; + const char *description; + void (*react)(char *msg); +}; +#endif + struct rv_monitor { const char *name; const char *description; @@ -31,6 +39,9 @@ struct rv_monitor { int (*enable)(void); void (*disable)(void); void (*reset)(void); +#ifdef CONFIG_RV_REACTORS + void (*react)(char *msg); +#endif }; bool rv_monitoring_on(void); @@ -39,5 +50,11 @@ int rv_register_monitor(struct rv_monitor *monitor); int rv_get_task_monitor_slot(void); void rv_put_task_monitor_slot(int slot); +#ifdef CONFIG_RV_REACTORS +bool rv_reacting_on(void); +int rv_unregister_reactor(struct rv_reactor *reactor); +int rv_register_reactor(struct rv_reactor *reactor); +#endif /* CONFIG_RV_REACTORS */ + #endif /* CONFIG_RV */ #endif /* _LINUX_RV_H */ diff --git a/kernel/trace/rv/Kconfig b/kernel/trace/rv/Kconfig index 6d127cdb00dd..3eb5d48ab4f6 100644 --- a/kernel/trace/rv/Kconfig +++ b/kernel/trace/rv/Kconfig @@ -10,3 +10,14 @@ menuconfig RV theorem proving). RV works by analyzing the trace of the system's actual execution, comparing it against a formal specification of the system behavior. + +config RV_REACTORS + bool "Runtime verification reactors" + default y + depends on RV + help + Enables the online runtime verification reactors. A runtime + monitor can cause a reaction to the detection of an exception + on the model's execution. By default, the monitors have + tracing reactions, printing the monitor output via tracepoints, + but other reactions can be added (on-demand) via this interface. diff --git a/kernel/trace/rv/Makefile b/kernel/trace/rv/Makefile index fd995379df67..8944274d9b41 100644 --- a/kernel/trace/rv/Makefile +++ b/kernel/trace/rv/Makefile @@ -1,3 +1,4 @@ # SPDX-License-Identifier: GPL-2.0 obj-$(CONFIG_RV) += rv.o +obj-$(CONFIG_RV_REACTORS) += rv_reactors.o diff --git a/kernel/trace/rv/rv.c b/kernel/trace/rv/rv.c index e1179a3bcc0c..45cf64eb2600 100644 --- a/kernel/trace/rv/rv.c +++ b/kernel/trace/rv/rv.c @@ -353,6 +353,10 @@ static int create_monitor_dir(struct rv_monitor_def *mdef) goto out_remove_root; } + retval = reactor_populate_monitor(mdef); + if (retval) + goto out_remove_root; + return 0; out_remove_root: @@ -748,6 +752,7 @@ int rv_unregister_monitor(struct rv_monitor *monitor) int __init rv_init_interface(void) { struct dentry *tmp; + int retval; rv_root.root_dir = rv_create_dir("rv", NULL); if (!rv_root.root_dir) @@ -771,6 +776,9 @@ int __init rv_init_interface(void) &monitoring_on_fops); if (!tmp) goto out_err; + retval = init_rv_reactors(rv_root.root_dir); + if (retval) + goto out_err; turn_monitoring_on(); diff --git a/kernel/trace/rv/rv.h b/kernel/trace/rv/rv.h index 47643840116e..a284a369fee0 100644 --- a/kernel/trace/rv/rv.h +++ b/kernel/trace/rv/rv.h @@ -18,13 +18,27 @@ struct rv_interface { #define rv_remove tracefs_remove #define MAX_RV_MONITOR_NAME_SIZE 32 +#define MAX_RV_REACTOR_NAME_SIZE 32 extern struct mutex rv_interface_lock; +#ifdef CONFIG_RV_REACTORS +struct rv_reactor_def { + struct list_head list; + struct rv_reactor *reactor; + /* protected by the monitor interface lock */ + int counter; +}; +#endif + struct rv_monitor_def { struct list_head list; struct rv_monitor *monitor; struct dentry *root_d; +#ifdef CONFIG_RV_REACTORS + struct rv_reactor_def *rdef; + bool reacting; +#endif bool task_monitor; }; @@ -32,3 +46,24 @@ struct dentry *get_monitors_root(void); int init_rv_monitors(struct dentry *root_dir); int rv_disable_monitor(struct rv_monitor_def *mdef); int rv_enable_monitor(struct rv_monitor_def *mdef); + +#ifdef CONFIG_RV_REACTORS +int reactor_populate_monitor(struct rv_monitor_def *mdef); +void reactor_cleanup_monitor(struct rv_monitor_def *mdef); +int init_rv_reactors(struct dentry *root_dir); +#else +static inline int reactor_populate_monitor(struct rv_monitor_def *mdef) +{ + return 0; +} + +static inline void reactor_cleanup_monitor(struct rv_monitor_def *mdef) +{ + return; +} + +static inline int init_rv_reactors(struct dentry *root_dir) +{ + return 0; +} +#endif diff --git a/kernel/trace/rv/rv_reactors.c b/kernel/trace/rv/rv_reactors.c new file mode 100644 index 000000000000..a6522c196382 --- /dev/null +++ b/kernel/trace/rv/rv_reactors.c @@ -0,0 +1,508 @@ +// SPDX-License-Identifier: GPL-2.0 +/* + * Copyright (C) 2019-2022 Red Hat, Inc. Daniel Bristot de Oliveira + * + * Runtime reactor interface. + * + * A runtime monitor can cause a reaction to the detection of an + * exception on the model's execution. By default, the monitors have + * tracing reactions, printing the monitor output via tracepoints. + * But other reactions can be added (on-demand) via this interface. + * + * == Registering reactors == + * + * The struct rv_reactor defines a callback function to be executed + * in case of a model exception happens. The callback function + * receives a message to be (optionally) printed before executing + * the reaction. + * + * A RV reactor is registered via: + * int rv_register_reactor(struct rv_reactor *reactor) + * And unregistered via: + * int rv_unregister_reactor(struct rv_reactor *reactor) + * + * These functions are exported to modules, enabling reactors to be + * dynamically loaded. + * + * == User interface == + * + * The user interface resembles the kernel tracing interface and + * presents these files: + * + * "available_reactors" + * - List the available reactors, one per line. + * + * For example: + * # cat available_reactors + * nop + * panic + * printk + * + * "reacting_on" + * - It is an on/off general switch for reactors, disabling + * all reactions. + * + * "monitors/MONITOR/reactors" + * - List available reactors, with the select reaction for the given + * MONITOR inside []. The default one is the nop (no operation) + * reactor. + * - Writing the name of an reactor enables it to the given + * MONITOR. + * + * For example: + * # cat monitors/wip/reactors + * [nop] + * panic + * printk + * # echo panic > monitors/wip/reactors + * # cat monitors/wip/reactors + * nop + * [panic] + * printk + */ + +#include + +#include "rv.h" + +/* + * Interface for the reactor register. + */ +static LIST_HEAD(rv_reactors_list); + +static struct rv_reactor_def *get_reactor_rdef_by_name(char *name) +{ + struct rv_reactor_def *r; + + list_for_each_entry(r, &rv_reactors_list, list) { + if (strcmp(name, r->reactor->name) == 0) + return r; + } + return NULL; +} + +/* + * Available reactors seq functions. + */ +static int reactors_show(struct seq_file *m, void *p) +{ + struct rv_reactor_def *rea_def = p; + + seq_printf(m, "%s\n", rea_def->reactor->name); + return 0; +} + +static void reactors_stop(struct seq_file *m, void *p) +{ + mutex_unlock(&rv_interface_lock); +} + +static void *reactors_start(struct seq_file *m, loff_t *pos) +{ + mutex_lock(&rv_interface_lock); + return seq_list_start(&rv_reactors_list, *pos); +} + +static void *reactors_next(struct seq_file *m, void *p, loff_t *pos) +{ + return seq_list_next(p, &rv_reactors_list, pos); +} + +/* + * available_reactors seq definition. + */ +static const struct seq_operations available_reactors_seq_ops = { + .start = reactors_start, + .next = reactors_next, + .stop = reactors_stop, + .show = reactors_show +}; + +/* + * available_reactors interface. + */ +static int available_reactors_open(struct inode *inode, struct file *file) +{ + return seq_open(file, &available_reactors_seq_ops); +}; + +static const struct file_operations available_reactors_ops = { + .open = available_reactors_open, + .read = seq_read, + .llseek = seq_lseek, + .release = seq_release +}; + +/* + * Monitor's reactor file. + */ +static int monitor_reactor_show(struct seq_file *m, void *p) +{ + struct rv_monitor_def *mdef = m->private; + struct rv_reactor_def *rdef = p; + + if (mdef->rdef == rdef) + seq_printf(m, "[%s]\n", rdef->reactor->name); + else + seq_printf(m, "%s\n", rdef->reactor->name); + return 0; +} + +/* + * available_reactors seq definition. + */ +static const struct seq_operations monitor_reactors_seq_ops = { + .start = reactors_start, + .next = reactors_next, + .stop = reactors_stop, + .show = monitor_reactor_show +}; + +static void monitor_swap_reactors(struct rv_monitor_def *mdef, struct rv_reactor_def *rdef, + bool reacting) +{ + bool monitor_enabled; + + /* nothing to do */ + if (mdef->rdef == rdef) + return; + + monitor_enabled = mdef->monitor->enabled; + if (monitor_enabled) + rv_disable_monitor(mdef); + + /* swap reactor's usage */ + mdef->rdef->counter--; + rdef->counter++; + + mdef->rdef = rdef; + mdef->reacting = reacting; + mdef->monitor->react = rdef->reactor->react; + + if (monitor_enabled) + rv_enable_monitor(mdef); +} + +static ssize_t +monitor_reactors_write(struct file *file, const char __user *user_buf, + size_t count, loff_t *ppos) +{ + char buff[MAX_RV_REACTOR_NAME_SIZE + 2]; + struct rv_monitor_def *mdef; + struct rv_reactor_def *rdef; + struct seq_file *seq_f; + int retval = -EINVAL; + bool enable; + char *ptr; + int len; + + if (count < 1 || count > MAX_RV_REACTOR_NAME_SIZE + 1) + return -EINVAL; + + memset(buff, 0, sizeof(buff)); + + retval = simple_write_to_buffer(buff, sizeof(buff) - 1, ppos, user_buf, count); + if (retval < 0) + return -EFAULT; + + ptr = strim(buff); + + len = strlen(ptr); + if (!len) + return count; + + /* + * See monitor_reactors_open() + */ + seq_f = file->private_data; + mdef = seq_f->private; + + mutex_lock(&rv_interface_lock); + + retval = -EINVAL; + + list_for_each_entry(rdef, &rv_reactors_list, list) { + if (strcmp(ptr, rdef->reactor->name) != 0) + continue; + + if (rdef == get_reactor_rdef_by_name("nop")) + enable = false; + else + enable = true; + + monitor_swap_reactors(mdef, rdef, enable); + + retval = count; + break; + } + + mutex_unlock(&rv_interface_lock); + + return retval; +} + +/* + * available_reactors interface. + */ +static int monitor_reactors_open(struct inode *inode, struct file *file) +{ + struct rv_monitor_def *mdef = inode->i_private; + struct seq_file *seq_f; + int ret; + + ret = seq_open(file, &monitor_reactors_seq_ops); + if (ret < 0) + return ret; + + /* + * seq_open stores the seq_file on the file->private data. + */ + seq_f = file->private_data; + + /* + * Copy the create file "private" data to the seq_file private data. + */ + seq_f->private = mdef; + + return 0; +}; + +static const struct file_operations monitor_reactors_ops = { + .open = monitor_reactors_open, + .read = seq_read, + .llseek = seq_lseek, + .release = seq_release, + .write = monitor_reactors_write +}; + +static int __rv_register_reactor(struct rv_reactor *reactor) +{ + struct rv_reactor_def *r; + + list_for_each_entry(r, &rv_reactors_list, list) { + if (strcmp(reactor->name, r->reactor->name) == 0) { + pr_info("Reactor %s is already registered\n", reactor->name); + return -EINVAL; + } + } + + r = kzalloc(sizeof(struct rv_reactor_def), GFP_KERNEL); + if (!r) + return -ENOMEM; + + r->reactor = reactor; + r->counter = 0; + + list_add_tail(&r->list, &rv_reactors_list); + + return 0; +} + +/** + * rv_register_reactor - register a rv reactor. + * @reactor: The rv_reactor to be registered. + * + * Returns 0 if successful, error otherwise. + */ +int rv_register_reactor(struct rv_reactor *reactor) +{ + int retval = 0; + + if (strlen(reactor->name) >= MAX_RV_REACTOR_NAME_SIZE) { + pr_info("Reactor %s has a name longer than %d\n", + reactor->name, MAX_RV_MONITOR_NAME_SIZE); + return -EINVAL; + } + + mutex_lock(&rv_interface_lock); + retval = __rv_register_reactor(reactor); + mutex_unlock(&rv_interface_lock); + return retval; +} + +/** + * rv_unregister_reactor - unregister a rv reactor. + * @reactor: The rv_reactor to be unregistered. + * + * Returns 0 if successful, error otherwise. + */ +int rv_unregister_reactor(struct rv_reactor *reactor) +{ + struct rv_reactor_def *ptr, *next; + + mutex_lock(&rv_interface_lock); + + list_for_each_entry_safe(ptr, next, &rv_reactors_list, list) { + if (strcmp(reactor->name, ptr->reactor->name) == 0) { + + if (!ptr->counter) { + list_del(&ptr->list); + } else { + printk(KERN_WARNING + "rv: the rv_reactor %s is in use by %d monitor(s)\n", + ptr->reactor->name, ptr->counter); + printk(KERN_WARNING "rv: the rv_reactor %s cannot be removed\n", + ptr->reactor->name); + return -EBUSY; + } + } + } + + mutex_unlock(&rv_interface_lock); + return 0; +} + +/* + * reacting_on interface. + */ +static bool __read_mostly reacting_on; + +/** + * rv_reacting_on - checks if reacting is on + * + * Returns 1 if on, 0 otherwise. + */ +bool rv_reacting_on(void) +{ + /* Ensures that concurrent monitors read consistent reacting_on */ + smp_rmb(); + return READ_ONCE(reacting_on); +} + +static ssize_t reacting_on_read_data(struct file *filp, + char __user *user_buf, + size_t count, loff_t *ppos) +{ + char *buff; + + buff = rv_reacting_on() ? "1\n" : "0\n"; + + return simple_read_from_buffer(user_buf, count, ppos, buff, strlen(buff)+1); +} + +static void turn_reacting_off(void) +{ + WRITE_ONCE(reacting_on, false); + /* Ensures that concurrent monitors read consistent reacting_on */ + smp_wmb(); +} + +static void turn_reacting_on(void) +{ + WRITE_ONCE(reacting_on, true); + /* Ensures that concurrent monitors read consistent reacting_on */ + smp_wmb(); +} + +static ssize_t reacting_on_write_data(struct file *filp, const char __user *user_buf, + size_t count, loff_t *ppos) +{ + int retval; + bool val; + + retval = kstrtobool_from_user(user_buf, count, &val); + if (retval) + return retval; + + mutex_lock(&rv_interface_lock); + + if (val) + turn_reacting_on(); + else + turn_reacting_off(); + + /* + * Wait for the execution of all events to finish + * before returning to user-space. + */ + tracepoint_synchronize_unregister(); + + mutex_unlock(&rv_interface_lock); + + return count; +} + +static const struct file_operations reacting_on_fops = { + .open = simple_open, + .llseek = no_llseek, + .write = reacting_on_write_data, + .read = reacting_on_read_data, +}; + +/** + * reactor_populate_monitor - creates per monitor reactors file + * @mdef: monitor's definition. + * + * Returns 0 if successful, error otherwise. + */ +int reactor_populate_monitor(struct rv_monitor_def *mdef) +{ + struct dentry *tmp; + + tmp = rv_create_file("reactors", RV_MODE_WRITE, mdef->root_d, mdef, &monitor_reactors_ops); + if (!tmp) + return -ENOMEM; + + /* + * Configure as the rv_nop reactor. + */ + mdef->rdef = get_reactor_rdef_by_name("nop"); + mdef->rdef->counter++; + mdef->reacting = false; + + return 0; +} + +/** + * reactor_cleanup_monitor - cleanup a monitor reference + * @mdef: monitor's definition. + */ +void reactor_cleanup_monitor(struct rv_monitor_def *mdef) +{ + lockdep_assert_held(&rv_interface_lock); + mdef->rdef->counter--; + WARN_ON_ONCE(mdef->rdef->counter < 0); +} + +/* + * Nop reactor register + */ +static void rv_nop_reaction(char *msg) +{ +} + +static struct rv_reactor rv_nop = { + .name = "nop", + .description = "no-operation reactor: do nothing.", + .react = rv_nop_reaction +}; + +int init_rv_reactors(struct dentry *root_dir) +{ + struct dentry *available, *reacting; + int retval; + + available = rv_create_file("available_reactors", RV_MODE_READ, root_dir, NULL, + &available_reactors_ops); + if (!available) + goto out_err; + + reacting = rv_create_file("reacting_on", RV_MODE_WRITE, root_dir, NULL, &reacting_on_fops); + if (!reacting) + goto rm_available; + + retval = __rv_register_reactor(&rv_nop); + if (retval) + goto rm_reacting; + + turn_reacting_on(); + + return 0; + +rm_reacting: + rv_remove(reacting); +rm_available: + rv_remove(available); +out_err: + return -ENOMEM; +}