1 /* SPDX-License-Identifier: GPL-2.0-only */
2 /*
3  * Tracing hooks
4  *
5  * Copyright (C) 2008-2009 Red Hat, Inc.  All rights reserved.
6  *
7  * This file defines hook entry points called by core code where
8  * user tracing/debugging support might need to do something.  These
9  * entry points are called tracehook_*().  Each hook declared below
10  * has a detailed kerneldoc comment giving the context (locking et
11  * al) from which it is called, and the meaning of its return value.
12  *
13  * Each function here typically has only one call site, so it is ok
14  * to have some nontrivial tracehook_*() inlines.  In all cases, the
15  * fast path when no tracing is enabled should be very short.
16  *
17  * The purpose of this file and the tracehook_* layer is to consolidate
18  * the interface that the kernel core and arch code uses to enable any
19  * user debugging or tracing facility (such as ptrace).  The interfaces
20  * here are carefully documented so that maintainers of core and arch
21  * code do not need to think about the implementation details of the
22  * tracing facilities.  Likewise, maintainers of the tracing code do not
23  * need to understand all the calling core or arch code in detail, just
24  * documented circumstances of each call, such as locking conditions.
25  *
26  * If the calling core code changes so that locking is different, then
27  * it is ok to change the interface documented here.  The maintainer of
28  * core code changing should notify the maintainers of the tracing code
29  * that they need to work out the change.
30  *
31  * Some tracehook_*() inlines take arguments that the current tracing
32  * implementations might not necessarily use.  These function signatures
33  * are chosen to pass in all the information that is on hand in the
34  * caller and might conceivably be relevant to a tracer, so that the
35  * core code won't have to be updated when tracing adds more features.
36  * If a call site changes so that some of those parameters are no longer
37  * already on hand without extra work, then the tracehook_* interface
38  * can change so there is no make-work burden on the core code.  The
39  * maintainer of core code changing should notify the maintainers of the
40  * tracing code that they need to work out the change.
41  */
42 
43 #ifndef _LINUX_TRACEHOOK_H
44 #define _LINUX_TRACEHOOK_H	1
45 
46 #include <linux/sched.h>
47 #include <linux/ptrace.h>
48 #include <linux/security.h>
49 #include <linux/task_work.h>
50 #include <linux/memcontrol.h>
51 #include <linux/blk-cgroup.h>
52 struct linux_binprm;
53 
54 /*
55  * ptrace report for syscall entry and exit looks identical.
56  */
ptrace_report_syscall(struct pt_regs * regs,unsigned long message)57 static inline int ptrace_report_syscall(struct pt_regs *regs,
58 					unsigned long message)
59 {
60 	int ptrace = current->ptrace;
61 
62 	if (!(ptrace & PT_PTRACED))
63 		return 0;
64 
65 	current->ptrace_message = message;
66 	ptrace_notify(SIGTRAP | ((ptrace & PT_TRACESYSGOOD) ? 0x80 : 0));
67 
68 	/*
69 	 * this isn't the same as continuing with a signal, but it will do
70 	 * for normal use.  strace only continues with a signal if the
71 	 * stopping signal is not SIGTRAP.  -brl
72 	 */
73 	if (current->exit_code) {
74 		send_sig(current->exit_code, current, 1);
75 		current->exit_code = 0;
76 	}
77 
78 	current->ptrace_message = 0;
79 	return fatal_signal_pending(current);
80 }
81 
82 /**
83  * tracehook_report_syscall_entry - task is about to attempt a system call
84  * @regs:		user register state of current task
85  *
86  * This will be called if %TIF_SYSCALL_TRACE or %TIF_SYSCALL_EMU have been set,
87  * when the current task has just entered the kernel for a system call.
88  * Full user register state is available here.  Changing the values
89  * in @regs can affect the system call number and arguments to be tried.
90  * It is safe to block here, preventing the system call from beginning.
91  *
92  * Returns zero normally, or nonzero if the calling arch code should abort
93  * the system call.  That must prevent normal entry so no system call is
94  * made.  If @task ever returns to user mode after this, its register state
95  * is unspecified, but should be something harmless like an %ENOSYS error
96  * return.  It should preserve enough information so that syscall_rollback()
97  * can work (see asm-generic/syscall.h).
98  *
99  * Called without locks, just after entering kernel mode.
100  */
tracehook_report_syscall_entry(struct pt_regs * regs)101 static inline __must_check int tracehook_report_syscall_entry(
102 	struct pt_regs *regs)
103 {
104 	return ptrace_report_syscall(regs, PTRACE_EVENTMSG_SYSCALL_ENTRY);
105 }
106 
107 /**
108  * tracehook_report_syscall_exit - task has just finished a system call
109  * @regs:		user register state of current task
110  * @step:		nonzero if simulating single-step or block-step
111  *
112  * This will be called if %TIF_SYSCALL_TRACE has been set, when the
113  * current task has just finished an attempted system call.  Full
114  * user register state is available here.  It is safe to block here,
115  * preventing signals from being processed.
116  *
117  * If @step is nonzero, this report is also in lieu of the normal
118  * trap that would follow the system call instruction because
119  * user_enable_block_step() or user_enable_single_step() was used.
120  * In this case, %TIF_SYSCALL_TRACE might not be set.
121  *
122  * Called without locks, just before checking for pending signals.
123  */
tracehook_report_syscall_exit(struct pt_regs * regs,int step)124 static inline void tracehook_report_syscall_exit(struct pt_regs *regs, int step)
125 {
126 	if (step)
127 		user_single_step_report(regs);
128 	else
129 		ptrace_report_syscall(regs, PTRACE_EVENTMSG_SYSCALL_EXIT);
130 }
131 
132 /**
133  * tracehook_signal_handler - signal handler setup is complete
134  * @stepping:		nonzero if debugger single-step or block-step in use
135  *
136  * Called by the arch code after a signal handler has been set up.
137  * Register and stack state reflects the user handler about to run.
138  * Signal mask changes have already been made.
139  *
140  * Called without locks, shortly before returning to user mode
141  * (or handling more signals).
142  */
tracehook_signal_handler(int stepping)143 static inline void tracehook_signal_handler(int stepping)
144 {
145 	if (stepping)
146 		ptrace_notify(SIGTRAP);
147 }
148 
149 /**
150  * set_notify_resume - cause tracehook_notify_resume() to be called
151  * @task:		task that will call tracehook_notify_resume()
152  *
153  * Calling this arranges that @task will call tracehook_notify_resume()
154  * before returning to user mode.  If it's already running in user mode,
155  * it will enter the kernel and call tracehook_notify_resume() soon.
156  * If it's blocked, it will not be woken.
157  */
set_notify_resume(struct task_struct * task)158 static inline void set_notify_resume(struct task_struct *task)
159 {
160 #ifdef TIF_NOTIFY_RESUME
161 	if (!test_and_set_tsk_thread_flag(task, TIF_NOTIFY_RESUME))
162 		kick_process(task);
163 #endif
164 }
165 
166 /**
167  * tracehook_notify_resume - report when about to return to user mode
168  * @regs:		user-mode registers of @current task
169  *
170  * This is called when %TIF_NOTIFY_RESUME has been set.  Now we are
171  * about to return to user mode, and the user state in @regs can be
172  * inspected or adjusted.  The caller in arch code has cleared
173  * %TIF_NOTIFY_RESUME before the call.  If the flag gets set again
174  * asynchronously, this will be called again before we return to
175  * user mode.
176  *
177  * Called without locks.
178  */
tracehook_notify_resume(struct pt_regs * regs)179 static inline void tracehook_notify_resume(struct pt_regs *regs)
180 {
181 	clear_thread_flag(TIF_NOTIFY_RESUME);
182 	/*
183 	 * This barrier pairs with task_work_add()->set_notify_resume() after
184 	 * hlist_add_head(task->task_works);
185 	 */
186 	smp_mb__after_atomic();
187 	if (unlikely(current->task_works))
188 		task_work_run();
189 
190 #ifdef CONFIG_KEYS_REQUEST_CACHE
191 	if (unlikely(current->cached_requested_key)) {
192 		key_put(current->cached_requested_key);
193 		current->cached_requested_key = NULL;
194 	}
195 #endif
196 
197 	mem_cgroup_handle_over_high();
198 	blkcg_maybe_throttle_current();
199 }
200 
201 #endif	/* <linux/tracehook.h> */
202