1 // SPDX-License-Identifier: GPL-2.0
2
3 #include <linux/context_tracking.h>
4 #include <linux/entry-common.h>
5 #include <linux/resume_user_mode.h>
6 #include <linux/highmem.h>
7 #include <linux/jump_label.h>
8 #include <linux/kmsan.h>
9 #include <linux/livepatch.h>
10 #include <linux/audit.h>
11 #include <linux/tick.h>
12
13 #include "common.h"
14
15 #define CREATE_TRACE_POINTS
16 #include <trace/events/syscalls.h>
17
syscall_enter_audit(struct pt_regs * regs,long syscall)18 static inline void syscall_enter_audit(struct pt_regs *regs, long syscall)
19 {
20 if (unlikely(audit_context())) {
21 unsigned long args[6];
22
23 syscall_get_arguments(current, regs, args);
24 audit_syscall_entry(syscall, args[0], args[1], args[2], args[3]);
25 }
26 }
27
syscall_trace_enter(struct pt_regs * regs,long syscall,unsigned long work)28 long syscall_trace_enter(struct pt_regs *regs, long syscall,
29 unsigned long work)
30 {
31 long ret = 0;
32
33 /*
34 * Handle Syscall User Dispatch. This must comes first, since
35 * the ABI here can be something that doesn't make sense for
36 * other syscall_work features.
37 */
38 if (work & SYSCALL_WORK_SYSCALL_USER_DISPATCH) {
39 if (syscall_user_dispatch(regs))
40 return -1L;
41 }
42
43 /* Handle ptrace */
44 if (work & (SYSCALL_WORK_SYSCALL_TRACE | SYSCALL_WORK_SYSCALL_EMU)) {
45 ret = ptrace_report_syscall_entry(regs);
46 if (ret || (work & SYSCALL_WORK_SYSCALL_EMU))
47 return -1L;
48 }
49
50 /* Do seccomp after ptrace, to catch any tracer changes. */
51 if (work & SYSCALL_WORK_SECCOMP) {
52 ret = __secure_computing();
53 if (ret == -1L)
54 return ret;
55 }
56
57 /* Either of the above might have changed the syscall number */
58 syscall = syscall_get_nr(current, regs);
59
60 if (unlikely(work & SYSCALL_WORK_SYSCALL_TRACEPOINT)) {
61 trace_sys_enter(regs, syscall);
62 /*
63 * Probes or BPF hooks in the tracepoint may have changed the
64 * system call number as well.
65 */
66 syscall = syscall_get_nr(current, regs);
67 }
68
69 syscall_enter_audit(regs, syscall);
70
71 return ret ? : syscall;
72 }
73
syscall_enter_from_user_mode_prepare(struct pt_regs * regs)74 noinstr void syscall_enter_from_user_mode_prepare(struct pt_regs *regs)
75 {
76 enter_from_user_mode(regs);
77 instrumentation_begin();
78 local_irq_enable();
79 instrumentation_end();
80 }
81
82 /* Workaround to allow gradual conversion of architecture code */
arch_do_signal_or_restart(struct pt_regs * regs)83 void __weak arch_do_signal_or_restart(struct pt_regs *regs) { }
84
85 /**
86 * exit_to_user_mode_loop - do any pending work before leaving to user space
87 * @regs: Pointer to pt_regs on entry stack
88 * @ti_work: TIF work flags as read by the caller
89 */
exit_to_user_mode_loop(struct pt_regs * regs,unsigned long ti_work)90 __always_inline unsigned long exit_to_user_mode_loop(struct pt_regs *regs,
91 unsigned long ti_work)
92 {
93 /*
94 * Before returning to user space ensure that all pending work
95 * items have been completed.
96 */
97 while (ti_work & EXIT_TO_USER_MODE_WORK) {
98
99 local_irq_enable_exit_to_user(ti_work);
100
101 if (ti_work & (_TIF_NEED_RESCHED | _TIF_NEED_RESCHED_LAZY))
102 schedule();
103
104 if (ti_work & _TIF_UPROBE)
105 uprobe_notify_resume(regs);
106
107 if (ti_work & _TIF_PATCH_PENDING)
108 klp_update_patch_state(current);
109
110 if (ti_work & (_TIF_SIGPENDING | _TIF_NOTIFY_SIGNAL))
111 arch_do_signal_or_restart(regs);
112
113 if (ti_work & _TIF_NOTIFY_RESUME)
114 resume_user_mode_work(regs);
115
116 /* Architecture specific TIF work */
117 arch_exit_to_user_mode_work(regs, ti_work);
118
119 /*
120 * Disable interrupts and reevaluate the work flags as they
121 * might have changed while interrupts and preemption was
122 * enabled above.
123 */
124 local_irq_disable_exit_to_user();
125
126 /* Check if any of the above work has queued a deferred wakeup */
127 tick_nohz_user_enter_prepare();
128
129 ti_work = read_thread_flags();
130 }
131
132 /* Return the latest work state for arch_exit_to_user_mode() */
133 return ti_work;
134 }
135
136 /*
137 * If SYSCALL_EMU is set, then the only reason to report is when
138 * SINGLESTEP is set (i.e. PTRACE_SYSEMU_SINGLESTEP). This syscall
139 * instruction has been already reported in syscall_enter_from_user_mode().
140 */
report_single_step(unsigned long work)141 static inline bool report_single_step(unsigned long work)
142 {
143 if (work & SYSCALL_WORK_SYSCALL_EMU)
144 return false;
145
146 return work & SYSCALL_WORK_SYSCALL_EXIT_TRAP;
147 }
148
syscall_exit_work(struct pt_regs * regs,unsigned long work)149 void syscall_exit_work(struct pt_regs *regs, unsigned long work)
150 {
151 bool step;
152
153 /*
154 * If the syscall was rolled back due to syscall user dispatching,
155 * then the tracers below are not invoked for the same reason as
156 * the entry side was not invoked in syscall_trace_enter(): The ABI
157 * of these syscalls is unknown.
158 */
159 if (work & SYSCALL_WORK_SYSCALL_USER_DISPATCH) {
160 if (unlikely(current->syscall_dispatch.on_dispatch)) {
161 current->syscall_dispatch.on_dispatch = false;
162 return;
163 }
164 }
165
166 audit_syscall_exit(regs);
167
168 if (work & SYSCALL_WORK_SYSCALL_TRACEPOINT)
169 trace_sys_exit(regs, syscall_get_return_value(current, regs));
170
171 step = report_single_step(work);
172 if (step || work & SYSCALL_WORK_SYSCALL_TRACE)
173 ptrace_report_syscall_exit(regs, step);
174 }
175
irqentry_enter_from_user_mode(struct pt_regs * regs)176 noinstr void irqentry_enter_from_user_mode(struct pt_regs *regs)
177 {
178 enter_from_user_mode(regs);
179 }
180
irqentry_exit_to_user_mode(struct pt_regs * regs)181 noinstr void irqentry_exit_to_user_mode(struct pt_regs *regs)
182 {
183 instrumentation_begin();
184 exit_to_user_mode_prepare(regs);
185 instrumentation_end();
186 exit_to_user_mode();
187 }
188
irqentry_enter(struct pt_regs * regs)189 noinstr irqentry_state_t irqentry_enter(struct pt_regs *regs)
190 {
191 irqentry_state_t ret = {
192 .exit_rcu = false,
193 };
194
195 if (user_mode(regs)) {
196 irqentry_enter_from_user_mode(regs);
197 return ret;
198 }
199
200 /*
201 * If this entry hit the idle task invoke ct_irq_enter() whether
202 * RCU is watching or not.
203 *
204 * Interrupts can nest when the first interrupt invokes softirq
205 * processing on return which enables interrupts.
206 *
207 * Scheduler ticks in the idle task can mark quiescent state and
208 * terminate a grace period, if and only if the timer interrupt is
209 * not nested into another interrupt.
210 *
211 * Checking for rcu_is_watching() here would prevent the nesting
212 * interrupt to invoke ct_irq_enter(). If that nested interrupt is
213 * the tick then rcu_flavor_sched_clock_irq() would wrongfully
214 * assume that it is the first interrupt and eventually claim
215 * quiescent state and end grace periods prematurely.
216 *
217 * Unconditionally invoke ct_irq_enter() so RCU state stays
218 * consistent.
219 *
220 * TINY_RCU does not support EQS, so let the compiler eliminate
221 * this part when enabled.
222 */
223 if (!IS_ENABLED(CONFIG_TINY_RCU) && is_idle_task(current)) {
224 /*
225 * If RCU is not watching then the same careful
226 * sequence vs. lockdep and tracing is required
227 * as in irqentry_enter_from_user_mode().
228 */
229 lockdep_hardirqs_off(CALLER_ADDR0);
230 ct_irq_enter();
231 instrumentation_begin();
232 kmsan_unpoison_entry_regs(regs);
233 trace_hardirqs_off_finish();
234 instrumentation_end();
235
236 ret.exit_rcu = true;
237 return ret;
238 }
239
240 /*
241 * If RCU is watching then RCU only wants to check whether it needs
242 * to restart the tick in NOHZ mode. rcu_irq_enter_check_tick()
243 * already contains a warning when RCU is not watching, so no point
244 * in having another one here.
245 */
246 lockdep_hardirqs_off(CALLER_ADDR0);
247 instrumentation_begin();
248 kmsan_unpoison_entry_regs(regs);
249 rcu_irq_enter_check_tick();
250 trace_hardirqs_off_finish();
251 instrumentation_end();
252
253 return ret;
254 }
255
raw_irqentry_exit_cond_resched(void)256 void raw_irqentry_exit_cond_resched(void)
257 {
258 if (!preempt_count()) {
259 /* Sanity check RCU and thread stack */
260 rcu_irq_exit_check_preempt();
261 if (IS_ENABLED(CONFIG_DEBUG_ENTRY))
262 WARN_ON_ONCE(!on_thread_stack());
263 if (need_resched())
264 preempt_schedule_irq();
265 }
266 }
267 #ifdef CONFIG_PREEMPT_DYNAMIC
268 #if defined(CONFIG_HAVE_PREEMPT_DYNAMIC_CALL)
269 DEFINE_STATIC_CALL(irqentry_exit_cond_resched, raw_irqentry_exit_cond_resched);
270 #elif defined(CONFIG_HAVE_PREEMPT_DYNAMIC_KEY)
271 DEFINE_STATIC_KEY_TRUE(sk_dynamic_irqentry_exit_cond_resched);
dynamic_irqentry_exit_cond_resched(void)272 void dynamic_irqentry_exit_cond_resched(void)
273 {
274 if (!static_branch_unlikely(&sk_dynamic_irqentry_exit_cond_resched))
275 return;
276 raw_irqentry_exit_cond_resched();
277 }
278 #endif
279 #endif
280
irqentry_exit(struct pt_regs * regs,irqentry_state_t state)281 noinstr void irqentry_exit(struct pt_regs *regs, irqentry_state_t state)
282 {
283 lockdep_assert_irqs_disabled();
284
285 /* Check whether this returns to user mode */
286 if (user_mode(regs)) {
287 irqentry_exit_to_user_mode(regs);
288 } else if (!regs_irqs_disabled(regs)) {
289 /*
290 * If RCU was not watching on entry this needs to be done
291 * carefully and needs the same ordering of lockdep/tracing
292 * and RCU as the return to user mode path.
293 */
294 if (state.exit_rcu) {
295 instrumentation_begin();
296 /* Tell the tracer that IRET will enable interrupts */
297 trace_hardirqs_on_prepare();
298 lockdep_hardirqs_on_prepare();
299 instrumentation_end();
300 ct_irq_exit();
301 lockdep_hardirqs_on(CALLER_ADDR0);
302 return;
303 }
304
305 instrumentation_begin();
306 if (IS_ENABLED(CONFIG_PREEMPTION))
307 irqentry_exit_cond_resched();
308
309 /* Covers both tracing and lockdep */
310 trace_hardirqs_on();
311 instrumentation_end();
312 } else {
313 /*
314 * IRQ flags state is correct already. Just tell RCU if it
315 * was not watching on entry.
316 */
317 if (state.exit_rcu)
318 ct_irq_exit();
319 }
320 }
321
irqentry_nmi_enter(struct pt_regs * regs)322 irqentry_state_t noinstr irqentry_nmi_enter(struct pt_regs *regs)
323 {
324 irqentry_state_t irq_state;
325
326 irq_state.lockdep = lockdep_hardirqs_enabled();
327
328 __nmi_enter();
329 lockdep_hardirqs_off(CALLER_ADDR0);
330 lockdep_hardirq_enter();
331 ct_nmi_enter();
332
333 instrumentation_begin();
334 kmsan_unpoison_entry_regs(regs);
335 trace_hardirqs_off_finish();
336 ftrace_nmi_enter();
337 instrumentation_end();
338
339 return irq_state;
340 }
341
irqentry_nmi_exit(struct pt_regs * regs,irqentry_state_t irq_state)342 void noinstr irqentry_nmi_exit(struct pt_regs *regs, irqentry_state_t irq_state)
343 {
344 instrumentation_begin();
345 ftrace_nmi_exit();
346 if (irq_state.lockdep) {
347 trace_hardirqs_on_prepare();
348 lockdep_hardirqs_on_prepare();
349 }
350 instrumentation_end();
351
352 ct_nmi_exit();
353 lockdep_hardirq_exit();
354 if (irq_state.lockdep)
355 lockdep_hardirqs_on(CALLER_ADDR0);
356 __nmi_exit();
357 }
358