1 /* $OpenBSD: syscall_mi.h,v 1.37 2024/12/27 11:57:16 mpi Exp $ */
2
3 /*
4 * Copyright (c) 1982, 1986, 1989, 1993
5 * The Regents of the University of California. All rights reserved.
6 *
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
9 * are met:
10 * 1. Redistributions of source code must retain the above copyright
11 * notice, this list of conditions and the following disclaimer.
12 * 2. Redistributions in binary form must reproduce the above copyright
13 * notice, this list of conditions and the following disclaimer in the
14 * documentation and/or other materials provided with the distribution.
15 * 3. Neither the name of the University nor the names of its contributors
16 * may be used to endorse or promote products derived from this software
17 * without specific prior written permission.
18 *
19 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
20 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
21 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
22 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
23 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
24 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
25 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
26 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
27 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
28 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
29 * SUCH DAMAGE.
30 *
31 * @(#)kern_xxx.c 8.2 (Berkeley) 11/14/93
32 */
33
34 #include <sys/param.h>
35 #include <sys/pledge.h>
36 #include <sys/acct.h>
37 #include <sys/tracepoint.h>
38 #include <sys/syscall.h>
39 #include <sys/signalvar.h>
40 #include <uvm/uvm_extern.h>
41
42 #ifdef KTRACE
43 #include <sys/ktrace.h>
44 #endif
45
46 #include "dt.h"
47 #if NDT > 0
48 #include <dev/dt/dtvar.h>
49 #endif
50
51 /*
52 * Check if a system call is entered from precisely correct location
53 */
54 static inline int
pin_check(struct proc * p,register_t code)55 pin_check(struct proc *p, register_t code)
56 {
57 extern char sigcodecall[], sigcoderet[], sigcodecall[];
58 struct pinsyscall *pin = NULL, *ppin, *plibcpin;
59 struct process *pr = p->p_p;
60 vaddr_t addr;
61 int error = 0;
62
63 /* point at start of syscall instruction */
64 addr = (vaddr_t)PROC_PC(p) - (vaddr_t)(sigcoderet - sigcodecall);
65 ppin = &pr->ps_pin;
66 plibcpin = &pr->ps_libcpin;
67
68 /*
69 * System calls come from the following places, checks are ordered
70 * by most common case:
71 * 1) dynamic binary: syscalls in libc.so (in the ps_libcpin region)
72 * 2a) static binary: syscalls in main program (in the ps_pin region)
73 * 2b) dynamic binary: syscalls in ld.so (in the ps_pin region)
74 * 3) sigtramp, containing only sigreturn(2)
75 */
76 if (plibcpin->pn_pins &&
77 addr >= plibcpin->pn_start && addr < plibcpin->pn_end)
78 pin = plibcpin;
79 else if (ppin->pn_pins &&
80 addr >= ppin->pn_start && addr < ppin->pn_end)
81 pin = ppin;
82 else if (PROC_PC(p) == pr->ps_sigcoderet) {
83 if (code == SYS_sigreturn)
84 return (0);
85 error = EPERM;
86 goto die;
87 }
88 if (pin) {
89 if (code >= pin->pn_npins || pin->pn_pins[code] == 0)
90 error = ENOSYS;
91 else if (pin->pn_pins[code] + pin->pn_start == addr)
92 ; /* correct location */
93 else if (pin->pn_pins[code] == (u_int)-1)
94 ; /* multiple locations, hopefully a boring operation */
95 else
96 error = ENOSYS;
97 } else
98 error = ENOSYS;
99 if (error == 0)
100 return (0);
101 die:
102 #ifdef KTRACE
103 if (KTRPOINT(p, KTR_PINSYSCALL))
104 ktrpinsyscall(p, error, code, addr);
105 #endif
106 KERNEL_LOCK();
107 /* XXX remove or simplify this uprintf() call after OpenBSD 7.5 release */
108 uprintf("%s[%d]: pinsyscalls addr %lx code %ld, pinoff 0x%x "
109 "(pin%s %d %lx-%lx %lx) (libcpin%s %d %lx-%lx %lx) error %d\n",
110 p->p_p->ps_comm, p->p_p->ps_pid, addr, code,
111 (pin && code < pin->pn_npins) ? pin->pn_pins[code] : -1,
112 pin == ppin ? "(Y)" : "", ppin->pn_npins,
113 ppin->pn_start, ppin->pn_end, ppin->pn_end - ppin->pn_start,
114 pin == plibcpin ? "(Y)" : "", plibcpin->pn_npins,
115 plibcpin->pn_start, plibcpin->pn_end, plibcpin->pn_end - plibcpin->pn_start,
116 error);
117 p->p_p->ps_acflag |= APINSYS;
118
119 /* Try to stop threads immediately, because this process is suspect */
120 if (P_HASSIBLING(p))
121 single_thread_set(p, SINGLE_UNWIND | SINGLE_DEEP);
122 /* Send uncatchable SIGABRT for coredump */
123 sigabort(p);
124 KERNEL_UNLOCK();
125 return (error);
126 }
127
128 /*
129 * The MD setup for a system call has been done; here's the MI part.
130 */
131 static inline int
mi_syscall(struct proc * p,register_t code,const struct sysent * callp,register_t * argp,register_t retval[2])132 mi_syscall(struct proc *p, register_t code, const struct sysent *callp,
133 register_t *argp, register_t retval[2])
134 {
135 uint64_t tval;
136 int lock = !(callp->sy_flags & SY_NOLOCK);
137 int error, pledged;
138
139 /* refresh the thread's cache of the process's creds */
140 refreshcreds(p);
141
142 #ifdef SYSCALL_DEBUG
143 KERNEL_LOCK();
144 scdebug_call(p, code, argp);
145 KERNEL_UNLOCK();
146 #endif
147 TRACEPOINT(raw_syscalls, sys_enter, code, NULL);
148 #if NDT > 0
149 DT_ENTER(syscall, code, callp->sy_argsize, argp);
150 #endif
151 #ifdef KTRACE
152 if (KTRPOINT(p, KTR_SYSCALL)) {
153 /* convert to mask, then include with code */
154 ktrsyscall(p, code, callp->sy_argsize, argp);
155 }
156 #endif
157
158 /* SP must be within MAP_STACK space */
159 if (!uvm_map_inentry(p, &p->p_spinentry, PROC_STACK(p),
160 "[%s]%d/%d sp=%lx inside %lx-%lx: not MAP_STACK\n",
161 uvm_map_inentry_sp, p->p_vmspace->vm_map.sserial))
162 return (EPERM);
163
164 if ((error = pin_check(p, code)))
165 return (error);
166
167 pledged = (p->p_p->ps_flags & PS_PLEDGE);
168 if (pledged && (error = pledge_syscall(p, code, &tval))) {
169 KERNEL_LOCK();
170 error = pledge_fail(p, error, tval);
171 KERNEL_UNLOCK();
172 return (error);
173 }
174 if (lock)
175 KERNEL_LOCK();
176 error = (*callp->sy_call)(p, argp, retval);
177 if (lock)
178 KERNEL_UNLOCK();
179
180 return (error);
181 }
182
183 /*
184 * Finish MI stuff on return, after the registers have been set
185 */
186 static inline void
mi_syscall_return(struct proc * p,register_t code,int error,const register_t retval[2])187 mi_syscall_return(struct proc *p, register_t code, int error,
188 const register_t retval[2])
189 {
190 #ifdef SYSCALL_DEBUG
191 KERNEL_LOCK();
192 scdebug_ret(p, code, error, retval);
193 KERNEL_UNLOCK();
194 #endif
195 #if NDT > 0
196 DT_LEAVE(syscall, code, error, retval[0], retval[1]);
197 #endif
198 TRACEPOINT(raw_syscalls, sys_exit, code, NULL);
199
200 userret(p);
201
202 #ifdef KTRACE
203 if (KTRPOINT(p, KTR_SYSRET))
204 ktrsysret(p, code, error, retval);
205 #endif
206 }
207
208 /*
209 * Finish MI stuff for a new process/thread to return
210 */
211 static inline void
mi_child_return(struct proc * p)212 mi_child_return(struct proc *p)
213 {
214 #if defined(SYSCALL_DEBUG) || defined(KTRACE) || NDT > 0
215 int code = (p->p_flag & P_THREAD) ? SYS___tfork :
216 (p->p_p->ps_flags & PS_PPWAIT) ? SYS_vfork : SYS_fork;
217 const register_t child_retval[2] = { 0, 1 };
218 #endif
219
220 TRACEPOINT(sched, on__cpu, NULL);
221
222 #ifdef SYSCALL_DEBUG
223 KERNEL_LOCK();
224 scdebug_ret(p, code, 0, child_retval);
225 KERNEL_UNLOCK();
226 #endif
227 #if NDT > 0
228 DT_LEAVE(syscall, code, 0, child_retval[0], child_retval[1]);
229 #endif
230 TRACEPOINT(raw_syscalls, sys_exit, code, NULL);
231
232 userret(p);
233
234 #ifdef KTRACE
235 if (KTRPOINT(p, KTR_SYSRET))
236 ktrsysret(p, code, 0, child_retval);
237 #endif
238 }
239
240 /*
241 * Do the specific processing necessary for an AST
242 */
243 static inline void
mi_ast(struct proc * p,int resched)244 mi_ast(struct proc *p, int resched)
245 {
246 if (p->p_flag & P_OWEUPC) {
247 KERNEL_LOCK();
248 ADDUPROF(p);
249 KERNEL_UNLOCK();
250 }
251 if (resched)
252 preempt();
253
254 /*
255 * XXX could move call to userret() here, but
256 * hppa calls ast() in syscall return and sh calls
257 * it after userret()
258 */
259 }
260