xref: /freebsd/sys/amd64/ia32/ia32_syscall.c (revision f0a75d274af375d15b97b830966b99a02b7db911)
1 /*-
2  * Copyright (C) 1994, David Greenman
3  * Copyright (c) 1990, 1993
4  *	The Regents of the University of California.  All rights reserved.
5  *
6  * This code is derived from software contributed to Berkeley by
7  * the University of Utah, and William Jolitz.
8  *
9  * Redistribution and use in source and binary forms, with or without
10  * modification, are permitted provided that the following conditions
11  * are met:
12  * 1. Redistributions of source code must retain the above copyright
13  *    notice, this list of conditions and the following disclaimer.
14  * 2. Redistributions in binary form must reproduce the above copyright
15  *    notice, this list of conditions and the following disclaimer in the
16  *    documentation and/or other materials provided with the distribution.
17  * 3. All advertising materials mentioning features or use of this software
18  *    must display the following acknowledgement:
19  *	This product includes software developed by the University of
20  *	California, Berkeley and its contributors.
21  * 4. Neither the name of the University nor the names of its contributors
22  *    may be used to endorse or promote products derived from this software
23  *    without specific prior written permission.
24  *
25  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
26  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
27  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
28  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
29  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
30  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
31  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
32  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
33  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
34  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
35  * SUCH DAMAGE.
36  */
37 
38 #include <sys/cdefs.h>
39 __FBSDID("$FreeBSD$");
40 
41 /*
42  * 386 Trap and System call handling
43  */
44 
45 #include "opt_clock.h"
46 #include "opt_cpu.h"
47 #include "opt_isa.h"
48 #include "opt_ktrace.h"
49 
50 #include <sys/param.h>
51 #include <sys/bus.h>
52 #include <sys/systm.h>
53 #include <sys/proc.h>
54 #include <sys/pioctl.h>
55 #include <sys/kernel.h>
56 #include <sys/ktr.h>
57 #include <sys/lock.h>
58 #include <sys/mutex.h>
59 #include <sys/ptrace.h>
60 #include <sys/resourcevar.h>
61 #include <sys/signalvar.h>
62 #include <sys/syscall.h>
63 #include <sys/sysctl.h>
64 #include <sys/sysent.h>
65 #include <sys/uio.h>
66 #include <sys/vmmeter.h>
67 #ifdef KTRACE
68 #include <sys/ktrace.h>
69 #endif
70 #include <security/audit/audit.h>
71 
72 #include <vm/vm.h>
73 #include <vm/vm_param.h>
74 #include <vm/pmap.h>
75 #include <vm/vm_kern.h>
76 #include <vm/vm_map.h>
77 #include <vm/vm_page.h>
78 #include <vm/vm_extern.h>
79 
80 #include <machine/cpu.h>
81 #include <machine/intr_machdep.h>
82 #include <machine/md_var.h>
83 
84 #define	IDTVEC(name)	__CONCAT(X,name)
85 
86 extern inthand_t IDTVEC(int0x80_syscall), IDTVEC(rsvd);
87 extern const char *freebsd32_syscallnames[];
88 
89 void ia32_syscall(struct trapframe *frame);	/* Called from asm code */
90 
91 void
92 ia32_syscall(struct trapframe *frame)
93 {
94 	caddr_t params;
95 	int i;
96 	struct sysent *callp;
97 	struct thread *td = curthread;
98 	struct proc *p = td->td_proc;
99 	register_t orig_tf_rflags;
100 	int error;
101 	int narg;
102 	u_int32_t args[8];
103 	u_int64_t args64[8];
104 	u_int code;
105 	ksiginfo_t ksi;
106 
107 	/*
108 	 * note: PCPU_LAZY_INC() can only be used if we can afford
109 	 * occassional inaccuracy in the count.
110 	 */
111 	PCPU_LAZY_INC(cnt.v_syscall);
112 
113 	td->td_pticks = 0;
114 	td->td_frame = frame;
115 	if (td->td_ucred != p->p_ucred)
116 		cred_update_thread(td);
117 	params = (caddr_t)frame->tf_rsp + sizeof(u_int32_t);
118 	code = frame->tf_rax;
119 	orig_tf_rflags = frame->tf_rflags;
120 
121 	if (p->p_sysent->sv_prepsyscall) {
122 		/*
123 		 * The prep code is MP aware.
124 		 */
125 		(*p->p_sysent->sv_prepsyscall)(frame, args, &code, &params);
126 	} else {
127 		/*
128 		 * Need to check if this is a 32 bit or 64 bit syscall.
129 		 * fuword is MP aware.
130 		 */
131 		if (code == SYS_syscall) {
132 			/*
133 			 * Code is first argument, followed by actual args.
134 			 */
135 			code = fuword32(params);
136 			params += sizeof(int);
137 		} else if (code == SYS___syscall) {
138 			/*
139 			 * Like syscall, but code is a quad, so as to maintain
140 			 * quad alignment for the rest of the arguments.
141 			 * We use a 32-bit fetch in case params is not
142 			 * aligned.
143 			 */
144 			code = fuword32(params);
145 			params += sizeof(quad_t);
146 		}
147 	}
148 
149  	if (p->p_sysent->sv_mask)
150  		code &= p->p_sysent->sv_mask;
151 
152  	if (code >= p->p_sysent->sv_size)
153  		callp = &p->p_sysent->sv_table[0];
154   	else
155  		callp = &p->p_sysent->sv_table[code];
156 
157 	narg = callp->sy_narg;
158 
159 	/*
160 	 * copyin and the ktrsyscall()/ktrsysret() code is MP-aware
161 	 */
162 	if (params != NULL && narg != 0)
163 		error = copyin(params, (caddr_t)args,
164 		    (u_int)(narg * sizeof(int)));
165 	else
166 		error = 0;
167 
168 	for (i = 0; i < narg; i++)
169 		args64[i] = args[i];
170 
171 #ifdef KTRACE
172 	if (KTRPOINT(td, KTR_SYSCALL))
173 		ktrsyscall(code, narg, args64);
174 #endif
175 	CTR4(KTR_SYSC, "syscall enter thread %p pid %d proc %s code %d", td,
176 	    td->td_proc->p_pid, td->td_proc->p_comm, code);
177 
178 	if (error == 0) {
179 		td->td_retval[0] = 0;
180 		td->td_retval[1] = frame->tf_rdx;
181 
182 		STOPEVENT(p, S_SCE, narg);
183 
184 		PTRACESTOP_SC(p, td, S_PT_SCE);
185 
186 		AUDIT_SYSCALL_ENTER(code, td);
187 		error = (*callp->sy_call)(td, args64);
188 		AUDIT_SYSCALL_EXIT(error, td);
189 	}
190 
191 	switch (error) {
192 	case 0:
193 		frame->tf_rax = td->td_retval[0];
194 		frame->tf_rdx = td->td_retval[1];
195 		frame->tf_rflags &= ~PSL_C;
196 		break;
197 
198 	case ERESTART:
199 		/*
200 		 * Reconstruct pc, assuming lcall $X,y is 7 bytes,
201 		 * int 0x80 is 2 bytes. We saved this in tf_err.
202 		 */
203 		frame->tf_rip -= frame->tf_err;
204 		break;
205 
206 	case EJUSTRETURN:
207 		break;
208 
209 	default:
210  		if (p->p_sysent->sv_errsize) {
211  			if (error >= p->p_sysent->sv_errsize)
212   				error = -1;	/* XXX */
213    			else
214   				error = p->p_sysent->sv_errtbl[error];
215 		}
216 		frame->tf_rax = error;
217 		frame->tf_rflags |= PSL_C;
218 		break;
219 	}
220 
221 	/*
222 	 * Traced syscall.
223 	 */
224 	if (orig_tf_rflags & PSL_T) {
225 		frame->tf_rflags &= ~PSL_T;
226 		ksiginfo_init_trap(&ksi);
227 		ksi.ksi_signo = SIGTRAP;
228 		ksi.ksi_code = TRAP_TRACE;
229 		ksi.ksi_addr = (void *)frame->tf_rip;
230 		trapsignal(td, &ksi);
231 	}
232 
233 	/*
234 	 * Check for misbehavior.
235 	 */
236 	WITNESS_WARN(WARN_PANIC, NULL, "System call %s returning",
237 	    (code >= 0 && code < SYS_MAXSYSCALL) ? freebsd32_syscallnames[code] : "???");
238 	KASSERT(td->td_critnest == 0,
239 	    ("System call %s returning in a critical section",
240 	    (code >= 0 && code < SYS_MAXSYSCALL) ? freebsd32_syscallnames[code] : "???"));
241 	KASSERT(td->td_locks == 0,
242 	    ("System call %s returning with %d locks held",
243 	    (code >= 0 && code < SYS_MAXSYSCALL) ? freebsd32_syscallnames[code] : "???",
244 	    td->td_locks));
245 
246 	/*
247 	 * Handle reschedule and other end-of-syscall issues
248 	 */
249 	userret(td, frame);
250 
251 	CTR4(KTR_SYSC, "syscall exit thread %p pid %d proc %s code %d", td,
252 	    td->td_proc->p_pid, td->td_proc->p_comm, code);
253 #ifdef KTRACE
254 	if (KTRPOINT(td, KTR_SYSRET))
255 		ktrsysret(code, error, td->td_retval[0]);
256 #endif
257 
258 	/*
259 	 * This works because errno is findable through the
260 	 * register set.  If we ever support an emulation where this
261 	 * is not the case, this code will need to be revisited.
262 	 */
263 	STOPEVENT(p, S_SCX, code);
264 
265 	PTRACESTOP_SC(p, td, S_PT_SCX);
266 }
267 
268 
269 static void
270 ia32_syscall_enable(void *dummy)
271 {
272 
273  	setidt(IDT_SYSCALL, &IDTVEC(int0x80_syscall), SDT_SYSIGT, SEL_UPL, 0);
274 }
275 
276 static void
277 ia32_syscall_disable(void *dummy)
278 {
279 
280  	setidt(IDT_SYSCALL, &IDTVEC(rsvd), SDT_SYSIGT, SEL_KPL, 0);
281 }
282 
283 SYSINIT(ia32_syscall, SI_SUB_EXEC, SI_ORDER_ANY, ia32_syscall_enable, NULL);
284 SYSUNINIT(ia32_syscall, SI_SUB_EXEC, SI_ORDER_ANY, ia32_syscall_disable, NULL);
285