linux_sysvec.c revision 14331
1/*-
2 * Copyright (c) 1994-1995 S�ren Schmidt
3 * All rights reserved.
4 *
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
7 * are met:
8 * 1. Redistributions of source code must retain the above copyright
9 *    notice, this list of conditions and the following disclaimer
10 *    in this position and unchanged.
11 * 2. Redistributions in binary form must reproduce the above copyright
12 *    notice, this list of conditions and the following disclaimer in the
13 *    documentation and/or other materials provided with the distribution.
14 * 3. The name of the author may not be used to endorse or promote products
15 *    derived from this software withough specific prior written permission
16 *
17 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
18 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
19 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
20 * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
21 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
22 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
23 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
24 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
25 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
26 * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
27 *
28 *  $Id: linux_sysent.c,v 1.4 1996/01/14 10:59:57 sos Exp $
29 */
30
31/* XXX we use functions that might not exist. */
32#define	COMPAT_43	1
33
34#include <sys/param.h>
35#include <sys/systm.h>
36#include <sys/sysproto.h>
37#include <sys/sysent.h>
38#include <sys/imgact.h>
39#include <sys/signalvar.h>
40#include <vm/vm.h>
41#include <vm/vm_param.h>
42#include <vm/vm_prot.h>
43#include <vm/lock.h>
44#include <vm/vm_kern.h>
45#include <vm/vm_object.h>
46#include <vm/vm_page.h>
47#include <vm/vm_map.h>
48#include <vm/vm_pager.h>
49#include <vm/vm_extern.h>
50#include <sys/user.h>
51#include <sys/exec.h>
52#include <machine/cpu.h>
53#include <machine/frame.h>
54#include <machine/reg.h>
55#include <machine/specialreg.h>
56#include <machine/psl.h>
57#include <machine/sysarch.h>
58#include <machine/md_var.h>
59
60#include <i386/linux/linux.h>
61#include <i386/linux/linux_proto.h>
62#include <i386/linux/linux_syscall.h>
63
64/*
65 * Linux syscalls return negative errno's, we do positive and map them
66 */
67int bsd_to_linux_errno[ELAST] = {
68      -0,  -1,  -2,  -3,  -4,  -5,  -6,  -7,  -8,  -9,
69     -10, -35, -12, -13, -14, -15, -16, -17, -18, -19,
70     -20, -21, -22, -23, -24, -25, -26, -27, -28, -29,
71     -30, -31, -32, -33, -34, -11,-115,-114, -88, -89,
72     -90, -91, -92, -93, -94, -95, -96, -97, -98, -99,
73    -100,-101,-102,-103,-104,-105,-106,-107,-108,-109,
74    -110,-111, -40, -36,-112,-113, -39, -11, -87,-122,
75    -116, -66,  -6,  -6,  -6,  -6,  -6, -37, -38,  -9,
76      -6,
77};
78
79int bsd_to_linux_signal[NSIG] = {
80    0, LINUX_SIGHUP, LINUX_SIGINT, LINUX_SIGQUIT,
81    LINUX_SIGILL, LINUX_SIGTRAP, LINUX_SIGABRT, 0,
82    LINUX_SIGFPE, LINUX_SIGKILL, LINUX_SIGBUS, LINUX_SIGSEGV,
83    0, LINUX_SIGPIPE, LINUX_SIGALRM, LINUX_SIGTERM,
84    LINUX_SIGURG, LINUX_SIGSTOP, LINUX_SIGTSTP, LINUX_SIGCONT,
85    LINUX_SIGCHLD, LINUX_SIGTTIN, LINUX_SIGTTOU, LINUX_SIGIO,
86    LINUX_SIGXCPU, LINUX_SIGXFSZ, LINUX_SIGVTALRM, LINUX_SIGPROF,
87    LINUX_SIGWINCH, 0, LINUX_SIGUSR1, LINUX_SIGUSR2
88};
89
90int linux_to_bsd_signal[LINUX_NSIG] = {
91    0, SIGHUP, SIGINT, SIGQUIT, SIGILL, SIGTRAP, SIGABRT, SIGEMT,
92    SIGFPE, SIGKILL, SIGUSR1, SIGSEGV, SIGUSR2, SIGPIPE, SIGALRM, SIGTERM,
93    SIGBUS, SIGCHLD, SIGCONT, SIGSTOP, SIGTSTP, SIGTTIN, SIGTTOU, SIGIO,
94    SIGXCPU, SIGXFSZ, SIGVTALRM, SIGPROF, SIGWINCH, SIGURG, SIGURG, 0
95};
96
97int linux_fixup(int **stack_base, struct image_params *imgp)
98{
99    int *argv, *envp;
100
101    argv = *stack_base;
102    envp = *stack_base + (imgp->argc + 1);
103    (*stack_base)--;
104    **stack_base = (int)envp;
105    (*stack_base)--;
106    **stack_base = (int)argv;
107    (*stack_base)--;
108    **stack_base = (int)imgp->argc;
109    return 0;			/* XXX */
110}
111
112extern struct sysent linux_sysent[LINUX_SYS_MAXSYSCALL];
113
114static void linux_sendsig(sig_t action,
115			  int sig,
116			  int returnmask,
117			  unsigned code);
118
119extern int _ucodesel, _udatasel;
120
121/*
122 * Send an interrupt to process.
123 *
124 * Stack is set up to allow sigcode stored
125 * in u. to call routine, followed by kcall
126 * to sigreturn routine below.  After sigreturn
127 * resets the signal mask, the stack, and the
128 * frame pointer, it returns to the user
129 * specified pc, psl.
130 */
131
132static void
133linux_sendsig(catcher, sig, mask, code)
134	sig_t catcher;
135	int sig;
136	int mask;
137	unsigned code;
138{
139	register struct proc *p = curproc;
140	register int *regs;
141	struct linux_sigframe *fp, frame;
142	struct sigacts *psp = p->p_sigacts;
143	int oonstack;
144
145	regs = p->p_md.md_regs;
146	oonstack = psp->ps_sigstk.ss_flags & SS_ONSTACK;
147
148#ifdef DEBUG
149	printf("Linux-emul(%d): linux_sendsig(%8x, %d, %d, %d)\n",
150		p->p_pid, catcher, sig, mask, code);
151#endif
152	/*
153	 * Allocate space for the signal handler context.
154	 */
155	if ((psp->ps_flags & SAS_ALTSTACK) && !oonstack &&
156	    (psp->ps_sigonstack & sigmask(sig))) {
157		fp = (struct linux_sigframe *)(psp->ps_sigstk.ss_sp +
158		    psp->ps_sigstk.ss_size - sizeof(struct linux_sigframe));
159		psp->ps_sigstk.ss_flags |= SS_ONSTACK;
160	} else {
161		fp = (struct linux_sigframe *)regs[tESP] - 1;
162	}
163
164	/*
165	 * grow() will return FALSE if the fp will not fit inside the stack
166	 *	and the stack can not be grown. useracc will return FALSE
167	 *	if access is denied.
168	 */
169	if ((grow(p, (int)fp) == FALSE) ||
170	    (useracc((caddr_t)fp, sizeof (struct linux_sigframe), B_WRITE) == FALSE)) {
171		/*
172		 * Process has trashed its stack; give it an illegal
173		 * instruction to halt it in its tracks.
174		 */
175		SIGACTION(p, SIGILL) = SIG_DFL;
176		sig = sigmask(SIGILL);
177		p->p_sigignore &= ~sig;
178		p->p_sigcatch &= ~sig;
179		p->p_sigmask &= ~sig;
180		psignal(p, SIGILL);
181		return;
182	}
183
184	/*
185	 * Build the argument list for the signal handler.
186	 */
187	if (p->p_sysent->sv_sigtbl) {
188		if (sig < p->p_sysent->sv_sigsize)
189			sig = p->p_sysent->sv_sigtbl[sig];
190		else
191			sig = p->p_sysent->sv_sigsize + 1;
192	}
193
194	frame.sf_handler = catcher;
195	frame.sf_sig = sig;
196
197	/*
198	 * Build the signal context to be used by sigreturn.
199	 */
200	frame.sf_sc.sc_mask   = mask;
201	__asm("movl %%gs,%w0" : "=r" (frame.sf_sc.sc_gs));
202	__asm("movl %%fs,%w0" : "=r" (frame.sf_sc.sc_fs));
203	frame.sf_sc.sc_es     = regs[tES];
204	frame.sf_sc.sc_ds     = regs[tDS];
205	frame.sf_sc.sc_edi    = regs[tEDI];
206	frame.sf_sc.sc_esi    = regs[tESI];
207	frame.sf_sc.sc_ebp    = regs[tEBP];
208	frame.sf_sc.sc_ebx    = regs[tEBX];
209	frame.sf_sc.sc_edx    = regs[tEDX];
210	frame.sf_sc.sc_ecx    = regs[tECX];
211	frame.sf_sc.sc_eax    = regs[tEAX];
212	frame.sf_sc.sc_eip    = regs[tEIP];
213	frame.sf_sc.sc_cs     = regs[tCS];
214	frame.sf_sc.sc_eflags = regs[tEFLAGS];
215	frame.sf_sc.sc_esp_at_signal = regs[tESP];
216	frame.sf_sc.sc_ss     = regs[tSS];
217	frame.sf_sc.sc_err    = regs[tERR];
218	frame.sf_sc.sc_trapno = code;	/* XXX ???? */
219
220	if (copyout(&frame, fp, sizeof(frame)) != 0) {
221		/*
222		 * Process has trashed its stack; give it an illegal
223		 * instruction to halt it in its tracks.
224		 */
225		sigexit(p, SIGILL);
226		/* NOTREACHED */
227	}
228
229	/*
230	 * Build context to run handler in.
231	 */
232	regs[tESP] = (int)fp;
233	regs[tEIP] = (int)(((char *)PS_STRINGS) - *(p->p_sysent->sv_szsigcode));
234	regs[tEFLAGS] &= ~PSL_VM;
235	regs[tCS] = _ucodesel;
236	regs[tDS] = _udatasel;
237	regs[tES] = _udatasel;
238	regs[tSS] = _udatasel;
239}
240
241/*
242 * System call to cleanup state after a signal
243 * has been taken.  Reset signal mask and
244 * stack state from context left by sendsig (above).
245 * Return to previous pc and psl as specified by
246 * context left by sendsig. Check carefully to
247 * make sure that the user has not modified the
248 * psl to gain improper privileges or to cause
249 * a machine fault.
250 */
251int
252linux_sigreturn(p, args, retval)
253	struct proc *p;
254	struct linux_sigreturn_args *args;
255	int *retval;
256{
257	struct linux_sigcontext *scp, context;
258	register int *regs;
259	int eflags;
260
261	regs = p->p_md.md_regs;
262
263#ifdef DEBUG
264	printf("Linux-emul(%d): linux_sigreturn(%8x)\n", p->p_pid, args->scp);
265#endif
266	/*
267	 * The trampoline code hands us the context.
268	 * It is unsafe to keep track of it ourselves, in the event that a
269	 * program jumps out of a signal handler.
270	 */
271	scp = args->scp;
272	if (copyin((caddr_t)scp, &context, sizeof(*scp)) != 0)
273		return (EFAULT);
274
275	/*
276	 * Check for security violations.
277	 */
278#define	EFLAGS_SECURE(ef, oef)	((((ef) ^ (oef)) & ~PSL_USERCHANGE) == 0)
279	eflags = context.sc_eflags;
280	/*
281	 * XXX do allow users to change the privileged flag PSL_RF.  The
282	 * cpu sets PSL_RF in tf_eflags for faults.  Debuggers should
283	 * sometimes set it there too.  tf_eflags is kept in the signal
284	 * context during signal handling and there is no other place
285	 * to remember it, so the PSL_RF bit may be corrupted by the
286	 * signal handler without us knowing.  Corruption of the PSL_RF
287	 * bit at worst causes one more or one less debugger trap, so
288	 * allowing it is fairly harmless.
289	 */
290	if (!EFLAGS_SECURE(eflags & ~PSL_RF, regs[tEFLAGS] & ~PSL_RF)) {
291    		return(EINVAL);
292	}
293
294	/*
295	 * Don't allow users to load a valid privileged %cs.  Let the
296	 * hardware check for invalid selectors, excess privilege in
297	 * other selectors, invalid %eip's and invalid %esp's.
298	 */
299#define	CS_SECURE(cs)	(ISPL(cs) == SEL_UPL)
300	if (!CS_SECURE(context.sc_cs)) {
301		trapsignal(p, SIGBUS, T_PROTFLT);
302		return(EINVAL);
303	}
304
305	p->p_sigacts->ps_sigstk.ss_flags &= ~SS_ONSTACK;
306	p->p_sigmask = context.sc_mask &~
307		(sigmask(SIGKILL)|sigmask(SIGCONT)|sigmask(SIGSTOP));
308	/*
309	 * Restore signal context.
310	 */
311	/* %fs and %gs were restored by the trampoline. */
312	regs[tES]     = context.sc_es;
313	regs[tDS]     = context.sc_ds;
314	regs[tEDI]    = context.sc_edi;
315	regs[tESI]    = context.sc_esi;
316	regs[tEBP]    = context.sc_ebp;
317	regs[tEBX]    = context.sc_ebx;
318	regs[tEDX]    = context.sc_edx;
319	regs[tECX]    = context.sc_ecx;
320	regs[tEAX]    = context.sc_eax;
321	regs[tEIP]    = context.sc_eip;
322	regs[tCS]     = context.sc_cs;
323	regs[tEFLAGS] = eflags;
324	regs[tESP]    = context.sc_esp_at_signal;
325	regs[tSS]     = context.sc_ss;
326
327	return (EJUSTRETURN);
328}
329
330static void
331linux_prepsyscall(struct trapframe *tf, int *args, u_int *code, caddr_t *params)
332{
333	int i;
334	args[0] = tf->tf_ebx;
335	args[1] = tf->tf_ecx;
336	args[2] = tf->tf_edx;
337	args[3] = tf->tf_esi;
338	args[4] = tf->tf_edi;
339	*params = NULL;		/* no copyin */
340}
341
342extern char linux_sigcode[];
343extern int linux_szsigcode;
344
345struct sysentvec linux_sysvec = {
346    sizeof (linux_sysent) / sizeof(linux_sysent[0]),
347    linux_sysent,
348    0xff,
349    NSIG,
350    bsd_to_linux_signal,
351    ELAST,
352    bsd_to_linux_errno,
353    linux_fixup,
354    linux_sendsig,
355    linux_sigcode,
356    &linux_szsigcode,
357    linux_prepsyscall,
358};
359