sys_machdep.c revision 32012
1/*-
2 * Copyright (c) 1990 The Regents of the University of California.
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 * 2. Redistributions in binary form must reproduce the above copyright
11 *    notice, this list of conditions and the following disclaimer in the
12 *    documentation and/or other materials provided with the distribution.
13 * 3. All advertising materials mentioning features or use of this software
14 *    must display the following acknowledgement:
15 *	This product includes software developed by the University of
16 *	California, Berkeley and its contributors.
17 * 4. Neither the name of the University nor the names of its contributors
18 *    may be used to endorse or promote products derived from this software
19 *    without specific prior written permission.
20 *
21 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
22 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
23 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
24 * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
25 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
26 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
27 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
28 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
29 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
30 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
31 * SUCH DAMAGE.
32 *
33 *	from: @(#)sys_machdep.c	5.5 (Berkeley) 1/19/91
34 *	$Id: sys_machdep.c,v 1.30 1997/11/26 22:45:47 joerg Exp $
35 *
36 */
37
38#include "opt_user_ldt.h"
39#include "opt_vm86.h"
40
41#include <sys/param.h>
42#include <sys/systm.h>
43#include <sys/sysproto.h>
44#include <sys/proc.h>
45
46#include <vm/vm.h>
47#include <sys/lock.h>
48#include <vm/pmap.h>
49#include <vm/vm_map.h>
50#include <vm/vm_extern.h>
51
52#include <sys/user.h>
53
54#include <machine/cpu.h>
55#include <machine/pcb_ext.h>	/* pcb.h included by sys/user.h */
56#include <machine/sysarch.h>
57
58#include <vm/vm_kern.h>		/* for kernel_map */
59
60#define MAX_LD 8192
61#define LD_PER_PAGE 512
62#define NEW_MAX_LD(num)  ((num + LD_PER_PAGE) & ~(LD_PER_PAGE-1))
63#define SIZE_FROM_LARGEST_LD(num) (NEW_MAX_LD(num) << 3)
64
65
66
67void set_user_ldt	__P((struct pcb *pcb));
68#ifdef USER_LDT
69static int i386_get_ldt	__P((struct proc *, char *));
70static int i386_set_ldt	__P((struct proc *, char *));
71#endif
72#ifdef VM86
73static int i386_get_ioperm	__P((struct proc *, char *));
74static int i386_set_ioperm	__P((struct proc *, char *));
75int i386_extend_pcb	__P((struct proc *));
76#endif
77
78#ifndef _SYS_SYSPROTO_H_
79struct sysarch_args {
80	int op;
81	char *parms;
82};
83#endif
84
85int
86sysarch(p, uap)
87	struct proc *p;
88	register struct sysarch_args *uap;
89{
90	int error = 0;
91
92	switch(uap->op) {
93#ifdef	USER_LDT
94	case I386_GET_LDT:
95		error = i386_get_ldt(p, uap->parms);
96		break;
97
98	case I386_SET_LDT:
99		error = i386_set_ldt(p, uap->parms);
100		break;
101#endif
102#ifdef VM86
103	case I386_GET_IOPERM:
104		error = i386_get_ioperm(p, uap->parms);
105		break;
106	case I386_SET_IOPERM:
107		error = i386_set_ioperm(p, uap->parms);
108		break;
109	case I386_VM86:
110		error = vm86_sysarch(p, uap->parms);
111		break;
112#endif
113	default:
114		error = EINVAL;
115		break;
116	}
117	return (error);
118}
119
120#ifdef VM86
121int
122i386_extend_pcb(struct proc *p)
123{
124	int i, offset;
125	u_long *addr;
126	struct pcb_ext *ext;
127	struct segment_descriptor sd;
128	struct soft_segment_descriptor ssd = {
129		0,			/* segment base address (overwritten) */
130		ctob(IOPAGES + 1) - 1,	/* length */
131		SDT_SYS386TSS,		/* segment type */
132		0,			/* priority level */
133		1,			/* descriptor present */
134		0, 0,
135		0,			/* default 32 size */
136		0			/* granularity */
137	};
138
139	ext = (struct pcb_ext *)kmem_alloc(kernel_map, ctob(IOPAGES+1));
140	if (ext == 0)
141		return (ENOMEM);
142	p->p_addr->u_pcb.pcb_ext = ext;
143	bzero(&ext->ext_tss, sizeof(struct i386tss));
144	ext->ext_tss.tss_esp0 = (unsigned)p->p_addr + ctob(UPAGES) - 16;
145        ext->ext_tss.tss_ss0 = GSEL(GDATA_SEL, SEL_KPL);
146	/*
147	 * The last byte of the i/o map must be followed by an 0xff byte.
148	 * We arbitrarily allocate 16 bytes here, to keep the starting
149	 * address on a doubleword boundary.
150	 */
151	offset = PAGE_SIZE - 16;
152	ext->ext_tss.tss_ioopt =
153	    (offset - ((unsigned)&ext->ext_tss - (unsigned)ext)) << 16;
154	ext->ext_iomap = (caddr_t)ext + offset;
155	ext->ext_vm86.vm86_intmap = (caddr_t)ext + offset - 32;
156	ext->ext_vm86.vm86_inited = 0;
157
158	addr = (u_long *)ext->ext_vm86.vm86_intmap;
159	for (i = 0; i < (ctob(IOPAGES) + 32 + 16) / sizeof(u_long); i++)
160		*addr++ = ~0;
161
162	ssd.ssd_base = (unsigned)&ext->ext_tss;
163	ssd.ssd_limit -= ((unsigned)&ext->ext_tss - (unsigned)ext);
164	ssdtosd(&ssd, &ext->ext_tssd);
165
166	/* switch to the new TSS after syscall completes */
167	need_resched();
168
169	return 0;
170}
171
172struct i386_ioperm_args {
173	u_short start;
174	u_short length;
175	u_char enable;
176};
177
178static int
179i386_set_ioperm(p, args)
180	struct proc *p;
181	char *args;
182{
183	int i, error = 0;
184	struct i386_ioperm_args ua;
185	char *iomap;
186
187	if (error = copyin(args, &ua, sizeof(struct i386_ioperm_args)))
188		return (error);
189
190        /* Only root can do this */
191        if (error = suser(p->p_ucred, &p->p_acflag))
192                return (error);
193	/*
194	 * XXX
195	 * While this is restricted to root, we should probably figure out
196	 * whether any other driver is using this i/o address, as so not to
197	 * cause confusion.  This probably requires a global 'usage registry'.
198	 */
199
200	if (p->p_addr->u_pcb.pcb_ext == 0)
201		if (error = i386_extend_pcb(p))
202			return (error);
203	iomap = (char *)p->p_addr->u_pcb.pcb_ext->ext_iomap;
204
205	if ((int)(ua.start + ua.length) > 0xffff)
206		return (EINVAL);
207
208	for (i = ua.start; i < (int)(ua.start + ua.length) + 1; i++) {
209		if (ua.enable)
210			iomap[i >> 3] &= ~(1 << (i & 7));
211		else
212			iomap[i >> 3] |= (1 << (i & 7));
213	}
214	return (error);
215}
216
217static int
218i386_get_ioperm(p, args)
219	struct proc *p;
220	char *args;
221{
222	int i, state, error = 0;
223	struct i386_ioperm_args ua;
224	char *iomap;
225
226	if (error = copyin(args, &ua, sizeof(struct i386_ioperm_args)))
227		return (error);
228
229	if (p->p_addr->u_pcb.pcb_ext == 0) {
230		ua.length = 0;
231		goto done;
232	}
233
234	iomap = (char *)p->p_addr->u_pcb.pcb_ext->ext_iomap;
235
236	i = ua.start;
237	state = (iomap[i >> 3] >> (i & 7)) & 1;
238	ua.enable = !state;
239	ua.length = 1;
240
241	for (i = ua.start + 1; i < 0x10000; i++) {
242		if (state != ((iomap[i >> 3] >> (i & 7)) & 1))
243			break;
244		ua.length++;
245	}
246
247done:
248	error = copyout(&ua, args, sizeof(struct i386_ioperm_args));
249	return (error);
250}
251#endif /* VM86 */
252
253#ifdef USER_LDT
254/*
255 * Update the GDT entry pointing to the LDT to point to the LDT of the
256 * current process.
257 */
258void
259set_user_ldt(struct pcb *pcb)
260{
261	gdt_segs[GUSERLDT_SEL].ssd_base = (unsigned)pcb->pcb_ldt;
262	gdt_segs[GUSERLDT_SEL].ssd_limit = (pcb->pcb_ldt_len * sizeof(union descriptor)) - 1;
263	ssdtosd(&gdt_segs[GUSERLDT_SEL], &gdt[GUSERLDT_SEL].sd);
264	lldt(GSEL(GUSERLDT_SEL, SEL_KPL));
265	currentldt = GSEL(GUSERLDT_SEL, SEL_KPL);
266}
267
268struct i386_get_ldt_args {
269	int start;
270	union descriptor *desc;
271	int num;
272};
273
274static int
275i386_get_ldt(p, args)
276	struct proc *p;
277	char *args;
278{
279	int error = 0;
280	struct pcb *pcb = &p->p_addr->u_pcb;
281	int nldt, num;
282	union descriptor *lp;
283	int s;
284	struct i386_get_ldt_args ua;
285	struct i386_get_ldt_args *uap = &ua;
286
287	if ((error = copyin(args, uap, sizeof(struct i386_get_ldt_args))) < 0)
288		return(error);
289
290#ifdef	DEBUG
291	printf("i386_get_ldt: start=%d num=%d descs=%x\n", uap->start,
292		uap->num, uap->desc);
293#endif
294
295	/* verify range of LDTs exist */
296	if ((uap->start < 0) || (uap->num <= 0))
297		return(EINVAL);
298
299	s = splhigh();
300
301	if (pcb->pcb_ldt) {
302		nldt = pcb->pcb_ldt_len;
303		num = min(uap->num, nldt);
304		lp = &((union descriptor *)(pcb->pcb_ldt))[uap->start];
305	} else {
306		nldt = sizeof(ldt)/sizeof(ldt[0]);
307		num = min(uap->num, nldt);
308		lp = &ldt[uap->start];
309	}
310	if (uap->start > nldt) {
311		splx(s);
312		return(EINVAL);
313	}
314
315	error = copyout(lp, uap->desc, num * sizeof(union descriptor));
316	if (!error)
317		p->p_retval[0] = num;
318
319	splx(s);
320	return(error);
321}
322
323struct i386_set_ldt_args {
324	int start;
325	union descriptor *desc;
326	int num;
327};
328
329static int
330i386_set_ldt(p, args)
331	struct proc *p;
332	char *args;
333{
334	int error = 0, i, n;
335 	int largest_ld;
336	struct pcb *pcb = &p->p_addr->u_pcb;
337	int s;
338	struct i386_set_ldt_args ua, *uap;
339
340	if ((error = copyin(args, &ua, sizeof(struct i386_set_ldt_args))) < 0)
341		return(error);
342
343	uap = &ua;
344
345#ifdef	DEBUG
346	printf("i386_set_ldt: start=%d num=%d descs=%x\n", uap->start, uap->num, uap->desc);
347#endif
348
349 	/* verify range of descriptors to modify */
350 	if ((uap->start < 0) || (uap->start >= MAX_LD) || (uap->num < 0) ||
351 		(uap->num > MAX_LD))
352 	{
353 		return(EINVAL);
354 	}
355 	largest_ld = uap->start + uap->num - 1;
356 	if (largest_ld >= MAX_LD)
357  		return(EINVAL);
358
359  	/* allocate user ldt */
360 	if (!pcb->pcb_ldt || (largest_ld >= pcb->pcb_ldt_len)) {
361 		union descriptor *new_ldt = (union descriptor *)kmem_alloc(
362 			kernel_map, SIZE_FROM_LARGEST_LD(largest_ld));
363 		if (new_ldt == NULL) {
364 			return ENOMEM;
365 		}
366 		if (pcb->pcb_ldt) {
367 			bcopy(pcb->pcb_ldt, new_ldt, pcb->pcb_ldt_len
368 				* sizeof(union descriptor));
369 			kmem_free(kernel_map, (vm_offset_t)pcb->pcb_ldt,
370 				pcb->pcb_ldt_len * sizeof(union descriptor));
371 		} else {
372 			bcopy(ldt, new_ldt, sizeof(ldt));
373 		}
374  		pcb->pcb_ldt = (caddr_t)new_ldt;
375 		pcb->pcb_ldt_len = NEW_MAX_LD(largest_ld);
376 		if (pcb == curpcb)
377 		    set_user_ldt(pcb);
378  	}
379
380	/* Check descriptors for access violations */
381	for (i = 0, n = uap->start; i < uap->num; i++, n++) {
382		union descriptor desc, *dp;
383		dp = &uap->desc[i];
384		error = copyin(dp, &desc, sizeof(union descriptor));
385		if (error)
386			return(error);
387
388		switch (desc.sd.sd_type) {
389 		case SDT_SYSNULL:	/* system null */
390 			desc.sd.sd_p = 0;
391  			break;
392 		case SDT_SYS286TSS: /* system 286 TSS available */
393 		case SDT_SYSLDT:    /* system local descriptor table */
394 		case SDT_SYS286BSY: /* system 286 TSS busy */
395 		case SDT_SYSTASKGT: /* system task gate */
396 		case SDT_SYS286IGT: /* system 286 interrupt gate */
397 		case SDT_SYS286TGT: /* system 286 trap gate */
398 		case SDT_SYSNULL2:  /* undefined by Intel */
399 		case SDT_SYS386TSS: /* system 386 TSS available */
400 		case SDT_SYSNULL3:  /* undefined by Intel */
401 		case SDT_SYS386BSY: /* system 386 TSS busy */
402 		case SDT_SYSNULL4:  /* undefined by Intel */
403 		case SDT_SYS386IGT: /* system 386 interrupt gate */
404 		case SDT_SYS386TGT: /* system 386 trap gate */
405 		case SDT_SYS286CGT: /* system 286 call gate */
406 		case SDT_SYS386CGT: /* system 386 call gate */
407 			/* I can't think of any reason to allow a user proc
408 			 * to create a segment of these types.  They are
409 			 * for OS use only.
410 			 */
411     	    	    	return EACCES;
412
413 		/* memory segment types */
414 		case SDT_MEMEC:   /* memory execute only conforming */
415 		case SDT_MEMEAC:  /* memory execute only accessed conforming */
416 		case SDT_MEMERC:  /* memory execute read conforming */
417 		case SDT_MEMERAC: /* memory execute read accessed conforming */
418                         /* Must be "present" if executable and conforming. */
419                         if (desc.sd.sd_p == 0)
420                                 return (EACCES);
421 			break;
422 		case SDT_MEMRO:   /* memory read only */
423 		case SDT_MEMROA:  /* memory read only accessed */
424 		case SDT_MEMRW:   /* memory read write */
425 		case SDT_MEMRWA:  /* memory read write accessed */
426 		case SDT_MEMROD:  /* memory read only expand dwn limit */
427 		case SDT_MEMRODA: /* memory read only expand dwn lim accessed */
428 		case SDT_MEMRWD:  /* memory read write expand dwn limit */
429 		case SDT_MEMRWDA: /* memory read write expand dwn lim acessed */
430 		case SDT_MEME:    /* memory execute only */
431 		case SDT_MEMEA:   /* memory execute only accessed */
432 		case SDT_MEMER:   /* memory execute read */
433 		case SDT_MEMERA:  /* memory execute read accessed */
434			break;
435		default:
436			return(EINVAL);
437			/*NOTREACHED*/
438		}
439
440 		/* Only user (ring-3) descriptors may be present. */
441 		if ((desc.sd.sd_p != 0) && (desc.sd.sd_dpl != SEL_UPL))
442 			return (EACCES);
443	}
444
445	s = splhigh();
446
447	/* Fill in range */
448 	error = copyin(uap->desc,
449 		 &((union descriptor *)(pcb->pcb_ldt))[uap->start],
450 		uap->num * sizeof(union descriptor));
451 	if (!error)
452  		p->p_retval[0] = uap->start;
453
454	splx(s);
455	return(error);
456}
457#endif	/* USER_LDT */
458