[dragonfly.git] / sys / platform / pc32 / i386 / sys_machdep.c

/*-
 * Copyright (c) 1990 The Regents of the University of California.
 * All rights reserved.
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions
 * are met:
 * 1. Redistributions of source code must retain the above copyright
 *    notice, this list of conditions and the following disclaimer.
 * 2. Redistributions in binary form must reproduce the above copyright
 *    notice, this list of conditions and the following disclaimer in the
 *    documentation and/or other materials provided with the distribution.
 * 3. All advertising materials mentioning features or use of this software
 *    must display the following acknowledgement:
 *	This product includes software developed by the University of
 *	California, Berkeley and its contributors.
 * 4. Neither the name of the University nor the names of its contributors
 *    may be used to endorse or promote products derived from this software
 *    without specific prior written permission.
 *
 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
 * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
 * SUCH DAMAGE.
 *
 *	from: @(#)sys_machdep.c	5.5 (Berkeley) 1/19/91
 * $FreeBSD: src/sys/i386/i386/sys_machdep.c,v 1.47.2.3 2002/10/07 17:20:00 jhb Exp $
 * $DragonFly: src/sys/platform/pc32/i386/sys_machdep.c,v 1.32 2008/01/06 16:55:53 swildner Exp $
 *
 */

#include <sys/param.h>
#include <sys/systm.h>
#include <sys/sysproto.h>
#include <sys/malloc.h>
#include <sys/thread.h>
#include <sys/proc.h>
#include <sys/priv.h>
#include <sys/thread.h>
#include <sys/memrange.h>

#include <vm/vm.h>
#include <sys/lock.h>
#include <vm/pmap.h>
#include <vm/vm_map.h>
#include <vm/vm_extern.h>

#include <sys/user.h>

#include <machine/cpu.h>
#include <machine/pcb_ext.h>	/* pcb.h included by sys/user.h */
#include <machine/sysarch.h>
#include <machine/smp.h>
#include <machine/globaldata.h>	/* mdcpu */

#include <vm/vm_kern.h>		/* for kernel_map */
#include <sys/thread2.h>

#define MAX_LD 8192
#define LD_PER_PAGE 512
#define NEW_MAX_LD(num)  ((num + LD_PER_PAGE) & ~(LD_PER_PAGE-1))
#define SIZE_FROM_LARGEST_LD(num) (NEW_MAX_LD(num) << 3)


static int ki386_get_ldt(struct lwp *, char *, int *);
static int ki386_set_ldt(struct lwp *, char *, int *);
static int ki386_get_ioperm(struct lwp *, char *);
static int ki386_set_ioperm(struct lwp *, char *);
static int check_descs(union descriptor *, int);
int i386_extend_pcb(struct lwp *);

/*
 * sysarch_args(int op, char *params)
 *
 * MPALMOSTSAFE
 */
int
sys_sysarch(struct sysarch_args *uap)
{
	struct lwp *lp = curthread->td_lwp;
	int error = 0;

	get_mplock();

	switch(uap->op) {
	case I386_GET_LDT:
		error = ki386_get_ldt(lp, uap->parms, &uap->sysmsg_result);
		break;
	case I386_SET_LDT:
		error = ki386_set_ldt(lp, uap->parms, &uap->sysmsg_result);
		break;
	case I386_GET_IOPERM:
		error = ki386_get_ioperm(lp, uap->parms);
		break;
	case I386_SET_IOPERM:
		error = ki386_set_ioperm(lp, uap->parms);
		break;
	case I386_VM86:
		error = vm86_sysarch(lp, uap->parms);
		break;
	default:
		error = EOPNOTSUPP;
		break;
	}
	rel_mplock();
	return (error);
}

int
i386_extend_pcb(struct lwp *lp)
{
	int i, offset;
	u_long *addr;
	struct pcb_ext *ext;
	struct soft_segment_descriptor ssd = {
		0,			/* segment base address (overwritten) */
		ctob(IOPAGES + 1) - 1,	/* length */
		SDT_SYS386TSS,		/* segment type */
		0,			/* priority level */
		1,			/* descriptor present */
		0, 0,
		0,			/* default 32 size */
		0			/* granularity */
	};

	ext = (struct pcb_ext *)kmem_alloc(&kernel_map, ctob(IOPAGES+1));
	if (ext == NULL)
		return (ENOMEM);
	bzero(ext, sizeof(struct pcb_ext)); 
	ext->ext_tss.tss_esp0 = (unsigned)((char *)lp->lwp_thread->td_pcb - 16);
	ext->ext_tss.tss_ss0 = GSEL(GDATA_SEL, SEL_KPL);
	/*
	 * The last byte of the i/o map must be followed by an 0xff byte.
	 * We arbitrarily allocate 16 bytes here, to keep the starting
	 * address on a doubleword boundary.
	 */
	offset = PAGE_SIZE - 16;
	ext->ext_tss.tss_ioopt = 
	    (offset - ((unsigned)&ext->ext_tss - (unsigned)ext)) << 16;
	ext->ext_iomap = (caddr_t)ext + offset;
	ext->ext_vm86.vm86_intmap = (caddr_t)ext + offset - 32;

	addr = (u_long *)ext->ext_vm86.vm86_intmap;
	for (i = 0; i < (ctob(IOPAGES) + 32 + 16) / sizeof(u_long); i++)
		*addr++ = ~0;

	ssd.ssd_base = (unsigned)&ext->ext_tss;
	ssd.ssd_limit -= ((unsigned)&ext->ext_tss - (unsigned)ext);
	ssdtosd(&ssd, &ext->ext_tssd);

	/* 
	 * Put the new TSS where the switch code can find it.  Do
	 * a forced switch to ourself to activate it.
	 */
	crit_enter();
	lp->lwp_thread->td_pcb->pcb_ext = ext;
	lp->lwp_thread->td_switch(lp->lwp_thread);
	crit_exit();
	
	return 0;
}

static int
ki386_set_ioperm(struct lwp *lp, char *args)
{
	int i, error;
	struct i386_ioperm_args ua;
	char *iomap;

	if ((error = copyin(args, &ua, sizeof(struct i386_ioperm_args))) != 0)
		return (error);

	if ((error = priv_check_cred(lp->lwp_thread->td_ucred, PRIV_ROOT, 0)) != 0)
		return (error);
	if (securelevel > 0)
		return (EPERM);
	/*
	 * XXX 
	 * While this is restricted to root, we should probably figure out
	 * whether any other driver is using this i/o address, as so not to
	 * cause confusion.  This probably requires a global 'usage registry'.
	 */

	if (lp->lwp_thread->td_pcb->pcb_ext == 0)
		if ((error = i386_extend_pcb(lp)) != 0)
			return (error);
	iomap = (char *)lp->lwp_thread->td_pcb->pcb_ext->ext_iomap;

	if (ua.start + ua.length > IOPAGES * PAGE_SIZE * NBBY)
		return (EINVAL);

	for (i = ua.start; i < ua.start + ua.length; i++) {
		if (ua.enable) 
			iomap[i >> 3] &= ~(1 << (i & 7));
		else
			iomap[i >> 3] |= (1 << (i & 7));
	}
	return (error);
}

static int
ki386_get_ioperm(struct lwp *lp, char *args)
{
	int i, state, error;
	struct i386_ioperm_args ua;
	char *iomap;

	if ((error = copyin(args, &ua, sizeof(struct i386_ioperm_args))) != 0)
		return (error);
	if (ua.start >= IOPAGES * PAGE_SIZE * NBBY)
		return (EINVAL);

	if (lp->lwp_thread->td_pcb->pcb_ext == 0) {
		ua.length = 0;
		goto done;
	}

	iomap = (char *)lp->lwp_thread->td_pcb->pcb_ext->ext_iomap;

	i = ua.start;
	state = (iomap[i >> 3] >> (i & 7)) & 1;
	ua.enable = !state;
	ua.length = 1;

	for (i = ua.start + 1; i < IOPAGES * PAGE_SIZE * NBBY; i++) {
		if (state != ((iomap[i >> 3] >> (i & 7)) & 1))
			break;
		ua.length++;
	}
			
done:
	error = copyout(&ua, args, sizeof(struct i386_ioperm_args));
	return (error);
}

/*
 * Update the TLS entries for the process.  Used by assembly, do not staticize.
 *
 * Must be called from a critical section (else an interrupt thread preemption
 * may cause %gs to fault).  Normally called from the low level swtch.s code.
 *
 * MPSAFE
 */
void
set_user_TLS(void)
{
	struct thread *td = curthread;
	int i;
#ifdef SMP
	int off = GTLS_START + mycpu->gd_cpuid * NGDT;
#else
	const int off = GTLS_START;
#endif
	for (i = 0; i < NGTLS; ++i)
		gdt[off + i].sd = td->td_tls.tls[i];
}

#ifdef SMP
static
void
set_user_ldt_cpusync(struct lwkt_cpusync *cmd)
{
	set_user_ldt(cmd->cs_data);
}
#endif

/*
 * Update the GDT entry pointing to the LDT to point to the LDT of the
 * current process.  Used by assembly, do not staticize.
 *
 * Must be called from a critical section (else an interrupt thread preemption
 * may cause %gs to fault).  Normally called from the low level swtch.s code.
 */   
void
set_user_ldt(struct pcb *pcb)
{
	struct pcb_ldt *pcb_ldt;

	if (pcb != curthread->td_pcb)
		return;

	pcb_ldt = pcb->pcb_ldt;
#ifdef SMP
	gdt[mycpu->gd_cpuid * NGDT + GUSERLDT_SEL].sd = pcb_ldt->ldt_sd;
#else
	gdt[GUSERLDT_SEL].sd = pcb_ldt->ldt_sd;
#endif
	lldt(GSEL(GUSERLDT_SEL, SEL_KPL));
	mdcpu->gd_currentldt = GSEL(GUSERLDT_SEL, SEL_KPL);
}

struct pcb_ldt *
user_ldt_alloc(struct pcb *pcb, int len)
{
	struct pcb_ldt *pcb_ldt, *new_ldt;

	MALLOC(new_ldt, struct pcb_ldt *, sizeof(struct pcb_ldt),
		M_SUBPROC, M_WAITOK);

	new_ldt->ldt_len = len = NEW_MAX_LD(len);
	new_ldt->ldt_base = (caddr_t)kmem_alloc(&kernel_map,
					        len * sizeof(union descriptor));
	if (new_ldt->ldt_base == NULL) {
		FREE(new_ldt, M_SUBPROC);
		return NULL;
	}
	new_ldt->ldt_refcnt = 1;
	new_ldt->ldt_active = 0;

	gdt_segs[GUSERLDT_SEL].ssd_base = (unsigned)new_ldt->ldt_base;
	gdt_segs[GUSERLDT_SEL].ssd_limit = len * sizeof(union descriptor) - 1;
	ssdtosd(&gdt_segs[GUSERLDT_SEL], &new_ldt->ldt_sd);

	if ((pcb_ldt = pcb->pcb_ldt)) {
		if (len > pcb_ldt->ldt_len)
			len = pcb_ldt->ldt_len;
		bcopy(pcb_ldt->ldt_base, new_ldt->ldt_base,
			len * sizeof(union descriptor));
	} else {
		bcopy(ldt, new_ldt->ldt_base, sizeof(ldt));
	}
	return new_ldt;
}

void
user_ldt_free(struct pcb *pcb)
{
	struct pcb_ldt *pcb_ldt = pcb->pcb_ldt;

	if (pcb_ldt == NULL)
		return;

	crit_enter();
	if (pcb == curthread->td_pcb) {
		lldt(_default_ldt);
		mdcpu->gd_currentldt = _default_ldt;
	}
	pcb->pcb_ldt = NULL;
	crit_exit();

	if (--pcb_ldt->ldt_refcnt == 0) {
		kmem_free(&kernel_map, (vm_offset_t)pcb_ldt->ldt_base,
			  pcb_ldt->ldt_len * sizeof(union descriptor));
		FREE(pcb_ldt, M_SUBPROC);
	}
}

static int
ki386_get_ldt(struct lwp *lwp, char *args, int *res)
{
	int error = 0;
	struct pcb *pcb = lwp->lwp_thread->td_pcb;
	struct pcb_ldt *pcb_ldt = pcb->pcb_ldt;
	unsigned int nldt, num;
	union descriptor *lp;
	struct i386_ldt_args ua, *uap = &ua;

	if ((error = copyin(args, uap, sizeof(struct i386_ldt_args))) < 0)
		return(error);

#ifdef	DEBUG
	kprintf("ki386_get_ldt: start=%d num=%d descs=%p\n",
	    uap->start, uap->num, (void *)uap->descs);
#endif

	crit_enter();

	if (pcb_ldt) {
		nldt = (unsigned int)pcb_ldt->ldt_len;
		num = min(uap->num, nldt);
		lp = &((union descriptor *)(pcb_ldt->ldt_base))[uap->start];
	} else {
		nldt = (unsigned int)(sizeof(ldt) / sizeof(ldt[0]));
		num = min(uap->num, nldt);
		lp = &ldt[uap->start];
	}

	/*
	 * note: uap->(args), num, and nldt are unsigned.  nldt and num
	 * are limited in scope, but uap->start can be anything.
	 */
	if (uap->start > nldt || uap->start + num > nldt) {
		crit_exit();
		return(EINVAL);
	}

	error = copyout(lp, uap->descs, num * sizeof(union descriptor));
	if (!error)
		*res = num;
	crit_exit();
	return(error);
}

static int
ki386_set_ldt(struct lwp *lp, char *args, int *res)
{
	int error = 0;
	int largest_ld;
	struct pcb *pcb = lp->lwp_thread->td_pcb;
	struct pcb_ldt *pcb_ldt = pcb->pcb_ldt;
	union descriptor *descs;
	int descs_size;
	struct i386_ldt_args ua, *uap = &ua;

	if ((error = copyin(args, uap, sizeof(struct i386_ldt_args))) < 0)
		return(error);

#ifdef	DEBUG
	kprintf("ki386_set_ldt: start=%d num=%d descs=%p\n",
	    uap->start, uap->num, (void *)uap->descs);
#endif

	/* verify range of descriptors to modify */
	if ((uap->start < 0) || (uap->start >= MAX_LD) || (uap->num < 0) ||
		(uap->num > MAX_LD))
	{
		return(EINVAL);
	}
	largest_ld = uap->start + uap->num - 1;
	if (largest_ld >= MAX_LD)
		return(EINVAL);

	/* allocate user ldt */
	if (!pcb_ldt || largest_ld >= pcb_ldt->ldt_len) {
		struct pcb_ldt *new_ldt = user_ldt_alloc(pcb, largest_ld);
		if (new_ldt == NULL)
			return ENOMEM;
		if (pcb_ldt) {
			pcb_ldt->ldt_sd = new_ldt->ldt_sd;
			kmem_free(&kernel_map, (vm_offset_t)pcb_ldt->ldt_base,
				  pcb_ldt->ldt_len * sizeof(union descriptor));
			pcb_ldt->ldt_base = new_ldt->ldt_base;
			pcb_ldt->ldt_len = new_ldt->ldt_len;
			FREE(new_ldt, M_SUBPROC);
		} else {
			pcb->pcb_ldt = pcb_ldt = new_ldt;
		}
		/*
		 * Since the LDT may be shared, we must signal other cpus to
		 * reload it.  XXX we need to track which cpus might be
		 * using the shared ldt and only signal those.
		 */
#ifdef SMP
		lwkt_cpusync_simple(-1, set_user_ldt_cpusync, pcb);
#else
		set_user_ldt(pcb);
#endif
	}

	descs_size = uap->num * sizeof(union descriptor);
	descs = (union descriptor *)kmem_alloc(&kernel_map, descs_size);
	if (descs == NULL)
		return (ENOMEM);
	error = copyin(&uap->descs[0], descs, descs_size);
	if (error) {
		kmem_free(&kernel_map, (vm_offset_t)descs, descs_size);
		return (error);
	}
	/* Check descriptors for access violations */
	error = check_descs(descs, uap->num);
	if (error) {
		kmem_free(&kernel_map, (vm_offset_t)descs, descs_size);
		return (error);
	}

	/*
	 * Fill in the actual ldt entries.  Since %fs or %gs might point to
	 * one of these entries a critical section is required to prevent an
	 * interrupt thread from preempting us, switch back, and faulting
	 * on the load of %fs due to a half-formed descriptor.
	 */
	crit_enter();
	bcopy(descs, 
		 &((union descriptor *)(pcb_ldt->ldt_base))[uap->start],
		uap->num * sizeof(union descriptor));
	*res = uap->start;

	crit_exit();
	kmem_free(&kernel_map, (vm_offset_t)descs, descs_size);
	return (0);
}

static int
check_descs(union descriptor *descs, int num)
{
	int i;

	/* Check descriptors for access violations */
	for (i = 0; i < num; i++) {
		union descriptor *dp;
		dp = &descs[i];

		switch (dp->sd.sd_type) {
		case SDT_SYSNULL:	/* system null */ 
			dp->sd.sd_p = 0;
			break;
		case SDT_SYS286TSS: /* system 286 TSS available */
		case SDT_SYSLDT:    /* system local descriptor table */
		case SDT_SYS286BSY: /* system 286 TSS busy */
		case SDT_SYSTASKGT: /* system task gate */
		case SDT_SYS286IGT: /* system 286 interrupt gate */
		case SDT_SYS286TGT: /* system 286 trap gate */
		case SDT_SYSNULL2:  /* undefined by Intel */ 
		case SDT_SYS386TSS: /* system 386 TSS available */
		case SDT_SYSNULL3:  /* undefined by Intel */
		case SDT_SYS386BSY: /* system 386 TSS busy */
		case SDT_SYSNULL4:  /* undefined by Intel */ 
		case SDT_SYS386IGT: /* system 386 interrupt gate */
		case SDT_SYS386TGT: /* system 386 trap gate */
		case SDT_SYS286CGT: /* system 286 call gate */ 
		case SDT_SYS386CGT: /* system 386 call gate */
			/* I can't think of any reason to allow a user proc
			 * to create a segment of these types.  They are
			 * for OS use only.
			 */
			return EACCES;

		/* memory segment types */
		case SDT_MEMEC:   /* memory execute only conforming */
		case SDT_MEMEAC:  /* memory execute only accessed conforming */
		case SDT_MEMERC:  /* memory execute read conforming */
		case SDT_MEMERAC: /* memory execute read accessed conforming */
			/* Must be "present" if executable and conforming. */
			if (dp->sd.sd_p == 0)
				return (EACCES);
			break;
		case SDT_MEMRO:   /* memory read only */
		case SDT_MEMROA:  /* memory read only accessed */
		case SDT_MEMRW:   /* memory read write */
		case SDT_MEMRWA:  /* memory read write accessed */
		case SDT_MEMROD:  /* memory read only expand dwn limit */
		case SDT_MEMRODA: /* memory read only expand dwn lim accessed */
		case SDT_MEMRWD:  /* memory read write expand dwn limit */  
		case SDT_MEMRWDA: /* memory read write expand dwn lim acessed */
		case SDT_MEME:    /* memory execute only */ 
		case SDT_MEMEA:   /* memory execute only accessed */
		case SDT_MEMER:   /* memory execute read */
		case SDT_MEMERA:  /* memory execute read accessed */
			break;
		default:
			return(EINVAL);
			/*NOTREACHED*/
		}

		/* Only user (ring-3) descriptors may be present. */
		if ((dp->sd.sd_p != 0) && (dp->sd.sd_dpl != SEL_UPL))
			return (EACCES);
	}
	return (0);
}

/*
 * Called when /dev/io is opened
 */
int
cpu_set_iopl(void)
{
	curthread->td_lwp->lwp_md.md_regs->tf_eflags |= PSL_IOPL;
	return(0);
}

/*
 * Called when /dev/io is closed
 */
int
cpu_clr_iopl(void)
{
	curthread->td_lwp->lwp_md.md_regs->tf_eflags &= ~PSL_IOPL;
	return(0);
}
Commit	Line	Data
984263bc MD	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	* $FreeBSD: src/sys/i386/i386/sys_machdep.c,v 1.47.2.3 2002/10/07 17:20:00 jhb Exp $
978400d3	35	* $DragonFly: src/sys/platform/pc32/i386/sys_machdep.c,v 1.32 2008/01/06 16:55:53 swildner Exp $
984263bc MD	36	*
	37	*/
	38
984263bc MD	39	#include <sys/param.h>
	40	#include <sys/systm.h>
	41	#include <sys/sysproto.h>
	42	#include <sys/malloc.h>
85100692	43	#include <sys/thread.h>
984263bc	44	#include <sys/proc.h>
895c1f85	45	#include <sys/priv.h>
cf4e130b	46	#include <sys/thread.h>
fc26c9f1	47	#include <sys/memrange.h>
984263bc MD	48
	49	#include <vm/vm.h>
	50	#include <sys/lock.h>
	51	#include <vm/pmap.h>
	52	#include <vm/vm_map.h>
	53	#include <vm/vm_extern.h>
	54
	55	#include <sys/user.h>
	56
	57	#include <machine/cpu.h>
984263bc MD	58	#include <machine/pcb_ext.h> /* pcb.h included by sys/user.h */
984263bc MD	59	#include <machine/sysarch.h>
984263bc	60	#include <machine/smp.h>
85100692	61	#include <machine/globaldata.h> /* mdcpu */
984263bc MD	62
984263bc MD	63	#include <vm/vm_kern.h> /* for kernel_map */
cf4e130b	64	#include <sys/thread2.h>
984263bc MD	65
	66	#define MAX_LD 8192
	67	#define LD_PER_PAGE 512
	68	#define NEW_MAX_LD(num) ((num + LD_PER_PAGE) & ~(LD_PER_PAGE-1))
	69	#define SIZE_FROM_LARGEST_LD(num) (NEW_MAX_LD(num) << 3)
	70
	71
	72
73f913ca MD	73	static int ki386_get_ldt(struct lwp , char , int *);
	74	static int ki386_set_ldt(struct lwp , char , int *);
	75	static int ki386_get_ioperm(struct lwp , char );
	76	static int ki386_set_ioperm(struct lwp , char );
806bf111	77	static int check_descs(union descriptor *, int);
065b709a	78	int i386_extend_pcb(struct lwp *);
984263bc	79
41c20dac MD	80	/*
41c20dac MD	81	* sysarch_args(int op, char *params)
3919ced0 MD	82	*
3919ced0 MD	83	* MPALMOSTSAFE
41c20dac	84	*/
984263bc	85	int
753fd850	86	sys_sysarch(struct sysarch_args *uap)
984263bc	87	{
065b709a	88	struct lwp *lp = curthread->td_lwp;
984263bc MD	89	int error = 0;
984263bc MD	90
3919ced0 MD	91	get_mplock();
3919ced0 MD	92
984263bc	93	switch(uap->op) {
984263bc	94	case I386_GET_LDT:
73f913ca	95	error = ki386_get_ldt(lp, uap->parms, &uap->sysmsg_result);
984263bc	96	break;
984263bc	97	case I386_SET_LDT:
73f913ca	98	error = ki386_set_ldt(lp, uap->parms, &uap->sysmsg_result);
984263bc	99	break;
984263bc	100	case I386_GET_IOPERM:
73f913ca	101	error = ki386_get_ioperm(lp, uap->parms);
984263bc MD	102	break;
984263bc MD	103	case I386_SET_IOPERM:
73f913ca	104	error = ki386_set_ioperm(lp, uap->parms);
984263bc MD	105	break;
984263bc MD	106	case I386_VM86:
065b709a	107	error = vm86_sysarch(lp, uap->parms);
984263bc MD	108	break;
	109	default:
	110	error = EOPNOTSUPP;
	111	break;
	112	}
3919ced0	113	rel_mplock();
984263bc MD	114	return (error);
	115	}
	116
	117	int
065b709a	118	i386_extend_pcb(struct lwp *lp)
984263bc MD	119	{
	120	int i, offset;
	121	u_long *addr;
	122	struct pcb_ext *ext;
	123	struct soft_segment_descriptor ssd = {
	124	0, /* segment base address (overwritten) */
	125	ctob(IOPAGES + 1) - 1, /* length */
	126	SDT_SYS386TSS, /* segment type */
	127	0, /* priority level */
	128	1, /* descriptor present */
	129	0, 0,
	130	0, /* default 32 size */
	131	0 /* granularity */
	132	};
	133
e4846942 MD	134	ext = (struct pcb_ext *)kmem_alloc(&kernel_map, ctob(IOPAGES+1));
e4846942 MD	135	if (ext == NULL)
984263bc	136	return (ENOMEM);
984263bc	137	bzero(ext, sizeof(struct pcb_ext));
065b709a	138	ext->ext_tss.tss_esp0 = (unsigned)((char *)lp->lwp_thread->td_pcb - 16);
984263bc MD	139	ext->ext_tss.tss_ss0 = GSEL(GDATA_SEL, SEL_KPL);
	140	/*
	141	* The last byte of the i/o map must be followed by an 0xff byte.
	142	* We arbitrarily allocate 16 bytes here, to keep the starting
	143	* address on a doubleword boundary.
	144	*/
	145	offset = PAGE_SIZE - 16;
	146	ext->ext_tss.tss_ioopt =
	147	(offset - ((unsigned)&ext->ext_tss - (unsigned)ext)) << 16;
	148	ext->ext_iomap = (caddr_t)ext + offset;
	149	ext->ext_vm86.vm86_intmap = (caddr_t)ext + offset - 32;
	150
	151	addr = (u_long *)ext->ext_vm86.vm86_intmap;
	152	for (i = 0; i < (ctob(IOPAGES) + 32 + 16) / sizeof(u_long); i++)
	153	*addr++ = ~0;
	154
	155	ssd.ssd_base = (unsigned)&ext->ext_tss;
	156	ssd.ssd_limit -= ((unsigned)&ext->ext_tss - (unsigned)ext);
	157	ssdtosd(&ssd, &ext->ext_tssd);
984263bc	158
df011816 MD	159	/*
	160	* Put the new TSS where the switch code can find it. Do
	161	* a forced switch to ourself to activate it.
	162	*/
	163	crit_enter();
	164	lp->lwp_thread->td_pcb->pcb_ext = ext;
	165	lp->lwp_thread->td_switch(lp->lwp_thread);
	166	crit_exit();
	167
984263bc MD	168	return 0;
	169	}
	170
	171	static int
73f913ca	172	ki386_set_ioperm(struct lwp lp, char args)
984263bc MD	173	{
	174	int i, error;
	175	struct i386_ioperm_args ua;
	176	char *iomap;
	177
	178	if ((error = copyin(args, &ua, sizeof(struct i386_ioperm_args))) != 0)
	179	return (error);
	180
9910d07b	181	if ((error = priv_check_cred(lp->lwp_thread->td_ucred, PRIV_ROOT, 0)) != 0)
984263bc MD	182	return (error);
	183	if (securelevel > 0)
	184	return (EPERM);
	185	/*
	186	* XXX
	187	* While this is restricted to root, we should probably figure out
	188	* whether any other driver is using this i/o address, as so not to
	189	* cause confusion. This probably requires a global 'usage registry'.
	190	*/
	191
065b709a SS	192	if (lp->lwp_thread->td_pcb->pcb_ext == 0)
065b709a SS	193	if ((error = i386_extend_pcb(lp)) != 0)
984263bc	194	return (error);
065b709a	195	iomap = (char *)lp->lwp_thread->td_pcb->pcb_ext->ext_iomap;
984263bc MD	196
	197	if (ua.start + ua.length > IOPAGES * PAGE_SIZE * NBBY)
	198	return (EINVAL);
	199
	200	for (i = ua.start; i < ua.start + ua.length; i++) {
	201	if (ua.enable)
	202	iomap[i >> 3] &= ~(1 << (i & 7));
	203	else
	204	iomap[i >> 3] \|= (1 << (i & 7));
	205	}
	206	return (error);
	207	}
	208
	209	static int
73f913ca	210	ki386_get_ioperm(struct lwp lp, char args)
984263bc MD	211	{
	212	int i, state, error;
	213	struct i386_ioperm_args ua;
	214	char *iomap;
	215
	216	if ((error = copyin(args, &ua, sizeof(struct i386_ioperm_args))) != 0)
	217	return (error);
	218	if (ua.start >= IOPAGES * PAGE_SIZE * NBBY)
	219	return (EINVAL);
	220
065b709a	221	if (lp->lwp_thread->td_pcb->pcb_ext == 0) {
984263bc MD	222	ua.length = 0;
	223	goto done;
	224	}
	225
065b709a	226	iomap = (char *)lp->lwp_thread->td_pcb->pcb_ext->ext_iomap;
984263bc MD	227
	228	i = ua.start;
	229	state = (iomap[i >> 3] >> (i & 7)) & 1;
	230	ua.enable = !state;
	231	ua.length = 1;
	232
	233	for (i = ua.start + 1; i < IOPAGES * PAGE_SIZE * NBBY; i++) {
	234	if (state != ((iomap[i >> 3] >> (i & 7)) & 1))
	235	break;
	236	ua.length++;
	237	}
	238
	239	done:
	240	error = copyout(&ua, args, sizeof(struct i386_ioperm_args));
	241	return (error);
	242	}
	243
984263bc	244	/*
806bf111 MD	245	* Update the TLS entries for the process. Used by assembly, do not staticize.
	246	*
	247	* Must be called from a critical section (else an interrupt thread preemption
	248	* may cause %gs to fault). Normally called from the low level swtch.s code.
923e4699 MD	249	*
923e4699 MD	250	* MPSAFE
806bf111 MD	251	*/
	252	void
	253	set_user_TLS(void)
	254	{
	255	struct thread *td = curthread;
	256	int i;
	257	#ifdef SMP
	258	int off = GTLS_START + mycpu->gd_cpuid * NGDT;
	259	#else
	260	const int off = GTLS_START;
	261	#endif
	262	for (i = 0; i < NGTLS; ++i)
4e7c41c5	263	gdt[off + i].sd = td->td_tls.tls[i];
806bf111 MD	264	}
806bf111 MD	265
4e2eba5b	266	#ifdef SMP
6819df07 MD	267	static
	268	void
	269	set_user_ldt_cpusync(struct lwkt_cpusync *cmd)
	270	{
	271	set_user_ldt(cmd->cs_data);
	272	}
4e2eba5b	273	#endif
6819df07	274
806bf111	275	/*
984263bc	276	* Update the GDT entry pointing to the LDT to point to the LDT of the
806bf111 MD	277	* current process. Used by assembly, do not staticize.
	278	*
	279	* Must be called from a critical section (else an interrupt thread preemption
	280	* may cause %gs to fault). Normally called from the low level swtch.s code.
984263bc MD	281	*/
	282	void
	283	set_user_ldt(struct pcb *pcb)
	284	{
	285	struct pcb_ldt *pcb_ldt;
	286
b7c628e4	287	if (pcb != curthread->td_pcb)
984263bc MD	288	return;
	289
	290	pcb_ldt = pcb->pcb_ldt;
	291	#ifdef SMP
7b95be2a	292	gdt[mycpu->gd_cpuid * NGDT + GUSERLDT_SEL].sd = pcb_ldt->ldt_sd;
984263bc MD	293	#else
	294	gdt[GUSERLDT_SEL].sd = pcb_ldt->ldt_sd;
	295	#endif
	296	lldt(GSEL(GUSERLDT_SEL, SEL_KPL));
85100692	297	mdcpu->gd_currentldt = GSEL(GUSERLDT_SEL, SEL_KPL);
984263bc MD	298	}
	299
	300	struct pcb_ldt *
	301	user_ldt_alloc(struct pcb *pcb, int len)
	302	{
	303	struct pcb_ldt pcb_ldt, new_ldt;
	304
	305	MALLOC(new_ldt, struct pcb_ldt *, sizeof(struct pcb_ldt),
	306	M_SUBPROC, M_WAITOK);
984263bc MD	307
984263bc MD	308	new_ldt->ldt_len = len = NEW_MAX_LD(len);
e4846942 MD	309	new_ldt->ldt_base = (caddr_t)kmem_alloc(&kernel_map,
e4846942 MD	310	len * sizeof(union descriptor));
984263bc MD	311	if (new_ldt->ldt_base == NULL) {
	312	FREE(new_ldt, M_SUBPROC);
	313	return NULL;
	314	}
	315	new_ldt->ldt_refcnt = 1;
	316	new_ldt->ldt_active = 0;
	317
	318	gdt_segs[GUSERLDT_SEL].ssd_base = (unsigned)new_ldt->ldt_base;
	319	gdt_segs[GUSERLDT_SEL].ssd_limit = len * sizeof(union descriptor) - 1;
	320	ssdtosd(&gdt_segs[GUSERLDT_SEL], &new_ldt->ldt_sd);
	321
	322	if ((pcb_ldt = pcb->pcb_ldt)) {
	323	if (len > pcb_ldt->ldt_len)
	324	len = pcb_ldt->ldt_len;
	325	bcopy(pcb_ldt->ldt_base, new_ldt->ldt_base,
	326	len * sizeof(union descriptor));
	327	} else {
	328	bcopy(ldt, new_ldt->ldt_base, sizeof(ldt));
	329	}
	330	return new_ldt;
	331	}
	332
	333	void
	334	user_ldt_free(struct pcb *pcb)
	335	{
	336	struct pcb_ldt *pcb_ldt = pcb->pcb_ldt;
	337
	338	if (pcb_ldt == NULL)
	339	return;
	340
cf4e130b	341	crit_enter();
b7c628e4	342	if (pcb == curthread->td_pcb) {
984263bc	343	lldt(_default_ldt);
85100692	344	mdcpu->gd_currentldt = _default_ldt;
984263bc	345	}
cf4e130b MD	346	pcb->pcb_ldt = NULL;
cf4e130b MD	347	crit_exit();
984263bc MD	348
984263bc MD	349	if (--pcb_ldt->ldt_refcnt == 0) {
e4846942 MD	350	kmem_free(&kernel_map, (vm_offset_t)pcb_ldt->ldt_base,
e4846942 MD	351	pcb_ldt->ldt_len * sizeof(union descriptor));
984263bc MD	352	FREE(pcb_ldt, M_SUBPROC);
984263bc MD	353	}
984263bc MD	354	}
	355
	356	static int
73f913ca	357	ki386_get_ldt(struct lwp lwp, char args, int *res)
984263bc MD	358	{
984263bc MD	359	int error = 0;
065b709a	360	struct pcb *pcb = lwp->lwp_thread->td_pcb;
984263bc	361	struct pcb_ldt *pcb_ldt = pcb->pcb_ldt;
6d2edbe1	362	unsigned int nldt, num;
984263bc	363	union descriptor *lp;
984263bc MD	364	struct i386_ldt_args ua, *uap = &ua;
	365
	366	if ((error = copyin(args, uap, sizeof(struct i386_ldt_args))) < 0)
	367	return(error);
	368
	369	#ifdef DEBUG
26be20a0	370	kprintf("ki386_get_ldt: start=%d num=%d descs=%p\n",
984263bc MD	371	uap->start, uap->num, (void *)uap->descs);
	372	#endif
	373
9acd5bbb	374	crit_enter();
984263bc MD	375
984263bc MD	376	if (pcb_ldt) {
6d2edbe1	377	nldt = (unsigned int)pcb_ldt->ldt_len;
984263bc MD	378	num = min(uap->num, nldt);
	379	lp = &((union descriptor *)(pcb_ldt->ldt_base))[uap->start];
	380	} else {
6d2edbe1	381	nldt = (unsigned int)(sizeof(ldt) / sizeof(ldt[0]));
984263bc MD	382	num = min(uap->num, nldt);
	383	lp = &ldt[uap->start];
	384	}
6d2edbe1 MD	385
	386	/*
	387	* note: uap->(args), num, and nldt are unsigned. nldt and num
	388	* are limited in scope, but uap->start can be anything.
	389	*/
	390	if (uap->start > nldt \|\| uap->start + num > nldt) {
9acd5bbb	391	crit_exit();
984263bc MD	392	return(EINVAL);
	393	}
	394
	395	error = copyout(lp, uap->descs, num * sizeof(union descriptor));
	396	if (!error)
90b9818c	397	*res = num;
9acd5bbb	398	crit_exit();
984263bc MD	399	return(error);
	400	}
	401
	402	static int
73f913ca	403	ki386_set_ldt(struct lwp lp, char args, int *res)
984263bc	404	{
806bf111	405	int error = 0;
984263bc	406	int largest_ld;
065b709a	407	struct pcb *pcb = lp->lwp_thread->td_pcb;
984263bc MD	408	struct pcb_ldt *pcb_ldt = pcb->pcb_ldt;
984263bc MD	409	union descriptor *descs;
806bf111	410	int descs_size;
984263bc MD	411	struct i386_ldt_args ua, *uap = &ua;
	412
	413	if ((error = copyin(args, uap, sizeof(struct i386_ldt_args))) < 0)
	414	return(error);
	415
	416	#ifdef DEBUG
26be20a0	417	kprintf("ki386_set_ldt: start=%d num=%d descs=%p\n",
984263bc MD	418	uap->start, uap->num, (void *)uap->descs);
	419	#endif
	420
	421	/* verify range of descriptors to modify */
	422	if ((uap->start < 0) \|\| (uap->start >= MAX_LD) \|\| (uap->num < 0) \|\|
	423	(uap->num > MAX_LD))
	424	{
	425	return(EINVAL);
	426	}
	427	largest_ld = uap->start + uap->num - 1;
	428	if (largest_ld >= MAX_LD)
	429	return(EINVAL);
	430
	431	/* allocate user ldt */
	432	if (!pcb_ldt \|\| largest_ld >= pcb_ldt->ldt_len) {
	433	struct pcb_ldt *new_ldt = user_ldt_alloc(pcb, largest_ld);
	434	if (new_ldt == NULL)
	435	return ENOMEM;
	436	if (pcb_ldt) {
	437	pcb_ldt->ldt_sd = new_ldt->ldt_sd;
e4846942 MD	438	kmem_free(&kernel_map, (vm_offset_t)pcb_ldt->ldt_base,
e4846942 MD	439	pcb_ldt->ldt_len * sizeof(union descriptor));
984263bc MD	440	pcb_ldt->ldt_base = new_ldt->ldt_base;
	441	pcb_ldt->ldt_len = new_ldt->ldt_len;
	442	FREE(new_ldt, M_SUBPROC);
806bf111	443	} else {
984263bc	444	pcb->pcb_ldt = pcb_ldt = new_ldt;
806bf111 MD	445	}
	446	/*
	447	* Since the LDT may be shared, we must signal other cpus to
	448	* reload it. XXX we need to track which cpus might be
	449	* using the shared ldt and only signal those.
	450	*/
984263bc	451	#ifdef SMP
6819df07	452	lwkt_cpusync_simple(-1, set_user_ldt_cpusync, pcb);
984263bc MD	453	#else
	454	set_user_ldt(pcb);
	455	#endif
	456	}
	457
	458	descs_size = uap->num * sizeof(union descriptor);
e4846942	459	descs = (union descriptor *)kmem_alloc(&kernel_map, descs_size);
984263bc MD	460	if (descs == NULL)
	461	return (ENOMEM);
	462	error = copyin(&uap->descs[0], descs, descs_size);
	463	if (error) {
e4846942	464	kmem_free(&kernel_map, (vm_offset_t)descs, descs_size);
984263bc MD	465	return (error);
	466	}
	467	/* Check descriptors for access violations */
806bf111 MD	468	error = check_descs(descs, uap->num);
806bf111 MD	469	if (error) {
e4846942	470	kmem_free(&kernel_map, (vm_offset_t)descs, descs_size);
806bf111 MD	471	return (error);
	472	}
	473
	474	/*
4e7c41c5 MD	475	* Fill in the actual ldt entries. Since %fs or %gs might point to
4e7c41c5 MD	476	* one of these entries a critical section is required to prevent an
806bf111 MD	477	* interrupt thread from preempting us, switch back, and faulting
	478	* on the load of %fs due to a half-formed descriptor.
	479	*/
	480	crit_enter();
	481	bcopy(descs,
	482	&((union descriptor *)(pcb_ldt->ldt_base))[uap->start],
	483	uap->num * sizeof(union descriptor));
	484	*res = uap->start;
	485
	486	crit_exit();
e4846942	487	kmem_free(&kernel_map, (vm_offset_t)descs, descs_size);
806bf111 MD	488	return (0);
	489	}
	490
	491	static int
	492	check_descs(union descriptor *descs, int num)
	493	{
	494	int i;
	495
	496	/* Check descriptors for access violations */
	497	for (i = 0; i < num; i++) {
984263bc MD	498	union descriptor *dp;
	499	dp = &descs[i];
	500
	501	switch (dp->sd.sd_type) {
	502	case SDT_SYSNULL: /* system null */
	503	dp->sd.sd_p = 0;
	504	break;
	505	case SDT_SYS286TSS: /* system 286 TSS available */
	506	case SDT_SYSLDT: /* system local descriptor table */
	507	case SDT_SYS286BSY: /* system 286 TSS busy */
	508	case SDT_SYSTASKGT: /* system task gate */
	509	case SDT_SYS286IGT: /* system 286 interrupt gate */
	510	case SDT_SYS286TGT: /* system 286 trap gate */
	511	case SDT_SYSNULL2: /* undefined by Intel */
	512	case SDT_SYS386TSS: /* system 386 TSS available */
	513	case SDT_SYSNULL3: /* undefined by Intel */
	514	case SDT_SYS386BSY: /* system 386 TSS busy */
	515	case SDT_SYSNULL4: /* undefined by Intel */
	516	case SDT_SYS386IGT: /* system 386 interrupt gate */
	517	case SDT_SYS386TGT: /* system 386 trap gate */
	518	case SDT_SYS286CGT: /* system 286 call gate */
	519	case SDT_SYS386CGT: /* system 386 call gate */
	520	/* I can't think of any reason to allow a user proc
	521	* to create a segment of these types. They are
	522	* for OS use only.
	523	*/
984263bc MD	524	return EACCES;
	525
	526	/* memory segment types */
	527	case SDT_MEMEC: /* memory execute only conforming */
	528	case SDT_MEMEAC: /* memory execute only accessed conforming */
	529	case SDT_MEMERC: /* memory execute read conforming */
	530	case SDT_MEMERAC: /* memory execute read accessed conforming */
	531	/* Must be "present" if executable and conforming. */
806bf111	532	if (dp->sd.sd_p == 0)
984263bc	533	return (EACCES);
984263bc MD	534	break;
	535	case SDT_MEMRO: /* memory read only */
	536	case SDT_MEMROA: /* memory read only accessed */
	537	case SDT_MEMRW: /* memory read write */
	538	case SDT_MEMRWA: /* memory read write accessed */
	539	case SDT_MEMROD: /* memory read only expand dwn limit */
	540	case SDT_MEMRODA: /* memory read only expand dwn lim accessed */
	541	case SDT_MEMRWD: /* memory read write expand dwn limit */
	542	case SDT_MEMRWDA: /* memory read write expand dwn lim acessed */
	543	case SDT_MEME: /* memory execute only */
	544	case SDT_MEMEA: /* memory execute only accessed */
	545	case SDT_MEMER: /* memory execute read */
	546	case SDT_MEMERA: /* memory execute read accessed */
	547	break;
	548	default:
984263bc MD	549	return(EINVAL);
	550	/NOTREACHED/
	551	}
	552
	553	/* Only user (ring-3) descriptors may be present. */
806bf111	554	if ((dp->sd.sd_p != 0) && (dp->sd.sd_dpl != SEL_UPL))
984263bc	555	return (EACCES);
984263bc	556	}
984263bc MD	557	return (0);
984263bc MD	558	}
fc26c9f1 MD	559
	560	/*
	561	* Called when /dev/io is opened
	562	*/
	563	int
	564	cpu_set_iopl(void)
	565	{
08f2f1bb	566	curthread->td_lwp->lwp_md.md_regs->tf_eflags \|= PSL_IOPL;
fc26c9f1 MD	567	return(0);
	568	}
	569
	570	/*
	571	* Called when /dev/io is closed
	572	*/
	573	int
	574	cpu_clr_iopl(void)
	575	{
08f2f1bb	576	curthread->td_lwp->lwp_md.md_regs->tf_eflags &= ~PSL_IOPL;
fc26c9f1 MD	577	return(0);
	578	}
	579