NetBSD/sys/arch/pmax/pmax/trap.c.sav2

1847 lines
44 KiB
Plaintext

/*
* Copyright (c) 1988 University of Utah.
* Copyright (c) 1992 The Regents of the University of California.
* All rights reserved.
*
* This code is derived from software contributed to Berkeley by
* the Systems Programming Group of the University of Utah Computer
* Science Department and Ralph Campbell.
*
* 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: Utah Hdr: trap.c 1.32 91/04/06
* from: @(#)trap.c 7.14 (Berkeley) 3/8/93
* $Id: trap.c.sav2,v 1.1.1.1 1993/10/12 03:22:38 deraadt Exp $
*/
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/proc.h>
#include <sys/kernel.h>
#include <sys/signalvar.h>
#include <sys/syscall.h>
#include <sys/user.h>
#include <sys/buf.h>
#ifdef KTRACE
#include <sys/ktrace.h>
#endif
#include <net/netisr.h>
#include <machine/trap.h>
#include <machine/psl.h>
#include <machine/reg.h>
#include <machine/cpu.h>
#include <machine/pte.h>
#include <machine/mips_opcode.h>
#include <vm/vm.h>
#include <vm/vm_kern.h>
#include <vm/vm_page.h>
#include <pmax/pmax/clockreg.h>
#include <pmax/pmax/kn01.h>
#include <pmax/pmax/kn02.h>
#include <pmax/pmax/kmin.h>
#include <pmax/pmax/maxine.h>
#include <pmax/pmax/kn03.h>
#include <pmax/pmax/asic.h>
#include <pmax/pmax/turbochannel.h>
#include <pmax/stand/dec_prom.h>
#include <asc.h>
#include <sii.h>
#include <le.h>
#include <dc.h>
/*
* This is a kludge to allow X windows to work.
*/
#define X_KLUGE
#ifdef X_KLUGE
#define USER_MAP_ADDR 0x4000
#define NPTES 550
static pt_entry_t UserMapPtes[NPTES];
static unsigned nUserMapPtes;
static pid_t UserMapPid;
#endif
struct proc *machFPCurProcPtr; /* pointer to last proc to use FP */
extern void MachKernGenException();
extern void MachUserGenException();
extern void MachKernIntr();
extern void MachUserIntr();
extern void MachTLBModException();
extern void MachTLBMissException();
extern unsigned MachEmulateBranch();
void (*machExceptionTable[])() = {
/*
* The kernel exception handlers.
*/
MachKernIntr, /* external interrupt */
MachKernGenException, /* TLB modification */
MachTLBMissException, /* TLB miss (load or instr. fetch) */
MachTLBMissException, /* TLB miss (store) */
MachKernGenException, /* address error (load or I-fetch) */
MachKernGenException, /* address error (store) */
MachKernGenException, /* bus error (I-fetch) */
MachKernGenException, /* bus error (load or store) */
MachKernGenException, /* system call */
MachKernGenException, /* breakpoint */
MachKernGenException, /* reserved instruction */
MachKernGenException, /* coprocessor unusable */
MachKernGenException, /* arithmetic overflow */
MachKernGenException, /* reserved */
MachKernGenException, /* reserved */
MachKernGenException, /* reserved */
/*
* The user exception handlers.
*/
MachUserIntr,
MachUserGenException,
MachUserGenException,
MachUserGenException,
MachUserGenException,
MachUserGenException,
MachUserGenException,
MachUserGenException,
MachUserGenException,
MachUserGenException,
MachUserGenException,
MachUserGenException,
MachUserGenException,
MachUserGenException,
MachUserGenException,
MachUserGenException,
};
char *trap_type[] = {
"external interrupt",
"TLB modification",
"TLB miss (load or instr. fetch)",
"TLB miss (store)",
"address error (load or I-fetch)",
"address error (store)",
"bus error (I-fetch)",
"bus error (load or store)",
"system call",
"breakpoint",
"reserved instruction",
"coprocessor unusable",
"arithmetic overflow",
"reserved 13",
"reserved 14",
"reserved 15",
};
#ifdef DEBUG
#define TRAPSIZE 10
struct trapdebug { /* trap history buffer for debugging */
u_int status;
u_int cause;
u_int vadr;
u_int pc;
u_int ra;
u_int code;
} trapdebug[TRAPSIZE], *trp = trapdebug;
#endif
static void pmax_errintr();
static void kn02_errintr(), kn02ba_errintr();
#ifdef DS5000_240
static void kn03_errintr();
#endif
static unsigned kn02ba_recover_erradr();
extern tc_option_t tc_slot_info[TC_MAX_LOGICAL_SLOTS];
extern u_long kmin_tc3_imask, xine_tc3_imask;
extern const struct callback *callv;
#ifdef DS5000_240
extern u_long kn03_tc3_imask;
#endif
int (*pmax_hardware_intr)() = (int (*)())0;
extern volatile struct chiptime *Mach_clock_addr;
/*
* Handle an exception.
* Called from MachKernGenException() or MachUserGenException()
* when a processor trap occurs.
* In the case of a kernel trap, we return the pc where to resume if
* ((struct pcb *)UADDR)->pcb_onfault is set, otherwise, return old pc.
*/
unsigned
trap(statusReg, causeReg, vadr, pc, args)
unsigned statusReg; /* status register at time of the exception */
unsigned causeReg; /* cause register at time of exception */
unsigned vadr; /* address (if any) the fault occured on */
unsigned pc; /* program counter where to continue */
{
register int type, i;
unsigned ucode = 0;
register struct proc *p = curproc;
u_quad_t sticks;
vm_prot_t ftype;
extern unsigned onfault_table[];
#ifdef DEBUG
trp->status = statusReg;
trp->cause = causeReg;
trp->vadr = vadr;
trp->pc = pc;
trp->ra = !USERMODE(statusReg) ? ((int *)&args)[19] :
p->p_md.md_regs[RA];
trp->code = 0;
if (++trp == &trapdebug[TRAPSIZE])
trp = trapdebug;
#endif
cnt.v_trap++;
type = (causeReg & MACH_CR_EXC_CODE) >> MACH_CR_EXC_CODE_SHIFT;
if (USERMODE(statusReg)) {
type |= T_USER;
sticks = p->p_sticks;
}
/*
* Enable hardware interrupts if they were on before.
* We only respond to software interrupts when returning to user mode.
*/
if (statusReg & MACH_SR_INT_ENA_PREV)
splx((statusReg & MACH_HARD_INT_MASK) | MACH_SR_INT_ENA_CUR);
switch (type) {
case T_TLB_MOD:
/* check for kernel address */
if ((int)vadr < 0) {
register pt_entry_t *pte;
register unsigned entry;
#ifndef ATTR
register vm_offset_t pa;
#endif
pte = kvtopte(vadr);
entry = pte->pt_entry;
if (entry & PG_RO) {
/* write to read only page in the kernel */
ftype = VM_PROT_WRITE;
goto kernel_fault;
}
entry |= PG_M;
pte->pt_entry = entry;
vadr &= PG_FRAME;
printf("trap: TLBupdate hi %x lo %x i %x\n", vadr,
entry, MachTLBUpdate(vadr, entry)); /* XXX */
#ifdef ATTR
pmap_attributes[atop(entry - KERNBASE)] |= PMAP_ATTR_MOD;
#else
pa = entry & PG_FRAME;
if (!IS_VM_PHYSADDR(pa))
panic("trap: kmod");
PHYS_TO_VM_PAGE(pa)->flags &= ~PG_CLEAN;
#endif
return (pc);
}
/* FALLTHROUGH */
case T_TLB_MOD+T_USER:
{
pmap_hash_t hp;
#ifndef ATTR
vm_offset_t pa;
#endif
#ifdef DIAGNOSTIC
extern pmap_hash_t zero_pmap_hash;
extern pmap_t cur_pmap;
if (cur_pmap->pm_hash == zero_pmap_hash ||
cur_pmap->pm_hash == (pmap_hash_t)0)
panic("tlbmod");
#endif
hp = &((pmap_hash_t)PMAP_HASH_UADDR)[PMAP_HASH(vadr)];
if (((hp->pmh_pte[0].high ^ vadr) & ~PGOFSET) == 0)
i = 0;
else if (((hp->pmh_pte[1].high ^ vadr) & ~PGOFSET) == 0)
i = 1;
else
panic("trap: tlb umod not found");
if (hp->pmh_pte[i].low & PG_RO) {
ftype = VM_PROT_WRITE;
goto dofault;
}
hp->pmh_pte[i].low |= PG_M;
printf("trap: TLBupdate hi %x lo %x i %x\n",
hp->pmh_pte[i].high, hp->pmh_pte[i].low,
MachTLBUpdate(hp->pmh_pte[i].high, hp->pmh_pte[i].low)); /* XXX */
#ifdef ATTR
pmap_attributes[atop(hp->pmh_pte[i].low - KERNBASE)] |=
PMAP_ATTR_MOD;
#else
pa = hp->pmh_pte[i].low & PG_FRAME;
if (!IS_VM_PHYSADDR(pa))
panic("trap: umod");
PHYS_TO_VM_PAGE(pa)->flags &= ~PG_CLEAN;
#endif
if (!USERMODE(statusReg))
return (pc);
goto out;
}
case T_TLB_LD_MISS:
case T_TLB_ST_MISS:
ftype = (type == T_TLB_ST_MISS) ? VM_PROT_WRITE : VM_PROT_READ;
/* check for kernel address */
if ((int)vadr < 0) {
register vm_offset_t va;
int rv;
kernel_fault:
va = trunc_page((vm_offset_t)vadr);
rv = vm_fault(kernel_map, va, ftype, FALSE);
if (rv == KERN_SUCCESS)
return (pc);
if (i = ((struct pcb *)UADDR)->pcb_onfault) {
((struct pcb *)UADDR)->pcb_onfault = 0;
return (onfault_table[i]);
}
goto err;
}
/*
* It is an error for the kernel to access user space except
* through the copyin/copyout routines.
*/
if ((i = ((struct pcb *)UADDR)->pcb_onfault) == 0)
goto err;
/* check for fuswintr() or suswintr() getting a page fault */
if (i == 4)
return (onfault_table[i]);
goto dofault;
case T_TLB_LD_MISS+T_USER:
ftype = VM_PROT_READ;
goto dofault;
case T_TLB_ST_MISS+T_USER:
ftype = VM_PROT_WRITE;
dofault:
{
register vm_offset_t va;
register struct vmspace *vm = p->p_vmspace;
register vm_map_t map = &vm->vm_map;
int rv;
#ifdef X_KLUGE
if (p->p_pid == UserMapPid &&
(va = pmax_btop(vadr - USER_MAP_ADDR)) < nUserMapPtes) {
register pt_entry_t *pte;
pte = &UserMapPtes[va];
MachTLBWriteRandom((vadr & PG_FRAME) |
(vm->vm_pmap.pm_tlbpid << VMMACH_TLB_PID_SHIFT),
pte->pt_entry);
return (pc);
}
#endif
va = trunc_page((vm_offset_t)vadr);
rv = vm_fault(map, va, ftype, FALSE);
if (rv != KERN_SUCCESS) {
printf("vm_fault(%x, %x, %x, 0) -> %x ADR %x PC %x RA %x\n",
map, va, ftype, rv, vadr, pc,
!USERMODE(statusReg) ? ((int *)&args)[19] :
p->p_md.md_regs[RA]); /* XXX */
printf("\tpid %d %s PC %x RA %x SP %x\n", p->p_pid,
p->p_comm, p->p_md.md_regs[PC],
p->p_md.md_regs[RA],
p->p_md.md_regs[SP]); /* XXX */
}
/*
* If this was a stack access we keep track of the maximum
* accessed stack size. Also, if vm_fault gets a protection
* failure it is due to accessing the stack region outside
* the current limit and we need to reflect that as an access
* error.
*/
if ((caddr_t)va >= vm->vm_maxsaddr) {
if (rv == KERN_SUCCESS) {
unsigned nss;
nss = clrnd(btoc(USRSTACK-(unsigned)va));
if (nss > vm->vm_ssize)
vm->vm_ssize = nss;
} else if (rv == KERN_PROTECTION_FAILURE)
rv = KERN_INVALID_ADDRESS;
}
if (rv == KERN_SUCCESS) {
if (!USERMODE(statusReg))
return (pc);
goto out;
}
if (!USERMODE(statusReg)) {
if (i = ((struct pcb *)UADDR)->pcb_onfault) {
((struct pcb *)UADDR)->pcb_onfault = 0;
return (onfault_table[i]);
}
goto err;
}
ucode = vadr;
i = (rv == KERN_PROTECTION_FAILURE) ? SIGBUS : SIGSEGV;
break;
}
case T_ADDR_ERR_LD+T_USER: /* misaligned or kseg access */
case T_ADDR_ERR_ST+T_USER: /* misaligned or kseg access */
case T_BUS_ERR_IFETCH+T_USER: /* BERR asserted to cpu */
case T_BUS_ERR_LD_ST+T_USER: /* BERR asserted to cpu */
i = SIGSEGV;
break;
case T_SYSCALL+T_USER:
{
register int *locr0 = p->p_md.md_regs;
register struct sysent *callp;
unsigned int code;
int numsys;
struct args {
int i[8];
} args;
int rval[2];
struct sysent *systab;
extern int nsysent;
#ifdef ULTRIXCOMPAT
extern struct sysent ultrixsysent[];
extern int ultrixnsysent;
#endif
cnt.v_syscall++;
/* compute next PC after syscall instruction */
if ((int)causeReg < 0)
locr0[PC] = MachEmulateBranch(locr0, pc, 0, 0);
else
locr0[PC] += 4;
systab = sysent;
numsys = nsysent;
#ifdef ULTRIXCOMPAT
if (p->p_md.md_flags & MDP_ULTRIX) {
systab = ultrixsysent;
numsys = ultrixnsysent;
}
#endif
code = locr0[V0];
switch (code) {
case SYS_indir:
/*
* Code is first argument, followed by actual args.
*/
code = locr0[A0];
if (code >= numsys)
callp = &systab[SYS_indir]; /* (illegal) */
else
callp = &systab[code];
i = callp->sy_narg;
args.i[0] = locr0[A1];
args.i[1] = locr0[A2];
args.i[2] = locr0[A3];
if (i > 3) {
i = copyin((caddr_t)(locr0[SP] +
4 * sizeof(int)),
(caddr_t)&args.i[3],
(u_int)(i - 3) * sizeof(int));
if (i) {
locr0[V0] = i;
locr0[A3] = 1;
#ifdef KTRACE
if (KTRPOINT(p, KTR_SYSCALL))
ktrsyscall(p->p_tracep, code,
callp->sy_narg, args.i);
#endif
goto done;
}
}
break;
case SYS___indir:
/*
* Like indir, but code is a quad, so as to maintain
* quad alignment for the rest of the arguments.
*/
code = locr0[A0 + _QUAD_LOWWORD];
if (code >= numsys)
callp = &systab[SYS_indir]; /* (illegal) */
else
callp = &systab[code];
i = callp->sy_narg;
args.i[0] = locr0[A2];
args.i[1] = locr0[A3];
if (i > 2) {
i = copyin((caddr_t)(locr0[SP] +
4 * sizeof(int)),
(caddr_t)&args.i[2],
(u_int)(i - 2) * sizeof(int));
if (i) {
locr0[V0] = i;
locr0[A3] = 1;
#ifdef KTRACE
if (KTRPOINT(p, KTR_SYSCALL))
ktrsyscall(p->p_tracep, code,
callp->sy_narg, args.i);
#endif
goto done;
}
}
break;
default:
if (code >= numsys)
callp = &systab[SYS_indir]; /* (illegal) */
else
callp = &systab[code];
i = callp->sy_narg;
args.i[0] = locr0[A0];
args.i[1] = locr0[A1];
args.i[2] = locr0[A2];
args.i[3] = locr0[A3];
if (i > 4) {
i = copyin((caddr_t)(locr0[SP] +
4 * sizeof(int)),
(caddr_t)&args.i[4],
(u_int)(i - 4) * sizeof(int));
if (i) {
locr0[V0] = i;
locr0[A3] = 1;
#ifdef KTRACE
if (KTRPOINT(p, KTR_SYSCALL))
ktrsyscall(p->p_tracep, code,
callp->sy_narg, args.i);
#endif
goto done;
}
}
}
#ifdef KTRACE
if (KTRPOINT(p, KTR_SYSCALL))
ktrsyscall(p->p_tracep, code, callp->sy_narg, args.i);
#endif
rval[0] = 0;
rval[1] = locr0[V1];
#ifdef DEBUG
if (trp == trapdebug)
trapdebug[TRAPSIZE - 1].code = code;
else
trp[-1].code = code;
#endif
i = (*callp->sy_call)(p, &args, rval);
/*
* Reinitialize proc pointer `p' as it may be different
* if this is a child returning from fork syscall.
*/
p = curproc;
locr0 = p->p_md.md_regs;
#ifdef DEBUG
{ int s;
s = splhigh();
trp->status = statusReg;
trp->cause = causeReg;
trp->vadr = locr0[SP];
trp->pc = locr0[PC];
trp->ra = locr0[RA];
trp->code = -code;
if (++trp == &trapdebug[TRAPSIZE])
trp = trapdebug;
splx(s);
}
#endif
switch (i) {
case 0:
locr0[V0] = rval[0];
locr0[V1] = rval[1];
locr0[A3] = 0;
break;
case ERESTART:
locr0[PC] = pc;
break;
case EJUSTRETURN:
break; /* nothing to do */
default:
locr0[V0] = i;
locr0[A3] = 1;
}
done:
#ifdef KTRACE
if (KTRPOINT(p, KTR_SYSRET))
ktrsysret(p->p_tracep, code, i, rval[0]);
#endif
goto out;
}
case T_BREAK+T_USER:
{
register unsigned va, instr;
/* compute address of break instruction */
va = pc;
if ((int)causeReg < 0)
va += 4;
/* read break instruction */
instr = fuiword((caddr_t)va);
#ifdef KADB
if (instr == MACH_BREAK_BRKPT || instr == MACH_BREAK_SSTEP)
goto err;
#endif
if (p->p_md.md_ss_addr != va || instr != MACH_BREAK_SSTEP) {
i = SIGTRAP;
break;
}
/* restore original instruction and clear BP */
i = suiword((caddr_t)va, p->p_md.md_ss_instr);
if (i < 0) {
vm_offset_t sa, ea;
int rv;
sa = trunc_page((vm_offset_t)va);
ea = round_page((vm_offset_t)va+sizeof(int)-1);
rv = vm_map_protect(&p->p_vmspace->vm_map, sa, ea,
VM_PROT_DEFAULT, FALSE);
if (rv == KERN_SUCCESS) {
i = suiword((caddr_t)va, p->p_md.md_ss_instr);
(void) vm_map_protect(&p->p_vmspace->vm_map,
sa, ea, VM_PROT_READ|VM_PROT_EXECUTE,
FALSE);
}
}
if (i < 0) {
i = SIGTRAP;
break;
}
p->p_md.md_ss_addr = 0;
goto out;
}
case T_RES_INST+T_USER:
i = SIGILL;
break;
case T_COP_UNUSABLE+T_USER:
if ((causeReg & MACH_CR_COP_ERR) != 0x10000000) {
i = SIGILL; /* only FPU instructions allowed */
break;
}
MachSwitchFPState(machFPCurProcPtr, p->p_md.md_regs);
machFPCurProcPtr = p;
p->p_md.md_regs[PS] |= MACH_SR_COP_1_BIT;
p->p_md.md_flags |= MDP_FPUSED;
goto out;
case T_OVFLOW+T_USER:
i = SIGFPE;
break;
case T_ADDR_ERR_LD: /* misaligned access */
case T_ADDR_ERR_ST: /* misaligned access */
case T_BUS_ERR_LD_ST: /* BERR asserted to cpu */
if (i = ((struct pcb *)UADDR)->pcb_onfault) {
((struct pcb *)UADDR)->pcb_onfault = 0;
return (onfault_table[i]);
}
/* FALLTHROUGH */
default:
err:
#ifdef KADB
{
extern struct pcb kdbpcb;
if (USERMODE(statusReg))
kdbpcb = p->p_addr->u_pcb;
else {
kdbpcb.pcb_regs[ZERO] = 0;
kdbpcb.pcb_regs[AST] = ((int *)&args)[2];
kdbpcb.pcb_regs[V0] = ((int *)&args)[3];
kdbpcb.pcb_regs[V1] = ((int *)&args)[4];
kdbpcb.pcb_regs[A0] = ((int *)&args)[5];
kdbpcb.pcb_regs[A1] = ((int *)&args)[6];
kdbpcb.pcb_regs[A2] = ((int *)&args)[7];
kdbpcb.pcb_regs[A3] = ((int *)&args)[8];
kdbpcb.pcb_regs[T0] = ((int *)&args)[9];
kdbpcb.pcb_regs[T1] = ((int *)&args)[10];
kdbpcb.pcb_regs[T2] = ((int *)&args)[11];
kdbpcb.pcb_regs[T3] = ((int *)&args)[12];
kdbpcb.pcb_regs[T4] = ((int *)&args)[13];
kdbpcb.pcb_regs[T5] = ((int *)&args)[14];
kdbpcb.pcb_regs[T6] = ((int *)&args)[15];
kdbpcb.pcb_regs[T7] = ((int *)&args)[16];
kdbpcb.pcb_regs[T8] = ((int *)&args)[17];
kdbpcb.pcb_regs[T9] = ((int *)&args)[18];
kdbpcb.pcb_regs[RA] = ((int *)&args)[19];
kdbpcb.pcb_regs[MULLO] = ((int *)&args)[21];
kdbpcb.pcb_regs[MULHI] = ((int *)&args)[22];
kdbpcb.pcb_regs[PC] = pc;
kdbpcb.pcb_regs[SR] = statusReg;
bzero((caddr_t)&kdbpcb.pcb_regs[F0], 33 * sizeof(int));
}
if (kdb(causeReg, vadr, p, !USERMODE(statusReg)))
return (kdbpcb.pcb_regs[PC]);
}
#else
#ifdef DEBUG
trapDump("trap");
#endif
#endif
panic("trap");
}
printf("trap: pid %d %s sig %d adr %x pc %x ra %x\n", p->p_pid,
p->p_comm, i, vadr, pc, p->p_md.md_regs[RA]); /* XXX */
trapsignal(p, i, ucode);
out:
/*
* Note: we should only get here if returning to user mode.
*/
/* take pending signals */
while ((i = CURSIG(p)) != 0)
psig(i);
p->p_pri = p->p_usrpri;
astpending = 0;
if (want_resched) {
int s;
/*
* Since we are curproc, clock will normally just change
* our priority without moving us from one queue to another
* (since the running process is not on a queue.)
* If that happened after we setrq ourselves but before we
* swtch()'ed, we might not be on the queue indicated by
* our priority.
*/
s = splstatclock();
setrq(p);
p->p_stats->p_ru.ru_nivcsw++;
swtch();
splx(s);
while ((i = CURSIG(p)) != 0)
psig(i);
}
/*
* If profiling, charge system time to the trapped pc.
*/
if (p->p_flag & SPROFIL) {
extern int psratio;
addupc_task(p, pc, (int)(p->p_sticks - sticks) * psratio);
}
curpri = p->p_pri;
return (pc);
}
/*
* Handle an interrupt.
* Called from MachKernIntr() or MachUserIntr()
* Note: curproc might be NULL.
*/
interrupt(statusReg, causeReg, pc)
unsigned statusReg; /* status register at time of the exception */
unsigned causeReg; /* cause register at time of exception */
unsigned pc; /* program counter where to continue */
{
register unsigned mask;
struct clockframe cf;
#ifdef DEBUG
trp->status = statusReg;
trp->cause = causeReg;
trp->vadr = 0;
trp->pc = pc;
trp->ra = 0;
trp->code = 0;
if (++trp == &trapdebug[TRAPSIZE])
trp = trapdebug;
#endif
cnt.v_intr++;
mask = causeReg & statusReg; /* pending interrupts & enable mask */
if (mask & MACH_INT_MASK_5) {
if (!USERMODE(statusReg)) {
#ifdef DEBUG
trapDump("fpintr");
#else
printf("FPU interrupt: PC %x CR %x SR %x\n",
pc, causeReg, statusReg);
#endif
} else
MachFPInterrupt(statusReg, causeReg, pc);
}
if (pmax_hardware_intr)
(*pmax_hardware_intr)(mask, pc, statusReg, causeReg);
if (mask & MACH_SOFT_INT_MASK_0) {
clearsoftclock();
cnt.v_soft++;
softclock();
}
/* process network interrupt if we trapped or will very soon */
if ((mask & MACH_SOFT_INT_MASK_1) ||
netisr && (statusReg & MACH_SOFT_INT_MASK_1)) {
clearsoftnet();
cnt.v_soft++;
#ifdef INET
if (netisr & (1 << NETISR_ARP)) {
netisr &= ~(1 << NETISR_ARP);
arpintr();
}
if (netisr & (1 << NETISR_IP)) {
netisr &= ~(1 << NETISR_IP);
ipintr();
}
#endif
#ifdef NS
if (netisr & (1 << NETISR_NS)) {
netisr &= ~(1 << NETISR_NS);
nsintr();
}
#endif
#ifdef ISO
if (netisr & (1 << NETISR_ISO)) {
netisr &= ~(1 << NETISR_ISO);
clnlintr();
}
#endif
}
}
/*
* Handle pmax (DECstation 2100/3100) interrupts.
*/
pmax_intr(mask, pc, statusReg, causeReg)
unsigned mask;
unsigned pc;
unsigned statusReg;
unsigned causeReg;
{
register volatile struct chiptime *c = Mach_clock_addr;
struct clockframe cf;
int temp;
/* handle clock interrupts ASAP */
if (mask & MACH_INT_MASK_3) {
temp = c->regc; /* XXX clear interrupt bits */
cf.pc = pc;
cf.sr = statusReg;
hardclock(&cf);
/* keep clock interrupts enabled */
causeReg &= ~MACH_INT_MASK_3;
}
/* Re-enable clock interrupts */
splx(MACH_INT_MASK_3 | MACH_SR_INT_ENA_CUR);
#if NSII > 0
if (mask & MACH_INT_MASK_0)
siiintr(0);
#endif
#if NLE > 0
if (mask & MACH_INT_MASK_1)
leintr(0);
#endif
#if NDC > 0
if (mask & MACH_INT_MASK_2)
dcintr(0);
#endif
if (mask & MACH_INT_MASK_4)
pmax_errintr();
return ((statusReg & ~causeReg & MACH_HARD_INT_MASK) |
MACH_SR_INT_ENA_CUR);
}
/*
* Handle hardware interrupts for the KN02. (DECstation 5000/200)
* Returns spl value.
*/
kn02_intr(mask, pc, statusReg, causeReg)
unsigned mask;
unsigned pc;
unsigned statusReg;
unsigned causeReg;
{
register unsigned i, m;
register volatile struct chiptime *c = Mach_clock_addr;
register unsigned csr;
int temp;
struct clockframe cf;
static int warned = 0;
/* handle clock interrupts ASAP */
if (mask & MACH_INT_MASK_1) {
csr = *(unsigned *)MACH_PHYS_TO_UNCACHED(KN02_SYS_CSR);
if ((csr & KN02_CSR_PSWARN) && !warned) {
warned = 1;
printf("WARNING: power supply is overheating!\n");
} else if (warned && !(csr & KN02_CSR_PSWARN)) {
warned = 0;
printf("WARNING: power supply is OK again\n");
}
temp = c->regc; /* XXX clear interrupt bits */
cf.pc = pc;
cf.sr = statusReg;
hardclock(&cf);
/* keep clock interrupts enabled */
causeReg &= ~MACH_INT_MASK_1;
}
/* Re-enable clock interrupts */
splx(MACH_INT_MASK_1 | MACH_SR_INT_ENA_CUR);
if (mask & MACH_INT_MASK_0) {
csr = *(unsigned *)MACH_PHYS_TO_UNCACHED(KN02_SYS_CSR);
m = csr & (csr >> KN02_CSR_IOINTEN_SHIFT) & KN02_CSR_IOINT;
#if 0
*(unsigned *)MACHPHYS_TO_UNCACHED(KN02_SYS_CSR) =
(csr & ~(KN02_CSR_WRESERVED | 0xFF)) |
(m << KN02_CSR_IOINTEN_SHIFT);
#endif
for (i = 0; m; i++, m >>= 1) {
if (!(m & 1))
continue;
if (tc_slot_info[i].intr)
(*tc_slot_info[i].intr)(tc_slot_info[i].unit);
else
printf("spurious interrupt %d\n", i);
}
#if 0
*(unsigned *)MACH_PHYS_TO_UNCACHED(KN02_SYS_CSR) =
csr & ~(KN02_CSR_WRESERVED | 0xFF);
#endif
}
if (mask & MACH_INT_MASK_3)
kn02_errintr();
return ((statusReg & ~causeReg & MACH_HARD_INT_MASK) |
MACH_SR_INT_ENA_CUR);
}
/*
* 3min hardware interrupts. (DECstation 5000/1xx)
*/
kmin_intr(mask, pc, statusReg, causeReg)
unsigned mask;
unsigned pc;
unsigned statusReg;
unsigned causeReg;
{
register u_int intr;
register volatile struct chiptime *c = Mach_clock_addr;
register volatile u_int *imaskp =
(volatile u_int *)MACH_PHYS_TO_UNCACHED(KMIN_REG_IMSK);
register volatile u_int *intrp =
(volatile u_int *)MACH_PHYS_TO_UNCACHED(KMIN_REG_INTR);
unsigned int old_mask;
struct clockframe cf;
int temp;
static int user_warned = 0;
old_mask = (*imaskp & kmin_tc3_imask);
if (mask & MACH_INT_MASK_4)
(*callv->halt)((int *)0, 0);
if (mask & MACH_INT_MASK_3) {
/* masked interrupts are still observable */
intr = (*intrp & old_mask);
*intrp = 0;
/*
* Set things up so that the asic
* interrupts can be flipped on after being serviced.
*/
*imaskp = 0;
MachEmptyWriteBuffer();
splx(MACH_INT_MASK_3 | MACH_SR_INT_ENA_CUR);
#ifdef notdef
if (intr & (KMIN_INTR_SCSI_PTR_LOAD |
KMIN_INTR_SCSI_OVRUN | KMIN_INTR_SCSI_READ_E)) {
asc_dma_intr(intr);
}
#endif
if (intr & KMIN_INTR_TIMEOUT)
kn02ba_errintr();
if (intr & KMIN_INTR_CLOCK) {
temp = c->regc; /* XXX clear interrupt bits */
cf.pc = pc;
cf.sr = statusReg;
hardclock(&cf);
}
*imaskp = KMIN_INTR_CLOCK; /* Re-enable clock */
if ((intr & KMIN_INTR_SCC_0) &&
tc_slot_info[KMIN_SCC0_SLOT].intr)
(*(tc_slot_info[KMIN_SCC0_SLOT].intr))
(tc_slot_info[KMIN_SCC0_SLOT].unit);
if ((intr & KMIN_INTR_SCC_1) &&
tc_slot_info[KMIN_SCC1_SLOT].intr)
(*(tc_slot_info[KMIN_SCC1_SLOT].intr))
(tc_slot_info[KMIN_SCC1_SLOT].unit);
/*
* Re-enable serial ports, since they are sensitive to
* interrupt latency.
*/
*imaskp = (KMIN_INTR_CLOCK | KMIN_INTR_SCC_0 | KMIN_INTR_SCC_1);
if ((intr & KMIN_INTR_SCSI) &&
tc_slot_info[KMIN_SCSI_SLOT].intr)
(*(tc_slot_info[KMIN_SCSI_SLOT].intr))
(tc_slot_info[KMIN_SCSI_SLOT].unit);
if ((intr & KMIN_INTR_LANCE) &&
tc_slot_info[KMIN_LANCE_SLOT].intr)
(*(tc_slot_info[KMIN_LANCE_SLOT].intr))
(tc_slot_info[KMIN_LANCE_SLOT].unit);
if (user_warned && ((intr & KMIN_INTR_PSWARN) == 0)) {
printf("%s\n", "Power supply ok now.");
user_warned = 0;
}
if ((intr & KMIN_INTR_PSWARN) && (user_warned < 3)) {
user_warned++;
printf("%s\n", "Power supply overheating");
}
}
if ((mask & MACH_INT_MASK_0) && tc_slot_info[0].intr)
(*tc_slot_info[0].intr)(tc_slot_info[0].unit);
if ((mask & MACH_INT_MASK_1) && tc_slot_info[1].intr)
(*tc_slot_info[1].intr)(tc_slot_info[1].unit);
if ((mask & MACH_INT_MASK_2) && tc_slot_info[2].intr)
(*tc_slot_info[2].intr)(tc_slot_info[2].unit);
/*
* Re-enable all asic interrupts.
*/
*imaskp = old_mask;
return ((statusReg & MACH_HARD_INT_MASK) |
MACH_SR_INT_ENA_CUR);
}
/*
* Maxine hardware interrupts. (Personal DECstation 5000/xx)
*/
xine_intr(mask, pc, statusReg, causeReg)
unsigned mask;
unsigned pc;
unsigned statusReg;
unsigned causeReg;
{
register u_int intr;
register volatile struct chiptime *c = Mach_clock_addr;
register volatile u_int *imaskp = (volatile u_int *)
MACH_PHYS_TO_UNCACHED(XINE_REG_IMSK);
register volatile u_int *intrp = (volatile u_int *)
MACH_PHYS_TO_UNCACHED(XINE_REG_INTR);
u_int old_mask;
struct clockframe cf;
int temp;
old_mask = *imaskp & xine_tc3_imask;
if (mask & MACH_INT_MASK_4)
(*callv->halt)((int *)0, 0);
/* handle clock interrupts ASAP */
if (mask & MACH_INT_MASK_1) {
temp = c->regc; /* XXX clear interrupt bits */
cf.pc = pc;
cf.sr = statusReg;
hardclock(&cf);
}
if (mask & MACH_INT_MASK_2)
kn02ba_errintr();
/*
* Service all I/O ASIC interrupts.
*/
if (mask & MACH_INT_MASK_3) {
/* masked interrupts are still observable */
intr = (*intrp & old_mask);
*intrp = 0;
/*
* Re-enable clock interrupts.
*/
*imaskp = 0;
MachEmptyWriteBuffer();
splx((statusReg & MACH_HARD_INT_MASK) |
MACH_SR_INT_ENA_CUR);
#ifdef notdef
if (intr & (XINE_INTR_SCSI_PTR_LOAD |
XINE_INTR_SCSI_OVRUN | XINE_INTR_SCSI_READ_E)) {
asc_dma_intr(intr);
}
#endif
if ((intr & XINE_INTR_SCC_0) &&
tc_slot_info[XINE_SCC0_SLOT].intr)
(*(tc_slot_info[XINE_SCC0_SLOT].intr))
(tc_slot_info[XINE_SCC0_SLOT].unit);
if ((intr & XINE_INTR_DTOP_RX) &&
tc_slot_info[XINE_DTOP_SLOT].intr)
(*(tc_slot_info[XINE_DTOP_SLOT].intr))
(tc_slot_info[XINE_DTOP_SLOT].unit);
/*
* Re-enable the serial and desktop bus interrupts.
*/
*imaskp = (XINE_INTR_SCC_0 | XINE_INTR_DTOP_RX);
if ((intr & XINE_INTR_FLOPPY) &&
tc_slot_info[XINE_FLOPPY_SLOT].intr)
(*(tc_slot_info[XINE_FLOPPY_SLOT].intr))
(tc_slot_info[XINE_FLOPPY_SLOT].unit);
if ((intr & XINE_INTR_TC_0) &&
tc_slot_info[0].intr)
(*(tc_slot_info[0].intr))
(tc_slot_info[0].unit);
if ((intr & XINE_INTR_TC_1) &&
tc_slot_info[1].intr)
(*(tc_slot_info[1].intr))
(tc_slot_info[1].unit);
if ((intr & XINE_INTR_ISDN) &&
tc_slot_info[XINE_ISDN_SLOT].intr)
(*(tc_slot_info[XINE_ISDN_SLOT].intr))
(tc_slot_info[XINE_ISDN_SLOT].unit);
if ((intr & XINE_INTR_LANCE) &&
tc_slot_info[XINE_LANCE_SLOT].intr)
(*(tc_slot_info[XINE_LANCE_SLOT].intr))
(tc_slot_info[XINE_LANCE_SLOT].unit);
if ((intr & XINE_INTR_SCSI) &&
tc_slot_info[XINE_SCSI_SLOT].intr)
(*(tc_slot_info[XINE_SCSI_SLOT].intr))
(tc_slot_info[XINE_SCSI_SLOT].unit);
/*
* Turn the interrupts all back on.
*/
*imaskp = old_mask;
}
return ((statusReg & MACH_HARD_INT_MASK) |
MACH_SR_INT_ENA_CUR);
}
#ifdef DS5000_240
/*
* 3Max+ hardware interrupts. (DECstation 5000/240) UNTESTED!!
*/
kn03_intr(mask, pc, statusReg, causeReg)
unsigned mask;
unsigned pc;
unsigned statusReg;
unsigned causeReg;
{
register u_int intr;
register volatile struct chiptime *c = Mach_clock_addr;
volatile u_int *imaskp = (volatile u_int *)
MACH_PHYS_TO_UNCACHED(KN03_REG_IMSK);
volatile u_int *intrp = (volatile u_int *)
MACH_PHYS_TO_UNCACHED(KN03_REG_INTR);
u_int old_mask;
struct clockframe cf;
int temp;
static int user_warned = 0;
old_mask = *imaskp & kn03_tc3_imask;
*imaskp = old_mask;
if (mask & MACH_INT_MASK_4)
(*callv->halt)((int *)0, 0);
/* handle clock interrupts ASAP */
if (mask & MACH_INT_MASK_1) {
temp = c->regc; /* XXX clear interrupt bits */
cf.pc = pc;
cf.sr = statusReg;
hardclock(&cf);
causeReg &= ~MACH_INT_MASK_1;
/* reenable clock interrupts */
splx(MACH_INT_MASK_1 | MACH_SR_INT_ENA_CUR);
}
if (mask & MACH_INT_MASK_0) {
intr = *intrp;
/* masked interrupts are still observable */
intr &= old_mask;
if (intr & KN03_INTR_SCSI_PTR_LOAD) {
*intrp &= ~KN03_INTR_SCSI_PTR_LOAD;
#ifdef notdef
asc_dma_intr();
#endif
}
if (intr & (KN03_INTR_SCSI_OVRUN | KN03_INTR_SCSI_READ_E))
*intrp &= ~(KN03_INTR_SCSI_OVRUN | KN03_INTR_SCSI_READ_E);
if (intr & KN03_INTR_LANCE_READ_E)
*intrp &= ~KN03_INTR_LANCE_READ_E;
if ((intr & KN03_INTR_SCC_0) &&
tc_slot_info[KN03_SCC0_SLOT].intr)
(*(tc_slot_info[KN03_SCC0_SLOT].intr))
(tc_slot_info[KN03_SCC0_SLOT].unit);
if ((intr & KN03_INTR_SCC_1) &&
tc_slot_info[KN03_SCC1_SLOT].intr)
(*(tc_slot_info[KN03_SCC1_SLOT].intr))
(tc_slot_info[KN03_SCC1_SLOT].unit);
if ((intr & KN03_INTR_TC_0) &&
tc_slot_info[0].intr)
(*(tc_slot_info[0].intr))
(tc_slot_info[0].unit);
if ((intr & KN03_INTR_TC_1) &&
tc_slot_info[1].intr)
(*(tc_slot_info[1].intr))
(tc_slot_info[1].unit);
if ((intr & KN03_INTR_TC_2) &&
tc_slot_info[2].intr)
(*(tc_slot_info[2].intr))
(tc_slot_info[2].unit);
if ((intr & KN03_INTR_SCSI) &&
tc_slot_info[KN03_SCSI_SLOT].intr)
(*(tc_slot_info[KN03_SCSI_SLOT].intr))
(tc_slot_info[KN03_SCSI_SLOT].unit);
if ((intr & KN03_INTR_LANCE) &&
tc_slot_info[KN03_LANCE_SLOT].intr)
(*(tc_slot_info[KN03_LANCE_SLOT].intr))
(tc_slot_info[KN03_LANCE_SLOT].unit);
if (user_warned && ((intr & KN03_INTR_PSWARN) == 0)) {
printf("%s\n", "Power supply ok now.");
user_warned = 0;
}
if ((intr & KN03_INTR_PSWARN) && (user_warned < 3)) {
user_warned++;
printf("%s\n", "Power supply overheating");
}
}
if (mask & MACH_INT_MASK_3)
kn03_errintr();
return ((statusReg & ~causeReg & MACH_HARD_INT_MASK) |
MACH_SR_INT_ENA_CUR);
}
#endif /* DS5000_240 */
/*
* This is called from MachUserIntr() if astpending is set.
* This is very similar to the tail of trap().
*/
softintr(statusReg, pc)
unsigned statusReg; /* status register at time of the exception */
unsigned pc; /* program counter where to continue */
{
register struct proc *p = curproc;
int sig;
cnt.v_soft++;
/* take pending signals */
while ((sig = CURSIG(p)) != 0)
psig(sig);
p->p_pri = p->p_usrpri;
astpending = 0;
if (p->p_flag & SOWEUPC) {
p->p_flag &= ~SOWEUPC;
ADDUPROF(p);
}
if (want_resched) {
int s;
/*
* Since we are curproc, clock will normally just change
* our priority without moving us from one queue to another
* (since the running process is not on a queue.)
* If that happened after we setrq ourselves but before we
* swtch()'ed, we might not be on the queue indicated by
* our priority.
*/
s = splstatclock();
setrq(p);
p->p_stats->p_ru.ru_nivcsw++;
swtch();
splx(s);
while ((sig = CURSIG(p)) != 0)
psig(sig);
}
curpri = p->p_pri;
}
#ifdef DEBUG
trapDump(msg)
char *msg;
{
register int i;
int s;
s = splhigh();
printf("trapDump(%s)\n", msg);
for (i = 0; i < TRAPSIZE; i++) {
if (trp == trapdebug)
trp = &trapdebug[TRAPSIZE - 1];
else
trp--;
if (trp->cause == 0)
break;
printf("%s: ADR %x PC %x CR %x SR %x\n",
trap_type[(trp->cause & MACH_CR_EXC_CODE) >>
MACH_CR_EXC_CODE_SHIFT],
trp->vadr, trp->pc, trp->cause, trp->status);
printf(" RA %x code %d\n", trp-> ra, trp->code);
}
bzero(trapdebug, sizeof(trapdebug));
trp = trapdebug;
splx(s);
}
#endif
#ifdef X_KLUGE
/*
* This is a kludge to allow X windows to work.
*/
caddr_t
vmUserMap(size, pa)
int size;
unsigned pa;
{
register caddr_t v;
unsigned off, entry;
if (nUserMapPtes == 0)
UserMapPid = curproc->p_pid;
else if (UserMapPid != curproc->p_pid)
return ((caddr_t)0);
off = pa & PGOFSET;
size = btoc(off + size);
if (nUserMapPtes + size > NPTES)
return ((caddr_t)0);
v = (caddr_t)(USER_MAP_ADDR + pmax_ptob(nUserMapPtes) + off);
entry = (pa & 0x9ffff000) | PG_V | PG_M;
if (pa >= MACH_UNCACHED_MEMORY_ADDR)
entry |= PG_N;
while (size > 0) {
UserMapPtes[nUserMapPtes].pt_entry = entry;
entry += NBPG;
nUserMapPtes++;
size--;
}
return (v);
}
vmUserUnmap()
{
int id;
nUserMapPtes = 0;
if (UserMapPid == curproc->p_pid) {
id = curproc->p_vmspace->vm_pmap.pm_tlbpid;
if (id >= 0)
MachTLBFlushPID(id);
}
UserMapPid = 0;
}
#endif
/*
*----------------------------------------------------------------------
*
* MemErrorInterrupts --
* pmax_errintr - for the DS2100/DS3100
* kn02_errintr - for the DS5000/200
* kn02ba_errintr - for the DS5000/1xx and DS5000/xx
*
* Handler an interrupt for the control register.
*
* Results:
* None.
*
* Side effects:
* None.
*
*----------------------------------------------------------------------
*/
static void
pmax_errintr()
{
volatile u_short *sysCSRPtr =
(u_short *)MACH_PHYS_TO_UNCACHED(KN01_SYS_CSR);
u_short csr;
csr = *sysCSRPtr;
if (csr & KN01_CSR_MERR) {
printf("Memory error at 0x%x\n",
*(unsigned *)MACH_PHYS_TO_UNCACHED(KN01_SYS_ERRADR));
panic("Mem error interrupt");
}
*sysCSRPtr = (csr & ~KN01_CSR_MBZ) | 0xff;
}
static void
kn02_errintr()
{
printf("erradr %x\n", *(unsigned *)MACH_PHYS_TO_UNCACHED(KN02_SYS_ERRADR));
*(unsigned *)MACH_PHYS_TO_UNCACHED(KN02_SYS_ERRADR) = 0;
MachEmptyWriteBuffer();
}
#ifdef DS5000_240
static void
kn03_errintr()
{
printf("erradr %x\n", *(unsigned *)MACH_PHYS_TO_UNCACHED(KN03_SYS_ERRADR));
*(unsigned *)MACH_PHYS_TO_UNCACHED(KN03_SYS_ERRADR) = 0;
MachEmptyWriteBuffer();
}
#endif /* DS5000_240 */
static void
kn02ba_errintr()
{
register int mer, adr, siz;
static int errintr_cnt = 0;
siz = *(volatile int *)MACH_PHYS_TO_UNCACHED(KMIN_REG_MSR);
mer = *(volatile int *)MACH_PHYS_TO_UNCACHED(KMIN_REG_MER);
adr = *(volatile int *)MACH_PHYS_TO_UNCACHED(KMIN_REG_AER);
/* clear interrupt bit */
*(unsigned int *)MACH_PHYS_TO_UNCACHED(KMIN_REG_TIMEOUT) = 0;
errintr_cnt++;
printf("(%d)%s%x [%x %x %x]\n", errintr_cnt,
"Bad memory chip at phys ",
kn02ba_recover_erradr(adr, mer),
mer, siz, adr);
}
static unsigned
kn02ba_recover_erradr(phys, mer)
register unsigned phys, mer;
{
/* phys holds bits 28:2, mer knows which byte */
switch (mer & KMIN_MER_LASTBYTE) {
case KMIN_LASTB31:
mer = 3; break;
case KMIN_LASTB23:
mer = 2; break;
case KMIN_LASTB15:
mer = 1; break;
case KMIN_LASTB07:
mer = 0; break;
}
return ((phys & KMIN_AER_ADDR_MASK) | mer);
}
/*
* Return the resulting PC as if the branch was executed.
*/
unsigned
MachEmulateBranch(regsPtr, instPC, fpcCSR, allowNonBranch)
unsigned *regsPtr;
unsigned instPC;
unsigned fpcCSR;
int allowNonBranch;
{
InstFmt inst;
unsigned retAddr;
int condition;
extern unsigned GetBranchDest();
#if 0
printf("regsPtr=%x PC=%x Inst=%x fpcCsr=%x\n", regsPtr, instPC,
*instPC, fpcCSR);
#endif
inst = *(InstFmt *)instPC;
switch ((int)inst.JType.op) {
case OP_SPECIAL:
switch ((int)inst.RType.func) {
case OP_JR:
case OP_JALR:
retAddr = regsPtr[inst.RType.rs];
break;
default:
if (!allowNonBranch)
panic("MachEmulateBranch: Non-branch");
retAddr = instPC + 4;
break;
}
break;
case OP_BCOND:
switch ((int)inst.IType.rt) {
case OP_BLTZ:
case OP_BLTZAL:
if ((int)(regsPtr[inst.RType.rs]) < 0)
retAddr = GetBranchDest((InstFmt *)instPC);
else
retAddr = instPC + 8;
break;
case OP_BGEZAL:
case OP_BGEZ:
if ((int)(regsPtr[inst.RType.rs]) >= 0)
retAddr = GetBranchDest((InstFmt *)instPC);
else
retAddr = instPC + 8;
break;
default:
panic("MachEmulateBranch: Bad branch cond");
}
break;
case OP_J:
case OP_JAL:
retAddr = (inst.JType.target << 2) |
((unsigned)instPC & 0xF0000000);
break;
case OP_BEQ:
if (regsPtr[inst.RType.rs] == regsPtr[inst.RType.rt])
retAddr = GetBranchDest((InstFmt *)instPC);
else
retAddr = instPC + 8;
break;
case OP_BNE:
if (regsPtr[inst.RType.rs] != regsPtr[inst.RType.rt])
retAddr = GetBranchDest((InstFmt *)instPC);
else
retAddr = instPC + 8;
break;
case OP_BLEZ:
if ((int)(regsPtr[inst.RType.rs]) <= 0)
retAddr = GetBranchDest((InstFmt *)instPC);
else
retAddr = instPC + 8;
break;
case OP_BGTZ:
if ((int)(regsPtr[inst.RType.rs]) > 0)
retAddr = GetBranchDest((InstFmt *)instPC);
else
retAddr = instPC + 8;
break;
case OP_COP1:
switch (inst.RType.rs) {
case OP_BCx:
case OP_BCy:
if ((inst.RType.rt & COPz_BC_TF_MASK) == COPz_BC_TRUE)
condition = fpcCSR & MACH_FPC_COND_BIT;
else
condition = !(fpcCSR & MACH_FPC_COND_BIT);
if (condition)
retAddr = GetBranchDest((InstFmt *)instPC);
else
retAddr = instPC + 8;
break;
default:
if (!allowNonBranch)
panic("MachEmulateBranch: Bad coproc branch instruction");
retAddr = instPC + 4;
}
break;
default:
if (!allowNonBranch)
panic("MachEmulateBranch: Non-branch instruction");
retAddr = instPC + 4;
}
#if 0
printf("Target addr=%x\n", retAddr);
#endif
return (retAddr);
}
unsigned
GetBranchDest(InstPtr)
InstFmt *InstPtr;
{
return ((unsigned)InstPtr + 4 + ((short)InstPtr->IType.imm << 2));
}
/*
* This routine is called by procxmt() to single step one instruction.
* We do this by storing a break instruction after the current instruction,
* resuming execution, and then restoring the old instruction.
*/
cpu_singlestep(p)
register struct proc *p;
{
register unsigned va;
register int *locr0 = p->p_md.md_regs;
int i;
/* compute next address after current location */
va = MachEmulateBranch(locr0, locr0[PC], 0, 1);
if (p->p_md.md_ss_addr || p->p_md.md_ss_addr == va ||
!useracc((caddr_t)va, 4, B_READ)) {
printf("SS %s (%d): breakpoint already set at %x (va %x)\n",
p->p_comm, p->p_pid, p->p_md.md_ss_addr, va); /* XXX */
return (EFAULT);
}
p->p_md.md_ss_addr = va;
p->p_md.md_ss_instr = fuiword((caddr_t)va);
i = suiword((caddr_t)va, MACH_BREAK_SSTEP);
if (i < 0) {
vm_offset_t sa, ea;
int rv;
sa = trunc_page((vm_offset_t)va);
ea = round_page((vm_offset_t)va+sizeof(int)-1);
rv = vm_map_protect(&p->p_vmspace->vm_map, sa, ea,
VM_PROT_DEFAULT, FALSE);
if (rv == KERN_SUCCESS) {
i = suiword((caddr_t)va, MACH_BREAK_SSTEP);
(void) vm_map_protect(&p->p_vmspace->vm_map,
sa, ea, VM_PROT_READ|VM_PROT_EXECUTE, FALSE);
}
}
if (i < 0)
return (EFAULT);
printf("SS %s (%d): breakpoint set at %x: %x (pc %x)\n",
p->p_comm, p->p_pid, p->p_md.md_ss_addr,
p->p_md.md_ss_instr, locr0[PC]); /* XXX */
return (0);
}
#ifdef DEBUG
kdbpeek(addr)
{
if (addr & 3) {
printf("kdbpeek: unaligned address %x\n", addr);
return (-1);
}
return (*(int *)addr);
}
#define MIPS_JR_RA 0x03e00008 /* instruction code for jr ra */
/*
* Print a stack backtrace.
*/
void
stacktrace()
{
unsigned pc, sp, fp, ra, va, subr;
int a0, a1, a2, a3;
unsigned instr, mask;
InstFmt i;
int more, stksize;
int regs[8];
extern setsoftclock();
extern char start[], edata[];
cpu_getregs(regs);
/* get initial values from the exception frame */
sp = regs[0];
pc = regs[2];
ra = 0;
a0 = regs[3];
a1 = regs[4];
a2 = regs[5];
a3 = regs[6];
fp = regs[7];
loop:
/* check for current PC in the kernel interrupt handler code */
if (pc >= (unsigned)MachKernIntr && pc < (unsigned)MachUserIntr) {
/* NOTE: the offsets depend on the code in locore.s */
printf("interrupt\n");
a0 = kdbpeek(sp + 36);
a1 = kdbpeek(sp + 40);
a2 = kdbpeek(sp + 44);
a3 = kdbpeek(sp + 48);
pc = kdbpeek(sp + 20);
ra = kdbpeek(sp + 92);
sp = kdbpeek(sp + 100);
fp = kdbpeek(sp + 104);
}
/* check for current PC in the exception handler code */
if (pc >= 0x80000000 && pc < (unsigned)setsoftclock) {
ra = 0;
subr = 0;
goto done;
}
/* check for bad PC */
if (pc & 3 || pc < 0x80000000 || pc >= (unsigned)edata) {
printf("PC 0x%x: not in kernel\n", pc);
ra = 0;
subr = 0;
goto done;
}
/*
* Find the beginning of the current subroutine by scanning backwards
* from the current PC for the end of the previous subroutine.
*/
va = pc - sizeof(int);
while ((instr = kdbpeek(va)) != MIPS_JR_RA)
va -= sizeof(int);
va += 2 * sizeof(int); /* skip back over branch & delay slot */
/* skip over nulls which might separate .o files */
while ((instr = kdbpeek(va)) == 0)
va += sizeof(int);
subr = va;
/* scan forwards to find stack size and any saved registers */
stksize = 0;
more = 3;
mask = 0;
for (; more; va += sizeof(int), more = (more == 3) ? 3 : more - 1) {
/* stop if hit our current position */
if (va >= pc)
break;
instr = kdbpeek(va);
i.word = instr;
switch (i.JType.op) {
case OP_SPECIAL:
switch (i.RType.func) {
case OP_JR:
case OP_JALR:
more = 2; /* stop after next instruction */
break;
case OP_SYSCALL:
case OP_BREAK:
more = 1; /* stop now */
};
break;
case OP_BCOND:
case OP_J:
case OP_JAL:
case OP_BEQ:
case OP_BNE:
case OP_BLEZ:
case OP_BGTZ:
more = 2; /* stop after next instruction */
break;
case OP_COP0:
case OP_COP1:
case OP_COP2:
case OP_COP3:
switch (i.RType.rs) {
case OP_BCx:
case OP_BCy:
more = 2; /* stop after next instruction */
};
break;
case OP_SW:
/* look for saved registers on the stack */
if (i.IType.rs != 29)
break;
/* only restore the first one */
if (mask & (1 << i.IType.rt))
break;
mask |= 1 << i.IType.rt;
switch (i.IType.rt) {
case 4: /* a0 */
a0 = kdbpeek(sp + (short)i.IType.imm);
break;
case 5: /* a1 */
a1 = kdbpeek(sp + (short)i.IType.imm);
break;
case 6: /* a2 */
a2 = kdbpeek(sp + (short)i.IType.imm);
break;
case 7: /* a3 */
a3 = kdbpeek(sp + (short)i.IType.imm);
break;
case 30: /* fp */
fp = kdbpeek(sp + (short)i.IType.imm);
break;
case 31: /* ra */
ra = kdbpeek(sp + (short)i.IType.imm);
}
break;
case OP_ADDI:
case OP_ADDIU:
/* look for stack pointer adjustment */
if (i.IType.rs != 29 || i.IType.rt != 29)
break;
stksize = (short)i.IType.imm;
}
}
done:
printf("%x+%x (%x,%x,%x,%x) ra %x sz %d\n",
subr, pc - subr, a0, a1, a2, a3, ra, stksize);
if (ra) {
if (pc == ra && stksize == 0)
printf("stacktrace: loop!\n");
else {
pc = ra;
sp -= stksize;
goto loop;
}
}
}
#endif /* DEBUG */