[deliverable/linux.git] / arch / um / os-Linux / skas / process.c

/*
 * Copyright (C) 2002- 2004 Jeff Dike (jdike@addtoit.com)
 * Licensed under the GPL
 */

#include <stdlib.h>
#include <string.h>
#include <unistd.h>
#include <errno.h>
#include <signal.h>
#include <sched.h>
#include "ptrace_user.h"
#include <sys/wait.h>
#include <sys/mman.h>
#include <sys/user.h>
#include <sys/time.h>
#include <sys/syscall.h>
#include <asm/types.h>
#include "user.h"
#include "sysdep/ptrace.h"
#include "kern_util.h"
#include "skas.h"
#include "stub-data.h"
#include "mm_id.h"
#include "sysdep/sigcontext.h"
#include "sysdep/stub.h"
#include "os.h"
#include "proc_mm.h"
#include "skas_ptrace.h"
#include "chan_user.h"
#include "registers.h"
#include "mem.h"
#include "uml-config.h"
#include "process.h"
#include "longjmp.h"
#include "kern_constants.h"

int is_skas_winch(int pid, int fd, void *data)
{
	if(pid != os_getpgrp())
		return(0);

	register_winch_irq(-1, fd, -1, data);
	return(1);
}

static int ptrace_dump_regs(int pid)
{
        unsigned long regs[MAX_REG_NR];
        int i;

        if(ptrace(PTRACE_GETREGS, pid, 0, regs) < 0)
                return -errno;
        else {
                printk("Stub registers -\n");
                for(i = 0; i < ARRAY_SIZE(regs); i++)
                        printk("\t%d - %lx\n", i, regs[i]);
        }

        return 0;
}

void wait_stub_done(int pid, int sig, char * fname)
{
	int n, status, err;

	do {
		if ( sig != -1 ) {
			err = ptrace(PTRACE_CONT, pid, 0, sig);
			if(err)
				panic("%s : continue failed, errno = %d\n",
				      fname, errno);
		}
		sig = 0;

		CATCH_EINTR(n = waitpid(pid, &status, WUNTRACED));
	} while((n >= 0) && WIFSTOPPED(status) &&
	        ((WSTOPSIG(status) == SIGVTALRM) ||
		 /* running UML inside a detached screen can cause
		  * SIGWINCHes
		  */
		 (WSTOPSIG(status) == SIGWINCH)));

	if((n < 0) || !WIFSTOPPED(status) ||
	   (WSTOPSIG(status) != SIGUSR1 && WSTOPSIG(status) != SIGTRAP)){
		err = ptrace_dump_regs(pid);
		if(err)
			printk("Failed to get registers from stub, "
			       "errno = %d\n", -err);
		panic("%s : failed to wait for SIGUSR1/SIGTRAP, "
		      "pid = %d, n = %d, errno = %d, status = 0x%x\n",
		      fname, pid, n, errno, status);
	}
}

extern unsigned long current_stub_stack(void);

void get_skas_faultinfo(int pid, struct faultinfo * fi)
{
	int err;

	if(ptrace_faultinfo){
		err = ptrace(PTRACE_FAULTINFO, pid, 0, fi);
		if(err)
			panic("get_skas_faultinfo - PTRACE_FAULTINFO failed, "
			      "errno = %d\n", errno);

		/* Special handling for i386, which has different structs */
		if (sizeof(struct ptrace_faultinfo) < sizeof(struct faultinfo))
			memset((char *)fi + sizeof(struct ptrace_faultinfo), 0,
			       sizeof(struct faultinfo) -
			       sizeof(struct ptrace_faultinfo));
	}
	else {
		wait_stub_done(pid, SIGSEGV, "get_skas_faultinfo");

		/* faultinfo is prepared by the stub-segv-handler at start of
		 * the stub stack page. We just have to copy it.
		 */
		memcpy(fi, (void *)current_stub_stack(), sizeof(*fi));
	}
}

static void handle_segv(int pid, union uml_pt_regs * regs)
{
	get_skas_faultinfo(pid, &regs->skas.faultinfo);
	segv(regs->skas.faultinfo, 0, 1, NULL);
}

/*To use the same value of using_sysemu as the caller, ask it that value (in local_using_sysemu)*/
static void handle_trap(int pid, union uml_pt_regs *regs, int local_using_sysemu)
{
	int err, status;

	/* Mark this as a syscall */
	UPT_SYSCALL_NR(regs) = PT_SYSCALL_NR(regs->skas.regs);

	if (!local_using_sysemu)
	{
		err = ptrace(PTRACE_POKEUSR, pid, PT_SYSCALL_NR_OFFSET,
			     __NR_getpid);
		if(err < 0)
			panic("handle_trap - nullifying syscall failed errno = %d\n",
			      errno);

		err = ptrace(PTRACE_SYSCALL, pid, 0, 0);
		if(err < 0)
			panic("handle_trap - continuing to end of syscall failed, "
			      "errno = %d\n", errno);

		CATCH_EINTR(err = waitpid(pid, &status, WUNTRACED));
		if((err < 0) || !WIFSTOPPED(status) ||
		   (WSTOPSIG(status) != SIGTRAP + 0x80)){
                        err = ptrace_dump_regs(pid);
                        if(err)
                                printk("Failed to get registers from process, "
                                       "errno = %d\n", -err);
			panic("handle_trap - failed to wait at end of syscall, "
			      "errno = %d, status = %d\n", errno, status);
                }
	}

	handle_syscall(regs);
}

extern int __syscall_stub_start;

static int userspace_tramp(void *stack)
{
	void *addr;
	int err;

	ptrace(PTRACE_TRACEME, 0, 0, 0);

	init_new_thread_signals();
	err = set_interval(1);
	if(err)
		panic("userspace_tramp - setting timer failed, errno = %d\n",
		      err);

	if(!proc_mm){
		/* This has a pte, but it can't be mapped in with the usual
		 * tlb_flush mechanism because this is part of that mechanism
		 */
		int fd;
		__u64 offset;
		fd = phys_mapping(to_phys(&__syscall_stub_start), &offset);
		addr = mmap64((void *) UML_CONFIG_STUB_CODE, UM_KERN_PAGE_SIZE,
			      PROT_EXEC, MAP_FIXED | MAP_PRIVATE, fd, offset);
		if(addr == MAP_FAILED){
			printk("mapping mmap stub failed, errno = %d\n",
			       errno);
			exit(1);
		}

		if(stack != NULL){
			fd = phys_mapping(to_phys(stack), &offset);
			addr = mmap((void *) UML_CONFIG_STUB_DATA,
				    UM_KERN_PAGE_SIZE, PROT_READ | PROT_WRITE,
				    MAP_FIXED | MAP_SHARED, fd, offset);
			if(addr == MAP_FAILED){
				printk("mapping segfault stack failed, "
				       "errno = %d\n", errno);
				exit(1);
			}
		}
	}
	if(!ptrace_faultinfo && (stack != NULL)){
		struct sigaction sa;

		unsigned long v = UML_CONFIG_STUB_CODE +
				  (unsigned long) stub_segv_handler -
				  (unsigned long) &__syscall_stub_start;

		set_sigstack((void *) UML_CONFIG_STUB_DATA, UM_KERN_PAGE_SIZE);
		sigemptyset(&sa.sa_mask);
		sigaddset(&sa.sa_mask, SIGIO);
		sigaddset(&sa.sa_mask, SIGWINCH);
		sigaddset(&sa.sa_mask, SIGALRM);
		sigaddset(&sa.sa_mask, SIGVTALRM);
		sigaddset(&sa.sa_mask, SIGUSR1);
		sa.sa_flags = SA_ONSTACK;
		sa.sa_handler = (void *) v;
		sa.sa_restorer = NULL;
		if(sigaction(SIGSEGV, &sa, NULL) < 0)
			panic("userspace_tramp - setting SIGSEGV handler "
			      "failed - errno = %d\n", errno);
	}

	os_stop_process(os_getpid());
	return(0);
}

/* Each element set once, and only accessed by a single processor anyway */
#undef NR_CPUS
#define NR_CPUS 1
int userspace_pid[NR_CPUS];

int start_userspace(unsigned long stub_stack)
{
	void *stack;
	unsigned long sp;
	int pid, status, n, flags;

	stack = mmap(NULL, PAGE_SIZE, PROT_READ | PROT_WRITE | PROT_EXEC,
		     MAP_PRIVATE | MAP_ANONYMOUS, -1, 0);
	if(stack == MAP_FAILED)
		panic("start_userspace : mmap failed, errno = %d", errno);
	sp = (unsigned long) stack + PAGE_SIZE - sizeof(void *);

	flags = CLONE_FILES | SIGCHLD;
	if(proc_mm) flags |= CLONE_VM;
	pid = clone(userspace_tramp, (void *) sp, flags, (void *) stub_stack);
	if(pid < 0)
		panic("start_userspace : clone failed, errno = %d", errno);

	do {
		CATCH_EINTR(n = waitpid(pid, &status, WUNTRACED));
		if(n < 0)
			panic("start_userspace : wait failed, errno = %d",
			      errno);
	} while(WIFSTOPPED(status) && (WSTOPSIG(status) == SIGVTALRM));

	if(!WIFSTOPPED(status) || (WSTOPSIG(status) != SIGSTOP))
		panic("start_userspace : expected SIGSTOP, got status = %d",
		      status);

	if (ptrace(PTRACE_OLDSETOPTIONS, pid, NULL, (void *)PTRACE_O_TRACESYSGOOD) < 0)
		panic("start_userspace : PTRACE_OLDSETOPTIONS failed, errno=%d\n",
		      errno);

	if(munmap(stack, PAGE_SIZE) < 0)
		panic("start_userspace : munmap failed, errno = %d\n", errno);

	return(pid);
}

void userspace(union uml_pt_regs *regs)
{
	int err, status, op, pid = userspace_pid[0];
	int local_using_sysemu; /*To prevent races if using_sysemu changes under us.*/

	while(1){
		restore_registers(pid, regs);

		/* Now we set local_using_sysemu to be used for one loop */
		local_using_sysemu = get_using_sysemu();

		op = SELECT_PTRACE_OPERATION(local_using_sysemu, singlestepping(NULL));

		err = ptrace(op, pid, 0, 0);
		if(err)
			panic("userspace - could not resume userspace process, "
			      "pid=%d, ptrace operation = %d, errno = %d\n",
			      pid, op, errno);

		CATCH_EINTR(err = waitpid(pid, &status, WUNTRACED));
		if(err < 0)
			panic("userspace - waitpid failed, errno = %d\n",
			      errno);

		regs->skas.is_user = 1;
		save_registers(pid, regs);
		UPT_SYSCALL_NR(regs) = -1; /* Assume: It's not a syscall */

		if(WIFSTOPPED(status)){
		  	switch(WSTOPSIG(status)){
			case SIGSEGV:
				if(PTRACE_FULL_FAULTINFO || !ptrace_faultinfo)
					user_signal(SIGSEGV, regs, pid);
				else handle_segv(pid, regs);
				break;
			case SIGTRAP + 0x80:
			        handle_trap(pid, regs, local_using_sysemu);
				break;
			case SIGTRAP:
				relay_signal(SIGTRAP, regs);
				break;
			case SIGIO:
			case SIGVTALRM:
			case SIGILL:
			case SIGBUS:
			case SIGFPE:
			case SIGWINCH:
				user_signal(WSTOPSIG(status), regs, pid);
				break;
			default:
			        printk("userspace - child stopped with signal "
				       "%d\n", WSTOPSIG(status));
			}
			pid = userspace_pid[0];
			interrupt_end();

			/* Avoid -ERESTARTSYS handling in host */
			if(PT_SYSCALL_NR_OFFSET != PT_SYSCALL_RET_OFFSET)
				PT_SYSCALL_NR(regs->skas.regs) = -1;
		}
	}
}

int copy_context_skas0(unsigned long new_stack, int pid)
{
	int err;
	unsigned long regs[MAX_REG_NR];
	unsigned long fp_regs[HOST_FP_SIZE];
	unsigned long current_stack = current_stub_stack();
	struct stub_data *data = (struct stub_data *) current_stack;
	struct stub_data *child_data = (struct stub_data *) new_stack;
	__u64 new_offset;
	int new_fd = phys_mapping(to_phys((void *)new_stack), &new_offset);

	/* prepare offset and fd of child's stack as argument for parent's
	 * and child's mmap2 calls
	 */
	*data = ((struct stub_data) { .offset	= MMAP_OFFSET(new_offset),
				      .fd	= new_fd,
				      .timer    = ((struct itimerval)
					            { { 0, 1000000 / hz() },
						      { 0, 1000000 / hz() }})});
	get_safe_registers(regs, fp_regs);

	/* Set parent's instruction pointer to start of clone-stub */
	regs[REGS_IP_INDEX] = UML_CONFIG_STUB_CODE +
				(unsigned long) stub_clone_handler -
				(unsigned long) &__syscall_stub_start;
	regs[REGS_SP_INDEX] = UML_CONFIG_STUB_DATA + PAGE_SIZE -
		sizeof(void *);
#ifdef __SIGNAL_FRAMESIZE
	regs[REGS_SP_INDEX] -= __SIGNAL_FRAMESIZE;
#endif
	err = ptrace_setregs(pid, regs);
	if(err < 0)
		panic("copy_context_skas0 : PTRACE_SETREGS failed, "
		      "pid = %d, errno = %d\n", pid, -err);

	err = ptrace_setfpregs(pid, fp_regs);
	if(err < 0)
		panic("copy_context_skas0 : PTRACE_SETFPREGS failed, "
		      "pid = %d, errno = %d\n", pid, -err);

	/* set a well known return code for detection of child write failure */
	child_data->err = 12345678;

	/* Wait, until parent has finished its work: read child's pid from
	 * parent's stack, and check, if bad result.
	 */
	wait_stub_done(pid, 0, "copy_context_skas0");

	pid = data->err;
	if(pid < 0)
		panic("copy_context_skas0 - stub-parent reports error %d\n",
		      -pid);

	/* Wait, until child has finished too: read child's result from
	 * child's stack and check it.
	 */
	wait_stub_done(pid, -1, "copy_context_skas0");
	if (child_data->err != UML_CONFIG_STUB_DATA)
		panic("copy_context_skas0 - stub-child reports error %ld\n",
		      child_data->err);

	if (ptrace(PTRACE_OLDSETOPTIONS, pid, NULL,
		   (void *)PTRACE_O_TRACESYSGOOD) < 0)
		panic("copy_context_skas0 : PTRACE_OLDSETOPTIONS failed, "
		      "errno = %d\n", errno);

	return pid;
}

/*
 * This is used only, if stub pages are needed, while proc_mm is
 * available. Opening /proc/mm creates a new mm_context, which lacks
 * the stub-pages. Thus, we map them using /proc/mm-fd
 */
void map_stub_pages(int fd, unsigned long code,
		    unsigned long data, unsigned long stack)
{
	struct proc_mm_op mmop;
	int n;
	__u64 code_offset;
	int code_fd = phys_mapping(to_phys((void *) &__syscall_stub_start),
				   &code_offset);

	mmop = ((struct proc_mm_op) { .op        = MM_MMAP,
				      .u         =
				      { .mmap    =
					{ .addr    = code,
					  .len     = PAGE_SIZE,
					  .prot    = PROT_EXEC,
					  .flags   = MAP_FIXED | MAP_PRIVATE,
					  .fd      = code_fd,
					  .offset  = code_offset
	} } });
	CATCH_EINTR(n = write(fd, &mmop, sizeof(mmop)));
	if(n != sizeof(mmop)){
		n = errno;
		printk("mmap args - addr = 0x%lx, fd = %d, offset = %llx\n",
		       code, code_fd, (unsigned long long) code_offset);
		panic("map_stub_pages : /proc/mm map for code failed, "
		      "err = %d\n", n);
	}

	if ( stack ) {
		__u64 map_offset;
		int map_fd = phys_mapping(to_phys((void *)stack), &map_offset);
		mmop = ((struct proc_mm_op)
				{ .op        = MM_MMAP,
				  .u         =
				  { .mmap    =
				    { .addr    = data,
				      .len     = PAGE_SIZE,
				      .prot    = PROT_READ | PROT_WRITE,
				      .flags   = MAP_FIXED | MAP_SHARED,
				      .fd      = map_fd,
				      .offset  = map_offset
		} } });
		CATCH_EINTR(n = write(fd, &mmop, sizeof(mmop)));
		if(n != sizeof(mmop))
			panic("map_stub_pages : /proc/mm map for data failed, "
			      "err = %d\n", errno);
	}
}

void new_thread(void *stack, jmp_buf *buf, void (*handler)(void))
{
	(*buf)[0].JB_IP = (unsigned long) handler;
	(*buf)[0].JB_SP = (unsigned long) stack +
		(PAGE_SIZE << UML_CONFIG_KERNEL_STACK_ORDER) - sizeof(void *);
}

#define INIT_JMP_NEW_THREAD 0
#define INIT_JMP_CALLBACK 1
#define INIT_JMP_HALT 2
#define INIT_JMP_REBOOT 3

void switch_threads(jmp_buf *me, jmp_buf *you)
{
	if(UML_SETJMP(me) == 0)
		UML_LONGJMP(you, 1);
}

static jmp_buf initial_jmpbuf;

/* XXX Make these percpu */
static void (*cb_proc)(void *arg);
static void *cb_arg;
static jmp_buf *cb_back;

int start_idle_thread(void *stack, jmp_buf *switch_buf)
{
	int n;

	set_handler(SIGWINCH, (__sighandler_t) sig_handler,
		    SA_ONSTACK | SA_RESTART, SIGUSR1, SIGIO, SIGALRM,
		    SIGVTALRM, -1);

	n = UML_SETJMP(&initial_jmpbuf);
	switch(n){
	case INIT_JMP_NEW_THREAD:
		(*switch_buf)[0].JB_IP = (unsigned long) new_thread_handler;
		(*switch_buf)[0].JB_SP = (unsigned long) stack +
			(PAGE_SIZE << UML_CONFIG_KERNEL_STACK_ORDER) -
			sizeof(void *);
		break;
	case INIT_JMP_CALLBACK:
		(*cb_proc)(cb_arg);
		UML_LONGJMP(cb_back, 1);
		break;
	case INIT_JMP_HALT:
		kmalloc_ok = 0;
		return(0);
	case INIT_JMP_REBOOT:
		kmalloc_ok = 0;
		return(1);
	default:
		panic("Bad sigsetjmp return in start_idle_thread - %d\n", n);
	}
	UML_LONGJMP(switch_buf, 1);
}

void initial_thread_cb_skas(void (*proc)(void *), void *arg)
{
	jmp_buf here;

	cb_proc = proc;
	cb_arg = arg;
	cb_back = &here;

	block_signals();
	if(UML_SETJMP(&here) == 0)
		UML_LONGJMP(&initial_jmpbuf, INIT_JMP_CALLBACK);
	unblock_signals();

	cb_proc = NULL;
	cb_arg = NULL;
	cb_back = NULL;
}

void halt_skas(void)
{
	block_signals();
	UML_LONGJMP(&initial_jmpbuf, INIT_JMP_HALT);
}

void reboot_skas(void)
{
	block_signals();
	UML_LONGJMP(&initial_jmpbuf, INIT_JMP_REBOOT);
}

void switch_mm_skas(struct mm_id *mm_idp)
{
	int err;

#warning need cpu pid in switch_mm_skas
	if(proc_mm){
		err = ptrace(PTRACE_SWITCH_MM, userspace_pid[0], 0,
			     mm_idp->u.mm_fd);
		if(err)
			panic("switch_mm_skas - PTRACE_SWITCH_MM failed, "
			      "errno = %d\n", errno);
	}
	else userspace_pid[0] = mm_idp->u.pid;
}
Commit	Line	Data
abaf6977 GS	1	/*
	2	* Copyright (C) 2002- 2004 Jeff Dike (jdike@addtoit.com)
	3	* Licensed under the GPL
	4	*/
	5
	6	#include <stdlib.h>
	7	#include <string.h>
	8	#include <unistd.h>
	9	#include <errno.h>
	10	#include <signal.h>
abaf6977 GS	11	#include <sched.h>
	12	#include "ptrace_user.h"
	13	#include <sys/wait.h>
	14	#include <sys/mman.h>
	15	#include <sys/user.h>
	16	#include <sys/time.h>
1f6f6164	17	#include <sys/syscall.h>
abaf6977 GS	18	#include <asm/types.h>
	19	#include "user.h"
	20	#include "sysdep/ptrace.h"
abaf6977 GS	21	#include "kern_util.h"
	22	#include "skas.h"
	23	#include "stub-data.h"
	24	#include "mm_id.h"
	25	#include "sysdep/sigcontext.h"
	26	#include "sysdep/stub.h"
	27	#include "os.h"
	28	#include "proc_mm.h"
	29	#include "skas_ptrace.h"
	30	#include "chan_user.h"
	31	#include "registers.h"
	32	#include "mem.h"
	33	#include "uml-config.h"
	34	#include "process.h"
1d7173ba	35	#include "longjmp.h"
1ffb9164	36	#include "kern_constants.h"
abaf6977 GS	37
	38	int is_skas_winch(int pid, int fd, void *data)
	39	{
	40	if(pid != os_getpgrp())
	41	return(0);
	42
	43	register_winch_irq(-1, fd, -1, data);
	44	return(1);
	45	}
	46
f30c2c98 JD	47	static int ptrace_dump_regs(int pid)
	48	{
	49	unsigned long regs[MAX_REG_NR];
	50	int i;
	51
	52	if(ptrace(PTRACE_GETREGS, pid, 0, regs) < 0)
	53	return -errno;
	54	else {
	55	printk("Stub registers -\n");
	56	for(i = 0; i < ARRAY_SIZE(regs); i++)
	57	printk("\t%d - %lx\n", i, regs[i]);
	58	}
	59
	60	return 0;
	61	}
	62
abaf6977 GS	63	void wait_stub_done(int pid, int sig, char * fname)
	64	{
	65	int n, status, err;
	66
	67	do {
	68	if ( sig != -1 ) {
	69	err = ptrace(PTRACE_CONT, pid, 0, sig);
	70	if(err)
	71	panic("%s : continue failed, errno = %d\n",
	72	fname, errno);
	73	}
	74	sig = 0;
	75
	76	CATCH_EINTR(n = waitpid(pid, &status, WUNTRACED));
	77	} while((n >= 0) && WIFSTOPPED(status) &&
	78	((WSTOPSIG(status) == SIGVTALRM) \|\|
	79	/* running UML inside a detached screen can cause
	80	* SIGWINCHes
	81	*/
	82	(WSTOPSIG(status) == SIGWINCH)));
	83
	84	if((n < 0) \|\| !WIFSTOPPED(status) \|\|
	85	(WSTOPSIG(status) != SIGUSR1 && WSTOPSIG(status) != SIGTRAP)){
f30c2c98 JD	86	err = ptrace_dump_regs(pid);
f30c2c98 JD	87	if(err)
abaf6977	88	printk("Failed to get registers from stub, "
f30c2c98	89	"errno = %d\n", -err);
abaf6977 GS	90	panic("%s : failed to wait for SIGUSR1/SIGTRAP, "
	91	"pid = %d, n = %d, errno = %d, status = 0x%x\n",
	92	fname, pid, n, errno, status);
	93	}
	94	}
	95
	96	extern unsigned long current_stub_stack(void);
	97
	98	void get_skas_faultinfo(int pid, struct faultinfo * fi)
	99	{
	100	int err;
	101
	102	if(ptrace_faultinfo){
	103	err = ptrace(PTRACE_FAULTINFO, pid, 0, fi);
	104	if(err)
	105	panic("get_skas_faultinfo - PTRACE_FAULTINFO failed, "
	106	"errno = %d\n", errno);
	107
	108	/* Special handling for i386, which has different structs */
	109	if (sizeof(struct ptrace_faultinfo) < sizeof(struct faultinfo))
	110	memset((char *)fi + sizeof(struct ptrace_faultinfo), 0,
	111	sizeof(struct faultinfo) -
	112	sizeof(struct ptrace_faultinfo));
	113	}
	114	else {
	115	wait_stub_done(pid, SIGSEGV, "get_skas_faultinfo");
	116
	117	/* faultinfo is prepared by the stub-segv-handler at start of
	118	* the stub stack page. We just have to copy it.
	119	*/
	120	memcpy(fi, (void )current_stub_stack(), sizeof(fi));
	121	}
	122	}
	123
	124	static void handle_segv(int pid, union uml_pt_regs * regs)
	125	{
	126	get_skas_faultinfo(pid, &regs->skas.faultinfo);
	127	segv(regs->skas.faultinfo, 0, 1, NULL);
	128	}
	129
	130	/To use the same value of using_sysemu as the caller, ask it that value (in local_using_sysemu)/
	131	static void handle_trap(int pid, union uml_pt_regs *regs, int local_using_sysemu)
	132	{
	133	int err, status;
	134
	135	/* Mark this as a syscall */
	136	UPT_SYSCALL_NR(regs) = PT_SYSCALL_NR(regs->skas.regs);
	137
	138	if (!local_using_sysemu)
	139	{
	140	err = ptrace(PTRACE_POKEUSR, pid, PT_SYSCALL_NR_OFFSET,
	141	__NR_getpid);
	142	if(err < 0)
	143	panic("handle_trap - nullifying syscall failed errno = %d\n",
	144	errno);
	145
	146	err = ptrace(PTRACE_SYSCALL, pid, 0, 0);
	147	if(err < 0)
	148	panic("handle_trap - continuing to end of syscall failed, "
	149	"errno = %d\n", errno);
	150
	151	CATCH_EINTR(err = waitpid(pid, &status, WUNTRACED));
	152	if((err < 0) \|\| !WIFSTOPPED(status) \|\|
f30c2c98 JD	153	(WSTOPSIG(status) != SIGTRAP + 0x80)){
	154	err = ptrace_dump_regs(pid);
	155	if(err)
	156	printk("Failed to get registers from process, "
	157	"errno = %d\n", -err);
abaf6977 GS	158	panic("handle_trap - failed to wait at end of syscall, "
abaf6977 GS	159	"errno = %d, status = %d\n", errno, status);
f30c2c98	160	}
abaf6977 GS	161	}
	162
	163	handle_syscall(regs);
	164	}
	165
	166	extern int __syscall_stub_start;
abaf6977 GS	167
	168	static int userspace_tramp(void *stack)
	169	{
	170	void *addr;
537ae946	171	int err;
abaf6977 GS	172
	173	ptrace(PTRACE_TRACEME, 0, 0, 0);
	174
e64bd134	175	init_new_thread_signals();
537ae946 JD	176	err = set_interval(1);
	177	if(err)
	178	panic("userspace_tramp - setting timer failed, errno = %d\n",
	179	err);
abaf6977 GS	180
	181	if(!proc_mm){
	182	/* This has a pte, but it can't be mapped in with the usual
	183	* tlb_flush mechanism because this is part of that mechanism
	184	*/
09ee011e JD	185	int fd;
	186	__u64 offset;
	187	fd = phys_mapping(to_phys(&__syscall_stub_start), &offset);
1ffb9164	188	addr = mmap64((void *) UML_CONFIG_STUB_CODE, UM_KERN_PAGE_SIZE,
09ee011e	189	PROT_EXEC, MAP_FIXED \| MAP_PRIVATE, fd, offset);
abaf6977	190	if(addr == MAP_FAILED){
09ee011e	191	printk("mapping mmap stub failed, errno = %d\n",
abaf6977 GS	192	errno);
	193	exit(1);
	194	}
	195
	196	if(stack != NULL){
abaf6977	197	fd = phys_mapping(to_phys(stack), &offset);
1ffb9164 JD	198	addr = mmap((void *) UML_CONFIG_STUB_DATA,
1ffb9164 JD	199	UM_KERN_PAGE_SIZE, PROT_READ \| PROT_WRITE,
abaf6977 GS	200	MAP_FIXED \| MAP_SHARED, fd, offset);
abaf6977 GS	201	if(addr == MAP_FAILED){
09ee011e	202	printk("mapping segfault stack failed, "
abaf6977 GS	203	"errno = %d\n", errno);
	204	exit(1);
	205	}
	206	}
	207	}
09ee011e	208	if(!ptrace_faultinfo && (stack != NULL)){
4b84c69b JD	209	struct sigaction sa;
4b84c69b JD	210
abaf6977 GS	211	unsigned long v = UML_CONFIG_STUB_CODE +
	212	(unsigned long) stub_segv_handler -
	213	(unsigned long) &__syscall_stub_start;
	214
1ffb9164	215	set_sigstack((void *) UML_CONFIG_STUB_DATA, UM_KERN_PAGE_SIZE);
4b84c69b JD	216	sigemptyset(&sa.sa_mask);
	217	sigaddset(&sa.sa_mask, SIGIO);
	218	sigaddset(&sa.sa_mask, SIGWINCH);
	219	sigaddset(&sa.sa_mask, SIGALRM);
	220	sigaddset(&sa.sa_mask, SIGVTALRM);
	221	sigaddset(&sa.sa_mask, SIGUSR1);
	222	sa.sa_flags = SA_ONSTACK;
	223	sa.sa_handler = (void *) v;
	224	sa.sa_restorer = NULL;
	225	if(sigaction(SIGSEGV, &sa, NULL) < 0)
	226	panic("userspace_tramp - setting SIGSEGV handler "
	227	"failed - errno = %d\n", errno);
abaf6977 GS	228	}
	229
	230	os_stop_process(os_getpid());
	231	return(0);
	232	}
	233
	234	/* Each element set once, and only accessed by a single processor anyway */
	235	#undef NR_CPUS
	236	#define NR_CPUS 1
	237	int userspace_pid[NR_CPUS];
	238
	239	int start_userspace(unsigned long stub_stack)
	240	{
	241	void *stack;
	242	unsigned long sp;
	243	int pid, status, n, flags;
	244
abaf6977 GS	245	stack = mmap(NULL, PAGE_SIZE, PROT_READ \| PROT_WRITE \| PROT_EXEC,
	246	MAP_PRIVATE \| MAP_ANONYMOUS, -1, 0);
	247	if(stack == MAP_FAILED)
	248	panic("start_userspace : mmap failed, errno = %d", errno);
	249	sp = (unsigned long) stack + PAGE_SIZE - sizeof(void *);
	250
	251	flags = CLONE_FILES \| SIGCHLD;
	252	if(proc_mm) flags \|= CLONE_VM;
	253	pid = clone(userspace_tramp, (void ) sp, flags, (void ) stub_stack);
	254	if(pid < 0)
	255	panic("start_userspace : clone failed, errno = %d", errno);
	256
	257	do {
	258	CATCH_EINTR(n = waitpid(pid, &status, WUNTRACED));
	259	if(n < 0)
	260	panic("start_userspace : wait failed, errno = %d",
	261	errno);
	262	} while(WIFSTOPPED(status) && (WSTOPSIG(status) == SIGVTALRM));
	263
	264	if(!WIFSTOPPED(status) \|\| (WSTOPSIG(status) != SIGSTOP))
	265	panic("start_userspace : expected SIGSTOP, got status = %d",
	266	status);
	267
	268	if (ptrace(PTRACE_OLDSETOPTIONS, pid, NULL, (void *)PTRACE_O_TRACESYSGOOD) < 0)
	269	panic("start_userspace : PTRACE_OLDSETOPTIONS failed, errno=%d\n",
	270	errno);
	271
	272	if(munmap(stack, PAGE_SIZE) < 0)
	273	panic("start_userspace : munmap failed, errno = %d\n", errno);
	274
	275	return(pid);
	276	}
	277
	278	void userspace(union uml_pt_regs *regs)
	279	{
	280	int err, status, op, pid = userspace_pid[0];
	281	int local_using_sysemu; /To prevent races if using_sysemu changes under us./
	282
	283	while(1){
	284	restore_registers(pid, regs);
	285
	286	/* Now we set local_using_sysemu to be used for one loop */
	287	local_using_sysemu = get_using_sysemu();
	288
	289	op = SELECT_PTRACE_OPERATION(local_using_sysemu, singlestepping(NULL));
	290
	291	err = ptrace(op, pid, 0, 0);
	292	if(err)
	293	panic("userspace - could not resume userspace process, "
	294	"pid=%d, ptrace operation = %d, errno = %d\n",
802e3077	295	pid, op, errno);
abaf6977 GS	296
	297	CATCH_EINTR(err = waitpid(pid, &status, WUNTRACED));
	298	if(err < 0)
	299	panic("userspace - waitpid failed, errno = %d\n",
	300	errno);
	301
	302	regs->skas.is_user = 1;
	303	save_registers(pid, regs);
	304	UPT_SYSCALL_NR(regs) = -1; /* Assume: It's not a syscall */
	305
	306	if(WIFSTOPPED(status)){
	307	switch(WSTOPSIG(status)){
	308	case SIGSEGV:
	309	if(PTRACE_FULL_FAULTINFO \|\| !ptrace_faultinfo)
	310	user_signal(SIGSEGV, regs, pid);
	311	else handle_segv(pid, regs);
	312	break;
	313	case SIGTRAP + 0x80:
	314	handle_trap(pid, regs, local_using_sysemu);
	315	break;
	316	case SIGTRAP:
	317	relay_signal(SIGTRAP, regs);
	318	break;
	319	case SIGIO:
	320	case SIGVTALRM:
	321	case SIGILL:
	322	case SIGBUS:
	323	case SIGFPE:
	324	case SIGWINCH:
	325	user_signal(WSTOPSIG(status), regs, pid);
	326	break;
	327	default:
	328	printk("userspace - child stopped with signal "
	329	"%d\n", WSTOPSIG(status));
	330	}
abaf6977 GS	331	pid = userspace_pid[0];
	332	interrupt_end();
	333
	334	/* Avoid -ERESTARTSYS handling in host */
	335	if(PT_SYSCALL_NR_OFFSET != PT_SYSCALL_RET_OFFSET)
	336	PT_SYSCALL_NR(regs->skas.regs) = -1;
	337	}
	338	}
	339	}
abaf6977 GS	340
	341	int copy_context_skas0(unsigned long new_stack, int pid)
	342	{
	343	int err;
b92c4f92	344	unsigned long regs[MAX_REG_NR];
e2216feb	345	unsigned long fp_regs[HOST_FP_SIZE];
abaf6977 GS	346	unsigned long current_stack = current_stub_stack();
	347	struct stub_data data = (struct stub_data ) current_stack;
	348	struct stub_data child_data = (struct stub_data ) new_stack;
	349	__u64 new_offset;
	350	int new_fd = phys_mapping(to_phys((void *)new_stack), &new_offset);
	351
	352	/* prepare offset and fd of child's stack as argument for parent's
	353	* and child's mmap2 calls
	354	*/
	355	*data = ((struct stub_data) { .offset = MMAP_OFFSET(new_offset),
	356	.fd = new_fd,
	357	.timer = ((struct itimerval)
	358	{ { 0, 1000000 / hz() },
	359	{ 0, 1000000 / hz() }})});
e2216feb	360	get_safe_registers(regs, fp_regs);
abaf6977 GS	361
	362	/* Set parent's instruction pointer to start of clone-stub */
	363	regs[REGS_IP_INDEX] = UML_CONFIG_STUB_CODE +
	364	(unsigned long) stub_clone_handler -
	365	(unsigned long) &__syscall_stub_start;
	366	regs[REGS_SP_INDEX] = UML_CONFIG_STUB_DATA + PAGE_SIZE -
	367	sizeof(void *);
	368	#ifdef __SIGNAL_FRAMESIZE
	369	regs[REGS_SP_INDEX] -= __SIGNAL_FRAMESIZE;
	370	#endif
	371	err = ptrace_setregs(pid, regs);
	372	if(err < 0)
	373	panic("copy_context_skas0 : PTRACE_SETREGS failed, "
2ace87b9	374	"pid = %d, errno = %d\n", pid, -err);
abaf6977	375
e2216feb JD	376	err = ptrace_setfpregs(pid, fp_regs);
	377	if(err < 0)
	378	panic("copy_context_skas0 : PTRACE_SETFPREGS failed, "
2ace87b9	379	"pid = %d, errno = %d\n", pid, -err);
e2216feb	380
abaf6977 GS	381	/* set a well known return code for detection of child write failure */
	382	child_data->err = 12345678;
	383
	384	/* Wait, until parent has finished its work: read child's pid from
	385	* parent's stack, and check, if bad result.
	386	*/
	387	wait_stub_done(pid, 0, "copy_context_skas0");
	388
	389	pid = data->err;
	390	if(pid < 0)
	391	panic("copy_context_skas0 - stub-parent reports error %d\n",
2ace87b9	392	-pid);
abaf6977 GS	393
	394	/* Wait, until child has finished too: read child's result from
	395	* child's stack and check it.
	396	*/
	397	wait_stub_done(pid, -1, "copy_context_skas0");
	398	if (child_data->err != UML_CONFIG_STUB_DATA)
802e3077	399	panic("copy_context_skas0 - stub-child reports error %ld\n",
abaf6977 GS	400	child_data->err);
	401
	402	if (ptrace(PTRACE_OLDSETOPTIONS, pid, NULL,
	403	(void *)PTRACE_O_TRACESYSGOOD) < 0)
	404	panic("copy_context_skas0 : PTRACE_OLDSETOPTIONS failed, "
	405	"errno = %d\n", errno);
	406
	407	return pid;
	408	}
	409
	410	/*
	411	* This is used only, if stub pages are needed, while proc_mm is
ef0470c0	412	* available. Opening /proc/mm creates a new mm_context, which lacks
abaf6977 GS	413	* the stub-pages. Thus, we map them using /proc/mm-fd
	414	*/
	415	void map_stub_pages(int fd, unsigned long code,
	416	unsigned long data, unsigned long stack)
	417	{
	418	struct proc_mm_op mmop;
	419	int n;
09ee011e JD	420	__u64 code_offset;
	421	int code_fd = phys_mapping(to_phys((void *) &__syscall_stub_start),
	422	&code_offset);
abaf6977 GS	423
	424	mmop = ((struct proc_mm_op) { .op = MM_MMAP,
	425	.u =
	426	{ .mmap =
	427	{ .addr = code,
	428	.len = PAGE_SIZE,
	429	.prot = PROT_EXEC,
	430	.flags = MAP_FIXED \| MAP_PRIVATE,
09ee011e JD	431	.fd = code_fd,
09ee011e JD	432	.offset = code_offset
abaf6977	433	} } });
a61f334f	434	CATCH_EINTR(n = write(fd, &mmop, sizeof(mmop)));
b4cf95c6	435	if(n != sizeof(mmop)){
a61f334f	436	n = errno;
b4cf95c6 JD	437	printk("mmap args - addr = 0x%lx, fd = %d, offset = %llx\n",
b4cf95c6 JD	438	code, code_fd, (unsigned long long) code_offset);
abaf6977	439	panic("map_stub_pages : /proc/mm map for code failed, "
a61f334f	440	"err = %d\n", n);
b4cf95c6	441	}
abaf6977 GS	442
	443	if ( stack ) {
	444	__u64 map_offset;
	445	int map_fd = phys_mapping(to_phys((void *)stack), &map_offset);
	446	mmop = ((struct proc_mm_op)
	447	{ .op = MM_MMAP,
	448	.u =
	449	{ .mmap =
	450	{ .addr = data,
	451	.len = PAGE_SIZE,
	452	.prot = PROT_READ \| PROT_WRITE,
	453	.flags = MAP_FIXED \| MAP_SHARED,
	454	.fd = map_fd,
	455	.offset = map_offset
	456	} } });
a61f334f	457	CATCH_EINTR(n = write(fd, &mmop, sizeof(mmop)));
abaf6977 GS	458	if(n != sizeof(mmop))
abaf6977 GS	459	panic("map_stub_pages : /proc/mm map for data failed, "
a61f334f	460	"err = %d\n", errno);
abaf6977 GS	461	}
	462	}
	463
3c917350	464	void new_thread(void stack, jmp_buf buf, void (*handler)(void))
abaf6977	465	{
3c917350 JD	466	(*buf)[0].JB_IP = (unsigned long) handler;
	467	(*buf)[0].JB_SP = (unsigned long) stack +
	468	(PAGE_SIZE << UML_CONFIG_KERNEL_STACK_ORDER) - sizeof(void *);
abaf6977 GS	469	}
abaf6977 GS	470
e2216feb	471	#define INIT_JMP_NEW_THREAD 0
3c917350 JD	472	#define INIT_JMP_CALLBACK 1
	473	#define INIT_JMP_HALT 2
	474	#define INIT_JMP_REBOOT 3
abaf6977	475
3c917350	476	void switch_threads(jmp_buf me, jmp_buf you)
abaf6977	477	{
3c917350 JD	478	if(UML_SETJMP(me) == 0)
3c917350 JD	479	UML_LONGJMP(you, 1);
abaf6977 GS	480	}
abaf6977 GS	481
ad28e029	482	static jmp_buf initial_jmpbuf;
abaf6977 GS	483
	484	/* XXX Make these percpu */
	485	static void (cb_proc)(void arg);
	486	static void *cb_arg;
ad28e029	487	static jmp_buf *cb_back;
abaf6977	488
3c917350	489	int start_idle_thread(void stack, jmp_buf switch_buf)
abaf6977	490	{
a5df0d1a	491	int n;
abaf6977 GS	492
	493	set_handler(SIGWINCH, (__sighandler_t) sig_handler,
	494	SA_ONSTACK \| SA_RESTART, SIGUSR1, SIGIO, SIGALRM,
	495	SIGVTALRM, -1);
	496
a5df0d1a	497	n = UML_SETJMP(&initial_jmpbuf);
abaf6977 GS	498	switch(n){
abaf6977 GS	499	case INIT_JMP_NEW_THREAD:
3c917350 JD	500	(*switch_buf)[0].JB_IP = (unsigned long) new_thread_handler;
	501	(*switch_buf)[0].JB_SP = (unsigned long) stack +
	502	(PAGE_SIZE << UML_CONFIG_KERNEL_STACK_ORDER) -
	503	sizeof(void *);
abaf6977 GS	504	break;
	505	case INIT_JMP_CALLBACK:
	506	(*cb_proc)(cb_arg);
ad28e029	507	UML_LONGJMP(cb_back, 1);
abaf6977 GS	508	break;
	509	case INIT_JMP_HALT:
	510	kmalloc_ok = 0;
	511	return(0);
	512	case INIT_JMP_REBOOT:
	513	kmalloc_ok = 0;
	514	return(1);
	515	default:
	516	panic("Bad sigsetjmp return in start_idle_thread - %d\n", n);
	517	}
3c917350	518	UML_LONGJMP(switch_buf, 1);
abaf6977 GS	519	}
	520
	521	void initial_thread_cb_skas(void (proc)(void ), void *arg)
	522	{
ad28e029	523	jmp_buf here;
abaf6977 GS	524
	525	cb_proc = proc;
	526	cb_arg = arg;
	527	cb_back = &here;
	528
	529	block_signals();
a5df0d1a	530	if(UML_SETJMP(&here) == 0)
ad28e029	531	UML_LONGJMP(&initial_jmpbuf, INIT_JMP_CALLBACK);
abaf6977 GS	532	unblock_signals();
	533
	534	cb_proc = NULL;
	535	cb_arg = NULL;
	536	cb_back = NULL;
	537	}
	538
	539	void halt_skas(void)
	540	{
	541	block_signals();
ad28e029	542	UML_LONGJMP(&initial_jmpbuf, INIT_JMP_HALT);
abaf6977 GS	543	}
	544
	545	void reboot_skas(void)
	546	{
	547	block_signals();
ad28e029	548	UML_LONGJMP(&initial_jmpbuf, INIT_JMP_REBOOT);
abaf6977 GS	549	}
	550
	551	void switch_mm_skas(struct mm_id *mm_idp)
	552	{
	553	int err;
	554
	555	#warning need cpu pid in switch_mm_skas
	556	if(proc_mm){
	557	err = ptrace(PTRACE_SWITCH_MM, userspace_pid[0], 0,
	558	mm_idp->u.mm_fd);
	559	if(err)
	560	panic("switch_mm_skas - PTRACE_SWITCH_MM failed, "
	561	"errno = %d\n", errno);
	562	}
	563	else userspace_pid[0] = mm_idp->u.pid;
	564	}