[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"

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);
}

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)){
		unsigned long regs[MAX_REG_NR];

		if(ptrace(PTRACE_GETREGS, pid, 0, regs) < 0)
			printk("Failed to get registers from stub, "
			       "errno = %d\n", errno);
		else {
			int i;

			printk("Stub registers -\n");
			for(i = 0; i < ARRAY_SIZE(regs); i++)
				printk("\t%d - %lx\n", i, regs[i]);
		}
		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))
			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, 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, 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, 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
 * availabl. 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
	} } });
	n = os_write_file(fd, &mmop, sizeof(mmop));
	if(n != sizeof(mmop)){
		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
		} } });
		n = os_write_file(fd, &mmop, sizeof(mmop));
		if(n != sizeof(mmop))
			panic("map_stub_pages : /proc/mm map for data failed, "
			      "err = %d\n", -n);
	}
}

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