[mirror_qemu.git] / bsd-user / freebsd / os-syscall.c

/*
 *  BSD syscalls
 *
 *  Copyright (c) 2003-2008 Fabrice Bellard
 *  Copyright (c) 2013-2014 Stacey D. Son
 *
 *  This program is free software; you can redistribute it and/or modify
 *  it under the terms of the GNU General Public License as published by
 *  the Free Software Foundation; either version 2 of the License, or
 *  (at your option) any later version.
 *
 *  This program is distributed in the hope that it will be useful,
 *  but WITHOUT ANY WARRANTY; without even the implied warranty of
 *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 *  GNU General Public License for more details.
 *
 *  You should have received a copy of the GNU General Public License
 *  along with this program; if not, see <http://www.gnu.org/licenses/>.
 */

/*
 * We need the FreeBSD "legacy" definitions. Rust needs the FreeBSD 11 system
 * calls since it doesn't use libc at all, so we have to emulate that despite
 * FreeBSD 11 being EOL'd.
 */
#define _WANT_FREEBSD11_STAT
#define _WANT_FREEBSD11_STATFS
#define _WANT_FREEBSD11_DIRENT
#define _WANT_KERNEL_ERRNO
#define _WANT_SEMUN
#include "qemu/osdep.h"
#include "qemu/cutils.h"
#include "qemu/path.h"
#include <sys/syscall.h>
#include <sys/param.h>
#include <sys/sysctl.h>
#include <utime.h>

#include "qemu.h"
#include "signal-common.h"
#include "user/syscall-trace.h"

#include "bsd-file.h"
#include "bsd-proc.h"

/* I/O */
safe_syscall3(ssize_t, read, int, fd, void *, buf, size_t, nbytes);
safe_syscall4(ssize_t, pread, int, fd, void *, buf, size_t, nbytes, off_t,
    offset);
safe_syscall3(ssize_t, readv, int, fd, const struct iovec *, iov, int, iovcnt);
safe_syscall4(ssize_t, preadv, int, fd, const struct iovec *, iov, int, iovcnt,
    off_t, offset);

safe_syscall3(ssize_t, write, int, fd, void *, buf, size_t, nbytes);
safe_syscall4(ssize_t, pwrite, int, fd, void *, buf, size_t, nbytes, off_t,
    offset);
safe_syscall3(ssize_t, writev, int, fd, const struct iovec *, iov, int, iovcnt);
safe_syscall4(ssize_t, pwritev, int, fd, const struct iovec *, iov, int, iovcnt,
    off_t, offset);

void target_set_brk(abi_ulong new_brk)
{
}

/*
 * errno conversion.
 */
abi_long get_errno(abi_long ret)
{
    if (ret == -1) {
        return -host_to_target_errno(errno);
    } else {
        return ret;
    }
}

int host_to_target_errno(int err)
{
    /*
     * All the BSDs have the property that the error numbers are uniform across
     * all architectures for a given BSD, though they may vary between different
     * BSDs.
     */
    return err;
}

bool is_error(abi_long ret)
{
    return (abi_ulong)ret >= (abi_ulong)(-4096);
}

/*
 * Unlocks a iovec. Unlike unlock_iovec, it assumes the tvec array itself is
 * already locked from target_addr. It will be unlocked as well as all the iovec
 * elements.
 */
static void helper_unlock_iovec(struct target_iovec *target_vec,
                                abi_ulong target_addr, struct iovec *vec,
                                int count, int copy)
{
    for (int i = 0; i < count; i++) {
        abi_ulong base = tswapal(target_vec[i].iov_base);

        if (vec[i].iov_base) {
            unlock_user(vec[i].iov_base, base, copy ? vec[i].iov_len : 0);
        }
    }
    unlock_user(target_vec, target_addr, 0);
}

struct iovec *lock_iovec(int type, abi_ulong target_addr,
        int count, int copy)
{
    struct target_iovec *target_vec;
    struct iovec *vec;
    abi_ulong total_len, max_len;
    int i;
    int err = 0;

    if (count == 0) {
        errno = 0;
        return NULL;
    }
    if (count < 0 || count > IOV_MAX) {
        errno = EINVAL;
        return NULL;
    }

    vec = g_try_new0(struct iovec, count);
    if (vec == NULL) {
        errno = ENOMEM;
        return NULL;
    }

    target_vec = lock_user(VERIFY_READ, target_addr,
                           count * sizeof(struct target_iovec), 1);
    if (target_vec == NULL) {
        err = EFAULT;
        goto fail2;
    }

    max_len = 0x7fffffff & MIN(TARGET_PAGE_MASK, PAGE_MASK);
    total_len = 0;

    for (i = 0; i < count; i++) {
        abi_ulong base = tswapal(target_vec[i].iov_base);
        abi_long len = tswapal(target_vec[i].iov_len);

        if (len < 0) {
            err = EINVAL;
            goto fail;
        } else if (len == 0) {
            /* Zero length pointer is ignored. */
            vec[i].iov_base = 0;
        } else {
            vec[i].iov_base = lock_user(type, base, len, copy);
            /*
             * If the first buffer pointer is bad, this is a fault.  But
             * subsequent bad buffers will result in a partial write; this is
             * realized by filling the vector with null pointers and zero
             * lengths.
             */
            if (!vec[i].iov_base) {
                if (i == 0) {
                    err = EFAULT;
                    goto fail;
                } else {
                    /*
                     * Fail all the subsequent addresses, they are already
                     * zero'd.
                     */
                    goto out;
                }
            }
            if (len > max_len - total_len) {
                len = max_len - total_len;
            }
        }
        vec[i].iov_len = len;
        total_len += len;
    }
out:
    unlock_user(target_vec, target_addr, 0);
    return vec;

fail:
    helper_unlock_iovec(target_vec, target_addr, vec, i, copy);
fail2:
    g_free(vec);
    errno = err;
    return NULL;
}

void unlock_iovec(struct iovec *vec, abi_ulong target_addr,
        int count, int copy)
{
    struct target_iovec *target_vec;

    target_vec = lock_user(VERIFY_READ, target_addr,
                           count * sizeof(struct target_iovec), 1);
    if (target_vec) {
        helper_unlock_iovec(target_vec, target_addr, vec, count, copy);
    }

    g_free(vec);
}

/*
 * All errnos that freebsd_syscall() returns must be -TARGET_<errcode>.
 */
static abi_long freebsd_syscall(void *cpu_env, int num, abi_long arg1,
                                abi_long arg2, abi_long arg3, abi_long arg4,
                                abi_long arg5, abi_long arg6, abi_long arg7,
                                abi_long arg8)
{
    abi_long ret;

    switch (num) {
        /*
         * process system calls
         */
    case TARGET_FREEBSD_NR_exit: /* exit(2) */
        ret = do_bsd_exit(cpu_env, arg1);
        break;

        /*
         * File system calls.
         */
    case TARGET_FREEBSD_NR_read: /* read(2) */
        ret = do_bsd_read(arg1, arg2, arg3);
        break;

    case TARGET_FREEBSD_NR_pread: /* pread(2) */
        ret = do_bsd_pread(cpu_env, arg1, arg2, arg3, arg4, arg5, arg6);
        break;

    case TARGET_FREEBSD_NR_readv: /* readv(2) */
        ret = do_bsd_readv(arg1, arg2, arg3);
        break;

    case TARGET_FREEBSD_NR_preadv: /* preadv(2) */
        ret = do_bsd_preadv(cpu_env, arg1, arg2, arg3, arg4, arg5, arg6);

    case TARGET_FREEBSD_NR_write: /* write(2) */
        ret = do_bsd_write(arg1, arg2, arg3);
        break;

    case TARGET_FREEBSD_NR_pwrite: /* pwrite(2) */
        ret = do_bsd_pwrite(cpu_env, arg1, arg2, arg3, arg4, arg5, arg6);
        break;

    case TARGET_FREEBSD_NR_writev: /* writev(2) */
        ret = do_bsd_writev(arg1, arg2, arg3);
        break;

    case TARGET_FREEBSD_NR_pwritev: /* pwritev(2) */
        ret = do_bsd_pwritev(cpu_env, arg1, arg2, arg3, arg4, arg5, arg6);
        break;

    default:
        qemu_log_mask(LOG_UNIMP, "Unsupported syscall: %d\n", num);
        ret = -TARGET_ENOSYS;
        break;
    }

    return ret;
}

/*
 * do_freebsd_syscall() should always have a single exit point at the end so
 * that actions, such as logging of syscall results, can be performed. This
 * as a wrapper around freebsd_syscall() so that actually happens. Since
 * that is a singleton, modern compilers will inline it anyway...
 */
abi_long do_freebsd_syscall(void *cpu_env, int num, abi_long arg1,
                            abi_long arg2, abi_long arg3, abi_long arg4,
                            abi_long arg5, abi_long arg6, abi_long arg7,
                            abi_long arg8)
{
    CPUState *cpu = env_cpu(cpu_env);
    int ret;

    trace_guest_user_syscall(cpu, num, arg1, arg2, arg3, arg4, arg5, arg6, arg7, arg8);
    if (do_strace) {
        print_freebsd_syscall(num, arg1, arg2, arg3, arg4, arg5, arg6);
    }

    ret = freebsd_syscall(cpu_env, num, arg1, arg2, arg3, arg4, arg5, arg6,
                          arg7, arg8);
    if (do_strace) {
        print_freebsd_syscall_ret(num, ret);
    }
    trace_guest_user_syscall_ret(cpu, num, ret);

    return ret;
}

void syscall_init(void)
{
}
Commit	Line	Data
66eed099 WL	1	/*
	2	* BSD syscalls
	3	*
	4	* Copyright (c) 2003-2008 Fabrice Bellard
	5	* Copyright (c) 2013-2014 Stacey D. Son
	6	*
	7	* This program is free software; you can redistribute it and/or modify
	8	* it under the terms of the GNU General Public License as published by
	9	* the Free Software Foundation; either version 2 of the License, or
	10	* (at your option) any later version.
	11	*
	12	* This program is distributed in the hope that it will be useful,
	13	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	14	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	15	* GNU General Public License for more details.
	16	*
	17	* You should have received a copy of the GNU General Public License
	18	* along with this program; if not, see <http://www.gnu.org/licenses/>.
	19	*/
	20
	21	/*
	22	* We need the FreeBSD "legacy" definitions. Rust needs the FreeBSD 11 system
	23	* calls since it doesn't use libc at all, so we have to emulate that despite
	24	* FreeBSD 11 being EOL'd.
	25	*/
	26	#define _WANT_FREEBSD11_STAT
	27	#define _WANT_FREEBSD11_STATFS
	28	#define _WANT_FREEBSD11_DIRENT
	29	#define _WANT_KERNEL_ERRNO
	30	#define _WANT_SEMUN
	31	#include "qemu/osdep.h"
	32	#include "qemu/cutils.h"
	33	#include "qemu/path.h"
	34	#include <sys/syscall.h>
	35	#include <sys/param.h>
	36	#include <sys/sysctl.h>
	37	#include <utime.h>
	38
	39	#include "qemu.h"
66eed099 WL	40	#include "signal-common.h"
	41	#include "user/syscall-trace.h"
	42
c5c84d16	43	#include "bsd-file.h"
9554d330	44	#include "bsd-proc.h"
c5c84d16	45
80da1b00 WL	46	/* I/O */
	47	safe_syscall3(ssize_t, read, int, fd, void *, buf, size_t, nbytes);
	48	safe_syscall4(ssize_t, pread, int, fd, void *, buf, size_t, nbytes, off_t,
	49	offset);
	50	safe_syscall3(ssize_t, readv, int, fd, const struct iovec *, iov, int, iovcnt);
	51	safe_syscall4(ssize_t, preadv, int, fd, const struct iovec *, iov, int, iovcnt,
	52	off_t, offset);
	53
770d8aba WL	54	safe_syscall3(ssize_t, write, int, fd, void *, buf, size_t, nbytes);
	55	safe_syscall4(ssize_t, pwrite, int, fd, void *, buf, size_t, nbytes, off_t,
	56	offset);
	57	safe_syscall3(ssize_t, writev, int, fd, const struct iovec *, iov, int, iovcnt);
	58	safe_syscall4(ssize_t, pwritev, int, fd, const struct iovec *, iov, int, iovcnt,
	59	off_t, offset);
	60
66eed099 WL	61	void target_set_brk(abi_ulong new_brk)
	62	{
	63	}
	64
deeff83b WL	65	/*
	66	* errno conversion.
	67	*/
	68	abi_long get_errno(abi_long ret)
66eed099	69	{
deeff83b WL	70	if (ret == -1) {
	71	return -host_to_target_errno(errno);
	72	} else {
	73	return ret;
	74	}
	75	}
66eed099	76
deeff83b WL	77	int host_to_target_errno(int err)
	78	{
	79	/*
	80	* All the BSDs have the property that the error numbers are uniform across
	81	* all architectures for a given BSD, though they may vary between different
	82	* BSDs.
	83	*/
	84	return err;
	85	}
	86
	87	bool is_error(abi_long ret)
	88	{
66eed099 WL	89	return (abi_ulong)ret >= (abi_ulong)(-4096);
	90	}
	91
1ed771b2 WL	92	/*
	93	* Unlocks a iovec. Unlike unlock_iovec, it assumes the tvec array itself is
	94	* already locked from target_addr. It will be unlocked as well as all the iovec
	95	* elements.
	96	*/
	97	static void helper_unlock_iovec(struct target_iovec *target_vec,
	98	abi_ulong target_addr, struct iovec *vec,
	99	int count, int copy)
	100	{
	101	for (int i = 0; i < count; i++) {
	102	abi_ulong base = tswapal(target_vec[i].iov_base);
	103
	104	if (vec[i].iov_base) {
	105	unlock_user(vec[i].iov_base, base, copy ? vec[i].iov_len : 0);
	106	}
	107	}
	108	unlock_user(target_vec, target_addr, 0);
	109	}
	110
	111	struct iovec *lock_iovec(int type, abi_ulong target_addr,
	112	int count, int copy)
	113	{
	114	struct target_iovec *target_vec;
	115	struct iovec *vec;
	116	abi_ulong total_len, max_len;
	117	int i;
	118	int err = 0;
	119
	120	if (count == 0) {
	121	errno = 0;
	122	return NULL;
	123	}
	124	if (count < 0 \|\| count > IOV_MAX) {
	125	errno = EINVAL;
	126	return NULL;
	127	}
	128
	129	vec = g_try_new0(struct iovec, count);
	130	if (vec == NULL) {
	131	errno = ENOMEM;
	132	return NULL;
	133	}
	134
	135	target_vec = lock_user(VERIFY_READ, target_addr,
	136	count * sizeof(struct target_iovec), 1);
	137	if (target_vec == NULL) {
	138	err = EFAULT;
	139	goto fail2;
	140	}
	141
	142	max_len = 0x7fffffff & MIN(TARGET_PAGE_MASK, PAGE_MASK);
	143	total_len = 0;
	144
	145	for (i = 0; i < count; i++) {
	146	abi_ulong base = tswapal(target_vec[i].iov_base);
	147	abi_long len = tswapal(target_vec[i].iov_len);
	148
	149	if (len < 0) {
	150	err = EINVAL;
	151	goto fail;
	152	} else if (len == 0) {
	153	/* Zero length pointer is ignored. */
	154	vec[i].iov_base = 0;
	155	} else {
156	vec[i].iov_base = lock_user(type, base, len, copy);
157	/*
158	* If the first buffer pointer is bad, this is a fault. But
159	* subsequent bad buffers will result in a partial write; this is
160	* realized by filling the vector with null pointers and zero
161	* lengths.
162	*/
163	if (!vec[i].iov_base) {
164	if (i == 0) {
165	err = EFAULT;
166	goto fail;
167	} else {
168	/*
169	* Fail all the subsequent addresses, they are already
170	* zero'd.
171	*/
172	goto out;
173	}
174	}
175	if (len > max_len - total_len) {
176	len = max_len - total_len;
177	}
178	}
179	vec[i].iov_len = len;
180	total_len += len;
181	}
182	out:
183	unlock_user(target_vec, target_addr, 0);
184	return vec;
185
186	fail:
187	helper_unlock_iovec(target_vec, target_addr, vec, i, copy);
188	fail2:
189	g_free(vec);
190	errno = err;
191	return NULL;
192	}
193
883808d8 WL	194	void unlock_iovec(struct iovec *vec, abi_ulong target_addr,
	195	int count, int copy)
	196	{
	197	struct target_iovec *target_vec;
	198
	199	target_vec = lock_user(VERIFY_READ, target_addr,
	200	count * sizeof(struct target_iovec), 1);
	201	if (target_vec) {
	202	helper_unlock_iovec(target_vec, target_addr, vec, count, copy);
	203	}
	204
	205	g_free(vec);
	206	}
	207
66eed099	208	/*
db697887 WL	209	* All errnos that freebsd_syscall() returns must be -TARGET_<errcode>.
	210	*/
	211	static abi_long freebsd_syscall(void *cpu_env, int num, abi_long arg1,
	212	abi_long arg2, abi_long arg3, abi_long arg4,
	213	abi_long arg5, abi_long arg6, abi_long arg7,
	214	abi_long arg8)
	215	{
	216	abi_long ret;
	217
	218	switch (num) {
9554d330 WL	219	/*
	220	* process system calls
	221	*/
	222	case TARGET_FREEBSD_NR_exit: /* exit(2) */
	223	ret = do_bsd_exit(cpu_env, arg1);
	224	break;
80da1b00 WL	225
	226	/*
	227	* File system calls.
	228	*/
	229	case TARGET_FREEBSD_NR_read: /* read(2) */
	230	ret = do_bsd_read(arg1, arg2, arg3);
	231	break;
	232
	233	case TARGET_FREEBSD_NR_pread: /* pread(2) */
	234	ret = do_bsd_pread(cpu_env, arg1, arg2, arg3, arg4, arg5, arg6);
	235	break;
	236
	237	case TARGET_FREEBSD_NR_readv: /* readv(2) */
	238	ret = do_bsd_readv(arg1, arg2, arg3);
	239	break;
	240
	241	case TARGET_FREEBSD_NR_preadv: /* preadv(2) */
	242	ret = do_bsd_preadv(cpu_env, arg1, arg2, arg3, arg4, arg5, arg6);
770d8aba WL	243
	244	case TARGET_FREEBSD_NR_write: /* write(2) */
	245	ret = do_bsd_write(arg1, arg2, arg3);
	246	break;
	247
	248	case TARGET_FREEBSD_NR_pwrite: /* pwrite(2) */
	249	ret = do_bsd_pwrite(cpu_env, arg1, arg2, arg3, arg4, arg5, arg6);
	250	break;
	251
	252	case TARGET_FREEBSD_NR_writev: /* writev(2) */
	253	ret = do_bsd_writev(arg1, arg2, arg3);
	254	break;
	255
	256	case TARGET_FREEBSD_NR_pwritev: /* pwritev(2) */
	257	ret = do_bsd_pwritev(cpu_env, arg1, arg2, arg3, arg4, arg5, arg6);
80da1b00 WL	258	break;
80da1b00 WL	259
db697887 WL	260	default:
	261	qemu_log_mask(LOG_UNIMP, "Unsupported syscall: %d\n", num);
	262	ret = -TARGET_ENOSYS;
	263	break;
	264	}
	265
	266	return ret;
	267	}
	268
	269	/*
	270	* do_freebsd_syscall() should always have a single exit point at the end so
	271	* that actions, such as logging of syscall results, can be performed. This
	272	* as a wrapper around freebsd_syscall() so that actually happens. Since
	273	* that is a singleton, modern compilers will inline it anyway...
66eed099 WL	274	*/
	275	abi_long do_freebsd_syscall(void *cpu_env, int num, abi_long arg1,
	276	abi_long arg2, abi_long arg3, abi_long arg4,
	277	abi_long arg5, abi_long arg6, abi_long arg7,
	278	abi_long arg8)
	279	{
db697887 WL	280	CPUState *cpu = env_cpu(cpu_env);
	281	int ret;
	282
	283	trace_guest_user_syscall(cpu, num, arg1, arg2, arg3, arg4, arg5, arg6, arg7, arg8);
	284	if (do_strace) {
	285	print_freebsd_syscall(num, arg1, arg2, arg3, arg4, arg5, arg6);
	286	}
	287
	288	ret = freebsd_syscall(cpu_env, num, arg1, arg2, arg3, arg4, arg5, arg6,
	289	arg7, arg8);
	290	if (do_strace) {
	291	print_freebsd_syscall_ret(num, ret);
	292	}
	293	trace_guest_user_syscall_ret(cpu, num, ret);
	294
	295	return ret;
66eed099 WL	296	}
	297
	298	void syscall_init(void)
	299	{
	300	}