2 files changed, 2024 insertions, 0 deletions

diff --git a/sys/compat/linux/linux_misc.c b/sys/compat/linux/linux_misc.c
new file mode 100644
index 000000000000..8b326582b11f
--- /dev/null
+++ b/sys/compat/linux/linux_misc.c
@@ -0,0 +1,1120 @@
+/*-
+ * Copyright (c) 1994-1995 Søren Schmidt
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ * 1. Redistributions of source code must retain the above copyright
+ *    notice, this list of conditions and the following disclaimer
+ *    in this position and unchanged.
+ * 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. The name of the author may not be used to endorse or promote products
+ *    derived from this software withough specific prior written permission
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``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 AUTHOR 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.
+ *
+ * $FreeBSD$
+ */
+
+#include "opt_compat.h"
+
+#include <sys/param.h>
+#include <sys/systm.h>
+#include <sys/sysproto.h>
+#include <sys/kernel.h>
+#include <sys/mman.h>
+#include <sys/proc.h>
+#include <sys/fcntl.h>
+#include <sys/imgact_aout.h>
+#include <sys/mount.h>
+#include <sys/namei.h>
+#include <sys/resourcevar.h>
+#include <sys/stat.h>
+#include <sys/sysctl.h>
+#include <sys/unistd.h>
+#include <sys/vnode.h>
+#include <sys/wait.h>
+#include <sys/time.h>
+#include <sys/signalvar.h>
+
+#include <vm/vm.h>
+#include <vm/pmap.h>
+#include <vm/vm_kern.h>
+#include <vm/vm_map.h>
+#include <vm/vm_extern.h>
+
+#include <machine/frame.h>
+#include <machine/limits.h>
+#include <machine/psl.h>
+#include <machine/sysarch.h>
+#include <machine/segments.h>
+
+#include <posix4/sched.h>
+
+#include <machine/../linux/linux.h>
+#include <machine/../linux/linux_proto.h>
+#include <compat/linux/linux_mib.h>
+#include <compat/linux/linux_util.h>
+
+#define BSD_TO_LINUX_SIGNAL(sig)	\
+	(((sig) <= LINUX_SIGTBLSZ) ? bsd_to_linux_signal[_SIG_IDX(sig)] : sig)
+
+struct linux_rlimit {
+	unsigned long rlim_cur;
+	unsigned long rlim_max;
+};
+
+static unsigned int linux_to_bsd_resource[LINUX_RLIM_NLIMITS] =
+{ RLIMIT_CPU, RLIMIT_FSIZE, RLIMIT_DATA, RLIMIT_STACK,
+  RLIMIT_CORE, RLIMIT_RSS, RLIMIT_NPROC, RLIMIT_NOFILE,
+  RLIMIT_MEMLOCK, -1
+};
+
+int
+linux_alarm(struct proc *p, struct linux_alarm_args *args)
+{
+    struct itimerval it, old_it;
+    struct timeval tv;
+    int s;
+
+#ifdef DEBUG
+    printf("Linux-emul(%ld): alarm(%u)\n", (long)p->p_pid, args->secs);
+#endif
+    if (args->secs > 100000000)
+	return EINVAL;
+    it.it_value.tv_sec = (long)args->secs;
+    it.it_value.tv_usec = 0;
+    it.it_interval.tv_sec = 0;
+    it.it_interval.tv_usec = 0;
+    s = splsoftclock();
+    old_it = p->p_realtimer;
+    getmicrouptime(&tv);
+    if (timevalisset(&old_it.it_value))
+	untimeout(realitexpire, (caddr_t)p, p->p_ithandle);
+    if (it.it_value.tv_sec != 0) {
+	p->p_ithandle = timeout(realitexpire, (caddr_t)p, tvtohz(&it.it_value));
+	timevaladd(&it.it_value, &tv);
+    }
+    p->p_realtimer = it;
+    splx(s);
+    if (timevalcmp(&old_it.it_value, &tv, >)) {
+	timevalsub(&old_it.it_value, &tv);
+	if (old_it.it_value.tv_usec != 0)
+	    old_it.it_value.tv_sec++;
+	p->p_retval[0] = old_it.it_value.tv_sec;
+    }
+    return 0;
+}
+
+int
+linux_brk(struct proc *p, struct linux_brk_args *args)
+{
+#if 0
+    struct vmspace *vm = p->p_vmspace;
+    vm_offset_t new, old;
+    int error;
+
+    if ((vm_offset_t)args->dsend < (vm_offset_t)vm->vm_daddr)
+	return EINVAL;
+    if (((caddr_t)args->dsend - (caddr_t)vm->vm_daddr)
+	> p->p_rlimit[RLIMIT_DATA].rlim_cur)
+	return ENOMEM;
+
+    old = round_page((vm_offset_t)vm->vm_daddr) + ctob(vm->vm_dsize);
+    new = round_page((vm_offset_t)args->dsend);
+    p->p_retval[0] = old;
+    if ((new-old) > 0) {
+	if (swap_pager_full)
+	    return ENOMEM;
+	error = vm_map_find(&vm->vm_map, NULL, 0, &old, (new-old), FALSE,
+			VM_PROT_ALL, VM_PROT_ALL, 0);
+	if (error)
+	    return error;
+	vm->vm_dsize += btoc((new-old));
+	p->p_retval[0] = (int)(vm->vm_daddr + ctob(vm->vm_dsize));
+    }
+    return 0;
+#else
+    struct vmspace *vm = p->p_vmspace;
+    vm_offset_t new, old;
+    struct obreak_args /* {
+	char * nsize;
+    } */ tmp;
+
+#ifdef DEBUG
+    printf("Linux-emul(%ld): brk(%p)\n", (long)p->p_pid, (void *)args->dsend);
+#endif
+    old = (vm_offset_t)vm->vm_daddr + ctob(vm->vm_dsize);
+    new = (vm_offset_t)args->dsend;
+    tmp.nsize = (char *) new;
+    if (((caddr_t)new > vm->vm_daddr) && !obreak(p, &tmp))
+	p->p_retval[0] = (int)new;
+    else
+	p->p_retval[0] = (int)old;
+
+    return 0;
+#endif
+}
+
+int
+linux_uselib(struct proc *p, struct linux_uselib_args *args)
+{
+    struct nameidata ni;
+    struct vnode *vp;
+    struct exec *a_out;
+    struct vattr attr;
+    vm_offset_t vmaddr;
+    unsigned long file_offset;
+    vm_offset_t buffer;
+    unsigned long bss_size;
+    int error;
+    caddr_t sg;
+    int locked;
+
+    sg = stackgap_init();
+    CHECKALTEXIST(p, &sg, args->library);
+
+#ifdef DEBUG
+    printf("Linux-emul(%ld): uselib(%s)\n", (long)p->p_pid, args->library);
+#endif
+
+    a_out = NULL;
+    locked = 0;
+    vp = NULL;
+
+    NDINIT(&ni, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE, args->library, p);
+    error = namei(&ni);
+    if (error)
+	goto cleanup;
+
+    vp = ni.ni_vp;
+    /*
+     * XXX This looks like a bogus check - a LOCKLEAF namei should not succeed
+     * without returning a vnode.
+     */
+    if (vp == NULL) {
+	error = ENOEXEC;	/* ?? */
+	goto cleanup;
+    }
+    NDFREE(&ni, NDF_ONLY_PNBUF);
+
+    /*
+     * From here on down, we have a locked vnode that must be unlocked.
+     */
+    locked++;
+
+    /*
+     * Writable?
+     */
+    if (vp->v_writecount) {
+	error = ETXTBSY;
+	goto cleanup;
+    }
+
+    /*
+     * Executable?
+     */
+    error = VOP_GETATTR(vp, &attr, p->p_ucred, p);
+    if (error)
+	goto cleanup;
+
+    if ((vp->v_mount->mnt_flag & MNT_NOEXEC) ||
+	((attr.va_mode & 0111) == 0) ||
+	(attr.va_type != VREG)) {
+	    error = ENOEXEC;
+	    goto cleanup;
+    }
+
+    /*
+     * Sensible size?
+     */
+    if (attr.va_size == 0) {
+	error = ENOEXEC;
+	goto cleanup;
+    }
+
+    /*
+     * Can we access it?
+     */
+    error = VOP_ACCESS(vp, VEXEC, p->p_ucred, p);
+    if (error)
+	goto cleanup;
+
+    error = VOP_OPEN(vp, FREAD, p->p_ucred, p);
+    if (error)
+	goto cleanup;
+
+    /*
+     * Lock no longer needed
+     */
+    VOP_UNLOCK(vp, 0, p);
+    locked = 0;
+
+    /*
+     * Pull in executable header into kernel_map
+     */
+    error = vm_mmap(kernel_map, (vm_offset_t *)&a_out, PAGE_SIZE,
+	    	    VM_PROT_READ, VM_PROT_READ, 0, (caddr_t)vp, 0);
+    if (error)
+	goto cleanup;
+
+    /*
+     * Is it a Linux binary ?
+     */
+    if (((a_out->a_magic >> 16) & 0xff) != 0x64) {
+	error = ENOEXEC;
+	goto cleanup;
+    }
+
+    /* While we are here, we should REALLY do some more checks */
+
+    /*
+     * Set file/virtual offset based on a.out variant.
+     */
+    switch ((int)(a_out->a_magic & 0xffff)) {
+    case 0413:	/* ZMAGIC */
+	file_offset = 1024;
+	break;
+    case 0314:	/* QMAGIC */
+	file_offset = 0;
+	break;
+    default:
+	error = ENOEXEC;
+	goto cleanup;
+    }
+
+    bss_size = round_page(a_out->a_bss);
+
+    /*
+     * Check various fields in header for validity/bounds.
+     */
+    if (a_out->a_text & PAGE_MASK || a_out->a_data & PAGE_MASK) {
+	error = ENOEXEC;
+	goto cleanup;
+    }
+
+    /* text + data can't exceed file size */
+    if (a_out->a_data + a_out->a_text > attr.va_size) {
+	error = EFAULT;
+	goto cleanup;
+    }
+
+    /*
+     * text/data/bss must not exceed limits
+     * XXX: this is not complete. it should check current usage PLUS
+     * the resources needed by this library.
+     */
+    if (a_out->a_text > MAXTSIZ ||
+	a_out->a_data + bss_size > p->p_rlimit[RLIMIT_DATA].rlim_cur) {
+	error = ENOMEM;
+	goto cleanup;
+    }
+
+    /*
+     * prevent more writers
+     */
+    vp->v_flag |= VTEXT;
+
+    /*
+     * Check if file_offset page aligned,.
+     * Currently we cannot handle misalinged file offsets,
+     * and so we read in the entire image (what a waste).
+     */
+    if (file_offset & PAGE_MASK) {
+#ifdef DEBUG
+printf("uselib: Non page aligned binary %lu\n", file_offset);
+#endif
+	/*
+	 * Map text+data read/write/execute
+	 */
+
+	/* a_entry is the load address and is page aligned */
+	vmaddr = trunc_page(a_out->a_entry);
+
+	/* get anon user mapping, read+write+execute */
+	error = vm_map_find(&p->p_vmspace->vm_map, NULL, 0, &vmaddr,
+		    	    a_out->a_text + a_out->a_data, FALSE,
+			    VM_PROT_ALL, VM_PROT_ALL, 0);
+	if (error)
+	    goto cleanup;
+
+	/* map file into kernel_map */
+	error = vm_mmap(kernel_map, &buffer,
+			round_page(a_out->a_text + a_out->a_data + file_offset),
+		   	VM_PROT_READ, VM_PROT_READ, 0,
+			(caddr_t)vp, trunc_page(file_offset));
+	if (error)
+	    goto cleanup;
+
+	/* copy from kernel VM space to user space */
+	error = copyout((caddr_t)(void *)(uintptr_t)(buffer + file_offset),
+			(caddr_t)vmaddr, a_out->a_text + a_out->a_data);
+
+	/* release temporary kernel space */
+	vm_map_remove(kernel_map, buffer,
+		      buffer + round_page(a_out->a_text + a_out->a_data + file_offset));
+
+	if (error)
+	    goto cleanup;
+    }
+    else {
+#ifdef DEBUG
+printf("uselib: Page aligned binary %lu\n", file_offset);
+#endif
+	/*
+	 * for QMAGIC, a_entry is 20 bytes beyond the load address
+	 * to skip the executable header
+	 */
+	vmaddr = trunc_page(a_out->a_entry);
+
+	/*
+	 * Map it all into the process's space as a single copy-on-write
+	 * "data" segment.
+	 */
+	error = vm_mmap(&p->p_vmspace->vm_map, &vmaddr,
+		   	a_out->a_text + a_out->a_data,
+			VM_PROT_ALL, VM_PROT_ALL, MAP_PRIVATE | MAP_FIXED,
+			(caddr_t)vp, file_offset);
+	if (error)
+	    goto cleanup;
+    }
+#ifdef DEBUG
+printf("mem=%08x = %08x %08x\n", vmaddr, ((int*)vmaddr)[0], ((int*)vmaddr)[1]);
+#endif
+    if (bss_size != 0) {
+        /*
+	 * Calculate BSS start address
+	 */
+	vmaddr = trunc_page(a_out->a_entry) + a_out->a_text + a_out->a_data;
+
+	/*
+	 * allocate some 'anon' space
+	 */
+	error = vm_map_find(&p->p_vmspace->vm_map, NULL, 0, &vmaddr,
+			    bss_size, FALSE,
+			    VM_PROT_ALL, VM_PROT_ALL, 0);
+	if (error)
+	    goto cleanup;
+    }
+
+cleanup:
+    /*
+     * Unlock vnode if needed
+     */
+    if (locked)
+	VOP_UNLOCK(vp, 0, p);
+
+    /*
+     * Release the kernel mapping.
+     */
+    if (a_out)
+	vm_map_remove(kernel_map, (vm_offset_t)a_out, (vm_offset_t)a_out + PAGE_SIZE);
+
+    return error;
+}
+
+int
+linux_newselect(struct proc *p, struct linux_newselect_args *args)
+{
+    struct select_args bsa;
+    struct timeval tv0, tv1, utv, *tvp;
+    caddr_t sg;
+    int error;
+
+#ifdef DEBUG
+    printf("Linux-emul(%ld): newselect(%d, %p, %p, %p, %p)\n",
+  	(long)p->p_pid, args->nfds, (void *)args->readfds,
+	(void *)args->writefds, (void *)args->exceptfds,
+	(void *)args->timeout);
+#endif
+    error = 0;
+    bsa.nd = args->nfds;
+    bsa.in = args->readfds;
+    bsa.ou = args->writefds;
+    bsa.ex = args->exceptfds;
+    bsa.tv = args->timeout;
+
+    /*
+     * Store current time for computation of the amount of
+     * time left.
+     */
+    if (args->timeout) {
+	if ((error = copyin(args->timeout, &utv, sizeof(utv))))
+	    goto select_out;
+#ifdef DEBUG
+	printf("Linux-emul(%ld): incoming timeout (%ld/%ld)\n",
+	    (long)p->p_pid, utv.tv_sec, utv.tv_usec);
+#endif
+	if (itimerfix(&utv)) {
+	    /*
+	     * The timeval was invalid.  Convert it to something
+	     * valid that will act as it does under Linux.
+	     */
+	    sg = stackgap_init();
+	    tvp = stackgap_alloc(&sg, sizeof(utv));
+	    utv.tv_sec += utv.tv_usec / 1000000;
+	    utv.tv_usec %= 1000000;
+	    if (utv.tv_usec < 0) {
+		utv.tv_sec -= 1;
+		utv.tv_usec += 1000000;
+	    }
+	    if (utv.tv_sec < 0)
+		timevalclear(&utv);
+	    if ((error = copyout(&utv, tvp, sizeof(utv))))
+		goto select_out;
+	    bsa.tv = tvp;
+	}
+	microtime(&tv0);
+    }
+
+    error = select(p, &bsa);
+#ifdef DEBUG
+    printf("Linux-emul(%ld): real select returns %d\n", (long)p->p_pid, error);
+#endif
+
+    if (error) {
+	/*
+	 * See fs/select.c in the Linux kernel.  Without this,
+	 * Maelstrom doesn't work.
+	 */
+	if (error == ERESTART)
+	    error = EINTR;
+	goto select_out;
+    }
+
+    if (args->timeout) {
+	if (p->p_retval[0]) {
+	    /*
+	     * Compute how much time was left of the timeout,
+	     * by subtracting the current time and the time
+	     * before we started the call, and subtracting
+	     * that result from the user-supplied value.
+	     */
+	    microtime(&tv1);
+	    timevalsub(&tv1, &tv0);
+	    timevalsub(&utv, &tv1);
+	    if (utv.tv_sec < 0)
+		timevalclear(&utv);
+	} else
+	    timevalclear(&utv);
+#ifdef DEBUG
+	printf("Linux-emul(%ld): outgoing timeout (%ld/%ld)\n",
+	    (long)p->p_pid, utv.tv_sec, utv.tv_usec);
+#endif
+	if ((error = copyout(&utv, args->timeout, sizeof(utv))))
+	    goto select_out;
+    }
+
+select_out:
+#ifdef DEBUG
+    printf("Linux-emul(%ld): newselect_out -> %d\n", (long)p->p_pid, error);
+#endif
+    return error;
+}
+
+int
+linux_getpgid(struct proc *p, struct linux_getpgid_args *args)
+{
+    struct proc *curp;
+
+#ifdef DEBUG
+    printf("Linux-emul(%ld): getpgid(%d)\n", (long)p->p_pid, args->pid);
+#endif
+    if (args->pid != p->p_pid) {
+	if (!(curp = pfind(args->pid)))
+	    return ESRCH;
+    }
+    else
+	curp = p;
+    p->p_retval[0] = curp->p_pgid;
+    return 0;
+}
+
+int     
+linux_mremap(struct proc *p, struct linux_mremap_args *args)
+{
+	struct munmap_args /* {
+		void *addr;
+		size_t len;
+	} */ bsd_args; 
+	int error = 0;
+ 
+#ifdef DEBUG
+	printf("Linux-emul(%ld): mremap(%p, %08x, %08x, %08x)\n",
+	    (long)p->p_pid, (void *)args->addr, args->old_len, args->new_len,
+	    args->flags);
+#endif
+	args->new_len = round_page(args->new_len);
+	args->old_len = round_page(args->old_len);
+
+	if (args->new_len > args->old_len) {
+		p->p_retval[0] = 0;
+		return ENOMEM;
+	}
+
+	if (args->new_len < args->old_len) {
+		bsd_args.addr = args->addr + args->new_len;
+		bsd_args.len = args->old_len - args->new_len;
+		error = munmap(p, &bsd_args);
+	}
+
+	p->p_retval[0] = error ? 0 : (int)args->addr;
+	return error;
+}
+
+int
+linux_msync(struct proc *p, struct linux_msync_args *args)
+{
+	struct msync_args bsd_args;
+
+	bsd_args.addr = args->addr;
+	bsd_args.len = args->len;
+	bsd_args.flags = 0;	/* XXX ignore */
+
+	return msync(p, &bsd_args);
+}
+
+int
+linux_time(struct proc *p, struct linux_time_args *args)
+{
+    struct timeval tv;
+    linux_time_t tm;
+    int error;
+
+#ifdef DEBUG
+    printf("Linux-emul(%ld): time(*)\n", (long)p->p_pid);
+#endif
+    microtime(&tv);
+    tm = tv.tv_sec;
+    if (args->tm && (error = copyout(&tm, args->tm, sizeof(linux_time_t))))
+	return error;
+    p->p_retval[0] = tm;
+    return 0;
+}
+
+struct linux_times_argv {
+    long    tms_utime;
+    long    tms_stime;
+    long    tms_cutime;
+    long    tms_cstime;
+};
+
+#define CLK_TCK 100	/* Linux uses 100 */
+#define CONVTCK(r)	(r.tv_sec * CLK_TCK + r.tv_usec / (1000000 / CLK_TCK))
+
+int
+linux_times(struct proc *p, struct linux_times_args *args)
+{
+    struct timeval tv;
+    struct linux_times_argv tms;
+    struct rusage ru;
+    int error;
+
+#ifdef DEBUG
+    printf("Linux-emul(%ld): times(*)\n", (long)p->p_pid);
+#endif
+    calcru(p, &ru.ru_utime, &ru.ru_stime, NULL);
+
+    tms.tms_utime = CONVTCK(ru.ru_utime);
+    tms.tms_stime = CONVTCK(ru.ru_stime);
+
+    tms.tms_cutime = CONVTCK(p->p_stats->p_cru.ru_utime);
+    tms.tms_cstime = CONVTCK(p->p_stats->p_cru.ru_stime);
+
+    if ((error = copyout((caddr_t)&tms, (caddr_t)args->buf,
+	    	    sizeof(struct linux_times_argv))))
+	return error;
+
+    microuptime(&tv);
+    p->p_retval[0] = (int)CONVTCK(tv);
+    return 0;
+}
+
+int
+linux_newuname(struct proc *p, struct linux_newuname_args *args)
+{
+	struct linux_new_utsname utsname;
+	char *osrelease, *osname;
+
+#ifdef DEBUG
+	printf("Linux-emul(%ld): newuname(*)\n", (long)p->p_pid);
+#endif
+
+	osname = linux_get_osname(p);
+	osrelease = linux_get_osrelease(p);
+
+	bzero(&utsname, sizeof(struct linux_new_utsname));
+	strncpy(utsname.sysname, osname, LINUX_MAX_UTSNAME-1);
+	strncpy(utsname.nodename, hostname, LINUX_MAX_UTSNAME-1);
+	strncpy(utsname.release, osrelease, LINUX_MAX_UTSNAME-1);
+	strncpy(utsname.version, version, LINUX_MAX_UTSNAME-1);
+	strncpy(utsname.machine, machine, LINUX_MAX_UTSNAME-1);
+	strncpy(utsname.domainname, domainname, LINUX_MAX_UTSNAME-1);
+
+	return (copyout((caddr_t)&utsname, (caddr_t)args->buf,
+			sizeof(struct linux_new_utsname)));
+}
+
+struct linux_utimbuf {
+	linux_time_t l_actime;
+	linux_time_t l_modtime;
+};
+
+int
+linux_utime(struct proc *p, struct linux_utime_args *args)
+{
+    struct utimes_args /* {
+	char	*path;
+	struct	timeval *tptr;
+    } */ bsdutimes;
+    struct timeval tv[2], *tvp;
+    struct linux_utimbuf lut;
+    int error;
+    caddr_t sg;
+
+    sg = stackgap_init();
+    CHECKALTEXIST(p, &sg, args->fname);
+
+#ifdef DEBUG
+    printf("Linux-emul(%ld): utime(%s, *)\n", (long)p->p_pid, args->fname);
+#endif
+    if (args->times) {
+	if ((error = copyin(args->times, &lut, sizeof lut)))
+	    return error;
+	tv[0].tv_sec = lut.l_actime;
+	tv[0].tv_usec = 0;
+	tv[1].tv_sec = lut.l_modtime;
+	tv[1].tv_usec = 0;
+	/* so that utimes can copyin */
+	tvp = (struct timeval *)stackgap_alloc(&sg, sizeof(tv));
+	if (tvp == NULL)
+		return (ENAMETOOLONG);
+	if ((error = copyout(tv, tvp, sizeof(tv))))
+	    return error;
+	bsdutimes.tptr = tvp;
+    } else
+	bsdutimes.tptr = NULL;
+
+    bsdutimes.path = args->fname;
+    return utimes(p, &bsdutimes);
+}
+
+#define __WCLONE 0x80000000
+
+int
+linux_waitpid(struct proc *p, struct linux_waitpid_args *args)
+{
+    struct wait_args /* {
+	int pid;
+	int *status;
+	int options;
+	struct	rusage *rusage;
+    } */ tmp;
+    int error, tmpstat;
+
+#ifdef DEBUG
+    printf("Linux-emul(%ld): waitpid(%d, %p, %d)\n",
+	(long)p->p_pid, args->pid, (void *)args->status, args->options);
+#endif
+    tmp.pid = args->pid;
+    tmp.status = args->status;
+    tmp.options = (args->options & (WNOHANG | WUNTRACED));
+    /* WLINUXCLONE should be equal to __WCLONE, but we make sure */
+    if (args->options & __WCLONE)
+	tmp.options |= WLINUXCLONE;
+    tmp.rusage = NULL;
+
+    if ((error = wait4(p, &tmp)) != 0)
+	return error;
+
+    if (args->status) {
+	if ((error = copyin(args->status, &tmpstat, sizeof(int))) != 0)
+	    return error;
+	tmpstat &= 0xffff;
+	if (WIFSIGNALED(tmpstat))
+	    tmpstat = (tmpstat & 0xffffff80) |
+		      BSD_TO_LINUX_SIGNAL(WTERMSIG(tmpstat));
+	else if (WIFSTOPPED(tmpstat))
+	    tmpstat = (tmpstat & 0xffff00ff) |
+		      (BSD_TO_LINUX_SIGNAL(WSTOPSIG(tmpstat)) << 8);
+	return copyout(&tmpstat, args->status, sizeof(int));
+    } else
+	return 0;
+}
+
+int
+linux_wait4(struct proc *p, struct linux_wait4_args *args)
+{
+    struct wait_args /* {
+	int pid;
+	int *status;
+	int options;
+	struct	rusage *rusage;
+    } */ tmp;
+    int error, tmpstat;
+
+#ifdef DEBUG
+    printf("Linux-emul(%ld): wait4(%d, %p, %d, %p)\n",
+	(long)p->p_pid, args->pid, (void *)args->status, args->options,
+	(void *)args->rusage);
+#endif
+    tmp.pid = args->pid;
+    tmp.status = args->status;
+    tmp.options = (args->options & (WNOHANG | WUNTRACED));
+    /* WLINUXCLONE should be equal to __WCLONE, but we make sure */
+    if (args->options & __WCLONE)
+	tmp.options |= WLINUXCLONE;
+    tmp.rusage = args->rusage;
+
+    if ((error = wait4(p, &tmp)) != 0)
+	return error;
+
+    SIGDELSET(p->p_siglist, SIGCHLD);
+
+    if (args->status) {
+	if ((error = copyin(args->status, &tmpstat, sizeof(int))) != 0)
+	    return error;
+	tmpstat &= 0xffff;
+	if (WIFSIGNALED(tmpstat))
+	    tmpstat = (tmpstat & 0xffffff80) |
+		  BSD_TO_LINUX_SIGNAL(WTERMSIG(tmpstat));
+	else if (WIFSTOPPED(tmpstat))
+	    tmpstat = (tmpstat & 0xffff00ff) |
+		  (BSD_TO_LINUX_SIGNAL(WSTOPSIG(tmpstat)) << 8);
+	return copyout(&tmpstat, args->status, sizeof(int));
+    } else
+	return 0;
+}
+
+int
+linux_mknod(struct proc *p, struct linux_mknod_args *args)
+{
+	caddr_t sg;
+	struct mknod_args bsd_mknod;
+	struct mkfifo_args bsd_mkfifo;
+
+	sg = stackgap_init();
+
+	CHECKALTCREAT(p, &sg, args->path);
+
+#ifdef DEBUG
+	printf("Linux-emul(%ld): mknod(%s, %d, %d)\n",
+	   (long)p->p_pid, args->path, args->mode, args->dev);
+#endif
+
+	if (args->mode & S_IFIFO) {
+		bsd_mkfifo.path = args->path;
+		bsd_mkfifo.mode = args->mode;
+		return mkfifo(p, &bsd_mkfifo);
+	} else {
+		bsd_mknod.path = args->path;
+		bsd_mknod.mode = args->mode;
+		bsd_mknod.dev = args->dev;
+		return mknod(p, &bsd_mknod);
+	}
+}
+
+/*
+ * UGH! This is just about the dumbest idea I've ever heard!!
+ */
+int
+linux_personality(struct proc *p, struct linux_personality_args *args)
+{
+#ifdef DEBUG
+	printf("Linux-emul(%ld): personality(%d)\n",
+	   (long)p->p_pid, args->per);
+#endif
+	if (args->per != 0)
+		return EINVAL;
+
+	/* Yes Jim, it's still a Linux... */
+	p->p_retval[0] = 0;
+	return 0;
+}
+
+/*
+ * Wrappers for get/setitimer for debugging..
+ */
+int
+linux_setitimer(struct proc *p, struct linux_setitimer_args *args)
+{
+	struct setitimer_args bsa;
+	struct itimerval foo;
+	int error;
+
+#ifdef DEBUG
+	printf("Linux-emul(%ld): setitimer(%p, %p)\n",
+	    (long)p->p_pid, (void *)args->itv, (void *)args->oitv);
+#endif
+	bsa.which = args->which;
+	bsa.itv = args->itv;
+	bsa.oitv = args->oitv;
+	if (args->itv) {
+	    if ((error = copyin((caddr_t)args->itv, (caddr_t)&foo,
+			sizeof(foo))))
+		return error;
+#ifdef DEBUG
+	    printf("setitimer: value: sec: %ld, usec: %ld\n",
+		foo.it_value.tv_sec, foo.it_value.tv_usec);
+	    printf("setitimer: interval: sec: %ld, usec: %ld\n",
+		foo.it_interval.tv_sec, foo.it_interval.tv_usec);
+#endif
+	}
+	return setitimer(p, &bsa);
+}
+
+int
+linux_getitimer(struct proc *p, struct linux_getitimer_args *args)
+{
+	struct getitimer_args bsa;
+#ifdef DEBUG
+	printf("Linux-emul(%ld): getitimer(%p)\n",
+	    (long)p->p_pid, (void *)args->itv);
+#endif
+	bsa.which = args->which;
+	bsa.itv = args->itv;
+	return getitimer(p, &bsa);
+}
+
+int
+linux_nice(struct proc *p, struct linux_nice_args *args)
+{
+	struct setpriority_args	bsd_args;
+
+	bsd_args.which = PRIO_PROCESS;
+	bsd_args.who = 0;	/* current process */
+	bsd_args.prio = args->inc;
+	return setpriority(p, &bsd_args);
+}
+
+int
+linux_setgroups(p, uap)
+	struct proc *p;
+	struct linux_setgroups_args *uap;
+{
+	struct pcred *pc;
+	linux_gid_t linux_gidset[NGROUPS];
+	gid_t *bsd_gidset;
+	int ngrp, error;
+
+	pc = p->p_cred;
+	ngrp = uap->gidsetsize;
+
+	/*
+	 * cr_groups[0] holds egid. Setting the whole set from
+	 * the supplied set will cause egid to be changed too.
+	 * Keep cr_groups[0] unchanged to prevent that.
+	 */
+
+	if ((error = suser(p)) != 0)
+		return (error);
+
+	if (ngrp >= NGROUPS)
+		return (EINVAL);
+
+	pc->pc_ucred = crcopy(pc->pc_ucred);
+	if (ngrp > 0) {
+		error = copyin((caddr_t)uap->gidset, (caddr_t)linux_gidset,
+			       ngrp * sizeof(linux_gid_t));
+		if (error)
+			return (error);
+
+		pc->pc_ucred->cr_ngroups = ngrp + 1;
+
+		bsd_gidset = pc->pc_ucred->cr_groups;
+		ngrp--;
+		while (ngrp >= 0) {
+			bsd_gidset[ngrp + 1] = linux_gidset[ngrp];
+			ngrp--;
+		}
+	}
+	else
+		pc->pc_ucred->cr_ngroups = 1;
+
+	setsugid(p);
+	return (0);
+}
+
+int
+linux_getgroups(p, uap)
+	struct proc *p;
+	struct linux_getgroups_args *uap;
+{
+	struct pcred *pc;
+	linux_gid_t linux_gidset[NGROUPS];
+	gid_t *bsd_gidset;
+	int bsd_gidsetsz, ngrp, error;
+
+	pc = p->p_cred;
+	bsd_gidset = pc->pc_ucred->cr_groups;
+	bsd_gidsetsz = pc->pc_ucred->cr_ngroups - 1;
+
+	/*
+	 * cr_groups[0] holds egid. Returning the whole set
+	 * here will cause a duplicate. Exclude cr_groups[0]
+	 * to prevent that.
+	 */
+
+	if ((ngrp = uap->gidsetsize) == 0) {
+		p->p_retval[0] = bsd_gidsetsz;
+		return (0);
+	}
+
+	if (ngrp < bsd_gidsetsz)
+		return (EINVAL);
+
+	ngrp = 0;
+	while (ngrp < bsd_gidsetsz) {
+		linux_gidset[ngrp] = bsd_gidset[ngrp + 1];
+		ngrp++;
+	}
+
+	if ((error = copyout((caddr_t)linux_gidset, (caddr_t)uap->gidset,
+	    ngrp * sizeof(linux_gid_t))))
+		return (error);
+
+	p->p_retval[0] = ngrp;
+	return (0);
+}
+
+int
+linux_setrlimit(p, uap)
+	struct proc *p;
+	struct linux_setrlimit_args *uap;
+{
+	struct __setrlimit_args bsd;
+	struct linux_rlimit rlim;
+	int error;
+	caddr_t sg = stackgap_init();
+
+#ifdef DEBUG
+	printf("Linux-emul(%ld): setrlimit(%d, %p)\n", (long)p->p_pid,
+	    uap->resource, (void *)uap->rlim);
+#endif
+
+	if (uap->resource >= LINUX_RLIM_NLIMITS)
+		return (EINVAL);
+
+	bsd.which = linux_to_bsd_resource[uap->resource];
+	if (bsd.which == -1)
+		return (EINVAL);
+
+	error = copyin(uap->rlim, &rlim, sizeof(rlim));
+	if (error)
+		return (error);
+
+	bsd.rlp = stackgap_alloc(&sg, sizeof(struct rlimit));
+	bsd.rlp->rlim_cur = (rlim_t)rlim.rlim_cur;
+	bsd.rlp->rlim_max = (rlim_t)rlim.rlim_max;
+	return (setrlimit(p, &bsd));
+}
+
+int
+linux_getrlimit(p, uap)
+	struct proc *p;
+	struct linux_getrlimit_args *uap;
+{
+	struct __getrlimit_args bsd;
+	struct linux_rlimit rlim;
+	int error;
+	caddr_t sg = stackgap_init();
+
+#ifdef DEBUG
+	printf("Linux-emul(%ld): getrlimit(%d, %p)\n", (long)p->p_pid,
+	    uap->resource, (void *)uap->rlim);
+#endif
+
+	if (uap->resource >= LINUX_RLIM_NLIMITS)
+		return (EINVAL);
+
+	bsd.which = linux_to_bsd_resource[uap->resource];
+	if (bsd.which == -1)
+		return (EINVAL);
+
+	bsd.rlp = stackgap_alloc(&sg, sizeof(struct rlimit));
+	error = getrlimit(p, &bsd);
+	if (error)
+		return (error);
+
+	rlim.rlim_cur = (unsigned long)bsd.rlp->rlim_cur;
+	if (rlim.rlim_cur == ULONG_MAX)
+		rlim.rlim_cur = LONG_MAX;
+	rlim.rlim_max = (unsigned long)bsd.rlp->rlim_max;
+	if (rlim.rlim_max == ULONG_MAX)
+		rlim.rlim_max = LONG_MAX;
+	return (copyout(&rlim, uap->rlim, sizeof(rlim)));
+}
+
+int
+linux_sched_setscheduler(p, uap)
+	struct proc *p;
+	struct linux_sched_setscheduler_args *uap;
+{
+	struct sched_setscheduler_args bsd;
+
+#ifdef DEBUG
+	printf("Linux-emul(%ld): sched_setscheduler(%d, %d, %p)\n",
+	    (long)p->p_pid, uap->pid, uap->policy, (const void *)uap->param);
+#endif
+
+	switch (uap->policy) {
+	case LINUX_SCHED_OTHER:
+		bsd.policy = SCHED_OTHER;
+		break;
+	case LINUX_SCHED_FIFO:
+		bsd.policy = SCHED_FIFO;
+		break;
+	case LINUX_SCHED_RR:
+		bsd.policy = SCHED_RR;
+		break;
+	default:
+		return EINVAL;
+	}
+
+	bsd.pid = uap->pid;
+	bsd.param = uap->param;
+	return sched_setscheduler(p, &bsd);
+}
+
+int
+linux_sched_getscheduler(p, uap)
+	struct proc *p;
+	struct linux_sched_getscheduler_args *uap;
+{
+	struct sched_getscheduler_args bsd;
+	int error;
+
+#ifdef DEBUG
+	printf("Linux-emul(%ld): sched_getscheduler(%d)\n",
+	       (long)p->p_pid, uap->pid);
+#endif
+
+	bsd.pid = uap->pid;
+	error = sched_getscheduler(p, &bsd);
+
+	switch (p->p_retval[0]) {
+	case SCHED_OTHER:
+		p->p_retval[0] = LINUX_SCHED_OTHER;
+		break;
+	case SCHED_FIFO:
+		p->p_retval[0] = LINUX_SCHED_FIFO;
+		break;
+	case SCHED_RR:
+		p->p_retval[0] = LINUX_SCHED_RR;
+		break;
+	}
+
+	return error;
+}
diff --git a/sys/compat/linux/linux_socket.c b/sys/compat/linux/linux_socket.c
new file mode 100644
index 000000000000..b7c9b7da8460
--- /dev/null
+++ b/sys/compat/linux/linux_socket.c
@@ -0,0 +1,904 @@
+/*-
+ * Copyright (c) 1995 Søren Schmidt
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ * 1. Redistributions of source code must retain the above copyright
+ *    notice, this list of conditions and the following disclaimer 
+ *    in this position and unchanged.
+ * 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. The name of the author may not be used to endorse or promote products
+ *    derived from this software withough specific prior written permission
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``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 AUTHOR 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.
+ *
+ * $FreeBSD$
+ */
+
+/* XXX we use functions that might not exist. */
+#include "opt_compat.h"
+
+#ifndef COMPAT_43
+#error "Unable to compile Linux-emulator due to missing COMPAT_43 option!"
+#endif
+
+#include <sys/param.h>
+#include <sys/proc.h>
+#include <sys/systm.h>
+#include <sys/sysproto.h>
+#include <sys/fcntl.h>
+#include <sys/socket.h>
+#include <sys/uio.h>
+
+#include <netinet/in.h>
+#include <netinet/in_systm.h>
+#include <netinet/ip.h>
+
+#include <machine/../linux/linux.h>
+#include <machine/../linux/linux_proto.h>
+#include <compat/linux/linux_util.h>
+
+static int
+linux_to_bsd_domain(int domain)
+{
+
+	switch (domain) {
+	case LINUX_AF_UNSPEC:
+		return (AF_UNSPEC);
+	case LINUX_AF_UNIX:
+		return (AF_LOCAL);
+	case LINUX_AF_INET:
+		return (AF_INET);
+	case LINUX_AF_AX25:
+		return (AF_CCITT);
+	case LINUX_AF_IPX:
+		return (AF_IPX);
+	case LINUX_AF_APPLETALK:
+		return (AF_APPLETALK);
+	}
+	return (-1);
+}
+
+static int
+linux_to_bsd_sockopt_level(int level)
+{
+
+	switch (level) {
+	case LINUX_SOL_SOCKET:
+		return (SOL_SOCKET);
+	}
+	return (level);
+}
+
+static int
+linux_to_bsd_ip_sockopt(int opt)
+{
+
+	switch (opt) {
+	case LINUX_IP_TOS:
+		return (IP_TOS);
+	case LINUX_IP_TTL:
+		return (IP_TTL);
+	case LINUX_IP_OPTIONS:
+		return (IP_OPTIONS);
+	case LINUX_IP_MULTICAST_IF:
+		return (IP_MULTICAST_IF);
+	case LINUX_IP_MULTICAST_TTL:
+		return (IP_MULTICAST_TTL);
+	case LINUX_IP_MULTICAST_LOOP:
+		return (IP_MULTICAST_LOOP);
+	case LINUX_IP_ADD_MEMBERSHIP:
+		return (IP_ADD_MEMBERSHIP);
+	case LINUX_IP_DROP_MEMBERSHIP:
+		return (IP_DROP_MEMBERSHIP);
+	case LINUX_IP_HDRINCL:
+		return (IP_HDRINCL);
+	}
+	return (-1);
+}
+
+static int
+linux_to_bsd_so_sockopt(int opt)
+{
+
+	switch (opt) {
+	case LINUX_SO_DEBUG:
+		return (SO_DEBUG);
+	case LINUX_SO_REUSEADDR:
+		return (SO_REUSEADDR);
+	case LINUX_SO_TYPE:
+		return (SO_TYPE);
+	case LINUX_SO_ERROR:
+		return (SO_ERROR);
+	case LINUX_SO_DONTROUTE:
+		return (SO_DONTROUTE);
+	case LINUX_SO_BROADCAST:
+		return (SO_BROADCAST);
+	case LINUX_SO_SNDBUF:
+		return (SO_SNDBUF);
+	case LINUX_SO_RCVBUF:
+		return (SO_RCVBUF);
+	case LINUX_SO_KEEPALIVE:
+		return (SO_KEEPALIVE);
+	case LINUX_SO_OOBINLINE:
+		return (SO_OOBINLINE);
+	case LINUX_SO_LINGER:
+		return (SO_LINGER);
+	}
+	return (-1);
+}
+
+/* Return 0 if IP_HDRINCL is set for the given socket. */
+static int
+linux_check_hdrincl(struct proc *p, int s)
+{
+	struct getsockopt_args /* {
+		int s;
+		int level;
+		int name;
+		caddr_t val;
+		int *avalsize;
+	} */ bsd_args;
+	int error;
+	caddr_t sg, val, valsize;
+	int size_val = sizeof val;
+	int optval;
+
+	sg = stackgap_init();
+	val = stackgap_alloc(&sg, sizeof(int));
+	valsize = stackgap_alloc(&sg, sizeof(int));
+
+	if ((error = copyout(&size_val, valsize, sizeof(size_val))))
+		return (error);
+
+	bsd_args.s = s;
+	bsd_args.level = IPPROTO_IP;
+	bsd_args.name = IP_HDRINCL;
+	bsd_args.val = val;
+	bsd_args.avalsize = (int *)valsize;
+	if ((error = getsockopt(p, &bsd_args)))
+		return (error);
+
+	if ((error = copyin(val, &optval, sizeof(optval))))
+		return (error);
+
+	return (optval == 0);
+}
+
+/*
+ * Updated sendto() when IP_HDRINCL is set:
+ * tweak endian-dependent fields in the IP packet.
+ */
+static int
+linux_sendto_hdrincl(struct proc *p, struct sendto_args *bsd_args)
+{
+/*
+ * linux_ip_copysize defines how many bytes we should copy
+ * from the beginning of the IP packet before we customize it for BSD.
+ * It should include all the fields we modify (ip_len and ip_off)
+ * and be as small as possible to minimize copying overhead.
+ */
+#define linux_ip_copysize	8
+
+	caddr_t sg;
+	struct ip *packet;
+	struct msghdr *msg;
+	struct iovec *iov;
+
+	int error;
+	struct  sendmsg_args /* {
+		int s;
+		caddr_t msg;
+		int flags;
+	} */ sendmsg_args;
+
+	/* Check the packet isn't too small before we mess with it */
+	if (bsd_args->len < linux_ip_copysize)
+		return (EINVAL);
+
+	/*
+	 * Tweaking the user buffer in place would be bad manners.
+	 * We create a corrected IP header with just the needed length,
+	 * then use an iovec to glue it to the rest of the user packet
+	 * when calling sendmsg().
+	 */
+	sg = stackgap_init();
+	packet = (struct ip *)stackgap_alloc(&sg, linux_ip_copysize);
+	msg = (struct msghdr *)stackgap_alloc(&sg, sizeof(*msg));
+	iov = (struct iovec *)stackgap_alloc(&sg, sizeof(*iov)*2);
+
+	/* Make a copy of the beginning of the packet to be sent */
+	if ((error = copyin(bsd_args->buf, packet, linux_ip_copysize)))
+		return (error);
+
+	/* Convert fields from Linux to BSD raw IP socket format */
+	packet->ip_len = bsd_args->len;
+	packet->ip_off = ntohs(packet->ip_off);
+
+	/* Prepare the msghdr and iovec structures describing the new packet */
+	msg->msg_name = bsd_args->to;
+	msg->msg_namelen = bsd_args->tolen;
+	msg->msg_iov = iov;
+	msg->msg_iovlen = 2;
+	msg->msg_control = NULL;
+	msg->msg_controllen = 0;
+	msg->msg_flags = 0;
+	iov[0].iov_base = (char *)packet;
+	iov[0].iov_len = linux_ip_copysize;
+	iov[1].iov_base = (char *)(bsd_args->buf) + linux_ip_copysize;
+	iov[1].iov_len = bsd_args->len - linux_ip_copysize;
+
+	sendmsg_args.s = bsd_args->s;
+	sendmsg_args.msg = (caddr_t)msg;
+	sendmsg_args.flags = bsd_args->flags;
+	return (sendmsg(p, &sendmsg_args));
+}
+
+struct linux_socket_args {
+	int domain;
+	int type;
+	int protocol;
+};
+
+static int
+linux_socket(struct proc *p, struct linux_socket_args *args)
+{
+	struct linux_socket_args linux_args;
+	struct socket_args /* {
+		int domain;
+		int type;
+		int protocol;
+	} */ bsd_args;
+	int error;
+	int retval_socket;
+
+	if ((error = copyin(args, &linux_args, sizeof(linux_args))))
+		return (error);
+
+	bsd_args.protocol = linux_args.protocol;
+	bsd_args.type = linux_args.type;
+	bsd_args.domain = linux_to_bsd_domain(linux_args.domain);
+	if (bsd_args.domain == -1)
+		return (EINVAL);
+
+	retval_socket = socket(p, &bsd_args);
+	if (bsd_args.type == SOCK_RAW
+	    && (bsd_args.protocol == IPPROTO_RAW || bsd_args.protocol == 0)
+	    && bsd_args.domain == AF_INET
+	    && retval_socket >= 0) {
+		/* It's a raw IP socket: set the IP_HDRINCL option. */
+		struct setsockopt_args /* {
+			int s;
+			int level;
+			int name;
+			caddr_t val;
+			int valsize;
+		} */ bsd_setsockopt_args;
+		caddr_t sg;
+		int *hdrincl;
+
+		sg = stackgap_init();
+		hdrincl = (int *)stackgap_alloc(&sg, sizeof(*hdrincl));
+		*hdrincl = 1;
+		bsd_setsockopt_args.s = p->p_retval[0];
+		bsd_setsockopt_args.level = IPPROTO_IP;
+		bsd_setsockopt_args.name = IP_HDRINCL;
+		bsd_setsockopt_args.val = (caddr_t)hdrincl;
+		bsd_setsockopt_args.valsize = sizeof(*hdrincl);
+		/* We ignore any error returned by setsockopt() */
+		setsockopt(p, &bsd_setsockopt_args);
+		/* Copy back the return value from socket() */
+		p->p_retval[0] = bsd_setsockopt_args.s;
+	}
+
+	return (retval_socket);
+}
+
+struct linux_bind_args {
+	int s;
+	struct sockaddr *name;
+	int namelen;
+};
+
+static int
+linux_bind(struct proc *p, struct linux_bind_args *args)
+{
+	struct linux_bind_args linux_args;
+	struct bind_args /* {
+		int s;
+		caddr_t name;
+		int namelen;
+	} */ bsd_args;
+	int error;
+
+	if ((error = copyin(args, &linux_args, sizeof(linux_args))))
+		return (error);
+
+	bsd_args.s = linux_args.s;
+	bsd_args.name = (caddr_t)linux_args.name;
+	bsd_args.namelen = linux_args.namelen;
+	return (bind(p, &bsd_args));
+}
+
+struct linux_connect_args {
+	int s;
+	struct sockaddr * name;
+	int namelen;
+};
+
+static int
+linux_connect(struct proc *p, struct linux_connect_args *args)
+{
+	struct linux_connect_args linux_args;
+	struct connect_args /* {
+		int s;
+		caddr_t name;
+		int namelen;
+	} */ bsd_args;
+	int error;
+
+	if ((error = copyin(args, &linux_args, sizeof(linux_args))))
+		return (error);
+
+	bsd_args.s = linux_args.s;
+	bsd_args.name = (caddr_t)linux_args.name;
+	bsd_args.namelen = linux_args.namelen;
+	error = connect(p, &bsd_args);
+	if (error == EISCONN) {
+		/*
+		 * Linux doesn't return EISCONN the first time it occurs,
+		 * when on a non-blocking socket. Instead it returns the
+		 * error getsockopt(SOL_SOCKET, SO_ERROR) would return on BSD.
+		 */
+		struct fcntl_args /* {
+			int fd;
+			int cmd;
+			int arg;
+		} */ bsd_fcntl_args;
+		struct getsockopt_args /* {
+			int s;
+			int level;
+			int name;
+			caddr_t val;
+			int *avalsize;
+		} */ bsd_getsockopt_args;
+		void *status, *statusl;
+		int stat, statl = sizeof stat;
+		caddr_t sg;
+
+		/* Check for non-blocking */
+		bsd_fcntl_args.fd = linux_args.s;
+		bsd_fcntl_args.cmd = F_GETFL;
+		bsd_fcntl_args.arg = 0;
+		error = fcntl(p, &bsd_fcntl_args);
+		if (error == 0 && (p->p_retval[0] & O_NONBLOCK)) {
+			sg = stackgap_init();
+			status = stackgap_alloc(&sg, sizeof stat);
+			statusl = stackgap_alloc(&sg, sizeof statusl);
+
+			if ((error = copyout(&statl, statusl, sizeof statl)))
+				return (error);
+
+			bsd_getsockopt_args.s = linux_args.s;
+			bsd_getsockopt_args.level = SOL_SOCKET;
+			bsd_getsockopt_args.name = SO_ERROR;
+			bsd_getsockopt_args.val = status;
+			bsd_getsockopt_args.avalsize = statusl;
+
+			error = getsockopt(p, &bsd_getsockopt_args);
+			if (error)
+				return (error);
+
+			if ((error = copyin(status, &stat, sizeof stat)))
+				return (error);
+
+			p->p_retval[0] = stat;
+			return (0);
+		}
+	}
+
+	return (error);
+}
+
+struct linux_listen_args {
+	int s;
+	int backlog;
+};
+
+static int
+linux_listen(struct proc *p, struct linux_listen_args *args)
+{
+	struct linux_listen_args linux_args;
+	struct listen_args /* {
+		int s;
+		int backlog;
+	} */ bsd_args;
+	int error;
+
+	if ((error = copyin(args, &linux_args, sizeof(linux_args))))
+		return (error);
+
+	bsd_args.s = linux_args.s;
+	bsd_args.backlog = linux_args.backlog;
+	return (listen(p, &bsd_args));
+}
+
+struct linux_accept_args {
+	int s;
+	struct sockaddr *addr;
+	int *namelen;
+};
+
+static int
+linux_accept(struct proc *p, struct linux_accept_args *args)
+{
+	struct linux_accept_args linux_args;
+	struct accept_args /* {
+		int s;
+		caddr_t name;
+		int *anamelen;
+	} */ bsd_args;
+	struct fcntl_args /* {
+		int fd;
+		int cmd;
+		long arg;
+	} */ f_args;
+	int error;
+
+	if ((error = copyin(args, &linux_args, sizeof(linux_args))))
+		return (error);
+
+	bsd_args.s = linux_args.s;
+	bsd_args.name = (caddr_t)linux_args.addr;
+	bsd_args.anamelen = linux_args.namelen;
+	error = oaccept(p, &bsd_args);
+	if (error)
+		return (error);
+
+	/*
+	 * linux appears not to copy flags from the parent socket to the
+	 * accepted one, so we must clear the flags in the new descriptor.
+	 * Ignore any errors, because we already have an open fd.
+	 */
+	f_args.fd = p->p_retval[0];
+	f_args.cmd = F_SETFL;
+	f_args.arg = 0;
+	(void)fcntl(p, &f_args);
+	p->p_retval[0] = f_args.fd;
+	return (0);
+}
+
+struct linux_getsockname_args {
+	int s;
+	struct sockaddr *addr;
+	int *namelen;
+};
+
+static int
+linux_getsockname(struct proc *p, struct linux_getsockname_args *args)
+{
+	struct linux_getsockname_args linux_args;
+	struct getsockname_args /* {
+		int fdes;
+		caddr_t asa;
+		int *alen;
+	} */ bsd_args;
+	int error;
+
+	if ((error = copyin(args, &linux_args, sizeof(linux_args))))
+		return (error);
+
+	bsd_args.fdes = linux_args.s;
+	bsd_args.asa = (caddr_t) linux_args.addr;
+	bsd_args.alen = linux_args.namelen;
+	return (ogetsockname(p, &bsd_args));
+}
+
+struct linux_getpeername_args {
+	int s;
+	struct sockaddr *addr;
+	int *namelen;
+};
+
+static int
+linux_getpeername(struct proc *p, struct linux_getpeername_args *args)
+{
+	struct linux_getpeername_args linux_args;
+	struct ogetpeername_args /* {
+		int fdes;
+		caddr_t asa;
+		int *alen;
+	} */ bsd_args;
+	int error;
+
+	if ((error = copyin(args, &linux_args, sizeof(linux_args))))
+		return (error);
+
+	bsd_args.fdes = linux_args.s;
+	bsd_args.asa = (caddr_t) linux_args.addr;
+	bsd_args.alen = linux_args.namelen;
+	return (ogetpeername(p, &bsd_args));
+}
+
+struct linux_socketpair_args {
+	int domain;
+	int type;
+	int protocol;
+	int *rsv;
+};
+
+static int
+linux_socketpair(struct proc *p, struct linux_socketpair_args *args)
+{
+	struct linux_socketpair_args linux_args;
+	struct socketpair_args /* {
+		int domain;
+		int type;
+		int protocol;
+		int *rsv;
+	} */ bsd_args;
+	int error;
+
+	if ((error = copyin(args, &linux_args, sizeof(linux_args))))
+		return (error);
+
+	bsd_args.domain = linux_to_bsd_domain(linux_args.domain);
+	if (bsd_args.domain == -1)
+		return (EINVAL);
+
+	bsd_args.type = linux_args.type;
+	bsd_args.protocol = linux_args.protocol;
+	bsd_args.rsv = linux_args.rsv;
+	return (socketpair(p, &bsd_args));
+}
+
+struct linux_send_args {
+	int s;
+	void *msg;
+	int len;
+	int flags;
+};
+
+static int
+linux_send(struct proc *p, struct linux_send_args *args)
+{
+	struct linux_send_args linux_args;
+	struct osend_args /* {
+		int s;
+		caddr_t buf;
+		int len;
+		int flags;
+	} */ bsd_args;
+	int error;
+
+	if ((error = copyin(args, &linux_args, sizeof(linux_args))))
+		return (error);
+
+	bsd_args.s = linux_args.s;
+	bsd_args.buf = linux_args.msg;
+	bsd_args.len = linux_args.len;
+	bsd_args.flags = linux_args.flags;
+	return (osend(p, &bsd_args));
+}
+
+struct linux_recv_args {
+	int s;
+	void *msg;
+	int len;
+	int flags;
+};
+
+static int
+linux_recv(struct proc *p, struct linux_recv_args *args)
+{
+	struct linux_recv_args linux_args;
+	struct orecv_args /* {
+		int s;
+		caddr_t buf;
+		int len;
+		int flags;
+	} */ bsd_args;
+	int error;
+
+	if ((error = copyin(args, &linux_args, sizeof(linux_args))))
+		return (error);
+
+	bsd_args.s = linux_args.s;
+	bsd_args.buf = linux_args.msg;
+	bsd_args.len = linux_args.len;
+	bsd_args.flags = linux_args.flags;
+	return (orecv(p, &bsd_args));
+}
+
+struct linux_sendto_args {
+	int s;
+	void *msg;
+	int len;
+	int flags;
+	caddr_t to;
+	int tolen;
+};
+
+static int
+linux_sendto(struct proc *p, struct linux_sendto_args *args)
+{
+	struct linux_sendto_args linux_args;
+	struct sendto_args /* {
+		int s;
+		caddr_t buf;
+		size_t len;
+		int flags;
+		caddr_t to;
+		int tolen;
+	} */ bsd_args;
+	int error;
+
+	if ((error = copyin(args, &linux_args, sizeof(linux_args))))
+		return (error);
+
+	bsd_args.s = linux_args.s;
+	bsd_args.buf = linux_args.msg;
+	bsd_args.len = linux_args.len;
+	bsd_args.flags = linux_args.flags;
+	bsd_args.to = linux_args.to;
+	bsd_args.tolen = linux_args.tolen;
+
+	if (linux_check_hdrincl(p, linux_args.s) == 0)
+		/* IP_HDRINCL set, tweak the packet before sending */
+		return (linux_sendto_hdrincl(p, &bsd_args));
+
+	return (sendto(p, &bsd_args));
+}
+
+struct linux_recvfrom_args {
+	int s;
+	void *buf;
+	int len;
+	int flags;
+	caddr_t from;
+	int *fromlen;
+};
+
+static int
+linux_recvfrom(struct proc *p, struct linux_recvfrom_args *args)
+{
+	struct linux_recvfrom_args linux_args;
+	struct recvfrom_args /* {
+		int s;
+		caddr_t buf;
+		size_t len;
+		int flags;
+		caddr_t from;
+		int *fromlenaddr;
+	} */ bsd_args;
+	int error;
+
+	if ((error = copyin(args, &linux_args, sizeof(linux_args))))
+		return (error);
+
+	bsd_args.s = linux_args.s;
+	bsd_args.buf = linux_args.buf;
+	bsd_args.len = linux_args.len;
+	bsd_args.flags = linux_args.flags;
+	bsd_args.from = linux_args.from;
+	bsd_args.fromlenaddr = linux_args.fromlen;
+	return (orecvfrom(p, &bsd_args));
+}
+
+struct linux_shutdown_args {
+	int s;
+	int how;
+};
+
+static int
+linux_shutdown(struct proc *p, struct linux_shutdown_args *args)
+{
+	struct linux_shutdown_args linux_args;
+	struct shutdown_args /* {
+		int s;
+		int how;
+	} */ bsd_args;
+	int error;
+
+	if ((error = copyin(args, &linux_args, sizeof(linux_args))))
+		return (error);
+
+	bsd_args.s = linux_args.s;
+	bsd_args.how = linux_args.how;
+	return (shutdown(p, &bsd_args));
+}
+
+struct linux_setsockopt_args {
+	int s;
+	int level;
+	int optname;
+	void *optval;
+	int optlen;
+};
+
+static int
+linux_setsockopt(struct proc *p, struct linux_setsockopt_args *args)
+{
+	struct linux_setsockopt_args linux_args;
+	struct setsockopt_args /* {
+		int s;
+		int level;
+		int name;
+		caddr_t val;
+		int valsize;
+	} */ bsd_args;
+	int error, name;
+
+	if ((error = copyin(args, &linux_args, sizeof(linux_args))))
+		return (error);
+
+	bsd_args.s = linux_args.s;
+	bsd_args.level = linux_to_bsd_sockopt_level(linux_args.level);
+	switch (bsd_args.level) {
+	case SOL_SOCKET:
+		name = linux_to_bsd_so_sockopt(linux_args.optname);
+		break;
+	case IPPROTO_IP:
+		name = linux_to_bsd_ip_sockopt(linux_args.optname);
+		break;
+	case IPPROTO_TCP:
+		/* Linux TCP option values match BSD's */
+		name = linux_args.optname;
+		break;
+	default:
+		name = -1;
+		break;
+	}
+	if (name == -1)
+		return (EINVAL);
+
+	bsd_args.name = name;
+	bsd_args.val = linux_args.optval;
+	bsd_args.valsize = linux_args.optlen;
+	return (setsockopt(p, &bsd_args));
+}
+
+struct linux_getsockopt_args {
+	int s;
+	int level;
+	int optname;
+	void *optval;
+	int *optlen;
+};
+
+static int
+linux_getsockopt(struct proc *p, struct linux_getsockopt_args *args)
+{
+	struct linux_getsockopt_args linux_args;
+	struct getsockopt_args /* {
+		int s;
+		int level;
+		int name;
+		caddr_t val;
+		int *avalsize;
+	} */ bsd_args;
+	int error, name;
+
+	if ((error = copyin(args, &linux_args, sizeof(linux_args))))
+		return (error);
+
+	bsd_args.s = linux_args.s;
+	bsd_args.level = linux_to_bsd_sockopt_level(linux_args.level);
+	switch (bsd_args.level) {
+	case SOL_SOCKET:
+		name = linux_to_bsd_so_sockopt(linux_args.optname);
+		break;
+	case IPPROTO_IP:
+		name = linux_to_bsd_ip_sockopt(linux_args.optname);
+		break;
+	case IPPROTO_TCP:
+		/* Linux TCP option values match BSD's */
+		name = linux_args.optname;
+		break;
+	default:
+		name = -1;
+		break;
+	}
+	if (name == -1)
+		return (EINVAL);
+
+	bsd_args.name = name;
+	bsd_args.val = linux_args.optval;
+	bsd_args.avalsize = linux_args.optlen;
+	return (getsockopt(p, &bsd_args));
+}
+
+int
+linux_socketcall(struct proc *p, struct linux_socketcall_args *args)
+{
+
+	switch (args->what) {
+	case LINUX_SOCKET:
+		return (linux_socket(p, args->args));
+	case LINUX_BIND:
+		return (linux_bind(p, args->args));
+	case LINUX_CONNECT:
+		return (linux_connect(p, args->args));
+	case LINUX_LISTEN:
+		return (linux_listen(p, args->args));
+	case LINUX_ACCEPT:
+		return (linux_accept(p, args->args));
+	case LINUX_GETSOCKNAME:
+		return (linux_getsockname(p, args->args));
+	case LINUX_GETPEERNAME:
+		return (linux_getpeername(p, args->args));
+	case LINUX_SOCKETPAIR:
+		return (linux_socketpair(p, args->args));
+	case LINUX_SEND:
+		return (linux_send(p, args->args));
+	case LINUX_RECV:
+		return (linux_recv(p, args->args));
+	case LINUX_SENDTO:
+		return (linux_sendto(p, args->args));
+	case LINUX_RECVFROM:
+		return (linux_recvfrom(p, args->args));
+	case LINUX_SHUTDOWN:
+		return (linux_shutdown(p, args->args));
+	case LINUX_SETSOCKOPT:
+		return (linux_setsockopt(p, args->args));
+	case LINUX_GETSOCKOPT:
+		return (linux_getsockopt(p, args->args));
+	case LINUX_SENDMSG:
+		do {
+			int error;
+			int level;
+			caddr_t control;
+			struct {
+				int s;
+				const struct msghdr *msg;
+				int flags;
+			} *uap = args->args;
+
+			error = copyin(&uap->msg->msg_control, &control,
+			    sizeof(caddr_t));
+			if (error)
+				return (error);
+
+			if (control == NULL)
+				goto done;
+
+			error = copyin(&((struct cmsghdr*)control)->cmsg_level,
+			    &level, sizeof(int));
+			if (error)
+				return (error);
+
+			if (level == 1) {
+				/*
+				 * Linux thinks that SOL_SOCKET is 1; we know
+				 * that it's really 0xffff, of course.
+				 */
+				level = SOL_SOCKET;
+				error = copyout(&level,
+				    &((struct cmsghdr *)control)->cmsg_level,
+				    sizeof(int));
+				if (error)
+					return (error);
+			}
+		done:
+			return (sendmsg(p, args->args));
+		} while (0);
+	case LINUX_RECVMSG:
+		return (recvmsg(p, args->args));
+	}
+
+	uprintf("LINUX: 'socket' typ=%d not implemented\n", args->what);
+	return (ENOSYS);
+}