NetBSD/sys/compat/ultrix/ultrix_misc.c

927 lines
22 KiB
C

/* $NetBSD: ultrix_misc.c,v 1.81 2003/01/18 08:49:22 thorpej Exp $ */
/*
* Copyright (c) 1995, 1997 Jonathan Stone (hereinafter referred to as the author)
* 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.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. All advertising materials mentioning features or use of this software
* must display the following acknowledgement:
* This product includes software developed by Jonathan Stone for
* the NetBSD Project.
* 4. The name of the author may not be used to endorse or promote products
* derived from this software without 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.
*/
/*
* Copyright (c) 1992, 1993
* The Regents of the University of California. All rights reserved.
*
* This software was developed by the Computer Systems Engineering group
* at Lawrence Berkeley Laboratory under DARPA contract BG 91-66 and
* contributed to Berkeley.
*
* All advertising materials mentioning features or use of this software
* must display the following acknowledgement:
* This product includes software developed by the University of
* California, Lawrence Berkeley Laboratory.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. All advertising materials mentioning features or use of this software
* must display the following acknowledgement:
* This product includes software developed by the University of
* California, Berkeley and its contributors.
* 4. Neither the name of the University nor the names of its contributors
* may be used to endorse or promote products derived from this software
* without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*
*
* @(#)sun_misc.c 8.1 (Berkeley) 6/18/93
*
* from: Header: sun_misc.c,v 1.16 93/04/07 02:46:27 torek Exp
*/
#include <sys/cdefs.h>
__KERNEL_RCSID(0, "$NetBSD: ultrix_misc.c,v 1.81 2003/01/18 08:49:22 thorpej Exp $");
#if defined(_KERNEL_OPT)
#include "opt_nfsserver.h"
#include "opt_sysv.h"
#endif
#include <sys/cdefs.h> /* RCS ID & Copyright macro defns */
/*
* Ultrix compatibility module.
*
* Ultrix system calls that are implemented differently in BSD are
* handled here.
*/
#if defined(_KERNEL_OPT)
#include "fs_nfs.h"
#endif
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/namei.h>
#include <sys/dirent.h>
#include <sys/proc.h>
#include <sys/file.h>
#include <sys/filedesc.h>
#include <sys/kernel.h>
#include <sys/exec.h>
#include <sys/malloc.h>
#include <sys/mbuf.h>
#include <sys/mman.h>
#include <sys/mount.h>
#include <sys/resource.h>
#include <sys/resourcevar.h>
#include <sys/signal.h>
#include <sys/signalvar.h>
#include <sys/socket.h>
#include <sys/vnode.h>
#include <sys/uio.h>
#include <sys/wait.h>
#include <sys/utsname.h>
#include <sys/unistd.h>
#include <sys/ipc.h>
#include <sys/sa.h>
#include <sys/syscallargs.h>
#include <compat/ultrix/ultrix_syscall.h>
#include <compat/ultrix/ultrix_syscallargs.h>
#include <compat/common/compat_util.h>
#include <netinet/in.h>
#include <miscfs/specfs/specdev.h>
#include <nfs/rpcv2.h>
#include <nfs/nfsproto.h>
#include <nfs/nfs.h>
#include <sys/socketvar.h> /* sosetopt() */
#include <compat/ultrix/ultrix_flock.h>
#ifdef __mips
#include <mips/cachectl.h>
#endif
static int ultrix_to_bsd_flock __P((struct ultrix_flock *, struct flock *));
static void bsd_to_ultrix_flock __P((struct flock *, struct ultrix_flock *));
extern struct sysent ultrix_sysent[];
extern const char * const ultrix_syscallnames[];
extern char ultrix_sigcode[], ultrix_esigcode[];
#ifdef __HAVE_SYSCALL_INTERN
void syscall_intern(struct proc *);
#else
void syscall __P((void));
#endif
const struct emul emul_ultrix = {
"ultrix",
"/emul/ultrix",
#ifndef __HAVE_MINIMAL_EMUL
0,
NULL,
ULTRIX_SYS_syscall,
ULTRIX_SYS_NSYSENT,
#endif
ultrix_sysent,
ultrix_syscallnames,
sendsig,
trapsignal,
ultrix_sigcode,
ultrix_esigcode,
setregs,
NULL,
NULL,
NULL,
#ifdef __HAVE_SYSCALL_INTERN
syscall_intern,
#else
syscall,
#endif
NULL,
NULL,
};
#define GSI_PROG_ENV 1
int
ultrix_sys_getsysinfo(l, v, retval)
struct lwp *l;
void *v;
register_t *retval;
{
struct ultrix_sys_getsysinfo_args *uap = v;
static short progenv = 0;
switch (SCARG(uap, op)) {
/* operations implemented: */
case GSI_PROG_ENV:
if (SCARG(uap, nbytes) < sizeof(short))
return EINVAL;
*retval = 1;
return (copyout(&progenv, SCARG(uap, buffer), sizeof(short)));
default:
*retval = 0; /* info unavail */
return 0;
}
}
int
ultrix_sys_setsysinfo(l, v, retval)
struct lwp *l;
void *v;
register_t *retval;
{
#ifdef notyet
struct ultrix_sys_setsysinfo_args *uap = v;
#endif
*retval = 0;
return 0;
}
int
ultrix_sys_waitpid(l, v, retval)
struct lwp *l;
void *v;
register_t *retval;
{
struct ultrix_sys_waitpid_args *uap = v;
struct sys_wait4_args ua;
SCARG(&ua, pid) = SCARG(uap, pid);
SCARG(&ua, status) = SCARG(uap, status);
SCARG(&ua, options) = SCARG(uap, options);
SCARG(&ua, rusage) = 0;
return (sys_wait4(l, &ua, retval));
}
int
ultrix_sys_wait3(l, v, retval)
struct lwp *l;
void *v;
register_t *retval;
{
struct ultrix_sys_wait3_args *uap = v;
struct sys_wait4_args ua;
SCARG(&ua, pid) = -1;
SCARG(&ua, status) = SCARG(uap, status);
SCARG(&ua, options) = SCARG(uap, options);
SCARG(&ua, rusage) = SCARG(uap, rusage);
return (sys_wait4(l, &ua, retval));
}
/*
* Ultrix binaries pass in FD_MAX as the first arg to select().
* On Ultrix, FD_MAX is 4096, which is more than the NetBSD sys_select()
* can handle.
* Since we can't have more than the (native) FD_MAX descriptors open,
* limit nfds to at most FD_MAX.
*/
int
ultrix_sys_select(l, v, retval)
struct lwp *l;
void *v;
register_t *retval;
{
struct sys_select_args *uap = v;
struct timeval atv;
int error;
/* Limit number of FDs selected on to the native maximum */
if (SCARG(uap, nd) > FD_SETSIZE)
SCARG(uap, nd) = FD_SETSIZE;
/* Check for negative timeval */
if (SCARG(uap, tv)) {
error = copyin((caddr_t)SCARG(uap, tv), (caddr_t)&atv,
sizeof(atv));
if (error)
goto done;
#ifdef DEBUG
/* Ultrix clients sometimes give negative timeouts? */
if (atv.tv_sec < 0 || atv.tv_usec < 0)
printf("ultrix select( %ld, %ld): negative timeout\n",
atv.tv_sec, atv.tv_usec);
#endif
}
error = sys_select(l, (void*) uap, retval);
if (error == EINVAL)
printf("ultrix select: bad args?\n");
done:
return error;
}
#if defined(NFS)
int
async_daemon(l, v, retval)
struct lwp *l;
void *v;
register_t *retval;
{
struct sys_nfssvc_args ouap;
SCARG(&ouap, flag) = NFSSVC_BIOD;
SCARG(&ouap, argp) = NULL;
return (sys_nfssvc(l, &ouap, retval));
}
#endif /* NFS */
#define SUN__MAP_NEW 0x80000000 /* if not, old mmap & cannot handle */
int
ultrix_sys_mmap(l, v, retval)
struct lwp *l;
void *v;
register_t *retval;
{
struct ultrix_sys_mmap_args *uap = v;
struct proc *p = l->l_proc;
struct sys_mmap_args ouap;
struct filedesc *fdp;
struct file *fp;
struct vnode *vp;
/*
* Verify the arguments.
*/
if (SCARG(uap, prot) & ~(PROT_READ|PROT_WRITE|PROT_EXEC))
return (EINVAL); /* XXX still needed? */
if ((SCARG(uap, flags) & SUN__MAP_NEW) == 0)
return (EINVAL);
SCARG(&ouap, flags) = SCARG(uap, flags) & ~SUN__MAP_NEW;
SCARG(&ouap, addr) = SCARG(uap, addr);
if ((SCARG(&ouap, flags) & MAP_FIXED) == 0 &&
SCARG(&ouap, addr) != 0 &&
SCARG(&ouap, addr) < (void *)round_page((vaddr_t)p->p_vmspace->vm_daddr+MAXDSIZ))
SCARG(&ouap, addr) = (void *)round_page((vaddr_t)p->p_vmspace->vm_daddr+MAXDSIZ);
SCARG(&ouap, len) = SCARG(uap, len);
SCARG(&ouap, prot) = SCARG(uap, prot);
SCARG(&ouap, fd) = SCARG(uap, fd);
SCARG(&ouap, pos) = SCARG(uap, pos);
/*
* Special case: if fd refers to /dev/zero, map as MAP_ANON. (XXX)
*/
fdp = p->p_fd;
if ((fp = fd_getfile(fdp, SCARG(&ouap, fd))) != NULL && /*XXX*/
fp->f_type == DTYPE_VNODE && /*XXX*/
(vp = (struct vnode *)fp->f_data)->v_type == VCHR && /*XXX*/
vp->v_rdev == zerodev) { /*XXX*/
SCARG(&ouap, flags) |= MAP_ANON;
SCARG(&ouap, fd) = -1;
}
return (sys_mmap(l, &ouap, retval));
}
int
ultrix_sys_setsockopt(l, v, retval)
struct lwp *l;
void *v;
register_t *retval;
{
struct ultrix_sys_setsockopt_args *uap = v;
struct proc *p = l->l_proc;
struct file *fp;
struct mbuf *m = NULL;
int error;
/* getsock() will use the descriptor for us */
if ((error = getsock(p->p_fd, SCARG(uap, s), &fp)) != 0)
return (error);
#define SO_DONTLINGER (~SO_LINGER)
if (SCARG(uap, name) == SO_DONTLINGER) {
m = m_get(M_WAIT, MT_SOOPTS);
mtod(m, struct linger *)->l_onoff = 0;
m->m_len = sizeof(struct linger);
error = sosetopt((struct socket *)fp->f_data, SCARG(uap, level),
SO_LINGER, m);
}
if (SCARG(uap, level) == IPPROTO_IP) {
#define EMUL_IP_MULTICAST_IF 2
#define EMUL_IP_MULTICAST_TTL 3
#define EMUL_IP_MULTICAST_LOOP 4
#define EMUL_IP_ADD_MEMBERSHIP 5
#define EMUL_IP_DROP_MEMBERSHIP 6
static int ipoptxlat[] = {
IP_MULTICAST_IF,
IP_MULTICAST_TTL,
IP_MULTICAST_LOOP,
IP_ADD_MEMBERSHIP,
IP_DROP_MEMBERSHIP
};
if (SCARG(uap, name) >= EMUL_IP_MULTICAST_IF &&
SCARG(uap, name) <= EMUL_IP_DROP_MEMBERSHIP) {
SCARG(uap, name) =
ipoptxlat[SCARG(uap, name) - EMUL_IP_MULTICAST_IF];
}
}
if (SCARG(uap, valsize) > MLEN) {
error = EINVAL;
goto out;
}
if (SCARG(uap, val)) {
m = m_get(M_WAIT, MT_SOOPTS);
error = copyin(SCARG(uap, val), mtod(m, caddr_t),
(u_int)SCARG(uap, valsize));
if (error) {
(void) m_free(m);
goto out;
}
m->m_len = SCARG(uap, valsize);
}
error = sosetopt((struct socket *)fp->f_data, SCARG(uap, level),
SCARG(uap, name), m);
out:
FILE_UNUSE(fp, p);
return (error);
}
#define ULTRIX__SYS_NMLN 32
struct ultrix_utsname {
char sysname[ULTRIX__SYS_NMLN];
char nodename[ULTRIX__SYS_NMLN];
char release[ULTRIX__SYS_NMLN];
char version[ULTRIX__SYS_NMLN];
char machine[ULTRIX__SYS_NMLN];
};
int
ultrix_sys_uname(l, v, retval)
struct lwp *l;
void *v;
register_t *retval;
{
struct ultrix_sys_uname_args *uap = v;
struct ultrix_utsname sut;
const char *cp;
char *dp, *ep;
memset(&sut, 0, sizeof(sut));
strncpy(sut.sysname, ostype, sizeof(sut.sysname) - 1);
strncpy(sut.nodename, hostname, sizeof(sut.nodename) - 1);
strncpy(sut.release, osrelease, sizeof(sut.release) - 1);
dp = sut.version;
ep = &sut.version[sizeof(sut.version) - 1];
for (cp = version; *cp && *cp != '('; cp++)
;
for (cp++; *cp && *cp != ')' && dp < ep; cp++)
*dp++ = *cp;
for (; *cp && *cp != '#'; cp++)
;
for (; *cp && *cp != ':' && dp < ep; cp++)
*dp++ = *cp;
*dp = '\0';
strncpy(sut.machine, machine, sizeof(sut.machine) - 1);
return copyout((caddr_t)&sut, (caddr_t)SCARG(uap, name),
sizeof(struct ultrix_utsname));
}
int
ultrix_sys_setpgrp(l, v, retval)
struct lwp *l;
void *v;
register_t *retval;
{
struct ultrix_sys_setpgrp_args *uap = v;
struct proc *p = l->l_proc;
/*
* difference to our setpgid call is to include backwards
* compatibility to pre-setsid() binaries. Do setsid()
* instead of setpgid() in those cases where the process
* tries to create a new session the old way.
*/
if (!SCARG(uap, pgid) &&
(!SCARG(uap, pid) || SCARG(uap, pid) == p->p_pid))
return sys_setsid(l, uap, retval);
else
return sys_setpgid(l, uap, retval);
}
#if defined (NFSSERVER)
int
ultrix_sys_nfssvc(l, v, retval)
struct lwp *l;
void *v;
register_t *retval;
{
#if 0 /* XXX */
struct ultrix_sys_nfssvc_args *uap = v;
struct emul *e = p->p_emul;
struct sys_nfssvc_args outuap;
struct sockaddr sa;
int error;
caddr_t sg = stackgap_init(p, 0);
memset(&outuap, 0, sizeof outuap);
SCARG(&outuap, fd) = SCARG(uap, fd);
SCARG(&outuap, mskval) = stackgap_alloc(p, &sg, sizeof sa);
SCARG(&outuap, msklen) = sizeof sa;
SCARG(&outuap, mtchval) = stackgap_alloc(p, &sg, sizeof sa);
SCARG(&outuap, mtchlen) = sizeof sa;
memset(&sa, 0, sizeof sa);
if (error = copyout(&sa, SCARG(&outuap, mskval), SCARG(&outuap, msklen)))
return (error);
if (error = copyout(&sa, SCARG(&outuap, mtchval), SCARG(&outuap, mtchlen)))
return (error);
return nfssvc(l, &outuap, retval);
#else
return (ENOSYS);
#endif
}
#endif /* NFSSERVER */
struct ultrix_ustat {
daddr_t f_tfree; /* total free */
ino_t f_tinode; /* total inodes free */
char f_fname[6]; /* filsys name */
char f_fpack[6]; /* filsys pack name */
};
int
ultrix_sys_ustat(l, v, retval)
struct lwp *l;
void *v;
register_t *retval;
{
struct ultrix_sys_ustat_args *uap = v;
struct ultrix_ustat us;
int error;
memset(&us, 0, sizeof us);
/*
* XXX: should set f_tfree and f_tinode at least
* How do we translate dev -> fstat? (and then to ultrix_ustat)
*/
if ((error = copyout(&us, SCARG(uap, buf), sizeof us)) != 0)
return (error);
return 0;
}
int
ultrix_sys_quotactl(l, v, retval)
struct lwp *l;
void *v;
register_t *retval;
{
#ifdef notyet
struct ultrix_sys_quotactl_args *uap = v;
#endif
return EINVAL;
}
int
ultrix_sys_vhangup(l, v, retval)
struct lwp *l;
void *v;
register_t *retval;
{
return 0;
}
/*
* RISC Ultrix cache control syscalls
*/
#ifdef __mips
int
ultrix_sys_cacheflush(l, v, retval)
struct lwp *l;
void *v;
register_t *retval;
{
struct ultrix_sys_cacheflush_args /* {
syscallarg(void *) addr;
syscallarg(int) nbytes;
syscallarg(int) flag;
} */ *uap = v;
struct proc *p = l->l_proc;
vaddr_t va = (vaddr_t)SCARG(uap, addr);
int nbytes = SCARG(uap, nbytes);
int whichcache = SCARG(uap, whichcache);
return (mips_user_cacheflush(p, va, nbytes, whichcache));
}
int
ultrix_sys_cachectl(l, v, retval)
struct lwp *l;
void *v;
register_t *retval;
{
struct ultrix_sys_cachectl_args /* {
syscallarg(void *) addr;
syscallarg(int) nbytes;
syscallarg(int) cacheop;
} */ *uap = v;
struct proc *p = l->l_proc;
vaddr_t va = (vaddr_t)SCARG(uap, addr);
int nbytes = SCARG(uap, nbytes);
int cacheop = SCARG(uap, cacheop);
return mips_user_cachectl(p, va, nbytes, cacheop);
}
#endif /* __mips */
int
ultrix_sys_exportfs(l, v, retval)
struct lwp *l;
void *v;
register_t *retval;
{
#ifdef notyet
struct ultrix_sys_exportfs_args *uap = v;
#endif
/*
* XXX: should perhaps translate into a mount(2)
* with MOUNT_EXPORT?
*/
return 0;
}
int
ultrix_sys_sigpending(l, v, retval)
struct lwp *l;
void *v;
register_t *retval;
{
struct ultrix_sys_sigpending_args *uap = v;
sigset_t ss;
int mask;
sigpending1(l->l_proc, &ss);
mask = ss.__bits[0];
return (copyout((caddr_t)&mask, (caddr_t)SCARG(uap, mask), sizeof(int)));
}
int
ultrix_sys_sigreturn(l, v, retval)
struct lwp *l;
void *v;
register_t *retval;
{
struct ultrix_sys_sigreturn_args *uap = v;
/* struct sigcontext13 is close enough to Ultrix */
return (compat_13_sys_sigreturn(l, uap, retval));
}
int
ultrix_sys_sigcleanup(l, v, retval)
struct lwp *l;
void *v;
register_t *retval;
{
struct ultrix_sys_sigcleanup_args *uap = v;
/* struct sigcontext13 is close enough to Ultrix */
return (compat_13_sys_sigreturn(l, uap, retval));
}
int
ultrix_sys_sigsuspend(l, v, retval)
struct lwp *l;
void *v;
register_t *retval;
{
struct ultrix_sys_sigsuspend_args *uap = v;
int mask = SCARG(uap, mask);
sigset_t ss;
ss.__bits[0] = mask;
ss.__bits[1] = 0;
ss.__bits[2] = 0;
ss.__bits[3] = 0;
return (sigsuspend1(l->l_proc, &ss));
}
#define ULTRIX_SV_ONSTACK 0x0001 /* take signal on signal stack */
#define ULTRIX_SV_INTERRUPT 0x0002 /* do not restart system on signal return */
#define ULTRIX_SV_OLDSIG 0x1000 /* Emulate old signal() for POSIX */
int
ultrix_sys_sigvec(l, v, retval)
struct lwp *l;
void *v;
register_t *retval;
{
struct ultrix_sys_sigvec_args *uap = v;
struct sigvec nsv, osv;
struct sigaction nsa, osa;
int error;
if (SCARG(uap, nsv)) {
error = copyin(SCARG(uap, nsv), &nsv, sizeof(nsv));
if (error)
return (error);
nsa.sa_handler = nsv.sv_handler;
#if 0 /* documentation */
/* ONSTACK is identical */
nsa.sa_flags = nsv.sv_flags & ULTRIX_SV_ONSTACK;
if ((nsv.sv_flags & ULTRIX_SV_OLDSIG)
/* old signal() - always restart */
|| (!(nsv.sv_flags & ULTRIX_SV_INTERRUPT))
/* inverted meaning (same bit) */
)
nsa.sa_flags |= SA_RESTART;
#else /* optimized - assuming ULTRIX_SV_OLDSIG=>!ULTRIX_SV_INTERRUPT */
nsa.sa_flags = nsv.sv_flags & ~ULTRIX_SV_OLDSIG;
nsa.sa_flags ^= SA_RESTART;
#endif
native_sigset13_to_sigset(&nsv.sv_mask, &nsa.sa_mask);
}
error = sigaction1(l->l_proc, SCARG(uap, signum),
SCARG(uap, nsv) ? &nsa : 0, SCARG(uap, osv) ? &osa : 0,
NULL, 0);
if (error)
return (error);
if (SCARG(uap, osv)) {
osv.sv_handler = osa.sa_handler;
osv.sv_flags = osa.sa_flags ^ SA_RESTART;
osv.sv_flags &= (ULTRIX_SV_ONSTACK | ULTRIX_SV_INTERRUPT);
native_sigset_to_sigset13(&osa.sa_mask, &osv.sv_mask);
error = copyout(&osv, SCARG(uap, osv), sizeof(osv));
if (error)
return (error);
}
return (0);
}
int
ultrix_sys_shmsys(l, v, retval)
struct lwp *l;
void *v;
register_t *retval;
{
#ifdef SYSVSHM
/* Ultrix SVSHM weirndess: */
struct ultrix_sys_shmsys_args *uap = v;
struct sys_shmat_args shmat_args;
struct compat_14_sys_shmctl_args shmctl_args;
struct sys_shmdt_args shmdt_args;
struct sys_shmget_args shmget_args;
switch (SCARG(uap, shmop)) {
case 0: /* Ultrix shmat() */
SCARG(&shmat_args, shmid) = SCARG(uap, a2);
SCARG(&shmat_args, shmaddr) = (void *)SCARG(uap, a3);
SCARG(&shmat_args, shmflg) = SCARG(uap, a4);
return (sys_shmat(l, &shmat_args, retval));
case 1: /* Ultrix shmctl() */
SCARG(&shmctl_args, shmid) = SCARG(uap, a2);
SCARG(&shmctl_args, cmd) = SCARG(uap, a3);
SCARG(&shmctl_args, buf) = (struct shmid_ds14 *)SCARG(uap, a4);
return (compat_14_sys_shmctl(l, &shmctl_args, retval));
case 2: /* Ultrix shmdt() */
SCARG(&shmat_args, shmaddr) = (void *)SCARG(uap, a2);
return (sys_shmdt(l, &shmdt_args, retval));
case 3: /* Ultrix shmget() */
SCARG(&shmget_args, key) = SCARG(uap, a2);
SCARG(&shmget_args, size) = SCARG(uap, a3);
SCARG(&shmget_args, shmflg) = SCARG(uap, a4)
& (IPC_CREAT|IPC_EXCL|IPC_NOWAIT);
return (sys_shmget(l, &shmget_args, retval));
default:
return (EINVAL);
}
#else
return (EOPNOTSUPP);
#endif /* SYSVSHM */
}
static int
ultrix_to_bsd_flock(ufl, fl)
struct ultrix_flock *ufl;
struct flock *fl;
{
fl->l_start = ufl->l_start;
fl->l_len = ufl->l_len;
fl->l_pid = ufl->l_pid;
fl->l_whence = ufl->l_whence;
switch (ufl->l_type) {
case ULTRIX_F_RDLCK:
fl->l_type = F_RDLCK;
break;
case ULTRIX_F_WRLCK:
fl->l_type = F_WRLCK;
break;
case ULTRIX_F_UNLCK:
fl->l_type = F_UNLCK;
break;
default:
return (EINVAL);
}
return (0);
}
static void
bsd_to_ultrix_flock(fl, ufl)
struct flock *fl;
struct ultrix_flock *ufl;
{
ufl->l_start = fl->l_start;
ufl->l_len = fl->l_len;
ufl->l_pid = fl->l_pid;
ufl->l_whence = fl->l_whence;
switch (fl->l_type) {
case F_RDLCK:
ufl->l_type = ULTRIX_F_RDLCK;
break;
case F_WRLCK:
ufl->l_type = ULTRIX_F_WRLCK;
break;
case F_UNLCK:
ufl->l_type = ULTRIX_F_UNLCK;
break;
}
}
int
ultrix_sys_fcntl(l, v, retval)
struct lwp *l;
void *v;
register_t *retval;
{
struct ultrix_sys_fcntl_args *uap = v;
struct proc *p = l->l_proc;
int error;
struct ultrix_flock ufl;
struct flock fl, *flp;
caddr_t sg;
struct sys_fcntl_args *args, fca;
switch (SCARG(uap, cmd)) {
case F_GETLK:
case F_SETLK:
case F_SETLKW:
error = copyin(SCARG(uap, arg), &ufl, sizeof(ufl));
if (error)
return (error);
error = ultrix_to_bsd_flock(&ufl, &fl);
if (error)
return (error);
sg = stackgap_init(p, 0);
flp = (struct flock *)stackgap_alloc(p, &sg, sizeof(*flp));
error = copyout(&fl, flp, sizeof(*flp));
if (error)
return (error);
SCARG(&fca, fd) = SCARG(uap, fd);
SCARG(&fca, cmd) = SCARG(uap, cmd);
SCARG(&fca, arg) = flp;
args = &fca;
break;
default:
args = v;
break;
}
error = sys_fcntl(l, args, retval);
if (error)
return (error);
switch (SCARG(uap, cmd)) {
case F_GETLK:
error = copyin(flp, &fl, sizeof(fl));
if (error)
return (error);
bsd_to_ultrix_flock(&fl, &ufl);
error = copyout(&ufl, SCARG(uap, arg), sizeof(ufl));
break;
}
return (error);
}