summaryrefslogtreecommitdiff
path: root/arch/parisc/hpux/sys_hpux.c
diff options
context:
space:
mode:
Diffstat (limited to 'arch/parisc/hpux/sys_hpux.c')
-rw-r--r--arch/parisc/hpux/sys_hpux.c985
1 files changed, 985 insertions, 0 deletions
diff --git a/arch/parisc/hpux/sys_hpux.c b/arch/parisc/hpux/sys_hpux.c
new file mode 100644
index 000000000000..a64fd48fbfb5
--- /dev/null
+++ b/arch/parisc/hpux/sys_hpux.c
@@ -0,0 +1,985 @@
+/*
+ * Implements HPUX syscalls.
+ *
+ * Copyright (C) 1999 Matthew Wilcox <willy with parisc-linux.org>
+ * Copyright (C) 2000 Philipp Rumpf
+ * Copyright (C) 2000 John Marvin <jsm with parisc-linux.org>
+ * Copyright (C) 2000 Michael Ang <mang with subcarrier.org>
+ * Copyright (C) 2001 Nathan Neulinger <nneul at umr.edu>
+ *
+ * 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, write to the Free Software
+ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
+ */
+
+#include <linux/file.h>
+#include <linux/fs.h>
+#include <linux/namei.h>
+#include <linux/sched.h>
+#include <linux/slab.h>
+#include <linux/smp_lock.h>
+#include <linux/syscalls.h>
+#include <linux/utsname.h>
+#include <linux/vfs.h>
+#include <linux/vmalloc.h>
+
+#include <asm/errno.h>
+#include <asm/pgalloc.h>
+#include <asm/uaccess.h>
+
+unsigned long hpux_brk(unsigned long addr)
+{
+ /* Sigh. Looks like HP/UX libc relies on kernel bugs. */
+ return sys_brk(addr + PAGE_SIZE);
+}
+
+int hpux_sbrk(void)
+{
+ return -ENOSYS;
+}
+
+/* Random other syscalls */
+
+int hpux_nice(int priority_change)
+{
+ return -ENOSYS;
+}
+
+int hpux_ptrace(void)
+{
+ return -ENOSYS;
+}
+
+int hpux_wait(int *stat_loc)
+{
+ return sys_waitpid(-1, stat_loc, 0);
+}
+
+int hpux_setpgrp(void)
+{
+ return sys_setpgid(0,0);
+}
+
+int hpux_setpgrp3(void)
+{
+ return hpux_setpgrp();
+}
+
+#define _SC_CPU_VERSION 10001
+#define _SC_OPEN_MAX 4
+#define CPU_PA_RISC1_1 0x210
+
+int hpux_sysconf(int which)
+{
+ switch (which) {
+ case _SC_CPU_VERSION:
+ return CPU_PA_RISC1_1;
+ case _SC_OPEN_MAX:
+ return INT_MAX;
+ default:
+ return -EINVAL;
+ }
+}
+
+/*****************************************************************************/
+
+#define HPUX_UTSLEN 9
+#define HPUX_SNLEN 15
+
+struct hpux_utsname {
+ char sysname[HPUX_UTSLEN];
+ char nodename[HPUX_UTSLEN];
+ char release[HPUX_UTSLEN];
+ char version[HPUX_UTSLEN];
+ char machine[HPUX_UTSLEN];
+ char idnumber[HPUX_SNLEN];
+} ;
+
+struct hpux_ustat {
+ int32_t f_tfree; /* total free (daddr_t) */
+ u_int32_t f_tinode; /* total inodes free (ino_t) */
+ char f_fname[6]; /* filsys name */
+ char f_fpack[6]; /* filsys pack name */
+ u_int32_t f_blksize; /* filsys block size (int) */
+};
+
+/*
+ * HPUX's utssys() call. It's a collection of miscellaneous functions,
+ * alas, so there's no nice way of splitting them up.
+ */
+
+/* This function is called from hpux_utssys(); HP-UX implements
+ * ustat() as an option to utssys().
+ *
+ * Now, struct ustat on HP-UX is exactly the same as on Linux, except
+ * that it contains one addition field on the end, int32_t f_blksize.
+ * So, we could have written this function to just call the Linux
+ * sys_ustat(), (defined in linux/fs/super.c), and then just
+ * added this additional field to the user's structure. But I figure
+ * if we're gonna be digging through filesystem structures to get
+ * this, we might as well just do the whole enchilada all in one go.
+ *
+ * So, most of this function is almost identical to sys_ustat().
+ * I have placed comments at the few lines changed or added, to
+ * aid in porting forward if and when sys_ustat() is changed from
+ * its form in kernel 2.2.5.
+ */
+static int hpux_ustat(dev_t dev, struct hpux_ustat __user *ubuf)
+{
+ struct super_block *s;
+ struct hpux_ustat tmp; /* Changed to hpux_ustat */
+ struct kstatfs sbuf;
+ int err = -EINVAL;
+
+ s = user_get_super(dev);
+ if (s == NULL)
+ goto out;
+ err = vfs_statfs(s, &sbuf);
+ drop_super(s);
+ if (err)
+ goto out;
+
+ memset(&tmp,0,sizeof(tmp));
+
+ tmp.f_tfree = (int32_t)sbuf.f_bfree;
+ tmp.f_tinode = (u_int32_t)sbuf.f_ffree;
+ tmp.f_blksize = (u_int32_t)sbuf.f_bsize; /* Added this line */
+
+ err = copy_to_user(ubuf, &tmp, sizeof(tmp)) ? -EFAULT : 0;
+out:
+ return err;
+}
+
+/*
+ * Wrapper for hpux statfs call. At the moment, just calls the linux native one
+ * and ignores the extra fields at the end of the hpux statfs struct.
+ *
+ */
+
+typedef int32_t hpux_fsid_t[2]; /* file system ID type */
+typedef uint16_t hpux_site_t;
+
+struct hpux_statfs {
+ int32_t f_type; /* type of info, zero for now */
+ int32_t f_bsize; /* fundamental file system block size */
+ int32_t f_blocks; /* total blocks in file system */
+ int32_t f_bfree; /* free block in fs */
+ int32_t f_bavail; /* free blocks avail to non-superuser */
+ int32_t f_files; /* total file nodes in file system */
+ int32_t f_ffree; /* free file nodes in fs */
+ hpux_fsid_t f_fsid; /* file system ID */
+ int32_t f_magic; /* file system magic number */
+ int32_t f_featurebits; /* file system features */
+ int32_t f_spare[4]; /* spare for later */
+ hpux_site_t f_cnode; /* cluster node where mounted */
+ int16_t f_pad;
+};
+
+static int vfs_statfs_hpux(struct super_block *sb, struct hpux_statfs *buf)
+{
+ struct kstatfs st;
+ int retval;
+
+ retval = vfs_statfs(sb, &st);
+ if (retval)
+ return retval;
+
+ memset(buf, 0, sizeof(*buf));
+ buf->f_type = st.f_type;
+ buf->f_bsize = st.f_bsize;
+ buf->f_blocks = st.f_blocks;
+ buf->f_bfree = st.f_bfree;
+ buf->f_bavail = st.f_bavail;
+ buf->f_files = st.f_files;
+ buf->f_ffree = st.f_ffree;
+ buf->f_fsid[0] = st.f_fsid.val[0];
+ buf->f_fsid[1] = st.f_fsid.val[1];
+
+ return 0;
+}
+
+/* hpux statfs */
+asmlinkage long hpux_statfs(const char __user *path,
+ struct hpux_statfs __user *buf)
+{
+ struct nameidata nd;
+ int error;
+
+ error = user_path_walk(path, &nd);
+ if (!error) {
+ struct hpux_statfs tmp;
+ error = vfs_statfs_hpux(nd.dentry->d_inode->i_sb, &tmp);
+ if (!error && copy_to_user(buf, &tmp, sizeof(tmp)))
+ error = -EFAULT;
+ path_release(&nd);
+ }
+ return error;
+}
+
+asmlinkage long hpux_fstatfs(unsigned int fd, struct hpux_statfs __user * buf)
+{
+ struct file *file;
+ struct hpux_statfs tmp;
+ int error;
+
+ error = -EBADF;
+ file = fget(fd);
+ if (!file)
+ goto out;
+ error = vfs_statfs_hpux(file->f_dentry->d_inode->i_sb, &tmp);
+ if (!error && copy_to_user(buf, &tmp, sizeof(tmp)))
+ error = -EFAULT;
+ fput(file);
+ out:
+ return error;
+}
+
+
+/* This function is called from hpux_utssys(); HP-UX implements
+ * uname() as an option to utssys().
+ *
+ * The form of this function is pretty much copied from sys_olduname(),
+ * defined in linux/arch/i386/kernel/sys_i386.c.
+ */
+/* TODO: Are these put_user calls OK? Should they pass an int?
+ * (I copied it from sys_i386.c like this.)
+ */
+static int hpux_uname(struct hpux_utsname *name)
+{
+ int error;
+
+ if (!name)
+ return -EFAULT;
+ if (!access_ok(VERIFY_WRITE,name,sizeof(struct hpux_utsname)))
+ return -EFAULT;
+
+ down_read(&uts_sem);
+
+ error = __copy_to_user(&name->sysname,&system_utsname.sysname,HPUX_UTSLEN-1);
+ error |= __put_user(0,name->sysname+HPUX_UTSLEN-1);
+ error |= __copy_to_user(&name->nodename,&system_utsname.nodename,HPUX_UTSLEN-1);
+ error |= __put_user(0,name->nodename+HPUX_UTSLEN-1);
+ error |= __copy_to_user(&name->release,&system_utsname.release,HPUX_UTSLEN-1);
+ error |= __put_user(0,name->release+HPUX_UTSLEN-1);
+ error |= __copy_to_user(&name->version,&system_utsname.version,HPUX_UTSLEN-1);
+ error |= __put_user(0,name->version+HPUX_UTSLEN-1);
+ error |= __copy_to_user(&name->machine,&system_utsname.machine,HPUX_UTSLEN-1);
+ error |= __put_user(0,name->machine+HPUX_UTSLEN-1);
+
+ up_read(&uts_sem);
+
+ /* HP-UX utsname has no domainname field. */
+
+ /* TODO: Implement idnumber!!! */
+#if 0
+ error |= __put_user(0,name->idnumber);
+ error |= __put_user(0,name->idnumber+HPUX_SNLEN-1);
+#endif
+
+ error = error ? -EFAULT : 0;
+
+ return error;
+}
+
+/* Note: HP-UX just uses the old suser() function to check perms
+ * in this system call. We'll use capable(CAP_SYS_ADMIN).
+ */
+int hpux_utssys(char *ubuf, int n, int type)
+{
+ int len;
+ int error;
+ switch( type ) {
+ case 0:
+ /* uname(): */
+ return( hpux_uname( (struct hpux_utsname *)ubuf ) );
+ break ;
+ case 1:
+ /* Obsolete (used to be umask().) */
+ return -EFAULT ;
+ break ;
+ case 2:
+ /* ustat(): */
+ return( hpux_ustat(new_decode_dev(n), (struct hpux_ustat *)ubuf) );
+ break ;
+ case 3:
+ /* setuname():
+ *
+ * On linux (unlike HP-UX), utsname.nodename
+ * is the same as the hostname.
+ *
+ * sys_sethostname() is defined in linux/kernel/sys.c.
+ */
+ if (!capable(CAP_SYS_ADMIN))
+ return -EPERM;
+ /* Unlike Linux, HP-UX returns an error if n==0: */
+ if ( n <= 0 )
+ return -EINVAL ;
+ /* Unlike Linux, HP-UX truncates it if n is too big: */
+ len = (n <= __NEW_UTS_LEN) ? n : __NEW_UTS_LEN ;
+ return( sys_sethostname(ubuf, len) );
+ break ;
+ case 4:
+ /* sethostname():
+ *
+ * sys_sethostname() is defined in linux/kernel/sys.c.
+ */
+ if (!capable(CAP_SYS_ADMIN))
+ return -EPERM;
+ /* Unlike Linux, HP-UX returns an error if n==0: */
+ if ( n <= 0 )
+ return -EINVAL ;
+ /* Unlike Linux, HP-UX truncates it if n is too big: */
+ len = (n <= __NEW_UTS_LEN) ? n : __NEW_UTS_LEN ;
+ return( sys_sethostname(ubuf, len) );
+ break ;
+ case 5:
+ /* gethostname():
+ *
+ * sys_gethostname() is defined in linux/kernel/sys.c.
+ */
+ /* Unlike Linux, HP-UX returns an error if n==0: */
+ if ( n <= 0 )
+ return -EINVAL ;
+ return( sys_gethostname(ubuf, n) );
+ break ;
+ case 6:
+ /* Supposedly called from setuname() in libc.
+ * TODO: When and why is this called?
+ * Is it ever even called?
+ *
+ * This code should look a lot like sys_sethostname(),
+ * defined in linux/kernel/sys.c. If that gets updated,
+ * update this code similarly.
+ */
+ if (!capable(CAP_SYS_ADMIN))
+ return -EPERM;
+ /* Unlike Linux, HP-UX returns an error if n==0: */
+ if ( n <= 0 )
+ return -EINVAL ;
+ /* Unlike Linux, HP-UX truncates it if n is too big: */
+ len = (n <= __NEW_UTS_LEN) ? n : __NEW_UTS_LEN ;
+ /**/
+ /* TODO: print a warning about using this? */
+ down_write(&uts_sem);
+ error = -EFAULT;
+ if (!copy_from_user(system_utsname.sysname, ubuf, len)) {
+ system_utsname.sysname[len] = 0;
+ error = 0;
+ }
+ up_write(&uts_sem);
+ return error;
+ break ;
+ case 7:
+ /* Sets utsname.release, if you're allowed.
+ * Undocumented. Used by swinstall to change the
+ * OS version, during OS updates. Yuck!!!
+ *
+ * This code should look a lot like sys_sethostname()
+ * in linux/kernel/sys.c. If that gets updated, update
+ * this code similarly.
+ */
+ if (!capable(CAP_SYS_ADMIN))
+ return -EPERM;
+ /* Unlike Linux, HP-UX returns an error if n==0: */
+ if ( n <= 0 )
+ return -EINVAL ;
+ /* Unlike Linux, HP-UX truncates it if n is too big: */
+ len = (n <= __NEW_UTS_LEN) ? n : __NEW_UTS_LEN ;
+ /**/
+ /* TODO: print a warning about this? */
+ down_write(&uts_sem);
+ error = -EFAULT;
+ if (!copy_from_user(system_utsname.release, ubuf, len)) {
+ system_utsname.release[len] = 0;
+ error = 0;
+ }
+ up_write(&uts_sem);
+ return error;
+ break ;
+ default:
+ /* This system call returns -EFAULT if given an unknown type.
+ * Why not -EINVAL? I don't know, it's just not what they did.
+ */
+ return -EFAULT ;
+ }
+}
+
+int hpux_getdomainname(char *name, int len)
+{
+ int nlen;
+ int err = -EFAULT;
+
+ down_read(&uts_sem);
+
+ nlen = strlen(system_utsname.domainname) + 1;
+
+ if (nlen < len)
+ len = nlen;
+ if(len > __NEW_UTS_LEN)
+ goto done;
+ if(copy_to_user(name, system_utsname.domainname, len))
+ goto done;
+ err = 0;
+done:
+ up_read(&uts_sem);
+ return err;
+
+}
+
+int hpux_pipe(int *kstack_fildes)
+{
+ int error;
+
+ lock_kernel();
+ error = do_pipe(kstack_fildes);
+ unlock_kernel();
+ return error;
+}
+
+/* lies - says it works, but it really didn't lock anything */
+int hpux_lockf(int fildes, int function, off_t size)
+{
+ return 0;
+}
+
+int hpux_sysfs(int opcode, unsigned long arg1, unsigned long arg2)
+{
+ char *fsname = NULL;
+ int len = 0;
+ int fstype;
+
+/*Unimplemented HP-UX syscall emulation. Syscall #334 (sysfs)
+ Args: 1 80057bf4 0 400179f0 0 0 0 */
+ printk(KERN_DEBUG "in hpux_sysfs\n");
+ printk(KERN_DEBUG "hpux_sysfs called with opcode = %d\n", opcode);
+ printk(KERN_DEBUG "hpux_sysfs called with arg1='%lx'\n", arg1);
+
+ if ( opcode == 1 ) { /* GETFSIND */
+ len = strlen_user((char *)arg1);
+ printk(KERN_DEBUG "len of arg1 = %d\n", len);
+
+ fsname = (char *) kmalloc(len+1, GFP_KERNEL);
+ if ( !fsname ) {
+ printk(KERN_DEBUG "failed to kmalloc fsname\n");
+ return 0;
+ }
+
+ if ( copy_from_user(fsname, (char *)arg1, len+1) ) {
+ printk(KERN_DEBUG "failed to copy_from_user fsname\n");
+ kfree(fsname);
+ return 0;
+ }
+
+ printk(KERN_DEBUG "that is '%s' as (char *)\n", fsname);
+ if ( !strcmp(fsname, "hfs") ) {
+ fstype = 0;
+ } else {
+ fstype = 0;
+ };
+
+ kfree(fsname);
+
+ printk(KERN_DEBUG "returning fstype=%d\n", fstype);
+ return fstype; /* something other than default */
+ }
+
+
+ return 0;
+}
+
+
+/* Table of syscall names and handle for unimplemented routines */
+static const char *syscall_names[] = {
+ "nosys", /* 0 */
+ "exit",
+ "fork",
+ "read",
+ "write",
+ "open", /* 5 */
+ "close",
+ "wait",
+ "creat",
+ "link",
+ "unlink", /* 10 */
+ "execv",
+ "chdir",
+ "time",
+ "mknod",
+ "chmod", /* 15 */
+ "chown",
+ "brk",
+ "lchmod",
+ "lseek",
+ "getpid", /* 20 */
+ "mount",
+ "umount",
+ "setuid",
+ "getuid",
+ "stime", /* 25 */
+ "ptrace",
+ "alarm",
+ NULL,
+ "pause",
+ "utime", /* 30 */
+ "stty",
+ "gtty",
+ "access",
+ "nice",
+ "ftime", /* 35 */
+ "sync",
+ "kill",
+ "stat",
+ "setpgrp3",
+ "lstat", /* 40 */
+ "dup",
+ "pipe",
+ "times",
+ "profil",
+ "ki_call", /* 45 */
+ "setgid",
+ "getgid",
+ NULL,
+ NULL,
+ NULL, /* 50 */
+ "acct",
+ "set_userthreadid",
+ NULL,
+ "ioctl",
+ "reboot", /* 55 */
+ "symlink",
+ "utssys",
+ "readlink",
+ "execve",
+ "umask", /* 60 */
+ "chroot",
+ "fcntl",
+ "ulimit",
+ NULL,
+ NULL, /* 65 */
+ "vfork",
+ NULL,
+ NULL,
+ NULL,
+ NULL, /* 70 */
+ "mmap",
+ NULL,
+ "munmap",
+ "mprotect",
+ "madvise", /* 75 */
+ "vhangup",
+ "swapoff",
+ NULL,
+ "getgroups",
+ "setgroups", /* 80 */
+ "getpgrp2",
+ "setpgid/setpgrp2",
+ "setitimer",
+ "wait3",
+ "swapon", /* 85 */
+ "getitimer",
+ NULL,
+ NULL,
+ NULL,
+ "dup2", /* 90 */
+ NULL,
+ "fstat",
+ "select",
+ NULL,
+ "fsync", /* 95 */
+ "setpriority",
+ NULL,
+ NULL,
+ NULL,
+ "getpriority", /* 100 */
+ NULL,
+ NULL,
+ NULL,
+ NULL,
+ NULL, /* 105 */
+ NULL,
+ NULL,
+ "sigvector",
+ "sigblock",
+ "sigsetmask", /* 110 */
+ "sigpause",
+ "sigstack",
+ NULL,
+ NULL,
+ NULL, /* 115 */
+ "gettimeofday",
+ "getrusage",
+ NULL,
+ NULL,
+ "readv", /* 120 */
+ "writev",
+ "settimeofday",
+ "fchown",
+ "fchmod",
+ NULL, /* 125 */
+ "setresuid",
+ "setresgid",
+ "rename",
+ "truncate",
+ "ftruncate", /* 130 */
+ NULL,
+ "sysconf",
+ NULL,
+ NULL,
+ NULL, /* 135 */
+ "mkdir",
+ "rmdir",
+ NULL,
+ "sigcleanup",
+ "setcore", /* 140 */
+ NULL,
+ "gethostid",
+ "sethostid",
+ "getrlimit",
+ "setrlimit", /* 145 */
+ NULL,
+ NULL,
+ "quotactl",
+ "get_sysinfo",
+ NULL, /* 150 */
+ "privgrp",
+ "rtprio",
+ "plock",
+ NULL,
+ "lockf", /* 155 */
+ "semget",
+ NULL,
+ "semop",
+ "msgget",
+ NULL, /* 160 */
+ "msgsnd",
+ "msgrcv",
+ "shmget",
+ NULL,
+ "shmat", /* 165 */
+ "shmdt",
+ NULL,
+ "csp/nsp_init",
+ "cluster",
+ "mkrnod", /* 170 */
+ "test",
+ "unsp_open",
+ NULL,
+ "getcontext",
+ "osetcontext", /* 175 */
+ "bigio",
+ "pipenode",
+ "lsync",
+ "getmachineid",
+ "cnodeid/mysite", /* 180 */
+ "cnodes/sitels",
+ "swapclients",
+ "rmtprocess",
+ "dskless_stats",
+ "sigprocmask", /* 185 */
+ "sigpending",
+ "sigsuspend",
+ "sigaction",
+ NULL,
+ "nfssvc", /* 190 */
+ "getfh",
+ "getdomainname",
+ "setdomainname",
+ "async_daemon",
+ "getdirentries", /* 195 */
+ NULL,
+ NULL,
+ "vfsmount",
+ NULL,
+ "waitpid", /* 200 */
+ NULL,
+ NULL,
+ NULL,
+ NULL,
+ NULL, /* 205 */
+ NULL,
+ NULL,
+ NULL,
+ NULL,
+ NULL, /* 210 */
+ NULL,
+ NULL,
+ NULL,
+ NULL,
+ NULL, /* 215 */
+ NULL,
+ NULL,
+ NULL,
+ NULL,
+ NULL, /* 220 */
+ NULL,
+ NULL,
+ NULL,
+ "sigsetreturn",
+ "sigsetstatemask", /* 225 */
+ "bfactl",
+ "cs",
+ "cds",
+ NULL,
+ "pathconf", /* 230 */
+ "fpathconf",
+ NULL,
+ NULL,
+ "nfs_fcntl",
+ "ogetacl", /* 235 */
+ "ofgetacl",
+ "osetacl",
+ "ofsetacl",
+ "pstat",
+ "getaudid", /* 240 */
+ "setaudid",
+ "getaudproc",
+ "setaudproc",
+ "getevent",
+ "setevent", /* 245 */
+ "audwrite",
+ "audswitch",
+ "audctl",
+ "ogetaccess",
+ "fsctl", /* 250 */
+ "ulconnect",
+ "ulcontrol",
+ "ulcreate",
+ "uldest",
+ "ulrecv", /* 255 */
+ "ulrecvcn",
+ "ulsend",
+ "ulshutdown",
+ "swapfs",
+ "fss", /* 260 */
+ NULL,
+ NULL,
+ NULL,
+ NULL,
+ NULL, /* 265 */
+ NULL,
+ "tsync",
+ "getnumfds",
+ "poll",
+ "getmsg", /* 270 */
+ "putmsg",
+ "fchdir",
+ "getmount_cnt",
+ "getmount_entry",
+ "accept", /* 275 */
+ "bind",
+ "connect",
+ "getpeername",
+ "getsockname",
+ "getsockopt", /* 280 */
+ "listen",
+ "recv",
+ "recvfrom",
+ "recvmsg",
+ "send", /* 285 */
+ "sendmsg",
+ "sendto",
+ "setsockopt",
+ "shutdown",
+ "socket", /* 290 */
+ "socketpair",
+ "proc_open",
+ "proc_close",
+ "proc_send",
+ "proc_recv", /* 295 */
+ "proc_sendrecv",
+ "proc_syscall",
+ "ipccreate",
+ "ipcname",
+ "ipcnamerase", /* 300 */
+ "ipclookup",
+ "ipcselect",
+ "ipcconnect",
+ "ipcrecvcn",
+ "ipcsend", /* 305 */
+ "ipcrecv",
+ "ipcgetnodename",
+ "ipcsetnodename",
+ "ipccontrol",
+ "ipcshutdown", /* 310 */
+ "ipcdest",
+ "semctl",
+ "msgctl",
+ "shmctl",
+ "mpctl", /* 315 */
+ "exportfs",
+ "getpmsg",
+ "putpmsg",
+ "strioctl",
+ "msync", /* 320 */
+ "msleep",
+ "mwakeup",
+ "msem_init",
+ "msem_remove",
+ "adjtime", /* 325 */
+ "kload",
+ "fattach",
+ "fdetach",
+ "serialize",
+ "statvfs", /* 330 */
+ "fstatvfs",
+ "lchown",
+ "getsid",
+ "sysfs",
+ NULL, /* 335 */
+ NULL,
+ "sched_setparam",
+ "sched_getparam",
+ "sched_setscheduler",
+ "sched_getscheduler", /* 340 */
+ "sched_yield",
+ "sched_get_priority_max",
+ "sched_get_priority_min",
+ "sched_rr_get_interval",
+ "clock_settime", /* 345 */
+ "clock_gettime",
+ "clock_getres",
+ "timer_create",
+ "timer_delete",
+ "timer_settime", /* 350 */
+ "timer_gettime",
+ "timer_getoverrun",
+ "nanosleep",
+ "toolbox",
+ NULL, /* 355 */
+ "getdents",
+ "getcontext",
+ "sysinfo",
+ "fcntl64",
+ "ftruncate64", /* 360 */
+ "fstat64",
+ "getdirentries64",
+ "getrlimit64",
+ "lockf64",
+ "lseek64", /* 365 */
+ "lstat64",
+ "mmap64",
+ "setrlimit64",
+ "stat64",
+ "truncate64", /* 370 */
+ "ulimit64",
+ NULL,
+ NULL,
+ NULL,
+ NULL, /* 375 */
+ NULL,
+ NULL,
+ NULL,
+ NULL,
+ "setcontext", /* 380 */
+ "sigaltstack",
+ "waitid",
+ "setpgrp",
+ "recvmsg2",
+ "sendmsg2", /* 385 */
+ "socket2",
+ "socketpair2",
+ "setregid",
+ "lwp_create",
+ "lwp_terminate", /* 390 */
+ "lwp_wait",
+ "lwp_suspend",
+ "lwp_resume",
+ "lwp_self",
+ "lwp_abort_syscall", /* 395 */
+ "lwp_info",
+ "lwp_kill",
+ "ksleep",
+ "kwakeup",
+ "ksleep_abort", /* 400 */
+ "lwp_proc_info",
+ "lwp_exit",
+ "lwp_continue",
+ "getacl",
+ "fgetacl", /* 405 */
+ "setacl",
+ "fsetacl",
+ "getaccess",
+ "lwp_mutex_init",
+ "lwp_mutex_lock_sys", /* 410 */
+ "lwp_mutex_unlock",
+ "lwp_cond_init",
+ "lwp_cond_signal",
+ "lwp_cond_broadcast",
+ "lwp_cond_wait_sys", /* 415 */
+ "lwp_getscheduler",
+ "lwp_setscheduler",
+ "lwp_getprivate",
+ "lwp_setprivate",
+ "lwp_detach", /* 420 */
+ "mlock",
+ "munlock",
+ "mlockall",
+ "munlockall",
+ "shm_open", /* 425 */
+ "shm_unlink",
+ "sigqueue",
+ "sigwaitinfo",
+ "sigtimedwait",
+ "sigwait", /* 430 */
+ "aio_read",
+ "aio_write",
+ "lio_listio",
+ "aio_error",
+ "aio_return", /* 435 */
+ "aio_cancel",
+ "aio_suspend",
+ "aio_fsync",
+ "mq_open",
+ "mq_unlink", /* 440 */
+ "mq_send",
+ "mq_receive",
+ "mq_notify",
+ "mq_setattr",
+ "mq_getattr", /* 445 */
+ "ksem_open",
+ "ksem_unlink",
+ "ksem_close",
+ "ksem_destroy",
+ "lw_sem_incr", /* 450 */
+ "lw_sem_decr",
+ "lw_sem_read",
+ "mq_close",
+};
+static const int syscall_names_max = 453;
+
+int
+hpux_unimplemented(unsigned long arg1,unsigned long arg2,unsigned long arg3,
+ unsigned long arg4,unsigned long arg5,unsigned long arg6,
+ unsigned long arg7,unsigned long sc_num)
+{
+ /* NOTE: sc_num trashes arg8 for the few syscalls that actually
+ * have a valid 8th argument.
+ */
+ const char *name = NULL;
+ if ( sc_num <= syscall_names_max && sc_num >= 0 ) {
+ name = syscall_names[sc_num];
+ }
+
+ if ( name ) {
+ printk(KERN_DEBUG "Unimplemented HP-UX syscall emulation. Syscall #%lu (%s)\n",
+ sc_num, name);
+ } else {
+ printk(KERN_DEBUG "Unimplemented unknown HP-UX syscall emulation. Syscall #%lu\n",
+ sc_num);
+ }
+
+ printk(KERN_DEBUG " Args: %lx %lx %lx %lx %lx %lx %lx\n",
+ arg1, arg2, arg3, arg4, arg5, arg6, arg7);
+
+ return -ENOSYS;
+}