kernel_optimize_test/arch/parisc/kernel/sys_parisc32.c

612 lines
14 KiB
C
Raw Normal View History

/*
* sys_parisc32.c: Conversion between 32bit and 64bit native syscalls.
*
* Copyright (C) 2000-2001 Hewlett Packard Company
* Copyright (C) 2000 John Marvin
* Copyright (C) 2001 Matthew Wilcox
*
* These routines maintain argument size conversion between 32bit and 64bit
* environment. Based heavily on sys_ia32.c and sys_sparc32.c.
*/
#include <linux/compat.h>
#include <linux/kernel.h>
#include <linux/sched.h>
#include <linux/fs.h>
#include <linux/mm.h>
#include <linux/file.h>
#include <linux/signal.h>
#include <linux/resource.h>
#include <linux/times.h>
#include <linux/utsname.h>
#include <linux/time.h>
#include <linux/smp.h>
#include <linux/smp_lock.h>
#include <linux/sem.h>
#include <linux/msg.h>
#include <linux/shm.h>
#include <linux/slab.h>
#include <linux/uio.h>
#include <linux/nfs_fs.h>
#include <linux/ncp_fs.h>
#include <linux/sunrpc/svc.h>
#include <linux/nfsd/nfsd.h>
#include <linux/nfsd/cache.h>
#include <linux/nfsd/xdr.h>
#include <linux/nfsd/syscall.h>
#include <linux/poll.h>
#include <linux/personality.h>
#include <linux/stat.h>
#include <linux/highmem.h>
#include <linux/highuid.h>
#include <linux/mman.h>
#include <linux/binfmts.h>
#include <linux/namei.h>
#include <linux/vfs.h>
#include <linux/ptrace.h>
#include <linux/swap.h>
#include <linux/syscalls.h>
#include <asm/types.h>
#include <asm/uaccess.h>
#include <asm/semaphore.h>
#include <asm/mmu_context.h>
#include "sys32.h"
#undef DEBUG
#ifdef DEBUG
#define DBG(x) printk x
#else
#define DBG(x)
#endif
/*
* sys32_execve() executes a new program.
*/
asmlinkage int sys32_execve(struct pt_regs *regs)
{
int error;
char *filename;
DBG(("sys32_execve(%p) r26 = 0x%lx\n", regs, regs->gr[26]));
filename = getname((const char __user *) regs->gr[26]);
error = PTR_ERR(filename);
if (IS_ERR(filename))
goto out;
error = compat_do_execve(filename, compat_ptr(regs->gr[25]),
compat_ptr(regs->gr[24]), regs);
if (error == 0) {
task_lock(current);
current->ptrace &= ~PT_DTRACE;
task_unlock(current);
}
putname(filename);
out:
return error;
}
asmlinkage long sys32_unimplemented(int r26, int r25, int r24, int r23,
int r22, int r21, int r20)
{
printk(KERN_ERR "%s(%d): Unimplemented 32 on 64 syscall #%d!\n",
current->comm, current->pid, r20);
return -ENOSYS;
}
#ifdef CONFIG_SYSCTL
struct __sysctl_args32 {
u32 name;
int nlen;
u32 oldval;
u32 oldlenp;
u32 newval;
u32 newlen;
u32 __unused[4];
};
asmlinkage long sys32_sysctl(struct __sysctl_args32 __user *args)
{
#ifndef CONFIG_SYSCTL_SYSCALL
return -ENOSYS;
#else
struct __sysctl_args32 tmp;
int error;
unsigned int oldlen32;
size_t oldlen, __user *oldlenp = NULL;
unsigned long addr = (((long __force)&args->__unused[0]) + 7) & ~7;
DBG(("sysctl32(%p)\n", args));
if (copy_from_user(&tmp, args, sizeof(tmp)))
return -EFAULT;
if (tmp.oldval && tmp.oldlenp) {
/* Duh, this is ugly and might not work if sysctl_args
is in read-only memory, but do_sysctl does indirectly
a lot of uaccess in both directions and we'd have to
basically copy the whole sysctl.c here, and
glibc's __sysctl uses rw memory for the structure
anyway. */
/* a possibly better hack than this, which will avoid the
* problem if the struct is read only, is to push the
* 'oldlen' value out to the user's stack instead. -PB
*/
if (get_user(oldlen32, (u32 *)(u64)tmp.oldlenp))
return -EFAULT;
oldlen = oldlen32;
if (put_user(oldlen, (size_t *)addr))
return -EFAULT;
oldlenp = (size_t *)addr;
}
lock_kernel();
error = do_sysctl((int __user *)(u64)tmp.name, tmp.nlen,
(void __user *)(u64)tmp.oldval, oldlenp,
(void __user *)(u64)tmp.newval, tmp.newlen);
unlock_kernel();
if (oldlenp) {
if (!error) {
if (get_user(oldlen, (size_t *)addr)) {
error = -EFAULT;
} else {
oldlen32 = oldlen;
if (put_user(oldlen32, (u32 *)(u64)tmp.oldlenp))
error = -EFAULT;
}
}
if (copy_to_user(args->__unused, tmp.__unused, sizeof(tmp.__unused)))
error = -EFAULT;
}
return error;
#endif
}
#endif /* CONFIG_SYSCTL */
asmlinkage long sys32_sched_rr_get_interval(pid_t pid,
struct compat_timespec __user *interval)
{
struct timespec t;
int ret;
KERNEL_SYSCALL(ret, sys_sched_rr_get_interval, pid, (struct timespec __user *)&t);
if (put_compat_timespec(&t, interval))
return -EFAULT;
return ret;
}
static int
put_compat_timeval(struct compat_timeval __user *u, struct timeval *t)
{
struct compat_timeval t32;
t32.tv_sec = t->tv_sec;
t32.tv_usec = t->tv_usec;
return copy_to_user(u, &t32, sizeof t32);
}
static inline long get_ts32(struct timespec *o, struct compat_timeval __user *i)
{
long usec;
if (__get_user(o->tv_sec, &i->tv_sec))
return -EFAULT;
if (__get_user(usec, &i->tv_usec))
return -EFAULT;
o->tv_nsec = usec * 1000;
return 0;
}
asmlinkage int
sys32_gettimeofday(struct compat_timeval __user *tv, struct timezone __user *tz)
{
extern void do_gettimeofday(struct timeval *tv);
if (tv) {
struct timeval ktv;
do_gettimeofday(&ktv);
if (put_compat_timeval(tv, &ktv))
return -EFAULT;
}
if (tz) {
extern struct timezone sys_tz;
if (copy_to_user(tz, &sys_tz, sizeof(sys_tz)))
return -EFAULT;
}
return 0;
}
asmlinkage
int sys32_settimeofday(struct compat_timeval __user *tv, struct timezone __user *tz)
{
struct timespec kts;
struct timezone ktz;
if (tv) {
if (get_ts32(&kts, tv))
return -EFAULT;
}
if (tz) {
if (copy_from_user(&ktz, tz, sizeof(ktz)))
return -EFAULT;
}
return do_sys_settimeofday(tv ? &kts : NULL, tz ? &ktz : NULL);
}
int cp_compat_stat(struct kstat *stat, struct compat_stat __user *statbuf)
{
[PATCH] VFS: Make filldir_t and struct kstat deal in 64-bit inode numbers These patches make the kernel pass 64-bit inode numbers internally when communicating to userspace, even on a 32-bit system. They are required because some filesystems have intrinsic 64-bit inode numbers: NFS3+ and XFS for example. The 64-bit inode numbers are then propagated to userspace automatically where the arch supports it. Problems have been seen with userspace (eg: ld.so) using the 64-bit inode number returned by stat64() or getdents64() to differentiate files, and failing because the 64-bit inode number space was compressed to 32-bits, and so overlaps occur. This patch: Make filldir_t take a 64-bit inode number and struct kstat carry a 64-bit inode number so that 64-bit inode numbers can be passed back to userspace. The stat functions then returns the full 64-bit inode number where available and where possible. If it is not possible to represent the inode number supplied by the filesystem in the field provided by userspace, then error EOVERFLOW will be issued. Similarly, the getdents/readdir functions now pass the full 64-bit inode number to userspace where possible, returning EOVERFLOW instead when a directory entry is encountered that can't be properly represented. Note that this means that some inodes will not be stat'able on a 32-bit system with old libraries where they were before - but it does mean that there will be no ambiguity over what a 32-bit inode number refers to. Note similarly that directory scans may be cut short with an error on a 32-bit system with old libraries where the scan would work before for the same reasons. It is judged unlikely that this situation will occur because modern glibc uses 64-bit capable versions of stat and getdents class functions exclusively, and that older systems are unlikely to encounter unrepresentable inode numbers anyway. [akpm: alpha build fix] Signed-off-by: David Howells <dhowells@redhat.com> Cc: Trond Myklebust <trond.myklebust@fys.uio.no> Cc: Al Viro <viro@zeniv.linux.org.uk> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-10-03 16:13:46 +08:00
compat_ino_t ino;
int err;
if (stat->size > MAX_NON_LFS || !new_valid_dev(stat->dev) ||
!new_valid_dev(stat->rdev))
return -EOVERFLOW;
[PATCH] VFS: Make filldir_t and struct kstat deal in 64-bit inode numbers These patches make the kernel pass 64-bit inode numbers internally when communicating to userspace, even on a 32-bit system. They are required because some filesystems have intrinsic 64-bit inode numbers: NFS3+ and XFS for example. The 64-bit inode numbers are then propagated to userspace automatically where the arch supports it. Problems have been seen with userspace (eg: ld.so) using the 64-bit inode number returned by stat64() or getdents64() to differentiate files, and failing because the 64-bit inode number space was compressed to 32-bits, and so overlaps occur. This patch: Make filldir_t take a 64-bit inode number and struct kstat carry a 64-bit inode number so that 64-bit inode numbers can be passed back to userspace. The stat functions then returns the full 64-bit inode number where available and where possible. If it is not possible to represent the inode number supplied by the filesystem in the field provided by userspace, then error EOVERFLOW will be issued. Similarly, the getdents/readdir functions now pass the full 64-bit inode number to userspace where possible, returning EOVERFLOW instead when a directory entry is encountered that can't be properly represented. Note that this means that some inodes will not be stat'able on a 32-bit system with old libraries where they were before - but it does mean that there will be no ambiguity over what a 32-bit inode number refers to. Note similarly that directory scans may be cut short with an error on a 32-bit system with old libraries where the scan would work before for the same reasons. It is judged unlikely that this situation will occur because modern glibc uses 64-bit capable versions of stat and getdents class functions exclusively, and that older systems are unlikely to encounter unrepresentable inode numbers anyway. [akpm: alpha build fix] Signed-off-by: David Howells <dhowells@redhat.com> Cc: Trond Myklebust <trond.myklebust@fys.uio.no> Cc: Al Viro <viro@zeniv.linux.org.uk> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-10-03 16:13:46 +08:00
ino = stat->ino;
if (sizeof(ino) < sizeof(stat->ino) && ino != stat->ino)
return -EOVERFLOW;
err = put_user(new_encode_dev(stat->dev), &statbuf->st_dev);
[PATCH] VFS: Make filldir_t and struct kstat deal in 64-bit inode numbers These patches make the kernel pass 64-bit inode numbers internally when communicating to userspace, even on a 32-bit system. They are required because some filesystems have intrinsic 64-bit inode numbers: NFS3+ and XFS for example. The 64-bit inode numbers are then propagated to userspace automatically where the arch supports it. Problems have been seen with userspace (eg: ld.so) using the 64-bit inode number returned by stat64() or getdents64() to differentiate files, and failing because the 64-bit inode number space was compressed to 32-bits, and so overlaps occur. This patch: Make filldir_t take a 64-bit inode number and struct kstat carry a 64-bit inode number so that 64-bit inode numbers can be passed back to userspace. The stat functions then returns the full 64-bit inode number where available and where possible. If it is not possible to represent the inode number supplied by the filesystem in the field provided by userspace, then error EOVERFLOW will be issued. Similarly, the getdents/readdir functions now pass the full 64-bit inode number to userspace where possible, returning EOVERFLOW instead when a directory entry is encountered that can't be properly represented. Note that this means that some inodes will not be stat'able on a 32-bit system with old libraries where they were before - but it does mean that there will be no ambiguity over what a 32-bit inode number refers to. Note similarly that directory scans may be cut short with an error on a 32-bit system with old libraries where the scan would work before for the same reasons. It is judged unlikely that this situation will occur because modern glibc uses 64-bit capable versions of stat and getdents class functions exclusively, and that older systems are unlikely to encounter unrepresentable inode numbers anyway. [akpm: alpha build fix] Signed-off-by: David Howells <dhowells@redhat.com> Cc: Trond Myklebust <trond.myklebust@fys.uio.no> Cc: Al Viro <viro@zeniv.linux.org.uk> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-10-03 16:13:46 +08:00
err |= put_user(ino, &statbuf->st_ino);
err |= put_user(stat->mode, &statbuf->st_mode);
err |= put_user(stat->nlink, &statbuf->st_nlink);
err |= put_user(0, &statbuf->st_reserved1);
err |= put_user(0, &statbuf->st_reserved2);
err |= put_user(new_encode_dev(stat->rdev), &statbuf->st_rdev);
err |= put_user(stat->size, &statbuf->st_size);
err |= put_user(stat->atime.tv_sec, &statbuf->st_atime);
err |= put_user(stat->atime.tv_nsec, &statbuf->st_atime_nsec);
err |= put_user(stat->mtime.tv_sec, &statbuf->st_mtime);
err |= put_user(stat->mtime.tv_nsec, &statbuf->st_mtime_nsec);
err |= put_user(stat->ctime.tv_sec, &statbuf->st_ctime);
err |= put_user(stat->ctime.tv_nsec, &statbuf->st_ctime_nsec);
err |= put_user(stat->blksize, &statbuf->st_blksize);
err |= put_user(stat->blocks, &statbuf->st_blocks);
err |= put_user(0, &statbuf->__unused1);
err |= put_user(0, &statbuf->__unused2);
err |= put_user(0, &statbuf->__unused3);
err |= put_user(0, &statbuf->__unused4);
err |= put_user(0, &statbuf->__unused5);
err |= put_user(0, &statbuf->st_fstype); /* not avail */
err |= put_user(0, &statbuf->st_realdev); /* not avail */
err |= put_user(0, &statbuf->st_basemode); /* not avail */
err |= put_user(0, &statbuf->st_spareshort);
err |= put_user(stat->uid, &statbuf->st_uid);
err |= put_user(stat->gid, &statbuf->st_gid);
err |= put_user(0, &statbuf->st_spare4[0]);
err |= put_user(0, &statbuf->st_spare4[1]);
err |= put_user(0, &statbuf->st_spare4[2]);
return err;
}
struct linux32_dirent {
u32 d_ino;
compat_off_t d_off;
u16 d_reclen;
char d_name[1];
};
struct old_linux32_dirent {
u32 d_ino;
u32 d_offset;
u16 d_namlen;
char d_name[1];
};
struct getdents32_callback {
struct linux32_dirent __user * current_dir;
struct linux32_dirent __user * previous;
int count;
int error;
};
struct readdir32_callback {
struct old_linux32_dirent __user * dirent;
int count;
};
#define ROUND_UP(x,a) ((__typeof__(x))(((unsigned long)(x) + ((a) - 1)) & ~((a) - 1)))
#define NAME_OFFSET(de) ((int) ((de)->d_name - (char __user *) (de)))
static int filldir32 (void *__buf, const char *name, int namlen,
loff_t offset, u64 ino, unsigned int d_type)
{
struct linux32_dirent __user * dirent;
struct getdents32_callback * buf = (struct getdents32_callback *) __buf;
int reclen = ROUND_UP(NAME_OFFSET(dirent) + namlen + 1, 4);
[PATCH] VFS: Make filldir_t and struct kstat deal in 64-bit inode numbers These patches make the kernel pass 64-bit inode numbers internally when communicating to userspace, even on a 32-bit system. They are required because some filesystems have intrinsic 64-bit inode numbers: NFS3+ and XFS for example. The 64-bit inode numbers are then propagated to userspace automatically where the arch supports it. Problems have been seen with userspace (eg: ld.so) using the 64-bit inode number returned by stat64() or getdents64() to differentiate files, and failing because the 64-bit inode number space was compressed to 32-bits, and so overlaps occur. This patch: Make filldir_t take a 64-bit inode number and struct kstat carry a 64-bit inode number so that 64-bit inode numbers can be passed back to userspace. The stat functions then returns the full 64-bit inode number where available and where possible. If it is not possible to represent the inode number supplied by the filesystem in the field provided by userspace, then error EOVERFLOW will be issued. Similarly, the getdents/readdir functions now pass the full 64-bit inode number to userspace where possible, returning EOVERFLOW instead when a directory entry is encountered that can't be properly represented. Note that this means that some inodes will not be stat'able on a 32-bit system with old libraries where they were before - but it does mean that there will be no ambiguity over what a 32-bit inode number refers to. Note similarly that directory scans may be cut short with an error on a 32-bit system with old libraries where the scan would work before for the same reasons. It is judged unlikely that this situation will occur because modern glibc uses 64-bit capable versions of stat and getdents class functions exclusively, and that older systems are unlikely to encounter unrepresentable inode numbers anyway. [akpm: alpha build fix] Signed-off-by: David Howells <dhowells@redhat.com> Cc: Trond Myklebust <trond.myklebust@fys.uio.no> Cc: Al Viro <viro@zeniv.linux.org.uk> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-10-03 16:13:46 +08:00
u32 d_ino;
buf->error = -EINVAL; /* only used if we fail.. */
if (reclen > buf->count)
return -EINVAL;
[PATCH] VFS: Make filldir_t and struct kstat deal in 64-bit inode numbers These patches make the kernel pass 64-bit inode numbers internally when communicating to userspace, even on a 32-bit system. They are required because some filesystems have intrinsic 64-bit inode numbers: NFS3+ and XFS for example. The 64-bit inode numbers are then propagated to userspace automatically where the arch supports it. Problems have been seen with userspace (eg: ld.so) using the 64-bit inode number returned by stat64() or getdents64() to differentiate files, and failing because the 64-bit inode number space was compressed to 32-bits, and so overlaps occur. This patch: Make filldir_t take a 64-bit inode number and struct kstat carry a 64-bit inode number so that 64-bit inode numbers can be passed back to userspace. The stat functions then returns the full 64-bit inode number where available and where possible. If it is not possible to represent the inode number supplied by the filesystem in the field provided by userspace, then error EOVERFLOW will be issued. Similarly, the getdents/readdir functions now pass the full 64-bit inode number to userspace where possible, returning EOVERFLOW instead when a directory entry is encountered that can't be properly represented. Note that this means that some inodes will not be stat'able on a 32-bit system with old libraries where they were before - but it does mean that there will be no ambiguity over what a 32-bit inode number refers to. Note similarly that directory scans may be cut short with an error on a 32-bit system with old libraries where the scan would work before for the same reasons. It is judged unlikely that this situation will occur because modern glibc uses 64-bit capable versions of stat and getdents class functions exclusively, and that older systems are unlikely to encounter unrepresentable inode numbers anyway. [akpm: alpha build fix] Signed-off-by: David Howells <dhowells@redhat.com> Cc: Trond Myklebust <trond.myklebust@fys.uio.no> Cc: Al Viro <viro@zeniv.linux.org.uk> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-10-03 16:13:46 +08:00
d_ino = ino;
if (sizeof(d_ino) < sizeof(ino) && d_ino != ino)
return -EOVERFLOW;
dirent = buf->previous;
if (dirent)
put_user(offset, &dirent->d_off);
dirent = buf->current_dir;
buf->previous = dirent;
[PATCH] VFS: Make filldir_t and struct kstat deal in 64-bit inode numbers These patches make the kernel pass 64-bit inode numbers internally when communicating to userspace, even on a 32-bit system. They are required because some filesystems have intrinsic 64-bit inode numbers: NFS3+ and XFS for example. The 64-bit inode numbers are then propagated to userspace automatically where the arch supports it. Problems have been seen with userspace (eg: ld.so) using the 64-bit inode number returned by stat64() or getdents64() to differentiate files, and failing because the 64-bit inode number space was compressed to 32-bits, and so overlaps occur. This patch: Make filldir_t take a 64-bit inode number and struct kstat carry a 64-bit inode number so that 64-bit inode numbers can be passed back to userspace. The stat functions then returns the full 64-bit inode number where available and where possible. If it is not possible to represent the inode number supplied by the filesystem in the field provided by userspace, then error EOVERFLOW will be issued. Similarly, the getdents/readdir functions now pass the full 64-bit inode number to userspace where possible, returning EOVERFLOW instead when a directory entry is encountered that can't be properly represented. Note that this means that some inodes will not be stat'able on a 32-bit system with old libraries where they were before - but it does mean that there will be no ambiguity over what a 32-bit inode number refers to. Note similarly that directory scans may be cut short with an error on a 32-bit system with old libraries where the scan would work before for the same reasons. It is judged unlikely that this situation will occur because modern glibc uses 64-bit capable versions of stat and getdents class functions exclusively, and that older systems are unlikely to encounter unrepresentable inode numbers anyway. [akpm: alpha build fix] Signed-off-by: David Howells <dhowells@redhat.com> Cc: Trond Myklebust <trond.myklebust@fys.uio.no> Cc: Al Viro <viro@zeniv.linux.org.uk> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-10-03 16:13:46 +08:00
put_user(d_ino, &dirent->d_ino);
put_user(reclen, &dirent->d_reclen);
copy_to_user(dirent->d_name, name, namlen);
put_user(0, dirent->d_name + namlen);
dirent = ((void __user *)dirent) + reclen;
buf->current_dir = dirent;
buf->count -= reclen;
return 0;
}
asmlinkage long
sys32_getdents (unsigned int fd, void __user * dirent, unsigned int count)
{
struct file * file;
struct linux32_dirent __user * lastdirent;
struct getdents32_callback buf;
int error;
error = -EBADF;
file = fget(fd);
if (!file)
goto out;
buf.current_dir = (struct linux32_dirent __user *) dirent;
buf.previous = NULL;
buf.count = count;
buf.error = 0;
error = vfs_readdir(file, filldir32, &buf);
if (error < 0)
goto out_putf;
error = buf.error;
lastdirent = buf.previous;
if (lastdirent) {
put_user(file->f_pos, &lastdirent->d_off);
error = count - buf.count;
}
out_putf:
fput(file);
out:
return error;
}
static int fillonedir32(void * __buf, const char * name, int namlen,
loff_t offset, u64 ino, unsigned int d_type)
{
struct readdir32_callback * buf = (struct readdir32_callback *) __buf;
struct old_linux32_dirent __user * dirent;
[PATCH] VFS: Make filldir_t and struct kstat deal in 64-bit inode numbers These patches make the kernel pass 64-bit inode numbers internally when communicating to userspace, even on a 32-bit system. They are required because some filesystems have intrinsic 64-bit inode numbers: NFS3+ and XFS for example. The 64-bit inode numbers are then propagated to userspace automatically where the arch supports it. Problems have been seen with userspace (eg: ld.so) using the 64-bit inode number returned by stat64() or getdents64() to differentiate files, and failing because the 64-bit inode number space was compressed to 32-bits, and so overlaps occur. This patch: Make filldir_t take a 64-bit inode number and struct kstat carry a 64-bit inode number so that 64-bit inode numbers can be passed back to userspace. The stat functions then returns the full 64-bit inode number where available and where possible. If it is not possible to represent the inode number supplied by the filesystem in the field provided by userspace, then error EOVERFLOW will be issued. Similarly, the getdents/readdir functions now pass the full 64-bit inode number to userspace where possible, returning EOVERFLOW instead when a directory entry is encountered that can't be properly represented. Note that this means that some inodes will not be stat'able on a 32-bit system with old libraries where they were before - but it does mean that there will be no ambiguity over what a 32-bit inode number refers to. Note similarly that directory scans may be cut short with an error on a 32-bit system with old libraries where the scan would work before for the same reasons. It is judged unlikely that this situation will occur because modern glibc uses 64-bit capable versions of stat and getdents class functions exclusively, and that older systems are unlikely to encounter unrepresentable inode numbers anyway. [akpm: alpha build fix] Signed-off-by: David Howells <dhowells@redhat.com> Cc: Trond Myklebust <trond.myklebust@fys.uio.no> Cc: Al Viro <viro@zeniv.linux.org.uk> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-10-03 16:13:46 +08:00
u32 d_ino;
if (buf->count)
return -EINVAL;
[PATCH] VFS: Make filldir_t and struct kstat deal in 64-bit inode numbers These patches make the kernel pass 64-bit inode numbers internally when communicating to userspace, even on a 32-bit system. They are required because some filesystems have intrinsic 64-bit inode numbers: NFS3+ and XFS for example. The 64-bit inode numbers are then propagated to userspace automatically where the arch supports it. Problems have been seen with userspace (eg: ld.so) using the 64-bit inode number returned by stat64() or getdents64() to differentiate files, and failing because the 64-bit inode number space was compressed to 32-bits, and so overlaps occur. This patch: Make filldir_t take a 64-bit inode number and struct kstat carry a 64-bit inode number so that 64-bit inode numbers can be passed back to userspace. The stat functions then returns the full 64-bit inode number where available and where possible. If it is not possible to represent the inode number supplied by the filesystem in the field provided by userspace, then error EOVERFLOW will be issued. Similarly, the getdents/readdir functions now pass the full 64-bit inode number to userspace where possible, returning EOVERFLOW instead when a directory entry is encountered that can't be properly represented. Note that this means that some inodes will not be stat'able on a 32-bit system with old libraries where they were before - but it does mean that there will be no ambiguity over what a 32-bit inode number refers to. Note similarly that directory scans may be cut short with an error on a 32-bit system with old libraries where the scan would work before for the same reasons. It is judged unlikely that this situation will occur because modern glibc uses 64-bit capable versions of stat and getdents class functions exclusively, and that older systems are unlikely to encounter unrepresentable inode numbers anyway. [akpm: alpha build fix] Signed-off-by: David Howells <dhowells@redhat.com> Cc: Trond Myklebust <trond.myklebust@fys.uio.no> Cc: Al Viro <viro@zeniv.linux.org.uk> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-10-03 16:13:46 +08:00
d_ino = ino;
if (sizeof(d_ino) < sizeof(ino) && d_ino != ino)
return -EOVERFLOW;
buf->count++;
dirent = buf->dirent;
[PATCH] VFS: Make filldir_t and struct kstat deal in 64-bit inode numbers These patches make the kernel pass 64-bit inode numbers internally when communicating to userspace, even on a 32-bit system. They are required because some filesystems have intrinsic 64-bit inode numbers: NFS3+ and XFS for example. The 64-bit inode numbers are then propagated to userspace automatically where the arch supports it. Problems have been seen with userspace (eg: ld.so) using the 64-bit inode number returned by stat64() or getdents64() to differentiate files, and failing because the 64-bit inode number space was compressed to 32-bits, and so overlaps occur. This patch: Make filldir_t take a 64-bit inode number and struct kstat carry a 64-bit inode number so that 64-bit inode numbers can be passed back to userspace. The stat functions then returns the full 64-bit inode number where available and where possible. If it is not possible to represent the inode number supplied by the filesystem in the field provided by userspace, then error EOVERFLOW will be issued. Similarly, the getdents/readdir functions now pass the full 64-bit inode number to userspace where possible, returning EOVERFLOW instead when a directory entry is encountered that can't be properly represented. Note that this means that some inodes will not be stat'able on a 32-bit system with old libraries where they were before - but it does mean that there will be no ambiguity over what a 32-bit inode number refers to. Note similarly that directory scans may be cut short with an error on a 32-bit system with old libraries where the scan would work before for the same reasons. It is judged unlikely that this situation will occur because modern glibc uses 64-bit capable versions of stat and getdents class functions exclusively, and that older systems are unlikely to encounter unrepresentable inode numbers anyway. [akpm: alpha build fix] Signed-off-by: David Howells <dhowells@redhat.com> Cc: Trond Myklebust <trond.myklebust@fys.uio.no> Cc: Al Viro <viro@zeniv.linux.org.uk> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-10-03 16:13:46 +08:00
put_user(d_ino, &dirent->d_ino);
put_user(offset, &dirent->d_offset);
put_user(namlen, &dirent->d_namlen);
copy_to_user(dirent->d_name, name, namlen);
put_user(0, dirent->d_name + namlen);
return 0;
}
asmlinkage long
sys32_readdir (unsigned int fd, void __user * dirent, unsigned int count)
{
int error;
struct file * file;
struct readdir32_callback buf;
error = -EBADF;
file = fget(fd);
if (!file)
goto out;
buf.count = 0;
buf.dirent = dirent;
error = vfs_readdir(file, fillonedir32, &buf);
if (error >= 0)
error = buf.count;
fput(file);
out:
return error;
}
/*** copied from mips64 ***/
/*
* Ooo, nasty. We need here to frob 32-bit unsigned longs to
* 64-bit unsigned longs.
*/
static inline int
get_fd_set32(unsigned long n, u32 *ufdset, unsigned long *fdset)
{
n = (n + 8*sizeof(u32) - 1) / (8*sizeof(u32));
if (ufdset) {
unsigned long odd;
if (!access_ok(VERIFY_WRITE, ufdset, n*sizeof(u32)))
return -EFAULT;
odd = n & 1UL;
n &= ~1UL;
while (n) {
unsigned long h, l;
__get_user(l, ufdset);
__get_user(h, ufdset+1);
ufdset += 2;
*fdset++ = h << 32 | l;
n -= 2;
}
if (odd)
__get_user(*fdset, ufdset);
} else {
/* Tricky, must clear full unsigned long in the
* kernel fdset at the end, this makes sure that
* actually happens.
*/
memset(fdset, 0, ((n + 1) & ~1)*sizeof(u32));
}
return 0;
}
static inline void
set_fd_set32(unsigned long n, u32 *ufdset, unsigned long *fdset)
{
unsigned long odd;
n = (n + 8*sizeof(u32) - 1) / (8*sizeof(u32));
if (!ufdset)
return;
odd = n & 1UL;
n &= ~1UL;
while (n) {
unsigned long h, l;
l = *fdset++;
h = l >> 32;
__put_user(l, ufdset);
__put_user(h, ufdset+1);
ufdset += 2;
n -= 2;
}
if (odd)
__put_user(*fdset, ufdset);
}
struct msgbuf32 {
int mtype;
char mtext[1];
};
asmlinkage long sys32_msgsnd(int msqid,
struct msgbuf32 __user *umsgp32,
size_t msgsz, int msgflg)
{
struct msgbuf *mb;
struct msgbuf32 mb32;
int err;
if ((mb = kmalloc(msgsz + sizeof *mb + 4, GFP_KERNEL)) == NULL)
return -ENOMEM;
err = get_user(mb32.mtype, &umsgp32->mtype);
mb->mtype = mb32.mtype;
err |= copy_from_user(mb->mtext, &umsgp32->mtext, msgsz);
if (err)
err = -EFAULT;
else
KERNEL_SYSCALL(err, sys_msgsnd, msqid, (struct msgbuf __user *)mb, msgsz, msgflg);
kfree(mb);
return err;
}
asmlinkage long sys32_msgrcv(int msqid,
struct msgbuf32 __user *umsgp32,
size_t msgsz, long msgtyp, int msgflg)
{
struct msgbuf *mb;
struct msgbuf32 mb32;
int err, len;
if ((mb = kmalloc(msgsz + sizeof *mb + 4, GFP_KERNEL)) == NULL)
return -ENOMEM;
KERNEL_SYSCALL(err, sys_msgrcv, msqid, (struct msgbuf __user *)mb, msgsz, msgtyp, msgflg);
if (err >= 0) {
len = err;
mb32.mtype = mb->mtype;
err = put_user(mb32.mtype, &umsgp32->mtype);
err |= copy_to_user(&umsgp32->mtext, mb->mtext, len);
if (err)
err = -EFAULT;
else
err = len;
}
kfree(mb);
return err;
}
asmlinkage int sys32_sendfile(int out_fd, int in_fd, compat_off_t __user *offset, s32 count)
{
mm_segment_t old_fs = get_fs();
int ret;
off_t of;
if (offset && get_user(of, offset))
return -EFAULT;
set_fs(KERNEL_DS);
ret = sys_sendfile(out_fd, in_fd, offset ? (off_t __user *)&of : NULL, count);
set_fs(old_fs);
if (offset && put_user(of, offset))
return -EFAULT;
return ret;
}
asmlinkage int sys32_sendfile64(int out_fd, int in_fd, compat_loff_t __user *offset, s32 count)
{
mm_segment_t old_fs = get_fs();
int ret;
loff_t lof;
if (offset && get_user(lof, offset))
return -EFAULT;
set_fs(KERNEL_DS);
ret = sys_sendfile64(out_fd, in_fd, offset ? (loff_t __user *)&lof : NULL, count);
set_fs(old_fs);
if (offset && put_user(lof, offset))
return -EFAULT;
return ret;
}
/* lseek() needs a wrapper because 'offset' can be negative, but the top
* half of the argument has been zeroed by syscall.S.
*/
asmlinkage int sys32_lseek(unsigned int fd, int offset, unsigned int origin)
{
return sys_lseek(fd, offset, origin);
}
asmlinkage long sys32_semctl(int semid, int semnum, int cmd, union semun arg)
{
union semun u;
if (cmd == SETVAL) {
/* Ugh. arg is a union of int,ptr,ptr,ptr, so is 8 bytes.
* The int should be in the first 4, but our argument
* frobbing has left it in the last 4.
*/
u.val = *((int *)&arg + 1);
return sys_semctl (semid, semnum, cmd, u);
}
return sys_semctl (semid, semnum, cmd, arg);
}
long sys32_lookup_dcookie(u32 cookie_high, u32 cookie_low, char __user *buf,
size_t len)
{
return sys_lookup_dcookie((u64)cookie_high << 32 | cookie_low,
buf, len);
}