nuttx/fs/nfs/nfs_vfsops.c
patacongo f5ad5f4bc2 NFS update
git-svn-id: svn://svn.code.sf.net/p/nuttx/code/trunk@4713 42af7a65-404d-4744-a932-0658087f49c3
2012-05-07 23:33:39 +00:00

1676 lines
42 KiB
C

/****************************************************************************
* fs/nfs/nfs_vfsops.c
*
* Copyright (C) 2012 Gregory Nutt. All rights reserved.
* Copyright (C) 2012 Jose Pablo Rojas Vargas. All rights reserved.
* Author: Jose Pablo Rojas Vargas <jrojas@nx-engineering.com>
*
* Leveraged from OpenBSD:
*
* Copyright (c) 1989, 1993, 1995
* The Regents of the University of California. All rights reserved.
*
* This code is derived from software contributed to Berkeley by
* Rick Macklem at The University of Guelph.
*
* 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. 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.
*
****************************************************************************/
/****************************************************************************
* Included Files
****************************************************************************/
#include <nuttx/config.h>
#include <sys/socket.h>
#include <sys/statfs.h>
#include <sys/stat.h>
#include <sys/types.h>
#include <stdint.h>
#include <stdbool.h>
#include <stdlib.h>
#include <queue.h>
#include <string.h>
#include <fcntl.h>
#include <assert.h>
#include <errno.h>
#include <debug.h>
#include <time.h>
#include <nuttx/kmalloc.h>
#include <nuttx/fs/dirent.h>
#include <nuttx/fs/fs.h>
#include <nuttx/fs/nfs.h>
#include <semaphore.h>
#include <net/if.h>
#include <netinet/in.h>
#include "nfs.h"
#include "rpc_v2.h"
#include "rpc.h"
#include "nfs_proto.h"
#include "nfs_node.h"
#include "nfs_mount.h"
#include "xdr_subs.h"
#include "nfs_socket.h"
/****************************************************************************
* Pre-processor Definitions
****************************************************************************/
#define NFS_DIRHDSIZ (sizeof (struct nfs_dirent) - (MAXNAMLEN + 1))
#define NFS_DIRENT_OVERHEAD offsetof(struct nfs_dirent, dirent)
/****************************************************************************
* Private Type Definitions
****************************************************************************/
struct nfs_dirent
{
uint32_t cookie[2];
struct dirent dirent;
};
/****************************************************************************
* Private Function Prototypes
****************************************************************************/
static int nfs_open(FAR struct file *filep, const char *relpath,
int oflags, mode_t mode);
static ssize_t nfs_read(FAR struct file *filep, char *buffer, size_t buflen);
static ssize_t nfs_write(FAR struct file *filep, const char *buffer,
size_t buflen);
static int nfs_readdir(struct inode *mountpt, struct fs_dirent_s *dir);
static int nfs_bind(FAR struct inode *blkdriver, const void *data,
void **handle);
static int nfs_unbind(void *handle, FAR struct inode **blkdriver);
static int nfs_statfs(struct inode *mountpt, struct statfs *buf);
static int nfs_remove(struct inode *mountpt, const char *relpath);
static int nfs_mkdir(struct inode *mountpt, const char *relpath,
mode_t mode);
static int nfs_rmdir(struct inode *mountpt, const char *relpath);
static int nfs_rename(struct inode *mountpt, const char *oldrelpath,
const char *newrelpath);
static int nfs_fsinfo(struct inode *mountpt, const char *relpath,
struct stat *buf);
/****************************************************************************
* External Public Data (this belong in a header file)
****************************************************************************/
extern uint32_t nfs_true;
extern uint32_t nfs_false;
extern uint32_t nfs_xdrneg1;
extern struct nfsstats nfsstats;
extern int nfs_ticks;
/****************************************************************************
* Public Data
****************************************************************************/
/* nfs vfs operations. */
const struct mountpt_operations nfs_operations =
{
nfs_open, /* open */
NULL, /* close */
nfs_read, /* read */
nfs_write, /* write */
NULL, /* seek */
NULL, /* ioctl */
NULL, /* sync */
NULL, /* opendir */
NULL, /* closedir */
nfs_readdir, /* readdir */
NULL, /* rewinddir */
nfs_bind, /* bind */
nfs_unbind, /* unbind */
nfs_statfs, /* statfs */
nfs_remove, /* unlink */
nfs_mkdir, /* mkdir */
nfs_rmdir, /* rmdir */
nfs_rename, /* rename */
nfs_fsinfo /* stat */
};
/****************************************************************************
* Private Functions
****************************************************************************/
/****************************************************************************
* Public Functions
****************************************************************************/
/****************************************************************************
* Name: nfs_open
*
* Description: if oflags == O_CREAT it creates a file, if not it
* check to see if the type is ok and that deletion is not in progress.
****************************************************************************/
static int
nfs_open(FAR struct file *filep, FAR const char *relpath,
int oflags, mode_t mode)
{
struct inode *in;
struct nfs_fattr vap;
struct nfsv3_sattr sp;
struct nfsmount *nmp;
struct nfsnode *np;
struct CREATE3args create;
struct CREATE3resok *resok;
void *datareply;
int error = 0;
/* Sanity checks */
DEBUGASSERT(filep->f_inode != NULL);
/* Get the mountpoint inode reference from the file structure and the
* mountpoint private data from the inode structure
*/
in = filep->f_inode;
nmp = (struct nfsmount*)in->i_private;
np = (struct nfsnode*)filep->f_priv;
DEBUGASSERT(nmp != NULL);
/* Check if the mount is still healthy */
nfs_semtake(nmp);
error = nfs_checkmount(nmp);
if (error != 0)
{
goto errout_with_semaphore;
}
if (oflags == O_CREAT)
{
/* Sanity checks */
DEBUGASSERT(filep->f_priv == NULL);
again:
nfsstats.rpccnt[NFSPROC_CREATE]++;
memset(&sp, 0, sizeof(struct nfsv3_sattr));
memset(&vap, 0, sizeof(struct nfs_fattr));
vap = nmp->nm_head->n_fattr;
sp.sa_modetrue = true;
sp.sa_mode = txdr_unsigned(mode);
sp.sa_uidfalse = nfs_xdrneg1;
sp.sa_gidfalse = nfs_xdrneg1;
sp.sa_sizefalse = nfs_xdrneg1;
sp.sa_atimetype = txdr_unsigned(NFSV3SATTRTIME_TOCLIENT);
sp.sa_mtimetype = txdr_unsigned(NFSV3SATTRTIME_TOCLIENT);
sp.sa_atime = vap.fa3_atime;
sp.sa_mtime = vap.fa3_mtime;
memset(&create, 0, sizeof(struct CREATE3args));
create.how = sp;
create.where.dir = np->n_fhp;
create.where.name = relpath;
error = nfs_request(nmp, NFSPROC_CREATE, &create, &datareply);
if (!error)
{
/* Create an instance of the file private data to describe the opened
* file.
*/
np = (struct nfsnode *)kzalloc(sizeof(struct nfsnode));
if (!np)
{
fdbg("Failed to allocate private data\n", error);
error = -ENOMEM;
goto errout_with_semaphore;
}
/* Initialize the file private data (only need to initialize
* non-zero elements)
*/
resok = (struct CREATE3resok *) datareply;
np->n_open = true;
np->nfsv3_type = NFREG;
np->n_fhp = resok->handle;
np->n_size = fxdr_hyper(&resok->attributes.fa3_size);
np->n_fattr = resok->attributes;
fxdr_nfsv3time(&resok->attributes.fa3_mtime, &np->n_mtime)
np->n_ctime = fxdr_hyper(&resok->attributes.fa3_ctime);
/* Attach the private date to the struct file instance */
filep->f_priv = np;
/* Then insert the new instance into the mountpoint structure.
* It needs to be there (1) to handle error conditions that effect
* all files, and (2) to inform the umount logic that we are busy
* (but a simple reference count could have done that).
*/
np->n_next = nmp->nm_head;
nmp->nm_head = np->n_next;
error = 0;
}
else
{
if (error == NFSERR_NOTSUPP)
{
goto again;
}
}
np->n_flag |= NMODIFIED;
}
else
{
if (np->nfsv3_type != NFREG && np->nfsv3_type != NFDIR)
{
fdbg("open eacces typ=%d\n", np->nfsv3_type);
return EACCES;
}
if (np->n_flag & NMODIFIED)
{
if (np->nfsv3_type == NFDIR)
{
np->n_direofoffset = 0;
}
}
}
/* For open/close consistency. */
NFS_INVALIDATE_ATTRCACHE(np);
errout_with_semaphore:
kfree(np);
nfs_semgive(nmp);
return error;
}
#undef COMP
#ifdef COMP
/****************************************************************************
* Name: nfs_close
****************************************************************************/
static int nfs_close(FAR struct file *filep) done
{
struct nfsmount *nmp;
struct nfsnode *np;
int error = 0;
fvdbg("Closing\n");
/* Sanity checks */
DEBUGASSERT(filep->f_priv != NULL && filep->f_inode != NULL);
/* Recover our private data from the struct file instance */
np = filep->f_priv;
nmp = filep->f_inode->i_private;
DEBUGASSERT(nmp != NULL);
if (np->nfsv3_type == NFREG)
{
error = nfs_sync(filep);
kfree(np);
filep->f_priv = NULL;
}
return error;
}
#endif
/****************************************************************************
* Name: nfs_read
****************************************************************************/
static ssize_t nfs_read(FAR struct file *filep, char *buffer, size_t buflen)
{
struct nfsmount *nmp;
struct nfsnode *np;
unsigned int readsize;
int bytesleft;
uint64_t offset;
void *datareply;
struct READ3args read;
struct READ3resok *resok;
uint8_t *userbuffer = (uint8_t*)buffer;
int error = 0;
int len;
bool eof;
fvdbg("Read %d bytes from offset %d\n", buflen, filep->f_pos);
/* Sanity checks */
DEBUGASSERT(filep->f_priv != NULL && filep->f_inode != NULL);
/* Recover our private data from the struct file instance */
np = (struct nfsnode*) filep->f_priv;
nmp = (struct nfsmount*) filep->f_inode->i_private;
eof = false;
offset = 0;
DEBUGASSERT(nmp != NULL);
/* Make sure that the mount is still healthy */
nfs_semtake(nmp);
error = nfs_checkmount(nmp);
if (error != 0)
{
fdbg("nfs_checkmount failed: %d\n", error);
goto errout_with_semaphore;
}
if (np->nfsv3_type != NFREG)
{
fdbg("read eacces typ=%d\n", np->nfsv3_type);
return EACCES;
}
if ((nmp->nm_flag & (NFSMNT_NFSV3 | NFSMNT_GOTFSINFO)) == NFSMNT_NFSV3)
{
(void)nfs_fsinfo(filep->f_inode, NULL, NULL);
}
/* Get the number of bytes left in the file */
bytesleft = np->n_size - filep->f_pos;
readsize = 0;
/* Truncate read count so that it does not exceed the number
* of bytes left in the file.
*/
if (buflen > bytesleft)
{
buflen = bytesleft;
}
len = nmp->nm_rsize;
if (len < buflen)
{
error = EFBIG;
goto errout_with_semaphore;
}
nfsstats.rpccnt[NFSPROC_READ]++;
again:
memset(&read, 0, sizeof(struct READ3args));
read.file = np->nfsv3_type;
read.count = buflen;
read.offset = offset;
error = nfs_request(nmp, NFSPROC_READ, &read, &datareply);
if (error)
{
goto errout_with_semaphore;
}
//bcopy (datareply, &resok, sizeof(struct READ3resok));
resok = (struct READ3resok *) datareply;
eof = resok->eof;
if (eof == true)
{
readsize = resok->count;
np->n_fattr = resok->file_attributes;
memcpy(userbuffer, resok->data, readsize);
}
else
{
goto again;
}
nfs_semgive(nmp);
return readsize;
errout_with_semaphore:
nfs_semgive(nmp);
return error;
}
/****************************************************************************
* Name: nfs_write
****************************************************************************/
static ssize_t
nfs_write(FAR struct file *filep, const char *buffer, size_t buflen)
{
struct inode *inode;
struct nfsmount *nmp;
struct nfsnode *np;
unsigned int writesize;
void *datareply;
struct WRITE3args write;
struct WRITE3resok *resok;
uint8_t *userbuffer = (uint8_t*)buffer;
int error = 0;
uint64_t offset;
int len;
enum stable_how commit;
int committed = NFSV3WRITE_FILESYNC;
/* Sanity checks */
DEBUGASSERT(filep->f_priv != NULL && filep->f_inode != NULL);
/* Recover our private data from the struct file instance */
np = (struct nfsnode *)filep->f_priv;
inode = filep->f_inode;
nmp = (struct nfsmount *)inode->i_private;
offset = 0;
DEBUGASSERT(nmp != NULL);
/* Make sure that the mount is still healthy */
nfs_semtake(nmp);
error = nfs_checkmount(nmp);
if (error != 0)
{
goto errout_with_semaphore;
}
/* Check if the file size would exceed the range of off_t */
if (np->n_size + buflen < np->n_size)
{
error = -EFBIG;
goto errout_with_semaphore;
}
len = nmp->nm_wsize;
if (len < buflen)
{
error = -EFBIG;
goto errout_with_semaphore;
}
writesize = 0;
nfsstats.rpccnt[NFSPROC_WRITE]++;
memset(&write, 0, sizeof(struct WRITE3args));
write.file = np->nfsv3_type;
write.offset = offset;
write.count = buflen;
write.stable = committed;
memcpy((void *)write.data, userbuffer, buflen);
error = nfs_request(nmp, NFSPROC_WRITE, &write, &datareply);
if (error)
{
goto errout_with_semaphore;
}
//bcopy (datareply, &resok, sizeof(struct WRITE3resok));
resok = (struct WRITE3resok *) datareply;
writesize = resok->count;
if (writesize == 0)
{
error = NFSERR_IO;
goto errout_with_semaphore;
}
commit = resok->committed;
np->n_fattr = resok->file_wcc.after;
/* Return the lowest committment level obtained by any of the RPCs. */
if (committed == NFSV3WRITE_FILESYNC)
{
committed = commit;
}
else if (committed == NFSV3WRITE_DATASYNC &&
commit == NFSV3WRITE_UNSTABLE)
{
committed = commit;
}
if ((nmp->nm_flag & NFSMNT_HASWRITEVERF) == 0)
{
bcopy((void*) resok->verf, (void*) nmp->nm_verf, NFSX_V3WRITEVERF);
nmp->nm_flag |= NFSMNT_HASWRITEVERF;
}
else if (strncmp((char*) resok->verf, (char*) nmp->nm_verf, NFSX_V3WRITEVERF))
{
bcopy((void*) resok->verf, (void*) nmp->nm_verf, NFSX_V3WRITEVERF);
}
fxdr_nfsv3time(&np->n_fattr.fa3_mtime, &np->n_mtime)
nfs_semgive(nmp);
return writesize;
errout_with_semaphore:
nfs_semgive(nmp);
return error;
}
/****************************************************************************
* Name: nfs_readdirrpc
*
* Description: The function below stuff the cookies in after the name.
****************************************************************************/
int nfs_readdirrpc(struct nfsmount *nmp, struct nfsnode *np,
bool end_of_directory, struct fs_dirent_s *dir)
{
int error = 0;
void *datareply;
struct READDIR3args readir;
struct READDIR3resok *resok;
/* Loop around doing readdir rpc's of size nm_readdirsize
* truncated to a multiple of NFS_READDIRBLKSIZ.
* The stopping criteria is EOF.
*/
while (end_of_directory == false)
{
nfsstats.rpccnt[NFSPROC_READDIR]++;
memset(&readir, 0, sizeof(struct READDIR3args));
readir.dir = np->n_fhp;
readir.count = nmp->nm_readdirsize;
if (nfsstats.rpccnt[NFSPROC_READDIR] == 1)
{
readir.cookie.nfsuquad[0] = 0;
readir.cookie.nfsuquad[1] = 0;
readir.cookieverf.nfsuquad[0] = 0;
readir.cookieverf.nfsuquad[1] = 0;
}
else
{
readir.cookie.nfsuquad[0] = dir->u.nfs.cookie[0];
readir.cookie.nfsuquad[1] = dir->u.nfs.cookie[1];
readir.cookieverf.nfsuquad[0] = np->n_cookieverf.nfsuquad[0];
readir.cookieverf.nfsuquad[1] = np->n_cookieverf.nfsuquad[1];
}
error = nfs_request(nmp, NFSPROC_READDIR, &readir, &datareply);
if (error)
{
goto nfsmout;
}
//bcopy (datareply, &resok, sizeof(struct READDIR3resok));
resok = (struct READDIR3resok *) datareply;
np->n_fattr = resok->dir_attributes;
np->n_cookieverf.nfsuquad[0] = resok->cookieverf.nfsuquad[0];
np->n_cookieverf.nfsuquad[1] = resok->cookieverf.nfsuquad[1];
dir->fd_dir.d_type = resok->reply.entries->fileid;
memcpy(&dir->fd_dir.d_name[NAME_MAX], &resok->reply.entries->name, NAME_MAX);
//dir->fd_dir.d_name = resok->reply.entries->name;//
dir->u.nfs.cookie[0] = resok->reply.entries->cookie.nfsuquad[0];
dir->u.nfs.cookie[1] = resok->reply.entries->cookie.nfsuquad[1];
if (resok->reply.eof == true)
{
end_of_directory = true;
}
//more_dirs = fxdr_unsigned(int, *dp);
/* loop thru the dir entries*/
/*
while (more_dirs && bigenough)
{
if (bigenough)
{
if (info_v3)
{
dir->u.nfs.cookie[0] = cookie.nfsuquad[0];
}
else
{
dir->u.nfs.cookie[0] = ndp->cookie[0] = 0;
}
dir->u.nfs.cookie[1] = ndp->cookie[1] = cookie.nfsuquad[1];
}
more_dirs = fxdr_unsigned(int, *ndp);
}
*/
}
/* We are now either at the end of the directory */
if (resok->reply.entries == NULL)
{
np->n_direofoffset = fxdr_hyper(&dir->u.nfs.cookie[0]);
/* We signal the end of the directory by returning the
* special error -ENOENT
*/
fdbg("End of directory\n");
error = -ENOENT;
}
nfsmout:
return error;
}
/****************************************************************************
* Name: nfs_readdir
*
* Description: Read from directory
*
****************************************************************************/
static int nfs_readdir(struct inode *mountpt, struct fs_dirent_s *dir)
{
int error = 0;
struct nfsmount *nmp;
struct nfsnode *np;
bool eof = false;
//struct nfs_dirent *ndp;
fvdbg("Entry\n");
/* Sanity checks */
DEBUGASSERT(mountpt != NULL && mountpt->i_private != NULL);
/* Recover our private data from the inode instance */
nmp = mountpt->i_private;
np = nmp->nm_head;
dir->fd_root = mountpt;
/* Make sure that the mount is still healthy */
nfs_semtake(nmp);
error = nfs_checkmount(nmp);
if (error != 0)
{
fdbg("romfs_checkmount failed: %d\n", error);
goto errout_with_semaphore;
}
if (np->nfsv3_type != NFDIR)
{
error = EPERM;
goto errout_with_semaphore;
}
dir->u.nfs.nd_direoffset = np->n_direofoffset;
/* First, check for hit on the EOF offset */
if (dir->u.nfs.nd_direoffset != 0)
{
nfsstats.direofcache_hits++;
//np->n_open = true;
return 0;
}
if ((nmp->nm_flag & (NFSMNT_NFSV3 | NFSMNT_GOTFSINFO)) == NFSMNT_NFSV3)
{
(void)nfs_fsinfo(mountpt, NULL, NULL);
}
error = nfs_readdirrpc(nmp, np, eof, dir);
if (error == NFSERR_BAD_COOKIE)
{
error = EINVAL;
}
if (!error && eof)
{
nfsstats.direofcache_misses++;
nfs_semgive(nmp);
return 0;
}
errout_with_semaphore:
nfs_semgive(nmp);
return error;
}
/****************************************************************************
* Name: nfs_decode_args
****************************************************************************/
void nfs_decode_args(struct nfsmount *nmp, struct nfs_args *argp)
{
int adjsock = 0;
int maxio;
#ifdef CONFIG_NFS_TCPIP
/* Re-bind if rsrvd port requested and wasn't on one */
adjsock = !(nmp->nm_flag & NFSMNT_RESVPORT)
&& (argp->flags & NFSMNT_RESVPORT);
#endif
/* Also re-bind if we're switching to/from a connected UDP socket */
adjsock |= ((nmp->nm_flag & NFSMNT_NOCONN) != (argp->flags & NFSMNT_NOCONN));
/* Update flags atomically. Don't change the lock bits. */
nmp->nm_flag =
(argp->flags & ~NFSMNT_INTERNAL) | (nmp->nm_flag & NFSMNT_INTERNAL);
if ((argp->flags & NFSMNT_TIMEO) && argp->timeo > 0)
{
nmp->nm_timeo = (argp->timeo * NFS_HZ + 5) / 10;
if (nmp->nm_timeo < NFS_MINTIMEO)
{
nmp->nm_timeo = NFS_MINTIMEO;
}
else if (nmp->nm_timeo > NFS_MAXTIMEO)
{
nmp->nm_timeo = NFS_MAXTIMEO;
}
}
if ((argp->flags & NFSMNT_RETRANS) && argp->retrans > 1)
{
nmp->nm_retry = (argp->retrans < NFS_MAXREXMIT)? argp->retrans : NFS_MAXREXMIT;
}
if (!(nmp->nm_flag & NFSMNT_SOFT))
{
nmp->nm_retry = NFS_MAXREXMIT + 1; /* past clip limit */
}
if (argp->flags & NFSMNT_NFSV3)
{
if (argp->sotype == SOCK_DGRAM)
{
maxio = NFS_MAXDGRAMDATA;
}
else
{
maxio = NFS_MAXDATA;
}
}
else
{
maxio = NFS_V2MAXDATA;
}
if ((argp->flags & NFSMNT_WSIZE) && argp->wsize > 0)
{
int osize = nmp->nm_wsize;
nmp->nm_wsize = argp->wsize;
/* Round down to multiple of blocksize */
nmp->nm_wsize &= ~(NFS_FABLKSIZE - 1);
if (nmp->nm_wsize <= 0)
{
nmp->nm_wsize = NFS_FABLKSIZE;
}
adjsock |= (nmp->nm_wsize != osize);
}
if (nmp->nm_wsize > maxio)
{
nmp->nm_wsize = maxio;
}
if (nmp->nm_wsize > MAXBSIZE)
{
nmp->nm_wsize = MAXBSIZE;
}
if ((argp->flags & NFSMNT_RSIZE) && argp->rsize > 0)
{
int osize = nmp->nm_rsize;
nmp->nm_rsize = argp->rsize;
/* Round down to multiple of blocksize */
nmp->nm_rsize &= ~(NFS_FABLKSIZE - 1);
if (nmp->nm_rsize <= 0)
{
nmp->nm_rsize = NFS_FABLKSIZE;
}
adjsock |= (nmp->nm_rsize != osize);
}
if (nmp->nm_rsize > maxio)
{
nmp->nm_rsize = maxio;
}
if (nmp->nm_rsize > MAXBSIZE)
{
nmp->nm_rsize = MAXBSIZE;
}
if ((argp->flags & NFSMNT_READDIRSIZE) && argp->readdirsize > 0)
{
nmp->nm_readdirsize = argp->readdirsize;
/* Round down to multiple of blocksize */
nmp->nm_readdirsize &= ~(NFS_DIRBLKSIZ - 1);
if (nmp->nm_readdirsize < NFS_DIRBLKSIZ)
{
nmp->nm_readdirsize = NFS_DIRBLKSIZ;
}
}
else if (argp->flags & NFSMNT_RSIZE)
{
nmp->nm_readdirsize = nmp->nm_rsize;
}
if (nmp->nm_readdirsize > maxio)
{
nmp->nm_readdirsize = maxio;
}
if ((argp->flags & NFSMNT_MAXGRPS) && argp->maxgrouplist >= 0 &&
argp->maxgrouplist <= NFS_MAXGRPS)
{
nmp->nm_numgrps = argp->maxgrouplist;
}
if ((argp->flags & NFSMNT_READAHEAD) && argp->readahead >= 0 &&
argp->readahead <= NFS_MAXRAHEAD)
{
nmp->nm_readahead = argp->readahead;
}
if (argp->flags & NFSMNT_ACREGMIN && argp->acregmin >= 0)
{
if (argp->acregmin > 0xffff)
{
nmp->nm_acregmin = 0xffff;
}
else
{
nmp->nm_acregmin = argp->acregmin;
}
}
if (argp->flags & NFSMNT_ACREGMAX && argp->acregmax >= 0)
{
if (argp->acregmax > 0xffff)
{
nmp->nm_acregmax = 0xffff;
}
else
{
nmp->nm_acregmax = argp->acregmax;
}
}
if (nmp->nm_acregmin > nmp->nm_acregmax)
{
nmp->nm_acregmin = nmp->nm_acregmax;
}
if (argp->flags & NFSMNT_ACDIRMIN && argp->acdirmin >= 0)
{
if (argp->acdirmin > 0xffff)
{
nmp->nm_acdirmin = 0xffff;
}
else
{
nmp->nm_acdirmin = argp->acdirmin;
}
}
if (argp->flags & NFSMNT_ACDIRMAX && argp->acdirmax >= 0)
{
if (argp->acdirmax > 0xffff)
{
nmp->nm_acdirmax = 0xffff;
}
else
{
nmp->nm_acdirmax = argp->acdirmax;
}
}
if (nmp->nm_acdirmin > nmp->nm_acdirmax)
{
nmp->nm_acdirmin = nmp->nm_acdirmax;
}
if (nmp->nm_so && adjsock)
{
nfs_disconnect(nmp);
if (nmp->nm_sotype == SOCK_DGRAM)
while (nfs_connect(nmp))
{
nvdbg("nfs_args: retrying connect\n");
}
}
}
/****************************************************************************
* Name: mountnfs
*
* Description: Common code for nfs_mount.
*
****************************************************************************/
int mountnfs(struct nfs_args *argp, void **handle)
{
struct nfsmount *nmp;
struct nfsnode *np;
int error;
/* Create an instance of the mountpt state structure */
nmp = (struct nfsmount *)kzalloc(sizeof(struct nfsmount));
if (!nmp)
{
fdbg("Failed to allocate mountpoint structure\n");
return -ENOMEM;
}
/* Initialize the allocated mountpt state structure. The filesystem is
* responsible for one reference ont the blkdriver inode and does not
* have to addref() here (but does have to release in ubind().
*/
sem_init(&nmp->nm_sem, 0, 0); /* Initialize the semaphore that controls access */
nfs_init();
nmp->nm_flag = argp->flags;
nmp->nm_timeo = NFS_TIMEO;
nmp->nm_retry = NFS_RETRANS;
nmp->nm_wsize = NFS_WSIZE;
nmp->nm_rsize = NFS_RSIZE;
nmp->nm_readdirsize = NFS_READDIRSIZE;
nmp->nm_numgrps = NFS_MAXGRPS;
nmp->nm_readahead = NFS_DEFRAHEAD;
nmp->nm_fhsize = NFSX_V3FHMAX;
nmp->nm_acregmin = NFS_MINATTRTIMO;
nmp->nm_acregmax = NFS_MAXATTRTIMO;
nmp->nm_acdirmin = NFS_MINATTRTIMO;
nmp->nm_acdirmax = NFS_MAXATTRTIMO;
nmp->nm_fh = argp->fh;
//strncpy(&mp->mnt_stat.f_fstypename[0], mp->mnt_vfc->vfc_name, MFSNAMELEN);
//memmove(hst, mp->mnt_stat.f_mntfromname, MNAMELEN);
//bcopy(pth, nmp->nm_mntonname, 90);
//memmove(argp, &mp->mnt_stat.mount_info.nfs_args, sizeof(*argp));
nmp->nm_nam = argp->addr;
nfs_decode_args(nmp, argp);
/* Create an instance of the file private data to describe the opened
* file.
*/
np = (struct nfsnode *)kzalloc(sizeof(struct nfsnode));
if (!np)
{
fdbg("Failed to allocate private data\n", error);
return -ENOMEM;
}
nmp->nm_head = np;
/* Set up the sockets and per-host congestion */
nmp->nm_sotype = argp->sotype;
nmp->nm_soproto = argp->proto;
/* For Connection based sockets (TCP,...) defer the connect until
* the first request, in case the server is not responding.
*/
if (nmp->nm_sotype == SOCK_DGRAM && (error = nfs_connect(nmp)))
{
goto bad;
}
/* Mounted! */
nmp->nm_mounted = true;
nmp->nm_so = nmp->nm_rpcclnt->rc_so;
*handle = (void*)nmp;
nfs_semgive(nmp);
return 0;
bad:
nfs_disconnect(nmp);
sem_destroy(&nmp->nm_sem);
kfree(nmp->nm_head);
kfree(nmp->nm_so);
kfree(nmp->nm_rpcclnt);
kfree(nmp);
return error;
}
/****************************************************************************
* Name: nfs_bind
*
* Description: This implements a portion of the mount operation. This
* function allocates and initializes the mountpoint private data and
* binds the blockdriver inode to the filesystem private data. The final
* binding of the private data (containing the blockdriver) to the
* mountpoint is performed by mount().
*
****************************************************************************/
static int nfs_bind(struct inode *blkdriver, const void *data, void **handle)
{
int error;
struct nfs_args args;
bcopy(data, &args, sizeof(struct nfs_args));
if (args.version == NFS_ARGSVERSION)
{
args.flags &= ~(NFSMNT_INTERNAL | NFSMNT_NOAC);
}
else
{
return -EINVAL;
}
if ((args.flags & (NFSMNT_NFSV3 | NFSMNT_RDIRPLUS)) == NFSMNT_RDIRPLUS)
{
return -EINVAL;
}
if (args.fhsize < 0 || args.fhsize > NFSX_V3FHMAX)
{
return -EINVAL;
}
error = mountnfs(&args, handle);
return error;
}
/****************************************************************************
* Name: nfs_unbind
*
* Description: This implements the filesystem portion of the umount
* operation.
*
****************************************************************************/
int nfs_unbind(void *handle, struct inode **blkdriver)
{
struct nfsmount *nmp = (struct nfsmount *) handle ;
fvdbg("Entry\n");
if (!nmp)
{
return -EINVAL;
}
nfs_semtake(nmp);
nfs_disconnect(nmp);
sem_destroy(&nmp->nm_sem);
kfree(nmp->nm_head);
kfree(nmp->nm_so);
kfree(nmp->nm_rpcclnt);
kfree(nmp);
return 0;
}
/****************************************************************************
* Name: nfs_statfs
*
* Description: Return filesystem statistics
*
****************************************************************************/
static int nfs_statfs(struct inode *mountpt, struct statfs *sbp)
{
struct nfs_statfs *sfp;
struct nfsmount *nmp;
int error = 0;
uint64_t tquad;
void *datareply;
struct FSSTAT3args fsstat;
/* Sanity checks */
DEBUGASSERT(mountpt && mountpt->i_private);
/* Get the mountpoint private data from the inode structure */
nmp = (struct nfsmount*)mountpt->i_private;
/* Check if the mount is still healthy */
nfs_semtake(nmp);
error = nfs_checkmount(nmp);
if (error < 0)
{
fdbg("romfs_checkmount failed: %d\n", error);
goto errout_with_semaphore;
}
/* Fill in the statfs info */
memset(sbp, 0, sizeof(struct statfs));
sbp->f_type = NFS_SUPER_MAGIC;
if ((nmp->nm_flag & NFSMNT_GOTFSINFO) == 0)
{
(void)nfs_fsinfo(mountpt, NULL, NULL);
}
nfsstats.rpccnt[NFSPROC_FSSTAT]++;
memset(&fsstat, 0, sizeof(struct FSSTAT3args));
fsstat.fsroot = nmp->nm_fh;
error = nfs_request(nmp, NFSPROC_FSSTAT, &fsstat, &datareply);
if (error)
{
goto errout_with_semaphore;
}
sfp = (struct nfs_statfs *) datareply;
nmp->nm_head->n_fattr = sfp->obj_attributes;
sbp->f_bsize = NFS_FABLKSIZE;
tquad = fxdr_hyper(&sfp->sf_tbytes);
sbp->f_blocks = tquad / (uint64_t) NFS_FABLKSIZE;
tquad = fxdr_hyper(&sfp->sf_fbytes);
sbp->f_bfree = tquad / (uint64_t) NFS_FABLKSIZE;
tquad = fxdr_hyper(&sfp->sf_abytes);
sbp->f_bavail = tquad / (uint64_t) NFS_FABLKSIZE;
tquad = fxdr_hyper(&sfp->sf_tfiles);
sbp->f_files = tquad;
tquad = fxdr_hyper(&sfp->sf_ffiles);
sbp->f_ffree = tquad;
sbp->f_namelen = NAME_MAX;
errout_with_semaphore:
nfs_semgive(nmp);
return error;
}
/****************************************************************************
* Name: nfs_remove
*
* Description: Remove a file
*
****************************************************************************/
static int nfs_remove(struct inode *mountpt, const char *relpath)
{
struct nfsmount *nmp;
struct nfsnode *np;
void *datareply;
struct REMOVE3args remove;
struct REMOVE3resok *resok;
int error = 0;
/* Sanity checks */
DEBUGASSERT(mountpt && mountpt->i_private);
/* Get the mountpoint private data from the inode structure */
nmp = (struct nfsmount*)mountpt->i_private;
np = nmp->nm_head;
/* Check if the mount is still healthy */
nfs_semtake(nmp);
error = nfs_checkmount(nmp);
if (error == 0)
{
/* If the file is open, the correct behavior is to remove the file
* name, but to keep the file cluster chain in place until the last
* open reference to the file is closed.
*/
/* Remove the file */
if (np->nfsv3_type != NFREG)
{
error = EPERM;
goto errout_with_semaphore;
}
/* Do the rpc */
nfsstats.rpccnt[NFSPROC_REMOVE]++;
memset(&remove, 0, sizeof(struct REMOVE3args));
remove.object.dir = np->n_fhp;
remove.object.name = relpath;
error = nfs_request(nmp, NFSPROC_REMOVE, &remove, &datareply);
/* Kludge City: If the first reply to the remove rpc is lost..
* the reply to the retransmitted request will be ENOENT
* since the file was in fact removed
* Therefore, we cheat and return success.
*/
if (error == ENOENT)
{
error = 0;
}
if (error)
{
goto errout_with_semaphore;
}
resok = (struct REMOVE3resok *) datareply;
np->n_fattr = resok->dir_wcc.after;
np->n_flag |= NMODIFIED;
}
NFS_INVALIDATE_ATTRCACHE(np);
errout_with_semaphore:
nfs_semgive(nmp);
return error;
}
/****************************************************************************
* Name: nfs_mkdir
*
* Description: Create a directory
*
****************************************************************************/
static int nfs_mkdir(struct inode *mountpt, const char *relpath, mode_t mode)
{
struct nfsv3_sattr sp;
struct nfsmount *nmp;
struct nfsnode *np;
struct MKDIR3args mkir;
struct MKDIR3resok *resok;
void *datareply;
int error = 0;
/* Sanity checks */
DEBUGASSERT(mountpt && mountpt->i_private);
/* Get the mountpoint private data from the inode structure */
nmp = (struct nfsmount*) mountpt->i_private;
np = nmp->nm_head;
/* Check if the mount is still healthy */
nfs_semtake(nmp);
error = nfs_checkmount(nmp);
if (error != 0)
{
goto errout_with_semaphore;
}
nfsstats.rpccnt[NFSPROC_MKDIR]++;
memset(&mkir, 0, sizeof(struct MKDIR3args));
mkir.where.dir = np->n_fhp;
mkir.where.name = relpath;
sp.sa_modetrue = nfs_true;
sp.sa_mode = txdr_unsigned(mode);
sp.sa_uidfalse = nfs_xdrneg1;
sp.sa_gidfalse = nfs_xdrneg1;
sp.sa_sizefalse = nfs_xdrneg1;
sp.sa_atimetype = txdr_unsigned(NFSV3SATTRTIME_TOCLIENT);
sp.sa_mtimetype = txdr_unsigned(NFSV3SATTRTIME_TOCLIENT);
memset(&sp.sa_atime, 0, sizeof(nfstime3));
memset(&sp.sa_mtime, 0, sizeof(nfstime3));
mkir.attributes = sp;
error = nfs_request(nmp, NFSPROC_MKDIR, &mkir, &datareply);
if (error)
{
goto errout_with_semaphore;
}
resok = (struct MKDIR3resok *) datareply;
np->nfsv3_type = NFDIR;
np->n_fhp = resok->handle;
np->n_fattr = resok->obj_attributes;
np->n_flag |= NMODIFIED;
NFS_INVALIDATE_ATTRCACHE(np);
errout_with_semaphore:
nfs_semgive(nmp);
return error;
}
/****************************************************************************
* Name: nfs_rmdir
*
* Description: Remove a directory
*
****************************************************************************/
static int nfs_rmdir(struct inode *mountpt, const char *relpath)
{
struct nfsmount *nmp;
struct nfsnode *np;
struct RMDIR3args rmdir;
struct RMDIR3resok *resok;
void *datareply;
int error = 0;
/* Sanity checks */
DEBUGASSERT(mountpt && mountpt->i_private);
/* Get the mountpoint private data from the inode structure */
nmp = (struct nfsmount *)mountpt->i_private;
np = nmp->nm_head;
/* Check if the mount is still healthy */
nfs_semtake(nmp);
error = nfs_checkmount(nmp);
if (error == 0)
{
/* Remove the directory */
if (np->nfsv3_type != NFDIR)
{
error = EPERM;
goto errout_with_semaphore;
}
/* Do the rpc */
nfsstats.rpccnt[NFSPROC_RMDIR]++;
memset(&rmdir, 0, sizeof(struct RMDIR3args));
rmdir.object.dir = np->n_fhp;
rmdir.object.name = relpath;
error = nfs_request(nmp, NFSPROC_RMDIR, &rmdir, &datareply);
if (error == ENOENT)
{
error = 0;
}
if (error)
{
goto errout_with_semaphore;
}
resok = (struct RMDIR3resok *) datareply;
np->n_fattr = resok->dir_wcc.after;
np->n_flag |= NMODIFIED;
}
NFS_INVALIDATE_ATTRCACHE(np);
errout_with_semaphore:
nfs_semgive(nmp);
return error;
}
/****************************************************************************
* Name: nfs_rename
*
* Description: Rename a file or directory
*
****************************************************************************/
static int nfs_rename(struct inode *mountpt, const char *oldrelpath,
const char *newrelpath)
{
struct nfsmount *nmp;
struct nfsnode *np;
void *datareply;
struct RENAME3args rename;
struct RENAME3resok *resok;
int error = 0;
/* Sanity checks */
DEBUGASSERT(mountpt && mountpt->i_private);
/* Get the mountpoint private data from the inode structure */
nmp = (struct nfsmount *)mountpt->i_private;
np = nmp->nm_head;
/* Check if the mount is still healthy */
nfs_semtake(nmp);
error = nfs_checkmount(nmp);
if (error != 0)
{
goto errout_with_semaphore;
}
if (np->nfsv3_type != NFREG && np->nfsv3_type != NFDIR)
{
fdbg("open eacces typ=%d\n", np->nfsv3_type);
error= -EACCES;
goto errout_with_semaphore;
}
nfsstats.rpccnt[NFSPROC_RENAME]++;
memset(&rename, 0, sizeof(struct RENAME3args));
rename.from.dir = np->n_fhp;
rename.from.name = oldrelpath;
rename.to.dir = np->n_fhp;
rename.to.name = newrelpath;
error = nfs_request(nmp, NFSPROC_RENAME, &rename, &datareply);
/* ENOENT => 0 assuming that it is a reply to a retry. */
if (error == ENOENT)
{
error = 0;
}
if (error)
{
goto errout_with_semaphore;
}
resok = (struct RENAME3resok *) datareply;
np->n_fattr = resok->todir_wcc.after;
np->n_flag |= NMODIFIED;
NFS_INVALIDATE_ATTRCACHE(np);
errout_with_semaphore:
nfs_semgive(nmp);
return error;
}
/****************************************************************************
* Name: nfs_fsinfo
*
* Description: Return information about a file or directory
*
****************************************************************************/
static int nfs_fsinfo(struct inode *mountpt, const char *relpath, struct stat *buf)
{
struct nfsv3_fsinfo *fsp;
struct FSINFOargs fsinfo;
struct nfsmount *nmp;
uint32_t pref, max;
int error = 0;
void *datareply;
/* Sanity checks */
DEBUGASSERT(mountpt && mountpt->i_private);
/* Get the mountpoint private data from the inode structure */
nmp = (struct nfsmount*)mountpt->i_private;
/* Check if the mount is still healthy */
nfs_semtake(nmp);
error = nfs_checkmount(nmp);
if (error != 0)
{
goto errout_with_semaphore;
}
memset(buf, 0, sizeof(struct stat));
nfsstats.rpccnt[NFSPROC_FSINFO]++;
fsinfo.fsroot = nmp->nm_fh;
error = nfs_request(nmp, NFSPROC_FSINFO, &fsinfo, &datareply);
if (error)
{
goto errout_with_semaphore;
}
fsp = (struct nfsv3_fsinfo *) datareply;
nmp->nm_head->n_fattr = fsp->obj_attributes;
pref = fxdr_unsigned(uint32_t, fsp->fs_wtpref);
if (pref < nmp->nm_wsize)
{
nmp->nm_wsize = (pref + NFS_FABLKSIZE - 1) & ~(NFS_FABLKSIZE - 1);
}
max = fxdr_unsigned(uint32_t, fsp->fs_wtmax);
if (max < nmp->nm_wsize)
{
nmp->nm_wsize = max & ~(NFS_FABLKSIZE - 1);
if (nmp->nm_wsize == 0)
nmp->nm_wsize = max;
}
pref = fxdr_unsigned(uint32_t, fsp->fs_rtpref);
if (pref < nmp->nm_rsize)
{
nmp->nm_rsize = (pref + NFS_FABLKSIZE - 1) & ~(NFS_FABLKSIZE - 1);
}
max = fxdr_unsigned(uint32_t, fsp->fs_rtmax);
if (max < nmp->nm_rsize)
{
nmp->nm_rsize = max & ~(NFS_FABLKSIZE - 1);
if (nmp->nm_rsize == 0)
{
nmp->nm_rsize = max;
}
}
pref = fxdr_unsigned(uint32_t, fsp->fs_dtpref);
if (pref < nmp->nm_readdirsize)
{
nmp->nm_readdirsize = (pref + NFS_DIRBLKSIZ - 1) & ~(NFS_DIRBLKSIZ - 1);
}
if (max < nmp->nm_readdirsize)
{
nmp->nm_readdirsize = max & ~(NFS_DIRBLKSIZ - 1);
if (nmp->nm_readdirsize == 0)
{
nmp->nm_readdirsize = max;
}
}
buf->st_mode = fxdr_hyper(&fsp->obj_attributes.fa_mode);
buf->st_size = fxdr_hyper(&fsp->obj_attributes.fa3_size);
buf->st_blksize = 0;
buf->st_blocks = 0;
buf->st_mtime = fxdr_hyper(&fsp->obj_attributes.fa3_mtime);
buf->st_atime = fxdr_hyper(&fsp->obj_attributes.fa3_atime);
buf->st_ctime = fxdr_hyper(&fsp->obj_attributes.fa3_ctime);
nmp->nm_flag |= NFSMNT_GOTFSINFO;
errout_with_semaphore:
nfs_semgive(nmp);
return error;
}
#ifdef COMP
/****************************************************************************
* Name: nfs_sync
*
* Description: Flush out the buffer cache
*
****************************************************************************/
int nfs_sync(struct file *filep)
{
struct inode *inode;
struct nfsmount *nmp;
struct nfsnode *np;
int error = 0;
/* Sanity checks */
DEBUGASSERT(filep->f_priv != NULL && filep->f_inode != NULL);
/* Recover our private data from the struct file instance */
np = filep->f_priv;
inode = filep->f_inode;
nmp = inode->i_private;
DEBUGASSERT(nmp != NULL);
/* Make sure that the mount is still healthy */
nfs_semtake(nmp);
error = nfs_checkmount(nmp);
if (error != 0)
{
goto errout_with_semaphore;
}
/* Force stale buffer cache information to be flushed. */
/* Check if the has been modified in any way */
if ((np->n_flag & NMODIFIED) != 0)
{
//error = VOP_FSYNC(vp, cred, waitfor, p);
}
return error;
errout_with_semaphore:
nfs_semgive(nmp);
return error;
}
#endif