Print this page
11083 support NFS server in zone
Portions contributed by: Dan Kruchinin <dan.kruchinin@nexenta.com>
Portions contributed by: Stepan Zastupov <stepan.zastupov@gmail.com>
Portions contributed by: Joyce McIntosh <joyce.mcintosh@nexenta.com>
Portions contributed by: Mike Zeller <mike@mikezeller.net>
Portions contributed by: Dan McDonald <danmcd@joyent.com>
Portions contributed by: Gordon Ross <gordon.w.ross@gmail.com>
Portions contributed by: Vitaliy Gusev <gusev.vitaliy@gmail.com>
Reviewed by: Rick McNeal <rick.mcneal@nexenta.com>
Reviewed by: Rob Gittins <rob.gittins@nexenta.com>
Reviewed by: Sanjay Nadkarni <sanjay.nadkarni@nexenta.com>
Reviewed by: Jason King <jbk@joyent.com>
Reviewed by: C Fraire <cfraire@me.com>
Change-Id: I22f289d357503f9b48a0bc2482cc4328a6d43d16


   7  *
   8  * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
   9  * or http://www.opensolaris.org/os/licensing.
  10  * See the License for the specific language governing permissions
  11  * and limitations under the License.
  12  *
  13  * When distributing Covered Code, include this CDDL HEADER in each
  14  * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
  15  * If applicable, add the following below this CDDL HEADER, with the
  16  * fields enclosed by brackets "[]" replaced with your own identifying
  17  * information: Portions Copyright [yyyy] [name of copyright owner]
  18  *
  19  * CDDL HEADER END
  20  */
  21 
  22 /*
  23  * Copyright 2007 Sun Microsystems, Inc.  All rights reserved.
  24  * Use is subject to license terms.
  25  */
  26 




  27 #include <fs/fs_subr.h>
  28 
  29 #include <sys/errno.h>
  30 #include <sys/file.h>
  31 #include <sys/kmem.h>
  32 #include <sys/kobj.h>
  33 #include <sys/cmn_err.h>
  34 #include <sys/stat.h>
  35 #include <sys/systm.h>
  36 #include <sys/sysmacros.h>
  37 #include <sys/atomic.h>
  38 #include <sys/vfs.h>
  39 #include <sys/vfs_opreg.h>
  40 
  41 #include <sharefs/sharefs.h>
  42 
  43 /*
  44  * sharefs_snap_create: create a large character buffer with
  45  * the shares enumerated.
  46  */
  47 static int
  48 sharefs_snap_create(shnode_t *sft)
  49 {
  50         sharetab_t              *sht;
  51         share_t                 *sh;
  52         size_t                  sWritten = 0;
  53         int                     iCount = 0;
  54         char                    *buf;
  55 
  56         rw_enter(&sharefs_lock, RW_WRITER);
  57         rw_enter(&sharetab_lock, RW_READER);
  58 
  59         if (sft->sharefs_snap) {
  60                 /*
  61                  * Nothing has changed, so no need to grab a new copy!
  62                  */
  63                 if (sft->sharefs_generation == sharetab_generation) {
  64                         rw_exit(&sharetab_lock);
  65                         rw_exit(&sharefs_lock);
  66                         return (0);
  67                 }
  68 
  69                 ASSERT(sft->sharefs_size != 0);
  70                 kmem_free(sft->sharefs_snap, sft->sharefs_size + 1);
  71                 sft->sharefs_snap = NULL;
  72         }
  73 
  74         sft->sharefs_size = sharetab_size;
  75         sft->sharefs_count = sharetab_count;
  76 
  77         if (sft->sharefs_size == 0) {
  78                 rw_exit(&sharetab_lock);
  79                 rw_exit(&sharefs_lock);
  80                 return (0);
  81         }
  82 
  83         sft->sharefs_snap = kmem_zalloc(sft->sharefs_size + 1, KM_SLEEP);
  84 
  85         buf = sft->sharefs_snap;
  86 
  87         /*
  88          * Walk the Sharetab, dumping each entry.
  89          */
  90         for (sht = sharefs_sharetab; sht != NULL; sht = sht->s_next) {
  91                 int     i;
  92 
  93                 for (i = 0; i < SHARETAB_HASHES; i++) {
  94                         for (sh = sht->s_buckets[i].ssh_sh;
  95                             sh != NULL;
  96                             sh = sh->sh_next) {
  97                                 int     n;
  98 
  99                                 if ((sWritten + sh->sh_size) >
 100                                     sft->sharefs_size) {
 101                                         goto error_fault;
 102                                 }
 103 
 104                                 /*
 105                                  * Note that sh->sh_size accounts
 106                                  * for the field seperators.
 107                                  * We need to add one for the EOL
 108                                  * marker. And we should note that
 109                                  * the space is accounted for in
 110                                  * each share by the EOS marker.


 115                                     sh->sh_path,
 116                                     sh->sh_res,
 117                                     sh->sh_fstype,
 118                                     sh->sh_opts,
 119                                     sh->sh_descr);
 120 
 121                                 if (n != sh->sh_size) {
 122                                         goto error_fault;
 123                                 }
 124 
 125                                 sWritten += n;
 126                                 iCount++;
 127                         }
 128                 }
 129         }
 130 
 131         /*
 132          * We want to record the generation number and
 133          * mtime inside this snapshot.
 134          */
 135         gethrestime(&sharetab_snap_time);
 136         sft->sharefs_snap_time = sharetab_snap_time;
 137         sft->sharefs_generation = sharetab_generation;
 138 
 139         ASSERT(iCount == sft->sharefs_count);
 140 
 141         rw_exit(&sharetab_lock);
 142         rw_exit(&sharefs_lock);
 143         return (0);
 144 
 145 error_fault:
 146 
 147         kmem_free(sft->sharefs_snap, sft->sharefs_size + 1);
 148         sft->sharefs_size = 0;
 149         sft->sharefs_count = 0;
 150         sft->sharefs_snap = NULL;
 151         rw_exit(&sharetab_lock);
 152         rw_exit(&sharefs_lock);
 153 
 154         return (EFAULT);
 155 }
 156 
 157 /* ARGSUSED */
 158 static int
 159 sharefs_getattr(vnode_t *vp, vattr_t *vap, int flags, cred_t *cr,
 160     caller_context_t *ct)
 161 {
 162         timestruc_t     now;
 163         shnode_t        *sft = VTOSH(vp);

 164 
 165         vap->va_type = VREG;
 166         vap->va_mode = S_IRUSR | S_IRGRP | S_IROTH;
 167         vap->va_nodeid = SHAREFS_INO_FILE;
 168         vap->va_nlink = 1;
 169 
 170         rw_enter(&sharefs_lock, RW_READER);
 171 
 172         /*
 173          * If we get asked about a snapped vnode, then
 174          * we must report the data in that vnode.
 175          *
 176          * Else we report what is currently in the
 177          * sharetab.
 178          */
 179         if (sft->sharefs_real_vp) {
 180                 rw_enter(&sharetab_lock, RW_READER);
 181                 vap->va_size = sharetab_size;
 182                 vap->va_mtime = sharetab_mtime;
 183                 rw_exit(&sharetab_lock);
 184         } else {
 185                 vap->va_size = sft->sharefs_size;
 186                 vap->va_mtime = sft->sharefs_snap_time;
 187         }
 188         rw_exit(&sharefs_lock);
 189 
 190         gethrestime(&now);
 191         vap->va_atime = vap->va_ctime = now;
 192 
 193         vap->va_uid = 0;
 194         vap->va_gid = 0;
 195         vap->va_rdev = 0;
 196         vap->va_blksize = DEV_BSIZE;
 197         vap->va_nblocks = howmany(vap->va_size, vap->va_blksize);
 198         vap->va_seq = 0;
 199         vap->va_fsid = vp->v_vfsp->vfs_dev;
 200 
 201         return (0);
 202 }
 203 
 204 /* ARGSUSED */
 205 static int
 206 sharefs_access(vnode_t *vp, int mode, int flags, cred_t *cr,
 207     caller_context_t *ct)
 208 {


 242 
 243         vp->v_flag |= VROOT | VNOCACHE | VNOMAP | VNOSWAP | VNOMOUNT;
 244 
 245         *vpp = vp;
 246         VN_RELE(ovp);
 247 
 248         sft = VTOSH(vp);
 249 
 250         /*
 251          * No need for the lock, no other thread can be accessing
 252          * this data structure.
 253          */
 254         atomic_inc_32(&sft->sharefs_refs);
 255         sft->sharefs_real_vp = 0;
 256 
 257         /*
 258          * Since the sharetab could easily change on us whilst we
 259          * are dumping an extremely huge sharetab, we make a copy
 260          * of it here and use it to dump instead.
 261          */
 262         error = sharefs_snap_create(sft);

 263 
 264         return (error);
 265 }
 266 
 267 /* ARGSUSED */
 268 int
 269 sharefs_close(vnode_t *vp, int flag, int count,
 270     offset_t off, cred_t *cr, caller_context_t *ct)
 271 {
 272         shnode_t        *sft = VTOSH(vp);

 273 
 274         if (count > 1)
 275                 return (0);
 276 
 277         rw_enter(&sharefs_lock, RW_WRITER);
 278         if (vp->v_count == 1) {
 279                 if (sft->sharefs_snap != NULL) {
 280                         kmem_free(sft->sharefs_snap, sft->sharefs_size + 1);
 281                         sft->sharefs_size = 0;
 282                         sft->sharefs_snap = NULL;
 283                         sft->sharefs_generation = 0;
 284                 }
 285         }
 286         atomic_dec_32(&sft->sharefs_refs);
 287         rw_exit(&sharefs_lock);
 288 
 289         return (0);
 290 }
 291 
 292 /* ARGSUSED */
 293 static int
 294 sharefs_read(vnode_t *vp, uio_t *uio, int ioflag, cred_t *cr,
 295                         caller_context_t *ct)
 296 {
 297         shnode_t        *sft = VTOSH(vp);
 298         off_t           off = uio->uio_offset;
 299         size_t          len = uio->uio_resid;
 300         int             error = 0;

 301 
 302         rw_enter(&sharefs_lock, RW_READER);
 303 
 304         /*
 305          * First check to see if we need to grab a new snapshot.
 306          */
 307         if (off == (off_t)0) {
 308                 rw_exit(&sharefs_lock);
 309                 error = sharefs_snap_create(sft);
 310                 if (error) {
 311                         return (EFAULT);
 312                 }
 313                 rw_enter(&sharefs_lock, RW_READER);
 314         }
 315 
 316         /* LINTED */
 317         if (len <= 0 || off >= sft->sharefs_size) {
 318                 rw_exit(&sharefs_lock);
 319                 return (error);
 320         }
 321 
 322         if ((size_t)(off + len) > sft->sharefs_size)
 323                 len = sft->sharefs_size - off;
 324 
 325         if (off < 0 || len > sft->sharefs_size) {
 326                 rw_exit(&sharefs_lock);
 327                 return (EFAULT);
 328         }
 329 
 330         if (len != 0) {
 331                 error = uiomove(sft->sharefs_snap + off,
 332                     len, UIO_READ, uio);
 333         }
 334 
 335         rw_exit(&sharefs_lock);
 336         return (error);
 337 }
 338 
 339 /* ARGSUSED */
 340 static void
 341 sharefs_inactive(vnode_t *vp, cred_t *cr, caller_context_t *tx)
 342 {
 343         gfs_file_t      *fp = vp->v_data;
 344         shnode_t        *sft;

 345 
 346         sft = (shnode_t *)gfs_file_inactive(vp);
 347         if (sft) {
 348                 rw_enter(&sharefs_lock, RW_WRITER);
 349                 if (sft->sharefs_snap != NULL) {
 350                         kmem_free(sft->sharefs_snap, sft->sharefs_size + 1);
 351                 }
 352 
 353                 kmem_free(sft, fp->gfs_size);
 354                 rw_exit(&sharefs_lock);
 355         }
 356 }
 357 
 358 vnode_t *
 359 sharefs_create_root_file(vfs_t *vfsp)
 360 {
 361         vnode_t         *vp;
 362         shnode_t        *sft;
 363 
 364         vp = gfs_root_create_file(sizeof (shnode_t),
 365             vfsp, sharefs_ops_data, SHAREFS_INO_FILE);
 366 
 367         sft = VTOSH(vp);
 368 
 369         sft->sharefs_real_vp = 1;
 370 
 371         return (vp);
 372 }
 373 
 374 const fs_operation_def_t sharefs_tops_data[] = {


   7  *
   8  * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
   9  * or http://www.opensolaris.org/os/licensing.
  10  * See the License for the specific language governing permissions
  11  * and limitations under the License.
  12  *
  13  * When distributing Covered Code, include this CDDL HEADER in each
  14  * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
  15  * If applicable, add the following below this CDDL HEADER, with the
  16  * fields enclosed by brackets "[]" replaced with your own identifying
  17  * information: Portions Copyright [yyyy] [name of copyright owner]
  18  *
  19  * CDDL HEADER END
  20  */
  21 
  22 /*
  23  * Copyright 2007 Sun Microsystems, Inc.  All rights reserved.
  24  * Use is subject to license terms.
  25  */
  26 
  27 /*
  28  * Copyright 2018 Nexenta Systems, Inc.
  29  */
  30 
  31 #include <fs/fs_subr.h>
  32 
  33 #include <sys/errno.h>
  34 #include <sys/file.h>
  35 #include <sys/kmem.h>
  36 #include <sys/kobj.h>
  37 #include <sys/cmn_err.h>
  38 #include <sys/stat.h>
  39 #include <sys/systm.h>
  40 #include <sys/sysmacros.h>
  41 #include <sys/atomic.h>
  42 #include <sys/vfs.h>
  43 #include <sys/vfs_opreg.h>
  44 
  45 #include <sharefs/sharefs.h>
  46 
  47 /*
  48  * sharefs_snap_create: create a large character buffer with
  49  * the shares enumerated.
  50  */
  51 static int
  52 sharefs_snap_create(sharetab_globals_t *sg, shnode_t *sft)
  53 {
  54         sharetab_t              *sht;
  55         share_t                 *sh;
  56         size_t                  sWritten = 0;
  57         int                     iCount = 0;
  58         char                    *buf;
  59 
  60         rw_enter(&sg->sharefs_lock, RW_WRITER);
  61         rw_enter(&sg->sharetab_lock, RW_READER);
  62 
  63         if (sft->sharefs_snap) {
  64                 /*
  65                  * Nothing has changed, so no need to grab a new copy!
  66                  */
  67                 if (sft->sharefs_generation == sg->sharetab_generation) {
  68                         rw_exit(&sg->sharetab_lock);
  69                         rw_exit(&sg->sharefs_lock);
  70                         return (0);
  71                 }
  72 
  73                 ASSERT(sft->sharefs_size != 0);
  74                 kmem_free(sft->sharefs_snap, sft->sharefs_size + 1);
  75                 sft->sharefs_snap = NULL;
  76         }
  77 
  78         sft->sharefs_size = sg->sharetab_size;
  79         sft->sharefs_count = sg->sharetab_count;
  80 
  81         if (sft->sharefs_size == 0) {
  82                 rw_exit(&sg->sharetab_lock);
  83                 rw_exit(&sg->sharefs_lock);
  84                 return (0);
  85         }
  86 
  87         sft->sharefs_snap = kmem_zalloc(sft->sharefs_size + 1, KM_SLEEP);
  88 
  89         buf = sft->sharefs_snap;
  90 
  91         /*
  92          * Walk the Sharetab, dumping each entry.
  93          */
  94         for (sht = sg->sharefs_sharetab; sht != NULL; sht = sht->s_next) {
  95                 int     i;
  96 
  97                 for (i = 0; i < SHARETAB_HASHES; i++) {
  98                         for (sh = sht->s_buckets[i].ssh_sh;
  99                             sh != NULL;
 100                             sh = sh->sh_next) {
 101                                 int     n;
 102 
 103                                 if ((sWritten + sh->sh_size) >
 104                                     sft->sharefs_size) {
 105                                         goto error_fault;
 106                                 }
 107 
 108                                 /*
 109                                  * Note that sh->sh_size accounts
 110                                  * for the field seperators.
 111                                  * We need to add one for the EOL
 112                                  * marker. And we should note that
 113                                  * the space is accounted for in
 114                                  * each share by the EOS marker.


 119                                     sh->sh_path,
 120                                     sh->sh_res,
 121                                     sh->sh_fstype,
 122                                     sh->sh_opts,
 123                                     sh->sh_descr);
 124 
 125                                 if (n != sh->sh_size) {
 126                                         goto error_fault;
 127                                 }
 128 
 129                                 sWritten += n;
 130                                 iCount++;
 131                         }
 132                 }
 133         }
 134 
 135         /*
 136          * We want to record the generation number and
 137          * mtime inside this snapshot.
 138          */
 139         gethrestime(&sg->sharetab_snap_time);
 140         sft->sharefs_snap_time = sg->sharetab_snap_time;
 141         sft->sharefs_generation = sg->sharetab_generation;
 142 
 143         ASSERT(iCount == sft->sharefs_count);
 144 
 145         rw_exit(&sg->sharetab_lock);
 146         rw_exit(&sg->sharefs_lock);
 147         return (0);
 148 
 149 error_fault:
 150 
 151         kmem_free(sft->sharefs_snap, sft->sharefs_size + 1);
 152         sft->sharefs_size = 0;
 153         sft->sharefs_count = 0;
 154         sft->sharefs_snap = NULL;
 155         rw_exit(&sg->sharetab_lock);
 156         rw_exit(&sg->sharefs_lock);
 157 
 158         return (EFAULT);
 159 }
 160 
 161 /* ARGSUSED */
 162 static int
 163 sharefs_getattr(vnode_t *vp, vattr_t *vap, int flags, cred_t *cr,
 164     caller_context_t *ct)
 165 {
 166         timestruc_t     now;
 167         shnode_t        *sft = VTOSH(vp);
 168         sharetab_globals_t *sg = sharetab_get_globals(vp->v_vfsp->vfs_zone);
 169 
 170         vap->va_type = VREG;
 171         vap->va_mode = S_IRUSR | S_IRGRP | S_IROTH;
 172         vap->va_nodeid = SHAREFS_INO_FILE;
 173         vap->va_nlink = 1;
 174 
 175         rw_enter(&sg->sharefs_lock, RW_READER);
 176 
 177         /*
 178          * If we get asked about a snapped vnode, then
 179          * we must report the data in that vnode.
 180          *
 181          * Else we report what is currently in the
 182          * sharetab.
 183          */
 184         if (sft->sharefs_real_vp) {
 185                 rw_enter(&sg->sharetab_lock, RW_READER);
 186                 vap->va_size = sg->sharetab_size;
 187                 vap->va_mtime = sg->sharetab_mtime;
 188                 rw_exit(&sg->sharetab_lock);
 189         } else {
 190                 vap->va_size = sft->sharefs_size;
 191                 vap->va_mtime = sft->sharefs_snap_time;
 192         }
 193         rw_exit(&sg->sharefs_lock);
 194 
 195         gethrestime(&now);
 196         vap->va_atime = vap->va_ctime = now;
 197 
 198         vap->va_uid = 0;
 199         vap->va_gid = 0;
 200         vap->va_rdev = 0;
 201         vap->va_blksize = DEV_BSIZE;
 202         vap->va_nblocks = howmany(vap->va_size, vap->va_blksize);
 203         vap->va_seq = 0;
 204         vap->va_fsid = vp->v_vfsp->vfs_dev;
 205 
 206         return (0);
 207 }
 208 
 209 /* ARGSUSED */
 210 static int
 211 sharefs_access(vnode_t *vp, int mode, int flags, cred_t *cr,
 212     caller_context_t *ct)
 213 {


 247 
 248         vp->v_flag |= VROOT | VNOCACHE | VNOMAP | VNOSWAP | VNOMOUNT;
 249 
 250         *vpp = vp;
 251         VN_RELE(ovp);
 252 
 253         sft = VTOSH(vp);
 254 
 255         /*
 256          * No need for the lock, no other thread can be accessing
 257          * this data structure.
 258          */
 259         atomic_inc_32(&sft->sharefs_refs);
 260         sft->sharefs_real_vp = 0;
 261 
 262         /*
 263          * Since the sharetab could easily change on us whilst we
 264          * are dumping an extremely huge sharetab, we make a copy
 265          * of it here and use it to dump instead.
 266          */
 267         error = sharefs_snap_create(sharetab_get_globals(vp->v_vfsp->vfs_zone),
 268             sft);
 269 
 270         return (error);
 271 }
 272 
 273 /* ARGSUSED */
 274 int
 275 sharefs_close(vnode_t *vp, int flag, int count,
 276     offset_t off, cred_t *cr, caller_context_t *ct)
 277 {
 278         shnode_t        *sft = VTOSH(vp);
 279         sharetab_globals_t *sg = sharetab_get_globals(vp->v_vfsp->vfs_zone);
 280 
 281         if (count > 1)
 282                 return (0);
 283 
 284         rw_enter(&sg->sharefs_lock, RW_WRITER);
 285         if (vp->v_count == 1) {
 286                 if (sft->sharefs_snap != NULL) {
 287                         kmem_free(sft->sharefs_snap, sft->sharefs_size + 1);
 288                         sft->sharefs_size = 0;
 289                         sft->sharefs_snap = NULL;
 290                         sft->sharefs_generation = 0;
 291                 }
 292         }
 293         atomic_dec_32(&sft->sharefs_refs);
 294         rw_exit(&sg->sharefs_lock);
 295 
 296         return (0);
 297 }
 298 
 299 /* ARGSUSED */
 300 static int
 301 sharefs_read(vnode_t *vp, uio_t *uio, int ioflag, cred_t *cr,
 302     caller_context_t *ct)
 303 {
 304         shnode_t        *sft = VTOSH(vp);
 305         off_t           off = uio->uio_offset;
 306         size_t          len = uio->uio_resid;
 307         int             error = 0;
 308         sharetab_globals_t *sg = sharetab_get_globals(vp->v_vfsp->vfs_zone);
 309 
 310         rw_enter(&sg->sharefs_lock, RW_READER);
 311 
 312         /*
 313          * First check to see if we need to grab a new snapshot.
 314          */
 315         if (off == (off_t)0) {
 316                 rw_exit(&sg->sharefs_lock);
 317                 error = sharefs_snap_create(sg, sft);
 318                 if (error) {
 319                         return (EFAULT);
 320                 }
 321                 rw_enter(&sg->sharefs_lock, RW_READER);
 322         }
 323 
 324         /* LINTED */
 325         if (len <= 0 || off >= sft->sharefs_size) {
 326                 rw_exit(&sg->sharefs_lock);
 327                 return (error);
 328         }
 329 
 330         if ((size_t)(off + len) > sft->sharefs_size)
 331                 len = sft->sharefs_size - off;
 332 
 333         if (off < 0 || len > sft->sharefs_size) {
 334                 rw_exit(&sg->sharefs_lock);
 335                 return (EFAULT);
 336         }
 337 
 338         if (len != 0) {
 339                 error = uiomove(sft->sharefs_snap + off,
 340                     len, UIO_READ, uio);
 341         }
 342 
 343         rw_exit(&sg->sharefs_lock);
 344         return (error);
 345 }
 346 
 347 /* ARGSUSED */
 348 static void
 349 sharefs_inactive(vnode_t *vp, cred_t *cr, caller_context_t *tx)
 350 {
 351         gfs_file_t      *fp = vp->v_data;
 352         shnode_t        *sft;
 353         sharetab_globals_t *sg = sharetab_get_globals(vp->v_vfsp->vfs_zone);
 354 
 355         sft = (shnode_t *)gfs_file_inactive(vp);
 356         if (sft) {
 357                 rw_enter(&sg->sharefs_lock, RW_WRITER);
 358                 if (sft->sharefs_snap != NULL) {
 359                         kmem_free(sft->sharefs_snap, sft->sharefs_size + 1);
 360                 }
 361 
 362                 kmem_free(sft, fp->gfs_size);
 363                 rw_exit(&sg->sharefs_lock);
 364         }
 365 }
 366 
 367 vnode_t *
 368 sharefs_create_root_file(vfs_t *vfsp)
 369 {
 370         vnode_t         *vp;
 371         shnode_t        *sft;
 372 
 373         vp = gfs_root_create_file(sizeof (shnode_t),
 374             vfsp, sharefs_ops_data, SHAREFS_INO_FILE);
 375 
 376         sft = VTOSH(vp);
 377 
 378         sft->sharefs_real_vp = 1;
 379 
 380         return (vp);
 381 }
 382 
 383 const fs_operation_def_t sharefs_tops_data[] = {