fs/nfs/nfs4_subr.c

4  * The contents of this file are subject to the terms of the
23  * Use is subject to license terms.
130  * Copy an nfs_fh4.  The destination storage (to->nfs_fh4_val) must already
135 nfs_fh4_copy(nfs_fh4 *from, nfs_fh4 *to)
137 	to->nfs_fh4_len = from->nfs_fh4_len;
138 	bcopy(from->nfs_fh4_val, to->nfs_fh4_val, to->nfs_fh4_len);
336 	case NFS4ERR_FHEXPIRED:	/* if got here, failed to get a new fh */
391 		 * EACCESS, which translates to "permission denied".
491 	/* Invalid to specify 1 billion (or more) nsecs */
534  * Converts a utf8 string to a valid null terminated filename string.
576  * Converts a utf8 string to a C string.
626  * str_to_utf8 - converts a null-terminated C string to a utf8 string
757  * First security flavor of choice is to use sv_secdata
771 	 * SV4_TRYSECINFO indicates to try the secinfo list from
773 	 * sv_currsec to the selected security mechanism for
802 		/* sv_currsec points to one of the entries in sv_secinfo */
843 	 * Find the correct entry in the cache to check for free
864 	 * If we did have a preallocated entry, then just add it to
889 	 * list, then move it to the front of the list to try to take
997 	 * Set read buffer size to rsize
1018 	 * clget4 calls authget() to get an auth handle. For RPCSEC_GSS
1019 	 * security flavor, the client tries to establish a security context
1072 	 * Add the free client handle to the front of the list.
1073 	 * This way, the list will be sorted in youngest to oldest
1082 #define	CL_HOLDTIME	60	/* time to hold client handles */
1098 	 * Need to reclaim some memory, so step through the cache
1106 	 * Here we step through each non-NULL quadruple and start to
1107 	 * construct the reclaim list pointed to by cp.  Note that
1111 	 * been appended to the rear of cp.
1119 		 * in youngest to oldest order so that when the
1146 	 * If cp is empty, then there is nothing to reclaim here.
1152 	 * Step through the list of entries to free, destroying each client
1181 	 * The system is low on memory; go through and try to reclaim some from
1193  * These units are in "eights" of a second to avoid multiplies
1252 	 * are guaranteed to reprocess the retry as a new request.
1268 	 * transmission so that it can tell if it needs to continue.
1270 	 * For zone shutdown, behave as above to encourage quick
1286 		 * decided to reboot.
1334 		 * even trying to send out the request because we
1335 		 * won't be able to block waiting for the response.
1358 			 * There is no way to recover from this error,
1371 			 * loop, so check to see if the specific
1375 			 * ECONNRESET to indicate why.
1533 		 * then it is okay to skip the down and printed
1642  * Routine to look up the filehandle for the given path and rootvp.
1645  * - success: returns zero and *statp is set to NFS4_OK, and *fhp is
1768 	 * It is too hard to remember where all the OP_LOOKUPs are
1814 	 * the rnode to find the fh for the new server.
1876 		 * If we ran into filehandle problems, we should try to
1878 		 * But we need to avoid loops.
1951 	 * so we would expect to mark the object dead.
1979 	 * If we got parent info, use it to update the parent
2028  * Our caller has a filehandle for ".." relative to a particular
2029  * directory object.  We want to find or create a parent vnode
2031  * a vnode from this filehandle, but we need to also make sure
2034  * we have a remap failure, we won't be able to reopen this
2037  * for a missing parent FH and try a remap to recover from it.
2042  * set it to TRUE.
2092 	 * Since we don't have the rnode, we have to go over the wire.
2164 	 * It is possible for remap_lookup() to return with no error,
2343  * We provide a set of interfaces to allow the rest of the system to utilize
2438  * rddir4_cache * - A pointer to a rddir4_cache is returned to the caller.
2456 	 * Check to see if the readdir cache has been disabled.  If so, then
2489 		 * the cache due to a lockless allocation.  If we find that no
2504 		 * First, try to allocate an entry without sleeping.  If that
2519 		 * additional overhead by having to search the cache again,
2528 		 * We need to take another pass through the cache
2529 		 * since we dropped our lock to perform the alloc.
2537 	 * Check to see if we need to free our entry.  This can happen if
2538 	 * another thread came along beat us to the insert.  We can
2540 	 * would have a reference to this entry.
2549 	 * Found something.  Make sure it's ready to return.
2555 	 * until this completes.  The RDDIRWAIT bit is set to indicate that
2557 	 * is done, it should do a cv_broadcast to wakeup all of the threads
2558 	 * waiting for it to finish. If the thread wakes up to find that
2559 	 * someone new is now trying to complete the the entry, go back
2560 	 * to sleep.
2646  * Increment the reference count to this cache element.
2659  * Release a reference to this cache element.  If the count is zero then
2670 	 * Check to see if we have any waiters.  If so, we can wake them
2717 		 * Currently only the global zone can write to kstats, but we
2729 		 * display them as-is to all zones since in fact they apply to
2800  * Called by endpnt_destructor to make sure the client handles are
2803  * (rather than relying on clfini_zone to clean up) because the ZSD
2804  * callbacks have no ordering mechanism, so we have no way to ensure
2835 	 * This needs to be done before we call zone_key_create().
2868  * VOP_RWLOCK() is held for write access to prevent a race condition
2872  * This is a copy of nfs_directio, changes here may need to be made
2934  * Return TRUE if the file has any pages.  Always go back to
2935  * the master vnode to check v_pages since none of the shadows
2952  * This table is used to determine whether the client should attempt
2955  * the error value that corresponds to the clnt_stat value in the
2956  * table is non-zero, then that is the error to be returned AND
2961  * allow the functions to detect the change and issue a warning.