iozone tests were failing intermittently against emc-2 due to out-of-order LAYOUTCOMMITs that both specified new_last_offset
nfs41_open_state now maintains a cached value of the last_offset, and avoids sending it with LAYOUTCOMMIT unless the new last_offset is greater than the cached offset. this cached value is initialized on open/delegation, and updated on setattr for size
Signed-off-by: Casey Bodley <cbodley@citi.umich.edu>
when marshalling of the open downcall or allocating buffer for the downcall
fails, upcall.status gets set but we still need to cancel the open we
just did. instead check for non-allocated open state as evidence that
handle_open() failed and we don't need to cancel vs the other failure.
added 2011 year to the copyright line
added authors info to the license
added UofM license to libtirpc files that we modified
(but i probably missed some)
delegation stateid arguments to DELEGRETURN and OPEN are now stateid_arg, for use with recover_stateid_delegation(). added a nfs41_delegation_state pointer to stateid_arg, for when a delegation stateid is used in the absence of nfs41_open_state (DELEGRETURN, SETATTR)
recovery during a call to nfs41_delegation_to_open() requires special attention; recover_stateid_delegation() has to handle the case where recover_open() already reclaimed the open stateid. it does this by returning BAD_STATEID instead of retrying the OPEN (which would generate yet another open stateid)
Signed-off-by: Casey Bodley <cbodley@citi.umich.edu>
recover_client_state() flags all delegations as revoked before starting open state recovery. if recover_open() finds that its delegation is revoked, it attempts to recover it using CLAIM_PREVIOUS. if its delegation has already been reclaimed by another open, it can skip reclaiming the open stateid (provided it has no byte-range locks to reclaim). after all opens have been reclaimed, any delegations still marked 'revoked' are passed to recover_delegation(). recover_delegation() also uses CLAIM_PREVIOUS (or CLAIM_NULL outside of the grace period) to reclaim the delegation, but has to throw away the open stateid with CLOSE
added a try_recovery argument to nfs41_delegreturn() and nfs41_delegation_granted(), so it can be called by recover_open() if OPEN grants an unexpected open
Signed-off-by: Casey Bodley <cbodley@citi.umich.edu>
nfs41_open() in open.c is no longer used for recovery, so made static and renamed to do_open(). renamed nfs41_rpc_open() back to nfs41_open()
Signed-off-by: Casey Bodley <cbodley@citi.umich.edu>
prevent an OPEN with OPEN4_CREATE from breaking a write delegation by sending a SETATTR instead
moved static function remove_unsupported_attrs() from setattr.c to nfs41_superblock_supported_attrs() in nfs41_superblock.c, now used by both setattr.c and delegation.c
Signed-off-by: Casey Bodley <cbodley@citi.umich.edu>
new function nfs41_delegation_return() for synchronous delegation return. uses a condition variable to wait if another thread is already returning the delegation
if nfs41_delegate_open() would conflict with a delegation, return it before sending the OPEN
return the delegation before sending LINK, RENAME, REMOVE, and SETATTR
all of this functionality is dependent on the preprocessor define DELEGATION_RETURN_ON_CONFLICT (on by default). if not defined, nfs41_delegation_return() is a noop
Signed-off-by: Casey Bodley <cbodley@citi.umich.edu>
combined nfs41_lock_stateid_arg() into nfs41_open_stateid_arg(). if a delegation is present, use the delegation stateid before looking at lock/open stateids. if a delegation recall is in progress, wait on its condition variable before falling back to the open stateid
made nfs41_lock_stateid_arg() static to lock.c because of its special semantics; open_to_lock_owner4 for LOCK won't accept a delegation stateid, so nfs41_delegation_to_open() is called to convert it
Signed-off-by: Casey Bodley <cbodley@citi.umich.edu>
added delegation.c and .h
nfs41_client stores a list of nfs41_delegation_state
new function nfs41_delegate_open() to look for a compatible delegation before calling nfs41_open()
if nfs41_open() is granted a delegation, call nfs41_delegation_granted() to register it with the client
client calls nfs41_client_delegation_free() on unmount to free list of delegations
Signed-off-by: Casey Bodley <cbodley@citi.umich.edu>
combined nfs41_open() and nfs41_open_reclaim() into nfs41_rpc_open() by factoring out the open_claim4 argument. new function nfs41_open() in open.c deals with the nfs41_open_state, adding it to the client's list, and handles any delegations granted
added xdr for OPEN CLAIM types CLAIM_DELEGATE_CUR, CLAIM_DELEG_CUR_FH, CLAIM_DELEGATE_PREV, CLAIM_DELEG_PREV_FH (the _FH types are new to 4.1, and not supported by linux server)
Signed-off-by: Casey Bodley <cbodley@citi.umich.edu>
proposes an alternate solution for attempts to set file size without an open sta
teid. instead of acquiring one by sending OPEN, fail the request with ACCESS_DE
NIED
according the MS File System Algorithms documentation for setting FileAllocation
Information and FileEndOfFileInformation [http://msdn.microsoft.com/en-us/librar
y/ff469355%28v=PROT.10%29.aspx]:
"If Open.GrantedAccess does not contain FILE_WRITE_DATA, the operation MUST be f
ailed with STATUS_ACCESS_DENIED"
-removes open_owner_id, access_mask, access_mode from setattr upcall arguments
-moves map_access_2_allowdeny() back to open.c as a static function, since handl
e_setattr() was its only other call site
if we are doing CREATE_NEW file creation, then based on whether or not
we have a persistent session, we'll send either GUARDED4 create for
persistent session and EXCLUSIVE4_1 create otherwise.
every upcall (except few) pass session and open_state pointer, so
add that to marshal_nfs41_header() in the driver. remove passing
of session and open_state elsewhere in marshal functions.
in the deamon, upcall.c now reads and stores pointers to session
and open_state in nfs41_upcall datastructure instead of having
each individual upcall store their own pointers. setattrl
and readdir args keeping pointer because the rest of the code
uses them a lot.
in upcall_parse() up refcounts on session and open_state if
valid handles were passed in. down refcounts upcall_cleanup() as
before. but need to be careful with count value for mount and open
upcalls. we need to take an extra ref because upcall_cleanup() now
will always decrement it.
nfs41_lock_stateid_arg() is now called only once in handle_read()/handle_write(), and pnfs_read()/pnfs_write() no longer depend on nfs41_open_state
Signed-off-by: Casey Bodley <cbodley@citi.umich.edu>
if we see NFS4ERR_NO_GRACE from recovery operations, it means we lost our state due to a lease expiration rather than a server reboot. in this case, it's possible that conflicting locks were granted to other clients, so we have to try normal OPEN/LOCK operations to recover our state. because they're sent during recovery, nfs41_open() and nfs41_lock() take a new 'bool_t try_recovery' argument so we can avoid recursion
if these operations fail due to conflicting locks, we have no choice but to return errors to the application. using a stateid that was revoked due to lease expiration results in NFS4ERR_EXPIRED, and we map this error to ERROR_FILE_INVALID: The volume for a file has been externally altered so that the opened file is no longer valid.
Signed-off-by: Casey Bodley <cbodley@citi.umich.edu>
nfs41_open_state maintains a list of outstanding byte-range locks by calling open_lock_add() and open_lock_remove() in lock.c
during client state recovery, after reclaiming each OPEN stateid, send LOCK requests with reclaim=TRUE for each lock it owns, and update the open's lock stateid with the result
added 'bool_t reclaim' argument to nfs41_lock(); when set, compound_encode_send_decode() is called with try_recovery=FALSE to avoid recursive recovery
Signed-off-by: Casey Bodley <cbodley@citi.umich.edu>
operations that require a stateid now take stateid_arg for recovery information. these operations include close, setattr, lock/unlock, layoutget, and read/write (including pnfs)
nfs41_open_stateid_arg() locks nfs41_open_state and copies its stateid into a stateid_arg
nfs41_lock_stateid_arg() locks nfs41_open_state.last_lock and copies its stateid into a stateid_arg; if there is no lock state, it falls back to nfs41_open_stateid_arg()
pnfs_read/write() now take nfs41_open_state so they can generate stateid_args
Signed-off-by: Casey Bodley <cbodley@citi.umich.edu>
after the client and session have been recovered, loop through the client's list of open state, calling nfs41_open_reclaim() and updating the stateid on success
nfs41_open_state saves the share_access and share_deny fields from the initial open, for use with nfs41_open_reclaim()
Signed-off-by: Casey Bodley <cbodley@citi.umich.edu>
open state is added to the client's list on a successful call to nfs41_open(), and removed from the list on nfs41_close() regardless of success
Signed-off-by: Casey Bodley <cbodley@citi.umich.edu>
very similar to the issue with nfs41_open_state, an abandoned upcall could outlive its mount. to prevent their nfs41_root from being freed, upcalls need to hold a reference until they're finished. this also keeps all of its clients/sessions/rpc connections alive
Signed-off-by: Casey Bodley <cbodley@citi.umich.edu>
added call to upcall_cleanup() after both upcall_marshall() and upcall_cancel()
individual upcall operations define their nfs41_upcall_op structs locally, instead of putting tons of function prototypes in upcall.c
made the upcall_marshall() function optional; most marshall functions are noops
Signed-off-by: Casey Bodley <cbodley@citi.umich.edu>
while following symlinks on open, after we break from the loop we need to respect that nfs41_lookup could have returned some kind of error value. thus only return error_reparse if nfs41_lookup returned success or file_not_found error.
similarly in after calling nfs41_symlink_follow() only set the reparse_error if function was successful.
fixed a few cases of warning 4242: possible loss of data
wincrypt.h appears to come with windows.h in later versions of the ddk, but nfs41_client.c fails to compile in WDK 6001 without #include <wincrypt.h>
Signed-off-by: Casey Bodley <cbodley@citi.umich.edu>
avoid calling map_disposition_2_nfsopen() for this, because FILE_CREATE->ERROR_FILE_EXISTS is the only case we care about
Signed-off-by: Casey Bodley <cbodley@citi.umich.edu>
/* msdn: There is a maximum of 31 reparse points (and
* therefore symbolic links) allowed in a particular path. */
#define NFS41_MAX_SYMLINK_DEPTH 31
also added checks for the return value of nfs41_symlink_target() on open/link/rename
Signed-off-by: Casey Bodley <cbodley@citi.umich.edu>
when rename or link call nfs41_lookup() for the destination directory, they need to be able to handle ERROR_REPARSE and find the real dest dir
open now does the same thing when it sees ERROR_REPARSE; previously, it was only replacing the first symlink in the path, and could require multiple reparses on a path
modified nfs41_symlink_target() to support the case where the source and destination paths are the same (used by rename/link)
Signed-off-by: Casey Bodley <cbodley@citi.umich.edu>
added nfs41_file_info.symlink_dir to replace readdir's info.cansettime hack. when nfs_file_info_to_attributes() finds info.type==NF4LNK, it adds the FILE_ATTRIBUTE_DIRECTORY flag if info.symlink_dir is set
renamed nfs41_symlink_follow() to nfs41_symlink_target()
generalized lookup_symlink() into nfs41_symlink_follow(), which is called by readdir and open (also avoids an extra lookup)
added queries for symlink target type when doing normal GETATTRs (getattr.c) and opens with OPEN_REPARSE_POINT set (open.c)
Signed-off-by: Casey Bodley <cbodley@citi.umich.edu>
returns symlink_embedded=TRUE when the symlink isn't the last component, which gets passed to RxPrepareToReparseSymbolicLink()
Signed-off-by: Casey Bodley <cbodley@citi.umich.edu>
added check in handle_open() to avoid calling CREATE/OPEN when we're creating a symlink:
if (args->disposition == FILE_CREATE &&
args->access_mask == (FILE_WRITE_ATTRIBUTES | SYNCHRONIZE | DELETE) &&
args->access_mode == 0 &&
args->create_opts & FILE_OPEN_REPARSE_POINT)
these are the open arguments we get from the CreateSymbolicLink() syscall. by avoiding the call to CREATE/OPEN on handle_open(), we save ourselves from having to REMOVE the file before creating the symlink
added a check to handle_symlink() in case the file was actually created on open (an application could open the file with different arguments, and send the FSCTL_SET_REPARSE_POINT manually), and removes the file first
Signed-off-by: Casey Bodley <cbodley@citi.umich.edu>
added optional symlink argument to nfs41_create(), used when type is NF4LNK
changed createttype4.u.lnk.linkdata from char[] to const char* and updated encoding in nfs41_xdr.c
Signed-off-by: Casey Bodley <cbodley@citi.umich.edu>
no longer depends on nfs41_open_state or open_upcall_args
renamed to nfs41_symlink_follow() and added prototype to nfs41_ops.h
Signed-off-by: Casey Bodley <cbodley@citi.umich.edu>
added nfs41_abs_path symlink to struct open_upcall_args. we can't write the symlink target back to args->path anymore, since it's a pointer into the upcall buffer
Signed-off-by: Casey Bodley <cbodley@citi.umich.edu>
because we no longer have to convert strings from unicode, we can avoid copying them out of the upcall buffer
Signed-off-by: Casey Bodley <cbodley@citi.umich.edu>
when open parsing fails, we were still returning upcall.status==NO_ERROR, so the driver assumed the open succeeded. other operations then sent up an open_state==NULL, and crashed the daemon. when upcall_parse() returns an error, set upcall.status to notify the driver
upcall_parse() prints a 'parsing of upcall <name> failed with <error>.' message on failure, so i removed redundant messages from the individual upcall parsing functions
Signed-off-by: Casey Bodley <cbodley@citi.umich.edu>
when handle_open() encounters a file of type NF4LNK and the FILE_OPEN_REPARSE_POINT flag is not present, it calls READLINK for the symlink's target path. it then calls abs_path_link() to update the filename (args->path), whether the symlink target is an absolute or relative path. abs_path_link() also takes into account the special characters . and .., though it doesn't allow traversing .. entries below the root of the server's namespace
Signed-off-by: Casey Bodley <cbodley@citi.umich.edu>
on last_error == ERROR_REPARSE, the daemon converts args->path back to wchar and passes it down to the driver
Signed-off-by: Casey Bodley <cbodley@citi.umich.edu>
caused memory leaks of nfs41_open_state. we have the FileDispositionInformation upcall for exactly this, so there's no need to complicate the close code path
Signed-off-by: Casey Bodley <cbodley@citi.umich.edu>
