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>
sorry, earlier Casey, but the patch 'threading by io unit instead of stripe' from 6/20/2010 was nuts! with PNFS_THREAD_BY_SERVER disabled, we definitely -don't- want to create a separate thread for each io unit (each READ/WRITE request to a ds). we just want the one per stripe, as the intended alternative to PNFS_THREAD_BY_SERVER
Signed-off-by: Casey Bodley <cbodley@citi.umich.edu>
cancel_lock() can't rely on checking upcall.status, because a handle_lock() success could be overwritten by upcall_marshall() or failure to allocate the upcall reply buffer. added new flag lock_upcall_args.acquired for this purpose
Signed-off-by: Casey Bodley <cbodley@citi.umich.edu>
now that cancel_lock() and handle_unlock() depend on finding lock state in the open, the comment /* ignore errors from open_lock_add(); they just mean we won't be able to recover the lock after reboot */ no longer applies. allocate the lock state at the top of handle_lock(), so we can bail immediately on failure
Signed-off-by: Casey Bodley <cbodley@citi.umich.edu>
removed unused variable 'grandparent'
factored out the call to name_cache_entry_update() from root/non-root paths
Signed-off-by: Casey Bodley <cbodley@citi.umich.edu>
renamed name_cache_entry_refresh() to name_cache_entry_updated(). this function resets the entry's expiration timer. new function name_cache_entry_accessed() deals with moving the entry to the end of exp_entries to prevent eviction
name_cache_entry_accessed() now does the same for its parents as well. cache consistency of the parents is maintained because their expiration timers are unchanged. by 'refreshing' the parents of each name cache entry inserted, the eviction policy will always prefer child nodes (which are far less costly to evict)
Signed-off-by: Casey Bodley <cbodley@citi.umich.edu>
open_update_cache() wasn't retrying insert after the open_delegation_return() error case
nfs41_name_cache_delegreturn() needs a check for if (attributes->delegated) to avoid an extra deref when the call to nfs41_name_cache_insert() failed
added a TODO comment to nfs41_client_delegation_return_lru() for improvements to the naive algorithm
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.
to prevent delegations from claiming all available slots in the attribute cache, nfs41_name_cache_insert() now returns ERROR_TOO_MANY_OPEN_FILES when the number of delegated entries reaches 50% of the cache capacity. see comment 'delegations and cache feedback' in name_cache.c for details
when nfs41_open() sees this error from nfs41_name_cache_insert(), it calls new function nfs41_client_delegation_return_lru() to return the least-recently-used delegation and, if successful, loops back to nfs41_name_cache_insert(). nfs41_client_delegation_return_lru() returns NFS4ERR_BADHANDLE if all delegations are currently in use (associated with an existing open), in which case nfs41_open() returns the newly-granted delegation
Signed-off-by: Casey Bodley <cbodley@citi.umich.edu>
on opens that grant a delegation, return the delegation immediately if attribute cache insertion fails
nfs41_name_cache_insert() now returns success in the 'out_err_deleg' case, where name cache insertion failed but attr cache insertion was successful
Signed-off-by: Casey Bodley <cbodley@citi.umich.edu>
once the attribute cache fills up with delegated entries, attr_cache_find_or_create() will start returning ERROR_OUTOFMEMORY. this is a problem for nfs41_name_cache_insert(), because name_cache_find_or_create() will succeed and then name_cache_entry_update() will fail. this leaves behind a name cache entry with attributes=NULL, which is treated like a negative entry and causes later lookups to fail with ERROR_FILE_NOT_FOUND
added a call to name_cache_entry_invalidate() to clean up this entry if name_cache_entry_update() fails
Signed-off-by: Casey Bodley <cbodley@citi.umich.edu>
use GetTickCount64() for EXCHANGE_ID's client verifier instead of time(). the higher resolution helps prevent conflicting verifiers
Signed-off-by: Casey Bodley <cbodley@citi.umich.edu>
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)
new flags for nfs41_lock_state: exclusive, delegated, and id
on lock, open_lock_delegate() attempts to register a delegated lock instead of sending a LOCK request to the server
on unlock, open_unlock_delegate() removes/frees delegated locks instead of sending a LOCKU request
on delegreturn, delegation_flush_locks() sends LOCK requests for all delegated locks before returning a delegation
on recovery, recover_locks() avoids lock state recovery for delegated locks
Signed-off-by: Casey Bodley <cbodley@citi.umich.edu>
for delegations without an associated open, try the optional WANT_DELEGATION operation with CLAIM_PREVIOUS before falling back to OPEN/CLAIM_PREVIOUS. the advantage of WANT_DELEGATION is that it doesn't generate an open stateid, so we don't need the corresponding CLOSE as in recover_delegation_open(). if WANT_DELEGATION fails with NFS4ERR_NOTSUPP, don't try it again during that round of state recovery
Signed-off-by: Casey Bodley <cbodley@citi.umich.edu>
10.2.1. Delegation Recovery (re: client restart/lease expiration)
A server MAY support claim types of CLAIM_DELEGATE_PREV and CLAIM_DELEG_PREV_FH, and if it does, it MUST NOT remove delegations upon a CREATE_SESSION that confirm a client ID created by EXCHANGE_ID. Instead, the server MUST, for a period of time no less than that of the value of the lease_time attribute, maintain the client's delegations to allow time for the client to send CLAIM_DELEGATE_PREV and/or CLAIM_DELEG_PREV_FH requests. The server that supports CLAIM_DELEGATE_PREV and/or CLAIM_DELEG_PREV_FH MUST support the DELEGPURGE operation.
if there's a delegation to reclaim, recover_open_no_grace() now tries CLAIM_DELEGATE_PREV (supported by emc server, but not linux) before falling back to CLAIM_NULL
nfs41_client_delegation_recovery() sends DELEGPURGE to indicate that we're finished reclaiming delegations
Signed-off-by: Casey Bodley <cbodley@citi.umich.edu>
while the server is required to grant us delegations we reclaim through CLAIM_PREVIOUS, it may set the flag recalled=TRUE if it's not ready to grant the delegation. the client is then responsible for flushing modified state to the server and returning the delegation
new function nfs41_client_delegation_recovery() cleans up after delegation recovery by a) returning any delegations flagged as recalled, and b) 'forgetting' any delegations that we failed to reclaim. this function is called under the client's state recovery lock, directly after open and delegation state is recovered
added 'try_recovery' argument to delegation_return(), allowing it to be called during client state recovery. split out the code that removes the delegation from the client and its opens into delegation_remove(), which is what nfs41_client_delegation_recovery() uses to 'forget' a delegation
Signed-off-by: Casey Bodley <cbodley@citi.umich.edu>
