- 17 Jun, 2009 40 commits
-
-
Andy Adamson authored
nfsv4.1 clientid is established via EXCHANGE_ID rather than SETCLIENTID{,_CONFIRM} This is implemented using a new establish_clid method in nfs4_state_recovery_ops. nfs41: establish clientid via exchange id only if cred != NULL >From 2.6.26 reclaimer() uses machine cred for setting up the client id therefore it is never expected to be NULL. Signed-off-by:Rahul Iyer <iyer@netapp.com> [removed dprintk] Signed-off-by:
Benny Halevy <bhalevy@panasas.com> [nfs41: lease renewal] [revamped patch for new nfs4_state_manager design] Signed-off-by:
Trond Myklebust <Trond.Myklebust@netapp.com>
-
Andy Adamson authored
Use the machine cred for sending SEQUENCE to renew the client's lease. [revamp patch for new state management design starting 2.6.29] [nfs41: support minorversion 1 for nfs4_check_lease] Signed-off-by:
Andy Adamson <andros@netapp.com> Signed-off-by:
-
Benny Halevy authored
[moved nfs4_get_renew_cred related changes to "nfs41: introduce get_state_renewal_cred"] Signed-off-by:
-
Benny Halevy authored
Send a NFSv4.1 SEQUENCE op rather than RENEW that was deprecated in minorversion 1. Use the nfs_client minorversion to select reboot_recover/ network_partition_recovery/state_renewal ops. Note: we use reclaimer to create the nfs41 session before there are any cl_superblocks for the nfs_client. Signed-off-by:
-
Andy Adamson authored
Define a new session reset state which is set upon a sequence operation error in both the sync and async error handlers. Place all new requests and all but the last outstanding rpc on the slot_tbl_waitq. Spawn the recovery thread when the last slot is free. Call nfs4_proc_destroy_session, reinitialize the session, call nfs4_proc_create_session, clear the session reset state, and wake up the next task on the slot_tbl_waitq. Return the nfs4_proc_destroy_session status to the session reclaimer and check for NFS4ERR_BADSESSION and NFS4ERR_DEADSESSION. Other destroy session errors should be handled in nfs4_proc_destroy_session where the call can be retried with adjusted arguments. Signed-off-by: Andy Adamson<andros@netapp.com> Signed-off-by:
-
Andy Adamson authored
Remove checking for any errors that the SEQUENCE operation does not return. -NFS4ERR_STALE_CLIENTID, NFS4ERR_EXPIRED, NFS4ERR_CB_PATH_DOWN, NFS4ERR_BACK_CHAN_BUSY, NFS4ERR_OP_NOT_IN_SESSION. SEQUENCE operation error recovery is very primative, we only reset the session. Remove checking for any errors that are returned by the SEQUENCE operation, but that resetting the session won't address. NFS4ERR_RETRY_UNCACHED_REP, NFS4ERR_SEQUENCE_POS,NFS4ERR_TOO_MANY_OPS. Add error checking for missing SEQUENCE errors that a session reset will address. NFS4ERR_BAD_HIGH_SLOT, NFS4ERR_DEADSESSION, NFS4ERR_SEQ_FALSE_RETRY. A reset of the session is currently our only response to a SEQUENCE operation error. Don't reset the session on errors where a new session won't help. Don't reset the session on errors where a new session won't help. [nfs41: nfs4_async_handle_error update error checking] Signed-off-by:
-
Andy Adamson authored
[nfs41: change nfs4_restart_rpc argument] [nfs41: check for session not minorversion] [nfs41: trigger the state manager for session reset] Signed-off-by:
-
Andy Adamson authored
Signed-off-by: Andy Adamson<andros@netapp.com> Signed-off-by:
-
Andy Adamson authored
Move the code to reset a session from the session_reclaimer to the nfs4_state_manager. Destroy the session, and create a new one. Treat NFS4ERR_BADSESSION and NFS4ERR_DEADSESSION as a successful nfs4_proc_destroy_session. Signal nfs4_proc_create_session that this is a session reset so that the session slot table is re-used. If the clientid is stale, set both NFS4CLNT_LEASE_EXPIRED and NFS4CLNT_SESSION_SETUP bits and retry. Use a switch statement in nfs4_session_recovery_handle_error for future patche which will add handling for other errors. Signed-off-by:
-
Andy Adamson authored
At mount, nfs_alloc_client sets the cl_state NFS4CLNT_LEASE_EXPIRED bit and nfs4_alloc_session sets the NFS4CLNT_SESSION_SETUP bit, so both bits are set when nfs4_lookup_root calls nfs4_recover_expired_lease which schedules the nfs4_state_manager and waits for it to complete. Place the session setup after the clientid establishment in nfs4_state_manager so that the session is setup right after the clientid has been established without rescheduling the state manager. Unlike nfsv4.0, the nfs_client struct is not ready to use until the session has been established. Postpone marking the nfs_client struct to NFS_CS_READY until after a successful CREATE_SESSION call so that other threads cannot use the client until the session is established. If the EXCHANGE_ID call fails and the session has not been setup (the NFS4CLNT_SESSION_SETUP bit is set), mark the client with the error and return. If the session setup CREATE_SESSION call fails with NFS4ERR_STALE_CLIENTID which could occur due to server reboot or network partition inbetween the EXCHANGE_ID and CREATE_SESSION call, reset the NFS4CLNT_LEASE_EXPIRED and NFS4CLNT_SESSION_SETUP bits and try again. If the CREATE_SESSION call fails with other errors, mark the client with the error and return. Signed-off-by:
-
Andy Adamson authored
Separated from nfs41: schedule async session reset Do not kfree the session slot table upon session reset, just re-initialize it. Add a boolean to nfs4_proc_create_session to inidicate if this is a session reset or a session initialization. Signed-off-by:
-
Andy Adamson authored
Implement the sequence operation conforming to http://tools.ietf.org/html/draft-ietf-nfsv4-minorversion1-26 Check returned sessionid, slotid and slot sequenceid in decode_sequence. If the server returns different values for sessionID, slotID or slot sequence number than what was sent, the server is looney tunes. Pass the sequence operation status to nfs41_sequence_done in order to determine when to increment the slot sequence ID. Free slot is separated from sequence done. Signed-off-by:
Ricardo Labiaga <ricardo.labiaga@netapp.com> Signed-off-by: Andy Adamson<andros@umich.edu> [nfs41: sequence res use slotid] Signed-off-by:
-
Andy Adamson authored
The session is per struct nfs_client, not per nfs_server. Allow the handler to be called with no nfs_server which simplifies the nfs4_proc_async_sequence session renewal call and will let it be used by pnfs file layout data servers. Signed-off-by:
-
Andy Adamson authored
Implement the destroy_session operation conforming to http://tools.ietf.org/html/draft-ietf-nfsv4-minorversion1-26Signed-off-by:
-
Andy Adamson authored
Set the mount points rsize and wsize to the negotiated session fore channel maximum response and requeset size. These values will be bound checked in nfs_server_set_fsinfo. Signed-off-by:
-
Andy Adamson authored
Invalidate the session if the server returns invalid fore or back channel attributes. Use a KERN_WARNING to report the fatal session estabishment error. Signed-off-by:
-
Andy Adamson authored
Implement the create_session operation conforming to http://tools.ietf.org/html/draft-ietf-nfsv4-minorversion1-26 Set the real fore channel max operations to preserve server resources. Note: If the server returns < NFS4_MAX_OPS, the client will very soon get an NFS4ERR_TOO_MANY_OPS. A later patch will handle this. Set the max_rqst_sz and max_resp_sz to PAGE_SIZE - we preallocate the buffers. Set the back channel max_resp_sz_cached to zero to force the client to always set csa_cachethis to FALSE because the current implementation of the back channel DRC only supports caching the CB_SEQUENCE operation. The client back channel server supports one slot, and desires 2 operations per compound. Signed-off-by:
-
Andy Adamson authored
get_lease_time uses the FSINFO rpc operation to get the lease time attribute. nfs4_get_lease_time() is only called from the state manager on session setup so don't recover from clientid or sequence level errors. We do need to recover from NFS4ERR_DELAY or NFS4ERR_GRACE. Use NFS4_POLL_RETRY_MIN - the Linux server returns NFS4ERR_DELAY when an upcall is needed to resolve an uncached export referenced by a file handle. [nfs41: sequence res use slotid] Signed-off-by: Andy Adamson<andros@netapp.com> Signed-off-by:
-
Benny Halevy authored
Implement the exchange_id operation conforming to http://tools.ietf.org/html/draft-ietf-nfsv4-minorversion1-26 Unlike NFSv4.0, NFSv4.1 requires machine credentials. RPC_AUTH_GSS machine credentials will be passed into the kernel at mount time to be available for the exchange_id operation. RPC_AUTH_UNIX root mounts can use the UNIX root credential. Store the root credential in the nfs_client struct. Without a credential, NFSv4.1 state renewal fails. [nfs41: establish clientid via exchange id only if cred != NULL] Signed-off-by: Andy Adamson<andros@umich.edu> Signed-off-by:
-
Andy Adamson authored
Separate delegreturn calls from nfs41: sequence setup/done support Implement the delegreturn rpc_call_prepare method for asynchronuos nfs rpcs, call nfs41_setup_sequence from respective rpc_call_validate_args methods. Call nfs4_sequence_done from respective rpc_call_done methods. Note that we need to pass a pointer to the nfs_server in calls data for passing on to nfs4_sequence_done. Signed-off-by: Andy Adamson<andros@netapp.com> Signed-off-by:
-
Andy Adamson authored
Separate commit calls from nfs41: sequence setup/done support Implement the commit rpc_call_prepare method for asynchronuos nfs rpcs, call nfs41_setup_sequence from respective rpc_call_validate_args methods. Call nfs4_sequence_done from respective rpc_call_done methods. Note that we need to pass a pointer to the nfs_server in calls data for passing on to nfs4_sequence_done. Signed-off-by: Andy Adamson<andros@netapp.com> Signed-off-by:
Dean Hildebrand <dhildeb@us.ibm.com> Signed-off-by:
-
Andy Adamson authored
Separate write calls from nfs41: sequence setup/done support Implement the write rpc_call_prepare method for asynchronuos nfs rpcs, call nfs41_setup_sequence from respective rpc_call_validate_args methods. Call nfs4_sequence_done from respective rpc_call_done methods. Note that we need to pass a pointer to the nfs_server in calls data for passing on to nfs4_sequence_done. Signed-off-by:
Andy Adamson <andros@umich.edu> Signed-off-by:
-
Andy Adamson authored
Implement the read rpc_call_prepare method for asynchronuos nfs rpcs, call nfs41_setup_sequence from respective rpc_call_validate_args methods. Call nfs4_sequence_done from respective rpc_call_done methods. Note that we need to pass a pointer to the nfs_server in calls data for passing on to nfs4_sequence_done. Signed-off-by:
-
Andy Adamson authored
Implement the rpc_call_prepare methods for asynchronuos nfs rpcs, call nfs41_setup_sequence from respective rpc_call_validate_args methods. Call nfs4_sequence_done from respective rpc_call_done methods. Note that we need to pass a pointer to the nfs_server in calls data for passing on to nfs4_sequence_done. Signed-off-by: Andy Adamson<andros@netapp.com> Signed-off-by:
-
Andy Adamson authored
Separate nfs4_locku calls from nfs41: sequence setup/done support Call nfs4_sequence_done from respective rpc_call_done methods. Note that we need to pass a pointer to the nfs_server in calls data for passing on to nfs4_sequence_done. Signed-off-by:
-
Andy Adamson authored
Separate nfs4_lock calls from nfs41: sequence setup/done support Call nfs4_sequence_done from respective rpc_call_done methods. Note that we need to pass a pointer to the nfs_server in calls data for passing on to nfs4_sequence_done. Signed-off-by: Andy Adamson<andros@netapp.com> Signed-off-by:
-
Andy Adamson authored
Separate nfs4_open calls from nfs41: sequence setup/done support Call nfs4_sequence_done from respective rpc_call_done methods. Note that we need to pass a pointer to the nfs_server in calls data for passing on to nfs4_sequence_done. Signed-off-by: Andy Adamson<andros@netapp.com> Signed-off-by:
-
Andy Adamson authored
Separate nfs4_close calls from nfs41: sequence setup/done support Call nfs4_sequence_done from respective rpc_call_done methods. Note that we need to pass a pointer to the nfs_server in calls data for passing on to nfs4_sequence_done. Signed-off-by: Andy Adamson<andros@netapp.com> Signed-off-by:
-
Andy Adamson authored
Implement nfs4.1 synchronous rpc_call_done method that essentially just calls nfs4_sequence_done, that turns around and calls nfs41_sequence_done for minorversion1 rpcs. Signed-off-by:
-
Andy Adamson authored
Handle session level errors, update slot sequence id and sessions bookeeping, free slot. [nfs41: sequence res use slotid] Signed-off-by: Andy Adamson<andros@netapp.com> Signed-off-by:
-
Andy Adamson authored
[from nfs41: separate free slot from sequence done] Don't free the slot until after all rpc_restart_calls have completed. Session reset will require more work. As noted by Trond, since we're using rpc_wake_up_next rather than rpc_wake_up() we must always wake up the next task in the queue either by going through nfs4_free_slot, or just calling rpc_wake_up_next if no slot is to be freed. [nfs41: sequence res use slotid] [nfs41: remove SEQ4_STATUS_USE_TK_STATUS] [got rid of nfs4_sequence_res.sr_session, use nfs_client.cl_session instead] Signed-off-by:
-
Andy Adamson authored
Free a slot in the slot table. Mark the slot as free in the bitmap-based allocation table by clearing a bit corresponding to the slotid. Update lowest_free_slotid if freed slotid is lower than that. Update highest_used_slotid. In the case the freed slotid equals the highest_used_slotid, scan downwards for the next highest used slotid using the optimized fls* functions. Finally, wake up thread waiting on slot_tbl_waitq for a free slot to become available. Signed-off-by:
-
Andy Adamson authored
Allocate a slot in the session slot table and set the sequence op arguments. Called at the rpc prepare stage. Add a status to nfs41_sequence_res, initialize it to one so that we catch rpc level failures which do not go through decode_sequence which sets the new status field. Note that upon an rpc level failure, we don't know if the server processed the sequence operation or not. Proceed as if the server did process the sequence operation. Signed-off-by:
-
Benny Halevy authored
Find a free slot using bitmap-based allocation. Use the optimized ffz function to find a zero bit in the bitmap that indicates a free slot, starting the search from the 'lowest_free_slotid' position. If found, mark the slot as used in the bitmap, get the slot's slotid and seqid, and update max_slotid to be used by the SEQUENCE operation. Also, update lowest_free_slotid for next search. If no free slot was found the caller has to wait for a free slot (outside the scope of this function) Signed-off-by:
-
Andy Adamson authored
Perform the nfs4_setup_sequence in the rpc_call_prepare state. If a session slot is not available, we will rpc_sleep_on the slot wait queue leaving the tk_action as rpc_call_prepare. Once we have a session slot, hang on to it even through rpc_restart_calls. Ensure the nfs41_sequence_res sr_slot pointer is NULL before rpc_run_task is called as nfs41_setup_sequence will only find a new slot if it is NULL. A future patch will call free slot after any rpc_restart_calls, and handle the rpc restart that result from a sequence operation error. Signed-off-by:
-
Andy Adamson authored
Implement stubs for encode and decode sequence, defined as no-ops when CONFIG_NFS_V4_1 is not defined. Add the nfsv41 encode and decode sizes. Add encode_sequence to all nfs4_enc_* routines and decode_sequence to all nfs4_dec_* routines as required by v41. [was nfs41: minorversion support for xdr] [added nfs_client argument to encode_sequence so not to use sequence_args to pass sa_session] Signed-off-by: Andy Adamson<andros@netapp.com> Signed-off-by:
-
Benny Halevy authored
Signed-off-by:
-
Benny Halevy authored
As Trond suggested, rather than passing a constant to xdr_inline_pages, keep a running count of the expected reply bytes. In preparation for nfs41, where additional op sequence are expteced when talking to nfs41 servers. [NFS: cb_compoundhdr.replen is in words not bytes] Signed-off-by:
-
Benny Halevy authored
Signed-off-by:
-
Benny Halevy authored
replen holds the running count of expected reply bytes. repl will then be used by encoding routines for xdr_inline_pages offset after which data bytes are to be received directly into the xdr buffer pages. NOTE: According to the nfsv4 and v4.1 RFCs, the replied tag SHOULD be the same is the one sent, but this is not required as a MUST for the server to do so. The server may screw us if it replies a tag of a different length in the compound result. [NFS: cb_compoundhdr.replen is in words not bytes] Signed-off-by:
-
