- 11 Mar, 2011 30 commits
-
-
Andy Adamson authored
If a data server is unavailable, go through MDS. Mark the deviceid containing the data server as a negative cache entry. Do not try to connect to any data server on a deviceid marked as a negative cache entry. Mark any layout that tries to use the marked deviceid as failed. Inodes with a layout marked as fails will not use the layout for I/O, and will not perform any more layoutgets. Inodes without a layout will still do layoutget, but the layout will get marked immediately. Signed-off-by:
Andy Adamson <andros@netapp.com> Signed-off-by:
Trond Myklebust <Trond.Myklebust@netapp.com>
-
Christoph Hellwig authored
No need for generic cache with only one user. Keep a simple hash of deviceids in the filelayout driver. Signed-off-by:
Christoph Hellwig <hch@infradead.org> Acked-by:
-
Andy Adamson authored
Use our own async error handler. Mark the layout as failed and retry i/o through the MDS on specified errors. Update the mds_offset in nfs_readpage_retry so that a failed short-read retry to a DS gets correctly resent through the MDS. Signed-off-by:
-
Andy Adamson authored
Attempt a pNFS file layout read by setting up the nfs_read_data struct and calling nfs_initiate_read with the data server rpc client and the filelayout rpc call ops. Error handling is implemented in a subsequent patch. Signed-off-by:
Andy Adamson <andros@citi.umich.edu> Signed-off-by:
Dean Hildebrand <dhildeb@us.ibm.com> Signed-off-by:
Fred Isaman <iisaman@citi.umich.edu> Signed-off-by:
Fred Isaman <iisaman@netapp.com> Signed-off-by:
Mingyang Guo <guomingyang@nrchpc.ac.cn> Signed-off-by:
Oleg Drokin <green@linuxhacker.ru> Signed-off-by:
Ricardo Labiaga <Ricardo.Labiaga@netapp.com> Tested-by:
Benny Halevy <bhalevy@panasas.com> Signed-off-by:
-
Fred Isaman authored
Prepare for filelayout_read_pagelist with helper functions that find the correct data server, filehandle, and offset. Signed-off-by:
Marc Eshel <eshel@almaden.ibm.com> Signed-off-by:
Mike Sager <sager@netapp.com> Signed-off-by:
Tao Guo <guotao@nrchpc.ac.cn> Signed-off-by:
Tigran Mkrtchyan <tigran@anahit.desy.de> Signed-off-by:
Tigran Mkrtchyan <tigran.mkrtchyan@desy.de> Signed-off-by:
-
Andy Adamson authored
Introduce a data server set_client and init session following the nfs4_set_client and nfs4_init_session convention. Once a new nfs_client is on the nfs_client_list, the nfs_client cl_cons_state serializes access to creating an nfs_client struct with matching properties. Use the new nfs_get_client() that initializes new clients. Signed-off-by:
-
Andy Adamson authored
Separate the rpc run portion of nfs_read_rpcsetup into a new function nfs_initiate_read that is called for normal NFS I/O. Add a pNFS read_pagelist function that is called instead of nfs_intitate_read for pNFS reads. Signed-off-by:
Boaz Harrosh <bharrosh@panasas.com> Signed-off-by:
-
Fred Isaman authored
Move the pnfs_update_layout call location to nfs_pageio_do_add_request(). Grab the lseg sent in the doio function to nfs_read_rpcsetup and attach it to each nfs_read_data so it can be sent to the layout driver. Signed-off-by:
-
Fred Isaman authored
Add a pg_test layout driver hook which is used to avoid coelescing I/O across layout stripes. Signed-off-by:
-
Fred Isaman authored
Prepare put_lseg and get_lseg to be called from the pNFS I/O code. Pull common code from pnfs_lseg_locked to call from pnfs_lseg. Inline pnfs_lseg_locked into it's only caller. Signed-off-by:
-
Andy Adamson authored
The DS only role cannot be used to mount. Signed-off-by:
-
Andy Adamson authored
Data servers cannot send nfs4_proc_get_lease_time. but still need to setup state renewal. Add the NFS_CS_CHECK_LEASE_TIME bit to indicate if the lease time can be checked. Signed-off-by:
-
Andy Adamson authored
Data servers not sharing a session with the mount MDS always have an empty cl_superblocks list. Replace the cl_superblocks empty list check to see if it is time to shut down renewd with the NFS_CS_STOP_RENEW bit which is not set by such a data server. Signed-off-by:
-
Andy Adamson authored
Data servers require a zero stateid seqid, and there is no advantage to not doing the same for all NFSv4.1 Signed-off-by:
-
Andy Adamson authored
Now nfs_get_client returns an nfs_client ready to be used no matter if it was found or created. Signed-off-by:
-
Andy Adamson authored
Prevents an Oops triggered by CB_LAYOUTRECALL and LAYOUTGET race on a pnfs_layout_hdr first pnfs_layout_segment. Signed-off-by:
-
Andy Adamson authored
Cleanup nfs_read_data. We also won't use CONFIG_NFS_V4_1 for additional NFSv4.1 fields in subsequent patches. Signed-off-by:
-
Fred Isaman authored
The return values are not used by any callers. Signed-off-by:
-
Fred Isaman authored
The code was doing nothing more in either branch of the if. Signed-off-by:
-
Fred Isaman authored
rpc_run_task can only fail if it is not passed in a preallocated task. However, that is not at all clear with the current code. So remove several impossible to occur failure checks. Signed-off-by:
-
Fred Isaman authored
queue_work() only returns 0 or 1, never a negative value. Signed-off-by:
-
Fred Isaman authored
The pnfs code was using throughout the lock order i_lock, cl_lock. This conflicts with the nfs delegation code. Rework the pnfs code to avoid taking both locks simultaneously. Currently the code takes the double lock to add/remove the layout to a nfs_client list, while atomically checking that the list of lsegs is empty. To avoid this, we rely on existing serializations. When a layout is initialized with lseg count equal zero, LAYOUTGET's openstateid serialization is in effect, making it safe to assume it stays zero unless we change it. And once a layout's lseg count drops to zero, it is set as DESTROYED and so will stay at zero. Signed-off-by:
-
Fred Isaman authored
We do not need to clear the NFS_LAYOUT_BULK_RECALL, as setting it guarantees that NFS_LAYOUT_DESTROYED will be set once any outstanding io is finished. Signed-off-by:
-
Fred Isaman authored
The code could violate the following from RFC5661, section 12.5.3: "Once a client has no more layouts on a file, the layout stateid is no longer valid and MUST NOT be used." This can occur when a layout already has a lseg, starts another non-everlapping LAYOUTGET, and a CB_LAYOUTRECALL for the existing lseg is processed before we hit pnfs_layout_process(). Solve by setting, each time the client has no more lsegs for a file, a flag which blocks further use of the layout and triggers its removal. This also fixes a second bug which occurs in the same instance as above. If we actually use pnfs_layout_process, we add the new lseg to the layout, but the layout has been removed from the nfs_client list by the intervening CB_LAYOUTRECALL and will not be added back. Thus the newly acquired lseg will not be properly returned in the event of a subsequent CB_LAYOUTRECALL. Signed-off-by:
-
Chuck Lever authored
There have been a number of recent reports that NFSROOT is no longer working with default mount options, but fails only with certain NICs. Brian Downing <bdowning@lavos.net> bisected to commit 56463e50 "NFS: Use super.c for NFSROOT mount option parsing". Among other things, this commit changes the default mount options for NFSROOT to use TCP instead of UDP as the underlying transport. TCP seems less able to deal with NICs that are slow to initialize. The system logs that have accompanied reports of problems all show that NFSROOT attempts to establish a TCP connection before the NIC is fully initialized, and thus the TCP connection attempt fails. When a TCP connection attempt fails during a mount operation, the NFS stack needs to fail the operation. Usually user space knows how and when to retry it. The network layer does not report a distinct error code for this particular failure mode. Thus, there isn't a clean way for the RPC client to see that it needs to retry in this case, but not in others. Because NFSROOT is used in some environments where it is not possible to update the kernel command line to specify "udp", the proper thing to do is change NFSROOT to use UDP by default, as it did before commit 56463e50. To make it easier to see how to change default mount options for NFSROOT and to distinguish default settings from mandatory settings, I've adjusted a couple of areas to document the specifics. root_nfs_cat() is also modified to deal with commas properly when concatenating strings containing mount option lists. This keeps root_nfs_cat() call sites simpler, now that we may be concatenating multiple mount option strings. Tested-by:
Brian Downing <bdowning@lavos.net> Tested-by:
Mark Brown <broonie@opensource.wolfsonmicro.com> Signed-off-by:
Chuck Lever <chuck.lever@oracle.com> Cc: <stable@kernel.org> # 2.6.37 Signed-off-by:
-
Huang Weiyi authored
Remove duplicated #include('s) in fs/nfs/nfs4proc.c Signed-off-by:Huang Weiyi <weiyi.huang@gmail.com> Signed-off-by:
-
Trond Myklebust authored
There are no more external users of nfs4_state_mark_reclaim_nograce() or nfs4_state_mark_reclaim_reboot(), so mark them as static. Signed-off-by: -
Trond Myklebust authored
Signed-off-by: -
Trond Myklebust authored
We want SEQUENCE status bits to be handled by the state manager in order to avoid threading issues. Signed-off-by: -
Trond Myklebust authored
nfs4_schedule_state_recovery() should only be used when we need to force the state manager to check the lease. If we just want to start the state manager in order to handle a state recovery situation, we should be using nfs4_schedule_state_manager(). This patch fixes the abuses of nfs4_schedule_state_recovery() by replacing its use with a set of helper functions that do the right thing. Signed-off-by:
-
- 10 Mar, 2011 10 commits
-
-
Andy Adamson authored
Signed-off-by:
-
Andy Adamson authored
Signed-off-by:
-
Ricardo Labiaga authored
Fix bug where we currently retry the EXCHANGEID call again, eventhough we already have a valid clientid. Instead, delay and retry the CREATE_SESSION call. Signed-off-by:
-
Ben Hutchings authored
xs_create_sock() is supposed to return a pointer or an ERR_PTR-encoded error, but it currently returns 0 if xs_bind() fails. Signed-off-by:
Ben Hutchings <bhutchings@solarflare.com> Cc: stable@kernel.org [v2.6.37] Signed-off-by:
-
Frank Filz authored
(try3-resend) Fix nfs_compat_user_ino64 so it doesn't cause problems if bit 31 or 63 are set in fileid The problem was use of an int32, which when converted to a uint64 is sign extended resulting in a fileid that doesn't fit in 32 bits even though the intent of the function is to fit the fileid into 32 bits. Signed-off-by:
Frank Filz <ffilzlnx@us.ibm.com> Reviewed-by:
Jeff Layton <jlayton@redhat.com> [Trond: Added an include for compat.h] Signed-off-by:
-
Jovi Zhang authored
this commit fix compilation warning as following: linux-2.6/fs/nfs/nfs4proc.c:3265: warning: comparison of distinct pointer types lacks a cast Signed-off-by:
Jovi Zhang <bookjovi@gmail.com> Signed-off-by:
-
Stanislav Fomichev authored
add kmalloc return value check in decode_and_add_ds Signed-off-by:
Stanislav Fomichev <kernel@fomichev.me> Signed-off-by:
-
Jesper Juhl authored
We leak the memory allocated to 'ctxt' when we return after 'ib_dma_mapping_error()' returns !=0. Signed-off-by:
Jesper Juhl <jj@chaosbits.net> Signed-off-by:
-
Jeff Layton authored
I've been adding in more artificial delays in the NFSv4 commit and close codepaths to uncover races. The kernel I'm testing has the patch to close the race in __rpc_wait_for_completion_task that's in Trond's cthon2011 branch. The reproducer I've been using does this in a loop: mkdir("DIR"); fd = open("DIR/FILE", O_WRONLY|O_CREAT|O_EXCL, 0644); write(fd, "abcdefg", 7); close(fd); unlink("DIR/FILE"); rmdir("DIR"); The above reproducer shouldn't result in any silly-renaming. However, when I add a "msleep(100)" just after the nfs_commit_clear_lock call in nfs_commit_release, I can almost always force one to occur. If I can force it to occur with that, then it can happen without that delay given the right timing. nfs_commit_inode waits for the NFS_INO_COMMIT bit to clear when called with FLUSH_SYNC set. nfs_commit_rpcsetup on the other hand does not wait for the task to complete before putting its reference to it, so the last reference get put in rpc_release task and gets queued to a workqueue. In this situation, the last open context reference may be put by the COMMIT release instead of the close() syscall. The close() syscall returns too quickly and the unlink runs while the d_count is still high since the COMMIT release hasn't put its dentry reference yet. Fix this by having rpc_commit_rpcsetup wait for the RPC call to complete before putting the task reference when FLUSH_SYNC is set. With this, the last reference is put by the process that's initiating the FLUSH_SYNC commit and the race is closed. Signed-off-by: -
Trond Myklebust authored
Although they run as rpciod background tasks, under normal operation (i.e. no SIGKILL), functions like nfs_sillyrename(), nfs4_proc_unlck() and nfs4_do_close() want to be fully synchronous. This means that when we exit, we want all references to the rpc_task to be gone, and we want any dentry references etc. held by that task to be released. For this reason these functions call __rpc_wait_for_completion_task(), followed by rpc_put_task() in the expectation that the latter will be releasing the last reference to the rpc_task, and thus ensuring that the callback_ops->rpc_release() has been called synchronously. This patch fixes a race which exists due to the fact that rpciod calls rpc_complete_task() (in order to wake up the callers of __rpc_wait_for_completion_task()) and then subsequently calls rpc_put_task() without ensuring that these two steps are done atomically. In order to avoid adding new spin locks, the patch uses the existing waitqueue spin lock to order the rpc_task reference count releases between the waiting process and rpciod. The common case where nobody is waiting for completion is optimised for by checking if the RPC_TASK_ASYNC flag is cleared and/or if the rpc_task reference count is 1: in those cases we drop trying to grab the spin lock, and immediately free up the rpc_task. Those few processes that need to put the rpc_task from inside an asynchronous context and that do not care about ordering are given a new helper: rpc_put_task_async(). Signed-off-by:
-
