- 10 Apr, 2018 39 commits
-
-
Trond Myklebust authored
Signed-off-by:
Trond Myklebust <trond.myklebust@primarydata.com> Signed-off-by:
Anna Schumaker <Anna.Schumaker@Netapp.com>
-
Trond Myklebust authored
Signed-off-by:
-
Trond Myklebust authored
Signed-off-by:
-
Trond Myklebust authored
Signed-off-by:
-
Trond Myklebust authored
Signed-off-by:
-
Trond Myklebust authored
Signed-off-by:
-
Trond Myklebust authored
Don't bother even recording an invalid change attribute if we hold a delegation since we already know the state of our attribute cache. We can rely on the fact that we will pick up a copy from the server when we return the delegation. Signed-off-by:
-
Trond Myklebust authored
Currently, if the NFS_INO_INVALID_ATTR flag is set, for instance by a call to nfs_post_op_update_inode_locked(), then it will not be cleared until all the attributes have been revalidated. This means, for instance, that NFSv4 writes will always force a full attribute revalidation. Track the ctime, mtime, size and change attribute separately from the other attributes so that we can have nfs_post_op_update_inode_locked() set them correctly, and later have the cache consistency bitmask be able to clear them. Signed-off-by:
-
Trond Myklebust authored
If we managed to revalidate all the attributes, then there is no reason to mark them as invalid again. We do, however want to ensure that we set nfsi->attrtimeo correctly. Signed-off-by:
-
Trond Myklebust authored
If we received weak cache consistency data from the server, then those attributes are up to date, and there is no reason to mark them as dirty in the attribute cache. Signed-off-by:
-
Trond Myklebust authored
Even if the change attribute is missing, it is still OK to mark the other attributes as being up to date. Signed-off-by:
-
Trond Myklebust authored
The cache validity attribute is unsigned long, so make sure that the flags are too. Signed-off-by:
-
Trond Myklebust authored
Starting with NFSv4.1, the server is able to deduce the client id from the SEQUENCE op which means it can always figure out whether or not the client is holding a delegation on a file that is being changed. For that reason, RFC5661 does not require a delegation to be unconditionally recalled on operations such as SETATTR, RENAME, or REMOVE. Note that for now, we continue to return READ delegations since that is still expected by the Linux knfsd server. Signed-off-by:
-
Trond Myklebust authored
Signed-off-by:
-
Trond Myklebust authored
Signed-off-by:
-
Trond Myklebust authored
Ensure that when we do finally delete the file, then we return the delegation. Signed-off-by:
-
Trond Myklebust authored
Move the delegation recall out of the generic code, and into the NFSv4 specific callback. Signed-off-by:
-
Trond Myklebust authored
Move the delegation return out of generic code and down into the NFSv4 specific unlink code. Signed-off-by:
-
Trond Myklebust authored
Move the delegation return out of generic code. Signed-off-by:
-
Trond Myklebust authored
The 'fmode' argument can take an FMODE_EXEC value, which we want to filter out before comparing to the delegation type. Signed-off-by:
-
Chuck Lever authored
Michal Kalderon has found some corner cases around device unload with active NFS mounts that I didn't have the imagination to test when xprtrdma device removal was added last year. - The ULP device removal handler is responsible for deallocating the PD. That wasn't clear to me initially, and my own testing suggested it was not necessary, but that is incorrect. - The transport destruction path can no longer assume that there is a valid ID. - When destroying a transport, ensure that ib_free_cq() is not invoked on a CQ that was already released. Reported-by:
Michal Kalderon <Michal.Kalderon@cavium.com> Fixes: bebd0318 ("xprtrdma: Support unplugging an HCA from ...") Signed-off-by:
Chuck Lever <chuck.lever@oracle.com> Cc: stable@vger.kernel.org # v4.12+ Signed-off-by:
-
Jeff Layton authored
If we have a RECLAIM_COMPLETE with a populated cl_lock_waitq, then that implies that a reconnect has occurred. Since we can't expect a CB_NOTIFY_LOCK callback at that point, just wake up the entire queue so that all the tasks can re-poll for their locks. Signed-off-by:
Jeff Layton <jlayton@redhat.com> Signed-off-by:
-
Jeff Layton authored
The task is expected to sleep for a while here, and it's possible that a new EXCHANGE_ID has occurred in the interim, and we were assigned a new clientid. Since this is a per-client list, there isn't a lot of value in vetting the clientid on the incoming request. Signed-off-by:
-
Jeff Layton authored
We may get a notification and lose the race to another client. Ensure that we wait again for a notification in that case. Signed-off-by:
-
Chuck Lever authored
This information can help track down local misconfiguration issues as well as network partitions and unresponsive servers. There are several ways to send a ping, and with transport multi- plexing, the exact rpc_xprt that is used is sometimes not known by the upper layer. The rpc_xprt pointer passed to the trace point call also has to be RCU-safe. I found a spot inside the client FSM where an rpc_xprt pointer is always available and safe to use. Suggested-by:
Bill Baker <Bill.Baker@oracle.com> Signed-off-by:
-
Chuck Lever authored
Introduce a low-overhead mechanism to report information about latencies of individual RPCs. The goal is to enable user space to filter the trace record for latency outliers, or build histograms, etc. Signed-off-by:
-
Chuck Lever authored
Clean up: struct rpc_task carries a pointer to a struct rpc_clnt, and in fact task->tk_client is always what is passed into trace points that are already passing @task. Signed-off-by:
-
Chuck Lever authored
If recording xprt->stat.max_slots is moved into xprt_alloc_slot, then xprt->num_reqs is never manipulated outside xprt->reserve_lock. There's no longer a need for xprt->num_reqs to be atomic. Signed-off-by:
-
Chuck Lever authored
Some RPC transports have more overhead in their send_request callouts than others. For example, for RPC-over-RDMA: - Marshaling an RPC often has to DMA map the RPC arguments - Registration methods perform memory registration as part of marshaling To capture just server and network latencies more precisely: when sending a Call, capture the rq_xtime timestamp _after_ the transport header has been marshaled. Signed-off-by:
-
Chuck Lever authored
Some RPC transports have more overhead in their reply handlers than others. For example, for RPC-over-RDMA: - RPC completion has to wait for memory invalidation, which is not a part of the server/network round trip - Recently a context switch was introduced into the reply handler, which further artificially inflates the measure of RPC RTT To capture just server and network latencies more precisely: when receiving a reply, compute the RTT as soon as the XID is recognized rather than at RPC completion time. Signed-off-by:
-
Chuck Lever authored
Since commit 33849792 ("xprtrdma: Detect unreachable NFS/RDMA servers more reliably"), the xprtrdma transport now has a ->timer callout. But xprtrdma does not need to compute RTT data, only UDP needs that. Move the xprt_update_rtt call into the UDP transport implementation. Signed-off-by:
-
Chuck Lever authored
Refactor: Both rpcrdma_create_req call sites have to allocate the buffer where the transport header is built, so just move that allocation into rpcrdma_create_req. This buffer is a fixed size. There's no needed information available in call_allocate that is not also available when the transport is created. The original purpose for allocating these buffers on demand was to reduce the possibility that an allocation failure during transport creation will hork the mount operation during low memory scenarios. Some relief for this rare possibility is coming up in the next few patches. Signed-off-by:
-
Chuck Lever authored
With FRWR, the client transport can perform memory registration and post a Send with just a single ib_post_send. This reduces contention between the send_request path and the Send Completion handlers, and reduces the overhead of registering a chunk that has multiple segments. Signed-off-by:
-
Chuck Lever authored
RPC-over-RDMA version 1 credit accounting relies on there being a response message for every RPC Call. This means that RPC procedures that have no reply will disrupt credit accounting, just in the same way as a retransmit would (since it is sent because no reply has arrived). Deal with the "no reply" case the same way. Signed-off-by:
-
Chuck Lever authored
Create fewer MRs on average. Many workloads don't need as many as 32 MRs, and the transport can now quickly restock the MR free list. Signed-off-by:
-
Chuck Lever authored
Currently, when the MR free list is exhausted during marshaling, the RPC/RDMA transport places the RPC task on the delayq, which forces a wait for HZ >> 2 before the marshal and send is retried. With this change, the transport now places such an RPC task on the pending queue, and wakes it just as soon as more MRs have been created. Creating more MRs typically takes less than a millisecond, and this waking mechanism is less deadlock-prone. Moreover, the waiting RPC task is holding the transport's write lock, which blocks the transport from sending RPCs. Therefore faster recovery from MR exhaustion is desirable. This is the same mechanism that the TCP transport utilizes when handling write buffer space exhaustion. Signed-off-by:
-
Chuck Lever authored
Clean up: The generic rq_connect_cookie is sufficient to detect RPC Call retransmission. Signed-off-by:
-
Chuck Lever authored
Clean up: We need to check only that the value does not exceed the range of the u8 field it's going into. Signed-off-by:
-
Chuck Lever authored
With v4.15, on one of my NFS/RDMA clients I measured a nearly doubling in the latency of small read and write system calls. There was no change in server round trip time. The extra latency appears in the whole RPC execution path. "git bisect" settled on commit ccede759 ("xprtrdma: Spread reply processing over more CPUs") . After some experimentation, I found that leaving the WQ bound and allowing the scheduler to pick the dispatch CPU seems to eliminate the long latencies, and it does not introduce any new regressions. The fix is implemented by reverting only the part of commit ccede759 ("xprtrdma: Spread reply processing over more CPUs") that dispatches RPC replies specifically on the CPU where the matching RPC call was made. Interestingly, saving the CPU number and later queuing reply processing there was effective _only_ for a NFS READ and WRITE request. On my NUMA client, in-kernel RPC reply processing for asynchronous RPCs was dispatched on the same CPU where the RPC call was made, as expected. However synchronous RPCs seem to get their reply dispatched on some other CPU than where the call was placed, every time. Fixes: ccede759 ("xprtrdma: Spread reply processing over ... ") Signed-off-by:
-
- 04 Mar, 2018 1 commit
-
-
Linus Torvalds authored
-
