| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
git://git.linux-nfs.org/projects/anna/nfs-rdma into linux-next
Pull NFS client RDMA changes for 3.19 from Anna Schumaker:
"NFS: Client side changes for RDMA
These patches various bugfixes and cleanups for using NFS over RDMA, including
better error handling and performance improvements by using pad optimization.
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>"
* tag 'nfs-rdma-for-3.19' of git://git.linux-nfs.org/projects/anna/nfs-rdma:
xprtrdma: Display async errors
xprtrdma: Enable pad optimization
xprtrdma: Re-write rpcrdma_flush_cqs()
xprtrdma: Refactor tasklet scheduling
xprtrdma: unmap all FMRs during transport disconnect
xprtrdma: Cap req_cqinit
xprtrdma: Return an errno from rpcrdma_register_external()
|
|
An async error upcall is a hard error, and should be reported in
the system log.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
|
|
Currently rpcrdma_flush_cqs() attempts to avoid code duplication,
and simply invokes rpcrdma_recvcq_upcall and rpcrdma_sendcq_upcall.
1. rpcrdma_flush_cqs() can run concurrently with provider upcalls.
Both flush_cqs() and the upcalls were invoking ib_poll_cq() in
different threads using the same wc buffers (ep->rep_recv_wcs
and ep->rep_send_wcs), added by commit 1c00dd077654 ("xprtrmda:
Reduce calls to ib_poll_cq() in completion handlers").
During transport disconnect processing, this sometimes resulted
in the same reply getting added to the rpcrdma_tasklets_g list
more than once, which corrupted the list.
2. The upcall functions drain only a limited number of CQEs,
thanks to the poll budget added by commit 8301a2c047cc
("xprtrdma: Limit work done by completion handler").
Fixes: a7bc211ac926 ("xprtrdma: On disconnect, don't ignore ... ")
BugLink: https://bugzilla.linux-nfs.org/show_bug.cgi?id=276
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
|
|
Restore the separate function that schedules the reply handling
tasklet. I need to call it from two different paths.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
|
|
When using RPCRDMA_MTHCAFMR memory registration, after a few
transport disconnect / reconnect cycles, ib_map_phys_fmr() starts to
return EINVAL because the provider has exhausted its map pool.
Make sure that all FMRs are unmapped during transport disconnect,
and that ->send_request remarshals them during an RPC retransmit.
This resets the transport's MRs to ensure that none are leaked
during a disconnect.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
|
|
Recent work made FRMR registration and invalidation completions
unsignaled. This greatly reduces the adapter interrupt rate.
Every so often, however, a posted send Work Request is allowed to
signal. Otherwise, the provider's Work Queue will wrap and the
workload will hang.
The number of Work Requests that are allowed to remain unsignaled is
determined by the value of req_cqinit. Currently, this is set to the
size of the send Work Queue divided by two, minus 1.
For FRMR, the send Work Queue is the maximum number of concurrent
RPCs (currently 32) times the maximum number of Work Requests an
RPC might use (currently 7, though some adapters may need more).
For mlx4, this is 224 entries. This leaves completion signaling
disabled for 111 send Work Requests.
Some providers hold back dispatching Work Requests until a CQE is
generated. If completions are disabled, then no CQEs are generated
for quite some time, and that can stall the Work Queue.
I've seen this occur running xfstests generic/113 over NFSv4, where
eventually, posting a FAST_REG_MR Work Request fails with -ENOMEM
because the Work Queue has overflowed. The connection is dropped
and re-established.
Cap the rep_cqinit setting so completions are not left turned off
for too long.
BugLink: https://bugzilla.linux-nfs.org/show_bug.cgi?id=269
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
|
|
The RPC/RDMA send_request method and the chunk registration code
expects an errno from the registration function. This allows
the upper layers to distinguish between a recoverable failure
(for example, temporary memory exhaustion) and a hard failure
(for example, a bug in the registration logic).
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
|
|
It's always set to whatever CONFIG_SUNRPC_DEBUG is, so just use that.
Signed-off-by: Jeff Layton <jlayton@primarydata.com>
Signed-off-by: Trond Myklebust <trond.myklebust@primarydata.com>
|
|
Commit 38ca83a5 added RDMA_CM_EVENT_TIMEWAIT_EXIT. But that status
is relevant only for consumers that re-use their QPs on new
connections. xprtrdma creates a fresh QP on reconnection, so that
event should be explicitly ignored.
Squelch the alarming "unexpected CM event" message.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Tested-by: Steve Wise <swise@opengridcomputing.com>
Tested-by: Shirley Ma <shirley.ma@oracle.com>
Tested-by: Devesh Sharma <devesh.sharma@emulex.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
|
|
Clean up.
RPCRDMA_PERSISTENT_REGISTRATION was a compile-time switch between
RPCRDMA_REGISTER mode and RPCRDMA_ALLPHYSICAL mode. Since
RPCRDMA_REGISTER has been removed, there's no need for the extra
conditional compilation.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Tested-by: Steve Wise <swise@opengridcomputing.com>
Tested-by: Shirley Ma <shirley.ma@oracle.com>
Tested-by: Devesh Sharma <devesh.sharma@emulex.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
|
|
Clean up: The return code is used only for dprintk's that are
already redundant.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Tested-by: Steve Wise <swise@opengridcomputing.com>
Tested-by: Shirley Ma <shirley.ma@oracle.com>
Tested-by: Devesh Sharma <devesh.sharma@emulex.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
|
|
Minor optimization: grab rpcrdma_tk_lock_g and disable hard IRQs
just once after clearing the receive completion queue.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Tested-by: Steve Wise <swise@opengridcomputing.com>
Tested-by: Shirley Ma <shirley.ma@oracle.com>
Tested-by: Devesh Sharma <devesh.sharma@emulex.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
|
|
Currently rpcrdma_buffer_create() allocates struct rpcrdma_mw's as
a single contiguous area of memory. It amounts to quite a bit of
memory, and there's no requirement for these to be carved from a
single piece of contiguous memory.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Tested-by: Steve Wise <swise@opengridcomputing.com>
Tested-by: Shirley Ma <shirley.ma@oracle.com>
Tested-by: Devesh Sharma <devesh.sharma@emulex.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
|
|
Clean up: Name frmr_wr after the opcode of the Work Request,
consistent with the send and local invalidation paths.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Tested-by: Steve Wise <swise@opengridcomputing.com>
Tested-by: Shirley Ma <shirley.ma@oracle.com>
Tested-by: Devesh Sharma <devesh.sharma@emulex.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
|
|
Instead of relying on a completion to change the state of an FRMR
to FRMR_IS_INVALID, set it in advance. If an error occurs, a completion
will fire anyway and mark the FRMR FRMR_IS_STALE.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Tested-by: Steve Wise <swise@opengridcomputing.com>
Tested-by: Shirley Ma <shirley.ma@oracle.com>
Tested-by: Devesh Sharma <devesh.sharma@emulex.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
|
|
Instead of relying on a completion to change the state of an FRMR
to FRMR_IS_VALID, set it in advance. If an error occurs, a completion
will fire anyway and mark the FRMR FRMR_IS_STALE.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Tested-by: Steve Wise <swise@opengridcomputing.com>
Tested-by: Shirley Ma <shirley.ma@oracle.com>
Tested-by: Devesh Sharma <devesh.sharma@emulex.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
|
|
Any FRMR arriving in rpcrdma_register_frmr_external() is now
guaranteed to be either invalid, or to be targeted by a queued
LOCAL_INV that will invalidate it before the adapter processes
the FAST_REG_MR being built here.
The problem with current arrangement of chaining a LOCAL_INV to the
FAST_REG_MR is that if the transport is not connected, the LOCAL_INV
is flushed and the FAST_REG_MR is flushed. This leaves the FRMR
valid with the old rkey. But rpcrdma_register_frmr_external() has
already bumped the in-memory rkey.
Next time through rpcrdma_register_frmr_external(), a LOCAL_INV and
FAST_REG_MR is attempted again because the FRMR is still valid. But
the rkey no longer matches the hardware's rkey, and a memory
management operation error occurs.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Tested-by: Steve Wise <swise@opengridcomputing.com>
Tested-by: Shirley Ma <shirley.ma@oracle.com>
Tested-by: Devesh Sharma <devesh.sharma@emulex.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
|
|
When a LOCAL_INV Work Request is flushed, it leaves an FRMR in the
VALID state. This FRMR can be returned by rpcrdma_buffer_get(), and
must be knocked down in rpcrdma_register_frmr_external() before it
can be re-used.
Instead, capture these in rpcrdma_buffer_get(), and reset them.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Tested-by: Steve Wise <swise@opengridcomputing.com>
Tested-by: Shirley Ma <shirley.ma@oracle.com>
Tested-by: Devesh Sharma <devesh.sharma@emulex.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
|
|
FAST_REG_MR Work Requests update a Memory Region's rkey. Rkey's are
used to block unwanted access to the memory controlled by an MR. The
rkey is passed to the receiver (the NFS server, in our case), and is
also used by xprtrdma to invalidate the MR when the RPC is complete.
When a FAST_REG_MR Work Request is flushed after a transport
disconnect, xprtrdma cannot tell whether the WR actually hit the
adapter or not. So it is indeterminant at that point whether the
existing rkey is still valid.
After the transport connection is re-established, the next
FAST_REG_MR or LOCAL_INV Work Request against that MR can sometimes
fail because the rkey value does not match what xprtrdma expects.
The only reliable way to recover in this case is to deregister and
register the MR before it is used again. These operations can be
done only in a process context, so handle it in the transport
connect worker.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Tested-by: Steve Wise <swise@opengridcomputing.com>
Tested-by: Shirley Ma <shirley.ma@oracle.com>
Tested-by: Devesh Sharma <devesh.sharma@emulex.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
|
|
If the rb_mws list is exhausted, clean up and return NULL so that
call_allocate() will delay and try again.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Tested-by: Steve Wise <swise@opengridcomputing.com>
Tested-by: Shirley Ma <shirley.ma@oracle.com>
Tested-by: Devesh Sharma <devesh.sharma@emulex.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
|
|
During connection loss recovery, need to visit every MW in a
buffer pool. Any MW that is in use by an RPC will not be on the
rb_mws list.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Tested-by: Steve Wise <swise@opengridcomputing.com>
Tested-by: Shirley Ma <shirley.ma@oracle.com>
Tested-by: Devesh Sharma <devesh.sharma@emulex.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
|
|
If posting a FAST_REG_MR Work Reqeust fails, revert the rkey update
to avoid subsequent IB_WC_MW_BIND_ERR completions.
Suggested-by: Steve Wise <swise@opengridcomputing.com>
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
|
|
Clean ups:
- make it obvious that the rl_mw field is a pointer -- allocated
separately, not as part of struct rpcrdma_mr_seg
- promote "struct {} frmr;" to a named type
- promote the state enum to a named type
- name the MW state field the same way other fields in
rpcrdma_mw are named
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Tested-by: Steve Wise <swise@opengridcomputing.com>
Tested-by: Shirley Ma <shirley.ma@oracle.com>
Tested-by: Devesh Sharma <devesh.sharma@emulex.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
|
|
xprtrdma is currently throwing away queued completions during
a reconnect. RPC replies posted just before connection loss, or
successful completions that change the state of an FRMR, can be
missed.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Tested-by: Steve Wise <swise@opengridcomputing.com>
Tested-by: Shirley Ma <shirley.ma@oracle.com>
Tested-by: Devesh Sharma <devesh.sharma@emulex.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
|
|
When the client uses physical memory registration, each page in the
payload gets its own array entry in the RPC/RDMA header's chunk list.
Therefore, don't advertise a maximum payload size that would require
more array entries than can fit in the RPC buffer where RPC/RDMA
headers are built.
BugLink: https://bugzilla.linux-nfs.org/show_bug.cgi?id=248
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Tested-by: Steve Wise <swise@opengridcomputing.com>
Tested-by: Shirley Ma <shirley.ma@oracle.com>
Tested-by: Devesh Sharma <devesh.sharma@emulex.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
|
|
Ensure ia->ri_id remains valid while invoking dma_unmap_page() or
posting LOCAL_INV during a transport reconnect. Otherwise,
ia->ri_id->device or ia->ri_id->qp is NULL, which triggers a panic.
BugLink: https://bugzilla.linux-nfs.org/show_bug.cgi?id=259
Fixes: ec62f40 'xprtrdma: Ensure ia->ri_id->qp is not NULL when reconnecting'
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Tested-by: Steve Wise <swise@opengridcomputing.com>
Tested-by: Shirley Ma <shirley.ma@oracle.com>
Tested-by: Devesh Sharma <devesh.sharma@emulex.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
|
|
seg1->mr_nsegs is not yet initialized when it is used to unmap
segments during an error exit. Use the same unmapping logic for
all error exits.
"if (frmr_wr.wr.fast_reg.length < len) {" used to be a BUG_ON check.
The broken code will never be executed under normal operation.
Fixes: c977dea (xprtrdma: Remove BUG_ON() call sites)
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Tested-by: Steve Wise <swise@opengridcomputing.com>
Tested-by: Shirley Ma <shirley.ma@oracle.com>
Tested-by: Devesh Sharma <devesh.sharma@emulex.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
|
|
Fix the following warning when DMA-API debug is enabled by checking ib_dma_map_single result:
[ 1455.345548] ------------[ cut here ]------------
[ 1455.346863] WARNING: CPU: 3 PID: 3929 at /home/yanb/kernel/net-next/lib/dma-debug.c:1140 check_unmap+0x4e5/0x990()
[ 1455.349350] mlx4_core 0000:00:07.0: DMA-API: device driver failed to check map error[device address=0x000000007c9f2090] [size=2656 bytes] [mapped as single]
[ 1455.349350] Modules linked in: xprtrdma netconsole configfs nfsv3 nfs_acl ib_ipoib rdma_ucm ib_ucm ib_uverbs ib_umad rdma_cm ib_cm iw_cm autofs4 auth_rpcgss oid_registry nfsv4 nfs fscache lockd sunrpc dm_mirror dm_region_hash dm_log microcode pcspkr mlx4_ib ib_sa ib_mad ib_core ib_addr mlx4_en ipv6 ptp pps_core vxlan mlx4_core virtio_balloon cirrus ttm drm_kms_helper drm sysimgblt sysfillrect syscopyarea i2c_piix4 i2c_core button ext3 jbd virtio_blk virtio_net virtio_pci virtio_ring virtio uhci_hcd ata_generic ata_piix libata
[ 1455.349350] CPU: 3 PID: 3929 Comm: mount.nfs Not tainted 3.15.0-rc1-dbg+ #13
[ 1455.349350] Hardware name: Red Hat KVM, BIOS 0.5.1 01/01/2007
[ 1455.349350] 0000000000000474 ffff880069dcf628 ffffffff8151c341 ffffffff817b69d8
[ 1455.349350] ffff880069dcf678 ffff880069dcf668 ffffffff8105b5fc 0000000069dcf658
[ 1455.349350] ffff880069dcf778 ffff88007b0c9f00 ffffffff8255ec40 0000000000000a60
[ 1455.349350] Call Trace:
[ 1455.349350] [<ffffffff8151c341>] dump_stack+0x52/0x81
[ 1455.349350] [<ffffffff8105b5fc>] warn_slowpath_common+0x8c/0xc0
[ 1455.349350] [<ffffffff8105b6e6>] warn_slowpath_fmt+0x46/0x50
[ 1455.349350] [<ffffffff812e6305>] check_unmap+0x4e5/0x990
[ 1455.349350] [<ffffffff81521fb0>] ? _raw_spin_unlock_irq+0x30/0x60
[ 1455.349350] [<ffffffff812e6a0a>] debug_dma_unmap_page+0x5a/0x60
[ 1455.349350] [<ffffffffa0389583>] rpcrdma_deregister_internal+0xb3/0xd0 [xprtrdma]
[ 1455.349350] [<ffffffffa038a639>] rpcrdma_buffer_destroy+0x69/0x170 [xprtrdma]
[ 1455.349350] [<ffffffffa03872ff>] xprt_rdma_destroy+0x3f/0xb0 [xprtrdma]
[ 1455.349350] [<ffffffffa04a95ff>] xprt_destroy+0x6f/0x80 [sunrpc]
[ 1455.349350] [<ffffffffa04a9625>] xprt_put+0x15/0x20 [sunrpc]
[ 1455.349350] [<ffffffffa04a899a>] rpc_free_client+0x8a/0xe0 [sunrpc]
[ 1455.349350] [<ffffffffa04a8a58>] rpc_release_client+0x68/0xa0 [sunrpc]
[ 1455.349350] [<ffffffffa04a9060>] rpc_shutdown_client+0xb0/0xc0 [sunrpc]
[ 1455.349350] [<ffffffffa04a8f5d>] ? rpc_ping+0x5d/0x70 [sunrpc]
[ 1455.349350] [<ffffffffa04a91ab>] rpc_create_xprt+0xbb/0xd0 [sunrpc]
[ 1455.349350] [<ffffffffa04a9273>] rpc_create+0xb3/0x160 [sunrpc]
[ 1455.349350] [<ffffffff81129749>] ? __probe_kernel_read+0x69/0xb0
[ 1455.349350] [<ffffffffa053851c>] nfs_create_rpc_client+0xdc/0x100 [nfs]
[ 1455.349350] [<ffffffffa0538cfa>] nfs_init_client+0x3a/0x90 [nfs]
[ 1455.349350] [<ffffffffa05391c8>] nfs_get_client+0x478/0x5b0 [nfs]
[ 1455.349350] [<ffffffffa0538e50>] ? nfs_get_client+0x100/0x5b0 [nfs]
[ 1455.349350] [<ffffffff81172c6d>] ? kmem_cache_alloc_trace+0x24d/0x260
[ 1455.349350] [<ffffffffa05393f3>] nfs_create_server+0xf3/0x4c0 [nfs]
[ 1455.349350] [<ffffffffa0545ff0>] ? nfs_request_mount+0xf0/0x1a0 [nfs]
[ 1455.349350] [<ffffffffa031c0c3>] nfs3_create_server+0x13/0x30 [nfsv3]
[ 1455.349350] [<ffffffffa0546293>] nfs_try_mount+0x1f3/0x230 [nfs]
[ 1455.349350] [<ffffffff8108ea21>] ? get_parent_ip+0x11/0x50
[ 1455.349350] [<ffffffff812d6343>] ? __this_cpu_preempt_check+0x13/0x20
[ 1455.349350] [<ffffffff810d632b>] ? try_module_get+0x6b/0x190
[ 1455.349350] [<ffffffffa05449f7>] nfs_fs_mount+0x187/0x9d0 [nfs]
[ 1455.349350] [<ffffffffa0545940>] ? nfs_clone_super+0x140/0x140 [nfs]
[ 1455.349350] [<ffffffffa0543b20>] ? nfs_auth_info_match+0x40/0x40 [nfs]
[ 1455.349350] [<ffffffff8117e360>] mount_fs+0x20/0xe0
[ 1455.349350] [<ffffffff811a1c16>] vfs_kern_mount+0x76/0x160
[ 1455.349350] [<ffffffff811a29a8>] do_mount+0x428/0xae0
[ 1455.349350] [<ffffffff811a30f0>] SyS_mount+0x90/0xe0
[ 1455.349350] [<ffffffff8152af52>] system_call_fastpath+0x16/0x1b
[ 1455.349350] ---[ end trace f1f31572972e211d ]---
Signed-off-by: Yan Burman <yanb@mellanox.com>
Reviewed-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
|
|
If an error occurs in the marshaling logic, fail the RPC request
being processed, but leave the client running.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
|
|
Clean up. Remove HCA-specific clutter in xprtrdma, which is
supposed to be device-independent.
Hal Rosenstock <hal@dev.mellanox.co.il> observes:
> Note that there is OpenSM option (enable_quirks) to return 1K MTU
> in SA PathRecord responses for Tavor so that can be used for this.
> The default setting for enable_quirks is FALSE so that would need
> changing.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
|
|
Devesh Sharma <Devesh.Sharma@Emulex.Com> reports that after a
disconnect, his HCA is failing to create a fresh QP, leaving
ia_ri->ri_id->qp set to NULL. But xprtrdma still allows RPCs to
wake up and post LOCAL_INV as they exit, causing an oops.
rpcrdma_ep_connect() is allowing the wake-up by leaking the QP
creation error code (-EPERM in this case) to the RPC client's
generic layer. xprt_connect_status() does not recognize -EPERM, so
it kills pending RPC tasks immediately rather than retrying the
connect.
Re-arrange the QP creation logic so that when it fails on reconnect,
it leaves ->qp with the old QP rather than NULL. If pending RPC
tasks wake and exit, LOCAL_INV work requests will flush rather than
oops.
On initial connect, leaving ->qp == NULL is OK, since there are no
pending RPCs that might use ->qp. But be sure not to try to destroy
a NULL QP when rpcrdma_ep_connect() is retried.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
|
|
While marshaling an RPC/RDMA request, the inline_{rsize,wsize}
settings determine whether an inline request is used, or whether
read or write chunks lists are built. The current default value of
these settings is 1024. Any RPC request smaller than 1024 bytes is
sent to the NFS server completely inline.
rpcrdma_buffer_create() allocates and pre-registers a set of RPC
buffers for each transport instance, also based on the inline rsize
and wsize settings.
RPC/RDMA requests and replies are built in these buffers. However,
if an RPC/RDMA request is expected to be larger than 1024, a buffer
has to be allocated and registered for that RPC, and deregistered
and released when the RPC is complete. This is known has a
"hardway allocation."
Since the introduction of NFSv4, the size of RPC requests has become
larger, and hardway allocations are thus more frequent. Hardway
allocations are significant overhead, and they waste the existing
RPC buffers pre-allocated by rpcrdma_buffer_create().
We'd like fewer hardway allocations.
Increasing the size of the pre-registered buffers is the most direct
way to do this. However, a blanket increase of the inline thresholds
has interoperability consequences.
On my 64-bit system, rpcrdma_buffer_create() requests roughly 7000
bytes for each RPC request buffer, using kmalloc(). Due to internal
fragmentation, this wastes nearly 1200 bytes because kmalloc()
already returns an 8192-byte piece of memory for a 7000-byte
allocation request, though the extra space remains unused.
So let's round up the size of the pre-allocated buffers, and make
use of the unused space in the kmalloc'd memory.
This change reduces the amount of hardway allocated memory for an
NFSv4 general connectathon run from 1322092 to 9472 bytes (99%).
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Tested-by: Steve Wise <swise@opengridcomputing.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
|
|
Sagi Grimberg <sagig@dev.mellanox.co.il> points out that a steady
stream of CQ events could starve other work because of the boundless
loop pooling in rpcrdma_{send,recv}_poll().
Instead of a (potentially infinite) while loop, return after
collecting a budgeted number of completions.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Acked-by: Sagi Grimberg <sagig@dev.mellanox.co.il>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
|
|
Change the completion handlers to grab up to 16 items per
ib_poll_cq() call. No extra ib_poll_cq() is needed if fewer than 16
items are returned.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
|
|
Skip the ib_poll_cq() after re-arming, if the provider knows there
are no additional items waiting. (Have a look at commit ed23a727 for
more details).
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
|
|
The current CQ handler uses the ib_wc.opcode field to distinguish
between event types. However, the contents of that field are not
reliable if the completion status is not IB_WC_SUCCESS.
When an error completion occurs on a send event, the CQ handler
schedules a tasklet with something that is not a struct rpcrdma_rep.
This is never correct behavior, and sometimes it results in a panic.
To resolve this issue, split the completion queue into a send CQ and
a receive CQ. The send CQ handler now handles only struct rpcrdma_mw
wr_id's, and the receive CQ handler now handles only struct
rpcrdma_rep wr_id's.
Fix suggested by Shirley Ma <shirley.ma@oracle.com>
Reported-by: Rafael Reiter <rafael.reiter@ims.co.at>
Fixes: 5c635e09cec0feeeb310968e51dad01040244851
BugLink: https://bugzilla.kernel.org/show_bug.cgi?id=73211
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Tested-by: Klemens Senn <klemens.senn@ims.co.at>
Tested-by: Steve Wise <swise@opengridcomputing.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
|
|
Clean up: rpcrdma_ep_destroy() returns a value that is used
only to print a debugging message. rpcrdma_ep_destroy() already
prints debugging messages in all error cases.
Make rpcrdma_ep_destroy() return void instead.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Tested-by: Steve Wise <swise@opengridcomputing.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
|
|
Clean up: All remaining callers of rpcrdma_deregister_external()
pass NULL as the last argument, so remove that argument.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Tested-by: Steve Wise <swise@opengridcomputing.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
|
|
If the selected memory registration mode is not supported by the
underlying provider/HCA, the NFS mount command reports that there was
an invalid mount option, and fails. This is misleading.
Reporting a problem allocating memory is a lot closer to the truth.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
|
|
An audit of in-kernel RDMA providers that do not support the FRMR
memory registration shows that several of them support MTHCAFMR.
Prefer MTHCAFMR when FRMR is not supported.
If MTHCAFMR is not supported, only then choose ALLPHYSICAL.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
|
|
All kernel RDMA providers except amso1100 support either MTHCAFMR
or FRMR, both of which are faster than REGISTER. amso1100 can
continue to use ALLPHYSICAL.
The only other ULP consumer in the kernel that uses the reg_phys_mr
verb is Lustre.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
|
|
The MEMWINDOWS and MEMWINDOWS_ASYNC memory registration modes were
intended as stop-gap modes before the introduction of FRMR. They
are now considered obsolete.
MEMWINDOWS_ASYNC is also considered unsafe because it can leave
client memory registered and exposed for an indeterminant time after
each I/O.
At this point, the MEMWINDOWS modes add needless complexity, so
remove them.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Tested-by: Steve Wise <swise@opengridcomputing.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
|
|
Clean up: This memory registration mode is slow and was never
meant for use in production environments. Remove it to reduce
implementation complexity.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Tested-by: Steve Wise <swise@opengridcomputing.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
|
|
An IB provider can invoke rpcrdma_conn_func() in an IRQ context,
thus rpcrdma_conn_func() cannot be allowed to directly invoke
generic RPC functions like xprt_wake_pending_tasks().
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Tested-by: Steve Wise <swise@opengridcomputing.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
|
|
Two memory region leaks were found during testing:
1. rpcrdma_buffer_create: While allocating RPCRDMA_FRMR's
ib_alloc_fast_reg_mr is called and then ib_alloc_fast_reg_page_list is
called. If ib_alloc_fast_reg_page_list returns an error it bails out of
the routine dropping the last ib_alloc_fast_reg_mr frmr region creating a
memory leak. Added code to dereg the last frmr if
ib_alloc_fast_reg_page_list fails.
2. rpcrdma_buffer_destroy: While cleaning up, the routine will only free
the MR's on the rb_mws list if there are rb_send_bufs present. However, in
rpcrdma_buffer_create while the rb_mws list is being built if one of the MR
allocation requests fail after some MR's have been allocated on the rb_mws
list the routine never gets to create any rb_send_bufs but instead jumps to
the rpcrdma_buffer_destroy routine which will never free the MR's on rb_mws
list because the rb_send_bufs were never created. This leaks all the MR's
on the rb_mws list that were created prior to one of the MR allocations
failing.
Issue(2) was seen during testing. Our adapter had a finite number of MR's
available and we created enough connections to where we saw an MR
allocation failure on our Nth NFS connection request. After the kernel
cleaned up the resources it had allocated for the Nth connection we noticed
that FMR's had been leaked due to the coding error described above.
Issue(1) was seen during a code review while debugging issue(2).
Signed-off-by: Allen Andrews <allen.andrews@emulex.com>
Reviewed-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
|
|
Some rdma devices don't support a fast register page list depth of
at least RPCRDMA_MAX_DATA_SEGS. So xprtrdma needs to chunk its fast
register regions according to the minimum of the device max supported
depth or RPCRDMA_MAX_DATA_SEGS.
Signed-off-by: Steve Wise <swise@opengridcomputing.com>
Reviewed-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
|
|
This patch enhances the IB core support for Memory Windows (MWs).
MWs allow an application to have better/flexible control over remote
access to memory.
Two types of MWs are supported, with the second type having two flavors:
Type 1 - associated with PD only
Type 2A - associated with QPN only
Type 2B - associated with PD and QPN
Applications can allocate a MW once, and then repeatedly bind the MW
to different ranges in MRs that are associated to the same PD. Type 1
windows are bound through a verb, while type 2 windows are bound by
posting a work request.
The 32-bit memory key is composed of a 24-bit index and an 8-bit
key. The key is changed with each bind, thus allowing more control
over the peer's use of the memory key.
The changes introduced are the following:
* add memory window type enum and a corresponding parameter to ib_alloc_mw.
* type 2 memory window bind work request support.
* create a struct that contains the common part of the bind verb struct
ibv_mw_bind and the bind work request into a single struct.
* add the ib_inc_rkey helper function to advance the tag part of an rkey.
Consumer interface details:
* new device capability flags IB_DEVICE_MEM_WINDOW_TYPE_2A and
IB_DEVICE_MEM_WINDOW_TYPE_2B are added to indicate device support
for these features.
Devices can set either IB_DEVICE_MEM_WINDOW_TYPE_2A or
IB_DEVICE_MEM_WINDOW_TYPE_2B if it supports type 2A or type 2B
memory windows. It can set neither to indicate it doesn't support
type 2 windows at all.
* modify existing provides and consumers code to the new param of
ib_alloc_mw and the ib_mw_bind_info structure
Signed-off-by: Haggai Eran <haggaie@mellanox.com>
Signed-off-by: Shani Michaeli <shanim@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
|
|
The xprtrdma FRMR mapping logic assumes that a segment is <= PAGE_SIZE.
This is not true for NFS4.
Signed-off-by: Tom Tucker <tom@ogc.us>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
|
|
* remove interrupt.g inclusion from netdevice.h -- not needed
* fixup fallout, add interrupt.h and hardirq.h back where needed.
Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
|
|
The RDMA CM currently infers the QP type from the port space selected
by the user. In the future (eg with RDMA_PS_IB or XRC), there may not
be a 1-1 correspondence between port space and QP type. For netlink
export of RDMA CM state, we want to export the QP type to userspace,
so it is cleaner to explicitly associate a QP type to an ID.
Modify rdma_create_id() to allow the user to specify the QP type, and
use it to make our selections of datagram versus connected mode.
Signed-off-by: Sean Hefty <sean.hefty@intel.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
|
