init_nfsd() should not unregister pernet subsys if the register fails
but should instead unwind from the last successful operation which is
register_filesystem().

Unregistering a failed register_pernet_subsys() call can result in
a kernel GPF as revealed by programmatically injecting an error in
register_pernet_subsys().

Verified the fix handled failure gracefully with no lingering nfsd
entry in /proc/filesystems.  This change was introduced by the commit
bd5ae928 ("nfsd: register pernet ops last, unregister first"),
the original error handling logic was correct.

Fixes: bd5ae928 ("nfsd: register pernet ops last, unregister first")
Cc: stable@vger.kernel.org
Signed-off-by: Patrick Ho <Patrick.Ho@netapp.com>
Acked-by: J. Bruce Fields <bfields@redhat.com>
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>

When the back channel enters SEQ4_STATUS_CB_PATH_DOWN state, the client
recovers by sending BIND_CONN_TO_SESSION but the server fails to recover
the back channel and leaves it as NFSD4_CB_DOWN.

Fix by enhancing nfsd4_bind_conn_to_session to probe the back channel
by calling nfsd4_probe_callback.
Signed-off-by: Dai Ngo <dai.ngo@oracle.com>
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>

Dai Ngo reports that, since the XDR overhaul, the NLM server crashes
when the TEST procedure wants to return NLM_DENIED. There is a bug
in svcxdr_encode_owner() that none of our standard test cases found.

Replace the open-coded function with a call to an appropriate
pre-fabricated XDR helper.
Reported-by: Dai Ngo <Dai.Ngo@oracle.com>
Fixes: a6a63ca5 ("lockd: Common NLM XDR helpers")
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>

When the NFS server receives a large gss (kerberos) credential and tries
to pass it up to rpc.svcgssd (which is deprecated), it triggers an
infinite loop in cache_read().

cache_request() always returns -EAGAIN, and this causes a "goto again".

This patch:
 - changes the error to -E2BIG to avoid the infinite loop, and
 - generates a WARN_ONCE when rsi_request first sees an over-sized
   credential.  The warning suggests switching to gssproxy.

Link: https://bugzilla.kernel.org/show_bug.cgi?id=196583Signed-off-by: NeilBrown <neilb@suse.de>
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>

alloc_pages_bulk_array() attempts to allocate at least one page based on
the provided pages, and then opportunistically allocates more if that
can be done without dropping the spinlock.

So if it returns fewer than requested, that could just mean that it
needed to drop the lock.  In that case, try again immediately.

Only pause for a time if no progress could be made.
Reported-and-tested-by: Mike Javorski <mike.javorski@gmail.com>
Reported-and-tested-by: Lothar Paltins <lopa@mailbox.org>
Fixes: f6e70aab ("SUNRPC: refresh rq_pages using a bulk page allocator")
Signed-off-by: NeilBrown <neilb@suse.de>
Acked-by: Mel Gorman <mgorman@suse.com>
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>

Unlike other filesystems, NFSv3 tries to use fl_file in the GETLK case.
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>

In the reexport case, nfsd is currently passing along locks with the
reclaim bit set.  The client sends a new lock request, which is granted
if there's currently no conflict--even if it's possible a conflicting
lock could have been briefly held in the interim.

We don't currently have any way to safely grant reclaim, so for now
let's just deny them all.

I'm doing this by passing the reclaim bit to nfs and letting it fail the
call, with the idea that eventually the client might be able to do
something more forgiving here.
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
Acked-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>

As in the v4 case, it doesn't work well to block waiting for a lock on
an nfs filesystem.

As in the v4 case, that means we're depending on the client to poll.
It's probably incorrect to depend on that, but I *think* clients do poll
in practice.  In any case, it's an improvement over hanging the lockd
thread indefinitely as we currently are.
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
Acked-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>

NFS implements blocking locks by blocking inside its lock method.  In
the reexport case, this blocks the nfs server thread, which could lead
to deadlocks since an nfs server thread might be required to unlock the
conflicting lock.  It also causes a crash, since the nfs server thread
assumes it can free the lock when its lm_notify lock callback is called.

Ideal would be to make the nfs lock method return without blocking in
this case, but for now it works just not to attempt blocking locks.  The
difference is just that the original client will have to poll (as it
does in the v4.0 case) instead of getting a callback when the lock's
available.
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
Acked-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>

We shouldn't really be using a read-only file descriptor to take a write
lock.

Most filesystems will put up with it.  But NFS, for example, won't.
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>

Update comment to reflect that we *do* allow reexport, whether it's a
good idea or not....
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>

Make this lookup slightly more concise, and prepare for changing how we
look this up in a following patch.
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>

It'll come in handy to get the whole nlm_lock.
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>

Locks have two sets of op arrays, fl_lmops for the lock manager (lockd
or nfsd), fl_ops for the filesystem.  The server-side lockd code has
been setting its own fl_ops, which leads to confusion (and crashes) in
the reexport case, where the filesystem expects to be the only one
setting fl_ops.

And there's no reason for it that I can see-the lm_get/put_owner ops do
the same job.
Reported-by: Daire Byrne <daire@dneg.com>
Tested-by: Daire Byrne <daire@dneg.com>
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>

Signed-off-by: Chuck Lever <chuck.lever@oracle.com>

Disconnect injection stress-tests the ability for both client and
server implementations to behave resiliently in the face of network
instability.

A file called /sys/kernel/debug/fail_sunrpc/ignore-server-disconnect
enables administrators to turn off server-side disconnect injection
while allowing other types of sunrpc errors to be injected. The
default setting is that server-side disconnect injection is enabled
(ignore=false).
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>

Disconnect injection stress-tests the ability for both client and
server implementations to behave resiliently in the face of network
instability.

Convert the existing client-side disconnect injection infrastructure
to use the kernel's generic error injection facility. The generic
facility has a richer set of injection criteria.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>

This directory will contain a set of administrative controls for
enabling error injection for kernel RPC consumers.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>

svc_xprt_free() already "puts" the bc_xprt before calling the
transport's "free" method. No need to do it twice.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>

This should use the network-namespace-wide client_lock, not the
per-client cl_lock.

You shouldn't see any bugs unless you're actually using the
forced-expiry interface introduced by 89c905be.

Fixes: 89c905be "nfsd: allow forced expiration of NFSv4 clients"
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>

The failure case here should be rare, but it's obviously wrong.
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>

Shared by client and server. See:

https://www.iana.org/assignments/rpc-authentication-numbers/rpc-authentication-numbers.xhtmlSigned-off-by: Chuck Lever <chuck.lever@oracle.com>

nsm_use_hostnames is a module parameter and it will be exported to sysctl
procfs. This is to let user sometimes change it from userspace. But the
minimal unit for sysctl procfs read/write it sizeof(int).
In big endian system, the converting from/to  bool to/from int will cause
error for proc items.

This patch use a new proc_handler proc_dobool to fix it.
Signed-off-by: Jia He <hejianet@gmail.com>
Reviewed-by: Pan Xinhui <xinhui.pan@linux.vnet.ibm.com>
[thuth: Fix typo in commit message]
Signed-off-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>

This is to let bool variable could be correctly displayed in
big/little endian sysctl procfs. sizeof(bool) is arch dependent,
proc_dobool should work in all arches.
Suggested-by: Pan Xinhui <xinhui@linux.vnet.ibm.com>
Signed-off-by: Jia He <hejianet@gmail.com>
[thuth: rebased the patch to the current kernel version]
Signed-off-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>

Some paths through svc_process() leave rqst->rq_procinfo set to
NULL, which triggers a crash if tracing happens to be enabled.

Fixes: 89ff8749 ("SUNRPC: Display RPC procedure names instead of proc numbers")
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>

Including one's name in copyright claims is appropriate.  Including it
in random comments is just vanity.  After 2 decades, it is time for
these to be gone.
Signed-off-by: NeilBrown <neilb@suse.de>
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>

Relieve contention on sc_rw_ctxt_lock by converting rdma->sc_rw_ctxts
to an llist.

The goal is to reduce the average overhead of Send completions,
because a transport's completion handlers are single-threaded on
one CPU core. This change reduces CPU utilization of each Send
completion by 2-3% on my server.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Reviewed-By: Tom Talpey <tom@talpey.com>

/proc/lock_stat indicates the the sc_send_lock is heavily
contended when the server is under load from a single client.

To address this, convert the send_ctxt free list to an llist.
Returning an item to the send_ctxt cache is now waitless, which
reduces the instruction path length in the single-threaded Send
handler (svc_rdma_wc_send).

The goal is to enable the ib_comp_wq worker to handle a higher
RPC/RDMA Send completion rate given the same CPU resources. This
change reduces CPU utilization of Send completion by 2-3% on my
server.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Reviewed-By: Tom Talpey <tom@talpey.com>

Because wake_up() takes an IRQ-safe lock, it can be expensive,
especially to call inside of a single-threaded completion handler.
What's more, the Send wait queue almost never has waiters, so
most of the time, this is an expensive no-op.

As always, the goal is to reduce the average overhead of each
completion, because a transport's completion handlers are single-
threaded on one CPU core. This change reduces CPU utilization of
the Send completion thread by 2-3% on my server.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Reviewed-By: Tom Talpey <tom@talpey.com>

After calling vfs_test_lock() the pointer to a conflicting lock can be
returned, and that lock is not guarunteed to be owned by nlm.  In that
case, we cannot cast it to struct nlm_lockowner.  Instead return the pid
of that conflicting lock.

Fixes: 646d73e9 ("lockd: Show pid of lockd for remote locks")
Signed-off-by: Benjamin Coddington <bcodding@redhat.com>
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>

Clean up.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>

Clean up.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>

There's a few cases that a string that is to be recorded in a trace event,
does not have a terminating 'nul' character, and instead, the tracepoint
passes in the length of the string to record.

Add two helper macros to the trace event code that lets this work easier,
than tricks with "%.*s" logic.

  __string_len() which is similar to __string() for declaration, but takes a
                 length argument.

  __assign_str_len() which is similar to __assign_str() for assiging the
                 string, but it too takes a length argument.

Note, the TRACE_EVENT() macro will allocate the location on the ring
buffer to 'len + 1', that will be used to store the string into. It is a
requirement that the 'len' used for this is a most the length of the
string being recorded.

This string can still use __get_str() just like strings created with
__string() can use to retrieve the string.

Link: https://lore.kernel.org/linux-nfs/20210513105018.7539996a@gandalf.local.home/Tested-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>

Large splice reads call put_page() repeatedly. put_page() is
relatively expensive to call, so replace it with the new
svc_rqst_replace_page() helper to help amortize that cost.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Reviewed-by: NeilBrown <neilb@suse.de>

Replacing a page in rq_pages[] requires a get_page(), which is a
bus-locked operation, and a put_page(), which can be even more
costly.

To reduce the cost of replacing a page in rq_pages[], batch the
put_page() operations by collecting "freed" pages in a pagevec,
and then release those pages when the pagevec is full. This
pagevec is also emptied when each RPC completes.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>

A few useful observations:

 - The value in @size is never modified.

 - splice_desc.len is an unsigned int, and so is xdr_buf.page_len.
   An implicit cast to size_t is unnecessary.

 - The computation of .page_len is the same in all three arms
   of the "if" statement, so hoist it out to make it clear that
   the operation is an unconditional invariant.

The resulting function is 18 bytes shorter on my system (-Os).
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Reviewed-by: NeilBrown <neilb@suse.de>

Pull powerpc fixes from Michael Ellerman:

 - Fix crashes coming out of nap on 32-bit Book3s (eg. powerbooks).

 - Fix critical and debug interrupts on BookE, seen as crashes when
   using ptrace.

 - Fix an oops when running an SMP kernel on a UP system.

 - Update pseries LPAR security flavor after partition migration.

 - Fix an oops when using kprobes on BookE.

 - Fix oops on 32-bit pmac by not calling do_IRQ() from
   timer_interrupt().

 - Fix softlockups on CPU hotplug into a CPU-less node with xive (P9).

Thanks to Cédric Le Goater, Christophe Leroy, Finn Thain, Geetika
Moolchandani, Laurent Dufour, Laurent Vivier, Nicholas Piggin, Pu Lehui,
Radu Rendec, Srikar Dronamraju, and Stan Johnson.

* tag 'powerpc-5.14-5' of git://git.kernel.org/pub/scm/linux/kernel/git/powerpc/linux:
  powerpc/xive: Do not skip CPU-less nodes when creating the IPIs
  powerpc/interrupt: Do not call single_step_exception() from other exceptions
  powerpc/interrupt: Fix OOPS by not calling do_IRQ() from timer_interrupt()
  powerpc/kprobes: Fix kprobe Oops happens in booke
  powerpc/pseries: Fix update of LPAR security flavor after LPM
  powerpc/smp: Fix OOPS in topology_init()
  powerpc/32: Fix critical and debug interrupts on BOOKE
  powerpc/32s: Fix napping restore in data storage interrupt (DSI)

Pull irq fixes from Thomas Gleixner:
 "A set of fixes for PCI/MSI and x86 interrupt startup:

   - Mask all MSI-X entries when enabling MSI-X otherwise stale unmasked
     entries stay around e.g. when a crashkernel is booted.

   - Enforce masking of a MSI-X table entry when updating it, which
     mandatory according to speification

   - Ensure that writes to MSI[-X} tables are flushed.

   - Prevent invalid bits being set in the MSI mask register

   - Properly serialize modifications to the mask cache and the mask
     register for multi-MSI.

   - Cure the violation of the affinity setting rules on X86 during
     interrupt startup which can cause lost and stale interrupts. Move
     the initial affinity setting ahead of actualy enabling the
     interrupt.

   - Ensure that MSI interrupts are completely torn down before freeing
     them in the error handling case.

   - Prevent an array out of bounds access in the irq timings code"

* tag 'irq-urgent-2021-08-15' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  driver core: Add missing kernel doc for device::msi_lock
  genirq/msi: Ensure deactivation on teardown
  genirq/timings: Prevent potential array overflow in __irq_timings_store()
  x86/msi: Force affinity setup before startup
  x86/ioapic: Force affinity setup before startup
  genirq: Provide IRQCHIP_AFFINITY_PRE_STARTUP
  PCI/MSI: Protect msi_desc::masked for multi-MSI
  PCI/MSI: Use msi_mask_irq() in pci_msi_shutdown()
  PCI/MSI: Correct misleading comments
  PCI/MSI: Do not set invalid bits in MSI mask
  PCI/MSI: Enforce MSI[X] entry updates to be visible
  PCI/MSI: Enforce that MSI-X table entry is masked for update
  PCI/MSI: Mask all unused MSI-X entries
  PCI/MSI: Enable and mask MSI-X early

Pull locking fix from Borislav Petkov:

 - Fix a CONFIG symbol's spelling

* tag 'locking_urgent_for_v5.14_rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  locking/rtmutex: Use the correct rtmutex debugging config option