Signed-off-by: NIsrael Rukshin <israelr@mellanox.com>
Reviewed-by: NSagi Grimberg <sagi@grimberg.me>
Reviewed-by: NMax Gurtovoy <maxg@mellanox.com>
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

As for now, we don't care about sq_head pointer updates anyway, so
at least allow the controller to micro-optimize by omiting this update.

Note that we will probably need to support it when a controller
that requires this comes along.
Signed-off-by: NSagi Grimberg <sagi@grimberg.me>
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

If the controller supports traffic based keep alive, we restart the keep
alive timer if any admin or io commands was completed during the kato
period.  This prevents a possible starvation of keep alive commands in
the presence of heavy traffic as in such case, we already have a health
indication from the host perspective.

Only set a comp_seen indicator in case the controller supports keep
alive to minimize the overhead for pci controllers.
Signed-off-by: NSagi Grimberg <sagi@grimberg.me>
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

We get the controller attributes in identify, cache them as we'll need
them for traffic based keep alive support.
Reviewed-by: NChaitanya Kulkarni <chaitanya.kulkarni@wdc.com>
Signed-off-by: NSagi Grimberg <sagi@grimberg.me>
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

Instead of directly poking into the struct device add a new numa_node
field to struct nvme_ctrl.  This allows fabrics drivers where ctrl->dev
is a virtual device to support NUMA affinity as well.

Also expose the field as a sysfs attribute, and populate it for the
RDMA and FC transports.
Signed-off-by: NHannes Reinecke <hare@suse.com>
Reviewed-by: NSagi Grimberg <sagi@grimberg.me>
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

In function nvme_setup_cmd() we call command specific setup function
for flush, rw, and discard. Instead of calling memset in each function
lets call it once in the parent function.

This is purely code cleanup patch and it does not change any existing
functionality.
Signed-off-by: NChaitanya Kulkarni <chaitanya.kulkarni@wdc.com>
Reviewed-by: NSagi Grimberg <sagi@grimberg.me>
Reviewed-by: NJohannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

This avoids having to have differnet mq_ops for different setups
with or without poll queues.
Reviewed-by: NSagi Grimberg <sagi@grimberg.me>
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

The ->poll_fn has been stale for a while, as a lot of places check for mq
ops.  But there is no real point in it anyway, as we don't even use
the multipath code for subsystems without multiple ports, which is usually
what we do high performance I/O to.  If it really becomes an issue we
should rework the nvme code to also skip the multipath code for any
private namespace, even if that could mean some trouble when rescanning.
Reviewed-by: NKeith Busch <keith.busch@intel.com>
Reviewed-by: NSagi Grimberg <sagi@grimberg.me>
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

The code was always a bit of a hack that digs far too much into
RDMA core internals.  Lets kick it out and reimplement proper
dedicated poll queues as needed.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

Now that we can't poll regular, interrupt driven I/O queues there
is almost nothing that can race with an interrupt.  The only
possible other contexts polling a CQ are the error handler and
queue shutdown, and both are so far off in the slow path that
we can simply use the big hammer of disabling interrupts.

With that we can stop taking the cq_lock for normal queues.
Reviewed-by: NSagi Grimberg <sagi@grimberg.me>
Reviewed-by: NKeith Busch <keith.busch@intel.com>
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

This is the last place outside of nvme_irq that handles CQEs from
interrupt context, and thus is in the way of removing the cq_lock for
normal queues, and avoiding lockdep warnings on the poll queues, for
which we already take it without IRQ disabling.
Reviewed-by: NKeith Busch <keith.busch@intel.com>
Reviewed-by: NSagi Grimberg <sagi@grimberg.me>
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

Pass the opcode for the delete SQ/CQ command as an argument instead of
the somewhat confusing pass loop.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Reviewed-by: NKeith Busch <keith.busch@intel.com>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

We have three places that can poll for I/O completions on a normal
interrupt-enabled queue.  All of them are in slow path code, so
consolidate them to a single helper that uses spin_lock_irqsave and
removes the fast path cqe_pending check.
Reviewed-by: NSagi Grimberg <sagi@grimberg.me>
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

This will allow us to simplify both the regular NVMe interrupt handler
and the upcoming aio poll code.  In addition to that the separate
queues are generally a good idea for performance reasons.
Reviewed-by: NKeith Busch <keith.busch@intel.com>
Reviewed-by: NSagi Grimberg <sagi@grimberg.me>
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

Use a bit flag to mark if the SQ was allocated from the CMB, and clean
up the surrounding code a bit.
Reviewed-by: NSagi Grimberg <sagi@grimberg.me>
Reviewed-by: NKeith Busch <keith.busch@intel.com>
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

This gets rid of all the messing with cq_vector and the ->polled field
by using an atomic bitop to mark the queue enabled or not.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Reviewed-by: NKeith Busch <keith.busch@intel.com>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

Having another indirect all in the fast path doesn't really help
in our post-spectre world.  Also having too many queue type is just
going to create confusion, so I'd rather manage them centrally.

Note that the queue type naming and ordering changes a bit - the
first index now is the default queue for everything not explicitly
marked, the optional ones are read and poll queues.
Reviewed-by: NSagi Grimberg <sagi@grimberg.me>
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

Some error paths in configuration of admin queue free data buffer
associated with async request SQE without resetting the data buffer
pointer to NULL, This buffer is also freed up again if the controller
is shutdown or reset.
Signed-off-by: NPrabhath Sajeepa <psajeepa@purestorage.com>
Reviewed-by: NRoland Dreier <roland@purestorage.com>
Signed-off-by: NChristoph Hellwig <hch@lst.de>

nvme_stop_ctrl can be called also for reset flow and there is no need to
flush the scan_work as namespaces are not being removed. This can cause
deadlock in rdma, fc and loop drivers since nvme_stop_ctrl barriers
before controller teardown (and specifically I/O cancellation of the
scan_work itself) takes place, but the scan_work will be blocked anyways
so there is no need to flush it.

Instead, move scan_work flush to nvme_remove_namespaces() where it really
needs to flush.
Reported-by: NMing Lei <ming.lei@redhat.com>
Signed-off-by: NSagi Grimberg <sagi@grimberg.me>
Reviewed-by: NKeith Busch <keith.busch@intel.com>
Reviewed by: James Smart <jsmart2021@gmail.com>
Tested-by: NEwan D. Milne <emilne@redhat.com>
Signed-off-by: NChristoph Hellwig <hch@lst.de>

Without CONFIG_NVME_MULTIPATH enabled a multi-port subsystem might
show up as invididual devices and cause problems, warn about it.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Reviewed-by: NSagi Grimberg <sagi@grimberg.me>

Split the command submission and the SQ doorbell ring, and add the
doorbell ring as our ->commit_rqs() hook. This allows a list of
requests to be issued, with nvme only writing the SQ update when
it's necessary. This is more efficient if we have lists of requests
to issue, particularly on virtualized hardware, where writing the
SQ doorbell is more expensive than on real hardware. For those cases,
performance increases of 2-3x have been observed.

The use case for this is plugged IO, where blk-mq flushes a batch of
requests at the time.
Reviewed-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

When a PCIe NVMe device is not present, nvme_dev_remove_admin() calls
blk_cleanup_queue() on the admin queue, which frees the hctx for that
queue.  Moments later, on the same path nvme_kill_queues() calls
blk_mq_unquiesce_queue() on admin queue and tries to access hctx of it,
which leads to following OOPS:

Oops: 0000 [#1] SMP PTI
RIP: 0010:sbitmap_any_bit_set+0xb/0x40
Call Trace:
 blk_mq_run_hw_queue+0xd5/0x150
 blk_mq_run_hw_queues+0x3a/0x50
 nvme_kill_queues+0x26/0x50
 nvme_remove_namespaces+0xb2/0xc0
 nvme_remove+0x60/0x140
 pci_device_remove+0x3b/0xb0

Fixes: cb4bfda6 ("nvme-pci: fix hot removal during error handling")
Signed-off-by: NIgor Konopko <igor.j.konopko@intel.com>
Reviewed-by: NKeith Busch <keith.busch@intel.com>
Signed-off-by: NChristoph Hellwig <hch@lst.de>

__nvme_fc_init_request() invokes memset() on the nvme_fcp_op_w_sgl structure, which
NULLed-out the nvme_req(req)->ctrl field previously set by nvme_fc_init_request().
This apparently was not referenced until commit faf4a44fff ("nvme: support traffic
based keep-alive") which now results in a crash in nvme_complete_rq():

[ 8386.897130] RIP: 0010:panic+0x220/0x26c
[ 8386.901406] Code: 83 3d 6f ee 72 01 00 74 05 e8 e8 54 02 00 48 c7 c6 40 fd 5b b4 48 c7 c7 d8 8d c6 b3 31e
[ 8386.922359] RSP: 0018:ffff99650019fc40 EFLAGS: 00000246 ORIG_RAX: ffffffffffffff13
[ 8386.930804] RAX: 0000000000000046 RBX: 0000000000000000 RCX: 0000000000000006
[ 8386.938764] RDX: 0000000000000000 RSI: 0000000000000082 RDI: ffff8e325f8168b0
[ 8386.946725] RBP: ffff99650019fcb0 R08: 0000000000000000 R09: 00000000000004f8
[ 8386.954687] R10: 0000000000000000 R11: ffff99650019f9b8 R12: ffffffffb3c55f3c
[ 8386.962648] R13: 0000000000000000 R14: 0000000000000000 R15: 0000000000000001
[ 8386.970613]  oops_end+0xd1/0xe0
[ 8386.974116]  no_context+0x1b2/0x3c0
[ 8386.978006]  do_page_fault+0x32/0x140
[ 8386.982090]  page_fault+0x1e/0x30
[ 8386.985786] RIP: 0010:nvme_complete_rq+0x65/0x1d0 [nvme_core]
[ 8386.992195] Code: 41 bc 03 00 00 00 74 16 0f 86 c3 00 00 00 66 3d 83 00 41 bc 06 00 00 00 0f 85 e7 00 000
[ 8387.013147] RSP: 0018:ffff99650019fe18 EFLAGS: 00010246
[ 8387.018973] RAX: 0000000000000000 RBX: ffff8e322ae51280 RCX: 0000000000000001
[ 8387.026935] RDX: 0000000000000400 RSI: 0000000000000001 RDI: ffff8e322ae51280
[ 8387.034897] RBP: ffff8e322ae51280 R08: 0000000000000000 R09: ffffffffb2f0b890
[ 8387.042859] R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000
[ 8387.050821] R13: 0000000000000100 R14: 0000000000000004 R15: ffff8e2b0446d990
[ 8387.058782]  ? swiotlb_unmap_page+0x40/0x40
[ 8387.063448]  nvme_fc_complete_rq+0x2d/0x70 [nvme_fc]
[ 8387.068986]  blk_done_softirq+0xa1/0xd0
[ 8387.073264]  __do_softirq+0xd6/0x2a9
[ 8387.077251]  run_ksoftirqd+0x26/0x40
[ 8387.081238]  smpboot_thread_fn+0x10e/0x160
[ 8387.085807]  kthread+0xf8/0x130
[ 8387.089309]  ? sort_range+0x20/0x20
[ 8387.093198]  ? kthread_stop+0x110/0x110
[ 8387.097475]  ret_from_fork+0x35/0x40
[ 8387.101462] ---[ end trace 7106b0adf5e422f8 ]---

Fixes: faf4a44fff ("nvme: support traffic based keep-alive")
Signed-off-by: NEwan D. Milne <emilne@redhat.com>
Reviewed-by: NHannes Reinecke <hare@suse.com>
Signed-off-by: NChristoph Hellwig <hch@lst.de>

Free the kobject name that was allocated for the controller device on
failure rather than its parent.

Fixes: d22524a4 ("nvme: switch controller refcounting to use struct device")
Signed-off-by: NKeith Busch <keith.busch@intel.com>
Signed-off-by: NChristoph Hellwig <hch@lst.de>

blk_poll() has always kept spinning until it found an IO. This is
fine for SYNC polling, since we need to find one request we have
pending, but in preparation for ASYNC polling it can be beneficial
to just check if we have any entries available or not.

Existing callers are converted to pass in 'spin == true', to retain
the old behavior.
Signed-off-by: NJens Axboe <axboe@kernel.dk>

We always pass in -1 now and none of the callers use the tag value,
remove the parameter.
Reviewed-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

If we want to support async IO polling, then we have to allow finding
completions that aren't just for the one we are looking for. Always pass
in -1 to the mq_ops->poll() helper, and have that return how many events
were found in this poll loop.
Signed-off-by: NJens Axboe <axboe@kernel.dk>

It's specifically looking for a given request, which we will not be
supporting going forward. Also kill the qla2xxx poll implementation
as that's the only user of the nvme-fc poll, and the now unused
->poll_queue() hook.
Reviewed-by: NChristoph Hellwig <hch@lst.de>
Reviewed-by: NJames Smart <jsmart2021@gmail.com>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

We currently only really support sync poll, ie poll with 1 IO in flight.
This prepares us for supporting async poll.

Note that the returned value isn't necessarily 100% accurate. If poll
races with IRQ completion, we assume that the fact that the task is now
runnable means we found at least one entry. In reality it could be more
than 1, or not even 1. This is fine, the caller will just need to take
this into account.
Reviewed-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

We need a better way of configuring this, and given that polling is
(still) a bit niche, let's default to using 0 poll queues. That way
we'll have the same read/write/poll behavior as 4.20, and users that
want to test/use polling are required to do manual configuration of the
number of poll queues.
Reviewed-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

If we have separate poll queues, we know that they aren't using
interrupts. Hence we don't need to disable interrupts around
finding completions.

Provide a separate set of blk_mq_ops for such devices.
Signed-off-by: NJens Axboe <axboe@kernel.dk>

At least on SPARC, if MSI/MSI-X isn't supported, we get EINVAL if
we ask for more than one vector. This isn't covered by our ENOSPC
check.

If we get EINVAL, decrease our ask to just one vector, instead of
bailing out in error.

Fixes: 3b6592f7 ("nvme: utilize two queue maps, one for reads and one for writes")
Reported-by: NGuenter Roeck <linux@roeck-us.net>
Tested-by: NGuenter Roeck <linux@roeck-us.net>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

With the legacy request path gone there is no real need to override the
queue_lock.
Reviewed-by: NHannes Reinecke <hare@suse.com>
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

If an io error occurs on an io issued while connecting, recovery
of the io falls flat as the state checking ends up nooping the error
handler.

Create an err_work work item that is scheduled upon an io error while
connecting. The work thread terminates all io on all queues and marks
the queues as not connected.  The termination of the io will return
back to the callee, which will then back out of the connection attempt
and will reschedule, if possible, the connection attempt.

The changes:
- in case there are several commands hitting the error handler, a
  state flag is kept so that the error work is only scheduled once,
  on the first error. The subsequent errors can be ignored.
- The calling sequence to stop keep alive and terminate the queues
  and their io is lifted from the reset routine. Made a small
  service routine used by both reset and err_work.
- During debugging, found that the teardown path can reference
  an uninitialized pointer, resulting in a NULL pointer oops.
  The aen_ops weren't initialized yet. Add validation on their
  initialization before calling the teardown routine.
Signed-off-by: NJames Smart <jsmart2021@gmail.com>
Signed-off-by: NChristoph Hellwig <hch@lst.de>

NVMe always asks for io_queues + 1 worth of IRQ vectors, which
means that even when we scale all the way down, we still ask
for 2 vectors and get -ENOSPC in return if the system can't
support more than 1.

Getting just 1 vector is fine, it just means that we'll have
1 IO queue and 1 admin queue, with a shared vector between them.
Check for this case and don't add our + 1 if it happens.

Fixes: 3b6592f7 ("nvme: utilize two queue maps, one for reads and one for writes")
Reported-by: NGuenter Roeck <linux@roeck-us.net>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

Whenever we update ns_head info, we need to make sure it is still
compatible with all underlying backing devices because although nvme
multipath doesn't have any explicit use of these limits, other devices
can still be stacked on top of it which may rely on the underlying limits.
Start with unlimited stacking limits, and every info update iterate over
siblings and adjust queue limits.
Signed-off-by: NSagi Grimberg <sagi@grimberg.me>
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

We have this functionality in sbitmap, but we don't export it in
blk-mq for users of the tags busy iteration. This can be useful
for stopping the iteration, if the caller doesn't need to find
more requests.
Reviewed-by: NMike Snitzer <snitzer@redhat.com>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

Adds support for defining a variable number of poll queues, currently
configurable with the 'poll_queues' module parameter. Defaults to
a single poll queue.

And now we finally have poll support without triggering interrupts!
Reviewed-by: NHannes Reinecke <hare@suse.com>
Reviewed-by: NKeith Busch <keith.busch@intel.com>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

NVMe does round-robin between queues by default, which means that
sharing a queue map for both reads and writes can be problematic
in terms of read servicing. It's much easier to flood the queue
with writes and reduce the read servicing.

Implement two queue maps, one for reads and one for writes. The
write queue count is configurable through the 'write_queues'
parameter.

By default, we retain the previous behavior of having a single
queue set, shared between reads and writes. Setting 'write_queues'
to a non-zero value will create two queue sets, one for reads and
one for writes, the latter using the configurable number of
queues (hardware queue counts permitting).
Reviewed-by: NHannes Reinecke <hare@suse.com>
Reviewed-by: NKeith Busch <keith.busch@intel.com>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

This is in preparation for allowing multiple sets of maps per
queue, if so desired.
Reviewed-by: NHannes Reinecke <hare@suse.com>
Reviewed-by: NBart Van Assche <bvanassche@acm.org>
Reviewed-by: NKeith Busch <keith.busch@intel.com>
Signed-off-by: NJens Axboe <axboe@kernel.dk>