linux-toradex.git/drivers/nvme, branch v4.9.49

nvme-rdma: remove race conditions from IB signalling

2017-07-27T22:08:03+00:00

commit 5e599d73c1c1816af07f94ddba879499aa39b43c upstream.

This patch improves the way the RDMA IB signalling is done by using atomic
operations for the signalling variable. This avoids race conditions on
sig_count.

The signalling interval changes slightly and is now the largest power of
two not larger than queue depth / 2.

ilog() usage idea by Bart Van Assche.

Signed-off-by: Marta Rybczynska 
Reviewed-by: Sagi Grimberg 
Signed-off-by: Christoph Hellwig 
Signed-off-by: Greg Kroah-Hartman

nvmet-rdma: Fix missing dma sync to nvme data structures

2017-06-17T04:41:55+00:00

[ Upstream commit 748ff8408f8e208f279ba221e5c12612fbb4dddb ]

This patch performs dma sync operations on nvme_command
and nvme_completion.

nvme_command is synced
(a) on receiving of the recv queue completion for cpu access.
(b) before posting recv wqe back to rdma adapter for device access.

nvme_completion is synced
(a) on receiving of the recv queue completion of associated
nvme_command for cpu access.
(b) before posting send wqe to rdma adapter for device access.

This patch is generated for git://git.infradead.org/nvme-fabrics.git
Branch: nvmf-4.10

Signed-off-by: Parav Pandit 
Reviewed-by: Max Gurtovoy 
Signed-off-by: Sasha Levin 
Signed-off-by: Greg Kroah-Hartman

nvme: avoid to use blk_mq_abort_requeue_list()

2017-06-07T10:07:48+00:00

commit 986f75c876dbafed98eba7cb516c5118f155db23 upstream.

NVMe may add request into requeue list simply and not kick off the
requeue if hw queues are stopped. Then blk_mq_abort_requeue_list()
is called in both nvme_kill_queues() and nvme_ns_remove() for
dealing with this issue.

Unfortunately blk_mq_abort_requeue_list() is absolutely a
race maker, for example, one request may be requeued during
the aborting. So this patch just calls blk_mq_kick_requeue_list() in
nvme_kill_queues() to handle this issue like what nvme_start_queues()
does. Now all requests in requeue list when queues are stopped will be
handled by blk_mq_kick_requeue_list() when queues are restarted, either
in nvme_start_queues() or in nvme_kill_queues().

Reported-by: Zhang Yi 
Reviewed-by: Keith Busch 
Reviewed-by: Johannes Thumshirn 
Signed-off-by: Ming Lei 
Signed-off-by: Christoph Hellwig 
Signed-off-by: Greg Kroah-Hartman

nvme: use blk_mq_start_hw_queues() in nvme_kill_queues()

2017-06-07T10:07:48+00:00

commit 806f026f9b901eaf1a6baeb48b5da18d6a4f818e upstream.

Inside nvme_kill_queues(), we have to start hw queues for
draining requests in sw queues, .dispatch list and requeue list,
so use blk_mq_start_hw_queues() instead of blk_mq_start_stopped_hw_queues()
which only run queues if queues are stopped, but the queues may have
been started already, for example nvme_start_queues() is called in reset work
function.

blk_mq_start_hw_queues() run hw queues in current context, instead
of running asynchronously like before. Given nvme_kill_queues() is
run from either remove context or reset worker context, both are fine
to run hw queue directly. And the mutex of namespaces_mutex isn't a
problem too becasue nvme_start_freeze() runs hw queue in this way
already.

Reported-by: Zhang Yi 
Reviewed-by: Keith Busch 
Reviewed-by: Johannes Thumshirn 
Signed-off-by: Ming Lei 
Signed-off-by: Christoph Hellwig 
Signed-off-by: Greg Kroah-Hartman

nvme-rdma: support devices with queue size < 32

2017-06-07T10:07:48+00:00

commit 0544f5494a03b8846db74e02be5685d1f32b06c9 upstream.

In the case of small NVMe-oF queue size (<32) we may enter a deadlock
caused by the fact that the IB completions aren't sent waiting for 32
and the send queue will fill up.

The error is seen as (using mlx5):
[ 2048.693355] mlx5_0:mlx5_ib_post_send:3765:(pid 7273):
[ 2048.693360] nvme nvme1: nvme_rdma_post_send failed with error code -12

This patch changes the way the signaling is done so that it depends on
the queue depth now. The magic define has been removed completely.

Signed-off-by: Marta Rybczynska 
Signed-off-by: Samuel Jones 
Acked-by: Sagi Grimberg 
Signed-off-by: Christoph Hellwig 
Signed-off-by: Greg Kroah-Hartman

nvme: unmap CMB and remove sysfs file in reset path

2017-05-25T13:44:46+00:00

commit f63572dff1421b6ca6abce71d46e03411e605c94 upstream.

CMB doesn't get unmapped until removal while getting remapped on every
reset. Add the unmapping and sysfs file removal to the reset path in
nvme_pci_disable to match the mapping path in nvme_pci_enable.

Fixes: 202021c1a ("nvme : Add sysfs entry for NVMe CMBs when appropriate")

Signed-off-by: Jon Derrick 
Acked-by: Keith Busch 
Reviewed-By: Stephen Bates 
Signed-off-by: Christoph Hellwig 
Signed-off-by: Jens Axboe 
Signed-off-by: Greg Kroah-Hartman

nvme: simplify stripe quirk

2017-04-12T10:41:18+00:00

[ Upstream commit e6282aef7b89a11d26e731060c4409b7aac278bf ]

Some OEMs believe they own the Identify Controller vendor specific
region and will repurpose it with their own values. While not common,
we can't rely on the PCI VID:DID to tell use how to decode the field
we reserved for this as the stripe size so we need to do something else
for the list of devices using this quirk.

The field was supposed to allow flexibility on the device's back-end
striping, but it turned out that never materialized; the chunk is always
the same as MDTS in the products subscribing to this quirk, so this
patch removes the stripe_size field and sets the chunk to the max hw
transfer size for the devices using this quirk.

Signed-off-by: Keith Busch 
Signed-off-by: Christoph Hellwig 
Signed-off-by: Sasha Levin 
Signed-off-by: Greg Kroah-Hartman

nvme/pci: Disable on removal when disconnected

2017-04-08T07:30:36+00:00

commit 6db28eda266052f86a6b402422de61eeb7d2e351 upstream.

If the device is not present, the driver should disable the queues
immediately. Prior to this, the driver was relying on the watchdog timer
to kill the queues if requests were outstanding to the device, and that
just delays removal up to one second.

Signed-off-by: Keith Busch 
Reviewed-by: Johannes Thumshirn 
Reviewed-by: Christoph Hellwig 
Signed-off-by: Sagi Grimberg 
Signed-off-by: Jens Axboe 
Signed-off-by: Greg Kroah-Hartman

nvme/core: Fix race kicking freed request_queue

2017-04-08T07:30:36+00:00

commit f33447b90e96076483525b21cc4e0a8977cdd07c upstream.

If a namespace has already been marked dead, we don't want to kick the
request_queue again since we may have just freed it from another thread.

Signed-off-by: Keith Busch 
Reviewed-by: Johannes Thumshirn 
Reviewed-by: Christoph Hellwig 
Signed-off-by: Sagi Grimberg 
Signed-off-by: Jens Axboe 
Signed-off-by: Greg Kroah-Hartman

nvme: apply DELAY_BEFORE_CHK_RDY quirk at probe time too

2017-01-19T19:18:05+00:00

commit b5a10c5f7532b7473776da87e67f8301bbc32693 upstream.

Commit 54adc01055b7 ("nvme/quirk: Add a delay before checking for adapter
readiness") introduced a quirk to adapters that cannot read the bit
NVME_CSTS_RDY right after register NVME_REG_CC is set; these adapters
need a delay or else the action of reading the bit NVME_CSTS_RDY could
somehow corrupt adapter's registers state and it never recovers.

When this quirk was added, we checked ctrl->tagset in order to avoid
quirking in probe time, supposing we would never require such delay
during probe. Well, it was too optimistic; we in fact need this quirk
at probe time in some cases, like after a kexec.

In some experiments, after abnormal shutdown of machine (aka power cord
unplug), we booted into our bootloader in Power, which is a Linux kernel,
and kexec'ed into another distro. If this kexec is too quick, we end up
reaching the probe of NVMe adapter in that distro when adapter is in
bad state (not fully initialized on our bootloader). What happens next
is that nvme_wait_ready() is unable to complete, except if the quirk is
enabled.

So, this patch removes the original ctrl->tagset verification in order
to enable the quirk even on probe time.

Fixes: 54adc01055b7 ("nvme/quirk: Add a delay before checking for adapter readiness")
Reported-by: Andrew Byrne 
Reported-by: Jaime A. H. Gomez 
Reported-by: Zachary D. Myers 
Signed-off-by: Guilherme G. Piccoli 
Acked-by: Jeffrey Lien 
Signed-off-by: Christoph Hellwig 
Signed-off-by: Greg Kroah-Hartman