linux-toradex.git/drivers/nvme, branch v4.9.78

nvme-loop: handle cpu unplug when re-establishing the controller

2017-12-25T13:23:37+00:00

[ Upstream commit 945dd5bacc8978439af276976b5dcbbd42333dbc ]

If a cpu unplug event has occured, we need to take the minimum
of the provided nr_io_queues and the number of online cpus,
otherwise we won't be able to connect them as blk-mq mapping
won't dispatch to those queues.

Reviewed-by: Christoph Hellwig 
Signed-off-by: Sagi Grimberg 
Signed-off-by: Sasha Levin 
Signed-off-by: Greg Kroah-Hartman

nvme: use kref_get_unless_zero in nvme_find_get_ns

2017-12-20T09:07:31+00:00

[ Upstream commit 2dd4122854f697afc777582d18548dded03ce5dd ]

For kref_get_unless_zero to protect against lookup vs free races we need
to use it in all places where we aren't guaranteed to already hold a
reference.  There is no such guarantee in nvme_find_get_ns, so switch to
kref_get_unless_zero in this function.

Signed-off-by: Christoph Hellwig 
Reviewed-by: Sagi Grimberg 
Reviewed-by: Hannes Reinecke 
Reviewed-by: Johannes Thumshirn 
Signed-off-by: Sasha Levin 
Signed-off-by: Greg Kroah-Hartman

nvmet-rdma: Fix a possible uninitialized variable dereference

2017-12-20T09:07:25+00:00

[ Upstream commit b25634e2a051bef4b2524b11adddfbfa6448f6cd ]

When handling a new recv command, we grab a new rsp resource and
check for the queue state being live. In case the queue is not in
live state, we simply restore the rsp back to the free list. However
in this flow we didn't set rsp->queue yet, so we cannot dereference it.

Instead, make sure to initialize rsp->queue (and other rsp members)
as soon as possible so we won't reference uninitialized variables.

Reported-by: Yi Zhang 
Reported-by: Raju Rangoju 
Reviewed-by: Christoph Hellwig 
Tested-by: Raju Rangoju 
Signed-off-by: Sagi Grimberg 
Signed-off-by: Sasha Levin 
Signed-off-by: Greg Kroah-Hartman

nvmet: confirm sq percpu has scheduled and switched to atomic

2017-12-20T09:07:25+00:00

[ Upstream commit d11ea004a458b982e19b188c386e25a9b66ec446 ]

percpu_ref_kill is not enough to prevent subsequent
percpu_ref_tryget_live from failing. Hence call
perfcpu_ref_kill_confirm to make it safe.

Reviewed-by: Christoph Hellwig 
Signed-off-by: Sagi Grimberg 
Signed-off-by: Sasha Levin 
Signed-off-by: Greg Kroah-Hartman

nvme-loop: fix a possible use-after-free when destroying the admin queue

2017-12-20T09:07:25+00:00

[ Upstream commit e4c5d3762e2d6d274bd1cc948c47063becfa2103 ]

we need to destroy the nvmet sq and let it finish gracefully
before continue to cleanup the queue.

Reviewed-by: Christoph Hellwig 
Signed-off-by: Sagi Grimberg 
Signed-off-by: Sasha Levin 
Signed-off-by: Greg Kroah-Hartman

nvmet: cancel fatal error and flush async work before free controller

2017-12-09T21:01:54+00:00

[ Upstream commit 06406d81a2d7cfb8abcc4fa6cdfeb8e5897007c5 ]

Make sure they are not running and we can free the controller
safely.

Signed-off-by: Roy Shterman 
Signed-off-by: Sagi Grimberg 
Reviewed-by: Christoph Hellwig 
Signed-off-by: Sasha Levin 
Signed-off-by: Greg Kroah-Hartman

nvme-pci: add quirk for delay before CHK RDY for WDC SN200

2017-12-05T10:24:34+00:00

commit 8c97eeccf0ad8783c057830119467b877bdfced7 upstream.

And increase the existing delay to cover this device as well.

Signed-off-by: Jeff Lien 
Signed-off-by: Christoph Hellwig 
Signed-off-by: Greg Kroah-Hartman

nvmet: fix KATO offset in Set Features

2017-11-30T08:39:15+00:00

[ Upstream commit 6c73f949300f17851f53fa80c9d1611ccd6909d3 ]

The Set Features implementation for Keep Alive Timer was using the wrong
structure when retrieving the KATO value; it was treating the Set
Features command as a Property Set command.

The NVMe spec defines the Keep Alive Timer feature as having one input
in CDW11 (4 bytes at offset 44 in the command) whereas the code was
reading 8 bytes at offset 48.

Since the Linux NVMe over Fabrics host never sets this feature, this
code has presumably never been tested.

Signed-off-by: Daniel Verkamp 
Signed-off-by: Christoph Hellwig 
Signed-off-by: Sasha Levin 
Signed-off-by: Greg Kroah-Hartman

nvme-pci: Use PCI bus address for data/queues in CMB

2017-10-12T09:51:25+00:00

commit 8969f1f8291762c13147c1ba89d46238af01675b upstream.

Currently, NVMe PCI host driver is programming CMB dma address as
I/O SQs addresses. This results in failures on systems where 1:1
outbound mapping is not used (example Broadcom iProc SOCs) because
CMB BAR will be progammed with PCI bus address but NVMe PCI EP will
try to access CMB using dma address.

To have CMB working on systems without 1:1 outbound mapping, we
program PCI bus address for I/O SQs instead of dma address. This
approach will work on systems with/without 1:1 outbound mapping.

Based on a report and previous patch from Abhishek Shah.

Fixes: 8ffaadf7 ("NVMe: Use CMB for the IO SQes if available")
Reported-by: Abhishek Shah 
Tested-by: Abhishek Shah 
Reviewed-by: Keith Busch 
Signed-off-by: Christoph Hellwig 
Signed-off-by: Greg Kroah-Hartman

nvme-rdma: handle cpu unplug when re-establishing the controller

2017-10-08T08:26:10+00:00

[ Upstream commit c248c64387fac5a6b31b343d9acb78f478e8619c ]

If a cpu unplug event has occured, we need to take the minimum
of the provided nr_io_queues and the number of online cpus,
otherwise we won't be able to connect them as blk-mq mapping
won't dispatch to those queues.

Reviewed-by: Christoph Hellwig 
Signed-off-by: Sagi Grimberg 
Signed-off-by: Sasha Levin 
Signed-off-by: Greg Kroah-Hartman