linux-stable.git/drivers/nvme, branch v5.4.63

nvme: multipath: round-robin: fix single non-optimized path case

2020-09-03T09:26:55+00:00

[ Upstream commit 93eb0381e13d249a18ed4aae203291ff977e7ffb ]

If there's only one usable, non-optimized path, nvme_round_robin_path()
returns NULL, which is wrong. Fix it by falling back to "old", like in
the single optimized path case. Also, if the active path isn't changed,
there's no need to re-assign the pointer.

Fixes: 3f6e3246db0e ("nvme-multipath: fix logic for non-optimized paths")
Signed-off-by: Martin Wilck 
Signed-off-by: Martin George 
Reviewed-by: Keith Busch 
Signed-off-by: Sagi Grimberg 
Signed-off-by: Jens Axboe 
Signed-off-by: Sasha Levin

nvme-fc: Fix wrong return value in __nvme_fc_init_request()

2020-09-03T09:26:55+00:00

[ Upstream commit f34448cd0dc697723fb5f4118f8431d9233b370d ]

On an error exit path, a negative error code should be returned
instead of a positive return value.

Fixes: e399441de9115 ("nvme-fabrics: Add host support for FC transport")
Cc: James Smart 
Signed-off-by: Tianjia Zhang 
Reviewed-by: Chaitanya Kulkarni 
Reviewed-by: Christoph Hellwig 
Signed-off-by: Sagi Grimberg 
Signed-off-by: Jens Axboe 
Signed-off-by: Sasha Levin

nvme-multipath: do not fall back to __nvme_find_path() for non-optimized paths

2020-08-19T06:15:59+00:00

[ Upstream commit fbd6a42d8932e172921c7de10468a2e12c34846b ]

When nvme_round_robin_path() finds a valid namespace we should be using it;
falling back to __nvme_find_path() for non-optimized paths will cause the
result from nvme_round_robin_path() to be ignored for non-optimized paths.

Fixes: 75c10e732724 ("nvme-multipath: round-robin I/O policy")
Signed-off-by: Martin Wilck 
Signed-off-by: Hannes Reinecke 
Reviewed-by: Sagi Grimberg 
Signed-off-by: Christoph Hellwig 
Signed-off-by: Sasha Levin

nvme-multipath: fix logic for non-optimized paths

2020-08-19T06:15:59+00:00

[ Upstream commit 3f6e3246db0e6f92e784965d9d0edb8abe6c6b74 ]

Handle the special case where we have exactly one optimized path,
which we should keep using in this case.

Fixes: 75c10e732724 ("nvme-multipath: round-robin I/O policy")
Signed off-by: Martin Wilck 
Signed-off-by: Hannes Reinecke 
Reviewed-by: Sagi Grimberg 
Signed-off-by: Christoph Hellwig 
Signed-off-by: Sasha Levin

nvme-rdma: fix controller reset hang during traffic

2020-08-19T06:15:59+00:00

[ Upstream commit 9f98772ba307dd89a3d17dc2589f213d3972fc64 ]

commit fe35ec58f0d3 ("block: update hctx map when use multiple maps")
exposed an issue where we may hang trying to wait for queue freeze
during I/O. We call blk_mq_update_nr_hw_queues which in case of multiple
queue maps (which we have now for default/read/poll) is attempting to
freeze the queue. However we never started queue freeze when starting the
reset, which means that we have inflight pending requests that entered the
queue that we will not complete once the queue is quiesced.

So start a freeze before we quiesce the queue, and unfreeze the queue
after we successfully connected the I/O queues (and make sure to call
blk_mq_update_nr_hw_queues only after we are sure that the queue was
already frozen).

This follows to how the pci driver handles resets.

Fixes: fe35ec58f0d3 ("block: update hctx map when use multiple maps")
Signed-off-by: Sagi Grimberg 
Signed-off-by: Christoph Hellwig 
Signed-off-by: Sasha Levin

nvme-tcp: fix controller reset hang during traffic

2020-08-19T06:15:58+00:00

[ Upstream commit 2875b0aecabe2f081a8432e2bc85b85df0529490 ]

commit fe35ec58f0d3 ("block: update hctx map when use multiple maps")
exposed an issue where we may hang trying to wait for queue freeze
during I/O. We call blk_mq_update_nr_hw_queues which in case of multiple
queue maps (which we have now for default/read/poll) is attempting to
freeze the queue. However we never started queue freeze when starting the
reset, which means that we have inflight pending requests that entered the
queue that we will not complete once the queue is quiesced.

So start a freeze before we quiesce the queue, and unfreeze the queue
after we successfully connected the I/O queues (and make sure to call
blk_mq_update_nr_hw_queues only after we are sure that the queue was
already frozen).

This follows to how the pci driver handles resets.

Fixes: fe35ec58f0d3 ("block: update hctx map when use multiple maps")
Signed-off-by: Sagi Grimberg 
Signed-off-by: Christoph Hellwig 
Signed-off-by: Sasha Levin

nvme: add a Identify Namespace Identification Descriptor list quirk

2020-08-19T06:15:53+00:00

commit 5bedd3afee8eb01ccd256f0cd2cc0fa6f841417a upstream.

Add a quirk for a device that does not support the Identify Namespace
Identification Descriptor list despite claiming 1.3 compliance.

Fixes: ea43d9709f72 ("nvme: fix identify error status silent ignore")
Reported-by: Ingo Brunberg 
Signed-off-by: Christoph Hellwig 
Tested-by: Ingo Brunberg 
Reviewed-by: Sagi Grimberg 
Cc: Plamen Lyutov 
Signed-off-by: Greg Kroah-Hartman

nvme-pci: prevent SK hynix PC400 from using Write Zeroes command

2020-08-11T13:33:37+00:00

[ Upstream commit 5611ec2b9814bc91f7b0a8d804c1fc152e2025d9 ]

After commit 6e02318eaea5 ("nvme: add support for the Write Zeroes
command"), SK hynix PC400 becomes very slow with the following error
message:

[  224.567695] blk_update_request: operation not supported error, dev nvme1n1, sector 499384320 op 0x9:(WRITE_ZEROES) flags 0x1000000 phys_seg 0 prio class 0]

SK Hynix PC400 has a buggy firmware that treats NLB as max value instead
of a range, so the NLB passed isn't a valid value to the firmware.

According to SK hynix there are three commands are affected:
- Write Zeroes
- Compare
- Write Uncorrectable

Right now only Write Zeroes is implemented, so disable it completely on
SK hynix PC400.

BugLink: https://bugs.launchpad.net/bugs/1872383
Cc: kyounghwan sohn 
Signed-off-by: Kai-Heng Feng 
Signed-off-by: Christoph Hellwig 
Signed-off-by: Sasha Levin

nvme-tcp: fix possible hang waiting for icresp response

2020-08-05T07:59:45+00:00

[ Upstream commit adc99fd378398f4c58798a1c57889872967d56a6 ]

If the controller died exactly when we are receiving icresp
we hang because icresp may never return. Make sure to set a
high finite limit.

Fixes: 3f2304f8c6d6 ("nvme-tcp: add NVMe over TCP host driver")
Signed-off-by: Sagi Grimberg 
Signed-off-by: Christoph Hellwig 
Signed-off-by: Sasha Levin

nvme-rdma: assign completion vector correctly

2020-07-16T06:16:36+00:00

[ Upstream commit 032a9966a22a3596addf81dacf0c1736dfedc32a ]

The completion vector index that is given during CQ creation can't
exceed the number of support vectors by the underlying RDMA device. This
violation currently can accure, for example, in case one will try to
connect with N regular read/write queues and M poll queues and the sum
of N + M > num_supported_vectors. This will lead to failure in establish
a connection to remote target. Instead, in that case, share a completion
vector between queues.

Signed-off-by: Max Gurtovoy 
Signed-off-by: Christoph Hellwig 
Signed-off-by: Sasha Levin