linux-stable.git/drivers/scsi, branch v4.4.27

scsi: ibmvfc: Fix I/O hang when port is not mapped

2016-10-22T10:26:56+00:00

commit 07d0e9a847401ffd2f09bd450d41644cd090e81d upstream.

If a VFC port gets unmapped in the VIOS, it may not respond with a CRQ
init complete following H_REG_CRQ. If this occurs, we can end up having
called scsi_block_requests and not a resulting unblock until the init
complete happens, which may never occur, and we end up hanging I/O
requests.  This patch ensures the host action stay set to
IBMVFC_HOST_ACTION_TGT_DEL so we move all rports into devloss state and
unblock unless we receive an init complete.

Signed-off-by: Brian King 
Acked-by: Tyrel Datwyler 
Signed-off-by: Martin K. Petersen 
Signed-off-by: Greg Kroah-Hartman

scsi: arcmsr: Simplify user_len checking

2016-10-22T10:26:56+00:00

commit 4bd173c30792791a6daca8c64793ec0a4ae8324f upstream.

Do the user_len check first and then the ver_addr allocation so that we
can save us the kfree() on the error path when user_len is >
ARCMSR_API_DATA_BUFLEN.

Signed-off-by: Borislav Petkov 
Cc: Marco Grassi 
Cc: Dan Carpenter 
Cc: Tomas Henzl 
Cc: Martin K. Petersen 
Reviewed-by: Johannes Thumshirn 
Reviewed-by: Tomas Henzl 
Signed-off-by: Martin K. Petersen 
Signed-off-by: Greg Kroah-Hartman

scsi: arcmsr: Buffer overflow in arcmsr_iop_message_xfer()

2016-10-22T10:26:55+00:00

commit 7bc2b55a5c030685b399bb65b6baa9ccc3d1f167 upstream.

We need to put an upper bound on "user_len" so the memcpy() doesn't
overflow.

Reported-by: Marco Grassi 
Signed-off-by: Dan Carpenter 
Reviewed-by: Tomas Henzl 
Signed-off-by: Martin K. Petersen 
Signed-off-by: Greg Kroah-Hartman

fnic: pci_dma_mapping_error() doesn't return an error code

2016-10-07T13:23:45+00:00

commit dd7328e4c53649c1c7ec36bc1cf5b229b8662047 upstream.

pci_dma_mapping_error() returns true on error and false on success.

Fixes: fd6ddfa4c1dd ('fnic: check pci_map_single() return value')
Signed-off-by: Dan Carpenter 
Signed-off-by: Martin K. Petersen 
Signed-off-by: Greg Kroah-Hartman

megaraid: fix null pointer check in megasas_detach_one().

2016-09-30T08:18:36+00:00

commit 546e559c79b1a8d27c23262907a00fc209e392a0 upstream.

The pd_seq_sync pointer can't be NULL, we have to check its entries
instead.

Signed-off-by: Maurizio Lombardi 
Acked-by: Sumit Saxena 
Reviewed-by: Tomas Henzl 
Signed-off-by: Martin K. Petersen 
Signed-off-by: Greg Kroah-Hartman

scsi: fix upper bounds check of sense key in scsi_sense_key_string()

2016-09-15T06:27:54+00:00

commit a87eeb900dbb9f8202f96604d56e47e67c936b9d upstream.

Commit 655ee63cf371 ("scsi constants: command, sense key + additional
sense string") added a "Completed" sense string with key 0xF to
snstext[], but failed to updated the upper bounds check of the sense key
in scsi_sense_key_string().

Fixes: 655ee63cf371 ("[SCSI] scsi constants: command, sense key + additional sense strings")
Signed-off-by: Tyrel Datwyler 
Reviewed-by: Bart Van Assche 
Signed-off-by: Martin K. Petersen 
Signed-off-by: Greg Kroah-Hartman

cxlflash: Move to exponential back-off when cmd_room is not available

2016-09-15T06:27:50+00:00

[ Upstream commit ea76543127da32dec28af0a13ea1b06625fc085e ]

While profiling the cxlflash_queuecommand() path under a heavy load it
was found that number of retries to find cmd_room was fairly high.

There are two problems with the current back-off:
a) It starts with a udelay of 0
b) It backs-off linearly

Tried several approaches (a higher multiple 10*n, 100*n, as well as n^2,
2^n) and found that the exponential back-off(2^n) approach had the least
overall cost. Cost as being defined as overall time spent waiting.

The fix is to change the linear back-off to an exponential back-off.
This solution also takes care of the problem with the initial
delay (starts with 1 usec).

Signed-off-by: Manoj N. Kumar 
Acked-by: Matthew R. Ochs 
Reviewed-by: Johannes Thumshirn 
Signed-off-by: Uma Krishnan 
Signed-off-by: Martin K. Petersen 
Signed-off-by: Sasha Levin 
Signed-off-by: Greg Kroah-Hartman

cxlflash: Fix to avoid virtual LUN failover failure

2016-09-15T06:27:49+00:00

[ Upstream commit d5e26bb1d812ba74f29b6bcbc88c3dbfb3eed824 ]

Applications which use virtual LUN's that are backed by a physical LUN
over both adapter ports may experience an I/O failure in the event of a
link loss (e.g. cable pull).

Virtual LUNs may be accessed through one or both ports of the adapter.
This access is encoded in the translation entries that comprise the
virtual LUN and used by the AFU for load-balancing I/O and handling
failover scenarios. In a link loss scenario, even though the AFU is able
to maintain connectivity to the LUN, it is up to the application to
retry the failed I/O. When applications are unaware of the virtual LUN's
underlying topology, they are unable to make a sound decision of when to
retry an I/O and therefore are forced to make their reaction to a failed
I/O absolute. The result is either a failure to retry I/O or increased
latency for scenarios where a retry is pointless.

To remedy this scenario, provide feedback back to the application on
virtual LUN creation as to which ports the LUN may be accessed. LUN's
spanning both ports are candidates for a retry in a presence of an I/O
failure.

Signed-off-by: Matthew R. Ochs 
Acked-by: Manoj Kumar 
Reviewed-by: Uma Krishnan 
Signed-off-by: Martin K. Petersen 
Signed-off-by: Sasha Levin 
Signed-off-by: Greg Kroah-Hartman

cxlflash: Fix to escalate LINK_RESET also on port 1

2016-09-15T06:27:49+00:00

[ Upstream commit a9be294ecb3b9dc82b15625631b153f871181d16 ]

The original fix to escalate a 'login timed out' error to a LINK_RESET
was only made for one of the two ports on the card. This fix resolves
the same issue for the second port (port 1).

Signed-off-by: Manoj N. Kumar 
Acked-by: Matthew R. Ochs 
Reviewed-by: Uma Krishnan 
Signed-off-by: Martin K. Petersen 
Signed-off-by: Sasha Levin 
Signed-off-by: Greg Kroah-Hartman

lpfc: Fix DMA faults observed upon plugging loopback connector

2016-09-15T06:27:48+00:00

[ Upstream commit ae09c765109293b600ba9169aa3d632e1ac1a843 ]

Driver didn't program the REG_VFI mailbox correctly, giving the adapter
bad addresses.

Signed-off-by: Dick Kennedy 
Signed-off-by: James Smart 
Signed-off-by: Martin K. Petersen 
Signed-off-by: Sasha Levin 
Signed-off-by: Greg Kroah-Hartman