openwrt/staging/blogic.git
8 years agoscsi: libfc: sanitize E_D_TOV and R_A_TOV setting
Hannes Reinecke [Thu, 13 Oct 2016 13:10:41 +0000 (15:10 +0200)]
scsi: libfc: sanitize E_D_TOV and R_A_TOV setting

When setting the FCP timeout we need to ensure a lower boundary
for E_D_TOV and R_A_TOV, otherwise we'd be getting spurious I/O
issues due to the fcp timer firing too early.

Signed-off-by: Hannes Reinecke <hare@suse.com>
Acked-by: Johannes Thumshirn <jth@kernel.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
8 years agoscsi: libfc: use configured rport E_D_TOV
Hannes Reinecke [Thu, 13 Oct 2016 13:10:40 +0000 (15:10 +0200)]
scsi: libfc: use configured rport E_D_TOV

If fc_rport_error_retry() is attempting to retry the remote
port state we should be waiting for the configured e_d_tov
value rather than the default.

Signed-off-by: Hannes Reinecke <hare@suse.com>
Acked-by: Johannes Thumshirn <jth@kernel.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
8 years agoscsi: libfc: use configured lport R_A_TOV
Hannes Reinecke [Thu, 13 Oct 2016 13:10:39 +0000 (15:10 +0200)]
scsi: libfc: use configured lport R_A_TOV

We should be using the configured R_A_TOV value when sending the
exchange.

Signed-off-by: Hannes Reinecke <hare@suse.com>
Acked-by: Johannes Thumshirn <jth@kernel.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
8 years agoscsi: libfc: spurious I/O error under high load
Hannes Reinecke [Thu, 13 Oct 2016 13:10:38 +0000 (15:10 +0200)]
scsi: libfc: spurious I/O error under high load

If a command times out libfc is sending an REC, which also
might fail (due to frames being lost or something).
If no data has been transferred we can simply retry
the command, but the current code sets a state of FC_ERROR,
which then is being translated into DID_ERROR, resulting
in an I/O error.
So to handle this properly we need to set a separate
state FC_TRANS_RESET and mapping it onto DID_SOFT_RETRY.

Signed-off-by: Hannes Reinecke <hare@suse.de>
Acked-by: Johannes Thumshirn <jth@kernel.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
8 years agoscsi: libfc: additional debugging messages
Hannes Reinecke [Thu, 13 Oct 2016 13:10:37 +0000 (15:10 +0200)]
scsi: libfc: additional debugging messages

Signed-off-by: Hannes Reinecke <hare@suse.com>
Acked-by: Johannes Thumshirn <jth@kernel.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
8 years agoscsi: libfc: Revert "[SCSI] libfc: use offload EM instance again instead jumping...
Martin K. Petersen [Fri, 14 Oct 2016 20:37:29 +0000 (16:37 -0400)]
scsi: libfc: Revert "[SCSI] libfc: use offload EM instance again instead jumping to next EM"

This reverts commit 3e22760d4db6fd89e0be46c3d132390a251da9c6.

This revert came about because of efforts by Ewan Milne, Curtis Taylor
and I.  In researching this issue, significant performance issues were
seen on large CPU count systems using the software FCOE stack.  Hannes
also weighed in.

The same was not apparent on much smaller low count CPU systems.  The
behavior introduced by commit 3e22760d4db6fd89e0be46c3d132390a251da9c6
lands sup with large count CPU systems seeing continual
blk_requeue_request() calls due to ML_QUEUE_HOST_BUSY.

fc_exch_alloc() used to try all the available exchange managers in the
list for an available exchange id, but this was changed in 2010 so that
if the first matched exchange manager couldn't allocate one, it fails
and we end up returning host busy.  This was due to commit:

Setting the ddp_min module parameter to fcoe to 128MB prevents the
->match function from permitting the use of the offload exchange manager
for the frame, and we no longer see the problem with host busy status,
since it uses the larger non-offloaded pool.

Reverting commit 3e22760d4db6fd89e0be46c3d132390a251da9c6 was tested to
also prevent the host busy issue due to failing allocations.

Suggested-by: Ewan Milne <emilne@redhat.com>
Suggested-by: Curtis Taylor <cjt@us.ibm.com>
Tested-by: Laurence Oberman <loberman@redhat.com>
Signed-off-by: Laurence Oberman <loberman@redhat.com>
8 years agoscsi: lpfc: lpfc version changed to 11.2.0.2
James Smart [Thu, 13 Oct 2016 22:06:18 +0000 (15:06 -0700)]
scsi: lpfc: lpfc version changed to 11.2.0.2

lpfc version changed to 11.2.0.2

Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <james.smart@broadcom.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
8 years agoscsi: lpfc: Fix fw download on SLI-4 FC adapters
James Smart [Thu, 13 Oct 2016 22:06:17 +0000 (15:06 -0700)]
scsi: lpfc: Fix fw download on SLI-4 FC adapters

Fix fw download on SLI-4 FC adapters

Driver performs a quick validation of magic numbers in the fw
download image. Driver needed to be updated for more recent
magic numbers.

Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <james.smart@broadcom.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
8 years agoscsi: lpfc: Synchronize link speed with boot driver
James Smart [Thu, 13 Oct 2016 22:06:16 +0000 (15:06 -0700)]
scsi: lpfc: Synchronize link speed with boot driver

Synchronize link speed with boot driver

Link speed settings set by the boot driver are reported by the hw.
Driver will attempt to read them, and if set, will respect their
values.
The driver can override the settings with its own if instructed by
user space (via bsg), with the new values being picked up by the
boot driver.

Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <james.smart@broadcom.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
8 years agoscsi: lpfc: Correct panics with eh_timeout and eh_deadline
James Smart [Thu, 13 Oct 2016 22:06:15 +0000 (15:06 -0700)]
scsi: lpfc: Correct panics with eh_timeout and eh_deadline

Correct panics with eh_timeout and eh_deadline

We were having double completions on our SLI-3 version of adapters.
Solved by clearing our command pointer before calling scsi_done.

The eh paths potentially ran simulatenously and would see the non-null
value and invoke scsi_done again.

Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <james.smart@broadcom.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
8 years agoscsi: lpfc: Fix lost target in pt-to-pt connect
James Smart [Thu, 13 Oct 2016 22:06:14 +0000 (15:06 -0700)]
scsi: lpfc: Fix lost target in pt-to-pt connect

Fix lost target in pt-to-pt connect

Change reject code to something that allows a retry

Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <james.smart@broadcom.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
8 years agoscsi: lpfc: Revise strings with full lpfc parameter name
James Smart [Thu, 13 Oct 2016 22:06:13 +0000 (15:06 -0700)]
scsi: lpfc: Revise strings with full lpfc parameter name

Revise strings with full lpfc parameter name

Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <james.smart@broadcom.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
8 years agoscsi: lpfc: Code cleanup for lpfc_sriov_nr_virtfn parameter
James Smart [Thu, 13 Oct 2016 22:06:12 +0000 (15:06 -0700)]
scsi: lpfc: Code cleanup for lpfc_sriov_nr_virtfn parameter

Code cleanup for lpfc_sriov_nr_virtfn parameter

Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <james.smart@broadcom.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
8 years agoscsi: lpfc: Code cleanup for lpfc_max_scsicmpl_time parameter
James Smart [Thu, 13 Oct 2016 22:06:11 +0000 (15:06 -0700)]
scsi: lpfc: Code cleanup for lpfc_max_scsicmpl_time parameter

Code cleanup for lpfc_max_scsicmpl_time parameter

Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <james.smart@broadcom.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
8 years agoscsi: lpfc: Code cleanup for lpfc_topology parameter
James Smart [Thu, 13 Oct 2016 22:06:10 +0000 (15:06 -0700)]
scsi: lpfc: Code cleanup for lpfc_topology parameter

Code cleanup for lpfc_topology parameter

Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <james.smart@broadcom.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
8 years agoscsi: lpfc: Code cleanup for lpfc_aer_support parameter
James Smart [Thu, 13 Oct 2016 22:06:09 +0000 (15:06 -0700)]
scsi: lpfc: Code cleanup for lpfc_aer_support parameter

Code cleanup for lpfc_aer_support parameter

Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <james.smart@broadcom.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
8 years agoscsi: lpfc: Code cleanup for lpfc_enable_rrq parameter
James Smart [Thu, 13 Oct 2016 22:06:08 +0000 (15:06 -0700)]
scsi: lpfc: Code cleanup for lpfc_enable_rrq parameter

Code cleanup for lpfc_enable_rrq parameter

Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <james.smart@broadcom.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
8 years agoscsi: lpfc: Code clean up for lpfc_iocb_cnt parameter
James Smart [Thu, 13 Oct 2016 22:06:07 +0000 (15:06 -0700)]
scsi: lpfc: Code clean up for lpfc_iocb_cnt parameter

Code clean up for lpfc_iocb_cnt parameter

Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <james.smart@broadcom.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
8 years agoscsi: lpfc: Make lpfc_prot_xxx params per hba parameters
James Smart [Thu, 13 Oct 2016 22:06:06 +0000 (15:06 -0700)]
scsi: lpfc: Make lpfc_prot_xxx params per hba parameters

Make lpfc_prot_mask and lpfc_prot_guard per hba parameters

Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <james.smart@broadcom.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
8 years agoscsi: lpfc: Set driver environment data on adapter
James Smart [Thu, 13 Oct 2016 22:06:05 +0000 (15:06 -0700)]
scsi: lpfc: Set driver environment data on adapter

Set driver environment data on adapter

Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <james.smart@broadcom.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
8 years agoscsi: lpfc: Fix sg_reset on SCSI device causing kernel crash
James Smart [Thu, 13 Oct 2016 22:06:04 +0000 (15:06 -0700)]
scsi: lpfc: Fix sg_reset on SCSI device causing kernel crash

Fix sg_reset on SCSI device causing kernel crash

Driver could reference stale node pointers in task mgmt call.
Changed to use resetting cmd and look up node pointer in task mgmt
function.

Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <james.smart@broadcom.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
8 years agoscsi: lpfc: Correct embedded io wq element size
James Smart [Thu, 13 Oct 2016 22:06:03 +0000 (15:06 -0700)]
scsi: lpfc: Correct embedded io wq element size

Correct embedded io wq element size. Embedded element sizes are
128 byte elements

Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <james.smart@broadcom.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
8 years agoscsi: lpfc: Fix few small typos in lpfc_scsi.c
Milan P. Gandhi [Thu, 13 Oct 2016 22:06:02 +0000 (15:06 -0700)]
scsi: lpfc: Fix few small typos in lpfc_scsi.c

This patch does a cleanup and fixes few small typos in lpfc_scsi.c

Signed-off-by: Milan P. Gandhi <mgandhi@redhat.com>
Signed-off-by: James Smart <james.smart@avagotech.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
8 years agoMAINTAINERS: Changing maintainer for ufs DWC.
Joao Pinto [Fri, 14 Oct 2016 15:59:23 +0000 (16:59 +0100)]
MAINTAINERS: Changing maintainer for ufs DWC.

I am going to leave Synopsys and so this patch changes the Maintainer
for UFS Synopsys' specific drivers to my colleagues Manjunath and Prabu.

Signed-off-by: Joao Pinto <jpinto@synopsys.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
8 years agoMAINTAINERS: Remove defunct iss storage mailing list
Javier Martinez Canillas [Thu, 13 Oct 2016 16:57:41 +0000 (13:57 -0300)]
MAINTAINERS: Remove defunct iss storage mailing list

It appears that the mailing list email address doesn't exist anymore:

<iss_storagedev@hp.com>: host smtp.hp.com[15.73.96.116] said: 550 5.1.1
    <iss_storagedev@hp.com>: Recipient address rejected: User unknown in
    virtual alias table (in reply to RCPT TO command)

Signed-off-by: Javier Martinez Canillas <javier@osg.samsung.com>
Acked-by: Don Brace <don.brace@microsemi.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
8 years agoscsi: hpsa: Remove unneeded void pointer cast
Javier Martinez Canillas [Thu, 13 Oct 2016 16:10:08 +0000 (13:10 -0300)]
scsi: hpsa: Remove unneeded void pointer cast

It's not necessary to cast the result of kmalloc, since void pointers
are promoted to any other type. This also fixes following coccinelle
warning:

casting value returned by memory allocation function to (BIG_IOCTL_Command_struct *) is useless.

Signed-off-by: Javier Martinez Canillas <javier@osg.samsung.com>
Acked-by: Don Brace <don.brace@microsemi.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
8 years agoscsi: mptscsih: Remove bogus interpretation of request->ioprio
Christoph Hellwig [Thu, 13 Oct 2016 08:46:41 +0000 (10:46 +0200)]
scsi: mptscsih: Remove bogus interpretation of request->ioprio

Having an I/O priority does not mean we should send all requests as HEAD
OF QUEUE tags.

Reported-by: Adam Manzanares <adam.manzanares@wdc.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Hannes Reinicke <hare@suse.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
8 years agoscsi: ncr5380: Suppress unhelpful "interrupt without IRQ bit" message
Finn Thain [Mon, 10 Oct 2016 04:46:53 +0000 (00:46 -0400)]
scsi: ncr5380: Suppress unhelpful "interrupt without IRQ bit" message

If a NCR5380 host instance ends up on a shared interrupt line then
this printk will be a problem. It is already a problem on some Mac
models: when testing mac_scsi on a PowerBook 180 I found that PDMA
transfers (but not PIO transfers) cause the message to be logged.

These spurious interrupts don't appear to come from the DRQ signal from
the 5380. And they don't happen at all on the Mac LC III. A comment in
the NetBSD source code mentions this mystery. Testing seems to show
that we can safely ignore these interrupts.

Signed-off-by: Finn Thain <fthain@telegraphics.com.au>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Tested-by: Ondrej Zary <linux@rainbow-software.org>
Tested-by: Michael Schmitz <schmitzmic@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
8 years agoscsi: ncr5380: Use correct types for DMA routines
Finn Thain [Mon, 10 Oct 2016 04:46:53 +0000 (00:46 -0400)]
scsi: ncr5380: Use correct types for DMA routines

Apply prototypes to get consistent function signatures for the DMA
functions implemented in the board-specific drivers. To avoid using
macros to alter actual parameters, some of those functions are reworked
slightly.

This is a step toward the goal of passing the board-specific routines
to the core driver using an ops struct (as in a platform driver or
library module).

This also helps fix some inconsistent types: where the core driver uses
ints (cmd->SCp.this_residual and hostdata->dma_len) for keeping track of
transfers, certain board-specific routines used unsigned long.

While we are fixing these function signatures, pass the hostdata pointer
to DMA routines instead of a Scsi_Host pointer, for shorter and faster
code.

Signed-off-by: Finn Thain <fthain@telegraphics.com.au>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Tested-by: Ondrej Zary <linux@rainbow-software.org>
Tested-by: Michael Schmitz <schmitzmic@gmail.com>
Acked-by: Russell King <rmk+kernel@armlinux.org.uk>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
8 years agoscsi: ncr5380: Expedite register polling
Finn Thain [Mon, 10 Oct 2016 04:46:53 +0000 (00:46 -0400)]
scsi: ncr5380: Expedite register polling

Avoid the call to NCR5380_poll_politely2() when possible. The call is
easily short-circuited on the PIO fast path, using the inline wrapper.
This requires that the NCR5380_read macro be made available before
any #include "NCR5380.h" so a few declarations have to be moved too.

Signed-off-by: Finn Thain <fthain@telegraphics.com.au>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Tested-by: Ondrej Zary <linux@rainbow-software.org>
Tested-by: Michael Schmitz <schmitzmic@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
8 years agoscsi: ncr5380: Pass hostdata pointer to register polling routines
Finn Thain [Mon, 10 Oct 2016 04:46:53 +0000 (00:46 -0400)]
scsi: ncr5380: Pass hostdata pointer to register polling routines

Pass a NCR5380_hostdata struct pointer to the board-specific routines
instead of a Scsi_Host struct pointer. This reduces pointer chasing in
the PIO and PDMA fast paths. The old way was a mistake because it is
slow and the board-specific code is not concerned with the mid-layer.

Signed-off-by: Finn Thain <fthain@telegraphics.com.au>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Tested-by: Ondrej Zary <linux@rainbow-software.org>
Tested-by: Michael Schmitz <schmitzmic@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
8 years agoscsi: ncr5380: Use correct types for device register accessors
Finn Thain [Mon, 10 Oct 2016 04:46:53 +0000 (00:46 -0400)]
scsi: ncr5380: Use correct types for device register accessors

For timeout values adopt unsigned long, which is the type of jiffies etc.

For chip register values and bit masks pass u8, which is the return type
of readb, inb etc.

For device register offsets adopt unsigned int, as it is suitable for
adding to base addresses.

Pass the NCR5380_hostdata pointer to the board-specific routines instead
of the Scsi_Host pointer. The board-specific code is concerned with
hardware and not with SCSI protocol or the mid-layer.

Signed-off-by: Finn Thain <fthain@telegraphics.com.au>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Tested-by: Ondrej Zary <linux@rainbow-software.org>
Tested-by: Michael Schmitz <schmitzmic@gmail.com>
Acked-by: Russell King <rmk+kernel@armlinux.org.uk>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
8 years agoscsi: ncr5380: Store IO ports and addresses in host private data
Finn Thain [Mon, 10 Oct 2016 04:46:53 +0000 (00:46 -0400)]
scsi: ncr5380: Store IO ports and addresses in host private data

The various 5380 drivers inconsistently store register pointers
either in the Scsi_Host struct "legacy crap" area or in special,
board-specific members of the NCR5380_hostdata struct. Uniform
use of the latter struct makes for simpler and faster code (see
the following patches) and helps to reduce use of the
NCR5380_implementation_fields macro.

Signed-off-by: Finn Thain <fthain@telegraphics.com.au>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Tested-by: Ondrej Zary <linux@rainbow-software.org>
Tested-by: Michael Schmitz <schmitzmic@gmail.com>
Acked-by: Russell King <rmk+kernel@armlinux.org.uk>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
8 years agoscsi: ncr5380: Improve hostdata struct member alignment and cache-ability
Finn Thain [Mon, 10 Oct 2016 04:46:53 +0000 (00:46 -0400)]
scsi: ncr5380: Improve hostdata struct member alignment and cache-ability

Re-order struct members so that hot data lies at the beginning of the
struct and cold data at the end. Improve the comments while we're here.

Signed-off-by: Finn Thain <fthain@telegraphics.com.au>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Tested-by: Ondrej Zary <linux@rainbow-software.org>
Tested-by: Michael Schmitz <schmitzmic@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
8 years agoscsi: ncr5380: Increase register polling limit
Finn Thain [Mon, 10 Oct 2016 04:46:52 +0000 (00:46 -0400)]
scsi: ncr5380: Increase register polling limit

If NCR5380_poll_politely() is called under irq lock, the polling time
limit is clamped to avoid a spike in interrupt latency. When not under
irq lock, the same polling time limit acts as the worst case delay
between schedule() calls.

During PDMA (under irq lock) I've found that the 10 ms time limit is
sometimes too short, and leads to the error message,
sd 0:0:0:0: [sda] tag#1 macscsi_pread: !REQ and !ACK

This particular target identifies itself as a QUANTUM DAYTONA514S. It
seems to be slower to assert ACK than the other targets I've tested.
This patch solves the problem by increasing the polling timeout.

Signed-off-by: Finn Thain <fthain@telegraphics.com.au>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Tested-by: Ondrej Zary <linux@rainbow-software.org>
Tested-by: Michael Schmitz <schmitzmic@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
8 years agoscsi: ncr5380: Simplify register polling limit
Finn Thain [Mon, 10 Oct 2016 04:46:52 +0000 (00:46 -0400)]
scsi: ncr5380: Simplify register polling limit

When polling a device register under irq lock the polling loop terminates
after a given number of jiffies. Make this timeout independent of the HZ
setting.

All 5380 drivers benefit from this patch, which optimizes the PIO fast
path, because they all use PIO transfers (for phases other than DATA IN
and DATA OUT). Some cards support only PIO transfers (even for DATA
phases). CPU cycles are scarce on some of these systems, so a small
improvement here makes a big difference.

Signed-off-by: Finn Thain <fthain@telegraphics.com.au>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Tested-by: Ondrej Zary <linux@rainbow-software.org>
Tested-by: Michael Schmitz <schmitzmic@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
8 years agoscsi: atari_scsi: Make device register accessors re-entrant
Finn Thain [Mon, 10 Oct 2016 04:46:52 +0000 (00:46 -0400)]
scsi: atari_scsi: Make device register accessors re-entrant

This patch fixes an old bug: accesses to device registers from the
interrupt handler (after reselection, DMA completion etc.) could mess
up a device register access elsewhere, if the latter takes place outside
of an irq lock (during selection etc.).

Signed-off-by: Finn Thain <fthain@telegraphics.com.au>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Tested-by: Michael Schmitz <schmitzmic@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
8 years agoscsi: cumana_1: Remove unused cumanascsi_setup() function
Finn Thain [Mon, 10 Oct 2016 04:46:52 +0000 (00:46 -0400)]
scsi: cumana_1: Remove unused cumanascsi_setup() function

Signed-off-by: Finn Thain <fthain@telegraphics.com.au>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Acked-by: Russell King <rmk+kernel@armlinux.org.uk>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
8 years agoscsi: g_NCR5380: Merge g_NCR5380 and g_NCR5380_mmio drivers
Ondrej Zary [Mon, 10 Oct 2016 04:46:52 +0000 (00:46 -0400)]
scsi: g_NCR5380: Merge g_NCR5380 and g_NCR5380_mmio drivers

Merge the port-mapped IO and memory-mapped IO support (with the help of
ioport_map) into the g_NCR5380 module and delete g_NCR5380_mmio.

Signed-off-by: Ondrej Zary <linux@rainbow-software.org>
Signed-off-by: Finn Thain <fthain@telegraphics.com.au>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Tested-by: Ondrej Zary <linux@rainbow-software.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
8 years agoscsi: ufshcd: Fix possible unclocked register access
Subhash Jadavani [Fri, 7 Oct 2016 04:48:22 +0000 (21:48 -0700)]
scsi: ufshcd: Fix possible unclocked register access

Vendor specific setup_clocks callback may require the clocks managed by
ufshcd driver to be ON. So if the vendor specific setup_clocks callback
is called while the required clocks are turned off, it could result into
unclocked register access.

To prevent possible unclock register access, this change adds one more
argument to setup_clocks callback to let it know whether it is called
pre/post the clock changes by core driver.

Signed-off-by: Subhash Jadavani <subhashj@codeaurora.org>
Reviewed-by: Kiwoong Kim <kwmad.kim@samsung.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
8 years agoscsi: fcoe: Harden CVL handling when we have not logged into the fabric.
Chad Dupuis [Fri, 30 Sep 2016 09:01:20 +0000 (11:01 +0200)]
scsi: fcoe: Harden CVL handling when we have not logged into the fabric.

If we haven't logged into the fabric yet we want to be a little more nuanced
with our CVL handling than what we've been:

- If the FCF has been selected, check the source MAC to make sure the frame is
from the FCF we've selected.
- If a FCF is selected and the CVL is from the FCF but we have not logged in
yet, then reset everything and go back to solicitation.

Signed-off-by: Chad Dupuis <chad.dupuis@cavium.com>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Acked-by: Johannes Thumshirn <jth@kernel.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
8 years agoscsi: libfc: don't advance state machine for incoming FLOGI
Hannes Reinecke [Fri, 30 Sep 2016 09:01:19 +0000 (11:01 +0200)]
scsi: libfc: don't advance state machine for incoming FLOGI

When we receive an FLOGI but have already sent our own we should
not advance the state machine but rather wait for our FLOGI to
return before continuing with PLOGI.

Signed-off-by: Hannes Reinecke <hare@suse.com>
Acked-by: Johannes Thumshirn <jth@kernel.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
8 years agoscsi: libfc: Do not login if the port is already started
Hannes Reinecke [Fri, 30 Sep 2016 09:01:18 +0000 (11:01 +0200)]
scsi: libfc: Do not login if the port is already started

When the port is already started we don't need to login; that
will only confuse the state machine.

Signed-off-by: Hannes Reinecke <hare@suse.com>
Acked-by: Johannes Thumshirn <jth@kernel.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
8 years agoscsi: libfc: Do not drop down to FLOGI for fc_rport_login()
Hannes Reinecke [Fri, 30 Sep 2016 09:01:17 +0000 (11:01 +0200)]
scsi: libfc: Do not drop down to FLOGI for fc_rport_login()

When fc_rport_login() is called while the rport is not
in RPORT_ST_INIT, RPORT_ST_READY, or RPORT_ST_DELETE
login is already in progress and there's no need to
drop down to FLOGI; doing so will only confuse the
other side.

Signed-off-by: Hannes Reinecke <hare@suse.com>
Acked-by: Johannes Thumshirn <jth@kernel.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
8 years agoscsi: libfc: Do not take rdata->rp_mutex when processing a -FC_EX_CLOSED ELS response.
Chad Dupuis [Fri, 30 Sep 2016 09:01:16 +0000 (11:01 +0200)]
scsi: libfc: Do not take rdata->rp_mutex when processing a -FC_EX_CLOSED ELS response.

When an ELS response handler receives a -FC_EX_CLOSED, the rdata->rp_mutex is
already held which can lead to a deadlock condition like the following stack trace:

[<ffffffffa04d8f18>] fc_rport_plogi_resp+0x28/0x200 [libfc]
[<ffffffffa04cfa1a>] fc_invoke_resp+0x6a/0xe0 [libfc]
[<ffffffffa04d0c08>] fc_exch_mgr_reset+0x1b8/0x280 [libfc]
[<ffffffffa04d87b3>] fc_rport_logoff+0x43/0xd0 [libfc]
[<ffffffffa04ce73d>] fc_disc_stop+0x6d/0xf0 [libfc]
[<ffffffffa04ce7ce>] fc_disc_stop_final+0xe/0x20 [libfc]
[<ffffffffa04d55f7>] fc_fabric_logoff+0x17/0x70 [libfc]

The other ELS handlers need to follow the FLOGI response handler and simply do
a kref_put against the fc_rport_priv struct and exit when receving a
-FC_EX_CLOSED response.

Signed-off-by: Chad Dupuis <chad.dupuis@cavium.com>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Acked-by: Johannes Thumshirn <jth@kernel.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
8 years agoscsi: libfc: Fixup disc_mutex handling
Hannes Reinecke [Fri, 30 Sep 2016 09:01:15 +0000 (11:01 +0200)]
scsi: libfc: Fixup disc_mutex handling

The list of attached 'rdata' remote port structures is RCU
protected, so there is no need to take the 'disc_mutex' when
traversing it.
Rather we should be using rcu_read_lock() and kref_get_unless_zero()
to validate the entries.
We need, however, take the disc_mutex when deleting an entry;
otherwise we risk clashes with list_add.

Signed-off-by: Hannes Reinecke <hare@suse.com>
Acked-by: Johannes Thumshirn <jth@kernel.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
8 years agoscsi: libfc: Revisit kref handling
Hannes Reinecke [Fri, 30 Sep 2016 09:01:14 +0000 (11:01 +0200)]
scsi: libfc: Revisit kref handling

The kref handling in fc_rport is a mess. This patch updates
the kref handling according to the following rules:

- Take a reference whenever scheduling a workqueue
- Take a reference whenever an ELS command is send
- Drop the reference at the end of the workqueue function
- Drop the reference at the end of handling ELS replies
- Take a reference when allocating an rport
- Drop the reference when removing an rport

Signed-off-by: Hannes Reinecke <hare@suse.com>
Acked-by: Johannes Thumshirn <jth@kernel.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
8 years agoscsi: hisi_sas: Add v2 hw support for different refclk
John Garry [Tue, 4 Oct 2016 11:11:11 +0000 (19:11 +0800)]
scsi: hisi_sas: Add v2 hw support for different refclk

The hip06 D03 and hip07 D05 boards have different reference clock
frequencies for the SAS controller.

Register PHY_CTRL needs to be programmed differently according to this
frequency, so add support for this.

The default register setting in PHY_CTRL is for 50MHz, so only update
this register when the refclk frequency is 66MHz.

For ACPI we expect the _RST handler to set the correct value for
PHY_CTRL (we're forced to take different approach for DT and ACPI as
ACPI does not support fixed-clock device).

Signed-off-by: John Garry <john.garry@huawei.com>
Signed-off-by: Xiang Chen <chenxiang66@hisilicon.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
8 years agoscsi: hisi_sas: Add device tree support for hip07
John Garry [Tue, 4 Oct 2016 11:11:10 +0000 (19:11 +0800)]
scsi: hisi_sas: Add device tree support for hip07

Chipset hip07 incorporates v2 hw.

Signed-off-by: John Garry <john.garry@huawei.com>
Signed-off-by: Xiang Chen <chenxiang66@hisilicon.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
8 years agoscsi: devicetree: bindings: hisi_sas: add hip07 support
John Garry [Tue, 4 Oct 2016 11:11:09 +0000 (19:11 +0800)]
scsi: devicetree: bindings: hisi_sas: add hip07 support

Add support for hip07 chipset to hisi_sas controller.

Chipset hip07 has v2 hw.

Signed-off-by: John Garry <john.garry@huawei.com>
Signed-off-by: Xiang Chen <chenxiang66@hisilicon.com>
Acked-by: Rob Herring <robh@kernel.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
8 years agoscsi: fnic: Use time64_t to represent trace timestamps
Deepa Dinamani [Sat, 1 Oct 2016 23:48:14 +0000 (16:48 -0700)]
scsi: fnic: Use time64_t to represent trace timestamps

Trace timestamps use struct timespec and CURRENT_TIME which are not
y2038 safe.  These timestamps are only part of the trace log on the
machine and are not shared with the fnic.  Replace then with y2038 safe
struct timespec64 and ktime_get_real_ts64(), respectively.

Signed-off-by: Deepa Dinamani <deepa.kernel@gmail.com>
Reviewed-by: Arnd Bergmann <arnd@arndb.de>
Cc: Hiral Patel <hiralpat@cisco.com>
Cc: Suma Ramars <sramars@cisco.com>
Cc: Brian Uchino <buchino@cisco.com>
Cc: "James E.J. Bottomley" <jejb@linux.vnet.ibm.com>
Cc: "Martin K. Petersen" <martin.petersen@oracle.com>
Cc: linux-scsi@vger.kernel.org
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
8 years agoscsi: ipr: Use pci_irq_allocate_vectors
Christoph Hellwig [Sun, 11 Sep 2016 13:31:24 +0000 (15:31 +0200)]
scsi: ipr: Use pci_irq_allocate_vectors

Switch the ipr driver to use pci_alloc_irq_vectors.  We need to two
calls to pci_alloc_irq_vectors as ipr only supports multiple MSI-X
vectors, but not multiple MSI vectors.

Otherwise this cleans up a lot of cruft and allows to use a common
request_irq loop for irq types, which happens to only iterate over a
single line in the non MSI-X case.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Acked-by: Brian King <brking@linux.vnet.ibm.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
8 years agoscsi: arcmsr: Use pci_alloc_irq_vectors
Christoph Hellwig [Sun, 11 Sep 2016 13:31:23 +0000 (15:31 +0200)]
scsi: arcmsr: Use pci_alloc_irq_vectors

Switch the arcmsr driver to use pci_alloc_irq_vectors.  We need to two
calls to pci_alloc_irq_vectors as arcmsr only supports multiple MSI-X
vectors, but not multiple MSI vectors.

Otherwise this cleans up a lot of cruft and allows to use a common
request_irq loop for irq types, which happens to only iterate over a
single line in the non MSI-X case.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Acked-by: Ching Huang <ching2048@areca.com.tw>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
8 years agoMerge tag 'iommu-fixes-v4.9-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Tue, 8 Nov 2016 18:07:13 +0000 (10:07 -0800)]
Merge tag 'iommu-fixes-v4.9-rc4' of git://git./linux/kernel/git/joro/iommu

Pull IOMMU fixes from Joerg Roedel:

 - Four patches from Robin Murphy fix several issues with the recently
   merged generic DT-bindings support for arm-smmu drivers

 - A fix for a dead-lock issue in the VT-d driver, which shows up on
   iommu hotplug

* tag 'iommu-fixes-v4.9-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/joro/iommu:
  iommu/vt-d: Fix dead-locks in disable_dmar_iommu() path
  iommu/arm-smmu: Fix out-of-bounds dereference
  iommu/arm-smmu: Check that iommu_fwspecs are ours
  iommu/arm-smmu: Don't inadvertently reject multiple SMMUv3s
  iommu/arm-smmu: Work around ARM DMA configuration

8 years agoiommu/vt-d: Fix dead-locks in disable_dmar_iommu() path
Joerg Roedel [Tue, 8 Nov 2016 14:08:26 +0000 (15:08 +0100)]
iommu/vt-d: Fix dead-locks in disable_dmar_iommu() path

It turns out that the disable_dmar_iommu() code-path tried
to get the device_domain_lock recursivly, which will
dead-lock when this code runs on dmar removal. Fix both
code-paths that could lead to the dead-lock.

Fixes: 55d940430ab9 ('iommu/vt-d: Get rid of domain->iommu_lock')
Signed-off-by: Joerg Roedel <jroedel@suse.de>
8 years agoiommu/arm-smmu: Fix out-of-bounds dereference
Robin Murphy [Mon, 7 Nov 2016 18:25:09 +0000 (18:25 +0000)]
iommu/arm-smmu: Fix out-of-bounds dereference

When we iterate a master's config entries, what we generally care
about is the entry's stream map index, rather than the entry index
itself, so it's nice to have the iterator automatically assign the
former from the latter. Unfortunately, booting with KASAN reveals
the oversight that using a simple comma operator results in the
entry index being dereferenced before being checked for validity,
so we always access one element past the end of the fwspec array.

Flip things around so that the check always happens before the index
may be dereferenced.

Fixes: adfec2e709d2 ("iommu/arm-smmu: Convert to iommu_fwspec")
Reported-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Robin Murphy <robin.murphy@arm.com>
Acked-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Joerg Roedel <jroedel@suse.de>
8 years agoiommu/arm-smmu: Check that iommu_fwspecs are ours
Robin Murphy [Wed, 2 Nov 2016 17:31:32 +0000 (17:31 +0000)]
iommu/arm-smmu: Check that iommu_fwspecs are ours

We seem to have forgotten to check that iommu_fwspecs actually belong to
us before we go ahead and dereference their private data. Oops.

Fixes: 021bb8420d44 ("iommu/arm-smmu: Wire up generic configuration support")
Signed-off-by: Robin Murphy <robin.murphy@arm.com>
Signed-off-by: Joerg Roedel <jroedel@suse.de>
8 years agoiommu/arm-smmu: Don't inadvertently reject multiple SMMUv3s
Robin Murphy [Thu, 3 Nov 2016 17:39:07 +0000 (17:39 +0000)]
iommu/arm-smmu: Don't inadvertently reject multiple SMMUv3s

We now delay installing our per-bus iommu_ops until we know an SMMU has
successfully probed, as they don't serve much purpose beforehand, and
doing so also avoids fights between multiple IOMMU drivers in a single
kernel. However, the upshot of passing the return value of bus_set_iommu()
back from our probe function is that if there happens to be more than
one SMMUv3 device in a system, the second and subsequent probes will
wind up returning -EBUSY to the driver core and getting torn down again.

Avoid re-setting ops if ours are already installed, so that any genuine
failures stand out.

Fixes: 08d4ca2a672b ("iommu/arm-smmu: Support non-PCI devices with SMMUv3")
CC: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>
CC: Hanjun Guo <hanjun.guo@linaro.org>
Signed-off-by: Robin Murphy <robin.murphy@arm.com>
Signed-off-by: Joerg Roedel <jroedel@suse.de>
8 years agoiommu/arm-smmu: Work around ARM DMA configuration
Robin Murphy [Mon, 17 Oct 2016 11:06:21 +0000 (12:06 +0100)]
iommu/arm-smmu: Work around ARM DMA configuration

The 32-bit ARM DMA configuration code predates the IOMMU core's default
domain functionality, and instead relies on allocating its own domains
and attaching any devices using the generic IOMMU binding to them.
Unfortunately, it does this relatively early on in the creation of the
device, before we've seen our add_device callback, which leads us to
attempt to operate on a half-configured master.

To avoid a crash, check for this situation on attach, but refuse to
play, as there's nothing we can do. This at least allows VFIO to keep
working for people who update their 32-bit DTs to the generic binding,
albeit with a few (innocuous) warnings from the DMA layer on boot.

Signed-off-by: Robin Murphy <robin.murphy@arm.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Joerg Roedel <jroedel@suse.de>
8 years agoMerge tag 'arm64-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux
Linus Torvalds [Mon, 7 Nov 2016 18:16:23 +0000 (10:16 -0800)]
Merge tag 'arm64-fixes' of git://git./linux/kernel/git/arm64/linux

Pull arm64 fix from Will Deacon:
 "It's been pretty quiet on the fixes side of things for us, but Artem
  reported a build failure introduced during the merge window that
  appears with older GCCs that do not support asm goto. The fix is
  bigger than I'd like, but it's a mechnical move of some constants to
  break an include dependency between atomic.h and jump_label.h when
  !HAVE_JUMP_LABEL.

  Summary:

   - Fix build failure on compilers without asm goto"

* tag 'arm64-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux:
  arm64: Fix circular include of asm/lse.h through linux/jump_label.h

8 years agoMerge tag 'openrisc-for-linus-v4.9-rc5' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Mon, 7 Nov 2016 18:14:47 +0000 (10:14 -0800)]
Merge tag 'openrisc-for-linus-v4.9-rc5' of git://git./linux/kernel/git/groeck/linux-staging

Pull openrisc fix from Guenter Roeck:
 "Fix openrisc crash caused by ro_init changes"

* tag 'openrisc-for-linus-v4.9-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/groeck/linux-staging:
  openrisc: Define __ro_after_init to avoid crash

8 years agoMerge tag 'hwmon-for-linus-v4.9-rc5' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Mon, 7 Nov 2016 18:13:10 +0000 (10:13 -0800)]
Merge tag 'hwmon-for-linus-v4.9-rc5' of git://git./linux/kernel/git/groeck/linux-staging

Pull hwmon fix from Guenter Roeck:
 "Fix resource leak on devm_kcalloc failure"

* tag 'hwmon-for-linus-v4.9-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/groeck/linux-staging:
  hwmon: (core) fix resource leak on devm_kcalloc failure

8 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jikos/hid
Linus Torvalds [Mon, 7 Nov 2016 18:05:39 +0000 (10:05 -0800)]
Merge branch 'for-linus' of git://git./linux/kernel/git/jikos/hid

Pull HID fixes from Jiri Kosina:

 - modprobe-after-rmmod load failure bugfix for intel-ish, from Even Xu

 - IRQ probing bugfix for intel-ish, from Srinivas Pandruvada

 - attribute parsing fix in hid-sensor, from Ooi, Joyce

 - other small misc fixes / quirky device additions

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jikos/hid:
  HID: sensor: fix attributes in HID sensor interface
  HID: intel-ish-hid: request_irq failure
  HID: intel-ish-hid: Fix driver reinit failure
  HID: intel-ish-hid: Move DMA disable code to new function
  HID: intel-ish-hid: consolidate ish wake up operation
  HID: usbhid: add ATEN CS962 to list of quirky devices
  HID: intel-ish-hid: Fix !CONFIG_PM build warning
  HID: sensor-hub: Fix packing of result buffer for feature report

8 years agoopenrisc: Define __ro_after_init to avoid crash
Guenter Roeck [Sat, 24 Sep 2016 14:15:02 +0000 (07:15 -0700)]
openrisc: Define __ro_after_init to avoid crash

openrisc qemu tests fail with the following crash.

Unable to handle kernel access at virtual address 0xc0300c34

Oops#: 0001
CPU #: 0
   PC: c016c710    SR: 0000ae67    SP: c1017e04
   GPR00: 00000000 GPR01: c1017e04 GPR02: c0300c34 GPR03: c0300c34
   GPR04: 00000000 GPR05: c0300cb0 GPR06: c0300c34 GPR07: 000000ff
   GPR08: c107f074 GPR09: c0199ef4 GPR10: c1016000 GPR11: 00000000
   GPR12: 00000000 GPR13: c107f044 GPR14: c0473774 GPR15: 07ce0000
   GPR16: 00000000 GPR17: c107ed8a GPR18: 00009600 GPR19: c107f044
   GPR20: c107ee74 GPR21: 00000003 GPR22: c0473770 GPR23: 00000033
   GPR24: 000000bf GPR25: 00000019 GPR26: c046400c GPR27: 00000001
   GPR28: c0464028 GPR29: c1018000 GPR30: 00000006 GPR31: ccf37483
     RES: 00000000 oGPR11: ffffffff
     Process swapper (pid: 1, stackpage=c1001960)

     Stack: Stack dump [0xc1017cf8]:
     sp + 00: 0xc1017e04
     sp + 04: 0xc0300c34
     sp + 08: 0xc0300c34
     sp + 12: 0x00000000
...

Bisect points to commit d2ec3f77de8e ("pty: make ptmx file ops read-only
after init"). Fix by defining __ro_after_init for the openrisc
architecture, similar to parisc.

Fixes: d2ec3f77de8e ("pty: make ptmx file ops read-only after init")
Cc: Kees Cook <keescook@chromium.org>
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Acked-by: Stafford Horne <shorne@gmail.com>
8 years agoLinux 4.9-rc4
Linus Torvalds [Sat, 5 Nov 2016 23:23:36 +0000 (16:23 -0700)]
Linux 4.9-rc4

8 years agoMerge branch 'i2c/for-current' of git://git.kernel.org/pub/scm/linux/kernel/git/wsa...
Linus Torvalds [Sat, 5 Nov 2016 22:30:12 +0000 (15:30 -0700)]
Merge branch 'i2c/for-current' of git://git./linux/kernel/git/wsa/linux

Pull i2c fix from Wolfram Sang:
 "A bugfix for the I2C core fixing a (rare) race condition"

* 'i2c/for-current' of git://git.kernel.org/pub/scm/linux/kernel/git/wsa/linux:
  i2c: core: fix NULL pointer dereference under race condition

8 years agoarm64: Fix circular include of asm/lse.h through linux/jump_label.h
Catalin Marinas [Thu, 3 Nov 2016 18:34:34 +0000 (18:34 +0000)]
arm64: Fix circular include of asm/lse.h through linux/jump_label.h

Commit efd9e03facd0 ("arm64: Use static keys for CPU features")
introduced support for static keys in asm/cpufeature.h, including
linux/jump_label.h. When CC_HAVE_ASM_GOTO is not defined, this causes a
circular dependency via linux/atomic.h, asm/lse.h and asm/cpufeature.h.

This patch moves the capability macros out out of asm/cpufeature.h into
a separate asm/cpucaps.h and modifies some of the #includes accordingly.

Fixes: efd9e03facd0 ("arm64: Use static keys for CPU features")
Reported-by: Artem Savkov <asavkov@redhat.com>
Tested-by: Artem Savkov <asavkov@redhat.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>
8 years agoMerge branches 'sched-urgent-for-linus' and 'core-urgent-for-linus' of git://git...
Linus Torvalds [Sat, 5 Nov 2016 18:46:02 +0000 (11:46 -0700)]
Merge branches 'sched-urgent-for-linus' and 'core-urgent-for-linus' of git://git./linux/kernel/git/tip/tip

Pull stack vmap fixups from Thomas Gleixner:
 "Two small patches related to sched_show_task():

   - make sure to hold a reference on the task stack while accessing it

   - remove the thread_saved_pc printout

  .. and add a sanity check into release_task_stack() to catch problems
  with task stack references"

* 'sched-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  sched/core: Remove pointless printout in sched_show_task()
  sched/core: Fix oops in sched_show_task()

* 'core-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  fork: Add task stack refcounting sanity check and prevent premature task stack freeing

8 years agoMerge tag 'md/4.9-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/shli/md
Linus Torvalds [Sat, 5 Nov 2016 18:34:07 +0000 (11:34 -0700)]
Merge tag 'md/4.9-rc3' of git://git./linux/kernel/git/shli/md

Pull MD fixes from Shaohua Li:
 "There are several bug fixes queued:

   - fix raid5-cache recovery bugs

   - fix discard IO error handling for raid1/10

   - fix array sync writes bogus position to superblock

   - fix IO error handling for raid array with external metadata"

* tag 'md/4.9-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/shli/md:
  md: be careful not lot leak internal curr_resync value into metadata. -- (all)
  raid1: handle read error also in readonly mode
  raid5-cache: correct condition for empty metadata write
  md: report 'write_pending' state when array in sync
  md/raid5: write an empty meta-block when creating log super-block
  md/raid5: initialize next_checkpoint field before use
  RAID10: ignore discard error
  RAID1: ignore discard error

8 years agoMerge tag 'scsi-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi
Linus Torvalds [Sat, 5 Nov 2016 18:28:21 +0000 (11:28 -0700)]
Merge tag 'scsi-fixes' of git://git./linux/kernel/git/jejb/scsi

Pull SCSI fixes from James Bottomley:
 "Two more important data integrity fixes related to RAID device drivers
  which wrongly throw away the SYNCHRONIZE CACHE command in the non-RAID
  path and a memory leak in the scsi_debug driver"

* tag 'scsi-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi:
  scsi: arcmsr: Send SYNCHRONIZE_CACHE command to firmware
  scsi: scsi_debug: Fix memory leak if LBP enabled and module is unloaded
  scsi: megaraid_sas: Fix data integrity failure for JBOD (passthrough) devices

8 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dtor/input
Linus Torvalds [Sat, 5 Nov 2016 18:26:11 +0000 (11:26 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/dtor/input

Pull input subsystem updates from Dmitry Torokhov.

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dtor/input:
  Input: psmouse - cleanup Focaltech code
  Input: i8042 - add XMG C504 to keyboard reset table

8 years agoMerge tag 'firewire-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/ieee1394...
Linus Torvalds [Sat, 5 Nov 2016 18:17:34 +0000 (11:17 -0700)]
Merge tag 'firewire-fixes' of git://git./linux/kernel/git/ieee1394/linux1394

Pull FireWire (IEEE 1394) fixes from Stefan Richter:

 - add missing input validation to the firewire-net driver. Invalid
   IP-over-1394 encapsulation headers could trigger buffer overflows
   (CVE 2016-8633).

 - IP-over-1394 link fragmentation headers were read and written
   incorrectly, breaking fragmented RX/TX with other OS's stacks.

* tag 'firewire-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/ieee1394/linux1394:
  firewire: net: fix fragmented datagram_size off-by-one
  firewire: net: guard against rx buffer overflows

8 years agoMerge tag 'media/v4.9-3' of git://git.kernel.org/pub/scm/linux/kernel/git/mchehab...
Linus Torvalds [Sat, 5 Nov 2016 18:15:09 +0000 (11:15 -0700)]
Merge tag 'media/v4.9-3' of git://git./linux/kernel/git/mchehab/linux-media

Pull media fixes from Mauro Carvalho Chehab:
 "A series of fixup patches meant to fix the usage of DMA on stack, plus
  one warning fixup"

* tag 'media/v4.9-3' of git://git.kernel.org/pub/scm/linux/kernel/git/mchehab/linux-media: (32 commits)
  [media] radio-bcm2048: don't ignore errors
  [media] pctv452e: fix semicolon.cocci warnings
  [media] flexcop-usb: don't use stack for DMA
  [media] stk-webcam: don't use stack for DMA
  [media] s2255drv: don't use stack for DMA
  [media] cpia2_usb: don't use stack for DMA
  [media] digitv: handle error code on RC query
  [media] dw2102: return error if su3000_power_ctrl() fails
  [media] nova-t-usb2: handle error code on RC query
  [media] technisat-usb2: use DMA buffers for I2C transfers
  [media] pctv452e: don't call BUG_ON() on non-fatal error
  [media] pctv452e: don't do DMA on stack
  [media] nova-t-usb2: don't do DMA on stack
  [media] gp8psk: don't go past the buffer size
  [media] gp8psk: don't do DMA on stack
  [media] dtv5100: don't do DMA on stack
  [media] dtt200u: handle USB control message errors
  [media] dtt200u: don't do DMA on stack
  [media] dtt200u-fe: handle errors on USB control messages
  [media] dtt200u-fe: don't do DMA on stack
  ...

8 years agoMerge tag 'pci-v4.9-fixes-2' of git://git.kernel.org/pub/scm/linux/kernel/git/helgaas/pci
Linus Torvalds [Sat, 5 Nov 2016 18:11:31 +0000 (11:11 -0700)]
Merge tag 'pci-v4.9-fixes-2' of git://git./linux/kernel/git/helgaas/pci

Pull PCI fixes from Bjorn Helgaas:

 - fix for a Qualcomm driver issue that causes a use-before-set crash

 - fix for DesignWare iATU unroll support that causes external aborts
   when enabling the host bridge

* tag 'pci-v4.9-fixes-2' of git://git.kernel.org/pub/scm/linux/kernel/git/helgaas/pci:
  PCI: designware: Check for iATU unroll support after initializing host
  PCI: qcom: Fix pp->dev usage before assignment

8 years agoMerge tag 'for-linus-20161104' of git://git.infradead.org/linux-mtd
Linus Torvalds [Sat, 5 Nov 2016 17:52:29 +0000 (10:52 -0700)]
Merge tag 'for-linus-20161104' of git://git.infradead.org/linux-mtd

Pull MTD fixes from Brian Norris:

 - MAINTAINERS updates to reflect some new maintainers/submaintainers.

   We have some great volunteers who've been developing and reviewing
   already. We're going to try a group maintainership model, so
   eventually you'll probably see pull requests from people besides me.

 - NAND fixes from Boris:
    "Three simple fixes:

      - fix a non-critical bug in the gpmi driver
      - fix a bug in the 'automatic NAND timings selection' feature
        introduced in 4.9-rc1
      - fix a false positive uninitialized-var warning"

* tag 'for-linus-20161104' of git://git.infradead.org/linux-mtd:
  mtd: mtk: avoid warning in mtk_ecc_encode
  mtd: nand: Fix data interface configuration logic
  mtd: nand: gpmi: disable the clocks on errors
  MAINTAINERS: add more people to the MTD maintainer team
  MAINTAINERS: add a maintainer for the SPI NOR subsystem

8 years agoMerge tag 'mmc-v4.9-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/ulfh/mmc
Linus Torvalds [Sat, 5 Nov 2016 17:49:28 +0000 (10:49 -0700)]
Merge tag 'mmc-v4.9-rc2' of git://git./linux/kernel/git/ulfh/mmc

Pull MMC fixes from Ulf Hansson:
 "MMC host:

   - sdhci-msm: Fix error path in probe
   - dw_mmc-pltfm: Avoid NULL pointer dereference"

* tag 'mmc-v4.9-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/ulfh/mmc:
  mmc: sdhci-msm: Fix error return code in sdhci_msm_probe()
  mmc: dw_mmc-pltfm: fix the potential NULL pointer dereference

8 years agoMerge tag 'gpio-v4.9-3' of git://git.kernel.org/pub/scm/linux/kernel/git/linusw/linux...
Linus Torvalds [Sat, 5 Nov 2016 17:42:20 +0000 (10:42 -0700)]
Merge tag 'gpio-v4.9-3' of git://git./linux/kernel/git/linusw/linux-gpio

Pull GPIO fixes from Linus Walleij:
 "Some GPIO fixes for the v4.9 series:

   - Fix a nasty file descriptor leak when getting line handles.

   - A fix for a cleanup that seemed innocent but created a problem for
     drivers instantiating several gpiochips for one single OF node.

   - Fix a unpredictable problem using irq_domain_simple() in the mvebu
     driver by converting it to a lineas irqdomain"

* tag 'gpio-v4.9-3' of git://git.kernel.org/pub/scm/linux/kernel/git/linusw/linux-gpio:
  gpio/mvebu: Use irq_domain_add_linear
  gpio: of: fix GPIO drivers with multiple gpio_chip for a single node
  gpio: GPIO_GET_LINE{HANDLE,EVENT}_IOCTL: Fix file descriptor leak

8 years agoHID: sensor: fix attributes in HID sensor interface
Ooi, Joyce [Thu, 3 Nov 2016 10:55:15 +0000 (18:55 +0800)]
HID: sensor: fix attributes in HID sensor interface

User is unable to access to input-X-yyy and feature-X-yyy where
X is a hex value and more than 9 (e.g. input-a-yyy, feature-b-yyy) in HID
sensor custom sysfs interface.
This is because when creating the attribute, the attribute index is
written to using %x (hex). However, when reading and writing values into
the attribute, the attribute index is scanned using %d (decimal). Hence,
user is unable to access to attributes with index in hex values
(e.g. 'a', 'b', 'c') but able to access to attributes with index in
decimal values (e.g. 1, 2, 3,..).
This fix will change input-%d-%x-%s and feature-%d-%x-%s to input-%x-%x-%s
and feature-%x-%x-%s in show_values() and store_values() accordingly.

Signed-off-by: Ooi, Joyce <joyce.ooi@intel.com>
Reviewed-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
Acked-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
8 years agoHID: intel-ish-hid: request_irq failure
Srinivas Pandruvada [Fri, 21 Oct 2016 22:48:41 +0000 (15:48 -0700)]
HID: intel-ish-hid: request_irq failure

On some platforms ISH interrupt is shared, which causes request_irq to
fail. This requires IRQF_SHARED irq flag.

But IRQF_NO_SUSPEND and IRQF_SHARED should not be used together, so
removed IRQF_NO_SUSPEND flag. Anyway this driver doesn't require
IRQF_NO_SUSPEND, as this interrupt is not required during "noirq" phases
of suspending and resuming devices as well as during the time when
nonboot CPUs are taken offline and brought back online.

Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
8 years agoHID: intel-ish-hid: Fix driver reinit failure
Even Xu [Fri, 21 Oct 2016 22:48:40 +0000 (15:48 -0700)]
HID: intel-ish-hid: Fix driver reinit failure

When built as a module, modprobe followed by rmmod can fail because
DMA was still active. So to fix this, DMA needs to be disabled during
module exit.

This change disables DMA during modules exit and change the ISH PCI
device status to D3.

Signed-off-by: Even Xu <even.xu@intel.com>
Acked-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
8 years agoHID: intel-ish-hid: Move DMA disable code to new function
Even Xu [Fri, 21 Oct 2016 22:48:39 +0000 (15:48 -0700)]
HID: intel-ish-hid: Move DMA disable code to new function

Add a new function ish_disable_dma() and move DMA disable operations
here, so that this functionality can be reused.

Signed-off-by: Even Xu <even.xu@intel.com>
Acked-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
8 years agoHID: intel-ish-hid: consolidate ish wake up operation
Even Xu [Fri, 21 Oct 2016 22:48:38 +0000 (15:48 -0700)]
HID: intel-ish-hid: consolidate ish wake up operation

Same operations are done in ish_hw_start() and _ish_hw_reset() to
wakeup ISH device. Consolidate them by introducing a new function
ish_wakeup() and move the code there.

Signed-off-by: Even Xu <even.xu@intel.com>
Acked-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
8 years agoMerge tag 'nfsd-4.9-1' of git://linux-nfs.org/~bfields/linux
Linus Torvalds [Sat, 5 Nov 2016 03:12:10 +0000 (20:12 -0700)]
Merge tag 'nfsd-4.9-1' of git://linux-nfs.org/~bfields/linux

Pull nfsd bugfixes from Bruce Fields:
 "Fixes for some recent regressions including fallout from the vmalloc'd
  stack change (after which we can no longer encrypt stuff on the
  stack)"

* tag 'nfsd-4.9-1' of git://linux-nfs.org/~bfields/linux:
  nfsd: Fix general protection fault in release_lock_stateid()
  svcrdma: backchannel cannot share a page for send and rcv buffers
  sunrpc: fix some missing rq_rbuffer assignments
  sunrpc: don't pass on-stack memory to sg_set_buf
  nfsd: move blocked lock handling under a dedicated spinlock

8 years agoMerge branch 'for-4.9-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/kdave...
Linus Torvalds [Sat, 5 Nov 2016 03:08:16 +0000 (20:08 -0700)]
Merge branch 'for-4.9-rc3' of git://git./linux/kernel/git/kdave/linux

Pull btrfs fixes from Chris Mason:
 "Some fixes that Dave Sterba collected.  We held off on these last week
  because I was focused on the memory corruption testing"

* 'for-4.9-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/kdave/linux:
  btrfs: fix WARNING in btrfs_select_ref_head()
  Btrfs: remove some no-op casts
  btrfs: pass correct args to btrfs_async_run_delayed_refs()
  btrfs: make file clone aware of fatal signals
  btrfs: qgroup: Prevent qgroup->reserved from going subzero
  Btrfs: kill BUG_ON in do_relocation

8 years agoMerge branch 'overlayfs-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mszer...
Linus Torvalds [Sat, 5 Nov 2016 03:03:14 +0000 (20:03 -0700)]
Merge branch 'overlayfs-linus' of git://git./linux/kernel/git/mszeredi/vfs

Pull overlayfs fixes from Miklos Szeredi:
 "Fix two more POSIX ACL bugs introduced in 4.8 and add a missing fsync
  during copy up to prevent possible data loss"

* 'overlayfs-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mszeredi/vfs:
  ovl: fsync after copy-up
  ovl: fix get_acl() on tmpfs
  ovl: update S_ISGID when setting posix ACLs

8 years agoMerge tag 'drm-fixes-for-v4.9-rc4' of git://people.freedesktop.org/~airlied/linux
Linus Torvalds [Fri, 4 Nov 2016 20:30:13 +0000 (13:30 -0700)]
Merge tag 'drm-fixes-for-v4.9-rc4' of git://people.freedesktop.org/~airlied/linux

Pull drm fixes from Dave Airlie:
 "Fixes for amdgpu, radeon, intel, imx and virtio-gpu.

  This is a bit larger than I'd like, but I had some stuff I meant to
  send for -rc3 but was waiting for the PAT regression fix to land. So
  this is really fixes for rc3 and rc4 in one go.

  There are a set of fixes for an oops we've been seeing around MST
  display unplug, along with more suspend/resume and shutdown fixes for
  amdgpu, one power management follow on fix for nouveau, and set of imx
  fixes, and a single virtio-gpu regression fix"

* tag 'drm-fixes-for-v4.9-rc4' of git://people.freedesktop.org/~airlied/linux: (54 commits)
  virtio-gpu: fix vblank events
  drm/nouveau/acpi: fix check for power resources support
  drm/i915: Fix SKL+ 90/270 degree rotated plane coordinate computation
  drm/i915: Remove two invalid warns
  drm/i915: Rotated view does not need a fence
  drm/i915/fbc: fix CFB size calculation for gen8+
  drm: i915: Wait for fences on new fb, not old
  drm/i915: Clean up DDI DDC/AUX CH sanitation
  drm/i915: Respect alternate_aux_channel for all DDI ports
  drm/i915/gen9: fix watermarks when using the pipe scaler
  drm/i915: Fix mismatched INIT power domain disabling during suspend
  drm/i915: fix a read size argument
  drm/i915: Use fence_write() from rpm resume
  drm/i915/gen9: fix DDB partitioning for multi-screen cases
  drm/i915: workaround sparse warning on variable length arrays
  drm/i915: keep declarations in i915_drv.h
  drm/amd/powerplay: fix bug get wrong evv voltage of Polaris.
  drm/amdgpu/si_dpm: workaround for SI kickers
  drm/radeon/si_dpm: workaround for SI kickers
  drm/amdgpu: fix s3 resume back, uvd dpm randomly can't disable.
  ...

8 years agoPCI: designware: Check for iATU unroll support after initializing host
Niklas Cassel [Fri, 14 Oct 2016 21:54:55 +0000 (23:54 +0200)]
PCI: designware: Check for iATU unroll support after initializing host

dw_pcie_iatu_unroll_enabled() reads a dbi_base register.  Reading any
dbi_base register before pp->ops->host_init has been called causes
"imprecise external abort" on platforms like ARTPEC-6, where the PCIe
module is disabled at boot and first enabled in pp->ops->host_init.  Move
dw_pcie_iatu_unroll_enabled() to dw_pcie_setup_rc(), since it is after
pp->ops->host_init, but before pp->iatu_unroll_enabled is actually used.

Fixes: a0601a470537 ("PCI: designware: Add iATU Unroll feature")
Tested-by: James Le Cuirot <chewi@gentoo.org>
Signed-off-by: Niklas Cassel <niklas.cassel@axis.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Acked-by: Joao Pinto <jpinto@synopsys.com>
Acked-by: Olof Johansson <olof@lixom.net>
8 years agoMerge tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm
Linus Torvalds [Fri, 4 Nov 2016 20:08:05 +0000 (13:08 -0700)]
Merge tag 'for-linus' of git://git./virt/kvm/kvm

Pull KVM updates from Paolo Bonzini:
 "One NULL pointer dereference, and two fixes for regressions introduced
  during the merge window.

  The rest are fixes for MIPS, s390 and nested VMX"

* tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm:
  kvm: x86: Check memopp before dereference (CVE-2016-8630)
  kvm: nVMX: VMCLEAR an active shadow VMCS after last use
  KVM: x86: drop TSC offsetting kvm_x86_ops to fix KVM_GET/SET_CLOCK
  KVM: x86: fix wbinvd_dirty_mask use-after-free
  kvm/x86: Show WRMSR data is in hex
  kvm: nVMX: Fix kernel panics induced by illegal INVEPT/INVVPID types
  KVM: document lock orders
  KVM: fix OOPS on flush_work
  KVM: s390: Fix STHYI buffer alignment for diag224
  KVM: MIPS: Precalculate MMIO load resume PC
  KVM: MIPS: Make ERET handle ERL before EXL
  KVM: MIPS: Fix lazy user ASID regenerate for SMP

8 years agoMerge branch 'upstream' of git://git.linux-mips.org/pub/scm/ralf/upstream-linus
Linus Torvalds [Fri, 4 Nov 2016 20:03:57 +0000 (13:03 -0700)]
Merge branch 'upstream' of git://git.linux-mips.org/ralf/upstream-linus

Pull MIPS fixes from Ralf Baechle:
 "A set of MIPS fixes for 4.9:

   - lots of fixes for printk continuations
   - six fixes for FP related code.
   - fix max_low_pfn with disabled highmem
   - fix KASLR handling of NULL FDT and KASLR for generic kernels
   - fix build of compressed image
   - provide default mips_cpc_default_phys_base to ignore CPC
   - fix reboot on Malta"

* 'upstream' of git://git.linux-mips.org/pub/scm/ralf/upstream-linus:
  MIPS: Fix max_low_pfn with disabled highmem
  MIPS: Correct MIPS I FP sigcontext layout
  MIPS: Fix ISA I/II FP signal context offsets
  MIPS: Remove FIR from ISA I FP signal context
  MIPS: Fix ISA I FP sigcontext access violation handling
  MIPS: Fix FCSR Cause bit handling for correct SIGFPE issue
  MIPS: ptrace: Also initialize the FP context on individual FCSR writes
  MIPS: dump_tlb: Fix printk continuations
  MIPS: Fix __show_regs() output
  MIPS: traps: Fix output of show_code
  MIPS: traps: Fix output of show_stacktrace
  MIPS: traps: Fix output of show_backtrace
  MIPS: Fix build of compressed image
  MIPS: generic: Fix KASLR for generic kernel.
  MIPS: KASLR: Fix handling of NULL FDT
  MIPS: Malta: Fixup reboot
  MIPS: CPC: Provide default mips_cpc_default_phys_base to ignore CPC

8 years agoMerge branch 'parisc-4.9-3' of git://git.kernel.org/pub/scm/linux/kernel/git/deller...
Linus Torvalds [Fri, 4 Nov 2016 20:01:13 +0000 (13:01 -0700)]
Merge branch 'parisc-4.9-3' of git://git./linux/kernel/git/deller/parisc-linux

Pull parisc updates from Helge Deller:
 "The first three patches are trivial and add some required KERN_CONT,
  ignore the new pkey syscalls on parisc and use the LINUX_GATEWAY_ADDR
  define instead of hardcoded values.

  The two patches from Dave Anglin are important.

  The first one avoids trashing the sr2 and sr3 space registers in the
  Light-weight syscall path. Especially the usage of sr3 is critical
  since it may get trashed by the interrupt handler.

  The second patch is even more important and tagged for stable series.
  It protects one critical section in the syscall entry path by
  disabling local interrupts. Without disabling interrupts, the sr7
  space register may not be in sync with the current stack setup and
  thus an incoming hardware interrupt may destroy memory in random
  userspace areas"

* 'parisc-4.9-3' of git://git.kernel.org/pub/scm/linux/kernel/git/deller/parisc-linux:
  parisc: Ignore the pkey system calls for now
  parisc: Use LINUX_GATEWAY_ADDR define instead of hardcoded value
  parisc: Ensure consistent state when switching to kernel stack at syscall entry
  parisc: Avoid trashing sr2 and sr3 in LWS code
  parisc: use KERN_CONT when printing device inventory

8 years agoi2c: core: fix NULL pointer dereference under race condition
Vladimir Zapolskiy [Mon, 31 Oct 2016 19:46:24 +0000 (21:46 +0200)]
i2c: core: fix NULL pointer dereference under race condition

Race condition between registering an I2C device driver and
deregistering an I2C adapter device which is assumed to manage that
I2C device may lead to a NULL pointer dereference due to the
uninitialized list head of driver clients.

The root cause of the issue is that the I2C bus may know about the
registered device driver and thus it is matched by bus_for_each_drv(),
but the list of clients is not initialized and commonly it is NULL,
because I2C device drivers define struct i2c_driver as static and
clients field is expected to be initialized by I2C core:

  i2c_register_driver()             i2c_del_adapter()
    driver_register()                 ...
      bus_add_driver()                ...
        ...                           bus_for_each_drv(..., __process_removed_adapter)
      ...                               i2c_do_del_adapter()
    ...                                   list_for_each_entry_safe(..., &driver->clients, ...)
    INIT_LIST_HEAD(&driver->clients);

To solve the problem it is sufficient to do clients list head
initialization before calling driver_register().

The problem was found while using an I2C device driver with a sluggish
registration routine on a bus provided by a physically detachable I2C
master controller, but practically the oops may be reproduced under
the race between arbitraty I2C device driver registration and managing
I2C bus device removal e.g. by unbinding the latter over sysfs:

% echo 21a4000.i2c > /sys/bus/platform/drivers/imx-i2c/unbind
  Unable to handle kernel NULL pointer dereference at virtual address 00000000
  Internal error: Oops: 17 [#1] SMP ARM
  CPU: 2 PID: 533 Comm: sh Not tainted 4.9.0-rc3+ #61
  Hardware name: Freescale i.MX6 Quad/DualLite (Device Tree)
  task: e5ada400 task.stack: e4936000
  PC is at i2c_do_del_adapter+0x20/0xcc
  LR is at __process_removed_adapter+0x14/0x1c
  Flags: NzCv  IRQs on  FIQs on  Mode SVC_32  ISA ARM  Segment none
  Control: 10c5387d  Table: 35bd004a  DAC: 00000051
  Process sh (pid: 533, stack limit = 0xe4936210)
  Stack: (0xe4937d28 to 0xe4938000)
  Backtrace:
  [<c0667be0>] (i2c_do_del_adapter) from [<c0667cc0>] (__process_removed_adapter+0x14/0x1c)
  [<c0667cac>] (__process_removed_adapter) from [<c0516998>] (bus_for_each_drv+0x6c/0xa0)
  [<c051692c>] (bus_for_each_drv) from [<c06685ec>] (i2c_del_adapter+0xbc/0x284)
  [<c0668530>] (i2c_del_adapter) from [<bf0110ec>] (i2c_imx_remove+0x44/0x164 [i2c_imx])
  [<bf0110a8>] (i2c_imx_remove [i2c_imx]) from [<c051a838>] (platform_drv_remove+0x2c/0x44)
  [<c051a80c>] (platform_drv_remove) from [<c05183d8>] (__device_release_driver+0x90/0x12c)
  [<c0518348>] (__device_release_driver) from [<c051849c>] (device_release_driver+0x28/0x34)
  [<c0518474>] (device_release_driver) from [<c0517150>] (unbind_store+0x80/0x104)
  [<c05170d0>] (unbind_store) from [<c0516520>] (drv_attr_store+0x28/0x34)
  [<c05164f8>] (drv_attr_store) from [<c0298acc>] (sysfs_kf_write+0x50/0x54)
  [<c0298a7c>] (sysfs_kf_write) from [<c029801c>] (kernfs_fop_write+0x100/0x214)
  [<c0297f1c>] (kernfs_fop_write) from [<c0220130>] (__vfs_write+0x34/0x120)
  [<c02200fc>] (__vfs_write) from [<c0221088>] (vfs_write+0xa8/0x170)
  [<c0220fe0>] (vfs_write) from [<c0221e74>] (SyS_write+0x4c/0xa8)
  [<c0221e28>] (SyS_write) from [<c0108a20>] (ret_fast_syscall+0x0/0x1c)

Signed-off-by: Vladimir Zapolskiy <vladimir_zapolskiy@mentor.com>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
Cc: stable@kernel.org
8 years agoMIPS: Fix max_low_pfn with disabled highmem
James Hogan [Tue, 1 Nov 2016 13:59:09 +0000 (13:59 +0000)]
MIPS: Fix max_low_pfn with disabled highmem

When low memory doesn't reach HIGHMEM_START (e.g. up to 256MB at PA=0 is
common) and highmem is present above HIGHMEM_START (e.g. on Malta the
RAM overlayed by the IO region is aliased at PA=0x90000000), max_low_pfn
will be initially calculated very large and then clipped down to
HIGHMEM_START.

This causes crashes when reading /sys/kernel/mm/page_idle/bitmap
(i.e. CONFIG_IDLE_PAGE_TRACKING=y) when highmem is disabled. pfn_valid()
will compare against max_mapnr which is derived from max_low_pfn when
there is no highend_pfn set up, and will return true for PFNs right up
to HIGHMEM_START, even though they are beyond the end of low memory and
no page structs will actually exist for these PFNs.

This is fixed by skipping high memory regions when initially calculating
max_low_pfn if highmem is disabled, so it doesn't get clipped too high.
We also clip regions which overlap the highmem boundary when highmem is
disabled, so that max_pfn doesn't extend into highmem either.

Signed-off-by: James Hogan <james.hogan@imgtec.com>
Cc: Paul Burton <paul.burton@imgtec.com>
Cc: linux-mips@linux-mips.org
Patchwork: https://patchwork.linux-mips.org/patch/14490/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
8 years agoMIPS: Correct MIPS I FP sigcontext layout
Maciej W. Rozycki [Mon, 31 Oct 2016 16:27:40 +0000 (16:27 +0000)]
MIPS: Correct MIPS I FP sigcontext layout

Complement commit 80cbfad79096 ("MIPS: Correct MIPS I FP context
layout") and correct the way Floating Point General registers are stored
in a signal context with MIPS I hardware.

Use the S.D and L.D assembly macros to have pairs of SWC1 instructions
and pairs of LWC1 instructions produced, respectively, in an arrangement
which makes the memory representation of floating-point data passed
compatible with that used by hardware SDC1 and LDC1 instructions, where
available, regardless of the hardware endianness used.  This matches the
layout used by r4k_fpu.S, ensuring run-time compatibility for MIPS I
software across all o32 hardware platforms.

Define an EX2 macro to handle exceptions from both hardware instructions
implicitly produced from S.D and L.D assembly macros.

Signed-off-by: Maciej W. Rozycki <macro@imgtec.com>
Cc: linux-mips@linux-mips.org
Cc: linux-kernel@vger.kernel.org
Patchwork: https://patchwork.linux-mips.org/patch/14477/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
8 years agoMIPS: Fix ISA I/II FP signal context offsets
Maciej W. Rozycki [Mon, 31 Oct 2016 16:27:01 +0000 (16:27 +0000)]
MIPS: Fix ISA I/II FP signal context offsets

Fix a regression introduced with commit 2db9ca0a3551 ("MIPS: Use struct
mips_abi offsets to save FP context") for MIPS I/I FP signal contexts,
by converting save/restore code to the updated internal API.  Start FGR
offsets from 0 rather than SC_FPREGS from $a0 and use $a1 rather than
the offset of SC_FPC_CSR from $a0 for the Floating Point Control/Status
Register (FCSR).

Document the new internal API and adjust assembly code formatting for
consistency.

Signed-off-by: Maciej W. Rozycki <macro@imgtec.com>
Cc: Paul Burton <paul.burton@imgtec.com>
Cc: linux-mips@linux-mips.org
Cc: linux-kernel@vger.kernel.org
Patchwork: https://patchwork.linux-mips.org/patch/14476/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
8 years agoMIPS: Remove FIR from ISA I FP signal context
Maciej W. Rozycki [Mon, 31 Oct 2016 16:26:24 +0000 (16:26 +0000)]
MIPS: Remove FIR from ISA I FP signal context

Complement commit e50c0a8fa60d ("Support the MIPS32 / MIPS64 DSP ASE.")
and remove the Floating Point Implementation Register (FIR) from the FP
register set recorded in a signal context with MIPS I processors too, in
line with the change applied to r4k_fpu.S.

The `sc_fpc_eir' slot is unused according to our current ABI and the FIR
register is read-only and always directly accessible from user software.

[ralf@linux-mips.org: This is also required because the next commit depends
on it.]

Signed-off-by: Maciej W. Rozycki <macro@imgtec.com>
Cc: linux-mips@linux-mips.org
Cc: linux-kernel@vger.kernel.org
Patchwork: https://patchwork.linux-mips.org/patch/14475/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
8 years agoMIPS: Fix ISA I FP sigcontext access violation handling
Maciej W. Rozycki [Mon, 31 Oct 2016 16:25:44 +0000 (16:25 +0000)]
MIPS: Fix ISA I FP sigcontext access violation handling

Complement commit 0ae8dceaebe3 ("Merge with 2.3.10.") and use the local
`fault' handler to recover from FP sigcontext access violation faults,
like corresponding code does in r4k_fpu.S.  The `bad_stack' handler is
in syscall.c and is not suitable here as we want to propagate the error
condition up through the caller rather than killing the thread outright.

Signed-off-by: Maciej W. Rozycki <macro@imgtec.com>
Cc: linux-mips@linux-mips.org
Cc: linux-kernel@vger.kernel.org
Patchwork: https://patchwork.linux-mips.org/patch/14474/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
8 years agoMIPS: Fix FCSR Cause bit handling for correct SIGFPE issue
Maciej W. Rozycki [Fri, 28 Oct 2016 07:21:03 +0000 (08:21 +0100)]
MIPS: Fix FCSR Cause bit handling for correct SIGFPE issue

Sanitize FCSR Cause bit handling, following a trail of past attempts:

* commit 4249548454f7 ("MIPS: ptrace: Fix FP context restoration FCSR
regression"),

* commit 443c44032a54 ("MIPS: Always clear FCSR cause bits after
emulation"),

* commit 64bedffe4968 ("MIPS: Clear [MSA]FPE CSR.Cause after
notify_die()"),

* commit b1442d39fac2 ("MIPS: Prevent user from setting FCSR cause
bits"),

* commit b54d2901517d ("Properly handle branch delay slots in connection
with signals.").

Specifically do not mask these bits out in ptrace(2) processing and send
a SIGFPE signal instead whenever a matching pair of an FCSR Cause and
Enable bit is seen as execution of an affected context is about to
resume.  Only then clear Cause bits, and even then do not clear any bits
that are set but masked with the respective Enable bits.  Adjust Cause
bit clearing throughout code likewise, except within the FPU emulator
proper where they are set according to IEEE 754 exceptions raised as the
operation emulated executed.  Do so so that any IEEE 754 exceptions
subject to their default handling are recorded like with operations
executed by FPU hardware.

Signed-off-by: Maciej W. Rozycki <macro@imgtec.com>
Cc: Paul Burton <paul.burton@imgtec.com>
Cc: James Hogan <james.hogan@imgtec.com>
Cc: linux-mips@linux-mips.org
Cc: linux-kernel@vger.kernel.org
Patchwork: https://patchwork.linux-mips.org/patch/14460/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
8 years agoMIPS: ptrace: Also initialize the FP context on individual FCSR writes
Maciej W. Rozycki [Fri, 28 Oct 2016 07:20:09 +0000 (08:20 +0100)]
MIPS: ptrace: Also initialize the FP context on individual FCSR writes

Complement commit ac9ad83bc318 ("MIPS: prevent FP context set via ptrace
being discarded") and also initialize the FP context whenever FCSR alone
is written with a PTRACE_POKEUSR request addressing FPC_CSR, rather than
along with the full FPU register set in the case of the PTRACE_SETFPREGS
request.

Signed-off-by: Maciej W. Rozycki <macro@imgtec.com>
Cc: Paul Burton <paul.burton@imgtec.com>
Cc: James Hogan <james.hogan@imgtec.com>
Cc: linux-mips@linux-mips.org
Cc: linux-kernel@vger.kernel.org
Patchwork: https://patchwork.linux-mips.org/patch/14459/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
8 years agoMIPS: dump_tlb: Fix printk continuations
James Hogan [Fri, 21 Oct 2016 19:06:40 +0000 (20:06 +0100)]
MIPS: dump_tlb: Fix printk continuations

Since commit 4bcc595ccd80 ("printk: reinstate KERN_CONT for printing
continuation lines") the output from TLB dumps on MIPS has been
pretty unreadable due to the lack of KERN_CONT markers. Use pr_cont to
provide the appropriate markers & restore the expected output.

Continuation is also used for the second line of each TLB entry printed
in dump_tlb.c even though it has a newline, since it is a continuation
of the interpretation of the same TLB entry. For example:

[   46.371884] Index:  0 pgmask=16kb va=77654000 asid=73 gid=00
        [ri=0 xi=0 pa=ffc18000 c=5 d=0 v=1 g=0] [ri=0 xi=0 pa=ffc1c000 c=5 d=0 v=1 g=0]
[   46.385380] Index: 12 pgmask=16kb va=004b4000 asid=73 gid=00
        [ri=0 xi=0 pa=00000000 c=0 d=0 v=0 g=0] [ri=0 xi=0 pa=ffb00000 c=5 d=1 v=1 g=0]

Signed-off-by: James Hogan <james.hogan@imgtec.com>
Cc: Maciej W. Rozycki <macro@imgtec.com>
Cc: linux-mips@linux-mips.org
Patchwork: https://patchwork.linux-mips.org/patch/14444/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
8 years agoMIPS: Fix __show_regs() output
Paul Burton [Wed, 19 Oct 2016 13:33:23 +0000 (14:33 +0100)]
MIPS: Fix __show_regs() output

Since commit 4bcc595ccd80 ("printk: reinstate KERN_CONT for printing
continuation lines") the output from __show_regs() on MIPS has been
pretty unreadable due to the lack of KERN_CONT markers. Use pr_cont to
provide the appropriate markers & restore the expected register output.

Signed-off-by: Paul Burton <paul.burton@imgtec.com>
Signed-off-by: Matt Redfearn <matt.redfearn@imgtec.com>
Cc: Maciej W. Rozycki <macro@imgtec.com>
Cc: James Hogan <james.hogan@imgtec.com>
Cc: linux-mips@linux-mips.org
Cc: linux-kernel@vger.kernel.org
Patchwork: https://patchwork.linux-mips.org/patch/14432/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>