OSDN Git Service

uclinux-h8/linux.git
6 years agofpga: mgr: add status for fpga-manager
Wu Hao [Sat, 30 Jun 2018 00:53:10 +0000 (08:53 +0800)]
fpga: mgr: add status for fpga-manager

This patch adds status sysfs interface for fpga manager, it's a
read only interface which allows user to get fpga manager status,
including full/partial reconfiguration error and other status
information. It adds a status callback to fpga_manager_ops too,
allows each fpga_manager driver to define its own method to
collect latest status from hardware.

The following sysfs file is created:
* /sys/class/fpga_manager/<fpga>/status
  Return status of fpga manager, including reconfiguration errors.

Signed-off-by: Wu Hao <hao.wu@intel.com>
Acked-by: Alan Tull <atull@kernel.org>
Acked-by: Moritz Fischer <mdf@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agofpga: mgr: add region_id to fpga_image_info
Wu Hao [Sat, 30 Jun 2018 00:53:09 +0000 (08:53 +0800)]
fpga: mgr: add region_id to fpga_image_info

This patch adds region_id to fpga_image_info data structure, it
allows driver to pass region id information to fpga-mgr via
fpga_image_info for fpga reconfiguration function.

Signed-off-by: Wu Hao <hao.wu@intel.com>
Acked-by: Moritz Fischer <mdf@kernel.org>
Acked-by: Alan Tull <atull@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agodocs: fpga: add a document for FPGA Device Feature List (DFL) Framework Overview
Wu Hao [Sat, 30 Jun 2018 00:53:08 +0000 (08:53 +0800)]
docs: fpga: add a document for FPGA Device Feature List (DFL) Framework Overview

Add a document for FPGA Device Feature List (DFL) Framework Overview.

Signed-off-by: Enno Luebbers <enno.luebbers@intel.com>
Signed-off-by: Xiao Guangrong <guangrong.xiao@linux.intel.com>
Signed-off-by: Wu Hao <hao.wu@intel.com>
Acked-by: Alan Tull <atull@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agocoresight: tpiu: Fix disabling timeouts
Robin Murphy [Wed, 11 Jul 2018 19:40:35 +0000 (13:40 -0600)]
coresight: tpiu: Fix disabling timeouts

Probing the TPIU driver under UBSan triggers an out-of-bounds shift
warning in coresight_timeout():

...
[    5.677530] UBSAN: Undefined behaviour in drivers/hwtracing/coresight/coresight.c:929:16
[    5.685542] shift exponent 64 is too large for 64-bit type 'long unsigned int'
...

On closer inspection things are exponentially out of whack because we're
passing a bitmask where a bit number should be. Amusingly, it seems that
both calls will find their expected values by sheer luck and appear to
succeed: 1 << FFCR_FON_MAN ends up at bit 64 which whilst undefined
evaluates as zero in practice, while 1 << FFSR_FT_STOPPED finds bit 2
(TCPresent) which apparently is usually tied high.

Following the examples of other drivers, define separate FOO and FOO_BIT
macros for masks vs. indices, and put things right.

CC: Robert Walker <robert.walker@arm.com>
CC: Mike Leach <mike.leach@linaro.org>
CC: Mathieu Poirier <mathieu.poirier@linaro.org>
Fixes: 11595db8e17f ("coresight: Fix disabling of CoreSight TPIU")
Signed-off-by: Robin Murphy <robin.murphy@arm.com>
Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agocoresight: catu: Plug in CATU as a backend for ETR buffer
Suzuki K Poulose [Wed, 11 Jul 2018 19:40:34 +0000 (13:40 -0600)]
coresight: catu: Plug in CATU as a backend for ETR buffer

Now that we can use a CATU with a scatter gather table, add support
for the TMC ETR to make use of the connected CATU in translate mode.
This is done by adding CATU as new buffer mode.

Cc: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Suzuki K Poulose <suzuki.poulose@arm.com>
Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agocoresight: catu: Add support for scatter gather tables
Suzuki K Poulose [Wed, 11 Jul 2018 19:40:33 +0000 (13:40 -0600)]
coresight: catu: Add support for scatter gather tables

This patch adds the support for setting up a SG table for use
by the CATU. We reuse the tmc_sg_table to represent the table/data
pages, even though the table format is different.

Similar to ETR SG table, CATU uses a 4KB page size for data buffers
as well as page tables. All table entries are 64bit wide and have
the following format:

        63                      12      1  0
        x-----------------------------------x
        |        Address [63-12] | SBZ  | V |
        x-----------------------------------x

Where [V] ->  0 - Pointer is invalid
 1 - Pointer is Valid

CATU uses only first half of the page for data page pointers.
i.e, single table page will only have 256 page pointers, addressing
upto 1MB of data. The second half of a table page contains only two
pointers at the end of the page (i.e, pointers at index 510 and 511),
which are used as links to the "Previous" and "Next" page tables
respectively.

The first table page has an "Invalid" previous pointer and the
next pointer entry points to the second page table if there is one.
Similarly the last table page has an "Invalid" next pointer to
indicate the end of the table chain.

Cc: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Suzuki K Poulose <suzuki.poulose@arm.com>
Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agodts: bindings: Document device tree binding for CATU
Suzuki K Poulose [Wed, 11 Jul 2018 19:40:32 +0000 (13:40 -0600)]
dts: bindings: Document device tree binding for CATU

Document CATU device-tree bindings. CATU augments the TMC-ETR
by providing an improved Scatter Gather mechanism for streaming
trace data to non-contiguous system RAM pages.

Cc: devicetree@vger.kernel.org
Cc: frowand.list@gmail.com
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Mathieu Poirier <mathieu.poirier@linaro.org>
Reviewed-by: Rob Herring <robh@kernel.org>
Signed-off-by: Suzuki K Poulose <suzuki.poulose@arm.com>
Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agocoresight: Introduce support for Coresight Address Translation Unit
Suzuki K Poulose [Wed, 11 Jul 2018 19:40:31 +0000 (13:40 -0600)]
coresight: Introduce support for Coresight Address Translation Unit

Add the initial support for Coresight Address Translation Unit, which
augments the TMC in Coresight SoC-600 by providing an improved Scatter
Gather mechanism. CATU is always connected to a single TMC-ETR and
converts the AXI address with a translated address (from a given SG
table with specific format). The CATU should be programmed in pass
through mode and enabled even if the ETR doesn't use the translation
by CATU.

This patch provides mechanism to enable/disable the CATU always in the
pass through mode.

We reuse the existing ports mechanism to link the TMC-ETR to the
connected CATU.

i.e, TMC-ETR:output_port0 -> CATU:input_port0

Reference manual for CATU component is avilable in version r2p0 of :
"Arm Coresight System-on-Chip SoC-600 Technical Reference Manual".

Cc: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Suzuki K Poulose <suzuki.poulose@arm.com>
Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agocoresight: Add helper device type
Suzuki K Poulose [Wed, 11 Jul 2018 19:40:30 +0000 (13:40 -0600)]
coresight: Add helper device type

Add a new coresight device type, which do not belong to any
of the existing types, i.e, source, sink, link etc. A helper
device could be connected to a coresight device, which could
augment the functionality of the coresight device.

This is intended to cover Coresight Address Translation Unit (CATU)
devices, which provide improved Scatter Gather mechanism for TMC
ETR. The idea is that the helper device could be controlled by
the driver of the device it is attached to (in this case ETR),
transparent to the generic coresight driver (and paths).

The operations include enable(), disable(), both of which could
accept a device specific "data" which the driving device and
the helper device could share. Since they don't appear in the
coresight "path" tracked by software, we have to ensure that
they are powered up/down whenever the master device is turned
on.

Cc: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Suzuki K Poulose <suzuki.poulose@arm.com>
Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agocoresight: Cleanup device subtype struct
Suzuki K Poulose [Wed, 11 Jul 2018 19:40:29 +0000 (13:40 -0600)]
coresight: Cleanup device subtype struct

Clean up our struct a little bit by using a union instead of
a struct for tracking the subtype of a device.

Cc: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Suzuki K Poulose <suzuki.poulose@arm.com>
Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agocoresight: Handle errors in finding input/output ports
Suzuki K Poulose [Wed, 11 Jul 2018 19:40:28 +0000 (13:40 -0600)]
coresight: Handle errors in finding input/output ports

If we fail to find the input / output port for a LINK component
while enabling a path, we should fail gracefully rather than
assuming port "0".

Cc: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Suzuki K Poulose <suzuki.poulose@arm.com>
Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agocoresight: Cleanup platform description data
Suzuki K Poulose [Wed, 11 Jul 2018 19:40:27 +0000 (13:40 -0600)]
coresight: Cleanup platform description data

Nobody uses the "clk" field in struct coresight_platform_data.
Remove it.

Cc: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Suzuki K Poulose <suzuki.poulose@arm.com>
Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agocoresight: Fix check in coresight_tmc_etr_buf_insert_barrier_packet
Suzuki K Poulose [Wed, 11 Jul 2018 19:40:26 +0000 (13:40 -0600)]
coresight: Fix check in coresight_tmc_etr_buf_insert_barrier_packet

We request for "CORESIGHT_BARRIER_PKT_SIZE" length and we should
be happy when we get that size.

Cc: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Suzuki K Poulose <suzuki.poulose@arm.com>
Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agocoresight: include vmalloc.h for vmap/vunmap
Arnd Bergmann [Wed, 11 Jul 2018 19:40:25 +0000 (13:40 -0600)]
coresight: include vmalloc.h for vmap/vunmap

The newly introduced code fails to build in some configurations
unless we include the right headers:

drivers/hwtracing/coresight/coresight-tmc-etr.c: In function 'tmc_free_table_pages':
drivers/hwtracing/coresight/coresight-tmc-etr.c:206:3: error: implicit declaration of function 'vunmap'; did you mean 'iounmap'? [-Werror=implicit-function-declaration]

Fixes: 79613ae8715a ("coresight: Add generic TMC sg table framework")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agocoresight: tmc: Add configuration support for trace buffer size
Suzuki K Poulose [Wed, 11 Jul 2018 19:40:24 +0000 (13:40 -0600)]
coresight: tmc: Add configuration support for trace buffer size

Now that we can dynamically switch between contiguous memory and
SG table depending on the trace buffer size, provide the support
for selecting an appropriate buffer size.

Cc: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Suzuki K Poulose <suzuki.poulose@arm.com>
Acked-by: Rob Herring <robh@kernel.org>
Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agocoresight: tmc-etr buf: Add TMC scatter gather mode backend
Suzuki K Poulose [Wed, 11 Jul 2018 19:40:23 +0000 (13:40 -0600)]
coresight: tmc-etr buf: Add TMC scatter gather mode backend

Add the support for Scatter-Gather mode to the etr-buf layer.
Since we now have two different modes, we choose the backend
based on a set of conditions, documented in the code.

Cc: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Suzuki K Poulose <suzuki.poulose@arm.com>
Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agocoresight: tmc-etr: Add transparent buffer management
Suzuki K Poulose [Wed, 11 Jul 2018 19:40:22 +0000 (13:40 -0600)]
coresight: tmc-etr: Add transparent buffer management

The TMC-ETR can use the target trace buffer in two different modes.
Normal physically contiguous mode and a discontiguous list pages in
Scatter-Gather mode. Also we have dedicated Coresight component, CATU
(Coresight Address Translation Unit) to provide improved scatter-gather
mode in Coresight SoC-600. This complicates the management of the
buffer used for trace, depending on the mode in which ETR is configured.

So, this patch adds a transparent layer for managing the ETR buffer
which abstracts the basic operations on the buffer (alloc, free,
sync and retrieve the data) and uses the mode specific helpers to
do the actual operation. This also allows the ETR driver to choose
the best mode for a given use case and adds the flexibility to
fallback to a different mode, without duplicating the code.

The patch also adds the "normal" flat memory mode and switches
the sysfs driver to use the new layer.

Cc: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Suzuki K Poulose <suzuki.poulose@arm.com>
Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agocoresight: Add support for TMC ETR SG unit
Suzuki K Poulose [Wed, 11 Jul 2018 19:40:21 +0000 (13:40 -0600)]
coresight: Add support for TMC ETR SG unit

This patch adds support for setting up an SG table used by the
TMC ETR inbuilt SG unit. The TMC ETR uses 4K page sized tables
to hold pointers to the 4K data pages with the last entry in a
table pointing to the next table with the entries, by kind of
chaining. The 2 LSBs determine the type of the table entry, to
one of :

 Normal - Points to a 4KB data page.
 Last   - Points to a 4KB data page, but is the last entry in the
          page table.
 Link   - Points to another 4KB table page with pointers to data.

The code takes care of handling the system page size which could
be different than 4K. So we could end up putting multiple ETR
SG tables in a single system page, vice versa for the data pages.

Cc: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Suzuki K Poulose <suzuki.poulose@arm.com>
Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agocoresight: Add generic TMC sg table framework
Suzuki K Poulose [Wed, 11 Jul 2018 19:40:20 +0000 (13:40 -0600)]
coresight: Add generic TMC sg table framework

This patch introduces a generic sg table data structure and
associated operations. An SG table can be used to map a set
of Data pages where the trace data could be stored by the TMC
ETR. The information about the data pages could be stored in
different formats, depending on the type of the underlying
SG mechanism (e.g, TMC ETR SG vs Coresight CATU). The generic
structure provides book keeping of the pages used for the data
as well as the table contents. The table should be filled by
the user of the infrastructure.

A table can be created by specifying the number of data pages
as well as the number of table pages required to hold the
pointers, where the latter could be different for different
types of tables. The pages are mapped in the appropriate dma
data direction mode (i.e, DMA_TO_DEVICE for table pages
and DMA_FROM_DEVICE for data pages).  The framework can optionally
accept a set of allocated data pages (e.g, perf ring buffer) and
map them accordingly. The table and data pages are vmap'ed to allow
easier access by the drivers. The framework also provides helpers to
sync the data written to the pages with appropriate directions.

This will be later used by the TMC ETR SG unit and CATU.

Cc: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Suzuki K Poulose <suzuki.poulose@arm.com>
Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agodts: bindings: Restrict coresight tmc-etr scatter-gather mode
Suzuki K Poulose [Wed, 11 Jul 2018 19:40:19 +0000 (13:40 -0600)]
dts: bindings: Restrict coresight tmc-etr scatter-gather mode

We are about to add the support for ETR builtin scatter-gather mode
for dealing with large amount of trace buffers. However, on some of
the platforms, using the ETR SG mode can lock up the system due to
the way the ETR is connected to the memory subsystem.

In SG mode, the ETR performs READ from the scatter-gather table to
fetch the next page and regular WRITE of trace data. If the READ
operation doesn't complete(due to the memory subsystem issues,
which we have seen on a couple of platforms) the trace WRITE
cannot proceed leading to issues. So, we by default do not
use the SG mode, unless it is known to be safe on the platform.
We define a DT property for the TMC node to specify whether we
have a proper SG mode.

Cc: Mathieu Poirier <mathieu.poirier@linaro.org>
Cc: Mike Leach <mike.leach@linaro.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: John Horley <john.horley@arm.com>
Cc: Robert Walker <robert.walker@arm.com>
Cc: devicetree@vger.kernel.org
Cc: frowand.list@gmail.com
Reviewed-by: Rob Herring <robh@kernel.org>
Signed-off-by: Suzuki K Poulose <suzuki.poulose@arm.com>
Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agocoresight: Add helper for inserting synchronization packets
Suzuki K Poulose [Wed, 11 Jul 2018 19:40:18 +0000 (13:40 -0600)]
coresight: Add helper for inserting synchronization packets

Right now we open code filling the trace buffer with synchronization
packets when the circular buffer wraps around in different drivers.
Move this to a common place. While at it, clean up the barrier_pkt
array to strip off the trailing '\0'.

Cc: Mathieu Poirier <mathieu.poirier@linaro.org>
Cc: Mike Leach <mike.leach@linaro.org>
Signed-off-by: Suzuki K Poulose <suzuki.poulose@arm.com>
Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agocoresight: tmc-etr: Disallow perf mode
Suzuki K Poulose [Wed, 11 Jul 2018 19:40:17 +0000 (13:40 -0600)]
coresight: tmc-etr: Disallow perf mode

We don't support ETR in perf mode yet. So, don't
even try to enable the hardware, even by mistake.

Cc: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Suzuki K Poulose <suzuki.poulose@arm.com>
Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agocoresight: tmc-etr: Do not clean trace buffer
Suzuki K Poulose [Wed, 11 Jul 2018 19:40:16 +0000 (13:40 -0600)]
coresight: tmc-etr: Do not clean trace buffer

We zero out the entire trace buffer used for ETR before it is enabled,
for helping with debugging. With the addition of scatter-gather mode,
the buffer could be bigger and non-contiguous.

Get rid of this step; if someone wants to debug, they can always add it
as and when needed.

Cc: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Suzuki K Poulose <suzuki.poulose@arm.com>
Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agocoresight: tmc: Hide trace buffer handling for file read
Suzuki K Poulose [Wed, 11 Jul 2018 19:40:15 +0000 (13:40 -0600)]
coresight: tmc: Hide trace buffer handling for file read

At the moment we adjust the buffer pointers for reading the trace
data via misc device in the common code for ETF/ETB and ETR. Since
we are going to change how we manage the buffer for ETR, let us
move the buffer manipulation to the respective driver files, hiding
it from the common code. We do so by adding type specific helpers
for finding the length of data and the pointer to the buffer,
for a given length at a file position.

Cc: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Suzuki K Poulose <suzuki.poulose@arm.com>
Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agocoresight: ETM: Add support for Arm Cortex-A73 and Cortex-A35
Suzuki K Poulose [Wed, 11 Jul 2018 19:40:14 +0000 (13:40 -0600)]
coresight: ETM: Add support for Arm Cortex-A73 and Cortex-A35

Add ETM PIDs of the Arm cortex-A CPUs to the white list of ETMs.
While at it add a helper macro to make it easier to add the new
entries.

Cc: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Suzuki K Poulose <suzuki.poulose@arm.com>
Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agocoresight: Remove function coresight_vpid_to_pid()
Mathieu Poirier [Wed, 11 Jul 2018 19:40:13 +0000 (13:40 -0600)]
coresight: Remove function coresight_vpid_to_pid()

Now that we prevent users from using contextID tracing when PID namespaces
are involved there is no client for function coresight_vpid_to_pid().  As
such simply remove it.

Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Reviewed-by: Kim Phillips <kim.phillips@arm.com>
Acked-by: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agocoresight: etm4x: Don't use contextID with PID namespaces
Mathieu Poirier [Wed, 11 Jul 2018 19:40:12 +0000 (13:40 -0600)]
coresight: etm4x: Don't use contextID with PID namespaces

As with ETM3x, the ETM4x tracers can trigger trace acquisition based on
contextID value, something that isn't useful when PID namespaces are
enabled.  Indeed the PID value of a process has a different representation
in the kernel and the PID namespace, making the feature confusing and
potentially leaking internal kernel information.

As such simply return an error when the feature is being used from a
PID namespace other than the default one.

Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Reviewed-by: Kim Phillips <kim.phillips@arm.com>
Acked-by: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agocoresight: etm3x: Don't use contextID with PID namespaces
Mathieu Poirier [Wed, 11 Jul 2018 19:40:11 +0000 (13:40 -0600)]
coresight: etm3x: Don't use contextID with PID namespaces

Tracers can trigger trace acquisition based on contextID value, something
that isn't useful when PID namespaces are enabled.  Indeed the PID value
of a process has a different representation in the kernel and the PID
namespace, making the feature confusing and potentially leaking internal
kernel information.

As such simply return an error when the feature is being used from a
PID namespace other than the default one.

Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Reviewed-by: Kim Phillips <kim.phillips@arm.com>
Acked-by: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agomei: fix ssize_t to int assignment in read and write ops.
Tomas Winkler [Thu, 12 Jul 2018 14:10:10 +0000 (17:10 +0300)]
mei: fix ssize_t to int assignment in read and write ops.

Use ssize_t for rets variables in mei_write(), mei_read(), and
mei_cl_write() as well as change the return type of mei_cl_write()
to ssize_t, to prevent assignment of possible 64bit size_t
to int 32 bit variable.

As by product also eliminate warning
drivers/misc/mei/client.c:1702:11: warning: comparison between signed and unsigned integer expressions [-Wsign-compare]

Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agomei: use correct type for counter variable in for loops
Tomas Winkler [Thu, 12 Jul 2018 14:10:09 +0000 (17:10 +0300)]
mei: use correct type for counter variable in for loops

In for loops use same type for counter variable
as has the limiting variable.

drivers/misc/mei/bus-fixup.c:489:16: warning: comparison between signed and unsigned integer expressions [-Wsign-compare]
drivers/misc/mei/hw-txe.c:725:13: warning: comparison between signed and unsigned integer expressions [-Wsign-compare]
drivers/misc/mei/hw-txe.c:744:16: warning: comparison between signed and unsigned integer expressions [-Wsign-compare]

Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agomei: check for error returned from mei_hbuf_empty_slots()
Tomas Winkler [Thu, 12 Jul 2018 14:10:08 +0000 (17:10 +0300)]
mei: check for error returned from mei_hbuf_empty_slots()

mei_hbuf_empty_slots() may return with an error in case
of circular buffer overflow. This type of error may
be caused only by a bug. However currently, the error
won't be detected due signed type promotion in comparison to u32.
We add explicit check for less then zero and explicit cast
in comparison to suppress singn-compare warning.

Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agomei: don't update offset in write
Alexander Usyskin [Mon, 9 Jul 2018 09:21:44 +0000 (12:21 +0300)]
mei: don't update offset in write

MEI enables writes of complete messages only
while read can be performed in parts, hence
write should not update the file offset to
not break interleaving partial reads with writes.

Cc: <stable@vger.kernel.org>
Signed-off-by: Alexander Usyskin <alexander.usyskin@intel.com>
Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agomei: bus: type promotion bug in mei_fwver()
Tomas Winkler [Wed, 11 Jul 2018 12:29:32 +0000 (15:29 +0300)]
mei: bus: type promotion bug in mei_fwver()

if __mei_cl_recv() returns a negative then "bytes_recv"
type is promoted to a high positive value in comparison with
size_t evaluated by MKHI_FWVER_LEN(1). It results in error condition
not to be detected.

Cc: Dan Carpenter <dan.carpenter@oracle.com>
Fixes: 9078ad92ef86 ("mei: expose fw version to sysfs")
Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agomei: bus: type promotion bug in mei_nfc_if_version()
Dan Carpenter [Wed, 11 Jul 2018 12:29:31 +0000 (15:29 +0300)]
mei: bus: type promotion bug in mei_nfc_if_version()

We accidentally removed the check for negative returns
without considering the issue of type promotion.
The "if_version_length" variable is type size_t so if __mei_cl_recv()
returns a negative then "bytes_recv" is type promoted
to a high positive value and treated as success.

Cc: <stable@vger.kernel.org>
Fixes: 582ab27a063a ("mei: bus: fix received data size check in NFC fixup")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agoMerge tag 'fsi-updates-2018-07-12' of git://git.kernel.org/pub/scm/linux/kernel/git...
Greg Kroah-Hartman [Thu, 12 Jul 2018 06:42:09 +0000 (08:42 +0200)]
Merge tag 'fsi-updates-2018-07-12' of git://git./linux/kernel/git/benh/linux-fsi into char-misc-next

Ben writes:

FSI fixes and updates:

 - Reported build fixes
 - Add configuration of send/echo delayus
 - Object lifetime fix
 - Re-arrange some definitions in preparation for adding the CF master

6 years agofsi: Move various master definitions to a common header
Benjamin Herrenschmidt [Sun, 10 Jun 2018 06:25:25 +0000 (16:25 +1000)]
fsi: Move various master definitions to a common header

This moves the definitions for various protocol details
(message & response codes, delays etc...) out of
fsi-master-gpio.c to fsi-master.h in order to share them
with other master implementations.

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Reviewed-by: Joel Stanley <joel@jms.id.au>
6 years agofsi: master-gpio: Add missing release function
Benjamin Herrenschmidt [Wed, 13 Jun 2018 00:05:17 +0000 (10:05 +1000)]
fsi: master-gpio: Add missing release function

The embedded struct device needs a release function to be
able to successfully remove the driver.

We remove the devm_gpiod_put() as they are unnecessary
(the resources will be released automatically) and because
fsi_master_unregister() will cause the master structure to
be freed.

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Reviewed-by: Joel Stanley <joel@jms.id.au>
6 years agofsi: Don't use device_unregister() in fsi_master_register()
Benjamin Herrenschmidt [Thu, 28 Jun 2018 06:26:19 +0000 (16:26 +1000)]
fsi: Don't use device_unregister() in fsi_master_register()

In the error path of fsi_master_register(), we currently
use device_unregister(). This will cause the last reference
to the structure to be dropped, thus freeing the enclosing
structure, which isn't what the callers want.

Use device_del() instead so that we return to the caller
with a refcount of 1. The caller can then assume that it
must use put_device() after a call to fsi_master_register()
regardless of whether the latter suceeded or failed.

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
6 years agofsi: master-gpio: Remove "GPIO" prefix on some definitions
Benjamin Herrenschmidt [Thu, 12 Jul 2018 02:04:24 +0000 (12:04 +1000)]
fsi: master-gpio: Remove "GPIO" prefix on some definitions

Some definitions are generic to the FSI protocol or any
give master implementation. Rename them to remove the
"GPIO" prefix in preparation for moving them to a common
header.

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Reviewed-by: Joel Stanley <joel@jms.id.au>
# Conflicts:
# drivers/fsi/fsi-master-gpio.c

6 years agofsi: master-gpio: Remove unused definitions
Benjamin Herrenschmidt [Sun, 10 Jun 2018 06:20:18 +0000 (16:20 +1000)]
fsi: master-gpio: Remove unused definitions

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Reviewed-by: Joel Stanley <joel@jms.id.au>
6 years agofsi: master-gpio: Add more tracepoints
Benjamin Herrenschmidt [Tue, 29 May 2018 09:28:38 +0000 (19:28 +1000)]
fsi: master-gpio: Add more tracepoints

This adds a few more tracepoints that have proven useful when
debugging issues with the FSI bus.

This also makes echo_delay() use clock_zeros() instead of
open-code it in order to share the tracepoint.

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Reviewed-by: Joel Stanley <joel@jms.id.au>
6 years agofsi: master-gpio: Add support for link_config
Benjamin Herrenschmidt [Tue, 29 May 2018 05:11:32 +0000 (15:11 +1000)]
fsi: master-gpio: Add support for link_config

To configure the send and echo delays

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Reviewed-by: Joel Stanley <joel@jms.id.au>
6 years agofsi: master-gpio: Rename and adjust send delay
Benjamin Herrenschmidt [Tue, 29 May 2018 05:09:45 +0000 (15:09 +1000)]
fsi: master-gpio: Rename and adjust send delay

What the driver called "FSI_GPIO_PRIME_SLAVE_CLOCKS" is what
the FSI spec calls tSendDelay and should be 16 clocks by
default.

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Reviewed-by: Joel Stanley <joel@jms.id.au>
6 years agofsi: Add mechanism to set the tSendDelay and tEchoDelay values
Benjamin Herrenschmidt [Tue, 29 May 2018 05:01:07 +0000 (15:01 +1000)]
fsi: Add mechanism to set the tSendDelay and tEchoDelay values

Those values control the amount of "dummy" clocks between commands and
between a command and its response.

This adds a way to configure them from sysfs (to be later extended to
defaults in the device-tree). The default remains 16 (the HW default).

This is only supported if the backend supports the new link_config()
callback to configure the generation of those delays.

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Reviewed-by: Joel Stanley <joel@jms.id.au>
---

6 years agofsi: Move code around to avoid forward declaration
Benjamin Herrenschmidt [Tue, 29 May 2018 04:44:08 +0000 (14:44 +1000)]
fsi: Move code around to avoid forward declaration

Move fsi_slave_set_smode() and its helpers to before it's
first user and remove the corresponding forward declaration.

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Reviewed-by: Joel Stanley <joel@jms.id.au>
6 years agofsi: sbefifo: Fix checker warning about late NULL check
Benjamin Herrenschmidt [Thu, 12 Jul 2018 01:53:37 +0000 (11:53 +1000)]
fsi: sbefifo: Fix checker warning about late NULL check

"dev" is dereferences before it's checked.

Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
6 years agofsi/sbefifo: Add dependency on OF_ADDRESS
Guenter Roeck [Tue, 3 Jul 2018 13:43:49 +0000 (06:43 -0700)]
fsi/sbefifo: Add dependency on OF_ADDRESS

The driver calls of_platform_device_create() which is only available
if OF_ADDRESS is enabled. When building sparc64 images, this results
in

ERROR: "of_platform_device_create" [drivers/fsi/fsi-sbefifo.ko] undefined!

Fixes: 9f4a8a2d7f9d ("fsi/sbefifo: Add driver for the SBE FIFO")
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
6 years agofsi: sbefifo: Add missing mutex_unlock
Eddie James [Tue, 3 Jul 2018 21:21:29 +0000 (16:21 -0500)]
fsi: sbefifo: Add missing mutex_unlock

There was no unlock of the FFDC mutex.

Fixes: 9f4a8a2d7f9d ("fsi/sbefifo: Add driver for the SBE FIFO")
Signed-off-by: Eddie James <eajames@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
6 years agoextcon: max3355: include mod_devicetable.h
Arnd Bergmann [Wed, 11 Jul 2018 15:50:29 +0000 (17:50 +0200)]
extcon: max3355: include mod_devicetable.h

Another driver turned up that is missing linux/mod_devicetable.h after
the device IDs are split out from linux/platform_device.h:

drivers/extcon/extcon-max3355.c:127:34: error: array type has incomplete element type 'struct of_device_id'
 static const struct of_device_id max3355_match_table[] = {

Fixes: ac3167257b9f ("headers: separate linux/mod_devicetable.h from linux/platform_device.h")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agoheaders: fix linux/mod_devicetable.h inclusions
Arnd Bergmann [Mon, 9 Jul 2018 15:19:02 +0000 (17:19 +0200)]
headers: fix linux/mod_devicetable.h inclusions

A couple of drivers produced build errors after the mod_devicetable.h
header was split out from the platform_device one, e.g.

drivers/media/platform/davinci/vpbe_osd.c:42:40: error: array type has incomplete element type 'struct platform_device_id'
drivers/media/platform/davinci/vpbe_venc.c:42:40: error: array type has incomplete element type 'struct platform_device_id'

This adds the inclusion where needed.

Fixes: ac3167257b9f ("headers: separate linux/mod_devicetable.h from linux/platform_device.h")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agoDrivers: HV: Send one page worth of kmsg dump over Hyper-V during panic
Sunil Muthuswamy [Sun, 8 Jul 2018 02:56:51 +0000 (02:56 +0000)]
Drivers: HV: Send one page worth of kmsg dump over Hyper-V during panic

In the VM mode on Hyper-V, currently, when the kernel panics, an error
code and few register values are populated in an MSR and the Hypervisor
notified. This information is collected on the host. The amount of
information currently collected is found to be limited and not very
actionable. To gather more actionable data, such as stack trace, the
proposal is to write one page worth of kmsg data on an allocated page
and the Hypervisor notified of the page address through the MSR.

- Sysctl option to control the behavior, with ON by default.

Cc: K. Y. Srinivasan <kys@microsoft.com>
Cc: Stephen Hemminger <sthemmin@microsoft.com>
Signed-off-by: Sunil Muthuswamy <sunilmut@microsoft.com>
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agoslimbus: stream: Fix htmldocs warnings
Srinivas Kandagatla [Sat, 7 Jul 2018 19:28:41 +0000 (20:28 +0100)]
slimbus: stream: Fix htmldocs warnings

This patch fixes below warning during building htmldoc:
slimbus.h:352: warning: Function parameter or member 'name'
 not described in 'slim_stream_runtime'

This patch also removes documentation for state variable
in struct slim_stream_runtime which was redundant and removed.

Reported-by: kbuild test robot <lkp@intel.com>
Signed-off-by: Srinivas Kandagatla <srinivas.kandagatla@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agoheaders: fix build error in arch/arm/plat-samsung/adc.c, add <linux/mod_devicetable.h>
Randy Dunlap [Sat, 7 Jul 2018 21:03:29 +0000 (14:03 -0700)]
headers: fix build error in arch/arm/plat-samsung/adc.c, add <linux/mod_devicetable.h>

arch/arm/plat-samsung/adc.c needs to #include <linux/mod_devicetable.h>
explicitly since that header file was removed from
<linux/platform_device.h> (it wasn't needed there).

Reported-by: kbuild test robot <lkp@intel.com>
Signed-off-by: Randy Dunlap <rdunlap@infradead.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agoheaders: separate linux/mod_devicetable.h from linux/platform_device.h
Randy Dunlap [Wed, 20 Jun 2018 05:47:28 +0000 (22:47 -0700)]
headers: separate linux/mod_devicetable.h from linux/platform_device.h

At over 4000 #includes, <linux/platform_device.h> is the 9th most
#included header file in the Linux kernel.  It does not need
<linux/mod_devicetable.h>, so drop that header and explicitly add
<linux/mod_devicetable.h> to source files that need it.

   4146 #include <linux/platform_device.h>

After this patch, there are 225 files that use <linux/mod_devicetable.h>,
for a reduction of around 3900 times that <linux/mod_devicetable.h>
does not have to be read & parsed.

    225 #include <linux/mod_devicetable.h>

This patch was build-tested on 20 different arch-es.

It also makes these drivers SubmitChecklist#1 compliant.

Signed-off-by: Randy Dunlap <rdunlap@infradead.org>
Reported-by: kbuild test robot <lkp@intel.com> # drivers/media/platform/vimc/
Reported-by: kbuild test robot <lkp@intel.com> # drivers/pinctrl/pinctrl-u300.c
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agomisc: ti-st: Allow compile test of GPIO consumers if !GPIOLIB
Geert Uytterhoeven [Thu, 21 Jun 2018 19:00:16 +0000 (21:00 +0200)]
misc: ti-st: Allow compile test of GPIO consumers if !GPIOLIB

The GPIO subsystem provides dummy GPIO consumer functions if GPIOLIB is
not enabled. Hence drivers that depend on GPIOLIB, but use GPIO consumer
functionality only, can still be compiled if GPIOLIB is not enabled.

Relax the dependency on GPIOLIB if COMPILE_TEST is enabled, where
appropriate.

Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
Acked-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agoandroid: binder: Drop dependency on !M68K
Geert Uytterhoeven [Wed, 6 Jun 2018 12:40:56 +0000 (14:40 +0200)]
android: binder: Drop dependency on !M68K

As of commit 7124330dabe5b3cb ("m68k/uaccess: Revive 64-bit
get_user()"), the 64-bit Android binder interface builds fine on m68k.

Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agover_linux: Do not check for ver_linux pattern in version function
Alexander Kapshuk [Thu, 31 May 2018 19:22:47 +0000 (22:22 +0300)]
ver_linux: Do not check for ver_linux pattern in version function

Checking whether output of commands matches the ver_linux pattern in
the version function is original shell implementation legacy code. When
the original implementation failed to locate a particular utility,
it generated error output along the lines of:

ver_linux:line number: command not found.

The awk implementation, does not contain the name of the script within the
body of the error message returned by the subshell when a given utility
fails to be located. The error message returned is along the lines of:

sh: name of utility: command not found

Safeguarding against the ver_linux pattern being found in the output
being parsed may thus be safely omitted.

Signed-off-by: Alexander Kapshuk <alexander.kapshuk@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agover_linux: Process input coming from procmaps that matches libc only
Alexander Kapshuk [Thu, 31 May 2018 19:22:46 +0000 (22:22 +0300)]
ver_linux: Process input coming from procmaps that matches libc only

Currently, input coming from /proc/self/maps is split into fields without
checking whether or not it matches libc.so. This is not efficient.
All text processing should only be performed on lines of input that
match libc.so.

Signed-off-by: Alexander Kapshuk <alexander.kapshuk@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agotsl2550: fix lux1_input error in low light
Matt Ranostay [Sat, 9 Jun 2018 06:58:15 +0000 (23:58 -0700)]
tsl2550: fix lux1_input error in low light

ADC channel 0 photodiode detects both infrared + visible light,
but ADC channel 1 just detects infrared. However, the latter is a bit
more sensitive in that range so complete darkness or low light causes
a error condition in which the chan0 - chan1 is negative that
results in a -EAGAIN.

This patch changes the resulting lux1_input sysfs attribute message from
"Resource temporarily unavailable" to a user-grokable lux value of 0.

Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Matt Ranostay <matt.ranostay@konsulko.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agomisc: mic: fix passing the current time
Arnd Bergmann [Mon, 18 Jun 2018 14:25:51 +0000 (16:25 +0200)]
misc: mic: fix passing the current time

I noticed that the mic driver passes a 'struct timespec64' as part of
a message into an attached device, where it is used to set the current
system time.

This won't actually work if one of the two sides runs a 32-bit kernel and
the other runs a 64-bit kernel, since the structure layout is different
between the two.

I found this while replacing calls to the deprecated do_settimeofday64()
interface with the modern ktime_get_real_ts() variant, but it seems
appropriate to address both at the same time here.

To make sure we have a sane structure, let's define our own structure
using the layout of the 64-bit kernel.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agosiox: don't create a thread without starting it
Uwe Kleine-König [Thu, 28 Jun 2018 07:57:42 +0000 (09:57 +0200)]
siox: don't create a thread without starting it

When a siox master device is registered a kthread is created that is
only started when triggered by userspace. So this thread might be in
TASK_UNINTERRUPTIBLE state for long and trigger a warning

[  241.130465] INFO: task siox-0:626 blocked for more than 120 seconds.

with the respective debug settings enabled. It might be right to put an
unstarted thread to TASK_IDLE (in kernel/kthread.c:kthread()) instead,
but independant of this discussion it is cleaner for
siox_master_register() to start the thread immediately. The effect is
that it enters its own waiting state and then stays in state TASK_IDLE
which doesn't trigger the above warning.

As siox_poll_thread() uses some variables of the device the
initialisation of these is moved before thread creation.

Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Acked-by: Gavin Schenk <g.schenk@eckelmann.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agosiox: treat type errors as status errors
Uwe Kleine-König [Tue, 19 Jun 2018 09:38:54 +0000 (11:38 +0200)]
siox: treat type errors as status errors

The type bits are part of the per-device status word. So it's natural to
consider an error in the type bits as a status error instead of only
resulting in an unsynced state.

Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Acked-by: Gavin Schenk <g.schenk@eckelmann.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agomisc: mic: mark expected switch fall-through
Gustavo A. R. Silva [Tue, 3 Jul 2018 22:36:29 +0000 (17:36 -0500)]
misc: mic: mark expected switch fall-through

In preparation to enabling -Wimplicit-fallthrough, mark switch cases
where we are expecting to fall through.

Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agosgi-xp: xpc_partition: mark expected switch fall-throughs
Gustavo A. R. Silva [Tue, 3 Jul 2018 22:43:43 +0000 (17:43 -0500)]
sgi-xp: xpc_partition: mark expected switch fall-throughs

In preparation to enabling -Wimplicit-fallthrough, mark switch cases
where we are expecting to fall through.

Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agoeeprom: at25: mark expected switch fall-throughs
Gustavo A. R. Silva [Tue, 3 Jul 2018 22:47:36 +0000 (17:47 -0500)]
eeprom: at25: mark expected switch fall-throughs

In preparation to enabling -Wimplicit-fallthrough, mark switch cases
where we are expecting to fall through.

Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agomisc: ibmvmc: Use GFP_ATOMIC under spin lock
Wei Yongjun [Sat, 26 May 2018 09:45:59 +0000 (09:45 +0000)]
misc: ibmvmc: Use GFP_ATOMIC under spin lock

The function alloc_dma_buffer() is called from ibmvmc_add_buffer(),
in which a spin lock be held here, so we should use GFP_ATOMIC when
a lock is held.

Fixes: 0eca353e7ae7 ("misc: IBM Virtual Management Channel Driver (VMC)")
Signed-off-by: Wei Yongjun <weiyongjun1@huawei.com>
Reviewed-by: Bryant G. Ly <bryantly@linux.vnet.ibm.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years ago/dev/mem: Mark expected switch fall-through
Gustavo A. R. Silva [Mon, 2 Jul 2018 17:57:24 +0000 (12:57 -0500)]
/dev/mem: Mark expected switch fall-through

In preparation to enabling -Wimplicit-fallthrough, mark switch cases
where we are expecting to fall through.

Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agovmci: type promotion bug in qp_host_get_user_memory()
Dan Carpenter [Wed, 4 Jul 2018 09:33:34 +0000 (12:33 +0300)]
vmci: type promotion bug in qp_host_get_user_memory()

The problem is that if get_user_pages_fast() fails and returns a
negative error code, it gets type promoted to a high positive value and
treated as a success.

Fixes: 06164d2b72aa ("VMCI: queue pairs implementation.")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agonvmem: core: remove unused nvmem_device ncells member
Srinivas Kandagatla [Tue, 26 Jun 2018 11:36:40 +0000 (12:36 +0100)]
nvmem: core: remove unused nvmem_device ncells member

nvmem ncells can be over written by calling nvmem_add_cells()
multiple times. I see there is no real point of maintaining count
of cells when we have a list of cell.

Remove this to avoid any confusion!

Signed-off-by: Srinivas Kandagatla <srinivas.kandagatla@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agonvmem: rave-sp-eeprom: Remove VLA usage
Kees Cook [Wed, 20 Jun 2018 18:26:00 +0000 (11:26 -0700)]
nvmem: rave-sp-eeprom: Remove VLA usage

In the quest to remove all stack VLA usage from the kernel[1], this
uses the maximum allocation size for the stack and adds a sanity check,
similar to what has already be done for the regular rave-sp driver.

[1] https://lkml.kernel.org/r/CA+55aFzCG-zNmZwX4A2FQpadafLfEzK6CC=qPXydAacU1RqZWA@mail.gmail.com

Signed-off-by: Kees Cook <keescook@chromium.org>
Reviewed-by: Andrey Smirnov <andrew.smirnov@gmail.com>
Tested-by: Andrey Smirnov <andrew.smirnov@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agoMAINTAINERS: Add file patterns for w1 device tree bindings
Geert Uytterhoeven [Fri, 22 Jun 2018 10:08:20 +0000 (12:08 +0200)]
MAINTAINERS: Add file patterns for w1 device tree bindings

Submitters of device tree binding documentation may forget to CC
the subsystem maintainer if this is missing.

Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agow1: fix w1_ds2438 documentation
Mariusz Bialonczyk [Thu, 5 Jul 2018 04:51:36 +0000 (06:51 +0200)]
w1: fix w1_ds2438 documentation

The previous documentation was wrongly stating about the order
of magnitude of CONVERT_V result files contents (vad, vdd).
This commit is correcting this.

Reported-by: Adam Stolarczyk <adam@stolarczyk.net.pl>
Signed-off-by: Mariusz Bialonczyk <manio@skyboo.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agods2490.c: Move struct open braces to previous line
Henriette Hofmeier [Thu, 14 Jun 2018 15:20:41 +0000 (17:20 +0200)]
ds2490.c: Move struct open braces to previous line

Move open braces of two structs to the declaration line,
as criticized by checkpatch.

Signed-off-by: Henriette Hofmeier <passt@h-hofmeier.de>
Signed-off-by: Florian Harbecke <florian.harbecke@fau.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agods2490.c: Remove unnecessary whitespace
Henriette Hofmeier [Thu, 14 Jun 2018 15:20:40 +0000 (17:20 +0200)]
ds2490.c: Remove unnecessary whitespace

Remove unnecessary whitespace criticized by
checkpatch.

Signed-off-by: Henriette Hofmeier <passt@h-hofmeier.de>
Signed-off-by: Florian Harbecke <florian.harbecke@fau.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agods2490.c: Add required spaces
Henriette Hofmeier [Thu, 14 Jun 2018 15:20:39 +0000 (17:20 +0200)]
ds2490.c: Add required spaces

Add missing spaces in for- and while-loops
reported missing by checkpatch.

Signed-off-by: Henriette Hofmeier <passt@h-hofmeier.de>
Signed-off-by: Florian Harbecke <florian.harbecke@fau.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agoslimbus: ngd: add stream support
Srinivas Kandagatla [Thu, 5 Jul 2018 13:54:26 +0000 (14:54 +0100)]
slimbus: ngd: add stream support

This patch adds support to stream support, this involve implementing
user specific implementation of Data channel management and channel
management SLIMbus messages.

Signed-off-by: Srinivas Kandagatla <srinivas.kandagatla@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agoslimbus: stream: add stream support
Srinivas Kandagatla [Thu, 5 Jul 2018 13:54:25 +0000 (14:54 +0100)]
slimbus: stream: add stream support

This patch adds support to SLIMbus stream apis for slimbus device.
SLIMbus streaming involves adding support to Data Channel Management and
channel Reconfiguration Messages to slim core plus few stream apis.
>From slim device side the apis are very simple mostly inline with other
stream apis.

Currently it only supports Isochronous and Push/Pull transport protocols,
which are sufficient for audio use cases.

Signed-off-by: Srinivas Kandagatla <srinivas.kandagatla@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agoslimbus: ngd: Add qcom SLIMBus NGD driver
Srinivas Kandagatla [Tue, 19 Jun 2018 16:13:01 +0000 (17:13 +0100)]
slimbus: ngd: Add qcom SLIMBus NGD driver

This patch adds suppor to Qualcomm SLIMBus Non-Generic Device (NGD)
controller driver.
This is light-weight SLIMBus controller driver responsible for
communicating with slave HW directly over the bus using messaging
interface, and communicating with master component residing on ADSP
for bandwidth and data-channel management

Based on intial work from
Karthikeyan Ramasubramanian <kramasub@codeaurora.org> and
Sagar Dharia <sdharia@codeaurora.org>

Signed-off-by: Srinivas Kandagatla <srinivas.kandagatla@linaro.org>
Tested-by: Craig Tatlor <ctatlor97@gmail.com>
Reviewed-by: Vinod Koul <vkoul@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agoslimbus: ngd: dt-bindings: Add slim ngd dt bindings
Srinivas Kandagatla [Tue, 19 Jun 2018 16:13:00 +0000 (17:13 +0100)]
slimbus: ngd: dt-bindings: Add slim ngd dt bindings

This patch adds bindings for Qualcomm SLIMBus NGD controller.
SLIMBus NGD controller is a light-weight driver responsible for
communicating with SLIMBus slaves directly over the bus using messaging
interface and communicating with master component residing on ADSP for
bandwidth and data-channel management

Signed-off-by: Srinivas Kandagatla <srinivas.kandagatla@linaro.org>
Reviewed-by: Rob Herring <robh@kernel.org>
Tested-by: Craig Tatlor <ctatlor97@gmail.com>
Reviewed-by: Vinod Koul <vkoul@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agoslimbus: messaging: add slim_alloc/free_txn_tid()
Srinivas Kandagatla [Tue, 19 Jun 2018 16:12:59 +0000 (17:12 +0100)]
slimbus: messaging: add slim_alloc/free_txn_tid()

This patch adds slim_alloc_txn_tid() and slim_free_txn_tid() api
to allow controllers like ngd to allocate tids for user specific
commands. This also cleans up the existing code to use single place
for tid allocations and free.

This patch also make the tid allocation cyclic one, its very useful
to track the transactions back during debug.

Signed-off-by: Srinivas Kandagatla <srinivas.kandagatla@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agoslimbus: core: rearrange slim_eaddr structure
Srinivas Kandagatla [Tue, 19 Jun 2018 16:12:58 +0000 (17:12 +0100)]
slimbus: core: rearrange slim_eaddr structure

Rearrange struct slim_eaddr so that the structure is packed correctly
to be able to send in SLIMBus messages.

Signed-off-by: Srinivas Kandagatla <srinivas.kandagatla@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agoslimbus: core: add of_slim_device_get() helper
Srinivas Kandagatla [Tue, 19 Jun 2018 16:12:57 +0000 (17:12 +0100)]
slimbus: core: add of_slim_device_get() helper

On SLIMBus controllers like Qcom NGD(non ported device), controller
can request logical address once the remote side is powered, having a
helper function like this to explicitly enumerate the bus is helpful.
Also codec drivers which are taking to interface device would need
such a helper too.

Signed-off-by: Srinivas Kandagatla <srinivas.kandagatla@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agoslimbus: qcom: remove redundant depends in Kconfig
Srinivas Kandagatla [Tue, 19 Jun 2018 15:12:05 +0000 (16:12 +0100)]
slimbus: qcom: remove redundant depends in Kconfig

QCOM SLIMBus controller is already under a 'if SLIMBUS' in Kconfig,
having depends on SLIMBUS is totally redundant. Just remove it.

Signed-off-by: Srinivas Kandagatla <srinivas.kandagatla@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agoslimbus: messaging: initialize completion correctly
Srinivas Kandagatla [Tue, 19 Jun 2018 15:12:04 +0000 (16:12 +0100)]
slimbus: messaging: initialize completion correctly

slim_val_inf can contain random value from stack, make sure the completion
is initialized to NULL while filling the msg.

Signed-off-by: Srinivas Kandagatla <srinivas.kandagatla@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agoslimbus: messaging: remove multiple calls to pm_runtime_mark_last_busy
Srinivas Kandagatla [Tue, 19 Jun 2018 15:12:03 +0000 (16:12 +0100)]
slimbus: messaging: remove multiple calls to pm_runtime_mark_last_busy

There seems to be a multiple calls to pm_runtime_mark_last_busy(),
which looks like a typo.
Fix this by properly adding pm_runtime_put_autosuspend to put controller
in auto suspend state.

Signed-off-by: Srinivas Kandagatla <srinivas.kandagatla@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agoslimbus: messaging: pass correct wbuf
Srinivas Kandagatla [Tue, 19 Jun 2018 15:12:02 +0000 (16:12 +0100)]
slimbus: messaging: pass correct wbuf

There seems to be a typo while filling msg for slim_write, wbuf is
set to NULL instead of rbuf.

Signed-off-by: Srinivas Kandagatla <srinivas.kandagatla@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agoipack: add error handling for ioremap_nocache
Zhouyang Jia [Tue, 12 Jun 2018 03:07:21 +0000 (11:07 +0800)]
ipack: add error handling for ioremap_nocache

When ioremap_nocache fails, the lack of error-handling code may
cause unexpected results.

This patch adds error-handling code after calling ioremap_nocache.

Signed-off-by: Zhouyang Jia <jiazhouyang09@gmail.com>
Acked-by: Samuel Iglesias Gonsalvez <siglesias@igalia.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agoMerge tag 'fsi-updates-2018-06-12' of https://git.kernel.org/pub/scm/linux/kernel...
Greg Kroah-Hartman [Sat, 7 Jul 2018 15:02:35 +0000 (17:02 +0200)]
Merge tag 'fsi-updates-2018-06-12' of https://git./linux/kernel/git/benh/linux-fsi into char-misc-next

Ben writes:

FSI updates and sbefifo driver

6 years agoDocumentation: fpga: cleanup
Alan Tull [Thu, 28 Jun 2018 01:56:19 +0000 (20:56 -0500)]
Documentation: fpga: cleanup

Minor fixes including:

* fix some typos
* correct use of a/an
* rephrase explanation of .state ops function
* s/re-use/reuse/ (use only one spelling of 'reuse' in these docs)
* s/cpu/CPU/

Signed-off-by: Alan Tull <atull@kernel.org>
Acked-by: Randy Dunlap <rdunlap@infradead.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agodt-bindings: vendor-prefixes: fix bad merge
Johan Hovold [Wed, 4 Jul 2018 06:46:26 +0000 (08:46 +0200)]
dt-bindings: vendor-prefixes: fix bad merge

Remove the vendor-prefix reject file which was accidentally added when
merging the gnss sirfstar binding. The wi2wi prefix had already been
added by commit a593bff82cae2 ("dt-bindings: define vendor prefix for
Wi2Wi, Inc.").

Fixes: 176193b7dd6e ("dt-bindings: gnss: add sirfstar binding")
Reported-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Johan Hovold <johan@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agomisc: sram: enable clock before registering regions
Johan Hovold [Tue, 3 Jul 2018 10:05:48 +0000 (12:05 +0200)]
misc: sram: enable clock before registering regions

Make sure to enable the clock before registering regions and exporting
partitions to user space at which point we must be prepared for I/O.

Fixes: ee895ccdf776 ("misc: sram: fix enabled clock leak on error path")
Signed-off-by: Johan Hovold <johan@kernel.org>
Reviewed-by: Vladimir Zapolskiy <vladimir_zapolskiy@mentor.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agomisc: sram: fix resource leaks in probe error path
Johan Hovold [Tue, 3 Jul 2018 10:05:47 +0000 (12:05 +0200)]
misc: sram: fix resource leaks in probe error path

Make sure to disable clocks and deregister any exported partitions
before returning on late probe errors.

Note that since commit ee895ccdf776 ("misc: sram: fix enabled clock leak
on error path"), partitions are deliberately exported before enabling
the clock so we stick to that logic here. A follow up patch will address
this.

Fixes: 2ae2e28852f2 ("misc: sram: add Atmel securam support")
Cc: stable <stable@vger.kernel.org> # 4.9
Cc: Alexandre Belloni <alexandre.belloni@free-electrons.com>
Signed-off-by: Johan Hovold <johan@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agomei: expose fw version to sysfs
Alexander Usyskin [Sun, 24 Jun 2018 21:11:41 +0000 (00:11 +0300)]
mei: expose fw version to sysfs

The ME FW version is constantly used by detection and update tools.
To improve the reliability and simplify these tools provide
a sysfs interface to access version of the platform ME firmware
in the following format:
<platform>:<major>.<minor>.<milestone>.<build>.
There can be up to three such blocks for different FW components.

Signed-off-by: Alexander Usyskin <alexander.usyskin@intel.com>
Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agomei: add optional timeout to internal bus recv
Alexander Usyskin [Sun, 24 Jun 2018 21:11:40 +0000 (00:11 +0300)]
mei: add optional timeout to internal bus recv

Add optional timeout to internal bus recv function to
enable break out of internal flows in case of no answer from FW.

Signed-off-by: Alexander Usyskin <alexander.usyskin@intel.com>
Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agomei: remove unused MEI_IAMTHIF_STALL_TIMER define
Alexander Usyskin [Wed, 6 Jun 2018 21:31:47 +0000 (00:31 +0300)]
mei: remove unused MEI_IAMTHIF_STALL_TIMER define

MEI_IAMTHIF_STALL_TIMER is unused now and can be safely removed.

Signed-off-by: Alexander Usyskin <alexander.usyskin@intel.com>
Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agoDrivers: hv: vmbus: Make TLFS #define names architecture neutral
Michael Kelley [Tue, 5 Jun 2018 20:37:54 +0000 (13:37 -0700)]
Drivers: hv: vmbus: Make TLFS #define names architecture neutral

The Hyper-V feature and hint flags in hyperv-tlfs.h are all defined
with the string "X64" in the name.  Some of these flags are indeed
x86/x64 specific, but others are not.  For the ones that are used
in architecture independent Hyper-V driver code, or will be used in
the upcoming support for Hyper-V for ARM64, this patch removes the
"X64" from the name.

This patch changes the flags that are currently known to be
used on multiple architectures. Hyper-V for ARM64 is still a
work-in-progress and the Top Level Functional Spec (TLFS) has not
been separated into x86/x64 and ARM64 areas.  So additional flags
may need to be updated later.

This patch only changes symbol names.  There are no functional
changes.

Signed-off-by: Michael Kelley <mikelley@microsoft.com>
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agoTools: hv: vss: fix loop device detection
Vitaly Kuznetsov [Tue, 5 Jun 2018 20:37:56 +0000 (13:37 -0700)]
Tools: hv: vss: fix loop device detection

Commit ea81fdf0981d ("Tools: hv: vss: Skip freezing filesystems backed by
loop") added skip for filesystems backed by loop device. However, it seems
the detection of such cases is incomplete.

It was found that with 'devicemapper' storage driver docker creates the
following chain:

NAME MAJ:MIN
loop0 7:0
..docker-8:4-8473394-pool 253:0
  ..docker-8:4-8473394-eac... 253:1

so when we're looking at the mounted device we see major '253' and not '7'.

Solve the issue by walking /sys/dev/block/*/slaves chain and checking if
there's a loop device somewhere.

Other than that, don't skip mountpoints silently when stat() fails. In case
e.g. SELinux is failing stat we don't want to skip freezing everything
without letting user know about the failure.

Fixes: ea81fdf0981d ("Tools: hv: vss: Skip freezing filesystems backed by loop")
Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agotools: hv: update lsvmbus to be compatible with python3
Olaf Hering [Tue, 5 Jun 2018 20:37:55 +0000 (13:37 -0700)]
tools: hv: update lsvmbus to be compatible with python3

Python3 changed the way how 'print' works.
Adjust the code to a syntax that is understood by python2 and python3.

Signed-off-by: Olaf Hering <olaf@aepfle.de>
Acked-by: Dexuan Cui <decui@microsoft.com>
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agoDrivers: hv: vmbus: Remove x86 MSR refs in arch independent code
Michael Kelley [Tue, 5 Jun 2018 20:37:53 +0000 (13:37 -0700)]
Drivers: hv: vmbus: Remove x86 MSR refs in arch independent code

In architecture independent code for manipulating Hyper-V synthetic timers
and synthetic interrupts, pass in an ordinal number identifying the timer
or interrupt, rather than an actual MSR register address.  Then in
x86/x64 specific code, map the ordinal number to the appropriate MSR.
This change facilitates the introduction of an ARM64 version of Hyper-V,
which uses the same synthetic timers and interrupts, but a different
mechanism for accessing them.

Signed-off-by: Michael Kelley <mikelley@microsoft.com>
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agoDrivers: hv: vmbus: Fix the offer_in_progress in vmbus_process_offer()
Dexuan Cui [Tue, 5 Jun 2018 20:37:52 +0000 (13:37 -0700)]
Drivers: hv: vmbus: Fix the offer_in_progress in vmbus_process_offer()

I didn't really hit a real bug, but just happened to spot the bug:
we have decreased the counter at the beginning of vmbus_process_offer(),
so we mustn't decrease it again.

Fixes: 6f3d791f3006 ("Drivers: hv: vmbus: Fix rescind handling issues")
Signed-off-by: Dexuan Cui <decui@microsoft.com>
Cc: stable@vger.kernel.org
Cc: Stephen Hemminger <sthemmin@microsoft.com>
Cc: K. Y. Srinivasan <kys@microsoft.com>
Cc: Stable <stable@vger.kernel.org> # 4.14 and above
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>