to #27305291

commit d6548ae4d16dc231dec22860c9c472bcb991fb15 upstream.

The generated dimm id is needed for the sysfs attribute as well as being
used as the identifier/description for the security key. Since it's
constant and should never change, store it as a member of struct nvdimm.

As nvdimm_create() continues to grow parameters relative to NFIT driver
requirements, do not require other implementations to keep pace.
Introduce __nvdimm_create() to carry the new parameters and keep
nvdimm_create() with the long standing default api.
Signed-off-by: NDave Jiang <dave.jiang@intel.com>
Signed-off-by: NDan Williams <dan.j.williams@intel.com>

[ Shile: fixed conflict in drivers/acpi/nfit/nfit.h ]
Signed-off-by: NShile Zhang <shile.zhang@linux.alibaba.com>
Reviewed-by: NYang Shi <yang.shi@linux.alibaba.com>

to #27305291

commit b3ed2ce024c36054e51cca2eb31a1cdbe4a5f11e upstream.

Add command definition for security commands defined in Intel DSM
specification v1.8 [1]. This includes "get security state", "set
passphrase", "unlock unit", "freeze lock", "secure erase", "overwrite",
"overwrite query", "master passphrase enable/disable", and "master
erase", . Since this adds several Intel definitions, move the relevant
bits to their own header.

These commands mutate physical data, but that manipulation is not cache
coherent. The requirement to flush and invalidate caches makes these
commands unsuitable to be called from userspace, so extra logic is added
to detect and block these commands from being submitted via the ioctl
command submission path.

Lastly, the commands may contain sensitive key material that should not
be dumped in a standard debug session. Update the nvdimm-command
payload-dump facility to move security command payloads behind a
default-off compile time switch.

[1]: http://pmem.io/documents/NVDIMM_DSM_Interface-V1.8.pdfSigned-off-by: NDave Jiang <dave.jiang@intel.com>
Signed-off-by: NDan Williams <dan.j.williams@intel.com>

[ Shile: fixed conflicts:
This patch updated the file "drivers/acpi/nfit/intel.h". The header file is
introduced by commit 0ead111 ("acpi, nfit: Collect shutdown status") in
upstream, which also update the test files. So let's fetch this part to fix
the conflict:
- tools/testing/nvdimm/test/nfit.c
- tools/testing/nvdimm/test/nfit_test.h ]
Signed-off-by: NShile Zhang <shile.zhang@linux.alibaba.com>
Reviewed-by: NYang Shi <yang.shi@linux.alibaba.com>

commit 78153dd45e7e0596ba32b15d02bda08e1513111e upstream.

Gate ARS result consumption on whether the OS issued start-ARS since the
previous consumption. The BIOS may only clear its result buffers after a
successful start-ARS.

Fixes: 0caeef63 ("libnvdimm: Add a poison list and export badblocks")
Cc: <stable@vger.kernel.org>
Reported-by: NKrzysztof Rusocki <krzysztof.rusocki@intel.com>
Reported-by: NVishal Verma <vishal.l.verma@intel.com>
Reviewed-by: NToshi Kani <toshi.kani@hpe.com>
Signed-off-by: NDan Williams <dan.j.williams@intel.com>
Signed-off-by: NSasha Levin <sashal@kernel.org>

commit 5479b2757f26fe9908fc341d105b2097fe820b6f upstream.

The ARS implementation implements exponential back-off on the poll
interval to prevent high-frequency access to the DIMM / platform
interface. Depending on when the ARS completes the poll interval may
exceed the completion event by minutes. Allow root to reset the timeout
each time it probes the status. A one-second timeout is still enforced,
but root can otherwise can control the poll interval.

Fixes: bc6ba808 ("nfit, address-range-scrub: rework and simplify ARS...")
Cc: <stable@vger.kernel.org>
Reported-by: NErwin Tsaur <erwin.tsaur@oracle.com>
Reviewed-by: NToshi Kani <toshi.kani@hpe.com>
Signed-off-by: NDan Williams <dan.j.williams@intel.com>
Signed-off-by: NSasha Levin <sashal@kernel.org>

commit e34b8252a3d2893ca55c82dbfcdaa302fa03d400 upstream.

In preparation for introducing new flags to gate whether ARS results are
stale, or poll the completion state, convert the existing flags to an
unsigned long with enumerated values. This conversion allows the flags
to be atomically updated outside of ->init_mutex.
Reviewed-by: NToshi Kani <toshi.kani@hpe.com>
Signed-off-by: NDan Williams <dan.j.williams@intel.com>
Signed-off-by: NSasha Levin <sashal@kernel.org>

commit 317a992ab9266b86b774b9f6b0f87eb4f59879a1 upstream.

The ars_start_flags property of 'struct acpi_nfit_desc' is no longer
used since ARS_REQ_SHORT and ARS_REQ_LONG were added.
Reviewed-by: NToshi Kani <toshi.kani@hpe.com>
Signed-off-by: NDan Williams <dan.j.williams@intel.com>
Signed-off-by: NSasha Levin <sashal@kernel.org>

commit d3abaf43bab8d5b0a3c6b982100d9e2be96de4ad upstream.

The Address Range Scrub implementation tried to skip running scrubs
against ranges that were already scrubbed by the BIOS. Unfortunately
that support also resulted in early scrub completions as evidenced by
this debug output from nfit_test:

    nd_region region9: ARS: range 1 short complete
    nd_region region3: ARS: range 1 short complete
    nd_region region4: ARS: range 2 ARS start (0)
    nd_region region4: ARS: range 2 short complete

...i.e. completions without any indications that the scrub was started.

This state of affairs was hard to see in the code due to the
proliferation of state bits and mistakenly trying to track done state
per-range when the completion is a global property of the bus.

So, kill the four ARS state bits (ARS_REQ, ARS_REQ_REDO, ARS_DONE, and
ARS_SHORT), and replace them with just 2 request flags ARS_REQ_SHORT and
ARS_REQ_LONG. The implementation will still complete and reap the
results of BIOS initiated ARS, but it will not attempt to use that
information to affect the completion status of scrubbing the ranges from
a Linux perspective.

Instead, try to synchronously run a short ARS per range at init time and
schedule a long scrub in the background. If ARS is busy with an ARS
request, schedule both a short and a long scrub for when ARS returns to
idle. This logic also satisfies the intent of what ARS_REQ_REDO was
trying to achieve. The new rule is that the REQ flag stays set until the
next successful ars_start() for that range.

With the new policy that the REQ flags are not cleared until the next
start, the implementation no longer loses requests as can be seen from
the following log:

    nd_region region3: ARS: range 1 ARS start short (0)
    nd_region region9: ARS: range 1 ARS start short (0)
    nd_region region3: ARS: range 1 complete
    nd_region region4: ARS: range 2 ARS start short (0)
    nd_region region9: ARS: range 1 complete
    nd_region region9: ARS: range 1 ARS start long (0)
    nd_region region4: ARS: range 2 complete
    nd_region region3: ARS: range 1 ARS start long (0)
    nd_region region9: ARS: range 1 complete
    nd_region region3: ARS: range 1 complete
    nd_region region4: ARS: range 2 ARS start long (0)
    nd_region region4: ARS: range 2 complete

...note that the nfit_test emulated driver provides 2 buses, that is why
some of the range indices are duplicated. Notice that each range
now successfully completes a short and long scrub.

Cc: <stable@vger.kernel.org>
Fixes: 14c73f99 ("nfit, address-range-scrub: introduce nfit_spa->ars_state")
Fixes: cc3d3458 ("acpi/nfit: queue issuing of ars when an uc error...")
Reported-by: NJacek Zloch <jacek.zloch@intel.com>
Reported-by: NKrzysztof Rusocki <krzysztof.rusocki@intel.com>
Reviewed-by: NDave Jiang <dave.jiang@intel.com>
Signed-off-by: NDan Williams <dan.j.williams@intel.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

When the ACPI UC error notifier gets called and ARS_REQ bit is set
with the passed in flag, we can receive -EBUSY if ARS_REQ bit is already
set for the nfit_spa->ars_state. When that happens, the ARS request is
dropped. That can potentially cause us to miss the unreported errors that
the on going ARS request does not receive. Add an ARS_REQ_REDO state that
will request short ARS upon ARS completion to grab any errors we missed.
Signed-off-by: NDave Jiang <dave.jiang@intel.com>
Reviewed-by: NVishal Verma <vishal.l.verma@intel.com>

The notification of scrub completion happens within the scrub workqueue.
That can clearly race someone running scrub_show() and work_busy()
before the workqueue has a chance to flush the recently completed work.
Add a flag to reliably indicate the idle vs busy state. Without this
change applications using poll(2) to wait for scrub-completion may
falsely wakeup and read ARS as being busy even though the thread is
going idle and then hang indefinitely.

Fixes: bc6ba808 ("nfit, address-range-scrub: rework and simplify ARS...")
Cc: <stable@vger.kernel.org>
Reported-by: NVishal Verma <vishal.l.verma@intel.com>
Tested-by: NVishal Verma <vishal.l.verma@intel.com>
Reported-by: NLukasz Dorau <lukasz.dorau@intel.com>
Signed-off-by: NDan Williams <dan.j.williams@intel.com>

ARS is an operation that can take 10s to 100s of seconds to find media
errors that should rarely be present. If the platform crashes due to
media errors in persistent memory, the expectation is that the BIOS will
report those known errors in a 'short' ARS request.

A 'short' ARS request asks platform firmware to return an ARS payload
with all known errors, but without issuing a 'long' scrub. At driver
init a short request is issued to all PMEM ranges before registering
regions. Then, in the background, a long ARS is scheduled for each
region.

The ARS implementation is simplified to centralize ARS completion work
in the ars_complete() helper. The timeout is removed since there is no
facility to cancel ARS, and this otherwise arranges for system init to
never be blocked waiting for a 'long' ARS. The ars_state flags are used
to coordinate ARS requests from driver init, ARS requests from
userspace, and ARS requests in response to media error notifications.

Given that there is no notification of ARS completion the implementation
still needs to poll. It backs off exponentially to a maximum poll period
of 30 minutes.
Suggested-by: NToshi Kani <toshi.kani@hpe.com>
Co-developed-by: NDave Jiang <dave.jiang@intel.com>
Signed-off-by: NDave Jiang <dave.jiang@intel.com>
Signed-off-by: NDan Williams <dan.j.williams@intel.com>

acpi_nfit_query_poison() is awkward in that it requires an nfit_spa
argument in order to determine what max_ars value to use. Instead probe
for the minimum max_ars across all scrub-capable ranges in the system
and drop the nfit_spa argument.

This enables a larger rework / simplification of the ARS state machine
whereby the status can be retrieved once and then iterated over all
address ranges to reap completions.
Signed-off-by: NDan Williams <dan.j.williams@intel.com>

In preparation for re-working the ARS implementation to better handle
short vs long ARS runs, introduce nfit_spa->ars_state. For now this just
replaces the nfit_spa->ars_required bit-field/flag, but going forward it
will be used to track ARS completion and make short vs long ARS
requests.
Reviewed-by: NDave Jiang <dave.jiang@intel.com>
Signed-off-by: NDan Williams <dan.j.williams@intel.com>

Some BIOSen do not handle 0-byte transfer lengths for the _LSR and _LSW
(label storage read/write) methods. This causes Linux to fallback to the
deprecated _DSM path, or otherwise disable label support.

Introduce acpi_nvdimm_has_method() to detect whether a method is
available rather than calling the method, require _LSI and _LSR to be
paired, and require read support before enabling write support.

Cc: <stable@vger.kernel.org>
Fixes: 4b27db7e ("acpi, nfit: add support for the _LS...")
Suggested-by: NErik Schmauss <erik.schmauss@intel.com>
Signed-off-by: NDan Williams <dan.j.williams@intel.com>

In ACPI 6.2a the platform capability structure has been added to the NFIT
tables. That provides software the ability to determine whether a system
supports the auto flushing of CPU caches on power loss. If the capability
is supported, we do not need to do dax_flush(). Plumbing the path to set the
property on per region from the NFIT tables.

This patch depends on the ACPI NFIT 6.2a platform capabilities support code
in include/acpi/actbl1.h.
Signed-off-by: NDave Jiang <dave.jiang@intel.com>
Reviewed-by: NRoss Zwisler <ross.zwisler@linux.intel.com>
Signed-off-by: NRoss Zwisler <ross.zwisler@linux.intel.com>

The NVDIMM_FAMILY_INTEL 'Enable Latch System Shutdown Status' command
indicates to the platform that system software has acknowledged the most
recent unsafe shutdown status.
Signed-off-by: NDan Williams <dan.j.williams@intel.com>

Per v1.6 of the NVDIMM_FAMILY_INTEL command set [1] some of the new
commands require rev-id 2. In addition to enabling ND_CMD_CALL for these
new function numbers, add a lookup table for revision-ids by family
and function number.

[1]: http://pmem.io/documents/NVDIMM_DSM_Interface-V1.6.pdfSigned-off-by: NDan Williams <dan.j.williams@intel.com>

For vendor specific commands that do not have a common kernel
translation, hide them from nmemX/commands. For example, the following
results from new enabling to probe for support of the new
NVDIMM_FAMILY_INTEL DSMs specified in v1.6 of the command specification
[1]:

    # cat /sys/bus/nd/devices/nmem0/commands
    smart smart_thresh flags get_size get_data set_data effect_size
    effect_log vendor cmd_call unknown unknown unknown unknown unknown
    unknown unknown unknown

[1]: https://pmem.io/documents/NVDIMM_DSM_Interface-V1.6.pdfSigned-off-by: NDan Williams <dan.j.williams@intel.com>

ACPI 6.2 adds support for named methods to access the label storage area
of an NVDIMM. We prefer these new methods if available and otherwise
fallback to the NVDIMM_FAMILY_INTEL _DSMs. The kernel ioctls,
ND_IOCTL_{GET,SET}_CONFIG_{SIZE,DATA}, remain generic and the driver
translates the 'package' payloads into the NVDIMM_FAMILY_INTEL 'buffer'
format to maintain compatibility with existing userspace and keep the
output buffer parsing code in the driver common.

The output payloads are mostly compatible save for the 'label area
locked' status that moves from the 'config_size' (_LSI) command to the
'config_read' (_LSR) command status.

Cc: Jeff Moyer <jmoyer@redhat.com>
Cc: Johannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: NDan Williams <dan.j.williams@intel.com>

Though nfit_test need to show what feature is supported via ND_CMD_CALL on
device/nfit/dsm_mask, currently there is no way to tell it.
This patch makes to enable it.
Signed-off-by: NYasunori Goto <y-goto@jp.fujitsu.com>
Reviewed-by: NVishal Verma <vishal.l.verma@intel.com>
Signed-off-by: NDan Williams <dan.j.williams@intel.com>

ACPI 6.2 defines in section 9.20.7.2 that the OSPM may call a Start
ARS with Flags Bit [1] set upon receiving the 0x81 notification.

  Upon receiving the notification, the OSPM may decide to issue
  a Start ARS with Flags Bit [1] set to prepare for the retrieval
  of existing records and issue the Query ARS Status function to
  retrieve the records.

Add support to call a Start ARS from acpi_nfit_uc_error_notify()
with ND_ARS_RETURN_PREV_DATA set when HW_ERROR_SCRUB_ON is not set.

Link: http://www.uefi.org/sites/default/files/resources/ACPI_6_2.pdfSigned-off-by: NToshi Kani <toshi.kani@hpe.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Rafael J. Wysocki <rjw@rjwysocki.net>
Cc: Vishal Verma <vishal.l.verma@intel.com>
Cc: Linda Knippers <linda.knippers@hpe.com>
Signed-off-by: NDan Williams <dan.j.williams@intel.com>

ACPI 6.2 defines a new ACPI notification value to NVDIMM Root Device
in Table 5-169.

 0x81 Unconsumed Uncorrectable Memory Error Detected
      Used to pro-actively notify OSPM of uncorrectable memory errors
      detected (for example a memory scrubbing engine that continuously
      scans the NVDIMMs memory). This is an optional notification. Only
      locations that were mapped in to SPA by the platform will generate
      a notification.

Add support of this notification value by initiating an ARS scan. This
will find new error locations and add their badblocks information.

Link: http://www.uefi.org/sites/default/files/resources/ACPI_6_2.pdfSigned-off-by: NToshi Kani <toshi.kani@hpe.com>
Cc: Rafael J. Wysocki <rjw@rjwysocki.net>
Cc: Vishal Verma <vishal.l.verma@intel.com>
Cc: Linda Knippers <linda.knippers@hpe.com>
Signed-off-by: NDan Williams <dan.j.williams@intel.com>

There are new types and helpers that are supposed to be used in new code.

As a preparation to get rid of legacy types and API functions do
the conversion here.
Reviewed-by: NDan Williams <dan.j.williams@intel.com>
Signed-off-by: NAndy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: NChristoph Hellwig <hch@lst.de>

The workqueue may still be running when the devres callbacks start
firing to deallocate an acpi_nfit_desc instance. Stop and flush the
workqueue before letting any other devres de-allocations proceed.
Reported-by: NLinda Knippers <linda.knippers@hpe.com>
Signed-off-by: NDan Williams <dan.j.williams@intel.com>

The nvdimm probe flushing mechanism gives userspace a sync point where
it knows all asynchronous driver probe sequences have completed.
However, it need not wait for other asynchronous actions, like
on-demand address-range-scrub. Track the init work separately from other
work in the workqueue, and only flush the former.
Signed-off-by: NDan Williams <dan.j.williams@intel.com>

Stop requiring dimms be successfully mapped into a
system-physical-address range. For provisioning and hardware remediation
purposes the kernel should account for failed devices in sysfs. If
possible it should still allow management commands to be sent to the
device.
Reported-by: NToshi Kani <toshi.kani@hpe.com>
Tested-by: NToshi Kani <toshi.kani@hpe.com>
Reported-by: NLinda Knippers <linda.knippers@hpe.com>
Signed-off-by: NDan Williams <dan.j.williams@intel.com>

A recent flurry of bug discoveries in the nfit driver's DSM marshalling
routine has highlighted the fact that we do not have unit test coverage
for this routine. Add a self-test of acpi_nfit_ctl() routine before
probing the "nfit_test.0" device. This mocks stimulus to acpi_nfit_ctl()
and if any of the tests fail "nfit_test.0" will be unavailable causing
the rest of the tests to not run / fail.

This unit test will also be a place to land reproductions of quirky BIOS
behavior discovered in the field and ensure the kernel does not regress
against implementations it has seen in practice.
Signed-off-by: NDan Williams <dan.j.williams@intel.com>

Starting a full Address Range Scrub (ARS) on hitting a memory error
machine check exception may not always be desirable. Provide a way
through sysfs to toggle the behavior between just adding the address
(cache line) where the MCE happened to the poison list and doing a full
scrub. The former (selective insertion of the address) is done
unconditionally.

Cc: linux-acpi@vger.kernel.org
Cc: Linda Knippers <linda.knippers@hpe.com>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: NVishal Verma <vishal.l.verma@intel.com>
Signed-off-by: NDan Williams <dan.j.williams@intel.com>

Trigger an nmemX/nfit/flags attribute to fire an event whenever a
smart-threshold DSM is received.
Reviewed-by: NVishal Verma <vishal.l.verma@intel.com>
Acked-by: NRafael J. Wysocki <rafael@kernel.org>
Signed-off-by: NDan Williams <dan.j.williams@intel.com>

Per "ACPI 6.1 Section 9.20.3" NVDIMM devices, children of the ACPI0012
NVDIMM Root device, can receive health event notifications.

Given that these devices are precluded from registering a notification
handler via acpi_driver.acpi_device_ops (due to no _HID), we use
acpi_install_notify_handler() directly.  The registered handler,
acpi_nvdimm_notify(), triggers a poll(2) event on the nmemX/nfit/flags
sysfs attribute when a health event notification is received.

Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Tested-by: NToshi Kani <toshi.kani@hpe.com>
Reviewed-by: NVishal Verma <vishal.l.verma@intel.com>
Acked-by: NRafael J. Wysocki <rafael@kernel.org>
Reviewed-by: NToshi Kani <toshi.kani@hpe.com>
Signed-off-by: NDan Williams <dan.j.williams@intel.com>

We have had a couple bugs in this implementation in the past and before
we add another ->notify() implementation for nvdimm devices, lets allow
this routine to be exercised via nfit_test.

Rewrite acpi_nfit_notify() in terms of a generic struct device and
acpi_handle parameter, and then implement a mock acpi_evaluate_object()
that returns a _FIT payload.

Cc: Vishal Verma <vishal.l.verma@intel.com>
Reviewed-by: NVishal Verma <vishal.l.verma@intel.com>
Acked-by: NRafael J. Wysocki <rafael@kernel.org>
Signed-off-by: NDan Williams <dan.j.williams@intel.com>

Commit 20985164 "acpi: nfit: Add support for hot-add" added
support for _FIT notifications, but it neglected to verify the
notification event code matches the one in the ACPI spec for
"NFIT Update". Currently there is only one code in the spec, but
once additional codes are added, older kernels (without this fix)
will misbehave by assuming all event notifications are for an
NFIT Update.

Fixes: 20985164 ("acpi: nfit: Add support for hot-add")
Cc: <stable@vger.kernel.org>
Cc: <linux-acpi@vger.kernel.org>
Cc: Dan Williams <dan.j.williams@intel.com>
Reported-by: NLinda Knippers <linda.knippers@hpe.com>
Signed-off-by: NVishal Verma <vishal.l.verma@intel.com>
Signed-off-by: NDan Williams <dan.j.williams@intel.com>

When a latent (unknown to 'badblocks') error is encountered, it will
trigger a machine check exception. On a system with machine check
recovery, this will only SIGBUS the process(es) which had the bad page
mapped (as opposed to a kernel panic on platforms without machine
check recovery features). In the former case, we want to trigger a full
rescan of that nvdimm bus. This will allow any additional, new errors
to be captured in the block devices' badblocks lists, and offending
operations on them can be trapped early, avoiding machine checks.

This is done by registering a callback function with the
x86_mce_decoder_chain and calling the new ars_rescan functionality with
the address in the mce notificatiion.

Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Cc: Tony Luck <tony.luck@intel.com>
Signed-off-by: NVishal Verma <vishal.l.verma@intel.com>
Signed-off-by: NDan Williams <dan.j.williams@intel.com>

With the arrival of x86-machine-check support the nfit driver will add a
(conditionally-compiled) source file.  Prepare for this by moving all
nfit source to drivers/acpi/nfit/.  This is pure code movement, no
functional changes.
Signed-off-by: NDan Williams <dan.j.williams@intel.com>

Normally, an ARS (Address Range Scrub) only happens at
boot/initialization time. There can however arise situations where a
bus-wide rescan is needed - notably, in the case of discovering a latent
media error, we should do a full rescan to figure out what other sectors
are bad, and thus potentially avoid triggering an mce on them in the
future. Also provide a sysfs trigger to start a bus-wide scrub.

Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: NVishal Verma <vishal.l.verma@intel.com>
Signed-off-by: NDan Williams <dan.j.williams@intel.com>

Pass the nfit buffer as a parameter rather than hanging it off of
acpi_desc.
Reviewed-by: N"Lee, Chun-Yi" <jlee@suse.com>
Signed-off-by: NDan Williams <dan.j.williams@intel.com>

acpi_evaluate_object() allocates memory. Free the buffer allocated
during acpi_nfit_add(). In order for this memory to be freed
acpi_nfit_init() needs to be converted to duplicate the nfit contents in
its internal allocation.  Use zero-length arrays to minimize the thrash
with the rest of the nfit driver implementation.

All of the add_<nfit-sub-table>() routines now validate a minimum table
size and expect hotplugged tables to match the size of the original
table to count as a duplicate. For variable length tables, like 'idt'
and 'flush', we calculate the dynamic size. Note that hotplug by
definition cannot change the interleave as it would cause data
corruption of in-use namespaces.

Cc: Vishal Verma <vishal.l.verma@intel.com>
Reported-by: NXiao Guangrong <guangrong.xiao@intel.com>
Reported-by: NHaozhong Zhang <haozhong.zhang@intel.com>
Signed-off-by: NDan Williams <dan.j.williams@intel.com>

The __pmem address space was meant to annotate codepaths that touch
persistent memory and need to coordinate a call to wmb_pmem().  Now that
wmb_pmem() is gone, there is little need to keep this annotation.

Cc: Christoph Hellwig <hch@lst.de>
Cc: Ross Zwisler <ross.zwisler@linux.intel.com>
Signed-off-by: NDan Williams <dan.j.williams@intel.com>

nvdimm_flush() is a replacement for the x86 'pcommit' instruction.  It is
an optional write flushing mechanism that an nvdimm bus can provide for
the pmem driver to consume.  In the case of the NFIT nvdimm-bus-provider
nvdimm_flush() is implemented as a series of flush-hint-address [1]
writes to each dimm in the interleave set (region) that backs the
namespace.

The nvdimm_has_flush() routine relies on platform firmware to describe
the flushing capabilities of a platform.  It uses the heuristic of
whether an nvdimm bus provider provides flush address data to return a
ternary result:

      1: flush addresses defined
      0: dimm topology described without flush addresses (assume ADR)
 -errno: no topology information, unable to determine flush mechanism

The pmem driver is expected to take the following actions on this ternary
result:

      1: nvdimm_flush() in response to REQ_FUA / REQ_FLUSH and shutdown
      0: do not set, WC or FUA on the queue, take no further action
 -errno: warn and then operate as if nvdimm_has_flush() returned '0'

The caveat of this heuristic is that it can not distinguish the "dimm
does not have flush address" case from the "platform firmware is broken
and failed to describe a flush address".  Given we are already
explicitly trusting the NFIT there's not much more we can do beyond
blacklisting broken firmwares if they are ever encountered.

Cc: Ross Zwisler <ross.zwisler@linux.intel.com>
Signed-off-by: NDan Williams <dan.j.williams@intel.com>

In preparation for triggering flushes of a DIMM's writes-posted-queue
(WPQ) via the pmem driver move mapping of flush hint addresses to the
region driver.  Since this uses devm_nvdimm_memremap() the flush
addresses will remain mapped while any region to which the dimm belongs
is active.

We need to communicate more information to the nvdimm core to facilitate
this mapping, namely each dimm object now carries an array of flush hint
address resources.
Signed-off-by: NDan Williams <dan.j.williams@intel.com>

Now that all shared mappings are handled by devm_nvdimm_memremap() we no
longer need nfit_spa_map() nor do we need to trigger a callback to the
bus provider at region disable time.
Signed-off-by: NDan Williams <dan.j.williams@intel.com>