When the NFIT provides multiple flush hint addresses per-dimm it is
expressing that the platform is capable of processing multiple flush
requests in parallel. There is some fixed cost per flush request, let
the cost be shared in parallel on multiple cpus.
Since there may not be enough flush hint addresses for each cpu to have
one, keep a per-cpu index of the last used hint, hash it with current
pid, and assume that access pattern and scheduler randomness will keep
the flush-hint usage somewhat staggered across cpus.
Cc: Ross Zwisler <ross.zwisler@linux.intel.com>
Signed-off-by: Dan 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: Dan 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: Dan Williams <dan.j.williams@intel.com>
For autodetecting a previously established dax configuration we need the
info block to indicate block-device vs device-dax mode, and we need to
have the default namespace probe hand-off the configuration to the
dax_pmem driver.
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Device DAX is the device-centric analogue of Filesystem DAX
(CONFIG_FS_DAX). It allows persistent memory ranges to be allocated and
mapped without need of an intervening file system. This initial
infrastructure arranges for a libnvdimm pfn-device to be represented as
a different device-type so that it can be attached to a driver other
than the pmem driver.
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Now that pmem internals have been disentangled from pfn setup, that code
can move to the core. This is in preparation for adding another user of
the pfn-device capabilities.
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
In preparation for providing an alternative (to block device) access
mechanism to persistent memory, convert pmem_rw_bytes() to
nsio_rw_bytes(). This allows ->rw_bytes() functionality without
requiring a 'struct pmem_device' to be instantiated.
In other words, when ->rw_bytes() is in use i/o is driven through
'struct nd_namespace_io', otherwise it is driven through 'struct
pmem_device' and the block layer. This consolidates the disjoint calls
to devm_exit_badblocks() and devm_memunmap() into a common
devm_nsio_disable() and cleans up the init path to use a unified
pmem_attach_disk() implementation.
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Pass the device performing the probe so we can use a devm allocation for
the btt superblock.
Cc: Vishal Verma <vishal.l.verma@intel.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Pass the device performing the probe so we can use a devm allocation for
the pfn superblock.
Cc: Ross Zwisler <ross.zwisler@linux.intel.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
We can derive the common namespace from other information. We also do
not need to cache it because all the usages are in slow paths.
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
When section alignment padding is in effect we need to shift / truncate
the range that is queried for poison by the 'start_pad' or 'end_trunc'
reservations.
It's easiest if we just pass in an adjusted resource range rather than
deriving it from the passed in namespace. With the resource range
resolution pushed out to the caller we can also push the
namespace-to-region lookup to the caller and drop the implicit pmem-type
assumption about the passed in namespace object.
Cc: Vishal Verma <vishal.l.verma@intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
If a write is directed at a known bad block perform the following:
1/ write the data
2/ send a clear poison command
3/ invalidate the poison out of the cache hierarchy
Cc: <x86@kernel.org>
Cc: Ross Zwisler <ross.zwisler@linux.intel.com>
Reviewed-by: Vishal Verma <vishal.l.verma@intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
In preparation for asynchronous address range scrub support add an
ability for the pmem driver to dynamically consume address range scrub
results.
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
If a device will ever have badblocks it should always have a badblocks
instance available. So, similar to md, embed a badblocks instance in
pmem_device. This reduces pointer chasing in the i/o fast path, and
simplifies the init path.
Reported-by: Vishal Verma <vishal.l.verma@intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
During region creation, perform Address Range Scrubs (ARS) for the SPA
(System Physical Address) ranges to retrieve known poison locations from
firmware. Add a new data structure 'nd_poison' which is used as a list
in nvdimm_bus to store these poison locations.
When creating a pmem namespace, if there is any known poison associated
with its physical address space, convert the poison ranges to bad sectors
that are exposed using the badblocks interface.
Signed-off-by: Vishal Verma <vishal.l.verma@intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
When setting aside capacity for struct page it must be aligned to the
largest mapping size that is to be made available via DAX. Make the
alignment configurable to enable support for 1GiB page-size mappings.
The offset for PFN_MODE_RAM may now be larger than SZ_8K, so fixup the
offset check in nvdimm_namespace_attach_pfn().
Reported-by: Toshi Kani <toshi.kani@hpe.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
The alignment constraint isn't necessary now that devm_memremap_pages()
allows for unaligned mappings.
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
The expectation is that the legacy / non-standard pmem discovery method
(e820 type-12) will only ever be used to describe small quantities of
persistent memory. Larger capacities will be described via the ACPI
NFIT. When "allocate struct page from pmem" support is added this default
policy can be overridden by assigning a legacy pmem namespace to a pfn
device, however this would be only be necessary if a platform used the
legacy mechanism to define a very large range.
Cc: Christoph Hellwig <hch@lst.de>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Enable the pmem driver to handle PFN device instances. Attaching a pmem
namespace to a pfn device triggers the driver to allocate and initialize
struct page entries for pmem. Memory capacity for this allocation comes
exclusively from RAM for now which is suitable for low PMEM to RAM
ratios. This mechanism will be expanded later for setting an "allocate
from PMEM" policy.
Cc: Boaz Harrosh <boaz@plexistor.com>
Cc: Ross Zwisler <ross.zwisler@linux.intel.com>
Cc: Christoph Hellwig <hch@lst.de>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Implement the base infrastructure for libnvdimm PFN devices. Similar to
BTT devices they take a namespace as a backing device and layer
functionality on top. In this case the functionality is reserving space
for an array of 'struct page' entries to be handed out through
pfn_to_page(). For now this is just the basic libnvdimm-device-model for
configuring the base PFN device.
As the namespace claiming mechanism for PFN devices is mostly identical
to BTT devices drivers/nvdimm/claim.c is created to house the common
bits.
Cc: Ross Zwisler <ross.zwisler@linux.intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
When a BTT is instantiated on a namespace it must validate the namespace
uuid matches the 'parent_uuid' stored in the btt superblock. This
property enforces that changing the namespace UUID invalidates all
former BTT instances on that storage. For "IO namespaces" that don't
have a label or UUID, the parent_uuid is set to zero, and this
validation is skipped. For such cases, old BTTs have to be invalidated
by forcing the namespace to raw mode, and overwriting the BTT info
blocks.
Based on a patch by Dan Williams <dan.j.williams@intel.com>
Signed-off-by: Vishal Verma <vishal.l.verma@intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Fix multiple build warnings when CONFIG_BTT is not enabled:
In file included from ../drivers/nvdimm/bus.c:29:0:
../drivers/nvdimm/nd.h:169:15: warning: return type defaults to 'int' [-Wreturn-type]
static inline nd_btt_probe(struct nd_namespace_common *ndns, void *drvdata)
^
Signed-off-by: Randy Dunlap <rdunlap@infradead.org>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: linux-nvdimm@lists.01.org
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
ACPI NFIT table has System Physical Address Range Structure entries that
describe a proximity ID of each range when ACPI_NFIT_PROXIMITY_VALID is
set in the flags.
Change acpi_nfit_register_region() to map a proximity ID to its node ID,
and set it to a new numa_node field of nd_region_desc, which is then
conveyed to the nd_region device.
The device core arranges for btt and namespace devices to inherit their
node from their parent region.
Signed-off-by: Toshi Kani <toshi.kani@hp.com>
[djbw: move set_dev_node() from region.c to bus.c]
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Upon detection of an unarmed dimm in a region, arrange for descendant
BTT, PMEM, or BLK instances to be read-only. A dimm is primarily marked
"unarmed" via flags passed by platform firmware (NFIT).
The flags in the NFIT memory device sub-structure indicate the state of
the data on the nvdimm relative to its energy source or last "flush to
persistence". For the most part there is nothing the driver can do but
advertise the state of these flags in sysfs and emit a message if
firmware indicates that the contents of the device may be corrupted.
However, for the case of ACPI_NFIT_MEM_ARMED, the driver can arrange for
the block devices incorporating that nvdimm to be marked read-only.
This is a safe default as the data is still available and new writes are
held off until the administrator either forces read-write mode, or the
energy source becomes armed.
A 'read_only' attribute is added to REGION devices to allow for
overriding the default read-only policy of all descendant block devices.
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
This is disabled by default as the overhead is prohibitive, but if the
user takes the action to turn it on we'll oblige.
Reviewed-by: Vishal Verma <vishal.l.verma@linux.intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Support multiple block sizes (sector + metadata) for nd_blk in the
same way as done for the BTT. Add the idea of an 'internal' lbasize,
which is properly aligned and padded, and store metadata in this space.
Signed-off-by: Vishal Verma <vishal.l.verma@linux.intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Support multiple block sizes (sector + metadata) using the blk integrity
framework. This registers a new integrity template that defines the
protection information tuple size based on the configured metadata size,
and simply acts as a passthrough for protection information generated by
another layer. The metadata is written to the storage as-is, and read back
with each sector.
Signed-off-by: Vishal Verma <vishal.l.verma@linux.intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
The libnvdimm implementation handles allocating dimm address space (DPA)
between PMEM and BLK mode interfaces. After DPA has been allocated from
a BLK-region to a BLK-namespace the nd_blk driver attaches to handle I/O
as a struct bio based block device. Unlike PMEM, BLK is required to
handle platform specific details like mmio register formats and memory
controller interleave. For this reason the libnvdimm generic nd_blk
driver calls back into the bus provider to carry out the I/O.
This initial implementation handles the BLK interface defined by the
ACPI 6 NFIT [1] and the NVDIMM DSM Interface Example [2] composed from
DCR (dimm control region), BDW (block data window), IDT (interleave
descriptor) NFIT structures and the hardware register format.
[1]: http://www.uefi.org/sites/default/files/resources/ACPI_6.0.pdf
[2]: http://pmem.io/documents/NVDIMM_DSM_Interface_Example.pdf
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Boaz Harrosh <boaz@plexistor.com>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Jens Axboe <axboe@fb.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Christoph Hellwig <hch@lst.de>
Signed-off-by: Ross Zwisler <ross.zwisler@linux.intel.com>
Acked-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
BTT stands for Block Translation Table, and is a way to provide power
fail sector atomicity semantics for block devices that have the ability
to perform byte granularity IO. It relies on the capability of libnvdimm
namespace devices to do byte aligned IO.
The BTT works as a stacked blocked device, and reserves a chunk of space
from the backing device for its accounting metadata. It is a bio-based
driver because all IO is done synchronously, and there is no queuing or
asynchronous completions at either the device or the driver level.
The BTT uses 'lanes' to index into various 'on-disk' data structures,
and lanes also act as a synchronization mechanism in case there are more
CPUs than available lanes. We did a comparison between two lane lock
strategies - first where we kept an atomic counter around that tracked
which was the last lane that was used, and 'our' lane was determined by
atomically incrementing that. That way, for the nr_cpus > nr_lanes case,
theoretically, no CPU would be blocked waiting for a lane. The other
strategy was to use the cpu number we're scheduled on to and hash it to
a lane number. Theoretically, this could block an IO that could've
otherwise run using a different, free lane. But some fio workloads
showed that the direct cpu -> lane hash performed faster than tracking
'last lane' - my reasoning is the cache thrash caused by moving the
atomic variable made that approach slower than simply waiting out the
in-progress IO. This supports the conclusion that the driver can be a
very simple bio-based one that does synchronous IOs instead of queuing.
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Boaz Harrosh <boaz@plexistor.com>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Jens Axboe <axboe@fb.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Neil Brown <neilb@suse.de>
Cc: Jeff Moyer <jmoyer@redhat.com>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Greg KH <gregkh@linuxfoundation.org>
[jmoyer: fix nmi watchdog timeout in btt_map_init]
[jmoyer: move btt initialization to module load path]
[jmoyer: fix memory leak in the btt initialization path]
[jmoyer: Don't overwrite corrupted arenas]
Signed-off-by: Vishal Verma <vishal.l.verma@linux.intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
NVDIMM namespaces, in addition to accepting "struct bio" based requests,
also have the capability to perform byte-aligned accesses. By default
only the bio/block interface is used. However, if another driver can
make effective use of the byte-aligned capability it can claim namespace
interface and use the byte-aligned ->rw_bytes() interface.
The BTT driver is the initial first consumer of this mechanism to allow
adding atomic sector update semantics to a pmem or blk namespace. This
patch is the sysfs infrastructure to allow configuring a BTT instance
for a namespace. Enabling that BTT and performing i/o is in a
subsequent patch.
Cc: Greg KH <gregkh@linuxfoundation.org>
Cc: Neil Brown <neilb@suse.de>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
After 'uuid', 'size', and optionally 'alt_name' have been set to valid
values the labels on the dimms can be updated.
Write procedure is:
1/ Allocate and write new labels in the "next" index
2/ Free the old labels in the working copy
3/ Write the bitmap and the label space on the dimm
4/ Write the index to make the update valid
Label ranges directly mirror the dpa resource values for the given
label_id of the namespace.
Cc: Greg KH <gregkh@linuxfoundation.org>
Cc: Neil Brown <neilb@suse.de>
Acked-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
A blk label set describes a namespace comprised of one or more
discontiguous dpa ranges on a single dimm. They may alias with one or
more pmem interleave sets that include the given dimm.
This is the runtime/volatile configuration infrastructure for sysfs
manipulation of 'alt_name', 'uuid', 'size', and 'sector_size'. A later
patch will make these settings persistent by writing back the label(s).
Unlike pmem namespaces, multiple blk namespaces can be created per
region. Once a blk namespace has been created a new seed device
(unconfigured child of a parent blk region) is instantiated. As long as
a region has 'available_size' != 0 new child namespaces may be created.
Cc: Greg KH <gregkh@linuxfoundation.org>
Cc: Neil Brown <neilb@suse.de>
Acked-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
A complete label set is a PMEM-label per-dimm per-interleave-set where
all the UUIDs match and the interleave set cookie matches the hosting
interleave set.
Present sysfs attributes for manipulation of a PMEM-namespace's
'alt_name', 'uuid', and 'size' attributes. A later patch will make
these settings persistent by writing back the label.
Note that PMEM allocations grow forwards from the start of an interleave
set (lowest dimm-physical-address (DPA)). BLK-namespaces that alias
with a PMEM interleave set will grow allocations backward from the
highest DPA.
Cc: Greg KH <gregkh@linuxfoundation.org>
Cc: Neil Brown <neilb@suse.de>
Acked-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
This on media label format [1] consists of two index blocks followed by
an array of labels. None of these structures are ever updated in place.
A sequence number tracks the current active index and the next one to
write, while labels are written to free slots.
+------------+
| |
| nsindex0 |
| |
+------------+
| |
| nsindex1 |
| |
+------------+
| label0 |
+------------+
| label1 |
+------------+
| |
....nslot...
| |
+------------+
| labelN |
+------------+
After reading valid labels, store the dpa ranges they claim into
per-dimm resource trees.
[1]: http://pmem.io/documents/NVDIMM_Namespace_Spec.pdf
Cc: Neil Brown <neilb@suse.de>
Acked-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
On platforms that have firmware support for reading/writing per-dimm
label space, a portion of the dimm may be accessible via an interleave
set PMEM mapping in addition to the dimm's BLK (block-data-window
aperture(s)) interface. A label, stored in a "configuration data
region" on the dimm, disambiguates which dimm addresses are accessed
through which exclusive interface.
Add infrastructure that allows the kernel to block modifications to a
label in the set while any member dimm is active. Note that this is
meant only for enforcing "no modifications of active labels" via the
coarse ioctl command. Adding/deleting namespaces from an active
interleave set is always possible via sysfs.
Another aspect of tracking interleave sets is tracking their integrity
when DIMMs in a set are physically re-ordered. For this purpose we
generate an "interleave-set cookie" that can be recorded in a label and
validated against the current configuration. It is the bus provider
implementation's responsibility to calculate the interleave set cookie
and attach it to a given region.
Cc: Neil Brown <neilb@suse.de>
Cc: <linux-acpi@vger.kernel.org>
Cc: Greg KH <gregkh@linuxfoundation.org>
Cc: Robert Moore <robert.moore@intel.com>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: Christoph Hellwig <hch@lst.de>
Acked-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
The libnvdimm region driver is an intermediary driver that translates
non-volatile "region"s into "namespace" sub-devices that are surfaced by
persistent memory block-device drivers (PMEM and BLK).
ACPI 6 introduces the concept that a given nvdimm may simultaneously
offer multiple access modes to its media through direct PMEM load/store
access, or windowed BLK mode. Existing nvdimms mostly implement a PMEM
interface, some offer a BLK-like mode, but never both as ACPI 6 defines.
If an nvdimm is single interfaced, then there is no need for dimm
metadata labels. For these devices we can take the region boundaries
directly to create a child namespace device (nd_namespace_io).
Acked-by: Christoph Hellwig <hch@lst.de>
Tested-by: Toshi Kani <toshi.kani@hp.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
A "region" device represents the maximum capacity of a BLK range (mmio
block-data-window(s)), or a PMEM range (DAX-capable persistent memory or
volatile memory), without regard for aliasing. Aliasing, in the
dimm-local address space (DPA), is resolved by metadata on a dimm to
designate which exclusive interface will access the aliased DPA ranges.
Support for the per-dimm metadata/label arrvies is in a subsequent
patch.
The name format of "region" devices is "regionN" where, like dimms, N is
a global ida index assigned at discovery time. This id is not reliable
across reboots nor in the presence of hotplug. Look to attributes of
the region or static id-data of the sub-namespace to generate a
persistent name. However, if the platform configuration does not change
it is reasonable to expect the same region id to be assigned at the next
boot.
"region"s have 2 generic attributes "size", and "mapping"s where:
- size: the BLK accessible capacity or the span of the
system physical address range in the case of PMEM.
- mappingN: a tuple describing a dimm's contribution to the region's
capacity in the format (<nmemX>,<dpa>,<size>). For a PMEM-region
there will be at least one mapping per dimm in the interleave set. For
a BLK-region there is only "mapping0" listing the starting DPA of the
BLK-region and the available DPA capacity of that space (matches "size"
above).
The max number of mappings per "region" is hard coded per the
constraints of sysfs attribute groups. That said the number of mappings
per region should never exceed the maximum number of possible dimms in
the system. If the current number turns out to not be enough then the
"mappings" attribute clarifies how many there are supposed to be. "32
should be enough for anybody...".
Cc: Neil Brown <neilb@suse.de>
Cc: <linux-acpi@vger.kernel.org>
Cc: Greg KH <gregkh@linuxfoundation.org>
Cc: Robert Moore <robert.moore@intel.com>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: Christoph Hellwig <hch@lst.de>
Acked-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Tested-by: Toshi Kani <toshi.kani@hp.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
* Implement the device-model infrastructure for loading modules and
attaching drivers to nvdimm devices. This is a simple association of a
nd-device-type number with a driver that has a bitmask of supported
device types. To facilitate userspace bind/unbind operations 'modalias'
and 'devtype', that also appear in the uevent, are added as generic
sysfs attributes for all nvdimm devices. The reason for the device-type
number is to support sub-types within a given parent devtype, be it a
vendor-specific sub-type or otherwise.
* The first consumer of this infrastructure is the driver
for dimm devices. It simply uses control messages to retrieve and
store the configuration-data image (label set) from each dimm.
Note: nd_device_register() arranges for asynchronous registration of
nvdimm bus devices by default.
Cc: Greg KH <gregkh@linuxfoundation.org>
Cc: Neil Brown <neilb@suse.de>
Acked-by: Christoph Hellwig <hch@lst.de>
Tested-by: Toshi Kani <toshi.kani@hp.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Dan Williams
Vishal Verma
Randy Dunlap
Toshi Kani
Ross Zwisler