libnvdimm fixes for 4.4-rc1
Subsystems (6)
Hi Linus, please pull from...
git://git.kernel.org/pub/scm/linux/kernel/git/nvdimm/nvdimm libnvdimm-fixes
...to receive:
1/ 3 fixes tagged for -stable including a crash fix, simple performance tweak, and an invalid i/o error.
2/ build regression fix for the nvdimm unit tests
3/ nvdimm documentation update
---
The following changes since commit 5d50ac70fe98518dbf620bfba8184254663125eb:
Merge tag 'xfs-for-linus-4.4' of git://git.kernel.org/pub/scm/linux/kernel/git/dgc/linux-xfs (2015-11-11 20:18:48 -0800)
are available in the git repository at:
git://git.kernel.org/pub/scm/linux/kernel/git/nvdimm/nvdimm libnvdimm-fixes
for you to fetch changes up to 152d7bd80dca5ce77ec2d7313149a2ab990e808e:
dax: fix __dax_pmd_fault crash (2015-11-12 18:33:54 -0800)
----------------------------------------------------------------
Dan Williams (4):
tools/testing/nvdimm, acpica: fix flag rename build breakage libnvdimm, e820: fix numa node for e820-type-12 pmem ranges libnvdimm, pmem: fix size trim in pmem_direct_access()
dax: fix __dax_pmd_fault crash
Konrad Rzeszutek Wilk (1):
libnvdimm: documentation clarifications
Documentation/nvdimm/nvdimm.txt | 49 +++++++++++++++++++++++----------------- drivers/nvdimm/e820.c | 15 +++++++++++- drivers/nvdimm/pmem.c | 15 ++----------
fs/dax.c | 7 ++++++ tools/testing/nvdimm/test/nfit.c | 2 +-
5 files changed, 52 insertions(+), 36 deletions(-)
commit f42957967fb435aef6fc700fbbd9df89533b9a2e
Author: Dan Williams <[email protected]>
Date: Tue Nov 10 15:50:33 2015 -0800
tools/testing/nvdimm, acpica: fix flag rename build breakage
Commit ca321d1ca672 "ACPICA: Update NFIT table to rename a flags field" performed a tree-wide s/ACPI_NFIT_MEM_ARMED/ACPI_NFIT_MEM_NOT_ARMED/ operation, but missed the tools/testing/nvdimm/ directory.
commit f7256dc0cdbc68903502997bde619f555a910f50
Author: Dan Williams <[email protected]>
Date: Wed Nov 11 16:46:33 2015 -0800
libnvdimm, e820: fix numa node for e820-type-12 pmem ranges
Rather than punt on the numa node for these e820 ranges try to find a better answer with memory_add_physaddr_to_nid() when it is available.
commit 589e75d15702dc720b363a92f984876704864946
Author: Dan Williams <[email protected]>
Date: Sat Oct 24 19:55:58 2015 -0700
libnvdimm, pmem: fix size trim in pmem_direct_access()
This masking prevents access to the end of the device via dax_do_io(), and is unnecessary as arch_add_memory() would have rejected an unaligned allocation.
commit 8de5dff8bae634497f4413bc3067389f2ed267da
Author: Konrad Rzeszutek Wilk <[email protected]> Date: Tue Nov 10 16:10:45 2015 -0800
libnvdimm: documentation clarifications
A bunch of changes that I hope will help in understanding it better for first-time readers.
commit 152d7bd80dca5ce77ec2d7313149a2ab990e808e
Author: Dan Williams <[email protected]>
Date: Thu Nov 12 18:33:54 2015 -0800
dax: fix __dax_pmd_fault crash
Since 4.3 introduced devm_memremap_pages() the pfns handled by DAX may optionally have a struct page backing. When a mapped pfn reaches vmf_insert_pfn_pmd() it fails with a crash signature like the following:
kernel BUG at mm/huge_memory.c:905!
[..]
Call Trace: [<ffffffff812a73ba>] __dax_pmd_fault+0x2ea/0x5b0 [<ffffffffa01a4182>] xfs_filemap_pmd_fault+0x92/0x150 [xfs] [<ffffffff811fbe02>] handle_mm_fault+0x312/0x1b50
Fix this by falling back to 4K mappings in the pfn_valid() case. Longer term, vmf_insert_pfn_pmd() needs to grow support for architectures that can provide a 'pmd_special' capability.
Documentation/nvdimm/nvdimm.txt +0 -0
@@ -62,6 +62,12 @@ DAX: File system extensions to bypass the page cache and block layer to
mmap persistent memory, from a PMEM block device, directly into a
process address space.
+DSM: Device Specific Method: ACPI method to to control specific
+device - in this case the firmware.
+
+DCR: NVDIMM Control Region Structure defined in ACPI 6 Section 5.2.25.5.
+It defines a vendor-id, device-id, and interface format for a given DIMM.
+
BTT: Block Translation Table: Persistent memory is byte addressable. Existing software may have an expectation that the power-fail-atomicity
of writes is at least one sector, 512 bytes. The BTT is an indirection
@@ -133,16 +139,16 @@ device driver:
registered, can be immediately attached to nd_pmem.
2. BLK (nd_blk.ko): This driver performs I/O using a set of platform
- defined apertures. A set of apertures will all access just one DIMM.
- Multiple windows allow multiple concurrent accesses, much like
+ defined apertures. A set of apertures will access just one DIMM.
+ Multiple windows (apertures) allow multiple concurrent accesses, much like tagged-command-queuing, and would likely be used by different threads or different CPUs.
The NFIT specification defines a standard format for a BLK-aperture, but the spec also allows for vendor specific layouts, and non-NFIT BLK
- implementations may other designs for BLK I/O. For this reason "nd_blk" - calls back into platform-specific code to perform the I/O. One such
- implementation is defined in the "Driver Writer's Guide" and "DSM
+ implementations may have other designs for BLK I/O. For this reason
+ "nd_blk" calls back into platform-specific code to perform the I/O.
+ One such implementation is defined in the "Driver Writer's Guide" and "DSM Interface Example".
@@ -152,7 +158,7 @@ Why BLK?
While PMEM provides direct byte-addressable CPU-load/store access to
NVDIMM storage, it does not provide the best system RAS (recovery, availability, and serviceability) model. An access to a corrupted -system-physical-address address causes a cpu exception while an access +system-physical-address address causes a CPU exception while an access
to a corrupted address through an BLK-aperture causes that block window
to raise an error status in a register. The latter is more aligned with
the standard error model that host-bus-adapter attached disks present.
@@ -162,7 +168,7 @@ data could be interleaved in an opaque hardware specific manner across several DIMMs.
PMEM vs BLK
-BLK-apertures solve this RAS problem, but their presence is also the +BLK-apertures solve these RAS problems, but their presence is also the
major contributing factor to the complexity of the ND subsystem. They complicate the implementation because PMEM and BLK alias in DPA space.
Any given DIMM's DPA-range may contribute to one or more
@@ -220,8 +226,8 @@ socket. Each unique interface (BLK or PMEM) to DPA space is identified by a region device with a dynamically assigned id (REGION0 - REGION5).
1. The first portion of DIMM0 and DIMM1 are interleaved as REGION0. A
- single PMEM namespace is created in the REGION0-SPA-range that spans
- DIMM0 and DIMM1 with a user-specified name of "pm0.0". Some of that
+ single PMEM namespace is created in the REGION0-SPA-range that spans most + of DIMM0 and DIMM1 with a user-specified name of "pm0.0". Some of that interleaved system-physical-address range is reclaimed as BLK-aperture accessed space starting at DPA-offset (a) into each DIMM. In that reclaimed space we create two BLK-aperture "namespaces" from REGION2 and @@ -230,13 +236,13 @@ by a region device with a dynamically assigned id (REGION0 - REGION5).
2. In the last portion of DIMM0 and DIMM1 we have an interleaved system-physical-address range, REGION1, that spans those two DIMMs as
- well as DIMM2 and DIMM3. Some of REGION1 allocated to a PMEM namespace - named "pm1.0" the rest is reclaimed in 4 BLK-aperture namespaces (for
+ well as DIMM2 and DIMM3. Some of REGION1 is allocated to a PMEM namespace + named "pm1.0", the rest is reclaimed in 4 BLK-aperture namespaces (for each DIMM in the interleave set), "blk2.1", "blk3.1", "blk4.0", and "blk5.0".
3. The portion of DIMM2 and DIMM3 that do not participate in the REGION1 - interleaved system-physical-address range (i.e. the DPA address below
+ interleaved system-physical-address range (i.e. the DPA address past offset (b) are also included in the "blk4.0" and "blk5.0" namespaces. Note, that this example shows that BLK-aperture namespaces don't need to be contiguous in DPA-space.
@@ -252,15 +258,15 @@ LIBNVDIMM Kernel Device Model and LIBNDCTL Userspace API
What follows is a description of the LIBNVDIMM sysfs layout and a corresponding object hierarchy diagram as viewed through the LIBNDCTL
-api. The example sysfs paths and diagrams are relative to the Example
+API. The example sysfs paths and diagrams are relative to the Example NVDIMM Platform which is also the LIBNVDIMM bus used in the LIBNDCTL unit test.
LIBNDCTL: Context
-Every api call in the LIBNDCTL library requires a context that holds the +Every API call in the LIBNDCTL library requires a context that holds the logging parameters and other library instance state. The library is
based on the libabc template:
-https://git.kernel.org/cgit/linux/kernel/git/kay/libabc.git/ +https://git.kernel.org/cgit/linux/kernel/git/kay/libabc.git
LIBNDCTL: instantiate a new library context example
@@ -409,7 +415,7 @@ Bit 31:28 Reserved
LIBNVDIMM/LIBNDCTL: Region
----------------------
-A generic REGION device is registered for each PMEM range orBLK-aperture
+A generic REGION device is registered for each PMEM range or BLK-aperture set. Per the example there are 6 regions: 2 PMEM and 4 BLK-aperture
sets on the "nfit_test.0" bus. The primary role of regions are to be a container of "mappings". A mapping is a tuple of <DIMM,
@@ -509,7 +515,7 @@ At first glance it seems since NFIT defines just PMEM and BLK interface types that we should simply name REGION devices with something derived
from those type names. However, the ND subsystem explicitly keeps the REGION name generic and expects userspace to always consider the -region-attributes for 4 reasons:
+region-attributes for four reasons:
1. There are already more than two REGION and "namespace" types. For PMEM there are two subtypes. As mentioned previously we have PMEM where @@ -698,8 +704,8 @@ static int configure_namespace(struct ndctl_region *region,
Why the Term "namespace"?
- 1. Why not "volume" for instance? "volume" ran the risk of confusing ND - as a volume manager like device-mapper.
+ 1. Why not "volume" for instance? "volume" ran the risk of confusing
+ ND (libnvdimm subsystem) to a volume manager like device-mapper.
2. The term originated to describe the sub-devices that can be created within a NVME controller (see the nvme specification:
@@ -774,13 +780,14 @@ block" needs to be destroyed. Note, that to destroy a BTT the media needs to be written in raw mode. By default, the kernel will autodetect
the presence of a BTT and disable raw mode. This autodetect behavior
can be suppressed by enabling raw mode for the namespace via the -ndctl_namespace_set_raw_mode() api.
+ndctl_namespace_set_raw_mode() API.
Summary LIBNDCTL Diagram
------------------------
-For the given example above, here is the view of the objects as seen by the LIBNDCTL api:
+For the given example above, here is the view of the objects as seen by the +LIBNDCTL API:
+---+ |CTX| +---------+ +--------------+ +---------------+ +-+-+ +-> REGION0 +---> NAMESPACE0.0 +--> PMEM8 "pm0.0" |
drivers/nvdimm/e820.c +0 -0
@@ -3,6 +3,7 @@
* Copyright (c) 2015, Intel Corporation.
*/
#include <linux/platform_device.h>
+#include <linux/memory_hotplug.h>
#include <linux/libnvdimm.h>
#include <linux/module.h>
@@ -25,6 +26,18 @@ static int e820_pmem_remove(struct platform_device *pdev) return 0;
}
+#ifdef CONFIG_MEMORY_HOTPLUG
+static int e820_range_to_nid(resource_size_t addr)
+{
+ return memory_add_physaddr_to_nid(addr);
+}
+#else
+static int e820_range_to_nid(resource_size_t addr)
+{
+ return NUMA_NO_NODE;
+}
+#endif
+
static int e820_pmem_probe(struct platform_device *pdev)
{
static struct nvdimm_bus_descriptor nd_desc;
@@ -48,7 +61,7 @@ static int e820_pmem_probe(struct platform_device *pdev) memset(&ndr_desc, 0, sizeof(ndr_desc)); ndr_desc.res = p;
ndr_desc.attr_groups = e820_pmem_region_attribute_groups;
- ndr_desc.numa_node = NUMA_NO_NODE;
+ ndr_desc.numa_node = e820_range_to_nid(p->start);
set_bit(ND_REGION_PAGEMAP, &ndr_desc.flags);
if (!nvdimm_pmem_region_create(nvdimm_bus, &ndr_desc)) goto err;
drivers/nvdimm/pmem.c +0 -0
@@ -105,22 +105,11 @@ static long pmem_direct_access(struct block_device *bdev, sector_t sector,
{
struct pmem_device *pmem = bdev->bd_disk->private_data; resource_size_t offset = sector * 512 + pmem->data_offset;
- resource_size_t size;
-
- if (pmem->data_offset) {
- /*
- * Limit the direct_access() size to what is covered by
- * the memmap
- */
- size = (pmem->size - offset) & ~ND_PFN_MASK;
- } else
- size = pmem->size - offset;
-
- /* FIXME convert DAX to comprehend that this mapping has a lifetime */
+
*kaddr = pmem->virt_addr + offset;
*pfn = (pmem->phys_addr + offset) >> PAGE_SHIFT;
- return size;
+ return pmem->size - offset;
}
static const struct block_device_operations pmem_fops = {
fs/dax.c +0 -0
@@ -627,6 +627,13 @@ int __dax_pmd_fault(struct vm_area_struct *vma, unsigned long address,
if ((length < PMD_SIZE) || (pfn & PG_PMD_COLOUR)) goto fallback;
+ /*
+ * TODO: teach vmf_insert_pfn_pmd() to support
+ * 'pte_special' for pmds
+ */
+ if (pfn_valid(pfn))
+ goto fallback;
+
if (buffer_unwritten(&bh) || buffer_new(&bh)) { int i;
for (i = 0; i < PTRS_PER_PMD; i++)
tools/testing/nvdimm/test/nfit.c +0 -0
@@ -1135,7 +1135,7 @@ static void nfit_test1_setup(struct nfit_test *t)
memdev->interleave_ways = 1;
memdev->flags = ACPI_NFIT_MEM_SAVE_FAILED | ACPI_NFIT_MEM_RESTORE_FAILED
| ACPI_NFIT_MEM_FLUSH_FAILED | ACPI_NFIT_MEM_HEALTH_OBSERVED - | ACPI_NFIT_MEM_ARMED; + | ACPI_NFIT_MEM_NOT_ARMED;
offset += sizeof(*memdev);
/* dcr-descriptor0 */