[PATCH 8/9] ndctl: fix ndctl_region_get_interleave_ways(), set minimum

2017-01-09 Thread Dan Williams
A pmem region that does not have any associated dimm mappings is still a x1 interleave width by definition. Given that tooling does division by interleave-ways, return 1 as a minimum so that users do not need to check for the zero case. If a user instead wants to know the number of mapping entries

[PATCH 9/9] ndctl, create-namespace: enforce --size must be even multiple of interleave-width

2017-01-09 Thread Dan Williams
If we don't check the size is properly aligned then the kernel will cause the create-namespace request to abort. When the size is not properly aligned suggest an alternative (rounded to the same units that user specified) and bail out early without disturbing the current configuration. Signed-off-

[PATCH 7/9] ndctl: return unit size from parse_size64()

2017-01-09 Thread Dan Williams
In preparation for suggesting namespace sizes that satisfy alignment requirements, add the ability to track the size suffix. The suggestions will be aligned to the size suffix by default. Signed-off-by: Dan Williams --- util/size.c | 18 +- 1 file changed, 17 insertions(+), 1 d

[PATCH 6/9] daxctl: add daxctl to rpm spec

2017-01-09 Thread Dan Williams
Signed-off-by: Dan Williams --- ndctl.spec.in |8 +++- 1 file changed, 7 insertions(+), 1 deletion(-) diff --git a/ndctl.spec.in b/ndctl.spec.in index 6453edd1954c..b4817622e0b2 100644 --- a/ndctl.spec.in +++ b/ndctl.spec.in @@ -114,9 +114,15 @@ make check %defattr(-,root,root) %licens

[PATCH 0/9] ndctl, create-namespace: updates and fixes

2017-01-09 Thread Dan Williams
Close some holes in namespace creation that allowed invalid configurations to leak through to the kernel rather than trapping and warning about them earlier. Add a man page and rpm specfile support for the recently added daxctl utility. --- Dan Williams (9): ndctl, create-namespace: skip i

[PATCH 1/9] ndctl, create-namespace: skip idle regions when scanning for capacity

2017-01-09 Thread Dan Williams
Let validate_namespace_options() check that the region is enabled before allowing namespace creation to proceed. Signed-off-by: Dan Williams --- ndctl/builtin-xaction-namespace.c | 17 + 1 file changed, 13 insertions(+), 4 deletions(-) diff --git a/ndctl/builtin-xaction-namesp

[PATCH 3/9] ndctl, create-namespace: improve failure message

2017-01-09 Thread Dan Williams
Distinguish failure to find suitable capacity with failures due to missing capabilities in the kernel. Signed-off-by: Dan Williams --- ndctl/builtin-xaction-namespace.c | 20 ++-- 1 file changed, 14 insertions(+), 6 deletions(-) diff --git a/ndctl/builtin-xaction-namespace.c

[PATCH 5/9] daxctl, list: add man page

2017-01-09 Thread Dan Williams
Document the 'daxctl list' command which emits the details of the Device-DAX regions and device instances in the system: # daxctl list -DRi { "id":1, "devices":[ { "chardev":"dax1.0", "size":3233808384 } ] } Signed-off-by: Dan Williams --- Documenta

[PATCH 4/9] ndctl, create-namespace: add an 'align' option

2017-01-09 Thread Dan Williams
When memory-mode support was added to the kernel it also included a facility to set the alignment of the namespace. By default it is set to 2M, but 4K and 1G are also valid options. Add this support to the utility. As Dave notes, this is prompted by the desire to add test support for device-dax usi

[PATCH 2/9] ndctl, create-namespace: skip blk regions when scanning for memory or dax mode

2017-01-09 Thread Dan Williams
Treat blk regions as invalid capacity when the mode is a pmem-only mode type. Signed-off-by: Dan Williams --- ndctl/builtin-xaction-namespace.c | 18 ++ 1 file changed, 10 insertions(+), 8 deletions(-) diff --git a/ndctl/builtin-xaction-namespace.c b/ndctl/builtin-xaction-nam

Great Opportunities: AI/Data-Mining/NLP, Blockchain CTO, etc.

2017-01-09 Thread Nicholas Meyler
Exciting Searches for January (1) NLP/Machine-Learning Data Scientist (Pleasanton, CA): My client has remained at the center of scientific discovery for more than 50 years, manufacturing and distributing a broad range of products for the life science research and clinical diagnostic markets. T

Re: [ndctl PATCH 0/7] introduce 'daxctl list', and 'ndctl list' updates

2017-01-09 Thread Verma, Vishal L
On Mon, 2017-01-09 at 14:00 -0800, Dan Williams wrote: > On Mon, Jan 9, 2017 at 1:56 PM, Vishal Verma > wrote: > > On 01/06, Dan Williams wrote: > > > * The 'ndctl list' command awkwardly prints out all the > > > corresponding > > >   device-dax information when a namespace is in 'dax' mode. > > >

Re: [ndctl PATCH 0/7] introduce 'daxctl list', and 'ndctl list' updates

2017-01-09 Thread Vishal Verma
On 01/06, Dan Williams wrote: > * The 'ndctl list' command awkwardly prints out all the corresponding > device-dax information when a namespace is in 'dax' mode. Conversely if > someone is only interested in listing device-dax information they need to > contend with libnvdimm data. > > Int

Re: [ndctl PATCH 0/7] introduce 'daxctl list', and 'ndctl list' updates

2017-01-09 Thread Dan Williams
On Mon, Jan 9, 2017 at 1:56 PM, Vishal Verma wrote: > On 01/06, Dan Williams wrote: >> * The 'ndctl list' command awkwardly prints out all the corresponding >> device-dax information when a namespace is in 'dax' mode. Conversely if >> someone is only interested in listing device-dax informatio