[RFC PATCH 1/4] libnvdimm/namespace: Make namespace size validation arch dependent

[Dan Williams]

On Tue, Nov 19, 2019 at 7:19 PM Aneesh Kumar K.V
<aneesh.kumar at linux.ibm.com> wrote:
>
> On 11/19/19 11:28 PM, Dan Williams wrote:
> > On Mon, Nov 18, 2019 at 1:52 AM Aneesh Kumar K.V
> > <aneesh.kumar at linux.ibm.com> wrote:
> >>
> >> Dan Williams <dan.j.williams at intel.com> writes:
> >>
> >>> On Sat, Nov 16, 2019 at 4:15 AM Aneesh Kumar K.V
> >>> <aneesh.kumar at linux.ibm.com> wrote:
> >>>>
> >>
> >> ....
> >>
> >>
> >>>>
> >>>> Considering the direct-map map size is not going to be user selectable,
> >>>> do you agree that we can skip the above step 0 configuration you
> >>>> suggested.
> >>>>
> >>>> The changes proposed in the patch series essentially does the rest.
> >>>>
> >>>> 1) It validate the size against the arch specific limit during
> >>>> namespace creation. (part of step 1)
> >>>
> >>> This validation is a surprise failure to ndctl.
> >>>
> >>>> 2) It also disable initializing a region if it find the size not
> >>>> correctly aligned as per the platform requirement.
> >>>
> >>> There needs to be a way for the user to discover the correct alignment
> >>> that the kernel will accept.
> >>>
> >>>> 3) Direct map  mapping size is different from supported_alignment for a
> >>>> namespace. The supported alignment controls what possible PAGE SIZE user want the
> >>>> namespace to be mapped to user space.
> >>>
> >>> No, the namespace alignment is different than the page mapping size.
> >>> The alignment is only interpreted as a mapping size at the device-dax
> >>> level, otherwise at the raw namespace level it's just an arbitrary
> >>> alignment.
> >>>
> >>>> With the above do you think the current patch series is good?
> >>>
> >>> I don't think we've quite converged on a solution.
> >>
> >> How about we make it a property of seed device. ie,
> >> we add `supported_size_align` RO attribute to the seed device. ndctl can
> >> use this to validate the size value. So this now becomes step0
> >>
> >> sys/bus/nd/devices/region0> cat btt0.0/supported_size_align
> >> 16777216
> >> /sys/bus/nd/devices/region0> cat pfn0.0/supported_size_align
> >> 16777216
> >> /sys/bus/nd/devices/region0> cat dax0.0/supported_size_align
> >> 16777216
> >
> > Why on those devices and not namespace0.0?
>
> sure.
>
> >
> >> We follow that up with validating the size value written to size
> >> attribute(step 1).
> >>
> >> While initializing the namespaces already present in a region we again
> >> validate the size and if not properly aligned we mark the region
> >> disabled.
> >
> > The region might have a mix of namespaces, some aligned and some not,
> > only the misaligned namespaces should fail to enable. The region
> > should otherwise enable successfully.
> >
>
> One misaligned namespace would mean, we get other namespace resource
> start addr wrongly aligned. If we allow regions to be enabled with
> namespace with wrong size, user would find further namespace creation in
> that regions failing due to wrongly aligned resource start. IMHO that is
> a confusing user experience.
>

Why would one wrongly aligned namespace prevent other namespaces from
being aligned? There's no requirement that consecutive namespaces are
allocated contiguously. Also consider a namespace that starts
misaligned, but ends aligned. That subsequent namespace can be enabled
without issue.