On Fri, Jun 16, 2017 at 10:14 AM, Jeff Moyer <jmoyer(a)redhat.com> wrote:
Dan Williams <dan.j.williams(a)intel.com> writes:
> On Fri, Jun 16, 2017 at 10:00 AM, Jeff Moyer <jmoyer(a)redhat.com> wrote:
>> Dave Jiang <dave.jiang(a)intel.com> writes:
>>> The daxctl io option allows I/Os to be performed between block/file to
>>> and from device dax files. It also provides a way to zero a device dax
>>> i.e. daxctl io --input=/home/myfile --output=/dev/dax1.0
>> Does that really belong in daxctl?
> I'm open to teaching other utilities how to do mmap based i/o, but is
> there any harm in carrying this dedicated utility for this purpose?
Well, what is the purpose of daxctl, exactly? Given the 'ctl' in the
name, I figured it was used for configuration.
> Certainly the error clearing interface to send custom nvdimm ioctls is
> not something I expect 'cp' or 'dd' to comprehend.
Sure, clearing errors makes sense. I don't see a need to reimplement dd
or xfs_io, though.
Neither dd nor xfs_io tool does what I want which is read from stdin
and write to a device-dax instance. This is explicitly not trying to
be a dd replacement as it enforces that one of the arguments must be a
device-dax instance. If neither input nor output is device-dax the
user is expected to just use dd.