Re: [PATCH v5 2/8] drivers/pmem: Allow pmem_clear_poison() to accept arbitrary offset and len

[Vivek Goyal <vgoyal@redhat.com>]

On Tue, Feb 25, 2020 at 02:49:30PM -0800, Dan Williams wrote:
[..]
> > > I'm ok with replacing blkdev_issue_zeroout() with a dax operation
> > > callback that deals with page aligned entries. That change at least
> > > makes the error boundary symmetric across copy_from_iter() and the
> > > zeroing path.
> >
> > IIUC, you are suggesting that modify dax_zero_page_range() to take page
> > aligned start and size and call this interface from
> > __dax_zero_page_range() and get rid of blkdev_issue_zeroout() in that
> > path?
> >
> > Something like.
> >
> > __dax_zero_page_range() {
> >   if(page_aligned_io)
> >         call_dax_page_zero_range()
> >   else
> >         use_direct_access_and_memcpy;
> > }
> >
> > And other callers of blkdev_issue_zeroout() in filesystems can migrate
> > to calling dax_zero_page_range() instead.
> >
> > If yes, I am not seeing what advantage do we get by this change.
> >
> > - __dax_zero_page_range() seems to be called by only partial block
> >   zeroing code. So dax_zero_page_range() call will remain unused.
> >
> >
> > - dax_zero_page_range() will be exact replacement of
> >   blkdev_issue_zeroout() so filesystems will not gain anything. Just that
> >   it will create a dax specific hook.
> >
> > In that case it might be simpler to just get rid of blkdev_issue_zeroout()
> > call from __dax_zero_page_range() and make sure there are no callers of
> > full block zeroing from this path.
> 
> I think you're right. The path I'm concerned about not regressing is
> the error clearing on new block allocation and we get that already via
> xfs_zero_extent() and sb_issue_zeroout().

Well I was wrong. I found atleast one user which uses __dax_zero_page_range()
to zero full PAGE_SIZE blocks.

xfs_io -c "allocsp 32K 0" foo.txt

In that case, I will add a new dax method say dax_zero_page_range() which will
only take PAGE_SIZE aligned range will clear known poison and call that
from __dax_zero_page_range() if I/O is PAGE_SIZE aligned.

For now I will limit this interface to take only single page at a time
because otherwise implementation becomes complex in dm/md stack where
range has to be broken across multiple devices and there are no users
because current iomap API passes one page at a time.

So once we have grown users, then one can also tackle the complexity of
modifying dm/md to break a page range across multiple devices.

Will post a patch series for this.

Thanks
Vivek