Hi Matthew,
On Tue, May 19, 2020 at 08:16:32AM -0700, Matthew Wilcox wrote:
> On Tue, May 19, 2020 at 06:06:19PM +0800, Gao Xiang wrote:
> > In addition, I found some limitation of new {attach,detach}_page_private
> > helper (that is why I was interested in this series at that time [1] [2],
> > but I gave up finally) since many patterns (not all) in EROFS are
> >
> > io_submit (origin, page locked):
> > attach_page_private(page);
> > ...
> > put_page(page);
> >
> > end_io (page locked):
> > SetPageUptodate(page);
> > unlock_page(page);
> >
> > since the page is always locked, so io_submit could be simplified as
> > set_page_private(page, ...);
> > SetPagePrivate(page);
> > , which can save both one temporary get_page(page) and one
> > put_page(page) since it could be regarded as safe with page locked.
>
> It's fine to use page private like this without incrementing the refcount,
> and I can't find any problematic cases in EROFS like those fixed by commit
> 8e47a457321ca1a74ad194ab5dcbca764bc70731
>
> So I think the new helpers are not for you, and that's fine. They'll be
> useful for other filesystems which are using page_private differently
> from the way that you do.
Yes, I agree. Although there are some dead code in EROFS to handle
some truncated case, which I'd like to use in the future. Maybe I
can get rid of it temporarily... But let me get LZMA fixed-sized
output compression for EROFS in shape at first, which seems useful
as a complement of LZ4...
Thanks,
Gao Xiang
