On Wed, Jun 13, 2018 at 05:58:01AM -0400, Kent Overstreet wrote:
> On Mon, Jun 11, 2018 at 09:48:00PM +0200, Christoph Hellwig wrote:
> > Hi all,
> > this series cleans up various places where bcache is way too intimate
> > with bio internals. This is intended as a baseline for the multi-page
> > biovec work, which requires some nasty workarounds for the existing
> > code.
> > Note that I do not have a bcache test setup, so this will require
> > some careful actual testing with whatever test cases are used for
> > bcache.
> > Also the new bio_reused helper should be useful at least for MD raid,
> > but that work is left for later.
> Strong nak on the patch series (except for the patch to not clone in
> bch_data_verify, that patch looks fine).
> Unless something has seriously changed with how multi-page bvecs work since I
> started that code nothing in bcache should be broken by it - Ming, can you
> confirm or deny if that's still correct?
Yes, the multi-page bvecs implementation didn't have big change, and
previously I run xfstests on bcache0, and no regression was observed.
> It is true that bch_bio_map() will be suboptimal when multi page bvecs go in,
> but it won't be broken. The way it's used now is really conflating two
> things, but where it's used for mapping a bio to a single buffer (i.e. not
> before bio_alloc_pages) that can be switched to something that just creates a
> single bvec.
Yes, multipage bvec shouldn't break any driver or fs.