On Tue, Jul 05, 2022 at 04:54:18PM +0100, Al Viro wrote:
> On Tue, Jul 05, 2022 at 08:45:06AM -0700, Keith Busch wrote:
> > From: Keith Busch <kbusch@xxxxxxxxxx>
> >
> > The size being added to a bio from an iov is aligned to a block size
> > after the pages were gotten. If the new aligned size truncates the last
> > page, its reference was being leaked. Ensure all pages that were not
> > added to the bio have their reference released.
> >
> > Since this essentially requires doing the same that bio_put_pages(), and
> > there was only one caller for that function, this patch makes the
> > put_page() loop common for everyone.
> >
> > Fixes: b1a000d3b8ec5 ("block: relax direct io memory alignment")
> > Reported-by: Al Viro <viro@xxxxxxxxxxxxxxxxxx>
> > Signed-off-by: Keith Busch <kbusch@xxxxxxxxxx>
>
> I still very much dislike this. Background: iov_iter_get_pages should
> advance the fucking iterator by the amount it has grabbed. It's
> really much cleaner that way. So your round-down-then-fuck-off-if-zero
> is going to be a clumsy.
I currently don't see a better way to do it, but I'll be happy if you come up
with something.
> Whatever; I can deal with that on top of your patch. Where would you
> have it go wrt tree? Could you do a branch based at the last of your
> original series, so that both Jens and I could pull from it?
Branch pushed here:
https://git.kernel.org/pub/scm/linux/kernel/git/kbusch/linux.git/log/?h=alignment-fixes-rebased
> One thing I would really like to avoid is having the entire #for-5.20/block
> in ancestors of those commits; that would make for a monumental headache
> with iov_iter series ;-/
I'm sorry this is clashing with your work. Please let me know if there's
anything else I can do to help.
