[Drbd-dev] Behaviour of verify: false positives -> true positives

[Lars Ellenberg]

On Wed, Oct 01, 2008 at 01:28:02PM +0200, Thomas Schoebel-Theuer wrote:
> > as private_bio is a clone of master_bio,
> > it shares its bvec, thus the pages.
> 
> Oops, I didn't realize that (and I tested the copy version only briefly 
> because I was more interested in the checksumming because of the kernel log 
> messages). The implications are clear to me.
> 
> IMHO best solution would be the existence of a bio_copy() in addition to 
> bio_clone() in bio.h, but even if we would get it into the upstream it would 
> not help for elder kernels.
> 
> > you need to remember the original pages, and you need to restore them
> > before completion of the master bio.
> 
> Well, one problem is that the length of bvec could be nearly arbitrary (in 
> theory),

BIO_MAX_PAGES
DRBD_MAX_SEGMENT_SIZE

> so preclaiming "enough" space in struct drbd_request is probably no 
> good idea. Well, kmalloc() would be possible to keep the allocation dynamic 
> without wasting space.

I meant the pre-allocating in drbd_pp_alloc. number of pages.

> But what about simply generating a completely new bio and copying over all the 
> stuff by hand? This would mean to implement some sort of bio_copy() in the 
> local code which could then later be lifted upstreams if other people liked 
> it too. What do you think is better?

no, I think it would be enough to just set (pseudo code)
copy_page->private = orig_page;
bvec->bv_page = copy_page;

then, later before completion,
bio_for_each_segment
  memcmp,
  restore orig
  drbd_pp_free copy

-- 
: Lars Ellenberg
: LINBIT | Your Way to High Availability
: DRBD/HA support and consulting http://www.linbit.com

DRBD® and LINBIT® are registered trademarks of LINBIT, Austria.