Re: [dm-devel] [PATCH v5 12/12] block: Only clone bio vecs that are in use

[Muthu Kumar <muthu.lkml-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>]

Tejun,

On Thu, Aug 9, 2012 at 12:01 AM, Tejun Heo <tj-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org> wrote:
> Hello,
>
> On Wed, Aug 08, 2012 at 04:47:46PM -0700, Muthu Kumar wrote:
>> You are changing the meaning of __bio_clone() here. In old code, the
>> number of io_vecs, bi_idx, bi_vcnt are preserved. But in this modified
>> code, you are mapping bio_src's bi_iovec[bi_idx] to bio_dests
>> bi_iovec[0] and also restricting the number of allocated io_vecs of
>> the clone. It may be useful for cases were we would like a identical
>> copy of the original bio (may not be in current code base, but this
>> implementation is definitely not what one would expect from the name
>> "clone").
>
> Implementation details changed somewhat but the high-level semantics
> didn't change at all.  Any driver not messing with bio internals - and
> they shouldn't - shouldn't notice the change.

The reason for doing this change is because the code in question is
messing with bio internals.

No in-kernel drivers
> seem to be broken by the change.  If you ask me, this looks more like
> a bug fix to me where the bug is a silly behavior restricting
> usefulness of the interface.
>
>> May be, call this new implementation some thing else (and use it for bcache)?
>
> This doesn't only change __bio_clone() but all clone interface stacked
> on top of it, so, no way.

>This ain't windows.

ah... when you put it this way, it gets a different perspective :)

Anyway, my point is, we shouldn't make it non-obvious ("clone" should
be just "clone"). But, we can always add more comments i guess.

Regards,
Muthu


>
> Thanks.
>
> --
> tejun