Re: [Qemu-block] [Qemu-devel] [PATCH 03/12] block: Introduce BlockBackendPublic

[Kevin Wolf]

Am 22.03.2016 um 22:53 hat Eric Blake geschrieben:
> On 03/22/2016 09:33 AM, Kevin Wolf wrote:
> > Some features, like I/O throttling, are implemented outside
> > block-backend.c, but still want to keep BlockBackends in a list. In
> > order to avoid exposing the whole struct layout in the public header
> > file, this patch introduces an embedded public struct where list entry
> > structs can be added and a pair of functions to convert between
> > BlockBackend and BlockBackendPublic.
> > 
> > Signed-off-by: Kevin Wolf <address@hidden>
> > ---
> >  block/block-backend.c          | 17 +++++++++++++++++
> >  include/sysemu/block-backend.h |  9 +++++++++
> >  2 files changed, 26 insertions(+)
> > 
> > diff --git a/block/block-backend.c b/block/block-backend.c
> > index df8f717..4394950 100644
> > --- a/block/block-backend.c
> > +++ b/block/block-backend.c
> > @@ -33,6 +33,7 @@ struct BlockBackend {
> >      DriveInfo *legacy_dinfo;    /* null unless created by drive_new() */
> >      QTAILQ_ENTRY(BlockBackend) link;         /* for block_backends */
> >      QTAILQ_ENTRY(BlockBackend) monitor_link; /* for monitor_block_backends 
> > */
> > +    BlockBackendPublic public;
> 
> Any reason to not put the public struct at offset 0?

No, but also no reason to put it there. :-)

> > +
> > +/*
> > + * Returns a BlockBackend given the associated @public fields.
> > + */
> > +BlockBackend *blk_by_public(BlockBackendPublic *public)
> > +{
> > +    return container_of(public, BlockBackend, public);
> > +}
> 
> At least container_of() doesn't care, so I guess it doesn't matter.
> 
> > +/* This struct is embedded in (the private) BlockBackend struct and 
> > contains
> > + * fields that must be public. This is in particular for QLIST_ENTRY() and
> > + * friends so that BlockBackends can be kept in lists outside 
> > block-backend.c */
> > +typedef struct BlockBackendPublic {
> > +} BlockBackendPublic;
> 
> No fields?  So really all we need this for is so that we can have an
> address of a member of the larger struct, so that we can do list
> operations based on that address?

Well, a list still needs a QLIST_ENTRY, otherwise I could have directly
used BlockBackend. I just tried to keep the introduction of .public
separate from the addition of a specific list.

I think it's strictly speaking invalid C to have an empty struct, but
gcc compiles it and so I thought it should be acceptable to have one for
a single commit until something is added there.

Kevin

pgpOe5nm85kDf.pgp
Description: PGP signature