Re: [Qemu-devel] [PATCH v4 2/7] block: Introduce op_blockers to BlockDriverState

[Fam Zheng]

On 2013年11月26日 01:13, Paolo Bonzini wrote:
> Il 22/11/2013 06:24, Fam Zheng ha scritto:
> > +void bdrv_op_block(BlockDriverState *bs, BlockOpType op, Error *reason)
> > +{
> > +    BdrvOpBlocker *blocker;
> > +    assert(op >=0 && op < BLOCK_OP_TYPE_MAX);
> > +
> > +    blocker = g_malloc0(sizeof(BdrvOpBlocker));
> > +    blocker->reason = reason;
> > +    QLIST_INSERT_HEAD(&bs->op_blockers[op], blocker, list);
> > +}
> > +
> > +void bdrv_op_unblock(BlockDriverState *bs, BlockOpType op, Error *reason)
>
> What about making BlockOpType a bitmask, and having bdrv_op_{,un}block
> take multiple ORed BlockOpTypes?
>
> bdrv_op_{,un}block_all then are not necessary, you only need a
> BLOCK_OPERATION_ALL value.

Bitmap is not enough, at least it should be an array. For example when 
we enable multiple block jobs, there're two stoppers for drive_del, right?

And a bool or int is not as friendly as an error message, because when 
an operation is blocked we can't print a specific reason when we don't 
have this information at all. So let's be consistent with migration 
blockers.

Fam