Re: [Qemu-block] [PATCH v10 03/16] block: Use block_job_add_bdrv() in mirror_start_job()

[Kevin Wolf]

Am 11.10.2016 um 15:46 hat Alberto Garcia geschrieben:
> On Mon 10 Oct 2016 06:03:41 PM CEST, Kevin Wolf wrote:
> 
> >> Use block_job_add_bdrv() instead of blocking all operations in
> >> mirror_start_job() and unblocking them in mirror_exit().
> >> 
> >> Signed-off-by: Alberto Garcia 
> >
> > Compared to the old code, this unblocks BLOCK_OP_TYPE_DATAPLANE, i.e.
> > you can now run a dataplane device on a BDS used as the mirror target.
> >
> > This means that the target could require a different AioContext than
> > the source, which we can't support. So it seems unlikely to me that we
> > can lift this restriction.
> 
> Thanks, I'll fix it.
> 
> What happens if you run a dataplane on the source, though? That's
> currently allowed as far as I'm aware. Wouldn't that have a similar
> effect?

The block job takes care to put the target into the same dataplane
AioContext then. The job doesn't really care whether it works in the
main thread or a separate I/O thread, it just requires that it's a
single context, which is currently defined by the source.

Kevin

Re: [Qemu-block] [PATCH v10 03/16] block: Use block_job_add_bdrv() in mirror_start_job()

[Alberto Garcia]

On Mon 10 Oct 2016 06:03:41 PM CEST, Kevin Wolf wrote:

>> Use block_job_add_bdrv() instead of blocking all operations in
>> mirror_start_job() and unblocking them in mirror_exit().
>> 
>> Signed-off-by: Alberto Garcia 
>
> Compared to the old code, this unblocks BLOCK_OP_TYPE_DATAPLANE, i.e.
> you can now run a dataplane device on a BDS used as the mirror target.
>
> This means that the target could require a different AioContext than
> the source, which we can't support. So it seems unlikely to me that we
> can lift this restriction.

Thanks, I'll fix it.

What happens if you run a dataplane on the source, though? That's
currently allowed as far as I'm aware. Wouldn't that have a similar
effect?

Berto

Re: [Qemu-block] [PATCH v10 03/16] block: Use block_job_add_bdrv() in mirror_start_job()

[Paolo Bonzini]

On 10/10/2016 18:03, Kevin Wolf wrote:
>> > Use block_job_add_bdrv() instead of blocking all operations in
>> > mirror_start_job() and unblocking them in mirror_exit().
>> > 
>> > Signed-off-by: Alberto Garcia 
> Compared to the old code, this unblocks BLOCK_OP_TYPE_DATAPLANE, i.e.
> you can now run a dataplane device on a BDS used as the mirror target.
> 
> This means that the target could require a different AioContext than the
> source, which we can't support. So it seems unlikely to me that we can
> lift this restriction.

Indeed, this is in the pipeline, but still quite far away!

Paolo

Re: [Qemu-block] [PATCH v10 03/16] block: Use block_job_add_bdrv() in mirror_start_job()

[Kevin Wolf]

Am 06.10.2016 um 15:02 hat Alberto Garcia geschrieben:
> Use block_job_add_bdrv() instead of blocking all operations in
> mirror_start_job() and unblocking them in mirror_exit().
> 
> Signed-off-by: Alberto Garcia 

Compared to the old code, this unblocks BLOCK_OP_TYPE_DATAPLANE, i.e.
you can now run a dataplane device on a BDS used as the mirror target.

This means that the target could require a different AioContext than the
source, which we can't support. So it seems unlikely to me that we can
lift this restriction.

Kevin