Re: [PATCH 6/8] virtio-blk: mark IO_CODE functions

[Stefan Hajnoczi]

On Fri, Jul 08, 2022 at 11:19:43AM +0200, Emanuele Giuseppe Esposito wrote:
> 
> 
> Am 05/07/2022 um 16:39 schrieb Stefan Hajnoczi:
> > On Thu, Jun 09, 2022 at 10:37:25AM -0400, Emanuele Giuseppe Esposito wrote:
> >> Just as done in the block API, mark functions in virtio-blk
> >> that are called also from iothread(s).
> >>
> >> We know such functions are IO because many are blk_* callbacks,
> >> running always in the device iothread, and remaining are propagated
> >> from the leaf IO functions (if a function calls a IO_CODE function,
> >> itself is categorized as IO_CODE too).
> >>
> >> Signed-off-by: Emanuele Giuseppe Esposito <eesposit@redhat.com>
> >> ---
> >>  hw/block/dataplane/virtio-blk.c |  4 ++++
> >>  hw/block/virtio-blk.c           | 35 +++++++++++++++++++++++++++++++++
> >>  2 files changed, 39 insertions(+)
> > 
> > The definition of IO_CODE() is:
> > 
> >   I/O API functions. These functions are thread-safe, and therefore
> >   can run in any thread as long as the thread has called
> >   aio_context_acquire/release().
> > 
> > I'm not sure it matches with the exact semantics you have in mind. Are
> > they really allowed to be called from any thread and even from multiple
> > threads? Or maybe just from the BlockBackend's AioContext thread?
> 
> I think it is just from the BlockBackend's AioContext thread. But I
> classified blk_* functions as IO_CODE.
> 
> What is your opinion on that?

There is a difference between blk_*() APIs and device emulation code.
Device emulation code controls exactly where it runs (vCPU thread, main
loop, IOThread). blk_*() APIs may be called from more of contexts and
they have no control over it.

I'd like to make sure that the annotations match the actual usage that
the code was designed for.

Stefan

signature.asc
Description: PGP signature