Re: [PATCH 02/26] qcow2: remove incorrect coroutine_fn annotations

[Stefan Hajnoczi]

On Tue, Apr 19, 2022 at 01:07:19PM -0500, Eric Blake wrote:
> On Fri, Apr 15, 2022 at 03:18:36PM +0200, Paolo Bonzini wrote:
> > Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
> > ---
> 
> Again, a sentence on why this is correct would be helpful.
> 
> >  block/qcow2-refcount.c | 4 ++--
> >  block/qcow2.h          | 4 ++--
> >  2 files changed, 4 insertions(+), 4 deletions(-)
> > 
> > diff --git a/block/qcow2-refcount.c b/block/qcow2-refcount.c
> > index b91499410c..b6f90b2702 100644
> > --- a/block/qcow2-refcount.c
> > +++ b/block/qcow2-refcount.c
> > @@ -1206,7 +1206,7 @@ void qcow2_free_any_cluster(BlockDriverState *bs, 
> > uint64_t l2_entry,
> >      }
> >  }
> >  
> > -int coroutine_fn qcow2_write_caches(BlockDriverState *bs)
> > +int qcow2_write_caches(BlockDriverState *bs)
> >  {
> >      BDRVQcow2State *s = bs->opaque;
> >      int ret;
> > @@ -1226,7 +1226,7 @@ int coroutine_fn qcow2_write_caches(BlockDriverState 
> > *bs)
> >      return 0;
> >  }
> >  
> > -int coroutine_fn qcow2_flush_caches(BlockDriverState *bs)
> > +int qcow2_flush_caches(BlockDriverState *bs)
> >  {
> >      int ret = qcow2_write_caches(bs);
> 
> Both of these eventually hit qcow2_cache_write, which is not marked
> coroutine, so these should not be either.

coroutine_fn may call non-coroutine_fn, so this alone is not a reason
for removing it from qcow2_write_caches().

There must be a call chain where qcow2_write_caches() and
qcow2_flush_caches() are is invoked from outside coroutine_fn.

Stefan

signature.asc
Description: PGP signature