Re: [PATCH v2 00/26] block: fix coroutine_fn annotations

[Daniel P . Berrangé]

On Mon, May 09, 2022 at 12:29:53PM +0200, Paolo Bonzini wrote:
> This is the initial result of reviving Marc-André's series at
> https://patchew.org/QEMU/20170704220346.29244-1-marcandre.lureau@redhat.com/.
> A lot of the patches are similar to the ones that Marc-André wrote,
> but due to the changes in the code it was easier to redo them.
> 
> For nbd, the patch is on top of "nbd: mark more coroutine_fns" that
> I sent a few days ago and that (AIUI) Eric has already queued; only
> one function was missing, much to my surprise.
> 
> Apart from this, I also identified the following functions that
> can be called both in coroutine context and outside:

snip

> - qio_channel_readv_full_all_eof
> - qio_channel_writev_full_all

I'm trying to understand what criteria makes those two functions
liable for the annotation, but not others in the I/O code ?

What is the actual rule for when to apply 'coroutine_fn' annotation
to a function, and does it apply transitively to up and/or down the
call stack ? 


With regards,
Daniel
-- 
|: https://berrange.com      -o-    https://www.flickr.com/photos/dberrange :|
|: https://libvirt.org         -o-            https://fstop138.berrange.com :|
|: https://entangle-photo.org    -o-    https://www.instagram.com/dberrange :|