Re: [Qemu-devel] [PATCH RFC 0/6] virtio: refactor host notifiers

[Fam Zheng]

On Thu, 03/17 11:01, Cornelia Huck wrote:
> As discussed in the virtio-blk dataplane start/stop thread, the
> various transports contain a lot of similar/identical boilerplate
> code for setting up ioeventfds. This makes the code hard to follow.
> 
> Let's drag all of the common handling into the virtio-bus layer,
> hopefully reducing both the boilerplate code and the number of
> indirections.
> 
> This RFC patchset is the result of hacking this up quickly, and I
> was rather surprised that it did not die immediately on my s390
> system (which is all I 'tested'). So be warned :)
> 
> Feedback about the approach and the interface would be good.

I tested this series with the two assertions (Paolo's "test" reentrancy
detection and my "assert(is_in_iothread())", in virtio-blk request handler),
and it seems to fix the crash . (A few seconds to crash on master vs.  several
hours survival).

Fam

> 
> Cornelia Huck (6):
>   virtio-bus: common ioeventfd infrastructure
>   virtio-bus: have callers tolerate new host notifier api
>   virtio-ccw: convert to ioeventfd callbacks
>   virtio-pci: convert to ioeventfd callbacks
>   virtio-mmio: convert to ioeventfd callbacks
>   virtio-bus: remove old set_host_notifier callback
> 
>  hw/block/dataplane/virtio-blk.c |   6 +-
>  hw/s390x/virtio-ccw.c           | 133 
> ++++++++++++++--------------------------
>  hw/scsi/virtio-scsi-dataplane.c |   9 ++-
>  hw/virtio/vhost.c               |  13 ++--
>  hw/virtio/virtio-bus.c          | 108 ++++++++++++++++++++++++++++++++
>  hw/virtio/virtio-mmio.c         | 128 +++++++++++++-------------------------
>  hw/virtio/virtio-pci.c          | 124 +++++++++++++------------------------
>  include/hw/virtio/virtio-bus.h  |  15 ++++-
>  8 files changed, 263 insertions(+), 273 deletions(-)
> 
> -- 
> 2.6.5
>