On 14 July 2017 at 10:42, Paolo Bonzini <pbonz...@redhat.com> wrote: > On 14/07/2017 04:14, Fam Zheng wrote: >> v4: Add a pointer type param to DEFINE_PROP_LINK, don't change pointer types >> to >> "Object *". [Paolo] >> >> v3: Include Igor's patch, and fix virtio-crypto too. >> Always use qdev_prop_allow_set_link_before_realize and >> OBJ_PROP_LINK_UNREF_ON_RELEASE. >> Include as many applicable devices as possible and some more clean-ups. >> There are still more left but they use uncommon check and flags >> parameters, >> it's better to take care of them in following batches. >> >> v2: Create a new header for link properties. [Paolo] >> Don't wrap, use PropertyInfo.create() (much better diffstat, yay!). >> [Paolo] >> >> Link properties of devices created with object_property_add_link() are not >> reflected in HMP "info qtree". For example, whether a virtio-blk device has >> an >> iothread (i.e. has enabled data plane) can not be introspected easily. >> >> Introduce a new type of qdev property macro to fix that.
> I'm leaving the ARM patches to Peter and respinning my pull request with > the others. OK. Fam, you'll need to send a patchset with whatever the stuff I need to pick up is, then (and it'll need to get to me by monday given the freeze date is tuesday). thanks -- PMM
