On 05/01/2022 16:18, Daniel P. Berrangé wrote:
On Wed, Jan 05, 2022 at 04:00:54PM +0100, Laurent Vivier wrote:
On 05/01/2022 15:55, Daniel P. Berrangé wrote:
On Wed, Jan 05, 2022 at 03:49:12PM +0100, Laurent Vivier wrote:
On 05/01/2022 13:38, Daniel P. Berrangé wrote:
The -device JSON syntax impl leaks a reference on the created
DeviceState instance. As a result when you hot-unplug the
device, the device_finalize method won't be called and thus
it will fail to emit the required DEVICE_DELETED event.
A 'json-cli' feature was previously added against the
'device_add' QMP command QAPI schema to indicated to mgmt
apps that -device supported JSON syntax. Given the hotplug
bug that feature flag is no unusable for its purpose, so
Not sure to understand: do you mean "now unusable"?
An application wants to known whether QEMU can support JSON
syntax with -device. If they look for the 'json-cli' feature
as a witness, they'll end up using JSON with QEMU 6.2 which
is giving them broken hotplug. This is unusable for any
non-trivial use cases. So we need a new witness to indicate
whether JSON is viable with -device, that only the newly
fixed QEMU will report.
I understand that, my problem was with your sentence:
"Given the hotplug bug that feature flag is no unusable for its purpose"
What's the problem with that ? It is reasonabled to say a -device impl
which is broken for hotplug is not usable for non-toy use cases.
The problem for me is the double negation: "no" and "unusable"
Thanks,
Laurent
