Sebastien Boeuf

> @sboeuf we haven't been working on this lately, will probably pick it up again sometime in the next few weeks, I'll keep you in the loop. Sounds good, thanks...

Hi @tmakatos, A friendly ping to ask if you made any progress with this topic?

We need to define clearly what it means to pull security fixes for the dependencies. Are we gonna update the dependencies based on the patch version or the minor version?...

> Thanks to @weltling we have some good documentation on releases. Can we restart the conversation about LTS releases. I'm thinking about making v25.0 since we have some deprecations that...

I'm in favor of using `virtio` as the way to expose a vTPM to the guest. Since it's already there for `crosvm`, we could simply pull their virtio driver into...

This could directly benefit from the approach proposed [here](https://github.com/cloud-hypervisor/cloud-hypervisor/issues/2594) of using `vfio-user`.

@rbradford Just a note that TPM drivers expect a TPM device to be located on an i2c bus and not a PCI one. In order to overcome this limitation, we...

>There's nothing specific to PCI in the vfio-user spec, for what it's worth. libvfio-user is based around PCI, certainly. I know, but let's be honest, both VFIO and vfio-user have...

@liuw so this proxy code in Cloud Hypervisor takes care of placing a device on the MMIO address space? About the ACPI part, that's tricky as it means we either...

> > For the datapath, that means the virtqueues are handled through DMA directly between the hardware and the guest, while the control path goes through the vDPA framework. >...