Acked-by: Wang, Yu1 <yu1.wang@...>

Acked-by: Wang, Yu1 <yu1.wang@...>

Then how acrn-dm knows if the asynio is really enabled or not? We need some logs to indicate the asynio status.

There's a concern. If the asynio kernel hasn't merged, then the upstream kernel can not work with the new acrn-dm, right?...

vm_init_asyncio others are LGTM. Acked-by: Wang, Yu1 <yu1.wang@...>

Ok. /* Mitigate the epoll_wait repeat cycles by read out the events as more * as possile.*/ Acked-by: Wang, Yu1 <yu1.wang@...>

I don't understand why it occupied 100% cpu. It get the event and find out nothing needs to be done, right?

What's the real impact of processing repeatly?

But you can not guarantee all data can be read out as there has race.

Let's keep the verb-object fashion. load_pci_rombar Others are LGTM. Acked-by: Wang, Yu1 <yu1.wang@...>

Acked-by: Wang, Yu1 <yu1.wang@...>

Then why do we need to read in loop? read on time is enough?

We don't emulate virtual root port for user VM. It is only used for SOS to prevent physical ACS be disabled by SOS then do P2P attack for Pre-launched VM. Yes. Let's remove it. And the log can be chan

Do we have bus1/bus2 for virtual pci device? I think that we emulate all devices as RCiEPs?

Then this log doesn't make sense... Why we need to mention the virtual BDF here? Checking the class is enough? Ok. this part is OK to me.

Reviewed-by: Wang, Yu1 <yu1.wang@...>

Acked-by: Wang, Yu1 <yu1.wang@...>

The GPU on Bus0 is the iGPU only, right?.. And usually the discrete GPU has the physical ROMBAR, then it should be done by ROMBAR passthrough instead of ROMBAR virtualization?..

Please add a log here. Others are LGTM. Acked-by: Wang, Yu1 <yu1.wang@...>

PCI ROM only supports Intel iGPU. Can we swap the logic to positive flow. if ((rom_header[0] == 0x55) && (rom_header[1] == 0xaa)) { ... } else { pr_warn("Incorrect Rom file format "); ptdev->need_romb