Re: [PATCH] Add log to Xenfilt DriverAddDevice

[Paul Durrant <xadimgnik@xxxxxxxxx>]

On 13/07/2021 09:25, Martin Harvey wrote:
> Yes, we definitely need to log all filter AddDevice calls. Here's why:
>
> Cut a long story short, the customer likes to do all the things to a VM which 
> they normally like to do to windows machines. This includes upgrading windows 
> editions via windows update on Win10, and also in some cases, mounting the ISO 
> for a newer windows server version, and upgrading the version of windows server.
>
> These upgrade cases do some driver migration steps, which often (but not 
> always) work perfectly. Unfortunately, we've seen cases where class UpperFilter 
> registry keys go missing / don't work, such that XenFilt no longer binds to 
> ACPI/PNP0A03 (PCIBus if memory serves).
>
> If XenFilt can't filter the appropriate devices, **Everything** breaks - 
> particularly, Plug/Unplug doesn't work right, and the machine continually 
> prompts for reboot. I can go back thru case reports and give more details if 
> required.
>
> Tracking this down as the newbie back in early 2020 took **weeks**, where it 
> could have taken a day or so. That debug print would make debugging similar 
> cases very much easier, and it's a problem we're likely to run into again.
>
> Please include the debug print for developers sanity.
That sounds like a justification :-) Care to send a v2 of the patch with 
(at least some of) this in the commit message?
  Cheers,

    Paul

> MH.
>
>
> -----Original Message-----
> From: win-pv-devel <win-pv-devel-bounces@xxxxxxxxxxxxxxxxxxxx> On Behalf Of 
> Paul Durrant
> Sent: 12 July 2021 20:06
> To: win-pv-devel@xxxxxxxxxxxxxxxxxxxx
> Subject: Re: [PATCH] Add log to Xenfilt DriverAddDevice
>
> > Log all calls to DriverAddDevice, and the type of the device if
> > possible
> Any particular reason existent logging is insufficient? I think we need more 
> justification here.