[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [RFC PATCH 0/2] xenbus: Batch foreign page mappings


  • To: Rafał Wojdyła <omeg@xxxxxxxxxxxxxxxxxxxxxx>, win-pv-devel@xxxxxxxxxxxxxxxxxxxx
  • From: Tu Dinh <ngoc-tu.dinh@xxxxxxxxxx>
  • Date: Tue, 10 Nov 2026 19:40:07 +0100
  • Authentication-results: eu.smtp.expurgate.cloud; dkim=pass header.s=selector1 header.d=vates.tech header.i="@vates.tech" header.h="From:Subject:Date:Message-ID:To:MIME-Version:Content-Type:In-Reply-To:References:Feedback-ID"
  • Delivery-date: Tue, 10 Nov 2026 18:40:14 +0000
  • Feedback-id: default:8631fc262581453bbf619ec5b2062170:Sweego
  • List-id: Developer list for the Windows PV Drivers subproject <win-pv-devel.lists.xenproject.org>

On 10/11/2026 18:32, Rafał Wojdyła wrote:
[...]
> 
> I've iterated a lot on the xeniface part but I'll send what I have now. 
> I'll be on vacation for two weeks and it's quite a bit of code, so there 
> will be some time for review. My main point of uncertainty is what to do 
> if repeated unmap retries still fail. I'm partial to treating that as a 
> fatal failure because we can't really do much there: we could quarantine 
> the memory, but the peer would still have pending pages, and we can't 
> just defer infinitely.
> 

IMO, it's better to just keep the IRP stuck in canceling until the 
backend map is released. Otherwise a mapping created between userspace 
and backend could cause the guest to crash. Windows seems to know how to 
handle long-running cancellation anyway, and holding the IRP in 
cancellation is a good way to keep the memory quarantined:

https://learn.microsoft.com/en-us/windows-hardware/drivers/kernel/canceling-irps


--
Ngoc Tu Dinh | Vates XCP-ng Developer

XCP-ng & Xen Orchestra - Vates solutions

web: https://vates.tech

 


Rackspace

Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.