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

Re: [PATCH v3 0/6] x86/intr: HVM guest interrupt handling fixes/cleanup

  • To: Jan Beulich <jbeulich@xxxxxxxx>, Ian Jackson <iwj@xxxxxxxxxxxxxx>
  • From: Roger Pau Monné <roger.pau@xxxxxxxxxx>
  • Date: Wed, 27 Jan 2021 10:21:28 +0100
  • Arc-authentication-results: i=1; mx.microsoft.com 1; spf=pass smtp.mailfrom=citrix.com; dmarc=pass action=none header.from=citrix.com; dkim=pass header.d=citrix.com; arc=none
  • Arc-message-signature: i=1; a=rsa-sha256; c=relaxed/relaxed; d=microsoft.com; s=arcselector9901; h=From:Date:Subject:Message-ID:Content-Type:MIME-Version:X-MS-Exchange-SenderADCheck; bh=hJFLVXrzh51xs6M57WFMzfYS+J4wnad27iXTxQ/rp4w=; b=I9c2k7MHByaXPtgMdIr0SkGZlX1OJ9XyjxzlfYQWKNbzMyd5JXT15/WYl4mcHqXtdmjqO5NlX8x85dSePD07zdsgK8SbgBfNS36aCKeR06gdZnQf/YsRsu3CxKx/TtBXas9cBjz0fSkr0CsoYSLkUSlqEnkqmeQSS5rbwin3vO2kiIs9drZOZ3zicVAMHkqF7Sn9T7xK6YzQUUanZXcpSEfmHu4w6Fcpbf5aQt/qejLmwjDDK5JCdmzAT1pOltXGxXowgC8+ryStGJdatJCIVoDB/PnUGaWWgzI5c/zorTMXcf3uXFgsDWrG2ivEssO55FIdj2FK0/EzadwhvutLZw==
  • Arc-seal: i=1; a=rsa-sha256; s=arcselector9901; d=microsoft.com; cv=none; b=FXcfxMx7PjNeFi8eunFz6QK3MkxDvhny6dJpkrwkMvSPjalE3Hbw3unPpdZVSwoe/UbggBl1BKtHx/0GzBrcGBv7rO33LcCzhC0uH40+bl2u7h0Jz76bL47iwAgX6PpnII7gX9B1Gu0eql4LEnM3QUqiKcr795C0Apwvw3jrZ06iNIeIBBDx4x+j30wK9aXQfTaJimjHVOoZv5Pa88Z0Mg+dzQSHvapGnH109jj0wm6UCN1FVTV6bv7clb0vOqOBZWNc/2brGtUFUTCvB2p0M65KcF61wUVMyQTgMo4tTexXzikZglbhzHwRzLSJqtTfPv/V9Gtjxz64A2CDOJF+WQ==
  • Authentication-results: esa6.hc3370-68.iphmx.com; dkim=pass (signature verified) header.i=@citrix.onmicrosoft.com
  • Cc: Andrew Cooper <andrew.cooper3@xxxxxxxxxx>, Wei Liu <wl@xxxxxxx>, Kevin Tian <kevin.tian@xxxxxxxxx>, Paul Durrant <paul@xxxxxxx>, <xen-devel@xxxxxxxxxxxxxxxxxxxx>
  • Delivery-date: Wed, 27 Jan 2021 09:21:44 +0000
  • Ironport-sdr: ASrrKvq9f07i+3Dy8XzJkYDL0h7hcQ4H1WtqRM7hrVZ7KALCu8+Sjv9sbIhlkG5RmrvJhsBp/G PIsiiQryPQc6+2h5mQ9PIzM62oqOhHdVJxKGbAaemFHw4wFErJby/xN7SiMCQoPmE7Vs4i6b41 TzKveqNAJkkJKnZ4Tv+lMhU+vTLD92kk5mOB9KeEKuDYpgJZV4ywsR2/F8usGlLCsWJ3PzJ62n PuSpj1ivDEFWOtkxhlMM5WBOHEZLnf9hH/XqUXOA02SJ2LMiA91/hIwlglZPVbJ/kKEGmqhw5c OeQ=
  • List-id: Xen developer discussion <xen-devel.lists.xenproject.org>

On Tue, Jan 26, 2021 at 06:07:23PM +0100, Jan Beulich wrote:
> Ian,
> On 26.01.2021 14:45, Roger Pau Monne wrote:
> > The following series aims to fix some shortcomings of guest interrupt
> > handling when using passthrough devices. The first 5 patches are
> > bugfixes or cleanups, which I think should solve the issue(s) that
> > caused the dpci EOI timer to be introduced. However neither me nor
> > others seem to be able to reproduce the original issue, so it's hard to
> > tell.
> > 
> > It's my opinion that we should remove the timer and see what explodes
> > (if anything). That's the only way we will be able to figure out what
> > the original issue was, and how to fix it without introducing yet
> > another per-guest-irq related timer.
> > 
> > Thanks, Roger.
> > 
> > Roger Pau Monne (6):
> >   x86/vioapic: top word redir entry writes don't trigger interrupts
> >   x86/vioapic: issue EOI to dpci when switching pin to edge trigger mode
> >   x86/vpic: force int output to low when in init mode
> >   x86/vpic: don't trigger unmask event until end of init
> >   x86/vpic: issue dpci EOI for cleared pins at ICW1
> >   x86/dpci: remove the dpci EOI timer
> while half of this series was still submitted in time, I'd still
> like to raise the question of including part or all of it in
> 4.15. In particular the last change is one which I would prefer
> to see happen early in a release cycle. Risk assessment is
> pretty difficult, I'm afraid (Roger can correct me here), as at
> least some of what gets adjusted are cases we don't normally
> expect to be exercised. (FAOD patch 5 is still pending a R-b
> tag.)
> Roger, if you could give your own judgement on which of the
> changes you would view as more or less clear 4.15 candidates,
> this may help Ian take a decision.

I agree, the only risky patch is the last one, mainly because we have
no way to reproduce the issue that code was fixing. It's my assumption
that all the fixes prior to the last patch should address the same
issues the timer was trying to address.

So my recommendation would be to not commit the last patch unless all
the prior ones have been committed, and that would include 5/6 which
is still missing a R-b.

Thanks, Roger.



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