Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Date:   Tue, 21 Jun 2022 10:24:30 +0100
Message-ID: <87r13i2wtt.wl-maz@kernel.org>
From:   Marc Zyngier <maz@kernel.org>
To:     Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc:     Sebastian Ene <sebastianene@google.com>,
        Rob Herring <robh+dt@kernel.org>,
        Arnd Bergmann <arnd@arndb.de>,
        Dragan Cvetic <dragan.cvetic@xilinx.com>,
        linux-kernel@vger.kernel.org, devicetree@vger.kernel.org,
        will@kernel.org, vdonnefort@google.com,
        Guenter Roeck <linux@roeck-us.net>
Subject: Re: [PATCH v7 2/2] misc: Add a mechanism to detect stalls on guest vCPUs
In-Reply-To: <YrGHS48LR759stoL@kroah.com>
References: <20220621080308.3952915-1-sebastianene@google.com>
        <20220621080308.3952915-3-sebastianene@google.com>
        <YrGBBFW2d/scKDeN@kroah.com>
        <6b5bb5e69888f69fcfdcb8c9c2fd2660@kernel.org>
        <YrGHS48LR759stoL@kroah.com>
User-Agent: Wanderlust/2.15.9 (Almost Unreal) SEMI-EPG/1.14.7 (Harue)
 FLIM-LB/1.14.9 (=?UTF-8?B?R29qxY0=?=) APEL-LB/10.8 EasyPG/1.0.0 Emacs/27.1
 (x86_64-pc-linux-gnu) MULE/6.0 (HANACHIRUSATO)
MIME-Version: 1.0 (generated by SEMI-EPG 1.14.7 - "Harue")
Content-Type: text/plain; charset=US-ASCII
Precedence: bulk

On Tue, 21 Jun 2022 09:54:35 +0100,
Greg Kroah-Hartman <gregkh@linuxfoundation.org> wrote:
> 
> On Tue, Jun 21, 2022 at 09:44:35AM +0100, Marc Zyngier wrote:
> > On 2022-06-21 09:27, Greg Kroah-Hartman wrote:
> > > On Tue, Jun 21, 2022 at 08:03:09AM +0000, Sebastian Ene wrote:
> > > > This driver creates per-cpu hrtimers which are required to do the
> > > > periodic 'pet' operation. On a conventional watchdog-core driver, the
> > > > userspace is responsible for delivering the 'pet' events by writing to
> > > > the particular /dev/watchdogN node. In this case we require a strong
> > > > thread affinity to be able to account for lost time on a per vCPU.
> > > > 
> > > > This part of the driver is the 'frontend' which is reponsible for
> > > > delivering the periodic 'pet' events, configuring the virtual
> > > > peripheral
> > > > and listening for cpu hotplug events. The other part of the driver
> > > > handles the peripheral emulation and this part accounts for lost
> > > > time by
> > > > looking at the /proc/{}/task/{}/stat entries and is located here:
> > > > https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/3548817
> > > > 
> > > > Signed-off-by: Sebastian Ene <sebastianene@google.com>
> > > > ---
> > > >  drivers/misc/Kconfig               |  12 ++
> > > >  drivers/misc/Makefile              |   1 +
> > > >  drivers/misc/vcpu_stall_detector.c | 222
> > > > +++++++++++++++++++++++++++++
> > > >  3 files changed, 235 insertions(+)
> > > >  create mode 100644 drivers/misc/vcpu_stall_detector.c
> > > > 
> > > > diff --git a/drivers/misc/Kconfig b/drivers/misc/Kconfig
> > > > index 41d2bb0ae23a..e15c85d74c4b 100644
> > > > --- a/drivers/misc/Kconfig
> > > > +++ b/drivers/misc/Kconfig
> > > > @@ -483,6 +483,18 @@ config OPEN_DICE
> > > > 
> > > >  	  If unsure, say N.
> > > > 
> > > > +config VCPU_STALL_DETECTOR
> > > > +	tristate "VCPU stall detector"
> > > > +	select LOCKUP_DETECTOR
> > > > +	help
> > > > +	  Detect CPU locks on a kvm virtual machine. This driver relies on
> > > > +	  the hrtimers which are CPU-binded to do the 'pet' operation.
> > > > When a
> > > > +	  vCPU has to do a 'pet', it exits the guest through MMIO write and
> > > > +	  the backend driver takes into account the lost ticks for this
> > > > +	  particular CPU.
> > > > +	  To compile this driver as a module, choose M here: the
> > > > +	  module will be called vcpu_stall_detector.
> > > 
> > > Should this depend on KVM_GUEST?
> > 
> > Not all architectures have KVM_GUEST, and arm64 has no use for it.
> 
> Ah, I thought this was a requirement (or created a better guest image)
> for use under KVM.  Nevermind then...

It really depends whether an architecture relies on non-architectural
extensions to support KVM guests. PPC does most of the time, x86
certainly works better with the knowledge that this is a KVM guest.

KVM on arm64 implements the architecture itself, and hardly anything
else (if something sucks in virt, it also likely sucks bare metal).
The couple of KVM-specific options we support are definitely not worth
a KVM_GUEST, as they only cover pretty esoteric stuff that nobody
enables, such as PTP_1588_CLOCK_KVM.

	M.

-- 
Without deviation from the norm, progress is not possible.