Re: [PATCH v4 2/2] misc: Add a mechanism to detect stalls on guest vCPUs

From: Rob Herring <robh@kernel.org>
To: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Sebastian Ene <sebastianene@google.com>,
	linux-kernel@vger.kernel.org,
	Derek Kiernan <derek.kiernan@xilinx.com>,
	Dragan Cvetic <dragan.cvetic@xilinx.com>,
	Arnd Bergmann <arnd@arndb.de>,
	devicetree@vger.kernel.org, qperret@google.com, will@kernel.org,
	Guenter Roeck <linux@roeck-us.net>
Subject: Re: [PATCH v4 2/2] misc: Add a mechanism to detect stalls on guest vCPUs
Date: Fri, 29 Apr 2022 16:03:45 -0500	[thread overview]
Message-ID: <YmxSsSxncvMak+L1@robh.at.kernel.org> (raw)
In-Reply-To: <YmuyLCdpnCzGnILI@kroah.com>

On Fri, Apr 29, 2022 at 11:38:52AM +0200, Greg Kroah-Hartman wrote:
> On Fri, Apr 29, 2022 at 09:26:26AM +0000, Sebastian Ene wrote:
> > On Fri, Apr 29, 2022 at 10:51:14AM +0200, Greg Kroah-Hartman wrote:
> > > On Fri, Apr 29, 2022 at 08:30:33AM +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/vm-watchdog.c | 206 +++++++++++++++++++++++++++++++++++++
> > > >  3 files changed, 219 insertions(+)
> > > >  create mode 100644 drivers/misc/vm-watchdog.c
> > > > 
> > > > diff --git a/drivers/misc/Kconfig b/drivers/misc/Kconfig
> > > > index 2b9572a6d114..26c3a99e269c 100644
> > > > --- a/drivers/misc/Kconfig
> > > > +++ b/drivers/misc/Kconfig
> > > > @@ -493,6 +493,18 @@ config OPEN_DICE
> > > >  
> > > >  	  If unsure, say N.
> > > >  
> > > > +config VM_WATCHDOG
> > > > +	tristate "Virtual Machine Watchdog"
> > > > +	select LOCKUP_DETECTOR
> > > > +	help
> > > > +	  Detect CPU locks on the 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.
> > 
> > Hi,
> > 
> > > 
> > > There's nothing to keep this tied to a virtual machine at all, right?
> > > You are just relying on some iomem address to be updated, so it should
> > > be a "generic_iomem_watchdog" driver as there's nothing specific to vms
> > > at all from what I can tell.
> > > 
> > > thanks,
> > > 
> > > greg k-h
> > 
> > That's right although I might think of using the term "generic lockup detector"
> > instead of watchdog. The only reason why I would keep "virtual machine"
> > word in, is that there is no actual hardware for this.
> 
> That doesn't really matter, it's just a memory location in device tree
> that you are needing, odds are some hardware device could use it just
> like this.

Such as a shared on-chip memory that both a system control processor and 
the main processors can access. Of course, those also typically already 
have a comnunication channel.

But for a VM-hypervisor interface, why isn't one of the existing 
communications interfaces being used? One that is discoverable would be 
better than using DT.

Rob