[Qemu-devel] Unix signal to send ACPI-shutdown to Guest

[Qemu-devel] Unix signal to send ACPI-shutdown to Guest

[Andrew Wood]

Ive been looking into the possibility of using a unix signal to send an 
acpi shutdown request to a VM, and came across a posting on this l in 
March 1. See 
https://lists.nongnu.org/archive/html/qemu-devel/2017-03/msg04169.html


To summarise to refresh memories  there was a patch proposed using 
SIGINT followed by discussion as to why SIGINT wasnt ideal because it 
would cause regressions for anyone relying on the existing handling of 
SIGINT, and various other signals were considered including HUP & USR1 & 
USR2 all of which are already used for something else.

Is SIGPWR a suitable candidate, I have made a patch to use SIGPWR but as 
Im new to QEMUs internals Im not sure if there are any circumstances in 
which an alternative handler is registered for SIGPWR which might conflict?

Regards

Andrew

Re: [Qemu-devel] Unix signal to send ACPI-shutdown to Guest

[Stefan Hajnoczi]

[-- Attachment #1: Type: text/plain, Size: 1266 bytes --]

On Sat, Apr 28, 2018 at 11:59:41PM +0100, Andrew Wood via Qemu-devel wrote:
> Ive been looking into the possibility of using a unix signal to send an acpi
> shutdown request to a VM, and came across a posting on this l in March 1.
> See https://lists.nongnu.org/archive/html/qemu-devel/2017-03/msg04169.html
> 
> 
> To summarise to refresh memories  there was a patch proposed using SIGINT
> followed by discussion as to why SIGINT wasnt ideal because it would cause
> regressions for anyone relying on the existing handling of SIGINT, and
> various other signals were considered including HUP & USR1 & USR2 all of
> which are already used for something else.
> 
> Is SIGPWR a suitable candidate, I have made a patch to use SIGPWR but as Im
> new to QEMUs internals Im not sure if there are any circumstances in which
> an alternative handler is registered for SIGPWR which might conflict?

A quick grep of the Linux kernel code suggests that SIGPWR is very
rarely raised by the kernel.  It's only sent to init on s390 and Silicon
Graphics SN systems.  QEMU could use the signal on Linux without risk of
collision.

It's worth sending a patch.  Here are the patch submission guidelines:
https://wiki.qemu.org/Contribute/SubmitAPatch

Stefan

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 455 bytes --]