[v2,0/2] xen/x86: implement NMI continuation as softirq

Message

Jürgen Groß Oct. 7, 2020, 1:30 p.m. UTC

Move sending of a virq event for oprofile to the local vcpu from NMI
to softirq context.

This has been tested with a small test patch using the continuation
framework of patch 1 for all NMIs and doing a print to console in
the continuation handler.

Version 1 of this small series was sent to the security list before.

Juergen Gross (2):
  xen/x86: add nmi continuation framework
  xen/oprofile: use set_nmi_continuation() for sending virq to guest

 xen/arch/x86/oprofile/nmi_int.c |  9 +++++++-
 xen/arch/x86/traps.c            | 37 +++++++++++++++++++++++++++++++++
 xen/include/asm-x86/nmi.h       |  8 ++++++-
 xen/include/xen/softirq.h       |  5 ++++-
 4 files changed, 56 insertions(+), 3 deletions(-)

Comments

Roger Pau Monne Oct. 15, 2020, 10:49 a.m. UTC | #1

On Wed, Oct 07, 2020 at 03:30:09PM +0200, Juergen Gross wrote:
> Move sending of a virq event for oprofile to the local vcpu from NMI
> to softirq context.
> 
> This has been tested with a small test patch using the continuation
> framework of patch 1 for all NMIs and doing a print to console in
> the continuation handler.
> 
> Version 1 of this small series was sent to the security list before.
> 
> Juergen Gross (2):
>   xen/x86: add nmi continuation framework
>   xen/oprofile: use set_nmi_continuation() for sending virq to guest

Apart from the comments in patch 1, I think this is a fine approach if
it allows us to restore to the previous state of the event lock.

I think we should be expecting a v3 with the nmi callback prototype?

Thanks, Roger.

Jürgen Groß Oct. 15, 2020, 10:52 a.m. UTC | #2

On 15.10.20 12:49, Roger Pau Monné wrote:
> On Wed, Oct 07, 2020 at 03:30:09PM +0200, Juergen Gross wrote:
>> Move sending of a virq event for oprofile to the local vcpu from NMI
>> to softirq context.
>>
>> This has been tested with a small test patch using the continuation
>> framework of patch 1 for all NMIs and doing a print to console in
>> the continuation handler.
>>
>> Version 1 of this small series was sent to the security list before.
>>
>> Juergen Gross (2):
>>    xen/x86: add nmi continuation framework
>>    xen/oprofile: use set_nmi_continuation() for sending virq to guest
> 
> Apart from the comments in patch 1, I think this is a fine approach if
> it allows us to restore to the previous state of the event lock.

This will not be enough to do that, but it is clearly removing a
potential deadlock.

> I think we should be expecting a v3 with the nmi callback prototype?

And using an IPI instead of a softirq, yes.


Juergen